CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   natural convection (http://www.cfd-online.com/Forums/fluent/92000-natural-convection.html)

 ravisalgar August 30, 2011 01:40

natural convection

hi frienz!!!
m trying to model a natural convection problem and need some help regarding what boundary conditions i need to give in FLUENT.

Problem Definition:

A square chamber containing fluid(air initially @ atm. conditions) with a small opening at the top and a heat source( exactly opposite to the openinig but at the bottom maintained at some temp higher than fluid),rest all the walls are adiabatic.

kindly help me out...

waiting for rply..

 jlefevre76 August 31, 2011 18:01

If you expect flow to be mainly coming in from a surface, use pressure inlet. If you expect flow to be mainly leaving a surface, use pressure outlet. Pressures should be set to ambient pressure I think. I was a little confused though, is it free convection from a surface or is it free convection inside a cavity, or a little bit of each?

 ravisalgar August 31, 2011 23:56

thanks for rplying jlefevre76

Sir it is a natural convection inside the cavity but net mass flow rate will b zero i mean due to density difference as hot air rises a very small amount of it will flow out through the outlet vent but in return the same amount of outside air will enter the cavity and the circulation continues.
for your reference i m defining below my outlet conditions for this particular problem plz look through and suggest me which boundary condition in fluent fit for these conditions.

@ outlet : a) dT/dy = 0
b)x velocity component u = 0
c)y velovity component v not equal to zero.

 jlefevre76 September 1, 2011 00:38

So, you'll need some vents in the sides or somewhere to bring mass back into the cavity. If you expect mass to be coming into a vent, use a pressure inlet BC. If you expect mass to be leaving a vent, use pressure outlet. Everything else will be a wall boundary condition and you'll specify the thermal properties of each wall (adiabatic, isothermal, etc).

A pressure outlet boundary condition, as I understand it, will give you dT/dx = 0, dT/dy = 0, and du/dy = 0, dv/dy = 0 (in other words, nodes on either side of the boundary will have equal velocities and temperatures).

The pressure inlets will try to keep the pressure in the cavity at whatever you set them to. I would set the gauge pressures at the inlets and outlets to 0, and just make sure the operating pressure for the volume is correct (define, operating conditions). Or maybe make sure the reference value pressure is what you want (I have to be honest, I'm not sure which one fluent references for the gauge pressure boundary conditions). I hope that at least gives you a starting point.

 ravisalgar September 1, 2011 08:54

well thank u sir..
but with pressure outlet condition i m not able to give v component of velocity not equal to zero,so do u feel should i go for writing an UDF for that particular boundary.

 jlefevre76 September 1, 2011 15:58

I'm reading up on flow boundary conditions now in the fluent user guide, and maybe I gave you some bad information. zero gradients are for outflow boundary condition. If I find another setup makes more sense, I'll let you know as soon as I do.
)

 jlefevre76 September 1, 2011 19:14

Okay, so at least for my problem, pressure inlet where I expected the most flow in, and pressure outlet for where I expected the most flow out, was the best option. However, it didn't work until I defined the direction of the expected flow. You can define it in cartesian coordinates or cylindrical. Mine needed cylindrical, and that was my problem.
Like you said though, if your boundary condition had an extra specification, you'll need to use a udf. Start looking under the examples in the udf manual under DEFINE_PROFILE (or it might be listed as a boundary condition example, I'm not sure). It's chapter 8 of the udf if I remember right.

 ravisalgar September 1, 2011 23:59

thank u very much ur information was worth noting...so now m thinking to go for writing UDF for that particular boundary...

:rolleyes::rolleyes::rolleyes:..

 jlefevre76 September 6, 2011 22:23

Oh, the command DEFINE_WALL_FUNCTION might be most appropriate for what you're trying to do, not DEFINE_PROFILE

I'm in the middle of trying to write one myself, and it's a bit harder than I initially thought, but if you're doing something simple, Fluent UDF manual gives some good example code.
https://www.sharcnet.ca/Software/Flu...f/main_pre.htm

 Rahul123 December 26, 2012 07:24

Can you plz help in writing a boundary udf for du/dy=0
The problem is I cannot understand how to differentiate between du/dy=0 and du/dx=0 in UDF.
Thank you

 jlefevre76 December 26, 2012 14:32

Ok, so, at any wall, the default for Fluent is for du/dx and du/dy to be equal to zero (that is the no slip boundary condition). It sounds like you're referring to slug flow? Or are you talking about inviscid flow?

So, here are some options that may help you (or maybe not).
Fluent has a moving wall boundary condition. It is still no slip, but you can match the flow velocity and wall velocity to get the same effect. Perhaps easier would be to change the shear condition (select boundary conditions, then the boundary you want to edit, then edit, then in the shear condition box, change to either specified shear or marangoni stress). If you were hoping for a boundary with zero slip, I think you could simulate that by using specified shear and setting the one you want to zero (which I think would be both maybe? I'm not sure there).

If you were wanting to just do inviscid flow, that's easy to change. Just go to models, viscous, hit edit, and change the viscosity model to inviscid.

Default, like I said, is du/dx = 0, du/dy = 0, and that is the no slip boundary condition.

I hope that helps somewhat. :D

 jlefevre76 December 26, 2012 14:45

If for some reason you do need to create a UDF, here is the way to do so: https://www.sharcnet.ca/Software/Flu...df/node231.htm

 All times are GMT -4. The time now is 13:59.