Effect of Earthquake Duration on Reinforced Concrete Bridge Columns
AuthorMohammed, Mohammed S.
StatisticsView Usage Statistics
Recent earthquakes in Chile (2015, 2014, and 2010), Japan (2011), China (2008), and Indonesia (2004) are reminders of the possible occurrence of a long-duration, large-magnitude earthquake in the Cascadia subduction zone along the Pacific Northwest Coast of the United States. Although the duration of an earthquake is expected to affect the response of structures, current seismic design specifications use response spectra to identify the hazard and do not consider duration effects. A few reasons contribute to this situation. One is that earthquakes with long durations are rare in practice. Most events are shorter, and duration for these events does not present design issues not already captured in the response spectrum. Also, research to date has not presented a coherent and persuasive case that duration should be included. Results have supported both sides of the case, and not left provision specification writers with a compelling case for adding duration. The available records from recent long-duration earthquakes, such as 2011 Tohoku earthquake in Japan and 2010 Maule earthquake in Chile, made it possible to better study the topic.Experimental program and analytical investigations were conducted in this study to investigate the influence of earthquake duration on structural response of reinforced concrete (RC) bridge columns. The main objectives were to quantify the effect of duration on collapse capacity and to recommend whether this effect should be included in seismic design provisions. The first phase of the analytical investigation, i.e. the pre-test analysis phase, used an OpenSees model that was calibrated against previous experimental results for a Column designated as NF-2. This model was used to select the experimental program ground motions. Five ground motions were selected for the five identical specimens, four long-duration motions and one with a short-duration. The main objective of the pre-test analysis was to choose motions with different durations that have similar or close response spectra. These motions were selected with a target to impose maximum displacement demands on the specimens that are less than the maximum displacement capacity, 9.8 in. (13.6 % drift ratio), previously achieved from Column NF-2. Imposing less displacement demands than the column capacity allowed for studying the effect of the increased number of displacement cycles resulting from long-duration motions. Additionally, experimental fragility curves were developed using data from over 25 bridge column models that had been tested on shake tables or under lateral quasi-static loads. The developed experimental fragility curves were used along with the pre-test OpenSees analysis to predict the expected damage for the specimens before testing.The experimental program consisted of two phases. In phase I, three columns were tested (Columns LD-J1, SD-L, and LD-C1) on a shake table using ground motions from the 2011 Tohoku (long-duration), 1989 Loma Prieta (short-duration), and 2010 Maule (long-duration) earthquakes. The response spectra of the three motions were modified to match a target response spectrum. Therefore, the three motions had the same acceleration response spectra but different durations. The difference in the collapse capacity between the three specimens could be attributed to the different ground motion durations used. In phase II, two additional columns were tested. The fourth specimen in the program and the first in phase II was tested using a long-duration motion from the 2011 Tohoku, Japan earthquake (Column LD-J2). The motion was not modified, but it was selected because its response spectrum was very close, around the specimen’s fundamental period, to the final motion response spectra used for the first three columns in phase I. Comparisons could be made between the results of the four specimens because the motions had similar or close acceleration response spectra. The fifth and final specimen was tested using a long-duration motion from the 2010 Maule, Chile earthquake (ColumnLD-C2). The response spectrum was similar to the previous ones at the specimen’s fundamental period but with higher spectral accelerations at the long periods. The reason of selecting this motion was to investigate the effect of ground motion duration with poor soil conditions. All five columns began with 100% of the selected motions then followed by an aftershock. The main motions were then incrementally amplified until failure (125%, 150%, ..., and so on).In the post-test analysis, two OpenSees models (Model I and Model II) with different assumptions were utilized. Including low-cycle fatigue in OpenSees Model II was the basic difference between the two models. Both models were calibrated against the experimental test results. Incremental dynamic analysis (IDA) was employed using different sets of spectrally equivalent long and short-duration motions. The motions were scaled until failure occurred in the column. Three ground motion sets were used for comparison purposes; one long-duration and two short-duration sets. Each set consisted of 112 ground motion records. For each of the 112 records in the long-duration set, a corresponding short-duration record that has a closely matching response spectrum was selected. For the first short-duration set, amplitude matching was used, where the short-duration motion was scaled by a factor to match the response spectrum of the corresponding long-duration motion. In the second short-duration set, spectral matching using wavelets was utilized. The preservation of frequency content was quantified by correlating the wavelet coefficients of the original and the modified time series. Comparative collapse analysis was achieved by developing collapse fragility curves using different spectral acceleration intensity measures and the different sets of ground motions. The experimental and analytical studies concluded that ground motion duration has a significant effect on the collapse capacity of RC bridge columns. Columns subjected to long-duration motions were damaged more than those with spectrally equivalent short duration motions. The reinforcing steel in the columns subjected to longer motions had a significantly larger number of plastic strain cycles, which caused the bars to fracture at lower displacements compared to the columns subjected to short-duration motions; i.e. ground motion duration affected the displacement collapse capacity of bridge columns. The shake table tests showed that spectral accelerations at collapse for columns subjected to long-duration motions were lower by 21% to 29% than the column subjected to short-duration motion. The geometric mean of the displacement capacities of the long-duration specimens was 32% lower than the maximum displacement capacity of the short-duration specimen. The post-test analysis showed a reduction in spectral accelerations at collapse for long-duration columns by about 20% compared to columns subjected to short-duration motions. A reduction in the displacement capacity of about 25% was also found. Therefore, it is recommended that duration effects be considered in the seismic design of bridges, especially in regions where long-duration earthquakes may occur. Based on the work in this study, a preliminary design recommendation is proposed: consider the displacement capacity of RC bridge columns lower by 25% for locations where long-duration earthquakes are expected.