Solution
We developed an agentic workflow that acts as an information and simulation-based decision-support system for process experts and operators. It combines two core capabilities: (1) intelligent access to operational knowledge and (2) simulation-driven exploration of process behaviour. The system can be deployed securely on-premises, ensuring full data confidentiality.
By embedding all relevant model documentation, operational manuals, and data sources into a Retrieval-Augmented Generation (RAG) framework, the solution transforms static documents into a dynamic, searchable knowledge base. Process experts can query the system in natural language to:
- Retrieve targeted information about process operation and modelling from knowledge sources, and
- Configure and run process simulations for different operating scenarios.
This approach offers a promising path to bring advanced simulation models to process experts and operators for industrial operational decision support.
Successful POC for industry
For the proof of concept, we used the Søderberg electrode model developed by SINTEF and ERAMET Norway (in NextGenSøderberg project). To run this model, the operators/metallurgists need to convert several operational parameters into the bespoke input format required by the model – which requires extensive training.
The expert system integrates existing model documentation, simulation tools, and operational guidelines into a unified interface. Operators can ask questions like “How can I give the inputs for the heat transfer around current clamps?” and receive responses grounded in the knowledge sources. In addition, the operators can use the framework to configure and run simulation via prompts like “run simulation” and visualise simulation results.
This demonstrates how language model-based expert systems can support operational decision-making without requiring users to become simulation experts themselves.
The solution is designed to run securely on-premises, ensuring full data confidentiality while enabling integration into existing workflows.
Business impact
- Reduces training time for using simulation tools
- Accelerates decision-making with simulation-backed insights
- Improves efficiency by integrating simulations into daily workflows
- Ensures data confidentiality through secure on-premises deployment
Benefits
- Simplifies the setup and execution of computational simulations.
- Enhances decision support for industrial operations.
- Improves operational efficiency and process understanding.
More information
Top image was created using Microsoft Copilot and extended with Adobe Firefly.
