[Edgar Yako]

Hello,

I have been designing a radial compressor while trying to achieve certain operational standards. This includes a pressure ratio of 4.5 and an efficiency of at least 80%.

Before I started simulating my radial compressor I did some pre-liminary calculations where I iterated my design parameters (i.e mass flow rate, pressure ratio) to obtain my required power to run the compressor. Once these calculations where done, I used Vista CCD to state my parameters, these were,

Duty and Aerodynamic Data Tab
Overall Pressure ratio :4.5
Mass flow rate : 0.2 kg/s
Rotational Speed : 150000 rpm
Temperature inlet:293.15K
Pressure : 101000 Pa
Isentropic Efficiency: 0.8
(Everything else left default)

Geometry Tab
Hub diameter : 20 mm
Shroud Diameter : 40 mm
Number of vanes : 7 (for both main vane and intervane)
(Everything else left default)

I then passed this into BladeGen and consequently into turbogrid. Here, I just changed within Mesh data, the size factor to 0.5 and the Near wall element size specification method to y+ with a reynolds number of 1.0e6.

After this, I sent my results into CFX where I used turbomode with 300 iterations. Until now, the best results that I obtained were 61% efficiency with a pressure ratio of 2.02. I obtained these by using the tables in CFD-Post. The worst results were obtained with a number of blades of 11.

I don't know where the problem lies. Is the lack of inefficiency coming from using default parameters along the process? Or is the problem related to my geometry dimensions of the compressor? Mass flow rate/Pressure ratio? Should I improve the blade design?

Maybe I am missing something very obvious but I don't know where to look first. Until now, the only parameter that has made a difference is the number of blades. I have attached an image with both Pressure/Velocity distributions along the blades.

Thank you for your time
Edgar

[ghorrocks]

A good place to start is the FAQ: https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F

[Edgar Yako]

Thank you for the link, it looks helpful.

Sorry, isn't this the CFX results section? On Which section should I post then?

[ghorrocks]

Yes, this is the CFX section. But the FAQ covers all the basic accuracy checks you should first as they are standard practise.

So you should first confirm your simulation is accurate with the techniques described in the FAQ. In particular check your mesh and convergence criteria.

If you show the simulation is accurate then your design is the problem. You then look at the simulation results in detail and try to work out where your inefficiency is coming from. But this only makes sense once you have an accurate simulation.