Modeling anisotropic material properties in FDS

[vivashchenko]

I'm working on an FDS simulation that requires modeling materials with anisotropic properties, specifically:

• Direction-dependent thermal conductivity
• Other orientation-dependent material characteristics

From reviewing the FDS documentation, it appears most material properties are defined isotropically. However, I need to account for scenarios where thermal behavior differs significantly along different axes (e.g., composite materials, wood grain, or layered structures).

Does FDS/SMV currently support native implementation of anisotropic material properties?

If not directly supported, are there established workarounds such as:

• Leveraging SURF definitions with directional dependencies?
• Custom modifications via user-defined functions?
• Would this require implementing new subroutines in material.f90 or similar core files?

Any guidance or examples of anisotropic material implementations would be greatly appreciated.

[johodges]

I have a fork where I made this change linked below. It is a pretty simple change which adds 3 parameters/ramps for CONDUCTIVITY_X, CONDUCTIVITY_Y, CONDUCTIVITY_Z. When we discussed this before @mcgratta wasn't sure it was needed since you could get a similar effect by changing the conductivity on 1-D surfaces on difference sides of an OBST. That may work, but I think it is cleaner to have as separate parameters if it does not add too much computational overhead.

master...johodges:fds:anisotropic_conductivity

[vivashchenko]

Thank you for clarifying the anisotropic material capabilities in FDS! This is extremely helpful for our project.

A related question arises when dealing with inclined surfaces where the material’s surface axes don’t align with the global coordinate system (e.g., a diagonal timber panel with grain direction at 45° to X/Y axes).

[mcgratta]

I would like to see some validation simulations where it is demonstrated that this has a significant impact. The solid phase routines are already complicated, and we don't want to add more complexity unless there is a demonstrated impact.

[vivashchenko]

We plan to verify this implementation in our ongoing work. If significant differences are observed between isotropic and anisotropic modeling approaches (particularly for heat transfer rates), we’d be happy to share it