Load Rating of Bridges and Culverts with Missing or Incomplete As-Built Information (2022)

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.

6 This chapter reviews the literature identified as relevant to the question of load rating bridges and culverts with missing or incomplete as-built information. Such information can generally be categorized as having to do with diagnostic or proof load testing, NDT, and other pertinent studies. The review in this section is organized accordingly. Of course, numerous other sources on bridge load rating are not focused on bridges with missing information but rather assume that all information necessary to complete an analytical load rating is available. These sources are not included here. Furthermore, no published sources in the scholarly literature address the “engineering judgment” approach to load rating; that is, a rating is based on the rating engineer’s experience and knowledge of the condition of the bridge, traffic, and ratings of other similar structures. As noted previously, the synthesized state of practice in Chapter 3 considers DOT manuals and documents that address load rating BCMI, many of which are available online; materials obtained as a result of the survey; and policy memoranda published by FHWA that impact load rating BCMI. Thus, these are not discussed here but are introduced in Chapter 3. Diagnostic and Proof Load Testing In 1993, Yen and Zhou [7] reported on a diagnostic load test of a reinforced concrete spandrel arch built in 1918. No plans were available for the bridge, which was posted for 15 tons and carried heavy truckloads from a nearby quarry. The bridge had strain gauges mounted primarily on the floor beams and deck of the deteriorated sections, with some of the strain gauges mounted directly on rebar. Two different truck configurations with several different weights made load passes. The maximum measured rebar stress was 5.0 ksi in tension, and the maximum con- crete stress was 230 psi in tension. Yen and Zhou conducted finite element analyses with three models defined by three different assumptions: (1) no deterioration, (2) 20% section loss in the rebar, and (3) parapet that is not an integral load-carrying component. The models displayed reasonably good agreement with the measured data. Load ratings for the rebar in the floorbeam (assumed to be controlling) used the measured live load strains and were 2.16 and 1.24 for the two different loads. Yen and Zhou concluded that the bridge was safe to carry HS-20 AASHTO loads and ML80 PA loads. In 1996, Saraf, Sokolik, and Novak [8] described proof load testing of a concrete bridge and a steel girder bridge in Michigan. The concrete bridge was 77 years old at the time, consisted of a slab on six reinforced concrete T-beam girders, had no plans, and carried an H15 load rating before testing of 0.79. The steel girder bridge was 69 years old at the time, with an H15 rating of 0.60. The bridges were equipped with strain gauges and displacement transducers. Army tanks on trailers achieved the target proof loads needed. The bridges exhibited linear behavior in the C H A P T E R 2 Literature Review

Literature Review 7   tests. Based on the results of the tests, the state updated the H15 load ratings to 1.24 and 1.97, respectively, for the concrete and the steel bridges. In their 1997 report, Klaiber, Wipf, and Streeter [9] reported on the diagnostic testing and load rating of six old reinforced concrete bridges in Iowa. This process measured strains and deflections of the bridges, with two loaded dump trucks as the test vehicles. The calculations of load ratings based on the test results used the procedure outlined in 1993 by Lichtenstein [10], which is the source of the load rating procedure in the AASHTO MBE [2]. Plans were available for five of the six bridges. Five of the six tests showed that the bridges displayed higher capacity than the analytical calculations predicted, resulting in higher HS-20 load ratings. The modified load rating from one test (on the bridge with no plans) ranked lower than the analytical rating for the HS-20 vehicle. In 2002, Eitel, Huckelbridge, and Capaldi [11] presented a method for evaluating and load rating Ohio reinforced concrete slab bridges that lacked complete information. The approach relied on measuring the midspan bridge deflection caused by a loaded dump truck. The com- putation of the ratio of the service load moment to the cracking moment used the measured deflection and assumed properties. The reinforcement ratio was assumed, as was a flexibility coefficient that determines the support fixity. A chart enabled estimation of the ratio of dead load to live load slab effective width, with a total of 18 slab bridges diagnostically tested. Eitel, Huckelbridge, and Capaldi did not directly calculate a load rating factor but instead computed a service live load capacity in tons. The idealized service live load capacity exceeded the Ohio load limit for all of the bridges tested. Following the diagnostic test, two of the bridges were destructively tested by using hydraulic jacks. The load was insufficient to cause yielding in one bridge: coring revealed the slab to be thicker than originally estimated. The second destructive test produced a nonlinear response. In 2004, Chajes, Shenton, and Thompson [12] proposed two methods for load rating con- crete slab bridges without existing plans; both methods rely on measuring strains or deflections of the bridge in a load test. The two methods are referred to as the steel area method (SAM) and the simplified method (SM). The SAM estimates the quantity of reinforcing steel in the slab by using strain or deflection measurements at two different load levels. Once the steel area is pro- jected, a standard BRASSTM rating is completed, applying the other standard assumptions (e.g., transverse load distribution, dead load, impact). The SM employs the live load strain measured in a diagnostic load test, the allowable strain in the steel, and the calculated dead load strain to compute the RF directly. The first test of the procedures entailed small beam specimens in the laboratory and data from a previous diagnostic test of a concrete box culvert with 2 feet of fill. Shenton, Chajes, and Huang in 2007 [13] and Huang and Shenton in 2010 [14] extended this analysis. The first test of the SAM utilized laboratory tests of four large beams. Both the SAM and the SM tests addressed a bridge with available plans. The strain-based SAM under- estimated the area of steel in the bridge, and the deflection-based SAM overestimated the area of the steel. Load ratings for the HS-20 vehicle and several Delaware legal loads using both methods showed that the strain-based SAM values were lower than the theoretically computed ratings and that the load ratings based on the deflection-based SAM exceeded the theoretically computed ratings. The SM ratings were all greater than the theoretical ratings and also higher than the SAM deflection-based ratings. In 2006, Hag-Elsafi and Kunin [15] reported on a diagnostic load test conducted on a New York prestressed concrete girder bridge with no available plans. Before testing, the posted bridge limit was 12 tons. The load test measured the strain at numerous locations, along with three deflections. Utilizing three loaded dump trucks of varying weights, several truck passes enabled estimates of the transverse load distribution, support fixity, and section modulus. Because the details of the prestressing were unknown, the test included conducting analyses to arrive at low

8 Load Rating of Bridges and Culverts with Missing or Incomplete As-Built Information and high values for the initial prestressing force and the eccentricity for both H20 and HS-20 design loads, based on satisfying code requirements for midspan top and bottom stresses under initial and service load conditions. Using the load test results, ratings were completed for H20 and HS-20 loads under the low and high prestressing force and eccentricity assumptions. The project successfully increased the load ratings of the bridge: the recommended H20 inventory and operating ratings were 24.6 and 40.8 tons, respectively, while the recommended HS-20 inventory and operating ratings were 40.3 and 67.7 tons, respectively. Aguilar et al. [16], in 2015, described a detailed New Mexico project aimed at load rating a prestressed concrete double T-beam bridge that had no plans. To analyze the bounds on the prestressing, the project employed Magnel diagrams and assumed material properties, supple- mented by Hilti Ferroscan images of the strands to estimate the harping and tendon eccentricity (it helped that the girders were T-beams, so the sides of the webs could be scanned at different locations). Aguilar et al. first performed a diagnostic load test with limited loads to determine the transverse location of the loads for the proof load test. This step was necessary because of the uncertainty surrounding the integrity of the shear keys between the beams. The maximum load for the proof test was based on the Service III criteria to avoid cracking the girders, which yielded a limiting strain of 532 microstrain for the test. The proof load test was stopped when it reached a maximum recorded strain of 528 microstrain. The test demonstrated that the shear keys were not functioning properly at higher loads. The load rating was computed for three AASHTO and three New Mexico legal loads based on the proof test results. Before the test, the New Mexico DOT rated the bridge by using historical information, producing an HS-20 operating rating of 1.3 (and no bridge posting); as an outcome of the test, the bridge was posted. In 2016, Cuaron, Jauregui, and Weldon [5] reported on a project undertaken by New Mexico State University to investigate the load rating of concrete slab bridges without plans. The project included a national survey and load ratings of 23 slab bridges. The project staff sent a survey to 50 state agencies and received 33 responses. This paper observed some overall results and described the procedures used by eight states based on documents obtained from the states. The load rating procedure began with field measurements and Windsor Probe tests to estimate concrete strength. The successful probe tests on 11 of the 23 bridges yielded concrete compres- sive strengths ranging from 4970 to 8230 psi, which were significantly higher than the assumed strength. A Hilti P200 PS Ferroscan generated estimates of the size and spacing of mild steel reinforcement. In addition to quick scans, the project included more detailed block scans to obtain the size, spacing, and depth of cover of the bottom reinforcement. Cuaron, Jauregui, and Weldon noted that top reinforcement in continuous spans is more difficult to assess; the quick scan produced estimates of spacing and depth of cover, with the size approximated based on his- torical records. A review of plans showed top reinforcement as typically 10% higher than bottom reinforcement. The area of steel per foot was also estimated based on the AASHTO Standard Specifications for Highway Bridges [81]. The field measurements and reinforcement estimates then served as the basis for creating the as-built drawings. Subsequent load ratings used AASHTO BrR for the HS-20; AASHTO Type 3, 3-3, and 3S2; and six New Mexico legal loads (based on an assumed compressive strength of 3 ksi and the strength measured by the Windsor Probe). The average ratings increased 16% with the strengths from the probe, counter to that reported in Miller, Shahrooz, and Gearhart [17] and in Gearhart [18]; however, Cuaron, Jauregui, and Weldon [5] did not indicate the limit state that controlled the load ratings. Comparing the New Mexico load ratings to the ratings obtained with the Idaho and Oregon procedures, the latter load ratings were as much as 69% and 62% smaller, respectively—but on average, 35% and 30% less than the New Mexico State University load ratings. The 2016 master’s degree thesis by Gunter [19] presented a detailed study of a prestressed concrete channel bridge (adjacent inverted double-T beams) with no existing plans. The work

Literature Review 9   focused on the investigation of a specific bridge in South Carolina. The study included a diag- nostic load test of the bridge, laboratory testing to failure of one channel beam, and load rating of the bridge based on the test results. The relatively standard diagnostic test used Bridge Diag- nostics, Inc. strain transducers and displacement sensors. The bridge load totaled a maximum of 48.7 kips—just over half of the South Carolina legal load limit. The load test results served as the basis for updating the load rating of the bridge, using the procedure outlined in the AASHTO MBE [2]. Details of the prestressing strands of the lab test beam were in question. The measured cracking moment and ultimate strengths exceeded the theoretical values. All of the load rating factors were well under 1 before the tests. The HL-93 inventory and operating ratings were higher but still below 1 with the test modified factors; the updated legal load rating equaled 1.14. The 2016 master’s degree thesis by Subedi [20] reported on development of a method for load rating reinforced concrete flat slab bridges based on NDT, finite element modeling, and diag- nostic load testing. Ultimately, the load rating factor is given by the ratio of the vehicle weight required to generate a midspan deflection equal to the AASHTO deflection limit state divided by the deflection produced by the design vehicle. Bagheri et al. [21] reported in 2017 on a method for estimating the stiffness of skewed reinforced concrete slab bridges based on measurements of ambient and traffic-induced vibration of the bridge. This method involved a parametric study using finite elements to calculate the funda- mental frequency of more than 14,000 slab bridges to determine the relationship between the frequency and the geometry and material properties of the bridge. Next, an artificial neural net- work determined a coefficient that relates the frequency to the slab stiffness. In the last step, the team demonstrated the approach by conducting a test in Virginia of a bridge with plans. Bagheri et al. suggest that given an estimate of the stiffness of the bridge, a load rating can be performed (this was not, however, demonstrated in the paper). Jauregui, Weldon, and Aguilar [22], in 2019, described in great detail a procedure for load rating prestressed concrete bridges that are missing plans; this article was a more comprehensive presentation of the work reported by Aguilar et al. [16] in 2015. This procedure incorporated four steps: (1) estimating material properties based on past specifications and Magnel diagrams; (2) verifying the steel, using rebar scanning; (3) performing field load testing at the diagnostic or proof level; and (4) rating the bridge by using proof load test results. Jauregui, Weldon, and Aguilar reported on a national survey that they conducted; 35 out of 50 states responded. The article provided a flowchart of the proposed procedure and then described three case studies of a T-beam bridge, a box-beam bridge, and an I-girder bridge. Details were provided, including the final load ratings for HS-20; AASHTO Types 3, 3-3, and 3S2; and three New Mexico legal loads. In 2019, the Transportation Research Board published Transportation Research Circular E-C257: Primer on Bridge Load Testing [23]. The primer outlined in detail how to conduct diagnostic and proof load tests and explained how to interpret the results and how to use the results in a load rating. While it focused not just on testing BCMI to calculate load ratings, this publication represents an important source for owners considering these types of tests. Nondestructive Testing NDT methods utilize noninvasive measurement techniques to assess the condition and material properties of various bridge members. NDT methods—individually, in combination with each other, or in combination with limited use of minimally invasive inspection methods— can be employed to determine unknown parameters that are essential to the analysis of the structural performance of a bridge. NDT can offer useful information, including location and size of internal steel reinforcement, concrete strength, and member dimensions. NDT also can

10 Load Rating of Bridges and Culverts with Missing or Incomplete As-Built Information determine in situ member conditions (e.g., concrete degradation and spalling, corrosion of internal steel reinforcement). A thorough discussion of NDT falls outside of the scope of this report. However, Chapter 3 of the AASHTO MBE [2] lists the NDT methods that are most applicable to bridges. In addi- tion, the ACI [24] publishes a detailed overview of various NDT techniques for use in evaluating concrete structures (ACI 228.2R, first released in 1998 and then updated in 2013). In this report, the focus is on past applications and research studies that report on applications of NDT in load rating BCMI. NDT methods relevant to BCMI can generally be divided into the following multiple catego- ries based on the applied stimulus (ACI 228.2R [24]): 1. Stress wave methods such as impact-echo, specified in ASTM C1383 [25]; rebound number (ASTM C805 [26]); ultrasonic pulse-echo (ASTM E797 [27]); pulse velocity (ASTM C597 [28]); and spectral analysis of surface waves. 2. Nuclear methods such as backscatter radiometry, direct transmission radiometry, and radio graphy. 3. Magnetic and electrical methods such as cover meters; half-cell potential, specified in ASTM C876 [29]; and concrete resistivity. 4. Infrared thermography, specified in ASTM D4788 [30]. 5. Electromagnetic wave methods such as the eddy current array (ASTM E3052 [31]) and ground-penetrating radar (GPR) (ASTM D6087 [32]). Table 1 lists examples of possible applications of various NDT methods and minimally invasive techniques to estimate the missing information for bridges. Many of the available tech- niques are standardized, with commercially available instrumentation, while others are still at the research and development stage or are reserved for specialized applications. A 2017 report by Miller, Shahrooz, and Gearhart [17] reviewed NDT methods for evaluating reinforced concrete bridges; a companion presentation in 2016 [33] included 54 slides. The pre- sentation covered NDT tests for estimating bar spacing, bar size, and cover depth. The authors addressed methods for estimating concrete strength but found that concrete strength does not affect load rating significantly. Steel strength, however, is more important, and the presenta- tion covered methods for determining steel strength. Several flowcharts showed the process for evaluation and load rating. The report included field studies on two different groups of bridges. The Gearhart master’s thesis [18] served as the basis for Miller, Shahrooz, and Gearhart [17, 33]. Gearhart focused specifically on evaluating NDT methods that are effective for estimating the reinforcement and material strengths for slab bridges. Gearhart studied and tested NDT methods for estimating concrete strength, steel strength, cover depth, bar spacing, and bar size. The author also conducted a national survey, obtaining 62 responses from 29 states and 31 counties in Ohio, and performed blind tests of four Ohio bridges with available plans. Gearhart noted that moment capacity is very insensitive to concrete strength, so concrete strength measure- ments do not need to be extremely accurate, and strength should not be the primary focus of rating slabs without plans. The Schmidt hammer represents the easiest and least expensive of the methods for estimating concrete strength, but the blind tests showed an average error of 24%. Gearhart estimated rebar size, spacing, and cover depth by using a magnetometer; the literature reports errors in the ± 10% to 20% range for this approach. The author noted that cover depths greater than about 2 inches are difficult to determine accurately with a magnetometer. Cover depth in the blind tests recorded an average error of 13.5% while the bar size error was 5.7% and the bar spacing error was 5.2%. Gearhart reported three methods for measuring steel rebar strength: standard tension test, compression test, and hardness test. Because the recovery of a steel rebar sample of suitable length from a slab is not easy, a true tension test is difficult. Results

Literature Review 11   from compression and hardness tests are more easily obtained because compression tests require relatively smaller bar size, and hardness tests are typically performed in situ. Gearhart reported the steel strength from these tests to be within 5 ksi of the actual strength. In 2020, Karshenas and Naghavi [34] performed a state-of-the-art review of available NDT methods as part of the implementation of load rating of Louisiana bridges with missing infor- mation. Their report primarily focused on potential applications of NDT in the load rating of concrete slab bridges, prestressed concrete channel bridges, and prestressed girder bridges. The thorough overview classified NDT methods according to cost, ease of use, and reliability and recommended direct measurements, GPR, ultrasonic echo, cover meter, and radiography to determine as-built geometric information (e.g., member dimensions, concrete cover, location of internal steel reinforcement). The review suggested determining the strength of concrete by core testing, rebound hammer, or penetration resistance testing. In addition, Karshenas and Naghavi recommended the visual inspection, impact-echo, impulse response, half-cell potential, and radiography methods to assess the extent of deterioration in bridge components (e.g., corrosion and section loss, concrete spalling, and rebar debonding). According to the report, important load rating information—such as yield strength of internal steel reinforcement, prestress in steel strands, and condition of strand bond zone—cannot be reliably determined with existing NDT methods. The report included 21 tables that rate the cost, ease of use, and reliability of various NDT methods for estimating the geometric and strength parameters of bridges with missing information, organized by the type of bridge. Missing Information NDT Technique Minimally Invasive Measurements Span length Field measurements Surveying equipment, such as the Global Positioning System (GPS) Terrestrial laser scanning, as described by the California DOT [35] N/A Slab/deck thickness Ultrasonic echo per ASTM E797, described by Kozlov, Samokutov, and Shevaldykin [36] Impact-echo per ASTM C1383, noted by Gibson and Popvics [37] GPR per ASTM D4748, noted by Clemena, Malhotra, and Carino [38] Slab through-depth coring Cross-sectional geometry Field measurements Terrestrial laser scanning, as described by the California DOT [35] N/A Condition at the support Visual inspection Simple support as a conservative assumption N/A Concrete compressive strength Evaluation of historical data (when available) Penetration resistance, as discussed by ACI [24] and per ASTM C803, as described by Nasser and Al-Manaseer [39], [40] Rebound hammer, per ASTM C805 [26] and as discussed by ACI [24] Ultrasonic pulse velocity, per ASTM C597 [28] and as noted by ACI [24] Compression test on field-drilled cores Yield strength of steel Historical records (when available) Evaluation of mill mark data (when accessible) In situ hardness test on exposed rebar per ASTM E140 [41] Tensile testing per ASTM A370 [42] Compression testing on short steel bars per ASTM E9 [43] Hardness test on manually exposed rebar surface per ASTM E140 [41] Tensile strength of the prestressing strand AASHTO MBE [2], which allows estimates of 232 ksi for bridges built before 1963 and 250 ksi for bridges erected after 1963 In situ hardness test on exposed strand per ASTM E140 [41] Tensile testing (which requires strand extracted from structure) per ASTM A1061 [44] Effective depth of steel reinforcement/cover or the strand layout Magnetic resistance meter, as discussed by ACI [24] and Carino [45] Eddy current meter, as described by ACI [24] GPR per ASTM D4748 [46] Radiography, as noted by ACI [24] and Redmer, Weise, and Ewert [47] Coring Area of steel/bar diameter Magnetic resistance meter, as discussed by ACI [24] and Carino [45] Eddy current meter, as described by ACI [24] GPR per ASTM D6432 [48], as noted by Bhaskar and Ramanjaneyulu [49] Coring Internal steel reinforcement corrosion Visual inspection Half-cell potential per ASTM C876 [29] GPR, as discussed by Hasan [50] Coring Exposure of rebar surface Table 1. Examples of possible applications of NDT to determine missing information in bridges.

12 Load Rating of Bridges and Culverts with Missing or Incomplete As-Built Information Other Studies In 2011, Taylor, Amini, and Lindt [51] reported load ratings of 16 Colorado county-owned concrete bridges with missing information—14 prestressed girder bridges and two reinforced concrete slab bridges. The concrete strength for one bridge was measured from three cubes (2″ × 2″ × 2″) fabricated from a piece of concrete that fell off the bridge, but the results were too variable for use in the load ratings. Therefore, assumed material properties were based on the AASHTO MBE [2], with the bridge geometries determined by field measurements. The creation of Magnel diagrams showed the possible prestressing forces and eccentricities for each bridge. Ratings were then computed using AASHTO distribution factors for specific loads: HL-93, HS-20, Type 3 with lane load, and Type 3 without the lane load. Many of the bridges did not rate for the various loads (i.e., RF < 1). Taylor, Amini, and Lindt discussed the very conservative nature of the many unknowns (prestressing force and tendon layout, concrete strength, loss of prestress, unknown mild reinforcement, diaphragm effects, deck composite action, distribution factors, controlling force) as contributing factors to the many load ratings of less than 1. In 2015, Alipour et al. [52] took a data-driven approach to load posting and safety evaluation of concrete slab bridges. They mined the NBI for slab bridges that satisfied certain criteria. The data were used to train various models, then validated with an independent validation set of bridges. The models predicted the load posting well. Alipour et al. applied the models to assess the posting of a set of more than 5,000 reinforced concrete slab bridges with missing plans. The results predicted that 8.4% of the posted bridges do not need posting and that 17% of the unposted bridges need further investigation and may need to be posted. Harris et al. [53] followed up on the work of Alipour et al. [52] but focused only on bridges in Virginia. Harris et al. used a few different types of machine learning models—linear regres- sion, polynomial regression, and artificial neural network (ANN)—to assess and predict bridge performance. In the end, they reported that 38% of the ANN ratings for slabs without plans were lower than the engineering judgment ratings (i.e., the Virginia DOT ratings)—and that 37% of the ANN ratings were conservative relative to the DOT ratings. The objective of the 2018 work by Armendariz and Bowman [54] was to develop a meth- odology for load rating bridges with missing plans. They reviewed load rating methodologies and previous work. The authors outlined a four-step process: bridge characterization, bridge database, field survey and inspection, and load rating. A flowchart tracked the overall process. In three case examples, the load rating of bridges in Indiana utilized the procedure. Two of the case examples are particularly important because they are under fill: one is a metal pipe arch, and the other is an earth-filled reinforced concrete arch. The metal pipe is considered flexible, and the rating is controlled by thrust, seam resistance, buckling stability, and minimum cover. For the example presented, the final LRFR and LFR inventory and operating ratings were con- trolled by thrust and were well above 1.0. The reinforced concrete arch is rigid; the authors noted that the controlling limit state is usually strength (from combined compression and flexure). The load rating process for it relies on an iterative procedure that involves the construction of a moment interaction diagram. A flowchart summarized the procedure. The final LFR inventory and operating load ratings were greater than 1.0. Armendariz and Bowman then reported on the procedure for load rating a buried reinforced concrete box culvert; however, no example was presented. In the last stage, the authors applied the proposed procedure to load rate a 70-foot- long reinforced concrete open-spandrel arch bridge. They performed a diagnostic load test and detailed 3D modeling on the spandrel arch. The analytical inventory load rating and operating LFR were both less than 1.0. Incorporating the results of the load test, the inventory rating was higher but still below 1.0, and the operating rating increased to 1.37. The finite element model results alone yielded an inventory rating of 1.25. The Armendariz Briones dissertation [55] served as the basis for the Armendariz and Bowman work and provided some additional details on the study.

Literature Review 13   Bagheri et al. [56], in 2018, presented a method for load rating reinforced concrete bridges without plans. They developed and analyzed a suite of finite element models to determine the dynamic modal properties of the bridges. They then created an ANN to estimate the flexural stiffness of a bridge based on measured dynamic properties. Next, load tests were employed to estimate other properties, with NDT as needed. Bagheri et al. verified the approach against a Virginia bridge with existing plans. The proposed method yielded lower inventory and operat- ing rating factors, but still within 4% of the AASHTO ratings for the tested bridge. In 2019, Lequesne and Collins [6] provided a synthesis of the state of practice for load rating concrete bridges without plans. The authors discussed NDT methods for estimating reinforcing steel size, depth, and spacing. Lequesne and Collins conducted a national survey, with 24 of the 49 contacted states returning the survey (although some were not fully completed); an appendix contained the full results of their survey. The authors reported methods for load rating falling into the following categories: applying assigned rating factors, rating bridges with historic design loads, performing calculations with assumed or measured properties, and conducting load testing. The Lequesne and Collins report offered three recommendations: (1) FHWA should provide notice to local agencies to retain bridge plans for the life of the structure; (2) for bridges that have been in service for several years and are expected to continue to see similar traffic, the loads, age, and condition should be the primary means for assigning load ratings; and (3) for bridges that undergo a change in use or require special evaluation, a proof load test or modeling with assumed or measured properties is a suitable approach for load rating. Summary The literature search identified multiple studies that utilized diagnostic and proof load testing to aid in load rating of bridges with missing or incorrect information. Table 2 summarizes load test results from previous studies. A comparison of the results shows that, in many instances, the load ratings based on the results of a load test are higher than the assigned load ratings based on the AASHTO MBE. However, in some cases, the load testing resulted in a lower load rating. Although a wealth of literature addresses NDT, very few studies employed NDT methods in the load rating of bridges and culverts with missing or incomplete as-built information. The use of NDT was primarily confined to field measurements, cover meters, GPR, and Schmidt hammer tests. A handful of other studies investigated applying machine learning and sophisti- cated modeling techniques to more fully understand the issue of load rating BCMI. All of the studies identified in this synthesis report have contributed to a better under- standing of the process of load rating BCMI, but a select few are viewed as key, particularly for practitioners. Table 3 lists these key sources and their specific contributions.

14 Load Rating of Bridges and Culverts with Missing or Incomplete As-Built Information Study Bridge Type Plans Methods Rating Factors Yen and Zhou (1993) [7] Open-spandrel concrete arch No Field measurements; diagnostic load test; analytical modeling RF of 1.24 (HS-20) Saraf, Sokolik, and Novak (1996) [8] Bridge 1: Reinforced concrete girder bridge Bridge 2: Composite steel girder bridge No Yes Proof load test Bridge 1: RF = 1.24 (H15) from proof load test vs. RF = 0.79 before load test Bridge 2: RF = 1.97 (H15) from proof load test vs. RF = 0.6 before load test Klaiber, Wipf, and Streeter (1997) [9] Bridge 1: Concrete spandrel arch Yes Diagnostic load test; concrete cores; tensile test on steel Bridge 1: RF = 0.94 (H20): did not change after load test because of lack of strain gauges on deck Bridge 2: Concrete open-spandrel arch Yes Bridge 2: RF = 0.79 (H20): did not change after load test because of lack of strain gauges on deck Bridge 3: Concrete slab Yes Bridge 3: RF = 3.02 (HS-20) from diagnostic load test vs. RF = 0.69 before the load test Bridge 4: Concrete filled spandrel arch Yes Bridge 4: RF = 8.89 (HS-20) from diagnostic load test vs. RF = 7.18 before the load test Bridge 5: Concrete stringer No Bridge 5: RF = 0.95 (HS-20) from diagnostic load test vs. RF = 1.61 before the load test Bridge 6: Concrete filled spandrel arch Yes Bridge 6: RF = 21.29 (HS-20) from diagnostic load test vs. RF = 5.75 before the load test Eitel, Huckelbridge, and Capaldi (2002) [11] 18 RC slabs Not reported Diagnostic load test Estimated service live load capacity greater than Ohio legal load limit for 16 of 18 bridges Chajes, Shenton, and Thompson (2004) [12] RC slab Yes Diagnostic load test; concrete cores Rating increased based on measured strains Hag-Elsafi and Kunin (2006) [15] Post-tensioned bulb-T girder No Diagnostic load test RF = 1.23 or 24.6 tons (H20) after diagnostic load test, which allowed removal of the 12-ton posting Shenton, Chajes, and Huang (2007) [13] RC slab Yes Diagnostic load test Rating depended on the method Aguilar et al. (2015) [16] Prestressed concrete double-T No Diagnostic load test followed by proof load test; Hilti Ferroscan to determine internal steel reinforcement location RF = 0.81 (Type 3) after proof load test vs. RF = 1.30 (HS-20) before load test Gunter (2016) [19] RC channel No Diagnostic load test HL-93: IR = 0.75, OR = 0.98, Legal = 1.14 Jauregui, Weldon, and Aguilar (2019) [22] Bridge 1: Prestressed concrete double-T No Diagnostic test; proof load test; Hilti Ferroscan to determine internal steel reinforcement location Bridge 1: RF = 1.09 (HS-20) after proof load test vs. RF = 1.02 before load test (AASHTO LRFR) Bridge 2: Prestressed concrete box beam No Bridge 2: RF = 1.17 (HS-20) after proof load test vs. RF = 1.40 before load test (AASHTO LFR) Bridge 3: AASHTO Type 3 I-girder No Bridge 3: RF = 1.17 (HS-20) after proof load test vs. RF = 1.64 before load test (AASHTO LFR) Table 2. Summary of load test results from the literature.

Literature Review 15   Source Title Contribution Jauregui, Weldon, and Aguilar [22] Load testing of bridges: proof load testing and the future of load testing Outlines a multistep process for load rating prestressed concrete bridges with missing information; includes three examples (double-T beam, box beam, and I-girder bridges) Transportation Research Board [23] Primer on Bridge Load Testing (Transportation Research Board Circular E-C257) Provides detailed procedures for conducting diagnostic and proof load tests and explains how the test data are used for load rating ACI [24] Report on nondestructive test methods for evaluation of concrete structures Makes a comprehensive presentation of nondestructive technologies for use with concrete structures; includes a summary with applications of 23 NDT methods Miller, Shahrooz, and Gearhart [33, 17] Synthesis Study on Load Capacity of Concrete Slab Bridges without Plans (Presentation) [33] Synthesis of load capacity of concrete slabs without plans [17] Determines methods to find or reasonably estimate slab bridge properties, including bridge dimensions, concrete strength, reinforcing bar size, reinforcing bar spacing, distance from compression face of slab to reinforcing bar, yield strength of reinforcing bar, and support conditions Armendariz and Bowman [54] Bridge Load Rating (Joint Transportation Research Program Publication No. FHWA/IN/JTRP- 2018/07) Addresses load rating buried bridges, including flexible conduit, earth-filled concrete arch, and box culvert Karshenas and Naghavi [34] Investigating Available State-of- the-Art Technology for Determining Needed Information for Bridge Rating Strategies Provides a state-of-the-art review of available NDT methods for implementation in load rating of bridges with missing information; uses a three-tier icon system to rate the cost, ease of use, and reliability of various NDT methods for determining geometric and strength parameters of bridges with missing information, organized by type of bridge Table 3. Key sources for load rating BCMI.