Statistical Analysis of The Reproducibility of The Standard Penetration Test: A Case Study
One of the most widespread survey methods of geotechnical investigation is the Standard Penetration Test - SPT. Although this test is regulated by standards in several countries, even the standardized methodology may lead to highly variable data. In this case, the data variability is not caused by sampling different stratigraphic horizons; it is caused by disturbances of three other sources, namely: human, equipment and procedure. This research, based on a case study and in a statistical analysis, isolate the influence of soil properties to clarify the relevance of these other variability sources. The analysis, based on Bootstrap resampling method, considers 8 SPT surveys executed by the same workers using the same equipment and on a pre-established soil profile. As part of the paper methodology, each blow was monitored for its height and impact velocity with a high-speed camera and a topographical target. It was possible to note that, despite the standardizations described, the data are significantly different. In the sample studied, the energy transmitted to the soil is statistically higher than the Standardized one. Consequently, parameters of soil resistance might be underestimated and, thus, engineering solutions might become more expensive than necessary. Therefore, the results show that the use of SPT may not be sufficient, depending on the risk to be assumed by stakeholders.
Index Terms- Bootstrap, Reproducibility of SPT, Risk in engineering, Statistical Analysis.