In addi-
tion, a device called the NAP (Normal
Oscillation and Pitch) chair was built
that simulated the vertical and pitching
motion of an airplane over a range of
vertical motion of about 6 feet. These
simulators allowed covering a range of
conditions systematically, rather than
obtaining different conditions by consid-
ering results on a number of different
airplanes. At the time the space pro-
gram started, a three-axis rotational
simulator was under construction in
which a cockpit was mounted to provide
angular motion in pitch, roll, and yaw.
This simulator was intended to study
combined rolling and yawing oscillations
of an airplane, but it found use for other
purposes during the space program. In
contrast with modern simulators that
use electronic displays and general-pur-
pose motion bases, these simulators
used mechanical systems to simulate
accurately the motion of the vehicle. The
output of the mechanical system was
amplied by a hydraulic servomecha-
nism based on a variable displacement
hydraulic pump. Servomechanisms of
this type were available in Naval gun tur-
rets. I have since learned that the devel-
opment of these servomechanisms was
largely attributed to Charles Manley, the
same man who earlier perfected the
excellent radial motor used in Samuel
Langleys aerodromes.
Almost simultaneously with the start of
the space program, all testing of high-
speed airplanes was transferred to the
High-Speed Flight Research Center at
Edwards Air Force Base, now called
the

Dryden Flight Research Center.
I was left with the problem of deciding
on the best use for the engineers under
my supervision, who had been trained
in studying the stability and control of
airplanes.
Airplanes, of course, had been ying for
many years, and there was little need
for studying the basic principles of sta-
bility and control. Most simulation work
on airplanes was devoted to studying
optimal stability and control characteris-
tics or to nding the characteristics that 36
Monographs in Aerospace History Number 40Journey Into Space Research
Space Simulation Studies
would provide the most desirable han-
dling qualities for the pilot. At the start of
the space program, before the rst
manned orbital ight, there was much
less condence in the ability of a human
pilot to perform the tasks required for
space operations. Many engineers
expressed the view that it would be
better to design spacecraft with com-
pletely automatic control. Test pilots, on
the other hand, who had at least
approached orbital ight conditions in
tests of very high-altitude airplanes,
usually felt condent that they could
control the entire ight of a spacecraft
just as they had controlled high-altitude
airplanes.
To resolve some of these questions, I
felt that the conditions encountered in
the various phases of a space vehicle
ight should be simulated as accurately
as possible to give the astronauts
experience with the new problems of
space ight. The following discussion
describes some of the work done in this
period.
After the successful completion of John
Glenns rst orbital ight, most doubts
concerning the effects of weightless-
ness were dispelled. Soon after this
time, however, denite space programs,
such as the Gemini and Apollo mis-
sions, were planned. The simulation
work then focused on specic problems
encountered in the launching, ight,
entry, and landing of the spacecraft
designed for these missions. These sim-
ulations are described in reference 5.1.
Lunar Landing Research
Facility
Landing on the surface of the Moon was
known to be one of the most critical
phases of the Apollo program. Control
by an astronaut, at least during the nal
phases of the descent, was considered
mandatory because the nature of the
lunar surface was not known in suf-
cient detail to plan the exact spot for
touchdown. Several conditions present
a control problem considerably different
from that of landing an airplane on
Earth. The lunar gravity is one-sixth that
on the Earth. All control of lift and atti-
tude is provided by rockets, which often
provide a discontinuous, on-off control
rather than a linear variation of control
force familiar on airplane controls. The
complete lack of an atmosphere on the
Moon makes it impossible to use any
type of aerodynamic control.
When President Kennedy made his
announcement of a major program to
send men to the Moon on May 25, 1961,
I immediately started thinking about
how this operation could be simulated.
I wrote a memorandum on this sub-
ject in May 1961 and discussed the
subject with the Associate Director,
Lawrence K. Loftin, Jr. on June 26,
1961.
The trajectory of the lunar vehicle would
be different from that on Earth because,
as stated previously, the gravitational
attraction of the Moon is only one-sixth
that of the Earth. To simulate the
reduced gravity, I visualized a suspen-
sion system for the simulated vehicle
that would exert a constant force in the
vertical direction equal to ve-sixth the
weight of the vehicle. The force on the
cable on which the vehicle was sus-
pended could be measured by a strain-
gauge balance at the vehicle and used
to control the output of a servomecha-
nism that reels the cable in and out as
required to apply the desired constant
force to the top of the cable. To provide
for horizontal motions of the vehicle in
the fore-and-aft and lateral directions,
sensors would measure the tilt of the
cable from the vertical and would be
used to control servomechanisms that
moved the suspension point to keep it
directly over the vehicle. Lunar Landing Research Facility
Monographs in Aerospace History Number 40Journey Into Space Research
37
The motions of the vehicle in response
to pilot commands would be provided by
rockets. As in an actual lunar vehicle, a
rocket sufciently powerful to support
the weight of the vehicle in the lunar
environment, plus some extra power to
maneuver, is required. Smaller rockets
are used to provide pitching, rolling, and
yawing moments. Previous studies had
found that a system using a platinum
catalyst to decompose hydrogen perox-
ide into steam and oxygen provided a
convenient and relatively safe means to
make a controllable rocket.
My rst concern in designing the lunar
landing facility was to analyze the servo-
mechanism used to maintain a constant
force in the suspension cable while the
vehicle was going through the maneu-
vers of landing. While the technical
details of this analysis are too involved
to present in this discussion, a brief
review of the problems involved in ser-
vomechanism design may be of interest.
An example of a simple type of servo-
mechanism is an autopilot to hold an
airplane on a desired constant heading.
If the heading deviates from the desired
value, a compass or other heading
detector measures the error in heading.
The error may be converted to an elec-
tric voltage that is fed to an electronic
amplier. The output of the amplier
drives an electric motor that moves the
rudder of the airplane in the direction to
reduce the error. As the heading error is
reduced to zero, the rudder is returned
to its neutral position. The ratio between
the rudder angle and the error in head-
ing is called the gain of the servomech-
anism. Increasing the gain increases
the speed with which the heading error
is reduced, but it may cause the rudder
to overshoot its neutral position and
oscillate about zero. Beyond a certain
value of the gain, the oscillations
increase with time, a condition called
dynamic instability. The gain must be
kept to a value safely below that which
produces instability.
In the case of the lunar landing research
facility, a servomechanism maintains a
constant force in the suspension equal
to ve-sixth of the vehicle weight while
the pilot controls the rocket that pro-
vides an additional one-sixth of the
weight plus whatever additional force is
required to control the rate of descent or
to slow the vehicle down for landing.
These force variations supplied by the
rocket act as disturbances to the force in
the cable. The error in the cable force is
corrected by the servomechanism by
varying the speed at which the cable is
reeled in or out.
If a steady disturbance is applied to a
system controlled by a servomecha-
nism, a steady error may result. This
condition may be corrected by placing
an integrator in the feedback loop. The
integrator builds up a signal that
increases with time to offset the steady
error. An integrator also has a gain to
determine its speed of operation. If the
gain is too large, dynamic instability will
again result. In general, the tendency to
instability is greater with an integrator in
the circuit.
In the case of the lunar landing facility, a
typical mode of operation is to let the
vehicle fall freely under a steady accel-
eration of one-sixth g until the descent
rate reaches the desired value to start
the landing run. In this case, the cable
tension should remain at its desired
value during this period of constant
acceleration. To maintain this steady
value during conditions of constant
acceleration, servomechanism theory
shows that a double integration of the
error is required. Such an arrangement
is called a type 2 servomechanism. I
therefore realized that a double integra-
tion should be in