Numerical and experimental study of the impact of temperature on relative permeability in an oil and water system.

Relative permeability is affected by several flow parameters, predominantly operating temperature and fluid viscosity. Fluid viscosity changes with temperature, which correspondingly affects the relative permeability. Temperature is believed to have a considerable effect on oil–water relative permeability and thus is a vital input parameter in petroleum reservoir development modelling. The actual effect of temperature on oil–water relative permeability curves has been a subject of debate within the scientific community. This is based on contradictory experimental and numerical results concerning the effect of temperature on oil–water relative permeability in literature. This study investigates the effect of temperature on multiphase flow physics in a porous media under varying temperature conditions. A computational fluid dynamics approach was adopted for a pore-scale study of the temperature effect on oil recovery factor under a water- and oil-wet condition. For the oil–water relative permeability investigation, a series of coreflooding experiments were conducted with well-sorted unconsolidated silica sandpacks, adopting the unsteady-state relative permeability method. The series of experiments were performed at different temperatures (range between 40 to 80 °C). Three levels of injection flowrates (0.5, 0.75 and 1.0 mL/min) and two oil viscosities (43 cP motor and 21 cP mineral oil – at 60 oC) were used in the study. A history matching approach using the commercial software Sendra was used to determine the oil-water relative permeability for each respective temperature, flowrate and oil viscosity. A support vector regression algorithm was later implemented for the machine learning modelling aspect of this work, which can predict reliable temperature dependent oil–water relative permeability. The pore-scale results showed that the displacement behaviour of water and oil-wet system is strongly affected by the contact angle with a profound effect on the oil recovery factor. The water-wet system resulted in about 35 – 45 % more oil recovery than the oil-wet system, with the unrecovered oil mainly adhering to the wall region of the pore bodies of the oil–water system. The results from all the experimental cases showed that the oil–water relative permeability is a function of temperature, water injection flowrate and oil viscosity. In addition, the experimental findings show a decreasing residual oil saturation of the more viscous fluid with increasing injection flowrate. The end-point water relative permeability varies slightly for the set of experiments, with the values higher for the less viscous oil under the same flowrate condition. Generally, the profile of oil and water relative permeability curve changes with varying oil viscosity and water injection flowrate at the same operating condition. This behaviour shows that the viscosity of oil is an important factor to be considered when selecting displacement flowrate to guarantee high oil production. Furthermore, an increment in temperature results in a corresponding rise in the relative permeability of both oil and water. Comparison of the experimental and machine learning results showed a good match, with consistency across all datasets. In addition to the machine learning model, this study proposes a modified empirical model using nonlinear least square regression for application in unconsolidated porous media. The output from this model can be applied for relative permeability prediction, preliminary evaluation in experimental design and as a valuable benchmarking tool for future laboratory experiments under varying temperature conditions