Sign In
Forgot Password?
Sign In | | Create Account

Fan Selection Utilizing a Closed-Form Solution for Blade Chassis



A logical checklist of allowable specifications, combined with appropriate analysis, can determine which fan is the best for your application.

There are many factors that go into fan selection for fan array applications. Among the many factors are fan size and arrangement, to overall airflow and fault tolerance.

What You Will Learn

  • How to confidently choose the best fan for your chassis through a structured analysis that systematically compares the gamut of factors to consider when deciding which fan to use.

Who Should Attend

  • Design Engineers
  • Mechanical Engineers
  • System Architects
  • Product Managers

About the Presenter

Presenter Image Joe Proulx

Joe Proulx has worked for Mentor Graphics Mechanical Analysis, formerly Flomerics, as a consultant engineer since July 2005 and is currently an application engineer. He also participates in training, customer support and consulting.

Joe's prior employment was with Sycamore Networks where he was responsible for all aspects of designing the cooling of the company's products. Along with the use of thermal simulation software, Joe was able work in a thermal lab complete with a thermal imaging camera, wind tunnel, multiple hot wire anemometers, thermocouples and other ubiquitous lab equipment.

Before that, Joe was employed by Degree Controls as thermal consultant. Degree Controls is a complete thermal solutions provider. Joe was responsible for handling customer's thermal needs from computational thermal analysis to physical flow and temperature measurement.

Prior to his position at Degree Controls, Joe was employed by Cabletron Systems in Rochester, NH. While there, Joe worked as a thermal analyst utilizing FloTherm and a complete thermal lab.

Joe graduated from the University of New Hampshire in 1995 with a Bachelor of Science degree in mechanical engineering.

Related Industries

Related Resources


Other Related Resources

Q and A Transcript

Q: What about operating at the fan centrex?
A: The level of detail of this approach was simplified a bit in order to compare the fans as quickly as possible. Once a fan has been selected, it would be prudent to pursue the robustness of the design and understand where all of the individual fans in the system are operating on the fan curve and determine if there will be any problems.
Q: Why did you ignore the heat produced by the fans when comparing the push/pull?
A: Good point. That is worth considering but what I've seen is that the power compared to the total flow of air creates a very small temperature rise.
Q: What was the equation for total acoustic noise of multiple fans in parallel? I wasn't fast enough in copying it down.
A: You can download a pdf of the presentation and it will be in there. Too hard to type the equation in this space.
Q: Can CFD give system resistance curve or is experimentation necessary ?
A: CFD can generate a system resistance curve. Just set the experiment up virtually, as you would in a lab on a wind tunnel. I would say that virtually is much easier.
Q: You used 2 resistor models ? How accurate are they for detailed analysis?
A: I only used them to avoid having a block which can have a varied temperature throughout. I wanted it to be more like a node instead. As far as their accuracy a 2r is better than a block but certainly not as good as a multi-resistor component or a detailed component.
Q: At some point of the presentation, can you do a quick check of the System Pressure Drop and compare with the Fan curve to see where the Operating point is?
A: You'll see this on slide 14
Q: What was the benefit of (4) data points on the fan curve? How did you use the system curve?
A: The 4 points are simply to reduce the complexity of the fan curve. As you know, fan curves are made of many points. To quickly handle the data of many fans I have elected to boil the curves down to 4 points. If the engineer felt that more detail was necessary, then of course it should be added. I’m of the opinion that this 4 point technique will give the designer the ability to decide which fan is better.
Q: Most of the times, vendors only give information enough for a 2-R model only. In those cases what is the best way to get a good model?
A: The 2r I was using was not a particular component, only values I picked based on the general values I have seen. As far as a discussion on representing actual components, that’s well beyond the scope of this seminar. But, I will say, the most accurate way to model a component is in detail. If the vendor does not provide the geometries, opening and recreating the device might work. Or, use a product like FloPACK that will help you get a good representation of any standard chip package without a lot of work and effort.
Q: Have you got a good correlation between the virtual and real system curves?
A: Yes
Online Chat