Question on implementation of Di Felice Drag

Submitted by shawnwuch on Thu, 07/02/2015 - 22:43

HI all,

1. I read the paper of Di Felice (1994) and found the drag coefficient correlation had Re = dp*magUr/nuf as the definition of particle Reynolds number (without the voidfraction term), but found the voidfraction term was coded in DiFeliceDrag.C. Could someone clarify?

Original Code in .C file:
// calc particle Re Nr
Rep = ds/scaleDia_*voidfraction*magUr/(nuf+SMALL);

// calc fluid drag Coeff
Cd = sqr(0.63 + 4.8/sqrt(Rep));

2. Also in the same file, when calculating drag coefficient, pow(voidfraction,(2-Xi)) was used instead of voidfraction^(Xi) as mentioned by Di Felice (I saw Zhu et al. (2007) used voidfraction^-(1+Xi), which was also confusing). Did I miss something here?

Thank you for helping.

Regards,
Shawn

Forums:

NTT1508 | Fri, 07/03/2015 - 00:42

Hi mate,

I think the current code is based on the paper of Zhou (2010). Take a look there and you can find the answer.

Regards,

shawnwuch | Fri, 07/03/2015 - 23:39

Thanks, I've found the reference in Zhou's paper.

e.derakhshani | Wed, 07/08/2015 - 19:55

Hello everyone,

According to Zhou 2010, Di Felice drag force is defined as follows:
fd=0.125 Cd rho_f pi d^2 ef^2 |u-v|(u-v) pow(voidfraction,X)

In the manual of CFDEM is referred to Zhou et al. (2010)-JFM for Di Felice drag force (http://www.cfdem.com/media/CFDEM/docu/forceModel_DiFeliceDrag.html) and the .C file is as below:
fd=0.125 Cd rho_f pi d^2 ef^2 |u-v|(u-v)pow(voidfraction,(2-Xi))

I think something is going wrong or I made a mistake.

P.S. :
Z. Y. ZHOU, S. B. KUANG, K. W. CHU and A. B. YU, Discrete particle simulation of particle–fluid flow: model formulations and their applicability, Journal of Fluid Mechanics / Volume 661 / October 2010, pp 482- 510

NTT1508 | Thu, 07/09/2015 - 01:04

Hi mate,

I suppose this is really optional for users. It might be acceptable for several versions of Drag force based on Di Felice concept to exist. I have not seen any paper discussing about this difference (????). You can test which one is the most suitable for you study e.g. But I think if CFDEM states that their reference is respect to Zhou, 2010, They should do like that.

Regards,

NTT1508 | Thu, 07/09/2015 - 01:15

And please check the formula used by Zhou (2010) again. I think CFDEM code is consistent with that (epsilon^(2-Xi))

e.derakhshani | Thu, 07/09/2015 - 11:47

Hi,

Do you mean that the CFDEM code is based on Zhou (2010)?

Do you mean by zhou (2010) is this "Z. Y. ZHOU, S. B. KUANG, K. W. CHU and A. B. YU, Discrete particle simulation of particle–fluid flow: model formulations and their applicability, Journal of Fluid Mechanics / Volume 661 / October 2010, pp 482- 510"? if yes, I think Zhou (2010) is proposed (epsilon^-Xi)

Best ragards

shawnwuch | Thu, 07/23/2015 - 17:27

Hi,

Glad someone is joining the discussion. As title, there is a epsilon_i^2 in that equation. So CFDEM code is following that well. However, the form is different from what Di Felice (1994) and Zhu (2007) had proposed; they differ mainly in the definition of Re of particle and f(epsilon). I haven't find any derivation of the form provided by Zhou (2010).

Regards,
Shawn

NTT1508 | Sun, 07/26/2015 - 23:45

Hi Shawnwuch,

Have you tried to test how they are different in result? We can create a simple case and use different forms of epsilon to check how the results are different. I think there is no paper for derivation of Zhou (2010). He just changed the form to fit his experiment, I think.

Regards,

shawnwuch | Tue, 08/04/2015 - 23:03

I think it would be interesting to create a simple case and compare the results with existing experimental data. Particle settling might be a place to start. We can look into more details if you're also interested in this idea.