• Computational fluid dynamics model of two-phase heavy oil and air flow in a horizontal pipe

      Sanders, Nicholas E.; Ahmadi, Mohabbat; Awoleke, Obadara; Dandekar, Abhijit (2020-05)
      The production of heavy oil resources is becoming more prevalent as the conventional resources of the world continue to deplete. These heavy oil resources are being produced from horizontal wells and need to be transported in pipeline to processing facilities as a two-phase flow. Two-phase flow is important to the oil industry with the general focus being placed on light oil or water and gas flows. With little work having been done on two-phase heavy oil flow this study will examine these two-phase flows by recreating experimental data generated for heavy oil and air flow in a 1.5-inch diameter pipe and expand this data to include larger 2.875-inch and 3.5-inch pipes. A computational fluid dynamics model was generated to mimic the 1.5-inch diameter pipe used in the experiments. This model was validated for laminar and turbulent flow by using the same heavy oil properties from the original experiment and air respectively. The model was then run to simulate the given two-phase oil-air flows provided from the experimental data for the flow velocities that had pressure drop and liquid holdup data available. The two-phase results were compared to both the experimental data and the Beggs and Brill values for both pressure drop and liquid holdup. A 2.875-inch and 3.5-inch model were generated and the same process was followed for laminar and turbulent validation and then with a subset of four two-phase flow velocities. Without the availability of experimental data for the two larger size pipes the two-phase results were only compared to the Beggs and Brill values. Overall the results showed a good correlation to the laminar and turbulent flow in all three models with the turbulent flow showing the largest error for the pressure drop when the flow was in the laminar to turbulent transition zone for Reynolds numbers. The two-phase results showed to be in between the experimental and Beggs and Brill method values for the original 1.5-inch model and showed that as the gas flow velocity increased in the system the error grew for all three models. Given that the Beggs and Brill method values were generated based on experiments for water-air flow in a 1.0-inch pipe the values for the pressure drop in the 2.875-inch pipe and the 3.5-inch pipe were not unexpected and seemed to match well with an extrapolation of the experimental values. This study shows that a model can be generated to examine the two-phase flow behavior in horizontal sections of well and in pipelines on a computational basis. While these models are time consuming to generate and run with the increase in computing capacity available easily they can become more suitable than generating experimental setups for finding the same information. There will need to be more work done on heavy oil two-phase flow and additional experiments run for larger size pipes and two-phase flow to help tune these models but they do show promise for the future.