Initial temperature field for thermal expansion

Austrian87

Dear code_aster community!

I want to calculate stresses caused by thermal expansion (actually shrinkage) and to use an initial temperature field.

The question is whether it is possible to set the reference temperature of the thermal expansion coefficient by the initial temperature field or not. As fas as I know, there is only the possibility to define a single reference value for each material.

The application is a curing analysis of a thermoset.

Thank you in advance!

Austrian87

In ELAS_FO TEMP_DEF_ALPHA is a single value. But I have an initial temperature field, thus, the initial temperature field is different at each node. The assumption is that there is no thermal strain at time step zero.

Austrian87

Yes I have a temperature gradient at time step zero. The temperature field comes from a molding simulation. Thus, it is assumed that thermal expansion at time step zero is zero for the different temperatures. Or let’s say, the mesh already represents the expansion of the material at time step zero.

Can I use a pre_epsi to compensate the expansion of the first time step?

Austrian87

What I need is something like described in the Abaqus manual.

The second term in the equation after Fig. 20.1.2-2. can the pre_epsi be used for this compensation?

Austrian87

I know how to define a temperature dependent alpha that’s not the problem. The question is whether Temp_ref could be somehow a temperature field and not a simple constant or not.

To be honest, your comments are not very useful.

mf

Hello,

having a field of TEMP_REF doesn't even make sense, because coefficients of thermal expansion are not defined in that way. Having seen hundreds of CTE diagrams in my life, I have never seen one in reference to another temperature than 20 or 25°C. And also, you don't have the material data to do this, you would need CTE data for a lot of different TEMP_REFs.

And Maurizio is absolutely right: when there is a temp difference in your part, there will be thermal stresses, always. Also in your real part, these stresses will act on the part when it is ejected from the mold and distort it.

What could work is defining the temp field as initial state for the simulation (at t=0) and start with the mechanical simulation at a later t=1 (or so). When no stresses are given in the ETAT_INIT, they are assumed = 0. Then, only the differences from the ETAT_INIT act on your part. There is for sure also a possibility to set stresses to = 0 at a certain point in time of a simulation,

Mario.

Austrian87

Thank you, Mario!

I agree, that it won’t make sense for a solid part. However, as mentioned in the first post, I try to implement a curing process in code_aster.

I will try to set the initial state for the temperature by using ETAT_INIT. I did not know that it can be used for the temperature.

mf

Yes, in a thermal simulation it is possible. So, in that case, you have to do both, mechanical and thermal simulation and project the temps onto the mechanical mesh. I don't know exactly what is the best way, but it should be possible.

Mario.

Austrian87

I checked the document U4.51.03 and the temperature is not used in ETAT_INIT. Thus, I will try to generate an EVOL_NOLI with zero stresses and displacements at time 0 for ETAT_INIT and I will set the temperature field used in AFFE_MATERIAU at time 0 to the temperature gradient used at time 1 where I start the analysis. I am not sure if code_aster will then use a deltaT of 0. But I will try it.

mf

Well I said thermal simulation didn't I? That would be a THER_NON_LINE.

Austrian87

I am just wondering how ETAT_INIT in the thermal analysis affects the EVOL_THER result because it is used in AFFE_MATERIAU and not in STAT_NON_LINE. Thank you for the pointer. I understand your point. I will check it.