Discussion Groups

Simulation Mechanical and Multiphysics

Active Member
7 Posts
0 Kudos
Registered: ‎10-23-2011

Choosing the proper fluid flow for flowbench replication...

132 Views, 3 Replies
04-24-2013 01:14 PM

Alright, I would like to replicate a cylinder head flow test to analyze intake and exhaust runners of a cylinder head design. I'm not a complete novice to the software but would like to start from base 1 for guidance. The flowtest uses a 28in H2O depression across the cylinder head to produce air flow. The flowrate values get as high as 650cfm through an inner diameter of 4.5in resulting in turbulent air flow. Also when testing, the measured air velocity will always reach a maximum value and stay centered on that value. I have completed a solid model of intake port, combustion chamber, intake valve, and a small portion of the cylinder bore. 


Proposed Boundary Conditions:

Inlet: Zero Pressure, Zero Velocity, Enter the Prescribed Turbulence Entry Conditions

Exit: 28inH20 (about 1.01psi)  of depression, Turbulence Exit Conditions


Intended Output Data:

1. Air Velocity at exit

2. CFM  at exit



Boundary Layer, Exclude Inlet/Outlet from mesh



0. Are these boudary conditions okay

1. Should any changes be made to the solver or is there another solver that should be used?

2. Should I use Steady of Unsteady Analysis?

3. When inputting Analysis Parameters, how should a structure my load curve? What values and how many time steps should I input?

Note: I know that the curve of the air as it reaches its maximum velocity will look like an inverse function(x^1/4 or so). Ramps fast then slows as it approaches maximum velocity.


4. With the steady analysis parameters, can you offer specific instruction as to what values would be good to start with in the time step boxes? Don't quite understand exaclty what I should input here.

The average air velocity values attained should be between about 720-3800in/s or little more if that helps in structuring the proper inputs for the steady analysis.


5. I'm an engineer but this program can often times go over my head with its nuances so if you would be detailed and  clear with you explanations. Please don't assume my intution will naturally take over as it fails us all in moments everyday. Thanks for the help very much, it is appreciated.



Distinguished Contributor
141 Posts
34 Kudos
Registered: ‎06-13-2012

Re: Choosing the proper fluid flow for flowbench replication...

04-26-2013 01:36 AM in reply to: aparadis

Hi, what version of the software are you using? Here's what I think,


0. The Zero Velocity BC looks fishy. How would the fluid enter the domain then? Are there multiple inlets? BCs play a very important role in defining your analysis, as well as helping convergence. Someone once told me that the ideal BC for Multiphysics is a velocity inlet and pressure outlet.

2. Ideally, choosing steady state or transient would be based on the intended analysis. If the intent is to study the flow after a long period of time, then steady stateit is. If you intend to study how the flow develops over time, then choose transient. I'm sure this is obvious, however, if you have convergence issues for steady state analysis, such as the flow would not reach steady state due to oscillation, then consider performing a transient analysis. If you do this, ramp up you load curve from zero then leave it constant for some time to let the flow stabilize.

3. If it were me, I would ramp up constantly from zero. If the analysis faces convergence issues at 20%, I would lessen the slope around that region or increase the number of time steps. Basically its trial and error for me.


I don't have answers to all your questions, but I hope this helps.




Active Member
7 Posts
0 Kudos
Registered: ‎10-23-2011

Re: Choosing the proper fluid flow for flowbench replication...

04-26-2013 06:40 AM in reply to: ilyas

Thanks for the help!


In response to the zero velocity, it wouldn't be zero, but it should be very close to that since the vacuum is initiating air flow. I thought it would be safe to assume a zero initial value because of that. I'm guessing it would be better to just assume a value close to zero in the direction of flow, say 0.1in/s or so. 


I'm actually not as interested in how the flow develops as much as I am in seeing what the flow is doing once it reaches its steady state velocity. That gives me a better picture of what the air is doing at its peak steady state velocity(when the valve opens to a certain distance). 


I did play with the Steady State anlysis yesterday and one of the problems I have is that it doesn't seem to complete analysis without doing much. It will either fail to solve where the anylsis window doesn't run or if I have say 10 steps, it will complete its analysis in 1 step. That seems odd given that fact that most of these softwares like to run for 10min or longer which is why I had questions about setting up the load curve for the steady analysis. I will do a little bit more with it this weekend to see if  I can make any progress. It gets hard to try to sort things out during the work week, just not enough time. 


As far as the model, there is one openinig, and one outlet. I have attached a schematic that gives a better picture of whats going on. 


Active Member
7 Posts
0 Kudos
Registered: ‎10-23-2011

Re: Choosing the proper fluid flow for flowbench replication...

04-26-2013 07:02 AM in reply to: aparadis

Oh, I'm using Autodesk Simulation Multiphysics 2013. Trying to learn all I can before applying this knowledge to CFD 2014  since fluid flow will be removed in the the 2014 version of Simulation MP.