No surprise – advanced diagnostic tools like PlantESP constantly crunch data in order to uncover important trends and provide actionable insights. As the automation market’s leading controller performance monitoring solution PlantESP typically performs complex calculations on each PID control loop numerous times each day. Such data crunching involves significant computer processing power. While all that processing may be the cost of admission to Big Data analytics and process optimization, there are proven ways of performing rigorous calculations efficiently and equipping users with the valuable information they require. Boxcar averaging is one method and it affords PlantESP users unique control over their plant-wide optimization effort.
Boxcar averaging is a highly effective method for intensive data calculation. It is used broadly by PID controller monitoring technologies such as PlantESP. The method assumes that the average of a series of consecutive data points (i.e. a boxcar of data) provides a more accurate measure of performance than any of the individual points. When applied to loop monitoring the method calculates a value for each KPI as a form of moving average. Doing so mutes the effect of outliers such as noise and emphasizes a PID’s core performance data. Given 1) the highly variable nature of industrial process dynamics, 2) the magnitude of data involved with each individual calculation, and 3) the large quantity of KPIs typically employed, this approach to data analysis is ideally suited for control loop performance assessment technologies.
Following are a few notable attributes of the boxcar averaging method and how it’s applied within PlantESP:
Process, Not Product
Generally speaking, controller performance monitoring solutions compliment a production facility’s other event-focused technologies. Whereas a DCS manages the flow of materials through a production process within established limits (i.e. Hi, Lo, HiHi, LoLo) and triggers alarms when a given threshold is eclipsed, controller monitoring solutions train their analysis on how the performance of PID loops evolves over time. In this way the two solutions are highly complementary. In a sense PlantESP looks beyond the flow of materials to the health of the underlying production processes. It identifies essential changes in the manner by which a loop’s PID controller performs, and it detects mechanical issues such as valve Stiction that negatively affect how process control is regulated.
Windows of Assessment
A boxcar is essentially a window of a fixed size that establishes the duration for which historical data is then polled by PlantESP. To be clear this is not the time it takes to calculate the many KPIs within PlantESP. Rather, it is the amount of data contained within each boxcar. While the recommended duration is four (4) hours individual loops within PlantESP can be configured to poll data for any fixed length of time. It’s worth noting that shorter durations allow loop monitoring technologies to more readily see changes in process performance. Shorter durations are also better suited for events that might otherwise be “averaged out” if contained in a broader sampling of data. In contrast, longer durations are helpful in establishing interactions that are shared between loops. Advanced forensic tools within PlantESP such as Power Spectrum rely on the frequency data contained within larger durations and use the data to infer loop interaction. Within PlantESP the duration setting can either be tailored loop-by-loop to accommodate unique behavior or a single duration value can be applied universally across all loops.
Flexibility of Frequency
Another customizable attribute within PlantESP is the frequency with which boxcars of data are calculated. As with duration a frequency can be assigned either universally or on a loop-by-loop basis. For most loops the recommended frequency is to have calculations performed every one (1) hour. That allows controller performance monitoring solutions like PlantESP to capture the contours of change as it both unfolds and affects the performance of individual PIDs. If boxcars are allowed to overlap as recommended, then they can more effectively capture subtle changes in performance much like a moving average.
As a plant-wide PID monitoring tool PlantESP capitalizes on process data stored within a production facility’s historian. PlantESP does not duplicate the data. Instead, it accesses the data temporarily in order to perform calculations. Once KPI values are calculated, only those values are retained by PlantESP.
Just as many facilities have a DCS to manage the flow of material through a production process, so too most facilities have a historian to capture and store essential process data. PlantESP provides a valuable complement. The boxcar averaging method makes meaningful use of that data and equips process manufacturers with insight into the health of their underlying production processes.