A created model may be used as initial data to stochastic simulation based analysis. In addition to logical structure the mean durations of states of roots needs to be defined. The roots are the nodes of the model that do not have any causes. For the roots that model failures the states are failed state and normal state so the defined values are Mean Time To Failure (MTTF) and Mean Time To Repair (MTTR). ELMAS produces several methods for defining these values from various types of source information or from experts’ best judgments.

Simulation of the model creates data that is like usage history of thousands of years. The created data is analyzed and several types of useful and concrete results are obtained. Basic state results give lots of information about reliability and availability of the studied system.

Conditional results make it possible to find out events or combinations of events that cause most often some interesting event e.g. System down. The importance results may be used for example to find out the events or root causes that alone have most potential for improving the reliability of the whole system.

Risk analysis may be done for the model if expenses are related to the events. In this small example case System down is defined to cost 20000 €/break and 4000 €/h for downtime. Repair parts for Valve failure costs 5000 € in average and repairing 500 €/h. Repair parts for Pump failure costs 300 € in average and repairing 100 €/h. With these values the average costs during defined time period (Total risk) caused by malfunctions is obtained.

With the analysis it is easy to compare how different design solutions affect to the reliability, availability or risks of the system. If in this example case tree pumps are placed parallel instead of two, the average costs in 10 years will be about 19000 € lower. These type of results help to make the decision if for example investing in the new pump would be profitable.