What is a KPI anyway?

Recently, there has been an upsurge in the level of interest in management by objective, or in establishing performance measures to track progress towards asset maintenance goals. One of the terms that are regularly misused and poorly defined is that of the Key Performance Measure, or KPI.

It seems that almost any indicator is referred to as a KPI, so what are they really? As the name states, a KPI is an indicator of key performance, not just of performance. In order to try to better explain this point we will look at an example of performance measurement as applied to a fleet of haulage trucks within a mine site. However, the same principles can be applied to fixed equipment within a process plant, manufacturing facility, or even to a distributed asset company such as utilities or rail.

Within a mine site there are three or four different operations, exploration drilling and blasting, mining the ore, processing it or filtering it, and then loading it for sale or transport.

While all of these are important processes the management of the fleet of haulage units or trucks is one of the key lynch pins in terms of operational cost control and meeting production targets. The fact that these items are all independent of each other, consume large quantities of energy, and cost a fortune for spare parts adds to the complexity of managing them, as does the fact that every eight or 12 hours they generally have a different person at the wheel.

So how do we know they are working well? There are a few standard measures that can be applied to any form of fleet equipment, but it really depends on what it is that we want to know. For example; a truck can be working well from a energy consumption standpoint, but not from an operational standpoint; or it could be working well from the point of view of operations but not from the point of view of cost effectiveness.

We could represent these by a range of metrics, each one of them important in their own right, but with no real indication of how each one relates to the other, or of how important they are with relation to other aspects of machine performance. This is at the heart of the problem; when we confuse metrics with KPIs we miss out on a range of additional useful information relating to overall balanced performance of the asset. Let’s look again at the fleet management issues in more detail.

As we started to discuss earlier a haul truck has several areas where direct performance matters, these could be:

• Energy performance
• Cost effectiveness
• Productivity

For the sake of this example we will leave out other aspects such as safety and the environment for now.

If we take productivity there are a range of performance measures that could be applied within this area. These take on the standard sorts of direct performance measures that you regularly see, e.g. availability, reliability or failure rate (MTBF), cycle time for a specific run, tons during a specific period, average load rates etcetera. All of these are useful indicators of the performance of this asset, but not all of them are directly related to this asset alone.

For example, tons during a specific period relates to how the truck is loaded as well as the speed with which it works, load rates are similar in that they rely on other assets to make up the measure. If we were really going to construct a scorecard these figures would be a must-use, but in this case we will work without them.

One measure that most people readily understand is that of availability, or the amount of time that an asset is available for duty when it is required to be available. The definition is important because it is no good if an asset is available for work when nobody needs it, particularly if it is not available when they do need it.

By itself it tells us a fair amount about the asset; it tells us that it was ready to work when we needed it to work. But the asset could still be performing poorly, even though the availability is high.

Let’s take the example of a pump that is required for ten hours and is available for 9 hours. This would give us 90% availability; good for some industries but not so great for others.

But what if the pump had actually broken down 20 times over the ten hour period, with each breakdown taking two minutes to correct? If this were the case, then the mean time between failure, or the MTBF would be:

10 hours required
20 failures
= 30 minutes

So on average the pump would be running for about 30 minutes prior to breaking down and needing attention again.

So by itself the availability figure is not telling us the entire story, we need to look at other figures such as the MTBF in this case. Are we done? What other issues could there be regarding availability of the asset? What about quality of workmanship? Not so important if everything is going well, but pretty vital if things are not going well. So maybe there could be a good case for including a measure such as Mean Time To Repair, (MTTR) a proxy for quality or for speed of work. We could also include a measure of rework if anybody could agree on what the definition of rework is.

So now we have three metrics that make up our view of the trucks productivity. Although we ruled out load rates and other indicators earlier there is probably another indicator that we could also include; that of Unit Costs.

Unit Cost is a great measure in any industry as long as it is defined as the costs of maintenance, operations or both; against the unit of production. Unit Costs are sometimes defined differently in utilities and infrastructure organizations and this can be misleading.

In this case the unit costs shows us how much our availability is costing us, availability by itself is great and allows us to have an asset that is contributing to the production targets, but how much is this costing? For example; a truck that is working but cannot take heavy loads due to developing issues with the wheel suspension system, is lowering productivity rates and raising unit costs of production.

So now we have four, for ease of understanding these four indicators could be ar-ranged like the graphic below.
 
Here we see that Availability remains the key indicator we use to tell us how we are going in terms of productivity, but it is not the only indicator of productivity. Other indicators are also included in various other perspectives, E.g. unit cost in the cost ef-fectiveness perspective and MTBF in the quality of performance perspective.

So we can star to see that a KPI is not just a performance indicator, it is the indicator that tells us the key information we need to know in relation so a specific strategic theme or area, in this case productivity.

However, unit cost may end up being the KPI for cost effectiveness.

Sound like semantics? It has been my experience that the correct use and definition of KPI’s, as opposed to indicators or general metrics, allows people to go directly to the agreed “most important” indicator of a specific area of performance. And if this is also tied in with a scorecard type approach, as it is in the graphic above, then they will get a picture of how the asset is performing as well as a good indication as to why.

There has also been a recent trend away from several indicators and towards one or two significant indicators. If you correctly use a KPI approach, then you can use all of the indicators that your data will support. Every time you look at the scorecard your eye is taken to the KPI, and then it is an easy task to look only at those indicators that are leading you towards why the asset is performing poorly.

This sort of graphical dashboard is useful tool for managing large scale asset perform-ance in any situation, particularly given the ease of use of today’s technologies.