Simulation of the performance and remediation of imperfect pile groups
3 COMPARISONS BETWEEN PIGS AND OTHER PROGRAMS FOR IDEAL
PILE GROUPS
Comparisons were made between the results from PIGS and from two established and commercially available pile group programs, PIGLET (Randolph, 1987) and DEFPIG (Poulos, 1980). For this compari-son, an ideal pile group in which all piles are identical was analyzed. The group contained nine piles, in a symmetrical 3 × 3 configuration, each 20 m long and 1 m in diameter, with a centre-to-centre spacing of 3 m in both directions. The idealized soil profile con-sisted of a uniform stiff clay layer 20 m deep, under-lain by a very stiff layer extending to considerable depth. The drained Young’s modulus values of the two layers were 50 MPa and 200 MPa respectively, while the ultimate skin friction value for the upper layer was 50 kPa. The ultimate end bearing pressure within the lower layer was 2.7 MPa. In the PIGS analysis, a hyperbolic factor Rf of 0.5 was assumed for the single piles. Two cases of loading were considered:
1. A central vertical load;
2. A central vertical load V together with a moment (equal to 2V) in the “x” direction.
The computed load-settlement behaviour obtained by PIGS, PIGLET and DEFPIG is shown in Figure 2 for the central vertical loading case. At relatively low loads, the three solutions agree closely, but as the load increases towards failure, the solution from the linear PIGLET analysis departs from the other two solutions, as would be expected. The non-linear PIGS
solution gives greater settlements than the non-linear DEFPIG solution, because of the assumed hyperbolic single pile response in PIGS, whereas the DEFPIG analysis assumes, in effect, an elastic-plastic single pile response.
Figure 3 compares the computed maximum pile load within the group from a purely elastic analysis.
The agreement between the three solutions is reasonably close.
Figure 4 compares the load-settlement curves for the case of combined vertical and moment load-ing. The agreement between the PIGS and DEFPIG load-settlement curves is again reasonable. The three analyses are also found to give similar results for the maximum pile loads.
From these comparisons, it is concluded that the simplified PIGS analysis provides a reasonable means of prediction the load-settlement behaviour of ideal pile groups.
4 APPLICATION TO BUILDINGS IN HONG KONG
4.1 Introduction
Construction of two high-rise residential blocks, each planned to contain about 650 apartments, was com-menced in Shatin, Hong Kong, in mid-1999. Around that time, there had been revelations of improper pil-ing practices in some projects in Hong Kong, and this gave rise to a heightened sense of awareness about building performance. In October 1999, a program of settlement monitoring was commenced on the two buildings, labelled as Blocks D and E, both of which by that stage had reached the 17th Floor. About 2 months later, it was observed that both buildings were settling unevenly, with each block tending to tilt as a rigid body. Of particular concern was the fact that there was a 6-storey secondary school very close (about 6 m) to the two blocks. Consequently, a program of investigative coring was commenced in late 1999 through a number of the bored piles supporting the buildings. As a result of this coring program, it was deduced that many of the piles for these buildings had not been constructed as per the design requirements, and not in accordance with the documented piling records. These deductions led to a thorough investigation of the foundations of the buildings and to an assessment of possible future courses of action.
The following sections describe one of the inde-pendent assessments carried out, including an evalu-ation of the as-built foundevalu-ations, an assessment of proposed remedial measures, and a consideration of various risk factors involved in undertaking remedial works in such circumstances.
Figure 2. Comparison of load-settlement curves for verti-cal loading.
Figure 3. Comparison of computed maximum loads for vertical loading only.
MAXIMUM PILE LOAD FOR VERTICAL LOAD ONLY
0 0.05 0.1 0.15
ANALYSIS METHOD PROPORTION OF LOAD ON MOST-HEAVILY-LOADED PILE
PIGS9-3Q DEFPIG PIGLET
Figure 4. Comparison of load-settlement curves for combined vertical load and moment.
VERTICAL LOAD + MOMENT
0
4.2 Geotechnical conditions
Prior to commencement of construction of Blocks D and E, a total of twenty two exploratory boreholes had been drilled in the vicinity of Blocks D and E, and Figure 5 shows a plan of the foundation piles and the borehole locations. A typical geological section is presented in Figure 6, the location of which is shown on Figure 5.
Based on the available information, the subsurface conditions for Blocks D and E comprised the follow-ing strata:
1. Fill—Highly variable ranging from silty fine sand to gravel and cobbles of granite and concrete, described as loose to dense.
2. Marine Deposit—Typically described as loose silty fine to coarse sand with some shell fragments, and also described as soft grey silty clay/clayey silt.
3. Alluvium—Highly variable unit ranging from soft sandy silt, loose silty sand to medium dense to dense gravel. This unit was typically described as firm to stiff sandy, silty clay and loose to medium dense silty, fine to coarse sand.
4. Colluvium—Material described as Colluvium was encountered in two of the boreholes and com-prised sandy angular to sub-angular medium to coarse gravel.
5. Completely to highly decomposed quartz monzonite—Saprolite that was generally des-cribed as being medium dense to very dense, slightly gravelly, silty sand/sandy silt and stiff to very stiff sandy clay. Corestones of Grade II Quartz Monzonite were encountered within the Grade V/IV material in a number of the bore-holes, in particular those drilled in the vicinity of Block E.
6. Moderately to slightly decomposed quartz monzonite—typically described as moderately strong to strong with medium- to widely-spaced joints.
The site was in relatively close proximity to a river, and the groundwater was typically about 2 m to 3 m below existing ground level.
4.3 Foundation details
As shown in Figure 5, each of the 41-storey blocks was supported on 18 bored piles, designed to be 2.3 m in shaft diameter, with a base belled out top 3.8 m. The piles were designed as end bearing piles, to be founded in Grade II (slightly weathered) granite, at a depth ranging between about 35 m and 46 m.
4.4 Measured settlements
When the buildings had reached the 17th storey, a program of settlement monitoring was commenced on the buildings, and Figure 7 shows typical time- settlement readings for one of the blocks. The non-uniform settlements are clearly evident from this figure. It is interesting to note that the coring inves-tigation works caused a temporary acceleration in settlement of the buildings, which played an influen-tial role in the ultimate decision made with respect to the buildings.
Figure 5. Plan of pile and borehole locations.
13
Figure 6. Geotechnical profile for section 1–1.
Fill Marine Deposit Alluvium BHD-12 B-148 B-152 B-155 B-158 B-161 B-164
Extremely Weathered Rock (V)
Sl. weathered rock (II)
Figure 7. Time-settlement plot for typical points in Block E.
016Oct99 20Nov99 25Dec991Jan00Date 19Feb0026Feb00 17Jun00
-100
4.5 Investigative coring results
A program of investigatory coring through the bored piles was instigated in December 1999, and this pro-gram revealed a number of deficiencies in both the pile length and the founding conditions at the base of the piles, and construction was halted at the 34th floor, very early in 2000. The conclusions of the independent consultant carrying out the coring and the assessment of pile condition are summarized in Table 1.
Only four of the total of 36 piles were compliant with the construction specifications which required founding on Grade II rock. The pile lengths were up to 16 m or more short of bedrock level. In addi-tion to the geotechnical deficiencies, six of the con-crete cores that were taken from the piles and tested showed low strengths, ranging between 22.5 MPa and 2 MPa, compared to the nominal strength of 35 MPa.
The findings of the coring investigation led to more detailed investigations of the foundation perform-ance and of measures that could be taken to enhperform-ance this performance. One of the independent analyses, undertaken by the author, is described below.
4.6 Analysis of building performance 4.6.1 Approach adopted
The computer program PIGS was used to carry out the analyses of foundation performance for Blocks D and E. The following procedure was used:
• From the available geotechnical information, sepa-rate geotechnical models were developed for each of the 18 bored piles supporting each block.
• The geotechnical parameters for each of the layers within each geotechnical model were assessed initially from conventional correlations with Standard Penetration Test (SPT) data or with rock type.
• From the available coring information, models were developed for each of the piles.
• The geotechnical and pile models were input into PIGS, and the foundation behaviour under the dead loading, up to Floor 34, was analyzed.
• Because of the apparent effects of the core drilling in causing additional building settlements, an attempt was made to exclude these settlements from those due to the building load. This was done on the basis of extrapolating the measured settlement-time rela-tionship, prior to the effects of the core drilling, to the end of January 2000, when it was assumed that the settlements due to the loads from the 34 floors of building would have been completed. For each of the settlement markers, these settlements were subtracted from the measured settlements to obtain an estimate of the settlement due to the core drill-ing. These settlements should be treated separately from those due to the building loads, as they would not recur if no further core drilling were done.
• The inferred pile settlements due to core drill-ing were then incorporated as “free-field” ground movements into the PIGS analysis for the founda-tion settlement behaviour. The computed settlement behaviour of the foundation, from Floor 17 (when the measurements began) to Floor 34 (when the con-struction halted) was compared with the measured values. The geotechnical parameters of the soil lay-ers were modified, and the analyses repeated until a reasonable match was found between the measured and computed settlement behaviour.
• Once a reasonable match was obtained between measured and computed settlements, the foundation behaviour for various cases was analyzed via PIGS.
The settlements due to core drilling were estimated at the various settlement marker locations, for each of the two blocks. From these values, the settlement at each pile location due to the core drilling was obtained. On this basis, the assessed additional pile settlement due to the core drilling ranged between 2 and 10 mm for the piles in Block D, and between 0 and 25 mm for the piles in Block E.
It was assumed that the shaft diameter of the exist-ing piles was 2.3 m and that there was no bell at the base of the shaft. The length of each pile was taken as the value indicated during the investigative coring, and the Young’s modulus of the pile was estimated from the measured unconfined compressive strength of the concrete cores.
For the remedial piles, it was assumed that, for an option involving additional bored piles, the shaft Table 1. Summary of inferred pile founding conditions
from coring.
Pile founding
Number of piles
condition Block D Block E
Pile founded on 3 1 inferior rock
Pile founded on 1 12
inferior rock, without sediments
Thick layer of soft 4 0
material below pile base
and base diameters would be 2.3 m and 3.5 m respec-tively, and then piles would extend 0.8 m into slightly decomposed rock. For an alternative option involving prebored H-piles, the shaft diameter of the remedial piles was assumed 0.61 m. In the latter case, the piles were to be founded 9 m below the top of the slightly decomposed rock.
The assumed loadings were as follows:
Dead load—11.612 MN per floor.
Live load—2.551 MN per floor.
Moment due to wind load at 34th floor level—35.5 MNm and 529.2 MNm in the two horizontal directions.
4.6.2 Calibration of PIGS models
Figures 8 and 9 show the measured and computed settlements for various sections across Blocks D and E, as obtained from the final calibration runs.
In order to reproduce the observed tilt, it was found necessary to introduce some eccentricity of applied loading.
While the agreement was not perfect, it was found possible to obtain a reasonable fit by adopting the fol-lowing correlations with the average SPT value N:
Young’s modulus: Es= αN MPa
where α = 2.8 for Block D, and 4.2 for Block E.
Ultimate shaft friction: fs= 1.0 N kPa
Ultimate end bearing pressure: fb= 0.1 N MPa.
The loading eccentricities, with respect to the geometric center of the foundation system, were as follows:
For Block D: 0.05 m in the “x” (east–west) direction, and –2.4 m in the “y” (north–south) direction.
For Block E: −1.5 m in the “x” (east–west) direction, and 0.0 m in the “y” (north–south) direction.
4.7 Assessment of as-designed foundations
For reference purposes, a PIGS analysis was carried out for the hypothetical case where it was assumed that the bored piles in the original foundation layout for Blocks D and E were founded 0.8 m into Grade III or better rock. In this case, no investigative core drilling settlements were applied to the piles. The fol-lowing stages were modelled:
• Stage 0—Building constructed to 34 storeys
• Stage 1—Building constructed to 41 storeys (full dead load of 476.1 MN applied)
• Stage 2—Full live load of 104.6 MN applied.
Figure 8. Measured and computed settlement profiles—
Block D.
D/BP12 D/BP7 D/BP6 D/BP15
30
D/BP17 D/BP4 D/BP6 D/BP14
30
* SETTLEMENT FROM 17/F TO 34/F PILE No
D/BP18 D/BP5 D/BP7 D/BP13
Figure 9. Measured and computed settlement profiles—
Block E.
E/BP15 E/BP6 E/BP7 E/BP12
30
E/BP18 E/BP5 E/BP7 E/BP13
30
* SETTLEMENT FROM 17/F TO 34/F PILE No
E/BP17 E/BP4 E/BP6 E/BP14
The results of the analyses are summarized in Tables 2 and 3. It can be seen that had the build-ing foundations been constructed as-designed (i.e.
founded 0.8 m into Grade III or better rock), the foundation system factor of safety under full dead and live load would have been greater than 3 and the maximum overall tilts of the foundations would have been significantly less than 1/400 (in fact, less than 1/1800) for both Blocks D and E. The maximum set-tlement of Block D and E were estimated to be 26 mm and 20 mm respectively.
4.8 Assessment of as-built foundations
The PIGS program was also used to compute the behaviour of the as-built foundation systems had construction proceeded without any enhancement works. This was done by simulating the addition of dead loading for the final 7 floors to the 41st floor, and then the addition of live loading, simulating the full occupancy loading situation. Analyses were also carried to simulate the case of wind loading when the buildings were at the 34th floor level.
Table 4 shows the computed maximum and mini-mum settlements, the overall tilts and the overall factor of safety against failure, for each block. The following observations can be made from this table:
1. Foundation system factor of safety is less than 3 for Block D under dead loading from 41 floors.
2. Foundation system factor of safety is less than 3 for Block E under dead loading from 34 floors.
3. Failure of the Block E model occurs (overturning) under dead + live loading from 41 floors.
4. Overall tilt of Block D is greater than 1/400 under