Process monitoring for fault detection and fault identification
The need
Much has been written on the advantages of real-time fault detection and fault identification. These systems are similar to the alarm systems and interlocks on existing processes - but they take it a step further.
The main extension is to detect faults in the process multivariately, rather than univarietly. (More information that highlights the problem with univariate alarms can be found here.)
Our monitoring systems also include fault identification that is relevant to the problem. With univariate alarms the fault is identified as the tag that generated the alarm. However, when multiple tags start to alarm, then it makes identification and understanding of the problem very difficult.
Applying multivariate process monitoring
ProSensus has been involved in several projects where multivariate monitoring systems have been deployed in the metals, petrochemicals, and materials industries. Multivariate monitoring is right for your application when some of these conditions exist:
- subtle problems go undetected for long periods of time (days, or weeks);
- the current alarm system does not always correctly diagnose the faults;
- the current system provides too many false alarms;
- you have real-time data available, and historical data in a database.
The ProSensus advantage
ProSensus will work with your team to develop a monitoring solution. We use historical data to develop and demonstrate the monitoring system. We combine our modeling expertise and your process insight to develop and fully test the final system. System implementation is performed jointly, with your IT staff, to create a solution that meets your needs.
The ProSensus team has experience with real-time implementation on several data systems. Please contact us to request additional information and we can discuss your intended application(s).
