topic Re: Velocity with very short traces in CFD Forum

Velocity with very short traces

Anonymous — Tue, 30 Apr 2019 13:07:06 GMT

Hello,

I'm trying to simulate the flow of a polymer and I want to see the velocity magnitude in general and also, determine the residence time with the Traces option. However, I assigned a Mass flow rate of 350 kg/h and a pressure of 170 bar at the inlet with 0 psi pressure at outlet.

Running the simulation with a time step of 1s I see the velocity magnitude in a plane (and looks fine), but when I add traces, they don't go from the inlet to the outlet.

Also, I don't understand why there's a huge difference in velocity in the initial length of the inlet and then it goes down. The velocity value seems very unreal, since the mass flow is 350kg/h...

Boundary Conditions:

Traces issue:

Can anyone help me and does anyone know what am I doing wrong?
(I hope you help me Fred :D)

Re: Velocity with very short traces

frederic.gaillard.7 — Tue, 30 Apr 2019 15:33:02 GMT

Hello André,

Yeah sure i can help you !
The trace problem might be related to the autodesk version, i need to verify first! However if you are interested about the resident time you can check the box LMA (local mean age) in the result quantities tab. The result will be available in the summary file.

For your boundary conditions it seems that they are to close of your region of interest. You should extend your intlet and your outlet from your computational domain.

Can you share your CFZ file ?
Fred

Re: Velocity with very short traces

Anonymous — Tue, 21 May 2019 14:39:44 GMT

Hi @frederic.gaillard.7 ,

I'm sorry for the late response, but I've been busy with some experimental testing and I've not been using Autodesk CFD too much.

In the last couple of days I redesigned the geometry with caution and created a whole new setup. This worked out very nicely! I've also spoken with one of my teachers and I understood what the 2nd Order Polymer was doing on the simulation and then I used Excel to get the 6 coefficients of the equation and the viscosity values given by the equation compared to the experimental ones are very very close! (Avg. error : 0,2%)

I've also extended the inlet 5x D (diameter) and it worked out!
The simulations seem to be accurate, however I don't think that the viscous heating, generated by the energy released by the shear stress applied on the rubber coumpound are being taken into consideration. I mean, I should see a temperature increase of the rubber at the inlet... But I don't see that!

My Resid in and Resid out plot are looking "pretty" meaning that everything should be relatively ok.

Re: Velocity with very short traces

frederic.gaillard.7 — Fri, 24 May 2019 15:17:10 GMT

Hello André,

The viscous heating is only supported by the k-epsilon turbulence model. Generally speaking the heat is transfered to the fluid at the end of the kolmogorov scale (energy cascade).

I never try this before but i suppect that you will need a fine mesh to modelize the TED properly.

Ultimately it will be the heated fluid that will transfered the heat to the rubber part.

Fred

Re: Velocity with very short traces

Anonymous — Fri, 24 May 2019 15:36:34 GMT

Hi @frederic.gaillard.7 ,

I'm struggling to find more information about this subject!

This link had some information about this, but unfortunately it's down now...

https://knowledge.autodesk.com/support/cfd/learn-explore/caas/sfdcarticles/sfdcarticles/Simulation-CFD-Defining-Plastic-Viscosity.html

I'm solving with the K-Epsilon turbulence model, but I don't see changes in temperature due to the viscous heating as I was expecting...

Also I'm using the 2nd Order polymer method to define the rubber viscosity in dependency of the shear stress.

To fine the mesh, do I need to refine only the fluid part or also the surrounding parts where the rubber flows?

I'm having success with my simulations, except for this part... The viscous heating seems not to be working, and I'm a bit "lost" because I don't know how to fix this.

Re: Velocity with very short traces

frederic.gaillard.7 — Fri, 24 May 2019 17:21:26 GMT

Hello André

Which information you're struggling with ??
Can you share your CFZ i will have a look.

Thank
fred

Re: Velocity with very short traces

Anonymous — Fri, 24 May 2019 19:05:31 GMT

Hi Fred,

The information about the viscous heating... I mean, it's very very very hard to find any software that calculates viscous heating. From what I've seen Autodesk CFD has this capability but i don't see the temperature increasing due to this viscous heating, and it's driving me crazy!

Again, the visocisty magnitude in general seems very weird! i think that the viscosity in average is too low from what it should be...

Re: Velocity with very short traces

frederic.gaillard.7 — Fri, 24 May 2019 19:31:58 GMT

quick suggestion here
your wall roughness is 0, try with a none zero values. I'm not sure but i might help to produces friction and increase heat.

Fred

Re: Velocity with very short traces

Anonymous — Sat, 25 May 2019 15:57:50 GMT

Oh!! I set the surface roughness on the rubber material (B163) instead of the steel...

Also I will take a look at the thermal conductivity of the rubber. Maybe it’s now correct, but I will confirm this data.

Do you know how to use the Carreau model? I’ve heard about it but I don’t know how to use it...

thanks Fred.

Re: Velocity with very short traces

frederic.gaillard.7 — Mon, 27 May 2019 16:11:52 GMT

Hello André,

I have the time to run through your simulation. Really nice project involving a rich physic !
I think you need to take a step back first. Your model is to complicated to achieve what you looking for : a temperature rise in the rubber fluid caused by the friction

The parameter that you are trying to caracterize is very small compared to other thermal source in your model. What i recommand in the first place is to focus only on the borracha block with his boundary condition. Suppress the rest of your model, it is not worth it for the moment !
By supressing the rest of your model you will be able to allocate more ressource into your ROI and have rapid result in term of the rubber temperature rise. I think you will need a good mesh to have accurate result here.

I was able to see a subtle rise of temperature 85degC to 85.01 degC after 165 second

Futhermore

I can't give you a complete answer but here is what I know :

1. k-epsilon turbulence model is the only model that support effective wall conductivity calculation regarding to the surface roughness. (here is a discussion about roughness and heat transfer)

2. It's not clear what is the roughness parameter that Autodesk use for his calculation but I think it's different from the Ra, Rz or Rrms surface measurement.

Autodesk 2014 Guide :
1. Wall Roughness

2. Roughness Height

Roughness Measurement Conversion
http://ijmem.avestia.com/2012/008.html
http://rsta.royalsocietypublishing.org/content/365/1852/699.full
http://arrow.dit.ie/cgi/viewcontent.cgi?article=1030&context=engschmecart

i find those links in this thread

Hope it helps

FRED

Re: Velocity with very short traces

frederic.gaillard.7 — Mon, 27 May 2019 18:34:06 GMT

Here is the result
Surface TemperatureTemperature inside the pipe

Re: Velocity with very short traces

Anonymous — Tue, 28 May 2019 10:42:18 GMT

Hi @frederic.gaillard.7 ,

Thanks a lot for the information, you're incredible! You know what? I think the the problem is me... I mean, I'm always assuming that I will see an increase of 10 ºC (for example) on the entire flow... But looking at what you sent I think that maybe I'm making a bad assumption. Maybe the rubber actually has a small increase in temperature, and I'm thinking the otherwise.

I'll take a look at your .cfz file and try to do some simulations.

Just a small detail. I don't want to see the temperature increase due to the viscous heating. My main goal is to see best way to homogenize the temperature of the rubber to avoid premature vulcanization... I want to compare the heating of the steel parts with water and with thermal resistances and see what works best.

This question about the viscous heating came in my head because when I was analyzing the results of the simlations I found very very very small increases in temperature and I was counting on seeing more differences 🙂

Thank you so much Fred, your help has been crucial to my project!!! 😄

Re: Velocity with very short traces

Anonymous — Thu, 30 May 2019 08:58:02 GMT

Hi @frederic.gaillard.7 ,

I've been looking at your .cfz file and I've been wondering if by supressing all parts except the fluid, the program will "supress" the effect of friction caused by the surface rougness of the steel walls?

Re: Velocity with very short traces

frederic.gaillard.7 — Thu, 30 May 2019 17:46:00 GMT

Probably but not 100% sure !

Most of the heat generated from the flow came from the internal friction of the fluid. On the previous picture you see a higher temperature in the middle of the duct.

What CFD guideline said about that ?
might be worth it to test it and evaluated his contribution.

Fred

Re: Velocity with very short traces

Anonymous — Fri, 31 May 2019 07:36:53 GMT

Hi @frederic.gaillard.7 ,

Thanks!
That's what I though... But anyway, if this is accurate, I think that I was wrong by thinking that the internal friction would generate relatively high power, but there's a slight increase in temperature.

I think I know why the heat generated by friction might be small! If we look closely, we have a non-newtonian fluid, and since there's higher shear stress near the walls, the fluid will decrease it's viscosity and that should decrease. So the higher viscosity part of the fluid (middle part, far from the wall) will slide on the fluid near the wall (which has low viscosity) and so the rubber near the wall will be almost stationary when comparing with the rubber on the center for example, am I right?
And if I'm right, the friction heat will have a small impact because the fluid velocity on the wall layer will be almost 0 and so, there's no heat generated by the interaction between fluid/solid. This meaning that we will only have internal friction, and as you showed, it will be almost negletable.

Sorry if I didn't explained clearly... I don't know if I'm right, I'm just trying to imagine the flow in my mind according to the results obtained in the simulations.

Here are some picture that show what I was talking about

Shear Stress

Viscosity

Re: Velocity with very short traces

frederic.gaillard.7 — Tue, 04 Jun 2019 13:27:51 GMT

Hello André,

Generally speaking heat disspisation through turbulence is always small no matter the type of fluid your dealing with. The velocity magnitude of your flow will have the greatest influence on the temperature rise of your fluid. Unfortunately i can't comment on the physical explanation of your result, i leave this one to you, but thank for sharing your thought on the matter, really interesting.

Fred

Re: Velocity with very short traces

Anonymous — Tue, 04 Jun 2019 15:12:04 GMT

Hi @frederic.gaillard.7 ,

Thank you for confirming that! People here in the company always mentioned the viscous heating as the major factor to the increase of the temperature on the rubber. However, from what I've study and from the tests I've done, I think that this is a wrong assumption! The rubber doesn't increase that much due to the shear stress!
And today I found something interesting and that might confirm this. When I have news, I will post it 😉

Anyway, thanks for you help, I'll mark it as solves/accepted solution.

Re: Velocity with very short traces

frederic.gaillard.7 — Tue, 04 Jun 2019 15:48:36 GMT

Thanks
keep me posted once you have your result 🙂

Re: Velocity with very short traces

Anonymous — Wed, 12 Jun 2019 12:40:58 GMT

Hi @frederic.gaillard.7 ,

Sorry for the late response but I had a small surgery and I was out of office in the last days.

So basically initially I set the temperature at the inlet of 85ºC based on what the .log file of the machine. But I used a thermocouple to measure the temperature of the rubber and I found that in that place it was 107ºC (approx.) meaning that probably the temperature sensor on the machine is malfunctioning or it's not assembled correctly...
This is a game changer because I was always looking for a temperature increase in the flow to match the real scenario. However with this data I'm almost 100% sure that the rubber is going to cooldown instead of heating.

I've been looking at the CFZ file that you've sent me and I've resumed the steel parts to simulate different surface roughness. But now I'm getting velocity profiles completely different from previous simulations, i.e, the velocity near the walls is greater than the velocity in the middle of the flow (seems not correct).

Basically I wan't to see how much temperature increase actually exists due to the friction on the walls and the viscous heating.

This is a log file. Basically I have two measuring points (1 and 2) and as you can see the point nº2 has a temperature value much different than point nº1. However I'm quite sure that the point n.º1 thermocouple is not giving "real" results. The red line represents the power output of the resistances (Max: 1800W for each one, on/off controller).

Re: Velocity with very short traces

frederic.gaillard.7 — Wed, 12 Jun 2019 13:00:02 GMT

Hello André,

Can you start another thread, with the new input, new data and what parameter you are interested in. It will easier to following up for the forum user 🙂
Thank you very much
Fred