Abstract:
The unconfined compressive strength (qu) is a critical design parameter in the design of jet-grouted piles
for ground improvement projects. This paper presents an evaluation of index properties affecting the
unconfined compressive strength of jet-grouted samples for the Bangkok soils in Thailand. A database
containing results of field investigation and laboratory testing for jet-grouted samples was created for this
evaluation. The jet grouted samples were collected using a large diameter jet grouting technique from the
MRTA Blue Line and Orange Line Projects in Bangkok, Thailand. This study focuses on two parts of
analysis: (1) prediction of qu for the jet grouted samples using index properties of the soils obtained from
the database, and (2) correlation between qu and Modulus of Elasticity (E50) of the jet grouted samples. The
prediction of qu for the jet grouted samples using index properties of the soils was conducted using the
Artificial Neural Network (ANN) model, which is a well-known data mining algorithm. Various index
properties of the soil samples were selected in the ANN model to determine the most critical parameters in
the prediction of qu. It was concluded that unit weight, grading curve, and plasticity index could be used to
reasonably predict the qu using the ANN model. A correlation of E50 = 100×qu adopted by the Japan Jet
Grout Association is commonly utilized in Southeast Asia for the design of jet-grouting projects. This
correlation could largely depend on soil types and properties, jet-grouting techniques, and construction
procedures. A correlation of E50 = 50×qu was found in this study for the jet-grouted soil samples collected
in the database. It is recommended that this correlation be used for future jet-grouted projects in Thailand.