Viscoelastic Fluid Flows using OpenFOAM The solver viscoelasticFluidFoam

jovani · February 6, 2009, 16:29

Hello Foamers,

My name is Jovani and I want announce that will soon be available a solver for treatment of viscoelastic fluid in OpenFOAM. I was working in the development of a viscoelasticFluidFoam solver (with the great orientation of the Dr. Hrvoje Jasak, thank you very much Hrvoje!!) to simulate viscoelastic flows.

The follow models are implemented: Maxwell, Oldroyd-B, Giesekus, PTT linear and exponential,
FENE-P, FENE-CR and DCPP. There are others models already implemented but not tested (Pom-Pom, SXPP, DXPP, WM ...).

The metodology used to obtain solutions in high We numbers was the DEVSS. The test geometry was mainly the 4:1 planar contraction. The solver are able to simulate multimode cases with n modes.

For the interested: A presentation that will soon be available on line to give more informations.

Jovani

deepsterblue · February 6, 2009, 18:43

Brilliant work, Jovani!

Do you think I could take a look at the sources? This is something we were looking to implement, but it looks like you've saved us the trouble!

jovani · February 6, 2009, 19:02

Hello Menon,

Yes, the solver will be available in the -dev version of OpenFOAM, but you need wait untill the solver is put into the source. Don't be worried, this will be done soon.

Jovani

kerstin · February 7, 2009, 10:03

Hi Jovani,

Great work! This will speed up research in viscoelastic simulations a lot. 8-)
Really great!
Greetings
Kerstin

hjasak · February 7, 2009, 10:18

A piece of propaganda: Jovani did this work under a Masters project at Universidade Federal Rio Grande do Sul, Brazil and we are trying to convince him to stay on and do a PhD along similar lines.

Can I ask the audience to kindly help me in this.

Hrv

holger_marschall · February 7, 2009, 11:55

Hi Jovani,

great job! And the fact that it was made under a master's project is simply terrific! Great.

And... You should just listen to Hrvoje! Go on with CFD and do a PhD. So we can wait for more good to come in the field of viscoelastic flow simulations!

best
Holger

jovani · February 7, 2009, 17:38

Thanks everyone!!

Thanks Hrvoje for the propaganda!

Marschall, I'm thinking to a PhD. There are a lot of studies to be done in this area.
As there isn't much in this area into commercial softwares, I think that if we do any efforts, with the help of Hrvoje of course, we can create something top of line.

Jovani

nzy102 · March 4, 2009, 16:08

Hello Jovani,

Thank you for your good work. When will the next solver be released?

Ning

max · March 5, 2009, 04:46

Hi Jovani,

that all looks very promising.

However, could you give some details about the stability of your solver:

- up to which We (or De)-Number stable solutions can be obtained in your 4to1 planar geometry?

- any tests on realistic 3-D geometries and if, what is the performance in terms of CPU-time?

cheers Max

jovani · March 5, 2009, 15:59

Hello All,

Ning, the solver will released in this month or the maximum in April. I believe I will not have time to test some models (white-Metzner and derivations, PP, XPP and Feta-PTT) but their will be into the solver too. Is necessary validate these models.

Max,

About the De numbers: I make any test to explore this question, but only for some models (PTT and Giesekus). But to give to you a number I simulate a case that I simulate was a LDPE flow with De=250 using Giesekus 8-mode model, this was't the limit for this case, I simply don't tried higher values. A complete study to this maximum Deborah values need to be done. Other, the DEVSS methodology sugest an arbitrary value for the the term added to moment equation, generally is used the value of etaP (value that is used in the solver), but we know that the better value for this term is related with each model and your parameters, this need more study too and I think with it's possible obtain any improvements. The interpolation shemes used also afect it.

About 3-D simulations: I do a test with a realistic case that was the flow into a capillary with a contraction. The mesh was about 150000 volumes, considering 2 simetry planes. As is make the transient simulation the Courant number control the time step, but the simulation time is acceptable for me. I used multigrid and parallelization tecniques for it.

file:///home/jovani/Desktop/PresentationSolver/pictures/magtautri2.png

file:///home/jovani/Desktop/PresentationSolver/pictures/meshtri.png

Jovani

jovani · March 5, 2009, 16:04

Hello All,

Ning, the solver will released in this month or the maximum in April. I believe I will not have time to test some models (white-Metzner and derivations, PP, XPP and Feta-PTT) but their will be into the solver too. Is necessary validate these models.

Max,

About the De numbers: I make any test to explore this question, but only for some models (PTT and Giesekus). But to give to you a number I simulate a case that I simulate was a LDPE flow with De=250 using Giesekus 8-mode model, this was't the limit for this case, I simply don't tried higher values. A complete study to this maximum Deborah values need to be done. Other, the DEVSS methodology sugest an arbitrary value for the the term added to moment equation, generally is used the value of etaP (value that is used in the solver), but we know that the better value for this term is related with each model and your parameters, this need more study too and I think with it's possible obtain any improvements. The interpolation shemes used also afect it.

About 3-D simulations: I do a test with a realistic case that was the flow into a capillary with a contraction. The mesh was about 150000 volumes, considering 2 simetry planes. As is make the transient simulation the Courant number control the time step, but the simulation time is acceptable for me. I used multigrid and parallelization tecniques for it.

file:///home/jovani/Desktop/PresentationSolver/pictures/magtautri2.png

file:///home/jovani/Desktop/PresentationSolver/pictures/meshtri.png

Jovani

jovani · March 5, 2009, 19:37

Ops, Sorry!!!

An error occurred, I sent twice the same message and the pictures do not is showed. I believe the pictures appears now.

U magnitude:

http://img24.imageshack.us/img24/174/magutri2.png][IMG]http://img24.imageshack.us/img24/174/magutri2.th.png[/IMG][/URL]

Stress magnitude:

http://img6.imageshack.us/img6/6859/magtautri2.png][IMG]http://img6.imageshack.us/img6/6859/magtautri2.th.png[/IMG][/URL]

Mesh:

http://img22.imageshack.us/img22/9122/meshtri.png][IMG]http://img22.imageshack.us/img22/9122/meshtri.th.png[/IMG][/URL]

Jovani

michaelb · April 8, 2009, 17:30

Hello Jovani

I am interested in knowing if the solver is released yet.

I also would like to know, before switching to OpenFOAM, if it possible to model two phase flows, one of the phase having specific viscoelastic properties.

Thank you

Mike

kerstin · April 9, 2009, 03:12

Hi Michael,
In principle nothing is impossible in OpenFOAM. ;-)
I guess handling the viscoelastic stresses correctly in the region, where you don't have a sharp interface is the most tricky part...I mean to make it really physical.
But how do you intend to do it, if you don't switch to OpenFOAM? Everything I can imagine to do as alternative (commercial code or own code) will be much more difficult than in OpenFOAM...
Greetings
Kerstin

jovani · May 7, 2009, 17:55

Hello Mike,

Kerstin is rigth about OpenFOAM

The solver was not release yet! I am finishing the tests Hrvoje ask me to do and I need talk with him about it in this days.
About free surface viscoelastic flow: Yes is possible in OpenFOAM, I am simulate rod-climming and die swell, but this work are just beginning, few tests for the methodology was done untill now!!

I advert you when the solver is released.

Best,
Jovani

wendywu · May 14, 2009, 18:27

Hi,
Jovani,

I am planing to use OpenFOAM to simulate aluminum extrusion which is viscoplastic constitutive model. Will you give the viscoplastic constitutive model in the future? I think maybe for you it is a very easy work to do. But I just don't know how to do it and where I can get started to do it by myself. Could you please give me some advice?
Attatchment is the constitutive model. Thank you.

Wendy.

kerstin · May 14, 2009, 23:58

Hi Wendy,

What do you mean mathematically with D^d in the .doc?
D is a tensor and what is meant with the power of d???

At the moment I don't see any viscoelastic part in your equation. Viscoelastic properties are related to the left hand side (time derivative) of a constitutive equation. You should have a look on the theory of linear viscoelasticity, where material behaviour and equations are explained with spring's and dash-pots...(keyword: Maxwell fluids... ) you only have an explicit equation for stress tensor as a function of deformation gradient tensor with non constant viscosity...

but please explain the D^d further, bacause your material law looks interesting...
Kerstin

wendywu · May 15, 2009, 13:47

Hi, thank you.

It is a viscoplastic model , not viscoelastic.
D is rate of deformation tensor, the superscript d represents deviator.
the equation for D is as attathment.

Thank you very much.

Wendy

wendywu · May 15, 2009, 14:41

Hi, I have another constitutive model , maybe it is easier to realize. please see the attathment.
Thank you .
Have a nice day.

Wendy

hjasak · May 16, 2009, 02:14

No pressure... but Jovani has got his examination on Monday - just registering support in public

Hrv

February 6, 2009, 16:29	Hello Foamers, My name is J	#1
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Hello Foamers, My name is Jovani and I want announce that will soon be available a solver for treatment of viscoelastic fluid in OpenFOAM. I was working in the development of a viscoelasticFluidFoam solver (with the great orientation of the Dr. Hrvoje Jasak, thank you very much Hrvoje!!) to simulate viscoelastic flows. The follow models are implemented: Maxwell, Oldroyd-B, Giesekus, PTT linear and exponential, FENE-P, FENE-CR and DCPP. There are others models already implemented but not tested (Pom-Pom, SXPP, DXPP, WM ...). The metodology used to obtain solutions in high We numbers was the DEVSS. The test geometry was mainly the 4:1 planar contraction. The solver are able to simulate multimode cases with n modes. For the interested: A presentation that will soon be available on line to give more informations. Jovani

February 6, 2009, 18:43	Brilliant work, Jovani! Do	#2
deepsterblue Senior Member Sandeep Menon Join Date: Mar 2009 Location: Amherst, MA Posts: 342 Rep Power: 13	Brilliant work, Jovani! Do you think I could take a look at the sources? This is something we were looking to implement, but it looks like you've saved us the trouble! __________________ Sandeep Menon University of Massachusetts Amherst http://www.ecs.umass.edu/~smenon

February 6, 2009, 19:02	Hello Menon, Yes, the solve	#3
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Hello Menon, Yes, the solver will be available in the -dev version of OpenFOAM, but you need wait untill the solver is put into the source. Don't be worried, this will be done soon. Jovani

February 7, 2009, 10:03	Hi Jovani, Great work! This	#4
kerstin New Member Kerstin Heinen Join Date: Mar 2009 Location: Ludwigshafen, Germany Posts: 27 Rep Power: 6	Hi Jovani, Great work! This will speed up research in viscoelastic simulations a lot. 8-) Really great! Greetings Kerstin

February 7, 2009, 10:18	A piece of propaganda: Jovani	#5
hjasak Senior Member Hrvoje Jasak Join Date: Mar 2009 Location: London, England Posts: 1,695 Rep Power: 17	A piece of propaganda: Jovani did this work under a Masters project at Universidade Federal Rio Grande do Sul, Brazil and we are trying to convince him to stay on and do a PhD along similar lines. Can I ask the audience to kindly help me in this. Hrv __________________ Hrvoje Jasak

February 7, 2009, 11:55	Hi Jovani, great job! And t	#6
holger_marschall Senior Member Holger Marschall Join Date: Mar 2009 Location: Darmstadt, Germany Posts: 105 Rep Power: 7	Hi Jovani, great job! And the fact that it was made under a master's project is simply terrific! Great. And... You should just listen to Hrvoje! Go on with CFD and do a PhD. So we can wait for more good to come in the field of viscoelastic flow simulations! best Holger __________________ Holger Marschall mailto:holgermarschall@yahoo.de

February 7, 2009, 17:38	Thanks everyone!! Thanks Hr	#7
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Thanks everyone!! Thanks Hrvoje for the propaganda! Marschall, I'm thinking to a PhD. There are a lot of studies to be done in this area. As there isn't much in this area into commercial softwares, I think that if we do any efforts, with the help of Hrvoje of course, we can create something top of line. Jovani

March 4, 2009, 16:08	Hello Jovani, Thank you for	#8
nzy102 Member Ning Yang Join Date: Mar 2009 Location: University Park, PA, USA Posts: 85 Rep Power: 6	Hello Jovani, Thank you for your good work. When will the next solver be released? Ning

March 5, 2009, 04:46	Hi Jovani, that all looks v	#9
max New Member markus gruber Join Date: Mar 2009 Posts: 1 Rep Power: 0	Hi Jovani, that all looks very promising. However, could you give some details about the stability of your solver: - up to which We (or De)-Number stable solutions can be obtained in your 4to1 planar geometry? - any tests on realistic 3-D geometries and if, what is the performance in terms of CPU-time? cheers Max

March 5, 2009, 15:59	Hello All, Ning, the solver	#10
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Hello All, Ning, the solver will released in this month or the maximum in April. I believe I will not have time to test some models (white-Metzner and derivations, PP, XPP and Feta-PTT) but their will be into the solver too. Is necessary validate these models. Max, About the De numbers: I make any test to explore this question, but only for some models (PTT and Giesekus). But to give to you a number I simulate a case that I simulate was a LDPE flow with De=250 using Giesekus 8-mode model, this was't the limit for this case, I simply don't tried higher values. A complete study to this maximum Deborah values need to be done. Other, the DEVSS methodology sugest an arbitrary value for the the term added to moment equation, generally is used the value of etaP (value that is used in the solver), but we know that the better value for this term is related with each model and your parameters, this need more study too and I think with it's possible obtain any improvements. The interpolation shemes used also afect it. About 3-D simulations: I do a test with a realistic case that was the flow into a capillary with a contraction. The mesh was about 150000 volumes, considering 2 simetry planes. As is make the transient simulation the Courant number control the time step, but the simulation time is acceptable for me. I used multigrid and parallelization tecniques for it. file:///home/jovani/Desktop/PresentationSolver/pictures/magtautri2.png file:///home/jovani/Desktop/PresentationSolver/pictures/meshtri.png Jovani

March 5, 2009, 16:04	Hello All, Ning, the solver	#11
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Hello All, Ning, the solver will released in this month or the maximum in April. I believe I will not have time to test some models (white-Metzner and derivations, PP, XPP and Feta-PTT) but their will be into the solver too. Is necessary validate these models. Max, About the De numbers: I make any test to explore this question, but only for some models (PTT and Giesekus). But to give to you a number I simulate a case that I simulate was a LDPE flow with De=250 using Giesekus 8-mode model, this was't the limit for this case, I simply don't tried higher values. A complete study to this maximum Deborah values need to be done. Other, the DEVSS methodology sugest an arbitrary value for the the term added to moment equation, generally is used the value of etaP (value that is used in the solver), but we know that the better value for this term is related with each model and your parameters, this need more study too and I think with it's possible obtain any improvements. The interpolation shemes used also afect it. About 3-D simulations: I do a test with a realistic case that was the flow into a capillary with a contraction. The mesh was about 150000 volumes, considering 2 simetry planes. As is make the transient simulation the Courant number control the time step, but the simulation time is acceptable for me. I used multigrid and parallelization tecniques for it. file:///home/jovani/Desktop/PresentationSolver/pictures/magtautri2.png file:///home/jovani/Desktop/PresentationSolver/pictures/meshtri.png Jovani

March 5, 2009, 19:37	Ops, Sorry!!! An error occu	#12
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Ops, Sorry!!! An error occurred, I sent twice the same message and the pictures do not is showed. I believe the pictures appears now. U magnitude: http://img24.imageshack.us/img24/174/magutri2.png][IMG]http://img24.imageshack.us/img24/174/magutri2.th.png[/IMG][/URL] Stress magnitude: http://img6.imageshack.us/img6/6859/magtautri2.png][IMG]http://img6.imageshack.us/img6/6859/magtautri2.th.png[/IMG][/URL] Mesh: http://img22.imageshack.us/img22/9122/meshtri.png][IMG]http://img22.imageshack.us/img22/9122/meshtri.th.png[/IMG][/URL] Jovani

April 8, 2009, 17:30		#13
michaelb New Member Michael Bruckner Join Date: Apr 2009 Location: France Posts: 27 Rep Power: 6	Hello Jovani I am interested in knowing if the solver is released yet. I also would like to know, before switching to OpenFOAM, if it possible to model two phase flows, one of the phase having specific viscoelastic properties. Thank you Mike

April 9, 2009, 03:12		#14
kerstin New Member Kerstin Heinen Join Date: Mar 2009 Location: Ludwigshafen, Germany Posts: 27 Rep Power: 6	Hi Michael, In principle nothing is impossible in OpenFOAM. ;-) I guess handling the viscoelastic stresses correctly in the region, where you don't have a sharp interface is the most tricky part...I mean to make it really physical. But how do you intend to do it, if you don't switch to OpenFOAM? Everything I can imagine to do as alternative (commercial code or own code) will be much more difficult than in OpenFOAM... Greetings Kerstin

May 7, 2009, 17:55		#15
jovani Member Jovani L. Favero Join Date: Mar 2009 Location: Rio de Janeiro, RJ, Brazil Posts: 40 Rep Power: 7	Hello Mike, Kerstin is rigth about OpenFOAM The solver was not release yet! I am finishing the tests Hrvoje ask me to do and I need talk with him about it in this days. About free surface viscoelastic flow: Yes is possible in OpenFOAM, I am simulate rod-climming and die swell, but this work are just beginning, few tests for the methodology was done untill now!! I advert you when the solver is released. Best, Jovani

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
VOF simulation of a viscoelastic fluid	sinah	OpenFOAM Running, Solving & CFD	10	November 25, 2010 11:02
FREE SURFACE VISCOELASTIC FLOWS	Valdemir G. Ferreira	Main CFD Forum	6	December 18, 2009 06:14
Viscoelastic flow modeling in OpenFOAM	vulda	OpenFOAM Running, Solving & CFD	1	March 17, 2008 08:32
Polyflow & OpenFoam on Viscoelastic flow modeling	Sumeshen	Main CFD Forum	0	March 14, 2008 09:29
Viscoelastic fluid codes	joel davison	Main CFD Forum	0	November 6, 2001 05:09

May 14, 2009, 23:58		#17
kerstin New Member Kerstin Heinen Join Date: Mar 2009 Location: Ludwigshafen, Germany Posts: 27 Rep Power: 6	Hi Wendy, What do you mean mathematically with D^d in the .doc? D is a tensor and what is meant with the power of d??? At the moment I don't see any viscoelastic part in your equation. Viscoelastic properties are related to the left hand side (time derivative) of a constitutive equation. You should have a look on the theory of linear viscoelasticity, where material behaviour and equations are explained with spring's and dash-pots...(keyword: Maxwell fluids... ) you only have an explicit equation for stress tensor as a function of deformation gradient tensor with non constant viscosity... but please explain the D^d further, bacause your material law looks interesting... Kerstin

May 16, 2009, 02:14		#20
hjasak Senior Member Hrvoje Jasak Join Date: Mar 2009 Location: London, England Posts: 1,695 Rep Power: 17	No pressure... but Jovani has got his examination on Monday - just registering support in public Hrv __________________ Hrvoje Jasak

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode