Gather the right data
Be sure to gather all the right information. Traditional process variable, setpoint, and output can tell you a lot. But if you also gather information about the controller mode, target values, tuning constants, operator actions, and alarms, you can get a complete picture of the plants dynamic performance. For more benefits on a broader scale, share data between the performance supervision system and your alarm management software, or condition-monitoring packages.
Giving context to the data can add a lot of value. One cement plant, for example, discovered that performance varied greatly between product grades. Organizing the data by unit operation, product codes, and batch codes can help you to see your process in a whole new way.
Loop structure information can help you to directly compare similar loops. You could, for example: compare between manufacturers for stiction on V-ball valves; look for flow loops that have unusually slow process dynamics; or uncover interactions between primary and secondary loops in cascade schemes. The cost of performing this powerful analysis is that you need to provide the loop structure information upfront.
Training is probably the single most important step in the entire process. Making efficient use of diagnostics requires process understanding, context, and a refresher on process dynamics and control. Dr. Hank Brittain of TOPControl has trained hundreds of people to use performance supervision systems. A user can be overwhelmed by the sophisticated diagnostics that are available, he says. But with just a few hours of training, users are better able to understand the specific diagnostics that apply directly to their specific role in their specific business.
Establishing a point of reference for process performance is often the most challenging step because goals and target variables must match. Every plant is different, and every plant is starting from a different point. Usually, a good place to start is with unit cost, reliability indices (equipment availability, time-to-repair, etc.), and production capacity, as well as more technical measures, such as time in normal mode, and process variance. Carefully documenting the current performance of a plant will quantify the benefits later. Quite often, quick improvement opportunities will be uncovered just by looking over the process baselines.
This final step is critical. Performance supervision will uncover and highlight opportunities to improve. But software cant fix valves! Resulting diagnostics should be reviewed and shared on a routine basis. Follow-up actions should be discussed at daily operations meetings and weekly maintenance planning sessions. By incorporating the analysis into routine business practices, diagnostics can have a direct impact on the bottom line.