‘of 75 m. The particle size distribution averaged 54% of sand, 42% of silt, and 4% of clay. Some more sandy layers as well as more clayey layers, including occasional lenses with gravel- ‘ize particles Were also present, The liquid limit of the finer zones ranged from 30 to 60%, the plastic limit was between 26 and 32%, and the plasticity index ranged from 0 10 29. Averaging about 12, The solid density was 2,670 kg/m’. (Ocdometer tests on Pitcher samples on the softer (siltstone) portions of the tuff indicated a quasi-preconsolidation stress lose to the effective overburden pressure, and a compression index C. ranging from 0.04 to 0.17, averaging about 0.1. The compression indices and void ratio values correspond to & Janbu modulus number of about 50, The initial slope of stress- sirain curves in unconsolidated-undrained compression tests gave an “undrained elastic” modulus that ranged from 7 to 27 MPa, averaging about 17 MPa. The average slope of un loading-reloading curves gave values ranging from 17 to 54 MPa, averaging about 30 MPa (the values are much smaller than the elastic modulus of 1,000 MPa used in the initial de sign). These values of "elastic" modulus correspond to a Janbu modulus number of about 204)~300 (Fellenius 1996). Triaxial tests on Pitcher samples on the softer portions of the tuff resulted in an undrained shear strength ranging from (69 to 147 kPa, with an average of 115 kPa. Unconfined compression tests on the NQ cores indicated an unconfined compressive strength ranging from 250 to 2.500 kPa, averaging about 750 kPa. While it is possible that the Strength of the core samples increased during the period be- tween sampling and testing due to inadvertent partial drying in the hot Manila climate, the core samples are also undoubt- edly indicative of the stronger portions of the Guadalupe for- mation. However, much of the formation consists ofthe softer ‘materials not recovered by NQ coring. Also, the levels at ‘which the harder and the softer portions of the formation occur vary considerably in adjacent borcholes. For these reasons, the Jower-strength, more compressible materials may control the bearing capacity and settlement of deep foundations, PRINCIPLES OF O-CELL TEST ‘The O-cell method (Osterberg 1998) incorporates a sacrifi- cial hydraulic jacklike device (Osterberg-cell) placed at or near the toe (base) of the pile (drilled-shaft or barrette) to be tested. When hydraulic pressure is increased, the O-cell ex- ppands, pushing the shaft upward and the base downward. The ‘upward movement of the O-cell top plate is the movement of the shaft at the O-cell location and it is measured by means ‘of tetales extending from the O-cell top plate to the ground surface. In addition, the separation of the top and bottom O- cell plates is measured by displacement transducers placed be- tween the plates. The downward movement of the O-cell base plate is obtained as the difference between the upward move- ment of the top plate and the cell plate separation. It is im- portant to realize that the upward and downward load move- ‘ments are not equal. The upward load movement is governed by the shear resistance characteristics of the soil along the shaft, whereas the downward load movement is governed by the compressibility of the soil below the pile toe. ‘AL the start of the test the pressure in the O-cell is O and the self-weight of the barrette at the location of the O-cell is ‘carried structurally by the O-cell assembly. The test consists ‘of applying load increments to the barrette by means of inere- mentally increasing pressure in the O-cell and recording the resulting plate separation and telltale movements. The first pressure increments transfer the barrete self-weight from the ‘assembly (o the O-cell uid, The O-cell load determined from the hydraulic pressure reading at completed transfer is the self- weight value and it is reached at minimal movement (i, sep aration of the O-cell plates). The self-weight consists oF the buoyant weight of the burrete plus any residual load in the pile atthe location of the O-cell ‘When the full self-weight of the barretle has been trans ferred to pressure in the O-cell, a further increase oF pressure expands the O-cell; that is, the top plate moves upward and the base plate moves downward. (The assembly is built with fan internal bond between the plates. « construction feature, Which breaks when the separation starts.) The O-cell load versus the upward movement is the load movement curve of the barrette shaft, The O-cell load versus the downward movement is the loa-movement curve of the barrette base. This separation of the load-movement behavior ‘of the shaft and base is not obtainable from a conventional Static loading test. OF course, the self-weight must be sub twacted to obtain the load-movement of the pile shaft, The self weight should be included in the load movement for the pile toe, however. ‘TEST BARRETTE AND TESTING PROGRAM ‘To minimize the magnitude of the required test load, one of the two smallest planned barrettes was selected for the O-cell test. The barrette was produced during the diaphragm wall construction in advance of the planned barrette construction, Fig. 2 shows a photograph of the reinforcement cage. O-cell assembly, and some of the instrumentation of the test barette while lying on the ground prior w insertion into the slurry trench. The test barretie had a eross section of 285 by 0.85 'm (area 2.42 m’) and was constructed to a depth of 28.2 m (base level) below the ground surface. The upper 12 m of the barrette was made from a very weak concrete t facilitate the Future excavation to this depth at the site The barrette was constructed by excavating the soit and soft rock using a 0.85-m-wide, mechanically operated clamshell while supporting the sides of the excavation with bentonitic slurry. Advancement of the barrete excavation through the sandstone layer encountered at a depth of about 20 m below the surface required the use of a chopping bit, and some con- ‘cems were expressed that, when the material changed again to silistone near the bottom of the barrette, some disturbance of the siltstone below the base of the barrette might have occurred sa result of this method of loosening the material. On reach ing the full depth, a reinforcing cage was inserted into the slurried hole. The O-cells andthe strain gauges forthe loading test were allached to the reinforcing cage. Sleeves for later insertion of telltale rods were also included. The slurry was then displaced by concrete, poured by tremie method. The con- struction was completed on October 12, 1996. ‘A.040-m-thick O-cell assembly was placed with the bottom (O-cell plate 1.0 m above the base. The assembly embodied ‘evo $50-mm (0.4. O-cells (the sacrificial jacks) connected in a series to a common pressure hose and pressure gauge in an 1568 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JULY 1909assembly of a total plan area of 0.55 by 2.48 m and a height of 0.40 m. The maximum stroke, that is, the separation of the -cell plates, was 150 mm. ° ‘To obtain information on the shaft resistance distribution, the barrette was instrumented with strain gauges installed in sroups of four separate gauges at depths of 15.6 m (Level 1), 21.1'm (Level 2), and 25.2 m (Level 3). The highest level was fat the approximate level of the boundary between the over- burden soils and the Guadalupe Tuff. The second level was approximately at the boundary between siltstone and sandstone layers within the tuff. Level 3 was placed 1.6 m above the O- cell top plate and 3.0 m above the barrette base. ‘The strain-gauge readings at the start of the test were taken to be 2070, Therefore, strain recorded during the test corre: sponds to the increase of load (for each gauge level) and re: flects the load generated by the O-cells after the transter of self-weight is completed, ‘The initial test was performed on October 18, 1996, 6 days. after completion of the construction. The testing procedure fol- lowed the quick test schedule, applying small, approximately ‘equal increments of load at equal short time intervals, The load, increments were applied by increasing the pressure in the O- cells in steps corresponding to about 600 kN, ‘The increments ‘were applied approximately every S min until the observations showed excessive movement of either the shaft (upward) fr the base (downward), or until the maximum capacity of the particular O-cell combination (i.e., 36,000 kN) hiad been reached. Readings of all gauges were taken at 1, 2, and 4 min into each foad increment. A graph of total cell separation ver- sus applied hydraulic pressure was plotted as the test pro- fressed, and the loading was terminated when distinct steep- tening of the load-displacement curve took place, characteristic ff imminent failure. At that time, there was difficulty in main- taining the hydraulic pressure. On December 5, 1996, the O- cell cest was repeated using the same loading schedule. ‘TEST RESULTS ‘The primary results of the O-cell test are the measured O- cell Toad versus the recorded plate movements. These results are plotted in two diagrams shown in Fig. 3, presenting the recorded load-movement data for the top and bottom O-cell plates for both the initial test and the retest. Notice that the top-plate diagram is not adjusted for the self-weight of the barette. The bottom diagram shows the measured expansion ‘of the O-cell, that is, the separation of the O-cell plates oc- curring below the 6,400-KN O-cell load dering the initial test. Because of leaks at the connection of the hydraulic hose to fone of the pressure gauges, the test had Co be interrupted twice, first in the very beginning of the test and a second time at an (O-cell load of 6,864) KN. The interruptions do not seen to have affected the general appearance of the load-movement curves, Doring the initial tes, at an applied load of 1,060 KN. sep- aration of the O-cell plates had not yet occurred. At the load, ff 1,664 KN, the next load level, a separation of 1.7 mm was, measured. Intersection of the straight portion of the line with, the abscissa at 1.400 KN identifies the self-weight to be ap- proximately equal to the calculated buoyant weight of the bar- reute, ‘The maximum applied O-cell toads during the initial test and retest were 10,300 and 11,600 KN. respectively. The a: Curmulated upward movements of the top plate at these loads, ‘were about 7 and 10 mim, respectively. The downward move~ ‘ment of the bottom plate, presented in the lower diagram, was, uch larger than the upward movement. The accumulated ‘movements at the maximum loads were 45 and 58 mm, cor- responding (0 about 6% of the barrette width, The shape of the load-displacement curve for the bortom plate is essentially, Tinear below 10,000 kN. Beyond this load, a progressive in- Total Oeil Load (KN) bi Movement at Bottom 0-Cel Pate (re) (0-cen Expansion (rm) Total Cell Load (KN) FIG, 3, Load versus Movement of 0-Cell Plates crease of the bottom: plate displacement is indicated; however, no ultimate load ean be defined from the curve. ‘Adjusting for the 1 400-kN self-weight, the maximum force in side shear is 8,900 KN during the initial test and 10,200 kN. daring the retest. The 7-mm upward movement of the O-cell top plate at the maximum load includes the elastic compres- sion of the barrette, which is estimated to be about 3mm. ‘Thus, the shaft resistance was fully mobilized at or before a 4mm relative movement at the top of the barrette—top of sound concrete 12 m below grade, The observation confirms the general wisdom that shaft resistance is mobilized at very small relative movement and is independent of the size (di- ameter) ofthe shaft "The strains recorded by means of the individual strain gauges at Levels 1-3 in the barrette are presented versus the recorded O-cell pressure in Fig. 4 (with the initial reading taken as the "zero" reading). Notice, the three diagrams em- ploy different scales for the ordinates. At gauge Level 1, the four gauges show similar development with increasing” cell pressure, However, this is not the case at Levels 2 and 3. At Level 2, the strain gauges agree and disagree in pairs; whereas, all strains recorded at Level 3 differ from each other. “The differences in strain gauge values suggest that the load ing induced by the two O-cells is uneven. This is what would JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JULY 1999 / 568