CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Steady State DPM (

K. Jagus June 21, 2005 03:21

Steady State DPM
Hello, I am doing a simulation of the flow in a combustion chamber with fuel injection. I am using DPM model. My simulation is steady state. I have following problems, maybe someone could shed some light on it. How do I limit injection number? That is if I set number of Iterations per DPM iteration to 1, Fluent injects new particles every time. I would like to inject them once and keep tracking. Second question is due to the weird distribution I get. I get an even spray from pressure swirl, then particles dissapear only to reappear further down the domain with no continuity. What may be the cause of it?

Thank You for any suggestions.

Alec Eiffel June 21, 2005 09:22

Re: Steady State DPM
Will try and answer some of those questions

How do I limit injection number? The easiest way to do this is to release the injection from a plane. When defining the plane check the "Bounded" and "Sample Points" boxes. The number of sample points gives no. of particles injected.

Your next question is setting the number of Iterations per DPM iteration. 1 is far too small. should be between 10 -20. Do you need to couple the flow? if not you can simply go to Display -> particle tracks and track the particles once the flow solution is converged.

As for the particles disappearing, this is because you simply havent tracked the particles far enough into the domain. The DPM model is not based on continuity. To fix this go to Define -> Models -> Discrete Phase and increase the Max Number of Steps.

K. Jagus June 21, 2005 10:04

Re: Steady State DPM
Thank You very much. I am now trying to follow your guidelines. I must admit that the plane creation is for a moment a mystery to me but I am working on it. If I release injection from the plane will I be able to use pressure-swirl injection type or any other for that matter? As for flow coupling I think I should use it. Thanks for Your patience, I am new to fluent and cfd,hence my questions.

Alec Eiffel June 21, 2005 10:50

Re: Steady State DPM
Sorry i didnt realise that you were using pressure-swirl injection. The plane injection is irrelevant for that as you said. I think you can control no. of particles using the "Number of Particle Streams" in the Define -> Injections panel but im not too sure about it.

K. Jagus June 21, 2005 10:59

Re: Steady State DPM
The number of particle streams can be controlled by the option You mentioned only for a single injection occurence.I mean if I set number of continuous phase iterations per DPM iteration to letīs say 20 fluent reports every 20 iterations that 40 particles injected (40 as set in number of particle streams). It happens every 20 iterations and the total amount of particles keeps growing. In my opinion it leads to strange situation where particles are not anymore less than 10% and DPM model cannot be applied anymore. Maybe it is the balance between those parameters and number of time steps that should be found by trial and error simply. I will keep trying and post the results.

Allan Walsh June 21, 2005 13:41

Re: Steady State DPM
If you couple the particle and continuous phases, then the interface source and sink terms will be updated each time the particles are tracked, not summed.

Not sure how the particles could pass out of the domain, and then reappear. I think you have a problem there.

K. Jagus June 22, 2005 02:34

Re: Steady State DPM
If they are not summed why every 20 iteration I get the message "injecting 20 particles, number tracked n*20 ..." where n is number of 20 iteration cycles. I checked this morning and I had 600 particles being tracked. As to passing out of the domain, I may have expressed myself not correctly. The probably sty in the domain only its display of residence time or velocity does not resemble continuous. There are very large gaps between concentrations of particles like sandwich between particles and continuous phase.

Jeff Moder June 24, 2005 17:10

Re: Steady State DPM
I am not a Fluent user, but it appears that Fluent is doing what i have seen other codes do regarding injecting particles.

While ultimately you want to continue tracking each particle injected until it leaves the domain or completely evaporates (if it is evaporating), an alternative is to follow all particles in the domain at a particular iteration for a fixed amount of time, and inject a new set of particles (with same injection properties) each time the particle solver is called. Eventually, if a true steady-state is reached, the above scenario will be equivalent to following a SINGLE set of injected particle for their entire lifetime (ie evaporate or leave domain) each call to the particle solver. For serial runs, the computational load should be the same, but depending on how the particle solver is parallelized, following particles only a certain fixed time and injecting a new set each call to the particle solver, can be more efficient.

All times are GMT -4. The time now is 14:51.