Thermal analisys

gemada

Hi.
Once again I'd like to say Mecway is a very pleasant software to use.
I'm a begginer so I have some questions about thermal analisys in Mecway:
1- Thermal transient is solved by finit difference method or finit element method? Is there any kind of simplification involved in the thermal analisys solver or is it identical to other fem software like ansys?
2- I'm trying to model a concrete beam heated in 3 surfaces (all except the top). The results seem to be the same if I apply convection and radiation or if I don't apply them. I believe I am missing something here, so I'd like to have some help. I've attached 2 files: one with convection and radiation and the other one without them.
3- In order to simulate a cold room above the beam (top not exposed to fire) I'd need to consider a convection on the top surface of 9w/m2/C. Should I select all the top nodes or shoul I select all the top nodes except the first and the last because they already have a convection value of 25w/m2/C related to the surface exposed to fire?
4- In order to include steel rebars how can I bound them to the concrete?
Thanks a lot.
Best regards.

Victor

Hello gemanda.

1. It uses the finite element method as described in common textbooks. It's probably the same as ANSYS and most other thermal FEA software.

2. The reason for no change between the two models is that the temperature constraint on the three edges prevents the convection and radiation from acting. Physically, the temperature constraint means heat flows into or out of the surface as needed to maintain that temperature on the surface, so any other heat flows can't influence it.

If you want to simulate a changing ambient temperature then this is unfortunately something that Mecway doesn't yet support but I hope to add it in a future version. You might be able to approximate that with an outer layer of elements representing the gas of the fire so the heat transfer occurs only by conduction instead of convection and radiation. Or use the CalculiX solver with Mecway version 4 that's still in beta.

3. Select all the top edges, including the corner nodes. Those corner nodes will get both heat flows combined if they also have the convection/radiation from the fire.

4. If the rebars are perpendicular to the mesh then you can model them as circular areas with different material properties. Either sharing nodes with the concrete, or connected with the bonded contact under Loads Constraints.

In the attached file, I also changed emissivity from 7 to 0.7 in case that was a mistake.

gemada

Victor thanks a lot for your support.
About your answers:
2. It would be great to simulate a changing ambient temperature. I believe this would atract a lot of students like me that are working with thermal experimental analisys and need a simple tool to model some scenarios before spending a lot of time and work in the experimental program.

I'll try your suggestion of an outer layer simulating the air heating inside the electrical furnace, but isn't it supposed to work with convection and radiation? Because conduction is already happening in the previous model I sent to you (temperature applied to the beam surface)

4. Is there any place in the operation manual that explains how you draw and mesh the rebar as you did? My cad program can't export as step file.

One more question:

Is it possible to export the results just for some nodes (to draw a temperature-time graphic in excel)?

Victor

2. I'll include this in a future version since other people have asked for it too. Thanks for the suggestion.

On second thought, using conduction instead won't give accurate results and will probably be difficult to set up since you'll need to modify the thermal conductivity of the "air" so that it seems more like convection and radiation, perhaps with temperature dependence.

4. I selected 12 nodes in a square pattern and used:

Section 5.2.6 Fit to sphere/cylinder/cone
"The Mesh tools → Fit to sphere/cylinder/cone... is used to smooth
circular features that look faceted after they have been refined."

to move them into a circular shape.

To export a subset of results for graphing, use the Table with the button on bottom right of toolbar while the solution is shown. You can choose a named selection or component and export a .csv file for Excel or copy and paste.

Victor

Hello gemada

It looks like you might be able to use a time dependent ambient temperature with convection and radiation in version 4 beta. I haven't tested this feature, but it might work. You would certainly need to verify the correctness yourself.

To use it, just define the ambient temperature as a function of t for time. For example, if it was increasing quadratically, enter something like 20*t^2 where t is probably seconds. Don't use minutes anywhere in the model if you're doing this, or it might treat t as being in minutes. It will show error messages but ignore them.

You might also be able to use a table of values like you already have but you'd need to edit the .liml file by hand to insert them.

Victor

Time dependent convection and radiation work correctly in version 4 beta despite the error messages. This will be a properly supported feature in the final release of version 4.

Two example files are attached.

The convection has a very high heat transfer coefficient so the object's temperature tracks the changing ambient temperature.

The radiation alternates asymmetrically so that the heat input during the heating stages equals the heat output during the cooling stages. The two balance, leading to no overall temperature change.

gemada

Hi Victor,
Thanks a lot for your answers. It was a great help.

About convection in beta version: it works despite the error message (that the analysis doesn't support time changing). Altough it seems to work I found the results very strange. I've input the ISO 834 heating curve in seconds and made the transient analysis also in seconds but the temperatures in the concrete surface are way bigger than the ISO curve values. I'm sending my example file.

About exporting the results: what I'd like to do is to export the results "Temperature x Time" just for one node. If I copy to clipboard I'll have a giant excel table with all the nodes throw all the step time so I must select the node values manually wich is very hard to do (I don't know if there's a better way to do this?)

Best regards

Victor

In the formulas, log means log base e, so change it to
20+345*log(8*t+1)/log(10)
The next version will show this in a graph as attached, so you can more easily check it.

To output temperature and time for just one node, make a node selection containing that node before solving, then after you solve, use the table and choose that node selection. Then it will only show results for that node.

Victor

Oh, it seems like t should be in minutes for the ISO 834 formula. If so, then use this instead:

20+345*log(8*t/60+1)/log(10)

Which looks like the graph attached

gemada

Thanks a lot. I forgot to divide for 60. Now convection and radiation work as they should

About the results exporting, I tried to do what you told (select node, solve, go to temperarure results and then table tool) but after the solving the node selection disapears and the table will show the results for all nodes.

Great to see version 4 is released.

Victor

Good to hear it's OK now. The table should be able to do one node. Does it look like this picture?

gemada

That helped. Now it's working!
Next step: I'll try to model some beams and compare the results with the ones obtained with supertempcalc and ansys or other software usually used by concrete researchers.