Abstract:
Shaft resistance is a major design factor for piles supporting structures such as
transmission towers, harbor structures and offshore platforms. Several studies have
been conducted to correlate the estimated capacity of piles with the actual capacity
determined from load test in compression. However, few studies focused on the
determination of the tension capacity of piles. The main objectives of this study are to
compare the estimated tension capacity of bored piles with results obtained from
uplift load tests and to evaluate the current design methods used to determine the
uplift capacity of bored piles.
In this study, uplift load tests were conducted on 21 bored piles in several
sites. Capacity of the piles was determined from load test data. Capacity of the piles
was also estimated from the sub-soil characteristics using five methods–Meyerhof
(1968), Murthy (1992), Tomlinson (1977), the German Code of Practice (DIN 4014),
and British/American Method (1974). The first three methods use two soil
parameters–cohesion, c and angle of internal friction, φ. The British/American
method uses δ (based on φ) while the German Code of Practice (DIN 4014) uses only
SPT N-value for determining the pile capacity.
The results of the pile load test correlate reasonably well with the results
estimated from theory. Linear regression of the experimental data showed that the
capacity estimated by Meyerhof and Murthy equation and the German Code of
Practice (DIN 4014) need to be multiplied by factors of 1.08, 1.22 and 1.28
respectively to get the actual pile capacity. The Tomlinson equation and the
British/American method over estimated the pile capacity. The capacity estimated by
the Tomlinson equation and the British/American method need to be multiplied by
factors of 0.86 and 0.81 respectively to get the actual pile capacity. As far as
regression is concerned, the British/American method provided the best regression.
The results of these experiments are based on 21 bored pile load test data. The
large scatter in the experimental data suggests that a larger sample size is required for
better correlation. Nevertheless, the study conducted in this thesis provides an
appropriate ground work for further study in this area.
