Best criteria to measure Corrective Action effectiveness - Poor Maintenance

#1
Hi everybody
In measuring the CA effectiveness, I face some doubts, hope you can give some ideas.
E.g. it resulted a NC for poor performance in a process. Such process is monitored monthly.
The root cause was, a poor maintenance of a machine.
The action plans were to redefine maintenance frequency and hire a new guy with more competency.
Now the process runs fine, and maintenance is given correctly.
Next, is to evaluate the CA effectiveness.
Here I could define, if in the next month, the production is ok, I could close out the CA, and say it was effective, but if same problem happens again, it was not effective, then I have to do a new analysis.
But if I decide to measure it, that if in three months in a row, the problem is not happening, ok, it was effective.
But if in the first month the problem appears, it was not effective, for that , I have to do a new analysis.
Finally I'm lost, don't know what to define (how and when) to measure the effectiveness ,mainly what effect has the short and long term measurements, if it seems it doesn't make any difference when is not effective.
Please give some insights.
Thanks
 

John Broomfield

Staff member
Super Moderator
#2
Firing an incompetent employee may not remove a root cause of ineffective maintenance from your system.

You may find that your process for recruiting, hiring and training employees remains ineffective.

And you may also find that your supervisory processes for monitoring and correcting ineffective work remain ineffective.

Your corrective action process should have identified and removed the root causes of ineffective maintenance to stop recurrence so broaden your search for evidence of effectiveness by taking a system view.
 
#3
Effectiveness would be based on hours of lost productivity, would it not? The verification of the plan itself would be based on maintenance records proving that the equipment was inspected and maintained as opposed to waiting for equipment failure and repair. Your goals and indicators of success should have been defined in the plan. If your plan isn't successful, then perform another analysis and uncover the true root cause of lost productivity, not just the root cause of that particular failure.

One of the systems of problem definition I prefer is the The Simplex Process by Min Basadur (I have no skin in the game but used it in my MBA program). It focuses on true problem definition through discovery, versus stamping out fires until another one erupts. Often, what you think is the problem is only the symptom of the problem.
 
Last edited:
#5
Firing an incompetent employee may not remove a root cause of ineffective maintenance from your system.

You may find that your process for recruiting, hiring and training employees remains ineffective.

And you may also find that your supervisory processes for monitoring and correcting ineffective work remain ineffective.

Your corrective action process should have identified and removed the root causes of ineffective maintenance to stop recurrence so broaden your search for evidence of effectiveness by taking a system view.
Thanks, John, I understand, let say I take that approach, fixing the root cause (recruiting), but In this case , what could be examples of criteria to measure effectiveness?
Thanks
 

Jim Wynne

Super Moderator
#6
Thanks, John, I understand, let say I take that approach, fixing the root cause (recruiting), but In this case , what could be examples of criteria to measure effectiveness?
Thanks
You haven't told us what "poor performance in a process" means, or how you determined that the cause of it was poor maintenance. The way that you verify effectiveness of corrective action is to see whether what you did keeps the bad thing from happening again. That could take days or months depending on the nature of the nonconforming condition. More details, please.
 
#7
You haven't told us what "poor performance in a process" means, or how you determined that the cause of it was poor maintenance. The way that you verify effectiveness of corrective action is to see whether what you did keeps the bad thing from happening again. That could take days or months depending on the nature of the nonconforming condition. More details, please.
Thanks Jim.
For the uncompliments, we found that direct cause, was a machine with failure , due to poor maintenance, for that, by improving the methods, and hiring a high competency operator, the problem disappeared.
We defined such goal in normal conditions but in the last three months , the production rate, started to decrease.
Then , we raise the AC.
Thanks
 

John Broomfield

Staff member
Super Moderator
#8
Thanks, John, I understand, let say I take that approach, fixing the root cause (recruiting), but In this case , what could be examples of criteria to measure effectiveness?
Thanks
Take a fresh look at your objectives for your recruiting process which, when combined with your training process should result in competent employees.

Newly acquired and newly trained employees usually work under close supervision until their competence is verified.
 

Tagin

Involved In Discussions
#9
As Jim pointed out, criteria and duration for effectiveness will depend on the particular situation.

I would be concerned that the action plan only included replacing an operator with a "better" operator. You have replaced a more fallible human with a less fallible human....but humans are still fallible.

In terms of PDCA, you have taken action to correct the DO phase. But, where is the action taken to prevent recurrence? This is what is missing. Is there a way that the process could be changed so that the issue is no longer dependent on the quality of the person (i.e., can you "build in" the proper maintenance to the process)?

You could also make improvements to the PLAN phase (e.g., recruiting/training criteria), as has been pointed out.

The CHECK phase is where it seems you need to implement something more (e.g., daily or weekly checks, etc.) that will detect an issue much sooner, so it won't have has big an impact before it is corrected, especially if the process can not be changed to prevent recurrence. This won't stop a N/C from occurring, but will potentially lessen the impact.

Finally I'm lost, don't know what to define (how and when) to measure the effectiveness ,mainly what effect has the short and long term measurements, if it seems it doesn't make any difference when is not effective.
The longer the duration of successful monitoring for effectiveness provides more evidence that the problem was really fixed. But I think the reason you are unsure about this is that you have not implemented any actions to prevent recurrence, so the only thing preventing recurrence is that the operator stays alert and never makes a mistake. And that could happen at any time, short term or long term. :)
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#10
so there are several ways to verify effectiveness, some involve monitoring to see if the problem recurs, some monitor to see if the cause recurs, others are based on certain knowledge of the causal mechanism and the solution.

The first two can be quite lengthy as you have to monitor for a period of time that exceeds the time for the original problem or cause to recur.

When there is no solid knowledge of the physics of the cause and solution, people often struggle with how to measure the effectiveness because all you really have is waiting to see if the problem recurs...I suspect that to be the case here.

I am concerned about trying to pursue the recruiting path as I'm not sure from what you have described that the poor maintenance stemmed from a person, particularly when you say that you also changed methods. we would need more details regarding the failure, what 'maintenance' wasn't done to create the fault in the machine, what was it about the maintenance that was missing, what new procedure did you put in place to improve the maintenance procedures and why was it felt that a different person was necessary (did the new procedure require a specific skill the previous person didn't have or was there a component of what the previous person did that contributed to the machine fault)
 

Top Bottom