What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Experience with Subaru at 358 Hours

SvingenB said:
From 0 to anything. That is (one of the main) problems using accelerometers for your application. If you take a closer look at Dans calculations, that is exactly what they tell you: http://www.vansairforce.com/community/showpost.php?p=742009&postcount=73

If you put an accelerometer on the coupling you might not even get the F2 (stiff coupling and steel wheel). Now, if you measure strain somewhere on the shaft, you will pick up everything. In addition you will get the exact strain/stress. Using accelerometers you will have to assume you have the correct mode shape (100% correct) and then calculate the forces. Then you want to get F3 as well. You need 2, preferably 3 accelerometers, you only need 1 strain gage and it will do a better job than the accelerometers.
Click to expand...

Please forgive my question if it seems foolish. I am trying to understand how one strain gauge on the prop shaft will give information on how well the coupler does its job damping spikes between the engine and gearbox? I can see it by making changes and observing the results in reduction at the prop shaft but I don't see how we can quantify things further back on the system the first time around.
Thanks.
 
Last edited:
rv6ejguy said:
Please forgive my question if it seems foolish. I am trying to understand how one strain gauge on the prop shaft will give information on how well the coupler does its job damping spikes between the engine and gearbox? I can see it by making changes and observing the results in reduction at the prop shaft but I don't see how we can quantify things further back on the system the first time around.
Thanks.
Click to expand...

Because the problem is tied to resonance of the complete system. At any other operation the torque is known and not a problem, except maybe start and shut down. The spikes are not the problem, they excite the system if they are small or large (of course smaller is better, but that's not a main issue). You can't remove resonance, but you can tune it so the frequencies are outside your normal operational range. The solution here is to lower the F2 without also lowering the F3 too much(considering Dan's model is a fairly good representation). Heavier flywheel and softer coupling will lower F2. But more stiffness in the propeller shaft may be necessary to prevent F3 to enter your envelope, or maybe not.

A strain gauge on the propeller shaft will show you all you need to know of both F2 and F3. Then you can use that to tune your mathematical model regarding stiffness and masses and odd effects from your gears, and when that is done you can redesign using your (corrected) model. If the process of starting and shutting down is the problem, which it may become by lowering F2, then maybe a slipper clutch or similar is needed.
 
Last edited:
Strain Gauges

+1 for using strain gauges. You want to make a direct measurement (with strain gauges) of the issue you are trying to diagnose, not an indirect measurement (with accelerometers) and then try to infer torsional vibrations from those results.

The engine and transmission torque is overlaid with the torsional vibrations. The best way to measure both is with a strain gauge on a convenient shaft. So if calibrated properly, you will measure actual torque at the propeller shaft (and thus be able to calcualate HP). You should also see the torsional vibrations, say on an oscilloscope, overlaid on this trace.

Accelerometers would be useful in diagnosing out of balance loads in the rotating components, but I don't think this is the object of the investigation.

I don't have this book but the table of contents appears relevent to this discussion. Have a look on the link and "click to look inside"
Machinery Vibration - Measurement and Analysis by Victor Wouk
http://www.amazon.com/Machinery-Vib...549305&sr=1-88&keywords=torsional+vibration#_
 
Last edited:
It is not practical to mount the strain gauge inside the propshaft nose I don't think, difficult to get it well bonded in there but also no way to get the signal out, wireless or otherwise as it is completely covered by metal parts of the prop hub and prop mechanism. There is about 1/2 inch of straight section available between the prop flange and redrive front seal. Not sure if that is enough space?

I posted a few new photos of the progress and of the flywheel mods today.
 
Sorry to be cryptic recently. Out of town, Ms. Patti's birthday.

rv6ejguy said:
That was my plan, yes. Do you see problems with that?
Click to expand...

Heeere little doggie......nice doggie......hold still, little doggie....;)

Sorry. I keed. Couldn't resist.

Seriously, stick to the known. The subject is hard enough without inventing your own measurement technology.

The strain gauge bridge is the least complex. The output is just a variable voltage. You need a radio or slip rings to get the signal off the rotating shaft, and a recording method which can be simple or elaborate. You get torque and frequency.

If you really, really want to look at the relative motions of the system elements you might consider magnetic or optical pickups. They are stationary installations so you don't need telemetry. Put a reluctor wheel and pickup behind the prop, another pickup at the big gear, a third near the flywheel's ring gear, and a rotary encoder or another reluctor/pickup on the accessory end of the crank. The rest is processing hardware and software; you're working with variations in rotational speed. Done right you get shaft twist in degrees and frequency, analogous to the calculated mode shape.

Here's the key: both methods have a huge industry knowledge base.

I am trying to understand how one strain gauge on the prop shaft will give information on how well the coupler does its job damping spikes between the engine and gearbox?
Click to expand...

Again, the coupler is a soft connecting stiffness, not a damper. A connecting stiffness is a torsional spring, expressed in terms of a torque at a given angular displacement. Torsional damping is a torque proportional to angular velocity.

Although a rubber coupler may have some damping coefficient due to hysteresis, it is not large.

I can see it by making changes and observing the results in reduction at the prop shaft but I don't see how we can quantify things further back on the system the first time around.
Click to expand...

The angular displacements (the shaft twists) are proportional. If you reduce the propshaft vibratory torque to half of its previous value, every other shaft torque in the system will also be reduced by half.
 
Last edited:
DanH said:
Sorry to be cryptic recently. Out of town, Ms. Patti's birthday.



Heeere little doggie......nice doggie......hold still, little doggie....;)

Sorry. I keed. Couldn't resist.

Seriously, stick to the known. The subject is hard enough without inventing your own measurement technology.

The strain gauge bridge is the least complex. The output is just a variable voltage. You need a radio or slip rings to get the signal off the rotating shaft, and a recording method which can be simple or elaborate. You get torque and frequency.

If you really, really want to look at the relative motions of the system elements you might consider magnetic or optical pickups. They are stationary installations so you don't need telemetry. Put a reluctor wheel and pickup behind the prop, another pickup at the big gear, a third near the flywheel's ring gear, and a rotary encoder or another reluctor/pickup on the accessory end of the crank. The rest is processing hardware and software; you're working with variations in rotational speed. Done right you get shaft twist in degrees and frequency, analogous to the calculated mode shape.

Here's the key: both methods have a huge industry knowledge base.
Click to expand...

Haha. Got it.:)

I checked the links you posted on telemetry gear. This stuff seems crude by modern standards as far as size in concerned. Lots of stuff as small as your thumbnail these days... 20 foot range which is fine for our purposes here.

I think the PowerSport PSRU was evaluated with optical encoders several years back, something similar may be a better option here. Unless a 1/2 wide WB is available and workable here, a strain gauge won't be in the recipe. I'll search for some commercially available stuff but frankly if it is expensive, that won't be in the cards either. I am the only guy flying this exact combination of parts so there is no way to share costs with others like a group flying Egg packages and my results won't be very useful to anyone else. While I am curious and want to learn more to improve what I have, I am not several thousand dollars curious.

We already have a Hall Effect sensor on the flywheel and process that information for ignition timing, I could easily add another to the prop hub and possibly even the drive ring. We could split this up in very fine increments with the micro. This could be easier to do than the accelerometers and the software is similar to what we do with EMS ignition control already. Might be able to log the data right into the ECU actually.

Right now, I'll concentrate on the inertia tests and getting the radiator installation and scoop done. I keep researching the measurement methods and hopefully decide which way to go, mount hardware and then get the engine back in.
 
Before I started reading this thread I thought I was a reasonably intelligent redneck. I now have serious doubts I am smart enough to keep ice in my cooler. :D
 
Inertia Tests

I did the bifilar suspension tests on the flywheel tonight, first with the flywheel as run for the last 9 years and then with the 16 steel weights added. I was at the shop 12 hours today and did not feel like looking at the equations and crunching the numbers to see if things made sense.

Stock flywheel 7 lbs. 14 oz, 57 cycles over 120 seconds

Modified flywheel 16 lbs. 1 oz, 57 cycles over 120 seconds

Wires are 12 inches apart, 6 feet long.

Will post some pix tomorrow and try to do the prop.
 
chaskuss said:
Not sure if any of the links below will help your education Ross. They may help those among us who are less learned, but interested.

http://www.epi-eng.com/piston_engine_technology/torsional_excitation_from_piston_engines.htm

http://www.epi-eng.com/propeller_technology/propeller_vibration_issues.htm

The link below shows that this is not Dan H's first rodeo :)

http://www.eaa.org/experimenter/articles/2009-10_beltdrive.asp

Charlie
Click to expand...

Yes, I've read almost everything on EPI's site and most of it makes good sense. I highly recommend it- there is some great stuff on there.

Dan, thanks for adding the last two clarifications. I am slowly getting it...
 
rv6ejguy said:
Haha. Got it.:)

I checked the links you posted on telemetry gear. This stuff seems crude by modern standards as far as size in concerned. Lots of stuff as small as your thumbnail these days... 20 foot range which is fine for our purposes here.

I think the PowerSport PSRU was evaluated with optical encoders several years back, something similar may be a better option here. Unless a 1/2 wide WB is available and workable here, a strain gauge won't be in the recipe. I'll search for some commercially available stuff but frankly if it is expensive, that won't be in the cards either. I am the only guy flying this exact combination of parts so there is no way to share costs with others like a group flying Egg packages and my results won't be very useful to anyone else. While I am curious and want to learn more to improve what I have, I am not several thousand dollars curious.

We already have a Hall Effect sensor on the flywheel and process that information for ignition timing, I could easily add another to the prop hub and possibly even the drive ring. We could split this up in very fine increments with the micro. This could be easier to do than the accelerometers and the software is similar to what we do with EMS ignition control already. Might be able to log the data right into the ECU actually.

Right now, I'll concentrate on the inertia tests and getting the radiator installation and scoop done. I keep researching the measurement methods and hopefully decide which way to go, mount hardware and then get the engine back in.
Click to expand...

Hardware vise this is probably more practical, but software vise this is an order of magnitude more difficult to get right. This is by nature a digital measurement and your resolution needs to be 10x your resonance frequency to assure good accuracy. In your case that means about 2000 Hz. Of course you have to log exactly simultaneous.

you can assure your rpm readings are done with enough resolution (1000+ Hz) by fine grading your pick ups, or you can use just a few (in theory you can get by with only one at each disk, but your ressonance frequency divided by rpm in Hz will be a better number). For the latter approach the time delay between the two disks is what you measure by cross correlation or similar. Your resolution will need to be the same. there be a problem with calibration.

I got all the software in the world to do analysis of this kind (and lots of other stuff) If you want I can analyse your data and see if I can shake out some frequencies and amplitudes.

I have to add that by measuring flywheel and prop, the f3 mode will be lost, you will not be able to pick it up.
 
Last edited:
rv6ejguy said:
I did the bifilar suspension tests on the flywheel tonight
Click to expand...

(Pssst....Ross.....57 seconds for both?)

Everyone, this is what Ross is doing...an accurate (and homebuilder friendly) method to establish the inertia of a component.

Materials: A roll of .032” safety wire, two screw-eyes, about 14 inches of 0.50” D x .032 aluminum tube, and a wristwatch with second hand.

Clear a spot in your shop under a ceiling joist. Mark two points on the underside of the joist exactly 12 inches apart. Install the screw eyes at these two points.

Hang two lengths of safety wire from the screw eyes. Each should be 76-80 inches long.

Mark and drill two .0625 holes in the aluminum tube, exactly 12 inches apart and 6 inches from the center.

Insert the ends of the safety wire through the holes in the ends of the aluminum tube, and fasten so that the length of the wires is exactly 72 inches, as measured from the top surface of the tube to the inside radius of the screw eyes.

You’ve now completed a fixture for determining “period of oscillation”. See the illustration. Using any handy lightweight hardware as required, fasten your subject mass (a crankshaft, flywheel, sprocket, propeller, whatever) to the underside of the aluminum tube. It should be concentric with the center point of the tube.

Get ready with the stopwatch. Slightly “wind” the hanging assembly a few degrees and let go. Note the time and start counting the rotating oscillations (rotate, stop, and return back to the start point is one oscillation). Record the total number of oscillations in 120 seconds. Now repeat the counting and timing process again for accuracy (or merely to convince yourself how dead accurate this method really is).

Remove the mass subject from the hanging fixture. Take it to a good set of scales, weigh it, and record weight. With time, count and weight in hand, go find a calculator.

The equation for mass moment of inertia is:

I = W [ r T / ( 2 pi ) ]^2 / L, where

I = Inertia in slugs/ft^2
W = weight in pounds
r = radius of the wires from center of rotation, in feet (0.5 for our fixture)
T = time in seconds for each oscillation (120 divided by the number of oscillations)
L = pendulum length in feet (6 for our fixture)

Notice everything is in feet, pounds and seconds so the answer is in slug-ft^2, which is what you'll need for most other non-metric torsional equations. For metric, 1 slug-ft^2 = 1.3554 kg m^2

Example: An actual 68" mahogany prop:
I = W [ r T / (2pi) ]^2 / L
I = 6.62 [ 0.5 x 5.5 / 6.28 ]^2 / 6
I = 0.2114 slug-ft^2

 
Last edited:
SvingenB said:
I got all the software in the world to do analysis of this kind (and lots of other stuff) If you want I can analyse your data and see if I can shake out some frequencies and amplitudes.
I have to add that by measuring flywheel and prop, the f3 mode will be lost, you will not be able to pick it up.
Click to expand...

Ross, that's a huge offer....Dr. Svingen is a professional. If you want to use the relative displacement method with fixed pickups, it would be well worth your investment to do it exactly as he may suggest.

I'm just sayin'.....
 
SvingenB, Thank you for your offer! I am not sure which way we will end up going here yet but if I get some decent data, I will forward it to you.:)

Dan,

Yes, 57 cycles for both tests. I expected a change in frequency...

Again, the coupler is a soft connecting stiffness, not a damper. A connecting stiffness is a torsional spring, expressed in terms of a torque at a given angular displacement. Torsional damping is a torque proportional to angular velocity.
Click to expand...

Dan, or anyone, do you know of any simple, dry damping mechanism that might be applied in this case? Rubber seems to be a double edged sword- soaks up the energy but re-releases 90+% of it back the other way between impulses.
 
The relative displacement method using time difference is a well known method of measuring torque on a shaft. It works more like a stroboscope (in theory, triggered by the rpm or the pickup itself). The problem with these TVs here is that the frequency that will be measured is higher than the frequency of rotation (rotation per second) and the rotation of the propeller is slower than the engine. It's the relative rate of change of pulse width within one rotation we are most interested in, not the actual pulse width, nor the number of pulses.

A simple timing/triggering mechanism or a counter will not work, some more elaborate cross correlation will have to be done. To make it simpler the number of pickups on the prop could be related to the number of pickups on the flywheel through the gear ratio. One rotation of the prop must give the same number of triggers on the prop and on the flywheel.

It's probably simpler to use lots (>60) "pickups" on one disk. Then you can directly record the pulse train (preferably at > 2000 Hz sampling frequency) and get the oscillation after much simpler analysis (no cross correlation or synchronization needed). This is probably the way to go if you are going to use this method.

In comparison, with a strain gage you get everything there and then :)
 
SvingenB said:
The relative displacement method using time difference is a well known method of measuring torque on a shaft. It works more like a stroboscope (in theory, triggered by the rpm or the pickup itself). The problem with these TVs here is that the frequency that will be measured is higher than the frequency of rotation (rotation per second) and the rotation of the propeller is slower than the engine. It's the relative rate of change of pulse width within one rotation we are most interested in, not the actual pulse width, nor the number of pulses.

A simple timing/triggering mechanism or a counter will not work, some more elaborate cross correlation will have to be done. To make it simpler the number of pickups on the prop could be related to the number of pickups on the flywheel through the gear ratio. One rotation of the prop must give the same number of triggers on the prop and on the flywheel.

It's probably simpler to use lots (>60) "pickups" on one disk. Then you can directly record the pulse train (preferably at > 2000 Hz sampling frequency) and get the oscillation after much simpler analysis (no cross correlation or synchronization needed). This is probably the way to go if you are going to use this method.

In comparison, with a strain gage you get everything there and then :)
Click to expand...

I can see it is not practical to get good data without a very high number of pulses per revolution to be able to catch the quick deflections using optical, magnetic or Hall Effect pickups. This would be the advantage of using accelerometers plus the signals are easy to work with and nothing will be missed assuming a high enough frequency response of the device plus no synching of signals because we don't need to perform counting of digital impulses or calculate time intervals. With a multi-trace recording scope we can superimpose the waveforms easily or manipulate in a PC.

I've looked at the Marcotte drive in detail and there just isn't a good place to install a WB on it.
 
rv6ejguy said:
Yes, 57 cycles for both tests. I expected a change in frequency...
Click to expand...

Cool. Not a fatigue misprint, so it means you added the additional mass at a location which did not change the radius of gyration (.707 x radius for a uniform disk, a little more for your flywheel because of the steel ring gear).

Remember, this is just another kind of pendulum. Changing the mass of a conventional pendulum bob doesn't change the period of oscillation. You do that by changing the length of the string. In the case of this flywheel, adding the mass further inboard (less radius, a shorter string) would make it oscillate faster, while a location further outboard would cause slower oscillation.

Dan, or anyone, do you know of any simple, dry damping mechanism that might be applied in this case? Rubber seems to be a double edged sword- soaks up the energy but re-releases 90+% of it back the other way between impulses.
Click to expand...

Let's not knock rubber just yet. An undamped rubber doughnut coupler straight from the catalog would do wonders for your system.

Dry friction damping? Sure. Nothing new. For example, Jeron Smith brought his system to OSH way back in 1998. It's a friction damper in parallel with a soft element, in the upper sprocket (http://www.raven-rotor.com/). Probably locked solid except at a resonant RPM.
 
Last edited:
Makes sense. My eye said all this weight near the outside would more than double the inertia but the steel ring gear contributes a lot being even further out. So it looks like 2.04 times the inertia if I did the calc right. Like you said before, science beats the eyeball in most cases.

So if I understand correctly, a well chosen Centaflex or Lovejoy coupler would spread out the relaxation response enough to coincide with the next torque impulse from the engine and not give so much torque reversal as the present thin walled rubber bushings?

The prop weighed 23 lbs. and did 20.5 cycles in 120 seconds BTW. So the prop has a ton of inertia as expected although it is turning 2.2 times slower than the flywheel.

Hmmm will also look into friction dampers too. I am on good terms with Jeron.
 
Rad Work

I am making slow progress on the ventral radiator installation. Lots of thinking before making any parts. A couple new photos here: http://www.sdsefi.com/rv16.htm

Many thanks to Russell Sherwood who shared information about his successful EG33/ Glasair installation. Some might remember him from the SARL races where he cleaned up in his displacement class.

For those interested, here is the basics of what was found to work well in this speed range:

Rad face area- 1 square inch per rated hp
Rad volume- 2.5 cubic inches per hp
Rad tubes- .080 X 1 or 1.25
Fins per inch- 14 to 19
Core depth- 2.5- 3 inches
Inlet area to rad face area- .12-.16
Exit area- variable geometry very important for drag reduction in cruise. Cruise .6 of inlet area, ground/ climb 2X inlet area.
Max wall divergent angle- 7 degrees

I will be instrumenting the rad duct for flight testing to check pressure drop and velocity at the exit.
 
use flywheel teeth + hall sensor?

Can you mount a hall effect sensor near your ring gear and use the ring gear teeth to generate the pulses? The differential time method would work to get the differential displacement.

In turbomachinery, they use fiber-optic pickups in the fan case to look at the fan blade tips passing. With several probes, they can pick up a couple of blade bending and torsional modes. Google NSMS and Bryan Hayes at AEDC.

rv6ejguy said:
I can see it is not practical to get good data without a very high number of pulses per revolution to be able to catch the quick deflections using optical, magnetic or Hall Effect pickups. This would be the advantage of using accelerometers plus the signals are easy to work with and nothing will be missed assuming a high enough frequency response of the device plus no synching of signals because we don't need to perform counting of digital impulses or calculate time intervals. With a multi-trace recording scope we can superimpose the waveforms easily or manipulate in a PC.

I've looked at the Marcotte drive in detail and there just isn't a good place to install a WB on it.
Click to expand...
 
Hi Ross,

I'm doing a Renesis on a -7. I spent several months digging for cooling info, so I feel your pain.

That inlet to core area ratio looks pretty extreme, compared to 'conventional wisdom' for a perpendicular duct. I wonder if the Glasair guy gets away with it due to higher climb speeds. The K&W book doesn't even discuss any ratio beyond .3, IIRC. I do know that there are some wedge diffusers flying with more extreme ratios. I think that the 7 degree number comes from using a flat sided duct. Using that number, your diffuser would be pretty long, right? The K&W 'streamline' duct is a more extreme variation on an exponential horn (basically a trumpet mouth curve), and the final angle at the core face is almost 90 degrees. It makes a much shorter diffuser, & supposedly, it's as efficient as the 7 degree diffuser.

The rotary guys have tried almost every conceivable technique for cooling. If you want the names of some that are flying successfully, PM or email me & I can put you in touch with a few.

Charlie
[email protected]
(Sorry if we're re-crossing ground you've already covered.)
 
scsmith said:
Can you mount a hall effect sensor near your ring gear and use the ring gear teeth to generate the pulses? The differential time method would work to get the differential displacement.

In turbomachinery, they use fiber-optic pickups in the fan case to look at the fan blade tips passing. With several probes, they can pick up a couple of blade bending and torsional modes. Google NSMS and Bryan Hayes at AEDC.
Click to expand...

I've thought about doing this as it offers pretty high resolution and we already use similar software to calculate time intervals for ignition timing. There are some issues with sensor response time varying with rpm that would have to be dealt with. I will do some more research into this.
 
rv7charlie said:
Hi Ross,

I'm doing a Renesis on a -7. I spent several months digging for cooling info, so I feel your pain.

That inlet to core area ratio looks pretty extreme, compared to 'conventional wisdom' for a perpendicular duct. I wonder if the Glasair guy gets away with it due to higher climb speeds. The K&W book doesn't even discuss any ratio beyond .3, IIRC. I do know that there are some wedge diffusers flying with more extreme ratios. I think that the 7 degree number comes from using a flat sided duct. Using that number, your diffuser would be pretty long, right? The K&W 'streamline' duct is a more extreme variation on an exponential horn (basically a trumpet mouth curve), and the final angle at the core face is almost 90 degrees. It makes a much shorter diffuser, & supposedly, it's as efficient as the 7 degree diffuser.

The rotary guys have tried almost every conceivable technique for cooling. If you want the names of some that are flying successfully, PM or email me & I can put you in touch with a few.

Charlie
[email protected]
(Sorry if we're re-crossing ground you've already covered.)
Click to expand...

I have not seen any rotary RV with a ventral radiator setup, just a lot of what I would term "massive compromise" designs. I see mostly really big rads and the aim seems to be just to cool it with little or no attention to minimizing cruise drag. A poor duct will mean poor pressure recovery, separation and unfavorable inlet/ exit to face area ratios. The often used P.L. ducts have to be the worst with flow turning through 180 degrees for massive momentum loss. Any liquid cooled aircraft not using a variable geometry rad exit flap is surely suffering a lot of cooling drag in cruise flight. The WW2 engineers knew this and every design had them.

We have several flying examples using similar design parameters and they cool very well, in even the hottest temperatures. In fact, Russell will be reducing the inlet area on his during his rebuild to reduce drag further, down to around .12 inlet to face area. While theory is great, there is no substitute for successful working examples and as I mentioned, Russell's plane is also a multiple race winner. To put this in perspective to those who believe that liquid cooling causes more drag than air cooling, Russell's design uses about half the inlet area per hp as an RV10- despite half the Delta T.

One of the main keys is a very long duct which is easy to do with a clean sheet design ventral setup. Mine will be about 48 inches long in total.

I've done extensive testing of various core types on my flow bench to test pressure drop (drag) as well as built a test rig to quantify core efficiency (temperature drop). There is a wide variation in core efficiencies (over 100%) and pressure drops (also over 100%). Many people are using very poor radiators to begin with. A poor rad means you need more area and volume and it creates more drag. By quantifying the temperature drop vs. pressure drop per unit area and unit volume we can select the best rad design to start with.

reg4_zpsd8ddf78b.jpg

Here is Reg Clarke's Subaru turbo powered Q2 showing a beautiful low drag ventral rad setup. Well proven with over 800 hours on it.

suberad8_zps545fe6d9.jpg

Here is another turbo Sube powered Q2 showing the very small inlet (about 22 square inches).

sherwood1_zpsde1ff75c.jpg

Here is an older photo of Russell's rad setup. Notice the 2 vertical guide vanes. There is also a horizontal one near the rad face.

We shall see how the RV works! :)
 
Last edited:
digidocs said:
Here's some great work on radiator ductwork design that I found today:
http://contrails.free.fr/tunnel_en.php#caracterisation

That site also has some near teardowns of Jabiru and Rotax engines and gearboxes.

Take care,
David
Click to expand...

This is a pretty interesting site. Russell and me have done some similar experiments. I see many people designing poorly shaped ducts and using no guide vanes to turn the air smoothly and wet the whole rad face. Both Russell and me found this highly necessary. It is a very simple solution and especially important when you have short ducts or larger divergent angles inside.
 
Last edited:
Ross,

I was just looking over your progress on your web site and want to thank you for sharing the information as you go along. I may actually be tempted to copy what you are doing once I get my turbo setup up and running again. I know what I am doing is all wrong as far as efficient airflows are concerned, but I am trying to make the best of having all up front and keeping the basic shape of the cowling somewhat close to original.

I am very curious about how you will direct the exhaust flow to avoid that heat from entering your radiator area. Perhaps the opening for air entry into the radiator will be low enough to avoid the exhaust?

Also, once you have your balsa wood shape completed, and you lay up the fiberglass on it, do you leave the balsa in place, or remove it somehow? I guess it's weight is not an issue and if you seal both sides it can become part of the structure as in foam and fiberglass construction?

I am looking forward to seeing more photos, and monitoring your progress. Thanks again for sharing..

One of these days perhaps we will finally get some fuel with proper octane ratings without the lead, or perhaps we will learn about some different materials we need to be using in key areas to avoid the damages it seems to cause in our Subes...

Randall Crothers
 
Q2??

When did Reg switch to a Q2??

Last I knew of he was in a Dragonfly------direct drive turbo Subie.

reg4_zpsd8ddf78b.jpg
 
Ross,

Repeating what I stated in post #4... I am really interested in how this cooling system works out.

I'll be installing an H6 Subie in my RV7. (No turbo). I realize your setup hasn't been flight tested yet, but as far as construction or design, would you do anything different if you were still building the airframe rather than adapting it to an existing aircraft?


Thanks,
 
Randy said:
Ross,

I was just looking over your progress on your web site and want to thank you for sharing the information as you go along. I may actually be tempted to copy what you are doing once I get my turbo setup up and running again. I know what I am doing is all wrong as far as efficient airflows are concerned, but I am trying to make the best of having all up front and keeping the basic shape of the cowling somewhat close to original.

I am very curious about how you will direct the exhaust flow to avoid that heat from entering your radiator area. Perhaps the opening for air entry into the radiator will be low enough to avoid the exhaust?

Also, once you have your balsa wood shape completed, and you lay up the fiberglass on it, do you leave the balsa in place, or remove it somehow? I guess it's weight is not an issue and if you seal both sides it can become part of the structure as in foam and fiberglass construction?

I am looking forward to seeing more photos, and monitoring your progress. Thanks again for sharing..

One of these days perhaps we will finally get some fuel with proper octane ratings without the lead, or perhaps we will learn about some different materials we need to be using in key areas to avoid the damages it seems to cause in our Subes...

Randall Crothers
Click to expand...

Exhaust exits at 45 degrees right now. The inlet to the ventral scoop is below the stock cowling exit and there will be a metal splitter plate attached to the top of the scoop inlet just like on the RV10 although the shape is a bit different. This prevents the hot exit air from entering the rad scoop inlet.

The balsa will stay in place with glass on both sides to form a lightweight and stiff sandwich. I am no guru with foam shaping and molds so I am doing it the old fashioned and painful way. This is very time consuming.

I am confident on 91 octane pump gas with my slightly lowered CR and some remapping of ignition timing at high MAP, I can safely run 38 inches for takeoff and 35 inches for climb like the old days on 100LL. Will still have the full advance of 32 degrees for cruise below 30 inches.

I almost forgot how much fun/work :rolleyes:;):( it was designing and building new stuff like this. One design decision affects 4 other things down the road and you need to have a solution for each one... This is like a 400 hour project... One thing is for sure, there is a good reason why not many people have ever put a belly rad on an RV6A before! Anyway, I look forward to instrumenting and testing for signs of the Meredith Effect when this is all done.:) See what happens to the cruise speeds and cooling.

Still to design and build:

1. Diffuser for the rad inlet inside the scoop

2. Mount the cabin heat exchanger/ 2nd rad above the nose gear mount and behind the oil pan

3. Rad exit door and control mechanism

4. Reposition and replumb the intercooler to the left cheek inlet

5. Hook up all the hoses and wires to the engine

6. Paint the scoop lime green

7. Paint and install all the cover plates where there used to be an air scoop

8. Install the gearbox and prop

9. Inspect/ clean, service the wheels, tires, bearing and brakes on the main gear.

10. Reweigh and calculate the new C of G. I hope to lose 35 lbs. in total.
 
Last edited:
Mike, I am really not sure of the differences between a Q2 and Dragonfly...:confused: I am sure someone here knows.
 
jcaplins said:
Ross,

Repeating what I stated in post #4... I am really interested in how this cooling system works out.

I'll be installing an H6 Subie in my RV7. (No turbo). I realize your setup hasn't been flight tested yet, but as far as construction or design, would you do anything different if you were still building the airframe rather than adapting it to an existing aircraft?


Thanks,
Click to expand...

I am hoping this setup works as well as Reg's or Russell's. I admit I am a bit chicken and did not go to the low end of the inlet to face area ratio that Russell has. Ground cooling is important when you get a long hold on a hot day. I figure with the exit door, I can trim most of the drag away in cruise that the larger inlet may cause. I don't want to build the scoop over again!

With the RV6A, options are really limited in doing anything a lot different that what you see here. On the RV10 with the flat floor and central tunnel, I ran the coolant pipes internally and overall it was much easier than the -6. I cheated and used a Pro-Stock scoop to house the rad with a lot of internal mods. You can see those parts on my RV10 page.

Is all this worth it? I frankly don't know. The Inkster/Getson RV7 turbo Subie had cheek mounted aftermarket rads and was blinding fast like that. Certainly it is way less work to go that way than what I am doing. I have a curious mind though and love to experiment. Very few others have documented before and after scoop characteristics other than Cozy builder Keith Spruerer: http://www.cozybuilders.org/Oshkosh_Presentations/2008_Columbia_Keith-Spreuer_P-51_Scoop.pdf. Keith shows why NACA ducts don't mix well feeding radiators and this mirrors my flight testing and ground testing with them.

I went this path because almost every aircraft which has switched to a proper ventral scoop has had no cooling issues and, well, it is cool looking too. I want to learn something from all this.

Of course if we were starting with a clean sheet design, we could do a much better low drag radiator setup, designing the structure around the liquid cooled engine and rad setup. The GP-5 running the Chevy V8 at Reno has a nice integrated setup. http://www.eaa.org/news/2010/2010-09-15_reno.asp

RVs are just plain hard to fit good rad setup too for a bunch of structural and layout reasons.
 
Last edited:
It looks like he sold the DF, and now has a Q2.

He used to fly the DF to the annual gathering at Eloy Az. Used to have big gills for hot air exhaust in the cowl cheek area as I recall. And I seem to remember the radiator hung out in the breeze under the fuse, with no fairing around it. Been a few years.......

The Q2 in your photo is wearing an N number, not a Canadian registration.
 
His brother has a composite repair business across the border in Montana. That might explain the N number. I believe he did most of the work on this plane.
 
rv6ejguy said:
I am hoping this setup works as well as Reg's or Russell's. I admit I am a bit chicken and did not go to the low end of the inlet to face area ratio that Russell has. Ground cooling is important when you get a long hold on a hot day. I figure with the exit door, I can trim most of the drag away in cruise that the larger inlet may cause. I don't want to build the scoop over again!

With the RV6A, options are really limited in doing anything a lot different that what you see here. On the RV10 with the flat floor and central tunnel, I ran the coolant pipes internally and overall it was much easier than the -6. I cheated and used a Pro-Stock scoop to house the rad with a lot of internal mods. You can see those parts on my RV10 page.

Is all this worth it? I frankly don't know. The Inkster/Getson RV7 turbo Subie had cheek mounted aftermarket rads and was blinding fast like that. Certainly it is way less work to go that way than what I am doing. I have a curious mind though and love to experiment. Very few others have documented before and after scoop characteristics other than Cozy builder Keith Spruerer: http://www.cozybuilders.org/Oshkosh_Presentations/2008_Columbia_Keith-Spreuer_P-51_Scoop.pdf. Keith shows why NACA ducts don't mix well feeding radiators and this mirrors my flight testing and ground testing with them.

I went this path because almost every aircraft which has switched to a proper ventral scoop has had no cooling issues and, well, it is cool looking too. I want to learn something from all this.

Of course if we were starting with a clean sheet design, we could do a much better low drag radiator setup, designing the structure around the liquid cooled engine and rad setup. The GP-5 running the Chevy V8 at Reno has a nice integrated setup. http://www.eaa.org/news/2010/2010-09-15_reno.asp

RVs are just plain hard to fit good rad setup too for a bunch of structural and layout reasons.
Click to expand...



Thanks for the info. I am still considering other possibilities as I build. Such as running the cooling lines inside, or even having the radiator inside, behind the baggage wall, and many others.

another idea is having the radiators in the cowling and exhaust it like the Pipistrel Pantherea:

7823
 
jcaplins said:
Thanks for the info. I am still considering other possibilities as I build. Such as running the cooling lines inside, or even having the radiator inside, behind the baggage wall, and many others.

another idea is having the radiators in the cowling and exhaust it like the Pipistrel Pantherea:

7823
Click to expand...

On all the RVs, submerging the rad aft of the baggage bay will involve cutting some big holes in the floor. To gain back the structural strength is going to involve quite a bit of new structure, ditto wherever you cut the holes for the exits. I looked into that extensively on both the 6A and -10. I am not a structural engineer so I didn't plus you have quite a problem getting the exit air aligned with the freestream. If you don't exit the air parallel to the freestream you will get plume drag and separation aft of the exit most likely.

You can't run the coolant lines inside on a -6 because the spar is solid top to bottom. On a -7 maybe but you are cutting holes in the center section structure so you'll need to add some doublers, insulate the lines because they are hot.

I spent many hours contemplating various layouts including wing rads. Every new placement involved multiple problems with piping coolant, structural changes, weight or unfavorable inlet or exit geometry.

Anything under the cowling on an RV will be quite a compromise as you just don't have the space/ length for properly shaped ducts. That being said, I am very interested to see what other people come up with! :)

I pulled off the scoop today and prayed it would not change shape much and it didn't! It only weighed 1 pound 11 oz. I will post a picture of it tomorrow. Turned out pretty cool. :)

Today I shaped and glued in the nose fairing blocks. Next step is to place some pour foam in the corners so I can shape a nice radius on the scoop before starting to glass the inside and outside, then on to the door, and shaping the diffuser which will not be fun. I want it to meet the core directly, not the tanks which just cause drag and turbulence. Most people get a pleasing outside shape but the inside is a haphazard mess. The whole interior needs to be treated like an internal aerodynamic structure.
 
Post some pictures of your scoop when you can, Ross...can't wait to see them. Are you planning to fab a cover for the rad supply and return lines to smooth the aerodynamics around them?
 
Michael White said:
Post some pictures of your scoop when you can, Ross...can't wait to see them. Are you planning to fab a cover for the rad supply and return lines to smooth the aerodynamics around them?
Click to expand...

You can see the whole tale here: http://www.sdsefi.com/rv16.htm

No plans to fair in the coolant pipes as the air is so messed up in this area (I know from video shot from a chase plane) that it would be a waste of time. Having the pipes mounted externally bugged me but it is the only real solution on a 6A. It works fine on several other flying examples and it is a short run of only a few inches on this installation before entering the the scoop.
 
Finally...

I have finally finished all the work and mods to my 6A, just have to re-weigh it today hopefully. 5 months and close to to 500 hours. Engine has been test run and soon we'll get to see how the mods worked out, hopefully the new rad setup cools better with less drag, we have lost some weight, oil and intercooling systems are improved and the airframe will be easier to work on and inspect in future. See the process here: http://www.sdsefi.com/rv16.htm

I plan to instrument the radiator system to see what pressures, temps and velocities we have. Is the Meridith Effect myth or reality at these speeds and coolant temps?

I hope to eventually tuft test and video the radiator and belly in flight and compare to the tests with the previous setup.
 
Ross, just took a cruise through the latest photos, looking good.

If TLAR engineering is accurate, you have a winner for sure.
 
More flywheel inertia is an absolute good for torsional behavior, as you've now seen for yourself. It will, however, add one oddity in the case of your Marcotte gearbox, which rotates the prop in the same direction as engine rotation. The additional inertia will add to gyroscopic precession. I think you'll notice the difference, although it should not be a big deal with the 6A. My scaled JN-4C (or an original Camel) was more interesting because engine/prop inertia was much higher in proportion to airframe mass.

An ideal PSRU would rotate the prop opposite the engine so the prop and flywheel gyroscopes offset each other. In theory you could have less precession than a Lycoming.
 
You must log in or register to reply here.
Back
Top