[CFDYourself]

Hi there,
I'm interested whether any of you have any tips for getting a solution for a very numerically stiff DPM model I have.

Do you have any tips for solving DPM problems where:
- Changes in the particle happen abruptly over very short distances (eg: 1 mesh element), occuring all over the model (can't be sloved by mesh refinement).
- Initial guess of the simulation (eg: the 1st iteration) is way off, leading to unphysical flow conditions etc.

I'm looking for anyone with info on:
- solver settings. I've read some ansys lit on this, but I can't get it to work.
- one-way vs. two-way coupling. is there an easy way to get ansys to consider say, species transport only, (without drag on the particle etc.) ?
- any tips for avoiding stiffness in UDFs etc? I've heard that avoiding logical statements ("if" functions) in your UDFs can improve convergence for example.

many thanks,
JimmyB

[`e`]

What causes the particles to change over very short distances? Fluent forces particles to have at least one step in each cell the particles passes through. Have you tried reducing the particle time step?

Is the initial guess inaccurate for the fluid flow or the discrete phase?

What particle tracking scheme are you using? Try using analytic or implicit instead of the higher order methods.

Ignore drag forces by specifying a custom drag law where you return zero via the DEFINE_DPM_DRAG macro.

I suppose numerical stiffness in UDFs depends on what the logical statements are enforcing and if they're using continuous / nicely behaved functions.

[CFDYourself]

Hi 'e', thankyou for the reply.

I tried the stuff you've mentioned, briefly, seemed to be an improvement

Cheers for the tip on deactivating momentum exchange using a UDF btw, that seems to have made things easier.

(here is the UDF for anyone searching this in the future. Needs to be interpreted (or compiled) and hooked as per any UDF.)
/************************************************** ***********
UDF for ignoring drag force
************************************************** ************/
#include "udf.h"
DEFINE_DPM_DRAG(zero_drag_force,Re,p)
{
real drag_force;
drag_force = 0;
return (drag_force);
}


As a side, do you know where I can find out more on the effect of "if" functions ?

[`e`]

I've not heard of people having issues with if statements in their UDFs. It'd depend on how you used said logical statements.

I suppose where an if statement is used with a random condition (say the particle has a chance of being affected by some force or otherwise) then the trajectory wouldn't be resolved with refinements (within one particle time step). A similar issue is encountered when including Brownian motion via a UDF. If there's no way of removing the random nature for each particle trajectory refinement, then set this limit to one (otherwise the particle time step tends to a very small number).