CROSSINGS Practical Information For The Bridge Industry
LFD vs. LRFD Whats loads to satisfy specified structural performance
requirements. A structural performance require- Up With the Letter R ment indicates what is required from a bridge at a given load level. With properly selected multiples Anyway? of the load, LFD can ensure a design allowing: 1. the expected number of passages of ordinary By Michael A. Grubb, P.E. vehicles during the life of the bridge, Over the past three decades, different disci- 2. occasional passages of overload vehicles without plines of structural engineering practice have been permanent damage, and gradually following a trend toward design for max- 3. in an extreme emergency, very few passages of imum strength under factored loads. In line with exceptionally heavy vehicles. The underlying this growing trend, two ultimate strength or limit philosophy is to ensure both safe and serviceable states design approaches have evolved to improve performance, while at the same time providing a steel-bridge design by providing more uniform lev- consistent live-load carrying capacity for all els of safety than are possible using the more tradi- bridges on the system. In Allowable Stress tional allowable or working stress design approach. Design (ASD), attention is focused on perfor- The first of these limit-states design approaches mance under service conditions only. LFD con- (LFD) was developed in the 1960s. The second siders performance in a broader context in that limit-states approach was introduced to bridge it deals with serviceability and safety separate- design in the early 1990s (LRFD) and parallels a ly. reliability-based approach that has been available LFD recognizes three basic and distinct load lev- for the design of steel building frames since 1986. els Service Load, Overload, and Maximum Load. This article examines the basic philosophy of each Service Load represents ordinary vehicles that of these design approaches in an attempt to eluci- may operate on the highways without special per- date the important differences and similarities mit. For design purposes, Service Load is repre- between the two methods. sented in AASHTO as the sum of the dead loads D and the standard live loads plus impact L+I. The Load Factor Design (LFD) primary structural performance requirements at In the mid 1960s, an advisory committee was Service Load are to provide adequate fatigue life formed by the American Iron and Steel Institute and to control live-load deflections and concrete (AISI) to review bridge-design practices and devel- deck cracking. If the design is adequate for fatigue op new design recommendations that would yield a and deflection under normal traffic loads, the more consistent and effective use of steel in high- absolute maximum stress due to these loads way bridges. The efforts of this committee resulted used for design in ASD is of little concern. in the publication in 1969 of the Tentative Overload is defined as the maximum live load that Criteria for Load Factor Design of Steel Highway can be allowed on the structure on infrequent occa- Bridges or AISI Bulletin No. 15. After a year of sions. Infrequent implies that the stresses caused study and some modifications, the tentative crite- by these loads are not subject to fatigue require- ria were adopted by AASHTO in 1970 as an alter- ments. The single structural performance require- nate method and published in the 1971 AASHTO ment at Overload is control of permanent deforma- Interim Specifications. Since that time, the use of tions caused by localized yielding and connection Load Factor Design (LFD) for steel bridges has slip to ensure good riding quality. For design pur- continued to increase; it is estimated that LFD is poses, Overload is taken as the load factor D times being used, either all or in part, by at least 40 the dead load plus the load factor L times the live State DOTs. load plus impact. The load factor D is to allow for possible increases in the dead load and is usually In recognition of the inherent ductility and taken as 1.0 on the assumption that the designer reserve strength of steel and an improved under- will allow for future additions to the dead load. standing of the structural behavior of steel bridges, The load factor L allows for possible overloads and LFD was developed as a method for proportioning is usually taken as 5/3 or 1.67 for live loadings structural members for multiples of the design
Reprinted from Modern Steel Construction
greater than or equal to AASHTO H20 loading. were approved by AASHTO for use as alternative specifications to the AASHTO Standard Service Load and Overload address serviceabili- Specifications for Highway Bridges, which contain ty requirements. To ensure adequate safety, the both the ASD and LFD provisions. The LRFD Maximum Load level is introduced. The single specifications were developed in response to a high structural performance requirement at Maximum level of interest amongst the members of the Load is that the bridge be able to safely resist the AASHTO Subcommittee on Bridges and Structures load. In LFD, this performance requirement is sat- in developing updated bridge specifications along isfied at Maximum Load through the following with an accompanying commentary. The goal was relationship: to develop more comprehensive specifications that (Maximum strength) > [DD + L(L+I)] would eliminate any gaps and inconsistencies in the Standard Specifications, incorporate the latest The load factor recognizes uncertainties that in bridge research, and achieve more uniform mar- exist in the loads and load analysis. The resistance gins of safety or reliability across a wide variety of factor represents several sources of uncertainty structures. The decision was then made to develop such as variations in materials and section size, these new specifications in a probability based variations in workmanship, and approximations LRFD format. made in strength calculations. In LFD, a value of equal to 1.0 was selected for members in flexure In the LRFD method, load and resistance factors and shear since the maximum strength equations are determined through statistical studies of the in LFD for flexure and shear represent the lower variability of loads and resistances, which is con- bounds of the test data. Lower values of are sidered to be a more realistic approach than the specified for column and connection design because application of judgment-based deterministic fac- of the greater consequences of failure of these ele- tors. In the calibration process, load and resis- ments. tance factors are calculated to provide a target level of reliability for a wide variety of structure For flexure and shear design in LFD, is shifted types and configurations. to the right-hand side of the preceding equation. The resulting / term, together with the load fac- The reliability theory on which the LRFD tors D and L, establishes the margin of safety method is based has been well-documented else- inherent in LFD for flexural members. The value where and will not be expanded on in depth here. of the / term was established based on past expe- Essentially, the level of reliability is measured rience using ASD practice as a guide. The safety of through the use of a reliability index. Although the ASD approach has been well established, but not strictly correct, the reliability index can be the live-load margin of safety is known to vary thought of in simple terms as a statistical indicator with the span because a single safety factor of of the fraction of times that a particular design cri- 1/0.55 or 1.82 is applied to both dead and live loads teria will be met or exceeded over the design life of in ASD. The minimum margin of safety in ASD is the structure. For example, according to this sim- associated with short spans. Therefore, it was ple definition, a reliability index of 3.5 indicates decided that in order to provide both safe and eco- that a particular design criteria may be exceeded nomical designs in LFD, a value of / would be in 2 out of 10,000 cases. The reliability index is selected that would yield the same steel section by currently based only on the design of individual ASD and LFD for a short simple-span bridge. It components of the bridge and does not represent a was determined that a value of / equal to 1.3 system reliability, which will typically be higher. would yield about the same minimum level of safe- In the calibration, a target reliability index is ty by ASD and LFD for an approximately 45-foot selected to provide a minimum acceptable margin long noncomposite simple-span bridge. As the of safety. Following the lead of LFD, past practice span length increases, the live-load margin of safe- was used as a guide in establishing the target reli- ty increases slightly in LFD since different load ability index. The primary goal during the devel- factors are applied to the dead and live loads, while opment of the LRFD specifications was not to the margin of safety remains nearly constant in cause a radical departure from the basic level of ASD. Since is equal to 1.0 for flexural members, safety inherent in the current highway system. does not explicitly appear in the strength equa- Rather, the primary objective was to increase the tions for flexural members in LFD. uniformity of the margin of safety across the various structure types that are utilized on Load and Resistance Factor Design the system. (LRFD) Reliability indices were calculated for a number In 1993, AASHTO adopted the Load and of sample bridge designs extracted from plans sup- Resistance Factor Design (LRFD) specifications for plied by State DOTs using: bridge design, which were developed under 1. the AASHTO HS live loading and lateral distri- NCHRP Project 12-33. The LRFD specifications bution factors given in the Standard Specifications in conjunction with the current While more specific differences between the load and resistance factors specified in LFD, and AASHTO LRFD and LFD specifications such as 2. a new HL live-load model and new lateral distri- differences in the live-load models, impact factors, bution factors introduced in the AASHTO LRFD lateral distribution factors, load combinations and specifications in conjunction with new load and the design for fatigue and their effects on the resistance factors determined from the calibra- overall design could be discussed and debated at tion process. some length, the purpose of this discussion is to emphasize that the differences in the basic under- The LFD designs were clustered around a relia- lying philosophies of the two methods are not all bility index of 3.5 with a large amount of scatter. that great. The primary difference in philosophy The use of the selected LRFD load and resistance boils down to the procedures used to perform the factors in conjunction with the new live-load model calibration in order to provide the minimum and distribution factors again resulted in a cluster- desired level of safety. LRFD calibration proce- ing of the indices around the target value of 3.5, dures allow for an improvement in the uniformity but with a greatly reduced amount of scatter indi- of the margin of safety across the system and also cating the attainment of a more uniform reliability provide a more realistic and rational framework for than provided by LFD procedures. performing future calibrations as more is learned about loads and material resistances. While proba- Similarities and Differences bilistic theories are employed in the LRFD calibra- Both LFD and LRFD are limit-states design tion process, it should be kept in mind that the approaches that strive to achieve more uniform user of the LRFD specification provisions need not live-load margins of safety for steel bridges, while be well-grounded in probability theory in order to still meeting established structural performance apply the provisions. Most of the LRFD resistance criteria for serviceability and safety. While tradi- equations for the design of steel-bridge components tional ASD considers performance under service are in fact very similar to the resistance equations conditions only, LFD and LRFD treat serviceability given in the current LFD provisions. Thus, design- and safety separately. In LFD, limit-state criteria ers who are proficient with LFD procedures for are specified to satisfy performance criteria at steel bridges should have little trouble converting three distinct load levels: Service Load, Overload to LRFD, once some level of familiarity is attained and Maximum Load. In LRFD, similar limit-state with the revisions to the load side of the basic criteria are specified to satisfy similar performance design equation that are presented in the LRFD objectives at four distinct limit states: the Service specifications. Limit State, the Fatigue and Fracture Limit State, the Strength Limit State and the Extreme Event Michael A. Grubb, P.E., is a Senior Steel Bridge Limit State. Design Specialist in the Pittsburgh office of HDR Engineering, Inc. Prior to 1994, he served as the Load and resistance factors are specified in each Assistant Manager of Bridge Engineering for AISC method to account for various sources of uncertain- Marketing, Inc. ty. In each method, lower load factors are applied to the dead loads. The values of the load and resis- tance factors are different in the two methods. Also, in LRFD, a single load factor is applied to each load component rather than applying sepa- NATIONAL STEEL BRIDGE ALLIANCE rate D and L factors. The resistance factors are One East Wacker Dr., Suite 3100 always explicitly applied in LRFD, while the resis- Chicago, IL 60601-2001 tance factor of 1.0 for flexural members is implicit- ph: 312/670-2400 fax: 312/670-5403 ly applied in LFD. In the LRFD specifications, an attempt is also made to treat redundancy, ductility The mission of The National Steel Bridge Alliance (NSBA), and importance more explicitly in the design by which was formed in 1995, is to enhance the art and applying subjective modifiers to the load side of the science of the design and construction of steel bridges. Its equation. activities include organizing meetings, conferences and national symposia, conducting the Prize Bridge Awards In the LRFD specifications, the load and resis- competition, supporting research, developing design aids, tance factors are determined from a probability- and providing assistance to bridge owners and designers. based calibration process to achieve a more uni- The NSBA membership includes representatives from all form reliability index for the various components of aspects of the steel bridge industry. the system than LFD. In the LFD specifications, the load and resistance factors are determined using a simpler calibration process based on judg- ment and experience to achieve a more uniform live-load carrying capacity than is possible using ASD.