Guidelines for Analysis Methods and Construction Engineering of Curved and Skewed Steel Girder Bridges (2012)

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.

3 1.1 Problem Statement At larger span lengths, tighter curvatures and/or sharper skews, assurance of fit-up, control of the component stresses, and control of the constructed geometry are critical attributes in the construction engineering of steel girder bridges. Significantly curved and/or skewed bridges generally exhibit significant torsional deformations, along with associated significant cross-frame forces, potential for uplift at bearings, and other effects. These attributes must be considered in the design, detailing, and construction of these structures. Conversely, straight bridges with negligible skew respond predominantly in a manner involving vertical girder displacements with little or no torsional response. Bridge engineers have a wide array of approximate and refined analysis and design tools at their disposal for the assessment of constructability. It is important that the right tool is selected for the job at hand. Furthermore, it is essential that construction plans and submittals adequately convey the information necessary to build a given structure safely without unnecessary delays or rework. With regard to these attributes, the key construction engineering considerations for steel I- and tub-girder bridges are as follows: •• Prediction of the deflected geometry at the intermediate and final stages of the construction. During steel erection stages, it can be necessary in some cases to limit the structural displace- ments to avoid fit-up difficulties. In addition, it is particularly important for the engineer to be able to predict the deflected geometry under the steel dead load, prior to the placement of the deck concrete, as well as under the total dead load, after placement of the deck and various appurtenances. It should be noted that, in general, there is no such thing as a “conservative” prediction of the structural displacements. Over-prediction of the displacements can be just as bad as under-prediction when considering the control of the constructed geometry. The deflections during the concrete deck placement generally need to be evaluated to assess that the deck thickness, cross-slopes, superelevations, and grade are within tolerances, the dead load rotations are limited at the bearings, the separate units are sufficiently aligned at deck joints, and the separate phases are matched in phased construction projects. Detailers and fabricators use long-established practices for various types of steel structures in which they detail and fabricate the steel components such that the parts do not fit together when they are in their unloaded (unstressed and undeformed) geometry. This initial lack of fit of the undeformed components is used to compensate for some of the displacements that occur under load, and it can facilitate or hinder the assembly of the unshored, partially shored, or shored structure depending on the procedures and the erection conditions. In curved and/or skewed I-girder bridges, the corresponding practices are commonly termed steel dead load fit (SDLF) C H A P T E R 1 Background

4 Guidelines for Analysis Methods and Construction Engineering of Curved and Skewed Steel Girder Bridges or total dead load fit (TDLF) detailing of the cross-frames. These detailing methods entail the fabrication of the cross-frames in a geometry that does not fit-up with the connection work points on the initially fabricated (cambered and plumb) girders. The corresponding internal locked-in forces twist the girders in a direction opposite to that corresponding to the torsional displacements under the bridge steel or total dead load. Due to the combined dead load and locked-in force effects, the girders deflect into a position where their webs are approximately plumb under the steel dead load, for SDLF, or under the total dead load, for TDLF. In certain cases, the dead load and locked-in force effects approximately cancel each other, such that the net final stresses due to the torsional deformations are approximately zero; however, in other cases these internal effects are additive (i.e., the locked-in forces increase the internal stresses). Numerous bridges also are built in which all the components are detailed ideally to fit-up in their undeformed geometry. This method of detailing is commonly referred to as no-load fit (NLF). When NLF detailing is used, the girders are plumb in the theoretical zero load condition when connected to the cross-frames, but due to the torsional deformations, they deflect into a position in which their webs are out of plumb, or laid over, under the action of the steel and total dead loads. There are various advantages and disadvantages to all of the above methods of detailing, and generally, different methods work well for different bridge types and geometries. Furthermore, it is important to note that the above descriptions are from the perspective of the structural analysis and behavior of steel I-girder bridges. However, the detailer and the fabricator do not conduct any structural analysis. When SDLF or TDLF detailing is used, the detailer and fabricator work solely with the specified steel dead load and total dead load cambers of the girders. The specified steel or total dead load cambers are subtracted from the initially fabricated (cambered and plumb) girder geometries and the cross-frames are detailed to fit between the girders in the anticipated plumb steel or total dead load final geometry. The torsional interactions between the individual girders and the overall structural system, via the attached cross-frames as the structure deforms under the loads, is only indirectly and approximately considered. SDLF and TDLF detailing are very effective at achieving approximately plumb steel girder webs at the targeted dead load condition. However, the resulting effects on the structural responses are quite complex and are generally not well understood. This has led to the current state of practice where the AASHTO LRFD Specifications (AASHTO, 2010) Article C6.7.2 state that for curved I-girder bridges, “ . . . the Engineer may need to consider the potential for any problematic locked-in stresses in the girder flanges or the cross-frames or diaphragms. . . .” However, due to the lack of detailed knowledge of the locked-in stresses that can be generated, no guidance is provided regarding when the influence of these stresses needs to be considered in the design. The de facto standard practice is that these effects are rarely, if ever, included in design calculations. That is, the implicit assumption in the structural design of steel I-girder bridges is no-load fit (NLF). The components are implicitly assumed to fit-up perfectly in their undeformed condition under zero load. As a result, regardless of the level of sophistication of the structural analysis, the structural displacements, internal forces, and internal stresses used in current practice are in error to the extent that the locked-in responses due to SDLF or TDLF detailing are important. The lack of understanding of SDLF and TDLF detailing effects has led, in some instances, to conflicting job requirements, such as stating that TDLF detailing should be used and that the I-girder webs should be plumb under the steel dead load condition, or stating that no significant locked-in forces shall be generated and that the I-girder webs should be plumb in the final dead load condition. The I-girder webs can be plumb only under one loading due to the fact that curved and skewed bridges displace torsionally under load. SDLF detailing targets approximately plumb webs in the steel dead load condition, while TDLF detailing targets approximately plumb webs under the final dead load. However, these detailing practices produce locked-in forces due to the corresponding fabricated initial lack of fit between the undeformed (cambered and plumb)

Background 5 no-load geometry of the girders and the fabricated geometry of the cross-frames. These forces can be both additive and subtractive with the dead load forces in the structure. Appendix A provides summary definitions of key terms pertaining to cross-frame detailing. It is essential that the reader understand these definitions to facilitate study and interpretation of the corresponding results and discussions throughout this report. •• Determination and assessment of cases where stability effects may be important. In curved and/or skewed structures, stability effects show up as significant second-order amplification of the displacements and the corresponding internal forces and stresses. In cases where they experience significant stability-related limit states, curved and skewed structures do not exhibit a “bifurcation” from a primary load-displacement response. Rather, the structural displacements increase at an increasing rate as the stability limit of the structure is approached. In cases where the structure is stability critical, second-order amplification can significantly impact the prediction and control of the constructed geometry. In girder bridge structures, large second-order amplification generally should be avoided in the structure’s final constructed condition as well as during the concrete deck placement. However, the engineer needs to be able to anticipate and/or predict a problem in order to prevent it. Lastly, it is important to note that large second-order amplification may not present any significant problem during intermediate stages of steel erection, unless the amplified displacements lead to difficulty with fit-up of the structural components. •• Identification and alleviation of situations where fit-up may be difficult during the erection of the structural steel. Due to a combination of (1) structural component or unit weights, (2) the deflections of the steel components under their self-weight during a specific erection stage, as well as (3) the stiffnesses of the components (i.e., the component resistances to being deformed by come-alongs, jacks, cranes, etc. such that their connections can be made), some situations involving tight curves, sharp skews, and/or long spans may be particularly problematic for the erector to fit the structural components together. These situations generally must be identified and addressed by the development of suitable erection plans. It is well known that TDLF detailing of the cross-frames in I-girder bridges tends to increase the forces required for fit-up. This is because the cross-frames do not fit together with the girders (without some force fitting) until the girder total dead load vertical deflec- tions have occurred in the final constructed configuration (including the influence of the concrete slab weight). The girders are not yet subjected to the total dead load, nor are they connected together in the final constructed geometry, during a given intermediate steel erection stage. In cases where cross-frames or other secondary framing must be included in shop assembly, the fabricator is not likely to choose TDLF. Inclusion of such framing in a shop assembly is rare and only necessary in complex framing situations, such as a single-point urban interchange (SPUI), where girders of varying lengths and curvature are joined by multiple short, stiff diaphragms. For such situations, the fabricator will likely choose SDLF or NLF so that the steel can be assembled in the yard without the weight of the deck present. In such cases, it is good for the erector to be aware of the assembly requirements so that the field assembly procedure can closely mimic the shop support conditions inasmuch as the jobsite conditions will allow. •• Estimation of component internal stresses during construction and in the final constructed condition. AASHTO LRFD Article 6.10.3 requires various checks of factored forces and stresses in steel girder bridges during construction. These include the following: 1. Prevention of any nominal yielding under factored loads (neglecting initial steel residual stress effects) during the construction. 2. Checking of strength limit states, which in some cases, can occur prior to nominal yielding of the structural components.

6 Guidelines for Analysis Methods and Construction Engineering of Curved and Skewed Steel Girder Bridges 3. Prevention of girder web bend buckling or shear buckling during the construction, such that the out-of-plane deflections of the (initially out-of-flat) girder webs are limited. 4. Limiting of girder flange lateral bending stresses (to 0.6Fy) to ensure the applicability of the AASHTO resistance equations for the girder strength limit states, and practically, to limit the magnitude of the flange lateral bending deformations. 5. Control of tensile stresses in the concrete deck, to limit the potential for significant deck cracking. Generally speaking, the structural analysis used for assessing the construction conditions must be sufficiently accurate such that, at the least, all major contributors to the structural responses are accounted for (including all major contributions to the structural displacements, e.g., any significant deformations in attachment details). It is important for engineers to understand if, and when, the responses of curved and/or skewed steel girder bridges are impacted significantly by (1) SDLF or TDLF detailing effects and/or (2) structural stability (i.e., second-order amplification) effects, in addition to the primary effects associated with the bending and twisting of these structures under load. •• Development of sufficient construction plans and submittals. Given the application of a sufficient level of structural analysis for a given job, it is also important that the construction plans and procedures contain adequate detail to properly convey the job requirements as a function of the bridge and construction complexity. Bridges with significant span lengths, curvature, and/or skew generally require detailed planning of the erection procedures and sequences such that lifting and assembly of their spatially deformed components is achievable. Longer bridges typically require placement of the deck concrete in multiple stages. Setup of the concrete from prior stages and, in some cases, during the current stage, can have a significant influence on the final geometry and the ultimate performance of the structure. Conversely, shorter bridges with minor curvature and skew can be built with less attention to the construction engineering. With respect to all of the above considerations, it is important that an appropriate level of effort is applied for the task at hand. More complete guidelines are needed in current practice (2012) regarding the level of construction analysis, plan detail, and submittals for curved and/or skewed steel girder bridge structures. 1.2 Current Knowledge Substantial progress has been achieved in recent years with the streamlining and unification of the AASHTO LRFD (2010a and b) provisions for general steel girder bridges. These Specifications provide more organized and explicit guidance on design for constructability than ever before. Also, recent AASHTO/NSBA Guidelines and Guide Specifications (AASHTO/NSBA, 2003, 2006, 2007, and 2011) provide numerous useful and important recommendations. In addition, many state DOTs have developed substantial constructability guidelines, such as PennDOT (2004), TxDOT (2005), and NCDOT (2006). However, while these documents provide important recommendations applicable to curved and/or skewed steel girder bridges, they target a broad range of steel bridge construction. The construction engineering of highly curved and/or skewed bridges is a highly specialized topic. NCHRP Project 12-79 seeks to develop recommendations that can be fully integrated with the present Specifications and Guidelines to better address the unique attributes of curved and/or skewed steel girder bridges. In recent years, the capabilities for simulation of physical tests using advanced 3D finite element analysis (FEA) has progressed to the point that, in numerous areas, the results from physical experiments can be reproduced readily and quite reliably. There is great potential for advanced 3D FEA simulation methods to be used as a tool for more comprehensive assessment of various levels of analysis and calculation suitable for design. Nevertheless, similar to the

Background 7 results from experimental testing, the results from an FEA test simulation are only as good as the accuracy of •• The detailed geometry (e.g., plate thicknesses, deck-slab thicknesses, haunch depths, girder web depths, bearing heights, bearing plan locations, etc.), •• The load and displacement boundary conditions, including any thermal loading conditions where important, and bearing restraints with finite stiffness or flexibility where important, •• The assumed initial conditions (e.g., initial residual stresses, geometric imperfections, any lack of fit between components in their unloaded condition, etc.), •• The constitutive relationships for the various constituent materials, including attributes such as early stiffness and strength gain of the deck concrete at a given casting stage, or between stages when the deck is placed sequentially in multiple stages, creep and shrinkage deformations of the concrete, concrete micro-cracking in tension, and concrete tension stiffening due to interaction with the deck reinforcing steel, and •• The kinematic assumptions and/or constraints imposed by structural theories and/or associated with the assumed interconnection between various components (e.g., the modeling of stay-in-place metal deck forms tied to the girders by flexible strap details; also, the composite interconnection between the steel girders and the concrete slab, including local short-term and creep deformation of the concrete in the vicinity of shear studs etc., particularly if accounting for early concrete stiffness gains). The consideration of above attributes should not detract from the use of advanced 3D FEA test simulations. In many respects, the above attributes are more easily specified, controlled, and quantified in sophisticated 3D FEA models than in physical tests. Also, in certain situations, many of the above attributes have an inconsequential effect on the structural response. However, similar to successful experimental testing procedures, the execution of refined test simulations requires great care in the creation and setup of the models. This is particularly the case where advanced simulation capabilities are not facilitated well by simplified computer user interfaces. As stated well by Hall et al. (1999), “3D FEA models are not all the same.” The current knowledge about the true accuracy of different methods of analysis for curved and/or skewed steel girder bridges is limited. NCHRP Project 12-79 provided an opportunity to gain substantial insights into the behavior of curved and skewed steel bridge structures, as well as the accuracy of various methods of analysis for these structures, by comparing the results from practical design-analysis methods to the results from refined 3D FEA test simulations. The NCHRP Project 12-79 research is the first time that the overall analysis and construction engineering of curved and/or skewed steel girder bridges has been studied in a systematic manner, considering a large sample of bridges representative of the range of structures encountered in practice, to develop improved guidelines for practice. 1.3 Objectives and Scope of This Research The objectives of NCHRP Project 12-79 are to provide the following: 1. An extensive evaluation of when simplified 1D or 2D analysis methods are sufficient and when 3D methods may be more appropriate for prediction of the constructability and of the constructed geometry of curved and/or skewed steel girder bridges, and 2. A guidelines document providing recommendations on the level of construction analysis, plan detail, and submittals for curved and skewed steel girder bridges suitable for direct incorporation into specifications or guidelines. Both I- and tub-girder bridges are addressed.

8 Guidelines for Analysis Methods and Construction Engineering of Curved and Skewed Steel Girder Bridges The first major objective starts with the assessment of the accuracy of “base” or “conventional” 1D (line-girder) and 2D-grid methods of analysis, representing current standards of care in the profession. These assessments lead to the identification of a number of important improvements that can be made to the current simplified methods of analysis. Various improvements are addressed that 1. Are easy to implement in structural engineering practice, and 2. Result in substantial improvements in the ability of the methods to capture the physical responses with minimal additional calculation effort. In recognition of the importance of integration with structural analysis and design software in structural engineering practice, specific considerations with respect to software implementation also are addressed. The identification of when stability effects (i.e., second-order amplification effects) are significant, as well as the calculation of these effects when they are important, is considered. In addition, a thorough evaluation of the influence of steel dead load fit (SDLF) and total dead load fit (TDLF) detailing of the cross-frames in steel I-girder bridges is conducted. The research focused on the first objective is summarized in the NCHRP Project 12-79 Task 8 report, “Evaluation of Analytical Methods for Construction Engineering of Curved and Skewed Steel Girder Bridges,” Appendix C of the contractors’ final report. The second major objective is addressed by the NCHRP Project 12-79 Task 9 report “Recommendations for Construction Plan Details and Level of Construction Analysis,” which is included as Appendix B of this document. The Task 9 report provides a detailed description of considerations necessary for the development of construction plans. This information is provided in a specification format, complete with a commentary. In addition, the Task 9 report synthesizes key recommendations from the Task 8 research into a specification form. 1.4 Organization of This Report Chapter 2 of this report provides a brief overview of the research approach used in NCHRP Project 12-79. Chapter 3 highlights the major findings from this research and their applications. Section 3.1 summarizes the results from the core NCHRP Project 12-79 research involving the assessment of the “base” or “conventional” 1D line-girder and 2D-grid methods of analysis, representing the current standards of care in the profession. A matrix of scores is provided, indicating the general accuracy of each of the methods for determining different types of responses. This section also gives several examples of how the matrix of scores should be applied. Section 3.2 discusses detailed results behind the assessment of the conventional analysis methods in Section 3.1 and focuses on key improvements that can be made to the current simplified methods of analysis identified in Task 8 of the NCHRP Project 12-79 research. Section 3.3 then summarizes essential results from the portion of the NCHRP Project 12-79 research focused on evaluating the influence of steel dead load fit (SDLF) and total dead load fit (TDLF) methods of detailing the cross-frames in steel I-girder bridges. This is followed by Section 3.4, which gives a synthesis of the overall pros and cons of no-load fit (NLF), steel dead load fit (SDLF), and total load fit (TDF) detailing of cross-frames, and Section 3.5, which provides a few basic recommendations for selection of cross-frame detailing methods in I-girder bridges. Chapter 3 concludes with Section 3.6, which highlights key construction engineering recommendations captured in NCHRP Project 12-79 Task 9. Chapter 4 emphasizes the most important findings of the NCHRP Project 12-79 research, provides specific recommendations for application and implementation of the findings, and describes areas where further research would be valuable.

Background 9 Appendix A provides summary definitions of key terms pertaining to cross-frame detailing. It is essential that the reader understand these definitions to facilitate study and interpretation of the corresponding results and discussions throughout this report. Appendixes B and C contain the reports for Tasks 8 and 9, addressing the two major objectives of the NCHRP Project 12-79 research. In addition, Appendix D contains a Task 7 report that provides specific written documentation on three of the 76 bridges considered in the NCHRP Project 12-79 studies. Appendix E provides a short summary of each of the bridges studied by the NCHRP Project 12-79 researchers, emphasizing the primary considerations addressed by each bridge. Appendixes F and G, respectively, show an early survey sent to owners/agencies in July 2008 and provide a brief summary of policies and practices pertaining to the analysis and design of curved and/or skewed steel girder bridges at the beginning of the NCHRP Project 12-79 research. Appendix H summarizes the criteria used for the parametric study bridges designed and evaluated during the core NCHRP Project 12-79 research. The parametric study designs were developed to reflect a comprehensive range of potential curved and/or skewed steel girder bridge attributes and geometries based on current practices. Appendix I provides a more detailed summary of results for each of the existing, example, and parametric bridges studied by the project, while Appendix J provides the engineering drawings for all of the bridges. Detailed electronic data from the complete set of analysis studies is available as one of the project’s Task 8 products. Finally, Appendix K ex- plains the organization of the project electronic data. Please note that Appendixes C through K are not published herein but are available at the TRB website by searching on NCHRP Report 725.