Gas Lift Optimization Study in an Iranian Oil Field- A Case Study

The pressure of the fluid inside the reservoir gradually drops as production from it rises. Oil output through production wells is reduced as a result of this pressure drop. Oil should be produced using artificial gas lift techniques when the fluid pressure in the reservoir drops to the point where it can no longer transfer and migrate to the earth's surface. By injecting gas into the well's fluid through the annular space, this technique lowers the well's pressure and enhances oil output. One issue with the gas lift is that there is a limited supply of gas. This means that the gas must be split and optimized among the wells so that the needed amount does not exceed the available gas; otherwise, production will be decreased and expenses will rise. In order to finish the optimization process and speed up the simulation, this article addresses the construction of proxy models that would replace these models with the intended simulator. The link between the input and output parameters is constructed using two potent neural networks: forward waterfalls and multilayer perceptrons. The gas injection rate is then optimized using the Ant colony technique. The objective is to maximize net present value. The optimization results showed a quicker convergence in a shorter amount of time. The results showed that the best and most accurate model for predicting the behavior of the model under study is the cascade forward model.

Keywords:

gas lift , proxy models , optimization , Maximizing net present value , Ant colony algorithm