hi evry body. i am modeling free convection in a room with dimensions 6X4X3 meters with fluent code 6.3 i want to know how fine should be my mesh to achive true results i am interested in velocity and temperature contours(i will be thankful if u tell me the interval size i should use in gambit). i should mention my medum is air, i use ideal gas model for density, presto model and the second guestin: if the number of cells is very much my mashine will get very slow what solution u suggest?

mesh fine near the boundry walls and any other critical area of interest. rest could be courser.

i know that but would u please help me with intervat size that should be used for such room?

intervel size u can check out in gambit.its not possible to tell. make fine mesh near all the walls and coarser in the core. try to use bousinesq law for density instead of ideal gas. check the Raleiygh no.

boussinesq or ideal gas for density - check Rayleigh as mentioned. Use PRESTO or Body force weighted for pressure disc. Use Gravity on for pressure field. May need to adjust operating density. Refined grid to capture plumes. Can adapt.

the rayleigh is about 10^11 so the flow is turbulant. the medium is partially an ideal gas(the medium is air) so i have used Ideal gas for density and presto for pressure. the pressure velocity coupling is simple and the number of cells used is 40000(for 72 m^3). with these settings i get very high temperature gradiants near hot surface (which is the room floor) and cold surface(which is ceiling) which dosent seem true and also no eddy is seen in the flow field?

I am modeling natural convection in a room. but i do not know what i boundary condition use for open window to environment. Is it true out/inlet pressure? thanks.

you can use pressure outlet if the window is far enough that the gradient of temperature is zero.

sajad! beacuse you model the room, your fine mesh must be small enough that you can get boundary condition near walls, so you should use boundary layer option in Gambit to fine your mesh. your mesh should be the order of mm to get boundary layer with respect to yout domain that is in meter. it is better to use Boussinesq App. instead of using ideal gas. beacuse in Ideal gas the operating pressure is very important and your answer is depended to amo

sajad! beacuse you model the room, your mesh must be small enough that you can get boundary layer near walls, so you should use boundary layer option in Gambit to fine your mesh. your mesh should be the order of mm to get boundary layer with respect to yout domain that is in meter. it is better to use Boussinesq App. instead of using ideal gas. beacuse in Ideal gas the operating pressure is very important and your answer is depended to amout of it.

but the temperature contour that you said is not true physicaly, are you sure you set boundary condition correctly?

dear mahyar. you mean I use successive ratio while i am creating mesh in gambit to fine my mesh or to use "create boundary layer" which has a separate icon in gambit? about the size of my mesh if it was in the order of mm then the total number of cells would add up to millions and my machine cannt run the project what you suggest? about the ideal gas do you say that Boussinesq App is better even for air? and if this is so then what i should use for Boussinesq parameters in operating conditions window?

about the temperature contours i am sure about the correctness of boundary layers. there is a severe temperature gradient near the the floor and ceiling which are the hat and cold surfaces respectively

thanks from your answer. but what Boussinesq parameters i use in operating conditions window? i use of density based or pressure based?

i know that the amount of mesh will be more that your machine can solve. bu you do not have any choise if you want to get answr exactly, but the largest mesh in the middle of your room could be the order of 5 to 10 cm. you can use boundary layer in the mesh section for creating mesh and also use function for creating triangular meshes that have a succesive ratio. for more information refer to gambit help. beacuse the total difference between the maximum and minimum of temperature doesn't exceed than 30, you can use boussinesq.yes i know that the boussinesq parameter can be set just one parameter, but it depend on your physical problem. if you are from Iran, you can call me with 0917-313-2176 for geting more information.

Dear Yasuri i must have more information about your physical domain to answer your question.

the reason of this may be beacuse of following. 1- your boundary condition set wrong. 2- the amount of mesh near wall is not enough.

dear mahyar yes i am from Iran and i will bother you as soon az possible on your cellphone...

dear mahyar yes i am from Iran and i will bother you as soon as possible on your cellphone...

thanks from your answer. my problem is very simle. It,s a building that it is heated by radiation, there is a difference temperature between into and out, i wnat to model natural convection in atria by temperature difference. i don,t know what i put for operation condition, or use density based or pressure based, coupling pressure condition...?

Simulation of natural convection from a heated element in an enclosed chamber. The enclosed chamber is a vertical 3D cylinder and a horizontal cylindrical 3D element is suspended in the chamber at the middle which is heated with a thermocouple. Please list down the boundary and the operating conditions...My solution is not getting converged. I am using the radiation model as well.