What Value is Hidden in a Plant’s Process Data

While a typical Distributed Control System (DCS) keeps a watchful eye on production and scheduling there are other important details that a DCS simply overlooks.  To be clear, a DCS consumes process data in order to measure and regulate the flow of material through a plant.  It interprets sensor data and orchestrates adjustments to Final Control Elements in order to maintain Set Points.  In this manner inputs are systematically transformed into outputs.

The role of the DCS is focused somewhat narrowly on managing the flow of material.  More specifically, it does not take into consideration the changing health of a plant’s means of production.  Hidden in the process data are essential details related to the valves and dampers, PID controllers, even the physical processes themselves such as tanks, furnaces, and heat exchangers which ultimately affect production.  Their health directly influences throughput capacity, impacts quality, and drives energy consumption.  It is these and other valuable insights that can be learned from a plant’s regulatory control systems.

Consider a few different insights that are readily available within a plant’s process data – all for the asking:

Dirty Little Secrets

Engineers routinely cite Stiction as the leading mechanical issue that hampers efficient production performance.  Indeed, when a valve is directed to adjust – whether to open or close – and it can’t due to excessive static friction that presents a problem.  The result of excessive Stiction is a process that is unable to accurately track Set Point.  While well-established metrics can provide the necessary intel, neither the identification nor the quantification of Stiction is in scope of a traditional DCS.

Uncovering Bad Behavior

When a fully automated process is shifted into manual mode a common knee-jerk reaction is to blame the operator in charge.  However, experience tells us that most operators take a control loop out of its designated mode (e.g. Automatic, Cascade, etc.) because of erratic performance and concerns over safety.  Such an action should prompt consideration of the process’ control strategy and design – not blame of the operator.  While not included in the metrics of a DCS, a simple KPI that tracks the time loops spend in their “normal” mode provides more than sufficient insight into the probability of this issue, and it can help process engineers to identify processes that require consideration.

Higher Order Questions

If the advertisers are to be believed, then manufacturers should invest in new optimization technologies rather than tune up their existing systems.  As most practitioners know controller tuning requires the same data that results from everyday output changes – a change in Controller Output that drives a response by the Process Variable.  Modeling output changes on a plant-wide basis can eliminate the need for performing bump tests.  What’s more, the findings can be used to isolate those loops that can benefit the most from adjustments to tuning coefficients.  That’s the equivalent of real-time plant-wide optimization.

Cross-Unit Comparisons

Oftentimes the alarms triggered by a DCS represent symptoms and not root-causes.  Consider a modest upstream deviation in performance that filters downstream and impacts a process that needs to be controlled within a tighter tolerance.  The resulting alert to the downstream process can mislead staff from addressing the true source of the problem.  What’s important to know is that the process data associated with the alarm contains valuable clues – frequency-based fingerprints that can be traced upstream and matched with the root-cause.  Unfortunately the standard DCS doesn’t include the forensic tools needed to extract such insights from the available process data.

The list of insights gained from the analysis of a plant’s regulatory control data goes beyond these few examples to include others such as comparisons of similar unit operations, the health of the processes themselves, and more.  While the DCS and other supervisory layer technologies are seen as the apex of control, it’s healthy to remember that they’re built on a foundation of regulatory control systems.