Each level in the organisation has one or more KPIs – each KPI:
When formulating a KPI, it is necessary to consider the objectives, the actions required to achieve the KPI and the team who will act on the information.
Although this is a very logical process, many organisations fail to translate objectives and goals into KPIs that are useful at the tactical level. In most cases, the KPI is merely the outcome or result of how the plant operated during a defined interval, masking intermittent periods of good and bad performance.
For example, the goal can be to produce concentrate at $x per ton. Although the KPIs provide feedback monthly or quarterly, it remains a retrospective measurement. In most cases, the KPI fails to provide insight into the sources affecting production cost because of how costs are aggregated and averaged within processing plants.
Because the KPIs lag the processing activities and fail to identify the sources of variation implies limited use to improve performance. It calls for the decomposition of the KPI into a metric that explains the real-time operating state of processes and equipment within processing plants.
The metric aims to generate actionable information to facilitate pro-active management of process and equipment, establishing circumstances that will lead to achieving the KPI.
A KPI combines various elements in a processing plant. As mentioned, the KPI value results from how the team operated the plant. Let’s use the production cost example. To realise a reduction in operating cost, what actions must the operating team take? Most operations rely on the production team’s experience to translate the goal into appropriate actions – with variable success. When depending on the team’s expertise to implement improvement, opinion and bias will inevitably play a role in the formulation of initiatives.
Successful performance improvement depends on various technical factors and the ability to implement change within the team affected by the improvement initiative. Employee engagement, motivation, team learning, and alignment toward a common goal are vitally important.
To decompose a KPI, therefore, requires a structured process that involves the client team. When planning the decomposition of a KPI, the focus is on understanding the decisions and actions that the team needs to implement at a tactical level. A ‘decision’ typically considers a logical grouping of variables to formulate an interpretation and matching activities. The grouping of input measurements and matching ‘decisions’ guide the breakdown of a KPI into modules.
To enable proactive management of operational risks requires early warning of deteriorating performance. The early warning provides an opportunity to mitigate or eliminate losses.
The Time-in-State® machine learning platform makes it possible to generate models for the modules associated with a KPI. A model is a digital twin defining the optimum operating envelope of each module. The model interprets the input variables and guides decision-making should the module deviate from its optimum envelope.
Data interpretation take place in real-time, and this facilitates proactive management of risk. Taking this approach delivers the following benefits:
Most Key Performance Indicators used within processing plants fail to provide actionable information. By decomposing the KPI using Time-in-State® models, it is possible to manage KPIs proactively and eliminate losses worth millions of dollar – this is a logical approach to performance management.
| For more information contact: |
Dr. Kobus van der Merwe – Kobus@tisMetrics.com