[s__s__s]

Hi Dear CFD Online members. I am currently preparing my master thesis about design and CFD simulations of a centrifugal compressor with vaneless diffuser and volute. I am using one single passage of rotor and whole geometry of diffuser and volute with stage interface. So far I got very promising converged results with exit corrected mass flow rate. But as it is expected, when i use the same corrected mass flow rate in different meshes, the outlet mass flow rate is different. I thought it can not be relatively compared. I have 3 questions right now:

1- After it converges with corrected mass flow rate I can apply mass flow rate as BC and it converges with exact mass flow, so can I use these results in mesh sensitivity study? Will it be trustable? (corrected mass flow rate gives 0.742... kg/s and mass flow rate at design point normally is 0.75 kg/s)

2-Should i find a way to converge the simulations directly by using mass flow rate?It was giving FINMES error thats why i prepared it via corrected mass flow rate without any convergence problem.Now I am trying to use small physical time steps (1e-6) to get converged results with mass flow BC instead of corrected one.(so far could not succeed)

3- If there is no way to use corrected mass flow rate in mesh sensitivity, what is the advantage of it in practical sense (except the theoretical advantages in user guide)?

Thank you very much in advance for your support. I have always checked the discussions in here so far to solve problems. You are doing great job here...

[Opaque]

The exit corrected mass flow rate boundary condition has two main advantages,

1. you can compute every point in the speed line of the machine w/o having to change the outlet boundary condition for specified pressure near choke, or specified mass flow near stall/surge. Using either of the last two will give you convergence problems as you approach the opposite regime

1. it is convenient to specify corrected mass flow directly than having to pre-computed the true mass flow before setting up the run.

On your observations, you are concerned about the true mass flow changing as you refine your mesh. I do not see that as a problem as long as it approaches an asymptotic value as the mesh is refined further. It is no different if you monitor the losses, enthalpy change, or efficiency of the device.

Hope the above helps,

[s__s__s]

Originally Posted by Opaque

The exit corrected mass flow rate boundary condition has two main advantages,

1. you can compute every point in the speed line of the machine w/o having to change the outlet boundary condition for specified pressure near choke, or specified mass flow near stall/surge. Using either of the last two will give you convergence problems as you approach the opposite regime

1. it is convenient to specify corrected mass flow directly than having to pre-computed the true mass flow before setting up the run.

On your observations, you are concerned about the true mass flow changing as you refine your mesh. I do not see that as a problem as long as it approaches an asymptotic value as the mesh is refined further. It is no different if you monitor the losses, enthalpy change, or efficiency of the device.

Hope the above helps,

Firstly thank you very much Opaque for reply in such short notice. Precisely my concern is about the mass flow rate because I designed everything myself by preparing an excel sheet starting from pressure ratio and mass flow rate; so i need to compare my results with my expected values at the design point, and mass flow rate is a strict design constraint. I was okay with such a small difference between 0.7425 and 0.75 but when I see all of the different meshes are giving as it expected different results for mass flow rate I got sceptical whether they can be compared. So finally if i understood right, I can compare the pressure, temperature and velocity of these results for mesh sensitivity, as soon as I use the same corrected mass flow rate for all by ignoring small differences in results of mass flow rate?

Bythe way I setup the simulations with corrected mass flow rate first, then after convergence I changed the BC into mass flow rate to force them converge exactly at design point and run for some more iterations. Is it a wrong approach?

[Opaque]

First, we must understand that solutions will change with mesh refinement until it becomes mesh independent. If you feel uncomfortable with the differences as you refine the mesh, you will need to refine further at a cost (longer simulations).

On the difference between the calculated mass flow and the expected mass flow computed corrected mass flow, it must be related to variables used to convert between the two. They are not fixed, and depend on the solution itself, correct ? So, the approach of using the exit corrected mass flow solution as an initial guess to the calculation using the specified mass flow is good; however, subject to failure if you are too close to choke.

Recall that near choke any minor variation in mass flow represents a major variation in pressure ratio, efficiency, etc (steep slope in the speed line) and the solution algorithm can become unstable.

Hope the above helps,

[s__s__s]

Originally Posted by Opaque

First, we must understand that solutions will change with mesh refinement until it becomes mesh independent. If you feel uncomfortable with the differences as you refine the mesh, you will need to refine further at a cost (longer simulations).

On the difference between the calculated mass flow and the expected mass flow computed corrected mass flow, it must be related to variables used to convert between the two. They are not fixed, and depend on the solution itself, correct ? So, the approach of using the exit corrected mass flow solution as an initial guess to the calculation using the specified mass flow is good; however, subject to failure if you are too close to choke.

Recall that near choke any minor variation in mass flow represents a major variation in pressure ratio, efficiency, etc (steep slope in the speed line) and the solution algorithm can become unstable.

Hope the above helps,

Thanks Opaque for the reply again. I have surprisingly got fully converged results with mass flow rate BC directly (1e-5 to 1e-6 in residuals tightly converged and also interrupt control with less than 1e-3 in efficiency and 1e-4 in pressure ratio changes) by modifying the turbo setup feature recommended values in time step. If I can complete these study with mass flow BC it will be more handy for my thesis. Because I dont need the performance assessment out of the design point. If I face with problems anyway, I will use corrected mass flow rate and use refined meshes till a tolerance reached in mass flow rate result. I totally got what you mean as you said it is a formula manipulate the mass flow rate so it is natural to have changes in mass flow rate by mesh refinement. Mass flow is nothing more than a variable like pressure or velocity anymore. Thank you very much for your support. I wish you the best.

Sedat