Chapter VI. Physical Model Testing
In Phase 1 of this research, a pile was loaded laterally in a static fashion by hanging weights from a cable that was attached near the top of the pile and run over a pulley. Increasing the load was done by stacking more weights on the hanger at the end of the cable. When the direction of loading had to be reversed, the weights were unstacked, the pulley apparatus was disassembled and then reassembled on the opposite side of the soil test vessel, and the weights were restacked as the testing procedure required.
This loading system was simple and worked well for the static load tests. For Phase 2, the initial ideas for laterally loading a pile group were based on Phase 1's system. Several other concepts were proposed for the loading system, and when financial needs were considered, it was decided to use as much on-hand equipment as possible. The hydraulic cylinders used for soil consolidation were not being used for anything at the time, but there was no means of using them to provide the lateral force for testing the piles. Since they were available, it was decided that an effort should be made to see if they could be used and once the decision was made to use the hydraulic cylinders for loading, design work began on how to mount the cylinders on the soil test vessel and how to power, operate, and control them.
The cylinders were mounted in a horizontal position on frames that were constructed to hold them in this manner and be sturdy enough to allow a large range of loads. The cylinders were mounted pointing toward the center of the soil test vessel. They loaded the pile group by pulling it back and forth as each of the cylinder rams was retracted in turn.
The hydraulic cylinders, as previously stated, were used to consolidate the test soil. All 10 of the cylinders were manufactured by Atlas Hydraulics and have a 101.6 mm (4.0 in.) bore diameter, a 50.8 mm (2.0 in.) ram diameter, and a 305 mm (12.0 in.) stroke. Two cylinders paired for the consolidation process were selected for the pile group testing. Changes had to be made to the plumbing of the two hydraulic fluid port fittings so that the cylinders would function as required. A male hydraulic hose quick-connect fitting was screwed into the retraction port on the cylinder to permit pressurized fluid to enter. An elbow with a short length of hydraulic hose was screwed into the extension port to allow free drainage of any fluid that might leak past the piston seals during testing. The foot plate at the end of each ram was drilled and fitted with a large U-bolt to which linkage for pulling on the pile group would be attached. Test runs on the cylinders could not be made until a pressurizing system was designed and assembled, but the cylinders were cleaned up and painted in preparation. U-frames used for mounting the pulley in the loading system from Phase 1 were the basic members of the new load frames on which the hydraulic cylinders were mounted. The U-frames were made from 76.2 mm (3 in.) steel channel, sized and welded to fit around the ends of the soil test vessel and bolt to end ribs of the tank. Two identical frames already were available, so two more of the frames were fabricated by USU Technical Services. Square structural tubing 89 mm x 89 mm (3.5 in. x 3.5 in.) with 6.35 mm (0.25 in.) thick walls was cut into four 1.22 m (4 ft.) lengths. Then they were welded to the U-frames at the corner positions so that two identical load frames were built up consisting of two of the lengths of structural tubing mated with two of the U-frames. The load frames were bolted to the soil test vessel with grade 8 bolts. Once in place, the frames were rigid. Figure 21 shows the loading system with the load frames in place.
Measurements were made of the hydraulic cylinder base plates so that mounting frames could be fabricated, which would allow the cylinders to be mounted on the load frames. Steel channel 76.2 mm x 38.1 mm (3.0 in. x 1.5 in.) was cut into four lengths, each 0.965 m (38 in.) long, to span the width of the load frames. These channel lengths were slotted with six 17.5 mm (0.688 in.) wide slots of two different lengths oriented perpendicular and parallel to the axis of the channel to allow for vertical and horizontal adjustment of the cylinders when aligning them with the pile group before testing. Steel strap was welded to the ends of the channels to bind them together at the proper distance to match the bolt holes in the base plates of the cylinders. Grade 8 bolts were purchased for mounting cylinders to the mounting frames and for bolting the mounting frames to the load frames. Figure 22 shows a view of a mounting frame.
With the load frames in place, placement of the mounting frames and hydraulic cylinders with respect to the top of the pile group was laid out by measuring from the top of the soil to a string that had been strung between the load frames. This string was leveled at the proper elevation, measurements were made and markings placed at the proper locations on the load frames where the bolt holes for mounting the frames were to be drilled. A portable 12.7 mm (0.5 in.) electromagnetic drill press was used to drill the holes for the 15.9 mm (0.625 in.) diameter bolts. A C-clamp was used to bolster the electromagnetic base on the drill press, which allowed it to be attached to the frames in a horizontal position for drilling. The drill bit used for cutting holes was long enough to drill through both sides of the box beam without having to reposition the drill press. Grade 8 bolts were purchased for bolting the hydraulic cylinders to the mounting frames and the mounting frames to the load frames.
Initially, hydraulic power for the cylinders was to be provided by the electric motor/hydraulic pump system used for consolidating the test soil, but calculations showed that a pressure change of only 36.5 kPa (5.3 p.s.i.) would produce a load change of 222 N (50 lb.), and a pressure of only 439 kPa (64 p.s.i.) would produce the target load of 2.67 kN (600 lb.). Controlling the pump at such low pressures and holding the pressure change steps to the desired 36.5 kPa (5.3 p.s.i.) would have been difficult, so it was decided that compressed air be used to provide hydraulic power to the loading system. A compressor provided compressed air to a tank holding hydraulic fluid which was connected to the cylinders by hoses. The pressure in the tank was controlled by an air pressure regulator mounted on the tank. Directional flow control was provided by a hydraulic control valve. Refer to Figure 21 for a diagram of the loading system.
A 0.113 m3 (30 gal.) upright steel tank that was pressure rated to 1.38 MPa (200 p.s.i.) was purchased to hold the hydraulic fluid. A port in the bottom of the tank was fitted with a brass elbow to which the outlet hose for the pressurized fluid was attached. Galvanized pipe fittings were assembled and screwed into two 19 mm (0.75 in.) ports in the tank, one in the top of the tank to which the air pressure regulator was mounted and served as an air inlet, and the other on the side of the tank near the top to serve as a filler spout for the hydraulic fluid. A 6.35 mm (0.25 in.) port in the top of the tank was fitted with a galvanized nipple to which a ball valve was attached. This valve served as the pressure release valve so that pressure in the tank could be blown down quickly. All other tank ports were sealed with screw-in plugs.
The air pressure regulator had a flow capacity of 0.03 m3/s (65 c.f.m.) and a maximum pressure rating of 2.07 MPa (300 p.s.i.) and was fitted with a pressure gauge that read from 34.5 kPa to 1034 kPa (5-150 p.s.i.). A male-end quick disconnect hose coupling was screwed into the regulator so the air supply hose could be attached easily.
For control of the hydraulic fluid flow, a D03 size, 120-volt AC solenoid operated directional control valve manufactured by Waterman Hydraulics was purchased. This valve had a three-position, blocked center, spring-centered spool with a flow capacity of 6.3 x 10-4 m3/s (10 g.p.m.) and a maximum operating pressure of 34.5 MPa (5000 p.s.i.). The solenoids in the valve were tested by wiring up a simple electrical circuit and listening for the tell-tale click as each solenoid was energized.
Flow of oil to and from various ports in the bottom of the solenoid valve was facilitated by a side ported subplate that was ordered with the valve. This subplate was bolted to the valve and permitted mounting of the valve to a steel channel section base plate, and also was the means by which four hydraulic hoses were connected to the valve. One hose supplied pressurized fluid to the valve, and one hose served as an exhaust oil dump. The two other hoses transported the pressurized oil, as directed by the valve, to the two hydraulic cylinders. All of the components had to be assembled after either being purchased or fabricated.
Once the loading system was assembled, it was tested in the CEE shop. The two hydraulic cylinders were hung from an A-frame by their U-bolts and the hoses were attached. Air pressure was supplied by hooking the air pressure regulator to the shop's air system. Pressure to the tank was slowly increased to determine the pressure required to make the cylinders start to move. The solenoid valve repeatedly was switched to send oil to each of the cylinders. Each cylinder lifted itself as the ram was retracted, but one of the cylinders would not extend when manually pulled, and the plumbing had to be switched and fluid sent to the other port to extend it. Another test determined the amount of force needed to extend the ram. This test showed that extension resistance in one of the cylinders was much greater than the other. Several different cylinders were tested until a pair of cylinders with similar working characteristics was found.
Upon seeing the results of the cylinder tests and seeing that the force needed to extend the cylinder rams was not negligible, an idea was proposed that the cylinders be linked so that the force needed to overcome the residual extension resistance of the rams by-passed the pile group and thus allow the pile group to be loaded only with the load intended for causing displacement. It was decided to link the cylinder foot plates with wire rope, then that they be tensioned and adjusted with turn buckles. Four brackets were cut from steel strap and drilled with two holes each for the needed connections. Two were then bolted to the back of each foot plate. The residual load by-pass cables were attached to the brackets. Figure 23 shows the residual load by-pass cables. Another benefit of using the by-pass cables was that forces in the load cells, on either end of the pile group, did not have to be compared, therefore the residual cylinder resistance could be subtracted and provide the net load on the pile group.
LabVIEW Software, coupled with the PC-LPM-16 A/D converter board, has the capability of acquiring data and sending analog signals. This ability was used to control the loading system. The solenoid valve was AC powered, and was switched externally using switching with relays that were controlled by LabVIEW.
Two solid-state switching relays, manufactured by Teledyne Inc., were mounted in the data acquisition box, wired to the fuse box, and powered by the 120 Volt AC supply cord. A relay-controlling logic circuit was wired between the A/D output line that was wired to the interface bus ribbon cable connector and the relays. This logic circuit was designed to prevent power from being sent to both solenoids on the valve at the same time, and thus prevent damage to the valve.
Power was delivered from the relays to the valve by a three-wire extension cord wired in a hot/hot/neutral configuration, rather than the standard hot/neutral/ground configuration. One wire from each of the solenoids was connected with the neutral lead. The other wires were each connected to a hot lead. This wiring scheme was used so that a single cord could be used to send power from the relays to the solenoids.
The cord leading to the solenoid valve was fitted with a plug with blades horizontally oriented rather than the normal vertical, and this plug mated with a socket of the same type coming from the data acquisition box. This allowed the solenoid valve to be detached from the data acquisition box as needed and also prevented the solenoid valve from being plugged in to a standard outlet so that it would not be damaged due to its wiring configuration.
This switching system was tested by replacing the solenoid valve with small red and green lights that were wired the same as the solenoid valve and turned on when the relays were triggered by LabVIEW.
One of the concerns for continued testing of model piles was how many tests could be done in the soil that had already been consolidated in the soil test vessel. In Phase 1, a single pile test was conducted in each end of the test vessel, with the pile located approximately 0.46 m (18 in.) from the end wall and centered between the side walls. Soil properties tests also were performed around the sites, thus disturbing soil in the ends of the vessel enough to prevent any further testing from being done.
A plot of soil in the vessel with a cushion of undisturbed soil of five pile diameters in every direction from the pile group would be large enough to conduct a group test without introducing any change in soil properties due to prior disturbance or boundary effects. Using this five pile diameter (5D) cushion as a guideline, it was determined that there was enough undisturbed soil to conduct four pile group tests. This information was necessary for planning where to install the pile group and how to set up the loading system.
Driving the pile group into the soil could have been done by either pushing them in connected as a group or inserting them individually and than pinning them together on the load rod. Because of the extremely tight fit between the load rod, pins, and piles, it was decided to pin the pile group together and drive it into the soil rather than push the piles separately and then try to pin them together.
To keep the piles aligned in 3D spacing during driving, two templates were made, through which the pile group was driven, while the proper spacing was maintained. Figure 24 shows the template design. Plywood was cut into four sections 230 mm (9 in.) wide and 610 mm (24 in.) long. The pieces were bound with wood screws in a stack, four boards high, so that the edges matched. The pile locations were marked on the top board, after which holes 34.9 mm (1.375 in.) in diameter were bored through the whole stack at the five locations. The boards were then split through the middle of the holes with a saw so that two identical pieces were made from each of the original boards. Two of the paired boards were used for templates, while the other two were stored for future use.
Two 2 x 6 planks, long enough to span the distance between the two yokes still attached to the soil test vessel, were split down the middle and used as mounts for the templates. After again laying out the four possible pile test locations on soil using the spacing template already mentioned, measurements were taken to determine the position where the piles should be placed with respect to the test vessel walls and the two yokes. The measurements were used to locate the pile spacing templates on the split planks, where they were fastened, so that when the planks were clamped to the yokes, the templates would be in the proper position for driving piles. Two blocks of wood were screwed on to one of the spacing template planks to act as spacers for clamping the template halves together, after which the templates were matched together with C-clamps.
The cylinder mounting frames were bolted to load frames and a string was strung through the bolt holes of one mount frame, across the soil and through the matching bolt holes on the other mount frame. When pulled tight, this string showed the axis for proper pile group alignment.
One of the spacing templates was clamped to the underside of the two yokes with C-clamps in a position roughly close to where the piles should be placed. A plumb bob string-guide was slipped into one of the end template holes, the plumb bob was suspended from the guide, and then lowered until it hung just above the string. The spacing template end nearest the plumb bob was moved until the plumb bob was centered over the string and then clamped in place. Next, the plumb bob was moved to the opposite end of the template and lowered until an alignment check could be made, after which the nearest end of the driving template was moved to center the plumb bob over the string. This procedure was repeated many times until the plumb bob could be lowered from either end of the spacing template and deflect the string downward with its pointed tip. The C-clamps were securely tightened to conclude the alignment procedure for the lower template. The other template was aligned in the same manner after it had been clamped on top of the yokes. To be considered satisfactorily aligned, the plumb bob had to be lowered from the upper template through the lower template in rapid succession several times without touching the lower template at all, and then rest its tip on the alignment string. Template alignment took several hours, but was crucial to guaranteeing pile group spacing and alignment.
With the templates set up for pile group placement, the piles and the load rod were pinned together in a prechosen order so as to place the piles with the highest number of reliable strain gages on the outside of the group, and the piles with a lesser number of reliable gages in the interior spots. Pile 4 was positioned as the lead pile on the left with Pile 2 next in line. Pile 1 was placed in the center with Pile 5 in the second position from the right. Pile 6 was the lead pile on the right. Since Pile 3 had two unreliable gages in the two uppermost positions, which are important for good boundary condition data, it was relegated to the role of thermal drift monitor.
The pile group was lifted up and slid down into the templates until the pile toes rested on the soil surface. Due to the soil resistance, the pile group had to be jacked into the soil with the ceiling of the Water Lab acting as the reaction member. A scissors-type automobile jack with a 230 mm (9 in.) travel distance was used to push the piles into the soil. Two redwood blocks 510 mm (20 in.) in length had been drilled to a shallow depth with large diameter bit in three equally spaced locations along the length of the blocks to accept 25.4 mm (1.0 in.) diameter ball bearing to make a one-directional moment-free joint. Then, if jack alignment was not true, the piles would not be pushed out of proper alignment during driving. One of the blocks was placed on top of the five pile caps while the ball bearing, second block, and jack were positioned on this block, after which jacking commenced. A torpedo level was used to help keep the piles level as they were forced into the soil. Each time the jack was fully extended and then compressed, a block of wood had to be inserted under the jack to take up space so that the pile group could be pushed farther into the soil.
When the piles were pushed in far enough that the uppermost spacing template interfered with movement, the clamps holding its ends were loosened and the template was removed. The lower spacing template was removed when the pile group reached that level as well. Pile driving ceased when the piles had been pushed 1.295 m (51 in.) into the soil. This left the top of the pile group 279 mm (11.0 in.) above the top of the soil at the load rod elevation. The pile caps were checked for level and the piles were checked for plumb during the entire process. Some cracking of the soil occurred during driving, but measurements with a hand held ruler showed that the cracks were only superficial.
Pile 3, the thermal drift monitoring pile, was pushed by hand into the soil in the end of the vessel where it would not disturb soil that could be used for further group testing.
With the pile group installed in the soil, further preparations for testing could be made, first of which was mounting the hydraulic cylinders to the soil test vessel. Actual bolting of the cylinders to the previously installed mounting frames was straight-forward. As stated earlier, grade 8 bolts were used for this attachment along with doubled-up high strength washers to help transfer the pulling force to a larger area of the web of the mounting frame channel sections. Other washers were placed between the cylinder base plates and the mounting frames to act as shims to level the cylinders, which was checked with a torpedo level. Getting the cylinders aligned longitudinally with the pile group was a much more difficult task. First a reference was needed, which could be aligned with the pile group and then used to align the cylinders. A 6.1 m (20 ft.) length of angle iron served as the reference. It was laid on the lower channel of the mounting frames and then aligned with the pile group by measuring the distance from its flat edge to each end of the pile group load rod. Adjustments were made in small increments until there was no perceptible difference in the distances from each end of the load rod to the angle iron. It was then firmly clamped to the mounting frames. The slots in the mounting frames allowed cylinders to be adjusted up and down and side to side. By measuring from the reference angle iron to two different spots on the cylinder bodies, the cyclinders were moved and shimmed until alignment requirements were met.
Once the hydraulic cylinders were aligned, they could be hooked to the pile group and to each other with the by-pass cables. Since the pile group was not centered between the cylinders, different linkage arrangements were needed on each end of the pile group to hook them to the cylinders. On the end nearest to a cylinder, a quick link was connected to the twisted clevis in the end of the load rod. The load cell was hooked onto the quick link on one end and to a turnbuckle on the other end. The turnbuckle was bolted to the foot plate on the end of the ram. A similar arrangement was used for the other side, except that a piece of steel strap 25 mm (1.0 in.) wide and 508 mm (20 in.) long was connected to the load cell by a quick link. The other end of the strap was hooked to a turnbuckle, which was connected to the ram foot plate with a quick link. These linkage assemblies were designed for versatility and ease of adjustment, which proved quite beneficial during testing.
A wide flange, thin section aluminum T-beam was aligned with the pile group and clamped to the underside of the soil vessel yokes. To this T-beam the LVDT mounting bracket was clamped, and then the LVDTs were hung in place between the pile group and the mounting bracket. They were leveled and aligned by adjusting the wing nuts on the pile group and the bracket.
Having accomplished installation, assembly, and alignment of the piles and other instruments, the input boards for all of them were plugged into the data acquisition box. Since the problem of voltage saturation still occurred when turning on the system, all boards were plugged in while the system was running. Pile input boards were inserted so the piles were arranged in the group, with boards from the thermal drift monitoring pile next in line, and finally, input boards from the load rod and load cells, and the LVDTs. With all input boards in place and taking readings, the large power supply heated up quickly and a warm-up time was needed to bring the data acquisition system into a stable equilibrium. The system was quite thermally sensitive, and opening the box could change the readouts on many of the strain gages, so the system was left running all the time.
Three different testing strategies were tried in an attempt to find a method that produced the most consistent, meaningful output. Time-controlled cycling of the pile group was used in the first attempt to perform a lateral load test. The cycle rate was 0.1 Hertz, during which time the testing program sampled and averaged three readings of all of the instruments right before the loading direction changed, thus picking up the interval of maximum load. Testing was halted after three 100-cycle test iterations at loads of 222.4 N (50 lb.), 444.8 N (100 lb.), and 667.2 N (150 lb.), because frictional differences between the two hydraulic cylinders caused quicker ram movement in one direction than in the other, and thus unequal and inconsistent loading of the pile group.
This unequal loading was highly undesirable, so the testing method and software were changed to a closed-loop configuration to allow cycling to be controlled by reading the load on the load cells and switching the solenoid valve when the target load was reached. The testing program was modified so that it continuously sampled all of the gages at a rate of three readings every five seconds. When the triggering load was reached, the last three samples taken were averaged and stored.
Since the test soil already had been disturbed by the previous test iterations, testing resumed at the load of 667.2 N (150 lb.), and continued in 222.4 N (50 lb.) increments of 100 cycles each until it was halted after 28 cycles at 1779.3 N (400 lb.) because the soil had deformed so much that the LVDTs were in danger of being damaged due to the excessive travel of the tops of the piles. This great amount of soil deformation was unexpected but the testing was deemed successful. Post test analysis of test data revealed that a small programming error had caused the baseline conditions data to be written repeatedly to the output file instead of the data gathered at the end of each half cycle, and so this test data was useless. During this initial testing, the load exerted by hydraulic cylinders was changed by adjusting the air pressure in the tank with the air pressure regulator. Hitting the target load proved to be quite difficult.
These initial tests proved to be valuable learning experiences, and out of them evolved a new approach to conducting testing. The lateral force on the pile group would be increased to a maximum of 1779.3 N (400 lb.) while continuously sampling all gages at the fastest rate possible so the desired data would be captured as the load passed through the targeted loads. All soil deformation would be unique to each cycle, and would not reflect accumulated disturbance of lower loadings seen in the other tests.
The piles were pulled from the spot where the earlier tests were conducted and reinstalled in a new location in the soil test vessel with the piles positioned in the group as shown in Figure 25. A new testing program, named One Shot.VI, was written for the new test procedure. It incorporated many features that made testing much easier. This program automatically named the output data files according to the cycle number and whether the pile group was being pulled to the right or the left. The load that was delivered to the pile group, as well as group displacement and pile-top angle as measured by the LVDTs, was displayed in real time on the computer monitor. All operations of the hydraulic system were controlled through signals sent to the solenoid valve. Streamlining of the program also allowed for faster data acquisition rate. A sample of some of the LabVIEW code used in this program is contained in Appendix B.
The procedure for conducting testing consisted of repeating the same basic steps many times to collect and store the data in half-cycle blocks, determined by whether the pile group was being pulled to the right or the left. To run a half-cycle test, the testing program One Shot.VI was called up on the computer and a baseline reading of all instruments was made. The program would not allow testing to begin without this initial reading. When this was accomplished, data acquisition was started by clicking the start button. Figure 26 shows the program control panel for One ShotVI. This started data acquisition and opened the solenoid valve. The air pressure regulator on the tank was slowly opened allowing the hydraulic fluid to be pressurized. As the cylinder being pressurized pulled on the pile group, the load, as monitored by the active load cell, was called out to help the person opening the regulator maintain a constant rate. The linkage on the inactive load cell was kept loose so that it would not register any load other than the weight of the linkage. As the load continued to rise, the pile group could be seen to deflect, and pile top displacement and angle were displayed on the computer. When the air pressure in the tank rose high enough to produce the desired 1779.3 N (400 lb.) load in the active load cell, the computer switched the solenoid valve, thus causing the hitherto unpressurized cylinder to pull the pile group in the opposite direction until the load on the active load cell fell below a given threshold, usually 89 N (20 lb.), at which point the solenoid valve was closed and data acquisition stopped. The pressure in the tank was released and the air regulator was reset in preparation for the next half-cycle.
Testing proceeded smoothly overall, except for the power going out for about three minutes throughout the whole city. Only the data being collected in one half-cycle were lost, but to resume testing without overwriting the existing data files, the program had to be modified. Testing was terminated after 50 full cycles because the soil had deformed to the point where the pile top movement threatened to damage the LVDTs. Nearly 53 megabytes of data were gathered during this test. The data acquisition rate was four samples from every channel per second. Continuously sampling all instrumentation at this rate produced data for 200 to 300 loads in each half-cycle. The data were checked for any readily noticeable problems and then saved on a set of six floppy disks. This concluded the testing phase of this project.
During the course of the testing and during the preliminary work to debug the system hardware and software, it was noticed that some of the strain gage readings seemed to be lower than expected, even when they were corrected by the calibration factors, so it was decided to recalibrate all of the instrumentation at the end of testing with everything configured as it was during the final test and under the same conditions. The piles were unplugged, pulled from the soil and cleaned off, and then reconnected to the data acquisition system.
Although the same procedure was followed for this calibration as was used in the initial calibration, a new calibration program was used that simplified and speeded up this somewhat tedious task. The new program, named Calibrate.VI, averaged 20 voltage readings for every gage at every load and converted them to stress values. The capabilities of LabVIEW were demonstrated during testing and instilled confidence in its ability to generate reliable calibration output without having to convert the voltage to stress in a spreadsheet. The LabVIEW output was imported into a Quattro Pro spreadsheet for regression analysis and graphical conversion. The calibration output for every strain gage in Pile 1, Pile 2, Pile 3, Pile 4, Pile 5, and Pile 6 can be viewed in Figures C-1 through C-6, respectively. Calibration output for the load rod is found in Figure C-7. Load cell calibration output is contained in Figure C-8, and LVDT calibration results are found in Figure C-9.
The wisdom of doing a second calibration was verified by the new calibration factors, which all came out to be roughly 20 to 25 percent greater than the initial factors. This was most likely because in the initial calibration work, only the input boards connected to the pile being calibrated were installed in the data acquisition system at the time of calibration, and so power demands and electrical noise conditions were much different than those experienced during testing. As stated earlier, the system heated up quite a lot when the full complement of instrumentation was being sampled, and the effects of this temperature difference also might have contributed to the discrepancy in the calibration factors.
The effects of allowing the system to come to equilibrium can be seen in the calibration output from Pile 4 when compared with the calibration output of the other piles. Pile 4 was calibrated first, and as is evidenced by the slightly erratic calibration output, the data acquisition system had not quite settled down to its best operating noise level. Pile 2 was calibrated second, and the output from all of its strain gages is consistent, showing that the system equilibrium had been reached.