Choosing the Optimum Maintenance Interval with Weibull++
Reliability and maintenance play an important role in the economic health and safety of most process industries. To illustrate how cumulative failure times data can be used to determine the best maintenance strategy for a component, this article uses some tools that are available in ReliaSoft's Weibull++ software to analyze data for a pump DE bearing failure in a cement plant.
In a cement plant, the DE bearing failure of a pump caused 12 shutdowns between 2002 and 2011. The plant's maintenance engineer collected and studied the following data.
|Shutdown No.||Date of Observation (from 2/2/2002)||Days Since Last Shutdown|
First, he performed a life data analysis of the data using Weibull++'s standard folio. Since this is a mechanical component failure, he decided to fit a lognormal distribution to the data, which seemed to be a good choice after examining the regression fit parameter (ρ = 0.99) and visually inspecting the probability-lognormal plot, shown next.
Figure 1 - Probability plot
Since the fitted lognormal distribution has an increasing failure trend, the engineer concluded that the best maintenance strategy would be a scheduled replacement of the bearing (i.e., a time-directed preventive maintenance strategy).
Determining the Maintenance Interval
Due to factors involving raw material waste and quality cost, the cost of unplanned maintenance is estimated to be four times the cost of planned maintenance. To determine the best maintenance interval, the engineer used the "Optimum Replacement" template that is available for Weibull++'s workbook reports. The resulting report is shown next.
Figure 2 - Optimum Replacement Time report
Based on this analysis, the engineer decided that the best interval for replacing DE bearings is 150 days because that interval has the lowest cost per unit of time (as highlighted in the report). He recommended this maintenance strategy to management.