Your CMMS, no matter its age or sophistication level, most likely is capable of slicing and dicing data in thousands of different ways. So where do you begin? How do you prioritize? Let’s look at one possible starting point: Pareto analysis. This is a simple and powerful tool that will quickly focus your asset management efforts, resulting in improved efficiency and effectiveness.
What is Pareto Analysis?
This technique is based on the Pareto Principle, sometimes referred to as the “80/20” rule. Pareto analysis is used to determine the input factors that are statistically most likely to influence an outcome. For example, about 80% of presenting problems with a given asset can be traced to roughly 20% of possible causes. Similarly, typically 20% of your spare-part stock-keeping units (SKUs) account for about 80% of your total inventory dollar value.
Of course, the actual percentages should be calculated using the CMMS rather than just using the 80/20 generalization, as there may be many cases where the 80/20 assumption is under- or overstated. A CMMS can assist in focusing on the true bottleneck and ultimately answer key questions as to what is driving poor quality, delays, or non-value-added cost. Is it asset availability, reliability, or performance? If so, which asset or component? Is it a parts problem? Does it have to do with operator or maintainer training?
How good is your data?
For any reporting or analysis tool to be useful, you need to start with good data. Data should be accurate, complete, timely, and collected in a consistent manner, in order to ensure statistical significance and support better decision making. To increase the likelihood that front-line employees will collect good data, there are two key factors to consider:
- Make sure the process for collecting data is quick, easy and accurate
- Maximize buy-in by making it clear what’s in it for those collecting the data (i.e., the reward for generating good data, and the consequences for not)
The two factors build on each other in that maintainers will have greater buy-in when they see that their efforts to collect good data lead to a safer working environment, less fire-fighting, better training, and other changes perceived as positive to them.
Asset hierarchy and failure tree
Although Pareto analysis may be a simple tool, the preparation required to get the most out of it focuses on establishing a comprehensive asset hierarchy and corresponding failure tree, as described below.
Asset hierarchy. Many companies struggle with establishing the asset hierarchy because they are not quite sure how granular to get in terms of parent/child relationships. Is a pump a part? Is the motor? Do the two parts form an asset? A key difference between an asset and a part is that you track work history against an asset, but not a part. Replacing a given part can be an action code on the work order charged to the parent asset and not the part itself.
“Rotating assets” such as motors are both parts and assets in that they can move among locations or sit in stores as a part would. A modern CMMS can track the work history and move history against any serialized rotating asset. A location hierarchy also should be established defining the position in which the asset or component resides. Examples of this would be “front passenger wheel well,” “Boston plant SW wing on Line 3 pack-off,” or a GPS coordinate. This is especially important when rotating assets are involved.
Failure tree. Once you have established asset and location hierarchies, failure trees can be created (i.e., the hierarchy of problem, cause, and action [PCA] codes). Here again, companies struggle with determining the appropriate level of granularity. Two key factors to consider here are:
• The likelihood of collecting useful data at a given level of granularity (e.g., there’s no sense in tracking PCA codes on a single light fixture)
• The number of likely PCA codes required at each level (e.g., five problem codes for “tires” at the component level makes more sense than burying those codes within a group of 75 codes at the “vehicle chassis system” or asset group level)
Analyzing the data
Today’s CMMS packages provide Pareto and other analysis tools with varying degrees of sophistication. Below are two examples of analyses you can perform.
1. Top 10 PCA codes. This is the simplest analysis but one of the most powerful. It can be generated after each shift, with results available for comparison by day, week, month, year to date, for multiple years, and relative to last year. For example, what were the 10 most expensive problem codes dealt with in the past year? The Pareto can be based on frequency of occurrence or total dollars derived from work orders. Another powerful option is to tie the Pareto analysis to the impact of each problem, if such information is recorded through or linked to your work orders.
2. Troubleshooting. When a maintainer investigates a problem code for a given asset or asset group as recorded on the work order, most CMMS packages provide a list of possible causes from which to choose. However, more-sophisticated CMMS packages also provide a Pareto showing the historical frequency and/or cost of each of the cause codes associated with that problem code.
For example, suppose the problem code is “flickering light”, the CMMS may report that the most likely cause code to investigate first is “faulty bulb” at say, 45% historical frequency (see Exhibit 1). Similarly, the maintainer can use the same Pareto to determine the most likely action required for a given cause code. Following the example above, suppose upon investigation the actual cause of the flickering light is a loose bulb connection that historically has been the most costly cause code according to the Pareto analysis. If the maintainer is tempted to fix the problem by repairing the socket for whatever reason, the Pareto will show that historically this action has been the most expensive, and that “replace bulb” has been a cheaper and more frequent action taken.
Note that further analysis is required to see if various actions have been more effective over time in reducing the frequency, cost, and impact of various problem/cause codes.
Exhibit 1 -- Pareto Analysis on Asset Failure Tree