The team then agreed on what defined acceptable performance. This was based on a combination of industry norms, available standards and past experience. The significant challenge was defining reasonable and achievable limits — targets that all the facilities could strive to achieve. For instance, if the number of oscillating loops at most facilities exceeded 15%, and it takes extraordinary effort to reduce the KPI by 1%, setting 5% as the limit wouldn't make sense.
We fully established the initiative at our Canadian sites — in Joffre, Alta., Corunna, Ont., Moore, Ont. and St. Clair, Ont. — in January 2010.
We use Matrikon's Control Performance Monitor to track control-loop service factor and oscillation. This tool already was deployed at a number of the facilities; the performance data could be extracted from the databases in a consistent manner, while providing a link back to the originating assets. This connection to physical assets was critical — to allow us to focus resources on the particular asset(s) causing poor performance.
For alarm management performance, we use Matrikon's Alarm Manager to provide the backbone for the KPI. Connectivity back to assets significantly impairing the performance comes from a suite of analysis tools. A member of our automation staff, Sean Ward, augmented these to enable visualizing the minute-to-minute performance of each tag's alarm actuation rates (Figure 2).
This additional programming helps practitioners pinpoint attention on those assets that frequently cause the highest audible alarm rates for operators. By applying knowledge on top of the standard product, we can determine which tags most significantly impact performance during a specified time slice, e.g., quarterly.
Connecting metrics back to a defensible economic basis is a widespread challenge. However, you must do this to know that you are acting in the areas that provide the best benefit for your effort. For KPIs for advanced process control (APC) and proportional-integral-derivative (PID) loops, we report the performance and economic data. This links the benefit attributed to the performance metric, as well as the opportunity missed when the asset isn't meeting the acceptable limit — providing clarity on the business impact of performance (Figure 3).
With this value visualized and contrasted to similar assets within NOVA Chemicals, we can decide where that performance issue fits in our priorities. With some additional effort, we can determine what's necessary to close the gap. Maybe we'd need to expend additional resources or refocus energy to get the desired performance.
VISUALIZATION IS VITAL
Calculating the metrics is only a portion of the effort. Metrics without a visualization interface don't have much impact — but a visualization interface with questionable data isn't very useful.
A significant challenge is determining how to visualize the KPIs in a meaningful way to stakeholders. The visualization must make sense to them personally, based on their work scope, and provide the tools to drill into the data with relative ease.
The use of tools such as pivot tables provides flexibility to view data from multiple perspectives with a few clicks of a mouse. We rely on pivot tables to summarize data and enable exposure of different levels of detail. Microsoft Excel, being so pervasive, is an ideal choice for visualizing the report data.
For NOVA Chemicals, this remains a work in progress. As advanced as today's computing technology is, providing the ability to view data in a summary format and then dynamically change the view based on user preference is still difficult — especially because the report is saved in PDF format rather than the native Excel format.