I am modeling a Tesla turbine with Fluent.

The tesla turbine has rotating disks inside and the disks rotate by drag force of injected air.

I have problem with drag coefficient convergence: the drag coefficient fluctuates a lot.

I guess two explanations for that situation.

1. Size of the turbine is too big (>million mesh)

2. The drag coefficient can fluctuates under some selected options.

Q1. I think first explanation looks more plausible. I am wondering if second explanation is possible.

Though the drag coefficient fluctuates, pressure and velocity distribution and power output are converged.

Q2. I am wondering if I can use the power and the pressure distribution though the drag coefficient fluctuates.

I appreciate your favors.

Drag coefficient is generally difficult to predict with CFD. It is influenced greatly by the choice of turbulence model and wall modelling.

In reply to 1: I don't think the number of cells is causing the fluctuation.

In reply to 2: Try using the coupled solver. This usually helps convergence for this type of problem. http://www.ansys.com/magazine/issues...p-n-tricks.pdf

how much is the relative fluctuations of the coefficient? that is an important parameter. generally it can be because of not covering enough the viscousity effected areas with the mesh, e.g. a lower-than-needed mesh refinement near the solid boundaries. But I agree with CDE. the drag coefficinet prediction using CFD is a troublesome problem. to get better accuracy, you should have very good quality meshes near the walls.

I really, really appreciate your answers, CDE and doki.

I am now refering to CDE's information and thinking about adequacy of generated mesh of wall.

The drag coefficient fluctuates from 0 to 0.2 for lower air velocity like 30m/s; and more than -1 to 1 for higher air velocities like 100m/s.

I think air velocity really accelerate fluctuation.

-

The problem is that though the drag coefficient is not converged, pressure distribution and power output become constant after 2000 iterations.

So my partner says we may neglect drag coefficient and just use with confidence the pressure and the power values.

Q. Do you think we can use that values?

Thanks for your favor, and wish every happiness.

frictional drag estimation in numerical methods is sth directly related to the wall treatment you have in the problem; and so you might have a solution of adequately accurate flow field but not drag coefficient. thus although there are still some physical aspects of the flow not captured, in some cases you can use the calculated flow variables. because the convergence rate of different terms in the momentum eqn is not the same, i.e. some variables converge sooner than the others!

Mr.John:which software your using(meshing), i think your taken mesh elements very high, size of model,your analysing only blade or total turbine

Thanks for your interests. I am using Gambit 2.2 and fluent 6.3 for meshing and simulation.

I am simulating a total feature of a Tesla turbine, but it is simple and small in contrast to a conventional turbine.

Actually I can not understand what you're meaning well. Do you mean that one million meshes are excess to simulate?

hi What i am telling yourmesh elements is very high. but your analysing disk only know or total turbine. what size of model you taken? Example: for 800mm disk---400 hexa mesh elements enough. Give me yourmail ID I will send all paramiters.

I wanted to simulate a disk turbine, which has a diameter of 10cm. Each disk surface involve about 5000 mesh elements.

As a matter of fact, when I examined the meshes, the Gambit represented good results. So I thought "The meshing work is now enough!" (I thought I might appreciate the function)

I doubted that a outlet area where velocity changes abruptly induced convergence problem. Because it is very small part, I decided to neglect it.

hi waht ever your taken correct. each disk 5000 elements means total how many blades total turbine(total blades) how many elements. i think very high mesh elements.will reduce mesh elements size. Boudary conditions: Inlet will give: 1)nozzle angle. 2)inlet tempareture Outlet will give: outlet Pressure your taking KW model or SST model your analysing will give KW model that is better results i have one doubt your analysing only disk or total turbine How many disks,disk gap

You mentioned KW model and SST model. As I know, k-w model and shear stress transfer(SST) model are for prediction of heat transfer on transonic turbine blade and also for turbulent flow.

But what I am dealing with is about a Tesla turbine, which has laminar flow inside and a kind of multiple-disk turbines without any blades.

I made 9 disks and each disk has 3mm thickness and 2mm gap.

In real time situation the disks are held by spacers\washers circomferencially.This washers apart from holding the disk blades rigidly,also developturbulence due to which forther energy from incomming fluide is absorbed there by enhancing the power of the shaft.Heace for real time situation we have to perform CFD simulation based on turbulence model for further refined results.Refer FOR THE DETAILS TEBA JURNALS.

[sivarama1]

Hi john,
i am also now tesla turbine simulating in cfx-11,i want some guidance,please reply soon.
Thankyou very much

Originally Posted by John
;153154

I am modeling a Tesla turbine with Fluent.

The tesla turbine has rotating disks inside and the disks rotate by drag force of injected air.

I have problem with drag coefficient convergence: the drag coefficient fluctuates a lot.

I guess two explanations for that situation.

1. Size of the turbine is too big (>million mesh)

2. The drag coefficient can fluctuates under some selected options.

Q1. I think first explanation looks more plausible. I am wondering if second explanation is possible.

Though the drag coefficient fluctuates, pressure and velocity distribution and power output are converged.

Q2. I am wondering if I can use the power and the pressure distribution though the drag coefficient fluctuates.

I appreciate your favors.

[sivarama1]

HI all,any body ,
I am simulating teslaturbine any guidence and any tutorials,please send to me,
MY mail ID:sivaramakrishnaiah@live.com,malayathsiva@gmail. com,sivaramakrishnaiah_22@yahoo.co.in

[sivarama1]

Hi kiran,
i have one doubt,regarding Tesla geometry.in that how to give to CFX,i am trying this one,please any example. Tesla turbine geometry is there please send to me.

[sivarama1]

Originally Posted by kiran
;153979

In real time situation the disks are held by spacers\washers circomferencially.This washers apart from holding the disk blades rigidly,also developturbulence due to which forther energy from incomming fluide is absorbed there by enhancing the power of the shaft.Heace for real time situation we have to perform CFD simulation based on turbulence model for further refined results.Refer FOR THE DETAILS TEBA JURNALS.

HI kiran,
I am simulating Tesla turbine any guidance and any tutorials,please send to me, any Example Geometry.
MY mail ID:sivaramakrishnaiah@live.com,malayathsiva@gmail. com,sivaramakrishnaiah_22@yahoo.co.in

[sivarama1]

Originally Posted by kiran
;153800

hi waht ever your taken correct. each disk 5000 elements means total how many blades total turbine(total blades) how many elements. i think very high mesh elements.will reduce mesh elements size. Boudary conditions: Inlet will give: 1)nozzle angle. 2)inlet tempareture Outlet will give: outlet Pressure your taking KW model or SST model your analysing will give KW model that is better results i have one doubt your analysing only disk or total turbine How many disks,disk gap

HI kiran,
I am simulating Tesla turbine any guidance and any tutorials,please send to me, any Example Geometry.
MY mail ID:sivaramakrishnaiah@live.com,malayathsiva@gmail. com,sivaramakrishnaiah_22@yahoo.co.in

[shahryar]

Hey john i hope you are doing well i am doing the same thing can we share simulation results please?

[gangadhar]

how can i improve my drag force and drag coeeficient values?