Cold sink in the element:Heat Transfer problem

[p282h]

I am trying to simulate LENS deposition process. But to start with simple I have been running a Heat transfer analysis where I have one deposit block which I removed in step-1 and added in step-2 and I am encountering a cold sink problem in the activated deposit block, although analysis completes without any errors. I have attached a detailed word document with few pictures and would be grateful if you can tell me where I am going wrong.

http://files.engineering.com/getfil...a23f-e34c86a728c3&file=ABAQUS_HT_problem.docx

 

[corus]

In severe transients you can find some instability in the solution that causes mysterious heat sinks and/or strange fluctuations in temperature. Your mesh seems coarse so you can refine the mesh close to where the severest temperatures occur, and decrease the initial time step to improve the solution.

Tata

 

[p282h]

Hi Corus:

I will run the analysis by refining mesh. But can you tell me what you mean by initial time step?.

 

[p282h]

Hi Corus

Refined the mesh:
Deposit block has 4000 elements and reduced the initial step time from 0.5 to 0.00005 and still the problem persists. Also, the step completes half the way (i.e. no conduction along the entire length of the block and of course not into substrate. If I add an additional step-3 and just carry HT it finishes with one increment only (i.e. no cooling).

http://files.engineering.com/getfil...9ee8-3f30499e2879&file=ABAQUS_HT_problem.docx

 

[corus]

You obviously have no heat transfer between the small and larger block. If you can't simply tie the two contacting surfaces together, or mesh as a single part, then chnage the value of the thermal conductance gap value to ensure conduction.
If you apply a heat flux to one surface and insulate everywhere else then temperatures will tend towards infinity over time. I think that is happening in this case.
You should also try and use symmetry to reduce the size of the problem, or even consider using a 2D mesh

Tata

 

[p282h]

Hi Corus:

I have tried to refine mesh in the regions where there is problem with the nodal temperature, tied the surfaces, gave very high gap conductance but the problem still persists. The file attached now has last two modifications made. I have no idea why this is happening. I have not used anything fancy here expect for model change.

http://files.engineering.com/getfil...a7e1-c5f28b914edc&file=ABAQUS_HT_problem.docx

 

[corus]

In your results you can see that where you have tied the surfaces together, heat is being conducted away but only at discrete positions where the nodes on the very coarse mesh connects to the refined mesh of the smaller block. In that case you'd need to refine the larger block.
Where you have apparently made a single part it appears that you can see that the temperatures aren't equal at the intersection of the 'two' parts, for some reason. The contours of NT11 should run across the 'two' parts but don't. It suggests that the single instance is a still two parts.
The instability in the solution is common in severe transients and does damp out as the solution continues through time. Try modelling it in 2D or 1D and put a very fine mesh, biased twards the point where the heat flux is applied and you should see an improvrmrnt. My guess is you'll still see some instability where, mathematically, temperatures appear to reduce below the position in which the flux is applied. I tend to ignore these discrepancies where they occur in the very initial period of the transient as they quickly disappear and the solution smoothes to a more reasonable result.

Tata