Water Production Surveillance Workflow using Neural Network and Bayesian Network Technology: A Case Study of Bongkot North Field, Thailand

Abstract

Abstract The PTTEP-operated Greater Bongkot North Area of the Bongkot gas field is selected for the pilot with around 200 active production wells and four disposal wells. High water production from the gas wells impacts well performance negatively and water-handling capacity constraints limit the volume of produced water. This warrants in many cases the shutting in of erstwhile productive wells. Potential water producing wells are identified in a time-consuming process by monitoring every well manually based on well head temperature (WHT) and pressure (WHP) trends. This leads to a delayed identification of water production wells. A new approach with a probabilistic predictive tool has been introduced for screening potential water production wells. A data mining approach with Self Organizing Maps and Bayesian Belief Networks is used in this case study. Well head data trends and well test data are used to train Neural Network and Bayesian Belief Network models to identify potential water production wells based on pre-defined trends. The end result is a probability of potential water production based on expert user knowledge and historical data trends. Based on the potential water production probability, well tests are run to confirm water production. Using this tool, the user can justify running production logging tools (PLTs) to identify water-contributing zones and decide on water shut off measures that can be taken. Self Organizing Maps and Bayesian Network models have been built for Greater Bongkot North. The predictability of the models was tested with historic data. The final models gave a reasonable success rate for detecting probable water production in wells. However, it was noted that knowledge of field behavior and extending the input data beyond the original well head data is required to further improve the prediction results. It is planned to test the workflow later this year for selected Greater Bongkot North platforms to verify the applicability of the methodology in a live environment. Introduction Bongkot field is located 150 km off the eastern coast of Thailand, approximately 180 km North East of Songkhla Province, in a water depth of around 80 meters. The field is intensively faulted and gas is located in several sand bodies in depth between 800 and 2800 meter. The field started production in 1993 and was originally operated by TOTAL. Since 1998 PTT Exploration & Production Plc. (PTTEP) is the operator of Bongkot gas and condensate field in the gulf of Thailand. The field has been subdivided into two main areas, Greater Bongkot North (GBN) and Greater Bongkot South (GBS). Current Bongkot hydrocarbon production comes from Greater Bongkot North Area (GBN).