This study presents the development and validation of a computational model which simulates the propagation of a smouldering front through a porous medium against unique experiments in coal tar and sand. The model couples a multiphase flow solver in porous media with a perimeter expansion module based on Huygens principle to predict the spread. A suite of two-dimensional experiments using coal tar-contaminated sand were conducted to explore the time-dependent vertical and lateral smouldering front propagation rates and final extent of remediation as a function of air injection rate. A thermal severity analysis revealed, for the first time, the temperature–time relationship indicative of coal tar combustion. The model, calibrated to the base case experiment, then correctly predicts the remaining experiments. This work provides further confidence in a model for predicting smouldering, which eventually is expected to be useful for designing soil remediation schemes for a novel technology based upon smouldering destruction of organic contaminants in soil.