Failure Mode and Effects Analysis (FMEA)

All products and services fail! Even the processes which have already achieved six sigma precision fail. However considering the possible sources of failures, the effects that they are likely to have and how prioritizing failure modes makes the product, service, process plan more robust. Considering possible failure modes and integrating them into the design goes a long way in reducing the fallibility of the product/service in question. The Failure Mode and Effects Analysis (FMEA) is a way to do the same.

The FMEA was first implemented by the aerospace industry in the 1960’s. Since then it has become an integral part of all projects where safety and reliability are major concerns. The automobile industry has extensively used FMEA. In fact, Ford Motor Company was one of the earliest users in the United States. These different industries have used different variations and versions of the FMEA analysis. These analysis vary from each other, however the crux remains the same i.e. to out-design potential problems before they can occur.

List out Potential Failures: The Pareto principle applies to failures too. Most of the failures that are caused have relatively few underlying causes. Therefore the proper management of these causes is imperative. The FMEA exercise relies on the experience of the people performing the FMEA analysis. The failure modes are listed down by means of brainstorming. The idea is to list down all the possible ways that one can think of in which the process will go wrong.

Attempt to Design Failures Out of the System: Once the possible failure modes in the system are identified efforts are made to prevent the failure from occurring. This can usually be done in one of the three ways mentioned below:

• Error Proofing: This is the strategy that needs to be followed when the failure mode in question is of high priority. This means that there is a high probability of the failure occurring and if it does occur, the entire system gets disrupted. In such cases, prevention is better than cure. Usually engineering and management teams are assembled and are urged to find solutions that will mitigate this risk.

• Increasing the Variability of the Process: A second strategy is to change the process so that the risk is eliminated. This may sometimes lead to operational losses. There must be a cost-benefit analysis to understand the implementation of this strategy.

  An example of the use of this strategy would be the production of Henry Ford’s famous Model T. Henry Ford ensured that the cars were produced only in black color. Thus he eliminated the complexities and failures that could have arisen if more than one colors was used.

• Control Planning: The last step is to create a failure control plan. This strategy relies on speedy detection of failure and setting control plan into motion as fast as possible. This strategy is implemented for smaller risks that are expected, anticipated an ddo not threaten the business.

Subjective

The problem with the FMEA analysis is that it is highly subjective. This means that it relies on the experts to solve the problem. Two people conducting the same analysis will come up with very different results. Hence the analysis is only as good as the person conducting it.

How to Implement FMEA

The FMEA analysis largely depends upon the people that are executing it. However, this does not mean that people cannot be trained to implement this analysis. It is just that the insight required to conduct this analysis requires the guidance and experience of a senior personnel. The step by step method to implement the FMEA analysis is given below:

Review the Process: The first step begins with understanding the process deeply. Under normal circumstances, we make a lot of assumptions. We assume the electricity supply will always be present, the raw material supply will be consistent and so on. The FMEA analysis focuses on explicitly stating the inputs and the pre-conditions that make the process work. By stating explicitly what is required, the executives are prepared for the next step.

Brainstorm for Failure Modes: In the above step, all the factors that could have possibly lead to the failure of the process are listed out. In this step, they need to be brainstormed for failure modes. That is the team needs to come up with ways that the process could go bust. They have to consider one factor, lets say electricity and then suppose what can go wrong. Lets say the supply could be turned off, there could be voltage fluctuations or the price of electricity could go up and many more! All these are failure modes of the process.

Rate for Occurrence: Once the failure modes have been listed, they need to be rated for probability of occurrence. This is usually done by assigning a score of 1,4 or 9. 1 represents very low chance of occurrence, 4 represents a medium chance and 9 represents an almost certain event.

Rate for Severity: The failure modes then need to be rated for severity of outcome. This is done by assigning a score of 1, 4 or 9. 1 represents very low disruption, 4 represents a medium disruption and 9 represents a show stopper.

Rate for Possibility of Detection: The failure modes are then rated for the possibility of detection. One needs to consider this from the point of view of the existing detection mechanism that the organization employs. Also the time frame in which the detection is done must be considered. Ratings of 1,4 and 9 must represent the increasing scale of severity.

Multiply to Get Risk Priority Number (RPN): The next step involves multiplication of the three numbers to come up with the Risk Priority Number (RPN). What has happened is that the three main characteristics of any failure modes have been separately rated. Now they need to be combined to understand the true threat that any individual failure mode faces.

Decide Cut-off and Prioritize: The final step is to arrange the failure modes in a descending order based on the scores which are generated. These scores are then prioritized to decide the cut-offs i.e. which must be error proofed, which must be controlled and where process must be changed to eliminate variation.