It is an established fact that substantial amounts of oil usually remain in a reservoir after primary and secondary processes. Therefore; there is an ongoing effort to sweep that remaining oil. Field optimization includes many techniques. Horizontal wells are one of the most motivating factors for field optimization. The selection of new horizontal wells must be accompanied with the right selection of the well locations. However, modeling horizontal well locations by a trial and error method is a time consuming method. Therefore; a method of Artificial Neural Network (ANN) has been employed which helps to predict the optimum performance via proposed new wells locations by incorporating reservoir properties and production data of previous wells. This study used the Artificial Neural Network (ANN) that has been programmed in a manner to predict the cumulative oil produced for a certain grid by providing the corresponding properties of the grid. The network has been validated with real data collected from a number of drilled hypothetical wells. Furthermore; the validated network used to simulate the field parts that have not been drilled yet, to predict the corresponding cumulative oil for each grid. Field-scale simulation has been carried out and new horizontal wells have been allocated using the validated prepared data by the Artificial Neural Network Algorithm and an approved Iraqi reservoir model. Finally, different optimization scenarios have been investigated on the overall field recovery performance.

In recent years the interest in fractured reservoirs has grown. The awareness has increased analysis of the role played by fractures in petroleum reservoir production and recovery. Since most Iraqi reservoirs are fractured carbonate rocks. Much effort was devoted to well modeling of fractured reservoirs and the impacts on production. However, turning that modeling into field development decisions goes through reservoir simulation. Therefore accurate modeling is required for more viable economic decision. Iraqi mature field being used as our case study. The key point for developing the mature field is approving the reservoir model that going to be used for future predictions. This can be achieved via History Matching. The production of this field is mainly from fractures, and it showed unfavorable decline during the production life from 1952. Thus, well modeling is necessary for developing the field, including history matching. This study employed a three-dimensional three-phase black-oil dual-porosity model by CMG-Builder/IMEX 2010 simulator using an Iraqi fractured, faulted reservoir model. History matching has been carried out to improve the model which was going to be used in the simulation study. Several modifications were required during the history matching, such as adjusting the vertical fracture permeability, the relative permeability, the aquifer properties, setting fault transmissibility, and the matrix and fracture compressibilities. The results of the History matching showed excellent agreement between the simulation and the historical profile. Furthermore the recent validated data by history matching have been used for the Iraqi reservoir model by running the CMG/IMEX simulator.