Oil foam interaction

Foam injection is a promising method to improve sweep efficiency for gas flooding in enhanced oil recovery (EOR). The effect of oil on foam is a complex phenomenon and for this reason three phase foam simulations have shown to be difficult in the past. Initially the objective of this simulation study was getting a better understanding of relating composition paths for foam displacements on the ternary diagram to the gas mobility reduction factor on the same diagram (Dharma, 2013). However, there were unexpected numerical problems during the course of the study and consequently the objective of the thesis has been changed. The new focus of the research is investigating the different problems that are encountered with three-phase foam simulations. Additionally an attempt is made to overcome the problems by changing the numerical method and adding a capillary diffusion coefficient. Two cases are examined, both showing oscillations using the STARST M foam model. Firstly, a Finite Volume Method (FVM) model has been created. This simple single point upstream scheme without diffusion is tested and used to study the challenging cases of Namdar Zanganeh et al. (2009) and Dharma (2013). In both researches oscillations between the foam and no-foam region appear. Research on the case of Namdar Zanganeh et al. is presented; the same difficulties occur using the FVM method. The oscillations cannot be overcome by using extremely small time steps and grid size. Hence this is a fundamental problem for finite volume grids with a single point up-wind scheme and no diffusion. The FVM script is extended with a capillary diffusion coefficient to approximate the effect of capillary pressure gradient and Finite Element Method (FEM) is in an attempt to overcome the problems encountered. The addition of the capillary diffusion coefficient makes the oscillations between the foam and no-foam region disappear and shocks to states predicted by Namdar Zanganeh et al. are present. The forward shock is broad, but it appears to be a shock. The velocity of the slow shock is too fast. For the FEM simulation COMSOL is used, with this numerical method no oscillations occur. However, a shock to a state not predicted by Namdar Zanganeh et al. (2009) is present. Moreover, a travelling ’shock’ does not advance at all, on the time scale studied. Additionally, research on the case of Dharma (2013) is presented. For this case the oscillations disappear using the FVM. This means the oscillations are a result of the higher order FVM used by Dharma. However, the result is not optimal since new complications arise: non-monotonically moving shocks. A study on this abnormality is carried out and results obtained are for a number of cases compared to the work of Dharma. The FVM results in this research show similar behaviour as Dharma found. An intermediate oil bank is found at a different saturation this most likely because of the use of a different oil relative permeability function.