Is it possible to have multiple material types in a model? I would like to model the cooldown time of a pipe filled with hot water as is cools in a constant outside skin temperature.

Administrator

This post was updated on .
Yes, just make a model with multiple subdomains and enter different material parameters in each subdomain. See for example the thermal shrinkfitting tutorial model.

Ok, I have done that, how do I do a transient analysis, or is that not possible with the free version?

Administrator

This post was updated on .
If you follow the linked tutorial it is a timedependent/transient example.
https://www.featool.com/doc/heat_transfer_03_heat_transfer3#tut_ht03 Step 76. Now that the problem is specified, press the Solve mode button to switch to solve mode. Since this is a time dependent study, open the solver settings and select the TimeDependent solver. Set the Time step to 0.25, Simulation time to 16, Time stopping criteria to 0, then press Solve to start the solution process.

Thank you. I think I am getting the hang of this a bit more now.
I have managed to run a model now, and I am getting values that appear to be similar to what I expect. I would like to know how i determine the Overall Heat Transfer Coefficient for the inside or outside of the pipe. How do I perform that calculation/analysis in post processing? Best regards Andrew Lindsay 
Administrator

This post was updated on .
That depends how you define "overall heat transfer coefficient" in your particular model. FEATool basically allows you to evaluate any general expressions of coefficients, in point, boundary and subdomain integrals (which can be found under the *Post* menu).
I would recommend that you work though a few of the heat transfer examples, which all illustrate how you can evaluate and postprocess expressions https://www.featool.com/doc/multiphysics_04_natural_convection1#tut_mph04 https://www.featool.com/doc/multiphysics_00_heat_exchanger1#tut_mph00 https://www.featool.com/doc/heat_transfer_02_heat_transfer2#tut_ht02

OK,
I am still looking into the overall heat transfer question, but I have another question regarding setting the initial conditions. Say I have three blocks of different materials, say they are blocks 1 2 and 3 from left to right. All sides of the assembly are insulated, with the exception of one face on the 3rd block. The first block is set to Temperature T1, and the Temperature, boundary condition for the uninsulated face of block three is set to T3. When I set up the equations for all the blocks, I set a temperature for Block 1 or T1, but what temperature do I set for Blocks 2 and 3 initially? If I leave it as 0, then the model assumes that this is a temperature of 0 Kelvin and the results are meaningless for a simple heat transfer example. Any assistance would be appreciated 
Administrator

This post was updated on .
If you have a static/stationary (linear) problem you can leave the initial conditions as zero because the solution will not depend on them (only the convergence for nonlinear problems depend on how close the initial conditions are to the final solution).
On the other hand, for instationary/timedependent problame you have to set initial conditions for all blocks/subdomains (even if they are completely isolated) as the future state depends on what was before (a different solution will ensue with different initial conditions (if the problem doesn't converge towards a fully stationary state)). An inverse problem could technically be posed, that is for what initial conditions do we get a given solution. This is beyond scope of the basic functionality but something along the line how to perform parameter minimization could probably be done.

OK, So I have had a little more time to examine my problem. I have got some results, but they are not consistent with what I would expect.
I get a region within my insulation that drops to 0, even though I have chosen starting values of 373K. FJ.fes Any ideas on how to correct this would be appreciated. 
Administrator

This post was updated on .
It seems you have loaded/opened the file "FieldJoint3.fea" while working on your model, and as ".fes" script files (which are simple text files with a log of the GUI commands perfomed) do not include binary attachments, or mesh/solution etc (it just points to the location of them) it is not possible to fully examine the model.
Could you attach the corresponding binary ".fea" file, or point to an online download location if it is too big to attach (such as Dropbox or Google Drive)?

Administrator

This post was updated on .
Dear Andrew,
I think this could possibly be a misunderstanding of how boundary and initial conditions work and the difference between them. Initial conditions (in your case T0 in the various subdomains) which are specified in the "Subdomain Settings" dialog box, sets the starting values or initial state of your model. The time dependent solver will use these values to start with and compute subsequent solutions (in time so that solution at u^(n+1) is a function of f(u^(n)), while the steady state solver just uses this an initial guess. So in general, different initial conditions will result in different solutions (unless your model converges to the same steadystate in time). In contrast, boundary conditions specify how the model interacts with the rest of the "world". For example, in the attached model the internal boundary (22) was prescribed with a "Temperature" value of 0 which fixes the temperature to zero at this boundary for all time (it does not mean no temperature, or insulation). The result in this case will be a zero temperature at this boundary that is increasing (or decreasing) to match the other prescribed temperatures. As the domains all are connected and solution continous there is no easy way to prescribe a "true" internal insulation condition. Instead if you really want to isolate two subdomains, you can make a thin subdomain (1) and simply remove it ( subtract) from the rest of the geometry (2) deactivate it (deselect the "active" button for the corresponding subdomain in the "Subdomain Settings" dialog box), in this way temperature will not flow through (3) set a really low thermal conductivity which essentially acts as a block, or vert thermally opaque temperature barrier I hope this somewhat clarifies things a little.

OK, I think I have found what was going wrong. For some reason, one of the domains (Boundary 16), was set as temperature 0, though when I had looked previously it had been a continuity boundary condition. I had gone through the model a number of times and had not been able to see where I had set that as a fixed temperature and not a continuity. But that is fixed now.
I have also (hopefully correctly), figured that I can get my starting position, by first setting Boundary 22 to the 373.15K temperature using a temperature boundary condition. Then, I run a steady state solution to determine my starting condition. I then set the boundary condition for Boundary 22, to be a continuity boundary condition and do my time dependent solution using the last computed solution as my initial condition. Now, I would love to be able to produce an animation of the temperature evolution with time of boundary 22, is there a way to automatically extract this information into a file, without needing Mathcad? Or is it going to be a horribly manual, labour intensive task? Best regards Andrew 
Administrator

This post was updated on .
Dear Andrew,
No, there are currently not any non horribly manual nor non labour intensive ways to generate animations without exporting to Matlab. Typically you would first select "Export FEA Struct To MATLAB" from the "File" menu to export your 'fea' data, and then use use a combination of the postplot and/or evalexpr functions in a loop to plot your data, combined with something like "print djpeg my_imagename.jpg" to export the images. An advanced example how to generate movies from FEA simulation data can be found in this post on advanced Matlab command line processing and visualization.

In reply to this post by Andrew
maybe but if you want to use a multibody mesh with boundary layer "like openfoam with more than one subdomains" its not possible I think. But in FEA with Fenics and no boundary layer "and only tetrahedral mesh" its maybe possible: https://i.imgur.com/FuCFpwI.png Best regards CFD2D
Marten Mehring
Madrigal Industrial Solutions UG 
Free forum by Nabble  Edit this page 