Use of data mining in design of soil improvement by jet grouting

Abstract

This paper addresses one of the main issues related to Jet Grouting (JG) technology, that is, the design of the mechanical properties of the soil-cement mixture. Thus, one of the most powerful Data Mining (DM) algorithms is applied, that is, Support Vector Machine (SVM), towards to the development of a new and more accurate approach for Uniaxial Compressive Strength (UCS) and stiffness prediction of both Jet Grouting Laboratory Formulations (JGLF) and soilcrete mixtures. The obtained results show that SVM algorithm can be used to accurately predict both strength and stiffness of JGLF. Related to soilcrete mixtures, it is shown that the SVM algorithm, despite some of the difficulties found, can give an important contribution for a better understanding of JG technology. Based on a detailed Sensitivity Analysis (SA) some important observations were made, which certainly will contribute for JG technical and economic efficiency improvement