Optimization modeling and variation reduction in a pharmaceutical production process by quality engineering and robust design

A key to be competitive in today's economy is to produce high-quality products at low production cost, to meet or exceed customer's requirements. Product and process variations cost manufacturing industry significant money in terms of high rework cost, scrap, and costly inspections. Reducing product and process variation in a production process is a vital issue in quality improvement programs, because variation grows into hundreds-of-thousands of dollars in added product cost per year. The objectives of this research are (1) to study the variability of a generic pharmaceutical filling process; (2) to generate the process capability and conduct process validation based on statistical process control (SPC); (3) to develop economic optimization models for the filling process; and then (4) to optimize the process mean (e.g., reduce the deviation of the average fill from the target value), as well as minimize the variability around the process mean in a generic liquid pharmaceutical filling operation. This research, motivated by the fact that many production processes are being run at sub-optimal settings, utilizes the combination of control charts and SPC to study the current variability, capability, and validation of our filling process. Then, we developed a model that accounts for both the controllable and uncontrollable factors, and the response variable. Based on the model, we used mixed-level factorial design and robust design methods to effectively determine the optimal level settings of controllable factors that minimize the variability in the fill weights, while keeping the mean fill weight on target. As a result, the response variable (the fill weight) was insensitive or robust to the variations in uncontrollable noise factors. We derived optimum specification limits for the filling process. This research provides consistent methods for process optimization and variation reduction that has been implemented to improve the performance of our filling operations. Consequently, we have decreased the amount of scrap, rework and the cost incurred by the firm. The approach and the models, based on Taguchi's robust design, can be applied to other similar production processes.