Improving operational efficiency through condition-based monitoring and Internet of Things (IoT) technologies
Mohamed Ibrahim A * , Michael Clark A and Will Castelnau A *A
Mohamed Ibrahim Hafez Mohamed, PhD, holds a Doctorate in Petroleum Engineering and is skilled in data science and machine learning (ML) with extensive experience in delivering innovative solutions to optimise and solve operational challenges to improve profitability and corporate reputation. His specialty is resolving complex problems through resolutions that integrate new technologies, innovation and strategic actions that are aligned with organisations’ goals and priorities. Mohamed thrives to grow relationships through feedback rich culture and formalise cross BUs engagements, and has the ability to present to executive levels and field operations. Currently Mohamed is the Data Science and Advanced Analytics Manager at Woodside Energy Ltd. |
Michael Clark holds an MSc in Petroleum Geoscience from Victoria University (Wellington) and a BSc (Hons) in Physics from the University of Canterbury. He is a ML specialist with extensive experience applying AI techniques to solve complex problems across the energy, mining, and healthcare sectors. His technical focus spans deep learning, time series forecasting, reinforcement learning, and applied computer vision. He contributes actively to the open-source community, develops educational resources for ML practitioners in the resource sector, and is a long-standing member of the Perth machine learning scene. Michael’s research interests center on safe AI, particularly alignment and reinforcement learning systems in real-world environments. |
Will Castelnau holds a Master’s in Applied Physics with a major in instrumentation. With extensive experience across defence, security, and oil and gas sectors, his career is marked by his adeptness at growing and maturing novel technologies through to adoption. Mr Castelnau has over 25 years’ experience in delivering complex software engineering solutions that facilitate data analysis, particularly in machine learning and computer vision systems. At Woodside, he has utilised his multidisciplinary background in electronics, mechanical, and software engineering to address and solve complex business problems through the application of advanced sensing technologies. |
Abstract
Woodside Burrup Pty Ltd (‘Woodside’) has operated the world-class Pluto liquefied natural gas (LNG) plants safely and reliably since 2012. This complex facility employs numerous heat exchangers to cool and liquefy natural gas. The heat transfer process is facilitated by large fin fans forcing ambient air through heat exchangers containing rows of hot finned tubes. Fin fans are crucial for the efficacy of this thermodynamic cycle. Hence, minimising their downtime and maintaining them is vital for achieving high production. Woodside developed low-resolution and low-cost wireless Internet of Things (IoT) vibration sensors coupled with advanced analytics techniques, to reduce maintenance costs and increase production efficiency. Data from the sensors, mounted on the rolling elements of fans and motors, detect performance changes and diagnose potential failures, reducing the need for human inspection. IoT technology has been instrumental in facilitating frequent and continuous data collection, enabling predictive maintenance. Shifting from reactive to proactive maintenance yields several benefits, primarily reducing operational costs and enhancing safety standards. This research highlights the successful implementation of minimum covariance determinant model and median absolute deviation for threshold setting to identify abnormal vibrations, with post-processing steps to manage alarm frequency. We demonstrate that the combination of low-cost sensing and high-density data coupled with our novel approach produces high accuracy by correlating real faults. This innovative exploitation of IoT presents profound possibilities to elevate operational efficiency through the strategic use of timeseries data, optimal anomaly detection, reducing potential failures and optimising maintenance. The insights are visualised through an interactive analytics platform, facilitating informed decision-making and enhancing operational effectiveness.
Keywords: condition based monitoring, data analytics, energy efficiency, IoT, low cost sensors, LNG, personal safety, predictive failure.
Mohamed Ibrahim Hafez Mohamed, PhD, holds a Doctorate in Petroleum Engineering and is skilled in data science and machine learning (ML) with extensive experience in delivering innovative solutions to optimise and solve operational challenges to improve profitability and corporate reputation. His specialty is resolving complex problems through resolutions that integrate new technologies, innovation and strategic actions that are aligned with organisations’ goals and priorities. Mohamed thrives to grow relationships through feedback rich culture and formalise cross BUs engagements, and has the ability to present to executive levels and field operations. Currently Mohamed is the Data Science and Advanced Analytics Manager at Woodside Energy Ltd. |
Michael Clark holds an MSc in Petroleum Geoscience from Victoria University (Wellington) and a BSc (Hons) in Physics from the University of Canterbury. He is a ML specialist with extensive experience applying AI techniques to solve complex problems across the energy, mining, and healthcare sectors. His technical focus spans deep learning, time series forecasting, reinforcement learning, and applied computer vision. He contributes actively to the open-source community, develops educational resources for ML practitioners in the resource sector, and is a long-standing member of the Perth machine learning scene. Michael’s research interests center on safe AI, particularly alignment and reinforcement learning systems in real-world environments. |
Will Castelnau holds a Master’s in Applied Physics with a major in instrumentation. With extensive experience across defence, security, and oil and gas sectors, his career is marked by his adeptness at growing and maturing novel technologies through to adoption. Mr Castelnau has over 25 years’ experience in delivering complex software engineering solutions that facilitate data analysis, particularly in machine learning and computer vision systems. At Woodside, he has utilised his multidisciplinary background in electronics, mechanical, and software engineering to address and solve complex business problems through the application of advanced sensing technologies. |
