Considerations in Engine Cowl Design

[facthunter]

Prop tip speed becoming supersonic makes great noises and uses power.  Good props are remarkably efficient at the top end (above 80%). Better take off performance than jets.  (Thrust).. Nev

[Mike Gearon]

On 27/6/2022 at 6:01 AM, APenNameAndThatA said:

I thought drag was proportional to velocity squared. And power required was proportional to velocity cubed. And the difference was because you need more power to go faster for a given amount of drag. P.S. I am amazed by how much you and others on here know. 

I was thinking the same. My job is to have these smart guys do the work and the experimenting and I’ll just apply the results.

 

A note on cowlings….. The Pipistrel range being derived from gliders are very serious about drag. The Pipistrel Virus I own in Oz  is a 145kn cruise with 100hp except it isn’t. Short version of the story is the 912 fuel injected was causing problems  with heat. The early versions ran 145kn cruise and they changed the cowling because of the heat issues in taxi and climb. Dropped 10kn off the cruise. 

[RFguy]

What I'd like to become accustomed with exactly how workwhile putting tape over the door gaps are, fixing up the struct-wing attachments (bit dirty) etc.

compared to minimising front area and optimising frontal shape etc.

The glider people do the tape thing of course.

 

The 230 while super smooth and really a super slick plane could have some improvements......is a bit untidy aerowise around strut attachments, main gear arch etc. 

It might not look like much area but when you consider thoise untidy bits are probably Cd=1 compare that to the drag coefficient Cd of the wing which might be 0.004  in S&L  !!!

 

so while the front area of the entire wing front area might be something like 1 square meter, its only 0.004 sq.meters as far as drag is concerned.

and a small flat plate area maybe 30cm x 10cm has a drag area of 0.03 sq. meters

 

so, makes sense to only open up your radiator area just to get just enough air you need to do the job. 

IE per the recent calcs - something like air inlet size 225cm2  (for the water) will do it for the rotax at 70 kts. have a big space behind the frontal opening  and use the full 33cm x 14.5cm radiator .

One third for the oil....

Going faster ? you might as well if you can organise, reduce the inlet size since there is at speed, more air coming in (its all about kg per second of air) .

 

Note that the 225cm2 number for an opening is for a square onto the airflow opening. half way down the cowl under the prop at 45deg to the airflow, suggest make it 1.4 x bigger.

 

Edited June 29, 2022 by RFguy

[pmccarthy]

Going from the Foxbat to the Vixxen they claimed increase in cruise from 85-90 kts to 110-115 kts with the same wing and engine, just cleaning up the fuselage and drag, with a lot of research.

[RFguy]

I've in recent times got a whole load of numbers for various fairly thin 30-35mm vehicle radiators , these are usualy two core deep.

pressure drop for our speeds is roughly linear to airspeed.

I'm working on something like 400 Pa (1.6")  drop for 36m/s airflow.

typical performance may be 520kW/m2 at 36m/s

if you look at the rotax radiator, 25kW it is speced for, and area is 33 x 15cm..(0.05sqM)  plugged into the above (0.05 x 520)  yields, surprise surprise, 25kW...!

However!

36m/s at the nose is not going to be 36m/s through the core . why- pressure drop !!

Benoulli says

velocity fluid = sqrt( 2x pressure difference / density)

If we have 36 m/s on the nose, that's 800Pa . And if the radiator drop at 36 m/s is 400 Pa, then you only have 400 Pa left over !

Plugging that into the equation yields, for that pressure drop - only 25 m/s airflow in the core . (it will be a bit different since airflow is down so pressure drop is down) . 

 

This is why cowling pressure is so critical and positive cowling pressure WRT atmo is bad for cooling ! (when the radiator is on the front or getting ram air) 

Fortunately, Pressure difference is proportional V squared , so increasing airspeed rapidly improves cooling.

 

this brings into thought that really deep radiators- they'll have much higher pressure drops.  They do have a higher 'efficiency' in terms of the amount of cooling that is done (Tdiff water) versus Tdiff air..

but at a cost of the need for much higher pressure difference. As pressure difference is V squared, the faster you go, they should work well. But that's likely one of the reasons why we dont see deep core radiators in town cars because the air speed is too slow to get the pressure up.  (my guess) 

glen.

[Ian]

I'm not sure cowling pressure per se is bad for cooling.

To create efficient low drag cooling you want small inlets and outlets and a large area of slow moving air across the radiator.

Essentially the reverse of the below.

http://hyperphysics.phy-astr.gsu.edu/hbase/imgmec/bernoul.gif

 

High velocity airflow across the radiator is bad it equals lots of drag.

Large thin radiators are more effective than smaller thicker radiators as the speed of the flow through the radiator can be lower for the same amount of cooling. Remember drag is proportional to v^2

Large inlets and outlets are bad (however you need them large enough to provide the requisite heat flow) and generally it's more effective throttling the outlet.

This means that the additional heat you add to the flow can do work (but not really at the velocities we've talking about) However it allows you to minimise drag.

 

WW2 was the epitome of radiator design as turbine engines made this field pretty much redundant. But there are some gems to be found in this area.

 

https://history.nasa.gov/SP-445/ch5-5.htm

https://reports.aerade.cranfield.ac.uk/bitstream/handle/1826.2/1425/arc-rm-1683.pdf

 

 

Edited July 2, 2022 by Ian
image wasn't working

[RFguy]

if there is too much pressure inside the cowl then you cant get enough pressure across the (forward) radiator and that will limit airflow.

I contest the statement that "High velocity airflow across the radiator is bad it equals lots of drag."

 

WHy ? because air will go through the radiator and exit somewhere useful, hopefully. 

once you stick a radiator in the airstream, on the nose of an airplane (not off the wing)  the difference in drag whether it has low airflow or high airflow through it  is peanuts. that's because once you have stuck a resriction there with about 25 to 50% pressure drop of the available pressure, it might as well be  a wall. That's whether it is fast or slow flow.  SO you might as well have fast flow, because with fast flow, and hence highest mass flow and pressure difference and thus  MINIMISE the frontal area to do the job. 

 

Now, what you do with what has come through the radiator is another story. yes for sure-  Then there is the opportunity to make good on some of that drag . 

 

Ideally the front facing radiator would have minimal pressure drop and be essentially transparent. But they're not,  The Cd of a perforated plate compared to a solid plate is not too different- they are all a very long way from a wing .

 

Ideally the outlet would be at the same energy that what came in. that is tricky because some work has to be done with pressure drop acrosss the radiator, the idea is to offset the energy added in the form of heat to make up for the energy lost doing the radiatior air work.

If we are talking the rotax radiator, 15x 33 = 0.05sqM, this amounts to at 51 m/s a drag cost of approx 4kW or 5.3HP if you dont gain any of it back.  My guess is most aircraft that are NOT tailored for expoiting the cowling eixt thrust are probably > 90% loss.   So, there is 5HP to get back, potentially. 

For the practical purposes of this discussion, we're talking about a Rotax setup. Not a scalding hot Lycoming or Jabiru. These have much higher air temperature rise across their exchangers. (which is why air cooling works well if you can tolerate the high temps- the temperature difference rules) 

 

 

Edited July 2, 2022 by RFguy

[Mike Gearon]

Reading all this I have the urge to create a lower cowling that has a large surface of stainless steel. Double skinned. Run your water pipes through it. Give it some fins. Added drag. Bugger all. 
 

Now, as with exercise I just need to wait until the urge subsides. I’ve made a promise not to invent in aviation. Just fly by the rules and follow advice.

[RFguy]

Hi Mike

great minds think alike . having the cowls, at least the top cowl as  one giant heat exchanger was originally on my mind.

Stainless is not good thermal conductivity too low, although if loads of copper/ally water tubes were bonded that might work. But Aly for the whole thing would be great.

 

Alas it's only going to be really effective in the high pressure region where the airflow is being deflected, not in the low pressure low velocity boundary region near the skins, the airflow is quite low and cooling performance per sqM is low.... It's amazing to think a tiny conventional radiator has several square meters of surface area ! but something like the leading edges, or the struts, they're useful..

there are some extrusions used for intercoolers etc that are thin flat channels , the problem is forming them into organic shapes like a cowl is more difficult (but possible)

I have suggested to SKippy that his exit airflow radiator should work, as long as he can pressurize the cowling to about 1.5kPa and it (with radiator blocked) holds pressure losing less than 100 litres per second. a bit like a leakdown test,,,,,, (airflow to do the job is about 1kg/sec !

 

 

 

 

[Geoff_H]

One consideration that does not seem to be taken into account in this thread is that lamina flow reduces heat exchanger transfer capability. Laminar flow creates an insulating layer next to the transfer 'metal'.  Velocity of fluids at a surface is zero for laminar flow.  I was taught that low pressure air acts like a liquid, I also have observed wall attachment at low flows that affects the flow through heat exchangers. I believe that the only way to study this problem is with finite element analysis.

[RFguy]

Geoff I'd think its unlikely we'd have laminar flow in any of the conceived areas I discussed above,

But, still the boundary layer (airflow =0 at the surface) is the big problem.

The trend these days in electronics is turbulent flow, 

 

What might be counter intuitive but highly effective  is if the cowl (used for cooling) was a massive pin - fin heatsink.

IE the surface could look like this : it is quite realisable with additive manufacturing techniques (metal printing) 

 

 

 

 

Edited July 3, 2022 by RFguy

[Geoff_H]

With properly designed  inlet I would expect laminar flow in, the inside would then  ensure turbulent flow in the heat exchanger (reynolds number drastically increasing) and exiting turbulent.  One of the biggest problems is a design that works well at climb out velocities then works just as well a  cruise velocities.  Mooney used an exhaust duct, its opening size adjustable from the cockpit.  Mooney always have temperature issues, it was a constant job to adjust the duct.

[RFguy]

Geoff are you u referring to the oil cooler on the mooney ? (this this discussion is about fluid based engine cooling - not air cooled cylinders) . 

All of my number exam are for max performance at best climb. the assumption (!) is that once at cruise, which might be 1.5x to 1.7x  climb that the extra mass air flow is sufficient to do the job even if the airflow goes awry.

 

 

 

[Geoff_H]

The oil cooler was part of it.  But there is a lot of similarities between fins and "radiator" flow.  Metal cooled by air.  Big difference is that the  'radiator' style has to be more complex owing to lower temperature of metal.  Yes the Mooney fight was with oil temperatures.

[Geoff_H]

In the early days of my career I did lots of work with low pressure airflow including heat exchangers.  It was all designed with mathematics, get the older guys to use experience to modify the maths.  The measure extensively when commissioning and alter as needed.  It was very impirical.  The airflow was with furnace airflow.  I am sure that aircraft manufacturers go through several interactions of cowling design, I would expect that with any one off design would need redesign, a reason among others that manufacturers charge so much.

[Ian]

On 02/07/2022 at 4:54 PM, RFguy said:

I contest the statement that "High velocity airflow across the radiator is bad it equals lots of drag."

 

Rather than argue about this I'd point you toward the article that I linked to and the section on "Classification of drag" being divided into 3 categories.

1. Skin friction drag of stream flow.
2. Drag due to eddying arising from separation of a stream from a surface.
3. Drag due to expansion losses without actual stream separation.

Points 2 and 3 are avoidable

Thus the ideal system will be designed to

• to avoid stream separation or severe expansions
• To reduce the stream velocity over the cooling surface, and
• to reduce the external surface to a minimum.

Even when following these principles it's hard to get the design right hence the design with maths and allow experienced people to fudge.

2 hours ago, Geoff_H said:

It was all designed with mathematics, get the older guys to use experience to modify the maths.  The measure extensively when commissioning and alter as needed.  It was very empirical

 

[RFguy]

Ian, what's your take on the ability for Skippy's exit flow radiator to work at anticipated ?

(see the beginning of this thread for pictures).

 

 

 

 

[spacesailor]

I have a nagging thought !,

That

At max power/climbout the airflow over That cowling would be at MAXIMUM velocity! .

Due to you propelling the air to the Rear of your aircraft, by that propeller. 

UNLESS 

your using a " ducted fan " propulsion system.

spacesailor

[RFguy]

u mean high AoA leading to less than ideal pressure state and flow vector on the outside of the cowl exit ?

[Ian]

20 hours ago, RFguy said:

Ian, what's your take on the ability for Skippy's exit flow radiator to work at anticipated ?

A problem with most small planes is the lack of space and it's even worse in the engine bay. You're making compromises simply to make things fit. Cooling is really hard to get right.

The aim is to accelerate the outflow as much as possible in the final duct recovering as much energy as possible. One of the things with these designs is that they tended to make the exit duct adjustable and not the inlet, higher pressure in the inlet duct makes the air spill around the intake so a static oversized inlet would appear to be OK. It's so difficult getting this right intially I'd be tempted to go down the variable geometry path as it would allow me to tune in flight rather than build 5 ducts before I found the best compromise.

The other thing which people have alluded to it that the cooling requirements in climb are different to those in cruise so with any static design you're either over-cooling in cruise or cooking in climb.

But it's another thing to build and another thing to manage in flight.

These pictures are from another forum it's worth noting that the mosquito's design opted for a very thick radiator even though it's less efficient simply to make it fit.

The P51

 

 

 

[RFguy]

Good post Ian

Ian, I'd say at the flying speeds of the Mosquito,  say 100 m/s , > 200 kts, there is plenty of pressure to run a deep (thick)  core radiator . at 100m/s ....6kPa.... and the X sectional area of the radiator if size is tight  can be kept down so exploiting the advantages of velocity squared. 

Doing all these numbers of radiator pressure drops, its quite easy to see why cooling is in the toilet at low (<60 kts) flying speeds in some setups. This is where the pressure available is similar to, or not much greater than the pressure drop through a radiator.

Edited July 4, 2022 by RFguy

[Ian]

It's not only pressure drop. The air warms as it flows through the radiator, warm air cools less efficiently, so that back half of the radiator cools less efficiently,  or you need to move the air through the radiator faster compared to a thin radiator which creates more drag.

 

But I do like the mosquito as a plane, it's an example of material design compromise where the design leveraged the properties of the material rather than simply building a wooden plane like a metal one.

Apparently this is a quote from Hermann Goring

'The British, who can afford aluminium better than we can, knock together a beautiful wooden aircraft that every piano factory over there is building, and they give it a speed which they have now increased yet again. What do you make of that? They have the geniuses and we have the nincompoops!'

[Geoff_H]

17 minutes ago, Ian said:

It's not only pressure drop. The air warms as it flows through the radiator, warm air cools less efficiently, so that back half of the radiator cools less efficiently

That's why well designed heat exchangers are contra flow.  Hot flows one direction through heat exchanger, cooling fluid the reverse.  

 

[facthunter]

Heat transfer is directly proportional to temperature difference. That's the advantage of hotter running air cooled motors but all of them will gain from airspeed. If it cools on low speed climb it will be overcooled in cruise and DESCENT and all are subject to ambient temp variation. Adjustable EXIT air is the most used.  Nev

[Ian]

59 minutes ago, Geoff_H said:

That's why well designed heat exchangers are contra flow.  Hot flows one direction through heat exchanger, cooling fluid the reverse.  

I was wondering what the efficiency gains were and wikipedia just makes thing easy. https://en.wikipedia.org/wiki/Heat_exchanger#Flow_arrangement