AI Tool Fixes Chemical Plant Blueprint Errors

Northeastern University industrial engineering student Sierre Ternoey has developed an artificial intelligence agent capable of diagnosing and correcting errors in chemical process simulation flowsheets. During a research co-op at RWTH Aachen University in Germany, Ternoey addressed a persistent industrial challenge: manufacturing plants rely on complex software to map production workflows, yet initial simulations frequently generate opaque error reports that demand extensive engineering expertise to resolve. Despite lacking formal chemical engineering coursework, Ternoey engineered an AI system trained to interpret diagnostic reports, isolate simulation failures, and automatically repair the underlying flowsheet architecture. In controlled trials, the agent successfully resolved 26 of 30 test cases while preserving the engineer's original design parameters. Research associate Jan Pyschik, who assigned the project, initially viewed it as a high-risk assignment due to its unpredictable outcomes, noting that previous students were discouraged from attempting it. Ternoey's successful execution relied heavily on the autonomous problem-solving framework taught in Northeastern's Cornerstone of Engineering curriculum under Professor Kathryn Schulte Grahame. The program emphasizes iterative milestone tracking, self-directed learning, and proactive initiation, training undergraduates to approach unstructured technical challenges without predefined solutions. Ternoey applied these methodologies to the Aachen research environment, translating foundational programming and systems analysis skills into a functional diagnostic tool. The project's success has prompted her supervisors to draft a peer-reviewed manuscript and adapt the prototype into a standalone utility for chemical engineers navigating simulation software bottlenecks. The initiative highlights a practical convergence of machine learning and industrial process engineering, demonstrating how automated diagnostic agents can significantly reduce troubleshooting time and eliminate manual guesswork in workflow optimization. Academic staff note that the project exemplifies a broader educational shift toward cross-disciplinary adaptability and engineer-led innovation. By treating simulation repair as a structured engineering problem rather than a static coding exercise, Ternoey effectively bridged data science with process engineering requirements. Researchers at RWTH Aachen are now refining the algorithm's architecture to improve accuracy across edge cases, with plans to commercialize the diagnostic tool and publish the underlying methodology. The outcome validates the efficacy of project-based learning in preparing students for complex industrial applications while underscoring AI's expanding role in streamlining traditional engineering workflows.