What are some unnecessarily detailed Wikipedia articles

Process FMEA

Process FMEA summary

Process managers dream of flawless processes. However, reality shows that errors (quality disruptions) can occur time and again during implementation. Process FMEA, for example, is suitable for predictively and methodically identifying possible quality weaknesses and taking measures.

The following article explains what an FMEA for processes is and how it is carried out. There are also references to common problems and application errors. The article is intended for users who need step-by-step instructions to maintain high quality standards in process management. In some places reference is made to Wikipedia articles for further research, in which further sources are named.

What is a process FMEA?
How does a P-FMEA work?
Template usage
Step 01 | Process detail analysis
Step 02 | Think about possible causes of failure
Step 03 | Predict possible effects
Step 04 | Determine the strength of the impact
Step 05 | Estimate frequency of occurrence
Step 06 | Assess the probability of detection
Step 07 | Calculate RPZ
Step 08 | Develop an action plan
Step 09 | Implement measures
Step 10 | Calculate the resulting RPN
Other possible activities
Difficulties in practice
Critical appraisal

What is a process FMEA?

FMEA stands for Failure Mode Effect Analysis. This stand-alone method is applied to processes. The process FMEA (hereinafter referred to as P-FMEA) is an established method with the aim of preventively identifying and evaluating process risks (especially after process adjustments or new introductions) and taking measures to minimize risks. Weak points in a process are usually the reason for carrying out a P-FMEA. It is therefore one of the quality-oriented analysis methods.

The P-FMEA elaboration at the right moment can have a very positive influence on the process development and also save a lot of money. A P-FMEA is most effectively used in the development phase, namely from the creation of the process concept to the full start of the implemented process. It is therefore often used when new products and thus new processes are introduced. The P-FMEA can also be used for other processes if quality problems arise due to changes to the process or changes to the tools or the environment.

The method is used with the aim of achieving the most error-free design of processes in compliance with all customer and quality requirements. This methodology is prescribed for ISO TS 16949 certified automotive suppliers, although the standard does not make any statements about the specific implementation. In this way it can be adapted to the specific company. It has proven its worth there to analyze potential process risks at an early stage, i.e. preventively, with the help of the P-FMEA.

Due to the complexity of processes, the P-FMEA aims in particular at a multidisciplinary consideration of possible process problems. That is why the team identifies potential risks in a methodically structured manner, records and sorts them and derives risk strategies. The application of the method is particularly useful if defined quality standards have to be achieved as a goal, but numerous influencing factors exist.

Use the P-FMEA ...

... when a process or service is designed or redesigned.

... when an existing process is carried out in a new environment or by new actors.

... when control plans are developed for a new or changed process.

... when improvement goals are planned for an existing process.

... when errors in an existing process are to be analyzed.

... periodically throughout the life of the process.

How does a process FMEA work?

A P-FMEA considers every single process step and checks possible sources of risk. For example, based on the Ishikawa method (also fishbone) there can be potential sources of risk M.man, M.machine, M.method, M.aterial, M.eating and M.ilieu (environment) are illuminated. Identified risks are then rated on a scale from 1 to 10 according to the following variables:

  1. Severity - Assessment of the impact (of the failure in the process), with 1 being the least concern and 10 being the most dangerous concern.
  2. Occurrence - Ratings the likelihood of occurrence, with 1 being the lowest and 10 being the highest.
  3. Detection - Evaluates the likelihood that an error will be detected, with 1 being the highest detection probability and 10 being the lowest detection probability.
  4. RPN - risk priority number = meaning x occurrence x detection. High RPN values ​​require corrective action. The corrective action ideally leads to a lower RPN value.

These values ​​are usually determined within a multifunctional team. A P-FMEA moderator keeps the team on track, i.e. scope, goal and time span for the implementation of the P-FMEA and leads the team through the method-guided process.

Template usage

Figure 1: The implementation of the P-FMEA using a form template is much easier.

The implementation of a P-FMEA is made much easier if a corresponding form is used. This ensures that essential steps are not skipped or details are ignored. A usable template can be found at https://www.mi-nautics.com/downloads. Their use of this document template will be discussed later.

The P-FMEA is a living document that should be initiated before the process is introduced and maintained throughout the entire product life cycle. In this way, ongoing risk assessment for quality management is also supported.

The following is a practical description of how a P-FMEA should be carried out in 10 steps.


Step 01 | Process detail analysis

Figure 2: Analytical process model as the basis for searching for possible errors

For a detailed consideration, a process representation is required, which at best is based on an analytical process model. For Figure 2, an operational process model was modeled in BPMN 2.0.

With the process flow diagram, the P-FMEA team members should become familiar with the process by physically going through the process. This is the time to ensure that everyone on the team understands the basic process flow and how the process components work. An initial look at such a model is helpful in bringing everyone together on the same details. Usually, not all team members are familiar with all the details of the process, so a joint inspection based on the model is helpful for everyone.

This model representation is ideally suited to identify individual process components, because it provides information about those involved, activities carried out, input and output data, decisions, rules, interfaces, tools as well as desired process results and thus ensures that really all components are received can be found in the P-FMEA to be carried out.

Note: If it can be assumed that the process model does not contain all the details (ie almost always), it is necessary to spend time on the spot (“gemba”) in order to understand the details and external circumstances; then the process model must be supplemented.

For the purpose of further analysis, it is advisable to assign a unique ID to all process elements in the process (Column B of the template). This can be a unique and not multiple use of labels, but also - as is often done in practice - a numbering of the elements. These unique identification attributes are used in the further course of the analysis and are used for referencing.

A list of all process components (with a unique ID) is helpful in order to identify their value contribution (Column C, gray lines of the template). Some elements will be value-added or non-value-added. Elements that meet regulatory requirements are often referred to as essential non-value-added as the third characteristic.

Step 02 | Think about possible causes of failure

A P-FMEA is a team analysis. This team is now brainstormed to identify potential sources of error (Column E of the template). The results of the possibly completed process detail analysis from step 1 are used as a guide. Possible error possibilities are now identified for each individual element. These are circumstances that can prevent individual elements of the process from achieving their desired results. It is not advisable to skip individual elements here. Completeness and thoroughness are required at this point. Findings from other sources can also be added to this collection. Perhaps support or quality events are known from the past, which can also be added here in order to complete the list of possible error possibilities as far as possible.

However, the bare identification of the error cases is not sufficient. The correct causes of the error must be identified (Column F of the template). However, this is not always easy. Here it is advisable, for example, to get to the bottom of the matter with the 5xWhy method to ensure that the reasons for the error can also be effectively processed.

Note: The compilation of the possible errors and sources of error, as well as the creation of the entire P-FMEA, is associated with effort. The elaboration of 2 complete P-FMEAs for two similar processes is of course a waste. A modular structure in which the final P-FMEA uses existing P-FMEA components is much more effective. This procedure considerably simplifies a later P-FMEA update. Some modeling tools offer database-supported attribution of model elements. This makes a simple modularization possible.

Step 03 | Predict possible effects

Now that the possible error possibilities and, above all, the sources of errors have been compiled in step 2, it is now time to consider what effects these potential errors can have (Column C in the template). It can also be the case here that one possibility of error can have several effects. These are now to be determined. However, only those effects are of interest that have a direct impact on process components and prevent them from fulfilling their intended function. Some errors are also noticeable for the (process) customer or influence the operational environment or the environment.

The collated effects should ideally be formulated in such a way that the concrete meaning for the process or the process result becomes clear. A description that is too vague makes it difficult to identify specific risks. Step 04 | Determine the strength of the impact

Not every impact is of constant importance (B). For this reason, it is first necessary to determine how severe the impact will weigh or, better still, how strong the impact will be felt. The team makes an estimate of the importance of each (Column D of the template) carried out how strong the effects will be felt for the (process) customers, subsequent process steps or the employees.

The severity of the impact is estimated and determined based on process effects. The severity rating is typically between 1 and 10. The typical severity levels for process effects can be as follows:

  1. no impact
  2. rare rework at the place of execution
  3. irregular rework at the place of execution
  4. regular rework at the place of execution
  5. Work interruption, rework with another tool
  6. Work interruption, rework at another location / by another actor
  7. There are major disruptions in the process and downstream work is affected
  8. Larger disruptions prevent implementation, noticeable delays or failures are noticeable for (process) customers
  9. Regulations are ignored, material or immaterial damage occurs
  10. There is danger to life and limb

Step 05 | Estimate frequency of occurrence

This step is about a further evaluation of the error possibilities and sources of error. In this case, the frequency of occurrence (A), i.e. the assessment of how often the error is likely to occur, must be carried out (Column H of the template). Every identified possibility of failure is to be evaluated accordingly. Sources for this can be disposal logs, rework reports, customer complaints, logs of preventive measures or others.

The assignment of frequencies of occurrence allows a ranking.

In practice, the following evaluation has proven to be practicable (although organization-specific deviating evaluation definitions can be found):

1 prevented by process design; checked for errors

2 occurs in> = 1 out of 1,000,000 process runs

3 occurs in> = 1 out of 100,000 process runs

4 occurs in> = 1 out of 10,000 process runs

5 occurs in> = 1 out of 2,000 process runs

6 occurs in> = 1 out of 500 process runs

7 occurs in> = 1 out of 100 process runs

8 occurs in> = 2 out of 100 process runs

9 occurs in> = 5 out of 100 process runs

10 occurs in> = 10 out of 100 process runs

Actions can be directed against failure causes that occur frequently (see step 10). Particular attention must be paid to points with a high degree of severity. These points need to be checked to ensure that due diligence is being performed.

Step 06 | Assess the probability of detection

Now it is a matter of checking how high the probability is that the error will be discovered within the organization before the customer does. For this purpose, current preventive measures are also required to avoid the errors listed (Column G of the template), analyze as well as detection measures (Column I of the template).

For each error possibility it is to be assessed which probabilities of detection represent the currently implemented avoidance and detection measures (Column J in the template).

Note: The detection measures currently being carried out can relate to possible errors, causes of errors or the effects of errors. Preventive measures cannot relate to impacts, as these cannot occur if successfully avoided.

The probability of detection can be rated and also includes values ​​on a relative scale from 1 to 10. In practice, a rating can look like this:

  1. Error (cause) was completely prevented:
    Error cannot occur; Poka Yoke / Validations grab
  2. Error detection immediately at the place of execution:
    The implementation of the activity is validated and generates recommendations for action; results in longer processing times
  3. Error detection immediately at the place of execution:
    The execution of the task has failed, unnecessary work has been carried out, process performance can no longer be provided; results in failure costs
  4. Error detection at the next location:
    causes inspection work / necessity and reassignment of tasks to the upstream location or causes considerable additional effort elsewhere due to correction work; results in non-value-adding work and consequently in capacity reduction
  5. Error detection at downstream quality gates (test points) takes place:
    unnecessary work has been done and causes considerable additional effort in other locations due to correction work; results in non-value-adding work, in a reduction in capacity and an increase in testing activities at the quality gate.
  6. Error detection at downstream quality gates (test points) takes place:
    unnecessary work has been done, process performance can no longer be provided; results in capacity reduction and failure costs
  7. Error detection only at the end of the process:
    unnecessary work has been done and causes considerable additional effort in other locations due to correction work; results in non-value-adding work, capacity reduction and renewed testing at quality gates
  8. Error detection only at the end of the process:
    unnecessary work has been done, process performance can no longer be provided; results in capacity reduction and failure costs
  9. No controls, the (process) customer recognizes the problem irregularly
    Unnecessary work has been done and causes considerable additional work in other locations due to complaint processing; results in activities that do not add value, error costs and capacity reduction
  10. No controls, the (process) customer regularly recognizes the problem
    Unnecessary work has been done and causes considerable additional work due to complaints elsewhere; results in activities that do not add value, strained customer-supplier relationships, error costs, capacity reduction and possibly damage to image

Step 07 | Calculate RPZ

The risk priority number (RPN) is the product of the three previously estimated ratings and reflects the relative risk (Column K of the template). The higher the RPN, the higher the potential risk. It is calculated as follows:Since each of the three relative rating scales is between 1 and 10, the RPN is always between ≥1 and ≤1000. The higher the RPN, the higher the relative risk. The RPZ offers us an excellent tool to prioritize targeted improvement efforts.

Step 08 | Develop an action plan

Derived measures always aim to reduce the RPN, which is multiplied from the meaning (B), the occurrence (A) and the discovery (E). These measures can be devised together with the team. The influencing of the variables mentioned can, however, be influenced with different amounts of effort.

A reduction in importance (B), i.e. the reduction in effects, is usually only necessary through comprehensive and far-reaching measures, which are often of a strategic nature. A change in the business model, the target group, but also a change in the scope of the process is conceivable here.

The reduction of the occurrence (A), i.e. the frequency of occurrence, requires the elimination (or at least close monitoring) of the causes of errors. In this context, error avoidance strategies such as poka yoke, data validation etc. are used for immediate error detection.

The reduction of the detection (E), i.e. the improvement of the detection probability, usually requires investments in technologies or other methods. The engineering sciences have developed processes here that address this variable. Statistical process controls (SPC), Design of Experiments (DoE) are mentioned here as examples, but work organization measures also help, such as work instructions, training courses, preventive maintenance, etc.

Which RPN is acceptable for an organization must be clarified individually. Corrective measures must be developed for unaccepted risks (Column L of the template)that bring about a lowering of the RPN.

In order to prevent a "You have to", responsibilities and deadlines are to be named (Column M of the template). Providing a name and a due date ensures a certain commitment and ensures future reporting. The nominated person has the task of reducing the RPN and should be measured against it.

Note: The RPN must be reduced to a tolerable risk level. The complete defense against risks (i.e. RPN = 1) is not an economically sensible goal (and probably impossible). It must be clarified within the organization which risks (i.e. which RPN) are acceptable.

Step 09 | Implement measures

The action plan developed by the team in step 8 must now be implemented. Dates and persons responsible were named for this. Depending on the measure, different efforts are now required. Poka Yoke measures can usually be implemented with a small budget, investments in test automation or software development in general are usually associated with projects. Classic project management methods such as Gantt diagrams or critical path diagrams are available to ensure that activities are kept up to date.

The implemented measures are logged (Column N of the template).

Step 10 | Calculate the resulting RPN

Figure 6: RPZ: triangular relationship for assessing risks

After individual corrective measures have been implemented, the parameters meaning (B), occurrence (A) and discovery (E) (Columns O, P, Q of the template) and therefore the recalculation of the RPN is necessary (Column S in the template)to prove the effectiveness of the measures.

In this way, it can be demonstrated to important stakeholders that corrective measures taken (and thus investments, if applicable) (Column N of the template) have achieved the desired effect.

Remaining risks will continue to be identified. It must then be assessed whether these are tolerable or whether further future measures need to be developed.

Other possible activities

In day-to-day business, things sometimes get stuck. To prevent this from happening, further measures should be considered in order to ensure the success of the derived measures. Below is a short but not exhaustive list of suggested actions:

  1. The follow-up of the derived measures creates the necessary pressure to act. For this reason, it is advisable not only to name responsibilities, but also to identify reporting channels and other stakeholders.
    In practice, a RACI (Responsable, Accountable, Consulted, Informed) matrix has proven to be a successful tool because it can be used for regular reporting to management.
  2. A responsible person and an interval at which the P-FMEA is to be updated at least should be named. Process owners or process managers could perhaps be responsible for this task and arrange for it to be carried out accordingly.
  3. it is conceivable to link process performance or quality to variable remuneration components for employees. This miraculously ensures continuous processing and further development.

The success of every FMEA lies in its successive addition and above all in its repetition.

The best organizations do this exercise regularly. They take actions and update the process FMEA scores. And again. And again.

Difficulties and Risks in Practice

Even if the procedure for the implementation was presented here in a straightforward manner, some difficulties are to be expected in practice. The following are some of the more common problems encountered when using a P-FMEA. "Caution!" Applies here. Attempts should be made to mitigate or circumvent the following problems:

  1. P-FMEAs should never be completed by just one person. The goal must be to involve a team that owns the process. A buy-in from every team member is required to prevent the P-FMEA from becoming incomplete and non-binding.
  2. It is a good idea to take the time to figure out which process really needs a P-FMEA. If you carry out too many P-FMEAs in non-critical processes, resources are tied up unnecessarily.
  3. Once the decision has been made to pursue a P-FMEA, it is important to include people on the team who have extensive experience with the process. Only they can exchange historical facts (!) That prove invaluable when assigning severity, occurrence and recognition values. On the one hand, this will help to obtain a more precise representation of the process elements and, on the other hand, to compile the wish list that the team would like to see implemented from a risk perspective.
  4. It is extremely important to spend time at the “gemba” or the actual location where the process takes place, as it is useful to understand in detail every single process and sub-process that is part of the P-FMEA.
  5. As a preliminary work for the P-FMEA, it is advisable to have the data for the field problems available with a preliminary analysis.
  6. Whenever the process, specification, design, material, etc. change, the P-FMEA must be updated and a new RPN value must be calculated.
  7. A fixed time interval for the review of the P-FMEA of the process must be defined, with the flexibility to analyze a change ad hoc. This must be a mandatory part of the organization's change management system.

Typical single errors


  • Different evaluation of the meaning "B" for errors with identical consequences

Update error:

  • Change of rating "A" when introducing additional detection measures and vice versa
  • P-FMEA does not correspond to a real process (no update after process change)
  • Measures no longer valid in P-FMEA
  • Handling of controls in P-FMEA does not correspond to the control plan

Formal errors:

  • P-FMEA Team is incomprehensible
  • P-FMEA history is incomprehensible - when and what (a rough description is sufficient) was changed in P-FMEA
  • Missing dates for the measures
  • It is not possible to distinguish between measures that have been introduced and those that have not yet been introduced.

Team mistakes:

  • destructive group dynamics
  • different level of knowledge of the team members
  • competing personal goals of team members

Critical appraisal

All methods, including the P-FMEA, have advantages and disadvantages. In order to present this clearly and to enable the reader to weigh them up, these are compared below.

  • P-FMEA has a preventive effect; one can visualize potential risks
  • potential weak points can be eliminated before they cause damage in the process
  • Repeated implementation builds up empirical knowledge in the organization
  • Know-how about error relationships and quality influences are systematically collected and made available
  • Reduction of errors in the early stages of process design
  • Improving the quality of processes
  • Detailed elaboration of the FMEA is relatively time-consuming and costly
  • The benefit of a P-FMEA is not certain, as numerous errors can be made during implementation, e.g. degradation of the P-FMEA as a point in a checklist, which does not justify any measures.
  • Use of many forms; Bureaucratic risk
  • Sufficient training and motivation for employees to use the P-FMEA.
  • Subjectivity of the risk priority number and thus uncertainty of the RPN; therefore do not use RPN as the only prioritization feature!