Hi,

Maybe because of the initial condition used is uniform size distribution...

Then it is logical to have zero sigma at the initial stages.

Yes this is correct. If you use a uniform initial condition, the EQMOM procedure will simply apply QMOM, using a weighted summation to discretize the distribution function. Also, it will automatically adapt the number of nodes to 1. We have examples of this case for pbeFoam, where the initial moments are of a uniform distribution.

Does sigma start changing however while the solution evolves? There is a setting in the solver to decide what is the minimum value of sigma above which the reconstruction with EQMOM is attempted. Below this value, the reconstruction is not performed, in order to avoid numerical problems in the root-finding procedure. You can see the default values in extendedMomentInversion.C (constructor method).

Regards,

AP

Hello Alberto,

Thanks for your kind reply.

Yes, the sigma changes after some time.

However, as we work on nano particles, mainly from 50 ~ 150 nm, so when I set the the initial values (t=0 s) as follows in single cell test (with only "sum ggregationKernel", no breakup):

moment.0=3.9806152e+20; moment.1=2.6129729e+13; moment.2=1716094.9; moment.3=0.11276348; moment.4=7.4133732e-09;

which corresponds to a initial condition: uniform particle size distribution of 65 nm (all the particles are set to 65 m).

When I plot the number density distribution at t=0.001 s (only output relevant parameters starting from t=0.001s), the resulted NDF plot shows the x axis is in the range from "o to 3" without unit,

the scale of the number density seems okay (not sure the distribution shape). So here raises my question: how to take into account the unit of particle size when reconstruct and plot the PSD?

In a test in the stirred tank with only aggregation kernel (in-house developed) at the same initial conditions for the moments (without growth and breakup):

moment.0=3.9806152e+20; moment.1=2.6129729e+13; moment.2=1716094.9; moment.3=0.11276348; moment.4=7.4133732e-09;

after the first "Solving for moment.0" the code crashed with "Floating point exception".

So after debugging efforts, it is found that all the values of the moment.0 (except those at wall boundaries) turned to negative after the first "Solving for moment.0" :

DICPCG: Solving for moment.0, Initial residual = 1, Final residual = 8.00144e-06, No Iterations 5

min moment.0: -5.38125e+20

max moment.0: 0 Which caused other parameters soaring, and finally reaching the floating point error.

I checked the values of the aggregation kernel (betta) in the process of first "solving for moment.0":

min betta: 2.75174e-33

max betta: 1.44799e-17It seems that values of "betta" were in the reasonable range, so it is weird.

How all the values of moment.0 (except moment.0=0 fixed value at the wall boundary) can become negative just after the first solving step?

My mesh is very good, and the PBE calculation is based on the pre-calculated converged flow field.

Hope my problems are clearly described.

It may hard to give any exact answers for this kind problem.

Could you please give some suggestions/clues/thoughts about what might be the reasons behind?

Thanks a million!

Best regards,

Dang.