Availability simulation modeling of a roasting process system

Abstract

Availability simulation modeling is a tool used by industry to assist maintenance decision-making. However, it can be time-consuming to conduct because of poor data quality and the complexity of real systems. In this paper, an availability simulation modeling approach that is suitable to the needs of industry is developed which balances the need for (a) an accurate reflection of system availability performance, and (b) efficiency in development. The approach was tested using data from a gold processing plant and considered 802 assets. The approach is used to construct two reliability block diagrams (RBD), a high level RBD at equipment level and a lower level at component level. The component level model indicates that the roasting system availability is at 94 %, within 3 % of the actual plant downtime; compared to 83 % for the equipment level model. These values are inclusive of equipment failures, as well as, corrective, and preventive maintenance tasks. The component model predicts that 50 % of system unscheduled downtime is caused by 14 failure modes. Potential production revenue savings of 460 h per year exists if these failure modes are addressed. The results suggest that the component-level approach is recommended although the time to construct the model was 8 weeks compared with 5 weeks for the equipment level model. This paper makes suggestions for how to improve the efficiency of development using production time-delay data. Usually maintenance work orders are used for process plant availability simulation studies. The alternative data set was found to improve modeling efficiency by approximately 80 %.