CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   dpm sample track (http://www.cfd-online.com/Forums/fluent/31436-dpm-sample-track.html)

Gael Corre May 8, 2003 02:13

dpm sample track
 
We have some problem using the Discrete Phase Modelling sample track. First,we wanted to use udf output function to print particles positions and time at given position we were interested in. The problem encountered is that using an udf when the stochastic tracking model is enabled gives us no results. We tried many time and we cannot understand why using a udf is impossible when the stochastic model is used. Then, when we use the particle sample track method without any udf output, but using lines we defined for printing in the output file the paticles data whenthey cross those lines, and the problem is that when time becomes high, we obtain many positions for the same time (up to 500 positions for the same time which is very incoherent). This thing occurs again when we use the default interior surface. If someone has already encountered this problem,or has any idea about why we obtain many position, we will be very interested,as we cannot work any more... thanks.

Alex Munoz May 8, 2003 14:43

Re: dpm sample track
 
Hi

I have been working with discrete randow walk model(DRWM) per years and I have ever hear this problem.

First at all, you email is very confussing, I really suggest that you write shorter sentences, with independent paragraphs.

Second, I am sure that you are doing something wrong

Third, which Fluent version are you using? I sugguest that you get fluent6.1.

Fourth, are you doing steady or unsteady particle tracking?

Fifth, It seems to me that you don't have the less idea about the foundation of the DRWM.

It you want that people help you in this forum. I suggest that you explain your case better.

Regards

Alex Munoz

Gael Corre May 9, 2003 01:23

Re: dpm sample track
 
Thanks for your response, I will try to make myself clear.

I am using fluent 5.5 but I am trying it now on Fluent 6.1

We are working on a 2D system (a driven cavity) with a steady flow. The idea is to track neutrally buoyant particles during a quite long time, so as to establish transfer functions between zones of the system. We are using the stochastic model for particle tracking once the flow field is solved.

We need to use a small length scale for the integration. Then, if we use a step by step report,we obtain files really too big because we are tracking the particles for a long time.

So, we try to use the Report>Trajectory>sample track option. We wanted to print the particle datas as they pass the lines that we have defined (here it is x=0.75). For high values of time, we obtain many coordinates for the same time, which is in many cases physically impossible. That occurs only for high values of time.

(line-10 10) ( X Y Z U V W diameter T mass-flow time name)

**Early data removed.

(( 7.4999e-01 3.1644e-01 0.0000e+00 -5.0512e-01 -1.0909e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3039e+03) injection-0:0)

(( 7.5001e-01 2.4768e-01 0.0000e+00 4.4392e-01 -5.1917e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3043e+03) injection-0:0)

(( 7.4999e-01 2.8565e-02 0.0000e+00 -3.8418e-01 -2.1176e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5001e-01 2.1889e-02 0.0000e+00 3.2552e-01 -3.8038e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 1.1009e-01 0.0000e+00 -6.3848e-01 2.1499e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5001e-01 2.5242e-01 0.0000e+00 9.3530e-01 7.1031e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 5.9257e-01 0.0000e+00 -2.0515e+00 -4.0922e-02 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5001e-01 7.8157e-01 0.0000e+00 4.2634e-01 1.1851e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 6.4539e-01 0.0000e+00 -1.1004e+00 -5.8345e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5000e-01 6.3429e-01 0.0000e+00 3.2369e-01 8.8266e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 6.9207e-01 0.0000e+00 -4.5330e-01 2.5844e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5001e-01 7.9841e-01 0.0000e+00 4.0404e-01 1.2834e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 7.6826e-01 0.0000e+00 -1.0250e+00 -6.2766e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.5000e-01 8.7466e-01 0.0000e+00 1.4805e-01 -5.2010e-01 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3047e+03) injection-0:0)

(( 7.4999e-01 4.3164e-01 0.0000e+00 -1.6911e-01 7.9524e-02 0.0000e+00 1.0000e-06 3.0000e+02 0.0000e+00 1.3051e+03) injection-0:0)

Why do we see this? What have we done wrong?

Alex Munoz May 9, 2003 13:41

Re: dpm sample track
 
Hi

Everything seem ok to me, Perhaps, you cannot see it because you have not yet understand was going on, let me explain you.

There are physically two ways to perform residence time distribution (RTD), pulse input or slug release of tracer and step input or continuous release of tracer.

A pulse input in a almost-perfect plug flow will move as band or piston through the reactor, until the band reaches the point of sampling, then RTD is a bell shape curve with a narrow dispersion.

A step input in a almost-perfect plug flow will dilute and move with this concentration to the outlet of the reactor, where you will register an increase of the tracer concentration until it reaches a maximum.

see this link http://cee.wpi.edu/ce3060_b02/reactors/intro.html

Now, coming back to your computational tracer analisis. The slug imput correspond to the steady particle tracking, whereas the step input correspond to the unsteady particle tracking.

As a result, the methodology to analysis the results from each particle tracking are different and well summarized by Levenspiel (search for a book about Reactor design by Levenspiel)

From the file that you send me I have some comment:

1) your particle diameter (dp) is too small as a result your particle time step(tp) is very small. I suggest that you increase your particle diameter to at least 100 microns. This will rise your particle time step in a order of magnitude of 4 because tp is proportional to the square of the dp. In this way your calculation will be much faster.

Second, determine what is your average residence time and get the files before this time.

In conclusion, in unsteady particle tracking particle are launched from the injection points every time step, whereas for steady particle tracking particles are launched just one time.

Regards

Alex Munoz


Alex Munoz May 9, 2003 13:57

Re: dpm sample track
 
Hi

I forget to tell you there is a Fluent users with the nick name Candy write to him (her) and ask for help she was doing unsteady particle tracking. personally I have not done one of this because demand to much time.

regards

Alex


All times are GMT -4. The time now is 00:20.