## Fluids

Topics relate to Fluent, CFX, Turbogrid and more

#### What is the reason for instability of source terms using DEFINE_DPM_SOURCE and DEFINE_DPM_LAW?

TAGGED:

• larvil
Subscriber
In my simulation I use the above law to generate mass and energy sources from the DPM model. While running the simulation, the DPM mass source gets unreasonably high, higher than possible with the injection mass flow. I assume there is some numeric problem when calculating the mass and energy source. nIn the DEFINE_DPM_LAW, the mass and the diameter of the particle are reduced or increased by calculating the mass flow mp_dot and apply it the following way (I also included a security that the particle can't get negative values):n TP_MASS(tp) -= mp_dot * TP_DT(tp); n if (TP_MASS(tp) < 0.)n {n TP_MASS(tp) = 0.;n }n TP_DIAM(tp) = pow(6.0 * TP_MASS(tp) / (TP_RHO(tp) * M_PI), 1./3.);nnIn the DEFINE_DPM_SOURCE, the sources get updated according to the example from the Ansys Fluent documentary:n mp_dot = (TP_MASS0(tp) - TP_MASS(tp)) * strength;n S->species[h2o_index] += mp_dot;n S->energy -= mp_dot * DPM_LATENT_HEAT(tp,TP_MATERIAL(tp));nI am assuming there are some controls that I can use to limit the source terms, but I don't find them, or there is another reason behind this.nThanks for some advice!nn
• Rob
Ansys Employee
Have you got a large amount of particle mass in a relatively small region? n
• larvil
Subscriber
What counts as large?nI could say that the particle density (1560kg/m3) is rather large compared to the density of the species (~2.4kg/m3) in the continuous phase.n
• Rob
Ansys Employee
High volume fraction, but also high source terms (ie very fast or very hot particles) in a fairly small volume. n
• larvil
Subscriber
The volume fraction is not very high, but the mass fraction is 0.44 in this particular case. The inlet velocities of the jet and the particles are at around 90 m/s. And the temperature of the particles and the inlet jet are at 194K, sourroundings are at 295K.
• larvil
Subscriber
Maybe this image also helps with my explanation. At iteration 200, injection starts.n
• DrAmine
Ansys Employee
You can adjust DPM URF and the frequency of updating the source terms.n
• DrAmine
Ansys Employee
What are you trying to describe with your law? What about using in addition to that a Switch UDF.n
• larvil
Subscriber
Isn't the underrelaxation only affecting the continous phase? nThe law is used so that the particles are able to sublimate and resublimate in the gas-jet that they're in. Basically, my goal is to find a steady-state of mass/enthalpy sources in the jet. I already use underrelaxation of DPM source terms because otherwise I would fight against the behaviour of the steady-state solver for the continous phase. Without underrelaxation it would switch sublimating and resublimating from iteration to iteration. This law is the only law I use currently, so what is the idea of a switch udf?nI found another post (https://forum.ansys.com/discussion/22240/local-urf-for-dpm-source-terms-fluent ) where I saw that I could use stochastic tracking and a larger number of tries. I will try that now, but maybe you have another idea.nThanks for helping!n
• larvil
Subscriber
Only stochastic tracking doesn't help, it eventually leads to instabilities.nCombining both, meaning to set the underrelaxation factor from .1 to .01 helped more. Instabilities did not occur as long as I tested.nBut still, I think there has to be a trick to not only keep it stable, but to let the solution converge faster.n
• DrAmine
Ansys Employee
What about first trying to under-relax the mass sources in the UDF? Or running with transient particle tracking (All my suggestions are blind as I do not know how you are modeling the mass transfer via the UDF).n
• larvil
Subscriber
I use a physical law for the heat transfer of a sphere in a gas. I divide the amount of heat that is transfered by the latent heat of the particle and so I get a delta mass that is either positive or negative depending on the temperature delta and that gets subtracted (or added) from the particle mass. nTransient particle tracking is active. I think that when I manually under-relax the mass source in my UDF by giving a URF to the delta mass I artificially make the particles live longer and thus not getting the correct positions of the source terms. Is that the case?n
• larvil
Subscriber
I made progress using the mixing law for density in the mixture material. For now it is quite stable.nRelated to this I have the following question, and it would be great if anyone knows this:nIs there a way to reset the dpm-sources with the command /solve/initialize/dpm-reset after an iteration and not before an iteration?nOr another trick to reset the dpm sources?n