FMEA is an analytical tool used in the design phase to help mitigate risk and failures during processes. Below we'll discuss the main types of FMEA and how to implement it into your maintenance processes.
Failure mode and effects analysis (FMEA) is a step-by-step process for anticipating things that could go wrong during the design stage by identifying all possible failures in the design, manufacturing and assembly processes. In other words, it's a structured approach to discover ways in which a process can fail (failures) and ways those failures lead to waste, defects or dangerous outcomes (effects). As part of the root cause analysis process, FMEA helps you minimize and limit these failures.
FMEA is broken down into two fairly broad categories: Design FMEA and Process FMEA. Each of these categories addresses failures in two different scenarios.
As the name implies, Design FMEA looks at potential risks in a new or changed design of a product or service. Design FMEA assigns severity or danger rankings to design functions, failure modes and the effects those failures could have on the customer. Once failure modes and severity rankings are identified, causes of the failures are sought out and identified to help with preventive maintenance schedules. For example, the occurrence ranking in the Design FMEA process (which we will discuss later) helps determine high-probability causes, initiating action to prevent failures from occurring. So, when should you use Design FMEA?
Design FMEA is great for identifying risks on a program as early as possible and mitigating failures as proactively as possible.
Process FMEA looks at potential failures impacting product quality, diminished process reliability, customer dissatisfaction, and environmental safety and hazards due to human error, materials and machines used, environmental factors, and more. Once potential failures are identified, severity rankings are assigned to each one. Process FMEA dissects all the steps in a current process and analyzes them individually to identify risks and possible errors. Use Process FMEA when:
Identifying risks of new technology and processes helps proactively prevent failure through preventive maintenance scheduling.
There are several instances or circumstances when performing FMEA is a good idea to ensure risk is mitigated and your operation is running smoothly, safely and at maximum capacity. Most people use Design and Process FMEA when:
It's also a good idea to do FMEA periodically throughout the life of the process. Consistently examining quality and reliability ensures processes are improved, giving you optimal results.
Now that you know what FMEA is, let's take a look at three criteria used to analyze potential failure modes. FMEA uses three criteria to search for potential problems: severity, occurrence and detection.
These criteria (and their ranking numbers) make up the equation to calculate a risk priority number (RPN) for each failure mode. The RPN calculation looks like this:
Let's take a look at a simple example to see how to calculate RPN. We've performed FMEA for a new espresso machine for our coffee shop. The management and staff determined if the espresso machine failed, it would greatly impact business by not producing coffee, which would lead to lost sales, angry customers and eventually loss of business from those angry customers.
They assign the espresso machine a severity rating of 10. The team decides the frequency in which a failure could occur on the espresso machine is fairly low, so they ranked the occurrence factor as a 4. Finally, the team determines by properly cleaning and maintaining the espresso machine each day and listening for abnormal noises or monitoring for quality production, they can detect a potential failure fairly easily. They rank the detection part of the equation as a 2. This gives an RPN of 80 (10 (s) x 4 (o) x 2 (d) = 80).
In most cases, severity can't be reduced, so concentrate on decreasing occurrence and increasing detection. It may be tempting to focus only on high RPN scores, but be sure you address anything with a high severity score, regardless of the total RPN. For example, you could have an equation like this:
10 (s) x 2 (o) x 2 (d) = 40
Even though the overall RPN is fairly low, the severity is at the maximum and should be assessed.
The FMEA process is conducted in a step-by-step fashion because each step builds on the previous one as you work through the analysis process. FMEA is done in seven steps, each one carefully designed to make the analysis quick and effective.
FMEA is a team effort, and a team approach is vital to its success. Your FMEA team should be led by a responsible manufacturing engineer or lead technician. Team members can include design and process engineers, materials suppliers and even customers. Once your team is assembled, it's time to do some pre-work – that is to collect and create key documents, such as information on past failures and prep documents. Prep documents should include:
Engineering consultant Quality-One International also recommends putting together a pre-work checklist to maximize FMEA efficiency. Your checklist can include things like:
Step one is a great time to consider all the ways each component could fail. Reviewing existing documentation and data likely will reveal several potential failures for each component in question. Brainstorm an exhaustive list initially and then pare down or combine items generated by the original list. Murphy's law states, "Anything that can go wrong will go wrong." During the step-one brainstorming session, keep this in mind when identifying functions, processes, systems and components that have the potential to fail.
Using an FMEA template or outline, it's time to add the functions, failure modes and effects to determine severity rankings. When listing the functions, make sure each one can be measured in some way. Functions may include:
Let's use the example of installing driver-side airbags in a car manufacturing assembly line. In this scenario, the function would be to properly orient and place the airbag into the assembly fixture.
Next, you'll list possible failures for each function. Think of failures as "anti-functions" and consider things like:
Using our example of installing the driver-side airbag, one possible failure mode could be not receiving the correct airbag for installation.
Now we need to list the possible effects our failure mode could have, making sure to give each effect a severity ranking (1-10). If the severity level is a 9 or 10 at this stage, actions should be considered. Potential effects for receiving the incorrect airbag for installation could be a delay in assembly while waiting for the correct airbag to arrive, or the wrong airbag being installed, causing a deployment malfunction in the event of a crash and leading to driver injury. Your team might decide to give this effect a severity ranking of 9 or 10.
Step three involves determining potential causes and prevention controls using an occurrence ranking. You can brainstorm causes using past failure data or by getting ideas and input from the design team. For example, why would we receive an incorrect airbag? One possible cause could be human error. An assembly-line stop prior to ours may have gotten the driver- and passenger-side airbags mixed up, or there may have been a discrepancy in the total number of airbags ordered, leading to more passenger-side airbags coming down the line. You and your team might decide there is a moderate chance this potential cause could occur and give it an occurrence ranking of 4.
Step four requires you to brainstorm and discuss controls or processes that ensure the design meets the requirements (Design FMEA) or, if a failure occurs, the likelihood an undetected failure mode would reach the customer. You can split this into two columns in your template: current process controls (prevention) and current process controls (detection).
Using our example, a process control currently in place might be a set of airbag assembly instructions. A current process control for detection may be a visual check of the airbags performed by the operator. You and your team may determine the likelihood of this failure being detected is fairly good and give it a detection ranking of 6.
Remember the RPN equation from earlier? This is where you assign an RPN number to each action established in steps two through four. The RPN number helps prioritize and assign follow-up action items. As discussed previously, RPN is calculated by multiplying the severity, occurrence and detection rankings for each possible failure, cause and control combination. After you've assigned RPNs for additional follow-up, assign the actions to the appropriate employees and make sure due dates are set for the completed actions.
Based on the assigned rankings in our example, the RPN is 240 (10 x 4 x 6).
Since the whole purpose of performing FMEA is to discover and mitigate risk, an action is only complete once it has been determined that it successfully reduces risk. In this step, failures should be listed in descending RPN order, so you can concentrate your efforts on the most critical areas. You may have heard of the Pareto principle, which states that 80 percent of issues come from 20 percent of the causes. This means the decision on where to focus your attention shouldn't be based strictly on the RPN alone, although it is a good starting point. The FMEA team leader should ensure actions are taken by the pre-determined due dates so a design review can occur.
Finally, you want to see if your actions truly did mitigate risk. Once all risk-mitigating actions have been implemented, the FMEA team should meet and re-rank each value (severity, occurrence and detection) and calculate a new RPN. The old and new RPNs should be compared, and if the risk factors are lower, new actions can be implemented into the design or process phases.
Below is an FMEA template using the airbag example from above. Listed topics are defined as follows:
Many people add additional columns to the template after the RPN column to help keep track of action items and track improvement. In the case of our airbag scenario, additional colums might look like this:
Now, you can add new columns for Severity, Occurrence, and Detection to see if the changes made a difference in the RPN.