Bridges for Service Life Beyond 100 Years: Innovative Systems, Subsystems, and Components (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

11 C h a p t e r 3 This chapter summarizes the project findings that ultimately led to the development of the main product of the project, the Design Guide for Bridges for Service Life (the Guide). Categories of Bridge Service Life Issues The results of the literature search and other activities (see Chapter 2 for details) that were performed in Phase 1 to iden- tify and comprehend a broad range of service life issues and develop a general framework for service life design were grouped into the following nine categories: • Concrete durability; • Bridge decks; • Substructure; • Bearings; • Expansion joints, joints, and jointless bridges; • Fatigue and fracture; • Structural steel corrosion protection; • Steel bridge systems; and • Concrete bridge systems. The main reason for dividing the research findings into these nine categories was to develop an organized approach that would assist in identifying and prioritizing the knowledge gaps related to service life design and to establish the scope of further research to be conducted in Phase 2 of the study. The following sections briefly summarize the information gathered under each of these general categories. Concrete Durability Concrete bridge structures are subjected to severe conditions, yet to be cost-effective they must provide satisfactory perfor- mance for long periods of time. The greatest durability chal- lenge identified for these types of structures was corrosion of reinforcing steel, which is due mainly to cracking or poor concrete cover (or both) that allows moisture and chloride intrusion. In addition to the key factor of chloride solutions, hydroxyl ions and the concentration and movement of oxygen are important factors in the corrosion of reinforcing steel. Cracking can be caused by loads or environmental effects. In a new bridge structure, cracking is mainly attributed to moisture loss and temperature change. As a structure ages, corrosion, physical effects, and chemical reactions can also contribute to cracking. Concretes that have low permeability and minimal cracks have reduced infiltration of water and aggressive solutions, enabling extended service life. Design, material selection, and construction practices should be properly executed to ensure minimal cracking and reduced infiltration of aggressive solutions and to further resist damage due to freeze–thaw cycles, abrasion, and chemical attack. In design, proper drainage details can minimize ponding and prolonged exposure of bridge components to aggressive solutions, thus minimizing corrosion and other environmen- tal distress. Similarly, proper cover depth is essential to slow the intrusion of chlorides to the reinforcing steel. In material selection, a very efficient way of resisting the intrusion of chlorides is using supplementary cementitious material (SCM) (pozzolans and slag) to reduce concrete per- meability, along with a proper water–cementitious materials ratio (w/cm). When concrete is exposed to cycles of freezing and thawing, it must be properly air entrained, have sound aggregates, and have the maturity to develop sufficient strength for long-lasting service. In general, the total air content of the mixture is determined. However, the air void parameters that indicate the distribution and size of the voids are needed for proper protection against freezing and thawing. A minimum compressive strength of 4,000 psi is commonly specified in bridge decks to resist freezing and thawing and salt scaling. Concrete with hard siliceous aggregates and satisfactory compressive strength is needed to ensure a safe road surface with proper wear and skid resistance. Findings and Applications

12 Good construction practices, mainly ensuring the proper location of reinforcing steel for proper cover depth, consolida- tion, and curing, are essential for longevity. Curing is generally achieved by external treatment of concrete, such as covering by wet burlap and plastic or by misting. However, the provision of moisture for the inside of the concrete through internal curing is also very beneficial, especially for concretes with low w/cm. Proper consolidation will minimize entrapped air voids that can reduce strength and durability. For areas that are hard to reach and areas that are congested with reinforcement, self-consolidating concrete may be the solution. The use of performance specifications, rather than prescrip- tive specifications, should be adopted to ensure the quality needed. In performance-type specifications, the characteristics of the mixture are specified rather than the mixture itself, allowing the material supplier and contractor to be innovative in the selection of materials and in the proportioning of the mixture. The material supplier, contractor, and the user share responsibility, as the material supplier and contractor are responsible for the development of the product and the user for its acceptance. To extend concrete bridge service life in existing structures, preventive maintenance should be emphasized and proper repairs should be performed before extensive damage occurs and costly rehabilitation is required. It is very important to keep concrete dry for improved durability when it is exposed outdoors. Ponding of the surfaces that keeps concrete saturated should be avoided by proper design and preventive mainte- nance. The scope of repair or rehabilitation work can vary significantly, from sealing of cracks to application of overlays to replacement of large components such as bridge decks. Preventive maintenance may also include tasks as simple as washing the structure to eliminate chloride buildup. Selection of concrete materials and mixture proportions is usually based on empirical relationships between concrete mixtures and laboratory and field performance. This approach assumes that the concrete selected supports the desired ser- vice life for the structure. The use of deterioration models involves predicting service life by using calculations in which degradation mechanisms and reaction rates for those mecha- nisms are considered. However, many service life predic- tion methods focus on the effect of one degradation process. Experience has shown that degradation results either when one or more degradation processes are operative or from the interaction of environment and loads. These combined effects complicate the prediction of service life for both new and existing structures; in addition, the environmental factors and loads may not be well defined. Corrosion of reinforcing steel can cause serious and costly damage to reinforced concrete structures. Concrete is a brittle material that is prone to cracking and as such may not be able to adequately protect the steel. For additional steel protection, corrosion-resistant steel reinforcement, cathodic protection, and electrochemical chloride extraction (ECE) should be seri- ously considered to increase the service life of bridges. Synopsis of Literature Search and Other Available Information on Concrete Durability In general, the study identified factors affecting concrete durability that included environment, design practice, material properties and their proportioning, and construction prac- tices and workmanship. Other factors such as specifications, cracking, and selection of protective systems were also found to affect durability. The study summarized various protective systems that included corrosion-resistant reinforcement, certain admixtures, cathodic protection, and ECE. Corrective measures for service life extension, predicting service life of new structures and existing structures, and life-cycle cost analysis were also investigated. The literature search also identified various types of concrete distress and deterioration and the possible solutions. Factors aFFecting concrete Durability • Environmental influences—Exposure to harmful environ- mental factors such as chemicals, cycles of freezing and thawing, temperature and moisture changes, vibration, and impact can significantly reduce the service life of concrete. • Design practice—Design and specification issues such as reinforcement types and detailing, joints, long slender components, inadequate creep provisions, and poor drain- age can affect reinforced concrete deterioration. Service life is greatly increased by selecting appropriate design features that provide satisfactory drainage, sufficient cross section and reinforcement, proper cover, adequate stability, and small deformations. • Materials and proportioning—Selection of proper ingredients and using them in proper proportions also affects durability. Cementitious materials should be of the sufficient amount and of the right type; for example, portland cements are selected with low tricalcium aluminate (C3A) if concrete is exposed to sulfates, and low alkali contents when alkali–silica reaction (ASR) is a concern. Aggregates with poor particle shape and poor gradation would have higher water demand, paste content, and increased w/cm adversely affecting the durability. The w/cm should not be too low; a range between 0.40 and 0.45 is commonly used for bridge decks. This, in combination with the use of SCM (pozzolans or slag) provides for low permeability (Ozyildirim 1998, 1999). • Construction practices and workmanship—Attention to good construction practices is crucial for achieving long-term durability of reinforced concrete. Proper cover depth is essential to slow the intrusion of chlorides to the level of reinforcing steel. Proper consolidation is needed to eliminate

13 entrapped air voids that would reduce strength and facilitate the penetration of aggressive solutions. Proper curing enables the hydration reactions to continue and reduces volumetric changes that can cause cracking. Adding extra water at the job site adversely affects the strength and durability. A well-qualified and trained workforce and well-executed workmanship increases productivity, reduces material waste, and provides expected service life. Proper use of test methods are needed to ensure that quality concrete is achieved. speciFications Current specifications generally are of a prescriptive type requiring a recipe of ingredients. They restrict innovation by limiting the use of many possible combinations of materials. In performance-type specifications, the characteristics of the mixture are specified rather than the mixture itself, allowing the producer to be innovative in the selection of materials and in the proportioning of the mixture. The contractor and the user share responsibility: the contractor is responsible for development of the product, and the user for its acceptance. Compensation can be adjusted by inclusion of bonus and penalty depending on the quality of the product. Bonuses would encourage the production of quality material and adherence to proper construction practices that would lead to longevity and minimal distress, such as reduced cracking in the structure. Penalties or rejections would get the attention of the contractor and encourage the contractor to take corrective actions. cracking Cracking can be due to loads or environmental effects (ACI 224R 2001; TRB 2006). In new bridge structures, crack- ing is primarily attributed to moisture loss and temperature change. With age, corrosion, physical and chemical reactions, and loads can also contribute to cracking. To reduce cracking, shrinkage should be reduced; however, cracking also depends on other factors such as restraint, elastic modulus, creep, and tensile strength. Proper selection of materials and proportion- ing can reduce moisture- and temperature-related factors. For example, reductions in water content and paste content reduce shrinkage; large aggregate sizes also help to reduce shrinkage; and reduced cement contents and replacement of cements with supplementary cementitious materials contribute to reduction in heat generation. For reduced thermal gradients and reduced contraction, aggregates with a low coefficient of thermal expansion can be used. In addition, aggregates with low modulus of elasticity are desirable to ensure lower stresses for a given deformation. Proper proportioning and curing are essential to reducing cracking. Internal curing through the use of reservoirs, such as prewetted lightweight aggregates, supplies water throughout a freshly placed cementitious mixture. This water maximizes hydration and minimizes self-desiccation and its accompanying stresses, which may produce early-age cracking (Bentz and Weiss 2011). Fibers are also used to control cracking. protective systems in reinForceD concrete Several types of protective systems such as corrosion-resistant reinforcement, cathodic protection, ECE, sealers, membranes, and overlays are available to improve the durability of rein- forced concrete. However, it is very important to use the proper materials and procedures to achieve the longevity sought. • Corrosion-resistant reinforcement—Corrosion-resistant reinforcement provides more resistance to corrosion through a higher corrosion threshold value. Thus, high amounts of chlorides can penetrate to the level of reinforcement without causing significant damage to the reinforcing steel. These systems include epoxy-coated reinforcement, galva- nized reinforcement, titanium reinforcement, stainless steel re inforcement, stainless steel–clad reinforcement, nickel-clad reinforcement, and copper-clad reinforcement. • Admixtures for corrosion protection—Chemical admixtures can be added to concrete during batching to protect against corrosion of embedded steel reinforcement due to chlorides. There are two main types: corrosion-inhibiting (protection for steel) and physical-barrier (reducing the ingress of harmful solutions) admixtures. Some corrosion-inhibiting admixtures also act as physical-barrier admixtures. • Cathodic protection—When it was first introduced, cathodic protection was used mainly to prevent further corrosion after repair of damaged structures. Later, however, according to Polder (1998), cathodic protection was incorporated into new construction in an effort to prevent corrosion from starting. For cathodic protection, both impressed current and sacrificial anode, or galvanic systems, have been used successfully on bridges in the United States. Impressed cur- rent systems are used most often on bridge decks. According to Kepler et al. (2000), sacrificial anode systems are generally used in substructures. • Electrochemical chloride extraction—ECE is applied to concrete structures containing reinforcement to extract chlorides from the concrete. Studies indicate that ECE can successfully remove substantial amounts of chloride from contaminated concrete and leads to an increase in the pH of the concrete and repassivation of corroding reinforcing steel (Kepler et al. 2000). A list of projects published (Sharp et al. 2002) demonstrates that ECE is a promising bridge restora- tion alternative. The systems worked best on bridge decks, but most agencies cannot afford to close a bridge deck or limit traffic flow for 4 to 8 weeks at a time, which is required for treatment. With ECE, the possibility exists of initiating the ASR as a side effect, as alkali metal ions are rearranged and hydroxyl ions are generated at the reinforcement. Studies have indicated that the ASR can be mitigated by lithium (Velivasakis et al. 1997).

14 correction measures For service liFe extension Various repair techniques are available to extend the service life of a concrete structure. In repair, the deteriorated materials, components, or elements are replaced or corrected (ACI 546 2006). The selection of a proper repair material is one of many interrelated steps; equally important are surface preparation, the method of application, construction practices, and inspection. Initially, a condition evaluation is needed that includes design and construction documents, visual observation, and destruc- tive and nondestructive testing. For repair materials, durability rather than high strength is the desired property. Due to high early strengths and high early modulus of elasticity, conven- tional repair materials are prone to cracking. Conventional portland cement concrete is widely used to repair structures. It is readily available and economical. However, in aggressive environments, portland cement concrete modified with silica fume, acrylics, styrene–butadiene latex, or epoxy may be needed for longevity. The specified repair material must be dimension- ally compatible with the existing concrete substrate to minimize the potential for failure. In repairs, due to traffic flow and time constraints, rapid-setting materials are sought that provide short setting time and high early compressive strength. Durability should be carefully checked. preDicting service liFe oF reinForceD concrete Many service life prediction methods focus on the effect of one degradation process (ACI 365 2000). Experience has shown, however, that degradation results when one or more degradation processes are operative or from the interaction of the environment and loads (Hookham 1990). Service life can be limited by many factors, such as the presence of chlorides, cycles of freezing and thawing, critical saturation, volume changes due to moisture and temperature, and to a lesser degree, by carbonation or aggressive chemicals, such as acids and sul- fates. For large elements and for those kept wet by contact with the ground or water, alkali aggregate reaction and delayed ettringite formation (high early temperatures) can be a prob- lem. Temperatures should be kept down during early curing, and supplementary cementitious materials should be used. Service life is also influenced by mechanical loads, such as fatigue, vibration, and local overloads. Methods that have been used for predicting the service lives of construction materials include estimates based on experience, deductions from performance of similar materials, accelerated testing results, mathematical modeling based on the chemistry and physics of expected degradation pro- cesses, and applications of reliability and stochastic concepts (Clifton and Knab 1989). Several mathematical models have been developed to predict the service life of concrete subjected to degradation processes such as corrosion, sulfate attack, leaching, and freeze–thaw damage (Clifton 1991; Williamson et al. 2007). preDiction oF remaining service liFe The methods for predicting the remaining service life of existing concrete structures are basically the same as those for new structures (ACI 365 2000). Most of the reported work on predicting remaining service lives of reinforced concrete structures has dealt with corrosion of the concrete reinforcement. Two major prediction approaches that have been pursued are the modeling approach and cor- rosion measurements (ACI 365 2000). Extrapolation techniques can be used to predict the remain- ing service life of a concrete structure. Based on a set of known points obtained by inspection or tests, new points can be pre- dicted, but this technique will often lead to a less meaningful result with greater uncertainty. commercial soFtware to preDict service liFe Commercial software programs are available for predicting service life. LIFECON is the Integrated and Predictive Life- Cycle Maintenance and Management Planning System, which includes life-cycle cost, life-cycle performance, life-cycle ecol- ogy, and health and comfort (Soderqvist and Vesikari 2003). STADIUM can be used to follow the transport of ions and liquids in reactive porous media. The successor model to STADIUM, SUMMA, is under development and will include a number of enhancements, including an improved user inter- face (Heffron 2007). SUMMA will also be able to handle cracks in concrete, representing an important refinement because concrete cracking is very common and facilitates pen- etration of chlorides. Life-365 is a standard model developed for predicting the service life and life-cycle cost of reinforced concrete exposed to chlorides. The DURACON software that can predict chloride profiles for future ages is available for use at both the design and construction phase. DIFFUZON combines diffusion transport and local chem- ical equilibrium for all species of a cementitious material. Multilayer-systems commercial software named FEMMASSE MLS incorporates pertinent physical and mechanical models governing the behavior of the structural elements under consideration. liFe-cycle cost analysis The National Institute of Standards and Technology (Fuller and Petersen 1995) defines life-cycle cost as “the total dis- counted dollar cost of owning, operating, maintaining and disposing of a building or a building system” over a period of time. Life-cycle cost analysis is an economic evaluation tech- nique that determines the total cost of owning and operating a facility over a period of time. types oF Distress anD Deterioration Distresses in concrete structures lead to deterioration that can cause premature failure and require costly repairs or reconstruction. There are three basic visual symptoms of

15 distress in a concrete structure: cracking, spalling, and dis- integration. The effects of deterioration may not be manifested visually. For example, internal cracking may not be visually observed for a long time even though the structural integrity may be reduced to an unacceptable level. Deterioration of material properties mostly sets in under the combined action of internal and external factors. Internal factors are those that determine a material’s quality (i.e., the way it is made, placed, and cured). The quality is governed by characteristics such as shrinkage, creep, and thermal behavior, which are inherent in its nature. External causes of deterioration are broadly grouped as physical, chemical, or mechanical. There are also the biological factors; however, they are not considered to be major concerns. The main physical factors are fluctuations of moisture content, temperature, carbonation, freezing and thawing, salt crystallization, wear, and fire; the main chemical factors are aggressive gases and liquids causing corrosion, alkali aggregate reactions, sulfate attack, and acid attack; the main mechanical factors are load, friction, and vibration. Analysis of DOT Survey Eighteen states and one Canadian province responded to the department of transportation (DOT) survey submitted by the R19A research team. A summary of the survey results for the various numbered requests and questions related to concrete durability follows. 1. Prioritize the following durability-related distresses that are most commonly found in existing bridge structures. Table 3.1 shows the range of rating values that cor re- sponded to various levels of distress frequency for the different structural elements. The rating scale considered was from 1 to 5. The most common deck-related stress was the cor- rosion of reinforcement, rated as 1.6. Freezing and thawing (2.6) and abrasion and wear (2.9) were found to be a prob- lem occasionally. ASR (3.9) was considered rare in occur- rence, and sulfate attack (4.7) was considered not to be a problem (see Figure 3.1). For beams, the corrosion of reinforcement (2.2) was occasionally considered to be a problem. Freezing and thawing (3.1) and ASR (3.7) were rarely a problem. Sulfate attack (4.8) and abrasion and wear (4.9) were considered not to be a problem (see Figure 3.2). The corrosion of reinforcement (2.1) in piers was considered to be a problem occasionally. Freezing and thawing (3.2), ASR (3.6), and abrasion and wear (4.3) were rarely a problem, and sulfate attack (4.7) was considered not to be a problem (see Figure 3.3). Finally, at the foundations, the corrosion of reinforce- ment (3.2), freezing and thawing (4.1), and ASR (4.1) were rarely found to cause any problem. Sulfate attack (4.5) and abrasion and wear (4.9) were considered not to be a problem (see Figure 3.4). Table 3.1. Rating Scale Considered in Study Rating Range Frequency 1.00–1.67 Routinely found 1.67–3.00 Occasionally found 3.00–4.33 Rarely found 4.33–5.00 Considered not found Figure 3.1. Survey results for deck-related distresses. Deck 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% corrosion of reinforcement freezing and thawing ASR sulfate attack abrasion, wear most sometimes least

16 Figure 3.2. Survey results for beam-related distresses. Beams 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% corrosion of reinforcement freezing and thawing ASR sulfate attack abrasion, wear most sometimes least Figure 3.3. Survey results for pier-related distresses. Piers 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% corrosion of reinforcement freezing and thawing ASR sulfate attack abrasion, wear most sometimes least Figure 3.4. Survey results for foundation-related distresses. Foundations 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% corrosion of reinforcement freezing and thawing ASR sulfate attack abrasion, wear most sometimes least

17 2. How do you detect corrosion? Eighty-nine percent of the states surveyed answered this question, with the exceptions of Pennsylvania and Vir- ginia. Fifty-eight percent of those states, which included California, Iowa, Michigan, Missouri, Nebraska, New Mexico, New York, Texas, and Washington, indicated that they perform visual inspections during which they usu- ally look for signs of concrete deterioration such as spall- ing, delamination, and cracking. In addition to visual inspection, 6% perform chloride content tests, another 6% do half-cell tests, 6% take cores to determine the loss of section, and 18% do both chloride and half-cell testing (see Figure 3.5). The survey results show that 40% did not mention explicitly that they perform visual inspection; 28% perform both chloride and half-cell tests, and only 12% perform half-cell testing alone. In addition to chloride and half-cell testing, Ontario mentioned linear polarization and Galvapulse as addi- tional tests. The threshold for chloride content was provided by only 26% of the states. The values varied from 1 lb/yd3 in Kansas and 1.4 lb/yd3 in Illinois to 2 lb/yd3 in California, Missouri, and Texas. Also, 26% provided information about the threshold for half-cell potential, with values varying from -200 mV in New York, -290 mV in Missouri, and -350 mV in California, Hawaii, and Kansas. 3. Classify as routinely, occasionally, rarely, not accepted, the type of reinforcement used in your past bridge projects. Table 3.2 shows the rating value for different types of rein- forcement. The considered rating scale was 1 = routinely, 2 = occasionally, 3 = rarely, and 4 = not used. Figure 3.6 shows the survey results for types of reinforcement used. 4. Indicate the frequency of encountering corrosion that requires corrective action of prestressing strands in precast concrete products. In spite of the different nomenclature adopted from each state, it was possible to distinguish seven types of structures. Prestressed I-girders were the most common; these were either occasionally (in Hawaii, Illinois, Iowa, New Mexico, New York, Ontario, and Pennsylvania) or rarely (in Arizona, Michigan, Missouri, Oregon, Texas, and Washington) found to need some corrective action. The second structures most mentioned were prestressed box beams, which were found to have some corrosion issues frequently (Illinois and New York), occasionally (New Mexico), and rarely (Missouri and Washington). This was followed by precast void slabs, which were either rarely (Arizona, Oregon, and Washington) or never (New Hampshire) found to have corrosion problems. The fourth was prestressed T-girders, for which there was occasion- ally (Hawaii and New York), rarely (Illinois), and never (New Hampshire) any need for corrective action due to corrosion. Post tensioned (Arizona, Oregon, and Texas) and prestressed double T (Missouri and Nebraska) struc- tures were occasionally to rarely found to have problems. The last structures mentioned (New Hampshire and Texas) were deck panels, without any problem reported to date. Figure 3.7 shows the survey results for corrosion encountered in prestressing strands. A high number of N/A answers were provided by the states. 5. What measures are taken to extend the life of girders with corroded pretensioned, prestressing strands? The answers varied from doing nothing to total replace- ment of the girder. Cleaning and patching were usually mentioned. Epoxy injection, flexible topical treatment, and waterborne sealer were some alternatives mentioned. Figure 3.5. Survey results for corrosion detection. Corrosion Detection 6% 6% 6% 18% 24% 12% 28% Inspection Only Inspection, Chloride and Half-cell Inspection and Half-cell Inspection and Chloride Inspection and Other Chloride and Half-cell Half-cell Only

18 Table 3.2. Rating Value for Different Types of Reinforcement Type Rating Routinely (%) Occasionally (%) Rarely (%) Not Used (%) Considered Routinely Used Epoxy-coated reinforcement 1.2 89 5 0 5 Prestressing strands (longitudinal) 1.3 89 0 0 11 Carbon steel (black) 1.4 84 5 0 11 Considered Occasionally Used Posttension tendons (longitudinal) 2.1 37 32 21 11 Prestressing strands (transverse) 2.2 37 26 16 21 Posttension tendons (transverse) 2.2 21 47 21 11 Considered Rarely Used Stainless steel 2.7 11 16 63 11 Galvanized steel 3.1 5 16 42 37 Black with sacrificial anode 3.1 5 11 53 32 Carbon-fiber reinforcement 3.3 0 16 42 42 Black with impressed current 3.3 5 5 42 47 Glass-fiber reinforcement 3.5 5 5 26 63 Considered Not Used Chromium alloy (A-1035) 3.5 5 5 21 68 Galvanized and epoxy coated 3.6 5 0 21 74 Stainless steel clad 3.6 5 0 21 74 Epoxy-coated prestressing strands 3.7 0 5 21 74 Chromium alloy (A-1035) 3.7 0 5 21 74 Carbon-fiber reinforcement prestressing 3.7 0 0 26 74 Galvanized prestressing strands 3.8 0 0 16 84 Other: MMFX 3.9 0 0 5 95 After rehabilitation, some states monitor the structure until it is replaced. Some states affirmed that they usually find cracks at the ends of girders, allowing water to pen- etrate and corrode the strands. Preventive measures such as increasing the overlay cover, keeping water off by adding deck grooves, using drains, and either sealing or eliminating joints were also mentioned. Cathodic protection with a low current was mentioned, as well. 6. Indicate the frequency of encountering corrosion that requires corrective action of posttensioning tendons structures. Corrosion is rarely found at most posttensioned struc- tures (see Figure 3.8). The survey showed no difference in the answers for different structures. Only Oregon had a different answer for the substructure, which had a slight effect on average results. N/A answers might be explained by the fact that some states affirmed that they do not have many posttensioned structures to experience distress. 7. What measures are taken to extend the life of girders with corroded posttensioning tendons? Some states affirmed that they do not have very many structures in this category. Therefore, they have not experienced corrosion yet. Measures are similar to pre- stressed structures (see Question 5). In regard to the question “Where corroded tendons have been found, indicate the percent repaired or replaced,” none of the surveyed states responded. The exceptions were New York, which stated that 100% were replaced, and Washington, which stated that 50% were repaired and 50% were replaced. 8. How are the following durability related issues addressed in structural design and detailing for new construction? Corrosion of reinforcement—Eighty-four percent of the states (excepting New Mexico, Texas, and Washington) mentioned additional cover, and 74% of the DOTs surveyed (including Arizona, California, Hawaii, Illinois,

19 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% ca rbo n ste el ep ox y c oa ted sta inle ss ste el ga lva niz ed ste el ga lva niz ed an d e po xy co ate d sta inle ss ste el c lad re inf . ch ro m ium allo y (A - 10 35 ) pre str es sin g s tra nd (lon g.) pre str es sin g s tra nd (tra ns v. ) ep ox y c oa ted pre str es sin g s tra nd ga lva niz ed pre str es sin g s tra nd cfr p p re str es sin g po st- ten sio n ten do n (lon g.) po st- ten sio n ten do n (tra ns . ) ch ro m ium allo y (A - 10 35 ) bla ck wit h i mp re ss ed cu rre nt bla ck wit h s ac rifi cia l a no de gla ss fib er- re inf . (GF R) ca rbo n- fib er re inf. (CF R) m m fx routinely occasionally rarely not used Figure 3.6. Survey results for type of reinforcement used. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% P/S I-girders P/S Box Beam Precast Slab P/S T-girders Post- tensioned P/S Double T Deck Panels Frequently Occasionally Rarely Never N/A Figure 3.7. Survey results for corrosion encountered in prestressing strands. N/A  not applicable.

20 Iowa, Kansas, Michigan, Missouri, Nebraska, New Mexico, New York, Ontario, Oregon, and Washington) mentioned epoxy-coated reinforcement as the protective measure taken to prevent corrosion of reinforcement. Corrosion inhibitors, silica fume, and fly ash are some of the supplementary cementitious materials mentioned by 26%. Drainage (mentioned by 21%) includes mea- sures such as proper design, grooving, removing joints, and placing catch basins high to keep water off. Only 16% mentioned high-performance concrete (HPC) and 11% w/cm ratio. In Figure 3.9, “ Others” represents 21% and includes measures such as cathodic protection, stainless steel, or galvanized rebar and glass fiber– reinforced polymer (GFRP) reinforcement. Freezing and thawing—Thirty-seven percent addressed freezing and thawing issues with air-entrainment specification. Twenty-six percent mentioned drainage and twenty-one percent mentioned sealer. “Others” included epoxy-coated reinforcement, cover, HPC, SCM, shallow foundation, and membranes; these were mentioned by 34%. Eleven percent did not consider freezing and thawing to be a problem. Figure 3.10 shows the survey results for freezing and thawing. Alkali–silica reaction—ASR is addressed by 42% of the states by using selected source of aggregate. SCM, such as pozzolan and fly ash, is used by 26%. “Others” included drainage, sealer, and air entrainment (11%). ASR is not considered to be an issue by 42% of the states. Figure 3.11 shows the survey results for addressing ASR. Sulfate attack—Sulfate attack is addressed by 16% of the states either by using Type V cement (low C3A), aggre- gate selection, or other measures such as HPC, drainage, and sealers. Fifty-eight percent consider sulfate attack an issue that should be addressed (see Figure 3.12). Figure 3.8. Survey results for corrosion encountered in posttensioning tendons. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Precast Concrete Cast-in-place Concrete Superstructure Substructure Frequently Occasionally Rarely Never N/A Figure 3.9. Survey results for addressing corrosion of reinforcement. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Add cover Epoxy-coated rebar Admixtures Drainage HPC w/c ratio Others

21 Figure 3.10. Survey results for addressing freezing and thawing. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Air entrainment Drainage Sealer Others None Figure 3.11. Survey results for addressing ASR. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Aggregate Admixture Others None Figure 3.12. Survey results for addressing sulfate attack. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cement type Aggregate Others None

22 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wearing surface Aggregate Others None Figure 3.13. Survey results for addressing abrasion and wear. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Epoxy-coated rebar Admixtures HPC Inhibitor Others Figure 3.14. Survey results for addressing corrosion of reinforcement. Abrasion and wear—Fifty-three percent of the respondents considered measures such as extra cover, overlay, and wearing surface to address abrasion and wear. Sixteen percent mentioned either hard aggregate selection or other measures such as membranes, SCM, and material selection. Twenty-one percent do not consider abrasion and wear to be a problem (see Figure 3.13). 9. How are material selection and proportioning addressed in new construction for the following durability-related distresses? Corrosion of reinforcement—Fifty-eight percent of the states mentioned epoxy-coated reinforcement and 53% mentioned SCM (fly ash and slag) and chemical admixture (i.e., shrinkage-reducing admixture) as measures they have taken. HPC was mentioned by 37%. Corrosion inhibitor was mentioned by 32%. The cate- gory “Others” included measures such as w/cm, stainless steel, cathodic protection, GFRP, carbon fiber–reinforced polymer, and water repellent; these measures were mentioned by 37% (see Figure 3.14). Freezing and thawing—Eighty-nine percent of states mentioned they address freeze–thaw distresses with air entrainment. Twenty-six percent mentioned w/cm ratio, and only 11% mentioned HPC with low perme- ability. “Others” (mentioned by 16%) included mea- sures such as uniform aggregate to minimize paste content, low-absorption aggregate and silane water repellents. Eleven percent did not address freezing and thawing as it is not a problem (see Figure 3.15). Alkali–silica reaction—SCM (fly ash, pozzolan, silica fume, and slag) was mentioned by 53%. Prequalified sources of aggregate were mentioned by 32% of the states. Twenty-one percent mentioned others such as state specification, HPC with low permeability, and drainage.

23 Twenty-one percent did not take any measures and do not consider it to be an issue (see Figure 3.16). Sulfate attack—Sulfate attack is addressed by 16% of the states by using Type V cement (low C3A). Eleven percent mentioned either pozzolan or “Others,” which included low-permeability concrete or state specification. The great majority, 63%, do not consider sulfate attack to be an issue (see Figure 3.17). Abrasion and wear—Hard aggregate is considered by 37% of the states. HPC is considered by 16%. “Others” (mentioned by only 11%) included measures such as either early curing or SCM (silica fume). Twenty-six percent of the states had no measures to address abrasion and wear (see Figure 3.18). 10. How are the following construction practices addressed in new construction for durability of concrete? Delivery and pumping—As shown in Figure 3.19, field test- ing, such as slump and air content determination, is performed by 47% of the states. Some states sample from the end of the pump hose. Inspection, including good practices, close monitoring, limiting free fall, and having a short interval between delivery and use (ensure continuous supply), was mentioned by 37%. “Others” (mentioned by only 11%) included self-consolidating concrete or adding superplasticizer, state specification, and the amount of air entrainment. The method of conveyance was not considered by 16% of the states. Consolidation—Inspection to ensure good practices was mentioned by 32%. Proper vibration by using an auto- matic, timed vibrator and vibration applied at vertical point was mentioned by 26%. “Others” included self- consolidating concrete, state specification, and ACI standard; these were mentioned by 42%. Consolidation was not considered by 21% (see Figure 3.20). Curing—Wet cure and application (burlap or plastic cover) were mentioned by 74% of the states. The curing time 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Air entrainment w/c ratio HPC Others None Figure 3.15. Survey results for addressing freezing and thawing. Figure 3.16. Survey results for addressing ASR. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Admixture Aggregate Others None

24 Figure 3.17. Survey results for addressing sulfate attack. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cement type Pozzolan Others None Figure 3.18. Survey results for addressing abrasion and wear. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Aggregate HPC Others None Figure 3.19. Survey results for practicing related to delivery and pumping. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Test Inspection Others None

25 varied from 4 to 14 days, but 7 to 10 days was the most common duration. Inspection to ensure proper curing procedures was mentioned by 16%. “Others” (men- tioned by 42%) included shorter time between finishing operation and cure application, uses of membrane, and state specification (see Figure 3.21). 11. Which form of specification (either prescriptive or performance) do you use for durability? Twenty-six percent of the states (Arizona, California, Hawaii, Illinois, and Virginia) mentioned either state specification or a contact. Hawaii will move to perfor- mance in the future. Iowa uses prescriptives for standard decks and performance for HPC. Kansas uses only per- formance. Michigan, New York, Texas, and Vermont use both, with limited performance specifications. Missouri uses more prescriptive but is moving to performance. New Mexico uses only performance and requires a detailed demonstration from the concrete supplier. New Hampshire uses prescriptive except for decks, for which the state specifies air entrainment, permeability, strength, and qual- ity control–quality assurance. Ontario looks mostly for end result with some prescribed limits. Oregon uses only prescriptive, and Pennsylvania uses only performance specifications. 12. How are the following durability-related distresses corrected in existing structures? Corrosion of reinforcement—Repair procedures (removing unsound concrete, cleaning, and removing and replac- ing the reinforcement) and patching were mentioned by 89% of the states. New Mexico said corrosion is not a problem, and Virginia did not answer, representing 11%. Beyond repairing, cathodic protection was mentioned by 21% of the states. Other measures (“Others”), such as corrosion inhibitor, zinc anode, and breathable sealers, were mentioned by 21% (see Figure 3.22). Freezing and thawing—As shown in Figure 3.23, repair procedures and water-repellent sealer were mentioned by 26% of the states. Overlay was mentioned by 21%, and 37% either reported that freezing and thawing is not a problem or did not answer. Alkali–silica reaction—As shown in Figure 3.24, repair and sealer application were mentioned by 37% of the states. Overlay was mentioned by 11%. “Others” (mentioned by 38%) included either exclusion of water moisture (drainage), confinement of the structure, or SCM. Thirty-two percent either did not answer or do not consider ASR to be a problem. Sulfate attack—Eighty-four percent of the states affirmed that either sulfate attack is not an issue or did not answer. Illinois mentioned overlay, scarification, and deck sealer; Michigan mentioned replacement; and Pennsylvania mentioned concrete removal and patching using vitrified clay liner plates. Figure 3.20. Survey results for practices related to consolidation. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Inspection Vibration Others None Figure 3.21. Survey results for practices related to curing. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wet curing Inspection Others

26 Figure 3.22. Survey results for correcting corrosion of reinforcement. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Repair / Patch Cathodic Protection Others None Figure 3.23. Survey results for correcting freezing and thawing. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Repair Sealer Overlay None Figure 3.24. Survey results for correcting ASR. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Repair Sealer Overlay Others None

27 Abrasion and wear—Fifty-eight percent of the states addressed the abrasion and wear issue by using over- lay, and 42% do not have abrasion and wear as an issue or did not answer. 13. If you have damage due to reinforcing steel corrosion, what is your method of repair? Ninety-five percent of the states surveyed affirmed that cleaning and patching are their method of repair for decks, except Pennsylvania, which indicated that decks are pro- grammed for replacement. The barrier is replaced by 89%. Oregon and Vermont did not mention any method. Beams are repaired by cleaning and patching by 89% of the respondents. Some states mentioned either replacing the reinforcement or the whole beam. New Hampshire and Vermont did not mention any method of repair. Substructure-related problems are addressed by 95%, with cleaning and patching considered as their method of repair. Table 3.3 shows the life of these component repairs. 14. Does your state use any models to predict the service life of structures in new construction? Ninety-five percent of the surveyed states do not use any models; Iowa was the sole exception. They affirm that epoxy-coated reinforcement is projected to reach 50 years before deck repair and overlay are needed. Hawaii has a model, but it is not used frequently. 15. Does your state or province use any models to predict the service life of existing structures (remaining life) based on condition surveys and concrete testing? Seventy-nine percent of the surveyed DOTs do not use any models. Michigan, Nebraska, Ontario, and Oregon, which represent 21%, indicated such use. Hawaii indi- cated interest for use in the future. 16. Please list any completed or ongoing research regard- ing concrete durability or service life of bridges. New Hampshire and Virginia did not answer. Eighty- nine percent provided contact information, such as the name of the researcher, contact number, link, and some detail about the project. Bridge Decks The primary function of a bridge deck is to provide a safe riding surface for traffic and direct structural support of wheel loads. In most regions of the United States, deteriora- tion of bridge decks has been a major cause of reported bridge deficiencies. To provide longer service life, the integrity of the bridge deck must be maintained. Degradation can result in loss of capacity or unsafe driving conditions, or both. With sufficient degradation, the reduction in carrying capacity can lead to failure, and unsafe driving conditions can lead to deadly traffic accidents. Most of the bridge decks in current use are made of cast- in-place (CIP) reinforced concrete. Other systems include full-depth CIP concrete slab superstructure, precast concrete deck panels, exodermic decks, steel orthotropic decks, fiber- reinforced polymer decks, and adjacent member superstructure systems. The types of deterioration experienced in bridge decks are mainly scaling, mortar flaking, freeze–thaw damage, abrasion damage, alkali–aggregate reactivity, excessive crack- ing and spalling, and delaminations due to corrosion of the reinforcement. A literature search was performed to establish the principal issues relating to bridge deck durability. This information was supplemented by highway department survey results, which identified additional research and procedures for cur- rent practice in the construction of bridge decks under their authority. Degradation of service life due to corrosion caused by cracking of bridge deck elements is identified as a significant issue to be addressed. The intrusion and concentration of chlorides allowed by deck cracking is identified as a concern nationwide, particularly in areas where deicing salts are used, and in marine environments. A dual strategy of corrosion protection incorporating improved concrete quality and rein- forcement protection systems has been adopted by most high- way agencies for their bridges. This strategy often consists of a low-shrinkage, low heat of hydration concrete with epoxy- coated reinforcement. Several systems present opportunities to have long service life. These include CIP concrete bridge decks using stainless steel reinforcing, orthotropic steel bridge decks with proper fatigue detailing and plate thickness, full-depth precast con- crete deck panels with appropriate joint details and closure of gaps over shear connectors, and adjacent member bridges with appropriate shear key joint design. Table 3.3. Component Repair Due to Corroded Steel Component Minimum (years) Maximum (years) Average (years) SD (years) Provided answer (%) Deck 1 40 17.2 9.20 84 Barrier 1 40 16.8 11.6 53 Beam 5 50 17.4 11.2 63 Substructure 5 30 16.3 6.20 74 Note: SD = standard deviation.

28 Description and Discussion of Bridge Deck Types The predominant bridge deck system in the United States consists of CIP reinforced concrete. Conventional concrete decks, either CIP or precast, were the major focus of this study; however, the service life of other bridge deck systems was also investigated. The service life of the following bridge deck types was studied: • CIP concrete bridge decks; • Full-depth CIP concrete slab superstructures; • Segmental concrete superstructure systems; • Precast concrete deck panels; • Steel orthotropic bridge decks; • Fiber-reinforced polymer (FRP) bridge decks and super- structures; and • Adjacent member superstructure systems. A summary of the information on the listed bridge deck systems is reported in the Guide. Causes of Bridge Deck Deterioration In most regions of the United States, deterioration of bridge decks has been a major cause of bridge deficiency. This dete- rioration has been shown to substantially increase with the use of deicing salts on bridges. Although deterioration occurs to concrete, the principal cause for the deficiencies is a direct result of bridge deck cracking, which significantly reduces the chloride intrusion protection of the reinforcement provided by concrete cover. Early detection and subsequent sealing of cracks, such as by epoxy injection, have been shown to increase durability of bridge decks; however, this procedure is not considered a primary defense strategy for addressing the cracking issue. Elimination of cracking is the primary goal. For CIP bridge decks on longitudinal stringers, one of the primary causes of cracking is the restraint developed when the composite deck is locked in with the supporting girders, occurring as the deck concrete reaches initial set. precast concrete Deck panels Precast deck systems provide many benefits over traditional CIP deck systems in bridge construction. The quality of precast deck systems is generally superior to field-cast concrete decks. The variability of construction due to environmental condi- tions is eliminated in the plant through the use of consistent casting operations and curing techniques. orthotropic steel briDge Deck Historically, orthotropic steel bridges have not been problem free. Their design and construction is more complex than con- ventional bridge construction, and they demand special mea- sures for routine inspection and maintenance. Fatigue cracking has been observed more frequently in such bridges throughout the world due to the numerous complicated welded details subject to complex stresses. Early analytical tools were limited in their ability to quantify the stress states at these details, limit- ing the experimental fatigue database. In addition, the fatigue performance of many of these details is highly sensitive to the construction techniques. Detailing practices have relied heavily, and continue to rely, on experience gained through trial and error. One clear advantage to the orthotropic steel deck is that it is a highly redundant system, and cracking is typically con- sidered a mere nuisance to be observed rather than a serious threat to the strength or integrity of the structure. Wearing sur- faces have also exhibited performance problems with cracking, rutting, shoving, delamination, or some combination of these stresses, which have often resulted in excessive maintenance and resurfacing. The delamination of the wearing surfaces is generally attributed to the small thickness of the top plates used in orthotropic decks. The corrosion resistance of orthotropic steel decks has been very good overall. The top side is protected by the wearing surface, and the bottom side can be protected with a conven- tional paint system. Like any steel bridge structure, it may require regular maintenance in terms of repainting. However, the coating on the underside of the deck can last a very long time if it is not subjected to direct saltwater spray. Orthotropic decks are typically made continuous, without joints, for extended lengths, which minimizes potential locations for water penetration. The individual ribs are typically sealed with end plates that prevent moisture from entering the interior of the rib. Outside the United States, a common approach has been to use a fully closed box girder cross section and employ an in-service dehumidification system on the interior to essentially eliminate the possibility of corrosion; thus, there is no need for an interior paint system. However, this system is expensive from an operational standpoint. One of the primary advancements for orthotropic bridges is in the technique for engineering analysis and design. For orthotropic bridges to become cost-effective in the United States, standardization is the goal. Fiber-reinForceD polymer briDge Deck Because FRP application is an emerging technology, the per- formance history for FRP bridge decks and superstructure systems is very limited and does not lend itself to credible projections of extended service life. Deterioration of these structures has included delamination of the FRP material (Alampalli et al. 2004) and cracking and delamination of the wearing surface. Causes for wearing surface cracking and delamination from FRP material may be related to material property differences (e.g., coefficient of thermal expansion) and response of the flexible structural system. The cause for the delamination of the FRP material was not reported.

29 aDjacent member superstructure systems Most of the deterioration of bridges with adjacent member superstructure systems occurs due to a breakdown of grout or concrete in the keyway between the girders, allowing water and chloride intrusion into the joints. It is possible that water may also leak into the voids inside the adjacent box members. In addition to creating a potential for deterioration due to corrosion and freeze–thaw damage, accumulation of water could create additional weight unaccounted for in the design. Similar experiences with the breakdown of these joints throughout the country give credence to the idea that the trans- verse design and detailing of these systems requires additional thought to ensure that the system transfers the required loads across the joint and remains watertight. Failure to transfer loads across the joint could potentially overstress the individual member, resulting in a lower-rated capacity. Little guidance is provided in the AASHTO LRFD Bridge Design Specifications (2004, 2010b) (LRFD Specifications) relating to the moments and the shears that are transferred across this joint. Protection of Existing Bridges Bridge protection systems to extend the life of existing struc- tures include epoxy injection into cracks; the use of coatings, such as sealers, membranes and overlays; cathodic protection; and chloride extraction systems. Although some of these sys- tems perform satisfactorily, their life is limited, and they are certainly not maintenance free. Additionally, their effectiveness in delaying damage is dependent on whether the conditions around the elements being protected have already begun to deteriorate. Detection is critical to providing adequate advance warning of deterioration to program and execute repair strategies. The following subsections describe approaches for pro- tecting bridge decks for extending service life. sealers Sealers are expected to minimize the intrusion of aggressive solutions into concrete. Today, silanes are popular sealers, followed by siloxanes, silicates, epoxies, gums, acrylics, ure- thanes, chlorinated rubber, silicones, and vinyls (NCHRP Syn- thesis 209). Sealers have been difficult to assess and have been reported to have a wide range of performance (NCHRP Syn- thesis 333 2004). membranes Membranes are perhaps the most effective way to protect bridge decks and obtain long service life. Agencies in Europe do not have the service life issues with bridge deck that U.S. agen- cies do. The main reason is the use of membranes. The quality and extent of service life that could be achieved by use of membranes depends on the quality of the material used and installation methods. Early experiences with membrane in the United States in the 1980s were unsuccessful, mainly because of lower-quality material and poor installation techniques. overlays Concrete overlays create a low-permeability protective layer over the conventional concrete on bridge decks (NCHRP Synthesis 333 2004). An overlay serves as a barrier to chloride ions, therefore increasing the time required for the concen- tration of chloride ions, at the level of the reinforcement, to reach the threshold for corrosion. Low-permeability overlays also decrease water penetration into a structure, allowing it to dry out, which reduces chloride ion mobility. Overlays can be applied to new decks or as a rehabilitation method on exist- ing decks (Kepler et al. 2000); however, overlays have limited effectiveness when applied to existing decks. If chloride ions are already present in the deck when the overlay is placed, then the only protection that the overlay can offer is a decrease in moisture infiltration (Sherman et al. 1993). The most common types of overlay include • Latex-modified concrete overlays; • Low-slump dense concrete overlays; • Silica fume concrete overlays; • Polymer concrete overlays; and • Very-high-early-strength concrete overlays. cathoDic protection Cathodic protection systems are designed to protect the reinforcement. The potential of the reinforcement is shifted in the negative direction either by impressed current or sac- rificial anodes. If steel is made cathodic, corrosion will stop (Virmani and Clemeña 1998). Impressed current is most com- mon and uses a titanium mesh anode in a concrete overlay (Kepler et al. 2000). Some states have reported successful use of the cathodic protection system, but others have cited difficulties with reli- ability and maintenance of the system. Cathodic protection systems have not been proven to be maintenance free or cost- effective (NCHRP Synthesis 333 2004). electrochemical chloriDe extraction ECE can be employed when high levels of chlorides are present at the level of reinforcing steel (Virmani and Clemeña 1998). This procedure enables the extraction of chlorides from concrete and leads to an increase in the pH of concrete that repassivates the steel. The life expectancy of a treated reinforced concrete is not known. The system is described in detail below under “Research Categories.” Analysis of DOT Survey The results of the survey conducted by the R19A research team indicate that the predominant type of bridge deck used

30 by the respondents is the CIP normal-weight concrete deck on stringers, followed by full-depth CIP concrete slab super- structures and then precast concrete slab–box beam super- structures (see Figure 3.25). Other systems, such as lightweight concrete, full-depth precast concrete on stringers, steel grid (exodermic), and steel orthotropic decks are rarely used. Three states indicated occasional to routine use of timber decks. Several states indicated use of precast concrete deck panels in composite and noncomposite configurations. Several respondents provided estimates of expected service life in years for the various bridge deck types identified, as shown in Figure 3.26. The service life varied significantly from state to state for many of the categories (e.g., 30 to 75 years for CIP normal-weight concrete decks on stringers without overlay). On average, most of the bridge deck systems averaged between a 40- to 50-year service life, except for steel grid and timber deck systems, which averaged around a 20-year service life. There was no indication that decks built under current standards exhibited substantially different service life with or without overlays. The predominant type of reinforcement used by the respon- dents is epoxy-coated reinforcement, with only one respondent indicating that this type of reinforcement is not acceptable (see Figure 3.27). This single state is not in an environment where deicing salts are used. The other reinforcing types are rarely used, but are being studied for use in several states. The predominant types of maintenance issues reported by the respondents for CIP concrete bridge decks that are rated poor or worse are illustrated in Figure 3.28. Four of the six dominant maintenance issues appear related, as deck cracking leads to chloride intrusion that results in deck delamination and spalling. The ranking of the routinely recurring maintenance issues encountered for CIP concrete bridge decks that are rated poor or worse indicates that chloride intrusion is the most impor- tant issue faced by the respondents, closely followed by deck cracking, deck delamination, and spalling (see Figure 3.29). Although the survey results for maintenance issues identified by the respondents ranked chloride intrusion slightly less than deck cracking, deck delamination, and spalling, two- thirds of the states providing a response to this answer ranked chloride intrusion as the number one concern for the subject bridge decks. Thicker concrete cover, use of epoxy-coated reinforcement and higher-quality concrete with air entrainment, lower heat of hydration, and proper curing were identified as means Figure 3.25. Survey results for bridge deck types used by the respondents.

31 Figure 3.26. Survey results for bridge deck life expectancy. Figure 3.27. Survey results for reinforcement types used by the respondents.

32 Figure 3.28. Survey results for bridge deck maintenance issues reported by the respondents. Figure 3.29. Survey results for bridge deck maintenance issues reported by the respondents.

33 used for increasing the durability of bridge decks in new construction. The issues associated with increased durability identified by those who responded are ranked in order of importance in Figure 3.30. Most of these issues are related to the quality of concrete and the reduction of deck cracking. Use of corrosion- resistant reinforcement is also identified as one of the most important durability strategies for bridge decks. The survey requested specific information regarding experience with practices and specifications that may affect bridge longevity. These questions dealt with deck-forming practices, different environmental conditions, concrete mix design, curing methods, and special maintenance procedures. A summary of the responses follows. The majority of respondents indicated they did not experi- ence differences in bridge deck longevity as a result of different forming systems (e.g., removable, composite or noncomposite steel stay-in-place, composite or noncomposite concrete stay- in-place). Some respondents had issues with reflective crack- ing over concrete stay-in-place forms, but three states indicated good performance of these systems. Other states do not allow steel stay-in-place forms due to the inability to inspect the bottom of the concrete slab. These states indicated that leakage can occur through cracking, which is hidden by the forms, delaying the response to the deterioration, thereby reducing the bridge deck service life. Respondents in states that experience differences in envi- ronmental conditions within their state indicated there are performance differences based on these conditions. California respondents indicated they experience twice the deck life for bridges in benign environments compared with those located in a more aggressive chloride environment (e.g., road salts). Abrasions from chain wear and truck traffic volumes were also identified as contributing factors to decreased bridge deck life. Respondents indicated that concrete mix design and cur- ing methods are very important to increasing the service life of concrete bridge decks. Numerous states have implemented high-performance concrete mix design aiming for low per- meability, low heat of hydration, and reduced cracking. Kansas has also developed a low-cracking HPC mix design for use in that state. Most respondents employ 7- to 14-day wet curing of decks with special requirements to quickly control evaporation from the deck surface. Most respondents indicated that they do not use overlays for new construction, although there are exceptions. The responses indicated that overlays are used primarily for Figure 3.30. Survey results for important durability issues to be considered in new construction.

34 extending the life of existing structures, with seven of the 17 respondents indicating that overlays are not used at all. Sealing of cracks with epoxy injection or sealers was identi- fied as a special maintenance procedure to increase the ser- vice life of bridge decks. The elimination of joints and drains was also indicated as beneficial. A majority of the respondents indicated that they have undertaken research and testing in the past or are currently undertaking research and testing of materials or construction procedures (or both) for bridge deck longevity. This research and testing addresses such issues as quality of concrete, use of GFRP reinforcement and other noncorroding reinforcement, and GFRP-reinforced decks and reports on problems associ- ated with their use. Respondents did not indicate that they considered the use of noncomposite or partially composite bridge decks to con- trol transverse cracking or to facilitate deck replacement. Respondents indicated a willingness to exchange a higher ini- tial bridge deck cost for increased durability, although there was a caution that this investment had to be cost-effective and rea- sonable considering budget constraints. Few respondents provided information relative to the pres- ent value of CIP decks when built as part of new structures and of CIP decks when built as replacement decks on existing structures. The information provided indicates that the cost to reconstruct a concrete bridge deck is about 160% to 225% of the cost for new construction of a similar deck. Substructure Substructure is defined as the structural elements required to support the superstructure; these elements typically include those from the top of support bearings down through the foundation. The function of these elements is to transfer all vertical and horizontal loads from the superstructure to the foundation supporting strata. To provide longer service life, the integrity of the substructure must be maintained. With sufficient degradation, the reduction in carrying capacity can lead to collapse. The principal issues relating to reduced durability of sub- structure elements were researched. This deterioration is due to environmental factors, such as seismic events and hydraulic conditions leading to bridge scour, and corrosion in extremely aggressive environments. As the designs of bridges begin to anticipate longer service life, the mean recurrence of forces due to conditions in the bridge environment, such as stream flow, vessel aberrance, seismic activity, and vehicle collision, needs to be reassessed to determine if the applied factors of safety are adequate for the increased probability of exceedance associated with a longer life. Degradation of service life due to corrosion on substructure elements is also identified as a significant issue. The intrusion and concentration of chlorides caused by joint and bridge drainage leakage is identified as a concern nationwide, par- ticularly in areas where deicing salts are used. Corrosion due to chloride intrusion is also identified as an issue in splash zones of extremely aggressive environments, whether in marine areas or adjacent to underpass roadways where deicing salts are used. Degradation of the substructure and bearings can also result in changes in stiffness that can adversely affect the distribution of forces within the superstructure and to the individual components of the substructure. The extension of the service life of new and existing substruc- ture components includes risk mitigation for mean recurrence– level events, improved materials and protection systems for substructure components, and proper maintenance and replacement (as needed) for proper bearing function. Description and Discussion of Substructure Types The substructure component includes all elements that support the superstructure and transfer vertical and horizontal loads from the superstructure to foundation materials through spread foundations, piles, or drilled shafts. The following substructure types and components were studied with respect to service life: • Substructure types 44 Abutments; 44 Piers; and 44 Bents. • Substructure components 44 Piles; 44 Drilled shafts; 44 Spread foundations; 44 Modular precast construction; and 44 Integral pier and abutment details. A summary of the information on the listed substructure systems is reported in the Guide. Causes of Substructure Deterioration The numerous causes of substructure deterioration, which are documented in the research for this project, can be categorized in three areas: • Improper consideration of design and service life in estab- lishing appropriate measure to address mean recurrence– level event forces; • Deterioration caused by corrosion and section loss, primarily from chloride intrusion; and • Seized bearings and unintended movement restraint.

35 mean recurrence–level event Forces As bridge service life increases, bridges are subjected to envi- ronmental conditions for a longer period of time. Many of these conditions are accounted for in design by the use of traditional recurrence values of extreme environmental con- ditions such as for hydraulic stages, wind loads, and seismic events. Design for vessel impact is treated similarly. The increased design life from 50-year (pre-load and resistance factor design or LRFD) to 75-year (LRFD) to 100-plus-year service life increases the statistical probability of exceeding the design recurrence–level event. These mean recurrence–level events include • Hydraulic flood events resulting in bridge scour; • Seismic events; • Vessel collision events; • Vehicle impact events; and • Fire events. substructure Deterioration Due to corrosion anD section loss Corrosion deterioration in substructure elements is due to numerous causes, including chloride intrusion due to leakage of expansion joints and bridge drainage where deicing salts are used to remove snow and ice from bridge decks; chloride intrusion due to direct salt splash from traffic traveling on roadways below the bridge where deicing salts are used to remove snow and ice from the pavement; chloride intrusion found in marine and brackish water environments affecting exposed elements; and corrosion due to concrete cracking induced by ASR and other concrete quality issues. Many of the issues affecting durability of the substructure are similar to the issues affecting the bridge in general. seizeD bearings anD unintenDeD movement restraint The structural design of the substructure is based on a distri- bution of longitudinal and transverse forces associated with the allowable movement of the superstructure. Fixed bear- ings provide an anchor that is intended to restrict “walking” of the superstructure that results from shrinkage and cycling of expansion and contraction; they are usually located longi- tudinally near the point of zero movement of a supported multispan superstructure unit. The remaining bearings are designed to allow the superstructure to either move or slide over the top of the substructure, thereby reducing restraining forces that would otherwise be required to resist the move- ment. Improper function, or seizing, of the bearings results in unintended movement restraint that can raise the force resisted by the substructure well above the intended design. This unintended restraint can cause unanticipated cracking with greater potential for corrosion. Protection of Substructure The extension of the service life of new and existing sub- structures includes risk mitigation for mean recurrence–level events, improved materials and protection systems for sub- structure components, and proper maintenance and replace- ment (as needed) for proper bearing function. The following subsections discuss approaches for protecting the substructure for the purpose of extending service life. risk mitigation Proper risk categorization of a structure should be performed based on longer life to establish proper force effects for new structures and retrofit requirements for existing structures. material protection Improvements to service life of new substructures can be achieved through the following material improvements and protection schemes: • Develop substructure components using specifications for construction materials that have strong resistance to saltwater environments. • Use HPC specifications to reduce concrete permeability in splash zones and its chemical resistance to salt water, including increased concrete quality and density, use of noncorroding reinforcement, and other methodologies. • Incorporate cathodic protection strategies. • Consider jacketing systems for piles and shafts in critical parts of the elements to protect these parts before damage and chloride intrusion can occur. • Eliminate bridge joints, and channel bridge drainage away from the substructure. Improvements to the service life of existing substructures can be achieved through retrofit and rehabilitation by using the following techniques: • Add corrosion protection with embedded sacrificial anodes. • Encapsulate piles with jacketing. • Improve the passivation around reinforcing steel through the removal of chloride ions. Analysis of DOT Survey The results of the survey performed by the R19A research team indicated that the predominant maintenance issues faced by state DOTs relate to substructures. These issues and their severity are shown in Figure 3.31. The primary issues faced routinely by the states are chloride intrusion issues relating to leakage of expansion joints and bridge drainage systems. This issue is addressed further in the

36 section titled “Expansion Joints, Joints, and Jointless Bridges.” Many states appear to be addressing these issues by moving toward jointless bridges, on which storm water is collected at bridge approaches. Chloride-intrusion issues relating to marine environments ranked equally high in states with marine environments. Little information was provided as to the methodology for addressing these concerns. Other areas of concern in which individual states indicated routine maintenance issues related to excessive expansion and contraction, including complete closure of joints, lock-up of bearings, scour at embankments and bridge pier foundations, and vehicle impact. All other items were ranked as occasional or rare. These issues are addressed further in the report sec- tions “Expansion Joints, Joints, and Jointless Bridges” and “Bearings.” In general, the adequacy of current inspection methods to identify maintenance issues in a timely manner for substruc- tures was ranked as excellent or acceptable. Areas in which individual states indicated unacceptable procedures included assessment of chloride intrusion, seismic susceptibility, scour detection, and foundation settlement. Bearings Bearings are an important bridge element that must be considered when evaluating ways to extend overall bridge service life. Bridge superstructures experience translational movements and rotations caused by traffic loading, ther- mal effects, creep and shrinkage, wind and seismic forces, initial construction tolerances, and other sources. Bridge bearings are designed and built to accommodate these movements and rotations while supporting required gravity loads, trans- mitting those loads to the substructure, and providing the necessary restraint to the structure. Proper functioning of bridge bearings is assumed in the analysis and design of overall bridge systems. Bearing failure, or improper behavior, can lead to significant changes in load distribution and over- all structural behavior that are not accounted for in the design and can significantly affect the superstructure–substructure interaction. Recent instances of rocker bearing rollover have nearly led to catastrophic span collapses. Clearly, bridge bear- ings play a critical role in overall bridge behavior and service life. Further, they represent a small cost in proportion to the overall structure cost, but can potentially cause big problems if they are not maintained or do not function properly. This part of the study focused on the various bearing types currently used, identifying related problems and deter- mining which options have the best potential for achieving 100-plus years of service life. It also looked at bridge systems that eliminate bearings by using integral girder–pier con- struction. The information compiled in this study was used to develop a detailed chapter in the Guide related to bridge bearings. Figure 3.31. Survey results for reported severity of issues relating to substructure durability.

37 Current Bearing Types and Usage The study identified the following main categories of bearings in use today: • Elastomeric bearings, which include steel-reinforced pads and plain elastomeric pads; • Cotton duck pads (CDPs); • Sliding bearings, using polytetrafluorethylene (PTFE) pads; • Manufactured high-load multirotational (HLMR) bearings, which include pot bearings, disc bearings, and spherical bearings; and • Mechanical, fabricated steel bearings, for fixed application and for expansion application using rollers or rockers. A literature review was conducted to compile existing research and relevant data regarding the performance of these bearing types and identify promising concepts. Design, materials, manufacturing, construction practices, main- tenance, and imposed loads were considered. The results of research and testing were evaluated to understand the mechanics and behavior of various bearings and evaluate future research needs. Phone seminars were conducted with Charles Roeder, bearing researcher at the University of Washington, and with representatives from D. S. Brown and R. J. Watson, both major bearing manufacturers, to gain further insight and ask specific questions related to bearing life. Research team members and various state DOT bridge engineers were invited to participate in these seminars. The DOT survey responses related to bearings were analyzed to determine trends in the industry. DOT data regarding use of various bearings, types of related problems, service life, extent of maintenance, and frequency of bearing replacement were compiled and compared. The following subsections summarize the compiled infor- mation related to service life for the various identified bear- ing types. elastomeric bearings From the DOT survey and other references, steel-reinforced elastomeric (SRE) bearings are the most widely used bearing type among nearly all states and have a good history of service. SRE bearings have become the most common type in recent years because of their desirable performance characteristics, durability, low maintenance requirements, and relative economy. Elastomeric bearings have no movable parts and accommodate movement and rotation by deformation of an elastomeric pad, which can be neoprene or natural rubber. Lateral and longitudinal movement is accommodated by the pad’s ability to deform in shear. These bearings can accom- modate combined movements in both longitudinal and transverse directions; circular elastomeric bearings have been used to efficiently accommodate multirotation and direction requirements. Plain, unreinforced elastomeric pads are used for short spans on which loads and movements can be accommodated by a single layer of elastomer. As vertical load and movement requirements increase, thin reinforcing plates are combined with multiple layers of elastomer to form a laminated, reinforced elastomeric assembly. Steel and fiberglass reinforcement layers have been used; however, fiberglass is weaker, more flexible, and does not bond as well to the elastomer as does steel reinforce- ment. Therefore, the use of thin steel plate reinforcement has become the norm. Elastomeric bearings have been used in the United States since the 1950s. A report published by DuPont in 1959, Design of Neoprene Bridge Bearing Pads, served as a design reference for many years, and included standard procedures to accommo- date compressive stress and shear deformation for various neo- prene hardness levels. Soon after, the 1961 American Association of State Highway and Transportation Officials (AASHTO) specification included the first formal bearing design procedure in the United States (Roeder and Stanton 1991). Although these early design provisions generally resulted in satisfactory designs, certain limitations became increasingly apparent; NCHRP Project 10-20 was established in 1981 to develop improved design procedures. In the 1980s, Stanton and Roeder conducted studies as part of this program to examine the complex nature of elastomeric bearing pads. A good summary of general research and practice up to 1980 is provided in NCHRP Report 248 by Stanton and Roeder (1982). NCHRP Project 10-20 also conducted tests to assess bearing compression, rotation, shear, stability, fatigue, and low-temperature behavior. Results of this work were reported in NCHRP Reports 298 and 325 (Roeder et al. 1987, 1989). Later, additional studies under NCHRP Project 10-51 were conducted on low-temperature behavior and on the require- ments for various materials tests; the results were reported in NCHRP Report 449 (Yura et al. 2001). This report recom- mended specifications and new test methods for evaluating essential properties of elastomeric bearings. NCHRP Report 596: Rotation Limits for Elastomeric Bearings (Stanton et al. 2008) provides a good overall review of the mechanics of SRE bearing behavior and potential failure modes. SRE bearings typically are a very robust system and problems have been rare, but errors in design have led to pad failure. Previous research and testing have identified various poten- tial failure modes and developed design requirements that address them (Stanton et al. 2008). Shear delamination between layers of elastomer and steel reinforcing plates is the most significant potential mode of failure. This problem can be caused by excessive shear strains due to combined axial load, rotation, and shear, and it is exacerbated by cyclic shear

38 strain due to traffic load. The LRFD Specifications now pro- vide updated requirements for considering combined shear strain, along with other design requirements that address shear deformation, pad instability, plate fracture, and compres- sive deflection. Two design methods have been established in the LRFD Specifications: a simplified and conservative Method A and a more detailed Method B. Method A is the most widely used method. Another reported field problem associated with elastomeric bearings, albeit in only few instances, has been “walking out,” or bearings slipping from their original position under the girder. Chen and Yura (1995) investigated slippage of elastomeric pads experienced on Texas bridges. McDonald et al. (2000) reported on the slippage of elastomeric pads and the results of a survey of state DOTs, bearing manufacturers, and other researchers to gain knowledge on the subject. Heymsfield et al. (2001) reported on the pad slippage experienced on Louisiana bridges and dis- cussed the causes and recommended practical guidelines to remedy the problem. Slippage was occurring primarily in elas- tomeric pads made of natural rubber because of paraffin that was added during the manufacturing process to meet required ozone degradation requirements established by AASHTO. Neoprene pads have inherently greater ozone resistance and do not need wax additives. When used, these waxes over time will bleed to the bearing surfaces and drastically reduce the coeffi- cient of friction between the bearing and its contact surface, which leads to slipping (Chen and Yura 1995). McDonald et al. (2000) reported on various remedial measures from a survey of state DOTs; these measures included • Only using neoprene pads; • Banning the use of antiozonant wax additives; • Vulcanizing bearing to sole plate for steel bridges, or hav- ing anchoring details; • Having positive attachment when low bearing stresses exist; • Inspecting bridges annually; and • Cleaning contact surfaces when replacing natural rubber bearings with neoprene bearings. In summary, when SRE bearings are adequately designed, manufactured, and installed it is believed that they have the best potential for achieving 100-plus years of service life with practically no long-term maintenance requirements. They have broad application to most bridges, both concrete and steel superstructures, that fall within the 300-ft span range for this study, and they are applicable for multidirectional move- ment (as with curved and skewed bridges). There is broad use of this type of bearing among the states, and there are over 50 years of service experience in the United States, and even more abroad, to verify these conclusions. Manufacturing quality has been very good and has improved significantly in the last 20 years. However, to achieve 100-plus-year service life, even greater emphasis must be placed on manufacturing quality assurance. Because of some past problems related to improper design, it is recommended that an updated design guide for SRE bearings be developed that would supplement the current LRFD Specifications. The guide would provide detailed guidance and understanding of the mechanics and behavior of these bearings and would help avoid design problems. Additional cyclic compression testing is still recommended, however, to supplement newer combined compression– rotation design recommendations. This additional fatigue testing, particularly for shear strain due to compressive load- ing, could also help develop a life prediction model. cotton Duck paDs CDPs are another type of elastomeric bearing that is occa- sionally used in some states. They have typically been used more for precast concrete I-girder bridges and have been used with span lengths up to the 150- to 180-ft range. CDPs are preformed, elastomeric pads consisting of very thin layers of elastomer (less than 1⁄60 in.) interlaid with fab- ric. The fabric can be cotton or polyester. The fabric layers are many, and closely spaced, which gives CDPs high compressive strength and stiffness. These closely spaced layers, however, provide high shear stiffness and much smaller shear deforma- tion capacity than other elastomeric bearing types provide. As a consequence, CDPs tolerate large compressive stresses but limited shear deformation because interlayer splitting occurs at relatively small shear strains. Thus, larger movement require- ments with CDPs must be accommodated by the addition of a PTFE sliding surface. Recent design recommendations have been developed and incorporated into the LRFD Specifications (Lehman et al. 2003). The overall performance of CDPs depends on their stiffness and deformation capacity, which can vary among manufac- turers. To ensure adequate performance from CDPs, quality control testing measures have also been developed. CDPs have not been widely used and their service history is limited, but there is a potential for this bearing type within a defined operational range (i.e., bridge systems that have short movement and rotation requirements). CDP applications using PTFE sliding surfaces for accommodating larger movements are subject to the limitations for PTFE wear, which are discussed in the next subsection. Because of their limited use and service history, and the need for PTFE sliding surfaces for movement, CDPs are less recommended than SRE bearings for long service life. polytetraFluorethylene sliDing bearings With certain combinations of high load, movement, and rota- tion, the capacity of elastomeric bearings to accommodate

39 the required translation through shear deformation can be exceeded. In these cases, additional movement capacity must be provided by the use of sliding surfaces. Further, all other types of bearings—including CDP and HLMR pot, disc, and spherical bearings—use sliding surfaces to accommodate expansion requirements. Currently, PTFE is the material used for sliding surfaces in the United States. Its low frictional characteristics, chemical inertness, and resistance to weathering and water absorption make it an attractive material for bridge bearing applications. The sliding movement is typically provided by a very smooth stainless steel plate sliding on a PTFE surface. The stainless steel surface is larger than the PTFE surface so that the full movement can be achieved without exposing the PTFE. The stainless steel is typically placed on top of the PTFE to prevent contamination with dirt or debris. PTFE sliding bearings may be guided, allowing movement in only one direction, or non- guided, allowing multidirectional movement. When PTFE sliding surfaces are combined with elastomeric pads, the elas- tomeric pad must be designed to accommodate the shear force that is needed to overcome the PTFE friction resistance. Sliding surfaces develop a frictional force that acts on the superstructure, substructure, and bearing. As a result, friction is an important design consideration; the low frictional resistance of PTFE makes it useful for this application. The coefficient of friction of PTFE increases with decreasing temperature and with decreasing contact pressure. It also increases if the mating surface is rough or contaminated with dust or dirt. Proper design, fabrication, and field installation are all essential for proper performance. Plain, unfilled PTFE is the most common material used for sliding bearings. However, plain PTFE wears under certain service conditions, particularly when subjected to combina- tions of high contact pressure, high rates of movement, and low temperatures, and may require replacement after a period of time. Fast sliding speeds, especially those associated with traffic movements, have been shown to be much more critical for PTFE wear than slow movements due to temperature (Stanton et al. 1999). Therefore, wear of the PTFE sliding sur- face is one of the critical factors affecting service life for these types of bearings, and it is not recommended to use plain PTFE as a sliding surface for bearings subject to relatively high sliding speeds and low temperatures. Further, there are minimal data to develop a life prediction model for sliding surfaces. Designers need to have the capability of predicting the expected service life of sliding surfaces for maintenance and replace- ment purposes. Woven or glass-filled PTFE surfaces provide much higher overall wear resistance, especially at higher sliding speeds (Stanton et al. 1999). However, these surfaces have higher friction coefficients that must be taken into account in the bridge system design. Dimpled and lubricated PTFE also provides exceptional wear resistance and low friction, but the long-term reliability and effectiveness of lubrication is questionable (Stanton et al. 1999). Dimples are spherical indentations machined into the PTFE surface to act as reservoirs for storage of lubrication. Silicone greases are specified because they are effective at low temperatures and do not attack the sliding material. Dimpled and lubricated PTFE has been used in Europe, but in the United States it has been used only in special cases on large spherical bearings where very low coefficient of friction requirements are needed to reduce friction loads on substructures. PTFE may creep (or cold flow) laterally when subjected to high compressive stress and shorten the life of the bearing. The reduction in PTFE thickness may also allow hard contact between metal components. Thus, although the compressive stress should be high to reduce friction, it must also be limited to control creep. PTFE is frequently recessed for one-half its thickness and bonded to control creep and permit larger compressive stress. Filled PTFE, which is reinforced with fiberglass or carbon fibers, has significantly greater resistance to creep and is sometimes used to resist creep or cold flow. Maurer sliding material (MSM) is an alternative sliding material developed in Germany as a higher-performing substitute for current PTFE-based sliding material. The new material is an ultrahigh-molecular-weight polyethylene that has performed well in recent field applications and experi- mental testing in Europe. It has been shown to provide excep- tional wear resistance, but when used in a dry condition (without lubrication) it has a much higher friction coefficient than plain PTFE. The friction coefficient reduces, however, with higher contact pressures. When used in a dimpled and lubricated condition, its friction coefficient reduces consider- ably and is more comparable to lubricated PTFE. Testing was performed as part of this study on high- performance sliding materials to compare friction coefficients, wear rates, and accumulated wear by considering a number of parameters, including contact pressure, speed of movement, and total accumulated movement. The goal was to evaluate these materials for improved wear resistance and to determine if it was feasible to develop a deterioration model. Materials tested included • Plain PTFE as a base; • Fluorogold, which is a fiberglass-reinforced PTFE; and • MSM, which is an ultrahigh-molecular-weight poly- ethylene. The results of this study are included in Appendix D. The testing showed significantly improved wear resistance of these high-performance sliding materials over plain PTFE. It also

40 showed that a wear rate model could be established for PTFE- based materials considering a pressure–velocity factor. high-loaD multirotational bearings When design loads and rotations exceed the reasonable limits for elastomeric bearings, HLMR bearings have typically been considered. HLMR situations often occur with longer spans; with curved or highly skewed bridges; or with complex framing, such as with straddle bents. In these cases the axis of rotation or the direction of movement, or both, are either not fixed or may be difficult to determine. The current LRFD Specifications now incorporate research findings on HLMR bearings reported in NCHRP Report 432: High-Load Multi-Rotational Bridge Bearings (Stanton et al. 1999). HLMR bearings include pot, disc, and spherical bearings. Each HLMR bearing type is unique in how it accommodates large loads and rotations. All are fabricated in fixed and expan- sion versions. The expansion versions accommodate transla- tional movement by means of PTFE sliding elements. Expansion versions may be guided, allowing movement in only one direc- tion, or nonguided, allowing multidirectional movement. The following paragraphs describe and compare the three types of HLMR bearings. Pot bearings have been the most-used HLMR bearing. The main elements of these bearings include a shallow steel cylinder, or pot, which contains a tight-fitting elastomeric disc that is thinner than the depth of the cylinder. A machined steel piston fits inside the cylinder and bears directly on the elastomeric disc. Brass rings are used to seal the elastomer between the piston and pot components. Vertical load is carried through the piston of the bearing and is resisted by compressive stress in the elastomeric pad. The pad is deformable, but almost incompressible in its confined condition; it is often idealized as behaving hydrostatically. Rotation can occur about any axis and is accommodated by deformation of the elastomeric pad. Horizontal loads on a pot bearing are resisted by direct contact between the pot wall and the piston. Early on, pot bearings developed a bad reputation for seal- ing ring failure and leakage of elastomer, but improvements through testing, improved design requirements, and improved manufacturing tolerances have solved many of the early prob- lems. There have not been recent reported failures. However, they are still susceptible to internal wear and are less likely to reach long service life. The AASHTO LRFD Bridge Construction Specifications (AASHTO 2010a) require cyclic testing to con- firm long-term performance. Pot bearings require a high degree of quality control in the fabrication and field installation process and an accurate determination of design loads and displacements to achieve satisfactory performance. Adequate clearances must be pro- vided to allow proper displacement without binding. Through the years, they have been the most economical and common HLMR bearing and have been implemented on bridges throughout the United States. Disc bearings were proprietary until recently and are now becoming more popular. They consist of a hard polyether urethane circular disc sandwiched between upper and lower steel plates with a center shear pin device to resist horizontal load. The discs are stiff enough to support the compressive load, yet can deform to permit rotation. However, rotational stiffness for a disc bearing is several times that of a pot bearing. Disc bearings are reasonably economical, but widespread use has been limited because of their originally patented and proprietary status, which made them available only from a single source. Now, there are several reputable bearing manu- facturers that can supply disc bearings, and usage has increased. Since their first use in the early 1970s, these bearings have had good performance, and there have been few reported field problems. Potential service life problems would likely be associated with production or operation defects related to design and manufacturing. Current specifications are mini- mal regarding design, and are generally performance related. Therefore, performance testing and manufacturing quality control measures are necessary to verify compliance. Previous research and testing of disc bearings for combined load and rotation (Stanton et al. 1999) showed that rotation of disc bearings is partly accommodated by uplift of the steel plates from the urethane disc, especially if the compressive load is light. This uplift should not result in any problems with fixed bearings, but it could be a concern with sliding bearings, since uplift of the disc produces edge loading on the PTFE sliding surface. To mitigate this potential problem, the LRFD Specifications limit the edge contact stress on PTFE surfaces. High edge pressure can accelerate PTFE wear as described previously. Tests also showed the urethane disc to be somewhat deformed and abraded by cyclic rotations, but the damage was not severe enough to affect performance. The AASHTO LRFD Bridge Con- struction Specifications (AASHTO 2010a) require cyclic testing to confirm long-term performance. Spherical bearings are the most robust of all HLMR types and are used primarily for accommodating large rotations about one or more axes. Rotation is developed by sliding a convex stainless steel surface against a concave spherical PTFE surface. The rotation occurs about the center of the radius of the curved surface, and the maximum rotation is limited only by the geometry and clearances of the bearing. Translational movement is accomplished by incorporating a flat PTFE slid- ing surface. Horizontal loads may be partially resisted by the curved geometry, but large horizontal loads may require addi- tional external restraint. Spherical bearings require a high level of accurate machin- ing for proper mating of the convex and concave spherical surfaces. As a result, they are generally the most expensive

41 HLMR type. The advantage, however, is their ability to accom- modate higher gravity loads and rotations. They are also considered to be the most reliable. Spherical bearings are subject to PTFE wear, which can limit their service life. Variations in friction with different types of PTFE and under different temperature and load conditions cause variations in behavior that can lead to performance issues. Woven PTFE has often been used with spherical bear- ings in the United States, and dimpled and lubricated PTFE is often used in Canada and Europe. mechanical steel bearings Mechanical steel bearings were used extensively up through the 1970s for steel bridges and some concrete bridges for both fixed and expansion conditions. Many existing bridges still have these types of bearings, and some states still use them for new construction. Steel bearings transmit loads through direct metal-to-metal contact. Most fixed bearings rely on a pin or knuckle to allow rotation while restricting translational movement. Rockers, rollers, and sliding bearings are common expansion types historically used. Typically, steel bearings are expensive to fabricate, install, and maintain, which is why elastomeric bearings have become popular. Moreover, steel bearings typically provide unidirectional movement. These types of bearings are fully designed by the engineer to accom- modate loads, movements, and rotations and can accommodate large requirements. Bronze-lubricated plate bearings have been used in con- junction with steel bearings to accommodate smaller levels of movement at expansion ends, but they are not used very often today. PTFE sliding surfaces have replaced bronze sliding plates because of a much lower coefficient of friction and lower cost. When functioning properly, mechanical steel bearings gen- erally provide the closest representation of assumed structural end conditions. Mechanical steel bearings have been used extensively over the years, but corrosion and freezing have been problems, particularly when the bearings are located below leaking deck joints. Steel rocker bearings have also performed poorly in seismic events and have been replaced as part of seismic retrofit in many instances. Mechanical steel bearings require a high level of maintenance, but if properly protected and maintained their service life can match the superstructure they support. Elastomeric and HLMR bearings, however, have typically replaced these older steel mechanical types in new construction in most states. The literature regarding fabricated steel bearings typically relates to discovered field problems. Rockers that are over- rotated or corroded and frozen bearings causing other forms of distress in pier caps or at the ends of concrete beams are occasionally observed during maintenance inspections. On rare occasions, these problems have been encountered by the traveling public. Fyfe et al. (2006) reported that steel rocker and roller bear- ings, along with older-type metal sliding plates, are the most susceptible to freezing in position. When bearings are frozen, bridges have developed their own provisions for contraction and expansion such as pier cap cracking or rocking of piers or abutments. Modjeski and Masters (2008) reported on a rocker bearing failure on a 32-year-old bridge in Pittsburgh, Pennsylvania, on which the rocker bearings at one pier tipped over due to continual buildup of debris and corrosion below a leaking deck expansion joint. Many bridges have performed satisfac- torily with steel bearings; however, this failure illustrates the importance of continual maintenance, particularly in areas affected by leaking joints. A corrosion protection plan is also critical for these types of bearings, especially when they are located below deck joints. Galvanizing, metalizing, and use of zinc-rich primers provide the best forms of protection, in descending order. System Approaches to Improved Bearing Service Life From a bridge system point of view, use of integral construction that eliminates or minimizes the number of bearings is desir- able. Eliminating or minimizing the number of deck joints whenever possible is also desirable. Eliminating deck joints prevents deck drainage and debris from spilling on bearings and causing deterioration, particularly with steel bearings, or steel surfaces on components of other types. Integral girder–pier construction has been used in some cases for resolving clearance issues and avoiding sharp skews, but it can also be considered for eliminating bearings and resulting maintenance issues. Most integral girder–pier con- struction to date has been with steel girders and CIP post- tensioned concrete caps. However, many integral caps have also been constructed within prestressed concrete girder and concrete box girder systems using conventional reinforce- ment without posttensioning. These various integral pier cap subsystems have performed well to date. The next subsection, “Integral Construction,” discusses these systems in greater detail. Skewed, curved, and wide bridges subject bearings to multidirectional movements or rotations, or both. Improper bearing orientation or inadequate multidirectional move- ment capacity can lead to higher stresses, wear, and reduced service life. The National Steel Bridge Alliance (NSBA 2005) provides guidance for proper orientation of expansion bearings in curved and skewed bridges. Bridges wider than three lanes can experience significant transverse thermal movement. Guides and keeper assemblies should be limited to the interior portions of the bridge that do not experience large transverse movements. Bearing details

42 for outer portions on wide bridges should be designed to accommodate transverse movement. Proper inspection of bearings during their service life is critical to evaluate proper performance, wear, and deteriora- tion. Early detection of problems can allow maintenance or repair before more serious conditions can develop. Shallower bearing types can be difficult to properly inspect, particularly when limited headroom prevents close access. Consideration should be made in overall bridge system design to allow access for proper inspection and maintenance of bearings, including • Considerations to improve bearing service life; • Regular inspection of bearings to detect problems early; • Regular maintenance performance and general cleaning when bearings are exposed to dirt and debris, washing sur- faces, and cleaning troughs below open-finger dams; • Cleaning and painting of steel surfaces on mechanical steel bearings to prevent corrosion; and • Repairing or replacing bearings that are not functioning before other problems occur due to higher stress on girders or piers. Converting existing jointed construction systems to continu- ous systems to eliminate joints and thus protect bearings below (and other elements) can be very beneficial. States vary regard- ing how this is done. Standard recommended details should be developed to facilitate design, construction, and performance. Regardless of expected service life, bearings are subjected to severe service conditions and have a high potential for unintended consequences related to improper design, manu- facturing, installation, and maintenance that can lead to shorter service lives than other bridge elements. Consideration should be given in overall bridge system design to allow for easy replacement of bearings with minimal traffic disruption. AASHTO (2004) and NSBA (2005) provide recommended bearing details that facilitate replacement. The following items should be considered during design to allow for bearing replacement: • Jacking locations should be provided at every girder. An alternative is to provide for jacking under a diaphragm that lifts adjacent girders simultaneously. • An unattached bearing can easily be pulled from its position when the load is removed. Anchor bolts should be placed so that they do not impede the removal of the bearing. • Bearings should be detailed in a way that they can be replaced with only 0.25 in. of jacking so as not to cause a bump at the top of the deck that would affect traffic. Bridge inspections and inspection data collection for bearings should be expanded to identify the bridge system for bearings, bearing types, conditions, and other relevant data. These recommendations for more detailed bearing data col- lection can be used within the Federal Highway Administra- tion (FHWA) Long-Term Bridge Performance program, which is intended to study the deterioration and durability of bridges and the impacts of maintenance and repair. These rec- ommendations can also be used to supplement the types of data collected for use within bridge management systems such as PONTIS. Improved data collection for bearings can be use- ful in determining and scheduling required maintenance and for developing more accurate deterioration models. Integral Construction general concept Integral, CIP posttensioned bent caps have been in service in Tennessee since 1978 (Wasserman 1987) and were part of an interchange in Knoxville. This technique allows girders to pass directly through the pier’s cap, rather than over the top in the traditional manner, thus overcoming vertical clearance restrictions and avoiding extreme skews (Wasserman 1987). The reported bridge system used longitudinal steel girders with single-column reinforced concrete piers. The construction methods and overall performance were described. To date, the structure has exhibited good performance. Other benefits sited were reduction in fill height, seismic resistance, and aesthetics. Within interchanges, directional ramps often cross other roadways on a skew. In these cases, the ramp piers are often placed to match the skew, which minimizes the crossing span. An alternative is to keep the ramp piers normal to the upper roadway and to either lengthen the span or raise the profile to provide the necessary vertical and horizontal clearance. Use of integral pier caps with single-column piers allows the pier caps to be placed normal to the roadway, thereby avoiding the sharp skew, while still maintaining the minimum span length and minimum ramp profile. This arrangement has a distinct advantage when the abutments are also placed normal to the ramp. Having both the abutments and piers in a consistent orientation normal to the roadway eliminates complexity in design and construction. Integral pier cap construction also has the advantage of eliminating bearings. nchrp project 12-54 NCHRP Project 12-54 and the subsequent NCHRP Report 527: Integral Steel Box-Beam Pier Caps (Wassef et al. 2004) reported on research conducted by Modjeski and Masters, Inc. and Iowa State University to develop recommended design methodologies and details for integral connections between steel superstructures and concrete substructures. The study first evaluated data on the state of the art for integral connections. A questionnaire was sent to state DOTs, industry associations, and select domestic and foreign bridge designers to determine

43 the extent and type of use. The main reason reported for using integral pier caps was to increase underclearance and to avoid sharp skews (94% of cases). Enhancing seismic perfor- mance was cited in 33% of cases. Steel plate girders were used in most of the bridge superstructures using integral pier caps. Most integral pier caps (76%) were supported on single- column piers. The majority of integral pier caps (90%) were made from concrete, and the remaining caps were made from steel. No cost data were available, but the respondents esti- mated a reduction in weight of steel and an increase in cost of fabrication and erection. They also estimated a savings on approach roadway costs due to the reduced height of fill. Performance history was reported as good, but long-term service data on many of the reported bridges were not available because they were new. Fourteen pier cap systems were initially examined, with variations in steel and concrete superstructures and variations in reinforced concrete, posttensioned concrete, and steel caps. Criteria were developed to assist in scoring the various cap systems and selecting a system for further studies. Systems with the highest scores were as follows: • Steel I-girders on single-column piers and posttensioned concrete pier caps; • Steel I-girders on multicolumn piers with columns located under each girder; and • Steel I-girders on single-column piers and steel box beam pier caps. Previous research on integral cap connections by Caltrans at the University of California at San Diego (UCSD) evaluated several integral connection systems and selected a steel girder with a posttensioned concrete cap for study and testing that was similar to the first system identified by NCHRP 12-54. To avoid duplicating the UCSD studies, NCHRP 12-54 chose the steel I-girders with steel box beam cap for further study. The study team concluded it was feasible to integrally con- nect reinforced concrete columns to steel box beam caps by extending the column longitudinal reinforcement through holes in the pier cap flange and filling the pier cap compart- ment directly above the column with concrete. ucsD stuDies Performance of integral pier details for steel bridges in seismic regions was researched by Patty et al. (2002) at UCSD. They concluded that posttensioned concrete diaphragms provided the best behavior in seismic regions. According to the same study, posttensioning the concrete diaphragm is an effective way of controlling the torsional- type cracks that could form in the diaphragm when the bridge is subjected to ground motion parallel to the span length (longitudinal). DOT Survey for Bearings The following analysis summarizes the 19 state DOT responses to the R19A project survey for bridge bearings. The survey questions for bearings focused on several key issues, including • Bearing types used and frequency of use; • Problems experienced with the various types used; • Service life experienced and expected; • Types of bearing improvements that have been developed or used; • Frequency and types of maintenance performed; • Frequency and reasons for bearing replacement; • Provisions incorporated into new designs to facilitate bear- ing replacement; • Use of integral piers and abutment caps to eliminate bearings; • Orientation of bearings on curved or skewed bridges; • Accommodations for transverse movements on wide bridges; and • State DOT research activities related to bearings. These issues are discussed in the following sections. bearing types useD by Dots Figure 3.32 summarizes the reported use of the various bearing types by the responding number of state DOTs. The chart identifies use in three categories: used frequently, occasionally, or rarely. The following list identifies the abbreviations shown in Figure 3.32: • SRE: steel-reinforced elastomeric bearings; • Plain E: plain elastomeric pads; • PTFE: sliding bearings using PTFE pads; • CD: cotton duck pads; • HLMR P: pot bearings; • HLMR D: disc bearings; • HLMR S: spherical bearings; and • Steel: fabricated mechanical steel bearings. SRE bearings were reported as the most frequently used bearing type by all survey respondents. They were reported as routinely used by nearly all states for both steel and concrete bridges. Plain elastomeric pads were used routinely by a few and occasionally by several DOT respondents, but were mostly used for concrete bridges. PTFE sliding bearings were reported routinely used by several respondents for both steel and concrete bridges; a few DOTs reported using them occasionally. CDPs are not widely used and were reported routinely used only by Nebraska; only a few other states reported occa- sionally; most reported rarely or did not respond.

44 HLMR bearings are used when required by higher loads and rotations, typically for longer spans or for curved and skewed bridges. Pot, disc, and spherical bearings were all reported used. Pots were the most reported of this type, followed closely by discs and to a lesser extent by spherical bearings. HLMR bear- ings were reported used for both steel and concrete bridges, but many respondents indicated their use for steel bridges only. Fabricated steel bearings (fixed, rocker, or roller) are still used by many DOT respondents in varying degrees, primarily for steel bridges, but a few indicated use also for concrete bridges. One respondent who reported occasional use indicated that they are no longer used on new bridges. One indicated that these types used to be their standard. Five DOTs reported using them rarely now. A few respondents reported using seismic isolation bear- ings (not shown in the Figure 3.32) occasionally for both steel and concrete bridges. Bearing ProBlems exPerienced Bearing deterioration caused by deck drainage through open or leaking deck joints was reported by nearly all the responding DOTs. Steel bearings were predominantly affected. Bearing freezing (restricted lateral or rotational movement) and steel element corrosion, both associated with steel bearings, were the predominant bearing problems and were reported routinely or occasionally by nearly all respondents. However, corrosion on exposed steel elements on all other bearing types was also reported as occasionally by a few respondents. The following list describes the types of problems experi- enced with the various bearing types. Aside from freezing and corrosion, there were no routinely found bearing problems; all other problems noted were reported as occasionally or rarely. • Elastomeric bearings had elastomeric material failure due to excessive loading or deformation as the most commonly reported problem, but the frequency was reported as occasional by eight respondents, and rarely or never by 10 respondents. Material degradation from environmental effects occurred, but to a much lesser degree. Walking out from under girders and slipping was reported by three respondents, but this problem type was not specifically asked in the survey, so it may be possible that other respondents have experienced this but did not add it to their response. • CDPs had no routinely or occasionally reported problems. • PTFE material failure was reported as occasional by three respondents, but rarely or never by the rest. • HLMR component failure was also reported to be occasional by a few respondents, but it was mostly reported as rarely or never. 44 Pot bearing failure due to elastomer leakage through sealing rings was reported occasionally by three states, but four others reported rarely or never. 44 Disc bearings were reported to have had component prob- lems occasionally by two states, but six states reported rarely or never. 44 Spherical bearings were reported by two states to have component problems rarely or never. service life of Bearings There was some variation among the responding DOTs con- cerning expected and experienced service life for the various types of bearings that were reported used. Overall, expected service life typically ranged between 50 and 75 years, but some types, such as PTFE sliding bearings or HLMR bear- ings, were more typically expected to last at the lower end of the range (30 to 50 years). It should be mentioned, however, that experiments carried out within this research project (reported in Appendix D) indicated that PTFE has a much shorter service life. Elastomeric bearings were expected to last at the higher end of the range, closer to 75 years. Experienced service life varied more. In some cases, low experienced service life was reported, but the report was Figure 3.32. Bearing type use by respondents. 0 5 10 15 20 SRE Plain E PTFE CD HLMR P HLMR D HLMR S Steel Routine Occasi Rarel ly onally y

45 actually the current number of years that the bearing type has been in service, which did not represent its actual full service life. This was also the case with some HLMR bearing types. Typically, sliding PTFE bearings experience wear on the PTFE surfaces that can shorten service life. Manufactured HLMR bearings are typically more complicated, with more wearing components that can also affect service life. Early pot bearings experienced service problems. Fabricated steel bearings have experienced long service life when protected and maintained. When subjected to a buildup of debris and corrosion, they have exhibited very short service lives. Table 3.4 summarizes the ranges of service life reported. bearing improvements to improve service liFe The DOTs were asked if they had developed or used any bear- ing improvements that had the potential for extending bear- ing service life. Table 3.5 summarizes their responses. Some respondents identified various improvements, which included • Use of integral construction and jointless bridges, which eliminate bearings and prevent deterioration due to leaking joints; • Use of elastomeric bearings, which have had improved performance results; • Improved PTFE bearings with improved PTFE hybrids or thicker PTFE surfaces, and keepers or recesses to prevent PTFE material from creeping out (PTFE surfaces wear over time, and improved PTFE hybrids or thicker PTFE surfaces can provide longer life); and • Use of galvanized anchor bolts. special protection For steel elements in bearings to prevent corrosion The DOTs were asked if they used any special coatings, paint systems, or types of steel for steel elements in bear- ings to prevent corrosion and extend service life. Some respondents identified various methods of protection. These included • Hot-dip galvanizing; • Metalizing; • Inorganic zinc-rich primer applied in shop; • Use of the same paint system as steel girders; and • Some use of weathering steel. Table 3.4. Ranges of Service Life Reported for Different Bearing Types Bearing Types Expected Service Life (years) Experienced Service Life (years) Comments Elastomeric 50–75, closer to 75 15–50 Experienced service life limited by current years in service Cotton duck 75 35–50 Only two DOTs reported PTFE 30–75 30–50 HLMR 30–75, mostly 50 10+ pots 15–40 others Early pots had problems. Lower service life often limited by current years in service Fabricated steel 50–75 15–100 Oldest type of bearing in service Table 3.5. Bearing Improvement Methods Bearing Improvement No. of DOTs Reporting Comments Use of integral construction and jointless bridges 3 Eliminates bearings and bearing problems and prevents deterioration from deck drainage Use of elastomeric bearings 4 Illinois reported use of SRE bearings since the 1970s with excellent results. Iowa reported using simplified bearings with noncorrosive elements and preferred bearings with neoprene. Use of thicker or improved PTFE plates 2 For longer wear Use of keepers with PTFE sliding bearings 1 Prevents creeping out Use of etched, bonded, and recessed PTFE surfaces 1 Prevents creeping out Use of galvanized anchor bolts 1 Prevents component corrosion Limit bearing types specified 1 Only uses steel-reinforced elastomeric, spherical, or seismic isolation bearings

46 Ontario reported having no data to evaluate concerning how any special protection has extended service life. New Hampshire assumed a 10-year increase in service life with their painting program, but had no supporting data. Table 3.6 summarizes the responses concerning protection for steel elements in bearings. requirements For speciFying new hlmr bearings The DOTs were asked whether they specified certain manu- facturers or performance requirements when specifying new bearings, particularly HLMR types. Most respondents reported specifying performance requirements such as loads, move- ments, and rotations, and they have standard supporting spe- cial provisions. Some respondents detail a specific type on the contract plans, but allow substitutions based on provided loads, movements, and rotations. Frequency anD types oF bearing maintenance Some respondents reported routine bearing maintenance, but most reported maintenance as occasionally or rarely performed. Cleaning and washing, cleaning and painting, lubricating, resetting, unfreezing, and replacing were the types of maintenance reported. For some bearing types, the reported maintenance correlated with the types of problems associated with that bearing type. For example, maintenance for fabricated steel bearings included cleaning and painting, resetting and unfreezing (all associated with steel corrosion and tipping). Maintenance for elastomeric bearings included resetting (associated with walking out and slipping), and for PTFE bearings included unfreezing (associated with PTFE wear or binding). reasons For bearing replacement The reasons reported for bearing replacement also correlated well with the problems experienced and reported by the responding DOTs. Table 3.7 summarizes the responses. The most common reasons reported for bearing replacement were corrosion and freezing of older fabricated steel bearings, typically pertaining to steel bridges. Frequency was reported occasionally to rarely. It was not uncommon for these types of bearings to be replaced when older bridges were rehabilitated. When replaced, SRE bearings or other types were used. In some cases, excess rotation or tipping was reported, and resulted in either replacement or resetting. Replacing high-profile steel bearings due to seismic retrofit was also reported by some respondents. Elastomeric bearings have been replaced or reset, but frequencies are typically reported as rare. Walking out from under girders was the most reported reason, but still only in a few instances. Other reported reasons were excess vertical or horizontal deformation, being undersized for loads (which also results in excess deformations), and material failure or degradation (which was reported to occur rarely). California Table 3.6. Special Protection for Steel Elements in Bearings Steel Element Protection No. of DOTs Reporting Comments Hot-dip galvanizing 3 Hawaii galvanizes bearing plates with elastomeric bearings. Iowa reports most steel components are galvanized. Metalizing 2 Inorganic zinc shop coat 5 Ontario also reported using epoxy–polyurethane field paint system. Painting system same as steel girders 6 New York uses a paint system for bearings even with weathering steel bridges. Texas reports that steel bearings have always been painted. Use of weathering steel 5 Iowa reports only a few parts use weathering steel. Nebraska uses weathering steel for sole plates. New Hampshire and Texas use weathering steel in some cases. Virginia uses weathering steel when joints are eliminated. Table 3.7. Reasons for Bearing Replacement Bearing Types Reasons for Replacing No. of DOTs Reporting Elastomeric Excess vertical or horizontal deformation 3 Undersized for load capacity 1 Material failure 1 Walking out 3 PTFE Stainless or PTFE plates debonding 1 HLMR (pots) Leaking elastomer 1 (rarely) Fabricated steel bearings Corrosion 13 Freezing or nonfunctioning 5 Tipping or excess rotation 2 Seismic retrofit 5

47 reported having to replace elastomeric bearings on rare occasions due to curling or delamination due to prestress shortening (i.e., excessive shear deformation). For sliding PTFE bearings, stainless or PTFE plate debonding was reported by a few states. provisions in new Designs to Facilitate bearing replacement The DOTs were asked whether they incorporate any provisions in new bridge designs that would facilitate future bearing replacement. They were also asked whether their provisions for bearing replacement would allow the work to be done with minimal traffic disruption. Many respondents reported that new designs allow for bearing replacement by providing jacking points with sufficient capacity to raise the girders. Providing adequate clearance below to install jacks was also mentioned. Most DOTs reported that traffic would have to be moved away from the jacking location while jacking, but traffic could still be maintained on the bridge. Blocking the structure immediately after jacking would allow all traffic to resume while actually replacing the bearings with minimal disruption. Table 3.8 summarizes the various responses. use oF integral pier caps to eliminate bearings The DOTs were asked if they used integral pier–bent caps as a way to eliminate bearings and to indicate the frequency and purpose of their use of integral construction. Most of the respondents reported that they have rarely or never used integral pier caps as a way to eliminate bearings. The few that did indicate this technique as used routinely were primarily referring to use at integral abutments. Several respondents reported using integral piers occasion- ally for resolving vertical clearance issues. Other respondents reported use of integral piers for accommodating sharp skews, for eliminating deck joints, and for frame behavior in seismic events. Table 3.9 summarizes the responses. types oF integral pier–bent caps The DOTs were asked what types of integral pier–bent caps they have used. CIP reinforced concrete construction was the most commonly reported type, but most respondents were referring to construction at integral abutments. CIP post tensioned caps, steel boxes, and twin steel-plate I-girders were also reported as types for integral pier construction. Table 3.10 summarizes the various responses. Table 3.8. Provisions to Facilitate Bearing Replacement and Minimize Traffic Disruption Provisions to No. of DOTs Reporting Comments Facilitate Bearing Replacement Provide jacking points and adequate clearance 9 California supports bridge for dead load and live load to replace bearings. Provide recesses in steel plates versus welding to allow easier removal of parts 1 Use reinforced elastomeric bearings 1 They have had good performance with this bearing type and do not anticipate replacement. Do not incorporate any provisions 6 Minimize Traffic Disruption Use jacking and blocking with some traffic disruption 4 Traffic resumed after blocking structure. Use hydraulic jacks to lift deck 1 No report on traffic impact Allow jacking with traffic 1 In some instances Support bridge for dead load and live load 1 California allows 0.5-in. bump across deck joints. Replace bearings at night with temporary road closure 2 Remove live load 1 Normally close bridge for rehabilitation 1 Do not incorporate any provisions 5 Virginia said done on case-by-case basis. Table 3.9. Purpose for Using Integral Construction Purpose Used Routinely Used Occasionally Used Rarely or Never To eliminate bearings 4 1 12 For vertical clearance issues 1 7 11 Other 4 1

48 bearing orientation For curveD girDer anD skeweD briDges to accommoDate thermal expansion The DOTs were asked how they orient bearings on curved and highly skewed bridges to accommodate thermal expansion. Most DOTs responded to the curved girder bridge question, citing two methods of orientation. The first method was to orient the bearings to match the orientation of the girders at the particular expansion bearing location. The other method was to orient the bearings along a chord from the fixed location to the particular expansion bearing. Nearly equal numbers of respondents reported using each method. For sharply skewed bridges, those that responded typically oriented the bearings in line with the girders. California, how- ever, reported orienting bearings perpendicular to the skew. Table 3.11 summarizes the various responses for curved girder bearing orientation. accommoDating transverse thermal movements on wiDe briDges The DOTs were also asked how they accommodate transverse thermal movements on wide bridges. The most common response was to use elastomeric or multidirectional bearings that can accommodate the transverse movement and can also provide additional lateral clearance to guide bars or other lateral restraints. Some states reported using longitudinal deck joints that limit the amount of potential transverse movement. Some also reported that they do not have wide bridges and have not experienced a problem. Dot research programs Dealing with perFormance anD Durability oF briDge bearings The DOTs were asked if they have been involved in any research programs dealing with performance and durability of bridge bearings. A few indicated they have supported research, but most had not. The state of Washington has been very active with NCHRP projects at the University of Washington. Expansion Joints, Joints, and Jointless Bridges Bridge elements are subjected to various loads, including traffic and environmental loads. The applied loads result in movement of bridge elements. The most important factor affecting the service life of bridges is how thermal expansion and contraction of the bridge elements is addressed. Review of the available literature and discussions and meet- ings with other researchers and organizations were used to identify the status of the practice and identify the challenges. The following sources of information were used to achieve this objective: • Results of a literature search; • Analysis of the DOT survey responses; • Review of unpublished research data conducted by the Construction Technology Laboratories and sponsored by FHWA in the late 1990s; • Review of the details listed in the final draft of “Connection Details for Prefabricated Bridge Elements and Systems,” which was developed by a FHWA sponsored project. The research team had obtained the 2011 draft of this publication from FHWA. This report has since been published (2012) and is now available through the FHWA website; • Input from AASHTO bridge subcommittees; • Input from researchers outside the team; • Input from industry; and • Input from SHRP 2 review panel and staff. Available information indicated that within current practice, the thermal movement of the bridge is handled in two distinct ways. One option is to provide expansion joints at designated locations within the superstructure. By doing so, the designer forces the entire thermal deformation to take place at these discrete locations. The other option is to make the super- structure and deck continuous and assume that the thermal movement will be accommodated by flexibility of the sub- structure elements, such as piles. In such cases, the movable joints (commonly referred to as expansion joints) are usually moved away from the bridge abutment and placed near the end of the approach slab. The most common practice to accommodate thermal expansion and contraction has been to provide expansion Table 3.10. Types of Integral Pier Caps Types of Construction No. of DOTs Reporting CIP reinforced concrete caps 12 CIP posttensioned concrete caps 4 Steel box girders 5 Twin steel-plate I-girders 1 Steel integral caps 1 Inverted T caps with continuous deck 1 Table 3.11. Bearing Orientation for Curved Girder Bridges Bearing Orientation for Curved Girder Bridges No. of DOTs Reporting In line with girder 8 On chords 7 Use HLMR bearings to accommodate movements 1 No criteria: leave to design engineer 2

49 joints over the abutments and piers. Numerous expansion joint types have been developed, mainly by private companies. The fact that transportation agencies have not been able to agree on one or even a few expansion joint types is indicative of the challenges related to the effectiveness and service life of expansion joints. Almost all expansion joints leak, and most, even with proper maintenance, have a service life of less than 10 years. It is often said that the best joint is no joint. Expansion joints are perhaps the most important factor affecting the deterioration of bridge elements. Results of the survey con- ducted clearly indicate that elimination of expansion joints can greatly enhance the service life of bridges. A leaky joint allows salt and other chemicals to penetrate below the deck and causes many maintenance and deterioration problems. A FHWA study (Fincher 1983) reports that over a 5-year period, more than 60% of joints evaluated were leaking water, and that the other 40% had problems that were actually decreasing their service lives. Another study conducted by Wallbank (1989) evaluated 200 concrete bridges and found leaking expansion joints to be the major cause of bridge element deterioration. The trend in recent years has been to eliminate expansion joints or to reduce their numbers. Developments in jointless bridges and ideas such as link slab reflect this trend. Surpris- ingly, investigations carried out in the joints area of research have been relatively small compared with topics related to other bridge elements. Unfortunately, quantitative data on long-term performance of various expansion joints are, for the most part, not available. The service life of bridges could be enhanced significantly if expansion joints were eliminated altogether or moved away from the bridge superstructure and into the roadway (the end of the approach slab). The common assumption made in jointless bridges is that the flexibility of the substructure is the mechanism that allows thermal expansion and contraction, allowing the joint to be placed off the bridge at the end of the approach slab. Recent work in Australia by Bridge et al. (2005) and Griffiths et al. (2005) has resulted in the development of a completely jointless bridge, with no expansion joint anywhere in the system, even outside the bridge or at end of approach slab. In Australia, several such bridges have been constructed; to date, their performance has been excellent. A very important observation with respect to jointless bridges is the variation in design requirements among those states that use jointless bridges. Results of a survey and other related activities indicate that there is a wide range of variation in design approach used for jointless bridges. Most important variations deal with maximum bridge length as a function of the material type, effect of skew, design of piles, and details over abutment and pier. Most states using jointless bridges report observed cracking at the abutment location, especially for bridges with skew. These states report that they have not found an effective way to eliminate the observed cracking. Review of DOT practices and available design provisions for jointless bridges indicated a need for the development of uniform design provisions and accompanying details for joint- less bridges. Further, there is a need to develop methodologies and details using a scientific approach that could allow extending maximum bridge lengths for jointless bridges. The need for having a scientific approach for the design and con- struction of jointless bridges along with appropriate details is magnified by the fact that the most effective way to enhance the service life of existing or new bridges is to eliminate joints as much as possible. Contrary to common belief, jointless integral abutment bridges have lower initial cost, in addition to many other advantages. Therefore, there is no major dis- advantage in using jointless bridges. In the immediate future, bridges in many parts of the country will continue to incorporate joints, despite the many long-term advantages of jointless bridges. Thus, it was con- cluded that it was important to address the service life design of expansion joints as well as to address the design, construction, and maintenance of jointless bridges. The current trend in bridge engineering is to have modular bridge systems. One of the most important characteristics of modular bridge systems is the inclusion of many different joint types. FHWA has recently published a manual containing different possible joints that could be used in conjunction with prefabricated bridge elements. The integrity of the modu- lar bridge system depends, to a large extent, on the durability and service life of the various joints used in these systems. Following are brief descriptions of some of the major find- ings during the initial phase of the R19A project with respect to jointless bridges, expansion devices, and completely joint- less or seamless bridge systems; a summary of DOT surveys related to expansion joints and jointless systems is also pro- vided. These findings influenced the scope of research carried out in the area of jointless systems and expansion devices. Jointless Bridge Systems Henry Derthick, former engineer of structures at the Tennessee DOT, once stated, “The only good joint is no joint.” Most states have adopted design procedures that reduce or eliminate the use of joints for short to moderate span lengths. However, design methods and construction details vary significantly from state to state. In jointless bridges, joints are eliminated over the abutment or pier, or both. For long-span bridges, the flexibility of piers is sometimes used to accommodate bridge expansion and contraction within the bridge length. In these situations, the piers have to be flexible enough to accommodate expansion and contraction movements within the span lengths. When pier flexibility is relied on to accommodate bridge movement between piers, the formation of cracks in concrete piers needs

50 to be a design consideration. Current practice in jointless bridge design demands placing a joint outside the bridge, that is, either just outside the abutment or at the end of the approach slab. Various details could be used to accomplish this objective. Details connecting the superstructure to the piers in jointless bridges can also vary. Three main options are identified: • The superstructure could be made integral with the sub- structure, which has the advantage of eliminating the need for bearings altogether. However, movement of the super- structure can result in the development of cracking in the piers, which could result in service life problems, especially if piers are located within splash zones. • Connections between the pier and the superstructure could be made to act as a pin (commonly referred to as “fixed” in design drawings). This option will demand bearings and will require designing the pier for bridge movement. The bearing design for this option will require accommodating rotation only. • Connections between the pier and the superstructure could be made to act as a roller (commonly referred to as “expansion” in design drawings). This option will require having bearings and designing them to accommodate hor- izontal movement as well as slight rotation. The pier design does not, however, require accommodating deflection in the traffic direction. The design of piers and piles at the abutment is one area in which there are differences of opinion on details and design provisions and for which a need for further research was identified. Some states orient steel piles so that bending is about a minor axis; others orient them so that bending takes place about a strong axis. The effect of soil surrounding the piles at the abutments is accounted for by determining the point of fixity, where piles could be assumed to act as a fixed cantilever and behave as a flagpole. Pile designs are generally dictated by capacity, ductility, and stability. Piles at the abutment for integral systems behave as a flagpole carrying a vertical load. Therefore, capacity and stability should be a design consideration. Extreme thermal movement could also create a low cycle fatigue concern. As a result, ductility should also be a design item. The important factor that limits the maximum length of jointless bridges is the capacity, stability, and ductility of piles at the abutments. As much as possible, attempts should be made to use integral abutments. In current practice, DOTs resort to semi-integral abutment details when they think that piles are not capable of providing adequate capacity, stability, and ductility. The most desirable detail over the pier would be to make the superstructure and substructure integral. This will require the pier to accommodate thermal movements without causing any adverse consequence, such as the formation of cracks and resultant durability problems. The following is a discussion of current practice related to jointless systems and details. Details useD over abutments Experimental studies of integral abutment bridges began in the 1930s (McCullough 1930). Early bridges of this type were relatively short, ranging in length from 50 to 100 ft. Subsequent increases in allowable length were based empirically on reports of successful performance of prototype bridges. Over time, various highway agencies have developed their own design criteria along with concomitant limits on length, skew, and horizontal curvature. In January 1980, FHWA released Technical Advisory T 5140.13, Integral, No-Joint Structures and Required Provisions for Movement. This advisory was aimed at providing state and local highway agencies with data and state-of-the-art information pertaining to integral abutments, continuous bridge lengths, and specification-oriented movement require- ments. Since this time, further progress has occurred in the development of various jointless bridge concepts, including • Full integral abutments, in which bridge girders are cast into a concrete end diaphragm that is connected to a concrete pile cap typically supported by a single row of piles; • Semi-integral abutments, in which the concrete end dia- phragm is not rigidly connected to the substructure; • Deck extensions, in which the end of the deck slab is simply extended over a traditional backwall and into the adjoining approach pavement; and • Shifted joint details, in which the joint is shifted between a semi-integral diaphragm and a fixed backwall. This detail is not currently used in practice. Full integral abutments are used commonly in many states for bridges having maximum total lengths from 250 to 400 ft. Thermal movements are accommodated within the foundation and are typically assumed to be unrestrained in the design of the superstructure. Within current practice, measures to accom- modate thermal movements vary significantly (Burke, Jr. 1990). These measures include • Limiting the bridge length, skew, horizontal curvature, or some combination of these factors; • Use of select backfill materials or uncompacted backfill, or both; • Spanning the area disturbed by the foundation movements immediately behind the abutments with the approach slab, thus avoiding settlement of the slab and associated surcharge loads; • Limiting the foundations to a single row of vertical piles; • Limiting the pile type and requiring a minimum pile length;

51 • Orienting vertical H-piles such that they are subjected to weak-axis bending due to longitudinal movements; • Providing a hinge detail within the abutment to limit the moment developed at the tops of the piles; • Anchoring the approach slab to the superstructure with a detail that allows rotation of the approach slab at the abut- ment to accommodate settlement of the approach fill; and • Providing an expansion joint at the roadway end of the approach pavement. Some of the items listed are, in fact, the opposite of what should be done. For instance, turning the piles about a weak axis of bending has a net effect of reducing the maximum bridge length that could be used as a result of reduction in pile capacity, ductility limit, and stability limit. Details For jointless briDges at interior supports oF multispan briDges Various details were found to be used over the interior supports of multispan bridges to eliminate joints, including simple spans for dead load made continuous for live load, link spans, and engineered cementitious composites. Dead Load Made Continuous for Live Load One of the concepts implemented by owner agencies for concrete bridges has been the use of simple spans for dead load made continuous for live load (Freyermuth 1969; Oesterle et al. 1989). A common implementation of this type of construction involves the use of precast, prestressed girders connected with a continuous CIP deck slab. Girders are simply supported for dead load, but continuity is achieved with deck steel as negative moment reinforcement over the piers. In addition, the girders are made integral with the interior pier diaphragms. The concept of a simple span made continuous has also been applied to eliminate interior joints and improve the construction speed and design economy for short- and medium-span steel girder bridges (Azizinamini et al. 2008; Azizinamini 2013). The concept of a simple span for dead load and continuous for live load for steel bridges has many advantages, among them completely eliminating joints and accelerating construction. Unfortunately, some simple made continuous prestressed concrete girder bridges have experienced severe cracking in the girders near the interior diaphragms. One example that has been studied extensively was on the Francis Case Memorial Bridge spanning the Washington Channel of the Potomac River in the District of Columbia (Telang and Mehrabi 2003). The prime cause of this distress was the restraint of upward creep of the prestressed girders under the influence of pre- stressing. According to Telang and Mehrabi, “By providing a large amount of positive moment reinforcement at the dia- phragms, designers inadvertently make the diaphragm area stronger than the adjacent girder sections, thereby forcing the cracking to occur in far more critical but weaker areas of the girder span.” The article states, “In closing, it is important to note that this seemingly simple transformation of simple-span prestressed girders to continuous spans should be attempted with caution, and significant attention must be paid during analysis and design to include loading conditions that can cause counterintuitive behavior such as secondary positive moments at the piers. Most importantly, positive moment reinforcement should be designed and detailed such that any cracking, if it occurs, should be limited to the relatively less critical diaphragm region of this type of structural system.” Further discussion of this problem and solutions to avoid it have been published by Oesterle et al. (2004) and Arockiasamy and Sivakumar (2005). Link Slab Link slab is a type of detail that is used in conjunction with existing or new bridges having girders that act as simple beams for both dead load and live load. In this type of deck detail, the slab spans continuously over the length between the adjacent girders while the adjacent girders are kept as simple spans (see Figure 3.33). The length of the deck con- necting the two adjacent simple-span girders is called the link slab (Caner and Zia 1998). Link slabs generally require less deck reinforcement, but they have more girder positive moment Figure 3.33. Conceptual detail for link slab.

52 demands than simple-made-continuous designs. It is believed that Caner and Zia (1998) were the first to develop the link slab idea. Limited analysis and laboratory experiments were carried out, and design recommendations were provided (Caner and Zia 1998). The use of this promising detail has been very limited due to field-observed cracking. Link slabs are not common in the snow belt states. A crack is invariably formed due to deck slab rotation as the bridge is loaded with live load. Recently, Wing and Kowalsky (2005) described the monitor- ing and assessment of a pilot study link slab bridge in North Carolina. To eliminate the observed cracking problem, Kim et al. (2004) have researched the application of engineered cementitious composites (ECC) to enhance crack-width control, deformation capacity, fatigue performance, and con- struction and placement of link slabs. From 1995 through 1998, Caner and Zia published several reports and papers that synthesized results from research conducted over a number of years focusing on the behavior of jointless bridge decks supported by simple-span girders. They addressed instantaneous effects due to live load plus impact, thermal effects due to temperature variations, and time-dependent effects due to creep and shrinkage of the con- crete. Both experimental and analytical studies were conducted, and a method of design for the deck region between simply sup- ported girders, referred to as the link slab, was developed based on the research results. The majority of the studies and the development of the design recommendations are documented in Caner (1996) and Zia et al. (1995). Caner and Zia (1998) provide a summary of the developments. The experimental program involved two composite speci- mens, including a continuous reinforced concrete deck slab cast on two simple-span steel beams, and a similar slab cast on two simple-span precast concrete beams. The steel bridge was tested with four support configurations: HRRH, RHRH, RRRR, and RHHR, where H stands for hinge and R stands for roller. The first and last letters indicate the conditions at two exterior supports, and the other letters indicate the con- ditions at two interior supports. The concrete bridge was tested with the same support conditions, except for the RRRR configuration, which is an unlikely condition in the field. The goal in testing different support conditions was to observe if there were any differences in the behavior of the jointless deck under different support conditions. In all cases, the beams were loaded to no more than 40% of the estimated ultimate load capacity to observe behavior in the elastic range. The final ultimate load test was performed using the RHHR support configuration. For each specimen, test results showed that the load–deflection behavior at the midspans of the girders, the link slab rebar stresses, and the crack widths in the link slab were essentially the same for all support conditions. The researchers concluded that the link slab offered negli- gible rotational end restraint to the bridge girders and that the link slab can be analyzed as a beam subjected to the same end rotations as the adjacent girders. It was also found that the actions within the link slabs were predominantly flexural, which was contrary to a number of previous analytical solu- tions that assumed that the link slabs acted primarily as an axially loaded element. The researchers found that under service-load conditions, the link slab would crack, primarily due to bending. In addition, the authors found that earlier programs developed by Gastal (1986) and El-Safty (1994) were capable of predicting the forces, stresses, and crack widths in the link slab due to thermal and creep and shrinkage effects. Caner (1996) modified the programs developed by Gastal and El-Safty to properly capture the link slab actions. All the solutions were based on beam theory. The reinforcing bar stresses compared reasonably well with the data mea- sured from the experimental tests. The predicted crack widths were somewhat larger than the measured crack widths. The researchers concluded that bending and cracking under live load plus impact were the governing factors that must be con- sidered in the design of the link slab. Although the program as modified by Caner (1996) was believed to be capable of predicting the forces, stresses, and crack widths in the link slab due to thermal and creep and shrinkage effects, the authors recommended that these predictions would need to be vali- dated by future tests, preferably under field conditions. Caner and Zia proposed a procedure for link slab design that focused on estimating the moment in the link slabs based on the rotations imposed by the simply supported girders and assuming cross-section properties for the link slab. Given this moment, the longitudinal reinforcement for the link slab is designed using a conservative working stress such as 40% of the yield strength of the rebar. Debonding of 5% of each girder span adjacent to the link slabs was recommended, although the impact of not debonding over this length does not appear to have been addressed. The crack-control criteria of the AASHTO Standard Specifications for Highway Bridges was used to limit the crack width at the surface of the deck to 0.013 in., considered appropriate for exterior exposure (AASHTO 1996). Five hypothetical jointless bridges were considered by Caner (1996) to illustrate the recommended design procedure. In all cases, loading a single span with a single HS20-44 truck load did not cause the link slabs to crack. However, loading two adjacent spans caused cracking in the link slabs. Caner observed that longer spans produced larger link slab stresses due to larger end rotations. When thermal and creep and shrinkage effects were included in the analysis, the crack widths and rebar stresses in the link slab depended on the number and location of the horizontal restraints on the bridge. When the bridge was hor- izontally restrained at only one of its supports, the link slab stresses due to thermal or time-dependent effects were very small because the bridge was assumed free to move horizontally

53 on unrestrained supports. If horizontally restrained bridges were subjected to thermal or creep and shrinkage effects, it was found that the link slab rebar stresses could exceed the yield strength of Grade 60 bars. In such cases, it was recom- mended that the size of the rebars could be increased or the spacing of the bars could be decreased to better control the rebar stresses and crack widths. Caner and Zia suggested a design of the link slab using only one layer of rebar placed near the top of the deck, but suggested that two layers could be used to improve performance in bridges having horizontal restraints. Caner (1996) also suggested con- sideration of the use of fiber-reinforced concrete in the link slab region to improve the tensile capacity of the concrete, as well as to improve crack control. He also suggested the use of galva- nized or epoxy-coated rebar in the link slabs when corrosion protection is warranted. Tests under repeated loading were also suggested to evaluate the link slab response. In more recent work, the North Carolina DOT has imple- mented the recommendations by Caner and Zia in a single field study. The results of this research are documented in Wing and Kowalsky (2005). Kim et al. (2004) have considered the use of fiber-reinforced concrete in link slab construction. In summary, within this project it was concluded that behavior of link slab was not well understood and there was a need for additional work. Engineered Cementitious Composites ECC is a fiber-reinforced, cement-based composite material that achieves high ductility under tensile and shear loading. Maximum ductility in excess of 3% under uniaxial loading can be attained with 2% fiber content by volume. One obvious use of ECC in bridges is in negative moment regions over inte- rior supports where the concrete slab is in tension, such as the link slab method. Crack size greatly affects the permeability of the concrete, which is responsible for allowing deleterious agents to reach the reinforcement, resulting in corrosion. This corrosion, in turn, causes expansive forces resulting in spalling of the concrete cover. The self-limiting crack-size property of ECC prevents this process from occurring. Permeability tests show that ECC loaded to the strain hardening stage (1.5%) tends to behave like sound concrete (Li 2003). Completely Jointless Bridge systems Having a structure with no joints is feasible. Rail bridges with continuous welded rails, in which the rails contain absolutely no joints, are a case in point. They are designed to take some movement and resist some force without any adverse perfor- mance consequence. Very recently, a few bridges in Australia have been con- structed with absolutely no joints. The system has been in operation for more than 7 years and has demonstrated excel- lent performance. Thus far, the concept has been used with continuously reinforced concrete pavements (CRCP), which is used in a few states, such as Oregon. Figure 3.34 shows a representation of the bridge system constructed in Australia. This system replaces the conventional system shown in Figure 3.35. In a jointless bridge system, the approach slab and the joints over the abutment and approach slab are eliminated. Figure 3.34. The CRCP jointless bridge system constructed in Australia. Figure 3.35. The conventional joint system in bridges consists of two joints on both ends of the approach slab.

54 The concrete deck is made continuous over the abutment and is connected to the pavement. The system is designed to accommodate the resulting tension and compression forces generated by making the deck and pavement seamless and continuous. The transition zone in the system, which has a length equal to approximately the bridge length, is used to accommodate movements associated with the bridge superstructure when it contracts. The transition zone is also designed to take the result- ing compression loads without buckling. A detailed descrip- tion of the system is provided by Griffiths et al. (2005). Synopsis of Literature Search and Other Available Information on Expansion Joint Systems The amount of research on the durability, behavior, and perfor- mance of expansion joints is surprisingly limited. In recent years, however, more attention has been paid to expansion joints than previously. This is evidenced by a 1994 British col- loquium (Pritchard 1994), a 1996 FHWA workshop (Burke, Jr. 1996), and a 2005 FHWA conference on the topic (Arockiasamy and Sivakumar 2005), in addition to various research studies being funded by different states (Arockiasamy et al. 2004; Fennema et al. 2004; Brena et al. 2007; Civjan et al. 2007). A review of the available literature demonstrated that an ideal expansion joint should be able to (Lee 1994): • Accommodate thermal expansion and contraction; • Accommodate movement due to traffic-induced loads; • Provide a smooth ride; • Prevent the creation of hazards and safety issues; • Accommodate needs during snow removal; • Prevent leaking of moisture and other chemicals to elements below the superstructure; • Have a long service life; • Be maintenance free or require minimal maintenance; and • Be cost-effective. Summaries of the major findings for various expansion devices are provided in the Guide. Analysis of DOT Survey The observed or estimated service lives of various expansion joints are fairly short and range between 1 and 30 years. The answers provided by responding agencies varied significantly. Part of this variation is attributed to the different climate and traffic conditions to which bridges are subjected. Nevertheless, the maximum service life is well below 100 years. Figure 3.36 summarizes the observed or estimated service life for the compression seal joint type along with maximum permitted longitudinal movements. One of the questions asked of the states in the general section of the survey regarded which freeze–thaw zones best described their climate. The three options provided were high freeze– thaw, medium freeze–thaw, and low freeze–thaw regions. The information provided in Figure 3.36 is organized based on the freeze–thaw regions the states are located in. Figures 3.37, 3.38, and 3.39 show the same information for strip seal expansion joints, modular expansion joints, and finger-plate joints, respectively. A review of the information provided in Figures 3.36 through 3.39 indicates • The variation in observed or estimated service life is the greatest for compression seal joints. • The maximum longitudinal movement permitted is the smallest for compression seal joints. • The maximum observed or estimated service life is more consistent for strip seal, modular, and finger-plate expansion joints. • The maximum observed or estimated service life is the high- est for finger-plate joint types, with some states reporting it to be about 50 years. • There are no obvious trends between the freeze–thaw region that the state is located in and the observed or estimated service life or the maximum longitudinal movement allowed for certain joint types. One of the consequences of having joints is damage to the ends of steel and concrete girders in existing bridges. Of par- ticular interest was the identification of effective retrofitting alternatives for existing bridges. Survey results indicated that states that do not use salt experience little or no damage to girder ends due to expansion joint presence. Hawaii and Arizona are among these states. This may suggest that in regions where bridges are not salted, the use of expansion joints along with proper maintenance and painting or metalizing girder ends may be allowed while achieving long service life. The previous scenario aside, many states have bridges with damaged girder ends due to the presence of expansion joints. A question was asked inquiring what techniques are used to repair these damaged ends. All respondents gave similar answers. These techniques included painting the ends of steel bridges, removing the damaged portion of the concrete and patching it with cementitious material, and sealing the expan- sion joints. Ontario was the only agency that also uses cathodic protection in addition to other techniques for repairing the damaged ends. This is a good practice, as simply patching the damaged concrete may amount to nothing more than hiding the internal corrosion problems that might, with greater con- sequences, resurface. Typical damages reported due to the presence of expansion joints included damage to girder ends; various types of damage

55 Figure 3.36. Summary of service life and maximum permitted movement for compression seal joint type. to bearings, including failure of the bearings to accommodate anticipated movements; deterioration of the substructure; deterioration of the abutment backwall and wingwall; and shifting of the bridges with skews. Figure 3.40 summarizes typical damages to various bridge systems as a direct result of expansion joint presence. This figure summarizes the frequency of typical damage as a function of freeze–thaw region. The ver- tical dimension is the number of states observing certain failure modes, and the two horizontal axes are the freeze–thaw region and the type of damage typical to various bridge elements. As expected, the frequency of occurrence of any given damage type is the highest in high freeze–thaw regions. This further confirmed that leakage of moisture and other chemicals through expansion joints, especially where roads are salted, is a major source of bridge deterioration. A large number of states (about 50%) eliminate expansion joints by making simple-span bridges continuous over piers. However, the design procedure and details used vary consid- erably among these states. States using jointless bridges agree on one issue: the types of damage seen in jointless bridges are minor, and the expected or observed service life ranges between 50 and 75 years. However, states using jointless bridges have significantly dif- ferent design provisions. The maximum bridge length that could be used in conjunction with jointless bridges varies sig- nificantly for both steel and concrete bridges. The maximum skew allowed also varies. There is, however, a trend in design provisions for jointless bridges. Every state using jointless bridges specifies a smaller maximum bridge length for steel bridges than for concrete bridges. Some states tie the maximum

56 Figure 3.37. Summary of service life and maximum permitted movement for strip seal expansion joint type. bridge length to the skew angle. As skew angle increases, the maximum allowed bridge length is reduced. Some states, especially in the West Coast area, allow jointless bridges for concrete structures and do not have specific design provisions for steel bridges. In summary, the lack of a national standard for jointless bridges has resulted in limited use, with significantly different design provisions among states using them. One approach to eliminate joints is to make the concrete deck continuous over piers and still have the girders act as simply supported spans by using a link slab. This idea was originally developed in North Carolina and was further researched in Michigan by incorporating the use of fiber- reinforced concrete. The number of states using a link slab as a means for eliminating joints in simple-span bridges is very small. Part of this trend could be related to the short history of link slab use, field-observed cracking problems, and a lack of AASHTO-approved design provisions. None of the states responding reported use of a link slab with fiber-reinforced concrete in continuous girder bridges. Ideas similar to link slab have the potential to replace the reinforcing requirement in Section 6.10.1.7 of the LRFD Specifications, which is meant for crack control over predominately negative flexural regions. In addition, link slab appears to be the only method for con- verting existing simple-span bridges into simple-span bridges with a continuous deck over pier. Analysis of the published data, discussion with other researchers, and survey analysis indicates that almost all expansion joints have a relatively short service life and will eventually leak. Finger-plate joints are reported to have the best performance; however, it is important to remember that finger-plate joints are used on relatively longer-span bridges and are typically subjected to better maintenance programs. However, joint- less bridges are reported to have satisfactory service life, with an observed or expected service life from 50 to 75 years. Indica- tions are that observed damage at abutments is small in jointless

57 Figure 3.38. Summary of service life and maximum permitted movement for modular expansion joint type. bridges and can be attributed to lack of available, scientifically based procedures to accommodate movement. Fatigue and Fracture Fatigue and fracture can result when tensile stresses are applied to members and connections containing inherent flaws at weld toes or bolt holes due to fabrication. These flaws could be mitigated during design through selection of more fatigue-resistant details or during construction by eliminating or carefully selecting locations for field welding. Cracks can grow from welded details, as well as bolted details. Base metal does not typically govern designs, as the flaws in welded and bolted details are more critical and propagate before base- metal flaws. The governing stress parameter for fatigue is stress range; fracture is controlled by total tensile stress. Details exhibit varying degrees of stress concentration with inherent flaws of varying size. Stress concentration is more predominate in details with lower fatigue resistance, and flaw size is more predominate in details with higher fatigue resistance. Fatigue of metal structures is the steady state propagation of preexisting flaws due to repetitive loads below the critical loads for strength. Steel bridges do not fail in fatigue, but in fracture, which is the unstable propagation of a larger flaw (most likely the result of fatigue). Fatigue of steel structures is categorized in the LRFD Specifications as either load-induced fatigue (6.6.1.2) or distortion-induced fatigue (6.6.1.3). According to the LRFD Specifications, When proper detailing practices are not followed, fatigue crack- ing has been found to occur due to strains not normally com- puted in the design process. This type of fatigue cracking is called distortion-induced fatigue. Distortion-induced fatigue often occurs in the web near a flange at a welded connection plate for a cross-frame where a rigid load path has not been

58 provided to adequately transmit the force in the transverse member from the web to the flange. Fatigue due to the stresses computed during the design process (typically in-plane stresses) is called load-induced fatigue. Synopsis of Published Literature and Other Search Activities Research studies beginning in the 1970s have resulted in the development of better details and design approaches that have greatly reduced, if not eliminated, the potential for fatigue and fracture of new steel bridges. Typical modern steel bridges, designed in accordance with the LRFD Specifications, rigorously address fatigue and fracture issues. Bridges designed before 1974 may still experience load-induced fatigue cracking; those designed before 1984 may still experience distortion-induced fatigue cracking. Fracture resistance is addressed in the current LRFD Speci- fications by material toughness requirements that vary with certain climatic temperature zones. Description and Discussion of Details Reported to Cause Fatigue Cracking Details consisting of connections between, and attachments to, steel bridge components can be susceptible to fatigue cracking. In-service fatigue cracking is the propagation of preexisting inherent flaws due to fabrication. Table 6.6.1.2.3-1 of the of the LRFD Specifications describes typical steel bridge details, both welded and bolted, which are susceptible to load- induced fatigue cracking if not properly designed. These details are grouped in eight detail categories (A, B, B′, C, C′, D, E, and E′) Figure 3.39. Summary of service life and maximum permitted movement for finger-plate joint type.

59 according to fatigue resistance, with letters higher in the alphabet denoting details with lower fatigue resistance. For example, fillet welded connections of girder flanges to their webs is defined as detail Category B. Transverse connections plates welded to girder webs and flanges are detail Category C′. Causes of Observed Fatigue Cracking Modern steel bridges should not exhibit fatigue cracking or fractures if designed and detailed in accordance with the LRFD Specifications. Bridges designed before 1974 for load- induced fatigue and before 1984 for distortion-induced fatigue may exhibit cracking, as these types of fatigue were not con- sidered in their design and detailing as they are today. Load-induced fatigue cracking in older bridges (designed before the mid 1970s) is a result of the lack of design criteria when they were originally designed. These cracks result when the fatigue stress range exceeds the fatigue resistance. Distortion-induced fatigue cracking in older bridges (designed before the mid 1980s) is a result of the lack of prop- erly codified detailing practices. These cracks typically occur when girders are not detailed to act as a complete system. For example, transverse connection plates not rigidly connected to girder flanges can result in fatigue cracks in the unreinforced web gaps. Ranking of Details with Respect to Reported Performances The detail categories of Article 6.6.1.2.3 of the LRFD Specifica- tions rank the fatigue resistance of typical steel bridge details. They are ranked according to fatigue resistance (or limiting stress range to achieve a 75-year life), although no detail is ulti- mately better than the next as the applied stress range defines their acceptability. For example, an overstressed Category B detail is worse than an understressed Category E detail. Analysis of DOT Survey DOTs report fatigue cracking of older bridges designed before the adoption of the current load-induced or distortion-induced fatigue design provisions. Figure 3.40. Deterioration types as a direct consequence of having bridge joints.

60 Methods for Improving Service Life of New Structures Bridge systems, subsystems, or components alone do not exhibit fatigue cracking. It is the welded and bolted details of these components that can be susceptible to fatigue. Bridge details designed in accordance with LRFD Specifications should not experience fatigue crack propagation during the specified 75-year design life. Bridges with detail Categories A through C′ typically are designed for infinite life. There is no improve- ment necessary for these bridge details. In some cases, bridges cannot be designed with only detail Categories A through C′; finite-life design may be necessary for detail Categories D, E, or E′. To extend the 75-year design life to 100 years, the num- ber 100 should be substituted into LRFD Equation 6.6.1.2.5-2 in place of the number 75, as suggested in the commentary to Article 6.6.1.2.5. Thus, new bridge systems, subsystems, and components should only use these details with well-defined fatigue resistance that can be accommodated through proper design and detailing. Methods for Improving the Service Life of Existing Structures When a refined remaining life calculation suggests low remain- ing fatigue life, which may occur for bridges that were designed before the establishment of modern fatigue provisions, ultra- sonic impact treatment (UIT) can be applied to details to enhance their remaining load-induced fatigue life through the introduction of beneficial compressive residual stresses. Note, however, that in many cases, remaining fatigue-life calculations yield low or negative lives due to crude estimates of stress along with the high degree of uncertainty associated with fatigue resistance. Research has demonstrated that UIT can improve the load-fatigue resistance of an end-welded cover-plated beam (detail Category E) up to detail Category C or better, and the fatigue resistance of transverse stiffeners (detail Category C) up to detail Category B or better. AASHTO is currently considering codifying a one-category fatigue- resistance improvement for welded details properly treated with UIT. For bridges with details susceptible to distortion-induced fatigue cracking (details currently prohibited by AASHTO), two retrofit solutions exist: eliminating the driving force of the distortion or providing a positive attachment (welded or bolted) across the short flexible gaps. A catalogue of distortion-induced fatigue cracking retrofits could be compiled. Methods for Improving the Efficiencies of Safety Inspection Procedures Normal safety inspection procedures have been efficient in locating early-stage fatigue cracks in existing bridges during routine biennial inspections. However, newer enhanced moni- toring systems using real-time methodologies can be very effective in monitoring critical locations. Methods for Improving the Efficiencies of Maintenance Procedures Maintenance or lack of maintenance has little or no effect on fatigue resistance. Active corrosion does not allow the devel- opment of propagating flaws as the corrosion in effect blunts the crack tips by “eating away” the stress concentrations of the sharp crack tips. Only arrested corrosion could cause accelerated crack propagation through increased stress concentrations. In such cases, any severe stress concentrations due to corrosion should be ground smooth before painting. Conclusions Steel highway bridges designed since the adoption of the current fatigue provisions should not exhibit cracking during their 75-year design lives. Extending this design life can easily be accomplished by modifying the finite-life equation. For most common bridge details (detail Categories A through C′), infinite life is often already provided. Bridges consisting of new systems, subsystems, and components should continue to use these welded and bolted details with well-defined fatigue resistances. If the details of any new systems, subsystems, or components cannot be categorized according to the detail categories of the LRFD Specifications, fatigue-resistance exper- imentation would be necessary to quantify the resistance required to achieve 100 years or more of service life. Protection of Steel Bridges for Corrosion Corrosion is a fundamental limitation of steel as a material of construction, and it is one of the primary causes of reduced service life for steel bridges. In its simplest form, corrosion of steel is the result of its exposure to oxygen and moisture and is accelerated in the presence of chloride ions from roadway deicing salt or from other sources, like seawater. The use of deicing agents composed chiefly from materials with readily soluble chloride ions creates an atmosphere in which unpro- tected steel corrodes quickly. The precursors of steel for metal bridges, cast and wrought iron, were largely immune to cor- rosion because the very high or alternately very low carbon content caused them to corrode much more slowly than steel. To achieve long-term service life, the corrosion potential of steel for a given environment must be addressed as part of an overall corrosion protection plan. The R19A research program conducted a literature and industry survey along with a survey of state DOTs to identify current approaches to structural steel corrosion protection.

61 Results of these surveys combined with research team experi- ence were further used in developing Chapter 6 of the Guide, which relates to corrosion protection of steel bridges. The following subsections summarize the literature and industry surveys, and a summary of the DOT survey is given at the end. Current strategies used to prevent corrosion of steel in bridges are discussed, and further recommendations for improving service life are presented. Synopsis of Published Literature Briefly, the literature search conducted as part of this research revealed that • FHWA has an active research, development, and testing pro- gram for coatings. Recent confirmatory research has been reported on two-coat systems and their ability to perform as well as the traditional three-coat system in use since approx- imately 1965. Recent promising work on the testing of one- coat system candidate materials for steel bridges has also been reported. Future research on the use of nanotechnology in coatings is under consideration. • For new construction, the coating systems and procedures, although improved, are basically unchanged since the late 1960s. New steel receives a coating system consisting of a zinc-rich primer, an epoxy midcoat, and a urethane topcoat, all applied over steel cleaned in accordance with SSPC-SP 10 Near-White Metal Blast Cleaning. • Maintenance overcoating is a process in which a new coating is applied over an existing coating. Based on industry knowl- edge and the DOT survey information obtained in this study, the coating systems include acrylic, calcium sulfonate, epoxy sealer–epoxy–urethane, epoxy sealer–urethane, poly- ester, and polyaspartic. • Maintenance recoating is a process in which a new coating system is applied over a surface from which all old coating material has been removed. Based on the literature search, industry knowledge, and the DOT survey information obtained during this study for maintenance recoating, the most commonly used system consists of an organic or inorganic zinc-rich primer with an epoxy midcoat and a urethane topcoat. • A recent special grade of weathering steel has been used in a small bridge in California. The bridge, discussed in more detail in the subsection on using corrosion-resistant steels, is fabricated from steel meeting the requirements of ASTM A1010-01 (2001). Among other requirements, this special grade of weathering steel contains between 10.5% to 12.5% chromium. A study conducted by experts from a major steel producer indicates that uncoated steel coupons were free of corrosion when exposed at the 25-m line (from high tide) at Kure Beach, North Carolina. Synopsis of Other Search Activities Interviews with coating suppliers, resin producers, consultants, and other engineering professionals from various foreign countries were also performed and are briefly summarized as follows. It is apparent that for coating systems, the best practice abroad is similar to the best practice in the United States, which was substantiated through conversations with representatives from China, Korea, Japan, and Western Europe. Although the actual coating systems are similar, one thing notably lacking in the United States is a codification of a comprehensive set of best practices for coating steel bridges. At least one country, the United Kingdom, appears to have a good start in the codi- fication of some of the necessary practices needed to specifi- cally target achieving improvements in bridge durability via protective coatings (Volume 1, Highway Structures Approved Procedures, Section 3 General Design, Part 8 Ba57/01 Design for Durability, Parts 5 and 6). One major coating supplier interviewed did not believe that it would be possible with current technology to have a coating system not containing zinc that could last for 100 years. That same supplier did believe that a coating system contain- ing a zinc primer could last 100 years. There are evolutionary changes under way in zinc coating. One supplier uses zinc flakes in coatings to obtain both barrier effects and galvanic effects in a combined effort to fight corrosion. Another vendor cited the reported success of flake glass–reinforced polyester coating material applied in multiple layers to drilling plat- forms in the North Sea. Some vendors are supplying advanced fluoropolymer-based coatings. These materials are said by these suppliers to provide a quantum improvement in color, gloss, and resistance to ultraviolet light. The march forward in the art and science of advancing the type and durability of various coating types continues. Strategies Used to Prevent Steel Corrosion Two main strategies are used to prevent corrosion of steel in bridges: the use of coatings and the use of corrosion-resistant materials. use oF coatings Engineers have historically used coatings as one way to mitigate the negative impact of the environment on steel. Fundamentally, a sure way to protect steel from corrosion is to keep it from getting wet. A major way that coatings protect steel from moisture is by providing a barrier to the elements. The key to corrosion protection by the use of coat- ings lies in the ability of the primer to inhibit corrosion when (or if) water penetrates the barrier layer(s) and the steel surface is subjected to repeated wet–dry cycles.

62 It is generally recognized that in an effective, multicoat coating system, the primary purpose of the coating layer closest to the steel surface is corrosion protection to the steel surface beneath. Any special aesthetic considerations can be accommodated in the subsequent coating layers. In some dry climates of the country where corrosion is not an issue, aesthetic considerations can play a more compelling role. The DOT survey confirms that states such as Arizona expect 50-plus years of service life for the same system expected to last 20 to 30 years elsewhere. From the very earliest years in the steel bridge era, beginning in ~1874 in the United States with the construction of the Eads Bridge in St. Louis, Missouri, lead and chromium rust-inhibitive pigments were added to paint to supplement the barrier protection offered by a coating film. For almost 100 years, until ~1965, the use of lead- or chromium-pigmented, multilayer coatings was the norm during new bridge con- struction and maintenance overcoating and in maintenance repainting. After 1965, bridge coating engineers turned away from coatings containing these toxic heavy metal pigments and settled on coatings containing metallic zinc as the corrosion- resistant pigment. The DOT survey indicated that all respon- dents use a system that includes a zinc-rich primer. As long as the zinc pigment in the coating is in close metal-to-metal contact with the steel substrate, the coating can provide gal- vanic protection to the steel. Galvanic protection is provided when zinc and steel (iron) are connected in the presence of air (oxygen) and moisture. In this coupling of materials, zinc (the more noble metal) will oxidize (corrode) in preference to the iron (steel). The preferential oxidation of zinc provides protection for the steel as long as there is nearby zinc left to consume. When the zinc is consumed, the steel beneath will be subject to corrosion. The method of attack taken since the mid 1960s has been to use a so-called “belt and suspenders” approach. When these coating systems are applied, the zinc- containing primers provide corrosion protection to the steel and, to protect the zinc-containing coating layer from oxidation due to oxygen, moisture, and air pollutants, additional coating layers are applied over the zinc-rich primer. When a steel surface protected by a coating system using a zinc-rich primer is bathed in saltwater and subjected to many wet–dry cycles, inevitably discontinuities in the coating provide a pathway through the coating for moisture to reach the steel surface beneath. As a result, the zinc begins to react to protect the steel from corroding. Eventually, the zinc in the zinc-rich primer is consumed, and corrosion in the form of red rust is evident. It may take many decades for the corrosion protec- tion offered by zinc to be consumed before corrosion of the steel is evident. Thereafter, the rate of corrosion is dictated by the local factors surrounding the steel (e.g., wet–dry cycles, chloride contamination, humidity). In a landmark literature survey compiled by Albias, von Loden, Onderzock, and Advies, which was published in the Journal of Protective Coatings and Linings in February 1997, the researchers opined that “[i]t has been clearly established that soluble salts on the surface of steel will increase the rate of corrosion and paint breakdown for many [coating] systems now in use.” This conclusion is as true today as it was then. The authors offered several conclusions: • From available data, it is not possible to establish a definitive allowable level of chloride contaminants. • In relation to the durability of the paint system, a maximum chloride level of 10 to 50 mg/m2 is allowed, depending on the use and exposure conditions. This is only a rough guideline. • Under specific conditions, higher maximum levels of chloride (up to hundreds of milligrams per square meter) are allowed for special, durable paint systems (e.g., zinc silicate). • Exposure to marine conditions or industrial environments considerably increases the chloride contamination of steel. • Abrasive blast cleaning does not remove all the chloride. • Results of detection methods for soluble chlorides are affected by temperature, mechanical forces, and the chemi- cals and type of analytical method used. • The effect on steel of the hydrochloric acid generated as a consequence of the corrosion reaction is devastating. Therefore, the removal of as much chloride as possible during blast cleaning and other surface preparation efforts is crucial. Although there is still not complete agreement as to the precise level of chloride residue that is acceptable, Appendix A, Table A-2 of the current version of the Society for Protective Coatings (formerly Steel Structures Painting Council or SSPC) surface preparation standard SSPC SP-12 (Surface Preparation and Cleaning of Metals by Waterjetting Prior to Recoating) identifies three levels of chloride removal: • NV-1: 0 µg/cm2; • NV-2: 7 µg/cm2; and • NV-3: up to 50 µg/cm2. The level most commonly specified by the respondents to the DOT survey conducted as part of this research was 7 µg/cm2. It is believed that complete removal of all chlorides via only dry blast cleaning is at best unlikely, and at worst, provides a false sense of security, even if white metal blast cleaning (SSPC SP-5, White Metal Blast Cleaning) is specified. Cleaning efforts beyond abrasive blast cleaning are usually needed. Various types of coatings including painting, galvanizing, and metalizing, are discussed in Chapter 6 of the Guide. This chapter was developed as a best practices guide for preventing corrosion of exposed structural steel for bridges.

63 use oF corrosion-resistant steels Weathering steel (coated or uncoated) has been the subject of much research since its initial use on bridges in about 1970. When small amounts of copper, chromium, nickel, phospho- rous, silicon, manganese or combinations of these metals are added to carbon steel, a low-alloy carbon steel results that has improved corrosion resistance. These weathering steels enjoy vastly enhanced corrosion resistance that can render them impervious to corrosion in many exposure environments. The degree of corrosion resistance afforded is dependent on a number of variables including climatic conditions, pollution levels, and the degree of sheltering from the atmosphere, as well as the composition of the steel itself. These variables influence the locations where its use is recommended. When weathering steel is properly exposed, a rusty red- orange to brown or purple-tinted patina forms. When the steel is exposed to the atmosphere, the corrosion rate becomes stabilized within about 3 to 5 years. In rural areas with little pollution, a somewhat longer period may be required to form this well-adhered protective patina. The formation of the protective patina requires a series of wet and dry periods. In areas where the steel is sheltered from the rain, the dark patina is also not able to form. Rather, a layer of light rust forms and provides protection for the steel beneath. However, when weathering steel is used in locations where regular wet–dry cycles occur, the steel is corrosion resistant to the point that no coating is necessary. There are some limitations to the advantage and recom- mended use of weathering steel in certain environments. For example, in areas with high concentrations of corrosive industrial or chemical fumes, the weathering steels exhibit a much higher corrosion rate and, instead of forming the tight corrosion-resistant patina, loose rust particles are formed on the steel surface. Under these conditions, the film formed will likely provide little protection. In a saltwater marine environ- ment or in a salt-rich area exposed to chloride-containing deicing materials, the protective patina will never form, and weathering steel offers little advantage, as the corrosion rate will equal that of regular carbon steel. With continuous wetness, such as in immersion applications or underground, weathering steel also offers no advantage. There were concerns about the effect of acid rain and diesel exhaust on the formation of the protective patina. Investigations by steel producers and the steel industry have indicated that neither has an effect on the corrosion rate of weathering steel. Guidance regarding the use of weathering steel has been offered by FHWA in Technical Advisory T5140.22: Uncoated Weathering Steel in Structures (1989). This document pro- vides general information about the history, limitations, and recommendations for the use of weathering steel. In the DOT survey conducted for this research, about half the respondents using weathering steel also paint all or parts of their weathering steel bridges. The use of painted weather- ing steel appears to offer significant “belt and suspenders” protection to steel in locations where corrosion issues are not present and provides paint protection in areas with conditions that would otherwise limit the use of weathering steel. A weathering-type steel that could withstand wet conditions including water containing chlorides and still resist corrosion would present a desirable solution to the limitations cited above. Such a material would constitute a true noncoating corrosion-resistant option. One potential steel that meets these requirements, which is ASTM A1010 steel, has recently been developed. The corrosion resistance has been accomplished by chemically augmenting the steel production formula to produce a material said to be a “corrosion-proof ” weathering steel. ASTM A1010 contains 10.5% to 12.5% chromium and is said by its vendors to be immune to corrosion based on tests per- formed in Kure Beach, North Carolina, in the 25-m test site. This “corrosion-proof” weathering steel has been used on one structure in 2004 in Colusa County, California, and was reported in a presentation given at the 2004 Prefabricated Bridge Elements and Systems Conference. The bridge design was a prefabricated lightweight technology called a multicell box girder. Reportedly, less than 23 tons of A1010 were needed to form the structure of this short-span bridge with overall dimensions of 72 ft long by 32 ft wide. Other bridge structures have been reported in Illinois, Texas, and Oregon. Chapter 6 of the Guide discusses the recommended uses of corrosion-resistant steels. Recommendations for Improved Service Life In addition to the use of coatings and corrosion-resistant steel described above, the R19A research team identified a number of areas that hold great promise in the battle to resist corrosion of steel. First, they include a recommendation to immediately require that corrosion resistance, as a crucial element of bridge dura- bility, be a “hold-point design requirement” in every new and rehabilitated steel structure. “Hold point” in this context means that further progress on the design would depend on having a corrosion review performed and a corrosion-resistance plan established. In addition to the first recommendation to design corrosion resistance into every steel bridge, other considerations include • Building in quality through the use of best practices in painting; • Researching and developing methods to remove chloride ions from steel surfaces that have already been contaminated with chloride ions; • Developing new superdurable coating materials by both basic and applied research efforts, including industry-to- industry technology transfer;

64 • Using a corrosion-resistance maintenance plan for every steel bridge, including painting priority, cost estimate, and timetable; • Using coatings to seal precast and cast-in-place (CIP) concrete bridge elements against the intrusion of water containing chlorides from deicing materials and from sea salt spray in seacoast locations; • Researching whether the very costly removal of mill scale is a necessity for corrosion protection of hot-rolled steel (the necessity of completely removing mill scale is questioned); • Evaluating the possible development and expansion of the use of cathodic protection methods on structural steel on bridges; • Continuing funding for corrosion protection and resistance during all phases of the life of the structure including design, construction, and remedial maintenance over the 100-year life of a structure, as it is simply not possible to build it and leave it alone with no maintenance for 100 years; • Evaluating the widespread use of alternative corrosion- resistance approaches, such as thermal spray metallizing, hot-dip galvanizing, and weathering steel alone or in com- bination, both with and without additional coating; and • Performing research leading to the development and use of alternative noncorrosive deicing materials. In addition to the considerations listed above, one truly novel concept was put forth during this research. That suggestion is to apply some form of zinc coating to the steel during the hot-rolling process. One obvious benefit is that at hot-rolling temperatures, there is already a great deal of thermal energy in the steel. Application of a coating of zinc at that point will eliminate the need for reheating the steel or the wire in the case of thermal spraying or the application of a molten zinc bath during hot-dip galvanizing. If a metal more active than iron in the corrosion series, like zinc, were employed, then long-term corrosion protection could be built in rather than added on. If this method were to prove possible, it could obviate the need for any subsequent treatment for corrosion resistance. The savings could prove to be very substantial. No mention of this approach to corrosion protection was noted in the literature survey part of this research project. It is recommended that these areas be further investigated through additional research. There is little doubt that the least cost and biggest payback is found in the first recommendation: building in quality through the use of best practices in paint- ing. Designing corrosion protection into every project from the very beginning of the project design stages will cause a dramatic lengthening of the maintain–repair–replace cycle. Despite the conservative DOT expectation of only 30 years, there are many steel bridges coated with zinc coating–based systems that are up to about 45 years old. If corrosion resistance were intentionally designed into a structure by managing the configuration and details of bridge design and detailing, such structures may well last 100 years before major recoating is needed. Of course, some maintenance painting, including incidental or spot or zone repairs to mitigate nicks from traffic- propelled or storm-driven debris or vehicular or other impact, will likely be required at least once or twice. The identification and uniform adoption of painting best practices is doubtless another major area of inquiry. If there were an actual “AASHTO Bridge Corrosion Protection Guide Speci- fication” that cited best practices in bridge painting, repainting, and overcoating, a tremendous savings and lengthening of the bridge painting maintenance service life would result. Analysis of DOT Survey The following issues were apparent from analysis of the DOT survey: • Painting is widely used among most respondents. • Weathering steel is used somewhat, with five of 16 respon- dents reporting use on over 50% of their bridge structures. • Most states are following the still current, but dated, FHWA guidance document T5140.22: Uncoated Weathering Steel in Structures (FHWA 1989). • Only one state (Michigan) uses no weathering steel. • On weathering steel bridges that are painted, virtually all respondents say that they paint the ends of the steel beneath joint areas. • All states surveyed that responded use zinc-rich paint sys- tems. Most expect to get 20 to 30 years of service life. A few states use water-based versions of the protective system including an acrylic topcoat. They indicate that they expect 20 to 30 years performance before removal or reworking. • AASHTO’s National Transportation Product Evaluation Program is a set of test data on coatings that is free and could be beneficial in placing coatings on a qualified prod- ucts list. The testing is funded by coating suppliers. Seven respondents indicated that they do not use the data, one state has its own testing standard, and the remaining states can access a wealth of free test data. There are opportunities for state DOTs to mix the means of protection so that more resistive means of protection are explored to protect steel in the most aggressively exposed areas. Most states do not allow the submission of alternative coatings on projects. • Some form of training is supplied or required for coating inspectors in most of the states surveyed (11 of 19). Painting contractor certification is required by about one-third of the DOT respondents. There may be an opportunity for others to consider specifying the existing SSPC painting contractor certification program. • All states responding (17 of 19) require the complete removal of mill scale, and only one specifies a coating to apply

65 over mill scale. A two-coat system is approved on the North East Protective Coating Committee (NEPCOAT) qualified products list. Most states do not allow the use of a two-coat system. • The coatings on all NEPCOAT qualified products lists have been extensively tested both in the NEPCOAT program and the National Transportation Product Evaluation Program. These materials, although promising, do not have a long proven field history. Cost savings by using two coats instead of three are estimated at 3% of the steel cost (free on board shop) of a new steel bridge. • Striping, although optional, is considered an excellent prac- tice and is required by most DOTs responding. However, they do not mandate its use. • Maintenance repainting strategies and materials used vary widely among the DOTs. Most use a zinc-rich primer. • Overcoating material selection varies widely among the states. There are at least nine systems used by the 17 states responding. • Chloride removal practices vary widely, as do acceptance criteria. Most states responding require the criteria described in SSPC SP-12, Waterjetting NV-2 (<7 µg/cm2). There is room for improvement in chloride removal techniques before a feasible method will emerge. • The decision to overcoat versus recoat is generally made at the district level. There is an opportunity for successful research in the evaluation of maintenance planning. • Wash water is retained or contained by about half the states. It is assumed that these practices are in response to urging by the department of natural resources or department of environmental protection in those states. The retention process, as dictated by others, can control the maintenance painting cost of both bridge maintenance repainting and overcoating. • Most states contain solid waste even when lead paint is not involved on the project. • Finally, it appears that painting funds and lead-based paint continue to be big problems in most states. There are still thousands of bridges coated with lead-based paint, which means that the lead-based paint problem will extend well into the future. Steel Bridge Systems Steel bridge systems have the potential for achieving a service life well over 100 years, but they must be designed, constructed, and maintained properly to achieve this goal. The primary component of a steel bridge system is the steel superstructure. However, the overall system also includes other important components and elements such as the deck, joints, bearings, and substructure, all of which are discussed in more detail in separate sections of this report. All the subsystems and components that make up the overall bridge system are affected differently by external factors, such as heavy loading or harsh environment, and require varying levels of maintenance to achieve 100-plus years of service life. In areas of severe impact, some elements may have to be replaced before the service life of the overall system is reached. All the various elements that make up a steel bridge system are addressed in this report, and recommendations are made to improve service life. The design must first consider the durability of all system components and address the individual maintenance and replacement needs as part of an overall approach to service life design. This study identified the most common steel bridge systems used in practice along with newer promising systems that have incorporated accelerated bridge construction concepts and details. Service life issues were identified, and recommendations for improving service life were proposed. The findings and results of this study were incorporated into the Guide in Chapter 2, Bridge System Selection. Typical Steel Bridge Systems The most common steel bridge systems used today are composite multigirder deck systems using rolled beams, plate girders, or tub girders. These systems can be single- or multi- span and are either straight or curved. Simple-span systems were often used in the past, but most multispan systems today are continuous. Rolled-beam bridges using W-shapes are used in shorter spans, up to about 100 ft for simple spans and up to about 120 ft for continuous spans. Welded plate girders are mostly used for spans over 120 ft (NSBA 2005). Until the 1970s, many bridges were designed with systems using two deck girders, transverse floor beams, and longitudinal stringers. Poor fatigue details in these earlier welded bridges, however, led to cracking, and the issue of fracture criticality became a major concern. Multigirder bridges with inherent redundancy became the desired configuration (NSBA 2005). Steel bridge systems have typically used composite CIP con- crete decks, but other deck types, including precast concrete panels with or without posttensioning are also common, par- ticularly when used with accelerated construction techniques. Substructures generally consist of reinforced concrete piers and abutments. A variation to the typical multigirder system is the girder– substringer system, which has been used as an economical concept for longer spans (beyond about 275 ft). This system uses several heavy girders with wide girder spacing and rolled-beam stringers supported midway between the main girders by truss K-type cross frames. For many years, bridges were designed as a series of simple spans with expansion joints at each pier because they were easy to design and construct. Leaking joints, however, became a leading cause of structural deterioration, and the desire to

66 eliminate joints became prevalent. Multispan steel girder systems were also shown to be much more efficient when designed as continuous systems, so continuous design became commonplace. Multispan systems have typically been fully continuous for both dead load and live load, but new systems, typically with spans up to 150 ft, have been introduced with a simple for dead load and continuous for live load concept. These systems combine the advantage of simple-span construction with the efficiency of live load continuity and the durability that comes from not having joints. Continuous steel bridge systems using integral abutments were developed as a way to eliminate joints altogether. Integral pier cap construction was also developed as a way to avoid sharp skews or longer spans in interchange ramp bridges. Integral pier caps also have the advantage of eliminating bearings, which can minimize future maintenance requirements. Steel plate girder systems have been used for spans up to about 500 ft. Spans up to 400 ft have been designed economi- cally with parallel flanges. Variable-depth haunched girders have been used in the 350- to 500-ft range. Use of high- performance steel (HPS 70W) has shown economy for plate girder and tub girder systems in most span ranges over 150 ft. For plate girders, a hybrid combination using HPS 70W in negative moment top and bottom flanges and positive moment bottom flanges has been shown to be the most economical system. Trusses, arches, cable-stayed, and suspension systems have also been used for longer span applications, typically over 500 ft. For spans up to 300 ft, which is the limit for this research, deck girder systems are the most applicable. Causes of Deterioration in Steel Bridge Systems The National Bridge Inventory (NBI 2013) database shows clearly that steel bridge systems have the potential for achiev- ing service life well over 100 years. However, if they are not designed, constructed, and maintained properly, they can be affected by certain environmental and loading conditions that can lead to serviceability problems and reduced service life. Most major steel bridges that are in the 100-plus-year cate- gory have undergone deck replacements and other types of major rehabilitation during their service lives that have kept them alive. The major causes of deterioration for steel elements within a steel bridge system are fatigue and fracture, as well as corrosion. Fatigue anD Fracture Early welded steel structures have a history of cracking at certain types of weld details due to load- and distortion-induced fatigue. Cracking at I-beam cover plate terminations or at other longitudinal weld terminations in tension zones was particularly evident. Cracking in girder webs due to out- of-plane bending within stiffener web gap regions next to cross-frame attachments also became a common problem. Subsequently, extensive research and laboratory testing have provided an understanding of fatigue behavior, and different weld detail types were found to have varying levels of fatigue susceptibility. Newer design provisions and recommended details were developed that provide solutions for both load- and distortion-induced fatigue to achieve desired service life. Steel bridges do not fail in fatigue, but in fracture, which is the rapid, unstable propagation of a larger flaw (most likely the result of fatigue). Fatigue crack initiation is independent of steel type and strength, but possible brittle fracture is influenced by steel toughness, among other variables. Early steels were more susceptible to brittle fracture, but in recent years, new high- performance steels—HPS 50W, 70W and 100W—have been developed with very high toughness characteristics. Although somewhat more costly than conventional-grade steels, high- performance steels are now encouraged where applicable, particularly in nonredundant or fracture-critical applications. Use of high-performance steel allows time for fatigue cracks, if developed, to be found during regular bridge safety inspec- tions before fracture can occur. Fatigue and fracture should not be an issue in new steel bridges designed in accordance with the latest guidelines of the LRFD Specifications. Extensive research has been done in recent years to identify causes and solutions for fatigue- and fracture- related problems. When using proper details and fabrication methods, both load- and distortion-induced fatigue problems should not be an issue in achieving desired service life. corrosion Corrosion, the result of exposure to oxygen and moisture, is one of the fundamental limitations of steel as a main con- struction material. Corrosion is greatly accelerated in the presence of chloride ions from roadway deicing salt or salt spray in a marine environment. Deck drainage with deicing salt leaking through open deck joints is a leading cause of steel element corrosion in bridges. To achieve long-term bridge durability, a corrosion-resistance plan must be a design requirement for every new or rehabili- tated steel structure. This plan should include the use of painting best practices and a maintenance plan that addresses painting priorities and timetables. Details that serve to protect and keep the steel dry must be included in the design. Among these are bridge system solutions that eliminate deck joints, preventing salt-contaminated drainage from reaching steel elements below. Painting best practices now include paint systems that contain metallic zinc as the corrosion-resistant pigment. Zinc coatings provide galvanic protection to the steel, in which zinc (the more noble metal) will oxidize (corrode) in preference

67 to the steel. To protect the zinc coating layer from oxidation, additional coating layers are applied over the zinc-rich primer. Many studies demonstrate the value of zinc coatings as a steel protection system. In addition to zinc-rich paint, these zinc coatings also include galvanizing and metalizing. Their use should be considered carefully as part of the best plan for achieving a 100-plus-year service life. Weathering steel has also found widespread use in steel bridges, and has been used in both unpainted and painted applications. Although no formal research exists to support the practice, painted weathering steel following a “belt and suspenders” approach is commonly used in locations where corrosion issues would otherwise limit the use of weathering steel, such as directly below deck joints. In addition to weath- ering steel, a new stainless steel for bridges, ASTM A1010, has been developed for use in severely corrosive environments. other issues Other deterioration types are related to damage caused by external sources such as truck impact or fire, which are hazard related. These issues are further addressed in the Guide. Strategies for Improving the Service Life of Steel Bridge Systems new briDges In addition to the previous discussion regarding fatigue and corrosion protection, additional solutions for addressing steel system service life for new bridges include • Integral abutments to eliminate deck joints. • New steel systems that eliminate deck joints at piers; these systems include a concept for simple for dead load, continu- ous for live load. • New steel systems that provide for accelerated construction, including modular construction with predecked panels. These modular systems require special attention to both transverse and longitudinal connection details for achieving long-term durability. • Modular orthotropic deck systems are also a consideration. These systems also require special attention to both transverse and longitudinal connection details. existing briDges In addition to regularly scheduled maintenance, additional solutions for addressing steel system service life for existing bridges include (1) retrofitting existing simple-span steel bridges to continuous systems and (2) retrofitting existing fatigue-sensitive details or using special techniques that improve fatigue resistance, such as UIT. The first solution, retrofitting existing simple-span steel bridges to continuous systems, has been implemented by many state DOTs, who have developed guidelines for convert- ing existing simple-span steel bridges to continuous by splic- ing the girders over piers when they were being rehabilitated, particularly when decks were being replaced. Benefits of con- verting to continuous spans include reducing the potential of continued deterioration due to leaking joints, increasing resistance to seismic displacements, and slightly improving the load-carrying capacity of the superstructure. However, according to NYSDOT (2008), the design for continuity retrofit needs to consider the system behavior of continuous girders as opposed to original simple-span behavior. For multispan continuous systems, larger movements may be realized at abutments or at interior piers if original fixed conditions are converted to full expansion conditions. In addition, increased vulnerability to fatigue may result due to portions of the retrofitted beams being subject to stress reversals and higher stress ranges than the original simple- span construction. The end regions of retrofitted girders originally designed for small, simple-span positive moments are subjected to larger-magnitude negative moments. Although the deck joints over the piers are eliminated, the retrofitted deck in this area is subjected to tension under service loads, and crack control measures must be considered. Continuity can also increase seismic loads on individual piers depending on retrofitted bearing fixity configurations. Deck replacement projects provide excellent opportunities to include girder retrofit because girders will be readily accessible, and future costs of maintaining the joints will be eliminated. States use several methods to convert existing simple-span bridges to continuous, including full continuity, continuous for live load only, and continuous deck only (link slab concept). The second solution for addressing steel system service life for existing bridges involves retrofitting fatigue-sensitive details or using special techniques to improve fatigue resis- tance, such as UIT. Load-induced fatigue cracking has been one of the biggest problems with older welded steel bridges, particularly those with Category E and Category E′ details. UIT, which is designed to improve the fatigue resistance of an existing welded detail (Fisher et al. 2002), imparts impact energy to the surface of the material in areas in which fatigue cracking is likely to occur. It is performed with a tool that contains an ultrasonic transducer and a set of small steel trans- fer pins that are free to move relative to the transducer. The ultrasonic transducer generates resonant high-frequency pulses that impart energy to the material surface through the steel transfer pins. The imparted energy results in a zone of plastic deformation near the surface of the material, the introduction of compressive stresses up to 2 mm (0.08 in.) deep, and the relaxation of residual tensile stresses due to the original welding operation. All these material changes combine to prevent the formation of a surface crack under cyclic loading.

68 Testing performed by Fisher et al. (2002) at Lehigh University has demonstrated that UIT is effective in improving the fatigue resistance of many of the details found on structural steel bridges in the United States. For example, testing has shown that UIT can greatly increase the fatigue resistance of a cover plate termination, a Category E detail, and achieve a Category C or better fatigue strength. UIT is applicable to new construction and as a retrofit for maintenance and repair procedures. Chapter 2 of the Guide provides more detailed information about the performance of steel bridges. Concrete Bridge Systems The performance of unreinforced concrete as a durable material has been demonstrated in many historic structures, including the Pantheon in Rome, which was constructed around AD 126. Although unreinforced concrete provides excellent strength and durability in compression, its tension capacity is extremely limited and has a reduced ductility due to cracking. The intro- duction of steel bar reinforcement to enhance the tension capacity of unreinforced concrete allowed it to be used in more conventional beam applications other than elements in compression. The use of reinforced concrete in bridge systems has a relatively short history. Its first recorded application in the United States was in the Alvord Lake Bridge in Golden Gate Park (San Francisco, California), which was constructed in 1889 and is still in service today. As advancements in concrete and steel reinforcement technology have occurred over the years, so has the use of this material in bridge systems throughout the United States. The introduction of prestressing into con- crete bridges in the 1950s has resulted in a startling increase in its ability to meet the needs of a growing transportation network. The longevity of concrete bridge systems is affected by many environmental factors. Corrosion of steel reinforcement remains one of the principal causes of deterioration of these structures. Advancements in reinforcement-protection sys- tems have enhanced the durability of concrete bridges tre- mendously, particularly when exposed to harsh environments. Concrete bridge systems have the potential for achieving long service life if properly designed, constructed, inspected, and maintained. As with steel bridge systems, the overall concrete bridge system also includes other important components and ele- ments such as deck, joints, bearings, and substructure, which are all discussed in more detail in the Guide. The design must first consider the durability of all system components and address the individual maintenance and replacement needs as part of an overall approach to service life design. Chapter 2 of the Guide provides a more detailed discussion of concrete bridges. Common Concrete Bridge Systems Several common reinforced concrete bridge systems are used in the United States. The type of system implemented at a par- ticular site is generally dictated by economy and the system’s ability to provide the required span or geometric requirements, such as curvature. The most commonly used concrete bridge superstructures are discussed in the following subsections. cast-in-place slab systems CIP slab superstructures commonly span less than 50 ft and are typically used over minor water crossings. They were tra- ditionally constructed as a series of simple spans, but in recent years, the use of continuous spans has gained favor, eliminating the joints over substructure units. precast concrete aDjacent members Precast concrete adjacent member superstructure systems consist of multiple precast beam elements set side by side and connected either with a CIP closure joint, posttensioning, composite concrete topping, or a combination of these con- nection methodologies. The precast elements commonly consist of either conventionally reinforced or prestressed solid rectangular sections, rectangular boxes, or double Ts. Precast solid and voided slabs also fall into this category. Precast adjacent box beams are the most prevalent adjacent box girder superstructure for short- and medium-span bridges (typically 20 to 127 ft), especially on secondary roadways. precast concrete i-girDers Precast I-girders are the most commonly used concrete bridge superstructure. These girders are made of high-performance plant-produced materials and are generally very durable. I-girders with a composite deck slab behave distinctly differ- ently from adjacent box girders, which are described above. Adjacent box girders are not typically covered with a CIP composite slab, and the sections are torsionally stiffer. In a bridge system consisting of I-girders with composite CIP slabs, commonly referred to as beam–slab bridges, the longitudinal stringers are typically prestressed concrete girders. The use of precast prestressed concrete girders traditionally has been limited to simple spans up to approximately 150 ft; however, with recent optimization of girder sections, precast girders with spans greater than 200 ft can be achieved. The girders can also be made continuous for live load. cast-in-place posttensioneD box girDers A superstructure of CIP posttensioned box girders is cast continuously on falsework. This type of superstructure has become very popular in several states, particularly California, Arizona, and Nevada, and has been used in spans up to 350 ft. This type of construction lends itself to local construction

69 industry practices in areas where contractors can economi- cally provide the required falsework. Similar to segmental construction, CIP on falsework offers the advantage of longer spans than conventional girders and can easily accommodate curved alignments. CIP construction also allows clean lines and architectural finishes that improve the aesthetics. The use of posttensioning further enhances concrete durability. Designing the structures as a frame and using monolithic connections between the superstructure and piers also elimi- nates bearings, which eliminates associated future maintenance. A potential disadvantage of this type of construction is diffi- culty in replacing the deck or widening the bridge. segmental posttensioneD concrete box girDers Specialty systems using segmental posttensioned concrete box girders have been used for spans greater than those that can be achieved with stringer-type girders and for bridge geometries with horizontal curvature. They are further divided into cantilever construction and span-by-span construction, and can be either precast or CIP. One of the advantages of seg- mental concrete bridges is the type of construction technique used, which minimizes the interruption to traffic during con- struction. Recently, however, there have been major concerns about corrosion of steel strands in grouted duct (Azizinamini and Gull 2012a, 2012b). This concern is especially important as there is no reliable technique to detect corrosion in steel strands embedded in concrete. Inspections of external ducts are rela- tively easier. However, development of a reliable methodology to detect corrosion in duct embedded in concrete remains an active research topic. segmental spans Segmental spans can be cast to match the shape of the align- ment, making them particularly suited to curvature. Spliced girders are typically used on relatively straight alignments; however, they have also been used for curved alignments in recent years in Nebraska and Colorado. Segmental and spliced posttensioned girder bridges have been observed to have improved deck performance due to the precompression of the deck. moDular pretoppeD concrete girDer units These superstructures using modular pretopped concrete girder units are primarily developed for accelerated bridge construction. They consist of precast beams or girders con- structed monolithically with sections of the deck, which are then connected with CIP joints in the field. Causes of Deterioration The literature search supplemented by the results of the DOT surveys performed during this study revealed numerous causes of deterioration in concrete systems, subsystems, and components. The various causes for concrete deterioration are discussed in detail in the sections of this report dealing with concrete durability and bridge decks. The degree and severity of concrete deterioration depends on the environmen- tal influences to which the bridge is subjected. The principal influences include • Environmental volumetric influences, including physical influences, such as the effects from moisture and temperature change; wear; and freezing and thawing; • Chemical influences, including exposure to chlorides, sulfates, carbon dioxide, alkalis, and various acids; and • Loading influences, including vibration and impact. The durability of concrete exposed to these influences is highly dependent on design practice, materials, and their pro- portioning and workmanship during construction. Although all these influences are important, the principal deterioration of concrete systems, subsystems, and components is the cor- rosion of the steel reinforcement, which results in severe crack- ing, spalling, and delamination of the surrounding concrete. The protection provided by the depassivated zone around the steel reinforcement is often compromised by cracking of the concrete. Concrete bridge decks are particularly affected by environ- mental influences. Bridge deck cracks are typically parallel to the steel reinforcement in the top layer. This cracking orientation can cause the concrete surrounding the steel reinforcement to reach the corrosion threshold limit within 6 months to a year in environments where the top of the bridge deck is exposed to chlorides, such as deicing salts. Concrete superstructures and substructures in marine environments are not exposed to the same concentration of chlorides as the top of a bridge deck directly in contact with deicing salts. They are, however, susceptible to the same type of corrosion, only at a slower rate through the same mechanism of failure. Strategies to Enhance Service Life of Concrete Bridges Several strategies have been developed to address the durabil- ity of concrete systems, subsystems, and components. These strategies, which are fully described in the concrete durability section of this report, include • Proportioning of concrete to provide low permeability and low cracking potential; • Use of noncorrosive materials (e.g., stainless steel) and protective coatings (e.g., epoxy); • Prestressing or posttensioning of elements to eliminate cracking;

70 • Use of more sophisticated strategies, such as cathodic protection and electrochemical chloride extraction; • Use of overlays and membranes on bridge decks; and • Various combinations of the above strategies. The use and application of these strategies is highly depen- dent on the environment in which the concrete systems are exposed. A single strategy that fits all conditions within the United States does not exist. These strategies must be reviewed by each governing agency for applicability. research Categories Following the literature search and other related tasks previ- ously described, a number of research topics were identified to fill the knowledge gap with respect to design of bridges for service life. Service life issues needing further research were divided into three categories, with Category 1 research topics having the highest priority. Category 1 and 2 research topics were selected to conduct proof of concept tests within the R19A project. As required by the original scope of work, and recognizing the limited resources within the project, proof of concept tests were intended to examine the feasibility of the ideas and leave detailed research to others. Greater effort was spent on Category 1 research topics. Category 3 research topics (see Appendix A) were identified and are recommended to be carried out by others. Table 3.12 lists Category 1 and 2 research projects as approved by SHRP 2 to carry out proof of concept tests. The following sections briefly describe the scope of work carried out and the results and findings for the Category 1 and 2 topics listed in Table 3.12. Based on priority, project resources, and other limitations, the level of efforts on each topic varied. When possible, the following information is provided for each topic: • Problem statement; • Objectives; • Scope of work; • Results; and • Recommendations for future research. The final major section of this chapter describes the devel- opment and provides an outline of the Guide. Link Slab Problem Statement A link slab is a type of detail used in conjunction with existing or new bridges in which the girders act as simple beams for both dead and live loads. In this type of deck detail, the slab spans continuously over the length between the adjacent simple- span girders (see Figure 3.41). The use of link slabs, however, has been limited primarily to short-span bridges in warm climates. The Phase 1 report provides a detailed description of past work and current status with respect to link slabs. Objectives The limited use of link slabs in cold climate areas are due to two factors: (1) lack of understanding of the force transfer Table 3.12. List of Category 1 and 2 Research Topics Research Topic Category Joints Link slab 2 Converting simple spans to continuous 2 Jointless bridges 1 Seamless bridge systems 2 Expansion joints 2 Joints in modular systems - Adjacent box - Closure pour 1 Bearings High-performing sliding surface 2 Enhancing corrosion resistance of concrete bridges Corrosion-resistant reinforced concrete structures 2 New galvanic systems to achieve long-term corrosion protection 2 Bridge deck Self-stressing deck systems: CIP 2 Self-stressing deck systems: precast 1 Delayed composite systems 2 Membranes for bridge deck 1

71 mechanism and load demand for which the link slab should be designed and (2) lack of thorough information about feasible details that can prevent formation of extensive cracking and leakage of moisture onto the bridge elements below the deck. The main objective of the link slab study was to comprehend the behavior of link slabs and recommend the scope of work that could lead to the development of appropriate details. Scope of Work In this research, several multispan bridges containing link slabs were instrumented to determine the motion at their expan- sion joints, located at the bridge abutments, and at the link slab details between the girders at the bridge piers. Finite element analyses were performed to analytically investigate the link slab behavior, and mechanistic models were developed to represent the overall link slab response. The primary focus of the field instrumentation was to determine the bridge dis- placements and the deformations within the vicinity of the link slab under thermal loads. The researchers instrumented the bridges with dial gauges to measure longitudinal movements and with DEMEC (DEmountable MEChanical strain gauge) points to measure expansion and contraction across the construction joint or control joint at the link slab. The finite element analyses were conducted to evaluate the potential magnitude and distribution of strains in the link slab region. Emphasis was placed on a particular link slab detail currently used in the state of Georgia; however, variations on the Georgia detail, including different details used by other states, were also considered. The detailed studies are documented in Davidson et al. (2012) and Snedeker et al. (2011). Results The field results showed that girder ends resting on fixed bearings, which for prestressed concrete girders have round holes for dowel bars at the bearings, have essentially the same ability to displace longitudinally due to daily thermal loads as those that sit on expansion bearings, which have slotted holes for the dowel bars. Furthermore, because of the annulus around the dowel bars at the bearings, little to no force is generated in the bearings due to the girder rotations. Even if the dowel bars are in contact with the holes within the girders, the force that can be induced in these bars is not sufficient to develop any significant stress or strain within the link slabs. In addition, the longitudinal forces induced by the deformation of typical elastomeric bearings are not sufficient to develop any significant stress or strain within the link slabs. Thus, little to no force occurs in the link slab reinforcement crossing the construction joint. Estimates of shrinkage strains in the link slab showed that a crack of sufficient width could easily exist in the link slab at the top of the deck so that the concrete on one side of the crack would not contact concrete on the other side of the crack. A finite element analysis showed that the deformations due to thermal and live loadings are not sufficient to close such a shrinkage crack that may exist in the deck. Therefore, essentially no moment is created in the link slab due to girder rotations. Debonding of the concrete slab on top of the girders adjacent to the link slab is not needed to achieve good performance of the link slab. Good performance of the link slab detail is contingent on the proper forming of a controlled joint near the midlength of the link slab and the proper sealing and main- tenance of this joint. Damage was observed in a small number of instances from evaluation of a large set of bridges in Georgia. In all cases where damage was observed within the link slab, the damage involved either (1) an unraveling of the edge of the slab due to inadequate chamfering or rounding of the edge of the slab at the construction or controlled joint or (2) spalling of concrete from a patch located adjacent to the controlled joint. The reinforcing bars crossing the link slab provide bridge deck continuity for longitudinal and transverse loads due to vehicle braking, wind, and earthquake, and they act as dowel bars transferring shear between one side of the deck crack to the other. However, there is little shear transfer when trans- verse edge beams supporting the deck are placed between the girders at the ends of the girders. The research observed that in bridges with minor skew angles up to approximately 20° Figure 3.41. Conceptual detail for link slab.

72 from the nonskewed orientation, extension of the top mat of reinforcing bars across the link slab is sufficient. Conclusions and Recommendations Based on the research study carried out, the following conclu- sions and recommendations are made with respect to design and construction of link slabs. DebonDing versus no DebonDing Many bridge organizations require debonding of the link slab from the girders and any edge beams or diaphragms over a length of up to 5% of the girder span length on each side of the link. Common practice is then to assume that the girders impose a rotation of q1 + q2 at the ends of the link slab, where q1 and q2 are the end rotations of each of the girders, and that the link slab is subjected to pure bending between the ends of the debonded zone. Basic kinematics does not allow a uniform curvature of (q1 + q2)/Ld in the link slab, where Ld is the length of the debonded zone. The link slab can only reach the curva- ture of the girders at the ends of the girders, since it is in contact with the top of the girders. Furthermore, the girders have essentially zero curvature at their simply supported ends. Obviously, the true three-dimensional response of the link slab is more complex than the simple (q1 + q2)/Ld curvature assumption. However, if there is a relatively wide top flange of the girder, such as in various precast sections, the restraint from the girders would appear to force a concentration in curvature across a short length of the link slab, located between the ends of the girders. Based on the above findings, debonding of the slab from the girder should not degrade the performance; however, based on the research conducted in this study, the impact of debonding is expected to be minimal. Good performance of link slabs has been observed without the use of debonding. Note that for precast concrete construction, the shear stirrups generally are placed close to the ends of the girders, and the slab concrete is used in the calculation of the composite girder shear strengths. Therefore, for precast concrete construction, it is not practical to debond the slab from the girders. It can be surmised that in many cases, the behavior of the link slab under live load and negative temperature gradients may involve the development of compression toward its bot- tom surface due to bending about a neutral axis near the top mat of the rebar. This compressive force can be reacted by the girder shear connectors at the ends of the link. In addition, these forces can be relieved by restrained shrinkage strains that occur due to early curing of the concrete. Hence, the link slab strains due to the girder end rotations may contribute mostly to the closing of small, unavoidable shrinkage cracks that occur due to early restrained deformations. The behavior under positive temperature gradients is similar, but the top of the slab is in compression due to the upward movement of the simply supported girders. inFluence oF skew Due to the differential vertical displacements between the girders at a given bridge cross section, as one moves away from a skewed bearing line, and when end diaphragms are used, due to the relatively large in-plane stiffness of the diaphragms, the tendency is for the girders to move (twist) in opposite direc- tions on each side of the skewed bearing line. This movement generates a tendency for displacements normal to the skewed bearing line either at the slab level or at the bearing level. The neutral axis of the composite girders is generally located in the slab or close to the top of the girders. As noted above, the mea- sured rotations also tend to be about an axis within the depth of the link slab at the ends of the girders. Therefore, the greatest tendency for displacements is normal to the skewed bearing line and at the bottom of the girders. The typical bearing details are flexible enough to allow these movements. selection oF link slab reinForcing A wide variety of reinforcing layouts are used by different bridge organizations. In many cases, concerns are expressed about terminating any of the reinforcing steel within the link slab region. Some organizations terminate all the slab- reinforcing steel at the link slab, place a construction joint at the center of the link slab, and continue one top mat of reinforcing steel across the joint. The most important attributes of the rebar continued across the joint are that (1) there is sufficient steel area such that the rebar is not yielded due to any defor- mations experienced by the link slab, and (2) the bar spacing is sufficiently small such that any cracks within the link slab other than at a formed or saw-cut joint remain small and well distributed. Lepech and Li (2009a) point out that “[t]he large majority of concrete link slabs within Michigan that have shown distress or required maintenance were found to have too little reinforcement included in the design, or the reinforcement was not installed properly by the contractor.” For link slabs at bearing lines having sharp skew angles, it is recommended that at least one mat of rebar be continued through the joint in both the longitudinal and transverse directions. For link slabs at bearing lines having a minor skew, some organizations orient the transverse reinforcing steel parallel to the bearing line. This technique avoids the need to terminate transverse reinforcing steel at a construction joint or the need to continue these bars through a construction joint. use oF a saw-cut or FormeD joint One common practice is to place a saw-cut or a formed (troweled) joint at the midlength of the link slab. The intent is to force a larger crack to occur at this specific location, where

73 the crack can be sealed and maintained. A number of cases have been observed by the investigators in which, due to inad- equate preparation of this joint, spalling or unraveling of the concrete occurred at the edge of the joint. In addition, it is important that if a saw cut is placed in the link slab, this cut must be made shortly after the concrete placement to ensure that the intended crack control occurs. By specifying a formed joint rather than a saw cut, the engineer can have some addi- tional confidence that the joint will be placed early enough for it to be effective. At highly skewed bearing lines, it is recommended that the saw cut or formed joint be placed over the pier, parallel to the skew. However, over a short distance at the edges of the slab, the joint should be turned at 90° to the edge of the deck. Because it is expected that saw cut or formed joints can be easily damaged by snow plows, their use is not recommended in cold regions. Waterproofing One option that alleviates the need for a saw cut or formed joint is to provide a waterproofing membrane for the deck. When a membrane is employed, there is no longer any need for a saw cut or formed joint. The reinforcing steel can be designed to control the crack size within the link slab, and the membrane guards against penetration of moisture into the deck. Separation of edge BeamS or diaphragmS In cases in which edge beams or diaphragms are placed at the ends of the girders, it is important that sufficient spacing be provided between these components on each side of the joint to accommodate the anticipated sum of the maximum girder rotations on each side of the joint, assuming that the center of rotation is about the top mat of reinforcing steel in the link slab. application of fiBer-reinforced concrete Recent research by Lepech and Li (2009a, 2009b) and Qian et al. (2009) has developed and demonstrated the use of fiber- reinforced concrete (or engineered cementitious composite [ECC] material) to attain excellent durability of link slabs while accommodating the deformations imposed on them by simply supported girder end rotations due to live and thermal loading conditions. Figure 3.42 shows a schematic of the link slab detail proposed by these investigators. Lepech and Li (2009a) provide a complete design procedure based on their findings. The design involves the use of a debond zone in which the shear connectors between the girder and the deck are removed to prevent composite action. Outside the debond zone on either end of the link slab are transition zones in which the shear connection and composite action between the girder and the deck are reestablished. The ECC is placed using a separate closure pour after the decks have been placed on each side of the link slab. Lepech and Li calculate a maximum girder end rotation demand based on the maximum girder service vertical deflections for use in their design procedure. If a deflection limit of L/800 is used, the design rotations are 0.00375 radians. The link slab is assumed to be subjected to uniform bending via these end rotations, and steel reinforce- ment is provided in the link to develop the corresponding induced moment. The ECC material is used to address the crack width serviceability requirements. The investigators recognize that in warm climates, link slab details using conventional reinforced concrete perform well, such that the cost of the advanced material, the extra construc- tion step, and the joint between the concrete slab and the ECC are not required. Converting Simple Spans to Continuous Problem Statement In the past, a large number of steel bridges were constructed as a series of simple spans with deck expansion joints at each pier. This was a popular system concept because it was easy to design and construct. Leaking deck joints, however, have become a leading cause of deterioration and subsequent reduced service life for all types of bridges. Converting existing simple-span systems to continuous and eliminating deck joints has been used as a way to extend service life of existing bridges. Some state DOTs have developed specific guidelines for splicing girders over piers; others have addressed this on a Source: Lepech and Li 2009a. Figure 3.42. Schematic of ECC link slab detail.

74 case-by-case design basis. As a result, many variations exist in the industry. There is a need to provide more uniform and con- sistent guidelines regarding design criteria, details, and perfor- mance that will result in cost-effective service life extension. Objectives The objectives of this study were to review current criteria and details in the industry for converting existing simple-span steel bridges to continuous and to develop recommendations for consistent cost-effective approaches. Scope of Work The scope of work for this study included • Collecting, reviewing, and summarizing information from various states and from the current literature; • Developing recommended criteria for continuity retrofit, including guidelines for analysis considering system behav- ior; and • Recommending appropriate details. Results Additional detail on this study is given in Appendix B. A literature survey and a brief survey of national HDR, Inc. offices were conducted to identify current industry and state DOT practices for converting existing simple-span steel bridges to continuous. It was found that a number of states have performed such conversions, primarily during deck replacement rehabilitations, but few have specific design procedures or standard details. New York State DOT has detailed guidance in their design manual. Tennessee and New Mexico have reported on specific project applications. Some states have guidelines for new design, but not for retrofit of existing design. These details for new design, however, could also be adapted for continuity retrofit. There are two primary reasons for continuity retrofit: eliminating deck joints and increasing live load–carrying capacity. Details used for achieving continuity over piers can vary depending on the desired outcome. Three methods have typically been used for continuity retrofit: • Converting girders to fully continuous for dead load and live load (dead load in this case refers to concrete slab and composite dead load); • Converting to continuous for live load only (partially con- tinuous); and • Converting by using a continuous slab over the joint and not connecting the girders (link slab). In the fully continuous retrofit, the girder flanges and web are spliced, and a single bearing is used under the converted girder. Except for steel dead load, the converted girder is fully continuous for all subsequent dead and live loads, and it behaves like a conventional continuous girder. Because of the significant additional negative moment at piers, splice plates have to be extended to accommodate the negative moment region. Top flange splice plates are typically bolted. Bottom flange cover plates can be field welded, and a butt connection splice made in compression with wedges. Although the greater amount of field work required for this type of conversion makes it the most expensive, it permits greater carrying capacity. In the continuous for live load retrofit, only the flanges are spliced, and separate bearings are used under the existing girder ends at the piers, similar to the existing configuration. Top and bottom flange splice plates are typically used, but the extension is less than required for the fully continuous retrofit. Splice plate connections are made similar to that described above. In these instances, the girder continuity connection at the piers is made after the deck is poured. In reality, composite dead load is then carried in the continuous system. An economical and functional modification to this concept is to eliminate the top flange splice plate and use a CIP concrete diaphragm along with the reinforced concrete deck to carry the negative live load moment. The bottom flange connection is still made as described above. This concept has shown to be a reliable solution. The link slab concept has been used by several states as part of retrofits to eliminate deck joints; it is discussed at length in the previous section on link slabs. With the link slab, girders are assumed to carry the same dead and live load as in the simple- span configuration. This is the most economical solution, but it may result in deck cracking over the piers, which may be objectionable in northern climates with heavy use of deicing chemicals. It also provides the least ability to increase load- carrying capacity. In all cases when converting to full or partial continuity, the resulting system behavior must be considered. Effects of negative moment over piers with respect to strength and fatigue must be considered and accounted for. Revised bridge end movement resulting from continuity retrofit and converting existing bearings from fixed to expansion at some locations must also be considered and accounted for. Conclusions and Recommendations Continuity retrofit is typically performed to eliminate deck joints or to increase girder carrying capacity. Three general methods of converting existing simple-span steel bridges to continuous have been used: fully continuous, partially continu- ous, and link slab. Fully continuous retrofits tend to be costly but provide the greatest potential for increasing load-carrying capacity. Partially continuous retrofits are generally more

75 economical and recommended. Newer tested details that elimi- nate the top flange splice are also recommended. System behavior of the retrofitted continuous system in regard to girder strength and fatigue and revised bridge move- ment must be considered and accounted for. Jointless Bridges Problem Statement Leaking deck joints have been a major cause of bridge dete- rioration and reduced service life, especially where roadway drainage carrying deicing chemicals can spill onto bridge elements below. Elimination of bridge deck expansion joints is therefore an important consideration in bridge system selection to provide long-term service life. A jointless bridge, also commonly referred to as an integral bridge, has a con- tinuous deck with no expansion joints over the superstructure, abutments, and piers. In this type of bridge structure, all movement due to thermal effects, creep, and shrinkage strain is accommodated either within the system itself or at the ends of the approach slabs where the slabs meet the roadway pave- ment. Unfortunately, at present there are no complete and scientific guidelines for designing jointless bridges. Under the SHRP 2 R19A project, a study was carried out to develop comprehensive guidelines for the design, construction, and maintenance of jointless bridges. This section provides a brief overview of the results obtained for this task. Chapter 8 of the Guide provides a detailed discussion of the design, construction, and maintenance issues related to jointless bridges, as well as various details associated with jointless bridges. types oF jointless briDges Three main types of jointless bridges are described in this subsection: integral and semi-integral jointless bridges, which are commonly used in practice, and a new class of jointless bridges referred to as seamless jointless bridges. The main char- acteristic of the seamless system is that expansion joints are eliminated altogether, and the bridge deck is connected to the approach road pavement with no expansion joint. Integral bridges have a superstructure constructed mono- lithically with the abutments, encasing the ends of the super- structure within the backwall. The main characteristics of integral bridges are their jointless construction and flexible abutment foundations. The system is structurally continuous, and the abutment foundation is flexible longitudinally. The movement of the superstructure is accommodated by the foundation. The main elements of an integral bridge system consist of bridge deck, girders, integral cast abutments, and approach slabs. The bridge movement is accommodated at the ends of the approach slabs. In addition, sleeper slabs are commonly used to provide vertical support for the ends of the approach slab where the slabs abut the roadway pavement. Semi-integral bridges are defined as having an end diaphragm serving as the abutment backwall; the end diaphragm is cast encasing the superstructure ends. In this system, the super- structure rests on expansion bearings, and the end diaphragm is not restrained longitudinally with respect to the pile cap or abutment stem. The deck may be sliding or cast monolithically with the backwall, but it does not have a joint above the abut- ment. The foundation is rigid longitudinally, where super- structure movement is accommodated through bearings. The main elements of a semi-integral bridge system consist of bridge deck, girders, abutment stem and bearing seat, inte- gral cast diaphragm backwall, approach slab, and sleeper slab. The bridge movement is accommodated at the ends of the approach slabs. jointless briDges in use A detailed history of jointless bridges is provided by Burke, Jr. (2009) and summarized here. On the basis of a nationwide mail survey of state and provincial transportation departments, it appears that the Ohio highway department was one of the first agencies to initiate the routine use of continuous construction for multispan bridges. However, these bridges had expansion joints at abutments; Ohio began this practice in 1930. The Tennessee DOT is now leading the way in construction of continuous bridges. For example, the Long Island Bridge of Kingsport was constructed in 1980 using 29 continuous spans without a single intermediate deck expansion joint. Continuous integral bridges with steel main members in the 300-ft range have performed successfully for years in North Dakota, South Dakota, and Tennessee. Continuous integral bridges with concrete main members 500 to 800 ft long have been constructed in Kansas, California, Colorado, and Tennessee. As of 1987, 11 states reported building continuous integral bridges in the 300-ft range. Missouri and Tennessee reported even longer lengths. Missouri reported steel and concrete bridges of 500 and 600 ft, respectively. Tennessee reported lengths of 400 and 800 ft for similar bridges. Sixty percent of those departments responding to the 1987 survey were using integral construction for continuous bridges. More recently, the Tennessee DOT completed the Happy Hollow Creek Bridge, a seven-span prestressed concrete curved integral bridge with a total length of over 1,175 ft. aDvantages oF jointless briDges Henry Derthick, former engineer of structures at the Tennessee DOT, once stated, “The only good joint is no joint.” In keeping with this statement, the known advantages of the jointless bridge systems include • Lower initial cost; • Lower maintenance cost;

76 • Prevention of leakage of moisture to bridge elements below deck, resulting in longer service life; • Improved ride quality; • Easier and faster construction; • Easier inspection; • Simplified bridge details; • Elimination of bearings (except for semi-integral); • Ideal suitability for bridges with skew and curvature or located in high seismic areas; and • Enhanced buoyancy resistance of the bridge. Because of these advantages, many DOTs have started using jointless bridges; however, the design provisions vary significantly from one state to another. cost-eFFectiveness oF jointless briDges Jointless bridges have a significant cost savings advantage compared with traditional bridges with expansion joints. Cost savings are realized both during initial construction and throughout the life of the bridge because of reduced main- tenance. This is particularly true for bridges with integral abutments. The Guide provides detailed estimates of cost savings that could be achieved by using jointless bridges as compared with including expansion devices in the superstructure. Factors aFFecting perFormance oF jointless briDges Results of the study indicated that DOTs are very satisfied with performance of jointless bridges. The only service life issue reported was the existence of cracking in the case of skewed bridges. The Guide provides extensive coverage of this issue and provides design recommendations. Factors affecting the performance of jointless bridges include • Curvature; • Skew; • Bearings; • Connection of superstructure and substructure; and • Other conditions, including 44 Site conditions, 44 Deterioration of piling, and 44 Abutment wall type (i.e., reinforced concrete versus mechanically stabilized earth walls). The Guide provides detailed discussion and recommenda- tions to address each of the above factors. new concepts to extenD applicability oF jointless briDges Flexible foundations are typically used for jointless bridges to reduce the induced forces and increase the length of jointless bridges (Burke, Jr. 2009). Deep foundations are widely used to support jointless bridges as they are relatively flexible and not prone to scour. The main criterion limiting the application of jointless bridges is the lateral displacement capacity of their supporting piles. Flexible pile cap connection Research conducted revealed that providing a pin-type con- nection at the pile head can effectively reduce the stiffness and thereby lower the moments developed in the pile as a result of lateral movement, because the pile will deform in a single rather than a double curvature shape. Some researchers have studied different details to provide rotational capacity over the pile head. However, the reported details can only handle relatively small movements and in some cases require additional investigation. Within the SHRP 2 R19A project, a study was carried out and a new pin head detail was devel- oped that provides significantly larger lateral displacement capabilities (Sherafati 2011; Sherafati and Azizinamini 2013). This new pin head pile detail consists of encasing the ends of the pile embedded in the pile cap beam with an elastomeric material, which, in turn, allows the pile head to act as a pin. The consequence is to have the pile bend in a single curvature, reduce the maximum moment in the pile, and move the loca- tion of maximum moment from the pile head end to within the length of the pile. The end result is allowing the pile to have larger lateral movement capabilities and consequently allowing use of larger bridge lengths in conjunction with jointless bridges. Details of this new pile head detail and corresponding design provisions are provided in Chapter 8 of the Guide. A flexible pile cap connection can be used to reduce the stiff- ness of the piles to lateral displacement. In this case, the devel- oped moments as a result of the pile head deflection are reduced, and eventually lateral displacement capacity can be increased. To achieve a flexible pile cap connection, the embedded portion of the pile head in the stub cap beam can be encased in a soft material casing. This material will eventually be embedded in the concrete cap. This condition allows for relative rotation of the pile head with respect to the cap. A proposed detail is shown in Figure 3.43. An elastomer casing is used to allow for rotation. The end plate is welded to the end of the pile to reduce the stress concentration at the pile head, and another plate is embedded in the concrete cap, using shear studs, to spread the applied end load and prevent crushing of the concrete in the vicinity of the pile head. These two plates can slide freely against each other. The main advantage of this detail is that it can increase the pile head lateral displacement capacity by reducing the stiffness of the system, which allows for construction of longer bridges using jointless bridge systems. Further, because of reduced lateral stiffness, smaller forces are developed in the cap, as well as in the superstructure. A prefabricated cap with circular holes can be constructed for accelerated construction, as shown in Figure 3.44. This will eliminate the need for forming and cast- ing concrete for the cap, which results in considerable savings in construction time and cost.

77 ExtEnsion to CurvEd GirdEr BridGEs Another new concept is related to extending the application of the jointless bridge systems to the case of curved girder bridges. Detailed parametric studies were carried out to identify the factors that affect the performance of jointless curved girder bridges (Doust 2011; Doust and Azizinamini 2011a, 2011b). The study was confined to steel girder bridges. The Guide pro- vides the design steps that should be considered when using jointless bridges for curved girder bridges. Objectives Results of the survey indicated that the design of jointless bridges is mainly based on trial and error and that there is a lack of scientifically based design provisions for jointless bridges. The objective of this study was to develop a comprehensive guideline for the design of jointless bridges in a format usable in design office. Scope of Work The scope of the work included a review of the available pub- lished and unpublished studies. Further tasks were carried out to extend the application of jointless bridges to longer bridges, as well as curved bridges. An experimental study was carried out to comprehend the performance of the proposed detail. Using the results of experimental data and by conducting numerical and analyti- cal studies, a design provision was then developed, which is reported elsewhere (Sherafati and Azizinamini 2013). Experimental Program Experimental studies were conducted on two pile cap con- nections. The objective of these tests was to comprehend the merits of the concept and develop suggestions for further investigation. The first specimen involved a relatively rigid connection and represented the current practice. The second specimen, however, consisted of the proposed connection detail and represented the flexible connection. The first test specimen (with rigid pile head) was used as a reference point to assess the merits of the proposed pile head detail. The specimens were oriented horizontally during testing, such that the load representing the lateral movements of the bridge was applied using the two vertical actuators, as shown in Figure 3.45. The test specimens were positioned perpen- dicular to the loading frame and elevated using concrete blocks, posttensioned to the floor (see Figure 3.45). Axial load, simulating the gravity loads, was applied using Dywidag rods passed through a small-diameter plastic duct placed in the middle of the concrete-filled tube (CFT) specimens. Figure 3.43. Proposed detail of flexible pile cap connection. CFT  concrete-filled tube. Figure 3.44. Prefabricated pile cap.

78 Results Figure 3.46 shows the load–deflection curves for cycling of both specimens at increasing displacement levels obtained from the ultimate loading of Specimen 1 and the initial loading of Specimen 2. The displacement in these curves corresponds to the lateral displacement under the actuators. The apparent boldness of the line in Figure 3.46 is due to the fact that five individual cycles were performed, and the results overlie each other. This is true for all but the last set of cycles of Specimen 1, as noted in the figure, when fracture had occurred and was progressively growing with each cycle. Note that the nonlinearity in Specimen 1 started around a ±1.00-in. deflection, which corresponds to the point at which strain gauges indicated plastic straining close to the connection. Specimen 2 displayed lower initial stiffness, which was the intent of the design. In this specimen the stiffness then dropped at a displacement of approximately ±2.0 in. of deflection. This change in stiffness was related to slippage of elastomeric casing with respect to the pipe surface. Although the curves appear somewhat similar, there are different underlying mechanisms that account for the behavior. Envelopes of moment–rotation curves are presented in Figure 3.47a for both specimens. The moment is evaluated by multiplying the lateral load by its arm plus the second-order moments developed by the axial load due to the deflection of the pile. The rotation is calculated using the deflections recorded by the first row of potentiometers on the top and the bottom. These curves are then fit by bilinear curves for further analyses. The initial stiffness in Specimen 1 is almost four times the corresponding value in Specimen 2. Using the results of moment–rotation analysis, the pile is modeled in SAP2000 by using beam elements. A nonlinear link element representing the rotational stiffness of the connection is assigned to one end. A 120-kip axial load is applied to the free end of the beam. The lateral load is then applied, and the corresponding displacement is extracted from the analysis. Figure 3.46. Load–deflection curves for (a) Specimen 1 and (b) Specimen 2. (a) (b) -50 -40 -30 -20 -10 0 10 20 30 40 50 -5 -4 -3 -2 -1 0 1 2 3 4 5 La te ra l L o a d (ki p) Deflection (in) UpDown -30 -20 -10 0 10 20 30 -5 -4 -3 -2 -1 0 1 2 3 4 5 La te ra l L oa d (ki p) Displacement (in) UpDown Figure 3.45. Test setup. Actuator CAP Support Block Hydraulic RamCFT

79 Note that the P - D effect is also considered in the analysis. Figure 3.47b shows the load deflection results from the testing compared with the analysis results from SAP2000 using rota- tional springs representing the pile cap connection. As shown, the analysis results match well with the test results, validating the moment–rotation stiffness calculations. The steel casing of the pile in the rigid connection began to yield at a displacement level of ±1.00 in. In the flexible con- nection, however, yielding was not present up to a displacement level of ±4.25 in., which is the displacement level at which lifetime cyclic testing was performed. Both specimens were then subjected to a number of cycles that would be expected from a lifetime of service equal to 100 years. Unfortunately, the displacement level applied to the flexible connection was overly optimistic, and although the exposed elements showed no distress, failure occurred after 13,000 cycles. The failure appeared to be caused by localized issues at the embedded end of the column that can be easily mitigated through improved detailing. Analytical investigations using the experimental results indicate that the flexible pile cap connection can increase the lateral displacement capacity of the piles by a factor of up to 3.8 in. for medium clay. Conclusions and Recommendations SHRP 2 R19A has developed detailed design, construction, and maintenance provisions for using jointless bridges. The com- plete summary of these provisions is provided in Chapter 8 of the Guide. The provisions allow, for the first time, design and use of jointless bridges without resorting to the trial-and-error approach currently practiced. The application of jointless bridges is extended to longer-length bridges by introducing a detail capable of making pile ends act like a pin compared with a fixed head. Further, design provisions are provided to extend the application of jointless bridges to the case of curved girder bridges. However, there is a need to carry out additional work, as outlined in the Guide, before applying the jointless bridge details to the case of curved girder bridges in practice. Seamless Bridge Systems Problem Statement For current U.S. practice, integral abutment bridges represent the best approach to eliminate joints, which are a major con- tributor to the reduced service life of bridges. In the current system, shown in Figure 3.48, expansion devices are placed at the end of approach slab. The level of maintenance required when expansion devices are placed at the end of the approach slab is significantly less than when expansion devices are placed over the abutment. Nevertheless, these expansion devices placed at the end of approach slab still need to be maintained. IntroductIon A seamless bridge system is presented that provides for bridges with long service lives by eliminating the joints over the bridge length, approach slab, and a segment of the roadway. (a) (b) -5,000 -4,000 -3,000 -2,000 -1,000 0 1,000 2,000 3,000 4,000 5,000 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 M om en t ( ki p- in .) Rotation (rad) Specimen 1 (Test) Specimen 1 (Fitted) Specimen 2 (Test) Specimen 2 (Fitted) -50 -40 -30 -20 -10 0 10 20 30 40 50 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 La te ra l L oa d (ki p) Deflection (in.) Specimen 1 (Test) Specimen 1 (SAP 2000) Specimen 2 (Test) Specimen 2 (SAP 2000) Figure 3.47. (a) Moment–rotation curves for specimens from the test and fitted bilinear curves, and (b) load–deflection curves from the test compared with the results of SAP2000 analysis. Figure 3.48. Schematic of the current U.S. practice for bridge–roadway interface.

80 The original idea of the seamless system was developed in Australia for use with continuously reinforced concrete pave- ments (CRCP) (Bridge et al. 2005). The proposed modifications have been made to the Australian system to adapt it to accom- modate typical U.S. practice, in which most pavements are either jointed plain concrete or flexible pavement. Description oF proposeD system In a seamless bridge expansion devices are eliminated entirely, even at the end of the approach slab, such that the bridge deck and pavement become continuous (seamless). The concept was first developed by Bridge et al. (2005) in Australia, has no joints at all, and was intended to be used with CRCP (Figure 3.49). In the case of CRCP, the transition region and CRCP are seam- lessly connected. There is no movement where the transition region ends and the CRCP starts. However, there will be com- pressive forces (during bridge expansion) or tensile forces (dur- ing bridge contraction). In the case of CRCP, because the pavement is reinforced and continuous, the level of tensile and compressive forces that are generated can be tolerated. In the case of flexible or jointed pavement, at the end of transi- tion region (see Figure 3.50), the level of forces generated must be near zero. Further, the movement must be small to eliminate any need for expansion devices (i.e., less than 0.25 in.). There- fore, all that is needed are dowel bars to achieve a smooth transition-zone roadway. In both the Australian and proposed systems, the transition zone is defined as the region between the bridge approach slab and the end joint (or a segment of the CRCP roadway in the Australian system). In the case of the Aus- tralian version of the system, all bridge movements are dissi- pated throughout the transition zone (via friction between the paving and base soil in the Australian system with CRCP). Both in the Australian and proposed versions, during bridge contrac- tion, the opening and closing of microcracks in the transition zone contribute to a reduction in the end joint movements. The advantages of the seamless bridge systems include low maintenance costs, longer service life, elimination of leakage of moisture to bridge elements below deck, improved ride quality, reduced noise, and elimination of grade beams at the end of approach slab. The seamless bridge system also pro- vides a good mechanism for resisting the lateral loads in the pavement associated with skewed or curved girder bridges. One of the best uses of seamless bridges could be in seismic applications because of the continuity between bridge deck and pavement. This continuity provides a good mechanism to resist the longitudinal forces generated during major seismic events. Objectives The objective of this research topic was to adapt the seamless bridge system, originally developed in Australia, for use in the United States. The original idea of the seamless system was developed in Australia for use with CRCP (Bridge et al. 2005). The specific objective therefore, was to modify the Australian system to accommodate jointed plain concrete or flexible pavement, which are the dominant pavement types used in the United States. Scope of Work The key factor in adapting the seamless bridge system to U.S. practice was to develop an appropriate transition system and associated design provisions. At the end of the transition region, the goal was to achieve minimal end movements. Further, within the transition region the goal was to achieve a predictable and controlled crack pattern and controlled axial forces. The system that was developed to meet these needs is shown in Figure 3.51. The transition slab is connected to a secondary slab, which is embedded below grade. The two slabs are connected by a series of small piles. The secondary slab increases the stiffness of the transition region, resulting in the desired short transition length. A similar system, without the secondary slab, may lose its effectiveness after multiple cycles due to compaction of the soil surrounding the small piles. A combination of experimental, numerical, and analytical studies was carried out to develop the design provisions for the seamless system shown in Figure 3.51. A special reinforcement reduction detail is used over the length of the transition zone to achieve a controlled crack Figure 3.49. CRCP–bridge interface as developed in Australia. Bridge Approach Abutment Transition Zone JPCP Figure 3.50. The seamless bridge concept for U.S. practice to be used with jointed plain concrete pavement (JPCP).

81 pattern when the bridge system is in tension. The system behav- ior in tension (temperature reduction with bridge contraction) is an important factor because the crack pattern plays a major role in design life and maintenance costs. Figure 3.52 shows a transition in which the reinforcement detail helps to maintain the desirable crack pattern (Jung et al. 2007). The amount of reinforcement is reduced over the length of the transition region as the force is reduced. Experimental Program The experimental phase of the investigation consisted of constructing a segment of the newly developed transition system in the laboratory and subjecting it to cyclic horizontal loads, simulating the axial loads that would be developed in the transition region due to thermal loads. The experiment revealed the general behavior of the proposed system under cyclic loading and its effectiveness in reducing the bridge end-joint movements, the stiffness of the transition region, and its capability in effectively reducing the bridge movements at the end joint. In addition, the development of cracks, the behavior of the proposed small pile–concrete slab connection detail, and the effect of controlled density fill (CDF) on the transition behavior were investigated. The test specimen represented a segment of the transition system for the prototype bridge. Based on the results of the parametric study (Ala and Azizinamini 2013a), short W10 × 49 piles spaced 4 ft longitudinally were used to connect the CRC PavementSaw Cuts or Induced Design Crack 100% Steel Zone 60% Steel Zone 30% Steel Zone Transition Transition JC Pavement Figure 3.52. Gradual transition from continuously reinforced to jointed pavement. Figure 3.51. Schematic and rendering of the recommended practice for bridge–roadway interface. TransitionBridge Road pavement Transition Zone Approach Zone Soil-nails Embedded Slab Bridge Approach Abutment Transition Zone JPCP Secondary Slab Small Piles

82 top slab (transition region) to the bottom secondary slab. The secondary slab was installed at a depth of 4 ft below grade. The top and bottom slabs were each 12 ft long, 5 ft wide, and 12 in. thick. To simulate the effect of soil surrounding the piles, this system was constructed in a 15-ft-long, 9-ft-wide, and 5-ft-deep concrete container filled with CDF (compacted sand and gravel mix). A large concrete anchor block was built next to the concrete box that provided a reaction for the hydraulic actuators. Figure 3.53 shows schematically the dif- ferent elements of the test setup. The lower, or secondary, slab was fixed against any movement by using posttensioning rods. The upper slab, representing the transition region, rested on the CDF. Loading consisted of moving this slab back and forth, horizontally, using the hydraulic ram, as shown in Figure 3.53. One important design aspect of the proposed seamless system is the pile-to-slab connection design. Failure of the connection between the piles and slabs must be prevented. Further, the entire assembly should be user friendly during construction. The connection detail between the piles and slabs is shown in Figure 3.54. The connection consisted of a plate welded to the pile ends with shear studs. W10x49 4.0" 6. 0" 3. 0" 2. 0" PL16x16x1.0" 4.0" 2.5" CL 2.5" 5/8" Both Flange 3/8" 4.0"6.0" 16" 16" 58.0" Figure 3.54. Small piles used in the test to connect the upper and lower slabs. Figure 3.53. Schematic of the experiment.

83 Figure 3.55 shows a photograph of the concrete box, reinforcement and forming of the bottom slab, and the three small piles. The soil compaction procedure consisted of pouring and spreading the soil in the concrete box (in 4-in.-thick increments before the compaction), and then watering while compacting the soil using a vibratory plate or hand tamper (or both). Figure 3.56 shows the test specimen after the completion of placing compacted soil inside the concrete box. The length of piles extending beyond the compacted soil and to be embedded in the top slab was about 5 in. The final step in construction of the test specimen consisted of placing the reinforcement for the top slab, as shown in Figure 3.57, and casting the top slab. Figure 3.58 shows the final test setup. Displacement-controlled loading was used to move the top slab in a horizontal direction to simulate the thermal move- ments that are applied to a bridge transition slab from the bridge. Results After the test was concluded, the soil beneath the top slab was removed to observe the condition of the piles and their connections to the top and bottom slabs. Figure 3.59 through Figure 3.55. Concrete box, three small piles, concrete forms for the bottom slab, and the bottom slab reinforcement. Figure 3.56. The relative density of the compacted soil is measured using a nuclear density device. Figure 3.57. Reinforcement of the top slab.

84 Figure 3.61 show photographs of the piles and slabs after the conclusion of the test and following the removal of the soil in the concrete box. As shown in Figure 3.59, voids had developed on both sides of the piles in the vicinity of the connections. The widths of the voids were equal to the width of the piles and were about 5 in. deep. Development of such voids may necessitate designing the top slab for punching shear. This aspect of the problem needs additional research before the concept is used in practice. The ideal connection between the piles and the top and bottom slabs is one that would perform elastically without (a) (b) Figure 3.58. Views of the final test setup: (a) west and (b) east. Figure 3.59. Compaction of CDF around a small pile. Figure 3.60. Failure of pile-to-slab connection.

85 sustaining any level of damage. Test results indicated that the type of connection used was not capable of completely achiev- ing this objective given the large displacements to which the specimen was subjected. The type of failure observed in the vicinity of the pile-to-slab connections is shown in Figure 3.60. Inspection of the piles after conclusion of the test revealed formation of yield lines in the webs and flanges of the piles, as shown in Figure 3.61. Figure 3.62 shows the resulting load–displacement hysteresis curves (positive values represent when the slab was pushed, and negative values represent when the slab was pulled). Figure 3.63 is the skeleton curve, constructed by connecting the apexes of the load–displacement curves at each applied displacement level. Two skeleton curves were constructed. One skeleton curve corresponded to the first cycle at each applied displace- ment level and another corresponded to the last (fifth) cycle at each applied displacement level. The skeleton curve indicates that there is no abrupt failure point. Rather, the secant stiffness of the system decreases fol- lowing a peak value. This behavior is related to gradual failure of the detail used to connect the pile to the slab. Conclusions and Recommendations Results of the experimental work, along with extensive numer- ical and analytical work carried out by Ala (2011) and Ala and Azizinamini (2013a, 2013b) resulted in the following major conclusions: 1. The proposed modification to the existing Australian system has promise for developing a completely jointless system for U.S. practice. Additional work needs to be carried out before the system is put into practice. 2. The type of connection detail used in this study to connect the small piles to the top and bottom slabs needs to be modified. The connection needs to be protected from any damage, and yielding and failure should happen outside the connection. This is similar to the concept used in earth- quake design, in which some of the elements of the structure Figure 3.61. Yield lines of the web on a small pile. Figure 3.62. Hysteresis load–displacement curves of the test specimen.

86 are “protected” elements and must remain elastic during an entire seismic event. TenTaTive Design Provisions Based on the work carried out, there are two tentative design provisions for the proposed seamless system: 1. The initial design of the system is an iterative process in which the length of the transition and secondary slabs; the shape, size, and spacing of small piles; and the embedment depth of the secondary slab are determined through struc- tural analyses of various system configurations. The initial design will also determine the demand in all components. 2. In the component design, various elements of the system will be designed according to the LRFD Specifications. The reduced cracked stiffness of the system in tension may be neglected during the initial design. Relevant pavement design loadings are the longitudinal strains (thermal effects, creep, and shrinkage) and the out-of- plane effects (traffic, settlement of approach embankments, and rotation transferred from the bridge deck). sTrucTural analysis Structural analysis of the proposed seamless bridge should be carried out as a system analysis in which all bridge components, including the transition zone, are incorporated in the analysis model. Only the effect of uniform temperature change needs to be considered in initial design of the transition system. The calculation of uniform temperature changes should be in accordance with Article 3.12.2 of the LRFD Specifications (AASHTO 2010b). The soil–pile interaction can be modeled in the structural analysis by using linear springs. The linear spring stiffness depends on the relative density of the CDF surrounding the small piles and the confinement pressure. Because the CDF is manually compacted, its relative density should be measured during the compaction process, and this compaction should be related to the soil stiffness (Greimann et al. 1987). The connection of the small piles to the slabs can be assumed rigid for analysis. The structural analyses should take into account the effects of longitudinal stiffness reduction due to development of tensile cracks in the transition slab (temperature reduction with bridge contraction). To do so, an iterative structural analysis of the system in conjunction with cracked section analyses is required. For the first iteration, the tensile forces in the structure can be assumed equal to the compressive forces due to thermal expansion. Cracked section analyses should be conducted for various segments of the transition slab, and the axial stiffness of the slab segments should be modified in each step. The structure is then analyzed using the modified in-plane stiffness to determine the updated in-plane tensile axial forces. This procedure will be repeated until convergence of the axial forces. Design of aPProach slab anD briDge Deck 1. The resistance against thermal contraction due to the presence of a transition zone will create tensile forces and hence concrete cracking in the bridge deck, approach slab, and transition zone. Additional reinforcement may be required in these areas to control cracking and to achieve desired crack spacing and width. Figure 3.63. Skeleton curve of the hysteresis load–displacement curves.

87 2. The approach slab axial strength should be adequate to resist compressive thermal stresses to avoid crushing of concrete. 3. The approach slab should be designed for the differential settlement between the bridge abutment and the transition system. Embankment settlement is another important cri- terion to check for the approach slab. The bridge approach embankments should be designed to achieve a long-term settlement of less than 0.75 in. to minimize the impacts on the motorway pavement (Bridge et al. 2005). To account for the probable geotechnical and construction variations, however, a more conservative approach embankment settlement of 1.5 in. should be assumed for the seamless pavement design (Bridge et al. 2005). Bending analysis should be considered (Bridge et al. 2005). Design oF transition slab 1. The transition slab should be designed to avoid concrete crushing in compression and also to achieve a uniform cracking pattern in tension. 2. The transition slab should have adequate punching shear capacity against the design truck heavy wheels and adequate moment transfer capacity between the small piles and the slabs. 3. The thickness of the transition slab is determined from the punching shear requirements, connection to the small piles, and also to some extent the in-plane horizontal stiffness of the system to reduce the movement of the end joint. 4. Reinforcement of the transition slab should be determined from cracked section analysis under tensile in-plane forces. 5. The transition slab should be checked for the maximum bending moments between the rows of small piles. 6. The slab should be designed for two-way (punching shear) and one-way shear. Design oF seconDary slab (bottom layer slab) 1. The thickness of the secondary slab is governed by punching shear and connection to the small piles (the secondary slab thickness will most likely be equal to the top slab). 2. The secondary slab is designed for the bending moment and one-way shear exerted from the small piles to the slab. 3. The secondary slab should be controlled for the bending moment due to the soil pressure underneath. Design oF small piles connecting top anD bottom slabs 1. The required stiffness of the small piles is determined in the initial design phase. 2. The small piles should be designed for the moment, shear, and maximum drift between the two top and bottom slabs. 3. The connection of the pile to the top and bottom slabs should remain elastic for maximum probable longitudinal movement. 4. The design of small piles should consider “overstrength” to prevent failure of connection. In the event that the strength of the pile is larger than the required strength by the designer, the piles should be able to achieve yielding before the failure of the connection. connection oF small piles to slabs Figure 3.64 shows the type of connection detail used to connect the piles to the top and bottom slabs. Figure 3.65 shows a portion of the connection that will be embedded in the top and bottom slabs. The connection should be capable of resisting cyclic loading at maximum probable longitudinal movement of the top slab relative to bottom slab without any damage. The experimental results showed that the detail used needs modifications and is not suitable for the proposed seamless system. Design oF controlleD Density Fill Design of the CDF consists of selecting the geomaterial type and compaction requirement. The required compaction depends on the stiffness requirement around the small piles and prevention of settlement under the top slab. It is very important to achieve the required compaction around the small piles. Granular material is recommended for ease of W10x49 4.0" 6. 0" 3. 0" 2. 0" PL16x16x1.0" 4.0" 2.5" CL 2.5" 5/8" Both Flange 3/8" 4.0"6.0" 16" 16" 58.0" Figure 3.64. Small piles used in the test to connect the upper and lower slabs.

88 compaction, low long-term settlement, and reduced potential for gap development around the piles caused by repeated pile movements. The moisture–density relationship of the soil, maximum and minimum density (relative density) tests, and in-place moisture content and density determinations during placement of the backfill (using a nuclear density gauge) are the recom- mended soil mechanics tests. CraCked SeCtion analySiS The ACI Committee 224 document, Cracking of Concrete Members in Direct Tension, provides methods of determining the maximum probable crack width and stiffness reduction for an axially tensioned concrete member (ACI 1997). The maximum probable crack width in a fully cracked member can be determined from Equation 3.1: = × −0.10 10 (3.1)max 3 3W f d As c where dc = distance from the center of the bar to the extreme tension fiber (in.), fs = service stress in the reinforcement (ksi), and A = effective tension area of concrete surrounding the tension reinforcement, and having the same centroid as that reinforcement, divided by the number of bars (in.2). Once the allowable crack width is determined from Equa- tion 3.1, the service stress–strain in the reinforcement can be determined. The LRFD Specifications define an exposure factor (ge) that is 1.00 for a Class 1 exposure condition (crack width of 0.017 in.) and 0.75 for a Class 2 exposure condition. Areas for a Class 2 exposure condition include decks and substructures exposed to water. The crack width is directly proportional to ge, which can be adjusted directly to be 0.75 for a Class 2 exposure condition or an approximate crack width of 0.012 in. The amount of required reinforcing steel (As) can be deter- mined from the axial tensile force in the member (P = fs × As, and so As = P/fs). The direct tensile strength of the concrete is f ′t. The stress in the reinforcing bars after the crack occurs is determined from Equation 3.2: ′ = ′ ρ − +     1 1 (3.2),f f ns cr tl The axial load that causes first cracking in the axially ten- sioned member is given by Equation 3.3: = ′ × (3.3),P f Acr s cr s The average strain in the tensile member can be calculated from the CEB Model Code, as shown by Equation 3.4: ε = ε −         1 (3.4), 2 k f f m s s cr s where es = fs/Es, and k = 1.0 for first loading and 0.5 for repeated or sustained loading (k = 0.5 for this study). The effective modulus of elasticity of steel bars is given by Equation 3.5: = −         1 (3.5)sm , 2E E k f f s c cr s Equation 3.5 is used to increase the stiffness of the reinforce- ments in the tensile member to compensate for the presence of some concrete after cracking. The effective axial cross-sectional stiffness of the tensile concrete member is (EA)eff = EsmAs. The ratio (EA)eff/(EA) is the section modification factor that should be used in the structural analysis to modify the axial stiffness of the member in tension. Figure 3.66 shows the general flow- chart for the cracked section analysis. Figure 3.66. Cracked section analysis flowchart. Concrete SlabStirrups Bars Studs Baseplate Figure 3.65. Recommended small pile–concrete slab connection.

89 In conclusion, the combination of experimental, numerical, and analytical work conducted provided information to adapt the seamless bridge system to U.S. practice. Tentative details and design provisions are provided. It is recommended that the following scope of work be carried out to further develop the system: 1. Develop connection details for connecting top and bottom slabs. These connection details should be such that under maximum probable loading, the connection should remain completely elastic without sustaining any damage. 2. Carry out additional laboratory tests to further comprehend the behavior of the system using the new connection details. 3. Refine the tentative design provisions. 4. Conduct a field demonstration project. Expansion Joints Although the best option is to eliminate their use, expansion joints may still be needed when the total bridge length exceeds practical limits for jointless bridges. An evaluation of current practice regarding the use of expansion joints was carried out, and recommendations for the best practices were developed and summarized in Chapter 9 of the Guide. The following is a brief summary of information provided in the Guide for enhancing service life of expansion joints. Introduction Bridge elements are subjected to various loads, including traffic and environmental loads, which result in movement of bridge elements. One of the key factors affecting the service life of bridges is how to address thermal expansion and contraction of the bridge elements. This design issue is handled in two distinct ways. One option is to provide expansion joints at designated locations within the superstructure. By doing so, the designer forces all thermal deformation to take place at these discrete locations. The other option is to make the super- structure and deck continuous and assume that the thermal movement will be accommodated by the flexibility of the substructure, such as with the use of integral abutments. In such cases, the joint is usually moved away from the bridge abutment and placed near the end of the approach slab. The following main issues should be considered when addressing the service life design of various expansion devices: • Identification of appropriate design methodologies; • Selection of durable expansion device types considering life-cycle costs; • Specification of best practices for construction; and • Development of an effective maintenance plan. An ideal expansion joint should be able to (Lee 1994) • Accommodate thermal expansion and contraction; • Accommodate movement due to traffic-induced loads; • Provide a smooth ride; • Prevent the creation of hazards and safety issues; • Accommodate needs during snow removal; • Prevent leaking of moisture and other chemicals to elements below the superstructure; • Have a long service life; • Be maintenance free or require minimal maintenance; and • Be cost-effective. Advantages and disadvantages were identified for several typical expansion device types used in practice. The Guide provides a summary of these findings. Expansion joint devices can be classified into three categories based on the maximum longitudinal movements. The cate- gories and specific devices that were considered within each category are as follows: • Expansion joints capable of accommodating small longi- tudinal movements (less than about 3 in.) 44 Compression seal, 44 Poured sealants, 44 Asphaltic plug joint, 44 Sheet seals, 44 Sliding plate joint, and 44 Open joints; • Expansion joints capable of accommodating medium longitudinal movements (between 3 and 5 in.) 44 Strip seals; • Expansion joints capable of accommodating large longitu- dinal movements (in excess of 5 in.) 44 Modular expansion joints, and 44 Finger-plate joints. Factors and Considerations Influencing Expansion Joint Service Life Factors that affect the service life of expansion joints were identified and summarized in the form of a fault tree. These summaries are provided in Chapter 9 of the Guide. Chapter 1 of the Guide includes a description of a fault tree. Overall Strategies for Enhanced Expansion Joint Service Life Providing bridge decks with enhanced service life requires a full understanding of the potential deterioration mecha- nisms. These mechanisms are associated with load-induced conditions, local environmental hazards, production-created

90 deficiencies, and lack of effective operational procedures. The study concluded that consideration of the following issues can lead to expansion joint devices with enhanced service life: • Design methodology; • Expansion joint system selection; • Life-cycle cost analysis; • Construction practice specifications; • Maintenance plan; and • Retrofit practices for expansion devices. A detailed discussion of each topic is provided in Chapter 9 of the Guide. Service Life Design Aids for Various Expansion Devices Two types of tables were developed that can assist in compre- hending the service life issues associated with various expansion joint devices and selecting the optimum device. These tables, which are referred to as technology and strategy tables, are briefly described below; additional details are provided in Chapter 9 of the Guide. technology tables Chapter 9 of the Guide provides technology tables for various expansion joint devices. These tables summarize service life and durability issues related to expansion joints and aid the design engineer in selecting the proper expansion device. The information could be used in several ways at the design or maintenance level. The purpose of these technology tables is to identify the most common types of service life problems encountered in commonly used expansion joints. This infor- mation can subsequently be used to provide possible solutions to service life problems encountered with expansion joints. An example of a technology table is given in Table 3.13. strategy table For expansion joint Devices A strategy table was developed to assist in selecting expansion joint devices. This table lists important parameters related to service life of expansion joint devices. Because of the lack of available quantitative data, qualitative information is provided for most categories in the strategy table, which is shown in Table 3.14. Joints in Modular Systems: Adjacent Concrete Box Beams This section summarizes the research topic that evaluated the performance of joints in adjacent box beam bridges, specifically with regard to the transverse connection between box elements. A complete report of this study is given in Appendix C. Problem Statement Precast adjacent box beams are commonly used for short- and medium-span bridges (typically 20 to 120 ft), especially on secondary roadways. These bridges consist of multiple precast box beams that are butted against each other to form the bridge deck and superstructure. Adjacent box beams are generally connected using partial- or full-depth grouted shear keys along the sides of each box. Transverse ties are typically used in addition to the grouted shear keys; these ties may vary from a limited number of threaded rods to several posttensioned tendons. In some cases, no additional wearing surface is applied to the structure, but in other cases, a noncomposite topping or a composite structural slab is added. Bridges constructed using adjacent box beams have been in service for many years and have generally performed well. However, a recurring problem in noncomposite systems with bituminous wearing surfaces is cracking in the grouted joints between adjacent units, resulting in reflective longitudinal cracks in the wearing surface. The development of these cracks over the shear keys allows chloride-laden water to penetrate to the sides and bottom of the beams and cause corrosion of the nonprestressed and prestressed reinforcement. In addition, the load distribution among the beams is adversely affected, meaning the loaded beams are required to carry more load than originally intended. These conditions led to severe beam deterioration and actual midspan fracture and collapse of an exterior beam on a bridge in Pennsylvania. There is a need to improve the transverse beam-to-beam connections in adjacent box beam systems to avoid these types of problems. Objectives The overall objective of this task was to improve the perfor- mance of adjacent box beam bridges, specifically with regard to the transverse connection. The specific objectives were to improve the structural capacity of this system in the trans- verse direction and to prevent longitudinal joint leakage. This was done by developing a detail that has the ability to transfer both moment and shear, instead of only shear, in the transverse direction. Scope of Work This study included the following tasks: • Literature and industry practice review; • Literature review; • Posttensioned concept development; • Testing program; • Design recommendations; and • Recommendations for further study.

91 Table 3.14. Strategy Table for Expansion Joints Maximum Longitudinal Movements (in.) Strategy Potential Deterioration Mode Options to Improve Service Life Life-Cycle Cost Difficulty Associated with Replacement Performance Record Expected Service Life (years)Initial Maintenance <1 in. Field-molded or other equivalent joint types See technology tables Provide sec- ond layer protection Low High Low Good 1 to 3 1 to 3 in. Strip seal joints See technology tables Provide sec- ond layer protection Medium High High Very good 3 to 30 Compression seal joint See technology tables Provide sec- ond layer protection Medium High High Very good 3 to 30 >4 in. Finger-plate joints See technology tables See technol- ogy tables High High High Very good 10 to 50 Modular expansion joints See technology tables See technol- ogy tables Very high Very high Very high Good 10 to 50 Table 3.13. Bridge Joints Technology Table Field Molded (or Field Formed) Joints Range of Movement: Less than 1 in. Expected Range of Service Life: 8–9 years Service Life Problems Solutions Advantages Disadvantages System Preservation Requirements Spalling and cracking of adjacent concrete Deeper notch Field molded joints are inexpensive and easy to install. They are best suited for single- span bridges with a maximum length of about 100 ft. This detail is an economical solution for simple-span bridges with spans of less than about 100 ft, especially in low-traffic areas where a few hours of interruption to traffic is not a major problem. It requires mainte- nance and has a short service life. In most cases, it needs complete replacement every 2 to 3 years. It requires regular 3-year inspection and replacement in most cases. Filling with silicone Beveling edges of concrete Placement of correct silicone thickness Correctly mixing silicone material Installation problems Training installation technicians Providing detailed installation plans by the manufacturer or contractor. Having a represen- tative of the joint supplier onsite during installation. Snowplow damage Not allowing inferior quality of bonding agents Installation slightly below the top of the deck elevation Debris accumulation Inspection and cleaning regularly Water leakage Draining water away from the joint Insulation Hot-weather damage Using high-quality silicone sealer

92 Results literature anD practice review A detailed literature review identified and evaluated various design and detailing practices for transverse connections in adjacent box girder bridges. Current practices were identified in Ontario, Canada; Japan; South Korea; and several states in the United States. There have been a variety of shear key details used along with various connection details using post- tensioning ties or threaded rods. Most practices in the United States have used partial-depth shear keys, with threaded rod connectors at diaphragm locations. In Japan, cast-in-place (CIP) concrete is placed in wide, full-depth joints between boxes, and posttensioning is applied through several ducts located at different elevations. All boxes are then covered with an asphalt concrete wearing surface. Longitudinal cracking and concrete deterioration have rarely been reported when this practice is used. In South Korea, the transverse connection between boxes is achieved by the use of a middepth shear key completely filled with CIP concrete, and heavy transverse posttensioning is applied, similar to the Japanese practice. Research conducted by the West Virginia DOT on several bridges with joint fracture and longitudinal wearing surface cracking revealed that vertical shear failures in the keys were due to inadequate grout installation and inadequate trans- verse tie force. Consequently, the West Virginia DOT changed its practice to install posttensioned high-strength ties, use pourable epoxy instead of nonshrink grout in shear keys, and sandblast surfaces to be grouted to improve the bonding of shear key material. In New York, two major changes were adopted in the state DOT’s design standards to address longitudinal cracking: shear keys were increased to almost the full depth of the precast boxes, and the number of transverse tendons was increased to three for spans less than 50 ft and 5 ft for longer spans. The Precast/Prestressed Concrete Institute subcommittee on adjacent member bridges conducted a survey on the current practices in the design and construction of adjacent box girder bridges in the United States and Canada. Most of the transpor- tation agencies reporting had experienced premature reflective cracking in the wearing surfaces on bridges built in the late 1980s and early 1990s. These agencies emphasized the impor- tance of eliminating these cracks, which allow the penetration of water and deicing chemicals leading to the corrosion of reinforcing steel in the sides and bottoms of the concrete boxes. The following preventive actions, among others, were recommended based on lessons learned in the last two decades: • Use of a CIP deck on top of the adjacent boxes to prevent water leakage and to uniformly distribute the loads on adjacent boxes; • Use of full-depth shear keys, due to their superior perfor- mance over the traditional top flange keys; and • Use of transverse posttensioning to improve load distribu- tion and minimize differential deflections among adjacent boxes. Several nonposttensioned alternatives have been proposed to emulate monolithic construction and eliminate the prob- lems associated with posttensioning and shear keys. In these alternatives, a reinforced concrete connection is used along the entire bridge length instead of the posttensioned diaphragms to transfer moment, shear, and torsion. The Texas DOT developed a unique adjacent member system that eliminates the necessity of either grouting or posttensioning operations to connect adjacent girders. In this system, wide, nearly full-depth shear keys are filled with con- crete while placing a reinforced CIP concrete overlay on top. The ordinary slab concrete is allowed to flow into the large shear keys provided by the I-shaped sides of the adjacent box girders. Similar to the Texas DOT, the state of Illinois also adopted a new connection detail but with a narrower, more conventional shear key. However, they increased the required number of transverse ties to provide better performance of the shear key and assist the beams in working together. A 5-in. CIP concrete overlay was required to help distribute the loads uniformly across all beams and protect the shear keys against cracking. previous university oF nebraska–lincoln research Previous research by the University of Nebraska–Lincoln (UNL) investigated the benefits of nonposttensioned systems and developed two new connection details referred to as the wide joint system and the narrow joint system. The wide joint system has a similar beam and joint shape to the one just described from Texas, but differs in that the UNL concept eliminated the concrete overlay and the diaphragms. Top and bottom transverse reinforcement was used in a wide shear key filled with concrete to connect adjacent boxes instead of the reinforced concrete composite topping. This monolithic joint with top and bottom reinforcement provides a continuous connection that transfers shear and moment between boxes and eliminates the need for intermediate or end diaphragms. The elimination of the CIP topping and diaphragms would significantly speed production and construction operations and would reduce the material, labor, and erection costs. Shear keys used in the wide joint system are wide, full-length and full-depth, and require a slight modification to the AASHTO standard box cross section and consequently the forms. The narrow joint system has a similar beam and joint shape to the one referred to above from Illinois, but differs in that it eliminated diaphragms and replaced the single middepth transverse tie with top and bottom transverse tie rods every

93 8 ft to connect each pair of adjacent boxes at the top and bottom flanges (see Figure 3.67). These rods provide continuous con- nection that transfers shear and moment between adjacent boxes more efficiently than the middepth transverse ties at discrete diaphragm locations. A slight modification is made to the standard box cross section by developing full-length horizontal and full-depth vertical shear keys. The wide joint and narrow joint specimens were tested using a similar loading scheme of 6.14 kips fatigue loading for 2,000,000 cycles followed by 18.4 kips fatigue loading for 5,000,000 cycles. The two connections had superior perfor- mance as there was no cracking or water leakage during testing. The two connections were also tested up to failure under static loading to determine their ultimate flexure and shear capacities. The two systems’ capacities were significantly higher than that of current connections. Posttensioned system develoPment A modified version of the UNL narrow joint system was devel- oped to allow for posttensioning with transverse high-strength steel rods. The system continues to eliminate the typical use of diaphragms and a concrete overlay for load distribution, which are details common to most adjacent box girder appli- cations. The system maintains the use of top and bottom reinforcement; however, ducts have been moved inward so that posttensioning extends through the box beam voids (see Figure 3.68). This arrangement simplifies inspection of the posttensioning as it is now unbonded with the box section and shear keys. Construction time is reduced due to the elimi- nation of grouted ducts, and venting is no longer an issue with girder production. A duct-within-a-duct setup is proposed so that post- tensioning can be applied after grouting of the shear key. The interior duct is necessary, as otherwise the duct would close during grouting, and posttensioning could not be threaded between adjacent members. Furthermore, posttensioning should be applied after the shear key is placed for the joints to be placed under initial compression. To calculate the load effects on the proposed connection, a three-dimensional (3-D) computer model was developed Elevation View Figure 3.67. UNL narrow joint system with threaded rods. 714" 734" Spherical Nut Dished PL Coupling Nut Threaded Rod @ 8 ft Spa Elevation View Interior Duct @ 8 ft Spa Exterior Duct @ 8 ft Spa Grout Shear Key Figure 3.68. Posttensioned system concept.

94 using SAP2000 to replicate a simple-span adjacent box beam bridge and to conduct a parametric study to determine the effects of box girder depth, span width, and span-to-depth ratio. The loads applied for analysis included the dead load of concrete curbs and railing and an HL-93 live load with a dynamic load allowance of 0.33. Single- and multiple-lane loadings were applied to determine the most critical loading case for the design of the transverse connection. Dead load due to self-weight and wearing surface was not considered as it is uniformly distributed over the bridge and, therefore, does not generate load effects in the transverse direction. The weight of the concrete curb and railing was assumed to be 0.48 kips/ft applied to the outside box girders. Design charts were developed for the posttensioned system based on the results of the 3-D model analysis. The effects of box girder depth, span width, and span-to-depth ratio were considered. The effect of the box depth and bridge width on the tension force in the transverse connection for the posttensioned system is shown in Figure 3.69. This figure is developed for zero skew and a span-to-depth ratio of 30. The figure indicates that the tension force increases by increasing the bridge width and decreases by increasing box depth. Also, it can be noticed that the effect of bridge width on the required tension force is higher on narrow bridges (width less than 52 ft) than on wide bridges (width greater than 52 ft). Forces for this chart result from factored loads (1.25 DL [dead load] + 1.75 LL [live load]) and can be divided by the effective prestress of the posttensioning to determine the required area of reinforcement. Figure 3.70 shows the effect of span-to-depth ratio and box depth on the transverse tension force in the posttensioned connection. This figure was developed for a bridge width of 52 ft and zero skew. The figure indicates that the higher the span-to-depth ratio is, the higher the required transverse tension force will be. Experimental Program An experimental program tested the performance of the post- tensioned system under fatigue and ultimate loading. A test specimen was designed using four 48-in.-wide by 27-in.-deep by 8-ft-long box beam elements positioned side by side and connected transversely by grouting and posttensioning. This created an overall specimen 16 ft long by 8 ft wide by 27 in. deep for testing. The length of the specimen consisted of the four box sections connected on each side. Fatigue was represented by a 5,000,000 cyclic load applied to create tension in the top and bottom transverse ties (tested separately). The performance of the joint was evaluated by data collected from strain gauge, deflection, and leakage monitoring at the critical location of tensile stress along the transverse connection. Two test frame setups were necessary to develop tension in both the top and bottom elements of the specimen. Top tension was deemed the critical case to test first, because it results in cracking at the top of the member where an actual girder is most susceptible to environmental corrosion. Cracks were monitored by leakage from a dam built around the middle joint, as well as white latex paint to aid with visibility. Beam supports for both cases were placed at stiff locations (i.e., at webs and joints). To test for top tension, a support beam was placed under the middle joint and at one end of the specimen so that the Figure 3.69. Effect of box depth and bridge width on the posttensioned system.

95 opposite end was cantilevered. Thus, a load placed at the cantilevered end would cause tension in the top of the joint directly over the center support. Testing for bottom tension in the middle joint was done by placing support beams at either end of the specimen with the load applied at the center. A neoprene bearing pad was placed between the specimen and the support beams to provide additional flexibility for the system. test loaDing The loads for testing were determined by applying the theory from the previously developed 3-D models of an adjacent box beam bridge system. First, a model was set up to determine the axial force produced by ultimate and fatigue loads in the trans- verse direction of a prototype bridge with a 64-ft span and thir- teen 4-ft-wide by 27-in.-deep adjacent box beams, which totaled 52 feet in width. Bridge loads included an assumed curb and rail load equal to 0.48 kips/ft and 12-ft lanes with standard HL-93 design vehicular and fatigue truck loading. The weight of the structure is uniform and does not affect the transverse direction, thus it was not included in the design analysis. Curb and rail loads were applied to the exterior edge of the exterior box beams. Ultimate loading conditions included single and multiple lanes; fatigue conditions considered a single lane only. Single lanes were placed at the center of the bridge width to create maximum tension in the bottom flange of the box girders and at the edge to create maximum tension in the top flange. The LRFD Specifications include a 33% dynamic load factor for ultimate analysis and a 15% dynamic load factor for fatigue analysis. Long-term effects were accounted for with an infinite life factor of 2 for fatigue. This results in load combinations of 1.25 DL + 1.75 LL for ultimate and 1.25 DL + (0.75)(2) LL for fatigue. testing The top tension test consisted of applying an 18.4-kip load for 5,000,000 cycles at the cantilevered exterior joint of the four- box specimen with supports at the center and opposite edge. Ponded water over the joint was used to monitor joint integrity. Strain gauges were placed next to the load and on either side of the center joint. No joint cracking occurred during the course of the experiment. Supports were then rearranged to both ends of the member, and the load was repositioned to the center joint for tension in the bottom flange. A 17.4-kip load was applied for 5,000,000 cycles, and again no joint cracking occurred. For ultimate capacity testing, the test setup for bottom tension was used. Ultimate moment capacity was calculated as 280 kip-ft using strain compatibility and was found to be slightly higher during testing. An applied load of 67 kips, in addition to the self-weight of the specimen, reached 300 kip-ft before failure. Cracking did not propagate until reaching a 60-kip total load, and then deflection increased exponentially from 0.1 to 3.31 in. over the next 30 kips applied. Design provisions A method was outlined for determining the required post- tensioning force using Figure 3.70, based on the bridge width and box depth for a given span-to-depth ratio. An example is provided in Appendix C. A construction sequence was also developed for placing adjacent box beams and post - tensioning. Conclusions and Recommendations The general objective of this study was to improve the perfor- mance of transverse connections currently used in adjacent box girder bridge applications. This was done by developing 0.0 20.0 40.0 60.0 80.0 100.0 120.0 25 30 35 40 Te ns io n fo rc e (k ip s) Span to Depth Ratio 27 42 Box Depth (in.) PT connection every 8 ft Bridge Width = 52 ft Figure 3.70. Effect of span-to-depth ratio and box depth on the posttensioned system.

96 a detail capable of transferring moment and shear in the transverse direction. UNL combined the efforts of this research with a previous study at the University of Nebraska–Omaha campus in which intermediate and end diaphragms and a concrete overlay were proven unnecessary for adequate load distribution in the transverse direction of nonposttensioned adjacent box girder connections. The benefits of this system, as well as the modified system that allows for posttensioning, include significantly simplifying box production, improving the rate of construction, easier inspection of voids, and reduced project costs. It was found that posttensioning also increases the capacity and efficiency of the section because joints are placed under compression and are less likely to experience reflective cracking and leakage. Based on test results, it can be concluded that a posttensioned transverse connection without diaphragms or concrete overlay can be designed and detailed to have comparable performance to typical connections while being more economical and prac- tical. The tested specimen had excellent performance under both static and cyclic loads. It is proposed that the transverse, posttensioned connection developed in pursuit of this research be applied in a full-scale application for observation of the joint performance under actual weathering and loading conditions. Joints in Modular Systems: Closure Pour Details Problem Statement The rebar detail used in the closure region between adjacent slabs has been investigated in the past. This study concen- trated on the development of an alternative detail, which is discussed in this section. The proposed detail costs less than a headed bar detail, which is commonly used. Objectives The work carried out under this topic is a good example of what is meant by proof of concept testing. The main objective of this study was to show that rebar in the closure region of modular bridge systems having a precast girder–slab can be developed by 90° hooks. Limited experimental work was carried out. Results demonstrated the feasibility of the concept; however, additional work will be needed before the detail is used in practice. Scope of Work Six specimens containing closure regions and using various details were subjected to both positive and negative moment loading to investigate their behavior and failure modes under ultimate load. Experimental Program Figure 3.71 shows a portion of the deck the test specimens were intended to represent. The test specimen modeled an 8-ft-wide and 3.5-ft-long portion of a modular bridge system having a precast slab, as shown in Figure 3.71. Six specimens were constructed to investigate the closure region behavior: two specimens had straight bars without a closure region and were used as control specimens, and the other four specimens contained a closure region between adjacent slabs that was used to analyze different rebar details. The total width of the closure region for the slabs was 12 in.; the rebar positioning was staggered to avoid constructability issues. The slabs were cast in two stages, the first stage being a 6.5-in. pour of only the outside regions, leaving an open space between them. One month later, the second stage was cast, which included the closure regions and a 2-in. topping over the previously cast regions. Table 3.15 provides important dimensions and aspects of each specimen. Figure 3.71. Slab section from adjacent slabs. Model Specimen Closure Region

97 Two slab specimens were built as a control group. These slabs did not have the closure region and used straight rebar as would typically be used in bridge construction. Two test specimens using a headed bar were formed. The heads at the end of the rebar come in multiple shapes and sizes. Tests performed by the Headed Reinforcement Cor- poration (HRC) showed that the circular heads provided better connection to the rebar and had a higher ultimate strength than the rectangular heads. Therefore Type 220, No. 4– and No. 5–headed bars were used in the current testing program. Details of the headed bars are given in the HRC200 specification (HRC 2011). Confinement reinforce- ment was used on both faces of the slab to prevent the vertical punch-out of any of the headed bars in the closure area. Figure 3.72b shows the construction details for the headed bar connection. The final two specimens were formed using hooked bars. Hooked bars may be obtained from any local steel fabricator, Table 3.15. Slab Specimen Summary Specimen Closure Region (Yes/No) Rebar Type Moment Applied Concrete Cover at Tension Face (in.) Concrete Cover at Compression Face (in.) S_N No Straight Negative 3.0 1.5 S_P No Straight Positive 1.5 3.0 HD_N Yes Headed Negative 3.0 1.5 HD_P Yes Headed Positive 1.5 3.0 H_N Yes Hooked Negative 3.0 1.5 H_P Yes Hooked Positive 1.5 3.0 (a) (b) 0.375" 12" 2" 2" 1" 1.5" #4 Confinement Bars 2"2" 12" 1" #4 Confinement Bars 1.5" Figure 3.72. (a) Hooked and (b) headed construction details.

98 greatly reducing the time and cost of fabrication and shipment to the work site. The hooked bar also provides greater clear- ance on both faces compared with the headed end detail. As with the headed rebar, confinement bars were used on both faces of the slab. Figure 3.72a shows the construction details for the hooked bar connection. Additional information and the reinforcement plan for the hooked bar construction can be seen in Figure 3.73. The bend diameter for the hooks was two in. Photographs of the headed and hooked specimens before casting the closure region are shown in Figure 3.74. Results negative moment benDing Deflections were found to be consistent across the width of the slab. However, the end with three bars consistently (a) (b) Figure 3.74. (a) Headed and (b) hooked specimens before closure region casting. Figure 3.73. Hooked bar construction details. #5 @ 1 2" ct rs . (b ott om ) # 4 @ 1 2" c trs . (to p) 3. 5' #4 Confinement Bars (top & bottom) 8' #4 @ 12" ctrs. (top) #4 @ 9" ctrs. (top) #5 @ 12" ctrs. (bottom) #5 @ 9" ctrs. (bottom) # 4 @ 1 2" c trs . (to p) #5 @ 1 2" ct rs . (b ott om ) 9" 9" 3.5'3.5' 1'1" 2" 1.5" 8.5"

99 experienced slightly higher deflections than the other side with four bars, for both the hooked and headed bar specimens. The moment–deflection relationship at the center location of the specimens for each of the three specimens tested in negative bending is shown in Figure 3.75. The following observations were made from Figure 3.75: • All the specimens tested in negative bending showed a similar moment–deflection relationship. The small dif- ference in moment–deflection behavior of the different specimens can be partly attributed to differences in material strengths. • All the specimens show sufficient ductility to give a warning before failure and thereby follow the design philosophy of preventing a brittle mode of failure. • Hooked rebars, which cost much less and are easier to con- struct than headed rebars, show behavior similar to headed bars in terms of both strength and ductility. positive moment benDing The bending behavior of the specimens under positive moment was similar to the bending behavior of the specimens under negative moment. The moment–deflection relationship at the center location of the specimen for each of the three speci- mens tested in positive bending is shown in Figure 3.76. The following observations were noted from Figure 3.76: • All the specimens tested in positive bending show a similar moment–deflection relationship. The observed differences 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.1 0.2 0.3 0.4 0.5 0.6 M /M n Deflection (inch) Straight Hooked Headed Figure 3.75. Moment versus deflection for the center location (Pot 5) of specimens in negative bending. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.1 0.2 0.3 0.4 0.5 0.6 M /M n Deflection (inch) Straight Hooked Headed Figure 3.76. Moment versus deflection for the center location (Pot 5) of the specimens in positive bending.

100 in moment–deflection curves between different specimens for the positive section are smaller than the differences observed for the negative section. • All the specimens show sufficient ductility to give a warn- ing before failure and thereby follow the design philosophy of preventing a brittle mode of failure. • The results obtained from hooked rebars, which cost much less and are easier to construct than headed rebars, match well with the results obtained with straight rebar, which indicates the capability of the hooked detail to develop the reinforcement in the closure pour regions. The ultimate moment capacity Mu of the, along with the ratio Mu/Mn, is shown in Table 3.16. In positive bending, the straight bar specimen and headed bar specimen have the same Mu; however, Mn for the headed bars specimen is higher than the straight bars specimen due to higher fy reinforce- ment used in the headed bar specimens. Therefore, Mu/Mn for the headed bar specimen is lower than Mu/Mn of the straight bar specimen. Mu/Mn of the hooked bar specimen is almost the same as the headed bar specimen in positive bending. For negative bending, the headed bar specimen has the lowest value of Mu and Mu/Mn ratio. The straight bar speci- men has the highest value of Mu and Mu/Mn ratio in negative bending. The hooked bar specimen lies in between the two extremes for the case of negative bending (see Table 3.16). Test results have shown that the longitudinal connection with a hooked bar detail provides adequate flexural capacity and ductility ratio. In addition, a similar cracking pattern was observed in the specimens detailed with hooked bars and headed bars. Conclusions and Recommendations Limited experimental work was carried out to evaluate the merits of using a hooked bar detail in the closure pour region versus a headed bar detail. Although the headed bars provide sufficient strength and ductility when subjected to both posi- tive and negative moment, the small cover between the heads of the bars is often a concern to bridge owners because of the potential for development of corrosion. The headed bars are also expensive and not readily available. The hooked bars were also found to provide adequate strength and ductility for both positive and negative moment. The main advantages of the hooked bar detail are cost and uniform cover over the reinforcement. High-Performing Sliding Surfaces for Bridge Bearings This section summarizes the research topic that evaluated the use of high-performing sliding surfaces for bridge bearings that would increase service life through greater wear resis- tance. A complete report of this study is given in Appendix D. At the time of writing, results of this research have been sub- mitted for publication (Ala et al. 2013a, 2013b). Problem Statement Selecting the optimum bearing type depends on the load, movement requirements, and economics. Steel-reinforced elastomeric (SRE) bearings have shown very good performance over the past 40 years due to their low cost and relatively long service life. As a result, SRE bearings are used extensively. However, with certain combinations of load and movement, the capacity of the SRE pad to accommodate the required translation through shear deformation can be exceeded. In these cases, additional movement capacity must be provided by the use of sliding surfaces. All other types of bearings— including cotton duck pads and high-load multirotational pot, disc, and spherical bearings—use sliding surfaces to accom- modate expansion requirements. Currently, polytetrafluorethylene (PTFE) is the material used for sliding surfaces in the United States. PTFE has low frictional characteristics, chemical inertness, and resistance to weathering and water absorption that make it an attractive material for bridge-bearing applications. However, plain PTFE wears under certain service conditions, particularly when subjected to combinations of high contact pressure, high rates of movement, and low temperatures. Fast sliding speeds, especially those associated with traffic movements, have been shown to be much more critical for PTFE wear than slow movements due to temperature. Therefore, wear of the PTFE sliding surface is one of the critical factors affecting the service life of these types of bearings. For improved service life, there is a need to consider other higher-performing sliding materials. Minimal data exist to develop a life prediction model for sliding surfaces. Designers need to have the capability of pre- dicting the expected service life of sliding surfaces for main- tenance and replacement purposes. Table 3.16. Ultimate Strength of the Specimens Specimen Positive Bending Negative Bending Mu Mn Mu/Mn Mu Mn Mu/Mn Straight 705 478 1.47 398 274 1.45 Headed 705 539 1.31 338 300 1.13 Hooked 642 478 1.34 350 274 1.28 Note: Ultimate strength = kips-in.

101 Objectives The main objectives of this study were to determine (1) the feasibility of achieving increased service life of sliding bear- ings that use PTFE through the use of alternative high- performing sliding materials that have greater wear resistance and (2) whether a life prediction model was feasible for PTFE and other higher-performance sliding surfaces. Scope of Work The scope of this research included analytical studies followed by the design, fabrication, and implementation of a limited experimental program to evaluate sliding surfaces used for bridge bearings. The experimental program tested small-scale specimens aimed at comparing the life of sliding materials commonly used in bridge structures against certain alterna- tive high-performance sliding materials. Performance charac- teristics such as coefficient of friction (COF), rate of wear, and total wear over a specified number of movement cycles were measured and compared. The study was developed to compare performance and wear of alternative high-performance sliding surfaces with conven- tional PTFE sliding surfaces over a range of bearing pressures, cycles of movement, and total cumulative travel distances. Discussions were held with various industry representa- tives to identify potential materials for consideration. Materi- als were selected for testing that would be available for bearing manufacturers to use in normal practice. The following three sliding materials were tested and com- pared in this study: 1. PTFE (plain, unreinforced) is the sliding material most commonly used in current bridge practice. PTFE was tested in dry and lubricated states and was used as a base- line to compare other materials. 2. MSM (Maurer sliding material) is an example of an inno- vative sliding material developed as a higher-performing substitute for current PTFE-based material. It is an ultrahigh-molecular-weight polyethylene developed by Maurer Söhne in Germany for use in high-speed rail bridges that experience fast rates of movement at expansion bear- ings. Recent field applications and experimental testing have shown very satisfactory results in Europe. Because these promising materials are new, no long-term data are available. However, the qualitative service life could be eval- uated by conducting cyclic wear testing. MSM was tested in dry and lubricated states for comparison with PTFE. 3. Fluorogold is an example of a reinforced PTFE-based material that was developed by Seismic Energy Products as a higher-performance material for sliding bridge bearings. It is an engineered product composed of virgin PTFE and special glass fiber reinforcing agents that improve wear. Again, the qualitative service life could be evaluated by conducting cyclic wear testing. Fluorogold was tested in only the dry state. The study also evaluated the feasibility of developing life prediction models for sliding materials based on the results of this testing and previous research. Analytical studies were performed to evaluate the magni- tude of horizontal movement due to girder–end rotation at expansion bearings under cyclic truck load. These studies would assist in determining realistic movement testing speeds in which to evaluate the various sliding surface materials. Bearing movement due to truck load is low-amplitude, high- cycle movement with fast movement speed, but movement due to temperature load is high-amplitude, low-cycle move- ment with low movement speed. Another objective was to evaluate the feasibility of devel- oping life prediction models for sliding materials that could be used for service life design. Design provisions were further developed for calculating the required thickness of the sliding surface based on the available literature, theoretical studies, and the results of the experimental program. Experimental Program A proof of concept test program was conducted. Sliding material specimen sizes (3-in. diameter), initial testing speed (25 in./min), and initial contact pressure (3,000 psi) were established to be consistent with tests performed and reported in NCHRP Report 432 (Stanton et al. 1999). The overall program was conducted in four major phases: 1. Phase 1 developed an analysis of a prototype bridge that computed and evaluated bridge movements and move- ment speed at expansion bearings due to truck loads. 2. Phase 2 developed the concept of an experimental pro- gram that was aimed at constructing a system capable of performing wear tests on multiple sliding materials to simulate various travel speeds and contact pressures. Test specimens were also designed as part of this phase. 3. Phase 3 constructed the test apparatus. 4. Phase 4 conducted the experimental testing. A 3-in.-diameter specimen size was selected for all tests in this experimental program to be consistent with previous PTFE testing performed in Europe (Campbell and Kong 1987) and with testing reported in NCHRP Report 432 (Stanton et al. 1999). Previous MSM studies were performed on samples with diameters of 3.0 and 6.1 in., with most tests performed on 3.0-in.-diameter samples. Previous studies have also shown that the size of the sliding surface specimen has minimal effect on its behavior, including the COF.

102 The test setup for the sliding surface wear tests used an MTS cyclic actuator, which was vertically installed in a large steel frame. A specially designed sliding material test fixture was installed below and connected to the actuator. The test fixture included a center moving plate (attached to the MTS actuator) with stainless steel surfaces attached to each side. This moving plate was sandwiched between two station- ary material test specimens mounted on steel backing plates. Two horizontal hydraulic jacks applied horizontal pressure, perpendicular to the surface of the sliding specimens, to create the required contact pressure. In all tests, the cyclic displacement was applied on a sine wave with a stroke length of 1 in. In other words, the center plate with stainless steel surfaces was moved upward from its initial central position 1 in., it was then moved back down for 2 in. (past the initial central position by 1 in.), and then back up 1 in. to the original position. The total movement per com- plete cycle was 4 in. Plain PTFE specimens were tested first to establish a base- line for comparison of MSM and Fluorogold results. This was followed by MSM (dimpled), Fluorogold (plain), and dim- pled PTFE. MSM specimens and the dimpled set of PTFE specimens were first tested for a prescribed number of cycles in a lubricated condition. Testing was then stopped, the speci- mens cleaned, and the testing resumed in the unlubricated condition. All tests were conducted at room temperature. Sliding speed was chosen to simulate the fast sliding speed associated with truck passage, which is low amplitude and high frequency, and results in a very large accumulative move- ment. However, the sliding of a bearing due to daily thermal movements is high amplitude, low frequency, and has consid- erably less accumulated movement. An initial sliding speed of 25 in./min was chosen for this study to be consistent with the maximum sliding speed for PTFE wear tests reported by Stanton et al. (1999) in NCHRP Report 432. The cycle frequency was initially set to 0.1 Hz (25 in./min) for all tests and was maintained for initial PTFE testing until the entire thickness was worn down. However, because of time limitations, this frequency was increased to 0.2 Hz (50 in./min) after about 116,000 cycles for the MSM and Fluorogold specimens. An initial contact pressure of 3,000 psi was chosen to be compatible with previous wear rate tests reported in NCHRP Report 432 (Stanton et al. 1999). Results Performance TesTs This research confirmed that high-performance sliding materials could be used effectively to increase the service life of bearings that use sliding surfaces by providing substantially improved wear resistance over long-term accumulated move- ment, particularly when subjected to high movement speeds and high contact pressures. However, these materials also exhibited higher COFs than plain PTFE at the contact pressures tested, which needs to be considered in the overall selection of a sliding surface. Figure 3.77 shows a plot of thickness reduction versus accu- mulated travel distance for PTFE, MSM, and Fluorogold for various combinations of contact pressure and sliding speed. Figure 3.77. Thickness reduction for sliding surfaces versus travel distance and numbers of cycles.

103 “PV” in the figure refers to the PV factor, a useful parameter for examining the wear rate that is obtained by multiplying pressure and velocity. The following observations were made concerning the per- formance of the MSM, Fluorogold, and PTFE samples: 1. Tests confirmed that both MSM and Fluorogold can pro- vide considerably greater wear resistance than conventional plain PTFE and can be used to increase service life when sliding surfaces are subject to high movement speed and high contact pressure. 2. Plain PTFE was shown to wear at a very high consistent rate under the initial combination of high sliding speed and contact pressure, resulting in 80% thickness loss within less than 2 mi of accumulated sliding length. 3. Fluorogold exhibited the best wear resistance of all materials tested, with no material thickness loss over about 8 mi of accumulated sliding length within the initial high values of pressure and velocity, but it started to show some wear at a much higher travel speed (50 in./min). 4. Plain PTFE and Fluorogold (a PTFE-based material) both exhibited rather constant rates of wear that varied for dif- ferent combinations of pressure and velocity. The wear rate for Fluorogold was very close to zero in the initial PV zone, but exhibited a constant wear rate (albeit very low) in higher PV zones. Plain PTFE exhibited consistent wear behavior, and showed very low wear in a relatively low PV zone (about a third of the initial PV). This indicated that a correlation may exist between wear rate and PV factor for PTFE-based materials. 5. MSM exhibited some initial thickness reduction of about 10% (described in item No. 6), but showed only about 10% additional loss over 13 mi (20.9 km) of accumulated travel. 6. MSM exhibited different behavior with respect to thickness reduction than plain PTFE or Fluorogold. MSM samples did not exhibit constant wear rates within PV zones; instead, they appeared to exhibit initial rapid thickness reduction at the beginning of each zone, after which the rates tended to slow down or stabilize as the tests proceeded. The initial thickness reduction was attributed to compressive defor- mation and creep and was not necessarily due to wear. This behavior made determining consistent wear rates for MSM difficult. Because of inconsistent wear rates, it was con- cluded that the PV factor may not be a suitable wear char- acteristic for MSM, as it is for PTFE-based materials. 7. The results of the SHRP 2 study, combined with the results reported in NCHRP Report 432, indicate the potential of service life prediction based on the PV factor and further confirm the potential of the PV limit transition between zones of low wear and severe wear. It was concluded that the PV factor could be used in a service life design method for PTFE-based sliding materials based on wear over an accumulated length of travel. 8. Fluorogold exhibited a COF about 2% higher than plain PTFE at the initial contact pressure of 3,000 psi. Unlubri- cated MSM, however, had a COF that was considerably higher. The COF of MSM was found to reduce consider- ably with increased contact pressure. At 5,000-psi contact pressure, MSM COF was comparable to Fluorogold COF at 3,000 psi. Lubricated MSM had a relatively low COF that was comparable with lubricated PTFE. Based on the limited proof of concept testing, Fluorogold (a glass-reinforced PTFE sliding material) exhibited the best overall high performance with a combination of high wear resistance and relatively low unlubricated COF. liFe preDiction results The results of this study, combined with the findings in NCHRP Report 432, indicated the potential of service life prediction based on the PV factor for PTFE-based sliding materials and further confirmed the potential of the PV limit transition between zones of low wear and severe wear. It was concluded that the PV factor could be used in a service life design method based on wear over an accumulated length of travel. Figure 3.78 shows the wear rate versus PV data for plain PTFE tests performed in this study combined with data from Low Wear Regime NCHRP Report 432 Tests SHRP 2 Tests Dimpled, unlubricated SHRP 2 Tests Plain Figure 3.78. Variation of wear rate with PV for plain PTFE.

104 the NCHRP Report 432 tests. From this plot, there appears to be a PV limit beyond which the wear rate increases signifi- cantly. It also appears that a service life design curve for wear rate is possible, which would be a function of PV as shown in the figure. It is recognized that the results shown are from very limited test data, and further testing would be required to establish more reliable curves and a true PV limit. Similar to the plain PTFE test results, the Fluorogold (glass-reinforced PTFE) test results also indicated the poten- tial of using the PV factor as a means of predicting service life. Like plain PTFE, the Fluorogold had a PV limit differentiat- ing zones of higher wear and very low wear, albeit the wear rates are significantly lower than those of plain PTFE. As with plain PTFE, further testing is needed to establish a more reli- able curve and to establish the real PV limit. MSM exhibited different behavior with respect to thick- ness reduction than plain PTFE or Fluorogold. MSM samples did not exhibit constant wear rates within PV zones, and it was concluded that the PV factor may not be a suitable wear characteristic for MSM as it is with PTFE-based materials. sliDing surFace Design proceDure For service liFe A proposed procedure was developed for design of sliding surfaces for service life. The procedure first determines demand requirements, which are based on bridge loads, traf- fic, and temperature data. Supply requirements are then determined based on selected material properties for the pro- posed sliding surface. The method uses a PV deterioration model based on test results as described above and illustrated in Figure 3.78. For a sliding surface material being considered, the service life is determined for a trial thickness using a PV model; this service life is compared with the design service life of the bridge system. If the service life of the sliding surface is less than the design service life of the bridge system, there are three options: 1. Consider required replacement schedule. 2. Consider increased thickness of the sliding surface material, or increased area of sliding surface, which reduces contact pressure and reduces wear rate. 3. Consider a higher-performing sliding material with greater wear resistance. Depending on the options chosen, various demand and supply steps will be repeated with new parameter values to arrive at a final solution. Conclusions and Recommendations This research program, as part of SHRP 2 Project R19A, con- firmed the feasibility of achieving increased service life of sliding bearings that are subject to high sliding speeds through the use of alternative high-performing materials that have greater wear resistance than conventional plain PTFE. Only proof of concept testing was performed in this study. To establish more reliable wear rates for the materials sam- pled, additional testing with various contact pressures and velocity combinations is required. Such testing would allow the development of a more statistically reliable model of PV versus wear rate for various types of PTFE-based sliding materials for actual service life design. Test data developed in this study, combined with those reported in NCHRP Report 432, are adequate for confirming a trend, but they provide a limited basis for developing reliable design curves. The effects of temperature were not considered in this study, but were evaluated in the NCHRP Report 432 study. Further testing is required to properly evaluate temperature as part of the wear rate model for service life design along with pressure and velocity. Field testing is needed to determine actual bridge move- ments and movement speeds at sliding expansion bearings for different types of girders (steel and concrete) under truck load and thermal load. There are little current data to sub- stantiate these load responses, and the results of analytical studies need to be validated against actual conditions. Corrosion-Resistant Reinforced Concrete Structures This section summarizes the research topic that evaluated the feasibility of increasing the corrosion threshold of conven- tional steel by electrochemically treating concrete containing conventional steel (black bars). A complete report of this study is given in Appendix E. Problem Statement According to Wipf (2006), over the last decades, the princi- ple techniques for corrosion prevention in bridge decks have included increased concrete cover depth to minimize the intrusion of chlorides from salts or ocean spray to the level of reinforcement and the application of epoxy coating over the steel reinforcement to protect the steel from chlorides and corrosion. However, increasing concrete cover depth increases both dead load and construction costs and does not elimi- nate the occurrence of cracks, which facilitate the intrusion of chlorides. In addition, cracks appear wider at the surface with increased cover depth. Epoxy coatings limit the expo- sure of the steel to chlorides, oxygen, and moisture and add minimally to bridge construction costs. However, holes and breaks in the epoxy coating at cracked locations, in combina- tion with high chloride concentrations, can result in corro- sion of the steel reinforcement, which affects the overall performance of the bridge. Moreover, epoxy coatings in aging

105 bridge decks may become brittle and eventually delaminate from the steel reinforcement. Dense (low-permeability) concretes, corrosion inhibitors, and both nonmetallic and steel-alloy corrosion-resistant re- inforcement are among the most common techniques being considered as alternative measures for mitigating corrosion in reinforced concrete structures (Wipf 2006). This research topic evaluated the means of achieving corrosion resistance in reinforced concrete with conven- tional reinforcing steel. The chloride threshold of conventional steel reinforcement would be increased by pretreating the concrete electrochemically through the chloride extraction method. The chloride threshold is the minimum concentra- tion of chloride in the concrete immediately surrounding the steel to initiate corrosion. For comparison, stainless steel (316LN) and titanium bars were used, which are known to have high corrosion resistance. This research builds on work completed to date, including that sponsored by FHWA (Hartt et al. 2006, 2009). Objectives The main objective was to determine the corrosion resistance of electrochemically treated concrete with black bar compared with both the black bar and with corrosion-resistant material. Corrosion-resistant materials used for comparison included commercially available stainless steel rebar (316LN) and tita- nium bar. The corrosion resistance of treated and untreated concrete samples was compared with the performance of con- ventional black reinforcing steel, which would provide the low level of protection, and to corrosion-resistant reinforcement, which would provide the upper level of protection. Scope of Work To increase the threshold level (thus, the corrosion resistance) in reinforced concrete structures containing conventional reinforcing steel, electrochemical treatment was applied. Reinforced concrete specimens with different levels of pre- treatment were prepared and tested to determine their critical chloride (corrosion initiation) threshold. Reinforced con- crete structures that can tolerate a higher level of chlorides before corrosion initiates will provide longer service lives. The electrochemical treatment was applied at low and high levels of pretreatment to increase the threshold level of black bars (mild steel). Tasks carried out to achieve the objectives of this study were as follows: • Summarize the literature of different steel reinforcement. • Obtain material and equipment to be used during the testing stage. • Fabricate specimens containing the selected reinforcement (mild steel [black bar], stainless steel, or titanium). • Conduct the electrochemical treatment on the test speci- mens containing mild reinforcement. • Apply cycles of wetting and drying. The wet stage had chlo- ride solution ponded over the specimens. • Monitor the samples by measuring the macrocell current and half-cell potential. Analyze the results and provide new recommendations. • Prepare the final report and recommendations. Experimental Program Reinforced concrete specimens were prepared in the labora- tory and subjected to an accelerated testing regime under controlled conditions. The corrosion resistance of each com- bination within the test matrix was compared with each of the other combinations. All specimens were prepared and tested in triplicate to provide statistically significant results. The test matrix included conventional reinforcing steel (black bar), black bar with two levels of electrochemical treat- ment, stainless steel, and titanium bar. The specimen preparation and testing procedure consid- ered to test the different matrices followed ASTM G109, Stan- dard Test Method for Determining the Effects of Chemical Admixtures on the Corrosion of Embedded Steel Reinforce- ment in Concrete Exposed to Chloride Environments (ASTM G109 2007a). Five variables were evaluated by placing different bars in conventional concrete: • Black bar in untreated concrete (control); • Black bar in concrete subjected to low electrochemical treatment; • Black bar in concrete subjected to high electrochemical treatment; • Stainless steel (316LN) in untreated concrete (control); and • Titanium steel in untreated concrete (control). Three specimens of each variable were prepared for a total of 15 prepared specimens. The concrete samples were prepared in compliance with ASTM G109. The length of the wet–dry cycle was modified to optimize, and thereby accelerate, the corrosion of the reinforcement. Testing was conducted as follows: 1. Each sample was placed on two nonelectrically conduct- ing supports. The first reading was taken 1 month after the specimens were cast. 2. The specimens were ponded with a salt solution (approxi- mately 400 mL at a depth of 1.5 in.) for 4 days. A loose-fitting

106 plastic cover was used to minimize evaporation. After 4 days, the salt solution was vacuumed off, and the specimens were allowed to dry for 10 days. The cycle was repeated until the termination of the test program. 3. The voltage across a resistor connected between the top and bottom bars was measured with a voltmeter, and the corrosion potential was measured by a half-cell system with a silver–silver chloride electrode. Measurements were taken at three times during each cycle: before ponding with the salt solution, after removal of the solution, and in the middle of the dry cycle. 4. The test was to be terminated when the integrated macro- cell current was equal to or greater than 150 C, which is equivalent to a macrocell current of 10 µA, as specified by ASTM G109. Results The concrete had a compressive strength of 4,830 psi, which is an average of three cylinders. Due to time constraints, the testing was terminated after 26 cycles. Table 3.17 lists the labels given to each specimen. Figure 3.79 shows the half-cell potential readings for all specimens varying with time. For steel reinforcement, a potential value more positive than -200 mV indicates there is a greater than 90% probability that no corrosion is occurring. This condition was observed in all but three specimens (one each from the groups of no, low, and high treatment). For these three specimens, the potential value was between -200 and -300 mV, which indicates that corrosion activity is uncertain. Table 3.17. Sample Summary Label Type of Reinforcement Type of Treatment BB_1 Black bar None BB_2 Black bar None BB_3 Black bar None ECL_1 Black bar Low level of electrochemical treatment ECL_2 Black bar Low level of electrochemical treatment ECL_3 Black bar Low level of electrochemical treatment ECH_1 Black bar High level of electrochemical treatment ECH_2 Black bar High level of electrochemical treatment ECH_3 Black bar High level of electrochemical treatment SS_1 Stainless steel 316LN None SS_2 Stainless steel 316LN None SS_3 Stainless steel 316LN None Ti_1 Titanium None Ti_2 Titanium None Ti_3 Titanium None Figure 3.79. Half-cell potential for all 15 specimens. 350- - - - - - 300 250 200 150 100 -50 0 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 Co rr os io n Po te n ti al (m V) Time (weeks) Half-cell potential BB_1 BB_2 BB_3 ECL_1 ECL_2 ECL_3 ECH_1 ECH_2 ECH_3 SS_1 SS_2 SS_3 Ti_1 Ti_2 Ti_3

107 The integrated macrocell charge is calculated using Equa- tion 3.6: 2 (3.6)1 1 1TC TC t t i ij j j j j j[ ]( ) ( )= + − × +− − − where TC = total corrosion (C); tj = time (s) at which measurement of the macrocell is carried out; and ij = macrocell current (amperes) at time tj. The total integrated charge was calculated using the current measurements and Equation 3.6. If the charge was greater than 150 C, then the test was terminated (ASTM G109). The results, plotted in Figure 3.80, show the variation of total charge with time. All values are below 150 C. The macrocell data indicate that one specimen each from the three specimen groups containing black bars (plain, low treatment level, and high treatment level) showed an increase in current. This is consistent with the half-cell potential data, which indicated uncertain corrosion activity in the same three specimens. These three specimens were selected to remove the rebar for visual inspection along with two control specimens (one each containing stainless steel and titanium): BB_3, ECL_1, ECH_3, SS_1, and Ti_1. BB_3 showed approximately a 0.75-in. length of corrosion product; ECL_1 showed approximately a 1-in. length of corrosion product; and ECH_3 showed approximately 0.25-in. length of corrosion product. SS_1 and Ti_1 did not show any signs of corrosion. Conclusions and Recommendations During the time period available for this project, one of the black bar specimens and one of the electrochemically treated black bar specimens from each of the low and high treatment levels showed an increase in current or potential values indic- ative of uncertain corrosion activity. The top bars of these specimens and two additional specimens, one each containing stainless steel and titanium reinforcements, were removed for visual observation. The stainless steel and titanium bars did not exhibit corro- sion within the available time period. To discern differences between black bars and electro- chemically treated black bars, a longer test period is needed. Initial observations indicate that electrochemically treated black bars may not provide the protection expected of stain- less steel or titanium; however, whether they provide benefits over the black bars without treatment cannot be concluded from this study due to time constraints. To draw firm con- clusions, additional research and extended testing periods are needed. New Galvanic Systems to Achieve Long-Term Corrosion Protection The study evaluated different promising concepts associated with new galvanic systems to delay the onset of corrosion and reduce the rate of corrosion of the reinforcing steel. A com- plete report of this study is given in Appendix F. Figure 3.80. Integrated macrocell charge for all 15 samples. 0 10 20 30 40 50 60 70 80 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 Coulombs (C ) Time (weeks) Integrated Charge BB_1 BB_2 BB_3 ECL_1 ECL_2 ECL_3 ECH_1 ECH_2 ECH_3 SS_1 SS_2 SS_3 Ti_1 Ti_2 Ti_3

108 Problem Statement Cathodic protection has been used in marine and underground structures, storage tanks, and pipelines to protect steel from corrosion (Virmani and Clemeña 1998). Cathodic protection was not used in steel reinforced concrete structures until 1973, when it was applied to a bridge deck (Stratfall 1974). There are two main types of cathodic protection: impressed current cathodic protection and sacrificial galvanic protection. Galvanic systems can delay the onset of corrosion and reduce the rate of corrosion of the reinforcing steel in reinforced structures. Further, the degree of protection achieved and the extension in the service life of the reinforced concrete bridge elements can be extended by improving the performance of galvanic corrosion-protection systems. Concrete patch repairs are very common in reinforced concrete structures. Often these concrete structures are chlo- ride contaminated beyond the location of the patch repairs, leading to accelerated corrosion around the patch. This phe- nomenon is referred to as ring anode corrosion or the halo effect. Ring anode corrosion can be eliminated by including galvanic anodes within the repair. The galvanic anode cor- rodes instead of the reinforcing steel. This type of application now has a 10-year successful history (Sergi et al. 2008). The knowledge gained from this application has led to the devel- opment of higher-output galvanic anodes. Initial testing of such anodes has shown improved performance and an ability to globally protect the reinforcing steel. Historically, galvanic anodes provided a level of current output per unit that was sometimes too low to achieve the desired level of protection. The development of improved anode units that can produce a higher level of current or increasing the driving voltage between the anode and steel reinforcement would improve the performance and allow galvanic anodes to meet the full range of corrosion protection levels that may be desired. The advantages of these types of galvanic systems over impressed current cathodic protection are the self-regulating current output of the system and the much-reduced require- ment of monitoring and maintenance. Recent initial work has enabled increased performance of galvanic anodes by modification of the surface area of the metal by design and by an increase in the driving voltage of the anode unit. It is also possible to increase the current density output by using high-voltage anodes. Objective The objective was to evaluate different promising concepts associated with the new galvanic systems in order to delay the onset of corrosion and to reduce the rate of corrosion of the reinforcing steel. The galvanic systems evaluated had an ordi- nary anode and one with a larger anode surface (four times the amount of zinc than ordinary anodes) and two levels of high-voltage anode. Scope of Work Small-scale laboratory testing was conducted to evaluate the potential of using galvanic anodes to extend the service life of reinforced-concrete structures and the level of protection pro- vided by different anode systems (higher metal surface, higher voltage output). Data collection included the measurement of the current, current density, potentials, and polarization versus time. The following tasks were carried out to achieve the objec- tives of this topic: • Summarize the literature of cathodic protection, including the galvanic systems. • Obtain material and equipment to be used during the test- ing stage. • Fabricate samples using different galvanic systems. • Set up an automated monitoring system to collect the long- term data. • Analyze the collected data and destructively analyze the anodes and steel bars. • Prepare the final report and provide recommendations based on the results. Appendix F provides more detailed information about each task. Experimental Program The ability of embedded galvanic anodes to provide sufficient output to protect cathodically the reinforcing steel from further corrosion was determined using small concrete slab samples. The levels of protection of different anodes with varying output were studied. The specimens measured 18 × 18 in. and were 8 in. thick to simulate a portion of a bridge deck. The test matrix included conventional reinforcing steel and galvanic anodes with dif- ferent surface areas or voltage outputs. The concrete slabs were cast in two layers. The bottom layer had uncontaminated concrete, and the top layer had concrete contaminated with salt. Five variables were evaluated: • Black bar embedded in concrete (without anode, control); • Black bar embedded in concrete with ordinary anode (OA); • Black bar embedded in concrete, with a larger anode having a surface area four times the ordinary (OA4); • Black bar embedded in concrete with high-voltage anode (under development) at Level 1 (provides higher applied voltage than a standard galvanic anode) (HVAL); and

109 • Black bar embedded in concrete with high-voltage anode (under development) at Level 2 (higher voltage than Level 1) (HVAH). Three specimens for each variable were prepared. The test- ing procedures for the determination of the corrosion activity involved the following: 1. Measuring half-cell potentials (ASTM C876 2009) at nine locations on the specimen. 2. Measuring the current flowing from the anode to each pair of rebars (measured hourly), including a total of four pairs, two on top and two at the bottom of the specimen, while at the same time measuring the corrosion potential by using embedded electrodes. 3. Measuring the current flowing during depolarization (when the anode was disconnected). The corrosion potential was measured by both the embedded electrodes and the half-cell system. A second data logger with quick-reading capability (10 reads/s) was used during the procedure to capture the exact moment when the anode was turned off and on. The following steps were performed: 1. Seven days after casting, before connecting the reinforce- ment to the anode for the first time, the corrosion potential was measured to establish the baseline level of corrosion potential of the steel using a half-cell system with a silver– silver chloride electrode. The tip of the reference electrode was positioned on a small wet pad to stabilize the readings. The same procedure was repeated three times during each wet–dry cycle. The corrosion potential was measured before ponding with the water, after removal of the water, and at the middle of the dry cycle. 2. Immediately after the first potential reading, the first anode was connected to the steel. The current delivered by the anode was measured, and then the anode immediately disconnected. The second anode was then connected to the steel and the current was measured. If both anodes were active (i.e., produced currents on the order of milliamperes), one of them was chosen and permanently connected to the steel. The other anode remained redundant until the end of the experiment. If one anode appeared faulty (e.g., the electrical connection was problematic), the spare anode would be used instead. 3. The specimens were ponded with water (no salt) for 4 days. Approximately 2 L of water (around 10-mm height) were added to the reservoir. At the end of the wet period, water was removed by vacuum. 4. A data logger was used to take readings every hour. The automated acquisition system recorded the total current measurements. 5. For depolarization measurements, approximately every 10 weeks a map of the potential across the surface of the specimen was obtained while the anode and steel bars were connected. The anode was then disconnected (using the switch). 6. Four hours later, a new potential map was obtained. The difference between the instant-off potential and the 4-hour depolarization potential was the value of interest; its use is described in Appendix F. 7. Twenty-four hours after disconnection, another potential map was determined. The difference between the instant- off potential and this 24-hour depolarization potential was the value of interest. After the conclusion of the depo- larization test, the anode was reconnected to all steel bars. The anode was kept connected to the reinforcement until the next depolarization test. Results The following data were collected from each specimen: the half- cell potential system; the current flowing from the anodes to each pair of rebars and the corrosion potential at the location of two embedded electrodes in the each specimen; and the current flowing during the instant-off and -on procedure from the anodes to each pair of rebars and the corrosion potential at the location of two embedded electrodes. The first half-cell potential measurement was taken on Day 0. After establishing the base line readings, all samples were subjected to cycles of 4 days’ wet and 10 days’ dry periods. Aside from the half-cell readings taken at the end of each dry cycle, two extra sets of readings were conducted to better evaluate and observe the specimens during the wet–dry cycles. Each specimen was identified using the labels given in Table 3.18. Five variables were tested, and three specimens for each variable were made and tested (15 distinct specimens) to provide statistical significance. Two anodes were installed in each specimen, but only one was used in the tests. The output voltage of each anode was measured separately. The anode with the highest potential reading measured was selected and connected permanently to the rebars throughout the experiment. Figure 3.81 shows the half-cell potential readings for all specimens. This plot represents the average values of corrosion potential measured at nine distinct points on each slab. The measurement was conducted three times during each wet–dry cycle. Initially, all specimens had similar corrosion-potential values. The control specimens (BB), with no anode, had the lowest corrosion-potential values, which remained steady throughout the experiment. All specimens with anodes had their corrosion potential shift in the negative direction after the anode was connected to the rebars. The OA, OA4, and HVAL specimens showed steady potential readings. Two HVAH specimens produced variable results, which may be due to variation in production of the prototypes. The highest negative corrosion potentials were provided by the

110 HVAH specimens, followed by the HVAL and OA4 specimens, then by OA, and finally by the BB specimens. Higher negative- corrosion potentials indicate that more corrosion protection is being provided. The electrical current flowing from the anode to the four pairs of bars over a period of 56 weeks was measured. In addition, corrosion potential data were collected at the two locations of the embedded reference electrodes. A very good agreement was found between those of embedded values and the surface-potential values. The corrosion rate was estimated through a depolarization testing procedure. Four depolarization measurements were conducted at approximately 10-week intervals. The tests con- sisted of collecting a series of readings during the temporary disconnection of the anodes, at 4 and 24 h after disconnection, and finally during the anode reconnection (turn on). The procedure started with the initial reading taken before the anodes were disconnected; an intermediate reading was conducted 4 h after the anodes were disconnected (4 h off). Finally, 24 h after disconnection and before reconnection (turn on) of the anodes, a final reading was taken (24 h off). The data collected during this procedure were used to estimate the corrosion rate using the Butler–Volmer equa- tion. During the fourth depolarization period of 24 h, it was observed that the corrosion potential of the control specimens did not change; however, all the specimens with anodes showed a large shift in potential after the disconnection of the anodes. The largest change was observed for the HVAH specimens, followed by the HVAL and OA4 specimens, and finally by the OA specimens. The potential of the reinforcing steel 4 h after disconnection averaged -112 mV for all specimens. The corrosion potential of the reinforcing steel 24 h after dis- connection averaged -66 mV. Figure 3.82 shows normalized values of the corrosion potential during the fourth depolarization period. Table 3.19 shows the corrosion rate estimation. Even though the concrete used in the specimens had a high level of salt per unit volume of concrete, the results indicated no corrosion activity. Table 3.18. Sample Labels Label Type of Anode BB_1 None BB_2 None BB_3 None OA_1 Ordinary anode OA_2 Ordinary anode OA_3 Ordinary anode OA4_1 Anode with four times the surface area of the ordinary anode OA4_2 Anode with four times the surface area of the ordinary anode OA4_3 Anode with four times the surface area of the ordinary anode HVAL_1 High-voltage anode—low level HVAL_2 High-voltage anode—low level HVAL_3 High-voltage anode—low level HVAH_1 High-voltage anode—high level HVAH_2 High-voltage anode—high level HVAH_3 High-voltage anode—high level 800 700 600 500 400 300 200 100 0 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 Co rr os io n Po te nti al (m V) Time (weeks) Average values of Half cell potential BB_1 BB_2 BB_3 OA_1 OA_2 OA_3 OA4_1 OA4_2 OA4_3 HVAL_1 HVAL_2 HVAL_3 HVAH_1 HVAH_2 HVAH_3 Figure 3.81. Half-cell potential for all 15 samples.

111 100.00 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 0 4 8 12 16 20 24 Co rr os io n Po te nt ia l( m V) Time (hours) Normalized values of Half cell potential BB_1 BB_2 BB_3 XP_1 XP_2 XP_3 XP4_1 XP4_2 XP4_3 HVAL_1 HVAL_2 HVAL_3 HVAH_1 HVAH_2 HVAH_3 Figure 3.82. Normalized values of half-cell potential during the fourth depolarization. Table 3.19. Corrosion Rate Estimation Label iappl (mA/m2) DE (mV) bc (mV) ba (mV) icorr (mA/m2) Corrosion Rate (A/cm2) Rating OA_1 331.32 277.00 120 60 1.64 0.16 Negligible OA_2 161.54 227.00 120 60 2.08 0.21 Negligible OA_3 293.71 300.00 120 60 0.93 0.09 Negligible OA4_1 728.08 392.00 120 60 0.40 0.04 Negligible OA4_2 528.61 371.89 120 60 0.42 0.04 Negligible OA4_3 512.58 373.33 120 60 0.40 0.04 Negligible HVAL_1 526.21 384.56 120 60 0.33 0.03 Negligible HVAL_2 450.59 348.00 120 60 0.57 0.06 Negligible HVAL_3 480.73 364.11 120 60 0.45 0.04 Negligible HVAH_1 1201.89 540.22 120 60 0.04 0.00 Negligible HVAH_2 186.49 255.78 120 60 1.39 0.14 Negligible HVAH_3 2597.31 594.33 120 60 0.03 0.00 Negligible Note: icorr = corrosion rate; iappl = applied electrical current; DE = observed corrosion potential; bc = cathodic Tafel’s slope; and ba = anodic Tafel’s slope.

112 Conclusions and Recommendations The test results indicated that there was no corrosion in any of the specimens in the given time period. The testing further indicated that specimens with high-level, high-voltage anodes (HVAH) provided increased corrosion protection, which is indicated by higher current and the generation of more nega- tive potential values than the low-level, high-voltage anodes (HVAL). Anodes having four times the surface area (OA4) pro- vided more current and corrosion protection than OA anodes. HVAL anodes exhibited similar corrosion protection as the OA4 anodes. Both high-voltage anodes and OA4 anodes pro- vided better corrosion protection than OA anodes. Due to time constraints, the tests were terminated without observing corrosion in some of the specimens. Further research with an extended time frame is recommended. Self-Stressing Deck Systems This section summarizes the research topic that evaluated a system for imparting longitudinal prestress force over the interior support of a continuous structure using imposed sup- port settlement. Additional information may be found in the doctoral dissertation by Marcelo Ferreira da Silva (da Silva 2011). At the time of writing, papers detailing the results of the research have been submitted for publication (Silva et al. 2013; Yakel and Azizinamini 2013). Problem Statement Bridges often use continuity over the interior supports to reduce force effects within the spans. As a result, negative bending moment is produced over the interior supports. In structures built with composite construction, this moment produces ten- sile stresses in the concrete deck and compressive stress in the bottom flanges of the steel girders. The tensile stress in the deck often leads to cracking, which allows intrusion of moisture and road salt, causing corrosion of the reinforcement and support- ing girders. Continued maintenance is required to forestall the deterioration; however, eventual replacement of the deck is typically required. One accepted method of countering the negative effects due to tensile deck stress is to reduce or eliminate the tensile stress through precompression of the deck. The system being presented uses the self-weight of the structure to induce a com- pressive force in the deck. The compressive force is developed by first raising the interior supports above their final elevation while the deck is cast (CIP construction) or the panels are placed (precast construction). Once the concrete has cured, the supports are lowered to their final elevation. Continuity of the steel member and composite action with the deck produce a compressive stress in the concrete slab that is balanced by tensile stresses in the bottom of the steel member. As a result of the compression in the slab, cracking over the interior support is reduced, which then increases durability. In addition, the magnitude of the compression force in the bot- tom flange is reduced, which can lead to section size reduction and possibly splice elimination, making the overall bridge design more efficient and cheaper than conventional design. The characteristics described apply to any system that pre- compresses the deck over the interior support, such as longitu- dinal posttensioning using steel strands. However, one of the biggest advantages of the self-stressing system is that precom- pression of the deck is accomplished without the use of costly and time-consuming posttensioning techniques, thus reducing the cost and time for bridge construction. Because prestressing strands are not used, potential corrosion of strands and associ- ated loss of prestressing force are eliminated. Although the concept of using the continuity of a bridge to prestress the bridge deck is simple, the technique has never been extensively researched or used in practice. Laboratory testing on a CIP self-stressing system is reported by Yakel et al. (2007). The research reported here is largely a continuation of that effort. Nagai et al. (2000) report on a field application similar to the self-stressing method, which they refer to as the piano method. Both of these sources emphasize the need for time-dependent analysis to predict the creep and shrinkage effect in the induced compressive force. A Highways for LIFE program bridge constructed by the Oregon DOT effectively used self-stressing, although it was used to address an uplift issue at the abutments and not to improve the performance of the deck (Ardani et al. 2010). Finally, the Inverset system uses a deck that is cast on the ground while a steel girder is held in place above. On righting the system, gravity induces a com- pression within the slab (Fort Miller Company, Inc. 1998). Objective The objective of the research reported in this section was to develop guidelines of the self-stressing method system for use on multispan continuous structures with composite CIP or precast bridge decks. Consequently, the research focused on the development of guidelines for use of the self-stressing deck system in practice. The self-stressing system can be a cheaper alternative to posttensioning for imparting a longitu- dinal compressive stress in the bridge deck and can minimize the size, if not completely eliminate the extent, of cracking of the concrete bridge deck. Although steel girders are used throughout the discussion, the concept could be adapted for concrete girders, as well. Scope of Work The work carried out under this project was an extension of the CIP experimentation that had been conducted previously. To this end, a small-scale test specimen similar to the original was

113 tested; the main difference was that the deck was composed of precast panels rather than CIP. Additionally, the system had two steel girders, rather than one, to provide stability while the precast panels were placed. Finally, the magnitude of the initial prestressing force imparted on the structure was greater than in the CIP testing. A design procedure and guide was developed to assist in the use of the system. A design based on the proposed proce- dure was then compared with a structure to be constructed using conventional methods. Experimental Program A small-scale test specimen was built and tested to investigate the parameters that affect the performance of the self-stressing system when used in conjunction with a precast deck. The test specimen was composed of two W12 × 22 steel girders placed over three supports 15 ft apart. Ten 3-ft long, full-depth (6-in.) concrete precast panels were placed on top of the girders. The panels were 5 ft wide, and the girders were spaced 3 ft apart, resulting in a 1-ft overhang to each side. Shear studs welded to the top of the girder projected through block-outs in the pre- cast deck panels. Figure 3.83 shows a 3-D sketch of the test specimen. The test specimen was built in the structural laboratory and generally represents a quarter-scale model of a prototype structure. The first step was to lift the intermediate support to a prede- termined height. Two hydraulic jacks, shown in Figure 3.84a, were placed under the central beam to lift the girder 1 in. Steel washers were used as shim plates (Figure 3.84b) to keep the support lifted while the deck was constructed. The completed precast panels to be used in the construction of the specimen are shown in Figure 3.85. As the precast panels were placed over the girder (Figure 3.86a), the panel-to-panel connections were accomplished using high-strength epoxy spread over the match-cast shear key (Figure 3.86b). Once all panels were set in place, the shear pocket block-outs and the center panel-to-panel connection were filled with nonshrink grout. After allowing the grout to harden, the shim plates were removed. Lowering the interior support created compressive force (self-stressing) in the specimen. The amount of prestressing achieved was directly propor- tional to the amount of displacement initially introduced during the construction. Figure 3.87 shows the compressive stress mea- sured at the top surface of the concrete deck compared with the predicted values. The maximum compressive stress located at the interior support was 2.30 ksi, or approximately 30% of the concrete compressive strength. Good agreement between the measured and calculated values was observed. Within the deck, the time-dependent analysis predicted a reduction in the precompression force of 30%, meaning the Figure 3.84. (a) Pancake jacks used to shim the system and (b) central support shimmed 1.0 in. (a) (b) Figure 3.83. 3-D view of self-stressing system specimen.

114 Figure 3.85. (a) Precast concrete deck panels organized in sequence and (b) match-cast detail. (a) (b) Figure 3.86. (a) Placing precast concrete panels and (b) high-strength epoxy spread over the match-cast shear key. (a) (b) maximum compressive stress over the support would be reduced from 2.3 ksi to 1.6 ksi. Recall that the target value of precompression to achieve the desired beneficial effects was 0.25 ksi, so the loss was acceptable and was expected. The changes in these values demonstrate the need to compensate for losses when determining the amount of precompression to apply. The “estimated” line in Figure 3.87 reduced stresses at the top surface of the precast deck obtained from the time- dependent analysis. Note that only the initial values of stress from testing are reported, as the long-term changes in stress resulting from creep cannot be directly measured. It was observed after 63 days that the monitored strains within the system were no longer changing, so the decision was made to stop the monitoring and begin the ultimate load test. The ultimate load test was conducted to determine what effect the self-stressing method, as implemented with precast panels, would have on the maximum capacity of the system. The test would also provide in-depth information and better understanding of the critical sections of the bridge, such as at midspan (maximum positive moment) and at the interior support (maximum negative moment). The ultimate loading was conducted to determine the failure mode that may govern the design of a self-stressing method. The loading was applied to the specimen by two spreader beams located at the midpoint of each span. The beams were connected to four hydraulic rams (two per beam)

115 through high-strength rods, as shown in Figure 3.88. Elasto- meric pads were place under the spreader beams to transfer the load to the deck. During the test, the data acquisition system was used to record all strains in both the concrete and steel, deflections, and hydraulic pressure in the rams (applied load). Figure 3.89 shows the ultimate load test setup in the structural laboratory. The analysis of the recorded data and observed mode of failure are discussed in the following section. Results Load RequiRed to induce cRacking To illustrate the benefit of the precompressed deck, Figure 3.90 shows an estimation of cracking moment. Figure 3.90, b and c, show the stress analysis for conventional construction with no deck precompression and for the self-stressing test speci- men, respectively. The required moment (force “P”) to crack the deck at the interior support was three times greater. In these calculations, a 40% precompression loss was assumed to account for the time-dependent effects. Thus, a final com- pressive stress of 1.38 ksi (2.3 ksi × 60%) was assumed. SeLf-StReSSing uLtimate StRength teSt ReSuLtS Loading was applied as small increments of displacement, pausing for data acquisition and inspection. Figure 3.91 shows the load–displacement curve obtained for the self-stressing test specimen. The initial response of the system was linear up to a level of applied displacement of 0.3 in., corresponding to a load of 110 kips; this point is labeled as 1 in Figure 3.91. 2.50 2.00 1.50 1.00 0.50 0.00 0.50 0 5 10 15 20 25 30 St re ss (k si) Length (ft) Stress at top surface of precast deck Calculated Measured Estimated Tension Compression Figure 3.87. Stress at top surface of precast deck panels after shim removal. Figure 3.88. Sketch of the ultimate load test setup (side view).

116 Figure 3.89. Ultimate load test setup. Figure 3.90. (a) Schematic of required moment (force P) to crack the deck and stress analysis between (b) conventional construction and (c) self-stressing construction. (a) 7.5 ft 7.5 ft 7.5 ft 7.5 ft P P (b) 3.00 2.50 2.00 1.50 1.00 0.50 0.00 0.50 1.00 0 3.75 7.5 11.25 15 18.75 22.5 26.25 30 St re ss (k si) Length (ft) Crack 0 kip 40 kip Tension Compression (c) 3.00 2.50 2.00 1.50 1.00 0.50 0.00 0.50 1.00 0 3.75 7.5 11.25 15 18.75 22.5 26.25 30 St re ss (k si) Length (ft) Crack 0 kip 40 kip 80 kip 120 kip Tension Compression Beyond this point, the load displacement began to display a nonlinear response. Strain data indicated that this change in response corresponded with the onset of yielding in the bottom flange. At an applied displacement level of 1.5 in., the load was removed to allow adjustment of a support when a sliding element was nearing the end of the available travel. The speci- men showed residual deformation (0.75 in.) after unloading. Although the girder had undergone some degree of plastic deformation, the slope of the load–displacement curve during the reloading stage (Box 2) was nearly equal to that at the start of the test. The specimen reached a maximum loading capacity of 230 kips (Box 3), which is over twice the load observed at the linear elastic limit of 110 kips. The bridge displayed good ductility at this maximum load value with a vertical displace- ment of 2.3 in., which is approximately seven times the deflec- tion at the elastic limit. The drop in load from the maximum value corresponded to development of localized crushing of the concrete near the west loading point. It is important to note that beyond this point, the specimen sustained a reduced level of load while subjected to a very large value of deflection (Box 4). The maximum predicted ultimate capacity based on strain compatibly method was 217 kips, which compares well with the observed experimental maximum load of 230 kips. cast-in-place ultimate strength test results For comparison, the results obtained from the CIP specimen are shown in Figure 3.92. The load–deflection response of the system was nearly linear until a deflection of approximately 0.25 in., which corresponded to an applied load of 55 kips. The unloading of the system seen in Figure 3.92 at about 60 kips was done to correct the loading frame. After correction, the loading was resumed. At 72 kips, yielding of the web occurred near the pier. This corresponded to a deflection of 1.1 in. at midspan. It can also be seen in Figure 3.92 that the specimen exhibited a sufficient amount of ductility. There were very few notable events observed before the ulti- mate load was reached. At a load of 65 kips the transverse crack pattern was first observed. The load ceased to increase

117 0 50 100 150 200 250 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Lo ad (k ip s) Displacement (in.) Load displacement curve 2 1 3 4 Figure 3.91. Load–displacement curve for ultimate load testing. Figure 3.92. Load–deflection curve during the ultimate load test. 0 10 20 30 40 50 60 70 80 90 100 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Lo ad (k ips ) Deflection (in.) E 2" Pot W 2" Pot E 10" Pot

118 after reaching 75 kips, at which point the vertical deflection at the middle of the east span was 1.65 in. The deflection continued to increase with no increase in load until crushing was observed at the top of the deck. Beyond this point, the specimen was able to absorb additional deflection, but at a reduced level of loading. The maximum deflection at the end of the test was approximately 3.25 in. at midspan. The observed ultimate load was slightly lower than the predicted failure load of 81 kips. This difference can likely be attributed to the fact that design calculations were made assuming the yield strength for the beam of 50 ksi. The material testing showed that the actual yield strength of the beam was slightly lower at 47.6 ksi. Failure moDes Although the bridge was constructed using an innovative method, the observed modes of failure were similar to those of a two-span continuous composite steel–concrete bridge con- structed using conventional details, as listed below: • Cracking in the vicinity of interior support; • Yielding of girder bottom flange (large deformation); • Combination of web and bottom flange buckling; and • Concrete crushing. Figure 3.93 shows the only tensile crack located at the closure region. Because steel reinforcement was not con- tained in this region, the crack width is much greater than if reinforcement were used to distribute the cracks. This is certainly a feature of the behavior that needs to be addressed. One solution would be to have two closure regions within the spans, such as at points of inflection, rather than a single clo- sure over the interior support. The panel positioning would then be adjusted such that there would be no joint directly over the support. Figure 3.94 shows the final deformed shape of the bridge after the ultimate load testing. It is clearly visible that the bridge had undergone large deformations due to yielding of the girder bottom flange. Conclusions and Recommendations The self-stressing method was used to construct a prototype bridge using the precast concrete panel system, and the proof of concept was successfully conducted and validated. The testing was deemed a success because the cracking was satisfactorily delayed. Because the bottom flange stress at the interior support region was completely eliminated, a single girder cross section could be used throughout the bridge length. The self-stressing method applied to precast concrete decks is proposed as an alternative to the conventional posttensioned concrete deck system used for preventing transverse deck cracking. If a CIP deck is considered, the self-stressing method can reduce or eliminate shrinkage cracks, which is often an issue even before the bridge is opened to traffic. Both analytical and numerical solutions have shown good agreement with the experimental results. Thus, both methods can be used to design a bridge using the self-stressing method. Time-dependent analysis should be carried out to determine the amount of precompression loss and additional stress induced in the girder due to the time-dependent effect. Guidelines were developed to facilitate the dissemination of the method, and a design example is provided to aid bridge engineers considering the self-stressing method. The design example illustrates advantages of the self-stressing method over Figure 3.93. Crack at the closure region due to tensile force over the interior support. Figure 3.94. Large deformation due to yielding of bottom flange at midspan.

119 the conventional method, such as the reduction of compressive stress at the girder bottom flange (mitigating buckling) and the development of compressive stress in the deck (reducing cracking). Delayed Composite Systems Problem Statement The majority of bridge decks for beam–slab bridges are constructed using field-cast concrete. Although accelerated construction and rapid renewal is one of the foci of SHRP 2, addressing the durability of this commonly used bridge deck construction method is extremely important. CIP bridge deck systems are known to exhibit transverse cracking before bridges are opened to traffic. This cracking is partly attributed to restraining forces that are provided by shear studs preventing fresh concrete from shrinking. Closure of these cracks through the introduction of compression from posttensioning should greatly enhance the durability of these bridge decks. This compression can be maximized by using the delayed composite system. One of the main features of the delayed composite system is that it isolates the shear studs over the steel beams and prevents development of the composite action until the bridge deck is precompressed through posttensioning. The proposed research builds on previous pilot studies of delayed composite action for precast full-depth deck systems for steel superstructures (Azizinamini and Yakel 2006) and concrete superstructures (Fallaha et al. 2004). The previously developed systems propose openings in the top of the deck that must be filled with a nonshrinking grout or concrete. This leaves a joint exposed to the deck surface that is not precompressed by the posttensioning. From a durability per- spective, this exposed joint could be a potential risk. The pro- posed research addresses increased durability through the minimization or elimination of surface joints in the deck. The proposed systems would provide delayed composite action by incorporating a continuous void with a completed CIP concrete deck over it with small grout tubes penetrating the deck surface. Objectives The main objective of this research was to determine the via- bility of the delayed composite system for CIP systems. CIP bridge deck concrete is subjected to early shrinkage and to shortening caused by a drop in temperature in the hydration cycle and cold weather. These two effects combine to create demand for deck concrete shortening. When concrete sets, it is anchored to the supporting girders through immediate composite action, with studs for steel girders and shear bars for concrete girders. The interaction restrains free shortening of the deck and creates tensile stresses beyond the tensile capacity of the young, weak concrete. This causes cracking in the first few days of deck life. Allowing the deck to slide relative to the girders with very little frictional effect would allow for shrinkage and temperature shortening to take place freely in the deck before it is “locked in” with the girders. It is important to have a system that will allow for low friction before composite action takes effect. Thus, this project focused on methods to achieve delayed composite action as effectively as possible. The fundamental concept is to provide open channels over the girders where shear connectors (studs or reinforcing bars) are located. These channels isolate the shear connectors while fresh concrete is placed and is allowed to hydrate, undergoing volume changes due to temperature drop in the hydration cycle and shrinkage. The next step in the process is to post- tension the deck. The amount of posttensioning needed to eliminate crack-inducing tension in the deck is considerably reduced by two effects: (1) the deck is free to slide relative to the girders through means of low friction bearing, and (2) the deck is posttensioned before development of the composite action; thus the entire posttensioning force is taken up by the deck, not the deck–girder system. Also, the much stiffer girders do not share in a prestressing force they do not need or that can possibly harm them. The final step is to develop the composite action through grouting of the channels. Scope of Work The scope of work for the research was proposed in four tasks as follows: 1. Research the components of the encapsulated channel concept to determine the best configuration to isolate the shear studs, minimize friction restraint between the girder top flange and the deck form supports, provide space for the posttensioning, and allow for complete grouting. Select the more promising components of the CIP alternative for further investigation in the remaining tasks. 2. Develop sufficient design and detailing for the encapsulated channel concept selected in Task 1 for further investigation. (The design concentrated on the design of the frictionless contact interface between the girder top flange and the deck form supports and on complete grouting of the channel.) 3. Perform laboratory testing of details developed in Tasks 1 and 2 using a small-scale laboratory test to prove the fea- sibility of the concept. (The dimensions and geometry of the small-scale laboratory test specimen were to be final- ized after conducting Tasks 1 and 2.) 4. Develop the scope of future research to develop detailing and design guidelines based on the results of Tasks 1, 2, and 3.

120 Results A bridge deck system was developed for delayed composite action over interior girders. The purpose of the system is to allow the concrete deck to move relative to the girder, which will prevent creep and shrinkage cracks shortly after the deck is poured. This is done by isolating the top flange of the steel girder with an enclosed apparatus that rests on a low friction bearing, allowing for movement relative to the steel girder. The channel is filled with grout after initial creep and shrinkage have occurred. Figure 3.95 shows the proposed system for interior beams. The support angles were made from bent 12-gauge steel sheet. The channel top plate, bent plate for bearing, and lateral support strips consisted of 16-gauge steel sheet. The top plate was bent and had holes drilled for reinforcement and grout and air holes for pouring of the concrete. Various plastics were researched as possible low-friction materials to be used between the supporting straps of the channel and the top flange of the girder. The results of the investigation identified the three most promising options as high-density polyethylene sheet, ultrahigh-molecular-weight sheet, and PTFE sheet. Test specimens were prepared as shown in Figure 3.96. During the preparation of the test specimens it was noticed that the bent angle plate at the base of the void would rotate and bind against the sides of the top flange, increasing the friction. Greater clearance between the bent angle plate and the side of the top flange is proposed to reduce this binding friction. The concept introduced here is applicable to interior gird- ers only. An alternate system for delayed composite action over exterior girders has not been developed and is currently considered not viable. Conclusions and Recommendations The delayed composite action system for CIP systems is determined not to be viable at this time. Further research is not recommended. Membranes for Bridge Decks As part of the research study, a manual devoted to water- proofing was developed. The objective was to include this manual as an appendix to the Guide. However, it was deter- mined that additional work was necessary before publishing the waterproofing manual. Following is a brief summary of the information in the manual. Problem Statement The major contributors to concrete deck deterioration are salt, water, and direct contact from traffic. One solution to these harmful factors is to protect the concrete surface by separating it from these elements with a waterproofing membrane. This solution has been addressed in both European and Canadian bridge codes. One of the most comprehensive European specifications is the guidelines on waterproofing membranes included in the handbook on bridge decks published by the Figure 3.95. Proposed delayed composite system for interior beams. Figure 3.96. Test specimen for interior girders.

121 Norwegian Public Roads Administration (2009). The use of membranes in Europe is an accepted practice, but only a limited use exists in the United States. European practice does not have the level of service life issues that exists in the United States. The major difference between U.S. and European practices is that the membranes used in Europe are of a much higher quality and have higher initial costs. Only a very limited amount of coordinated research and analysis into the waterproofing of bridge decks has been conducted in the United States. The recently published NCHRP Synthesis of Highway Practice 425: Waterproofing Membranes for Concrete Bridge Decks (Russell 2012) is an update of NCHRP Synthesis 220 (Manning 1995) with the same title as NCHRP Synthesis 425. The survey and literature review for NCHRP Synthesis 425 reported that most Canadian provinces and many European countries require the use of waterproofing membranes on all new bridge decks. In contrast, only 60% of U.S. state agencies reported the use of water- proofing membranes. Bridge decks are subject to traffic wear and chloride penetration due to winter deicing salt. Concrete decks will typically develop transverse and longitudinal cracks even before they are opened to traffic. These decks are horizontal elements and, as a result, salt-laden water can easily intrude to the level of the reinforcement. In many cases, it takes only a few months before the chloride threshold at the level of the reinforcement is exceeded at these crack locations, resulting in initiation of corrosion when unprotected reinforcement is used. Asphalt overlays alone can protect concrete decks against wear; however, they do not waterproof the deck. Use of a water- proofing system is, therefore, needed to protect the concrete bridge deck from intrusion of these harmful chlorides when asphalt overlays are used. There are several principle classes of waterproofing for bridge decks. The level of traffic (wear) and climate (chlorides) in which the bridge is located are among the two major factors dictating the type of waterproofing system that should be selected. The following systems can be considered: • Asphalt overlay directly applied to concrete deck is subject to leakage, and therefore provides resistance against wear from traffic only. • Low-viscosity epoxy, silane, or siloxane sealers can seal the cracks, but they provide a limited lifespan for protection against chloride ingress and corrosion. • Concrete-based overlays provide a layer of material over the bridge deck with low permeability that can provide resistance to chloride intrusion. However, these overlays are prone to cracking and delamination. Research studies are being carried out to examine the feasibility of using a very thin layer of ultrahigh-performance concrete as overlay (Harris et al. 2011). • Membrane systems combined with an asphalt overlay pro- vide resistance against wear and chloride intrusion. A good-quality membrane together with an asphalt overlay can be very effective in providing resistance to both wear and chloride intrusion. The life of the overlay system and mem- brane will be much less than 100 years; however, this system has many advantages: • The asphalt overlay can be milled and resurfaced as needed without replacement of the membrane, depending on the wear life of the asphalt. • The replacement or repair of the membrane may be a rela- tively simple task. • Concrete protected by a membrane can provide a very long service life. • If designed and installed correctly, the membrane can protect the concrete deck for a long period. The European experience indicates that the service life of a well-designed and correctly installed membrane can be more than 30 years and may extend to 60 years with a special system (NCHRP 1996). In the United States, the expected service life of water- proofing membranes is generally 16 to 20 years when installed on new bridge decks and anywhere between 6 and 20 years when installed on existing bridge decks (Russell 2012). The level of traffic and climate are not the only factors that should be used for selecting waterproofing. A particular construction type may result in the development of weak areas within a concrete deck. In phase-constructed bridges, construction joints or closure pours (or both) between phases can result in the development of longitudinal and trans- verse cracking. Use of a membrane system in the closure pour region could be an effective approach for eliminating many corrosion problems reported in conjunction with phase- constructed and modular system bridges. Objective The objective of this topic was to develop the best strategies for waterproofing concrete deck systems with membranes for various conditions and to develop a methodology for pre- dicting the service life of different systems. Scope of Work The scope of the work to achieve the objective of this topic included collecting the available information with respect to use of membrane systems in the United States and Europe and developing the first edition of a best practices manual for waterproofing concrete deck systems.

122 Results A draft version of the best practices manual for membrane use in U.S. practice was developed in cooperation with indus- try. However, the research team felt that additional work needs to be carried out before publishing the work. Following is a summary of some of the findings. Concrete bridge decks are structurally compromised, and their longevity reduced, by exposure to liquid water, chloride, and carbon dioxide ingress. Their vulnerability is increased in more northern regions where deicing salts are used in winter, and the concrete becomes more susceptible to cracking from thermal cycling. The cost of maintenance of a bridge deck over time can be significantly reduced if a greater initial investment is made by the use of a higher standard of membrane at the time of con- struction or rehabilitation. Many U.S. state agencies do not use any waterproofing on their bridges. To provide optimum performance, the waterproofing system must be compatible with the surface quality of the substrate and with the asphalt paving. The NCHRP Synthesis 425 survey revealed several types of defects observed with waterproofing membranes, predomi- nantly a lack of adhesion between the membrane and the concrete deck, a lack of adhesion between the membrane and the asphalt surface, and moisture penetrating through the membrane. All types of defects were more prominent with membranes applied to existing bridge decks than with mem- branes applied to new bridge decks. Most manufacturers rec- ommend a primer on the concrete deck and a tack coat on the waterproofing membrane to improve the adhesion between the layers. Development and Brief Content of Design Guide for Bridges for Service Life The results and essence of the entire research effort were incor- porated into a single document for service life design of bridges, the Design Guide for Bridges for Service Life (the Guide). This section of the final report provides a summary of the philosophy, content, and general framework used in the Guide. The Guide is a stand-alone document and the proj- ect’s main product. Close collaboration was maintained with the AASHTO T-9 bridge subcommittee while developing the Guide. Introduction The design for service life should be approached in a system- atic manner and must be transparent to the owner. It must provide the owner with a clear picture of the costs required to keep the bridge functional during the entire target service life of the bridge, with a clear roadmap for timely interventions and any other actions needed. The traditional approaches for enhancing the service life of bridges used in various codes, such as the AASHTO LRFD Specifications, Eurocode, or British Standards, are mainly in an indirect form, specifying use of certain details such as cover thickness, crack width, and concrete compressive strength. Although such information is useful, it is not adequate to design for service life in a systematic manner. The design for service life for bridges should be approached in a systematic manner and not as a series of isolated tasks, each addressing the service life of a particular portion of the bridge. The bridge maintenance program, retrofit or replace- ment options, and an overall management plan should all be part of this systematic service life design approach. It was concluded that the major missing element for design- ing bridges for service life was the absence of a framework that would allow the problem to be approached in a systematic manner and provide a complete solution in a format that could ensure long-lasting bridges. Individual solutions for given details that have historically reduced service life of bridges, a maintenance plan, a retrofit plan or replacement plan, and bridge management and life-cycle cost analysis are components of this systematic framework, but they are not the framework itself. The framework steps should begin during the design stage and should provide the owner with complete information for ensuring the serviceability of the bridge for a specified target service life. It is important that the resulting systematic plan be transparent and identify the challenges for the period of speci- fied service life at the design stage. It should also eliminate surprises for the owner for the period of specified service life. Guide Approach to Design for Service Life A major aspect of assessing the merits of any feasible alter- native bridge system for a given job is addressing the service life design considerations. A general framework for a systematic approach to design for service life requires the following main elements: • Identifying a feasible bridge system; • For each feasible bridge system, identifying the element, component, and subsystems that should be considered for service life design; • Identifying the factors that can affect the service life of different elements, components, and subsystems; • Identifying various strategies that could mitigate adverse effect(s) of factors capable of reducing service life; • Selecting the optimum strategies considering interaction among various strategies;

123 • Predicting the service life of various bridge elements, components, or subsystems; and • Using life-cycle cost analysis to develop maintenance, retro- fit, or replacement plans for bridge elements, components, and subsystems, such that the bridge as a system can provide the desired service life. The Guide uses the above approach and, when appropriate, communicates the information using flowcharts to ensure the transparency of the process. Figure 3.97 shows an example of a flowchart used in the Guide that can facilitate the service life design process. The factors affecting the service life of particular elements, components, or subsystems of a bridge can be identified using a fault tree, which provides a systematic method of identifying factors in various categories and successive subcategories. The Guide uses fault tree analysis to identify the factors that can affect the service life of a particular bridge. Figure 3.97. Example of flowcharts that can be used in service life design processes.

124 Figure 3.98. Starting point for fault tree. Reduced Service Life of Bridge Deck Caused by Deficiency Caused by Obsolescence Natural or Man made HazardsLoad Induced Production/ Operation Defects Figure 3.99. Continuation of fault tree for load-induced factor shown in Figure 3.98. Load Induced WearFatigue System Dependent Loads Differential Shrinkage System Framing Restraint Traffic Induced Loads ThermalOverload Figure 3.98 and Figure 3.99 show portions of a fault tree used in the Guide to assist in the design of bridge decks for service life. In Figures 3.98 and 3.99, the factors inside the circles are the most basic factors capable of reducing the service life. For these basic factors, an array of strategies capable of mitigating adverse service life effects should be identified, and the optimum strategy selected. Based on the results of the study, the Guide provides strategies capable of mitigating various factors that can reduce the service life of bridges. As an example, Table 3.20 shows some of the strategies and their advantages and disad- vantages, listed in the Guide, for some of the basic factors identi- fied in Figure 3.99. Constructing the fault tree and identifying feasible and cost-effective strategies and solutions for a particular service life issue are dependent on many factors that vary from loca- tion to location and state to state, and also depend on local practices and preferences. Consequently, the end result of the service life design process does not necessarily need to be the development of a unique solution. Identifying possible solutions and strategies to mitigate factors affecting service life could be based on data collected by local DOTs or agencies responsible for the bridge. The life- cycle cost analysis should be included in the strategy selection process and should include design, maintenance, inspection, retrofit, replacement, and user cost to be complete. It is important that the entire service life design process be communicated and shared with the owner, especially with respect to the life-cycle cost analysis portion of the process. The entire process should be transparent. In identifying the array of strategies to mitigate factors capa- ble of adversely affecting service life, it should be noted that there are different ways of achieving enhanced service life of existing and new bridges. Enhancement of service life can be accomplished by using improved, more durable materials and systems during original construction that will require minimal maintenance or by improving techniques and optimizing the

125 timing of interventions such as preventive maintenance actions. Interventions can be planned and carried out based on an assessment of individual bridge conditions and needs or based on a program of preventive maintenance actions. By acknowl- edging that service life can be extended by either using more durable, deterioration-resistant materials or by planned inter- vention, a cost comparison can be made to determine the most cost-effective approach for various environmental exposure levels and various levels of available maintenance and preserva- tion actions. Guide Approach for Predicting Service Life of Bridge Elements, Components, or Subsystems An important step in the design for service life specified in the Guide is the ability to predict the service life of various bridge elements, components, or subsystems. If deterioration models are available, they can be used to predict service life. Other approaches could include the use of data collected over the years by local agencies capable of developing deterioration models, expert opinion, or use of strategies that could avoid deterioration (avoidance of deterioration method) altogether. Summary of Guide Steps for Design for Service Bridge elements, components, and subsystems deteriorate at different rates and have different service lives. The service life of a bridge system is governed by the service life of its critical elements, components, and subsystems. The service life of a bridge system is reached when the service life of critical bridge elements, components, or subsystems is reached, including the ability to repair or replace them economically, or because of other considerations. The Guide provides very detailed instruction on steps that are needed in the design for service life process. The following is a brief summary of these steps. Step 1. Identify the project requirements, particularly those that will influence the service life. Step 2. Identify feasible bridge systems capable of meeting the project demand. Step 3. Select each feasible bridge system one at a time and carry out Steps 4 through 10. Step 4. Identify the factors that influence the service life of bridge elements, components, and subsystems, such as traffic and environmental factors. Step 5. Identify modes of failures and consequences (e.g., the corrosion of reinforcement, causing corrosion-induced cracking and loss of strength). Step 6. Identify suitable approaches for mitigating the fail- ure modes or assessing the risk of damage (e.g., through life-cycle cost analysis) of the use of higher-performing materials for sliding surfaces in bearings or the use of material prone to deterioration at lower initial cost. Step 7. Modify the bridge element, component, or subsystem under consideration by using the selected strategy, and ensure the compatibility between different strategies used for various bridge elements, components, or subsystems. Step 8. Estimate the service life of the bridge element, com- ponent, or subsystem using available data. Step 9. Compare the service life of the bridge element, com- ponent, or subsystem to the service life of the bridge sys- tem and develop appropriate maintenance, retrofit, and replacement plans. Step 10. Develop the design, fabrication, construction, opera- tion, maintenance, replacement, and management plans for achieving the specified design life for the bridge system. Step 11. Conduct life-cycle cost analysis for each feasible bridge system meeting strength and service life requirements, and select the optimum bridge system. Step 12. When specified by the owner or for unique bridges, document the entire design for service life processes in a document called an Owner’s Manual. Conduct an inde- pendent review of the document and provide it to bridge owner at the time of opening the bridge to traffic. Table 3.20. Possible Mitigating Strategies for Various Service Life Issues for Traffic-Induced Loads Service Life Issue Mitigating Strategy Advantages Disadvantages Fatigue Design per AASHTO LRFD Bridge Design Specifications Minimizes possibility of reinforce- ment failure May increase area of steel Overload Increase deck thickness Minimizes cracking Adds weight to bridge structure, increases cost Overload Minimize bar spacing for given amount of steel Improves crack control Does not add to strength. Higher labor. More prone to corrosion and difficult to construct Wear and Abrasion Implement concrete mix design strategies Identified in Chapter 4, Materials Identified in Chapter 4, Materials Wear and Abrasion Implement membranes and overlays Protects surface from direct con- tact with tires Requires periodic rehabilitation every 5 to 10 years

126 Owner’s Manual for Service Life Design The Guide suggests providing an Owner’s Manual for unique bridges or when required by the owner that summarizes the entire design for service life process and recommendations. The purpose of the manual is to equip owners with the complete knowledge necessary to keep the bridge operational for the specified service life period. The Owner’s Manual should be provided to the bridge owner at the time of opening the bridge to traffic. The entire design for service life must be well documented and must include assumptions, limitations, and other informa- tion of which the owner should be aware. The Guide suggests that the Owner’s Manual include complete information with respect to “hot spots” for more detailed or frequent inspections, as well as maintenance, retrofit, or replacement information for various bridge features. The manual should also include a com- plete management plan with respect to service life that provides information on timely maintenance actions and identifies the replacement parts and methodologies for replacement with information on the level of interruption, if any, to traffic. For unique bridges the Guide suggests that a bridge instrumentation and monitoring plan be developed and be correlated with the bridge service life management plan. Additional information in the Owner’s Manual should include the material properties in critical bridge elements, components, or subsystems as used during construction versus the assumed values during the design process. This information will be important for rating bridges. For unique bridges the Guide suggests that the designer exercise good engineering judgment for incorporating detail and the extent of information to be included in the owner’s manual. The bridge Owner’s Manual may be considered as an equivalent of a design calculation document, customarily provided to the bridge owner, except that it contains much more detailed information. The Guide also suggests that an independent review of the Owner’s Manual be carried out before submitting it to the Owner.