[Yanlu]

Hello everyone, I am curious about how the velocity and pressure are coupled in ANSYS CFX. Can I change to other solutions? When I verified the mesh independence, I found that after the number of my meshes increased, the pressure changed significantly, while the velocity and flow rate did not have a big effect. I want to know why this is the case.

[ghorrocks]

In CFX the pressure-velocity coupling is inside the matrix, as it is a coupled solver. In other words, the velocity and pressure equations are all solved in the same matrix. This means no coupling step needs to be done, it is taken care of.

You cannot turn this off in CFX and use an uncoupled approach (eg SIMPLE). If you want to try an uncoupled approach you will need to use Fluent, StarCD or a different solver.

[Yanlu]

Originally Posted by ghorrocks

In CFX the pressure-velocity coupling is inside the matrix, as it is a coupled solver. In other words, the velocity and pressure equations are all solved in the same matrix. This means no coupling step needs to be done, it is taken care of.

You cannot turn this off in CFX and use an uncoupled approach (eg SIMPLE). If you want to try an uncoupled approach you will need to use Fluent, StarCD or a different solver.

Thank you for your answer. So how should I adjust it when my pressure distribution is abnormal? In my three cases, the only difference is the mesh size, and the rest of the settings are the same.

[ghorrocks]

Then it sounds like your mesh is too coarse, you need to refine it further. For some general tips see FAQ: https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F

[Yanlu]

Originally Posted by ghorrocks

Then it sounds like your mesh is too coarse, you need to refine it further. For some general tips see FAQ: https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F

In fact, my velocity, flow rate, and wall shear stress are all consistent. And the convergence performance of the three grids is excellent. However, the average outlet pressure I monitor is not just a numerical increase or decrease, but a change in distribution trend. So I am wondering whether the pressure is automatically corrected during the coupling of velocity and pressure. Why does the pressure have a different trend when my velocity remains unchanged?

[ghorrocks]

Pressure changes can be caused by very small changes in the velocity field in some cases.

Can you post images of what you are seeing, including something which shows the problem you are having, your geometry and attach your output file.

[Yanlu]

Originally Posted by ghorrocks

Pressure changes can be caused by very small changes in the velocity field in some cases.

Can you post images of what you are seeing, including something which shows the problem you are having, your geometry and attach your output file.

I have added my grid diagram and the monitoring curves of the three sets of results. You can see that the pressure trends are very different. As for the out file, it is too large and the website does not support uploading.

90W.jpg

160W.jpg

371W.jpg

Mesh.jpg

[ghorrocks]

The website does support uploading, it just has a maximum file size. The CCL section at the top of the output file is the most important bit, cut that out and post that as an attachment please.

Thanks for the image of your geometry, but I have no idea what those graphs are showing. You will have to explain what we are looking at.

[Yanlu]

Originally Posted by ghorrocks

The website does support uploading, it just has a maximum file size. The CCL section at the top of the output file is the most important bit, cut that out and post that as an attachment please.

Thanks for the image of your geometry, but I have no idea what those graphs are showing. You will have to explain what we are looking at.

There is a picture showing my geometry and meshing strategy. The other three pictures are the test results of three different meshes, including the average pressure, average velocity and flow distribution at the inlet and outlets, and the wall shear stress of the total wall. I hope to use the wall shear stress to verify the mesh independence. However, as shown in these figures, my pressure distribution varies greatly.

[ghorrocks]

I see from your CCL that this is a transient model, you have complex functions controlling some boundary conditions and the fluid is blood, which you model as non-Newtonian. How are we expected to help you when you don't even tell us these critical and basic information?

Please explain what you are doing, what you are modelling, how you are modelling it and what the problem you are getting is. Your previous post shows many charts but I have no idea what any of them mean. I see your geometry and mesh, but what are the boundary conditions you are using on it?

[Yanlu]

Originally Posted by ghorrocks

I see from your CCL that this is a transient model, you have complex functions controlling some boundary conditions and the fluid is blood, which you model as non-Newtonian. How are we expected to help you when you don't even tell us these critical and basic information?

Please explain what you are doing, what you are modelling, how you are modelling it and what the problem you are getting is. Your previous post shows many charts but I have no idea what any of them mean. I see your geometry and mesh, but what are the boundary conditions you are using on it?

Hello, in my last reply which contains the ccl file, I have explained in detail the four pictures I uploaded before. My simulation is an ideal model of the human vertebral artery. It is a transient simulation, the inlet is a periodic flow inlet, and the outlet condition is related to the pressure and mass flow of the last timestep. For more specific numerical boundaries, please refer to my previously published paper: https://doi.org/10.1063/5.0206906
The simulation in my paper and the new simulation I am having problems with now only modified the geometric model and the inlet flow, but now I have encountered such a problem. I am still looking for why the problem occurs.

[ghorrocks]

OK, so back to your original question - if the pressure distribution at the outlet is important to you and it is changing with mesh refinement, then you need to do a mesh refinement study to refine it further until you get a mesh independent solution. The fact that other parameters are stable is good, but until all parameters of interest are stable then you have not achieved mesh independence.

Note that as you refine the mesh you will need to change the time step as well. The easiest way to handle this is to use adaptive time stepping, homing in on 3-5 coeff loops per iteration. Then the solver will find its own time step size.

[Yanlu]

Originally Posted by ghorrocks

OK, so back to your original question - if the pressure distribution at the outlet is important to you and it is changing with mesh refinement, then you need to do a mesh refinement study to refine it further until you get a mesh independent solution. The fact that other parameters are stable is good, but until all parameters of interest are stable then you have not achieved mesh independence.

Note that as you refine the mesh you will need to change the time step as well. The easiest way to handle this is to use adaptive time stepping, homing in on 3-5 coeff loops per iteration. Then the solver will find its own time step size.

Can you explain how to set the adaptive time step?

[ghorrocks]

It is in the time step settings, when you choose between a steady state and transient simulation. When you select a transient simulation it is one of the time step options.

[Yanlu]

Originally Posted by ghorrocks

It is in the time step settings, when you choose between a steady state and transient simulation. When you select a transient simulation it is one of the time step options.

I found this setting and I will try to use it. Thank you！