Quantitative FTA using Monte Carlo analyses in a pharmaceutical plant

Abstract

The evaluation of faults in a multipurpose pharmaceutical pilot plant used for production of polymer particles was performed, integrating traditional Fault Tree Analyses (FTA) and Monte Carlo procedures and employing tools of the quality risk management methodology for production of medicines. The plant was divided into four basic processes: (i) receipt and sampling of materials; (ii) treatment of purified water; (iii) reaction; and (iv) lyophilization and purification. For each process, the most critical failure was selected, and the FTA was built. Selection of basic events considered the most important effects on the final quality of the medicine. Then, the FTA was reduced to basic events using Boolean algebra. The quantitative assessment was made by assigning failure rate values for each event. The reliability data of the failure rates were based on the literature that deals with similar processes. The frequencies for each fault were determined through Monte Carlo simulations, considering that fault probability distributions followed the exponential distribution. When failure rate (ʎ) data are available, the quality management can establish a prediction of plant behavior over a period. This scenario is consistent and coherent with practices of pharmaceutical sites, since occurrence of high rates of failure must be corrected immediately in order to preserve the safety of the operation.