Maintenance & Metrics
In the environment of maintenance, there is a progressive hierarchy and general thinking that needs to be followed in order to achieve the best results. It starts out with frequency histograms, then it moves up to meantime calculations, from there it moves up to condition monitoring, and then ultimately predictive algorithms. Every one of these is usually progressively more expensive, whether in terms of how data is collected from a labor force, or whether there are automated collection systems set up.
“It can’t be that simple, you might argue- but psychologists and economists will tell you it is. Human beings adjust behavior based on the metrics they’re being held against. Anything you measure will impel a person to optimize his score on that metric. What you measure is what you’ll get. Period.” -Dan Ariely, Harvard Business Review.
Frequency Histograms → Meantime Calculations → Condition Monitoring → Predictive Algorithms
Developing frequency histograms is the cheapest and easiest thing you can do with a CMMS. What are the top ten things that are going on in my factory? Which asset has the most interruptions? Which asset has the most unplanned events? Which department struggles with the most downtime? Who has the most emergency breakdowns? These are all forms and types of frequency histograms.
The meantime between failures and the meantime to repair are fairly common metrics that most people are familiar with. The mean times between interruptions are defined as those that are not about unplanned events, but planned events. The number of preventive maintenance (PM) processes performed on equipment will not show up on a meantime failure or a meantime to repair because they are not unplanned events. Some schools of thought say that the more PMs, the better, but there are certainly situations where that is not true. Sometimes the best thing a company can do is stop performing certain PMS. However, there is a fair amount of data analysis required before that is a possibility.
Condition monitoring gets a little bit more expensive because collecting and analyzing data before determining a condition is required, such as tomography, acoustic emission, and vibration analysis. There is certainly some effort and money spent on collecting information if you have the right equipment to do it.
The last step from there is the predictive algorithms, which means taking the condition monitored data and running it through a series of calculations. If organizations skip a level of this hierarchy, they will not achieve the fullest result.
Where do you begin?
Start With The End In Mind – What Does Success Look Like?
- What data or reports would make life easier?
- What information does the boss consistently ask for?
- What is the current PM:CM ratio?
- What type of PM:CM ratio does the business need?
- What are the current maintenance budget busters?
- What is the on-time PM performance or compliance?
Some Basic Startup Metrics
Lagging Indicators – Performance Measurements
- CM to PM ratios (World Class = 1 to 4)
- Mean Time Between Failure (MTBF)
- Mean Time To Repair (MTTR)
- Mean Time Between Interruption (MTBI)
- Top 10 Analysis (Cost, Frequency, Duration)
Leading Indicators – Performance Drivers
- PM on-time Compliance
- Estimates vs. Actuals
- Percent Failures undergoing RCA
Asset Structure at a minimum must include the following:
- Parent Grouping
- Child Specific Assets
- Child Miscellaneous Asset – Only one per parent
The Work Order serves two critical purposes and must contain the following elements for each purpose:
What Work NEEDS To Be Done:
What Work WAS To Be Done:
- Child Asset – Mandatory
- Parent Asset Information
- Work Type – Planned / Unplanned
- Work Status – Open / Closed
- Primary Craft
- Required Craft & Hours
- Material Needed
- Safety Information / Requirements
- Due Date
- Actual Labor Used
- Date job was completed
- Time it took to complete this job
- Comments on significant deviation
- Parts actually consumed
- Problem code
Condition Monitoring (CM) Analysis
Detect a failure early enough to provide time to plan and schedule the needed work without panic, reactivity, and increasing cost.
Basic CMMS Data Collection
Start out by identifying which asset you want to remotely collect data from. Set up a data collection device, and get the data from that device into a local server. The local server will run the algorithm from the condition monitoring max/min levels and then sends a signal to the automated work order system. This triggers an alarm which will then auto-generate a work order for investigation from the CMMS system. That will allow the work order to be tracked for completion. The investigation will then discover whether the asset is in a true condition for repair. From there you will be able to plan and schedule the repair.
- Begin with the end in mind!
- Set up asset hierarchy
- Create supporting fields in work order form
- Monitor the work order process
- Monitor the asset performance
- Prioritize asset improvement areas
- Move up the P-F curve