CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   how to get the particle relative acceleration? (https://www.cfd-online.com/Forums/fluent/31435-how-get-particle-relative-acceleration.html)

 winnie May 7, 2003 22:24

how to get the particle relative acceleration?

Hi,

In fluent dpm, particle force balance can include the virtual mass force which is written as:

F=1/2*rho_l/rho_p*(du/dt-dup/dt)

This is also mentioned in fluent document. But in fluent, how can we get this force using udf since we can't access particle acceleration??

anyone has dealt with it??

Regards

winnie

 tuki72 June 2, 2003 03:48

Re: how to get the particle relative acceleration?

I calculated this one as follows:

1)I know u_f and u_p ate certain point and time 2) Then I know these after delT 3) I use d(u_f2-u_f1)/delT=du_f/dt and same for particles 4) Then subtract these two to find VMS.

I made it automatic, so its easier for me.

How do you calculate SLF (Saffman Lift Force) in fluent ??

I know how to, but I am just wondering how do?

tuki72@hotmail.com

 winnie June 2, 2003 09:08

Re: how to get the particle relative acceleration?

Hi, my friend

Thank you so much for your reply in the fluent online forum since this problem confused me a long time.

But I have tried this method, and the result is that when I include the virtual mass force, each dpm iteration will cost a long time which seems to me that this method is unsuitable because the time step is so small that the ultimate virtual mass force magnitude can be too large which made the particle have a great displacement during a time step. So I give it up. Now, since you have tried this method and it seems good to you, I will try once again, maybe I neglect something. Thank you.

As for the Safman lift force, since my case is the bubbly flow in which the bubble is about 1-5mm and safman lift force is important in low Reynold number case, I don't include saffman lift force in my case, instead, I include magnus force.

Thank you again for your help!!

Best regards

winnie

 winnie June 2, 2003 09:39

Re: how to get the particle relative acceleration?

Hi, my friend

Now I have some question about the method you provide to me.

1. for continuous phase, namely the fluid, we can get the velocity at the current location and time C_U(c,t), C_V(c,t), C_W(c,t) and the velocity at the previous time step C_U_M1(c,t), C_V_M1(c,t), C_W_M1(c,t), but it seems that only in unsteady calculation the continuous phase is related to the time, but my case is the steady one, how can I get the fluid accelaration in steady calculation.

2. for particle, we can get particle velocity at current position and at entry to current cell, and fluent provide time step P_DT(p) and two time P_TIME(p) and P_TIME0(p), my question is that is P_DT(p) equal to (P_TIME(p) - P_TIME0(p)), if they are equal, does the velocity varience between particle velocity at current position P_VEL(p)[i] and particle velocity at entry to current cellP_VEL0(p)[i] happen during this time step? if they are not equal, which one should be chosen?

Sorry for so many questions. Thank you again for your help!

Best regards

winnie

 tuki72 June 2, 2003 10:17

Re: how to get the particle relative acceleration?

1) for steady state fluid flow:du/dt=0, so in way you asked a stupid q :) when its converge, its down, no du/dt [please note that I understand in a SS case you are trying to calculate du/dt for the fluid flow 2) for a Unsteady state fluid flow u have du/dt, some times u may want:a) from a SS fluid flow, you swithc to US as u inject particles, in that case can:i) freez the fluid flow or ii) US fluid flowa as the particulate flow b) From the beginnning, u can have US fluid flow and/or u can inject particle as US case. In reality SS particulate flow is rare, however if u do such a simulation, du_p/dt=0 as well.

Now did I answer ur query?

Here are my questions:

1) At certain point and time, for a US case, u want to calculate VMF for all the points the particles are at the moment? So if u have 10,000 points (particles), what do u do? 2) U ignored SLF calculation, could you really prove (by calculating by the above procedure) that SLF is negligible (wrt to gravity force for example)?

I have more difficults questions, I have just started with the easiest ones :)

tuki72@hotmail.com

 winnie June 2, 2003 21:08

Re: how to get the particle relative acceleration?

Hi...

I think there are some concepts we should share.

In discrete phase model, the continuous phase is described in Eulerian method and the particle is described in Lagranian method. These two methods give us different results: the former gives the parameter of the point in the field and the later gives the parameter of the particle during its trajectory, so when we calculate du_f/dt-du_p/dt , we should choose either Eulerian method or Lagranian method, but not both.

In Eulerian method, the accelaration Du/Dt is calculated as:

Du/Dt = du/dt + du/dx * u + du/dy * v + du/dz * w

In lagranian method, the ccelaration Du/Dt is calculated as:

Du/Dt = [u of particle at (t + ¡÷t) - u of particle at t] / ¡÷t

best regards

winnie

So, for steady problem, whether we use Eulerian method or lagranian method, the acceleration can't be equal to zero.

 winnie June 2, 2003 22:08

Re: how to get the particle relative acceleration?

Hi, my friend

Now I use following expressions to calculate the acceleration:

for continuous phase:

lut=lux*lu+luy*lv+luz*lw;

lvt=lvx*lu+lvy*lv+lvz*lw;

lwt=lwx*lu+lwy*lv+lwz*lw;

in which lut means du_l/dt , lux means du_l/dx and so on...they can access from:

lu=C_U(c,t);

lv=C_V(c,t);

lw=C_W(c,t);

lux=C_U_G(c,t)[0];

luy=C_U_G(c,t)[1];

luz=C_U_G(c,t)[2];

lvx=C_V_G(c,t)[0];

lvy=C_V_G(c,t)[1];

lvz=C_V_G(c,t)[2];

lwx=C_W_G(c,t)[0];

lwy=C_W_G(c,t)[1];

lwz=C_W_G(c,t)[2];

and for particle:

but=(bu-bu0)/dt;

bvt=(bv-bv0)/dt;

bwt=(bw-bw0)/dt;

in which

bu=P_VEL(p)[0];

bv=P_VEL(p)[1];

bw=P_VEL(p)[2];

bu0=P_VEL0(p)[0];

bv0=P_VEL0(p)[1];

bw0=P_VEL0(p)[2];

dt=P_DT(p);

Although they are got using different method(the former: Eulerian method; the later: Lagranian method), the result of two method is equal to each other in steady problem.

When I use it in my case calculation, fluent can't go on when the first dpm iteration comes. So I think there must be something wrong that leads to some problems fluent can't deal with.

But what's the matter??