COOL










water























A look at liquid-cooled engine installations Neal Willford & David Lednicer

PHOTOS BY JIM KOEPNICK

The radiator inlet for the Merlin-powered P-51D (above) incorporates a number of subtle improvements over the original design on the Allison-powered XP-51. The most noticeable is that the P-51D's inlet is farther away from the wing and the turbulent boundary layer that degrades the cooling airflow to the radiator.

t seems like every EAA AirVenture ture, we’ll look at how the liquid-cooled

i










Oshkosh harvests a crop of new installation has evolved over the years and aircraft engines under develop-

examine the details of the P-51 Mustang

ment. Whether diesel or gasoline, installation. two stroke or four stroke, auto con-

As with the previous articles, a new

version or a clean sheet design, many rely spreadsheet is available to download from























on liquid cooling. In “Cool It” (see EAA the EAA Sport Aviation page on the EAA

Sport Aviation, August 2003) we reviewed website at www.eaa.org that will help you

the design of air-cooled engine installa- estimate the size of the radiator or oil cool-

tions, and now it’s time to do the same for er as well as the inlet and exit areas needed

PHOTO ILLUSTRATION BY JIM KOEPNICK

liquid-cooled installations. As a bonus fea- for a given installation.

Sport Aviation

35

Cooling System Details A liquid-cooled engine offers the flexibility of locating coolers in different locations, and the usual payment for this flexibility is increased weight and complexity compared to an equivalent air-cooled engine. A radiator, coolant, hoses, and reservoir tank all add to the airplane’s weight and cost. A simple equation computes how much engine heat a liquid cooling system will absorb: Heat Removed = coolant flow rate x CP x temperature difference. Engine manufacturers usually specify how much heat the cooling system must remove, and it’s roughly half of the engine’s horsepower. The coolant flow rate depends on the size of the engine’s water pump, and the pump speed depends on engine rpm (higher rpm equals greater coolant flow). CP is the coolant’s heat capacity, measured by how much heat it absorbs for a 1-degree change in coolant temperature. With a CP of 1, water absorbs heat better than air, which has a CP of 0.24. The engine oil also absorbs some engine heat, but its CP is only about half that of water. Temperature difference is the spread between the temperature of the coolant and the air flowing






The maximum coolant temperature is determined by the coolant’s boiling temperature.

through the radiator. The greater the difference the more heat the system removes; or, for a given heat removal, the smaller radiator you’ll need. The maximum coolant temperature is determined by the coolant’s boiling temperature. A water/ethylene glycol mixture circulates in most coolant systems because it has a much higher boiling temperature than water. This more than offsets the mixture’s lower CP value. Pressurizing the coolant system is another way to increase the maximum coolant temperature, which many motorists have learned the hard way by opening the car radiator cap prematurely. The Radiator How much heat the radiator removes from the coolant is governed by the same equation. In this case, the coolant flow rate is the amount of air passing through the radiator, and for a particular air density it will be equal to some volume flow of air in cubic feet per second. The heat equation doesn’t care how it gets the required volume flow of air; it can be moving

Approximate Cooler Drag Coefficient

Figure 1 Cooler Drag Coefficient versus Cooler Airspeed Radiator Installation Oil Cooler Installation Maximum Possible Cooler Drag Coefficient

Cooler Airspeed/Airplane Airspeed

36

JANUARY 2004

slowly through a large radiator or quickly through a small radiator. Designers do care, however, because this flow affects drag. Using a representative radiator and oil cooler installation, Figure 1 shows how the drag coefficient varies with the airspeed at the radiator face. If the speed of the air going through a radiator increases from 10 to 20 percent of the airplane’s cruise speed, then the drag coefficient radiator installation would be 8 times greater. Because doubling the airspeed through the cooler means twice the airflow, designers could achieve the same heat removal with a smaller radiator, but the smaller radiator’s cooling drag would still be substantially higher than the larger radiator’s. The red line in Figure 1 shows the maximum possible drag coefficient for a given radiator airspeed. Where this line crosses the radiator and oil cooler drag curves marks the maximum possible airspeed that each can handle. At these speeds the available dynamic pressure created by the airplane’s speed is completely used up overcoming the pressure loss through the duct and cooler, so it is not possible to have the cooling flow go any faster than this limit. The unfortunate truth is that a low drag, liquid-cooled installation favors a large radiator. There are practical limits to this conclusion. Installing a large radiator in an existing cowl may be difficult. However, radiators are pretty tolerant to being installed at an angle to the oncoming flow at the end of a duct—a good way to reduce the size of a radiator installed perpendicular to the airplane’s flight path. Although low cooling drag favors a larger radiator, the second practical limit is that small airplanes favor low external drag. In other words, designers don’t want a radiator so

big that it requires a large external duct to house it. In striving for a good balance between low internal and low external drag, successful low drag installations in the past suggest that the radiator should be sized to properly cool with a radiator airflow at 10 percent of the airplane’s speed at cruise, and that it be no higher than 30 percent of the airplane’s speed in climb. As air flows past the fins of a cooler, it absorbs heat from the fluid circulating through it. Typically, air passing through a radiator or oil cooler will experience a temperature increase of 50 to 70 percent of the temperature difference in between the coolant and air temperature. The dimensions of the cooler— its height, width, depth, and solidity (the ratio of the free passage area to the frontal area)—are all factors in how much the air temperature increases as it passes through the cooler. A low solidity means that the cooler has more cooling fins, or possibly thicker fins. For a given cooler size, a lower solidity will result in a greater cooling air temperature increase, or for the same amount of heat removal, a smaller cooler. The downside of using a cooler with low solidity is that it results in a greater pressure drop across the cooler. A higher pressure drop means higher cooling drag, but in some cases this drag increase can be partially offset by using a smaller cooler. Typically oil coolers have a lower solidity than radiators. The likely reason is that in the past, engine and airframe manufacturers have preferred small coolers for packaging reasons. Internal Aerodynamics Since many of the aerodynamic details are similar to an air-cooled installation, we’ll focus on the differences. (For the big picture of cooling aerodynamics, please review “Cool It,” referenced earliSport Aviation

37

If you are looking for performance, we sell it! FAA certified Hydraulic or Electric Constant Speed Propellers for all high performance aircraft, like LANCAIR, GLASAIR, CIRRUS, EXPRESS, QUESTAIR, VAN’s, VELOCITY, etc. 2-, 3-, 4-, 5- and 6-blade designs available up to 2000 hp. National, European and World Aerobatic Champions win with our specially designed, light weight, low noise and high performance propellers.

FAA STC for Pitts S2A and S2B, Cessna 172, 182, T/P210, Piper PA 46-310P/350P, PA60-601, 602 and more. Headquarters:

Service Center:

MT-Propeller Entwicklung GmbH MT-Propeller USA, Inc. Flugplatz Straubing-Wallmühle 1180 Airport Terminal Drive D-94348 Atting, Germany De Land, FL 32724 Tel: 011-49-9429-94090, Fax: 011-49-9429-8432 Tel: (386) 736-7762, Fax: (386) 736-7696 e-mail: [email protected] e-mail: [email protected] Internet: www.mt-propeller.com

Fly straight

to final assembly! Jump-Start Options

Jump Start Wings: • Leading edge and lower skins installed. • All flight control brackets and pulleys installed. • Anti-corrosion primer applied to all interior aluminum surfaces prior to assembly. • Upper skins fitted, drilled, dimpled and ready to rivet. • Strut beam assemblies installed.

Ailerons and Flaps: • Completed (Included in Wing Jump Start)

ALL NEW WEBSITE! www.newglastar.com New GlaStar Ph: 360.435.8533 x232 Fax: 360.435.9525

Jump Start Fuselage: • Composite shells fitted, laminated together and fastened to cage. • External seams fitted, gel-coated and buffed to match. • Top deck installed. • All 5 bulkheads installed. • COM antenna installed. • Inside visible shell surfaces finished with Zolatone grey speckle interior paint.

Jump Start Tail: • Rudder: Complete, less trim tab and tip. • Stabilizer: Complete, less tips. • Elevator: Complete, less trim tab and tips.

er.) Figure 2 is a schematic of an ideal liquid cooling installation. Each part of the system plays a role in how well the system cools the engine and how high the system drag is. In a real installation, finding room for a properly shaped diffuser and nozzle is often a challenge; the resulting compromises are in lost cooling performance and higher cooling drag. With this in mind, let’s examine Figure 2 to see what designers can do to keep the losses and drag to a minimum. Often, the radiator inlet is located behind the propeller, either directly behind it, like a P-40, or farther back, like a P-51. Locating the inlet in the propeller slipstream can greatly aid engine cooling on the ground and in climb. During World War II, testing by NACA (National Advisory Committee for Aeronautics) showed that cooling inlets had the best overall perform-

Figure 2

Radiator Installation Diagram

NOZZLE

INLET

EXIT DIFFUSER

ance when they were sized so that the inlet airspeed ratio was between 40 and 75 percent of the airplane speed. The inlet speed ratio is one of those paradoxes where the higher percentage corresponds to low air-

RADIATOR

plane speeds and vice versa. Because the inlet airflow speed remains fairly constant, comparing it to a lower (climb) speed gives a higher ratio, and comparing it to a faster (cruise) speed results in a lower ratio. NACA testing showed that when

This shows the calculated streamlines in the cooling system ducting of the P-51. The lines are color coded to indicate the velocity at any given location. Hot colors are higher speed flow; cool colors are lower speed flow. The calculated separation locations in the inlet ducting are also indicated. 38

JANUARY 2004

the inlet speed ratio was below 40 percent, the airflow started separating on the outside of the inlet causing higher external drag. When the ratio got above 75 percent, separation from the inlet duct wall was the problem; this resulted in higher internal drag and a loss in radiator effectiveness. Figure 1 shows that for low internal drag, the cooling air must slow down before it goes through the radiator. Using a tapered duct called a diffuser does this effectively, and a well-designed diffuser minimizes pressure losses while slowing the air. Viewed in profile, the diffuser’s shape plays a major role in its ability to slow the incoming air without it separating from the diffuser wall. This airflow separation is undesirable because it results in pressure losses that lead to higher drag and reduced radiator performance. A straight diffuser can only have a maximum angle of 7 degrees between the walls before pressure losses start getting excessive. This is a pretty shallow angle and can result in a rather long diffuser. American and German researchers independently found that a curved wall, or streamlined, diffuser (Figure 2) can have a much higher duct expansion angle before separation occurs. The maximum allowable expansion angle depends on the radiator’s pressure drop characteristics: a low solidity radiator can have a higher expansion angle and vice versa. Note that the ideal streamline diffuser looks like a trumpet and not an inlet on a jet engine nacelle. The trumpet shape delays airflow separation for as long as possible, thereby minimizing pressure losses. Reference 2 provides information for determining the ideal streamline diffuser shape and is included in this month’s spreadsheet. The downside of the ideal shape is that it needs to be at 2 to 3 times as long as the radiator’s height or width (whichever is greater). Sport Aviation

39

with Engine RPM OP MP EGT CHT OAT OT Voltage and current Carburetor temperature Fuel pressure and flow Fuel remaining, endurance Clock, timer, engine hours Programmable alarms

Simplifies your scan and replaces the usual engine instruments. Performs 22 different functions. Fits in a standard stack, weighs just 11⁄2 lb, and is customizable to your particular airplane. $999 for the full kit. Also available fully assembled.

Download full information about all of our products at our web site. Phone 307-864-9300 • Fax 307-864-9264 P O B o x 6 8 3 • T h e r m o p o l i s , Wy o m i n g 8 2 4 4 3

hese are the first tools you need to buy T when you re-cover your airplane. Anyone who has used them will tell you they’re the next best thing to having one of our staff right beside you. The VHS tape and the DVD will give you the Big Picture, and the manual will walk you step by step through every part of the process. You’re never on your own when you’re using Poly-Fiber.

www. polyfiber. com e-mail: info@polyfiber . com

800-362-3490







Some liquid-cooled installations do not have the room to install the proper diffuser shape. Therefore, pressure losses can be severe unless the air is slowed down sufficiently before it enters the cowling. An installation like this will need to slow the air down before it enters the duct, requiring larger inlets (which may result in higher external drag). If the inlet is located farther (like on a P-51), it should be far enough away from the fuselage surface to avoid ingesting the boundary layer, a thin layer of air next to the skin created by the friction of the air on the surface. This friction causes the boundary layer to flow quite slowly. The speed increases with distance from the skin, until it moves at the local flight speed. Because the friction disrupts the flow, there are large losses in a boundary layer and it is therefore advantageous to not ingest boundary layer air into a cooling system. The thickness of the boundary layer grows thicker as you move farther back along the fuselage. This is why

zle is not as critical as the diffuser’s. Testing has shown that the nozzle should not start pinching down severely before a distance of at least half the radiator’s height or width (whichever is greater), or severe losses can occur (due to back pressuring the radiator). The nozzle’s exit area controls the airflow through the system and the system drag. Installations without a nozzle, where the cooling air passes over the engine or other parts in the cowl before exiting, will likely have more pressure losses and higher cooling drag. As you can see, designing a low drag installation can be a challenge—especially if there are space constraints. History is often a good teacher, so the following is a brief overview of some of the lessons learned by earlier developers of liquid-cooled installations.

A liquid-cooled engine offers the flexibility of locating coolers in different locations, and the

usual payment for this flexibility is increased weight and com-





plexity compared to an equivalent air-cooled engine.

External calculated streamlines around the P-51 cooling duct.

40

JANUARY 2004

the cooling duct on the P-51 is located away from the fuselage. If possible, to reaccelerate the cooling air before exiting the airplane, use a nozzle downstream of the radiator. Just how fast the air will be able to accelerate depends on the losses upstream of the exit, and the higher the speed of the exiting cooling air, the lower the cooling losses. The nozzle’s exit area controls how much air flows through the cooling system, so for a low drag installation, it is desirable to use a cowl flap to regulate the airflow. Fortunately, the shape of the noz-

Liquid History The Wright brothers used a liquidcooled engine in their 1903 Flyer.

Sport Aviation

41

PROPELLERS

The “radiator” was a crude heat transfer device simply hung out in the breeze. By World War I, most liquid-cooled engines had radiators immediately behind the propeller. Hugo Junkers, the father of the modern, all-metal aircraft, debuted the ducted radiator installation on his Junkers J2, which earned him a patent (Reichspatent #299799) on January 15, 1915, and first flew in 1916. With his ducted radiator design, Junkers developed the concept of slowing the incoming air and converting the dynamic pressure to static pressure in a controlled fashion. Strangely enough, Junkers didn’t use the cooling configuration on his later designs. This is probably due to inadequacies in his ability to design efficient ducting. Additionally, the ducted radiator on the J2 was mounted on the aircraft’s belly and suffered total pressure losses due to the sluggish boundary layer it ingested. Two breakthroughs finally enabled designers to create an efficient ducted submerged radiator installation. The first came in 1935, when the Royal Aircraft Establishment examined the aerodynamics and thermodynamics of the ducted cooling system. This knowledge allowed the design of cooling systems for such aircraft as the Supermarine Spitfire and Messerschmitt Bf 109. The cooling systems on these aircraft were still quite inefficient because they ingested boundary layer air and internal ducting suffered from large regions of separated flow, which further decreased the available total pressure. The second major breakthrough came during World War II with the development of the streamline diffuser, arrived at independently by NACA and in Germany by the AVA. The diffuser shape minimized the losses in the inlet ducting, allowing for the maximum static pressure recovery at the radiator face. Finally, the development of bound-

for Experimental Aircraft best in safety, performance, durability constant speed propellers for 150 to 600 HP

www.aerocomposites.com Phone: (860) 829-6809

Email: [email protected]













ary layer diverters increased the available total pressure coming into a ducting system.

A cooling system works like a ramjet: the air

Cooling the P-51 The P-51 Mustang is widely regarded as a milestone in aircraft liquid cooling design because of its efficient, very low drag system. The early Allison-powered Mustang had a reasonably good cooling system.

rises, heat is added to the air, and the air is

is slowed down so that the static pressure then accelerated before exiting through a nozzle.

Most importantly, the radiator was positioned in the aft fuselage, submerged within the airframe, and was fed by a long entry diffuser duct. This reduced the airframe’s wetted area, which helped reduce the external drag. The Merlin-powered P-51B appears to be the first aircraft designed with a streamline diffuser duct, though this came about by accident. On the Allison Mustang the duct’s upper wall was shaped like a streamline diffuser, probably to avoid structural components. The duct’s lower wall didn’t have the characteristics of the streamlined duct, and when they put the Merlin in the P-51B, the airplane suffered from duct rumble—airframe vibrations caused by separated flow in the inlet ducting striking the radiator/intercooler. NACA tried a number of different diffuser shapes to see if they would cure the Mustang’s duct rumble. Because the oil cooler was mounted in front of and below the radiator/intercooler, these duct shapes had larger expansions near the end of their paths. While not optimal, analysis has shown that the duct that resulted from this study was quite good. The P-51B was also one of the first aircraft to use a boundary layer diverter, which ensures that only air with higher total pressure entered the cooling system. In its duct-rumble testing, NACA moved the inlet away from the Mustang’s belly by what was estimated to be the boundary layer thickness in this area. As a result, the duct rumble went away as the separation on the 42

JANUARY 2004

top wall of the duct now occurred far later. Germany made the same breakthrough. The Bf 109F introduced a boundary layer diverter in its radiator ducting, which kept the cooling system from ingesting the boundary layer on the bottom of the wing. The British never made this breakthrough during the war. They studied ways of improving the Spitfire’s cooling, but never seemed to fully understand that ingesting air from the boundary layer reduces cooling system effectiveness. Analysis has shown that because of this, approximately one-third to one-half of the radiator was immersed in air with reduced total pressure. Whether the P-51 cooling system produces thrust is an endless debate. A cooling system works like a ramjet: the air is slowed down so that the static pressure rises, heat is added to the air, and the air is then accelerated before exiting through a nozzle. The difference is that a cooling system adds much less heat and, because the flight speed is usually lower, the static pressure at the time of heat addition is lower. As a result, a cooling system is a very, very inefficient ramjet. Examination shows that the Mustang’s cooling system can produce tiny amounts of thrust at high power settings and high altitudes. P-51s racing at Reno often have redesigned cooling systems better suited to their circular mission. If there’s another breakthrough in cooling technology, it’s the water spray bar developed for Reno’s unlimited racers. Spraying water over the radiators (and oil coolers) increases their heat transfer. (Because airplanes for everyday use fly longer than the average race, this isn’t a practical solution for the average liquid-cooled aircraft.) Some years ago, Strega, a much modified P-51 racer, received new cooling system ducting. It did away with the boundary layer diverter and incorporated a non-optimal Sport Aviation

43

diffuser shape. Analysis predicted that it wouldn’t work, but it did. After some probing questions, it was learned that it worked because the team doubled the spray-bar water rate to compensate for the new duct’s inefficiencies. In the case of an unlimited racer, this is probably a reasonable trade-off.

References Fluid Dynamic Drag, Hoerner, Sighard, published by author, 1951. Aerodynamics of Propulsion, Kuchemann, Dietrich and Weber, Johanna, McGraw-Hill, 1953.




more at www.eaa.org

Undeniably the Best! Ultimate Bar Graph Engine Analyzer (UBG-16)

Still: The sharpest, most stable display The easiest to use and install The most powerful features STC’d & PMA’d

EI’s New Fast Response EGT Probes EI’s New Fast Response Probe Probe Tips

(Actual Size)

EI’s Standard Probe The FASTEST, most reliable ungrounded EGT Probes available!

The best engine analyzer starts with an ungrounded EGT probe! Ask us why.

Electronics International Inc. Phone: (541) 318-6060 Fax: (541) 318-7575 [email protected] For more details visit: www.Buy-EI.com