ACE: LLM-Based Bowtie Diagram Generation from FMEA – eBook

HomeResourcesArticleFMEA, Bowtie & Risk

Download UReason’s free eBook on AI-driven failure analysis and discover how Large Language Models can automate causal extraction from engineering documentation. Based on research conducted at the University of Amsterdam, the guide explores how FMEA data can be transformed into Bowtie diagrams to support faster and more scalable reliability analysis.

Tablet displaying UReason eBook cover titled “ACE – Automating Causal Extraction: Leveraging Large Language Models for Bowtie Diagram Generation in Failure Analysis and System Dynamics.”

Gain more information about:

  • Automating Causal Extraction
    Discover how Large Language Models can identify causal relationships in engineering documentation and translate them into structured risk models.
  • From FMEA to Bowtie Diagrams
    Learn how structured tables and narrative descriptions from FMEA documentation can be transformed into Bowtie diagrams using AI.
  • LLM Pipelines for Engineering Data
    Explore different approaches for extracting causal information, including retrieval-augmented generation, OCR-based processing, and vision-enabled pipelines.
  • Prompting Strategies for Reliable Results
    Understand how prompting techniques and model configurations influence the accuracy and consistency of generated Bowtie diagrams.
  • Challenges in AI-Assisted Failure Analysis
    See where LLMs perform well—such as structured data extraction—and where further improvements are needed, particularly with unstructured engineering narratives.

What’s Inside:

  • A Framework for Automating Bowtie Diagram Generation
    An overview of the ACE approach for extracting causal relationships from FMEA documentation.
  • Evaluation of Multiple LLM Models
    Insights from experiments using instruction-tuned models including LLaMA, Mistral, and Qwen.
  • Sensitivity Analysis of Prompting Strategies
    Results showing how prompt design and strict schema constraints affect diagram quality.
  • Structured vs. Narrative Data Extraction
    A comparison of model performance when processing structured FMEA tables versus unstructured engineering descriptions.
  • Future Opportunities for AI in Reliability Engineering
    Key lessons and directions for applying large language models to failure analysis and risk modeling.
UReason ebook on automating causal extraction from FMEA using large language models to generate Bowtie diagrams for failure analysis and risk modeling.

Download the eBook

This field is for validation purposes and should be left unchanged.

Fill in the form to access the document

Scroll to Top