Mechanically induced residual stresses passed to a thermal analysis

[Hurricanes]

Hi,

I was hoping someone might be able to point me in the right direction to find resources on how to do a mechanical to thermal analysis in Ansys.

Basically I want to yield and permanently deform a steel pipe in the transient mechanical solver, resulting in a solution with residual stress and strain. I then want to be able to pass this solution to the transient thermal solver coupled with another transient mechanical solver and watch as heat relieves the locked in stresses.

I cannot seem to find anything that points me in the direction of how to go about this. Everything seems to be just thermal to mechanical, not mechanical to thermal.

 

[Erik Panos Kostson]

Never done this, but perhaps one can perhaps use direct coupled field elements that have both translational degrees of freedom and a temperature formulation inbuilt in one element. Then one solves the coupled problem rather than transferring temperature data from a thermal to a structural analysis.
See for instance the SOLID226 element.

In this way one can do a plastic analysis first, and in the same model followed by a thermal step. So first load step is mechanical and is followed in a second increment/step, by a thermal load. These elements should include plasticity also, and are available in APDL. Not sure if it is possible to model this though, so it is best to ask the software developers directly and see if it is posiible (also to find out if there are any benchmarks/ case studies).

 

[mronlinetutor]

when you want to solve mechanical stress and thermal stress simutaneously, each finite element must satisfy both the stress law which is basically S(mechanical stiffness matrix)=integral ((B^T)E(B)dV and S(thermal stiffness matrix)=integral((B^T)K(B))dV. Qin must equal to Qout. Most FEA program solves them individually but the problem can be deal with by adding the Stiffness matrices of mechanical and thermal together and then solve for sigma which is stress.

disclaimer: all calculations and comments must be checked by senior engineers before they are taken to be acceptable.

 

[Mustaine3]

In Abaqus I would just run a fully coupled analysis with one or multiple steps. I guess Ansys can do something similar.