Stepwise Simulation of Pipe Flow

h17 · January 19, 2017, 09:21

Hello!

I am simulating a steady state pipe flow with radiative heat transfer (Monte Carlo). Due to the high computational demand, I simulate 10 cm of the flow, which works well.

Now I want to extend the simulation to a longer geometry. I could simply enlarger the geometry which is ideal beacuse of the maximum possible number of histories.

I see two possibilities for which I would like to ask for help.

1. Translational periodicity. I could insert such boundary condition. If it worked, it would surely be the easiest solution. However, I do not understand, how to setup a condition like "do 10 simulations in a row, as to have a result after 1 m pipe flow". In fact, I do not know at all, if it is possible?

2. I load my 10 cm simulation in CFX post, press File-> export -> BC profile and select the necessary values to load in into another simulation as an inlet for the next 10 cm. Though this is way more complicated, I experience problems: After the next run I compared the outlet plot of simulation 1 with the inlet plot of simulation 2 and they deviated in temperature (around 5 K !), pressure and velocity (slightly).

Is there another (better) way to to what I am trying? Do you have hints?

Best regards and thanks for any comment!

ghorrocks · January 19, 2017, 18:11

I do not understand your Q1 comment. Translational periodicity means, in effect, that the periodic pair repeats forever. It does not repeat only a finite number of times. If you want to do a simulation 10 times as long then just draw a domain 10 times as long and run that.

Are you looking into this because you hit the limit of the maximum number of histories? If so, then why not just increase the limit? Also, is Monte Carlo actually required? If you are just getting heat transfer between surfaces with the fluid being transparent then a discrete transfer model is much quicker, less memory and easier to get working accurately.

h17 · January 20, 2017, 04:16

Thanks for your reply. It's a pitty the problem cannot be solves with periodic BC.

Actually yes, I am hitting the limit. I set number of histories to 500 Millions, which my system cannot exceed. Radiation field and standard deviation are promising. Unfortunately, only the Monte Carlo model is suitable, beacuse I have a participating medium and need to consider absorption as well as scattering (anisotropic). Because of that I qould like not to upscale my geometry since my number of histories would 'dilute' in that new geometry.

I trief to export outlet results (velocity, pressure, temperature, turbulence) and import it for another case as inlet. Besides the fact that this is some work doing it by hand: If i plot the inlet properties, the do show some deviation to my exported values. Also, they shot some 'spots' where temperature or velocity values are wrong.

Do you have any idea how to proceed?

h17 · January 20, 2017, 05:41

I could solve the problem with the wrong values. It was due to a misdefined 'spatial fields' setting in the export options tab

I could solve the case now by calculating 10 cm sections and hand the result values on to each new case. Do you have an idea how to to this more elegant/quickly?

ghorrocks · January 20, 2017, 06:13

Can you take the irradiation flux from the 10cm model and then not do a radiation model in the 100cm model, but instead impose the radiation heat flux as a source term? There will be some manual processing in doing this, but it might work for you. Just an idea, let me know if I am on the wrong track.

h17 · January 26, 2017, 11:06

Quote:

Originally Posted by ghorrocks

Can you take the irradiation flux from the 10cm model and then not do a radiation model in the 100cm model, but instead impose the radiation heat flux as a source term? There will be some manual processing in doing this, but it might work for you. Just an idea, let me know if I am on the wrong track.

Thanks a lot. The idea works quite well, though import/export caused some trouble, which is why an answer took so long. For anybody trying to to the same: Do not include source points, but introduce a subdomain as a source.

@ghorrocks: Do you have any information about how exactly the anisotropy factor A is defined in MonteCarlo radiation models? In manuals I can only find the equation, without any information whether or not A may exceed unity. Calculation with such values is however possible. In different textbooks, this value is named and defined differently (e.g. Modest: Radiative Heat Transfer). Do you have any hint how this is done in CFX for MonteCarlo?

ghorrocks · January 26, 2017, 16:44

Sorry, I cannot help you with the details of radiation models, it is not my field of expertise.

h17 · January 27, 2017, 02:38

Quote:

Originally Posted by ghorrocks

Sorry, I cannot help you with the details of radiation models, it is not my field of expertise.

Thank you anyway. Do you have any idea where to find information about the way linear anisotropy is handled in CFX MonteCarlo? Any recommendable textbook, documentation, ... ?

ghorrocks · January 27, 2017, 15:49

Others on the forum have more experience with radiation modelling than me (Opaque

).

The CFX theory manual is quite detailed and includes references. It would be worth looking into some of the referenced papers.

January 19, 2017, 09:21	Stepwise Simulation of Pipe Flow	#1
h17 New Member Join Date: Jan 2017 Posts: 5 Rep Power: 9	Hello! I am simulating a steady state pipe flow with radiative heat transfer (Monte Carlo). Due to the high computational demand, I simulate 10 cm of the flow, which works well. Now I want to extend the simulation to a longer geometry. I could simply enlarger the geometry which is ideal beacuse of the maximum possible number of histories. I see two possibilities for which I would like to ask for help. 1. Translational periodicity. I could insert such boundary condition. If it worked, it would surely be the easiest solution. However, I do not understand, how to setup a condition like "do 10 simulations in a row, as to have a result after 1 m pipe flow". In fact, I do not know at all, if it is possible? 2. I load my 10 cm simulation in CFX post, press File-> export -> BC profile and select the necessary values to load in into another simulation as an inlet for the next 10 cm. Though this is way more complicated, I experience problems: After the next run I compared the outlet plot of simulation 1 with the inlet plot of simulation 2 and they deviated in temperature (around 5 K !), pressure and velocity (slightly). Is there another (better) way to to what I am trying? Do you have hints? Best regards and thanks for any comment!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Y pipe flow simulation	iv1985	CFX	2	June 24, 2014 01:35
Two-phase flow in a circular horizontal pipe	DmitryS	Fluent Multiphase	0	May 17, 2014 16:22
Blockage in pipe ( using solidwork flow simulation)	jchow	FloEFD, FloWorks & FloTHERM	1	January 16, 2012 16:03
About Turbulence Intensity (Pipe flow assimilated)	gRomK13	Main CFD Forum	1	July 10, 2009 03:11
FDTD Simulation of flow through tapered pipe	Jim	Main CFD Forum	3	December 25, 2006 10:56

January 19, 2017, 18:11		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	I do not understand your Q1 comment. Translational periodicity means, in effect, that the periodic pair repeats forever. It does not repeat only a finite number of times. If you want to do a simulation 10 times as long then just draw a domain 10 times as long and run that. Are you looking into this because you hit the limit of the maximum number of histories? If so, then why not just increase the limit? Also, is Monte Carlo actually required? If you are just getting heat transfer between surfaces with the fluid being transparent then a discrete transfer model is much quicker, less memory and easier to get working accurately.

January 20, 2017, 04:16		#3
h17 New Member Join Date: Jan 2017 Posts: 5 Rep Power: 9	Thanks for your reply. It's a pitty the problem cannot be solves with periodic BC. Actually yes, I am hitting the limit. I set number of histories to 500 Millions, which my system cannot exceed. Radiation field and standard deviation are promising. Unfortunately, only the Monte Carlo model is suitable, beacuse I have a participating medium and need to consider absorption as well as scattering (anisotropic). Because of that I qould like not to upscale my geometry since my number of histories would 'dilute' in that new geometry. I trief to export outlet results (velocity, pressure, temperature, turbulence) and import it for another case as inlet. Besides the fact that this is some work doing it by hand: If i plot the inlet properties, the do show some deviation to my exported values. Also, they shot some 'spots' where temperature or velocity values are wrong. Do you have any idea how to proceed?

January 20, 2017, 05:41		#4
h17 New Member Join Date: Jan 2017 Posts: 5 Rep Power: 9	I could solve the problem with the wrong values. It was due to a misdefined 'spatial fields' setting in the export options tab I could solve the case now by calculating 10 cm sections and hand the result values on to each new case. Do you have an idea how to to this more elegant/quickly?

January 20, 2017, 06:13		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Can you take the irradiation flux from the 10cm model and then not do a radiation model in the 100cm model, but instead impose the radiation heat flux as a source term? There will be some manual processing in doing this, but it might work for you. Just an idea, let me know if I am on the wrong track.

January 26, 2017, 16:44		#7
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Sorry, I cannot help you with the details of radiation models, it is not my field of expertise.

January 27, 2017, 15:49		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Others on the forum have more experience with radiation modelling than me (Opaque ). The CFX theory manual is quite detailed and includes references. It would be worth looking into some of the referenced papers.