Field Production Optimization in Constrained Water Temperature at Surface Facility Using Machine Learning and Genetic Algorithm

Abstract

Wastewater from all oil and gas industries in Indonesia must comply with the quality standard set by the environmental minister regulation Number 19 Year 2010 before it can be released into the water bodies. One requirement is for the temperature of wastewater to be below 113°F (45 deg Celsius). Production fluid from Seruni field is being processed at Seruni GS. The produced water from the field will be treated until the quality meets the requirement before being discharged to the nearby canal. If the outlet temperature of the produced water at the compliance point is about to exceed the limit, the operator will try to reduce the temperature by shutting down several wells, to reduce the overall flow rate of the produced water being treated in Seruni GS. This action will result in production loss and lower the overall oil production to be delivered from Seruni GS. Considering the importance of outlet temperature of produced water concerning the potential production loss and compliance to the permit, it is critical to building a model to understand the relationship between outlet temperature and the operational condition (such as produced water flow rate, ambient temperature, and others). A proposed solution is to combine the implementation of machine learning prediction and genetic optimization to predict minimum adjustment to the operational condition and the oil loss while still meeting the required temperature limit. Using the method, the developed model has achieved a Root Mean Square Error (RMSE) of about 0.38degF and a Coefficient of Determination (R2) of 0.97. In addition, the optimization result shows better decisions compared to current best practices in suggesting the well shut-in candidates. The system is run daily to estimate future outlet temperature and recommend the amount of water reduction, including the list of wells to be shut-in. Another utilization of this system is for assisting the evaluation of the potential impact of increased produced water flow rate from sizing up the subsurface pump and reactivating existing idle producer wells. As the company operates many other fields similar to Seruni field, there is an opportunity to replicate the approach implemented in Seruni field in other fields operated by the company.