- 3:55 PM
146c

Analysis of Management Actions, Human Behavior, and Process Reliability in Chemical Plants

Anjana Meel1, Warren D. Seider1, and Ulku Oktem2. (1) University of Pennsylvania, 311 A towne building, 220 south 33rd street, Philadelphia, PA 19104, (2) Risk Management and Decision Center, Wharton School,University of Pennsylvania, Philadelphia, PA 19104

Abstract: Management actions and human decisions have a significant impact on process reliability. Thus far, these factors have been ignored in calculating plant risks. The methods for plant-wide, dynamic risk assessment have been developed to predict: (1) the frequencies of occurrence of abnormal events in plant units, (2) the propagation of abnormal events through their safety systems, and (3) their subsequent consequences utilizing accident precursor data, helping to achieve inherently safer operations [1, 2]. In this work, the integration of management factors, human actions, and process systems with previously developed dynamic risk strategies for chemical plants is targeted.

Herein, the management decisions and human behavior patterns under various stressful conditions are examined to determine their impacts on fault detection and diagnosis, and risk assessment for chemical plants. The impacts of abnormal events originating in management, human, and physical systems on failure states are examined. For example, the impacts of poor training, maintenance problems, operator's inabilities, control system failures, and excessive feed quantities on failure states are studied. Also, decision-making techniques, such as expected utility theory and cost-benefit analysis, have been exploited to assist operators in choosing from among available actions under adverse scenarios. For example, when an alarm is triggered in a plant, operators can carry out cost-benefit analysis to select from among the following actions: (i) continue operation by correcting the problem, (ii) shutdown the plant manually, and (iii) wait for the high temperature alarm to be triggered.

Furthermore, the Near-miss Management System (NMMS) developed by the Wharton Risk Management and Decision Center has shown that identifying, reporting, and taking corrective action based on near-misses (abnormal events) has the potential to reduce the frequency of accidents [3]. Herein, a Game Theoretic decision model is developed for a specific plant to seek the advantages and disadvantages of having a NMMS in a plant by accounting for the tradeoffs among management, operators, and engineers. Each player must choose between having and not having a NMMS. The management's payoff is formulated as a function of NMMS cost and insurance incentives. For the engineer, system reliability, safety, and productivity are used to obtain the payoff. The operator's payoff is determined as a function of incentives, extra workload, and attitude.

Reference: 1. Meel A, Seider WD. Dynamic risk assessment of chemical processes: An accident precursor approach. 20th CCPS International Conference 2005. 2. Meel A, Seider WD. Use of Bayesian theory for dynamic failure assessment of an exothermic reactor. Submitted to Chem. Eng. Sci. 2005. 3. Phimister JR, Oktem U, Kleindorfer PR, Kunreuther H. Near-miss incident management in the chemical process industry. Risk Analysis 2003; 23:445-459.