Chapter II. Literature Review
Most research dealing with lateral loading of piles has been done on single piles even though piles are most frequently used in groups. Until Dr. Dan Brown performed his full-scale cyclic lateral load test on a pile group, few well instrumented, carefully performed full-scale tests had been completed (Brown and Reese, 1985).
In Brown's 1985 full-scale test, nine closed-ended steel pipe piles, 273 mm (10.75 in.) in diameter, with 9.27 mm (0.365 in.) thick walls, were used. These piles had been driven to a depth of about 12.2 m (40 ft.) below grade in stiff overconsolidated clays that were kept saturated for his testing. The piles were installed in a 3 x 3 arrangement with a center-to-center spacing of three pile diameters. An instrumented single pile was tested and used as a reference to which the pile test data could be compared.
All nine of the piles were instrumented with strain gages that had been attached to pipes that were inserted in the piles and grouted in place. Four gages, in a full bridge configuration and spaced at 90° intervals, were mounted at eight locations starting 0.3 m (1 ft.) below the soil surface and extending downward at 0.3 m (1 ft.) intervals for 2.4 m (8 ft.). Three more levels of strain gages were located at depths of 2.9 m (9.5 ft.), 3.4 m (11.5 ft.), and 4.0 m (13 ft.).
Linear potentiometers were used to make deflection measurements at the point of loading and also 1.2 m (4 ft.) above the point of loading. From data collected at these different elevations, slope of the top of the pile was determined. Eighteen potentiometers were used to monitor the movement of the nine piles.
Power for all instrumentation was provided by a Kepco power supply. Two Hewlett-Packard 3497A Data Acquisition/Control units were used for analog-to-digital conversion and were controlled by a Hewlett-Packard 85 microprocessor. Estimated calibration factors were used in computing bending moments and experimentally determined calibration factors were used for the load cells and potentiometers. Reading of all 132 channels of instrumentation required 20 seconds.
Loading of the pile group was accomplished with a 305 mm (12.0 in.) bore diameter double acting hydraulic cylinder, pressurized by a hydraulic pump. A servo valve operated by a servo controller controlled hydraulic fluid flow. A closed loop system consisting of a digital function generator and a feedback linear potentiometer was used to provide the desired loading pattern. A load frame with moment-free hinged connections with integral load cells was used to transmit lateral loads to the pile group.
Testing was done under deflection controlled conditions. The first loading consisted of 100 cycles at a deflection near 2.5 mm (0.1 in.). Two hundred cycles were done at each of the four other deflection settings, which ranged up to approximately 53 mm (2.1 in.) for the fifth setting. The load on the pile group varied from a low of 120 kN (27 kip) at the end of the first set of cycles to a high of about 740 kN (167 kip) at the beginning of the last group of cycles.
Results of this testing showed that the piles in a group take less of a load per pile than does a single pile similarly loaded. The group piles deflect more than does a single pile at the equivalent load per pile, and the bending moment in the group piles is greater than the bending moment in a single pile subjected to the same average load per pile. After cycling, the difference between the group behavior and the single pile behavior only slightly decreases. Trends from the loading showed that distribution of load through the group followed a front row to back row pattern, with the leading row taking more of the load. In the leading row, the center pile took more of the load, while the corner piles on the trailing row took more load than did the center pile on the trailing row.
A large-scale test was conducted by Townsend et al. (1997) at the Roosevelt Bridge replacement at Stuart, Fla. The test group consisted of 16 760 mm (30 in.) prestressed concrete piles driven into sand to a depth of around 14 m (46 ft.) and subjected to lateral loads with a fixed head production pile group acting as the reaction. This test compared actual test data to predictions made for the lateral loading of the pile groups with the Florida Pier software package.
Ten of 16 piles in the test group were fully instrumented. Instrumentation was installed in the piles by inserting an instrumented 355 mm (14 in.) steel pipe into the center void of each pile and grouting it in place. The instrumented pipes had nine levels of strain gages, configured for measuring bending stresses only, attached to the outside surface and spaced every 914 mm (36 in.) below the mudline, except for the last gage, which was 1.83 m (72 in.) below the next to last gage level. Each level used four gages in a full bridge configuration. Slopes and deflections were measured at the top of the pile with slope inclinometers and potentiometers. Of the 16 piles in the reaction group, only six of them were instrumented and were used to gather data on bending as well as axial strains.
Ten load cells of 445 kN (100 kip) capacity were used on each of the instrumented piles in the test group, with an additional load cell on each of three other piles so that more information could be gathered. A 4.44 MN (1000 kip) load cell was used between the loading jack and the load frame of the test group.
The data acquisition system was a System 4000, provided by Measurements Group. Readings were taken every 15 seconds during the test, which lasted more than five hours. Each reading sampled 232 channels.
Thirteen load steps were made, nine going up in steps of about 556 kN (125 kip) each, and four going down. The maximum load was about 4.8 MN (1080 kip).
By comparing measured pile group response with predictions made using Florida Pier, it was found that Florida Pier modeled the load deflection response of both pile groups quite well, and did a good prediction even after pile cracking. It also was found that the leading and trailing rows of piles had similar load deflection curves and that the piles in the leading row took more load than did those on the trailing row. Florida Pier predictions for bending moment agreed well with measured moments in the piles. Maximum bending moment was higher for the lead piles than for the trailing piles in the test.
Another full-scale lateral load test was conducted on a pile group at the Salt Lake International Airport by Rollins, Peterson, and Weaver (1996). The pile group was comprised of nine steel pipe piles arranged in a 3 x 3 configuration. The piles were 32.4 mm (12.75 in.) in diameter with 9.5 mm (0.375 in.) thick walls, and were filled with concrete. They were driven to a depth of 9.1 m (30 ft.) in a mixed soil strata consisting of clay, silt and sandy silt, and sand, and were spaced approximately three pile diameters apart. An isolated single pile was driven and tested for comparison with the results from the group tests.
Full details of instrumentation were not available, but strain gages and inclinometers were used for measuring bending moment and displacement of the piles. The piles were pin-connected to a loading frame for static load testing, and the load to each pile was measured through strain gage instrumented tie rods.
Lateral loading on the pile group was performed under four different conditions. The conditions were:
Results from the testing showed that dynamic resistance is greater than static resistance for the group, and approaches the value found through the static loading of a single pile. The ratio of the average load carried by a pile in each row as compared with the load taken by a single pile showed that piles in the front row carried about 80 percent, the piles in the middle row carried about 50 percent, and the piles in the trailing row carried about 60 percent. Maximum bending moments for the dynamic loading were within about 15 percent of those measured in the static loading. Load distribution among the piles was similar for static and dynamic loading. Further analysis is being conducted on this pile group experiment.
To cut back on the expense of doing lateral pile group testing and to simplify the tests, pile groups have been modeled using dimensional analysis to scale down the piles and loads to much more manageable levels. Model tests involving sandy soils often are conducted in a centrifuge so that the effective stress associated with overburden can be simulated.
McVay et al. (1994) developed a device to drive model piles individually in a nine pile group and measure row contributions and group displacements, all while the centrifuge was moving.
McVay's centrifuge tests were used to study the effects of pile spacing and soil density on load distribution by row, as well as the lateral resistance of the group. A load cell mounted between each row of piles on the pile cap was used to measure the load distribution in the group. An LVDT was used to measure lateral deflection of the group.
Test results showed that the front row took more of the load than the other rows, with the middle row taking the next and the trailing row taking the least. Distribution was 41 percent, 32 percent, and 27 percent, respectively. Soil density also changes the amount of load taken by each row, with the lead pile taking more at higher densities and a near even distribution among the rows at lower densities. Results also showed that changing the spacing to five diameters caused the lateral load resistance to increase with a corresponding decrease in the effects one pile had on another due to proximity.
Further testing by McVay, Casper, and Shang (1995) on laterally loaded, 3 x 3 pile groups with 3-D and 5-D pile spacing was conducted and compared with results from a lateral load test on a single pile. Results showed that the shadowing effect is a function of soil density and pile spacing. In a dense soil, load distribution for a pile group with 3-D pile spacing was 45 percent, 32 percent, and 23 percent for the leading, middle, and trailing rows, respectively. In medium loose sands, distribution was 37 percent, 33 percent, and 30 percent for the leading to trailing rows. Pile group efficiency for lateral loading with a pile spacing of 3-D was approximately 0.74. For a pile group with 5-D spacing, the group efficiency was approximately 0.93 with load distribution being 36 or 35 percent for the leading row, 33 percent for the middle row, and 31 percent for the trailing row.
Instrumented model pile group tests were conducted in clay by Rao, Ramakrishna, and Raju (1996). Model piles 21.5 mm (0.85 in.) in diameter and 1,000 mm (3.3 ft.) long were instrumented with strain gages spaced 150 mm (5.9 in.) apart at seven locations along the length of the piles, and connected in six different group configurations of either two or three piles per group at various center to center spacings. The tests were conducted in a soil test tank with dimensions of 1.2 m (47.2 in.) x 0.8 m (31.5 in.) x 1.1 m (43.3 in.), where soil was placed around the piles. Loading was accomplished by placing weights on a hanger attached to the pile group by wire rope passed over a pulley. The applied load was measured by a load cell, and lateral displacement was measured with an LVDT. Strain gage data were acquired through an ORIONA (3530A) data logging system. LVDT and load cell measurements were collected using a frequency amplifier and computer data acquisition system. Tests were conducted on a single pile; a two-pile group in series with pile spacings of 3-D, 4-D, 5-D, and 6-D; a two-pile group in parallel with pile spacings of 3-D and 4-D; a three-pile group in series with 3-D and 4-D spacing; and three-pile group tests with spacing set at 3-D in triangular arrangements with either the apex or flat side leading.
From the tests conducted on the two-pile groups arranged in series at different spacings, it was shown that group capacity increased with increased spacing between the piles and that this capacity reached twice that of a single pile when the spacing was 6-D. Results from loading in the parallel configuration showed that larger capacities are possible than for a series arrangement for the same number of piles.
The results of tests conducted on three pile groups in the series, parallel, and triangular arrangements showed that the series arrangement carried the least load of all, with the triangular-apex side leading configuration taking the most load for a given deflection. Measured bending moment for two- and three-pile groups arranged in the series configuration and with 4-D spacing was shown to be greater in the trailing piles than in the leading piles.