Hello,

I am trying to optimize a centrifugal fan impeller in CFD with a set inner radius, outer radius and angular velocity. I am dubious to include any form of volute or casing in my model as I do not want to influence the performance of the impeller as much as possible. My understanding is that the following equation can be used to evaluate the performance of the fan design.

Efficiency = Q Î"P/W

Q = Volumetric flow rate (m/&rho Î"P = Total pressure increase across the fan (P2-P1) W = Work (ωT)

My question is where exactly upstream and how far down stream should the pressures be measured? I am curious because I am currently measuring just up stream of the impeller leading edge and directly aft of the impeller trailing edge. I am worried that measuring the total pressures at these locations is over predicting fan performance. My models are suggesting forward-curved blades are superior to backward-curved blades which is not the case according to literature.

I would really appreciate some guidance as to how the optimization process of an impeller design should be conducted.

Thank you,

Simon

I think static pressure will give more realistic values both in up and downstream. Since total pressure has kinetic terms, statos pressure only considered for calculating efficiency.

I have seen fan efficiency calculated in this way, unfortunately I disagree. Using only static pressure does not account for the total energy loss through the system and therefore will not result in a true efficiency, that is work output over work input.

I originally used this method though the results simply did not follow any logical pattern.

I'm working at a place where the product we make is basically fan generated air flow into a tube. I've had to do a bunch of work with fans recently, and it's more complicated than just modeling the fan. The scroll casing is very important to the performance. So is the inlet cone - a very small change has a huge effect on performance. As far as measuring output, again, it's all about what it's going into. It seems an accurate solution would be to duplicate the way it gets measured with a pitot tube.