A review on the advancements and challenges of artificial intelligence based models for predictive maintenance of water injection pumps in the oil and gas industry

The oil and gas industry (OGI) plays a crucial role in the global energy sector, providing the primary sources of fuel for various sectors, such as transportation, industry, and residential use. The OGI encompasses a wide array of activities, from the exploration, extraction, and refinement to the distribution of hydrocarbon resources, such as crude oil and natural gas [ 1 ]. Within this industry, the complexity of operations, vast infrastructure, and the presence of high-value assets underline the critical importance of efficient maintenance and reliability management. This is pivotal for achieving optimal production, ensuring safety, and enhancing cost-effectiveness [ 2 ]. Predictive maintenance (PdM) is the methodology employed in this context, incorporating diverse approaches like data analysis, machine learning (ML) algorithms, and sensor technologies to gather and analyze real-time operational and sensor data from crucial equipment [ 3 ]. Through the application of predictive models, it becomes feasible to discern intricate patterns and anomalies in equipment performance. This foresight enables the early identification of potential failures, thus adopting a proactive stance in mitigating the repercussions of unexpected downtime. PdM has proven to be a valuable asset to the oil and gas industry, helping to optimize its performance and cost-effectiveness [ 2 ]. PdM is crucial in the OGI, improving operational efficiency, reducing downtime, and optimizing maintenance strategies. Water Injection Pumps (WIPs) are vital for smooth operations and maximizing production concerning reservoir pressure and oil recovery [ 4 ]. PdM leverages advanced technologies and data analytics to predict failures and enable proactive maintenance based on real-time conditions [ 5 ]. This is especially critical for demanding equipment like Water Injection Pumps (WIPs) in the Oil and Gas Industry (OGI). Implementing AI-based PdM for these pumps offers enhanced strategies through cutting-edge AI techniques, including Machine Learning (ML) and deep learning (DL). These AI approaches can analyze extensive data to identify early warning indicators of potential failures, enhancing operational efficiency and reducing downtime [ 6 ]. Proactive maintenance planning, efficient resource allocation, and optimized spare parts inventory can be achieved based on real-time equipment conditions. Traditional maintenance approaches often result in unnecessary actions and costly disruptions, while reactive maintenance poses safety and environmental risks [ 7 ]. PdM addresses these challenges by leveraging advanced technologies and data analytics to monitor the real-time health and performance of upstream rotating equipment in the OGI [ 6 ]. Through the continuous collection, harnessing, and analysis of real-time data sourced from a diverse array of outlets, encompassing sensors, control systems, and historical maintenance records, PdM models possess an inherent capability to identify anomalies, decipher patterns, and recognize early signals that may signify potential equipment failures [ 3 ]. By means of this extensive data scrutiny, PdM models facilitate the prompt detection of potential issues, thus enabling proactive maintenance interventions. This comprehensive data-driven approach equips organizations to preemptively anticipate and address impending challenges, effectively curtailing downtime, optimizing maintenance strategies, and augmenting operational efficiency [ 7 ]. The integration of PdM models instates a proactive maintenance paradigm that bolsters reliability, curtails costs, and extends the lifecycle of critical assets. The early identification and resolution of potential issues before they escalate into substantial failures empower operators to avert expensive breakdowns, mitigate production losses, and ensure the continuity of operations. Moreover, PdM offers the potential for more efficient resource planning and allocation [ 8 , 9 ]. With the capability to accurately forecast maintenance needs, operators can optimize spare parts inventories, diminish the necessity for unplanned repairs, and streamline their maintenance schedules. PdM for upstream rotating equipment is increasingly vital due to industry complexity and criticality [ 4 ]. Digitalization and IoT provide abundant data for monitoring and optimizing equipment, enabling proactive maintenance and operational efficiency. Advanced analytics, ML, and AI in PdM unveil hidden patterns for more accurate maintenance recommendations [ 10 ]. This research aims to comprehensively review AI-based PdM of WIPs in the OGI. It provides insights into the theoretical foundations and applicability of AI-based models for WIPs. To achieve this objective, the following specific research objectives have been identified: