The Quest for Speed

In the relentless quest for speed, I have spent many hours, and many dollars, on my airplane. I knew when I built her I would be racing her, and lucky for me, I was right. Unfortunately, since I first started the kit, several races across the country have been shut down by the insurance companies. At this writing, only two remain that I compete in, the Colorado Pilot Assn. Denver to Jackpot, NV Air Classic, and the Airventure Cup race from Dayton, OH to Oshkosh. Both are fun cross country races, but not as exciting as the Sun 100 race that was run at Sun 'n' Fun every year. No doubt Reno is a real rush, but way beyond my pocketbook.

Testing

The first step in developing speed mods is proper testing. By proper, I mean the tests have to be run in a way that provides repeatable results over a wide variety of conditions, since our quest for speed is a year-round pursuit.

To achieve this, we must eliminate all variables, so that the tests are run under exactly the same conditions each time. The same weight and c.g., identical power settings, running at the same density altitude, and consistent piloting are essential.

To eliminate winds aloft from our variables, I use the spreadsheet at the National Test Pilot School website. By flying 4 legs at approximately 90 degree different headings, and recording GPS groundspeed and ground track for each leg, the spreadsheet will calculate the average ground speed for 4 sets of three legs, and also give a standard deviation for the data set. The standard deviation is a measure of quality of your data, and should be below 0.5 for a good run. I will usually fly 3 or 4 sets (I call them "squares") during each test flight. I will fly 3 or 4 different days for each mod, and take an average of all (so 12 to 15 squares worth) to find a resultant speed for a particular mod.

I always set the power at 2500 rpm, wide open throttle, 75 degrees rich of peak. I always fill the tanks before the flight, and fly by myself. I use the Tru-Track autopilot to fly headings and hold altitude (this provides the added benefit of giving me time to watch for other traffic. If you have to hand fly the airplane, I strongly suggest a co-pilot to record data and help watch).. I fly at 9000' density altitude, using the E6B function in my GPS to calculate the altitude to fly based on barometric pressure and outside air temperature. I try to fly in the middle of the barometer, that is, somewhere near 29.92 in Hg. If the barometer is very high (often 30.50 or so in the summer in this area) or very low, I get skewed results.

*note  I live in SW Colorado. My home airport elevation is 7650 msl. so I have to choose a density altitude that is higher than optimum. Even so, my mid-summer testing can only be done during cold spells, or 9000' d.a. will result in a below ground level flight.

As you can read on other parts of this website, I have been doing this testing since I first flew the airplane. It has helped me state with confidence, for instance, that the MT prop was 4 kts slower than the old Hartzell, and that the new Hartzell was two knots faster yet. I have also used the test results to test various mods that I have tried. Some of these were well known to provide a speed increase, and some resulted from my own crackpot ideas.

Mods That Work

Propeller - Among the four props I have had on the airplane, the new Hartzell blended airfoil is the fastest. See the prop test page.

Horsepower - Knowing I would be racing, I ordered an O-360 parallel valve engine from Lycon with 10:1 pistons, port and polish, ceramic coatings, and fuel injection. I built a custom ram air inlet system. I used a tuned 4 into 1 exhaust from Sky Dynamics, and Lightspeed Plasma III ignition.

Drag reduction - This is what this page is really about. From the beginning, I did as much as I could to reduce parasitic drag, and to add lightness. I used Grove streamlined gear legs. Massey "hot tips" wing tips with internal antennas. Sam James cowl and wheel pants. A careful installation of the Sam James plenum. I even left the outside handle off the canopy. I have a fairing on the tailwheel. Many of these I did not test separately, since it was built that way from the beginning.

Last spring, I decided I didn't like the shape of the lower cowl. From the side, it looked like the bottom of a wing, with the flaps partially down. Now it looks like the bottom of a wing with the flaps up. Also, my theory is that the cooling air that is going down through the cylinder fins has a lot of motion downward. This causes eddies and turbulence when it exits the cowl and impacts the airstream. So, I extended the bottom of the cowl back about 2" to force this air into a rearward motion, blending into the airstream.

            

Testing indicates I gained about 3 kts with this mod. It was a lot of work, so it was very satisfying to actually gain some speed from it.

 

Engine Cooling Drag

It has been long recognized that engine cooling airflow can represent 30% or more of the total drag of the airplane. There is a landmark paper that was done on the subject by a group at Mississippi State in 1981.

It is available in PDF form from the NASA server

http://ntrs.nasa.gov/index.cgi?method=advanced

In the title block, paste in the following title without the quotes.  Hit 'begin search' and
a link to the pdf download will appear at the bottom of the page.

"An experimental investigation of the aerodynamics and cooling of a horizontally-opposed air-cooled aircraft engine installation"
 

 

From the initial build I have had a plenum and round inlets as suggested by the above study. Recently (spring '06) I have been working on cooling airflow. I have decided that anywhere air flows, it must be guided. I started with the air coming out of the oil cooler. Rather than let the air just blast out of the cooler and against the firewall, I made a scoop to guide it to the exit.

            

Then I added some baffles to the lower cowl, to guide the air exiting from the bottom of the cylinders. On the right side you can see the flange where the scat tube from the oil cooler attaches.

    

I have not been able to test this mod due to the warm summer temps, so I will have to continue the report later.

May 13, 2007

Testing over the winter did not show any increases in speed until I reduced the size of the inlets. Now it looks like I gained a knot or so, and also I have been able to leave the inlet reducers in during warm weather.

 

ECI Cold Air Intake and Sump

Over the winter I decided to remove my stock warm air sump with elbow adapter and replace it with the ECI cold air intake plenum and sump. The magnesium plenum and sump weighed 6 pounds less than the old setup. This caused my 4 into 1 exhaust system to no longer fit. After several phone calls and emails to Kevin Murray at Sky Dynamics, with never a call back, I gave up and bought a 4 pipe exhaust system from Larry Vetterman. This saved me about $900, and the service from Larry has been second to none. This allowed me to rework the bottom of the cowl for a little more streamlined cross section.

            

Flight testing shows a solid 3 kt. speed increase. This of course at the expense of a higher fuel flow, so it is not as popular with me as drag reduction.

While I was at it I decided to take the heat muff intake air off the outlet of the oil cooler...

    

And this did increase my cabin heat temps even in the winter to where the airplane is actually comfortable at 20 dF outside.

 

Exhaust pipes

I have tried a few configurations of the Vetterman 4 pipe exhaust. At first, I left them stock. Then I cut them off almost flush with the firewall. This seemed to gain little in speed, but it sure did make for a dirty belly. The exhaust soot covered the belly all the way back to the rudder in less than 5 hours. Now, I want to go fast, but this made me question my dedication.

I called Larry and had him send me 4 6" pieces of pipe with a 40 degree bend at a 6" radius, and with the straight end swaged out so it would slip on to the cut-off ends of my pipes. This cost all of $50. When I first flew it, the airplane vibrated terribly. I though the pipes had cracked from the weight, but inspection showed no damage. Another flight showed that while the edge of the glare shield was a blur from the vibration, the spinner was running smoothly. I finally decided the pipe extension was causing turbulence at the rear of the cowl. The previous coating of soot showed that I had a very smooth airflow there, and the downward wash of the exhaust was causing airframe buffeting. To add insult to injury, I had lost 3 knots airspeed.

I cut a couple of inches off the pipes, and changed the angle so they weren't pointed down so much, but still a little in an attempt to preserve my clean belly. Amazingly, theory proved true, and I was back to a smooth running airplane. I regained 2 of the 3 knots I had lost. I am willing to sacrifice the extra knot if I don't have to lie on my back and clean the belly every time I fly.

The picture on the left is with the longer pipes, and on the right the shorter ones. The devil is in the details!

           

 

Gap Seal Tape

Sealing gaps with tape is very popular among the racers. For several years I have been sealing seams on general principle. I concentrate on the lateral seams on the theory that they will disrupt airflow more than the longitudinal seams. So, I tape around the rear of the cowl, but not along the seam between the cowl halves. I tape the NACA vent under the wing, and the one on the fuse (the pilot's vent) if it is cool outside. I have never measured the effect of this taping. I do it just because all the fast guys seem to do it.

 

It is hard to find a good tape to use. I have experimented with vinyl and poly tape, but no clear winner has emerged. I don't want to use a real strong tape that will remove paint when I take it off, and I would like one that doesn't leave a residue. For this year's Rocket 100 I used vinyl tape, but because it rained the night before and the plane wasn't real dry, the tape didn't stick, and I flew the course with tape flapping away.

This summer I experimented with taping the gaps in the elevator, rudder, and ailerons where there are access holes for the attachment bolts.

          

I gained two knots from this. This is the most cost efficient speed mod I have ever made.

Later, in prep for the Rocket 100, I taped the whole gap of the elevators and rudders. I have been told by both Tracey Saylor and Dave Anders not to tape the ailerons. Both cited a stiffening of the controls to the point that the airplane was barely controllable.

I used Mylar tapes that are used by sailplane competitors, bought from Wings and Wheels.  http://www.wingsandwheels.com/

After hearing the stories, I started by taping half of the elevators, top side only. This caused no difference in flying, so I taped the second half, flew again, taped the bottoms, flew again, and I felt it flew the same as always. I followed the sequence with the rudder gaps.

I was not able to detect a speed gain from the additional tape, and so I won't be doing it again. When the airplane got a little dirty from the rain and sitting out, the seals were rubbing and scratching the paint.

 

 

An Engineer's Perspective on Cooling Drag

Recently (April, 2007) there was an excellent thread on the Lancair email list about cooling drag. I have obtained permission from the author, Fred Moreno, to post it here. Fred is a Californian Mechanical Engineer who is living in Australia, and flying a normally aspirated, non-pressurized Lancair IV. You will see that his calculation revolve around a 230 Kt cruise speed, but the numbers are scalable to RV speeds using the square of the indicated airspeeds.

“Has anyone done any work with cooling drag?”

Walter and the GAMI boys have done nice experiments and learned a lot, particularly about reversed flow in the cowl, leakage out from inlets and other neat stuff.  I have spent a lot of time over the last few years trying to learn more since my original training included a lot of heat transfer, internal fluid flow and such, and I think cooling drag is about the last place to reduce drag on a Lancair IV in cruise.    (Reno racing with high G turns creates different conditions and requires more mods, but I digress.)

I have corresponded off line with several of our members on this topic, and a lot of good work is being done, much of it covertly to gain some benefits in racing.

Cooling drag is at times a counter intuitive business, and like much in aviation, there is a lot of mis-information because it can be a bit complex. 

Observation: if you want to wring the most out of the airplane, it will take a lot of work.  I am into cowl rebuilds, cowl flaps, cooling plenums, discharge nozzles and such for more than 300 hours and I am not done.  I have had to make a lot of tooling and scratch build a lot of composite parts, all time consuming.   So this is a business for fanatics only.    Benefits?  On a Lancair IV, my current guess is plus 10-12 knots. 

Perspective: One an aspirated C210 or Bonanza class airplane at 170 knot cruise, cooling drag is reported at about 7%.  These planes have a flat plate drag area of about 4.5 square feet, more or less.  A Lancair IV has a flat plate drag area of about 2.1-2.2 square feet, about half.  I think the Legacy comes in at about 1.7-1.8 square feet.  Therefore, if you use the same method of cooling as for the spam cans, cooling drag will be approximately 14-15% of total drag. 

If you are turbocharged, pressurized, and at 25,000 feet, it is probably more like 20-25%, maybe more because of cool air and added flow from intercoolers.  And it is harder to minimize.

Rule of thumb for small changes: a 2% reduction in drag will give about a 1% increase in speed.  So if you can get that 14% of drag down to, say, 4% (probably not possible, but reach for the stars!) then the 10% drag reduction gives a 5% speed increase.  For the Legacy and aspirated LIV guys (like me) who are in the 200-240 knot range at maximum cruise, 8000 feet, you could get 10-12 knots.  So that brackets expectations.

Some other rules of thumb.

1)     Flow only the air the engine needs to stay cool.  That means no leakage (and there are leaks everywhere, big and small in a stock installations) and restricting the airflow to the necessary amount in cruise.  The easy and less effective way favored by most is to choke down the inlets.  The more effective but much more complicated way is to use cowl flaps to throttle the exits.  Also, this means don’t overcool the engine.  I will let Walter speak, here, but I think that the choke in the cylinders is sized assuming that the CHT for cruise is around 350-400F.  I think 380F is a good trade off number, but I may be wrong (since it happens all the time).  J

2)     No leakage means a top plenum, attention to leakage detail, and in particular extreme attention to detail eliminating leakage out front to behind the spinner, usually one of the largest offender areas.

3)     Hard core fanatics will do the following:

Fred Moreno

Then, the next day, in response to questions, Fred posted this follow-up report...

I have received questions both on the forum and off concerning my earlier comments about cooling drag.  They sent me scurrying back to my books and calculator. 

I spent some time generating a self-consistent set of numbers for cooling drag, pressures, temperatures, etc. with a uniform set of assumptions.  Some guesstimates are required, particularly concerning losses from friction in various locations, so I have listed everything assumed and calculated, and invite investigation and questions.  The results are summarized in the ugly sketch below.  It shows the flow conditions at various points along the cooling air pathway assuming you have the ultimate set up, no leaks, cowl flaps perfectly adjusted to get the right output area, etc.   

Air enters from the left, is partially compressed in front of the inlet by the slowing of the air flow, goes through the inlet, then through a short diffuser where it is slowed more, and then dumps into a large plenum over the engine where the flow is a fairly slow moving mess.  Then the air flows through cylinders and oil cooler (which can take a lot of flow) into the space below the engine (where the flow is again a mess, but a slow moving mess), and then accelerates through the outlets, the area of which is controlled by cowl flaps which are optimally adjusted.   I have assumed 9000 feet, 200 knots IAS, 230 knots TAS as an interesting speed for us aspirated engine guys to investigate.  I have used pressures in inches of water since the engine pressure drop charts use these units of measure.  The inlet velocity ratio is assumed to be 0.4 (velocity through cowl inlets divided by TAS) which seems to be roughly what the latest and fastest designs are using in the fastest production aircraft.

 

Comments:

1)     Flow is controlled at the outlet with cowl flaps, so we can get a lot of compression in front of the inlets where the compression process is frictionless.  With an inlet velocity of 0.4, we get about 86% of the total dynamic pressure in front of the inlet. 

2)     We are forced to accept a short diffuser behind the inlets which further slows the flow, but friction losses eat up an assumed 2 inches of water pressure.

3)     The velocity in the plenum is anybody’s guess as the flow is a turbulent mess, but the velocity is low, so losses are low.  But there are losses in the plenum, particularly with the 550 Continental engines because of the intake manifold and other clap trap above the cylinders.

4)     Note that the compression raises the temperature about 12F so the assumed temperature above the fins is about 40F.  We can use this when we go to the engine cooling charts to make an estimate of the air flow required.

5)     Engine cooling charts are published assuming full power and 475F CHT (red line) and cruise with 435F CHT.  The charts suggest that the engine pressure drop will be about 5 inches of water.  But we like 350F or thereabouts for our CHT and that requires more airflow for more cooling.  A real rough guess is about 20-25% more air flow, and this requires 8 inches of pressure drop across the engine which I have assumed.

6)     The flow below the engine is chaotic, but slow so losses are low.  It is also hot.  With 350F CHT, it is probably 100-120F higher than ambient, or perhaps 150F.  Because there is a lot of clap trap below the engine, there are more losses which I guesstimate at around 1.0 inch of water.

7)     As you approach the exit (“discharge nozzle”) the flow is accelerated, hopefully smoothly, efficiently, and directly aft to regain as much momentum as possible.  However, our exit nozzle is filled with exhaust pipes, pipe brackets, and other clap trap in the higher velocity area as well as friction losses along the walls.  My guess is we lose another 2 inches of water pressure.  This leaves about 12 inches of static pressure converted into dynamic pressure in the exiting flow.  With the warm, lower density air I get about 170 knots discharge velocity directly aft for our perfect cooling system.   The loss in velocity for the cooling flow is then 230 KTAS minus 170 KTAS outlet, or 60 knots. 

Cooling drag is equal to the momentum loss of the cooling air.  It is cooling air mass flow (pounds per second) times the velocity lost during passage through the cowl and engine.  If we had lower pressure losses, we would have more pressure at the exit to accelerate the flow aft faster. 

If we could get the exit velocity = free stream velocity (the true air speed), cooling drag would be zero. 

If exit flow was faster than the free stream velocity we would get thrust. 

If we are moving really fast with lots of ram pressure and low losses and we raised the temperature of the air stream up very high to lower the air density a lot and allow it to be accelerated aft really fast, we would have a ram jet.

But for our air cooled engine with lots of internal friction inside the cowl and engine and only modest temperature rises, we can not get thrust, even for our prefect case above.

Now for a cross check.  Let’s assume our IO-550 is pumping out about 225 HP (about 73%).  If you go to a Lycoming power chart and assume it is the same for the Continental (pretty good assumption), for 40F above the engine and 435F CHT the chart shows you need 3 lbs/sec. of cooling air and 5 inches of water pressure drop across the engine to get this flow. 

But we want 350F CHT, so the required air flow is (rough guesstimate) 3.7 lbs/sec. and then the required pressure drop to get this flow is about 8 inches of water, the figure I used above.   And if you calculate the cowl inlet area required to admit this flow rate at 40% of the free stream velocity (TAS), you get two 6 inch diameter inlets.

Moving right along, if we assume that the prop efficiency at 230 KTAS and 225 HP is 85%, then the thrust = drag = 272 pounds.  Now we compute the cooling drag which is the momentum loss which is the cooling air flow rate times the velocity loss of 60 knots times some constants to make the units work and we get 12 pounds for the cooling drag or about 4% for our perfect zero leakage cooling system. 

In my earlier message I noted that your run-of-the-mill spam can had cooling drag of about 6-7%.  Our hot rods have about half the drag as a typical Bonanza or C210, so cooling drag becomes maybe 14%.  If we work really hard maybe we can get it down to 4%.  You can now see what it takes.

Your turbo guys at 25,000 feet have a much more serious problem because there is little to work with.  Assuming 280 knots TAS, the IAS = 190 knots and you thus have about 10% less ram pressure available to start with, maybe 23 inches of water.  The air is colder, but it is also a lot less dense so if I really extrapolate off my power chart, and assume 262 HP (75% for a TSIO 550) and 350 CHT, the pressure drop required across the engine is 12-14 inches of water, call it 5 inches more than at low altitude.  So the high altitude means that you lose 3+5= 8 inches of water pressure compared to the low altitude case above as the flow approaches the exit.  Instead of 12 inches of water to accelerate the flow aft, you have about 4 inches, so the exit velocity is much lower requiring a larger exit area,  and the momentum loss much higher which means cooling drag is much higher. 

There is not much you can do about it because we already assumed a perfect, leak free cooling system.  However, if you run this configuration (big exit area) at lower altitude, you will have a lot of excess cooling drag you could shed – if you could close cowl flaps, shrink the exit area and thereby accelerate the flow aft to a higher velocity (lower momentum loss).

Closing comments: the above discussion and figures assume a perfect leak-free system with optimal inlet and exit areas.  Optimal exit area for cruise flight is too small for slow, high power climb on a hot day, so it requires adjustable exit area meaning cowl flaps.  And low leakage means a plenum, careful attention to plugging leaks, and also plugging leaks from inside the cowl to the outside. 

Bottom line: the best you can get over stock is probably 10-12 knots.  If you just minimize leakage with a plenum and closing off leaks, you might get 5 knots.  If you don’t want cowl flaps, you can close down the inlets (velocity ratio much higher than 0.4) to minimize flow rate and get another 2-3 knots, but you will have to climb at high IAS to stay cool.  If you want all you can get (10-12 knots) then you will have to do it all, and do it to perfection.

Questions and comments welcome.

Fred Moreno