In this article:
In the last article on FMEA we explored the structure of FMEA, and defined the meaning of each one of the elements that constitute the FMEA. We now turn to actions that would help us overcome, or at least reduce the risk.
Prioritize the risk
FMEA highlights risks. A risk is the possibility of a Failure. The comprehensive nature of FMEA tells us that we need to identify all risks; big and small.
Maximum effectiveness of the work you do comes from giving a priority to risks that cause maximum damage. Hence we need to focus on the effects of Failure first. Failures with most severe effects must be acted upon on priority.
Risk priority Number that we identified in the previous article is a multiplication of Severity, Occurrence and Detection rankings associated with the cause-Failure mode combination. However, focusing on the RPN alone could paint an incomplete picture of the failure. Consider the following 2 situations where RPN works out to 200.
Situation 1: 200 = Severity 5 * Occurrence 5 * Detection 8
Clearly, the situation 2 is far more serious. Severity 10 signals a potentially fatal consequence of failure. It signals a fatality that occurs without warning. Situation 2 therefore demands an immediate action over all others. A logical approach to prioritizing actions is therefore:
For a given failure mode, the Risk Priority Number must be calculated using the highest severity ranking among all effects associated with that failure mode.
Once a Failure Mode occurs, all Effects of this Failure Mode can potentially crop up. This includes the most serious effect as well. There is no way that effects can be made to occur selectively. It is only logical therefore to evaluate the risk with the most serious of all effects…leading us to the use of highest Severity.
The priority for actions is therefore given to Severity first, Occurrence next and Detection last. In fact, FMEA practice experts concur on the fact that actions should not be oriented towards improving Detection ranking at all. Priority should be given to preventing failures rather than detecting the failure and trying to water down its consequences. Detection focused actions can be taken only as an interim measure, till a more permanent Prevention measure is found. The quest for Prevention measures should go on in cases where you are forced to temporarily bank on Detection as a remedy to the problem.
To site an example, a seal bonding process in a valve manufacturing operation may be experiencing a large quantum of rejection. Banking on detection measures like 100% inspection can be only a short-term measure. A more permanent measure would be to determine the root-cause and bring down the occurrence of rejection through an intelligent process design.
All said, the fact remains that Actions can be taken on causes alone, and never on effects.
In summary, Actions need to focus first on Severity, then on Occurrence and if inevitable last on Detection. On account of this there have been instances where analysts have calculated the Risk Priority as a multiplication of the Severity and Occurrence alone, completely ignoring the Detection ranking.
Focus on Failure Mode elimination
A lot of literature suggests that Severity ranking can be changed only through actions on Design. This has lead to a widespread belief that Severity ranking can be changed only in Design FMEA.
The truth is that Severity ranking cannot be changed at all. No matter what you do.
If Severity of a Failure Mode has to be addressed, it can be done by either summarily eliminating the Failure Mode or by eliminating the Effect with which the Severity ranking is associated. This can be done through design. However, if the Failure Mode occurs (albeit with a very small probability of occurrence), and the Effect occurs, there is no way to change the Severity ranking associated with the Effect. A high Severity ranking can thus be addressed through measures of elimination of Failure Modes and Effects. It cannot be reduced.
Actions pertaining to bringing down the Occurrence and Detection rankings
Actions related to reducing Occurrence ranking should focus on making process improvements and mistake-proofing the process. These should not be confused with process improvement techniques that focus on establishing cause-effect relationship.
The exploratory technique of Design of Experiments that is widely used in Process improvement is aimed at establishing cause-effect relationships. This is focused on detection of relationships and not on prevention of errors. Similar, is the case with the technique of Statistical Process Control (SPC). SPC has two components. One is ‘Process Capability’ management that is done using Histograms and process capability indices of Cp and Cpk, and the other is ‘Process Control’ that monitors control charts that detect process drifts and changes in process variation. Process Capability management is a prevention mechanism, whereas Process Control is a detection mechanism. A clear understanding about the distinction between prevention and detection mechanisms will help you take more effective actions.
With the narration of causes are the descriptions of Current Controls of prevention and Current Controls of detection. Also documented are the currently identified Occurrence and Detection rankings.
Here, the Occurrence ranking should be identified based on the probability of occurrence of the Failure in spite of the Current Prevention control already being in place. Similarly the Detection ranking should be identified based on probability of detecting Failure in spite of the Current Detection control being in place. The rankings after improvement actions are taken are documented as the closing rankings that come out after the results of the actions are evaluated.
FMEA is correctly described as a living document. This means that nothing that has been done in a FMEA should be ever deleted or overwritten.
In the past practices, analysts used to overwrite the causes and actions and update the same ‘rows’ in the FMEA as the analysis, actions and improvements progressed. This leads to the wiping out of the precious trail of what has been done in the past, and how model improvement cycles have worked. All the learning of the past FMEAs was thus wiped out by overwriting. Thus to maintain the history, intermediate stages of FMEA used to be preserved as revisions. It was a herculean task to sift through the several revision documents to make out what had happened in the past.
The FMEA standards recognized this problem and eliminated revision numbers in the FMEA process. You will find that the fourth edition of the AIAG FMEA standard does not talk of the Revision number at all.
The correct and effective practice in FMEA is therefore to append everything that happens anew, to the existing FMEA, without scoring out any part of the historical work. If a new facet of failure is found in a cause that has been addressed in the past, simply add a new branch, leaving the past work intact.
As you progress, model process and product improvement cycles start getting built in your FMEA, learning from the past becomes handier, and your problem resolution cycle becomes leaner and faster.
To get the best mileage out of the FMEA practice, plan what actions are needed to be taken and take such actions in the planned time frame.
This aspect of learning from the past is vital in any FMEA practice.
This series of articles has attempted to explain
Readers are welcome to interact, make suggestions and discuss the article with the author.
Author: Ravindra Khare is a Founder and Director of Symphony Technologies. He is a qualified Mechanical and Industrial Engineer and a keen student of Quality & Productivity Technology for the past 25 years. He can be contacted at e-mail address: firstname.lastname@example.org or through us at email@example.com