U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-14-066    Date:  September 2014
Publication Number: FHWA-HRT-14-066
Date: September 2014


Fatigue Testing of Galvanized and Ungalvanized Socket Connections


Fatigue testing was performed in a specially fabricated load frame shown in figure 2. The methodology was based on a concept first performed at UT-Austin by bolting two pole specimens to a loading box and then simply supporting the ends of poles and loading through the loading box.(6) For this project, the loading box was replaced with a large concrete block. The block had dimensions of 45 by 45 by 72 inches and was cast of concrete, thus weighing just over 12.6 kips. Six 2-inch by 4.5UNC B7 threaded rods passed through the entire length of the block and stuck out approximately 12 inches on each side. These threaded rods were used to bolt the pole specimens to the loading block with double-nut moment joints. The leveling nuts were spaced approximately one rod diameter off the concrete surface. The benefit of using the concrete block to load the poles was to accurately recreate the boundary conditions typical for socket connections bolted to a concrete foundation. The threaded rods were orientated such that one rod was at the point of maximum bending stress. Specimens were randomly selected for installation, and the installation orientation was also randomized, that is, no attempt was made to orient the tube seam weld into a beneficial configuration. Since the socket connections had a six-bolt pattern and the support ends had a four-bolt pattern, the specimen could be installed in only one of two orientations.


During fatigue cycling the actuator would pull up on the loading block to induce a dead load stress in the pole. The actuator would cycle in between loads of 11.63 and 15.73 kips. This equated to a calculated mean (dead load) of approximately 19.3 ksi because the round and round-like poles had slightly different moments of inertia. This load range induced a stress range of 5.85 ksi in the fabricator 1 specimens and 5.73 ksi in the fabricator 2 specimens. The specimens were cycled either in load or displacement control. When running in load control, the actuator could not cycle faster than 0.8 Hz without going unstable. To expedite testing, most cycling occurred in displacement control with the actuator cycling at about 2.7 Hz, which was near the natural frequency of the system. The peak displacement targets were attained when the actuator had come to operating temperature and running in load control. Once a day, the displacement targets were verified and altered accordingly. However, once one of the two cycling specimens started to grow a crack, the test was run exclusively in load control to ensure a consistent stress range. Throughout the testing program, two poles never simultaneously reached failure; one pole always failed before another.

The specimens were cycled at a low stress range to ensure the threaded rods were below their fatigue threshold. If the threaded rods had failed, a new loading block would have to be fabricated. This was the fundamental drawback to using this concrete loading block versus the steel loading box used in the UT-Austin research.


Strength tests were performed on eight of the cracked pole sections. This was done to assess the remaining capacity of the cracked section, for instance, as a way to determine the risk associated with keeping a knowingly cracked pole in service. The same loading system was used in the strength tests as in the fatigue tests, except the tests were run monotonically until failure.

Two parameters were investigated in the strength tests: loading rate and temperature. The loading rate was analyzed in terms of the applied displacement from the actuator. The first two tests were conducted at 0.002 and 0.04 inches/s at room temperature. The first represented a true static loading rate and the second represented a rate closer to that of a wind loading event on a pole. The second two tests were conducted at 0.02 and 0.2 inches/s at -30 °F, which is the AASHTO Zone 2 lowest anticipated service temperature. To attain -30 °F, a rigid foam box was built around each connection and injected with liquid nitrogen through a solenoid valve controlled by a temperature controller and a thermocouple attached to the base plate of the connection. The difference in the loading rates was accidental. The first tests were performed using old analog controllers, while the second tests were run with digital controllers, which were upgraded through the duration of the project. The different loading rates were an artifact of not understanding the new digital controller functionality. All the poles tested statically were from fabricator 2.

The actuator had only a 6-inch stroke. In the tests to failure, the system had to be unloaded frequently, spacers had to be added at the support, and the system reloaded until failure occurred. Failure was defined as attaining a peak load or a fracture of the cross section.


Tensile coupon testing was used to characterize the steel plate used to make the specimen supplied by each of the fabricators. Testing was performed according to the ASTM E8 Specification.(10) The coupons were machined according to the schematic shown in figure 3. All pretest measurements and markings were also performed in accordance with ASTM E8.

Testing was performed on a four-post universal testing machine with hydraulic wedge grips and a 220 kip capacity. The testing machine was controlled with a digital controller. Strain measurements were made using a clip-on extensometer with a 1-inch measurement range but fitted with an extension bar to measure strain over an 8.000-inch gauge length.

Specimens were initially loaded at a rate of 0.0003 inch/s. Once the specimen had yielded, the static yield was attained by pausing the loading for a period of 90 seconds in three locations along the yield plateau. Once strain hardening had begun, the loading rate was gradually increased to 0.01 inch/s until the specimen fractured. Typically, it took approximately 20 min to fracture a specimen.

This illustration shows a three view drawing of the load frame. The 77 kip actuator hangs off a double C-channel section that is bolted to two columns. The concrete loading block sits between the two columns. The critical dimensions being conveyed in the drawing are 216 inches from the center of the actuator to the roller support at the end of one pole, 174 inches from the center of the roller support to the fillet weld of the socket connection. In addition, the concrete block is shown to be 72 inches long with a 45-inch square cross section.
Figure 2 . Illustration. Schematic of socket connection testing frame.

This illustration shows the dogbone shaped specimen used for tensile testing. The specimen is doubly symmetric, though reduced in width in the middle of the specimen. Specifically, the specimen is 18 inches in total length, with the ends being 2 inches wide. The middle of the specimen is 1.500 inches wide, transitioning to 1.506 inches wide at a distance of 4.5 inches from the center. There is a 1-inch radius fillet that transitions from the 1.506-inch width to the 2.00-inch end width.
Figure 3 . Illustration. ASTM E8 "plate-type"


Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101