Enhanced Predictive Maintenance for Offshore Equipment with AI/ML

January 21, 2025

The offshore oil and gas industry relies heavily on continuous operation of rotating equipment, including turbines, compressors, and pumps, as these machines are vital for maintaining production efficiency and minimizing costs. Traditional maintenance approaches, such as reactive and time-based methods, have often fallen short in addressing unique challenges posed by offshore environments. This inadequacy results in increased downtime and higher maintenance costs due to dependence on fixed schedules or post-breakdown responses. As a result, there is a growing need for more effective maintenance strategies to ensure the reliable function of critical equipment in the offshore sector.

Predictive maintenance, enhanced by artificial intelligence (AI) and machine learning (ML), offers a proactive strategy to address these challenges effectively. By leveraging real-time and historical data combined with ML algorithms, predictive maintenance systems can forecast equipment failures before they occur, allowing for optimized maintenance scheduling. This innovative method enhances equipment reliability, maximizes uptime, and reduces operational interruptions, providing significant benefits over traditional approaches. Murphy Oil Corporation’s project in the Gulf of Mexico showcases the tangible advantages of predictive maintenance in the offshore industry, offering valuable insights into how AI and ML can revolutionize equipment upkeep.

The Shift to Predictive Maintenance

Predictive maintenance signifies a fundamental shift from traditional maintenance strategies. Instead of waiting for equipment to fail or following a fixed maintenance schedule, predictive maintenance uses data-driven insights to foresee issues before they arise. This proactive approach is particularly advantageous in the offshore oil and gas industry, where equipment failure can lead to costly downtime and safety risks. By predicting and preventing potential failures, companies can enhance operational efficiency, maintain safety standards, and reduce unplanned shutdowns.

Murphy Oil Corporation’s initiative in the Gulf of Mexico serves as a prime example of the benefits of predictive maintenance. Over a 24-month period, Murphy implemented an AI/ML-based predictive maintenance system on its deepwater platforms, concentrating on production-critical rotating equipment. By integrating data from various sources and applying predictive models, the project aimed to enhance operational reliability. This integration not only empowered maintenance teams with valuable insights but also optimized the maintenance processes, ensuring that equipment remained functional and efficient. The project’s success underscores the potential of predictive maintenance in transforming maintenance strategies within the offshore industry.

Methodology of Implementation

Murphy Oil Corporation embarked on the project by partnering with a service provider to implement AI/ML-based predictive analytics on its rotating equipment. The execution was designed in several planned stages, each contributing to the overall goal of enhancing operational reliability and efficiency. The data flow began with sensors capturing process data, which was then stored in an offshore historian for secure storage and easy access. This initial step ensured that the data was readily available for analysis, laying the groundwork for the subsequent stages of the project.

In the first phase, data from export natural-gas compressors and main power-generation turbines was transferred via an Open Platform Communications (OPC) server to the service provider’s cloud. Predefined thresholds were used to set up alerts, which were reviewed manually by the service partner. If maintenance actions were required, the company was notified accordingly. For other rotating equipment, data was transferred from the offshore historian to the onshore historian, where it was reviewed by personnel as needed. The main power-generation turbines and export natural-gas compressors’ data was initially manually scraped from the service partner’s web interface, leading to the establishment of a dedicated data pipeline for periodic data transfer.

Developing and Deploying Predictive Models

To develop robust predictive models, time-series data from sensors was augmented with event-based data, including equipment history from the Computerized Maintenance Management System (CMMS) and daily progress reports. Initially, CMMS data transfer was performed manually using spreadsheets; later, a Representational State Transfer Application Programming Interface (REST API) was established to automate this process. This automation facilitated seamless data integration, allowing predictive models to be more accurate and responsive to real-time conditions.

Once the models were developed and deployed in the cloud, they began generating anomaly alerts. These alerts were classified as either true or false positives. True positives resulted in manual maintenance notifications in the CMMS, ensuring that necessary actions were taken promptly. Eventually, a REST API was developed to create CMMS notifications for true positives automatically, integrating seamlessly with Murphy’s offshore maintenance teams. This integration enabled proactive interventions, minimizing downtime, and ensuring that maintenance activities were timely and effective.

Observations and Results

Throughout the project, a total of 46 predictive models were deployed across the platforms, each comprising a unique set of models tailored to specific equipment types. Alerts generated by each model were meticulously recorded and dispositioned, with models for Platform 2 generating alerts earlier than those for Platform 1 due to differing implementation timelines. The number of alerts varied by equipment type and platform, reflecting the diverse operational conditions encountered during the project.

A cross-functional review team, including members from the AI/ML software service partner and Murphy’s facility and process engineering, reliability engineering, maintenance, and operations teams, reviewed and categorized the alerts. As more models were deployed, the incidence of false positives initially increased but decreased significantly after model retraining. This iterative process highlighted the importance of continuous model refinement to improve accuracy and reliability, demonstrating the dynamic nature of AI/ML-based predictive maintenance systems.

Challenges and Lessons Learned

Despite the project’s overall success, it encountered several initial challenges, including a lack of understanding of project prerequisites, which led to delays and misalignments. Future projects should ensure comprehensive assessments and understanding of prerequisites before commencement through detailed planning sessions, stakeholder meetings, and feasibility studies. Identifying clear objectives and expected outcomes for AI/ML-based predictive maintenance is crucial for aligning project goals with operational needs.

Another significant challenge was the availability and quality of data, with the first predictive model deployment delayed due to insufficient requisite data. Ensuring data readiness involves assessing data sources, quality, completeness, and reviewing equipment-related contracts. Murphy’s project revealed gaps in daily progress reports and CMMS event data, underscoring the need for thorough data-readiness checks at the project’s inception. Addressing these data prerequisites is essential for the successful deployment of predictive maintenance systems.

Importance of Domain Knowledge

One of the key takeaways from Murphy’s project was the critical importance of engaging service partners with relevant domain knowledge and experience in oil and gas equipment and processes. Proper vetting processes and competency evaluations for service partners are essential for ensuring that they bring specialized insights and best practices intrinsic to the industry. This domain knowledge is invaluable, as it enhances the accuracy of predictive models and ensures that maintenance strategies are well-aligned with operational realities.

Initial predictive models were rudimentary, similar to “check engine” lights, signaling anomalies without providing detailed specifics. Improving these models to detect specific failure modes based on available data is critical. This can be achieved by leveraging standards like ISO 14224 and Offshore Reliability Data, which provide comprehensive frameworks for equipment reliability and maintenance. Evaluating data sufficiency for detecting specific failure modes is crucial for developing models that are both accurate and reliable in diverse operational conditions.

Future Works

To enhance predictive maintenance efforts, several key areas need addressing:

  1. Enhancing CMMS Data Quality: High-quality CMMS (Computerized Maintenance Management System) data is essential. Implementing regular audits and data cleaning processes will ensure reliable inputs for predictive models.

  2. Ensuring Adequate Instrumentation: Proper instrumentation is necessary for precise operational data collection. Investing in modern sensors and data acquisition systems, supported by thorough cost-benefit analyses, should be prioritized. Accurate, real-time data collection is crucial for effective predictive maintenance.

  3. Developing In-House Knowledge: Building specialized expertise in rotating machinery and related technologies within the organization is vital. Training programs and knowledge-sharing initiatives should foster in-house expertise, equipping the maintenance staff to handle advanced predictive maintenance systems.

  4. Updating AI/ML Solutions: Keeping up with the latest advancements in AI and ML ensures state-of-the-art predictive maintenance strategies. Regularly reviewing and updating AI/ML solutions will refine model accuracy and predictive capabilities, enabling the maintenance system to evolve with technological advancements and operational demands.

By focusing on these areas, organizations can leverage improved data quality, effective instrumentation, domain-specific knowledge, and advanced AI/ML solutions to achieve greater success in predictive maintenance. These improvements will enhance operational reliability and efficiency, setting new benchmarks in the offshore industry’s maintenance practices.

In conclusion, implementing AI/ML-based predictive maintenance in offshore environments has shown significant potential for boosting operational reliability and efficiency. Murphy Oil Corporation’s experience emphasized the importance of understanding project prerequisites, ensuring data readiness, leveraging domain knowledge, and continuously refining predictive models and solutions. By addressing these key aspects, future projects can achieve more effective and reliable predictive maintenance, minimizing downtime and optimizing maintenance strategies in the offshore oil and gas industry.

Subscribe to our weekly news digest.

Join now and become a part of our fast-growing community.

Invalid Email Address
Thanks for Subscribing!
We'll be sending you our best soon!
Something went wrong, please try again later