Отправлено: 26.01.15 10:26. Заголовок: PREDICTION OF WATER-FLOODING PERFORMANCE IN CORE SCALE

PREDICTION OF WATER-FLOODING PERFORMANCE IN CORE SCALE: COMPARISON OF NUMERICAL SIMULATOR, NEURAL NETWORK AND CORRELATION
a Ghoodjani, E. 1; b Bolouri, S. H. a Sharif University of Technology, Tehran, Iran b Shahid Bahonar University, Kerman, Iran
ABSTRACT
A sensitivity analysis was performed on different parameters affecting recovery factor of water-flooding. A correlation was developed for estimation of the water-flood performance in core scale and constant injection rate experiments. The correlation was based on more than 230 numerical simulator runs for a wide range of porosity, permeability, viscosity ratio and pore size distribution in long core. Also, a neural network for prediction of water-flood recovery factor was created. The results show that the proposed correlation is reliable in a full range of parameters where the neural network fails to estimate water-flood performance in some intervals. The correlation may be used in reverse direction, if the recovery factor is known; the pore size distribution index can be estimated.