Questions to thermo-mechanical element (fld3d1tm)

[JStorm]

The calculation of heat capacity contribution is separated to thtrans3d. In thtrans3d quadrature and shape functions are determined again in order to calculate the integration point volume. I have not seen that a different quadrature or similar is used.

Is there a reason that heat capacity is separated and integration scheme is determined again?
I want to modify this part of the element and bring this few lines into main element subroutine, because this term depends on the constitutive in my material routine (instead of d(4) and d(64) only). But I worry that I have overlooked something.

[Prof. R.L. Taylor]

The only reason to split the transient terms is for use of elements like a 4-node tet ( or maybe any tet) where it is possible to fully integrate the other terms with a 1-point quadrature (or lower order) than that needed to integrate the heat capacity (or mass) array to obtain full rank of a consistent formulation.  If you are using hex elements then this type of split is not necessary. 

The elements in the couple3d are probably not fully developed as there are many possibilities for an implementation.  Be sure to check each step in your development to make sure the terms are present and that any linearization is fully implemented.

[JStorm]

Thank you Prof. Taylor, the element is working fine for my purposes. I have checked it and all I need is fully implemented. Now, I want to use the transient terms as well.

[Prof. R.L. Taylor]

In looking at the routine, it is quite old.  I would recomment you look at the iner3d.f routine in solid3d.f.  Also, I note that the tangent seems to be missing the ctan(2) to pick up the effect of the time integrator you use.  So check carefully.

[Prof. R.L. Taylor]

Also, one cannot compute the r(4,ii) as shown as the s(i1,j1) will have other terms in it from the thermal gradient before the call.  It is obvious the routine has not been used!

[JStorm]

Also, I note that the tangent seems to be missing the ctan(2) to pick up the effect of the time integrator you use.

For the transient heat conduction, ctan(2) is applied in thtrans3d at line 134 (FEAP 8.4).

Also, one cannot compute the r(4,ii) as shown as the s(i1,j1) will have other terms in it from the thermal gradient before the call.

As far as I see, the DOF 4 components in s are still zero, when it comes to thtrans3d. Before calling thtrans3d in fld3d1tm, only mechanical inertia and geometric tangent terms are added to DOF 1-3 of s.

I have created a 3D version of Ex6 (Ithmech_2d) with the 3D temperature element (see Ithmech_3d). Afterwards, I have applied the fld3d1tm element and it seems to give same results for steady state and transient case like with the 2D and 3D thermal element.

[Prof. R.L. Taylor]

Sorry for the misinformation, I saw it later.

Can you post the Ithmech_3d also

I think you can use the command: PARAmeter dt = dt*10

instead of the FUNC tinp, but I did not test in ver84;
or what ever you have in the function (saves having to post extra things).

[JStorm]

I think you can use the command: PARAmeter dt = dt*10

instead of the FUNC tinp, but I did not test in ver84;
or what ever you have in the function (saves having to post extra things).

tinp.fcn was not delivered with the FEAP 8.4 examples, but now i saw that it comes with 8.5.

Please, find attached my test environment for the thermal part of the thermo-mechanical element (fld3d1tm) ...

[JStorm]

... and the material interface.

I have modified umodelfc.f in order to have a generic material interface for coupled materials (umati*.f, umatl*c.f), like FEAP users are used for mechanical materials.