M4M: Management 4 Measurement
Process industry is under continuous pressure to reduce environmental footprint, including greenhouse gas emissions, chemical emissions, water consumption, and the use of scarce minerals—strongly motivated by national and international incentives. Meeting these goals requires educated operational process management decisions, based on the quantitative evidence provided by process data. Industry4.0 holds great promise for such predictive data-driven decisions with Data Science. Data Science with Deep Learning however has three major incompatibilities with industrial practice: (1) it treats all incoming data—however valuable—without curating, evaluating or improving its information content, (2) it currently predicts process performance, rather than more relevant indicators of economic performance or sustainability and (3) it creates predictions in independent computationally-driven exercises, rather than involving intelligence and knowledge from all stakeholders, which limits their acceptance as supportive management instruments on the workfloor.
Process data is an asset of considerable value, as it is generally accurately measured and comes with substantive process knowledge and intelligence from within the company. This knowledge may greatly enhance the information content and predictive value of the data—as it enables “retrospective Quality by Design” on historical process data. Multivariate Statistical Process Control has shown its considerable value for predictive analytics in the process industry, providing key insight into the critical attributes that drive process performance. We foresee that our latest developments in Path Analysis MSPC will greatly enhance further inclusion of both process knowledge and more accurate predictions focusing on the company Key Performance Indicators—both economic and sustainable. As these developments involve the entire company—from Workfloor to Boardroom—this creates great potential for the operator to contribute their insights into the development of industrial predictive analytics, which will in turn enhance the acceptance of the co-created solution as instrument for daily process operation.
- Increase the Value Proposition of (historical) process data as predictor for company performance.
- Enable Process management towards more economically valuable endpoints with novel predictive methods, historical data and existing process knowledge.
- Integrate and align KPIs from Workfloor to Boardroom to create co-ownership of company performance across all employees.