AI-Augmented DFMEA: Semi-Automated FMEA with Excel, Python, and ChatGPT Integration

SUMMARY & CONCLUSIONSFailure Mode and Effects Analysis (FMEA) is essential in reliability engineering but is often manual and time-consuming. This work implements local version of a semi-automated workflow that integrates Microsoft Excel, Python (using the xlwings package), and ChatGPT to accelerate and standardize FMEA generation while retaining engineering oversight.Engineers begin by entering the system context in an Excel worksheet; a Python back end then queries ChatGPT to generate an initial list of system components and candidate failure modes. Engineers review and refine these suggestions, define system boundaries and each component’s function, and approve or modify the items captured in designated Excel cells. The Python functions prompt ChatGPT with component context and repair/warranty data so that suggested failure modes, effects, and causes are grounded in organization-specific evidence rather than generic model knowledge.The implemented workflow produces FMEA entries that align more closely with observed field failure trends by incorporating repair and warranty data into prompt context. Auto-population of failure modes, causes, and effects substantially reduces routine authoring effort; initial tests indicate markedly faster completion times and broader coverage of plausible failures. Offloading routine content generation to the LLM allows engineers to focus on analysis, validation, and prioritization, improving consistency and the effective use of engineering judgment.Combining a familiar spreadsheet interface with Python automation and ChatGPT produces a practical, deployable semi-automated FMEA process. The approach enables reliability teams to generate richer, data-informed FMEA results more quickly, enhancing productivity and providing deeper insight into failure analysis. Planned Version 2 (cloud) work will include user-upload able, product- and system-specific knowledge bases (with macros to upload and tag documents to the cloud) so that subsequent Design Failure Mode and Effects Analysis (DFMEA) for the same system can limit retrieval-augmented generation (RAG) retrieval to a curated, relevant document set.