
[Sponsors] 
September 26, 2005, 04:54 
DPM mass problem

#1 
Guest
Posts: n/a

Hello!
I run an unsteady 3D simulation where water droplets are being injected into turbluent airflow (std ke model). I use a readfromfile injection where position, velocities, temperature, diameter and massflow is specified. Everything seems to be OK with the injection. I have also tried a single injection during my attempts to solve this problem and that didn't help. The problem is that as the calculations are advanced in time, the particles (which obey law 2 i.e. vaporization, where massflow is equal to or greater than 0 from the particles to the continuous phase) actually increase their mass! I want to use a UDF to monitor the decrease in mass, but as it is now, the ratio P_MASS(p)/P_INIT_MASS(p) is greater than 1... The decrease in mass can also be seen from the Particle Tracks window. After a few timesteps of continuous injection, there are particles in the system with mass equal to what you would expect from the specifications (density & diameter) and particles with greater mass than that, but no particles with smaller mass. Is there anyone who knows what may cause this problem? I suppose it must be some technical thing here, because this cannot make sense in a physical way, can it?? Thanks, Henrik 

September 26, 2005, 11:37 
Re: DPM mass problem  specification

#2 
Guest
Posts: n/a

Hello again!
I have investigated it more closely and this is exactly what happens: At injection, diameter and massflow is specified (density is constant). From density FLUENT will calculate the particle mass and from massflow and chosen timestep FLUENT will calculate the parcel mass. This gives at hand a number of particles in the parcel. So far so good. Then, if we look at the droplet just after injection, the parcel mass is behaving OK (slowly decreasing due to evaporation/boiling). But, the number of particles in the parcel is MUCH lower (about an order of magnitude). Instead (to compensate) the particle mass and diameter in the parcel has increased accordingly. Why is this? Is there some strange limit on how many particles that one parcel can represent? Or is it something else that I've missed? After the strange "jump" in parcel/particle properties at injection, everything is behaving just fine. Thank you for any help or ideas, Henrik 

September 27, 2005, 11:43 
Re: DPM mass problem  specification

#3 
Guest
Posts: n/a

Hello,
i dont do unsteady dpm calculation but i think i have got your mistake. You assume that particles injected = massflowrate/mass of single particle. So if your mass flow rate is 1kg/s and one particle weights 1g fluent should inject 1000 particles. This is not true. Mass flow rate and particle number are complete different stories. For a uniform size distribution one single particle is enough to track the entire stream. Just think of this particle as a representative of the whole stream. The particle motion is tracked and while the particle passes a cell it leaves source terms behind to account for momentum, energy and mass exchange. Since the single particle represents the entire stream, those source terms are calculated with the whole mass flow rate. This approach will lead to large errors since the flow will never go entirely through one single cell. Here comes stochastic tracking into play. The discrete random walk model allows you to split the single particle into n particles (n=number of tries) Lets say n=10, so fluent will now track 10 particles instead of one, each particle representing 1/10 of the stream mass flow rate. To estimate the numer of injected particles (nj) single injection: nj=number of tries group injection: jn=number of streams*number of tries surface injection: nj=number of cell faces on the surface*number of tries Sincerely RoM 

September 27, 2005, 14:26 
Re: DPM mass problem  specification

#4 
Guest
Posts: n/a

Thank you very much for your response!
However, I don't think this is the problem for this situation. I am aware of the difference between particles and parcels when doing DPM calculations. To try to isolate the problem, I now try a single injection. This means FLUENT injects one parcel per timestep, and that parcel represents a great number of particles of same diameter and mass. The problem is, that when looking at the injected parcels after, say, three timesteps, the diameters of the particles in the parcels injected will be BIGGER than the specified diameter in the injection. Since I simulate water drops, obeying laws 1, 2 or 3, it should not be possible for them to increase in mass (diameter and mass is related since density is constant in my calculations). The diameter gets bigger just after injection, then decreases like expected as the parcel/particles travel through the domain. I have now found that this seems to occur only when I use the Wave model for spray breakup. If I turn that off, or switch to the TAB model instead, there will be no parcels with diameter greater than the specified at injection, only smaller, which is what I would expect. But I have no idea why the Wave model would do such a thing? Any ideas are very welcome. Thanks, Henrik 

September 28, 2005, 01:39 
Re: DPM mass problem  specification

#5 
Guest
Posts: n/a

Do you use the droplet collision model? Maybe the parcel mass changes due to coalescence.
Sincelery RoM 

September 28, 2005, 03:43 
Re: DPM mass problem  specification

#6 
Guest
Posts: n/a

No, I have the collision model turned off. The reason I chose the Wave model was because Weber numbers are quite high in my system, and then the collision model may not be applicable (since it does not account for shattering of drops). I even tried to turn it on to see if that could help, but at least during the first few timesteps it had no effect.
Thank you once again, Henrik 

April 15, 2014, 16:27 

#7 
New Member
kourosh
Join Date: Aug 2013
Posts: 10
Rep Power: 5 
Does any one found a solution for this problem? I am facing the same issue here..I am using group injection with fixed diameter distribution and only after 3 injection my particle size distribution would change and particles diameter would increase.. I would appreciate if any one can help me with this , Thanks


September 10, 2014, 07:15 

#8 
New Member
Ali
Join Date: Jan 2014
Posts: 5
Rep Power: 4 
Hello.
I use Parcel number=1 so I expect, the particle diameter and mass, be equal to parcel diameter and mass respectively. but it is not! just after the first injection I check them and the result is shown below: Time /Diameter /Mass /ParcelMass /ParcelDiamet /ColorBy 2.61e04 /3.68733e04 /2.62030e08 /8.16545e06 /2.49989e03 /7.21577e+02 2.61e04 /7.31248e04 /2.04367e07 /8.16600e06 /2.49995e03 /7.28409e+02 2.61e04 /4.69963e04 /5.42510e08 /8.16614e06 /2.49996e03 /7.27910e+02 2.61e04 /2.54863e03 /8.65242e06 /1.17164e05 /2.81963e03 /7.29512e+02 2.61e04 /1.44134e03 /1.56499e06 /9.41391e06 /2.62131e03 /7.28597e+02 so my question is why the parcel mass and particle mass is not equal? the same quastion for diameters 

December 14, 2014, 12:44 
dpm

#9 
New Member
Devendra choudhary
Join Date: Dec 2014
Posts: 2
Rep Power: 0 
can any one please tell me how can i get dpm in cfx . i am not able to identify where i can get dpm option in cfx.
plese it is very important............... 

May 20, 2016, 11:00 

#10 
Senior Member
Andrea
Join Date: Mar 2009
Location: Torino, Italy
Posts: 101
Rep Power: 9 
I've run into the same problem recently, the advise is to check if WAVE model is acceptable in your simulation, because We number should be quite high in order to obtain consistent results. If We < 100, then go on with TAB.


Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
DPM Output Problem (Again...)  DLINSA  Fluent UDF and Scheme Programming  1  March 16, 2012 01:45 
Mass conservation problem in mixing tank multiphase simulation  rockewan  FLUENT  0  April 6, 2010 12:34 
mass flow in is not equal to mass flow out  saii  CFX  2  September 18, 2009 08:07 
DPM: How to find out the trapped particles mass  Prashant  Main CFD Forum  0  October 24, 2007 12:03 
CFXPost: problem with mass flow  Markus  CFX  5  February 19, 2005 18:30 