Forecasting Pipelines Operating Pressure: A Proactive Approach to Prevent Oil Congealing Using ARIMA, FFNN, and Hybrid Models

Effective management of operational pressure in oil pipeline is crucial for ensuring flow assurance and preventing disruptions such as oil congealing. This study explores the efficacy of Autoregressive Integrated Moving Average (ARIMA), Feedforward Neural Networks (FFNN), and a hybrid model combining both ARIMA and FFNN in forecasting operational pressure in a section of pipeline. Utilizing historical pressure data from an oil pipeline in Indonesia, the study compares these models based on their forecasting accuracy, evaluated using Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). The goal is to identify the most effective predictive model as a data- driven basis for Company in aiding preventive measures in pipeline operations. The study highlights how FFNN model outperformed other models, achieving an RMSE of 15.42 and a MAPE of 14.53. These result demonstrate the FFNN's potential to enhance operational pressure management, providing a reliable, data-driven basis for preventive measures in pipeline operations, thereby improving operational safety and reducing the risk of costly disruptions, indicating its potential utility in operational pressure management systems.