Axial Flow Supercharger

[devincox]

Has anyone considered making a model-sized version of an Axial Flow supercharger that could actually produce even a tiny amount of boost.
I saw a full sized verision at the links below, that is large enough for a car though still very small in size:

http://www.axialflow.com/products.htm
http://www.axialflow.com/engineering.htm

This looks like it would be a fun project. It wouldn't have to produce massive boost for it to be a worthwhile project, just a measureable bit.

I was just curious if anyone knew how complex this project might prove to be and some of the engineering problems to overcome. ie. The best way to mount the rotors and stators on a common shaft. Optimum blade size. Number of stages. etc.

I have a 4 axis sherline mill to cut the rotors and stators


Thanks,
Devin

[fyffe555]

interesting.. Is it just me or does something seem a little odd with this device?

Well designed and constructed Axial compressors can have adiabatic efficiencies of up to 80% if the path and the blades are well designed. However they can only develop pressure ratios of around 1.1 to 1.4 per stage so that's why this device has four stages.. Given this isn't General Electric lets say its 1.2 PR then the max pressure it can develop is twice ambient pressure (1.2^4) at ideal speeds. Now this device is an axial flow compressor which, because of the drive shaft turns the flow to radial at the end of the unit. That's going to increase pressure and reduce flow at the final stage but reduce the over all PR.

Centrifugal compressors can get around 2.8 pressure ratios at 85% which is why they are used so often in engine boost applications. Per volume of air and similar PR's a centrifugal compressor will take less power to turn than an equivalent multistage axial.

Axial flow compressors are used in aircraft to provide large volumes of air at a relatively low pressure at which combustion can occur. Thats not the ideal for the demand of a car engine. FYI Boosted, supercharged or forced induction engines usually run at less than ambient pressure. When you have your foot down you want the compressor to provide the volume of air the engine is drawing at the engines speed at the pressure you want. Since a car engine speeds change dramatically that's hard to achieve.

Compressor types are roughly split into positive displacement, centrifugal and axial compressors. PD compressors move a small vlume of air at high pressures and PR's, Axial move a lot of air at low pressures. Centrifugal sit somewhere in the middle. Trying to make an axial compressor work on an engine is difficult since it's flow and compression doesn't match a car engine well and axial flow works best at a fixed, high speed.

Having said all that, if the aim is to get a modest boost benefit why not? It would be difficult to get it to work as well as other alternatives and would be harder to make.

To actually build it I'd be concerned about the rotational speed. This will have to turn pretty fast, probably faster than a centrifugal compressor so at least over 30k rpm. Bearings would be an issue as would gearing up the drive. Positively locating the blades in the rotor disk is a must, maybe use a locking ring. Blades have to be of the corrrect shape and similar to a fairly high tolerence, both for PR and for balance. Efficiency and PR of the fan will be most affected by the airgap between the blade tip and the outer shroud. Sequencing the successive rotors in the correct position is important or blades can buffet and stall. I'd try and maintain the straight flow if possible which gives a challenge in driving the thing. To size it you'd need to work out the capacity of your engine and determine the volume and pressure required at a given speed. Each successive rotor will be of a smaller diameter and greater pitch and if you're going to get picky you need to consider the heat produced by compression and its effect on volume and pressure when sizing the thing.

Let us know how it goes.. interesting project, good luck.

Andrew

[devincox]

fyffe555,
thank you for your response, a few questions if I may.
1. You mentioned the accuracy of the blades. I was thinking I would cut them using my cnc/rotary table. I was going to get an airfoil shape and place it at the proper angle and then repeat around the circumference of the disk. Would this be accurate enough?

2. How do you size it? The engine is a 49cc four stroke.

3. Do I need to taper the inside of the housing, accounting for the sucessive shrinking of each rotor/stage?

Thanks,
Devin

[fyffe555]

Devin,

Accuracy;

Three problems - working out what the blade shape should be, machining each blade to be the same shape, accurately balancing the rotor including blades. CNC on a sherline from consistant material should fix the last two. You'll probably have to manually finish and balance them.

Blade shape should be airfoil section and have a twist to reduce angle of attack as the diameter increases. To be honest the airfoil isn't that critical, the twist is - within reason and for this I'd assume the pitch of the blade is a screw thread and both root and tip angle have to cause the root and tip to travel the same distance for the same rotation.

The accuracy of the blade mount so all the blades are at the same angle on the rotor and that the blades are located at the same diameter are the most critical.

Sizing;

The motor is 49cc. To calculate volume required - say it's running at 5k rpm, a four stroke means one induction cycle in two rpm so that 49cc x 2500 or 122.5 litres or 0.1225 cubic metres air per minute. You'd need to work out how you can make it and the limits you have on the size of the bearings, centre core etc. and so work out the inside rotor diameter. To figure the outside you need to work out the swept area and figure the pressures, volumes and temps. You need to look up Boyles law, Charles law and the Ideal gas law. Roughly speaking and ignoring all sorts of real thermodynamics Ideal gas law says there's a direct relationship of; (P1*V1)/T1 = (P2*V2)/T2 where P=pressure, V=volume, T=temperature, 1= one side of the system, 2= other side of the system

Note the air temp will increase as its compressed by the act of compression and the energy lost to the air in the devices 20-30% inefficiency and the stirring losses as a result of the blade shape, rotor position, blade to housing gap. To get, say, a twice ambient pressure as a good four stage axial then the temp could easily double.

The angle of attack of the blade determines the speed of air flow through the device. No easy mathematical way to work this out but assume a lot and its ok. Say the same 2 x ambient and double the temp. Plug that into the equation above. The volume will change with pressure and temp. You know its a PR of ~1.2 so you know volume is going to reduce by a the same amout at each stage an you know the end result needs to be 0.1225 m^2 per minute at twice ambient. If you know the volume at each stage you can then play with the blade angle to determine speed as a function of rpm and the swept area to give a given volume passed at each stage.. Rough but it should be close enough. note you have to have a high rotational speed for these things to get the PRs to work.


Housing; the shape of the outer surface must follow the diameters of the rotors and you need as small a gap between the blades and the outer housing as possible.

hope that helps.

Andrew

[devincox]

fyffe555,
I'm lost on how to calculate the twist.

I thought I could mount the rotors/stators on a common keyway that runs the length of the common power shaft ( except where the bearings are for the stators). This way they are positioned accurately.

Is there a program out there that could have these numbers input and give me a result?
I am not wanting it to be commercial grade so I couldn't I just make some approximations in the area of blade design and volume? Or is there a very small margin between working and not working on this type of compressor?

Thanks,
Devin

[fyffe555]

Devin, there's a wide margin between working and not working, sorry if I gave the impression otherwise, but the efficiencies do drop rapidly if the wrong choices are made or not calculated resulting in a lot less air than expected. You can assume a lot and just plan on changing the speed as necessary and make the rotors strong.

For the twist, think of the blade moving air at the same speed at the root and tip. Because the tip travels further and faster it's pitch should usually be less, just like a propeller. Another way calculate the twist angles is to ratio the distances travelled and since circumference = pi*D then the Tip Angle of Attack is (root diameter/tip diameter) * angle of root pitch. Say 2.5" root diameter, root angle 45 degrees, tip diameter 3.25" then Tip Angle of Attack =(2.5/3.25)*45 or 34.6 degrees. A straight line twist from 45 - 34.6 would be close enough.

NOTE; this is a gross approximation so real Mech/Aero or Thermo engineers please don't beat me up!

[devincox]

I will try to work something up tonight in a cad program and post the results. (2d only)

Thanks,
Devin

[devincox]

I finished the blade profile this weekend. I was going to cut some test blades but I just installed mach3 and forgot to setup my steppers with the new setup.
There is not much to the profile (so I did not post any pics) just a standard Joukowsky style airfoil profile for the blade. I am very curious to see how this will cut on my setup, givin the size constraints. I should have something this week.

Thanks,
Devin

[fyffe555]

Devin, Good job. Definitely be interested in seeing the blade. Couple of thoughts; Joukowsky airfoils have a cusp a the tail of the foil and because its a Joukowsky transform the tail is often extremely thin as well as curved. Be fun machining that... Second point was that the airfoil section in an axial compressor doesn't work quite the same as a free airfoil. Compressor section has much higher presssures on the underside and so the pitch is more important that the top foil section. Gives you much more licence to adapt the foil to machine well, such as thickening the tail section.

Andrew

[devincox]

fyffe555,
I know what your talking about on the tail of the airfoil. I was using geopath to generate the g-code and it did not like that thin section, especially since the entire blade is only about .750" in length. At its thinnest section, the tail was only .005" thick. So I took some liberty and fattened up the tail and added a small fillet to the end. We shall see.....

Devin

[fyffe555]

Another thought; how are you going to fuel this motor? Since its 49cc its probably not got Fuel injection so you've got to consider a carb. You can mount the blower after the carb and suck through but you can't blow through a carb unless you box in and pressure the entire carb. Either method will probably require different jetting from stock...

[devincox]

I had planned on trying suck-through first and then going from there based on results.
I set everything up on my machine lastnight, to cut the blades and found something I hadn't considered. I have my cnc rotary table mounted at 90 degrees and then a 4 jaw chuck and then a mandrel that the round blank is on. I will then cut a blade and then advance rotary table by a specific amount and cut the next blade. I am curious as to what the area around the blade root will look like. I think it may not be very smooth and round and am concerned what it will do to airflow. I am not sure how to program my 4th axis to compensate for this. Any ideas?

Thanks,
Devin