Ducted Fan Design

for a Cozy Mark IV

Here I'll attempt to design a ducted fan for a Cozy Mark IV powered by a converted automotive engine (Mazda 13b). The ducted fan should be well suited for this task for many reasons.

With all those advantages, you may wonder why there aren't more aircraft flying with ducted fans. I can think of a few reasons for this. Most aircraft have the prop in the front, pulling the aircraft through the air. As mentioned above, ducted fans don't work well in tractor aircraft. The reason for this is that the duct acts as a fixed stabilizer. You don't want a fixed stabilizer on the front of your aircraft any more than you want the fletching to be at the front of an arrow. Furthermore, an appropriately sized duct on the front of a tractor aircraft would obscure much more of the view than the mostly transparent spinning prop. That makes a very large percentage of aircraft for which it doesn't even pay to consider a ducted fan.

Most designers have the slower turning/higher torque aircraft engines in mind when designing their aircraft. I suspect few of these designers are willing to expend the time to design a duct which will, at best, realize a small performance drop (at cruise) when compared to a propeller on the same engine.

Ducted fans are far more involved to design and build and tune to a particular aircraft/powerplant combination. There are several free and commercial software packages to help a designer select the appropriate propeller pitch and length. Many props can be ordered a little oversized and then trimmed down until the desired engine RPM is achieved. Not so for the ducted fan. Like a prop, blade pitch and length must be considered, but so must the inlet and outlet diameter of the duct and the inlet shape. The fan blades can't be trimmed down to increase engine RPM without affecting the tip clearance that is so crucial for good duct performance. Furthermore, many applications will require the design and construction of a mount akin to an engine mount.

Generally speaking, a perfectly designed propeller will generally outperform a perfectly designed ducted fan at high speeds, where duct drag starts to dominate. It has been suggested that at the cruise speeds of a Cozy (200 mph), the difference will be small. How small this difference can be made remains the goal of this site. If the ducted fan performance can be expected to achieve 90-95% of that of the propeller, than I will likely try a ducted fan for my Cozy Mark IV project. If you have any questions or comments about this design, please feel free to e-mail me .

Much less design information is available for ducted fans than is available for props. Likewise, there are far fewer flying examples of ducted fans than of props in GA aircraft. There are, however, enough good sources of information and tools to get started:

Using the information in "Ducted Fan Design" I have created a MathCad sheet of calculations that attempts to design a ducted fan matched to the desired cruise speed and the power available. I also have an html file of the calculations for your browsing pleasure. In short, with 165 hp available at 75%, with a cruise speed of 200 mph, the duct would have a capture area of 7.1 sq.ft., a disk area of 4.6 sq.ft. and an exit area of 6.2 sq.ft. With a 29.5 inch blade, the blade tips would be traveling at Mach 0.7, right at the limit of compressibility for many airfoils. This geometry also gives an expanding duct exit, which helps static thrust, according to Raspet.

My apologies for the poor image below. I 'll get a better image/drawing up here soon.

The next step will be to use this geometry to layout the proper duct and aft cowl shape to feed the fan. For this first iteration, I am relying a lot on intuition, without much experience to back it up.

I am in need of a reasonably accurate model of a Mazda 13b engine so that I can shape the cowling correctly. For now, I am assuming that a cowling 18" in diameter at the flywheel face (approx. F.S. 162) will be adequate. From this point forward, I will round the shape into the plans firewall.

According to Raspet, the proper shape of the inlet should be a cambered ellipse, with a eccentricity of 2:0 or greater. This ensures that airflow around the inlet nose in static conditions will not separate from the duct wall. The high speed air around the inlet nose reduces the air pressure, thereby providing as much as half of the static thrust. I have chosen 3.5:1 as the ratio. The inlet diameter was chosen such that the outer boundary of the capture area stream tube at cruise passes through the stagnation point of the duct nose. This should minimize the duct drag at cruise. The inlet ellipse is set tangent to the free flow stream tube at the fan face. This ensures that the flow will be contracting into the fan and will start expanding immediately after the fan.

After the fan, the ellipse (and tapering tail cone) gradually increases the rate of expansion into the 6% conical duct exit. My research indicates that this angle of expansion is a conservative value to maintain attached flow - up to 8% is possible under the right conditions. This slope is carried backwards until the desired exit diameter is reached, thereby establishing the aft end of the duct. The rest is just connecting the dots.

Currently, I am trying out Gmsh to create the mesh blocks for ADPAC. I realize that some reformatting of the mesh will need to be done. However, I am really pleased with the modeler capabilities so far, and the price (free) can't be beat. It runs on everything from my Linux workstation to my Mac to the Windows boxes in the office. It is a programmable parametric modeler, and I have been able to create a fairly automated fan grid generator with it. To use this, you will also need to get the airfoil definition file. By simply editing a few values (blade count, number of blade stations, airfoils, AoA, chord, etc.) at the start of the file, you can quickly create various fan models. It only creates the surface mesh for now, but I am working on getting it to create the proper volume meshes. The image below shows the current status.

Fan Surface Model

This is a five blade model which has no real bearing on reality. As I mentioned above, by simply changing one parameter, the Gmsh script can create 3, 4, 5 or any number of blades you wish. I expect the final design will have 3-5 blades.

In addition to a surface representation of the fan, the boundaries need to be modeled as well. To understand what boundaries are needed, it helps to understand the type of grid used with ADPAC. The fan block utilizes an O-grid. Imagine a cartesian coordinate system (i,j,k) wrapped around the fan blade such that the i axis started at the trailing edge, followed the airfoil over the top, around the leading edge and back along the bottom to the trailing edge. The j axis is along the span of the the blade. The k axis radiates outward from the blade surface. A cut through the grid on the i-k plane would look a little like this:

O-grid Example

This section in under construction...

The i=0 and i=IMAX plane are the same surface, because the grid wraps around the blade.

i=0 and IMAX at trailing edge

The j=0 plane defines the hub, and the j=JMAX plane defines the duct wall within the thickness of the fan disc.

j=0 at fan hub j=JMAX at duct wall

The k=0 plane defines the boundary with the airfoil.

k=0 over top of airfoil k=0 over bottom of airfoil

The k=KMAX plane defines the outer boundary adjoining adjacent blade blocks as well as the fore and aft duct blocks.

k=KMAX at aft face of fan disc k=KMAX at boundary with trailing blade block k=KMAX at foreward face of fan disc k=KMAX at boundary with leading fan block k=KMAX at aft face of fan disc

Once this model is completed, it will be analyzed with ADPAC . Static, climb, cruise and maximum speeds will be looked at to try to characterize the performance envelope.