Please use this identifier to cite or link to this item:
|Title:||Characterisation of model uncertainties for laterally loaded rigid drilled shafts|
|Authors:||Phoon, K.-K. |
Limit state design/analysis
|Citation:||Phoon, K.-K., Kulhawy, F.H. (2005-02). Characterisation of model uncertainties for laterally loaded rigid drilled shafts. Geotechnique 55 (1) : 45-54. ScholarBank@NUS Repository. https://doi.org/10.1680/geot.22.214.171.124593|
|Abstract:||This paper presents a critical evaluation of model factors for laterally loaded rigid drilled shafts (bored piles). Both the lateral or moment limit and hyperbolic capacity are considered to make explicit the dependence of model factors on the criterion for interpreting 'capacity' from load test data. Although the hyperbolic capacity may be closest to the theoretical ultimate state or upper bound, results indicate that it generally does not produce a mean model factor of 1. When the measured capacity is interpreted consistently from load test data, the coefficient of variation (COV) appears to remain relatively constant between 30% and 40%. The range of the mean bias for the lateral or moment limit is 0·67 to 1·49, whereas that of the hyperbolic capacity is 0·98 to 2·28. Based on available data, a log-normal probability model appears adequate for reliability analysis. Laboratory-scale load tests conducted in uniform soil deposits prepared under controlled laboratory conditions are ideal for establishing benchmarks on the probable magnitude of uncertainty arising from model idealisations alone. However, the limited range of geometric and geotechnical parameters in a laboratory load test database may not produce a representative mean model factor. A field load test database typically contains more diverse geometric and geotechnical parameters, but it entails an unknown degree of extraneous uncertainties. A comparative study indicates that model statistics are surprisingly robust and appear not to be seriously affected by the above concerns (possibly because of normalisation). Model factors from drained analysis seem to be more variable than those from undrained analysis. A more detailed examination indicates that the higher COV of about 40% for these drained model factors arises because they are not completely random. There are reasons to believe that applying a more complete force system for drained analysis could minimise some of the undesired correlations and reduce the COV to a level comparable to undrained analysis.|
|Appears in Collections:||Staff Publications|
Show full item record
Files in This Item:
There are no files associated with this item.
WEB OF SCIENCETM
checked on Jul 16, 2018
checked on Jun 29, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.