*Test Methods in General*. As already explained, the behavior of a body
in an air stream cannot be predicted with any certainty by direct
mathematical calculation, and for this reason, each and every
aerodynamic body must be tested under conditions that are as nearly
similar to the actual working conditions as possible. Prior to Professor
Langley’s first experiments in 1887, mechanical flight with a heavier
than air machine was derided as an impossibility, even by such
scientists as Navier, Von Helmotz, Gay-Lussac, and others, who proved by
the most intricate calculations that a body larger than a bird could not
be supported by its own energy. Such calculations were, of course, based
on a wrong understanding of air flow, and as no experimental work had
been done up to that time, the flow was assumed according to the
individual taste and belief of the demonstrator. The presence of a
vacuum on the back of a plate was not understood, and as this
contributes full two-thirds of the lift, it is an easy matter to see why
all of the early predictions fell short of the actual lifting forces. To
quote one classic absurdity, the scientist Navier proved mathematically
that if mechanical flight were possible, then 17 swallows would be
capable of developing one horsepower.
In spite of these discouraging computations, Langley proceeded with a
very carefully conducted series of experiments, first investigating the
laws of surface sustenation on various forms of plates, and when the
data collected was sufficient for his needs, he started to construct a
number of model flyers with various wing arrangements and aerofoil
forms. It was Langley’s experiments upon aerofoils that cleared the way
for the Wright Brothers, who started a further and more complete
investigation in 1896. Experiments were made on the effect of curvature,
aspect ratio and angle of incidence, and the results obtained in their
"wind tunnel" were afterwards applied to their successful full size
machine. During 1901 to 1902 the Wrights investigated the properties of
at least 100 different aerofoil forms. Both Langley’s and Wrights’
experiments were with models, although they were made in a different
manner. It was in this way that experimental evidence gained precedence
over theory.
Langley’s specimens were mounted at the end of a revolving arm, so that
with the arm revolving, a relative air stream of known velocity could be
had. The aerofoil was mounted in such a way that the lift and drag could
be measured. In the early experiments of the Wrights, the models were
placed in an enclosed channel through which a stream of air was
maintained by a fan. The model was attached to a balance system so that
the lift and resistance could be measured. This is what is now known as
a "wind tunnel," and at present is almost exclusively used in model
tests. Several investigators immersed their model aerofoils in running
water so that the direction of flow could be visibly observed. While
this latter method is of great service in determining disturbances,
stream line flow, and general characteristics, it is qualitative rather
than quantitative, and cannot be used in obtaining accurate numerical
results. A more accurate method of mapping out the direction of flow,
eddies, etc., is to introduce smoke into the air stream.
*Full Size Experiments*. The old "rule of the thumb" method of building
a full size machine without model test data or other experimental
evidence to begin with has seen its day. It is not only exceedingly
expensive, but is highly dangerous, and many a flyer has met his death
in the endeavor to work out untried principles on a full size machine.
The first cost of the machine, the continual breakage and operating
expense, to say nothing of the damage suits and loss of time, make a
preliminary full size tryout an absurdity at the present time. Again,
the results of full size experiments are not always reliable, as so much
depends upon the pilot and weather conditions. The instruments used on a
large machine are far from being as accurate as those used in model
tests. These are also likely to be thrown out of adjustment unknowingly
by falls or collisions. The great number of variables that enter into
such a test make it almost an impossibility to obtain accurate data on
the result of minor alterations, and, in fact, it is almost impossible
to get the same results twice without further alterations than changing
the pilot. Full scale tests are necessary after sufficient data has been
obtained and applied in a scientific manner to the design of the
machine, but successful performance cannot be expected from a powered
machine built by guess work.
When performed in connection with a wind tunnel, or based on dependable
data from other sources, full size wing tests are very instructive and
useful if care is taken to have the tests conducted under uniform and
known conditions. Many full size experiments of this nature have been
carried out by Saint-Cyr University in France, and by the Royal Aircraft
Factory in Great Britain. Both of these institutions have a wind tunnel
and an almost unlimited fund of performance data, and last but not
least, have the services of skilled observers.
At Saint-Cyr, the full size wings, or the entire machine, are carried on
an electric car or "chariot." The speed of the car, the lift and drag,
can be determined at any moment during the run through suitable
recording devices. Actual flying tests have also been made, the
measurement of the propeller thrust giving the drag, while the lift is
known as being equal to the weight of the machine. The R.A.F. have
carried out a very extensive series of flight tests, the experiments on
the old "B. E.-2" probably being the best known.
The greater part of the experiments performed with the car at Saint-Cyr
differed considerably from the results obtained by model tests, and
apparently these differences followed no specific law. According to
theory, and the results obtained by different laboratories, the
performance of a full size wing should be better than with a model, but
the Saint-Cyr tests showed that such was not always the case. The center
of pressure movement differed in almost every case, and as a direct
result, the pressure distribution of the large wings was materially
different than with the model. The lift-drag ratio results varied,
sometimes being better for the model than for the large wing. These
differences can probably be explained as being due to variation in air
currents, side winds, etc.
*Model Tests*. Since lift and resistance are due to relative motion
between a body and the air stream, a model can either be towed through
the air, or it can be held stationary while the air is forced past it.
There has been some controversy on the relation between the results
obtained by the two methods, but for the present we will accept the
common belief that the results obtained by either method are the same.
In testing ship models, they are always towed through the tank, but in
the case of aero-dynamic bodies this is complicated and not desirable.
In towing models through the air a very high velocity is needed and this
necessitates either a very long track or a short time length for making
the observations. Again, it is almost impossible to avoid errors because
of vibration, inequality of movement due to uneven track, or air eddies
caused by differences in temperature and by the movement of the towing
device. In fact, the same difficulties apply to towed model tests as to
the full size "electric chariot."
The whirling arm method of testing as used by Langley, Maxim-Vickers,
and others, is a form of "towed testing," but is also open to serious
objections. Unless the arm is very long, every part of the model surface
will not move at the same velocity, the outer portions moving the
faster. As the forces produced by an air stream vary as the square of
the velocity, this may introduce a serious error. The fact that the body
passes repeatedly over the same path introduces error, as the body after
the first revolution is always working in disturbed air. The centrifugal
force, and the currents set up by the arm itself all reduce the accuracy
of the method.
When a model is placed in a uniform current of air in a properly
designed channel or tunnel, the greater part of the errors due to towed
tests are eliminated. The measuring instruments can be placed on a firm
foundation, the air stream can be maintained at a nearly uniform speed
and with little error due to eddies, and the test may be continued under
uniform conditions for an indefinite period. While there are minor
errors due to wall friction and slight variations in the velocity at
different points in the cross section of the tunnel, they are very small
when compared to the errors of towing. For this reason the wind tunnel
is the accepted means of testing.
*Eiffel’s Wind Tunnel*. The Eiffel Laboratory at Auteuil, France, is
probably one of the best known. The results in Chapters III and V were
obtained in this laboratory and thousands of similar experiments have
been carried out at this place. Two tunnels, a large and small, are
placed side by side in the main laboratory room, the tunnels being
supported midway between the floor and ceiling. The air is drawn from
this room into an airtight experimental chamber through a bell-mouthed
circular opening. A grill or honeycomb baffle is placed in the opening
to straighten out the flow, and from this point the air passes across
the chamber and exits through a circular duct to the suction side of a
large fan. From the fan the air is discharged into the room. The same
air thus circulates through the tunnel continuously. The test chamber is
considerably wider than the openings so that the walls do not influence
the flow around the model. A cylinder of air passes through the chamber
at a remarkably uniform velocity, and without any appreciable eddies.
Diameter of the stream approximates 6.6 feet in the large tunnel and 3
feet in the smaller. In the large tunnel the maximum velocity is 105
feet per second, and 131 feet per second is attained in the smaller. A
50-horsepower electric motor is used with a multiblade fan of the
"Sirocco" type.
The observer and weighing mechanism are supported above the air stream
on a sliding floor, and a standard extends from the model in the wind
stream to the balances on the weighing floor. These balances determine
the lift and drag of the models, the center of pressure, etc.
*The N. P. L. Tunnel*. The National Physical Laboratory at Teddington,
England, has a remarkably complete and accurate aerodynamic equipment.
This consists of a large tunnel of 7 square feet area, a small tunnel of
4 square feet, and a whirling table house. The large tunnel is 80 feet
in length with an air flow of 60 feet per second, the air being
circulated by a four-bladed propeller driven by an electric motor of 30
horsepower. The velocity is uniform within one-half per cent, and the
most accurate of results have been obtained. The smaller tunnel is about
56 feet long and the wind velocity is about 40 miles per hour maximum.
The propeller revolves at 600 revolutions and is driven by a
10-horsepower electric motor. There is no chamber and the models are
suspended in the passage half way between the "Diffuser" in the entering
end, and the baffles in the exit. The Massachusetts Institute of
Technology, and the Curtis Aeroplane Company both have similar tunnels.
*United States Navy Tunnel*. In this tunnel the air is confined in a
closed circuit, the return tunnel being much larger than the section in
which the tests are performed. The cross-sectional area is 8 square feet
at the point of test, and the stream is uniform within 2 per cent. The
balance and controls are mounted on the roof of the tunnel, with an arm
extending down through the air stream to the model, as in the Eiffel
tunnel. The balance is similar to Eiffel’s and is sensitive to less than
2/1000 pound. A velocity of 75 miles per hour may be attained by the
500-horsepower motor, but on account of the heating of the air stream
through skin friction, the tests are generally made at 40 miles per
hour. Models up to 36-inch span can be tested, while the majority of
models tested at M. I. T. are about 18 inches.