I have set up a model to simulate airflow through a canopied generator set.
The model is very simple with a fan pulling airflow through inlet ducts, over an engine, through a cooling core and out a 90 Degrees vertical outlet duct at the top of the canopy.
The geometry has been set up with 0 gage pressure only as a boundary condition, the fan has a fan curve from the supplier and the cooling core has an air restriction curve from the supplier.
I am looking to run an accurate model as possible but due to computational restraints, cubes represent engines etc.
The outcome I am looking for is a pressure restriction of the canopy and a volumetric flow at the outlet.
I know there is a summary report, but is there a way to set up a plane to show volumetric flow at the outlet? Or the pressure drop through the system?
Also, I know increasing meshing size will help accuracy of the model, but is there any other tips/help more experienced users can offer to better setup this model. For example, does setting a free air ratio as a resistance instead of a restriction curve better represent reality within Autodesk Simulation?
Thanks in advance
Hi,
Sounds like you have the right approach, there are a few things to mention that might help:
Thanks Mr. Wilde,
Does suppressing solid parts just stop heat transfer through them? Do they not have to be meshed for the CFD program to ascertain aiflow patterns around/ past/over solid objects?
Would this mean the only meshing I would be applying would be to the fan and the cooling core?
If I am modelling a fan, is ther any rules of thumb for the modelling? Currently I am using a cylinder, with no inner fixed hub modelled. It is set to the thickness of the suppliers drawing from one side of a blade tip to the other.
I feel I will have to play about with the resistive regions, at the moment I have a grill outlet setup as a pattern from the model which is creating alot of nodes. I think it might be better to set this as a resistive region with a free air ratio.
Thanks.
Not a problem 🙂
Does suppressing solid parts just stop heat transfer through them? Do they not have to be meshed for the CFD program to ascertain aiflow patterns around/ past/over solid objects? You are correct, it will just stop the heat transfer, the flow patterns around them would not be affected.
Would this mean the only meshing I would be applying would be to the fan and the cooling core? And the air, yes.
If I am modelling a fan, is ther any rules of thumb for the modelling? Currently I am using a cylinder, with no inner fixed hub modelled. It is set to the thickness of the suppliers drawing from one side of a blade tip to the other. What you have souds sensible, although it is worth including a solid part around the outer diameter to ensure that no flow can leak. also ensure it has a good uniform mesh.
I feel I will have to play about with the resistive regions, at the moment I have a grill outlet setup as a pattern from the model which is creating alot of nodes. I think it might be better to set this as a resistive region with a free air ratio Yes, this might be worthwhile. You can set the resistance to be different in each of the x.y,z directions to try to mimic the grill pattern.
Hi,
Many thanks for the reply.
I just have a couple more questions.
1. When I am modelling the fan, is it better to have a solid cylinder in the center to represent the hub? I have set a solid exterior to stop flow leak.
2. I have set up the geometry as a 0 Pa gage pressure. Is there any need to specify inlets as 0 Pa gage pressure as well?
Thanks in advance.
1. When I am modelling the fan, is it better to have a solid cylinder in the center to represent the hub? I have set a solid exterior to stop flow leak - It does not really matter althuogh you can do - suppress it from the mesh if you do.
2. I have set up the geometry as a 0 Pa gage pressure. Is there any need to specify inlets as 0 Pa gage pressure as well - I am unsure what this means. You only want to assign a p=0 to the fluid where there is an opening and only on the boundary of the model - nothing internally
Does that make sense?
Thanks,
Jon
Hi,
We usually have the slip at about 0.7.
ADV5 is normally the best scheme for this type of model too.
Do you find that the model eventually converges? Perhaps you need more mesh or difference BC's?
Hi John,
Thanks for replying.
I am using CFDesign 2011 which doesn't have ADV5. The meshing is as fine as can be for the fan.
The model eventually converges but the summary data does not read correctly - Please see attached screenshot. This is with the ADV set to 1 and the slip set to 1.0 (I stopped this into the run due to the convergence plot display). I have tried this with alip at 0.25, but the outcome is the same (Iterations converges around 1100).
The only boundary condition I have set is 0 Pa for the four walls and roof of the surrounding space. No other BC has been set.
Hi,
I would certainly say that ADV2 would be best then.
This convergence plot does not suggest convergence, if you ever see anything like this, just stop the model as it is not right.
Do you have some thin parts in there too (looks like you may do), resitances maybe? If so, are you using surface or volume resistances? Is the mesh tight on these also? 3-4 elements inlet to outlet (same as the fan)?
How large is the overall domain?