PRiSM Predictive Asset Analytics

Uses historical data to learn how a piece of an equipment should behave and generates a unique operational profile for that equipment to predict the operational conditions for that equipment during multiple operational and even transient conditions and make predictions every 1-5 minutes, depending on the equipment type.
Predictive Analytics
The prediction is based off all the variables that are correlated in that system, so a pump model will incorporate the current load, temperature, pressure, flow, vibrations, etc.  And, when the actual operating condition is found deviating from what PRiSM Predictive Asset Analytics predicts, an alert is sent to the appropriate analyst to diagnose the issue.  The early warning alert that an analyst receives from PRiSM Predictive Asset Analytics is before an alarm condition is triggered in DCS, SCADA or CBM system.
Picture of PRiSM Predictive Asset Analytics
Picture of PRiSM Predictive Asset Analytics
Use Predictive Models to Convert Data to Actionable Insights
Monitoring without advanced analytics lacks capabilities to convert raw data into actionable insights. How do you know which one of these trends is causing the problem? What if there we multiple trends that were causing a problem? It would be extremely difficult and most importantly time-consuming to search out problems with this trend alone.
Anomaly Prediction
Automated alarms are triggered earlier than typical alarms to notify users. PRiSM Predictive Asset Analytics automatically scans through all the trended data and provides an alert when one or many trends are deviating from expected behavior. without PRiSM Predictive Asset Analytics it’s very hard to tell if the bearing temperature is normal or abnormal. The actual value is shown first and should (for the most part) be difficult for anyone to see an anomaly.  After the PRiSM Predictive Asset Analytics prediction is overlaid we can see a deviation starting to form about mid-December.  A key aspect is that actual value reaches a max of ~160 F, a full 10 degrees lower than back in September when higher temperatures were expected.
Picture of PRiSM Predictive Asset Analytics
Differentiator: Sensor Relationships and Operational History
How is PRiSM Predictive Analytics different from other algorithms? Normal algorithms will have alarm band in red to accommodate variants/protection from false alarms. PRiSM Predictive Asset Analytics models determine what the band of operation is for normal under multiple conditions, so can have smaller alarm bands. This translates to providing more time to react to prevent equipment damage, loss of efficiency, loss in performance, etc.
Prevent False Alarms
  • A model will/will not run according to the filter definition for the model or for the signal
  • Include or exclude signals
  • Can be based on calculated points
  • Typically this allows for units not running and for bad instrumentation from causing false alarms
  • Deploy model
Picture of PRiSM Predictive Asset Analytics
Picture of PRiSM Predictive Asset Analytics
Fault Pattern Recognition
Another critical differentiator for PRiSM Predictive Asset Analytics is the ability to not only identify when the behavior has changed, but also highlight what the possible change might be indicating. So, when PRiSM Predictive Asset Analytics sends an alert it doesn’t just say that there is an alert on a temperature trend, rather it tells you which piece of equipment is affected, what the sensor or sensors are that have triggered the alert, and what the likely fault condition is. The top graph shows how often the fault condition occurred over the sample period and what the peak match was during that period. The bottom graph indicates which sensors were contributing to the alert. Imagine how much faster a plant engineer or an operations manager could diagnose a problem with this information.
Transient Analysis
Turbine startup and shutdown monitoring with predictive analytics.
Picture of PRiSM Predictive Asset Analytics
Picture of PRiSM Predictive Asset Analytics
Asset Health Check
Condition of all the assets in real-time.
Optimised Scenario
Ready to be deployed Predictive Analytics software with little to no software development required. Modelling technology that automatically builds a signature for each specific asset. Apt for large installations, supports transient module
Available on