Air Inlets

[Mark Wright]

I have the Sam James cowl, and made my own cf plenum for an RV-14. I've designed the diffusers to go from the round inlets to the engine using CAD and came up with what I thought was a good design. Before finishing the fabrication, I thought I would run it through a flow analysis, and discovered some NOT so great results. The variables were set to calculate using a velocity of 71m/s, which is about 160mph, with an atmospheric temp of 0 C. As shown in the pics, the air becomes extremely turbulent, and actually flows the wrong direction in the area towards the top of the diffuser. I did not include the model of the cylinder in this analysis; however, I suspect this will make the airflow more turbulent and contribute to even less air into the plenum, which is where I want it. Going to try some different iterations of the diffuser and hopefully, correct this issue. I'm open to ideas on design improvement.

BTW, this is the left side.

 

[Richard Connell]

Hi Mark
I have a SP cowl but it looks similar in some respects.
I spent a lot of time going down a similar path to you.
Some observations:
It looks to me that the transition from the ring to the ramp is too sharp.
After wasting way too much time I ended up butchering the SP cowl ramps and making something with a similar transition to theirs. Perhaps look at the vans stock ramp shape for inspiration too.

The constraint (at least in an RV10) is that there isnt much gap between the front of the cowl and the front of the cylinders. Particularly on #1. So you need to have some aggressive changes in direction in a short distance.
Most 4Cyl James cowls that I’m aware of use a prop extension to get some more distance. Are you doing that?
I know of one RV10 that has an extended hub prop to also lengthen out this distance (although I believe this was to make room for an AC compressor)
If I had my time again (which I won’t!) that’s what I’d do.
Here are some pics.
I have lots more if you want to Pm me.
Cheers

 

[DanH]

Is 140 knots through the inlet an accurate value? Tell us the inlet diameter and true airspeed.

 

[Mark Wright]

Thanks Richard for the pics. I'll redesign and soften some of the transitions. Unfortunately, I think that the inlets sit down too low in relation to where the air needs to be directed. Dan, I think I saw a change you made to your cowling in this regard, moving the inlets up compared to stock??

No prop extension, so yes the transition zone is too short. I'm wondering if some fins added to the inlet ring to direct the air flow would work? Something similar to the interior vents in a vehicle. I'll model this and run the simulation.

I ran several iterations with different velocities, but left the temp the same. The results were the same with lots of turbulence and air flowing in the wrong direction near the upper portion of the diffuser. The ring diameter is just shy of 5 inches. I'll run the the simulation using some different atmospheric pressures to see if the results change.

 

[Mark Wright]

Possible Solution

Adding a ramp from the front of the inlet to the back of the diffuser made a big difference in linear air flow without the turbulence. These flow vectors look much better. Going to experiment to find the optimal angle and shape for the ramp. The results show a reduction in velocity of about 30 percent from inlet to outlet. I don't know enough about fluid dynamics or aircraft engineering to know if this is a good result or not.

 

[DanH]

Dan, I think I saw a change you made to your cowling in this regard, moving the inlets up compared to stock??

I did, see below. However, note they are also 6" diameter, and for the 390, the cooling mass requirement (lbs per second) is less, so the exit is throttled. That means velocity through the inlet is probably quite a lot lower than what you've modeled. Not saying it will save a badly shaped duct, but it helps.

Fired up a little spreadsheet for you. For a 435 CHT at 75% the 540 needs about 3.25 lbs/sec mass flow. The 390 needs about 2.0 under the same conditions. Assume 125 KTAS and standard day at 2000 ft. Your stream tube diameter is 4.66" through a 5" inlet, so area ratio is 0.87. My stream tube diameter is 3.52 through a 6" inlet, so area ratio is 0.35.

If you want to stick with a high Vi/Vo design, you'll need to seriously improve the duct shape, as you're very much at the internal diffusion end of the scale. Even for my low Vi/Vo setup, I made sure there were no abrupt section changes of significance.

Long time ago we got pressure measurements of a short James cowl with poorly shaped plenum inlets. The coefficient of pressure was quite a lot lower than a stock RV-8 cowl measured at the same time, or my low Vi/Vo cowl. The indication nicely matches your simulation. The separation kills pressure recovery.

No prop extension, so yes the transition zone is too short.

Yep.

In my case, CG drove cooling design. The -8 tends to be nose heavy with an angle valve motor. A 390 and a Hartzell metal prop made it worse, so a prop extension was out of the question. No available length for ducts, so I went with low Vi/Vo, which doesn't really need a duct. The rubber ducts and plenum lid you see here are to eliminate leakage, not to improve pressure recovery.

 

[Mark Wright]

Ok, thanks Dan! I've done some math and calculated the mass flow at the end of the air duct, and it is roughly 12-13 lbs per second based on the simulation, if the airflow was constant throughout the entire cross section of the inlet; which it is not. Because of the poor design of the inlet, I would estimate about a 30% efficiency (just eye-balling the air flow vectors), which would mean the mass flow would be somewhere in the neighborhood of about 3-4 lbs per second, certainly enough to meet the requirements for cooling. But the Vi/Vo ratio would still not be good(?) since I'm still seeing a high velocity at the exit area of the inlet, is this correct? Aside from moving the inlet vents up, what are some design ideas to help mitigate the turbulence and high velocity at the outlet portion? I'll change all of the abrupt angles as much as possible, but the alignment is still poor. Putting a ramp in the bottom of the inlet influenced the directional issue, but comes at a price of keeping velocity high....

 

[DanH]

I've done some math and calculated the mass flow at the end of the air duct, and it is roughly 12-13 lbs per second based on the simulation if the airflow was constant throughout the entire cross section of the inlet; which it is not.

You have a 5" ring. Area is 19.625 sq in, or 0.13628 sq ft
Velocity (from your first post) is 160mph or 235 ft per second.
0.13628 x 235 = 32 cubic feet
Standard density at 2000 feet is 0.072 lbs per cubic foot.
So, 32 x 0.072 = 2.3 lbs per second for one inlet.
2 inlets = 4.6 lbs if there is no throttling downstream of the inlets. That would include flow restriction due to separation.

Either go low Vi/Vo, or do whatever it takes to go high Vi/Vo without separation...meaning a good internal diffuser. That's not the one with the ramp, and I don't think it can be done well without a propshaft extension.

But the Vi/Vo ratio would still not be good(?)

There is no good or bad Vi/Vo. The ratio of inlet velocity to freestream velocity merely indicates where the cooling flow is being slowed, either externally, out in front of the inlet (low Vi/Vo) or internally, after the air has passed through the inlet (high Vi/Vo). Properly designed, either works fine.

Dave Anders recently wrote a nice article focusing on high Vi/Vo design, although he didn't actually use the term. It's arguably the superior approach in terms of least external drag for a high and fast airplane in cruise. Get everything just right and there can't be any external separation around the inlets. But the internal diffuser shape can be critical.

A low Vi/Vo inlet must re-direct some of the freestream around the outside of the cowl, and if the inlet lip and cowl just aft of it are poorly shaped, it separates externally, just like your bad internal separation. However it need not do so in any significant degree. Take a look at the early NACA papers on radial cowl development. You'll find pressure plots, and see which leading edge shapes tended to separate. Classic radial cowls are just big examples of a low Vi/Vo inlet.

Aside from moving the inlet vents up, what are some design ideas to help mitigate the turbulence and high velocity at the outlet portion? I'll change all of the abrupt angles as much as possible, but the alignment is still poor.

It's a classic losing battle...pick a short cowl and plenum lid, and then try to couple them together as an afterthought. It's a better approach if a builder picks high or low Vi/Vo, and then optimizes the entire system to suit the choice.

Putting a ramp in the bottom of the inlet influenced the directional issue, but comes at a price of keeping velocity high....

...which is a heck of a lot better than losing the energy into turbulence. The energy is still in the airstream as velocity. It going to hit the fins and back wall, and slow in that way, raising pressure. It will work (lots of RV installs are doing precisely that), but it's not super efficient. For example, lots of lost energy in plain 'ole surface friction in the high velocity duct.

BTW, as my mentor liked to point out, low Vi/Vo diffusion is frictionless.

 

[Mark Wright]

Ok, dang. Live and learn, it makes sense, too late, that the air flow system really needs to be designed around the inlets. Sounds funny to say it now since it seems so obvious. With this high Vi/Vo system, I'll research and play with designs to find the most efficient diffuser. Silk purse from a sows ear at this point.

Earlier you talked about stream tube, which I understand the concept but I don't understand how you calculated it.

 

[DanH]

With this high Vi/Vo system, I'll research and play with designs to find the most efficient diffuser. Silk purse from a sows ear at this point.

Take a trip over to Chris Zvatson's site; nice application of a high Vi/Vo system:

https://www.n91cz.net/

Mentioned previously, Dave just did a nice article in Kitplanes focused on a high Vi/Vo system:

https://www.kitplanes.com/build-a-custom-plenum-that-will-work/

Good article, but with respect, I think the title is terrible. Plenum lids are easy. It's a system; inlet, ducting, baffles, and exit. Screw up one and none of it works well.

Personally I think a low Vi/Vo system is a better choice for all-around performance in the practical world. An RV need not be a one-trick pony.

 

[scsmith]

Another way to look at it

Mark,

Your flow analysis model is missing a key thing. You arbitrarily set an inlet velocity, but that's not how the system works. The presence of the cylinders, and all the internal flow paths, followed by an exit area, all have a big influence on what the inlet diffuser duct actually sees in terms of inlet velocity. The combination of all the pressure losses through the system with the exit area and pressure sets the mass flow rate through the system.

The ensuing discussion above gets to all that by comparing inlet velocity to exit velocity (Vi/Vo), but leaves out the question of what the overall mass flowrate is.

As the exit area is reduced, and exit velocity increased, the mass flow is also reduced. The inlet velocity is decreased because the flow starts slowing down out in front of the airplane. What comes in through the inlet is slower, higher pressure air that doesn't need to diffuse much more, so the performance of your diffusers becomes much less important. With an appropriately sized exit, your diffusers may work pretty well as they are. Not to say there isn't room for improvement in your shapes, but may not be necessary.

 

[DanH]

The combination of all the pressure losses through the system with the exit area and pressure sets the mass flow rate through the system.

The ensuing discussion above gets to all that by comparing inlet velocity to exit velocity (Vi/Vo)....

Minor note, but I use Vi/Vo as defined in CR3405, Vi being velocity through the inlet, and Vo being velocity approaching the inlet, i.e. aircraft velocity.

...but leaves out the question of what the overall mass flowrate is.

Post #6 worked backwards from Lycoming cooling charts. There is a grand assumption there, as I really don't what Lycoming's "standard" dyno room baffle looked like, or how restrictive it may have been. There has to be a good bit of variation, installation to installation, but the charts get us in the ballpark. 540 chart attached.

As the exit area is reduced, and exit velocity increased, the mass flow is also reduced.

And upper plenum pressure rises. At one point I had a pitot static probe in my primary exit, as well as piccolo tubes in the upper plenum. Old data below. Note upper plenum pressure rise and increased delta across the pitot static as the aux exit door is closed.

Upper plenum pressure above static pressure vs exit door position
Position in seconds of servo motor run. (picolo tubes in upper plenum)

Exit velocity vs door position (exit pitot-static probe)

All pressures are inches H2O

position.....pressure......% of Potential Q.....Exit delta
Open.........11.36..........0.7126.................0.91
2..............11.63...........0.7296.................1.15
4..............12.03...........0.7547.................1.27
6..............12.3.............0.7716.................1.35
8..............12.3.............0.7716.................1.45
closed.......12.35...........0.7747.................1.46

 

[DanH]

Earlier you talked about stream tube, which I understand the concept but I don't understand how you calculated it.

Divide required mass flow from cooling chart (lbs per second) by the number of inlets.

Convert to volume per second by dividing lbs per second by density (lbs per cubic ft) for the altitude of interest.

The volume of a cylinder is base area x height. Area is radius squared x pi.
Here height = length = distance traveled in one second at a given velocity. So, divide volume (cubic feet) by distance traveled in one second (feet per second), then divide by 3.14, then find the square root, then double it, then multiply by 12.

Or just download a spreadsheet here: https://www.danhorton.net/

 

[wirejock]

Inlet

Divide required mass flow from cooling chart (lbs per second) by the number of inlets.

Convert to volume per second by dividing lbs per second by density (lbs per cubic ft) for the altitude of interest.

The volume of a cylinder is base area x height. Area is radius squared x pi.
Here height = length = distance traveled in one second at a given velocity. So, divide volume (cubic feet) by distance traveled in one second (feet per second), then divide by 3.14, then find the square root, then double it, then multiply by 12.

Or just download a spreadsheet here: https://www.danhorton.net/

I recall you, Dan, mentioned the Vans inlets are low velocit/high volume. Is that right? Does that mean the plenum inlet shape is less critical?

Mine were totally unscientifically favricated. Basically an extension added to the RV Bits plenum.

 

[scsmith]

Minor note, but I use Vi/Vo as defined in CR3405, Vi being velocity through the inlet, and Vo being velocity approaching the inlet, i.e. aircraft velocity.

Ah, thanks for the correction Dan! I had mentally shifted gears and was thinking of Vi/Ve. (exit).

So the trick to calculating the Vi/Vo is estimating the system losses, as you did. Nice work. It would be fun to get a look at Lycoming's dyno setup wouldn't it. Aside from a few outliers (like Bill L) most RV's probably have very similar baffling around the cylinders, and it just comes down to the other 'leaks' and the oil cooler.

 

[DanH]

I recall you, Dan, mentioned the Vans inlets are low velocit/high volume. Is that right? Does that mean the plenum inlet shape is less critical?

Make a careful measurement of your stock inlet area, and apply it in the second column of the spreadsheet linked above. Area ratio tells where you are on the Vi/Vo scale.

Important note..."low velocity high volume" is incorrect in two ways.

We're interested in mass, not volume. For a given mass, volume varies with density, i.e. altitude.

Second, as Steve and I keep trying to tell everyone, mass in lbs per second is set by the baffle restriction and cowl exit area. A big inlet is slow flowing because of throttling downstream. A low velocity inlet configuration and a high velocity inlet configuration must flow the same mass if they are to equally cool the same engine.

Yes, plenum inlet duct shape becomes less critical as velocity through the inlet is reduced. A really slow inlet (low Vi/Vo) needs no duct at all. Ask Mark Frederick about changing his Rocket to low Vi/Vo. The inlets are 6" diameter, dumping directly into the upper space, no ducts. They're the same 6" inlet as on my cowl, but his velocity ratio is usually lower because the Rocket is operating at a higher airspeed.

I would not assume all Vans inlets are the same. For example, the inlets on 6's appear smaller than later models.

 

[Mark Wright]

Thanks Dan and Steve! Steve, thanks for the further explanation on the mass flow....it makes sense that the "air system" fills up and the equilibrium concept. Makes me feel a little better about the design I have. With the simulations I have run, the best case scenario is to have a direct line between the inlet and the plenum. Any alterations in direction vectors that go inward toward the primary flow or stream tube, cause an exponential increase in velocity. I needs to be said that these simulations provided valuable insights, real-world testing might introduce unforeseen factors, and usually do.

I'll post some results of design iterations and results and hopefully get some feedback from both of you to optimize the design further, considering both pressure losses and diffuser performance across different flow regimes. Thank you both very much!

 

[FireMedic_2009]

So have you determined about how much airspeed you were able to pickup by reducing the cooling drag compared to the regular stock cowls?

 

[DanH]

So have you determined about how much airspeed you were able to pickup by reducing the cooling drag compared to the regular stock cowls?

No way to compare. I didn't fly this airframe with a stock cowl.

When Marc Cook was to fly the then-new 119 version of the RV-14A, I asked him for a specific test. Go to altitude, set power, lock the autopilot, and record the TAS with the aux air door open, then closed. I wasn't there and thus have no opinion regarding test accuracy, but the reported difference was three knots. Bottom of the page here if you want to read the whole thing: https://www.kitplanes.com/vans-rv-14a/

Here's the important caveat. Part of that increase is truly a cooling drag reduction, which can be defined as mass x loss of momentum. Closing the aux tunnel door reduces mass flow, increases upper and lower plenum pressures, and increases velocity through the exits around the tailpipes. Less mass x less momentum loss = less cooling drag.

However, we can comfortably assume the above is only partly responsible for the three knot difference. When the aux tunnel door is open, it is a significant source of external drag. Think of it as a small version of the speed brake seen on early jet fighters, the F-86 being an example. Closing it eliminates a lot of drag.

I've been flying a variable exit about ten years now. The speed difference, open vs closed, is only about 1.5 knots. Why? The outlet door is wider, and opens at a shallow angle. It appears to generate less external drag when open, as compared to the -14's door.

Comparing the earlier RV-14 cowl to the 119 version, or a stock RV-8 cowl to mine, you'll see both lopped off a significant chunk of frontal area when the RV-standard "reverse coal shovel" exit was removed from the belly.

The subject here is inlets. Both of the above exit systems would work just fine with either a high or low velocity ratio inlet...if the particular inlet design is efficient. However, if the inlet and its attached ducting wastes a lot of the available dynamic pressure, rather than converting it to increased static pressure, the exit velocity won't increase very much when the aux door is closed. After the pressure drop across the engine baffles, there would be no significant pressure remaining in the lower plenum to drive exit velocity. It does not mean the poor inlet won't cool the engine. It just means the system isn't particularly efficient.
.

 

[gmcjetpilot]

Dan is all over this. I'll add there is no perfect only compromise based on the boundary condtion of a single engine air cooled Horz opposed direct drive tractor airplane configuration.

1) Computer modeling CFD, computational fluid dynamic does not account for how the prop beats the air at cowl inlet. The blade near hub is not an airfoil. So inlet flow in your transition is hard to model. There is empirical data. It varies widely for air speed and angle of attack. You can not optimize over a wide range... compromise.

My point is follow what others did. There is no real gain in that transition from cowl inlet to plenum over best practice dictated by form, fit, function of the fixed geometry. NO LEAKS is important.

2) CFD also does not reflect all the variables and wide conditions after your cowl inlet transition going into plenum and then cowl exit , complex, important. It's part of the "cooling system". It has to be sized and matched. The down draft air into lower cowl has to go around a lot of obstacles, engine mount, exhaust, gear, hoses, wires. Builders try to make fairings inside the cowl to redirect the air. I applaud that. Benifit? I personally would not do it. work, weight, cost, impeding access for maintenance, for small gain (typically not measurable I recall).

Again NO perfect air flow, CFD or not, in the madness that is cooling air inside your cowl. Minimize the drag and optimize and size as best you can. That is all you can do. The base line best practice is good enough for most. It is fun and educational to experiment. This was idea of the EAB category.

With hot rodded high compression engines insufficient cooling is common. Cowl flap or flaps is typical solution. It adds cooling drag, weight, complexity over fixed cowl air exit outlet. That is OK. Better than engine damage.

OIL cooler size, location and air flow through it is of big importance. Lycoming's are not only air cooled but oil cooled as you know... Due to available space oil coolers can be a challenge. Get the best largest one you can get. With your advanced understanding of air flow you can research it. I was the first to call out the poorly sized optional Van's oil cooler install kits of long ago.... 4" SCAT is a min. Think of this, you "steal air" from upper pressure plenum for oil cooler. However keep in mind that "stolen air" for the cooler dumping into lower cowl may reduce differential pressure across cylinders. Oil cooler exit air s/b into a low pressure area not increase lower cowl pressure or block bulk cowl exit air. I don't have a brilliant idea just a comment for thought. Some have made isolated dedicated oil cooler air inlet and/or exit. You see this on some certified planes. Radials might have a large cooler outside cowling in the breeze. This is overkill and too much drag for RV I think.

The links you got above, data, observations, experiments others have done are great. Of course this NASA funded cooling study 50 years ago gives insight.

https://ntrs.nasa.gov/citations/19810013485

Have Fun

 

[gmcjetpilot]

Dan is all over this. I'll add there is no perfect only compromise based on the boundary condtion of a single engine air cooled Horz oppised direct drive tractor airplane configuration.

1) Computer modeling CFD, computational fluid dynamic does not account for how the prop beats the air at cowl inlet. The blade near hub is not an airfoil. So inlet flow in your transition is hard to model. There is empirical data. It varies widely for air speed and angle of attack. You can not optimize over a wide range...

My point is follow what others did. There is no real gain in that transition from cowl inlet to plenum over best practice dictated by form fit function of the fixed geometry. NO LEAKS is important.

2) CFD does not reflect all the variables and wide conditions after your cowl inlet transition going into plenum and cowl exit , complex, important. It's part of the "cooling system". It has to be sized and matched. The down draft air into lower cowl has to go around a lot of obstacles, engine mount, exhaust, gear, hoses, wires. Builders try to make fairings inside the cowl to redirect the air. I applaud that. Benifit? I personally woukd not do it. Work, weight, cost, impeding access fir maintenance, for small gain (typically not measurable).

Again NO perfect air flow in the madness of air cooling inside your cowl. Minimize the drag and optimize and size as best you can. That is all you can do. The base line best practice is good enough.

However with hot rodded IO39O's insufficient cooling is common. Cowl flap or flaps is typical solution. It adds cooling drag, weight, complexityover fixed cowl outle. That is OK. Better than engine damage.

OIL cooler size, location and air flow through it is of big importance. Lycomings are not only air cooled but oil cooled as you know... Due to available space oil coolers can be a challenge. Get the best largest one you can get. With your advanced understanding of air flow you can research it. I was the first to call out the poorly sized Van's oil cooler install kits of long ago.... 4" SCAT is a min.

Think of this, you "steal air" from upper pressure plenum for oil cooler. Some have made isolated dedicated oil cooler air inlet and/or exit. You see this on some certified planes. Radials might have a large cooler outside cowling in the breeze. This is overkill and too much drag for RV I think. . However keep in mind that "stolen air" for the cooler dumping into lower cowl may reduce differential pressure across cylinders. Oil cooler exit aur s/b into a low pressure area not increase lower cowl pressure or block bulk cowl exit air. I don't have a brilliant idea just a comment for thought.

The links you got above, dara, observations, experiments others have done are great. Of course this NASA funded cooling study 50 years ago gives insight.

https://ntrs.nasa.gov/citations/19810013485

Have Fun

 

[DanH]

1) Computer modeling CFD, computational fluid dynamic does not account for how the prop beats the air at cowl inlet. The blade near hub is not an airfoil. So inlet flow in your transition is hard to model. There is empirical data. It varies widely for air speed and angle of attack. You can not optimize over a wide range...

I can't speak for modeling, but I have measured upper plenum pressures in slow climb at full power. The propeller can make a very significant contribution to available dynamic pressure, pushing upper plenum Cp (as defined in CR3405) well above 1.25 with good inlet design. "Prop beating" is irrelevant.

My point is follow what others did. There is no real gain in that transition from cowl inlet to plenum over best practice dictated by form fit function of the fixed geometry.

Both builders and manufacturers have made real gains with inlet changes. A lot of "do what others did" is simply not very good.

However with hot rodded IO39O's insufficient cooling is common. Cowl flap or flaps is typical solution. It adds cooling drag, weight, complexityover fixed cowl outle. That is OK. Better than engine damage.

Ummm, George, all the angle valve models are easy to cool in terms of CHT, as they have more fin area than a parallel valve. A builder suffering with with high CHT (>375) on a 390 is doing something wrong. The best installs are often too cool.

The "cowl flaps" posted previously (the 14-119 and my -8) are there to reduce exit area when closed, as compared to the prior standard configurations.

A current project (below) adds variable exit area to a first version -14 cowl. The goal was to enable closing off a bunch of exit area while not adding significant external drag. So far the results are good CHT control (i.e. it can be raised when desired), no significant speed change, and a little higher oil temperature.

Think of this, you "steal air" from upper pressure plenum for oil cooler.

Not really. A cooler isn't a giant leak. It offers resistance to flow, just like the baffled fins. See example below, a SW/Meggitt 10611, the cooler I'm using on a 390. A heat rejection of 500 BTU requires about 40 lbs per minute, so the pressure drop is about 6" H2O. The angle valve cooling chart says the same 6" drop across Lycoming's standard baffle flows about 2 lbs per second, or 120 lbs per minute. The cowl exit sets total mass flow. In this example the oil cooler simply flows 25% of total mass. We can shift that percentage with a larger or smaller cooler, or a change in cooler ducting, or a change in cylinder baffle restriction, or a change in cooler exit local pressure.
.

 

[gmcjetpilot]

The propeller can make a very significant contribution to available dynamic pressure, pushing upper plenum Cp (as defined in CR3405) well above 1.25 with good inlet design. "Prop beating" is irrelevant.

It can if you move cowl inlets outboard and why the round inlets popularized by LoPresti based on that NASA 1970's NASA study. The STOCK Van inlet has area right near the beat to death club part of the prop. The Book Speed with Economy, by Kent Paser showed placing the induction inlet closer to prop improved MAP.

Both builders and manufacturers have made real gains with inlet changes. A lot of "do what others did" is simply not very good.

Yes Sir.... I know as stated above.

Ummm, George, all the angle valve models are easy to cool in terms of CHT, as they have more fin area than a parallel valve. A builder suffering with with high CHT (>375) on a 390 is doing something wrong. The best installs are often too cool.

I have less angle valve experience. That is interesting. Yep I never had any real issue cooling 320/360's with parallel valves. Yes winter you have to block things off for sure, oil cooler and with stock Van's cowl I had plugs near the inner corners of Cowl inlet (speed taped om). I see these IO390's, with higher compression, ported polished I assume they are more of a challenge to cool.

The "cowl flaps" posted previously (the 14-119 and my -8) are there to reduce exit area when closed, as compared to the prior standard configurations.

Reducing cooling drag and not over cooling in cruise and having enough cooling in climb and/or hot weather. It works the trick

Fixed cooling system is sized as a compromise. I was flying the C182 this weekend. It has cowl flaps (two flaps one handle). I was moving the cowl flap handle. It was so cold I taxied, ran-up and did initial takeoff and climb with cowl flaps closed. Summer they would be wide open... It is really about NOT over cooling in the winter. In the summer it is all about keeping CHT and OP temps happy, cowl flap open all the time except cruise. The C182 has many cowl flap detents between full closed and open. In real hot temps you may cruise with cowl flaps partly open... It gives you control.

A current project (below) adds variable exit area to a first version -14 cowl. The goal was to enable closing off a bunch of exit area while not adding significant external drag. So far the results are good CHT control (i.e. it can be raised when desired), no significant speed change, and a little higher oil temperature.

Nice... I saw the picture... the coated exhaust inner surface. I calculated the time and work... I am going fixed outlet with my RV7 O360 180HP (8.5 to 1). If I need to add more area I can consider some type of cowl flap... Nice work.

Not really. A cooler isn't a giant leak. It offers resistance to flow, just like the baffled fins. See example below, a SW/Meggitt 10611, the cooler I'm using on a 390. A heat rejection of 500 BTU requires about 40 lbs per minute, so the pressure drop is about 6" H2O. The angle valve cooling chart says the same 6" drop across Lycoming's standard baffle flows about 2 lbs per second, or 120 lbs per minute. The cowl exit sets total mass flow. In this example the oil cooler simply flows 25% of total mass. We can shift that percentage with a larger or smaller cooler, or a change in cooler ducting, or a change in cylinder baffle restriction, or a change in cooler exit local pressure.

That is great data... however I do think having a good EXIT to the oil cooler, a diffuser can have benefits. You want lower pressure on the exit side of cooler. That cooler exit ducting on can be routed to near the cowl exit, a lower pressure area. The bonus is less internal flow "interference" drag, not interfering with differential down draft air across the cylinders.

The old twin I owned long ago, had the exhaust go through augmenter tube to induce airflow out of the cowl and mix the exhaust and cowl air for min drag. I think there is some data on this for way back. In the air it makes a nice unique exhaust note, over exhaust that dumps into the slip stream direct. On a twin you have room. On a RV a larger tunnel, taller and longer, under the plane to route exhaust might be needed. If you mount it all external, cooling efficiency and drag reduction would be mitigated by external parasitic drag and it would add weight.

Jet exhaust? I do have a longer secondary exhaust pipe on my 4 into 1... which in theory provides some thrust. However it beats the floor boards up. So I am adding a small down turn at the end of the exhaust, to redirect the exhaust slightly down. Less "thrust" but better for vibrations and noise in cockpit.

It is fun however to put theory to practice, and you do a great job documenting that. It may even "work". The issue is definition of "work", It may mean 0.5 mph increase in cruise or a few degrees lower CHT or OT.... Hard to measure. But that is a success. I find Van got it pretty close in stock form. As I said in my first post it is a compromise, always.