Can't achieve heat transfer

moranlouiemoranlouie Member
edited October 30 in Fluids

I have been dealing with this problem for some time now, and, after finally being able to create a solution with almost no errors, the simulator can't seem to create a meaningful temperature gradient:

A strange back pressure is found:

I am simulating a concentrated solar power generator in fluent:

  • pressure-based
  • steady-state
  • Discreet Ordinance Radiation model, solar radiation of 1,000 W/m^2 and 10,000 W/m^2 were tested, with the following vectors
  • Realizable k-epsilon viscous model with scalable wall functions

The cavity domain of the receiver, seen above the dish, is set to duratherm600. The walls of the domain were set to aluminum, while a wall that faces the dish was set to calcium carbonate. The tubes and it's walls, seen bellow, were set to copper, while the dish and its walls were set to aluminum. I am still unsure which absorption and scattering coefficients to use. The original literature this simulation was based on said the cavity would have an absorption coefficient of 0.9, so I put that in the duratherm600. I still can't find any source for the scattering and absorption coefficients of the other materials. I have experimented with these coefficients with little effect.

Dummy properties for duratherm600 were used:

R-12 enters a velocity inlet (@0.4 m/s and @10 m/s were tested) at 298.15 K and ejects at a pressure outlet at 0 Pa.

The calcium carbonate wall was set to semi transparent, radiation, 2 mm wall thickness, 300 K external radiation temperature, external/internal emissivity set to 1, and beam direction set to Y=1.

The aluminum walls of the cavity were set to convection, heat transfer coef. of 23 W/m^2-K, and 1.7 mm wall thickness.

Interfaces between domains of similar material (e.g. copper to copper or freon to freon) were set to matching, while interfaces between the freon, copper tubes, and cavity were set to mapped-coupled.

I used the coupled scheme at the pressure-velocity coupling, least squares cell based gradient, and second order whatever in the other settings, with pseudo transient enabled.


  • I experimented with the radiation intensity, the absorption/scattering coefficients of the materials, and values in the radiation tab in the boundary conditions. I found that, when messing with the radiation settings of the boundary conditions, anything beyond default settings causes fatal errors in the simulator, sometimes having to rebuild the simulation again. The absorption/scattering coefficients have little effect in the simulation, while increasing the radiation from 1,000 to 10,000 slightly increases temperature. I neglected to capture this, but I noticed the energy residual is the only line that failed to converged after 1,100 iterations. The simulation seems to create no fatal errors.

    The paper this is based on had conducted the same simulation on a non-Ansys program using the Monte Carlo RTE instead of DO RTE. It mentioned only the receiver absorbance, concentrator (dish) reflectance, and solar radiation intensity; as well as the geometry. So, I am wondering why I'm not getting the same results. These are the temperature gradients the paper was able to produce:

    Thank you for your time!

  • RobRob UKForum Coordinator

    Can you plot temperature on a slice through the whole domain, like the second image but the whole model.

  • The rest of the model has no domain, so here's the cavity and the dish for 10,000 W/m^2:

    It's just red the whole way with no deviations:

  • RobRob UKForum Coordinator

    You're trying to pass data from the flat screen to the collector system through a region with no mesh?

  • moranlouiemoranlouie Member
    edited October 30

    Yes. The only tutorial I could find used two geometries with no connecting domain. I'm talking about this one here:

Sign In or Register to comment.