High Turbulent Dissipation Residual at Domain Inlet

Asto1202 · July 1, 2024, 11:42

Hello,
I am trying to run an external hydrodynamics simulation for a stationary vehicle in subject to incoming flow at 12.3 m/s. The working fluid is water and the length of the vehicle is ~2.7m giving a length based Reynolds number of 3.2E+7.

I am running constant density, K-E, with Y+ of 50 over the the vehicle body, and many loads of interest (drag, lift, moments) seem to be converged.

I am having an issue with my turbulent dissipation rate residual converging. When I am visualizing where the residuals are high with thresholds and surfaces in a scene, I notice that the very high residuals are located in the wake of the vehicle and separation regions (as expected), but also at the inlet of the fluid domain, where the mesh becomes courser (see pictured).
TDR Visualization.png
Domain Mesh Visualization.jpg
TDR Threshold.jpg

I have played around with turbulence parameters (in the simulation pictured intensity = 0.6, viscosity ratio = 50, turbulent length scale = 2.7m and velocity scale =12.35) without much luck and was hoping someone knows how to solve this issue.

Happy to provide more info and screenshots as needed. Thanks!

ym92 · July 2, 2024, 10:03

Hi,

in Star-CCM+, you do not provide the turbulence intensity in %. So, if you choose 0.6, you use a turbulence intensity of 60%, which is certainly too high for external flows. Not sure if this is the problem, but try with 0.006 instead (or whatever is more suitable for this type of simulation).

Asto1202 · July 2, 2024, 12:14

Hi ym92,
My mistake when writing the post: the turbulent intensity is set to .06 (6%, not 60%). The rest of the info should be correct.

ym92 · July 3, 2024, 04:31

Hi Asto,

ah alright. Anyway, have you checked if your inlet values are reasonable if you use a calculator to estimate turbulence properties (e.g., http://www.wolfdynamics.com/tools.html?id=110). I am not sure what values are appropriate for your kind of simulation. But using your values for velocity and intensity/length scale gives huge values for the viscosity ratio.

Are you simulating a boat (or whatever vehicle) with 12.35m/s speed and stationary water or is the water really flowing at 12.35m/s and the boat is stationary?

Asto1202 · July 3, 2024, 09:02

Hi Ym,
Thanks for sharing that parameter calculator. I will attempt to play around with my turbulence parameters based on the calculator.

I now notice my turbulence length scale might not be reasonable for this simulation. The body length of the vehicle of interest is L = 2.7m, so the turbulence length scale might need to be Tu_l = 0.7*L = 0.203m.

This still yields a high turbulent viscosity ratio 1.84E5.

As for the flow scenario, I am looking at a vehicle attached to the hull of a ship, so from the point of view of the vehicle, the ground is stationary and the incoming flow is 12.35 m/s.

From what I have read in literature, high Re near wall turbulence modelling is quite computationally challenging. I'm curious to what degree I can trust my results even with a converging solution.

ym92 · July 3, 2024, 09:16

Regarding the scenario: from the point of view of the vehicle, the incoming flow might be at 12.35 m/s, however, if, in fact, the water is standing still (e.g., in a pool), the turbulence properties of the water at the inlet should not be calculated based on a velocity of 12.35 m/s. So be careful what values you are using

.

Asto1202 · July 3, 2024, 10:37

Do you have recommendations for what my velocity scale should be for this case?

One site recommended using the skin friction velocity back calculated based on the ship length based reynold's number. This gives me velocity of .3 m/s and with turbulent length scale of .203 m, yields a viscosity ratio of ~4500.

I'm finding that with high viscosity ratio my solution diverges quickly. Do you have any recommendations in this case?

ym92 · July 3, 2024, 10:45

To be honest, I don't have much experience with these kind of scenarios. Probably someone else. Otherwise I would recommend checking the viscosity ratio (and possibly k and E) somewhere in your domain (away from the inlet and the ship). From that you could back calculate reasonable inlet conditions. But that only works if you are not too far off with your IC and BC.

July 1, 2024, 11:42	High Turbulent Dissipation Residual at Domain Inlet	#1
Asto1202 New Member Join Date: Jul 2024 Posts: 5 Rep Power: 2	Hello, I am trying to run an external hydrodynamics simulation for a stationary vehicle in subject to incoming flow at 12.3 m/s. The working fluid is water and the length of the vehicle is ~2.7m giving a length based Reynolds number of 3.2E+7. I am running constant density, K-E, with Y+ of 50 over the the vehicle body, and many loads of interest (drag, lift, moments) seem to be converged. I am having an issue with my turbulent dissipation rate residual converging. When I am visualizing where the residuals are high with thresholds and surfaces in a scene, I notice that the very high residuals are located in the wake of the vehicle and separation regions (as expected), but also at the inlet of the fluid domain, where the mesh becomes courser (see pictured). TDR Visualization.png Domain Mesh Visualization.jpg TDR Threshold.jpg I have played around with turbulence parameters (in the simulation pictured intensity = 0.6, viscosity ratio = 50, turbulent length scale = 2.7m and velocity scale =12.35) without much luck and was hoping someone knows how to solve this issue. Happy to provide more info and screenshots as needed. Thanks!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Segmentation fault when using reactingFOAM for Fluids	Tommy Floessner	OpenFOAM Running, Solving & CFD	4	April 22, 2018 12:30
chtMultiRegionSimpleFoam turbulent case	Aditya Patil	OpenFOAM Running, Solving & CFD	6	April 24, 2017 22:13
simpleFoam error - "Floating point exception"	mbcx4jc2	OpenFOAM Running, Solving & CFD	12	August 4, 2015 02:20
Unstabil Simulation with chtMultiRegionFoam	mbay101	OpenFOAM Running, Solving & CFD	13	December 28, 2013 13:12
Upgraded from Karmic Koala 9.10 to Lucid Lynx10.04.3	bookie56	OpenFOAM Installation	8	August 13, 2011 04:03

July 2, 2024, 10:03		#2
ym92 New Member Yannick Join Date: May 2018 Posts: 16 Rep Power: 8	Hi, in Star-CCM+, you do not provide the turbulence intensity in %. So, if you choose 0.6, you use a turbulence intensity of 60%, which is certainly too high for external flows. Not sure if this is the problem, but try with 0.006 instead (or whatever is more suitable for this type of simulation).

July 2, 2024, 12:14		#3
Asto1202 New Member Join Date: Jul 2024 Posts: 5 Rep Power: 2	Hi ym92, My mistake when writing the post: the turbulent intensity is set to .06 (6%, not 60%). The rest of the info should be correct.

July 3, 2024, 04:31		#4
ym92 New Member Yannick Join Date: May 2018 Posts: 16 Rep Power: 8	Hi Asto, ah alright. Anyway, have you checked if your inlet values are reasonable if you use a calculator to estimate turbulence properties (e.g., http://www.wolfdynamics.com/tools.html?id=110). I am not sure what values are appropriate for your kind of simulation. But using your values for velocity and intensity/length scale gives huge values for the viscosity ratio. Are you simulating a boat (or whatever vehicle) with 12.35m/s speed and stationary water or is the water really flowing at 12.35m/s and the boat is stationary?

July 3, 2024, 09:02		#5
Asto1202 New Member Join Date: Jul 2024 Posts: 5 Rep Power: 2	Hi Ym, Thanks for sharing that parameter calculator. I will attempt to play around with my turbulence parameters based on the calculator. I now notice my turbulence length scale might not be reasonable for this simulation. The body length of the vehicle of interest is L = 2.7m, so the turbulence length scale might need to be Tu_l = 0.7*L = 0.203m. This still yields a high turbulent viscosity ratio 1.84E5. As for the flow scenario, I am looking at a vehicle attached to the hull of a ship, so from the point of view of the vehicle, the ground is stationary and the incoming flow is 12.35 m/s. From what I have read in literature, high Re near wall turbulence modelling is quite computationally challenging. I'm curious to what degree I can trust my results even with a converging solution.

July 3, 2024, 09:16		#6
ym92 New Member Yannick Join Date: May 2018 Posts: 16 Rep Power: 8	Regarding the scenario: from the point of view of the vehicle, the incoming flow might be at 12.35 m/s, however, if, in fact, the water is standing still (e.g., in a pool), the turbulence properties of the water at the inlet should not be calculated based on a velocity of 12.35 m/s. So be careful what values you are using .

July 3, 2024, 10:37		#7
Asto1202 New Member Join Date: Jul 2024 Posts: 5 Rep Power: 2	Do you have recommendations for what my velocity scale should be for this case? One site recommended using the skin friction velocity back calculated based on the ship length based reynold's number. This gives me velocity of .3 m/s and with turbulent length scale of .203 m, yields a viscosity ratio of ~4500. I'm finding that with high viscosity ratio my solution diverges quickly. Do you have any recommendations in this case?

July 3, 2024, 10:45		#8
ym92 New Member Yannick Join Date: May 2018 Posts: 16 Rep Power: 8	To be honest, I don't have much experience with these kind of scenarios. Probably someone else. Otherwise I would recommend checking the viscosity ratio (and possibly k and E) somewhere in your domain (away from the inlet and the ship). From that you could back calculate reasonable inlet conditions. But that only works if you are not too far off with your IC and BC.