In the present study, improved mathematical and numerical models are developed in order to simulate the influence of important reservoir (effective porosity and longitudinal dispersivity), fluid and microbial kinetic (Monod half-saturation constant, substrate to biomass yield coefficient and substrate inhibition) parameters on dynamics of in-Situ microbial enhanced oil recovery (MEOR) processes using Bacillus sp. in a typical sandstone reservoir. This numerical model is a novel attempt to consider the coupled effects of heat transport and substrate-inhibition kinetics on reactive-transport of microbes, substrates and biosurfactants as against the conventional models, thus making the prediction of MEOR processes more realistic and may assist in developing suitable strategies to enhance oil recovery at field scale. In addition, the present model also investigates the changes in relative permeabilities of reservoir fluids (oil and water) and fractional flow of water under the influence of nonisothermal condition, substrate inhibition, along with variations in pore-size distribution index, oil to water viscosity ratio and crude oil API gravity within reservoir during microbial flooding for estimating residual oil recovery. The present model results match significantly well, when verified, with the existing analytical and experimental results, thus proving the model's reliability. The results suggest that under nonisothermal conditions, biosurfactant production is maximum (77.8 mg ml−1) at effective reservoir porosity 23.4%, with low substrate half-saturation constant and high substrate-to-biomass yield coefficient values of 0.018 mg ml−1 and 0.8, respectively. Though biosurfactant production is reduced by more than 80% due to substrate inhibition, which is maximum at inhibition constant value 0.1 mg ml−1, residual oil-saturation is decreased by more than 90% at the end of MEOR process, thus enhancing oil displacement, mobility and hence, improved crude oil recovery. The reactive-transport of substrates, microbes and biosurfactants at injection mean fluid velocity of 1.68 m day−1, are also found to be highly sensitive to reservoir longitudinal dispersivity (αL) and maximum biosurfactants are produced (9.74 mg ml−1) at threshold αL value, 0.37 m, which corresponds to reservoir length of 50 m. Besides, a strategy is suggested to improve the oil recovery efficiency of the present in Situ MEOR approach by characterizing the reservoir for its pore size distribution index, and formation fluids (oil and water) for their viscosity ratios and crude oil API gravities, under nonisothermal and substrate inhibition conditions. © 2020 Elsevier B.V.