Modelling the Deferred Impact of Failures When Considering the Availability of Production Systems

Abstract

When considering the availability of a production system, it is likely that some equipment failures will not immediately impact production; that is to say, their impact is ‘deferred’. However, should this type of failure not be remedied, it will, in time affect production. The impact of a failure could be deferred for a number of reasons, e.g. transient cooling effects, chemical process dynamics, or the use of buffer storage capacity within a system. A system's ability to ‘adsorb’ critical equipment failures can have a significant effect on production availability. Often, the deferred impact of a failure is modelled as a time based delay, defining a fixed period of time before the failure has an effect on system performance. This approach is commonly used to model the delayed impact of failures of equipment Items such as heaters, chemical injection pumps, etc. However, deferred impact can also be related to the volume of buffer storage available within a system—in this case, when considering the benefit of such storage, using a purely time based approach can be an oversimplification. This is particularly true when the flow rate through a system varies significantly over time, or during the repair process. GL Noble Denton has developed a method whereby the deferred impact of failures can be incorporated into Monte Carlo Reliability Availability and Maintainability (RAM) models using a volume-based approach, in addition to the previously used time-based approach. This has provided significant benefits when modelling failures that are not immediately critical due to storage (e.g. ponds for water handling) and typical examples will be given. In addition, GL Noble Denton has also considered how the deferred impact of equipment failures can be restored following a failure and this will also be discussed. The paper also includes a case study to illustrate how the methods have been applied.