Perpustakaan Digital ITB

In the early life of reservoir, the hydrocarbons flow using natural energy and in effect, the pressure depletes periodically making it no longer able to flow naturally. This primary stage could only recover around 10% of the original oil in place. To recover the remaining oil in place, enhanced oil recovery is performed. Water injection has been widely used as EOR technique because of its convenient operation and efficiency in economical aspect. However, further studies of water injection found that there are significant differences in oil recovery caused by the water chemistry and salinity of injected water. Low salinity water injection is an enhancement of water injection in which the injected salinity is set at one point to achieve higher oil recovery. A lot of coreflood and field scale experiments were done to prove the effect of low salinity water injection. From those experiments, it is believed that performing low salinity water injection could add up to 25% incremental oil recovery. Low salinity becomes attractive in the past years because it has relatively low capital and operational cost, lower risk of souring and scaling, lower risk of corrosion, and its flexibility to be combined with other EOR techniques such as polymer flooding and other chemical EOR methods. In performing LSWI, it is needed to consider some injection parameters to prevent premature water breakthrough and formation damage. A sensitivity analysis is done to obtain the optimum low salinity water injection design for this case. However, sensitivity analysis could take a lot of time when performing a lot of possible cases. This paper aims to minimize simulation time and simplify future study of low salinity water injection by constructing a proxy model using commercial simulator CMG-CMOST. It is proven that the proxy model could help predicting LSWI performance better as it is easier and faster compared to conventional simulator. Although the approximation from proxy model is not highly accurate, but it is reliable enough to be used in selecting possible best cases of LSWI performance before running them in other simulators.