A new prediction model of the jet grouting column diameter for three jet grouting systems

Jet grouting column is widely used in ground improvements of hydraulic and geotechnical structures. Estimation of the diameter of the jet grouting column is indispensable to the design of those structures. Existing prediction models of the jet grouting column perform poorly when applied to the recently emerged double and triple-jet grouting systems and cannot quantify the prediction uncertainties. This study has compiled a global-wide jet grouting database and proposed an explicit model that is suitable for all three popular jet grouting systems (i.e., single, double, and triple). The Bayesian inference method was employed to estimate the regression parameters for the proposed model and quantify the prediction uncertainty. The proposed model is compared with a widely used empirical model through a cross-validation procedure. The comparisons show the proposed model is free of systematic errors and has much higher accuracy than the existing method, although the two have similar complexities in terms of the number of unknown regression parameters. The new model can reasonably quantify the prediction uncertainty in the sense that the 95% confidence interval of jet grouting column diameter covers the actual value with a probability of around 95%. To facilitate practice use, analytical formulas have been derived to calculate the statistics of jet grouting column diameter. This unique feature developed here to quantify the prediction uncertainty provides a potentially fundamentally new tool for engineers to estimate the jet grouting column diameter and assess the associated prediction errors.

Automated summary

This study compiles a global-wide jet grouting database and proposes an explicit model suitable for different jet grouting systems. The Bayesian inference method is used to estimate regression parameters and quantify prediction uncertainty. The proposed model is compared with an empirical model, showing higher accuracy and the ability to reasonably quantify prediction uncertainty.