Design of Concrete Structures Using High-Strength Steel Reinforcement (2011)

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.

51.1 Introduction TheAASHTOLRFDBridgeConstruction Specifications (2004) permit the use of uncoated reinforcing steel conforming to any of the specifications given in Table 1. Since all grades of rein- forcing steel (bars and wires) have an ASTM designation, ASTM designations will be used throughout this report. Recent revisions to §9.2 of the construction specifications and to MP 18 Standard Specification for Uncoated, Corrosion- Resistant, Deformed and Plain Alloy, Billet-Steel Bars for Con- crete Reinforcement and Dowels (AASHTO 2009) permit the specification of ASTM A1035 (2009) reinforcing steel. A1035 reinforcing bars are low-carbon, chromium steel bars charac- terized by a high tensile strength and a stress-strain relationship having no yield plateau. Yield strength is determined using the 0.2% offset method and is specified to exceed 100 ksi or 120 ksi. Because of their high chromium content (specified to be between 8–11%; slightly less than stainless steel), A1035 bars are reported to have superior corrosion resistance when com- pared to conventional reinforcing steel grades. For this reason, designers have specified A1035 as a direct, one-to-one replace- ment for conventional reinforcing steel as an alternative to stainless steel or epoxy-coated bars. The AASHTO LRFD Bridge Design Specifications (2007), however, limit the yield strength of reinforcing steel to 75 ksi for most applications. Therefore, although A1035 steel is being specified for its corro- sion resistance, its higher yield strength cannot be utilized. For many years, the design of reinforced-concrete structures in the United States was dominated by the use of steel rein- forcement with yield strength, fy, equal to 40 ksi and, since about 1970, 60 ksi. Design with steel having higher yield strength values has been permitted, but since the 1971 Ameri- can Concrete Institute (ACI) edition of ACI 318, yield strength values have been limited to 80 ksi. Currently, ACI 318 (2008) permits design using steel reinforcement with yield strength, defined as the stress corresponding to a strain of 0.0035, not exceeding 80 ksi. The exception is spiral transverse reinforce- ment in compression members where the use of yield strength up to 100 ksi is permitted. The AASHTO LRFD Bridge Design Specifications (2007) similarly limit the use of reinforcing steel yield strength in design to no less than 60 ksi and no greater than 75 ksi (exceptions are permitted with owner approval). Both ACI and AASHTO limits have been written and inter- preted to not exclude the use of higher strength grades of steel, but only to limit the value of yield strength that may be used in design. The limits on yield strength are primarily related to the prescribed limit on concrete compressive strain of 0.003 and to the control of crack widths at service loads. Crack width is a function of steel strain and consequently steel stress (Nawy 1968). Therefore, the stress in the steel reinforcement will always need to be limited to some extent to prevent cracking from affecting serviceability of the structure. However, with recent improvements to the properties of concrete, the ACI 318 limit of 80 ksi and AASHTO limit of 75 ksi on the steel reinforcement yield strength are believed to be unnecessarily conservative for new designs. Additionally, an argument can be made that if a higher strength reinforcing steel is used but not fully accounted for in design, there may be an inherent overstrength in the member that has not been properly taken into account. This concern is most critical in seismic applica- tions or when considering progressive collapse states. Steel reinforcement with yield strength exceeding 80 ksi is commercially available in the United States. If allowed, using steel with this higher capacity could provide various benefits to the concrete construction industry by reducing member cross sections and reinforcement quantities, which would lead to sav- ings in material, shipping, and placement costs. Reducing rein- forcement quantities would also reduce congestion problems leading to better quality of construction. Finally, coupling high- strength steel reinforcement with high-performance concrete should result in a much more efficient use of both materials. Additionally, much of the interest in higher strength rein- forcement stems from the fact that many of the higher strength C H A P T E R 1 Background

grades are more resistant to corrosion and therefore very attractive in reinforced-concrete applications. For instance, the A1035 reinforcing steel used in this study is reported to be between 2 to 10 times more resistant to corrosion than conventional A615 “black” reinforcing steel. In some appli- cations, A1035 reinforcing steel has replaced A615 steel on a one-to-one basis on the premise that it is more resistant to corrosion but not as costly as stainless steel grades. Clearly, if the enhanced strength of A1035 steel could be used in design calculations, less steel would be required, and this would result in a more efficient and economical structural system. 1.2 Objectives of NCHRP Project 12-77 The objective of the study presented in this report is to evaluate existing AASHTO LRFD Bridge Design Specifications relevant to the use of high-strength reinforcing steel and other grades of reinforcing steel having no discernable yield plateau. The focus of the experimental phase of this study is the use of ASTM A1035 (2009) reinforcing steel since it cap- tures both behavioral aspects of interest (i.e., it has a very high strength and has no discernable yield plateau). The analytical program of this study supplements the experimental data and evaluates design issues related to other grades of reinforcing steel with no distinct yield plateau. The project identified aspects of reinforced-concrete design and of the AASHTO LRFD Bridge Design Specifications that may be affected by the use of high-strength reinforcing steel. Design issues were prioritized and an integrated experimen- tal and analytical program was designed to develop the data required to permit the integration of high-strength reinforce- ment into the LRFD specifications. As described in Chapter 2, this program included parametric, experimental, and analytic studies in addition to a number of “proof tests” intended to validate existing LRFD provisions when applied to higher strength reinforcing steel. Thus, a crucial objective of the present work is to identify an appropriate steel strength and/or behavior model to ade- quately capture the behavior of high-strength reinforcing steel while respecting the tenets of design and the needs of the designer. As will be described throughout this report, a value of yield strength, fy, not exceeding 100 ksi was found to be permissible without requiring significant changes to the LRFD specifications or, more critically, to the design philosophy and methodology prescribed therein. Some limitations to this increase in permissible yield strength were identified and also are discussed. 1.3 Literature Review 1.3.1 Mechanical Properties of A1035 Reinforcing Steel A number of mechanical properties for reinforcing steel have been reported in the literature, although by far the most important are the tensile yield (fy) and ultimate strengths (fu); these parameters are discussed at length below. El-Hacha and Rizkalla (2002) report other mechanical properties of A1035 to be consistent with the higher tensile yield strength. Based on tests of #4, #6, and #8 bars, they report the following: • Compressive yield strength is the same as tensile yield, fy; • Poisson Ratio, ν = 0.26; 6 Table 1. Uncoated reinforcing steel permitted by 2004 AASHTO LRFD Bridge Construction Specifications. Designation Title Note Tested in This Study?* AASHTO M31 ASTM A615 Deformed and Plain Carbon- Steel standard reinforcing steel unless otherwise specified yes AASHTO M322 ASTM A996 Rail-Steel and Axle-Steel Plain Bars no ASTM A706 Low-Alloy Steel Deformed and Plain Bars “weldable” reinforcing steel yes AASHTO M225 ASTM A496 Deformed Steel Wire “cold-rolled” deformations on A82 plain wire yes AASHTO M55 ASTM A185 Welded Plain Wire Fabric welded A82 wire no AASHTO M32 ASTM A82 Plain Steel Wire yes AASHTO M221 ASTM A497 Deformed Steel Welded Wire Reinforcement welded wire fabric having wires conforming to A496 no ASTM A955 Deformed and Plain Stainless Steel Bars different types (allowable chemistries) of stainless steel are permitted within A955 yes *See Appendix A.

• Shear capacity exceeds the theoretical value of by a significant margin; and • The tensile capacity of bars is unaffected by the presence of standard 90° bends. 1.3.2 Tension Properties of A1035 Reinforcing Steel High-strength reinforcing bars often do not have a distinct yield plateau, as shown in Figure 1. For the representative case of an A1035 #5 reinforcing bar shown in this figure, the yield strength is determined to be 93 ksi or 114 ksi depending on whether the value is determined as that corresponding to a strain of 0.0035 or 0.005. The yield strength is determined to be 123 ksi if the 0.2% offset method is used to determine the yield point. If a simple definition of 1% strain is used (as is commonly used for prestressing steel, another type of steel without a well-defined yield plateau), the yield stress is approx- imately 140 ksi. Regardless of the method used for determining yield stress, the value of 68 ksi is found for the representative A615 #5 bar shown. A review of tensile test data from 16 previous studies of A1035 steel given in Appendix A results in the following conclusions: • Values of yield (fy) and ultimate (fu) strengths and the strain corresponding to the ultimate stress are relatively consistent among different test programs. • Values of rupture strain vary considerably although this may be an artifact of the test procedure where strain gages or extensometers typically do not capture ultimate behavior. • The use of the ASTM A1035-prescribed 0.2% offset method for establishing yield strength results in the greatest variabil- τ = f y 3 ity (COV = 10.3%) whereas stress based on absolute strain approaches to establishing the yield strength are consistent at each strain level considered (COV ≈ 7%). • There is little variation in material properties with bar size. Contrary to conventional wisdom, results from other studies appear to indicate that larger A1035 bars have mar- ginally greater strengths than smaller bars. Results from the present study, however, indicate a marginal reduction in bar strength with increasing bar size. • Regardless of the manner by which yield stress is determined, the condition that fu > 1.25fy is satisfied; this relationship is implicit in a number of AASHTO design articles including those relating to (1) mechanical couplers (AASHTO LRFD §5.11.5.2.2); and (2) element overstrength (AASHTO LRFD Appendix B3). For the purposes of modeling steel behavior, some litera- ture proposes “best fit” relationships for A1035 stress-strain behavior (Vijay et al. 2002, Rizkalla et al. 2005, Mast et al. 2008). In the present study, both Mast et al. (see Appendices C and D) and a Ramberg-Osgood (Ramberg and Osgood 1943) function are alternately adopted. Ramberg-Osgood (R-O) func- tions are commonly used to model prestressing strand and post-tensioning steel, and the parameters may be established directly from representative stress-strain curves. (R-O param- eters for the A1035 steel tested in this study are provided in Appendix A.) 1.3.2.1 Modulus of Elasticity, Es Regardless of yield or ultimate strength, all steel reinforcing bar grades have a reported modulus of elasticity, Es = 29,000 ksi. At stress levels below about 60 ksi, there is no evidence that the 7 (a) Complete Stress-Strain Curves (b) Various Determinations of fy Figure 1. Representative Stress-Strain Curves for A1035 and A615 Reinforcing Steel.

modulus varies from steel grade to steel grade. High-strength steel does, however, exhibit a “proportional limit” where the modulus begins to decrease as is evident in Figure 1a. Although this limit is partially a function of the steel capacity, it has been observed that A1035 steel behaves in an essentially linear man- ner to at least 70 ksi regardless of ultimate capacity (Mast et al. 2008 and this study). It is noted that while some empirical A1035 stress-strain relationships capture the behavior at large strains reasonably well, they fail to capture the initial linear behavior accurately and therefore may not be appropriate for design. An R-O formulation or a piece-wise formulation (Mast et al. 2008) overcomes this issue. 1.3.2.2 Fatigue Performance of High-Strength Steel Reinforcement DeJong and MacDougal (2006) and DeJong (2005) pres- ent a study of the fatigue behavior of high-strength reinforc- ing steel. DeJong conducted fatigue tests of ASTM A1035 steel having a reported (0.2% offset) yield value of 116 ksi and ultimate tensile strength of 176 ksi. Tests on #3, #4, and #5 bars demonstrated a fatigue strength (at N = 1 million cycles) of 45 ksi. Companion tests on Grade 60 reinforcing bars had a fatigue life of 24 ksi. El-Hacha and Rizkalla (2002) report fatigue tests of #4 and #6 A1035 reinforcing bars having a nominal yield strength, fy = 120 ksi. The endurance limit is not established in this study (no tests having N > 500,000 were conducted) although the behavior is reported to be generally superior to that expected for A615 bars. The tests were run with fmin = 0.2fy and the lowest stress range was S = 0.45fy = 54 ksi at which the observed fatigue life was approximately N = 500,000 and 360,000 for the #4 and #6 bars, respectively. Projecting these S-N results to greater values of N would lead to results similar to that reported by DeJong (2005) and superior to the behavior predicted by the present AASHTO requirements (Appendix E). No other known studies have examined the fatigue per- formance of high-strength reinforcing steel, although a num- ber of studies have reported fatigue properties of reinforcing steel having fy < 60 ksi. These investigations are summarized in Appendix E. Based on the data presented in Appendix E, it is seen that no studies report an endurance limit less than 24 ksi in tension-tension (i.e., fmin positive) tests. 1.3.3 Flexural Reinforcement To apply the higher material resistance factor, φ = 0.9 allowed by AASHTO (and ACI 318) in the design of tension-controlled reinforced-concrete flexural members, a member should exhibit a desirable ductile behavior. A desirable behavior implies that at service loads, the member should display small deflections and minimal cracking while at higher loads the member should display large deflections and sufficient crack- ing to provide warning before reaching its ultimate capacity. Both deflection and cracking are primarily a function of steel strain near the tension face of the member and, in general, desirable behavior of a member is related to ductility, which relates to yielding or inelastic deformation of the steel rein- forcement. For lower strength reinforcing materials, the only way to obtain high strains near the tension face at nominal strength is to ensure yielding of the tension steel; however, for high-strength reinforcement, yielding is no longer necessary (Mast et al. 2008). The objective of the work reported by Mast et al. (2008) was to assess the adequacy of a proposed 100 ksi reinforcement stress-strain relationship for A1035-compliant steel in order to establish acceptable strain limits for tension-controlled and compression-controlled sections reinforced with this high- strength steel. Mast et al. studied the behavior of concrete beams subject to flexural and axial loads at service level and nominal strength and determined the section behavior using a cracked section analysis that satisfied equilibrium and com- patibility. They assumed an elastic concrete stress distribu- tion under service load and used the ACI rectangular stress block to model concrete at nominal strength. Although they proposed a more complex empirical relationship for A1035 stress-strain behavior, Mast et al. adopted an elastic-perfectly plastic steel stress-strain relationship in their analysis. They used a steel modulus, Es = 29,000 ksi and defined a plastic yield plateau at fy = 100 ksi. This approach is equivalent to simply increasing the current code-prescribed limits on steel reinforcement yield strength to 100 ksi. For the nominal strength, Mast et al. performed a numer- ical analysis considering a rectangular, singly reinforced- concrete section having a number of different reinforcement ratios. They considered a concrete compressive strength fc′ = 6500 psi and an ultimate compression strain εcu = 0.003. Mast et al. considered elastic-perfectly plastic steel behavior hav- ing yield strengths of fy = 60, 80, and 100 ksi. They calculated balanced reinforcement ratios, ρb = 3.95%, 2.60%, and 1.85% for the values of fy = 60, 80, and 100 ksi, respectively. For ρb values greater than these limits, the section capacity was controlled by concrete compression and was therefore unaffected by the steel grade used. For sections with ρb < 1.75%, the use of the 100 ksi elastic-plastic model typically underestimated the nominal moment capacity of the section with respect to the actual behavior. On the other hand, for 1.75% < ρb < 2.7%, the use of the 100 ksi limitation over- estimated the capacity of the section by only a marginal amount (about 2.5%), which was considered insignificant for design purposes. Through a series of moment-curvature and deflection analyses, Mast el al. demonstrated that a simple beam designed 8

using 100 ksi steel at the tension-controlled strain limit of 0.0066 exhibited ductility behavior (as measured by steel strain and section curvature) similar to that exhibited by a 60 ksi design having a strain limit of 0.005. They demon- strated that the ratio of nominal to service deflections was indeed greater for the higher strength steel reinforced sec- tions. In addition, due to the higher tension strain in the high-strength reinforcement under service loading condi- tions, the beams may exhibit larger crack widths than if rein- forced with conventional steel. However, as shown in Mast et al. (2008), previous testing indicates that the measured crack width under service loading conditions is only slightly larger than the (so-called) acceptable crack widths for beams reinforced with conventional steel. It is proposed that since some high-strength steels have improved corrosion resist- ance, the increased crack widths may be acceptable as long as these are not aesthetically objectionable. Based on this work, Mast et al. proposed variation of the flexural resistance factor, φ, between 0.90 and 0.65 at strain limits greater than 0.009 and less than 0.005, respectively. These limits correspond to the tension-controlled limit of 0.005 and compression-controlled limit of 0.002 presently used for 60 ksi steel in AASHTO (2007). To help prevent compression- controlled failure, they suggest providing compression rein- forcement having a design yield strength, fy < 80 ksi. This limit is based on the maximum stress that can be developed at a strain of 0.003, which is the ultimate concrete strain at the extreme compression face of the concrete beam. A number of experimental studies (Seliem et al. 2006, McNally 2003, Malhas 2002, Vijay et al. 2002, Florida DOT 2002) of the flexural behavior of members reinforced with A1035 reinforcing steel support the conclusions of Mast et al. (2008). These studies all indicate that flexural members designed using the same simplified approach (i.e., elastic- perfectly plastic steel behavior at higher values of fy) will have flexural characteristics comparable to members having con- ventional reinforcement grades. Where reported, cracking and deflections at service loads are only marginally greater when using A1035 steel. One study (McNally) indicates a reduction in overall ductility when using an earlier formula- tion (since changed) of A1035 reinforcement. Other studies (Seliem and Florida DOT) report a marked increase in duc- tility likely resulting from the lower reinforcement ratio that may used in conjunction with the high-strength flexural reinforcement. 1.3.3.1 Applications in Bridge Decks Most extant applications of A1035 steel have been in bridge decks and its use is typically as a one-to-one replacement for less corrosion-resistant “black” steel. Bridge deck design is based more on serviceability criteria than on strength require- ments; therefore, it is not unexpected that experimental inves- tigations of A1035-reinforced decks exhibit no significant differences in behavior (particularly under service loads) compared to A615-reinforced counterparts (Rizkalla et al. 2005, Hill et al. 2003). 1.3.4 Shear Reinforcement The shear behavior of reinforced-concrete beams is not well understood and calculation of the shear strength is based on semi-empirical relationships. As a result, the cal- culated shear strength can vary significantly (up to 250%) among different code approaches (Hassan et al. 2008). Sim- ilarly, it is unclear whether current design approaches for shear may be extended to members having high-strength steel reinforcement. One concern is whether the high stress levels induced in the reinforcement may cause excessive cracking in the concrete resulting in degradation of the con- crete component of shear resistance. Sumpter (2007) sought to determine the feasibility of using high-strength steel as shear reinforcement for concrete members, particularly focusing on the member behavior under overload condi- tions where the steel experiences high stress levels. Sumpter reports tests of beams having shear span to depth ratios of approximately 3 alternately reinforced with A615 or A1035 longitudinal and transverse steel. Stirrup spacings used reflected the minimum and maximum permitted and an additional intermediate spacing between these limits. Due to the stiff nature of shear-critical sections, little differences between specimen behaviors were noted at service loads. As expected, observed capacity of these shear-critical members reflected the amount of shear reinforcement present. Mem- bers having A1035 shear reinforcement exhibited marginally greater capacity than those with A615 shear steel. Sumpter concludes that most observed behavior was dominated by concrete behavior and that stress in the shear reinforce- ment in any specimen never exceeded 80 ksi; thus, the high- strength steel (fy > 100 ksi) was not fully utilized, whereas the 60 ksi steel was. A study reported by Florida DOT (2002) draws the same conclusions with respect to the stress that may be developed in shear reinforcement. Sumpter also reports that all shear crack width values at service loads were less than the ACI-implied limit for flexural cracking of 0.016 in., regardless of the reinforcement grade or details. Indeed, Sumpter reports smaller crack widths in comparable members having high-strength steel than those with conven- tional steel. He attributes this behavior to enhanced bond characteristics of A1035 steel resulting from differences in rib configuration. This conclusion is curious because there is typically no difference between the rib configuration of A615 and A1035 reinforcing steels, and Sumpter does not report a difference in his test program. 9

1.3.5 Compression Members The concept of providing transverse reinforcement to concrete compression members is intended to improve the strength and ductility. As a concrete column is compressed axi- ally, it expands laterally. This lateral expansion is resisted by the transverse reinforcement and a lateral confining pressure in the concrete core is developed. Concrete strength and deformabil- ity are enhanced by the resulting state of multi-axial compres- sion (Richart et al. 1928 and countless researchers since). Current design philosophy for compression members equates the expected loss of axial load carrying capacity due to cover spalling to the expected strength gain of the remain- ing core due to the presence of confining reinforcement. This approach was developed and calibrated for columns fabricated with what today may only be considered normal, or moderate strength, concrete (fc′ < 8000 psi) and normal strength confin- ing steel (fy = 60 ksi). There is a perceived need for greater confinement for high-strength concrete than what is required for normal strength concretes (ACI 363 1992). Strength and deformability of concrete are known to be inversely propor- tional; therefore, more confinement is required in order for high-strength concrete columns to reach levels of deformation expected of well-detailed normal-strength concrete columns. In general, the degree of improvement in both axial capacity and ductility due to the provision of confinement is inversely proportional to the unconfined concrete strength (Pessiki et al. 2002, Carey and Harries 2005). The use of high-strength transverse reinforcement represents one manner by which this additional confinement may be realized. Confining pressures are generated from tensile forces in the transverse reinforcement that result from lateral expansion of the axially loaded concrete. As the lateral expansion is depend- ent on the mechanical properties of the concrete, the lateral strains, particularly in high-strength concrete, may be insuffi- cient to engage the higher confining pressures made possible by the use of high-strength transverse reinforcement (Martinez et al. 1982, Pessiki and Graybeal 2000). An additional, related consideration is that the transverse strains that engage the confining reinforcement must be limited to ensure continued resistance to shear. The maximum permitted transverse strain in this regard is often reported as 0.004 (Priestley et al. 1996). Previous research offers differing conclusions with respect to the use of high-strength transverse reinforcing steel. Ahmed and Shah (1982) demonstrated analytically that high-strength transverse reinforcement may enhance the ductility of a col- umn while having little effect on its strength. Martinez et al. (1982) propose limiting the strength of transverse reinforce- ment, based on their results showing that the higher steel strength was not fully utilized. Pessiki and Graybeal (2000) also conclude that the yield capacity of high-strength transverse reinforcement cannot be developed. Polat (1992) reported that ductility and strength enhancements were less than propor- tional to the strength of the transverse confining steel. Mugu- ruma et al. (1990) demonstrated very high axial ductilities using high-strength transverse reinforcement and reported yielding of transverse reinforcement having yield strengths of 198 ksi shortly after the peak axial load is achieved. Yong et al. (1988) observed two peaks on their axial load-deformation responses; the high-strength transverse reinforcement did not yield initially but had yielded at the second peak. Mugu- ruma et al. (1991) suggest that high-strength transverse rein- forcement offers better control of longitudinal bar buckling than normal strength confining steel. Cusson and Paultre (1994) report improvements in strength and ductility due to high-strength confining steel only for well-confined columns. Improvements in axial column behavior with high-strength transverse reinforcement have also been reported by Bjerkeli et al. (1990), Nagashima et al. (1992), Razvi and Saatcioglu (1994), and Nishiyama et al. (1993). Studies that have specif- ically used A1035 transverse reinforcement have provided similar conclusions (Restrepo et al. 2006, Stephan et al. 2003, El-Hacha and Rizkalla 2002). 1.3.6 Bond and Development Bond characteristics of ASTM A1035 reinforcing bars should not be expected to differ significantly from those of conventional reinforcing steel grades since neither the steel modulus nor bar deformations differ (Ahlborn and Den Hartigh 2002, Florida 2002). Studies that have reported load- slip relationships for A1035 steel have not concluded that these differ in any significant manner from similar relation- ships established for A615 bars (Ahlborn and DenHartigh 2002, El-Hacha et al. 2002 and 2006). Limited evidence (Sumpter 2007 and Zeno 2009) suggests modestly improved bond behavior that is believed to be associated with the rib geometry resulting from the rolling of the tougher A1035 material. Nonetheless, this effect is modest and cannot be generalized across material heats. Due to the higher bar stress to be developed, A1035 bars require a longer development length (ldb). However, simply increasing development length without providing confine- ment is an inefficient means of developing greater stresses (Seliem et al. 2006 and 2009, El-Hacha et al. 2006). With long development (or splice) lengths, the bond stress at the “front” of the development length is exhausted before the bond stress along the entire development length can be developed (Viwathanatepa et al. 1979). Confining reinforcement around development regions or splices is required to control the splitting cracks associated with a bond failure (Seliem et al. 2009). With higher strength steel, greater bar strain and slip will occur prior to development of the bar. The associated displacement of the bar lugs drives the splitting failure beyond that where yield of conventional bars 10

would occur; thus, confining reinforcement is critical in devel- oping higher strength bars. Seliem et al. (2009) assessed the present empirical develop- ment length equations prescribed by ACI 318 (2008) and ACI 408 (2003) when applied to developing A1035 bars. ACI 318 was found to underestimate development length requirements when no confining reinforcement was present and was only marginally improved when confining reinforcement was used. The ACI 408 recommendations were found to be adequate whether confinement was present or not. Present AASHTO requirements were not assessed although these can be shown to result in comparable development lengths to the require- ments of ACI 408 in cases where confinement is present. Peter- freund (2003), in a study of A1035 reinforcement for bridge decks (#4 and #5 bars only), concluded to the contrary, that is, the ACI 318 requirements for development length were ade- quate to develop A1035 bars with no confining reinforcement present. However, in his study, Peterfreund used the simplified ACI equation which results in development lengths almost twice as long as the more rigorous approach used by Seliem et al. and others. Seliem et al. recommended that confining reinforcement always be used when developing A1035 or other high-strength reinforcing steel. 1.3.6.1 Development of Standard Hooks Ciancone et al. (2008) evaluated the behavior of standard hooks made using #5 and #7 A1035 steel. No confinement reinforcement was provided in the specimens. While the #5 hooks were able to develop their yield capacity of 100 ksi, the #7 hooks were not. This result suggests an effect of bar size and supports the need for confining reinforcement when develop- ing A1035 bars. 1.3.7 Serviceability Considerations A fundamental issue in using A1035 or any other high- strength reinforcing steel is that the stress at service load (fs; assumed to be on the order of 0.6fy) may be greater than with conventional Grade 60 steel. Consequently, the service load reinforcing bar strains are greater (i.e., εs = fs/Es). This larger strain impacts deflection calculations and crack con- trol parameters. Regardless of this discussion, as discussed previously, most studies of members reinforced with A1035 steel exhibit serviceability performance, as measured by both deflections and crack widths, similar to that of members re- inforced with A615 bars. 1.3.7.1 Deflection Calculations Deflection of reinforced-concrete flexural members is most typically determined using an equivalent moment of inertia in an equation for elastic deflection. Both AASHTO (2007) and ACI 318 (2008) prescribe Branson’s Equation (Branson 1963) to determine an equivalent moment of inertia (Ie) of a cracked concrete section as follows: Where: Ig = moment of inertia of gross concrete section; Icr = moment of inertia of fully cracked concrete section; Mcr = moment to cause cracking; Ma = applied moment at which Ie is calculated; and m = factor as defined below. Setting m = 4 accounts for tension stiffening effects at the critical section along a span, while calculations are conven- tionally made setting m = 3 to reflect the “average” stiffness across the entire span. Equation 1 is found to be generally sat- isfactory for beams having typical amounts of non-prestressed reinforcement; indeed this equation was originally calibrated based on beams having a reinforcement ratio of ρ = 0.0165 (Branson 1963). The value of Ie calculated using Equation 1 is only slightly smaller than Ig in cases where Ma is only margin- ally larger than Mcr. This case generally happens in members having a low reinforcement ratio, typically ρ < 0.006. For such members, the calculated value of Ie is very sensitive to changes of Mcr (Gilbert 1999). Thus, Equation 1 may overestimate the effective moment of inertia for lightly reinforced flexural members having an Ig/Icr ratio greater than 3 (Scanlon et al. 2001, Bischoff 2005, and Gilbert 2006). As ρ decreases, Ig/Icr increases exponentially and Ma/Mcr decreases. The result is that the effective moment of inertia, Ie, is overestimated on the order of 200% when ρ = 0.007 but by only about 10% at ρ = 0.025 (Nawy and Neuwerth 1977). Bischoff (2005) reports that Branson’s Equation underestimates short-term deflection for concrete members when the reinforcing ratio is less than approximately 1% and the Ig/Icr ratio is greater than 3. Several attempts have been made by different investigators to modify Branson’s Equation, aiming to improve the accu- racy of the predicted deflection (Grossman 1981, Rangan 1982, Al-Zaid et al. 1991, Al-Shaikh and Al-Zaid 1993, Fikry and Thomas 1998). With the exception of Rangan (1982), none of these modifications has been adopted into building codes; Branson’s Equation remains the standard calculation for computing effective moment of inertia. The following two approaches have been proposed to modify Branson’s Equation to address its efficacy when used with lower reinforcing ratios: • Introduce a coefficient, β, into the first term of Equation 1 to modify Ig (Gao et al. 1998). β is less than unity and is calculated based on reinforcing bar modulus (for I M M I M M I Ie cr a m g cr a m cr= ⎛⎝⎜ ⎞⎠⎟ + − ⎛⎝⎜ ⎞⎠⎟ ⎡ ⎣⎢ ⎤ ⎦⎥ ≤1 g ( . )Eq 1 11

softer reinforcing materials such as FRP) (Theriault and Benmokrane 1998, Masmoudi et al. 1998) or reinforcing ratio relative to the balanced ratio (i.e., ρ/ρb) (Yost et al. 2003). The latter approach is necessary when considering high-strength steel reinforcement. • Adjust the exponent m (Dolan 1989) as a function of the reinforcing ratio (Toutanji and Saafi 2000; Al-Zaid et al. 1991) or simply increase the value of m (Brown and Bartholomew [1996] propose m = 5). Other methods involving finding an effective modulus of the beam have been proposed by Murashev (1940), Rao (1966), and CEB-FIP (1993). Finally, approaches involving integrating curvature along a beam have been proposed by Ghali (1993), Toutanji and Saafi (2000), Rasheed et al. (2004), and Razaqpur et al. (2000). Bischoff (2005), in addition to providing a thorough review of all deflection investigations briefly summarized above, pro- poses a method of calculating the effective moment of inertia at a section that better captures the effects of tension stiffening particularly for “soft” sections having low reinforcing ratios. This method is summarized in Equation 2. Where: n = modular ratio Es/Ec and ρ = reinforcing ratio. Moment-curvature relationships may then be predicted using M = EcIeφ. The derivation of Equation 2 is presented in Bischoff (2005) and is further shown to be essentially equiv- alent to the Murashev (1940) equation, of which the Branson Equation is a simplification. 1.3.7.2 Crack Control The traditional “z-factor” or Gergely-Lutz (1968) approach of directly assessing cracking behavior of concrete beams was dropped by ACI 318 in 1999 and by AASHTO in 2005 in favor of a simplified version of the alternative approach proposed by Frosch (1999 and 2001) that prescribed spacing limits for longitudinal reinforcing steel thereby indi- rectly controlling crack width. The empirically tuned Gergely- Lutz approach was considered inadequate to address cases k n n ncr = ( ) + −ρ ρ ρ2 2 I k n k bdcr cr cr= + −( )⎡⎣⎢ ⎤ ⎦⎥ 3 2 3 3 1ρ η = −1 I Icr g I I M M e cr cr a = − ⎛⎝⎜ ⎞⎠⎟1 2 η ( )Eq. 2 having very large concrete cover (ACI 224 2001). Addition- ally, Beeby (1983) showed no conclusive evidence linking reinforcement corrosion with crack width while Poursaee et al. (2010) show that a crack as small as 0.004 in. acts as a free surface with respect to water ingress. Despite the latter asser- tion, the simplified versions of the Frosch approach adopted by AASHTO and ACI implicitly assume a maximum crack width of 0.017 in. which was also the value assumed for exterior expo- sure conditions when applying the Gergely-Lutz approach prior to 1999. The ACI 318 version of the Frosch equation for determin- ing the maximum spacing of flexural reinforcement to affect adequate crack control is as follows: Where: cc = minimum concrete cover measured to center of rein- forcing bar closest to the extreme tension face and fs = service load stress in reinforcing bar closest to the extreme tension face. Equation 3 may be rewritten in terms of reinforcing bar strain (εs), assuming the material obeys Hooke’s Law, and cal- ibrated for any desired crack width (w) (Ospina and Bakis 2007), as follows: Thus, the relationship between crack width, reinforcing bar strain, and longitudinal bar spacing required to control cracking is demonstrated in a relatively simple format consis- tent with present design practice. The relationship is material independent, only assuming a linear behavior is present. Available data comparing the cracking behavior of steel and FRP-reinforced members confirm the implications of this approach (e.g., Creazza and Russo 2001, Bischoff and Paixao 2004). Ospina and Bakis conclude that the use of Equation 3 is valid, if not conservative, for beams having large elastic reinforcing bar strains. AASHTO Equation 5.7.3.4-1 (AASHTO 2007) takes the same form as the ACI equation, as follows: Where: dc = minimum concrete cover measured to center of rein- forcing bar; s f d f dd s s c s c≤ − ( )700 2γβ in ksi; in inches (Eq. 5) s w c w w s c s ≤ ⎛⎝⎜ ⎞⎠⎟ − ≤ ⎛⎝⎜ ⎞⎠⎟1 15 2 5 0 92. . .ε ε in inches (Eq. 4)( ) s f c f f s c s r ≤ ⎛⎝⎜ ⎞ ⎠⎟ − ≤ ⎛ ⎝⎜ ⎞ ⎠⎟15 40 000 2 5 12 40 000, . , in psi; in inches Eq. 3cc( ) ( ) 12

fs = service load stress in reinforcing bar; h = overall depth of the concrete section. This equation can therefore also be rearranged in a man- ner similar to Equation 4, resulting in the same conclusions and implications. For Class 1 exposure, Equation 5 is calibrated, through γd = 1, for a crack width of 0.017 in.; for Class 2 (γd = 0.75) or other exposures, the de facto crack width is γd × 0.017. In the commentary to §5.7.3.4, AASHTO describes the use of the γd term to calibrate Equation 5 for any desired crack width limitation. It is well established that crack control is improved by using a larger number of well-distributed, smaller diameter bars to make up the required area of flexural reinforcing steel. The number of bars that can be provided, however, is restricted by minimum spacing requirements (AASHTO LRFD §5.10.3). Thus, if the area of flexural steel is provided using the greatest number of bars that may be placed in a section, such a section, theoretically, should exhibit the best control of crack widths. This relationship is manifested in Equations 4 and 6 where the crack width (w) is propor- tional to the flexural bar spacing (s). Ward (2009) shows that for fs = 36 ksi (appropriate for bars having fy = 60 ksi), Class 1 and 2 exposure crack width limits (0.017 in. and 0.0128 in., respectively) are met for all permissible designs (see Appendix I). As seen in Equations 4 and 6, crack width (w) is also pro- portional to reinforcing bar stress (or strain, in this case, εs). Therefore, if fs = 60 ksi (appropriate for bars having fy = 100 ksi), crack widths are expected to increase. In this case, Ward (2009) shows that while the Class 1 exposure crack width limit (0.017 in.) is met for all practical beam design cases, the Class 2 limit (0.0128 in.) is generally only met with #5 bars and smaller (see Appendix I). The impli- cation of this is that accepted crack width limits may not be met with higher permitted reinforcing bar stress. Ward (2009) proposes the following two alternatives to addressing crack control for beam design in the context of AASHTO LRFD (§5.7.3.4): • Limit fs ≤ 50 ksi in order to satisfy present Class 2 require- ments; or • Limit fs ≤ 60 ksi and remove the Class 2 limit when consid- ering high-strength reinforcing steel. s w dd s s c≤ ⎛ ⎝⎜ ⎞ ⎠⎟ ⎛⎝⎜ ⎞⎠⎟ −1 34 2. γ β ε (Eq. 6) βs c c d h d = + −( )1 0 7. ; and Indeed, AASHTO LRFD Commentary C5.7.3.4 provides the following third alternative: The crack width is directly proportional to the γd factor, there- fore, if the individual Authority with jurisdiction desires an alter- nate crack width, the γd factor can be adjusted directly. Thus a value of γd > 1.0 may be permitted as appropriate. This approach is additionally deemed to be appropriate for deck slabs since the value of fs will be appreciably lower. It is also noted that deck slabs designed using the empirical approach of AASHTO LRFD §9.7.2 are not required to sat- isfy §5.7.3.4. 1.3.7.3 “Acceptable” Crack Widths ACI Committee 224 (2001) suggests that crack widths exceeding 0.016 in. may be unacceptable from the standpoint of aesthetics. Similarly, Halvorsen (1987) states that a case could be made that crack widths ranging from 0.006 to 0.012 in. could be considered unacceptable for aesthetic reasons as they are visible to the naked eye; hence, generating a sense of insecurity about structural distress. Beyond this, there is little consensus as to acceptable crack widths. 1.3.7.4 Analytical Assessment of Crack Widths Soltani (2010) conducted a detailed analytical assessment of expected crack widths. This approach accounted for non- linear stress transfer between the bar and surrounding con- crete along the development length and nonlinear bar slip relationships associated with the stress transfer. Soltani con- sidered a range of bar sizes and reinforcement ratios and used experimentally determined R-O stress-strain relationships to model the steel reinforcement. Figure 2 provides a represen- tative result showing anticipated average crack widths at the location of the reinforcing steel for a concrete tension zone having a reinforcing ratio of 2%. Soltani concluded that through reinforcing bar stresses of 72 ksi, average crack widths (it is only possible to consider average crack widths in an analytical context) remain below 0.016 in. for all but the largest bars considered (#10). The results were relatively insensitive to changes in reinforcing ratio. Finally, it is noted that crack widths expressed at the surface of a concrete mem- ber may be amplified from those at the reinforcing bar loca- tion due to the depth of concrete cover and/or the curvature of the member. 1.3.8 Corrosion Performance of Reinforcing Steel Grades The quantification of corrosion resistance is beyond the scope of the present work but is summarized here in the 13

interest of completeness and because enhanced corrosion resistance is a major factor behind the drive to adopt A1035 reinforcing steel. A1035 steel is a microcomposite Fe-C-Cr- Mn alloy that has an average chromium content of approxi- mately 9%, which is too low to be referred to as “stainless steel” (Cr > 10.5%) but sufficiently high to impart a degree of corrosion resistance when compared to “black steel” as represented by A615 or A706. A large number of studies have compared the corrosion resistance of A1035 steel with that of A615 and A706 black steel and A955 austenitic (304 and 316), duplex (2101), and ferritic stainless steels. Generally, the relative performance of these materials in terms of their corrosion resistance is ranked from most to least susceptible to corrosion in the order indicated in Table 2. Thus, micro- composite alloys tend to be 2 to 10 times more corrosion resistant than black steel while austenitic stainless steel may be a few orders of magnitude improved. A summary of cor- rosion performance of reinforcing steel is presented in Appendix A and provides quantitative data available in the literature. 1.4 Survey of Use of High-Strength Steel Reinforcement in Bridge Structures A written survey intended to assess the current practice and the use of high-strength reinforcing steel was disseminated in June 2007. In all, 65 surveys were distributed to U.S. state DOTs, Canadian Ministries of Transportation (MOTs), and a few other agencies. A copy of the survey instrument and “raw” responses are provided in Appendix J. Thirty-two surveys were returned—a response rate of 49%. Of these, 27 (84% of those returned) report no use of “steel reinforcement (not prestressing rods or tendons) with specified yield strengths greater than 60 ksi” (Question #1). The primary reason for not utilizing high-strength reinforce- ment (Question #1a) was not that it was not permitted per se but simply has not been used (15 of 27 respondents answer- ing “no” to Question #1). Despite this response, some respondents went on to cite the prohibition by AASHTO on reinforcing steel having strengths greater than 60 ksi (9 of 27 respondents). Additionally, five responding jurisdictions stated their specifications specifically “prohibit yield strengths above 60 ksi.” It is not clear whether the prohibition cited by these latter respondents is a specific prohibition or simply a pro- hibition by exclusion (such as, reinforcing steel strength not to exceed 60 ksi . . . ). In one case, the jurisdiction specifically requires the use of ASTM A706 Grade 60 reinforcing steel. Eleven of 27 respondents identified the lack of “data on per- formance to satisfy our performance requirements.” One such response specifically cited concerns about “strength, ductility for seismic [loading] and, to a lesser extent, weldability.” 14 Figure 2. Theoretical average crack widths for tension zone having   0.02 (Soltani 2010). Material Performance, where A615 = 1.0 A706 black steel 0.5 – 0.8 A615 black steel 1.0 A1035 microcomposite alloy 2 – 10 A955 2101 duplex stainless steel 2 – 10 A955 304 austenitic stainless steel >10 A955 316 austenitic stainless steel >20 Table 2. Relative corrosion performance of reinforcing steel grades.

Of the five jurisdictions reporting use of high-strength reinforcing steel, three report only the use of steel up to 75 ksi, and two report use of steel having a yield strength greater than 100 ksi (Question #2). Respondents indicated that high- strength reinforcement is not excluded from use in any appli- cation and, indeed, has been applied in all applications cited in Question #3 except “spirals in piers.” One jurisdiction reporting use of steel having fy greater than 100 ksi reports its use as only “main flexural reinforcement in beams,” although it is apparently “permitted” elsewhere. The second such juris- diction reports its use as only slab reinforcement. The reasons for incorporating this steel (Question #4) are reported as being to “improve durability by enhancing corrosion resist- ance of reinforcement.” Both jurisdictions having used fy greater than 100 ksi in flexural applications report this use as being on an “experimental/trial” basis. Three jurisdictions report fewer than 10 structures having high-strength steel reinforcement while one reports between 10 and 50 struc- tures (Question #5). Design using high-strength reinforcement was facilitated by the engineer of record’s best judgment (Question #6). Three of the five respondents reported simply using AASHTO design methods for 60 ksi reinforcement and replacing the steel, one-for-one, with high-strength steel bars. In one response, high-strength steel is simply used in place of 60 ksi steel for areas requiring corrosion resistance; nonetheless, increased lap lengths are prescribed in this case. A comment from this respondent follows (identifying information has been removed): [This jurisdiction] has mainly used [A1035] steel to aid with corrosion. We have been very conservative with its usage. Usu- ally designing as if we are using 60 ksi rebar or in some instances more. Then we will use longer development lengths to assist with ultimate capacities. Codes [are] not fully written to use such high-strength rebar properties; therefore [this jurisdiction] hasn’t generally designed for maximum strength usage in bars. How- ever, we do look at ultimate bending capacity and increase laps if deemed prudent. We have built about five slab bridges using [A1035] entirely. We have used it in some bridge decks—three to seven. Then we use it in our P/S girders for shear reinforcing [bars] in the ends of girders and as shear reinforcement between the girders and the bridge deck—substitute it for 60 ksi rebar— assuming 60 ksi properties. No problems or impediments to design were reported by any respondents (Questions 7 to 9). One respondent specifi- cally stated that they were satisfied with the results versus cost of high-strength steel when compared to 60 ksi although they had concerns over crack width with higher yield strength (paraphrased by authors of this report). We advertised one project where high-strength bars [A1035] were allowed as an alternative to epoxy-coated rebar. The con- tractor elected to use epoxy-coated bars. We would only allow the rebar to be designed for up to 75 ksi, until AASHTO has spec- ifications to account for higher strength bars. 1.4.1 Survey of Use of Stainless Steel Reinforcement in Bridge Structures A similar, abbreviated survey (Appendix J) addressing the use of stainless steel was also conducted. In this case, 28 responses were received. Thirteen jurisdictions reported the use of stainless steel reinforcing bars; in all but one case for slab reinforcement and in most cases on an experimental basis. Design for stainless steel bars was apparently a one-for- one substitution for conventional reinforcing bars. 1.4.2 Reported Use of A1035 Reinforcing Steel in Highway Bridge Infrastructure MMFX Inc., the only supplier of A1035 reinforcing steel, reports 25 U.S. and 4 Canadian jurisdictions that have used A1035 reinforcing steel in at least one bridge project as of December 2009. Most applications have been bridge decks. According to MMFX, most applications are simply one- to-one replacement of A615 with A1035 in order to take advantage of the improved corrosion resistance of the latter. Nonetheless, there are 17 known projects where a value greater than fy = 60 ksi was used in design; these are listed in Appendix J. Design values of fy of 75, 80, and 100 ksi are reported, although most were 75 ksi (thus, presumably tak- ing advantage of the upper limit on fy prescribed by AASHTO specifications). Cross-referencing existing projects with the survey indicate that most projects were experimental “demonstration” proj- ects. It is further noted that not all jurisdictions reported to have erected structures returned the survey. Additionally, five jurisdictions reporting in the survey no projects with high performance steel are revealed by the project list to have, in fact, one or two existing demonstration projects. 15