Experimental Investigation and Prediction Models for Thermal Conductivity of Biomodified Buffer Materials for Hazardous Waste Disposal

AbstractMicrobe-aided modification to certain geotechnical properties of soil has garnered appreciable amount of interest among the researchers in recent days. In particular, microbial influence on thermal conduction properties of soil is relevant in engineered-barrier applications as in case of a deep geological repository (DGR). In view of this, in the present investigation, a radiation resistant and extremophilic microbial species Deinococcus radiodurans has been studied for its possible influence on thermal conductivity of several geomaterial samples. Thermal conductivity measurements for varieties of geomaterials with varying moisture content and bacterial dosage are made using an indigenously fabricated thermal needle probe, efficacy of which has been established through calibration. In general, it was observed that microbial addition enhanced the thermal conductivity of soil. This observation can be attributed to any possible biogeochemical reactions in the geomaterials because of the bacterial addition. Furthermore, based on the experimental data, the following three artificial intelligence (AI) techniques are used to develop thermal conductivity prediction models for geomaterials and to quantify the influence of microbial addition on soil thermal conductivity: (1) multi-gene genetic programming (MGGP), (2) multivariate adaptive regression splines (MARS), and (3) functional networks (FN). The models proposed are found to deliver satisfactory performance in terms of statistical parameters and can be used to predict soil thermal conductivity.