Applying Artificial Neural Networks to Predict the Effects of Microbial Degredation on Physical Reservoir Parameters in Geothermal and Carbon Capture and Storage Settings

Abstract

In this research, the influence of microbial degradation on the porosity and permeability of sedimentary reservoirs is explored. Bentheimer Sandstone samples are exposed to a a total of nine different metabolite concentration and temperature conditions. Relative impact of initial porosity and permeability values and temperature and concentration conditions on changes in porosity and permeability are explored using an artificial neural network. Predictive ability of the neural network is limited but performance should improve with increased sample size and addition of more input parameters. Liquid permeability of Bentheimer sandstone decreases up to 40 percent. Both dry and liquid porosity decrease up to 10 percent. In situ mobilisation of fines could be a mechanism that explains decrease in effective permeability and porosity. Further insight into the physical and chemical processes that govern porosity and permeability change could be gained by expanding the research to include more controlled parameters and increase sample sizes.