Aerodynamics of Propulsion

Homebuilt Aircraft & Kit Plane Forum

Help Support Homebuilt Aircraft & Kit Plane Forum:

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Just received via inter-library loan. As pro asked, table-of-contents attached so folks can ask questions.

I’m going to be working through this with an example of interest to me — a 180 ho liquid-cooled aircraft with an aft duct with a nominal cruise-climb speed of 85 kts IAS. From billski’s guidance in the other thread, we expect this to correspond to an inlet area of roughly 500 cm^2. If reasonable, let’s use these numbers for discussion.
 

Attachments

  • CONTENTS.pdf
    894.4 KB · Views: 8

rv7charlie

Well-Known Member
Supporting Member
Joined
Nov 17, 2014
Messages
3,277
Location
Pocahontas MS
That's ~77"sq? Sounds excessive; most of the rotaries in that HP range that I know about are flying with a bit over half that, and many of their diffusers are sub-optimum. A lot of the early ones (admittedly with lower output engines) were flying with Van's stock RV cowl openings and truly terrible inlet diffusers.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Charlie, yeah, it's almost exactly twice the numbers I got using the guidance from your videos...

The 500 cm^2 number is based on Billski's discusson in cooling fans. With OAT of 100° F and a max coolant temp of 225 °F, we get (mixing units here deliberately) a 70 K temperature difference, derating to 50 K. With 135 kW of heat estimated, and sea level air density, this comes out to 500 cm^2 at 85 kts.

I can think of a few reasons this is higher than expected, and I'd like to understand which (possibly multiple) apply:
  1. The estimated heat generated is too high. This is very much a rough rule of thumb; but if anything it's probably more true for the case I'm interested in than for other designs, since the duct will contain an intercooler, radiator, and oil cooler; no separate oil cooler or partial air cooling taking some of the load.
  2. The max temperature is too low. With the oil cooler last in the stack, and with oil temperatures up to 250 °F reasonable for this type of engine, this can give another 10 K or so to play with, dropping 80 cm^2 or so.
  3. The derating of temperature drop is too conservative. Billski gave a pretty wide range of derating factors, and I'm on the less conservative side, but...
  4. The goal is too ambitious; although 85 kts is a reasonable cruise-climb speed for an aircraft with a 50 kts stall and a 150+ kts cruise, my experience with my Commander 112TC has certainly indicated that extended climb at 85 kts is ambitious, and 100 kts is more reasonable. Similarly targeting 100° F OAT for this condition (sea level) might be ambitious... but I live at 5k ft with OATs regularly above 105° F, so I don't want to be too reasonable here, either.
  5. Most interestingly, and what I think I'm most likely to learn from the book under discussion (take that, thread drift!), I wonder whether there's a useful amount of suction in a sane duct geometry that would mean that the inlet mass flow is higher than the mass through through the equivalent area of the free stream. Obviously at very low speed this is true due to propeller effects, but with my duct six ft behind my prop I don't expect this to be a big effect at speed.
Thoughts? Anything else I'm missing? I'd certainly love to convince myself that 250 cm^2 with a well-designed duct is doable... but I'd also love to take 1400 ft/minute at 85 kts up to 14k ft on a warm day.
 

rv7charlie

Well-Known Member
Supporting Member
Joined
Nov 17, 2014
Messages
3,277
Location
Pocahontas MS
Well, we were obviously starting with different assumptions. ;-) With only a few exceptions, the rotary guys (most alt engines in general) are flying normally aspirated engines with fixed pitch props. Therefore, a ~180 HP engine in something like an RV will only be making about 135 HP, at best, in climb because the prop is going to limit rpm at low airspeed. A lot of them also accept somewhat cooling-limited short term WOT operation in exchange for efficient cooling drag reduction at cruise. Note that this is quite common with traditional engines as well; few Lycs on fast planes like RVs can sustain extended full power climbs without getting oil and cyl head temps into the danger zone.

If you're turbocharging + intercooler, a controllable prop, operating in high/hot conditions, dumping every bit of the heat into one heat exchanger, and expect to do all this for extended full power climbs, then even 77 sq in might be marginal.

Oil temp is typically measured at the return from the cooler; if you're seeing 250 there, it will be a lot higher in the pan.

Climb speed is a bit wing planform dependent; the 2 seat RVs mostly have a calculated best climb in that range, but with a fixed pitch prop, the climb rate doesn't change much at all from around 85-90 mph up to around 120 mph. Van himself once wrote an article about it. IIRC, it has to do with the induced drag of the low aspect ratio wing on the acro planes. Obviously, unless you're trying to clear obstructions it makes more sense to climb at 120; you cool better and you cover more ground.

One thing that's come out of the experience of rotary drivers is the growing awareness of some old cooling research that discusses 'internal vs external diffusion', something that a good friend, a retired Pratt & Whitney engineer (who built a rotary powered RV) mentioned to me over 20 years ago. The redneck version is, if you shape the lips of the cooling inlet correctly (critical point), you can go a lot larger on the inlet; big enough to consistently cool in worst case situations, and then use a cowl flap or other method to 'throttle' the cooling exit. If executed well, the increase in cooling drag in cruise configuration is supposed to be negligible because the mass flowing through the system would be the same as with a cruise-optimized inlet & diffuser. It allows a modified inlet diffuser shape that is supposed to be less critical to achieve, but again, the inlet lip shape becomes critical.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Yeah, my daily driver is a 210 HP turbocharged engine, CS prop, and I can't really imagine going back to naturally aspirated (living at 5k ft) or fixed pitch (ever). One of the requirements for the design I've been working on is "don't find myself in situations where I'd rather be in the Commander than My Design."

Any paper references on the inlet lip shaping? If I picture what you're describing, the idea here is that at cruise, the exit flap is closed so much that the pressure at the intake is higher at ambient, so air spills around the duct; but with good lip shaping the cost of this spillage is low. Is that the idea?
 

rv7charlie

Well-Known Member
Supporting Member
Joined
Nov 17, 2014
Messages
3,277
Location
Pocahontas MS
I think that once you have experience in higher performance homebuilts, you'll realize that the only thing you might miss from the Commander is the air conditioning (if it has it). Years ago I bought a 180 HP Lyc out of a Thorp T18. The engine's 1st 50 hours were in the prototype Commander 112; removed before the plane was marketed because it wasn't enough power to make it fly safely. Most fast homebuilts are lighter and cleaner than their nearest equivalent from the certified world.

If I picture what you're describing, the idea here is that at cruise, the exit flap is closed so much that the pressure at the intake is higher at ambient, so air spills around the duct; but with good lip shaping the cost of this spillage is low. Is that the idea?
Exactly. The external diffusion idea means the conversion of dynamic to 'static' pressure happens out in front of the inlet, so the duct doesn't have to turn high speed air and also slow it at the same time to recover the pressure. The lip becomes critical as the exit is throttled, because air must flow smoothly around the external part of the lip without going turbulent. (Internal diffusion has a critical lip shape as well; it is much sharper than the shape for external diffusion.)

I'll try to find the doc & post it here, but IIRC, the basis for the idea is in chapter 4 of K&W.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Was just working through chapter 4, some good stuff there. I think the most interesting things to me were (a) the focus is on having a smooth enough pressure distribution to avoid Mach effects, leading to a wider duct outer area -- but as expected for Mach effects, this doesn't really matter at our speeds; (b) the only only thing limiting duct outer area (preventing a lower area) is off-design behavior, but off-design behavior isn't explored as thoroughly as Mach effects; and (c) two different families of "that looks about right" duct intakes designed in completely different ways are pretty darn close in performance to optimal. Super promising that a practical approach (using the most rounded section A, or something that looks a lot like it) is practical across expected flight conditions.

1661832354625.png
 

Monty

Well-Known Member
Joined
Jul 15, 2010
Messages
1,365
Location
Fayetteville, AR / USA
Off design inlet for us, just translates to external diffusion with a lower Vi/V0. You aren't going to be sucking in air....Most people who try to use this text fail to READ the book. Don't just start pulling equations and numbers off the graphs. You need to completely understand the assumptions behind the derivations and charts. It took me several passes to realize how they arrived at the streamline diffuser, and how to apply it.

As I've said before, you can design a banner towing airplane for hell if you like, but the cooling system is going to be way oversized for normal operation. It's ridiculous to make a point design for an exception to normal operation. You can't climb VX full power at 100f. This is a silly design point. If you do this the installation will over cool in all other situations and generate too much drag. Size for cruise and put a spray bar on it if you absolutely positively MUST have this capability. Or do like everyone else on the planet: lower your climb rate, up the speed, and go full rich on a very hot day. Anything else is airplane abuse by a bad operator.

The inlet lip should be a 2/1 ellipse. The inlet velocity ratio Vi/V0 number is a design point you must choose. You can't do this or size the inlet/exit without good pressure drop and heat exchanger effectiveness data. All of this works together. Trying to consider these in isolation will have you chasing your tail.

You should forget about having multiple heat exchangers in the same duct. That just won't work. They have different heat rejection temps, and different pressure drops. You'll get too much flow through one and not enough through the others. They go in seperate ducts sized for the pressure drops and Q values.
 
Last edited:

rv7charlie

Well-Known Member
Supporting Member
Joined
Nov 17, 2014
Messages
3,277
Location
Pocahontas MS
FWIW, Monty actually knows this stuff; I'm just a nearly blind man who's been grasping at vague shapes for a couple of decades.

Attached is the paper I think I'm remembering; a NASA doc from 1981 that mentions internal/external diffusion and lip shape, and has a couple of references back to K&W. It also discusses propeller effects. It obviously is dealing with air cooled piston engines, so...
 

Attachments

  • NASA CR 3405.pdf
    7.6 MB · Views: 3

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Don't worry, I'll read the book in detail before I pretend to have learned anything. I generally do a first pass pretty quickly so I understand the layout and what topics are covered, but I'm not treating anything I see this casually as true knowledge.
You should forget about having multiple heat exchangers in the same duct. That just won't work. They have different heat rejection temps, and different pressure drops. You'll get too much flow through one and not enough through the others. They go in seperate ducts sized for the pressure drops and Q values.
Monty, I've definitely gone back and forth on this, so interested in more input. Wouldn't surprise me at all to learn that I have to go with different ducts, just haven't reached that point in my knowledge to be convinced.

For the single duct, the idea was to size it for the coolant radiator, taking into account the flow restriction and heating effect of the other coolers. The idea was to have, from front to back, the air-to-air intercooler (oversized for this application, relatively thin core, inlet temperature ~200 °F, but low heat capacity fluid), the coolant radiator (inlet temperature ~195 °F), and the oil cooler (massively oversized, inlet temperature ~230 °F). So the temperature of the exchangers are increasing back to front, much like a counterflow cooler would. The exhaust flap would be manipulated to adjust coolant temperature (manually or not), and the oil cooler would be fed by a bypassing thermostat; the intercooler would be open loop, you get what you get.

In terms of heat rejection temps, I would expect that having increasing core temps as the air flow warms would be the most efficient way to transfer heat to a given amount of air; obviously the flow does need to be sized taking into account the pre-warming before each stage.

In terms of different pressure drops, the idea was to have these all be the same size (which massively oversizes the oil cooler even at single row, and somewhat oversizes the intercooler), and have them ducted in series, so that all air that flows through one must flow through all of them. The downside with this is obviously that I'm "paying" for the pressure drop on the oil cooler even when no oil is flowing through it, and even paying twice since it increases the pressure drop I need to force air through the radiator as well... but at least the flow through all of them is controlled.

So, with true respect for learning something from the answer -- why doesn't this make sense?
 

rv6ejguy

Well-Known Member
Supporting Member
Joined
Jun 26, 2012
Messages
4,824
Location
Calgary, Alberta, Canada
Just received via inter-library loan. As pro asked, table-of-contents attached so folks can ask questions.

I’m going to be working through this with an example of interest to me — a 180 ho liquid-cooled aircraft with an aft duct with a nominal cruise-climb speed of 85 kts IAS. From billski’s guidance in the other thread, we expect this to correspond to an inlet area of roughly 500 cm^2. If reasonable, let’s use these numbers for discussion.
This inlet area is way bigger than required for just the coolant given the hp and climb speed. My RV has best climb at 85 kts also, similar hp and I use an inlet area of 29.5 in2. I've never seen the coolant over 194F. Russell Sherwood's best climb is at 105 kts, 230hp with an inlet of 33 in2, restricted to 24 with an insert for racing.

We both found that the inlet lip radius doesn't need to be large as suggested by others here. Mine is .125 inches and Russell's at an estimated .200.

On the exit side, I can close down to about 19 in2, Russell to about 15 in2. He cruises with it in the range of 16-22.

Stacking HXs (charge air and oil) will result in much higher momentum loss than just the rad HX alone. Probably best to have separate, dedicated ducts for those items.

Guide vanes are your friend here if you're asking air to turn more than about 6 degrees. Russell and I both found them indispensable in reducing turbulence prior to the rad during our flow testing experiments.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
As I've said before, you can design a banner towing airplane for hell if you like, but the cooling system is going to be way oversized for normal operation. It's ridiculous to make a point design for an exception to normal operation. You can't climb VX full power at 100f. This is a silly design point.
Agreed, and perhaps I will be less conservative here... but to be clear, the design I'm looking at has estimated Vs ~= 50 kts, Vx ~= 66 kts, Vy ~= 76 kts, so 85 kts was significantly above these speeds. I admit it's informed by the Commander, which really wants 100 kts for cooling but can be so slow to climb at 100 kts that I need to advise ATC (and it has Vx ~= 72 kts, Vy ~= 86 kts). That said, due to extra power the design I'm exploring should have no problem keeping 500 ft/minute to this speed or above; perhaps this is a case where I need to rely less on a hunch about what I want, and instead look at the highest speed for 600 ft/minute climb (IFR desired + margin) at critical condition (altitude) and cool for that; will run numbers, but should be 100+ kts.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
My RV has best climb at 85 kts also, similar hp and I use an inlet area of 29.5 in2.
Ross,

(Oops, sorry Charlie, I confused you with Ross earlier when I said "your videos", two usernames starting with "rv".)

Both you and the Glassair have oil coolers mounted behind the cowling and are naturally aspirated, yes?

Just for comparison purposes -- I'm definitely getting a consistent message that I'll take to heart that 500 cm^2 is oversized for real operations, I just want to understand how much and what the consequences will be for erring on each side. Every drawing I've done so far has 240 cm^2 inlet / 140 cm^2 (adjustable) outlet based almost entirely on your videos, so believe me, I'd love to end up back there! (That's an inlet of 37 in^2, which matches yours pretty well given power differences at altitude between NA and turbo -- but then, it matches because you inspired it.)
 

rv6ejguy

Well-Known Member
Supporting Member
Joined
Jun 26, 2012
Messages
4,824
Location
Calgary, Alberta, Canada
My RV is turbocharged. Both airplanes use air to oil HXs for the oil, mounted in the cowlings, my intercooler is cowling mounted as well. You'll find it challenging to bring charge air to and from an aft mounted intercooler using a ventral scoop. Somewhat easier if all within the fuselage like a P47 but you still have long plumbing runs which have to pass through the cockpit area.

If your goal is just cooling, that's easy. If it's low drag plus cooling, you'll have to minimize momentum losses through the HXs and through proper duct design and a controllable exit.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
So, missing context here — I hinted at this in the “today’s sketch” thread but haven’t run a thread on this design in a while. This is an aft-mounted auto-conversion engine installation with drive shaft to a tractor prop. Aft-mounted belly duct is right near the engine. Intercooler as first in duct is minimal plumbing runs.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Yeah, exactly. I definitely know that propulsion integration (drive shaft and cooling) is the most ambitious part of this design. I’m not all /that/ worried about ground cooling, although perhaps I should be, since my taxis are short and run ups are limited — my intuition is that the heat capacity of the system is fine until take-off; but I’m definitely worried about climb, and like all of us I’m worried about cruise drag, although probably more in theory than in practice since this is in no way a racer, and a few knots is just a few knots.
 

addaon

Well-Known Member
Supporting Member
Joined
Feb 24, 2008
Messages
1,975
Location
Kanab, UT
Back to the book for a moment, new favorite section, although not super applicable to me.

Section 5-2 discusses thick fairings with uniform pressure distribution. It also introduces a method of determining approximate surfaces of uniform pressure — using cavitation in water flow as an analog computer. Analog computers are an ongoing hobby for me, and this is a great use of one, using the similarities in fluid flow between gasses and liquids to build the analogy, but using the differences — formation of a bubble in low pressure zones in liquid — to do the computation. Combine with some obvious properties of bubbles (interiors are constant pressure) and Bob’s your uncle.

I’m sure this was a standard approach at one time, but first I’ve seen it used this directly, and I like it!
 

rv6ejguy

Well-Known Member
Supporting Member
Joined
Jun 26, 2012
Messages
4,824
Location
Calgary, Alberta, Canada
Yeah, exactly. I definitely know that propulsion integration (drive shaft and cooling) is the most ambitious part of this design. I’m not all /that/ worried about ground cooling, although perhaps I should be, since my taxis are short and run ups are limited — my intuition is that the heat capacity of the system is fine until take-off; but I’m definitely worried about climb, and like all of us I’m worried about cruise drag, although probably more in theory than in practice since this is in no way a racer, and a few knots is just a few knots.
I think you have to adequately size the components for ground cooling in GA aircraft as you never know when you'll face a long hold somewhere. I've done 40 minute holds in some cases at 24C ambient and tested to 28C for 25 minutes where coolant temp stabilized at 194F. Cooling is better at 636 prop rpm, 1400 engine rpm than it is at 1000 engine rpm.

Racers and military aircraft can have less stringent ground cooling criteria.
 
Top