Perpustakaan Digital ITB

Steam flooding is a conventional thermal EOR method that has been proven to gain a significant recovery in many heavy oil reservoirs around the world. However, there are many challenges in order to optimize steam flood project. They are mainly associated with uncertainties of reservoir properties. This study will investigate the optimum well spacing scenario on the 5-spot pattern steam flooding by considering the technical parameters and economical aspect. The effect of each parameter on the performance or objective function of the steam flooding project is determined by constructing a predictive proxy model. The performance of the 5-spot pattern steam flooding project was determined through an objective function, in this case the profitability index (PI). It is an objective function to consider the economic aspect, which is indirectly influenced by production performance. By using CMG-CMOST, Latin Hyper Cube Sampling (LHS) method used to generate 2570 experiments. They were used to build a predictive proxy model using Multi-Layer Neural Network. Quality control of experimental data was also carried out to build a good predictive proxy model by considering the R-squared value in the verification and training data. From the experiment data, a reliable proxy model was found to have excellent accuracy between reservoir simulation and proxy model – the R-squared is 0.9495. In conclusion, the proposed proxy model can be an alternative tool to determine the optimal well spacing on a 5-spot pattern of steam flooding projects.