NGL Operation Strategy Using Predictive Analytics

Abstract Oil & Gas is a data-rich industry which is prime for data-driven and decision making. The significant growth witnessed in the digital transformation field and the new era of the industrial revolution 4 (IR 4.0), is a direct result of the need within the industry to utilize the large amount of data to make better decisions, improve operation strategy, plan better for preventive maintenance (PM), and process improvement. The uncertainty associated with estimating the incoming feed gas rate to NGL plants has resulted in deviation from optimal compressor recycle rate, missed opportunities of meeting planned PM and imposed urgency during operation. Utilizing machine learning algorithms, namely regression and decision tree model, the incoming feed gas can be predicted which result in the machine learning algorithms which results in the identification of the optimum number of running trains and recycle rate required for efficient operation. NGL-Operation Planner (NGL-OP) is the outcome of utilizing ML algorithms which provides the ability of predicting incoming feed gas, identifying optimum number of running trains required as well as estimating the optimal recycle rate. Adopting this approach is a new and strategical way to plan NGL operation. The developed tool also has the ability to advise whether to shut down, maintain existing operation or starting-up a new train. The implementation of the model resulted in a significant improvement in NGL operation. The improvement includes fuel gas consumption reduction of around 449 MMSCF/Year which resulted in a significant cost saving, reduction in emissions around 27 M tons/year, and 10% reduction in operating unnecessary running trains. These savings have been achieved through the utilization of the NGL-OP to minimize uncertainty and improve our planning strategy.