Abraham Lincoln once said, "History is not history unless it is the truth." How is your history - truthful or skewed? We're not talking about your family history, your country's history, or world history, but rather your process history! Is your historian telling you the truth? Would you know if it wasn't? There are several pitfalls that can plague a data historian - pitfalls that typically come to light only after an event occurrence in which we are left scratching our heads and asking, "what in the world happened to our process?"
Data historians are powerful tools that serve us in multiple ways. They record how our plants and processes run for the purpose of process analysis, process improvement, historical record keeping, regulatory compliance, and event reconstruction. But our analysis and records are no better than the quality of our data. Understanding where our historians can let us down is the first step in proactively addressing these potential pitfalls.
The first area to examine is simple analog trending. There is nothing more frustrating for the process engineer than going back in time to look at a process variable and finding a flat line trend - flat lined because the deadband or deviation exception settings were too large or because the analog input filtering was too excessive. Proper tuning of historical data collection involves balancing the needed resolution of the data with the available data collection bandwidth, online storage requirements, and available disk space. These criteria can be satisfied simultaneously, but it takes some upfront analysis and engineering to set the historical data collection parameters correctly - and we want them set correctly the first time - before the data is needed.
A second and more challenging area is a Sequence of Events (SOE) analysis. What was the triggering event for a plant evolution? What happened in what order? More than one process engineer has been confounded by their historical data until they realized their alarms were time-stamped in the process controller, but their analog and discrete data were time-stamped in the historian - making an exact correlation between the two impossible. For critical process evolutions that may require SOE analysis, different coding techniques need to be employed within the control configuration itself in order to accurately reconstruct a sequence of events.
Your historian is your tool for knowing your process history, but it must be configured properly in order to tell you "the truth, the whole truth, and nothing but the truth." If you need help analyzing your historian to ensure accuracy and functionality, we're happy to provide you with support in this area. We'll make sure you not only know your history, but you can be confident it's historically accurate as well.