Sand Control Prediction and Advisory Automation: Malaysia's Cost-Effective Digital Transformation that's Contributing to Sustainability

Abstract Malaysia Petroleum Management (MPM) has embarked on a digital transformation journey of cloud-based sand control prediction and advisory automation with the support of PETRONAS Carigali and data science teams to address the oil and gas wells sanding pain points. Leveraging on its local regulatory advantage, consisting of petroleum engineering expertise and mass data from all the operating fields, the digitalisation of offset data into machine learning (ML) has enabled the nation's predictive solution to improve sand management strategy decision making. The intelligent sand management system is developed through reliable ML interpretation principle to tackle the nation's sand control challenges. The system is trained to analyse rock mechanic data using several key parameters to predict unconfined compressive strength (UCS) value as the pre-requisite for sand production prediction. It analyses the subsurface sand management data lake and, with the system calculated UCS value, the suitable downhole sand control (DHSC) method will be advised based on well design inputs. The proposed DHSC design's high level well cost is also generated based on benchmarking. Sand encroachment from producing sand reservoirs is one of the key issues in oil and gas wells. The erosion consequences in production facility due to sand production normally result in well integrity concern, loss of primary containment (LOPC) risk and subsequently the domino effect of undesirable health, safety and environmental (HSE) incident(s) at operators' producing asset. Conventional well completion design addressing sand control requires significant time and resources to gather and perform sanding analyses and risk ranking from scattered subsurface data. Besides that, a costly reservoir sand core sample test and time-consuming studies for its strength normally will be performed to assist in better well design sanding risk profiling. It is common for the executed downhole sand control selection to be sub-optimum, resulting in earlier end of well productivity or costly remedial works. However, the developed system can significantly reduce the sand control selection analyses and decision-making duration. It serves as a solution for the unavailability of core samples for UCS tests, or cost saving for marginal operator on rock mechanic test avoidance. The optimised DHSC algorithm provides better long-term sand control success rate and well life longevity. The system has proven data analytics result of above 0.91 R-Squared (R2) accuracy. A similar approach can be referred by the industry to minimise sand control issue in current cost saving and digital business operating rhythm. This digital transformation approach has also contributed to reducing equivalent greenhouse emission from carbon dioxide (CO2) compared to traditional UCS laboratory test and lengthy DHSC method analysis processes.