Discussion Forum

Posted: 2 decades ago 2010/10/28 23:02 GMT-4

Kadir, I had the same problem before when I tried to compare calculated drag to analytical solution. Two observations: 1- drag on a body is extremely sensitive to "wall effects". So your computational domain should be very large compared to the size of the sphere. To ensure that computational domain boundaries are not affecting your results, I suggest you do a domain enlargement study and watch how calculated drag asymptotes to a fixed value. 2- accurate calculation of forces is important. You may want to use reacf operator rather than integrating the stress tensor. Good luck Ozgur

Posted: 2 decades ago 2010/10/29 4:34 GMT-4

Thanks Ozgur I will try but what you mean by reacf ?

Posted: 2 decades ago 2010/10/29 11:57 GMT-4

just run a search for reacf in the help documentation and you should be able to find plenty of information and examples.

Posted: 1 decade ago 2011/04/08 18:53 GMT-4

Hi, I want to find drag force over a sphere and I have the same problem, have you found the solution? How can I find the equation of reacf? Thanks

Posted: 1 decade ago 2011/04/11 15:25 GMT-4

If your flow is say in the Z-direction then the variable you need is reacf(w). You should sum that variable over the surface of the sphere. If you want to get the Stokes drag force (the equation Kadir provided in the beginning of the thread) you should select “Creeping Flow” for the fluid flow, otherwise another drag contribution proportional to the square of the velocity could be more dominant. Nagi Elabbasi Veryst Engineering

Posted: 1 decade ago 2011/05/26 5:05 GMT-4

Hi, I am working with a 3D model of a flow past a sphere as well. My computer lab doesn't have the CFD module, so I don't have the creeping flow option. I have done a very large domain vs the sphere size, sphere (r=0.05m), U_0=0.001m/s, rho=1e3kg/m^3, eta=0.9e-3Pa*s. My problem is that I still have a problem calculating the drag force, doing the integration over the sphere surface I always find exactly twice the analytic value of the drag force. I can't find what's wrong, Nagi you mention that there is a drag contribution proportional to the square of the velocity could be more dominant. My domain is a cylindric channel, I've tried with walls as boundaries and also with symmetric boundaries to make the model believe that the domain is essentially infinite and I always find the twice the analytic value. Any ideas? Thanks in advance.

Posted: 1 decade ago 2011/05/27 3:51 GMT-4

[QUOTE] Hi, I am working with a 3D model of a flow past a sphere as well. My computer lab doesn't have the CFD module, so I don't have the creeping flow option. I have done a very large domain vs the sphere size, sphere (r=0.05m), U_0=0.001m/s, rho=1e3kg/m^3, eta=0.9e-3Pa*s. My problem is that I still have a problem calculating the drag force, doing the integration over the sphere surface I always find exactly twice the analytic value of the drag force. I can't find what's wrong, Nagi you mention that there is a drag contribution proportional to the square of the velocity could be more dominant. My domain is a cylindric channel, I've tried with walls as boundaries and also with symmetric boundaries to make the model believe that the domain is essentially infinite and I always find the twice the analytic value. Any ideas? Thanks in advance. [/QUOTE] reacf() gives the most accurate solution, check. however, if you want to see if you've done a mistake or not, you can just integrate the fluid forces in the x direction, and have a rough estimate. The more finer mesh, the closer it is to reacf value (with large meshes it is far off as well). If still high, you may write your steps down and we can check (preferably in a new forum thread) -- Comsol 4.1 Ubuntu 10.04.1

Posted: 1 decade ago 2011/05/29 9:32 GMT-4

Getting exactly twice the drag force is most likely due to an input error or typo. It’s not the square velocity term because you are using creeping flow. Can you post the drag force value that you’re getting and the expression you use in COMSOL to calculate it? Nagi Elabbasi Veryst Engineering