HOW TO COOL YOUR WANKEL copyright 2000-2004 by Paul Lamar

Can an aircraft engine be improved by liquid cooling? Cooling an aircraft
engine with liquid under all ambient temperatures is not an easy thing to
do. Many aircraft designers and builders have failed. All aspects of the
cooling system design must be near perfect. There is no cooling but air
cooling. Ultimately all waste heat must be dissipated to the air surrounding
the aircraft.


Back Ground on Air Cooled Aircraft Piston Engines,

The current air cooled  aircraft piston engine's power to weight ratio will
be very very difficult to improve upon based on its hundred year detail
development. No helpful material breakthroughs have occurred in the last 70
years.  However strict attention must be paid to the cylinder head and
exhaust gas temperature of an aircraft engine or failure will occur. The
piston engine is finicky and the fuel air mixture must be set just right at all
power levels or the exhaust valves and cylinder heads will fail due to over
heating.

Temperature gradients in the metal are very important. A temperature
gradient is defined here as a large change in temperature from one part of a
solid to another. The air cooled engine by its very nature incorporates high
temperature gradients. High temperature gradients lead to cracking as part
of the metal wants to expand and this expansion is resisted by the cooler
parts. Cracked cylinder heads are a common problem with air cooled engines.

Pistons, of necessity, are almost always made of light weight aluminum by
the fact that they are starting up and stopping twice per engine revolution.
The modern light weight high HP piston engine would not be possible without
light weight aluminum pistons. Consequently they must be cooled by oil or
they too will soften, fail from over stress and may melt. Heavy duty
aluminum pistons sometimes have steel or cast iron inserts in the high
stress, high temperature ring land area.

Review of basic principals and history.

HP is a force times a velocity, Or put another way, force times a distance
traveled per second.  Watts (heat per unit time) and HP are interchangeable
(746 watts equals one HP). It takes HP at the power plant to generate the
electricity that is used to light your 100 watt light bulbs.

Therefore the critical thing when designing a cooling system is the amount
of continuous HP the engine is called on to develop. A cooling system that
works for a light duty car application might not work for the same 200 HP
engine when used as an aircraft engine. Cars generally average around 40 HP
continuously and are occasional required to climb a long steep hill where
the HP required might climb to as high as 100 HP. The same engine used in an
aircraft might be expected to generate 200 HP or max power for as long as it
takes to climb to cruising altitude. Then it may be called on to generate
150 HP continuously. The design of the aircraft engine cooling system is far
more demanding. You cannot merely pull over to the side of the road if the
cooling system fails.

Heat balance

All engines have a heat balance. Heat balance is a list of the portions of
the total energy in the fuel that is distributed to various purposes. The
intrinsic efficiency of the engine determines what fraction of the heat must
be rejected.  The heat rejected is the sum of what goes out the exhaust pipe
and what goes out the cooling surfaces.

A Rolls Royce Merlin used in the P51 has a heat balance of:

Mechanical HP out 29.6%
Waste cooling heat 18.9%
Waste exhaust heat 51.6%

One of the most efficient aircraft engines ever designed was the liquid
cooled O-200 Continental aircraft engine as used in Dick Rutan and Jeana
Yeager's around the world non stop and un-refueled Rutan Voyager.
Continental claimed this engine had a heat balance of:

Mechanical HP out 35%
Waste cooling heat 21%
Waste exhaust heat 44%

For the early 1981 Mazda 12A rotary engine it was
about:

Mechanical HP out 28%
Waste cooling heat 24%
Waste exhaust heat 48%

This differs little from any other typical intermittent internal combustion
engine but small efficiency improvements continue to be made in the Mazda
wankel rotary. The most recent Mazda RX8 engine is far better than this.
However I have no heat balance numbers yet.

The large amount of energy 'thrown away' as waste heat is an inescapable
result of the laws of thermodynamics.  Engineers try to keep a high
combustion temperature for maximum overall efficiency and a high coolant
temperature (by minimizing the heat rejected to the cooling system), while
not melting or softening the metal the engine is made of!

This is also why turbo compound engines with turbines in the exhaust stream
recovering some of that waste energy are able to drastically improve the fuel
efficiency of the internal combustion engine.

There are three main ways of transferring heat. Conduction, convection and
radiation. Conduction takes place between solids or within the same solid
while convection happens between a solid and a fluid such as air or water.
Heat flows from the hot parts to the cool parts. Radiation takes place by
one object "seeing" another hotter object. No not nuclear radiation.  You
can be warmed by the sun no matter how cold it is out side. You will not be
warmed if you are in the shade. This is a classic example of heat transfer
by radiation. We are mainly concerned with conduction and convection when
cooling an engine.

This is a bit subtle and concentrated so it will take some cogitation and
re-reading for this to sink in.

The continuous HP per cubic inch of displacement and the geometry of the
combustion chamber combine to determine how much heat must be rejected per
unit time through the walls of the combustion chamber. The higher the HP per
cubic inch the more severe the cooling problem. If your engine has 20 square
inches of combustion chamber surface to transmit 100 HP worth of waste heat
that is easier to do than transferring 200 HP worth of waste heat with the
same 20 square inches. This is the very crux of the situation in getting the
heat out of the metal and into the coolant when dealing with high continuous
HP per cubic inch engines. A high amount of coolant flowing past the
combustion chamber area is very important.

Aluminum combustion chambers melt at a lower temperature than cast iron but
they conduct heat better than cast iron. Most really heavy duty piston
engines use cast iron and or steel combustion chambers, heads and cylinders.
As you can imagine there is a severe compromise in the power to weight ratio
of any heavy duty piston ICE.  It is also one of the developmental
differences between a heavy duty truck engine, an aircraft engine and a
light duty car engine. A heavy duty, over the road, tractor trailer turbo
diesel engine can weigh 2000 pounds while generating only 350 HP. The life
is expected to be up to or over 500,000 miles. A 350 HP air cooled aircraft
engine has a similar duty cycle and expected life but it weighs only about
500 pounds. This is a remarkable power to weight ratio for a heavy duty
engine. This is the result of 100 years of detail refinement and
development. Only the finest materials are used. That is also one reason why
aircraft engines are so expensive. Remarkably a stock low cost Mazda rotary
engine has about the same power to weight ratio as an aircraft engine but is
far more robust. A stock Mazda engine block can withstand four times the
stock HP rating by upping the boost on the turbo. Try that with a piston
engine and you are looking at catastrophic failure in the hand grenade mode
within 30 seconds.

Air cooled aircraft engines have relatively low HP per cubic inch partly for
the durability reason and partly for the heat transfer reason. They are high
displacement low RPM engines. This is opposite to what we have been told for
many years. Most automotive journalist tout the high power to weight ratio
of high RPM low displacement piston engines. The key to this misconception
is automotive piston engines are very light duty engines. Particularly in
the higher HP versions.

A Mazda wankel rotary engine is indeed a small displacement engine as it
displace only about 80 cubic inches while generating 200 HP therefore its
cooling requirements appear to be severe. It is the exception in car engines
duty cycle wise. It dominated endurance car racing until it was banned by
car racing organizations as being unfair to piston engines.  It is the only
car engine that can withstand maximum power for hours on end. Unlike a
piston engine however the aluminum combustion chamber or rotor housing of a
Mazda wankel engine is lined with steel and is of very large area. The steel
is chrome plated which of course is shiny and reduces heat transfer by
radiation from the white hot combustion gases.

The other very important difference is the combustion chamber moves as the
rotor turns. This means the area the combustion heat is applied too is much
larger than a piston engine. This is not good from an efficiency point of
view but excellent from a cooling point of view and hence HP per cubic inch
of displacement point of view. Mazda rotary engines have generated as much
as 400 HP per rotor without melting the rotor housings. To my knowledge
nobody has ever melted a rotor housing which happens to be a rather common
problem with piston engine aluminum combustions chambers. Metal temperature
measured near this surface are much the same as normal temperatures in
piston engine combustion chambers. About 500 degrees F. The rotor is cast
iron and nearly impossible to melt. (See the side bar the Rotor Verses the
Piston.)

In a Mazda wankel rotary engine one third of the cooling requirement is oil
and two thirds is water or glycol/water mix.  This differs from most piston
engines. The piston must be cooled by the oil of course but the piston makes
up a smaller portion of the total metal surface exposed to the hot
combustion gases at any one time. The rotor face makes up a large portion of
the surface area in contact with the hot combustion gases during the entire
expansion cycle.  A piston top is a decreasing proportion of the total area
exposed to the hot gases as the piston moves down the bore. Oil cooling in a
wankel rotary is therefore very very important.

Pistons, of necessity are almost always made of light weight aluminum for
reasons mentioned above. The wankel rotary engine on the other hand can and
does use a heavy cast iron, impossible to melt rotor as the rotor never
stops and starts up in the opposite direction. It moves in a perfect circle
at all times. Never the less the rotor must also be cooled with oil.
 
In an air-cooled engine waste heat is transferred by the fins on heads and
cylinders directly to the air. This is done first by conduction and then by
convection. The rate of heat transfer is proportional to the difference in
temperature between the fins and the surrounding air and the total area of
the fins. This temperature difference, initially, is about 400 degrees. The
metal surface surrounding the combustion chamber is 500 degrees and the
ambient air temperature surrounding the fins is about 100 degrees F. If the
engine gets hotter the rate of heat transfer increases. In a sense this is a
stable cooling system. Higher temperature, faster heat transfer, more
cooling. The fin density or number of fins per inch of head or cylinder
surface must be such that the rate of heat transfer to the air keeps the
hottest part of the combustion chamber below the melting or softening
temperature of the metal.  Usually aluminum and usually about 500 degrees F.
The signs of too high a metal temperature in the combustion chamber is a
pock marked metal surface.

In a water cooled engine waste heat is first transferred to the surface of
the coolant jacket by conduction and then to a coolant by convection. It is
important to keep the coolant flow high to facilitate this heat transfer
process. The amount of heat removed from the metal is proportional to the
volume of flow and the increase in temperature of the coolant. It is stored
in the coolant and moved to a heat exchanger where it is then transferred to
the air by convection. The decrease of the coolant temperature in the heat
exchanger must match the increase in temperature that occurred in the block.

The coolant temperature is limited to about 250 degrees due to boiling. The
combustion chamber metal temperature can be 500 degrees so the difference is
only about 250 degrees initially and the rate of heat transfer is lower than
in an air cooled engine. If the engine metal gets hotter the rate of heat
transfer will at first be increased by a process called nucleate boiling.
Nucleate boiling is a benign form of small scale boiling with enhanced heat
transfer characteristics.  Tiny bubbles of vapor are formed and they work
their way to a vapor pocket at the very top of the cooling system by
buoyancy. Bubbles rise. If they cannot make their way to the top of the
cooling system steam pockets may occur which are very bad as the heat
transfer from metal to coolant in that local area is greatly reduced.

The internal shape of the water jacket around the combustion chamber and the
flow of coolant must be such that no local severe boiling takes place at the
maximum continuous power the engine is likely to generate. This is
regardless of the bulk temperature of the coolant.  If this rule is violated
parts of the combustion chamber may melt or crack. This is the limiting
factor in how much continuous HP you can get out of any liquid cooled ICE.

Electric coolant pumps.

Some have advocated replacing the mechanical pump with an electric pump. It
takes about three HP to drive the mechanical pump at 6000 RPM and provide
the coolant flow necessary to keep the metal temperature near the combustion
chamber within bounds. In the last 100 years there have been no break
troughs in impeller shapes that would allow a significant improvement in
pump efficiency. Therefore it would require about three HP from the electric
motor driving the pump.  There are 746 watts in one HP that is at minimum
2,238 watts. Since the electrical system voltage is 12 volts and watts
is equal to voltage times current it would require about 186 amps out
of the battery to provide the required coolant flow.

Back to reality.

If the temperature of the bulk coolant goes beyond the boiling temperature
overall boiling occurs of course.  All hell breaks loose and the coolant is
forced out of the water jacket through the pressure cap overflow valve by
massive amounts of hot vapor or steam. Once boiling of the coolant starts it
is all over in an instant. Less coolant, hotter metal, more steam, less
cooling. One might say the cooling system pops like a pop corn. Thus the
cooling system becomes very unstable indeed. Something we all have probably
experienced in our cars.  This is a major difference between the air cooled
engine and the liquid cooled engine. Needless to say this cannot be allowed
to happen in an aircraft engine.

Once the heat is in the coolant it must be transferred to the air by the
heat exchanger (unfortunately called a radiator) by convection. In this case
the heat exchanger is 250 degrees F and the air is 100 degrees F or a
difference of only 150 degrees so the rate of heat transfer is also
unfortunately still lower. Small differences in ambient air temperature make
relatively large differences in the cooling of liquid cooled aircraft
engines. This is much more pronounced and noticeable than it would be with
air cooled aircraft engines. Air cooled engines historically work better in
the desert as a ten or twenty degree F increase in ambient temperature is a
smaller percentage difference in an air cooled engine compared to a liquid
cooled engine. As a rule of thumb a one degree F increase if air temperature
is a one percent loss in cooling capacity.

The Hard Part. Getting The Air To Go Through The Heat Exchanger.

Here is where most builders have difficulty. It is not easy to design a low
aerodynamic drag aircraft liquid cooling system. The cooling drag can be as
much as 30% of the overall drag on the airplane. The scoops and ducts must
be designed properly. The job of the duct is to slow the air velocity and to
raise the pressure to the point that it optimise's radiator performance. 
Many lessons were learned in world War I & II on this subject. Some
important improvements did not have time to be applied during the war so
some of the best lessons CANNOT be learned by studying WW II aircraft. The
P51 seems to be the ultimate liquid cooled aircraft of WW II yet its duct
system has been improved upon by some work published after the war.  There
is an out of print book written by Kuchemann and Weber published in 1953
called The Aerodynamics of Propulsion wherein much of this has been
presented. Used copies are still available on the Internet.
Prior to this work most duct designers thought it necessary to design the
duct so there was a gradual transition from scoop opening to heat exchanger. 
See figure x. This is not the case and a more bell shaped transition is
called for such as figure y. The coordinates of such as duct are given in
table a. 

If the rad is perpendicular to the air flow and a bell shape is not used
there will be separation of the flow on the duct walls and most of the air
will impact the center of the rad with a consequently loss of air flow
around the edges of the rad. If the duct is NOT about 2-1/2 times longer
than the max dimension on the rad a small opening (25% to 30% of rad area)
can not be used. In that case a larger and larger opening as the duct gets
shorter and shorter will be required. The limit to this is no duct at all
with just the rad stuck out into the air stream perpendicular to the air
flow like a WW I fighter plane. Any air that cannot flow through the rad for
what ever reason (such as an object placed too close behind the rad) will
cause external diffusion and the air will be diverted around the rad with
high cooling drag. External diffusion happens anywhere flow into the duct is
restricted. A NACA duct is not recommended for heat exchangers as small
pressure increases in the duct trigger the air to bypass the opening.
A form of external diffusion so to speak.

External diffusion does not cause as high a drag  as intuition and flat
plate drag theory would have you believe. The external diffusion phenomena
reduces the drag somewhat.  The WW II Corsair cowl is a good example. The
Corsair was no faster when a spinner was tried because of external
diffusion. Later radial engines were made with longer crank cases and a
longer internal bell shaped radial engine cowl could then be employed with a
reduction in cooling drag over the Corsair configuration. These had a
smaller diameter opening of less area. The Lockheed P2V and late versions of
the Constellation among others are good examples.

Air Flow Measuring Tools.

An essential tool for troubling shooting a cooling system design is a water
manometer. Air flow cannot be seen. Particularly from inside the cockpit
when you are flying. Some way of determining what the air is doing in the
cowl is extremely important. A low cost water manometer can do it all. It
can measure static pressures and air velocity. The idea is to achieve as
high a static pressure in front of the rad as possible and as low a static
pressure as possible on the rear of the radiator. Water manometer are easily
fabricated from clear plastic tubing and low cost drip irrigation plastic
Tees and elbows. Drip irrigation vinyl tubing is about $4 for 50 feet and is
ideal for plumbing the measuring points in the cowl to the manometer. A
small piece of porous kitchen foam hot glued to the end of the vinyl tubing
insures you are measuring static pressure and not stray local air currents.
If you point the end of an open tube in the direction the air flow is coming
from you can measure the air velocity. This is the same principle used in a
pito tube. At sea level the formula is MPH = the square root of (pounds per
square foot /.0026). This is called dynamic pressure. One inch of water is
equal to .037 psi so you can see that a water manometer is extremely
sensitive device for measuring low pressures and air velocity.

Improvements in heat exchangers.

In recent years improvements have been made in heat exchangers themselves.
It has been found that the thinner the heat exchanger is the better it works
given a constant volume. In other words a heat exchanger that is twenty
inches by twenty inches by one inch thick works better than a heat exchanger
that is ten inches by ten inches by four inches thick. Both heat exchangers
have the same four hundred cubic inches of core volume. The rule of thumb is
the first half of the exchanger does three quarters of the work. This is
because as the cooling air passes through the heat exchanger it warms up and
the all important difference between the air temperature and the metal at
the back of the heat exchanger is reduced and so too is the the rate of heat
transfer.

Another advantage of the thin heat exchanger is the air pressure required to
push air through the exchangers is lower as the exchanger gets thinner and
thinner. The pressure available is an out growth of the dynamic pressure
caused by the aircraft moving through the air at high speed. All modern cars
and trucks use thin, large area heat exchangers mounted perpendicular to the
air flowing through the grill. They need to cool while climbing hills when
the HP requirements are highest at only sixty or so MPH where the dynamic
pressure is only 0.065 PSI. A good scoop and duct or grill will give up to
about 80% of this or only 0.052 PSI. Not much!!!  This is also the reason
large cooling fans are used in trucks and cars. At 100 MPH which is a
typical climb speed for a lot of airplanes the dynamic pressure is only 0.18
PSI. Again a properly designed scoop and duct will give about 80% of this or
0.14 PSI.

Unfortunately a large area thin heat exchanger is difficult to install
perpendicular to the air stream in an aircraft as it increases the
aircraft's frontal area. Let me define the term frontal area so there is no
mis-understanding. Technically frontal area is the maximum cross sectional
area of the aircraft perpendicular to the air stream regardless of where
along its fuselage length this occurs. It is also called flat plate area. 
It could occur half way back only the fuselage. Some people mis construe
this term to mean the shape of the front of the airplane. Not true.

Once the flow enters the duct and the rad is not perpendicular to the axis
of the duct the impact effect comes into play. The air pressure will build
at the back of the duct due to the inertia of the air while little air will
flow through the part of the rad closest to the front. In other words if the
duct is NOT a wedge shape, more or less, most of the air will try to flow
through the rad closest to the back of the duct. The part of the rad nearest
the front will have less air flow. More about wedge shaped ducts later.

Thick heat exchangers were successfully  used in fast airplanes during WW II
as the dynamic pressure available is proportional to the square of the air
speed. At two hundred MPH the dynamic pressure is four times what it is at
one hundred miles per hour.  Most home built airplanes climb out at about
one hundred MPH so a thin heat exchanger is called for. A P51's heat
exchangers were a total of eleven inches thick and at four hundred MPH the
dynamic pressure was sixteen times higher than it is at one hundred MPH. 
Never the less P51 and other liquid cooled WW II airplanes had problems
cooling while taxing for takeoff and climb out. Obviously a thin heat
exchanger is called for in home built aircraft. The problem is how do we fit
them in? The solution was mentioned in Kuchemann & Weber book Aerodynamics
of Propulsion and another book still in print published after WW II by Kays
& London simply called Compact Heat Exchangers.

The heat exchanger is installed at an oblique angle to the center line of
the aircraft as shown in figure z. This was referred to by K & M as an
oblique cooling block. Unfortunately it is difficult to apply the bell
shaped duct principle to such an arrangement so a modified shape was
developed by K & M and Kays & London shown in figure r. An approximation to
this shape that works fairly well is a simple wedge shaped duct as shown in
figure s. The distribution of the air flow through the heat exchanger can be
checked with a test rig. This consist of a blower, a heat exchanger and a
prototype duct. The air velocity distribution coming out of the back side of
the heat exchanger can be measured by a hand held digital wind speed meter
or an array of pito tubes.

Why not place the thin rad in the back of the fuselage perpendicular to the
air stream similar to the P51 arrangement? This will work of course but the
frontal area of the the aircraft will be increased by the P51 type scoop
required. If the P51 type scoop is on the aft end of the airplane where the
boundary layer has had considerably time to build up and there is
insufficient distance between duct opening and skin surface the airflow will
merely be diverted around the opening. If the P51 duct is well out of the
boundary layer and the flow into the duct is restricted, say by too thick of
a rad, external diffusion will also take place.

This in effect is what happens to the NACA duct when used with a flow
restriction like a radiator even if located where the boundary layer is
thin.  A flush duct such as a NACA duct was not recommended by NACA for
feeding air to a flow restriction such as found in a heat exchanger. Recent
full scale tests have verified this as fact. The NACA duct will fail to work
under those circumstances.

Furthermore the long plumbing lines required for a rear mounted heat
exchanger might tax the capabilities of the stock Mazda water pump. Not to
mention the extra weight of coolant required to fill the necessary long and
large diameter coolant lines. Speaking of water pumps Mazda mounted the pump
on the top of the engine. This means if the coolant level drops in the
system the pump will cavitate by sucking air. Some builders have installed
water pumps lower down on the right side of the engine to forestall this
potential problem.

It is better to fit obliquely mounted heat exchangers under the cowl.  This
is facilitated by the small size of the Mazda wankel engine compared to a
air cooled aircraft piston engine. The two rotor engine is twelve inches
wide by 14 inches high by eighteen inches long not including the PSRU. This
is much smaller than any piston engine of comparable HP. There is usually
plenty of room under the engine for an oblique mounted radiator with the air
exiting straight down through Venetian blind style cowl flaps. In airplanes
with retractable nose landing gear one or two obliquely mounted heat
exchangers, one on either side of the engine, work well exiting straight out
into the air stream also with Venetian blind or gill type cowl flaps. Fixed
louvers could be used if high speed drag is not an important factor.

It is also very important to make sure the air going through the heat
exchanger is as cool as possible so it is is unwise to pass the air over the
hot engine and then through the heat exchanger. This increases the
temperature of the air flowing through the heat exchanger reducing the all
important difference in air temperature. It is far better to duct the cool
air directly to the heat exchanger and then let it flow over the engine or,
better yet, directly to the outside of the cowl. Thus the duct serves two
purposes as it also shields the heat exchangers from radiant heat emanating
from the red hot exhaust system.

When the air arrives at the heat exchanger it should be slowed considerably
and the pressure should be about 80% of dynamic pressure if the duct and
scoop were designed properly. As the air passes through the heat exchanger
the pressure will be reduced still further. The air temperature will
increase up to about 40 degrees F above ambient air if the heat exchanger is
designed properly. The product of the increase in air temperature and the
volume of flow determines the amount of heat carried away.  The air will
then exit through the cowl outlet openings. As the air passes through these
relatively small openings it must speed back up to near or above the slip
stream velocity and be directed to the rear of the aircraft to recover as
much energy from the cooling system as possible. This energy recovery phase
reduces the over all cooling drag.

The other half or more of the equation is getting the air out the back side
of the heat exchanger. According to our test of off the shelf car radiators
using a duct mounted on top of a Dodge Caravan (see picture on the web site)
and a water manometer it is very important to have a clear space behind the
heat exchanger. K & M also discussed this factor. The air flow coming out of
the heat exchanger can be as high as 40% of the forward speed of the
airplane. If there is an object within ten inches behind the typical heat
exchanger this air velocity will convert to static pressure on the face of
the object. This in turn will impede the air flow through the heat exchanger
in the vicinity of the object. It is best to mount the heat exchanger on the
inside surface of the cowl and have a totally unobstructed back side by
exiting the air directly to the air stream.

If an opening is at the bottom or edge of the firewall a large radius must
be used on the firewall just before the opening in the cowl. This minimizes
the energy loss of the high speed cooling air exiting the cowling. This has
been well known for at least 70 years.

Cowl Flaps

Cowl flaps must be used on high speed airplanes as the dynamic pressure
varies widely over the speed range of an airplane. At 100 MPH it is 26
pounds per square foot and at 200 MPH it is over 100 pounds per square foot. 
The most severe cooling requirements are during a max angle max HP climb out
at about 100 MPH. At 200 MPH the dynamic pressure is four times higher than
it is at 100 MPH and so too is the potential volume of air available to cool
the engine. Therefore the cowl flaps are nearly closed increasing the
pressure in the entire cooling system and diverting the air flow around the
scoop and reducing the internal air flow to roughly what was needed at 100
MPH. This greatly reduces the cooling system drag. Probably worth about five
or ten MPH additional cruise speed around 200 MPH. More as aircraft cruise
speed increases.

Full power top speed runs such as cross country racing are another matter.
Top speed is proportional to HP cubed while dynamic pressure is proportional
only to the square of speed. Consequently cooling again becomes an issue. It
may then be necessary to partially open the cowl flaps again. There is
nothing like a long cross country race on a hot day to test your cooling
system.

In a car, a thermostat is used to control the engine temperature.  This is a
bad idea in an aircraft engine as it is better to use the cowl flaps for a
reduction in cooling drag as stated above.

Nitty Gritty Cooling Details.

A few bad experiences like this during climb out can be very discouraging
indeed. Forget all you have learned about dealing with car overheating. You
cannot just quickly reduce power and pull over to the side of the road and
stop. One must budget about $1000 for cooling plumbing when installing a
liquid cooled engine. The finest plumbing available should be used which
means all AN lines and fittings. To give you a feel for the problem a liquid
cooled aircraft cooling system can run 30 psi pressure. If the coolant hose
is 1-1/2 inch in diameter there is a force of 53 pounds trying to pull the
hose off the hose barb. If non AN hose is used dollar-an-inch high quality
silicone rubber hose must be used with constant torque hose clamps made by
the Breeze Co. Constant torque hose clamps keep a constant pressure on the
hose as the aluminum hose barbs expand and contract underneath. Automotive
black hose and 50 cent hose clamps are not going to cut it.  Several rotary
powered airplanes have crashed destroying years of work and untold amounts
of money because the builders ignored this very important detail.

Many people have given up on water cooled aircraft engines at this point. It
takes an extra amount of perseverance and determination to continue.

Don't be tempted to use an oil to water heat exchanger for oil cooling.
Several builders have tried this and failed miserably.  A fundamental factor
in the process of heat transfer is the difference in the temperature of the
two mediums. The rate of heat transfer in any heat exchanger is proportional
to this temperature difference. In an OIL TO WATER heat exchanger this
difference could be as little as 10 degrees Fahrenheit. Water 220 oil 230
degrees F.  In a OIL TO AIR heat exchanger this difference is usually about
100 degrees F or more.

An OIL TO WATER heat exchanger adds to the water cooling problem and reduces
the margins of safety between normal operation and unstable operation. (Bulk
coolant boiling). An OIL TO AIR heat exchanger oil cooling system on the
other hand is a stable system as oil boils at a temperature well above
normal oil operating temperatures. (Nothing worse in the middle ages than
being boiled in oil.) It is also a parallel independent cooling system and
will continue to function when the liquid cooling system fails giving a
slight additional time to get the aircraft on the ground. This is not
the case when an oil to water heat exchanger is used. In that case all your
eggs are in one basket.

Most aircraft liquid cooling  systems have a pressure gage. As the engine
warms up the pressure in the cooling system increases. If there is a slow
leak it will not necessarily be initially noticed by a reduction in
pressure. The leak might be such that the normal expansion of the coolant
will maintain the pressure despite the leak.

Consequently we have found that the best way to test for leaks is pressurize
the coolant system to about 30 psi while the engine is still cold before
flight. This can be done by installing an ordinary tire valve, called a
schrader valve, and using any convenient source of pressurized air.

There have been some recent improvements in materials used for sealing the
rotary, both oil and water. These improvements have made it more tolerant to
temporary over heating. These mainly consist of 500 degree high temperature
Teflon encapsulated silicone O rings used for sealing the water jacket and
the oil scraper rings on the sides of the rotor.

Unlike a two cycle piston engine the rotary will not normally seize. The
rotor, being made from cast iron, expands less then the rotor housing made
from aluminum. It will seize if your oil cooler plumbing fails and the
engine pumps all its oil overboard. In that case the bearings could fail or
the rotor could spin weld itself to the end housing.

In a typical water cooled automotive type engine the water jacket inside the
cylinder head is inaccessible. You can not open up the head and make
modifications and improvements to the water jacket. One car V8 aircraft
installation project that I know about actually drilled eight small holes in
the cylinder heads and ran eight small tubes to the pressure tank to vent
steam pockets.

This is not the case with a Mazda wankel engine. The rotor housing is a
separate part and when the engine is disassembled the interior part of the
rotor housing water jacket is readily accessible for modifications. The
development of the exact shape of the internal water jacket around the
wankel combustion chamber is still in a state of change after 40 years of
development as more and more power is extracted from the rotary.  Some turbo
charged Mazda rotary engines used in drag racing cars burning alcohol and
nitrous oxide are generating up to 400 HP per rotor or 800 HP for a two
rotor engine.

The normally aspirated four rotor Mazda engine that won the 1991 Le Mans 24
hour road race out right generated about 172 HP per rotor for a total of 690
HP. The engine was in excellent to new condition after the race.  Racing
Beat in Anaheim CA has about 200 hours on the dyno with a turbo three rotor
Mazda engine generating 900 HP on 100LL aviation fuel. That of course is 300
HP per rotor.  This engine is intended for use in an aircraft. I mention
these particular rotary engines to illustrate the importance of cooling
system development to these very high HP per cubic inch engines.

Several modifications have been tried such as grooving the walls of the
water jacket to improve the heat transfer from the metal to the liquid
coolant. We have discovered that this is unnecessary if you limit your
engine to around 200 HP for take off and cruise at 175 HP. However this
feature may come in handy in the future when we get more experience with
higher HP turbo charged Mazda two and three rotor engines in aircraft use.

We do know the two rotor Mazda  engine will currently generate about 200 HP
continuously with out melting or cracking for a bare engine weight of 235
pounds with PSRU for an excellent power to weight ratio. Tracy Crook near
Bell Florida has well over 1400 hours in his Mazda rotary powered RV4 with
no metal cracking or housing warping. Ken Welter near Portland Oregon has
run hundreds of pounds of nitrous oxide through his normally aspirated Mazda
wankel engine increasing the power from 200 HP to 300 HP momentarily helping
him get his 2900 pound modified Coot amphibian on the step and off the
water.

Having said all that and you are going forward anyway with your liquid
cooled aircraft engine it is best to suppress your creative urges and copy a
system that someone else has used and that you know works. Later on, after
you have flown your airplane for awhile, you can do a little experimenting.

Given this situation it is best to test your new water cooled engine powered
airplane at low use uncontrolled airports so you can get it back on the
ground right now!  Otherwise you may damage your engine or crash when the
steam starts spuing from the cowling.

I hope this clarifies and answers a few questions you may have had about
liquid cooling an an aircraft engine.

Paul Lamar