Modelling stone columns under a soil–cement bed using an artificial neural network. Issue 1 (25th February 2021)

• Record Type: Journal Article
• Authors: Das, Manita; Dey, Ashim Kanti
• Is Part Of: Proceedings of the Institution of Civil Engineers. Volume 174:Issue 1(2021:Feb.)
• Page Start: 42
• Page End: 58
• Publication Date: 2021-02-25
• Subjects: columns -- geotechnical engineering -- models (physical); Soil stabilization -- Periodicals; Geotechnical engineering -- Periodicals; Grouting (Soil stabilization) -- Periodicals; 624.1513605
• DOI: 10.1680/jgrim.18.00092
• Languages: English
• ISSNs: 1755-0750

Abstract

Abstract : The bulging effect of stone columns under a vertical load can be restricted by placing a stiff soil–cement compacted layer over the stone columns. The technique also improves the load-carrying capacity of the stone columns considerably. A series of laboratory experiments was conducted to obtain the load-carrying capacity of stone columns by varying parameters such as the thickness of the soil–cement bed, spacing between stone columns, length of stone columns and settlement due to loading to determine the bearing capacity. To avoid a complex interaction between the input variables, an artificial neural network model was adopted to predict the load-carrying capacity. The prediction efficiency of the model was found to be superior to that of a multi-variable regression model. Finally, a neural interpretation diagram was developed, from which the relative effect of an individual input parameter could be visualised.