Perpustakaan Digital ITB

Cyclic Steam Stimulation is one method to recover heavy oil. Cyclic Steam Stimulation is usually used as a pilot project before implementing steamflood in the heavy oil reservoir. Moreover, for a remote field, Cyclic Steam Stimulation is a powerful method for a field with a heavy oil reservoir. It is because the field usually lacks surface facilities surrounding that area. A numerical solution (dynamic simulation) is usually used to simulate the Cyclic Steam Stimulation. However, a dynamic simulator for Cyclic Steam Stimulation needs exhaustive data preparation and a comprehensive study, resulting in demands for intensive manpower and computing resources, leading to additional cost and time. Moreover, the thermal process's complexity and uncertainty in some parameters make a numerical solution take longer. This study presents a mathematical solution combined with a statistical approach instead of a numerical solution by implementing a predictive model, i.e., Polynomial Regression and an Artificial Neural Network. This study is a computation and cheap alternative solution for full numerical simulation. The objective function of the predictive model is to determine cumulative oil production of Cyclic Steam Simulation. The predictive model is also designed to optimize fourteen uncertainty parameters. The experimental design uses Latin Hypercube Sampling to generate representative data with a minimal sample. It is generated from a basecase model with fourteen uncertainty parameters using CMG-CMOST. They are also trained and tested to get a suitable predictive model to approximate the numerical solution result. This study can approximate the numerical result well. It has testing R-squared for Polynomial Regression around 0.931. Meanwhile, the testing R-squared for the artificial neural network is 0.970. The result of this study may be used to find the suitable reservoir characters and well performance to be applied a Cyclic Steam Stimulation. Furthermore, this tool can be a preliminary tool to evaluate the feasibility of Cyclic Steam Stimulation before applying full numerical simulation. It can also determine the suitable steam volume and condition needed, as does the injection length and soaking time before the well is put into production.