A DFMEA by Any Other Name
Together, AIAG and VDA approaches may bring the best of both worlds
by Matthew Barsalou
German and American automotive industries have used different approaches to design failure mode and effects analysis (DFMEA). American companies have followed the Automotive Industry Action Group (AIAG) failure mode and effects analysis (FMEA) handbook,1 while German companies have followed the Verband der Automobilindustrie (VDA) standard.2
These approaches now have been harmonized, and a new DFMEA form has been released as part of the AIAG/VDA Failure Modes and Effects Handbook.3 Although there is now a harmonized standard, the AIAG and VDA approaches to DFMEAs were fundamentally different. The VDA approach looks at the interaction between systems and subsystems, and the AIAG approach examines each assembly and component separately.4
Richard Harpster, a risk and FMEA expert, described the VDA approach as more inefficient due to the use of structure analysis, function analysis and failure analysis. He used the example of a twin bed assembly with a frame subassembly that, in turn, has a wooden leg subassembly and finally a wooden leg as the component at the lowest level.
The requirement in this example is resisting the shock impact of 14,600 bed entries and exits by a 250-pound person during a period of 10 years. Harpster wrote that it would be difficult to identify the function of the wooden leg, and the verification requirements of the wooden leg alone would be unknown because the impact on the leg alone is not the same as the impact on the entire system—that is, the bed assembly. Furthermore, Harpster stated that four DFMEAs would be needed to cover all parts of the system.5
How can you know whether the bed will survive if you don’t include its constituent parts in an analysis? Don’t worry about the number of reduced force impacts that the wooden leg must survive because its evaluation should be included as part of a system test. Test the entire bed and you’ll know whether the leg fails, even without knowing what force was transferred to the leg.
This is exactly how it would be done in a true VDA DFMEA.
FCs, FEs and FMs
Harpster stated that four DFMEAs would be needed.6 This is incorrect because only the focus element would be needed in the DFMEA form. In this case, it would be the frame assembly with a final failure effect at the bed and a failure cause at the wooden leg. Furthermore, this bed example is misleading because the VDA approach would assess the entire system with prevention actions for the failure cause (FC) at the lowest-level system element and detection actions for the entire system.
The FC is at the leg, and that’s where the prevention action would take place. A failure mode (FM) at the leg would be the FC at the next higher level, which would be the frame assembly. This failure effect (FE) for the leg would be the FM for the frame assembly, and the FE for the frame assembly would be the FM for the system.
The chain of causality, known as “failure net” in VDA terminology, goes up through the assemblies until the final effect at the system level. The detection actions would be to detect the final FM; therefore, the detection action would occur at the system level for the cause at the component level. The bed example uses four system levels. VDA DFMEAs, however, generally use only three, and this is compatible with the new standard.
Figure 1 shows the chain of causality from the component to the system level—with the cause of the leg breaking being that the leg diameter is too small. This, in turn, leaves the frame assembly unsupported, resulting in a bed collapse leading to a potential occupant injury.
A VDA-style DFMEA often has dedicated moderators who use software to build a tree structure, such as the one in Figure 2. Here, the connections among FCs, FMs and FEs are created. In this example, the FM for the frame assembly is selected, and the FEs and FCs are visible on the right side of the illustration.
FE, FM and FC are relative in a VDA DFMEA. The effect for one system element is the cause at the next higher-level system element. This means the failure mode for the next lower-system element also is the cause for the system element in focus.
The VDA approach is more complicated, but it will look comparable to an AIAG DFMEA when viewed in a DFMEA form. The main difference between an empty VDA DFMEA form and an empty AIAG DFMEA form is that the VDA form has the detection and prevention actions stacked. VDA-style DFMEAs can be switched to the AIAG form with two clicks in the DFMEA software. Table 1 shows the example in an AIAG-style DFMEA form.
VDA and AIAG
In a VDA DFMEA, the entire system is assessed so the final severity for a component in a system’s assembly is based on the system-level severity, which is what affects the customer. In the bed example, a broken wooden leg assessed as a single component is only a complete failure, which would have a severity of eight. Failing to consider the final effect will result in an unrealistically low severity rating.
Alternatively, an engineer may realize a broken wooden leg can lead to injury and therefore give it a severity of 10, but the reason a broken component was assessed as a 10 may be unclear to non-bed experts. Making clear the entire chain of causality will be beneficial to those who formally only performed AIAG-style DFMEAs.
Table 2 shows the bed example in the new DFMEA form. Here, the system elements, as well as their functions, are considered. The FC is at the lowest level, the FM is at the level that is in focus and the FE is at the system level.
This archives the basic steps in the chain of causality. But individual steps are missing. I’d prefer to see “bed collapse” added to the FE, and I’m curious to see how this will be implemented in industry.
The VDA style delivers a superior DFMEA due to clearly linking FEs through system elements. But software and an experienced moderator are required so I’ve often recommended the AIAIG approach because a superior approach that can’t be used is simply useless.
Hopefully, the new standard gives us the best of both worlds.
- Automotive Industry Action Group (AIAG), Potential Failure Modes and Effects: Reference Manual, fourth edition, 2008.
- Verband der Automobilindustrie (VDA) (or translated into English, Association of the Automobile Industry), Qualitätsmanagement in der Automobilindustrie: Sicherung der Qualität während der Produktrealisierung Methoden und Verfahren—System FMEA (or translated into English, Quality Management in the Automotive Industry: Quality Assurance During Product Implementation, Methods and Procedures—FMEA System), 2003.
- AIAG/VDA, Failure Modes and Effects Handbook, first edition, AIAG, 2019.
- Chad Kymal and Gregory F. Gruska, “AIAG and VDA Release Draft of Harmonized FMEA Manual,” Quality Digest, Dec. 13, 2017, https://tinyurl.com/digest-AIAG-VDA.
- Richard Harpster, “The Case Against the AIAG-VDA DFMEA,” Quality Digest, Jan. 31, 2018, https://tinyurl.com/The-Case-Against.
Matthew Barsalou is an extramural researcher at Poznan University of Technology in Poland. He has a master’s degree in business administration and engineering from Wilhelm Büchner Hochschule in Darmstadt, Germany, and a master’s degree in liberal studies from Fort Hays State University in Hays, KS. Barsalou is an ASQ senior member and holds several certifications.