Identifying Highly Variable and Energy Intensive Batch Manufacturing Processes Using Statistical Methodologies

Batch manufacturing processes are extremely energy intensive. These processes can benefit from statistical analytical methods to identify production phases that are highly variable or contain outliers. Outliers, in this sense, represent batch processes that take significantly longer time to complete, which in turn results in the consumption of much greater energy than other observed identical batch processes. This work presents a case study on the analysis of manually recorded written batch manufacturing records from a pharmaceutical facility. As batch records are currently recorded in writing, it poses a barrier to rapid identification of process variability and outliers. The benefit in identifying processing steps that are highly variable is the introduction of standard operating procedures that reduce variability. Outlier identification allows for batch processes to be further investigated so that root causes are identified and acted upon to prevent future occurrences. The highly variable process steps and outliers are identified using boxplots. These are further analysed to identify causes, which include transcription error in data recording, parallel processing between manufacturing locations sharing utilities and heuristic based decisions made by plant process technicians. By identifying and acting upon these causes, the facility can achieve greater energy efficiencies and have a more sustainable approach to batch manufacturing.