temperature dependent material property support?

[jaredmagnusson]

Hello,

Has a method been developed (or maybe in the works) to support temperature dependent material properties using the elemCallBack method of initializing pyTACS? I guess another alternative without elemCallBack would be supporting a MATT1 input card for a given BDF file?

[timryanb]

I think the best we have for temperature varying constitutive classes is the PhaseChangeMaterialConstitutive. This class has the capability to specify two distinct property phases that get switched between depending on the temperature of the element.
@markleader should be able to provide more detail if you have further questions on it.

[jaredmagnusson]

oh that could work, I'll look into that, thank you!

[markleader]

@jaredmagnusson there is a smoothing term that controls how quickly the material changes phases. If you don't want something as discrete as a phase change material, you could play with this to make it more linear. Happy to discuss more if you have any questions

[jaredmagnusson]

Hello Mark,
Would this be the variable returned from evalTransitionCoef? Are those functions of thicknesses specified?

If I wanted to model material property changes due to temperature, I'm thinking:

set mt to transition temperature
set lh to zero
unsure about t/tlb/tub and how those would affect the smoothing

I'm interested in using LinearThermoelasticity3D/HeatConduction3D, using SolidConstitutive. Would PhaseChangeMaterialConstitutive be available for direct substitution here?

[markleader]

@jaredmagnusson I agree with setting lh to 0 and 'mt' to where you want the transition to be centered on. If you want a smoother change of material properties, try setting dT to a larger value. The material will be in a transition phase between material 1 and material 2 when the temperature is between [mt-dT, mt+dT]. The t / tlb / tub are the material thickness and the corresponding lower/upper bounds, so these are not related to the material property transition.

PhaseChangeConstitutive is only implemented in 2D now so for now it only works with LinearThermoelasticity2D/HeatConduction2D. I would recommend starting there if possible. Adding 3D support should be doable if needed though.

[jaredmagnusson]

thank you for the clarification. We're currently using 3D, so if it really comes to it, maybe we could look into a generic TemperatureVaryingSolidConstitutive or something