(177c) Equipment Health Monitoring to Improve Reliability and Plant Performance through Application of Data Analytics and Predictive Modeling Techniques | AIChE

(177c) Equipment Health Monitoring to Improve Reliability and Plant Performance through Application of Data Analytics and Predictive Modeling Techniques


Thorat, S. - Presenter, Ingenero Inc.
Equipment health and reliability plays a critical role in overall ethylene plant operation strategy and consistently maintaining high production rates. Enhanced equipment availability and reliability can be achieved through effective data analytics by proper interpretation, analysis and use. Knowledge acquired through data can be further put to use for managing current problems and for determining strategy for future. This paper explains how application of data analytics through modeling techniques and systematic monitoring of equipment health is effectively utilized to identify anomalies and solutions to improve equipment reliability and performance.

An intelligent proactive approach needs to be followed, which entails a continuous assessment of measured and derived parameters against known engineering boundaries to detect and correct problems before failures occur. Apt techniques, tools and equipment models are necessary to track the performance and identify quantum of deviation from normal. Further, predictive and prescriptive analytics helps determine the impact of deviation, need for corrective action and reach challenging solutions.

Specific Techniques for Reliability Improvement

  • Statistical tools and machine learning models
    • Track and address deviations leading to equipment failures and loss of performance
  • Conditional monitoring
    • Enhance the equipment reliability and availability through effective data management
  • Life Estimation/Failure Prediction
    • Failure modes identification, typical life and mean time between failures etc.
  • Degradation Analysis and Loss of function of equipment
    • Based on operation bands, tracking loss of function based on process parameters to flag off abnormal operation that could lead to failure
  • Predictive Analysis for operations and maintenance decision making
    • Estimate the optimal time for scheduling maintenance based on the estimated loss in production while continuing to operate an unhealthy equipment

While effective multi-modeling is key, it has to be combined with the process of data collection, transforming it into information and knowledge using the models, visualization and dashboards to help interpretation and early detection and then an implementation and tracking process to ensure benefit realization. Successful application of the above process has significantly helped in improving the health of key equipment in ethylene units resulting in,

  • Longer mean time between machine overhauls/ replacement à extending T/A intervals
  • Improving efficiency and overall effectiveness à increasing plant production capacity depending on the severity of operating constraints