Failure Mode and Effects Analysis (FMEA) with R
Tony Gojanovic
June 2018
The following indicates the status of the most recent build with Travis:
This simple package creates a set of graphs for risk assessment and risk management. Specifically, severity, occurrence and detection as obtained through an FMEA exercise are plotted.
The libraries ggplot, ggrepel and dplyr are used with this package.
See vignette files for specific details on usage.
A disutility graph or hazard graph is a scatterplot in which the undesirability of a failure is based on failure severity and the occurrence (likelihood) of potential causes or mechanisms of that failure happening. In this package, random "jitter" is added to the severity and likelihood values in order to provide a more interpretable graph (severity and occurrence values are typically discrete).
Detection ability is plotted against occurrence orlikelihood of potential causes or mechanisms of failure. Essentially, it depicts the state of risk management controls. Random "jitter" is added to the detection and likelihood values in order to provide a more interpretable graph.
Ranking tables for severity, occurrence and detection are typically on a 1 to 10 scale. Definition of the differing levels can be found in FMEA literature or determined by the end user.
Package: FMEA
GitHub: https://github.com/cowboy2718/FMEA
Data for this function is input by the user (typically the FMEA session is captured in an Excel workbook). The following fields are what a typical FMEA worksheet possesses. Data must be tidy.
| Field | Description | Data type | Required by |
|---|---|---|---|
| item_no | risk item identifier | Character string | risk_disutility; risk_controls |
| process_step | process or design phase | Character string | risk_disutility; risk_controls |
| function | what the item's function does | Character string | |
| requirements | what are criteria for success | Character string | |
| failure | how the item might fail | Character string | |
| effect | what is the effect of failure | Character string | |
| severity | the failure severity rating | Integer (1 to 10) | risk_disutility; risk_controls |
| cause | cause of the failure | Character string | |
| occurrence | the likelihood of the cause | Integer (1 to 10) | risk_disutility; risk_controls |
| hazard | hazard = severity x occurrence | Calculated numeric (1 to 100) | |
| controls | what will detect the failure or cause | Character string | |
| detection | the effectiveness of the controls | Integer (1 to 10) | risk_disutility; risk_controls |
| rpn | rpn = severity x occurrence x detection | Calculated numeric (1 to 1000) | |
| mitigation | how to mitigate the failure | Character string | |
| responsibility | who owns the mitigation activity | Character string | |
| date | when the mitigation will be completed | Date | |
| mitigation_status | status of the mitigation activity | Character string | |
| sev_1 | updated severity after mitigation | Integer (1 to 10) | |
| occ_1 | updated occurrence after mitigation | Integer (1 to 10) | |
| haz_1 | updated hazard | Calculated numeric (1 to 100) | |
| det_1 | updated detection after mitigation | Integer (1 to 10) | |
| rpn_1 | updated risk priority number | Calculated numeric (1 to 1000) | |
| date_1 | update date for mitigation | date |
|
Validation and verification of risk mitigation is the key to a successful risk mitigation program. Specifically,
Verification assures that in fact a proper corrective action has been implemented regarding a specific risk mitigation action. Was the mitigation action completed and was the documentation completed to support all of the necessary changes? Was the change visible and trained, etc.? Note: Verification may involve recreation of the problems to determine if suspect root causes are had been addressed related to the risk. This step can be by-passed since some risk states may be too risky to recreate.
Validation requires proof that the risk mitigation action is sustainable and effective. Does the mitigation action actually work the way that it should.
• Systems are in place to sustain corrective action and prevent future occurrences.
• Reliable detection controls are in place and are tested.
• Does the mitigation action actually work the way that it should.
• Must be sustainable and robust. Makeshift or temporary fixes do not apply as sustainable.
Note: A “show me” will be a demonstration of validation i.e. that a sustainable system of reliable detection controls exist.
The following are useful resources in understanding Failure Modes and Effects Analysis and risk assessment and management.
Carlson, Carl S., Effective FMEAs: Achieving Safe, Reliable, and Economical Products and Processes Using Failure Mode and Effects Analysis, John Wiley and Sons, New York, 2012, p. 164.
Dussault, Heather B., "The Evolution and Practical Applications of Failure Modes and Effects Analyses", RADC-TR-83-73, Rome Air Development Center, Air Force Systems Command, Griffiss Air Force Base, New York, March 1983.
Haimes, Yacov Y., Risk Modeling, Assessment, and Management, 2nd edition, John Wiley & Sons, New York, 2004.
MacKenzie, Cameron, "Summarizing Risk Using Risk Measures and Risk Indices," peer-reviewed draft, Defense Resource Management Institute, Naval Post Graduate Schook, 2015, http://faculty.nps.edu/camacken/docs/MacKenzie%20Summarizing%20Risk%20Using%20Risk%20Measures%20and%20Risk%20Indices.pdf
Roland, Harold E., and Brian Moriarty, System Safety Engineering and Managements, John Wiley and Sons, New York, 1983, pp. 187 - 193.
Tversky, Amos, and Daniel Kahneman, "Judgement Under Uncertainty: Heuristics and Biases," Science, New Series, Vol. 185, No. 4157. (Sep. 27, 1974), pp. 1124-1131.
The following were very useful links used in the development of this project.
- Building R Packages in R Studio, John Muschelli
- Putting your R Package on Github, Karl Broman
- RStudio & Github Integration, James Dayhuff
- Creating RStudio projects from GitHub Repositories, Nicholas Reich
- Developing R Packages | Part 7: Unit Testing, Colin Pistell
- Travis CI Tutorial - How to Use Travis CI with Github for Continuous Integration, Full Stack Academy
- Create Awesome HTML Table with knitr::kable and kableExtra, Hao Zhu