Seismic Evaluation of Tower Components for Long Span Bridges

Seismic Evaluation of Long Span Bridge Tower Components

Ahmad M. Itani[1], Ph.D., P.E.

Abstract

Recent earthquakes exposed the vulnerability of bridge towers in transferring the loads during earthquakes. This paper discusses the seismic evaluation of tower components for three California long span bridges. Large-scale experiments were conducted on the components of these towers to evaluate their performance, ultimate strength and failure mode under large cyclic deformations. One-half scale riveted laced members with end restrained were tested under cyclic axial deformations. These members are similar to the tower members of the San Francisco-Oakland Bay Bridge and the Richmond San Rafael Bridge in the California Bay Area. Experimental results showed that the lacings play an important role in connecting the components of the built-up laced member. However, due to the flexibility of the lacings, the built-up member did not act as a one unit, causing it to buckle under less than expected buckling load.

Introduction

The towers of the SFOBB play a significant role in the overall seismic response and performance of the bridge. They are responsible of transferring the inertia forces that are generated during an earthquake in the superstructure to the foundation. Therefore, the components of these towers lie in the direct seismic load path and any premature failure in these members will interrupt the load path and might cause inadequate behavior.

To ensure adequate seismic performance, the components of the towers should be able to withstand severe deformation without experiencing early fracture or premature instability. The uncertainties that are generally associated with seismological and geotechnical studies make it extremely difficult to determine, with certainty, the magnitude of forces that should be resisted by the tower members. Therefore, limit state procedure provides a reliable alternative method in determining the seismic response of the tower. This procedure relates the ultimate capacity of the system to its individual member deformation capacity. Based on this, the rotational, tensile, and post buckling capacity of the tower diagonals should be known in order to determine the overall seismic response of the towers.

Axial Capacity of Built-up Sections

Bracing built-up members with various configurations were tested by Goel et al. [2, 3, 4, 5, and 6] in the last twenty years to determine their cyclic behavior. It was concluded from these tests that built-up members have the tendency of not working as a whole unit and usually experience individual buckling when the main section components are not adequately connected to each other. The results of their investigations were adopted by the LRFD Specification [7] to ensure that built-up section behave as solid sections. Two criteria were established for this purpose:

1. Compression members composed of two or more shapes shall be connected to one another at intervals such that the slenderness ratio of either shape does not exceed three fourth of the governing slenderness ratio of the built-up member

1. 
   If buckling involve relative deformation that produce shear forces in connectors between individual shapes, KL/r should be replaced by (KL/r)m:

Where

(KL/r)o is the column slenderness of built-up member acting as a unit

a/ri is the largest slenderness of individual components

(KL/r)m is the modified column slenderness of built-up member

a is the distance between connectors

ri is the radius of gyration of individual components

Figure 1 shows buckling around the Y-Y axis for a laced member. This mode buckling involves relative deformation between the two solid flanges of the section. If these flanges were not adequately connected, the section will experience individual buckling by bending around the weak local axis of each flange.

Objectives and Scope

The main objectives of this study are:

• Determine the cyclic behavior of built-up laced members with their end connections.

• Evaluate the ultimate tensile/compressive capacity and post-buckling behavior of laced members.

To achieve these objectives, experimental testing was conducted on one-half scale, built-up, laced specimen with its end conditions. The prototype of the specimen was the diagonal member of Pier E9 of the SFOBB Eastern Spans. Pier E9 is an anchor pier that resists longitudinal forces of the five, simple-span trusses from E4 to E9 and the two, simple-span trusses from E9 to E11. It is a rigid steel tower with double cross bracing, as shown in Figure 2.

Selection of Test Specimen

Member B9WG in Bent E9W (transverse bent of Pier E9) was chosen as a representative member of tower diagonals. This member is located at the upper left of the Bent E9W. It is connected to the upper beam TS9WA and tower leg TSW9D at the upper end and to the cross diagonal member B9WH as shown in Figure 2.

The gross and net area of member B9WG is equal to 52.5 in2 and 48 in2, respectively. The member is fabricated from Copper Bearing Carbon Steel [10]. According to the Design Specifications of SFOBB [11], the carbon steel should have minimum and maximum ultimate strength equal to 62 ksi and 70 ksi, respectively while the minimum yield point is equal to 37 ksi. The Carnegie Steel Company in Duquesn, PA, manufactured the steel that was used on the Eastern Spans [10]. Coupon tests conducted on the steel showed that the yield point varied between 37 ksi to 40 ksi, while the ultimate strength varied between 63 ksi to 69 ksi for the rolled angle sections. The yield point varied between 33 ksi to 39 ksi and the ultimate strength varied between 65 ksi to 69 ksi for the plates.

Therefore, based on these values the tensile capacity of B9WG member would be approximately 2100 kips, based on gross section yielding and 3312 kips, based on net section fracture. These values showed that it was neither feasible nor economical to perform full-scale, cyclic testing. Reducing the dimensions of B9WG member by one-half would reduce the force by one–fourth, which will make the tensile capacity of the member approximately 525 kips to 828 kips. Therefore, one-half scale testing of B9WG prototype was adopted in this study. Tables 1 and 2 list the section properties of the one-half scale section of the laced member and its individual components.

Ultimate Capacity of Test Specimen Using Current Code Procedures

The ultimate capacity of B9WG specimen with its end connections is divided into member and connection capacities. The LRFD equations were utilized in calculating these capacities.

Axial Capacity

• Tension yielding on gross area and tension fracture on net area

Yielding on gross area:

Fracture on net area

• In-plane and out-of-plane global buckling

In-plane global buckling around X-X (Section behave as solid section):

Out-of-plane buckling around Y-Y axis (Buckling involves relative deformation that produces shear forces in the connectors between individual shapes:

• Buckling of Individual Component

Buckling of individual component around local YY-YY, occurs between the lacing bars when:

Therefore, buckling of the member individual components will occur around the local YY-YY, because the section will not behave as an integral unit. The buckling capacity is equal to:

Segment buckling between stay plate and top connection and between stay plate and bottom connection; assuming the effective length factor is 1.25:

Therefore, the ultimate member buckling capacity is expected to be equal to 230 kips with bending around bottom connection of the laced member.

Test Set-up for B9WG Specimen with End Connections

The test set-up that was used to test the B9WG specimen, with its end connections, consisted of: two-hinge loading frame, end brackets, actuators/control, and instrumentation/data acquisition system. To maintain the geometrical configurations of the B9WG prototype, the test specimen was connected to two end brackets. The specimen, with the end brackets, was connected to the two-hinge loading frame and the fixed reaction blocks, as shown Figure 4. The in-plane kinematics of the two-hinge loading frame simulates the transverse movement of Pier E9. Therefore, imposing lateral displacement on the two-hinge loading frame; will subject the specimen to axial forces and end moments, according to the end rigidities of the double gusset plates.

Cyclic Testing of Built-up Laced Member

Cyclic displacement was applied to the two-hinge frame through its upper pin. As a result, the lacing member experienced axial push-pull displacement, which creates axial force and end moments due to connection flexural rigidities. The cyclic displacement consisted of stretching the specimen and then compressing it to cause buckling. The number and the amplitude of these cycles are usually called deformation history. This history governs the amount of deformation that is introduced in the specimen in each cycle and, thus, dictates the fatigue life of the metal.

The deformation history used in this study was based on the lateral displacement response of SFOBB bents. The time history analyses that were conducted by Caltrans have shown that the SFOBB steel towers have drifts range between 1 to 2% of the tower height. The 1% drift results in 4.6” of axial displacement in B9WG member. The yield displacement of B9WG member is equal to:

Therefore, the diagonal member will be subjected to axial deformation between 9 to 18 y. The yield displacement of B9WG specimen is equal to:

The cyclic deformation usually starts with tension excursion to reduce the initial crookedness of the member and to determine experimentally the yield displacement.

Cyclic Response of “As-Built” B9WG Specimen

The loading began with small, cyclic displacements, in the elastic range to ensure that the actuators and the instrumentation were working properly. The tensile displacement was increased incrementally to determine, experimentally, the yield displacement. At an axial displacement equal to 0.42” and corresponding load of 275 kips, the white wash at the ends of the member started to flake; therefore, it was decided to consider this value as the y for this specimen.

Figure 5 shows the hysteretic behavior of the test specimen. The specimen was subjected to one cycle equal to 0.5 y and 1.0 y. During the compression excursion of 1.0 y, at 0.18” and compressive load equal to 225 kips, a sudden local out-of-plane buckling occurred at the two end connections of the specimen. The local buckling initiated in the zone between the stay plates and the gusset plates. As a result, the compression capacity of the specimen dropped from 225 to 100 kips.

After the local buckling, two additional tensile excursions were conducted, because it was believed that no useful information could be obtained in the compression excursions. As the tensile load increased, it was apparent that some localized yielding was taking place in the buckled area causing the fracture of one rivet at the top member connection. Before damaging the member much further, it was decided to stop the test and modify the end conditions of the member to prevent the premature connection failure.

Due to member out-of-plane buckling, large localized strain concentrations resulted in two problems that required repair. Some rivets that connected the angles to the stay plate of the laced member were damaged. Furthermore, the outstanding leg of the laced member angle, experienced partial fracture around the sheared rivet.

End Retrofit of B9WG Specimen

Since the member did not suffer any damage due the cyclic testing, it was decided to retrofit the end connections and continue the test. The retrofit strategy was to extend the stay plates inside the gusset plate connections to force the laced member to act as one unit in that zone. It consisted of bolting a combination of angles, shim plates, and extension plates onto the stay plates and the ends of the member. ASTM A307 bolts, with threads excluded from the shear plane, were used to bolt the angles and extension plates onto the member and connection.

Cyclic Response of “Retrofitted” B9WG Specimen

Figure 6 shows the hysteretic behavior of the retrofitted specimen. During the second compression excursion, out-of-plane buckling occurred in the bottom connection at an axial load equal to 175 kips. Upon increasing the cyclic displacement, the severity of the out-of-plane buckling became more pronounced. Edge buckling of the bottom gusset plate caused rapid degradation of the compressive capacity of the member.

During the tension excursion of the fourth cycle, few rivets that connect the member’s angles to the top gusset plate sheared off. The axial load was equal to 323 kips. Shortly thereafter, the rest of the rivets connecting the splice plate to the gusset sheared off, causing the member to pull away completely from the gusset plate. This caused the complete failure of the member, since the load dropped instantly to zero.

Analysis of Test Results for “As-Built” and Retrofitted B9WG Specimen

Testing of “As-Built” and retrofitted B9WG specimen showed the consequences of the individual component buckling in the built-up section. The termination of the stay plate before the double gusset plate connection caused the individual components of the laced member to act separately. Buckling around the local YY-YY axis supports this conclusion. The spacing between the stay plate and the end of the laced member is equal to 13.1” and 9.63” for the bottom and top connection. The slenderness ratio of the individual components of the laced member was equal to 48.2 and 35.2 at the bottom and top connections, receptively. As mentioned earlier, that segment buckling might occur between the laces and at the top or bottom connection. The calculated load of segment buckling was equal to 278 kips, 246 kips, and 230 kips. The measured load during testing ‘As Built” B9WG was equal to 225 kips.

The current AISC LRFD Specifications [7] states in E4 under “Built-Up Members”:

“Individual components of compression members composed of two or more shapes shall be connected to one another at intervals, a, such that the effective slenderness ratio Ka/ri of each of the component shapes, between the connectors, does not exceed three-fourths times the governing slenderness ratio of the built-up member. The least radius of gyration ri shall be used in computing the slenderness ratio of each component part.”

“Open sides of compression members built-up from plates or shapes shall be provided with continuous cover plates perforated with a succession of access holes. As an alternative to perforated cover plates, lacing with tie plates is permitted at each end. Tie plates shall be as near the ends as practical.”

Therefore, the AISC specifications address the need of the stay plates to be extended to the ends of the member. In order to prevent any individual bulking in B9WG:

Where “a” is the distance between the connectors of the two segments, ri is the minimum radius of gyration of individual component, and KL/r is governing slenderness ratio of the built-up member. Therefore, the distance a should be:

Therefore, individual buckling continued to occur even when the stay plates were extended to the ends of the gusset plate connections.

The extension of the stay plates and the angles inside the gusset plate connection in retrofitted B9WG specimen has shifted the location of the bending to inside the gusset plate. Shifting the inelastic activity into the gusset plates exposed the edge-buckling problem in gusset plate connections. The edge compactness ratio of the bottom gusset plate was equal to 80, which is more than two times the value allowed by AASHTO.

The top gusset plate connection fractured at tensile load equal 323 kips. The ultimate shear capacity of the top connection was calculated to be equal to 683. The individual buckling of the laced member caused the top half of the member to engage before the bottom half causing the capacity to be reduced to one-half.

Due to the premature buckling at the end connections of the laced member, the member stayed intact and was not subjected to significant yielding. The first test showed low values of strains, because the inelastic activity concentrated in out-of-plane bending at the ends of the laced member. The second test showed localized yielding and flexure behavior at the bottom gusset plates. The maximum reported strain reached up to 4% at SG74, which is due to the out-of-plane the edge buckling of the bottom gusset plate.

The test results showed that the individual buckling of the laced member and the out-of-plane buckling of the gusset plates dominated the cyclic response of the built-up laced member. Extending the stay plates all the way through the double gusset plate connection did control the individual buckling of the laced member at the connections. The individual components of the built-up section buckled around YY-YY axis limiting the capacity of the built-up section to less than 40% of the code predicted value.

Based on the cyclic testing, the following conclusions may be drawn:

• The laced member suffered premature buckling at its end connections at a load less than 40% of its code predicated capacity.

• The AISC LRFD relationships between individual component buckling and section buckling can be used to predict the ultimate capacity of laced members.

• Edge buckling and out-of-plane buckling of gusset plates presented a failure mode for laced members. The end connections of the laced member were the weakest link in the system.

• Individual buckling of the laced member caused a differential distortion in the built-up section, thus, dropping the tensile capacity of the section by almost one-half.

• The stay plates of the laced member should extended as much as possible inside the double gusset plate connection.

References