Verification of geotechnical reliability using load tests and integrity tests - Risk and Reliability in Geotechnical Engineering - page 524

and afterward setup. Construction effects and setup effects are worth a particular mention
because these effects can introduce many mechanisms that cause a large scatter of data. One
example of construction effects is the influence of drilling fluids on the capacity of drilled
shafts. According to O'Neill and Reese (1999), improper handling of bentonite slurry alone
could reduce the beta factor for pile shaft resistance from a common range of 0.4-1.2 to
below 0.1. Setup effects refer to the phenomenon that the pile capacity increases with time
following pile installation. Chow et al. (1998) and many others revealed that the capacity of
driven piles in sand could increase from a few percent to over 200% after the end of initial
driving. Such effects make the pile capacity from a load test a “nominal” value.
The variability of properties of the soil and the pile concrete is affected by the space over
which the properties are estimated (e.g., Fenton and Griffiths 2003). The issue of spatial
variability of soils is the subject in several chapters of this topic. Note that, after including
various effects, the variability of the soils at a site may not be the same as the variability of
the pile capacity at the same site. Zhang and Dasaka (2010) investigated the spatial variabil-
ity of a weathered ground using random field theory and geostatistics. The scale of fluctua-
tion of the depth of competent rock (moderately decomposed rock) beneath a building block
is approximately 30 m, whereas the fluctuation scale of the as-build depth of the piles that
provide the same nominal capacity is only approximately 10 m.
Table 14.2 lists the values of the COV of the capacity of driven piles from load tests in nine
sites. These values range from 0.12 to 0.28. Note that the variability reported in the table is
among test piles at one site; the variability among production piles and among different sites
may be larger (Baecher and Rackwitz 1982). In this chapter, a mean value of COV = 0.20 is
adopted for analysis. This assumes that the standard deviation is proportional to the mean
pile capacity.
Kay (1976) noted that the possible values of the pile capacity within a site favor a log-
normal distribution. This may be tested with the data from 16 load tests in a site in southern
Italy reported by Evangelista et al. (1977). Figure 14.1 shows the measured and assumed
normal and log-normal distributions of the pile capacity in the site. Using the Kolmogorov-
Smirnov test, both normal and log-normal theoretical curves do not appear to fit the observed
cumulative curve very well. However, it is acceptable to adopt a log-normal distribution for
mathematical convenience, since the maximum deviation in Figure 14.1 is still smaller than
the 5% critical value. The tail part of the cumulative distribution of the pile capacity is of
more interest to designers since the pile capacity in that region is smaller. In Figure 14.1 , it
can be seen that the assumed log-normal distribution underestimates the pile capacity at a

Table 14.2 Within-site viability of the capacity of driven piles in eight sites

Number
of piles

Diameter
(m)

Length
(m) Site Soil COV Reference

Ashdod, Israel 12 - - Sand 0.22 Kay (1976)

Bremerhaven,
Germany

9 - - Sand 0.28 Kay (1976)

San Francisco, USA 5 - - Sand and clay 0.27 Kay (1976)

Southern Italy 12 0.40 8.0 Sand and gravel 0.25 Evangelista et al. (1977)

Southern Italy 4 0.40 12.0 Sand and gravel 0.12 Evangelista et al. (1977)

Southern Italy 17 0.52 18.0 Sand and gravel 0.19 Evangelista et al. (1977)

Southern Italy 3 0.36 7.3 Sand and gravel 0.12 Evangelista et al. (1977)

Southern Italy 4 0.46 7.0 Sand and gravel 0.14 Evangelista et al. (1977)

Southern Italy 16 0.50 15.0 Clay, sand, and

gravel

0.20 Evangelista et al. (1977)