AnsweredAssumed Answered

thermal analysis-cookie on greased baking pan

Question asked by Mark Kaiser on Aug 31, 2012
Latest reply on Sep 5, 2012 by Mark Kaiser

Like • Show 0 Likes0
Comment • 18

OK, so I'm not simulating a cookie on a greased baking pan, but something similar. I have an aluminum part resting on a steel bed. The bed is fixed, the aluminum part is resting freely on the bed. Meaning the aluminum part is not bolted, snapped, or held in place in any way, shape or form on the steel bed, it just sits there. We heat the bed and part to 400 F. I would like to know how the part deforms.

I know how to import thermal studys into static studys. I know how to set temperature at zero strain. What I don't know is how to set up the contacts between the part and bed, and have something realistic happen. I also don't run simulation much, and this would be my first venture into thermal simulation.

What else do I not know, and how can I set this up?

SW 2010

No one else has this question

Outcomes

18 Replies

Richard Wehmeyer Aug 31, 2012 5:11 PM
Mark Correct
Correct Answer
It depends on just what you want to get out of it. The differance in thermal resistance will greatly depend on the surface texture of each component. If it is really rough you may need to discover the amount of thermal resistance before you procede. If the surfaces are highly polished you could use a bonded contact. This bond is not a physical bond and will not be included in the static study.
Like • Show 0 Likes0
Actions
- Mark Kaiser @ Richard Wehmeyer on Sep 4, 2012 8:08 AM
  Mark Correct
  Correct Answer
  Machined aluminum on machined steel. Not rough, but not highly polished. I'm assuming bonded would be close enough. So bonded just means heat will transfer well between the two parts?
  Like • Show 0 Likes0
  Actions
Anthony Botting Aug 31, 2012 10:12 PM
Mark Correct
Correct Answer
Hi Mark: I assume this is a transient analysis, i.e., you need the deformation field as a function of time and temperature?
Like • Show 0 Likes0
Actions
- Mark Kaiser @ Anthony Botting on Sep 4, 2012 8:10 AM
  Mark Correct
  Correct Answer
  Actually, if I had the final deformation, at the end of a time period, that would be good enough. Looking for how much a 5 foot long piece of aluminum (cross section 2.38" x 3.00" overall) will twist, or twist and bow, when heated to 400 F.
  Like • Show 0 Likes0
  Actions
  - Anthony Botting @ Mark Kaiser on Sep 4, 2012 10:08 AM
    Mark Correct
    Correct Answer
    Hi Mark: I might be "reaching", here, but it sounds like you are needing a "Steady-state" analysis. If that is the case, all components should be at thermal equilibrium (at 400F) at the end of the time period, and since the aluminum component is resting w/o further restraints, I would expect it to expand uniformly, except normal to the surface on which it rests.
    In that case you could just use a plain "static" study and set the component temperature to 400F. Install a "normal to surface" fixture on the surface that is resting (this will allow the surface to expand parallel to itself, but not normal to itself); turn-on the "soft-spring" option to stabilize the model; set the zero strain temperature, and run the static study. You should be able to plot stress and observe zero-stress field, and then plot displacement and observe deflections. Does that help?
    Like • Show 0 Likes0
    Actions
    - Mark Kaiser @ Anthony Botting on Sep 4, 2012 1:03 PM
      Mark Correct
      Correct Answer
      Yes, that helps, but I can't seem to find the normal to surface fixture. I'm running SW 2010, is it possible this was added after SW 2010? Once I do find the normal to surface option, does this option make the surface remain planar? I assume there are situations where a surface would not remain planar at the end of the study. Therefore I am wondering if setting the part on the steel plate is more realistic?
      
      I did find the soft spring option.
      
      Really need to convince my employer to get me to some training, appreciate the help.
      Like • Show 0 Likes0
      Actions
      - Anthony Botting @ Mark Kaiser on Sep 4, 2012 3:46 PM
        Mark Correct
        Correct Answer
        Yes. There should be a Fixture type named "Symmetry" or "Roller/Slider" in 2010. Both will do the same thing as "normal to surface" for a flat surface. It holds a flat surface planar, and lets it contract or expand in-plane. {side note: roller/slider can act like a roller bearing on a curved surface}.
        Holding the surface planar can be a problem if you expect the surface to lift away from its resting position. The fixture will not let it lift away. But for this problem, since the component is at thermal equilibrium, I would expect the resting surface to remain flat.
        
        In the properties of the study, I found an option "use soft spring to stabilize model" (attached image). Is there something like or similar to that in properties of study, in SW 2010?
        Like • Show 0 Likes0
        Actions
        Mark Kaiser @ Anthony Botting on Sep 4, 2012 3:59 PM
        Mark Correct
        Correct Answer
        For the fixture type, I ended up hitting the advanced tab, selected the resting surfaces, and a reference plane for direction. Then I used the translation option set at zero for normal to the resting surfaces.
        
        2010 has both roller/slider and symmetry, and has the same solver options you show.
        Like • Show 0 Likes0
        Actions
    - Mark Kaiser @ Anthony Botting on Sep 4, 2012 3:13 PM
      Mark Correct
      Correct Answer
      I may have it, does this displacment result look more realistic?
      
      Like • Show 0 Likes0
      Actions
      - Anthony Botting @ Mark Kaiser on Sep 4, 2012 3:33 PM
        Mark Correct
        Correct Answer
        Hi Mark. It does, assuming the entire thermal mass of the component is held uniformly at one temperature (i.e., not only on outer surfaces, but all throughout the volume of the component). You might do a unidirectional hand calculation estimate just to check. It should be pretty-much spot-on with the FEA results.
        Try:
        [change in length from middle section to one end]=[coefficient of thermal expansion] X [original length from middle section to one end] X [change in temperature].
        Where "change in temperature" is presumably 400 F - "zero strain" temperature. So if the "zero strain" temperature is 75F, then the change in temperature would be 325.
        
        The "change in length from middle section to one end" should be equivalent to the magnitude (absolute value of) displacement value at either end of the component, in the longitudinal direction. So, if "X" is in the longitudinal direction, just create a displacement plot of component "UX".
        The middle section should show 0 (zero), and one end will be a positive change in length in the +X direction. The opposite end should show a negative change in length since it will be expanding in the -X direction.
        Like • Show 0 Likes0
        Actions
        Mark Kaiser @ Anthony Botting on Sep 4, 2012 3:52 PM
        Mark Correct
        Correct Answer
        Yes, the hand calculations match a plot along the same axis.
        Like • Show 0 Likes0
        Actions
        Anthony Botting @ Mark Kaiser on Sep 4, 2012 4:13 PM
        Mark Correct
        Correct Answer
        That's great. "You are DONE", as Sal Khan would say.
        Like • Show 0 Likes0
        Actions
  - Jerry Steiger @ Mark Kaiser on Sep 4, 2012 9:12 PM
    Mark Correct
    Correct Answer
    Mark,
    
    If the part is going to be at a uniform temperature, why would there be any twist or bow? It seems like that would only happen if you have internal stresses in the part.
    
    Jerry Steiger
    Like • Show 0 Likes0
    Actions
    - Mark Kaiser @ Jerry Steiger on Sep 5, 2012 7:56 AM
      Mark Correct
      Correct Answer
      The steady state uniform temp study was just step 1, to get me started in thermal analysis with SW. Realistically, the part gets cycled in and out of an oven, with another part resting on it each time it goes in the oven. If I had some more experience with thermal analysis, I might be able to say confidently what was going to happen before we stick a part in the oven, but that's what I get to find out. Will be fun eventually.
      Like • Show 0 Likes0
      Actions
Mark Kaiser Sep 4, 2012 10:10 AM
Mark Correct
Correct Answer
OK, I managed to do a thermal study, input the results into a static analysis, used global bonded, set the contact faces to no penetration, and ran the study, without errors. Here's my displacement results. If I understand correctly, all the red nodes moved approx. .03" from their original position. So the part in question grew approx. .03" away from where the two parts contact, and an approx. total of .06" wider and longer. Am I understanding the displacement plot correctly?
Like • Show 0 Likes0
Actions
- Richard Wehmeyer @ Mark Kaiser on Sep 4, 2012 10:32 AM
  Mark Correct
  Correct Answer
  Sort of. Use deformation in X and Y plots as well as the deformation magnitude plot. Instead of using the contacts though, I would do as Anthony suggested. You can also set the "pan" as "excluded from analysis" also you can use the plot properties to change the deformation scale. this will show the deformations drastically and using "animate" you can see it cycle (flex).
  Like • Show 0 Likes0
  Actions
- Anthony Botting @ Mark Kaiser on Sep 4, 2012 10:36 AM
  Mark Correct
  Correct Answer
  Hi Mark: yes it sounds like you're interpreting the plot correctly. However, I am suspect of the deformation field. I would expect the unit, if it's at thermal equilibrium (steady-state), it would be 400F all-over, and thus expand uniformly everywhere. Similar to how a cookie sheet placed in an oven (resting on the grating), would expand uniformly. From the plot, it appears the contacts are "catching" on the ends of the component and the steel plate, preventing it from expanding uniformly - thus it curls. If you are running a transient analysis, then yes, I would expect the deformation field to look like the plot as shown, at some time before thermal equilibrium.
  Like • Show 0 Likes0
  Actions
  - Mark Kaiser @ Anthony Botting on Sep 4, 2012 1:04 PM
    Mark Correct
    Correct Answer
    I applied 400 F only to the top surfaces of the part, and inserted the thermal study results into the static study, for the above displacement plot. Does that help the displacement plot seem more like you would expect?
    Like • Show 0 Likes0
    Actions

thermal analysis-cookie on greased baking pan

Outcomes

Related Content

Recommended Content