Behavior of a Steel Girder Bolted Splice Connection

Introduction

The American Association of State Highway Officials and later the American Association of State Highway and Transportation Officials (AASHTO) specifications have required for many years that all steel splices and connections be designed for the average of calculated stress and strength of the member but not less than 75 percent of the strength of the member. This approach is straightforward when applied to axial truss members where the stress is equal in the various components of the member. However, such an approach does have its disadvantages. For example, application of this approach to a composite steel girder is generally more complex because the stresses in the flanges are not equal, and the distribution of the stress in the web is a function of the loads applied to the composite and noncomposite sections. In most designs of steel splices and connections, member strength controls the design because the designer usually intentionally places the splice in a low-moment region near the point of dead load inflection. The present traditional method for designing bolted splices in these regions uses a combination of designing for 75 percent of the strength of each flange at a minimum and then designing the web for a design moment determined based on the corresponding design stress in the flange with the largest design stress. The resulting connection design is tedious and can result in a large number of bolts in the web connection.

Experimental research by Ibrahim at the University of Texas showed that a simpler method of design produced a connection with adequate design capacity.⁽¹⁾ Rather than trying to proportion the primary bending moment to flanges and web, this method suggests decoupling the design so the flanges resist the primary bending moment and the web resists shear force. An ad hoc group consisting of representatives from AASHTO, steel bridge design consultants, a steel bridge fabricator, and representatives from the Federal Highway Administration (FHWA) was formed in 2015 to implement the findings of Ibrahim and develop a simpler design method that could be proposed to AASHTO.⁽¹⁾ The simpler method included designing the bolted flange and web splice connections for 100 percent of the individual capacities of the flange and web; the flange splices were designed for the yield capacity of the flange, and the web splices were designed for the shear capacity of the web. Therefore, the method satisfied the AASHTO design criteria because the web and flange splices have strengths equal to the design strengths of the respective components. No further check of the shear capacity of the connection was required; however, additional forces in the web connection may need to be considered if the flanges are not adequate to develop the factored design moment.

To better understand the ramifications of implementing the simpler design method, the ad hoc group worked with 11 different continuous span arrangements and 2 different girder spacings and compared the bolted splice designs using the 2 design philosophies. The ad hoc group felt detailed finite element modeling on the most extreme outlier connection that had the largest disparity in the required number of bolts would lend credit to the simpler design philosophy. The work reported herein only covers the finite element modeling effort in support of the ad hoc group recommendations.

A ballot to revise the language covering splice design in the AASHTO Load and Resistance Factor Design (LRFD) Bridge Design Specifications was presented to the AASHTO Subcommittee on Bridges and Structures (SCOBS) in June 2016.⁽²⁾ A concurrent ballot was passed at the 2016 SCOBS meeting that revised the shear strength of high-strength bolts to align with the latest specification of the Research Council of Structural Connections.⁽³⁾ The revised shear strength of high-strength bolts was not originally considered when performing the finite element analysis for the ad hoc group. All references made throughout this report to the current design method are those published in the seventh edition of the AASHTO LRFD Bridge Design Specifications; references to the new method refer to those that appear in the eighth edition of the AASHTO LRFD Bridge Design Specifications.^(2,4)

Example Splice Connection

The outlier bolted splice was from a three-span continuous girder with 234-ft long back spans, a 300-ft long center span, and a girder spacing of 12 ft. The girder was designed using ASTM A709 grade 50 steel and ⁷/₈-inch-diameter ASTM A325 bolts.^(5,6) An 8-inch concrete deck was used in the design. At the spliced section, the girder flange and web plates had dimensions as shown in table 1.

A design program was used to iteratively proportion the girder and select the optimum location for the splice, which was 64 ft away from the first interior support in the first span. The unfactored moments required for AASHTO design at the splice location from the continuous girder analysis are as shown in table 2. However, the factored shear demand was only 853 kip based on an assumed shear panel dimension ratio (d_o/D) of 3.0 used in the program. The author herein was most interested in pushing the demand as far as possible for the connection, and the factored shear force was reevaluated for a d_o/D of 2.0, thus resulting in a factored shear of 1,312 kip.

Comparison of Splice Designs

This report does not delve into the detailed calculations of each design method; only the end result is discussed herein. A side-by-side comparison of the current and new design methods is shown in 0 and 0. The design methods resulted in no difference between the number of bolts in the flange splices; however, there was a significant difference between the required bolts for the web splices, with the new method requiring only two rows of bolts at a much larger spacing. Overall, the new method required 104 fewer bolts than the current method.

Source: FHWA
Figure 1. Illustration. Detailing of the field splice using current method.

Source: FHWA
Figure 2. Illustration. Detailing of the field splice using new method.