12 Replies Latest reply on Jun 2, 2014 11:23 PM by Jared Conway

Thermal analysis execution problem

Allan Douglas Santos Silva May 17, 2014 2:16 PM

Hi There.

I'm trying to perform a transciente analysis in a system with concentric cylinders and with a heat power at the center of the system. But the software could never finish the analysis; always appears an issue, like the error mensage: Status code 9999 PCGLS_FATAL. Or something like: the Diagonal coefficient matrix is zero.And further something like: the results not converged after 1000 iterations.

Beyond that the analysis take too long time. The program takes sometimes 24 hours to calculate the view factor and after he spends hours to make one iteraction. It happens that at the interactions the solidworks fails to converge the results; most of the time he could not even pass the first iteration.

This is analysis are par of a research project; it is very important realize them in the next weeks. If anyone can help, I would greatly appreciate it.

Above is the link of a zip archive of all the parts of the model. There are several simulations in the model, from the more simplified model to the more complete.At the zip archive are atached also the curve of ambient temperature and heat power, beyond of specific heat of Aluminium.

https://www.dropbox.com/s/54ly8zcsovkm9eh/C%C3%B3pia_ensaio%20t%C3%A9rmico-zip.zip

From already thank you

PS: I'm Brazilin, so sorry for my english.

Re: Thermal analysis execution problem

Allan Douglas Santos Silva May 19, 2014 10:23 AM (in response to Allan Douglas Santos Silva)

I inserted the link because he is too many bigger to attached it.The properties of studies are already set; only just run them. I'm new in simulations like this, therefore i think that i make some mistake tha i can't figure out.
Like Show 0 Likes(0)

Actions
Re: Thermal analysis execution problem

Allan Douglas Santos Silva May 26, 2014 11:53 AM (in response to Allan Douglas Santos Silva)

If someone solve this problem, we will not have problem in including your name as contributor on our next paper
Like Show 0 Likes(0)

Actions
- Re: Thermal analysis execution problem
  
  Jared Conway May 26, 2014 9:40 PM (in response to Allan Douglas Santos Silva)
  
  Can you give a synopsis of the analysis?
  
  How does heat go in?
  How does it go out?
  What do you want to learn?
  What are you expecting?
  
  My recommendation would to start really small and simple. Make sure it works and you understand the program. Them build up to the full analysis.
  
  The error looks similar to something I've run into before where there was no way for heat to leave the system.
  
  Like Show 0 Likes(0)
  
  Actions
  - Re: Thermal analysis execution problem
    
    Allan Douglas Santos Silva May 26, 2014 11:23 PM (in response to Jared Conway)
    
    Off Course!
    
    The system consists in five concentric cylindrical shells, with a heat power at the central position . All the system is in the vacuum, thus the heat exchange happens only for radiation between the cylinders and by the heat radiated by the heat power.So the heat power radiates the heat for the most internal cylinder, and this cylinder radiates heat for the next cylinder and so on, until the external cylinder that radiates the heat for the environment.
    It's a transient analysis and we want to find out how will be the temperature distribution at the cylinders. We already have the experimental results and we need to reproduce this results through the thermal analysis from SW. The problem is that the analysis don't work.
    
    I'm sure that I'm making some mistake, but I don't know what is it.I tried a very simple case, with only one cylinder and the heat power at the center;nevertheless the analysis take more of two days and after all this time the solver fails, with a message saying that the results not converged after 1000 iterations.
    
    What do you think?
    
    Like Show 0 Likes(0)
    
    Actions
    - Re: Thermal analysis execution problem
      
      Jared Conway May 26, 2014 11:39 PM (in response to Allan Douglas Santos Silva)
      
      You have 2 of the most computationally intensive features enabled. Radiation and tranisient. Long computations are expected.
      
      Have you checked that steady state works first? It sounds like you are starting simple otherwise.
      
      Do you have initial temps applied?
      
      On the outer sphere do you have radiation to the environment enabled?
      
      If either of those aren't met, you will have trouble.
      
      Like Show 0 Likes(0)
      
      Actions
      - Re: Thermal analysis execution problem
        
        Allan Douglas Santos Silva May 26, 2014 11:53 PM (in response to Jared Conway)
        
        I will chech first if the steady state works
        
        Yes I have initial temps apllied, they are next to -187K
        the radiation in the outher and in the inner cylinder are aplied.
        
        One other information is that the solver never pass the first iteration.
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Jared Conway May 27, 2014 12:19 AM (in response to Allan Douglas Santos Silva)
        
        what file is the simulation study in and what study should we be looking at?
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Jared Conway May 27, 2014 12:41 AM (in response to Jared Conway)
        
        just guessing, i took a look at modelo final, 1clindro
        
        a few things i noticed that don't make sense to me:
        1. the surface heat generation is applied to the faces of the component in the middle. i'd recommend applying it to the body since you chose total anyways.
        2. no initial temperature applied to the cylinder in the middle, only the shell
        3. you have open system selected for the inside face and also surface to ambient
        4. to get started, i'd recommend the coarsest mesh possible and also draft qualify meshing. (i haven't really dug in but fir some reason the mesh is getting refined in areas i'm not expecting, i may also recommend building a model specifically for analysis for testing your setup. ran into some other issues with the model wtih excluded components like simulation thinking material wasn't applied..etc which also points me to suggesting a new model)
        5. global and component cotact set to insulated? really shouldn't do anything because they aren't touching but wanted to note it as strange.i couldn't edit the component contact.
        6. you have the very small faces of the cylinder selected as part of the radiation. that part is probably not significant but every element is involved in the radiation.at minimum, make it part of the surface to ambient vs the surface to surface. that way it doesn't have to do view factor calculations to all the other faces.
        7. usually symmetry isn't possible with radiation but i'd think about if there is a way to do a slice of this vs the whole model. maybe insulated faces or something similar at the cut planes. i think we came up with a solution a few years ago at hawk ridge, but its not really coming to me right now.
        8. no surface to surface radiation between the shell and the cylinder
        
        made most of the fixes described here and running a steady state
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Jared Conway May 27, 2014 12:56 AM (in response to Jared Conway)
        
        solving started after about 14 mins
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Jared Conway May 27, 2014 1:23 AM (in response to Jared Conway)
        
        with the changes, transient also got through 3 steps in about 25 mins
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Allan Douglas Santos Silva May 30, 2014 12:42 PM (in response to Jared Conway)
        
        Thanks!!!
        
        Finally The Analysis works. Your Tips helped to solve the problem.
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: Thermal analysis execution problem
        
        Jared Conway Jun 2, 2014 11:23 PM (in response to Jared Conway)
        
        ?
        
        Like Show 0 Likes(0)
        
        Actions

Go to original post