Conduction Problem

[p282h]

Hi:

I have been having some conduction issues in my simulation. I have a part with solid-1 extruded, and another solid-2 extrude on the top of the former. I partitioned the two solids by cutting plane and partitioned the solid-2 into two cells using cutting plane. I have not applied any interaction, BCs, loads but applied only ICs to solid-1 300K, and 2000 K solid-2 and used Model change to add elements. I see no conduction into the solid-1 but something totally different is happening. Can someone suggest what to do?.

http://files.engineering.com/getfil...2c3-6973d45b57b3&file=Simulation_contours.doc

 

[corus]

The mass of 1 body is much larger than the mass of the extrusion so I'd expect the two bodies to reach some intermediate temperature, close to the temperature of the larger mass. I think that's what you see, though why it happens initially from one end only I don't know

Tata

 

[p282h]

Hi Corus:

The simulation pics I have put in the file are incomplete. In the sense that I have used model change and trying to simulate a moving heat source and that is why u see it initially happen from one end. But if this is the case how can I avoid it and let the conduction happen from hot to cold (2000 to 300 K).

Thanks,

 

[mrgoldthorpe]

There might be a *big* problem with using only one layer of elements in solid-2.

* On the boundary of 1/2 do they (i) share the same node numbers or, (ii) have separate node numbers, so you tie together the temperature dof using MPCs?

* If (i) above: which INITIAL CONDITION, TYPE=TEMPERATURE is last in the input file? This is the one which will take precedence for the boundary layer of nodes, so if that last IC refers to Solid-1 then you only have the nodes above the boundary in Solid-2 set to 2000K.

* If (ii) above, in which direction is the MPC? The node on Solid-1 should be tied to the node on Solid-2 to ensure the 3000K IC holds; not "2" tied to "1".

* In any case, there is an enormous temperature gradient across the single element height of Solid-2 - hardly the conditions for a stable solution!

* what kind of elements are you using: first or second order?

As Corus states, the thermal capacity of the two solids would appear to be enormously mis-matched, so without further heat added the final temperature of the two will be close to 300K (work it out using the relative heat capacities and volumes/masses).

As Corus suggested in the earlier LENS thread of yours (I've just looked at) to get reasonable results near the interface you need more mesh refinement and certainly more elements height-wise in Solid-2. As he also suggested in that thread, you can take advantage of two vertical planes of symmetry to cut down the model size - or do 2D trial solutions to get a feel for the problem. You'll make far more progress understanding what's going on rather than throwing everything into a large 3D model. You'll see and understand how the thermal transient progresses into the depth of solid.

I also notice in that LENS thread a very surface heat flux of 1e08!!! Presumably that's W/m^2. That's probably the reason for your strange results there. You would need either much smaller time steps or much more mesh refinement to get accurate results in the early and intermediate stages of the transient.

Now I'm not a thermal transient thermal analysis specialist, but I recall there's a recommended size of size step per increment depending on minimum mesh size and thermal conductivity at least. (Corus will know.) You must take account of that in setting your time steps, or refining your mesh.

 

[p282h]

Hi mrgoldthorpe:

As I said before, the solid-2 is extruded on top of solid-1 (both same part), I did not use any Interactions like MPC. But I am very sure that even though I have partitioned the two surface using cutting plane method, the nodes are the same. It means when I select solid-1 to give IC 300K the nodes common to both get selected too, similar is the case when I select Solid-2 to give IC 2000K. That is at the two imaginary interfaces the nodes are common and have both 300, 2000K. I did this to avoid any interactions conditions. When I had used the same strategy for only one solid, and part of it sectioned and applied ICs, and Model Change it worked perfectly fine. But one I have two solids (in one part), it works with no heat flow into the subtrate.

My question is: 1. Is my partition wrong, like using cutting plane method?. 2. Do I still have to apply MPCs to a partitioned surface?.

Thank you for your time

 

[p282h]

Hi mrgoldthorpe:

Answer to your questions:

1. I am using quad elements
2. the materials for the two solids are same. How can I work out using the relative heat capacities and volumes/masses?.
3. The last IC condition is 2000K on Solid-2.
4. I will try to add more layers(elements to solid-2), and run the analysis, to see if it works.

5. In the previous posts to which Corus replied on my questions related to LENS: I created two parts and merged in assembly module in CASE-1; and tied in interaction module in CASE-2. In both cases it didnot work. Eventually, I tried to mesh it finer at the surface in which load is applied and increased the time step but there was always the negative temperature gradient in the elements close to the top surface where load is applied.Therefore, I followed this current procedure to avoid using two parts and interactions.