Fatigue Performance of High-Frequency Welded Steel I-Beams

FATIGUE TESTING SETUP AND PROCEDURE

A 110-kip servo hydraulic actuator was used to load the simply supported beam specimens in four-point bending as illustrated in the load frame setup of figure 4. Beams were cyclically loaded with a sine wave function at rates between 1.5 and 1.75 Hz. A minimum load between 3.5 and 4 kip was used during cycling to prevent shifting and any unwarranted dynamic effects from occurring during testing. The maximum beam displacement was monitored throughout the test using a linear variable differential transducer within the actuator. Limits were set such that the actuator would turn itself off when the beam displacement increased 0.02 inch more than when the limit was turned on. Generally, it was possible to find budding cracks near the web–flange junction within the constant moment regions. Once cracks were discovered and marked, beams were cycled to failure. Typically, less than 5,000 additional cycles were enough at this point for a crack to extend well into the web of the beam specimen or for stability issues to occur due to asymmetrical loading caused by the ever-increasing loss of specimen cross section at the crack locations.

The first three beams tested were instrumented with strain gauges to ensure symmetrical loading. Strain gauges were placed on both sides of the web at three separate locations along the inside of the tension flange in the constant moment region and at midspan on the inside of the compression flange. Strains on average agreed with strains calculated using linear elastic beam theory.

The beam supports were 6 inches wide in the direction of the beam length, resulting in a span of 138 inches between support centers ( figure 4 ). For the first beam test (i.e., beam 1), the span between the centers of the spreader beam load bearings was 48 inches. However, due to initial imperfections in the specimens, web slenderness, and an insufficient amount of torsional stiffness in the spreader beam, there sometimes were elastic stability issues that caused the actuator to move out-of-plane. To combat this, lateral braces at the beam supports were constructed using a steel bar and C-clamps as shown in figure 5. Wooden stiffeners were used at the supports to provide additional torsional and overturning stiffness at these locations as can also be seen in the same figure. A lateral torsional support was provided by bracing the beam compression flanges at midspan using a slender steel plate strip that was clamped to the compression flange on one end and to a load frame column on the opposite end as shown in figure 6. After testing of beam 1 was complete, the center-to-center distance between spreader beam load bearings was decreased to 40 inches to further remedy the stability issue encountered.

When conducting the first two beam tests (i.e., beams 1 and 2), the spreader beam bearings were detailed, as shown in figure 7, by clamping the specimen between two thick plates. Clamping the plates prevented the loading points from moving under the cyclic loading, and wooden stiffeners were used to prevent distortion to the beam from the clamping force. This setup proved adequate for testing beam 1. However, during testing of beam 2, an audible squeaking developed at one of the spreader beam load bearings on several occasions from slight movement between the specimen tension flange and the bottom clamping plate. The squeaking was eliminated by tightening the high-strength rods, thus increasing the clamping effect. This eventually led to a fretting failure of beam 2 as will be discussed later. Since fretting was undesirable, the load-bearing setup was simplified for beams 3 through 6. The revised setup eliminated the wood stiffeners, high-strength rods, and lower clamping plate and instead modified the upper plate so it could directly clamp to the specimen compression flange as seen in figure 6.