[steveleung]

i am currently doing my dissertation about the ventilation system. i place the heat source which define as wall with heat flux. but after adjust the relaxation factor. the solution is convergence, but the air flow still not consistent as the experiment. could anyone help me to check my case and mesh file? any suggestion for me?

[k_k]

hi,

try to refine the grid, if it is too coarse

try shifting to higher order discretization/flux types

try advancing at little bigger time steps

cheers

[ximik]

the first two make sense, but "bigger time steps?" wouldn't this increase courant number?

[steveleung]

k_k:

i had tried to refine the grid already. the max skewness is 0.76. but the model still not consist the experiment data. the air velocity in space not uniform as the mixing ventilation experiment. Should i negligible the residual values and let the model continue with interaction?

[k_k]

hi,

sorry it was smaller time step. Ok if you still have the same problem, there should be some discrepancies between the
1. Operating conditions
2.Boundary conditions and
3. Fluid properties
of what you simulating and the experiment. Check if u have specified BC and fluid properties same as that of experiment.

Cheers

[meangreen]

A skewness of 0.76 is very high. I would try remeshing with a differnt scheme to avoid this. You may need to have more cells, but you should try to keep your skewness under 0.7, and your highest aspect ratio under 1.5. Refine the mesh (make the cell sizes half as large and then compare the results).

Now, have you tried changing your model at all? The problem could be something beyond the mesh. What parts of the model are empirical, what are theoretical? Have you tried using a different type of model if available (e.g. Reynolds stress turbulence model instead of K-epsilon model)?

Have you checked your boundary conditions? Have you sensitized your solution? Changed your convergence criteria? Are you making sure you are converging on every time step? Fluent reports that if you don't converge on the first few time steps it isn't a big deal, but i have shown it makes a drastic change in the solution.

Casey

[steveleung]

the main problem is the temperature layer hadn't formed in space. the buoyancy flow did not formed in space. can anyone help me to check the file?

btw, anyone have the file to calculate the PMV and PPD in fluent. is there any tutorial link can teach me how to input the formula through custom field function calculator?? thx so much