Automatic Well Selection Method for Acid-fracturing in Kazakhstan R Oilfield Carbonate Reservoir

摘要

In order to increase success rate and production of acid-fractured wells in Kazakhstan R oilfield, this paper compared various well-selection mathematic methods. Artificial neural network was chosen as a better way to predict production rate and finally choose target wells. Besides, some problems of this algorithm, such as normalization of input data, infection of order, were optimized to make it possible to be realized. With realistic data, sample database was set up and predictable model was also gotten. It's proven that the predicted results are very close to realistic data. Also, we analyzed the sensitivity of different parameters, and some useful conclusions were given: acid-fracturing is more effective for low and medium permeability reservoirs; if each layer could be treated separately, wells with large permeability differential will be better choice for acid-fracturing; bottom hole pressure is not the higher the better.