Part 7. KPI’s - you get what you measure

This is the 7th in a series of 10 articles that further explore the key concepts in “Simplifying Mine Maintenance” by Gerard Wood.

In my last article about the importance of culture in establishing good maintenance practices, I noted that an important step for achieving this is to recognise good work. There are two general ways this can be done - QA or KPI. QA is a means for identifying good work done as it is done, and providing recognition and feedback to the person(nel) involved. KPI’s are based on long term methods using metrics that are agglomerated and averaged over a period of time. Although KPI’s may indicate improved practices, feedback and recognition is often generalised and may not necessarily reach the tradesmen who do the work. I believe that both are important. Personalised feedback and recognition is motivating and drives ownership, generalised tracking of performance is important to identify overall trends.?

Weld repairs of cracking are generally the least well managed and measured maintenance activities on a mine. The lack of welding knowledge throughout the industry means that there is often an attitude that “it’s just welding”, and that as long as there is no crack visible it means that it is “repaired”. Job done.??

This is further compounded by the general lack of any viable way to actually measure and assess whether a weld repair has been done correctly. Attempting to use the typical compliance based QA is often not relevant or suitable for crack repairs. Furthermore, most repairs are carried out without a welding supervisor or inspector present - it is simply not feasible. (More on this in future articles). This issue is highlighted in Simplifying Mine Maintenance where it says

?"There is no KPI that says a guy did a weld that will last or that he cleaned out all the dirt from a component before he fitted it. You can’t measure those things, but they are the principal reason why equipment remains reliable or not"

Actually, “You can’t measure those things...” is the only thing that I disagree with in the whole book! You CAN measure and control whether “a guy did a weld that will last”. Welding is not a “black art”. There are clear and controllable steps that must be carried out in order to achieve an effective repair.?However, I believe this is more related to QA than KPI’s. Good QA provides a simple way to provide near term feedback and recognition, but is just one input for broader KPI’s for long term performance management.

Peter Drucker said,

“If you can’t measure it, you can’t manage it”

but as Gerard alludes to, perhaps this is more aptly put as

“If you don't measure it, you can’t manage it”

Therefore, in regards to KPI’s I think that Gerard makes a very good point - I have never come across a mine site that truly has any metric in relation to crack repair effectiveness.?I think there are many reasons why this is the case, but fundamentally I suspect that it comes back to the general attitude and maintenance practices in that providing cracking can be repaired during scheduled maintenance and does not cause lost production, it isn’t perceived to be a problem. In some respects this is true, but it does not take into account the effect of poor quality repairs over the life of the asset whereby there is a snowball effect of re-cracking and new cracks appearing and requiring to be repaired. It also doesn’t factor how poor quality repairs often end up creating even bigger cracks than were originally “repaired”. Eventually this requires more time and labour for repairs either during scheduled maintenance or as a “campaign”, or even a shutdown. With multiple poor quality repairs in the same region, the long-term effect is that the structural life becomes a limiting factor.?

Also contributing to this issue is that on many sites, the CMMS drives the wrong KPI's for maintenance. There is a focus on work order completion rather than quality. This leads to work that is “done” rather than “done right”. There is no focus on quality and craftsmanship.?

So what should be measured for KPI’s relating to crack repairs, and how can these be quantified? It really depends on the data sources available - poor or inadequate data may make it seem to be an impossible task. However, the challenge is often in knowing what to measure. Here are some KPI concepts that are relatively straightforward once the right metrics and systems are in place.

1. Adherence to the crack repair management framework. Each role from inspection, reliability, planning and supervision have their responsibilities and there needs to be a clear framework/process that is easily auditable to see if it is being followed.?QA is an essential part of this.
2. Accurately keep track of the total cost of crack repairs. This can be in a global sense based on expenditure for welding contractors, or it can be more nuanced and based on actual repair times/costs for each crack. Unplanned downtime and maintenance over-runs should also be taken into account.?
3. The effectiveness and quality of crack repairs can be assessed by the amount of cracking based on the condition severity. This sounds simple enough, but first of all requires consistent condition severity thresholds. It also requires NDT reporting to be able provide the raw data, and then post-processing to “normalise” the data over a given period or for a specific asset class.?Some NDT service providers can provide this information to some extent, but in general it is a metric that is still too hard to manage using the available data. However, even a review of NDT reports will reveal an indication of general asset health, provided that there is some point of reference for what is good/normal that can be used as a benchmark comparison.?
4. Measure re-cracking. Ideally, we would know how long crack repairs last. MTBF is generally used for parts and equipment management, and a similar concept should be used for crack repairs. To do this based on NDT data would require a significant advancement in reporting and crack tracking capabilities that currently does not exist. However, it is still possible. It is generally readily evident whether there has been a previous repair at the location of cracking. All that is needed is to record this at the time of the repair. The objective is to have fewer cracks that are re-cracking from prior repairs.

In the next article, I explore the need for suitable QA to provide improved maintenance outcomes, provide the basis for KPI's, and help reinforce culture change.