Analyzing Failures through your CMMS

Maintenance effectiveness is impossible without good information. Good information is the synthesis of tombstone data, "as-found" data, condition data, and operational and maintenance data (equipment events and minor maintenance). This data can reveal knowledge vital to optimal decisions about scheduled rework and condition based tasks.

There are several key issues:

1. Defining the probability of any failure (the age-reliability relationship) requires enough of the right data.
2. However serious consequences of that failure precludes the collection of data about the failure, and hence a conservative safe-life limit is selected
3. On the other hand where consequences of failure are small, lots of data exists - but of course it is less valuable, and is frequently expensive to collect.
4. The middle range between these extremes can pay huge dividends in terms of maintenance cost savings as this is where the bulk of resources are spent

Most equipment in today's industrial economy is defined as complex - i.e. they will experience failures based on a number of often unrelated component failures. The trick here is to separate the failure cycles of these components and record and analyze them. This concept was recognised by Nowlan and Heap* in what is now seen to be the original (and still the best!) discussion of reliability.

Data of this type is the most fertile territory in which to conduct statistical analyses. One key point to make is that to achieve satisfactorily high confidence levels in statistical terms, the required data sample need not be numerically very large. Many maintenance analytical tools such as Exakt, include procedures to compensate for small data sets.

The equipment's recorded history as collected in the CMMS thus can become the central data source for this analysis - but only if is it accessible, consistent and reliable. One of the essential difficulties in data analysis is that of reporting failures in a consistent manner. We are all familiar with the drop down lists of fault codes in the CMMS's and EAM's. And we all know that the top few on the list attract the most votes! In most cases, the tie-in between the RCM's consistent language of failure description and the CMMS fault code is simply not made - let alone made accessible by a system interface.

A second reason for paying close attention to CMMS data is that the RCM conclusions only remain valid if the operating context, performance requirements, failure modes, and effects remain essentially unchanged. If changes have taken place, then the RCM analysis needs to be revisited.

Implicit in the above diagram is that relative to the entire life of an item, the advanced warning period between the detection of deterioration and the actual failure is very short. That means that gathering data to establish the age reliability relationships using the potential failure rather than the functional failure helps to reduce the problem of lack of data for failures with serious consequences. This would then establish inspection intervals and the times at which to intensify such inspections. Graphing the data will also quantify the effectiveness of the CBM program as measured by the gap between the functional failure and potential failure curves of Nowlan and Heap's diagram reproduced below.

Furthermore, graphs based on statistical analysis of this type will help to identify the dominant failure modes which themselves could be managed by some form of scheduled maintenance or redesign, improving overall reliability. Additionally, knowledge that the age reliability relationship of an item is an exponential survival curve would point out whether current scheduled rework tasks are ineffective and in need of replacement by a more applicable form of maintenance.

* Nowlan and Heap, Reliability-Centred Maintenance 1978

Our partners in Equipment Reliability and Failure Analysis are at www.omdec.com, or contact us at info@datatrak.ca

Back

Contact us