You are on page 1of 14

LARGE SCALE TESTING OF BRIDGE ABUTMENT BACKFILL SOIL 2010 Ashley Oertel; University of California, Santa Barbara REU

site: University of California, Los Angeles; REU Advisor: Anne Lemnitzer

Table of Contents Abstract ........................................................................................................................................... 3 Introduction ..................................................................................................................................... 3 Methods........................................................................................................................................... 4 Results and Discussion ................................................................................................................... 7 Acknowledgements ......................................................................................................................... 8 Appendix ......................................................................................................................................... 9 A.1) Additional data.................................................................................................................... 9 A.2) Backwall and Reaction Block Failure Images .................................................................. 10 A.3) Sand Cone Testing ............................................................................................................ 11 A.4) Backfill Soil Lab Analysis ................................................................................................ 12 References ..................................................................................................................................... 14

LARGE SCALE TESTING OF BRIDGE ABUTMENT BACKFILL SOIL 2010 Ashley Oertel; University of California, Santa Barbara REU site: University of California, Los Angeles; REU Advisor: Anne Lemnitzer Abstract Bridge abutments typically contain a backwall that is designed in such a way that it will break free of the base when struck by the bridge deck during an earthquake. The backwall will be pushed into the backfill soil, which will resist displacement. A full height bridge abutment was built and tested. Boundary conditions on the wall allowed for only horizontal displacement. The backfill soil that was used was silty sand that is typical of California bridge design practice. Six large actuators, each affixed with an LVDT sensor pushed against the backwall to induce horizontal displacement while minimizing vertical displacement. This test was a repeat of a previous test. The goal was to determine how the height of a backwall affected the resistance the backfill soil provided. In the previous test, data had been flawed. This test was to correct the errors, and to show that a taller backwall will have a much higher maximum resistance than the shorter backwall previously tested. Introduction Bridges contain a deck and an abutment. The abutment, shown in Figure 1, is placed at either end of the bridge in order to provide support and absorb motion due to loading. It consists of a backwall, wingwalls, and backfill soil. In the event of an earthquake, a bridge deck may strike against the backwall. The backwall is designed to break free from the supports and push into the backfill soil. The backfill resists lateral displacement which, ideally, prevents extensive damage to the deck. The properties of the backfill soil, including water content and how densely it is packed, play a large role on the effect it has on the lateral motion of the bridge deck [Dicleli, 2010]. Large scale tests are performed in order to fully see the effects the backfill plays on the horizontal displacement of the backwall. The NEES@UCLA Figure 1: Bridge abutment large scale abutment test is constructing a full-height bridge abutment and applying a horizontal force in order to analyze the effects of the abutment on the horizontal displacement of the backwall, and also the fail properties of the backfill soil. Actuators prevent vertical and rotational motion since testing and post-earthquake bridge analysis show that displacement in these directions is negligible [Lemnitzer, 2009]. However, other sources claim that while the consequences of tilting and rotating of abutments often get overlooked, they do in fact largely impact the structure [Al-Homound, 1999]. There is obviously a strong need for research in this area, as increased knowledge of the relevance of rotational and vertical motion would impact the way testing of other bridge components are performed. A test similar to the current project was performed at the same site, underneath the I-405 and I-105 interchange, in 2006. In that test the backwall was only five and a half feet tall. The

data collected from this test allowed the stiffness of the backfill soil to be calculated [Stewart, 2007]. Later, in 2009, the test was repeated in order to determine the effect using an eight foot backwall would have. It was predicted that the eight foot wall would lead to the soil having a much higher maximum passive resistance. The data showed that the maximum passive resistance increased as desired, but the stiffness of the soil decreased. This shouldnt have happened as the soil properties hadnt significantly changed. Due to this questionable data, NEES@UCLA is performing a replica of the test that was performed in July, 2009. The 2010 test is also being conducted in the same site as the previous tests. Since, the project is aided by Caltrans, the specimen constructed follows Caltrans specifications for bridge abutments [California, 2006]. During the testing, the backfill soil is added in layers and compacted. The density of the packing and water content of small samples are collected with each layer in order to understand how the properties of the backfill soil affect the results of the testing. In addition to large scale testing, numerical modeling and small scale testing has been performed [Ellis, 2001]. While these methods provide good estimates, computer and numerical modeling can be a bit too ideal for testing something so large with so many factors affecting it. These methods are based on information that is assumed, and often cant show what is actually occurring. For example, in numerical modeling the fail plane of the backfill was often modeled as a linear surface until large scale testing of the abutment showed that it was in fact a log-spiral surface [Shamsabadi, 2007]. The previous version of the current UCLA large scale abutment test noted a log-spiral fail surface, and also observed multiple other fail surfaces as well [Lemnitzer, 2009]. Methods Testing took place at a Caltrans test site near the I-405 and I-105 junction. The specimen created for testing was a concrete block representing a bridge backwall which was 2.6m high by 4.5m wide and 0.9m thick. The specimen was full scale in height. Side panels were built to simulate wingwalls and contain the backfill and the remainder of the dug-out earth is filled with soil to prevent lateral displacement and buckling of the wingwalls. As mentioned, the backwall specimen is full scale in height, however, it is small scale in width since a full scale model would have to be built wide enough to support at least six lanes of traffic. The smaller scale Figure 2: Construction of wing walls width used in the test might lead to more effective friction forces, so the wingwalls were lined with plastic foil to reduce the friction on the backfill soil. This is shown in Figure 2. The ends of the wingwalls were approximately 0.15m away from the prototype backwall in order to ensure no direct contact between the two elements occurred during testing. This distance is also normal Caltrans practice.

The backwall specimen was 2.90m from the reaction block. The reaction block has dimensions of 7.30m x 3.65m x 1.83m and has two supporting columns that extend 13.72m below ground. Six actuators were placed in the space between the reaction block and the backwall specimen. Four were horizontal to apply a lateral force to the specimen and two were diagonal in order to prevent vertical motion. This configuration can be seen in Figures 3 and 4.

Figure 3: Horizontal and diagonal actuators

Figure 4: Top view of actuators

Linear variable differential transducers (LVDTs), shown in Figure 5 were mounted on a fixed truss near each of the actuators in order to monitor the backwall movement. Two vertical LVDTs (Figure 6) were also mounted in order to detect any vertical displacement of the backwall. Any displacement of the reaction block was also monitored with string potentiometers.

Figure 5: Horizontal LVDT

Figure 6: Vertical LVDT

The backfill consisted of silty sand known as Sand Equivalent 30 (SE30). The depth varied as the distance from the wall increased, but according to standard practice in California the nominal height above the base of the wall was 1.67m. Samples of the soil were collected and analyzed to determine the properties. This data can be found in the appendix in section A4. Compaction curves revealed optimum moisture content to be about 9%. The backfill soil was added in 8 inch lifts and compacted using a vibrating plate (Figure 7). Sand cone tests were performed at each layer to determine how densely the sand was compacted and the moisture content. A detailed description of the sand cone process can also be found in the appendix in section A3. Cone Penetration Testing (CPT) was Figure 7: Compacting each lift of backfill attempted in order to determine the axial and frictional forces of the soil with respect to soil depth. This test uses a cone-tipped sensor that penetrates the sand. It records the force required to push through the sand, and also the shear forces on the sides of the rod. These forces are plotted automatically against the depth. This all takes place inside of a truck that has been converted into a large soil probe. Photos of this process are shown in Figures 8 and 9. Unfortunately problems were encountered and the CPT wouldnt penetrate any deeper than 1m.

Figure 8: CPT penetrating backfill soil

Figure 9: Interior of CPT truck Seven holes were drilled along the centerline of the backfill to a depth of 11.5 feet. These holes were then filled with a gypsum mixture and left to set. Once dry, this mixture would become very brittle. The seven gypsum columns would crack easily when the soil failed, making the fail surfaces easy to detect. A grid was painted on the surface of the backfill soil to simplify

the monitoring of cracks that would appear during testing. The completed test site prior to testing is shown in Figures 10 and 11.

Figure 10: Diagram of test site

Figure 11: Gypsum columns and grid

The test was performed under displacement control in order to prevent rotation or vertical motion of the backwall specimen. If vertical movement was detected by the LVDTs, the diagonal actuators would correct the motion by adjusting the applied vertical force as necessary and the horizontal actuators pushed the wall to the desired displacement while preventing rotation. Displacements detected by the LVDTs were recorded using National Instruments Labview 7. Pretests were performed before the backfill was added to a displacement of 2.5 cm in order to estimate the shear resistance at the wall-soil interface. The loads were applied in halfcycles where the wall was displaced then pulled back to the original position. Then it was displaced again, but this time with a larger distance. This process was repeated after the backfill soil was added except after the wall was displaced it was pulled back to zero compressive force rather than zero displacement. This was to prevent any soil from collapsing into the gap that may have formed between the soil and the backwall.

Results and Discussion The data collected from the test showed that the problem with the low stiffness encountered in 2009 had been corrected. Figure 12 shows the horizontal, vertical, and total forces applied by the actuators vs. the horizontal displacement of the backwall. Using this data, the stiffness of the soil was later determined to be 52-72 kips/in/ft which is within the expected range. For a breakdown of the force applied by each actuator, see Figure 12: Applied force vs. horizontal displacement 7

Figure B in the Appendix. Figure 13 is a comparison of the 2010 data to the data from the 2006 and 2009 tests. It also shows some of the predicted curves. This plot makes it easy to see that while the maximum passive resistance in 2009 was expectedly higher than in 2006, the stiffness (found from the initial slope of the curve) was much lower. The red curve shows the data from the 2010 test. The stiffness matches very close to the stiffness calculated in 2006, and the force applied increases to a much higher value. Unfortunately, the maximum value could never be determined for this test as the forces were much higher than anticipated and the backwall and reaction block failed before the soil. The data collected before the termination of the test indicates that the maximum passive resistance from using an 8 foot backwall would, in fact, be significantly higher than when using a 5.5 foot backwall. To confirm these findings, the test will be repeated once more after the reaction block has been repaired and the backwall specimen has been replaced.

Figure 13: Comparison of 2010 data to predictions and data collected in previous tests Acknowledgements I would like to thank my mentors, Anne Lemnitzer and Alberto Salamanca; John Wallace, the PI of this project and Co-PI John Stewart. I would also like to thank Chris Hilson and Steve Keowan for their hard work and contributions to the setup of this project. I would also like to thank Caltrans, who supported this project under contract #59A0247 and Id like to acknowledge the support of the National Science Foundation through grant # ECC-1005054.

Appendix Here is a collection of additional images, data and plots. A.1) Additional data

Figure A: This plot shows the vertical displacement of the backwall with respect to the horizontal displacement. Zero vertical displacement was desired, and the actuators corrected for any vertical motion within a reasonable tolerance throughout the test. At a horizontal displacement of about 1.3 inches, the plot for Displ V2 becomes quite messy; this is when the diagonal actuators reached their maximum capacity.

Figure B: This plot is the breakdown of the forces applied to the backwall by each actuator. It appeared odd that the center horizontal actuators did not begin to apply any force until the backwall had displaced about 0.7 inches. There were problems early on,

when it was discovered that these actuators werent in contact with the backwall. This may have still been the case until that point. A.2) Backwall and Reaction Block Failure Images

Figure C: This image shows the damage due to the failing of the reaction block. The cracks have been highlighted to make them more visible.

Figure D: This figure shows another large break that occurred in the top of the reaction block.

Figure E: This image is of the base of the backwall where one of the diagonal actuators was connected. The cracks that formed have been highlighted to make them more visible.

10

A.3) Sand Cone Testing The sand cone tests were performed in order to quickly determine the compaction density water content of each lift of backfill soil that was added. Two tests were performed per lift. After the soil was added and compacted using a large vibrating plate, the sand cone test was performed. First the sand in the sand cone was weighed. A small area of backfill soil was flattened out and made level. The metal plate was pressed firmly into the sand and a hole was dug from the center of the plate about 10-15 cm deep. The sand Figure F: Sand cone soil collection cone was then placed on top of the hole and the valve opened so the sand in the sand cone could flow freely into the hole. The position of the test was measured and recorded. Since the density of the sand was known, the container was weighed again and the volume of soil removed could be determined. It was then weighed and the compaction density was determined. A small soil sample of known mass was then placed into a frying pan and the water was dried out. The sample was then weighed again and the water content could be calculated. The locations of each test were plotted on a grid which is shown in Figure I.

Figure G: Equipment needed for sand cone test

Figure H: Determining water content

Figure I: Sand cone test locations

11

A.4) Backfill Soil Lab-Analysis Soil samples were taken back to the lab to analyze. Gradation tests, compaction tests, and maximum and minimum density tests were performed for each soil sample. The following plots and tables are the results of those tests. Table A1: Maximum and minimum dry densities Sample: Back 6" unmixed Trial 1 Trial 2 Trial 3 Average 115.4 Max Dry unit Weight, g_d (pcf) 113.69 116.04 116.47 87.34 90.21 93.11 90.22 Min Dry unit Weight, g_d (pcf) Sample: Front 1' mixed Max Dry unit Weight, g_d (pcf) Min Dry unit Weight, g_d (pcf) Sample: Middle New Mixed Max Dry unit Weight, g_d (pcf) Min Dry unit Weight, g_d (pcf) Sample: Delivered 7/6/2010 Max Dry unit Weight, g_d (pcf) Min Dry unit Weight, g_d (pcf) Sample: Delivered 7/7/2010 Max Dry unit Weight, g_d (pcf) Min Dry unit Weight, g_d (pcf) Sample: Delivered 7/9/2010 Max Dry unit Weight, g_d (pcf) Min Dry unit Weight, g_d (pcf) Trial 1 Trial 2 Trial 3 Average 116.83 117.16 116.99 116.99 92 91.93 93.91 92.61 Trial 1 Trial 2 Trial 3 Average 115.81 118.84 119.97 118.21 87.77 89.17 91.13 89.36 Trial 1 Trial 2 Trial 3 Average 17.37 118.17 118.5 118.01 74.28 77.8 88.7 80.26 Trial 1 Trial 2 Trial 3 Average 111.11 108.35 112.99 110.81 88.4 88.58 90.05 89.01 Trial 1 Trial 2 Trial 3 Average 199.07 121.23 120.27 120.19 72.45 92.67 92.14 85.76

12

Figure J: Gradation curves for all six soil samples collected

Figure K: Compaction curves for all six samples of soil collected from the test site

13

References Al-Homound, A.S., Whitman, R.V. [1999]. Seismic Analysis and Design of Rigid Bridge Abutments Considering Rotation and Sliding Incorporating Non-linear Soil Behavior. Soil Dynamics and Earthquake Engineering 18(3), 247-277. California Dept. of Transportation CALTRANS. [2006]. Seismic design criteria, ver. 1.4. CALTRANS, Division of Engineering Services, Office of Structure Design, Sacramento, Calif. Dicleli, Murat, Erhan, Semih [2010]. Effect of Soil-Bridge Interaction on the Magnitude of Internal Forces in Integral Abutment Bridge Components Due to Live Load Effects. Engineering Structures 32(1), 129-145. Ellis, E.A. Springman, S.M. [2001]. Modelling of Soil-Structure Interaction for a Piled Bridge Abutment in Plane Strain FEM Analyses. Computers and Geotechnics 28(2), 79-98. Lemnitzer, Anne Ahlberg, Eric R., Nigbor, Robert L., Shamsabadi, Anoosh, Wallace, John W., Stewart, Jonathan P. [2009]. Lateral Performance of Full-Scale Bridge Abutment Wall with Granular Backfill. Shamsabadi, A., Rollins, K. M., and Kapuskar, M. [2007]. Nonlinear soil-abutment-bridge structure interaction for seismic performance based design. J. Geotech. Geoenviron. Eng., 707720. Stewart, Jonathan P.; Taciroglu, Ertugrul; Wallace, John W.; Ahlberg, Eric R.; Lemnitzer, Anne; Rha, Changsoon; Tehrani, Payman; Keowan, Steve; Nigbor, Robert L.; Salamanca, Alberto [2007]. Full Scale Cyclic Testing of Foundation Support Systems for Highway Bridges. Part II: Abutment Backwalls

14

You might also like