Can't obtain correct temperature contour in solids in Fluent

NinoNino United Kingdom Member Posts: 3

Hello guys,

 

I am trying to study the cooling of air that is flowing inside a steel pipe that sits in ambient air. I am examining the 2d-axisymmetric case.

 

I have modeled the 3 different domains (air-environment domain, steel pipe domain, air flowing domain) and have created a part of all three domains. 

I have a great structured mesh.

 

I have set the temperature of the air-environment at 273 K through the Fixed Values option in the Domains tab. The air is still and not moving so I have not defined any inlets/outlets.

I have set the steel pipe domain as a solid.

I have defined a velocity inlet (2 m/s and 330 K) and pressure outlet (0 pascals, 300 K backflow temperature) for the domain of the air flowing inside the pipe

 

I get a wall and shadow wall for the outer surface of the pipe, which is set as Coupled. I have the same for the inner surface of the pipe.

 

I am running the simulation and I am getting a result but when I get a temperature contour of a cross-section of the whole pipe, I get a specific temperature for the whole surface of the pipe instead of a linear temperature gradient that drops from the ambient temperature at the outter surface of the pipe to the temperature of the flow at the inner surface. I am supposed to get a regular conduction temperature profile inside the wall but instead I get a specific temperature throughout it's surface.

 

You can see the results here :

 

Does anybody know the issue here? Any help is greatly appreciated since this is for my bachelor thesis.

I get expected profiles for the flow and the ambient air since the temperature is decreasing and increasing respectively, but I am getting a weird contour for the pipe.

Nino

 

Does anybody know what's up with this? I am trying to get the temperature profile of the steel pipe without doing an FSI with Fluent and Static Thermal.

 

I even tested a regular wall with two air domains on each side, one hot and one cold, and I still don't get a conduction profile, just a specific value of the temperature (a specific color in the contour). I have created and meshed the solid geometry of the wall.

 

Thanks!

Comments

  • RobRob UKPosts: 11,730Forum Coordinator
    edited January 2020

    The temperature looks to change in the hot section along the length of the domain.  Can you post an image of the mesh?  Good quality cells don't necessarily mean you have a good mesh. 

    Check the heat flux balance from stationary fluid to solid to moving fluid, do these balance? 

    Have you resolved the near wall flow on the moving fluid side of the mesh?  

    Finally, create a surface on the solid zone (Create Surface in the Results section) and see if there is a subtle temperature gradient that's masked by the range. You may want to turn off node values to show the cell temperature rather than the smoothed contours too. 

  • NinoNino United KingdomPosts: 10Member
    edited January 2020

    Hello rwoolhou. Thanks for your comment, I have been trying to get help with this for some time but no one seems to know.

     

    Here is my mesh. It's a mapped face mesh with edge sizings. Even if I had a 3D case, I would mesh the pipe in this pattern as well by using the hexa method with edge sizings.

     

    Here is the temperature contour on the results tab. I don't see a change in the radial direction.

     

    I calculated the wall heat flux at the plots tab of the results section throughout the axial direction of the pipe. The innerint and innerint-shadow is the inner surface of the pipe and the outterint and outterint-shadow is the outter surafce of the pipe. As you can see the heat flux is not the same for the inner and outter surfaces but neither are the surface areas. Below I have plotted the total heat amounts plots and they are quite similar.

     

    What do you mean by resolving the near-wall flow on the moving fluid side?

     

    Here are the plots for the heat transfer coefficient as well. The inner one is very close to my theoretical value but the outter one seems quite wrong since you can'y really have a 200 W/m2 K value in ambient air without any flow. For ambient air I have a range between 1-30 W/m2K. This high value justifies the different results that I am getting from Fluent. If I get a 200 value in my theoretical model then the results are pretty much identical, but they seem wrong.     

     

    However this doesn't justify the temperature gradient in the wall

     

    The thing is that I tried a very simple model where I got 1 hot domain and 1 cold domain across a wall domain. I kept the flows still without any velocities, inlets or outlets. I defined the domain temperatures from the Fixed values domain tab and I got a similar result. A small temperature gradient in the hot and cold domains and a specific temperature across the wall instead of a linear temperature gradient (conduction-like).

     

     

    Nino

  • RobRob UKPosts: 11,730Forum Coordinator
    edited January 2020

    If you're using the FIX functions you can get some odd results. If there's no motion in a fluid region just model it as a solid and set the temperature using FIX, or just ignore it and use a wall condition. Then re-run the model. 

    Also try refining the solid zone mesh: think about the physics behind the heat transfer. How conductive is the solid?

     

  • NinoNino United KingdomPosts: 10Member
    edited January 2020

    When you say FIX functions, do you mean UDF's (user defined functions)? I have never hear of FIX functions. Could you please elaborate on this?

    If I model it as a solid, then I will have a conductive thermal resistance instead of a convective thermal resistance so I think I am going to get wrong results.

    I tried refining the mesh in the solid pipe domain and I still face the same issue. I think that unless you have a temperature boundary condition on each side of the wall, you can't get the expected temperature gradient in the pipe.

    I could simplify the model by using a convective wall boundary condition and skip the ambient air domain altogether but then I am not including the influence of the ambient air temperature drop around the pipe on the heat transfer.

     

     

  • RobRob UKPosts: 11,730Forum Coordinator
    edited January 2020

    FIX is in the cell zone conditions: it's a way to set a value that's retained in the solver. 

    If you model a fluid as a solid then, yes you lose the convective component. However, if you have no flow boundary and constant density where does convection come into it? 

  • NinoNino United KingdomPosts: 10Member
    edited January 2020

    Yes I understand what you mean. I have been using the Fixed values to define the temperature of the ambient air domain all this time, so it's not the solution.

    I guess I will have to stick with a convective wall boundary condition.

     

    Thank you rwoolhou. If anyone has any solutions, I would really appreciate it.

     

    Nino

  • NinoNino United KingdomPosts: 10Member
    edited January 2020

    Hello rwoolhou,

     

    I occured another issue with the simulation. Can you check out the last post I did?

    In short I have slpit the outer surface of the pipe into two surfaces and have applied two constant temperature, wall boundary conditions.

    When I see the results, the temperature on both these surfaces is not steady throughout the whole entirety. 

    Do you happen to know the issue? Not a lot of people in the forum know about CFD.

    Thanks,

    Nino

  • RobRob UKPosts: 11,730Forum Coordinator
    edited January 2020

    Please can you post a plot of temperature with node values off? Ie the cell values. 

  • vogtludwvogtludw Posts: 8Member

    I think i ran into a similar problem as Nino. In my case, a solid is surrounded by two fluid regions (cooling channel) and for the solid i would expect a temperature gradient in horizontal direction (moving direction) and in vertical direction. In vertical direction the temperature profil should be symmetric. Right now i´m getting a gradient in horizontal direction but not in vertical direction, so the temperature in a cross section is uniform but should be lower at the boundary and higher in the middle of the solid.

    I checked the temperature in a cross section to get a more detailed look at the temperatures but the temperature looks uniform.

    The heat flux from solid-fluid and fluid-solid balances each other out and i have inserted boundary layers to resolve the near wall flow in the fluid region. I have tried different mesh sizes for the solid region but till now i haven´t been able to get the correct temperature profil along the cross section. Does somebody have experience with a similar problem or any notes on how to solve this?

  • RobRob UKPosts: 11,730Forum Coordinator

    You've got heat transfer so all the boundaries are linked. What fluid are you using and what is the velocity?

  • vogtludwvogtludw Posts: 8Member

    I´m using the default air from the fluent database and I tried steel and aluminium as the solids but neither showed a temperature gradient. The solid is moving pretty slow, so i set the inlet velocity of the fluid to v = 0.2 m/s. I was also wondering if the temperature at the end is correct? From my point of view it looks like the temperature gradient is too steep in horizontal direction because i would suspect a higher temperature with our parameters. I was running the simulation for steel and got a temperature of ~600K at the end of the cooling section. The cooling section is 2m long, so the cooling time is 10s and to get a temperature drop of ~270 K seems not realistic.


    Because of that i was wondering if I choose the correct setup? The simulation is for a cooling channel of a solid which is continuously moving and for this i´m running a steady simulation. The inlet Temperature of the solid is realized with a wall boundary and a fixed starting temperature. I´m assigning the relative speed between the air and solid to the air instead of a moving reference frame to simplify the simulation (I know to get precise results both regions need their actual velocity). The results look like the temperature is only conducting from the inlet wall to the end of the solid and not like the solid is continuously cooled down (Even if i initialize the whole solid region with the temperature of the inlet). I tried using a material (wood) with an extreme small thermal conductivity to verify this and in this case the solid has reached the same temperature as the fluid at the midway point of the cooling channel and that shouldn´t be possible.

  • vogtludwvogtludw Posts: 8Member

    I think i ran into a similar problem as Nino. In my case, a solid is surrounded by two fluid regions (cooling channel) and for the solid i would expect a temperature gradient in horizontal direction (moving direction) and in vertical direction. In vertical direction the temperature profil should be symmetric. Right now i´m getting a gradient in horizontal direction but not in vertical direction, so the temperature in a cross section is uniform but should be lower at the boundary and higher in the middle of the solid.

    I checked the temperature in a cross section to get a more detailed look at the temperatures but the temperature looks uniform.

    The heat flux from solid-fluid and fluid-solid balances each other out and i have inserted boundary layers to resolve the near wall flow in the fluid region. I have tried different mesh sizes for the solid region but till now i haven´t been able to get the gradient along the cross section. Does somebody have experience with a similar problem or any notes on how to solve this?

  • RobRob UKPosts: 11,730Forum Coordinator
    edited November 22

    Zoom into the boundary region between fluid & solid and replot the temperature with node values off and make sure the mesh is displayed so we can see the boundary between the fluid & solid. Also please post a screen grab of the walls bounding the fluid at that position.

    To add, I think you've both fallen foul of the spam filters, in future if a post doesn't appear save the text somewhere and wait, one of us will clean the queue when we get on line. We're in different time zones so that may take a few hours, or days if it's the weekend.

  • vogtludwvogtludw Posts: 8Member

    Yeah sorry about the mishap. I made a cut after a few cms (~10) and zoomed into the top boundary. In a cut in the middle of the cooling section the colours are blurred and its even more difficult to identify a temperature gradient. Right now the size of the first inflation layer (solid) is 1mm which i suspect should be fine enough.

    I also plotted the temperature to this position in a plot and it looks like its curved just a little bit along the cross section (node values are turned off for both pictures).


  • RobRob UKPosts: 11,730Forum Coordinator

    You've got heat passing into the solid, but with the mesh resolution you're not seeing much of a gradient. Replot just on the fluid and separately solid zone and alter the temperature range.

  • vogtludwvogtludw Posts: 8Member

    The first on is from the fluid (bottom boundary):

    And the second one is from the solid:

    I think in the solid there is a very little temperature gradient visible, but i would suspect in reality it is more extreme than in the simulation.

  • RobRob UKPosts: 11,730Forum Coordinator
    edited November 25

    The top image makes sense if there's little/no motion. What is causing the fluid to move?

  • vogtludwvogtludw Posts: 8Member

    The inlet velocity of the fluid is set at the speed of the moving solid v_fluid = 0.2 m/s and the solid is given no movement v_solid = 0 m/s, so the movement is caused from the inlet velocity. In the bottom picture, a temperature drop of ~15K (moving direction) occurs for a really short cooling distance ~5cm. Do you think that possible or realistic with our parameters?

  • RobRob UKPosts: 11,730Forum Coordinator

    Increase the mesh resolution and see what you get. You can use adaption in Fluent to avoid remeshing.

Sign In or Register to comment.