Major objectives to reduce the economic and human losses in the chemical process industries are to improve the mechanisms for process safety as well as product quality control.  Quality defects are flagged by “special causes” (arising due to failures of hardware components, errors by operators in performing operating procedures, inefficient management [1], etc.) in the statistical process control mechanisms.  In many respects, these are closely related to the safety problems, which are flagged by “near-misses” (indicators of potential accidents).  Prior research has emphasized the importance of identifying these near-misses, high-probability, low-consequence events, in predicting and preventing the occurrence of catastrophic, low–probability, high-consequence events [2, 3].  This paper quantifies the relationship between special-causes and near-misses and proposes a novel approach to integrate and enhance techniques for safety and quality, thereby helping to improve the predictability of potential accidents.

An approach to identify potential special-causes detectable from product-defect information and determine the relative significance and probabilities of the occurrence of the near misses, including failure probabilities of safety systems, is presented.  This approach involves utilizing dynamically available product-quality and safety data, and exploiting the pattern information from near-misses, often overlooked in product defects.  Abnormal events that, otherwise, may lead to off-specification products and accidents, are often prevented using predictions of the occurrences and severity of potential abnormal events and product-quality defects. A multivariate Bayesian analysis framework is formulated for data analysis and predictive modeling. These predictions provide direction and emphasis to improve safety and product-quality performance.  As a case study, an acrylic fed-batch polymerization reactor is considered to show the application and reliability of this approach.

References

1. Meel, A., W. D. Seider, and U. Oktem, “Analysis of Management Actions, Human Behavior, and Process Reliability in Chemical Plants.  I. Impact of Management Actions,” Proc. Safety Prog., 27, 1, 7-14 (2008).

2. Phimister J. R., U. Oktem, P. R. Kleindorfer, and H. Kunreuther, “Near-miss Incident Management in the Chemical Process Industry,” Risk Analysis, 23, 445-459 (2003).

3. Meel, A., and W. D. Seider, “Plant-Specific Dynamic Failure Assessment using Bayesian Theory,” Chem. Eng. Sci., 61, 7036-7056 (2006).