Yplus value won’t drop below 5

pdhyani96 · March 15, 2021, 02:13

I am simulating a steady state flow over a generic turbine nozzle. I started with k-omega model, but my results were fluctuating in grid convergence study. So, I switched to k-epsilon model. Now, I am getting stable results. But no natter how small I keep the first element height, my y plus values won’t go under 5. I have ran multiple simulations with first element offset being a factor of 10 of the largest value.

Can anyone tell why this is happening? Is this a property of k epsilon model?

ghorrocks · March 15, 2021, 03:18

I don't know what it happening, it depends on the details of what you are doing.

The y+ value of the first element is dependant on the mesh you generated and the flow conditions. So if the flow conditions are the same than are you sure the mesh is generating the right sized elements for the y+ values you intend?

Note the k-e turbulence model uses wall functions, and you cannot integrate it to the wall. This means that if you use the k-e turbulence model with y+ values smaller than 11 then you are applying what may be an inappropriate boundary condition.

pdhyani96 · March 15, 2021, 17:56

I create my mesh in Turbogrid. I can change the overall mesh size and the first element size independently. My inlet conditions are the same, so I estimate the first element size with the chord based Reynolds no.

CFX uses automatic wall functions where the solver uses a solver yplus that stays around 11 irrespective of the actual y plus values. So, I was assuming that it would not cause a problem.

Another interesting fact, with k-omega model, the y plus value doesn't drop below 1.8, no matter how small wall element size I use. So, I am confused if this is the geometry that is causing the problems. or something else?

ghorrocks · March 15, 2021, 18:09

Your comment is not quite right. The k-e wall functions model the log-layer in the turbulent boundary layer which lies from y+ = ~11 to a much bigger y+. If your mesh actually has a y+ of < 11 then you are modelling it as if it is the log layer, and not the viscous sublayer which is the actual case.

The result of this is that you are not applying a mathematical model which predicts the viscous sublayer accurately for y+<11. But whether this is significant for your simulation or not depends on what you are trying to do and the accuracy you require. You are correct in saying that for many applications this issue does not create a large enough error to be important.

pdhyani96 · March 15, 2021, 18:12

My application is with jet cooling on turbine blades. And I need to have a yplus that can capture the jet effects. That is the reason why I was using k-omega model, but it has been giving me stability issues.

March 15, 2021, 02:13	Yplus value won’t drop below 5	#1
pdhyani96 New Member DhyaniBaba Join Date: Aug 2019 Posts: 23 Rep Power: 7	I am simulating a steady state flow over a generic turbine nozzle. I started with k-omega model, but my results were fluctuating in grid convergence study. So, I switched to k-epsilon model. Now, I am getting stable results. But no natter how small I keep the first element height, my y plus values won’t go under 5. I have ran multiple simulations with first element offset being a factor of 10 of the largest value. Can anyone tell why this is happening? Is this a property of k epsilon model?

March 15, 2021, 03:18		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,819 Rep Power: 144	I don't know what it happening, it depends on the details of what you are doing. The y+ value of the first element is dependant on the mesh you generated and the flow conditions. So if the flow conditions are the same than are you sure the mesh is generating the right sized elements for the y+ values you intend? Note the k-e turbulence model uses wall functions, and you cannot integrate it to the wall. This means that if you use the k-e turbulence model with y+ values smaller than 11 then you are applying what may be an inappropriate boundary condition. aero_head likes this. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

March 15, 2021, 18:09		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,819 Rep Power: 144	Your comment is not quite right. The k-e wall functions model the log-layer in the turbulent boundary layer which lies from y+ = ~11 to a much bigger y+. If your mesh actually has a y+ of < 11 then you are modelling it as if it is the log layer, and not the viscous sublayer which is the actual case. The result of this is that you are not applying a mathematical model which predicts the viscous sublayer accurately for y+<11. But whether this is significant for your simulation or not depends on what you are trying to do and the accuracy you require. You are correct in saying that for many applications this issue does not create a large enough error to be important. aero_head likes this. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Other] Contribution a new utility: refine wall layer mesh based on yPlus field	lakeat	OpenFOAM Community Contributions	58	December 23, 2021 02:36
yPlus twoPhaseEulerFoam	Andrea_85	OpenFOAM Post-Processing	4	January 11, 2021 05:01
OF 4.0 multiregion case calculate yPlus	pbnuclex	OpenFOAM Post-Processing	6	July 16, 2020 04:27
rhoSimpleFoam yPlus	tmik	OpenFOAM Post-Processing	1	September 16, 2018 15:53
heat transfer coefficients with "bad" yplus	Andrew	Main CFD Forum	4	April 8, 1999 04:43

March 15, 2021, 17:56		#3
pdhyani96 New Member DhyaniBaba Join Date: Aug 2019 Posts: 23 Rep Power: 7	I create my mesh in Turbogrid. I can change the overall mesh size and the first element size independently. My inlet conditions are the same, so I estimate the first element size with the chord based Reynolds no. CFX uses automatic wall functions where the solver uses a solver yplus that stays around 11 irrespective of the actual y plus values. So, I was assuming that it would not cause a problem. Another interesting fact, with k-omega model, the y plus value doesn't drop below 1.8, no matter how small wall element size I use. So, I am confused if this is the geometry that is causing the problems. or something else?

March 15, 2021, 18:12		#5
pdhyani96 New Member DhyaniBaba Join Date: Aug 2019 Posts: 23 Rep Power: 7	My application is with jet cooling on turbine blades. And I need to have a yplus that can capture the jet effects. That is the reason why I was using k-omega model, but it has been giving me stability issues.