A Step-Change in Alarm Rationalization and Management
Technology normally advances gradually but, occasionally and unpredictably, a step-change occurs. That happened recently when we recognized the parallel coordinate transformation as the basis for a new multi-variable method for analyzing, interrogating, and presenting data problems in the process industry, many of which hadn’t previously even been recognized as multi-variate.
Operator Alarms in process plants are an example. They are often a factor in serious plant disasters, but attention has been focused on avoiding too many alarms so as not to overload the operator processing and recognizing the significance of each alarm within the thousands of other variable values available.
The Rationalization process was performed by a committee of experts across disciplines, often including ten to fifteen people. They would discuss each alarm individually, determining what it protects, the consequences if action was not taken, the operator action, the priority to be given to the alarm and, finally, the value of the limit where the “trigger” should be set.
These meetings produce vigorous debate, but consensus, especially around the value for an alarm limit, rarely happens. The value is that proposed by the strongest personality or left unchanged. There has been no way to determine the correct value anyway!
The received wisdom (and documentation) was to “put alarm limits at the boundary of where the process normally operates” which sounds simple, but as there has been no means of determining the location of the boundary in a multi-dimensional space, it only encouraged interpretation and argument. Then we showed how easy it is, using the parallel coordinate transformation for an accurate 2-d representation of high-dimensional objects such as Operating Windows and Envelopes. It became easy to set alarm limits at the boundary of an operating envelope corresponding to operating objectives and predict alarm performance in advance.
Operations improve, energy usage and emissions decrease, product quality increases and the cost of the rationalization project is much reduced. And those endless big meetings shrink to a few days for the final alarm review. What’s not to like?
First presented on 18 & 19 October 2023.
Presenter: Alan Mahoney