# Aerodynamics of Propulsion

### Help Support Homebuilt Aircraft & Kit Plane Forum:

#### WINGITIS

##### Well-Known Member
Some ideas from 101 years ago to consider/ponder....

Added two more pics from the same article

Also at the bottom of picture 3 is the similar system subsequently used by the British Mosquito in WW2.

#### Attachments

• 1921-German aircraft cooling options discussed.jpg
129.8 KB · Views: 2
• 1921-Cooling 2.jpg
21.5 KB · Views: 2
• 1921-Cooling 3.jpg
55.9 KB · Views: 2
• De Havilland Mosquito DH98 NFll HJ711 Elvington.jpg
83.6 KB · Views: 1
217.5 KB · Views: 1
Last edited:

##### Well-Known Member
Supporting Member
Streamline Diffusers and that pesky Third Dimension

Table 12-1 gives a table of coordinates for a streamline diffuser of different expansion ratios (0.25, 0.30, 0.40, 0.50). The coordinate system here is a bit odd -- I'm sure it's explained somewhere else in the book, but at a flip through I can't find it. The main dimension Y_B is the height of the heat exchanger; but it looks like this coordinate runs from -1 to 1. (Logic: For the 0.50 expansion ratio, the final y coordinate is 0.5/Y_B; no other context I can think of makes sense here.)

This table is for a two dimensional diffuser, and the coordinates are modified from true streamlines to give a total recommended length of 3*Y_B. So, if I'm reading this right, for a 12" high diffuser, max(Y_B) = 6", so the diffuser length would be 18". I'd appreciate a check that I'm interpreting this correctly.

Where things get interesting is applying this to a three dimensional heat exchanger, with height and width. Suppose a moderate aspect ratio -- so 24" width, 8" height, irrelevant depth, as an example. With a 2D diffuser you can diffuse the air in two directions -- if the inlet is 24" x 4", you get an expansion ratio of 0.5 with a diffuser length, by the table, of 12"; if the inlet is 12" x 8" you have a diffuser length of 36".

If makes moderate intuitive sense to me that, for a 2D streamlined diffuser, you can build it either way, and will get different results, but I'd love a sanity check. The obvious follow-up question is which of these would be preferred.

It also seems obvious that, in these days of CFD, it would make sense to drop that 24" x 8" heat exchanger into CFD, modeled as a thin plate perpendicular to the flow with the appropriate pressure drop across it (no geometry modeling), and selecting 3D streamlines. There's still a lot of approximations here (no boundary layer), but it would give a 3D inlet shape with the streamline property as much as the 2D examples in the book have, and probably also give some nice visualization of how to to deal with e.g. corners. However, this will still give infinitely long streamlines, which have to be truncated somewhere. Is the appropriate length to truncate to here based on the maximum wall angle, or based on equivalent length (so 3/2*sqrt(24*8) = ~21" length)? Too long means excess wall friction losses; too short means separation losses. Ideas?

Last edited:

#### PMD

##### Well-Known Member
I suspect that once you size core volumes large enough to handle the real world heat rejection needed, you'll find that either the face area will be so large that you can't fit it on the plane, or the core(s) will get so thick that you can't force the air through them.
IMHO, THIS is the very core of the problem (sorry, couldn't resist that one). If you look at the same installations I seem to be reviewing - RED, Thielert, Austro and Rotax 9xx - you will soon see that the entire cowling outside of the core engine is surrounded by thin core, large cross section radiators and a tangle of hoses, supports and "blast tubes" and/or ductwork to feed them. Add an exhaust system with muffler(s) and not only do you have some serious space issues, you have a lot of vibrating garbage causing maintenance issues, places where you have heat being rejected that aren't blessed with "natural" airflow and finally the subject of this thread: an aerodynamic nightmare to get air in and out in an orderly and efficient flow. For those reasons, I am tending to lean towards keeping as much of the cooling heat exchange INSIDE of the engine (i.e. oil and charge air) and having only one (or bifurcated) heat exchanger ideally placed in ductwork to manage airflow.

#### rv7charlie

##### Well-Known Member
Supporting Member
However, this will still give infinitely long streamlines, which have to be truncated somewhere.
You've probably already absorbed more of the book than I did, and most of what I absorbed has evaporated over the years. But I don't understand the questions around this. From my fading memory, you specify the core's face size, then specify the face to inlet area ratio, then run the numbers using the formula. As I remember it, the formula's progression tells you the length. A smaller inlet to face ratio (bigger inlet) yields a shorter duct. I'm not seeing how you get infinite length streamlines from reasonable inlet sizes.

Also don't understand '3D inlet'. Probably a semantics/terms problem, but I can't see an inlet as being anything other than two dimensional; 3D in this context feels 'solid' to me. Are you describing a duct that expands in both the X & Y directions, instead of only one axis as expected in the book?

##### Well-Known Member
Supporting Member
Let me clarify, definitely was fuzzy on my side.

1) The table I'm looking at in the book actually has a length of "3" (1.5x the heat exchanger height) for all the expansion ratios; I'm not seeing a formula for streamline diffuser shape, and indeed there's even a comment that an attempt to fit standard curves to this didn't work particularly, well, so use the table. I'd be very curious what you're thinking of -- the length of 3 in this case is pitched as a good compromise, but has very little theoretical support, especially with varying expansion.

2) The "infinite" comment was in the context of understanding how these tables were generated. Basically, they solved a 2D flow with constant inflow velocity impinging on a line segment with the appropriate pressure drop. The streamlines they selected are then the streamlines that run from the end points of that line segment forward to infinity; this is the "natural" shape of the flow when hitting that (2D) heat exchanger. These streamlines are infinitely long, though, and never parallel; so they made a couple of adjustments. Basically, they chose arbitrarily to cut them off at a length of 1.5x the line segment length, then made slight adjustments to make them approach parallel at that distance. There's also slight tweaks at the heat exchanger face.

3) Yeah, 3D inlet is sloppy terminology. (Long answer follows. Short answer: "Are you describing a duct that expands in both the X & Y directions, instead of only one axis as expected in the book?" Yes.) The book numbers are for a 1D approximation of a heat exchanger (line segment) immersed in a 2D flow, so the diffuser walls are curves -- 1D surfaces in 2D space. I'm discussing using the same approach they used for generating those tables, but taking advantage of cheap CFD to model the heat exchanger as a 2D surface (plane segment / rectangle) immersed in a 3D flow; the selected streamlines then form a wall that's a curved surface in 3D space. The inlet is then a 2D plane in this 3D space as well.

#### wsimpso1

##### Super Moderator
Staff member
Hmm, if you work any serious expansion from inlet to HX in both width and hieght, you will end up needing either guide vanes in both axes to wet the HX or pretty long diffuser.

Ever look at the inlet nozzle of the F-100 Super Sabre. Seriously supersonic bird, flow must decrease to subsonic by the time it gets to the round compressor face, and the jet won't make much power if the whole face is not at big flow. LOONG duct and not much expansion. 3D system.

I suspect that doing most of the expansion in one axis and only a little in the other is way more practical unless you have a significant budget for analytical and wind tunnel work.

Billski

Last edited:

#### rv6ejguy

##### Well-Known Member
Supporting Member
As Bill said, given the space/ length constraints in most aircraft, you'll probably end up using a guide vane to efficiently turn the flow in most cases to wet the top of the HX unless very shallow. In my installation, I chose a rectangular inlet shape to avoid having to expand flow too much horizontally and installing guide vanes in two planes. I had to have hand access to lay cloth and resin to build the thing which set the inlet height at 2 inches. Many considerations here from a practical standpoint outside the theoretical.

#### Vigilant1

##### Well-Known Member
Supporting Member
Probably obvious but: Drag considerations exterior to the duct are also worthy of consideration. Even if it is possible to fabricate a round inlet to feed a rectangular heat exchanger, the air that doesn't go in will need to travel a long distance to get around the wide dimension of the HX, and that's more skin friction drag. For this and other practical packaging reasons, I'd think a thin rectangular inlet that has a long dimension almost equal to the long dimension of the HX does a good job of minimizing the angles inside and outside of the duct.
Obviously, everything depends on the actual situation.

#### rv6ejguy

##### Well-Known Member
Supporting Member
If you use a round or oval inlet to feed a rectangular HX, you'll have to transition that shape. Not sure how much you gain or lose there. Corner flow is often a mess anyway. Guide vanes can help smooth some of this too.

If you really want to learn something, 3D print a couple iterations of the duct, choose an HX, build a rig to put on the roof of your car and instrument it.

Applying theory with no solid numbers to plug in will result in an answer where you question the accuracy so why bother? Through experimentation you'll be able to get some actual data and then apply that to a theoretical model, permitting refinement from there. Follow up with a final 3D printed model to validate the design.

Last edited:

##### Well-Known Member
Supporting Member
Thanks for the input, all. Consistent. Don’t be over ambitious, don’t take theory further then it goes, experiment. Agreed.

The logic behind asking about expansion in both directions was (a) it fits my geometry reasonably well and (b) the wall angles would be less extreme, which seems like a good thing. But the vane challenges are real, got it.

##### Well-Known Member
Supporting Member
Another related, though less "Aerodynamics", question:

Have those of you with ducts given any thought to consequences / mitigations for water in the duct?

My duct will be significantly aft of my nominal CG, and my aircraft is very CG sensitive.

The case I'm most worried about is rain on a windy day, blowing in under the protection of the wing (which is possible, given the geometry), flowing to the bottom of the duct, and then potentially freezing.

Hopefully a pre-flight will notice this, but I can imagine a few pounds of ice at the very bottom escaping a glance without a flashlight due to the angles; and a few pounds of ice absolutely can put this design out of W&B.

So, mitigations? Ways to make it easier to detect? How big a drainage hole can one put on the bottom of a duct without noticeably interfering with the flow in normal use? (A drain valve like we use for fuel tanks would seal when not in use, but would only help with water, not ice.) Other ideas?

#### Vigilant1

##### Well-Known Member
Supporting Member
So, mitigations?
Well fitting duct covers? They'll be needed to keep birds out anyway. 'Remove before flight" and hopefully a reminder flag visible from the cockpit, if possible. If not, at least stow it inside the plane and make it part of your interior checklist to see it stowed before starting the engine.

##### Well-Known Member
Supporting Member
That's definitely part of the plan. I have those foam inlet ducts for the Commander that are pretty snug fitting, and I've definitely found water under the cowling in really driving rain (although it's possible that it's splashing up from the ground and through the nose gear door -- we get weather here), so the duct covers would either need to be even better seal than those, or just part of the plan.

#### rv6ejguy

##### Well-Known Member
Supporting Member
I have a couple 3/32 drain holes in mine for water.

Ice, well I park in the hangar so not a concern for me. If you have ice in there, would have to have a way to remove it like any other airframe ice. Wouldn't be fun in a duct without a lot of Iso.

#### mcrae0104

##### Well-Known Member
Supporting Member
The case I'm most worried about is rain on a windy day, blowing in under the protection of the wing (which is possible, given the geometry), flowing to the bottom of the duct, and then potentially freezing.
It rains in Utah?

##### Well-Known Member
Supporting Member
Yep. Not that often (15" a year where I am), but when it rains, it rains. 1" - 2" per hour is typical, and often combined with 45 mph gusts. Monsoon season is a real thing.

#### Vigilant1

##### Well-Known Member
Supporting Member
I suppose a small (1/4” dia) open drain hole at the low point wouldn't cost much leakage.
4 Functions:
1) If rain blows in, it can drain out before freezing
2) If snow blows in and melts, it will drain before refreezing as a block
3) If there's an icicle there during the walk around, it is a clue to look further
4) You can stick a pencil up into the hole during the preflight. If it won't go in, there's a problem.

Would it be possible to design the duct so it drains out of the lip? Avoiding the problem is better than hoping to detect it.

It's a good thing to have thought about.

ETA: Oops, I cross-posted with Ross.

Last edited:

##### Well-Known Member
Supporting Member
Drain hole seems like part of the solution too. 3/32" seems sane; going up to 1/4" would make it a way better drain, but I'd start to worry about the flow through it during flight. I think there's likely to be two low points (one in front and one behind the radiator), so two 3/32" - 1/8" holes make sense. The point about them not just acting as drains, but being a good thing to poke and check makes a lot of sense to me... easy enough to add to the walk around. Make the diameter match the fuel sump drains and can even use the fuel tester that's already in my hand.

I'd love the duct to drain forward, but the current geometry looks like there's a definitely a low spot in the middle.

#### WINGITIS

##### Well-Known Member
The MIG-29 is known for its great pressure recovery onto the face of its jet engine allowing very smooth throttle pickup even at large AOA, RECTANGLE TO ROUND..

Some pics, also one of a MIG-25:

#### Attachments

• images MIG 29 duct.jpg
5.2 KB · Views: 0
• MIG 29 ENGINE FRONT FACE.JPG
3.1 MB · Views: 0
• 0635 mig 29.jpg
72.3 KB · Views: 0
• 7065b820-2de2-4438-b7e6-6eee4a7b2364 MIG 29.jpeg
382.8 KB · Views: 0
• 14 MIG 25.jpg
62.9 KB · Views: 0
• images MIG 29.jpg
11.4 KB · Views: 0
• 8648311721_32818f4579_b NIG 29.jpg
126.4 KB · Views: 0