SEARCH

SEARCH BY CITATION

Abstract

Faced with the lack of historical data or access to specific plant maintenance records, the analyst can estimate failure rates for plant-specific vessels and pipes using the Thomas model. The model uses available information regarding the vessel shape, size, and thickness along with the construction details of the vessel or pipe (number and type of welds, etc.). Maintenance and inspection history of the equipment can also be factored into the model, allowing the model to reflect the analyst's state of knowledge about the impact of these observed praictices on predicted failure rates. The model was orginally developed using a large database for predicting vessel and pipe failures for the nuclear industry but can be equally applicable to the chemical industry. Any additional information about plant-specific failures can be easily factored into the analysis by using Bayesian techniques.

In the present application, the Thomas model was used to predict failure rates of process vessels storing corrosive and hazardous chemicals. Engineering judgement was used in making the model reflect the corrosive properties of the stored chemicals and the effectiveness of the inspection programs. Finally, the results of the model were compared with available industry wide data sources and were found to be reasonable.