Best practices

Best Practices for Working with the Full-Order Model (FOM)

The Full-Order Model (FOM) is the cornerstone of the Virtual Thermal Sensor (VTS) design process in Twin Fabrica. As the high-fidelity thermal representation of your system, it governs the accuracy of all subsequent stages, including Model Order Reduction (MOR), calibration, and simulation.

This section outlines critical best practices to maintain consistency, traceability, and performance throughout the VTS workflow.

1. Treat the FOM as the Foundation of the VTS

The Full-Order Model is not a temporary step—it is the core identity of your virtual sensor. It defines the underlying physics of your system and directly impacts every downstream operation. Any change to the physics setup—no matter how small—redefines the thermal environment and resets the workflow.

🔁 Key FOM Dependencies:

• Material definitions (e.g., conductivity, density, specific heat)
• Heat sources (e.g., distribution, intensity, location)
• Boundary conditions (e.g., Dirichlet, Neumann, convective)

⚠️ Any modification to these elements invalidates existing reductions or calibrations, requiring a full rebuild and reevaluation of the model.

Even subtle changes, such as tweaking a single parameter or switching a boundary condition type, produce a new thermal response and must be treated as a distinct configuration.

2. Maintain Versioned Project Copies

Once the physics configuration is complete and the FOM has been successfully built, it is highly recommended to save a dedicated copy of the project to preserve the model's state.

✅ Benefits of Versioning:

• Provides a stable reference for each thermal configuration
• Prevents accidental overwrites during experimentation
• Enables efficient comparative analysis between physical variants

For every significant change to the setup (e.g., new boundary condition set, adjusted heat input), create a separate project file. This keeps each version clean and lightweight, improving both performance and manageability.

3. Use Project Variants to Explore Sensor Behavior

If you're designing multiple Virtual Thermal Sensors (VTS) based on different physical setups:

• Maintain separate projects for each variant
• Perform reduction and calibration individually
• Export the resulting models for cross-comparison

Twin Fabrica’s export tools allow you to:

• Archive Reduced Order Models (ROMs) or Calibrated Models
• Benchmark sensor performance across configurations
• Validate thermal estimation robustness under varying thermal conditions

By following these practices, you ensure that every version of your model is physically consistent, traceable, and simulation-ready—making your VTS design process more reliable, repeatable, and efficient.

Would you like all best practices compiled into a single reference section? I can also help generate templates for version control or experimental documentation workflows.