Dear All,

Im currently trying to simulate heat tranfer in the annulus of a duct.. My inner duct has dimensions of 1cm radius with a thickness of 0.5 annulus(solid) and length 1cm. My outer duct has dimensions 4cm radius and length 8cm; the outer duct is united with a frustrum of length 2cm..meaning its a contracting duct geometry..The inner duct is placed in the centre of the outer duct. i have to generate some amount of volumtric heat in the annulus of the inner duct..So far i have been using default values of in the boundary conditions, exp for the velocity which is 5m/s and a corresponding pressure outflow(calculated by using Bernoulli's eqn and inlet velocity) Im quite unsure of wht kind of solver to use..whther segregated or coupled..which is the best? want to see heat distributed througout the outer duct too... Also, how do i calculate the Reynolds number for such a flow?..i want to see if i have to increase my turbulent viscosity ratio as i got the limitation warning in my first few iterations.

In segregated solver first mass and momentun equations are solved then energy equations are solved. But in coupled solver all are solved simultaneously, so this takes longer time than segregated solver.

Solver selection should depend on type of problem and Fluent manual has guidelines on this.

Reynold's number you can compute based on some characteristic dimension, eg inlet of the annulus.

Try following some of the problems given in training manual.

Hope this helps. Ashu

You mean to say i dun have to compute the Reynold's number for the whole duct?..just calc the Reynold's number for the annulus region which is filled with volumetric heat source is it? If so, could you explain to me why? Also, when i reduced the velocity to 1m/s, i seem to be getting lots of reverse flow, which may be due to the fact that the air particales are being hit back in.. but the reverse flow occurs only for about 20 iterations..then stops. I wonder if it would affect my mass balance and velocity out of the contracting duct... Plz help...

I don't agree with your statement 'so this takes longer time than segregated solver'

On iteration will take indeed longer with the coupled solver, but often the coupled solver needs less iterations to converge. So sometimes the coupled solver is faster than the segregated solver.

In some cases the segregated solver does not converge, while the coupled solver gives fast convergence.

Laika, still orbiting

So in my case should i use coupled or segregated..? very unsure of the solvers... By the way, is it alrite to get turbulent viscosity limited in 38 cells during my 2nd iteration, followed by a bit of reversed flow?Reverse flow happens for the first 20 iterations Subsequently i dun get tht though...

plz help..Thnx in advance... in a dilemma now cos report is due soon...

puck

You can run the model with Segegated solver and monitor the convergence. If the results make sense after convergence feel good. Else try using Coupled solver. If the results using both the solvers are different then there is something wrong.

Try with coarse mesh, get convergence using segregated solver. Refine mesh and solve again and check whether there is some change in results due to mesh refinement. Mesh refinement would help in convergence also.

Enjoy Ashu

Alrite..but is there a way to solve for turbulent heat transfer?(trying to see the amount of heat transfered from a hair dryer) Im using kepsilon method...can i input all my boundary conditions into and jus iterate or do i have to follow a different method for turbulent flow?

I think Fluent manual talks about what turbulent model is good for which kind of problem. But simple k-e model can be used initially and then if results are not okay go for variants for k-e model.

Bye Ashu

I did read..but dun really know wht to use.. I trying k-e now...but results arent very good.. Will tryu\ other models then.. Thnk you..

puck

Puck, 'Results are not not very good', you're telling us.

Don't blame the turbulence model to easy. The standard k-e-models are very good.

--Do you have realistic values for the inlet conditions? (I'm thinking of the turbulence at the inlet) --Is your mesh ok? What about the quality? Do you have a sufficiently fine mesh near walls? --Do you have good thermal conditions at the walls. Did you model the heat loss through your walls?

How can you tell the results are not very good? What's wrong with the results? Are you not happy with the pressure field? The velocities? The heat transfer?

please tell us more,

Laika, still orbiting

Hi,

Thnx for replying...anyways, i dun know whether u read about the geometry im drawing in my initial messages... Anyways,im trying to model a hair dryer, but since its hard to draw the wiring inside the hair dryer using Gambit, i was asked to draw a contracting duct with a solid tube(with an annulus) inside..the annulus has volumetric heat generation.i did this by drawing cylinders and splitting the volumes with two faces..then i meshed the inidividual sections with a spacing of 0.08 when i input the heat generation inside the annulus and run it, i find that there isnt much convection of heat into the surrounding air of the outside duct. Furthermore, the temperature inside the solid tube goes as high to 2000K, when i input a velocity of 5m/s.Im initialising with the velocity input... Im happy with the mass balance and pressure..but just dun understand why i dun see proper heat convection by the annulus into the surrounding air...i need to get at least 340K at the outlet of the hair dryer... Thnx alot for your reply.. Hiope you understand...

OK, now I understand a bit more of what you want to model.

--Your volumetric heat source approach is not a very good one. Just create your large duct and subtract the annulus from it. The annulus is not a part of your CFD-calculation domain.

--Give tha wall of the annulus a wall boundary type in Ganmbit. --In Fluent for the annulus-BC: leave the default thermal setting of specified heat flux, but change the value from zero to your actual value. You actual value is, of course, the total thermal power of the dryer devided by the area of the annulus. Unless you changed the units-settings, you must specify in Watts per square metres. I estimate this to be around 200000 Watts per square metres.

--check the energy-equation is turned on

--run!!

--are you sure about these 5 m/s. I think that's pretty high.

good luck, Laika, still orbiting

hi..sorry to bother ya.. but the annulus is part of the smaller duct which lies in the middle of the large duct.. Therefore, i used split volumes method.. now my problem is that, i dun get symmetrical conditions on either side of the duct..meaning the heat profile on the top section varies from the section and my professor immediately refused to accpet the results..

I was asked to reduce the convergencel limit but still dun get that accuarate symmetry.. Im thinking of doing a symmetrical plane instead in gambit... Is tht a good idea?..even when i split volumes in gambit to get the faces defined for the plane, im not sure if its a good method..cos i define a lot of faces as symmetry..hard to get one plane when the faces are split... Im really lost..cos i have been doing this for 2 mths and my results are not pleasing.. Thnk you

Puck