Proceedings of the EuroGNC 2013, 2nd CEAS Specialist Conference on FrAT3.

1
Guidance, Navigation & Control, Delft University of Technology, Delft,
The Netherlands, April 10-12, 2013

TECS Generalized Airplane Control System Design – An

Update
Antonius A. Lambregts *

Abstract. The Total Energy Control System (TECS) was developed in the early eighties to overcome well
known safety/design deficiencies of traditional Single Input/Single output (SISO) based Flight Guidance
and Control (FG&C) systems. TECS uses generalized Multi Input/Multi Output (MIMO) based airplane
control strategies to functionally integrate all desired automatic and augmented manual control modes and
achieve consistently high performance for airplane maneuvering in the vertical plane. This paper
documents further insights gained over the past years on TECS design details for achieving precision
control decoupling, integration of augmented manual control modes, embedded envelope protection
functions and innerloop design using airplane dynamic model inversion. Additionally, non-linear design
aspects are discussed, including thrust limiting, energy management, maneuver rate limiting and mode
logic.

1 Introduction
Automatic Flight Guidance and Control (FG&C) systems have evolved into highly capable systems. These
systems have contributed immensely to the improvement of aviation safety. Unfortunately, these systems
still use traditional SISO control strategies that do not provide full 6 degrees of freedom airplane control.
Therefore, airplanes equipped with these systems are still vulnerable to Loss of Control (LOC).
Furthermore, these systems have become exceedingly complex, due to an excessive number of modes,
mode overlap and mode idiosyncrasies, making it a challenge for the flight crew to avoid mistakes using
these systems that can jeopardize operational safety. Most of the FG&C system modes are considered
“non-flight critical”. This means that the flight crew is assumed to recognize and safely manage any failure
of function of such modes. However, too often this assumption has proven to be unwarranted. As a result
there have been too many automation related incidents and accidents, due to stall, roll divergence after an
engine failure, icing etc. The current generation of FG&C systems do not take full advantage of modern
MIMO control and functional integration strategies provide simpler, more efficient and less costly designs.

TECS and THCS Development

FG&C system design and safety deficiencies were well recognized as long ago as the late seventies. In the
early eighties NASA initiated research to address these deficiencies. This work resulted in the Total Energy
Control System (TECS), which uses a generalized MIMO-based energy control strategy to functionally
integrate all vertical flight path and speed control modes. This approach provides inherent envelope
protection and avoids open ended SISO mode operations, thereby largely eliminating LOC safety risks.
System complexities are reduced sharply by eliminating mode overlap, simplifying mode processing and
providing more intuitive Man Machine Interfaces (MMI). Design generalization makes the system directly
reusable, thereby reducing development costs for new applications. The system was successfully
implemented and flight tested on the NASA B737 in 1985. The counterpart to TECS is the Total Heading
Control System (THCS) which integrates all lateral directional control modes. Its design objectives and
strategies are analogous to TECS. It was developed in the late eighties on the Condor High Altitude Long
Endurance autonomous UAV program. TECS and THCS were successfully applied on the Condor and
flight tested to demonstrate autonomous control capability under all operational and variety of failure
conditions.
The basic TECS concepts are described in [1, 2]. This paper describes TECS design updates since the early
nineties. A companion paper [3] describes design updates to the THCS design. It also describes a simplified
TECS/THCS-based Mode Control Panel concept and a Primary Flight Displays concept that incorporates

*
Chief Scientific and Technical Adviser for Advanced Control

1
1284
 FrAT3.1

the TECS/THCS control and guidance strategies. Another companion paper [4] provides more details on
Flight Envelope protection strategies.
Design objectives. The TECS/THCS design objectives include:
 use of one pilot-like MIMO-based control strategy for all automatic and manual control modes,
 full envelope protection to prevent LOC
 generalized functionally integrated design, consistency of operation between modes
 energy-efficient vertical flight path/speed control (elimination of stand-alone Autothrottle)
 decoupled Mode Command responses, reduced controller activity
 reduced design complexity by elimination of function overlap and using modular design
 simpler, more intuitive Mode Control Panel (MCP), clearer Flight Mode Annunciation (FMA)
 large cost reductions by generalized/reusable design, minimal application specific development,
reduction in laboratory and flight testing and shorter application development cycle.

2 TECS –Architecture and Conceptual Design

TECS Design
TECS uses a generalized MIMO based energy control strategy to provide all vertical flight path and
airspeed control mode functions. Thrust is used to control the airplane’s Total Energy requirement, the
elevator is used to distribute the Total Energy between Potential and Kinetic energy.

.
COLUMN
MODE CONTROL THROTTLE

PANEL
BACK-DRIVE

CDU INTEGRATED FLIGHT GUIDANCE FEED FORWARD

& CONTROL COMPUTER COMMANDS
PROCESSING

FMC PATH MODES

 C

FEEDBACK MIMO
THRUST
SCALING
E
E ENGINE T
C
NORMALIZATION Core Controller

IRU v C
COMMANDS

g COORDINATION
SPEED MODES
FEEDBACK
PITCH
INNER
ELEVATOR
ACTUATOR δe
ADC NORMALIZATION LOOP

Figure 2.1 TECS Architecture

A detailed discussion about the energy based control and design generalization strategy can be found in [1].
The general architecture is shown in Figure 2.1

Outerloop Modes Signal Processing. The outerloop speed mode error is first converted to a true
airspeed error and this error is multiplied by a factor KV / g to produce the non-dimensional longitudinal
acceleration signal ( Vc / g ). The outerloop path mode error (path deviation) is multiplied by a factor
K h / Vˆtrue to produce the non-dimensional flight path angle signal (  c ). Speed and vertical path energy
errors need to be weighted equally. Thus the gains KV and K h should have the same numerical value.
The V / g and  c signals are used as the standard inputs to the TECS Core Controller, see figure 2.1. The
c
Vˆtrue signal is a filtered true airspeed signal. This simple outerloop mode signal processing does not

2
1285
 FrAT3.1

include integral control signal paths to assure transient free mode switching and avoid the need for mode
integrator initialization logic.

Core Controller. In the Core Controller the associated error signals (V / g ) and (  ) are formed.
The sum of these error signals (  V / g ) represents the airplane’s specific Total Energy Rate error
 
ˆ
signal. This signal is used in an integral control signal path, together with V / g and ˆ feedbacks used in
proportional signal paths, to develop a normalized thrust command. Likewise, the difference between these
error signals (   V / g ) represents the airplane’s Differential Energy Rate error signal. In the original

concept this signal was used in an integral control signal path, together with proportional Vˆ / g and ˆ
feedback signal paths, to develop the elevator command during operations with the thrust command
between Tmin and Tmax. In the current design, explained below,   is used instead of (   V / g )
during operations with the thrust command between Tmin and Tmax. Because the error signals are used
only in integral control signal paths, the control effectors respond to a step command from any of the
outerloop modes with a control effector rate, resulting in smooth airplane dynamics.

Avoiding outerloop mode tracking errors. Since the integrators reside in the core controller, the
feedback signals ( ˆ and Vˆ / g ), used in the TECS Core Controller, must be re-referenced in the low
frequency range to the outerloop mode true airspeed and vertical path feedback signals respectively, to
avoid possible outerloop mode command tracking offset due to bias errors in these feedback signals. This
is done by using free running complementary filters, designed to take into account turbulence and
windshear effects on system performance.

Command Response Decoupling. In order to achieve decoupled outerloop command responses,

the Core Controller must be designed so that in response to a Vc / g or a c command, the
(   V / g ) and (   V / g ) quantities go to zero simultaneously with identical dynamics.

Speed Envelope Protection – automatic modes. Generally, for most automatic mode
operations, whenever the thrust command is at the upper or lower limit, a Speed-on-Elevator Control
Priority (SoECP) is used to maintain the commanded airspeed. For those cases there is no need for
separate speed envelope protection functions. However, for the Glide Slope mode a Path-on Elevator
Control Priority (PoECP) strategy is used to handle conditions with thrust command at the upper or lower
limit. This strategy assures the Glide Slope will be captured when the airplane is at the correct position to
do so, often while the thrust is at idle. Momentary open loop speed responses are protected by the Vmin
and Vmax control. Vmin/Vmax and Normal Load factor protection is also provided as an integrated part of
the augmented manual control modes and are used to protect automatic mode operations for such rare
events.

Normal Load Factor Protection - automatic modes. In the original TECS design, normal
acceleration limiting was achieved by placing a rate limiting function on the  c and Vc / g signals. These
rate limiting circuits introduce new system states which must be initialized at mode engagement and at any
time the input to the rate limiter reverses direction, in order to avoid a response delays. To avoid this extra
complexity an alternate method for normal load factor limiting, using amplitude limiting on (  ) and
(V / g ) was developed. This method is described below.

Energy Management during execution of simultaneous flight path and airspeed

commands. The rate limiters on the  c and Vc / g signals (or the amplitude limiters on (  ) and

3
1286
 FrAT3.1

(V / g ) signals in the updated design) also provide effective rate limiting on the thrust and elevator
commands. In addition, these functions provide efficient airplane energy management during execution of
large simultaneous vertical flight path and airspeed commands. If the commands have opposing energy
demands, the (  ) and (V / g ) error signal inputs to the TECS Core thrust control channel will initially
cancel, so thrust will stay constant until the elevator control has exchanged kinetic energy and potential
energy to the extend possible. This depends on the relative amplitude of the speed and flight path mode
commands. After the energy exchange is completed a change in thrust will be commanded to reduce any
remaining airplane total energy error to zero. Thus, thrust is always used efficiently.

If the commands require a substantial energy change in the same direction, the thrust command will quickly
go the upper or lower limit with double the rate limit of a single command. After the thrust command
reaches the limit, a SoECP will be used and a Control Authority Allocation (CAA) amplitude limit is
applied to the (Vc / g ) signal. This CAA-limit, defined as (Vc / g )limit  K em (  Vˆ / g ) , effectively limits
ˆ
energy rate used to execute the speed command. (The quantity (  V / g ) represents the airplane’s total
energy rate.) Therefore the remaining part of the available energy rate will by default be used to satisfy the
flight path command. For example, if during a climb at maximum thrust and therefore (  V / g )  0 a
ˆ
value of K em  .5 is selected, an accelerate command will be executed with half of the available energy
rate. The other half is then allocated to continue the climb at ~half the initial rate. After the speed command
is captured the climb rate will return to its original value, or if the altitude command is captured first, the
acceleration will increase to capture the commanded speed, see the simulation results in Figure 3.7.
ˆ
Similarly, if during idle descent when (  V / g )  0 a value of K em  1 is selected, a deceleration
command temporarily reduce the climb rate to ~zero to capture the commanded speed and then the idle
descent rate will be reestablished. This strategy facilitates the operational requirement for reducing the
airplane speed to 250 knots or less, before descending below 10,000ft. In the original design, the required
logic for this Energy Management function was complex and not without flaws. These flaws and the fixes
developed are discussed below.

TECS Performance in turbulence and windshear

Balancing the control command tracking performance and control effector activity for operation in
turbulence and windshear conditions is a difficult problem for any flight control design. Reducing control
effector activity inevitably results in deterioration of the command tracking in windshear. The performance
objective used here for command tracking in a 1 knot/sec windshear is a peak IAS-error < 5 knots and a
vertical path deviation h  20 ft. The Vˆ , ĥ , ̂ , ˆ filter gains are determined to achieve the preferred
compromise between control effector activity in turbulence and reducing induced vertical and side
acceleration and path deviation in windshear. It is beyond the scope of this paper to describe details related
to choices of controller architecture, feedback signal processing and gains.

3 TECS -Design Updates

Flight path and speed control decoupling - revisited

From the start of the TECS and THCS development it has been a design objective to avoid ad hoc design
and experimental tuning and instead use design solutions based on first principles of physics, whenever
possible. Since then a better theoretical insight has been gained into achieving improved decoupling of the
flight path response from the execution of a speed command. In the original TECS concept the differential

4
1287
 FrAT3.1

energy rate error (   V / g ) was used as the input to the elevator control channel during operation with
the thrust command between the lower and the upper limit. This conceptual architecture did not achieve the
desired decoupling of the vertical path response during execution of an airspeed command. Therefore in the
early TECS design and ad hoc method was used to improve path decoupling during speed command
execution, with limited success. Later it was realized that an acceleration at constant flight path angle does
not require an immediate change in pitch attitude, as is the case when changing flight path angle while
maintaining airspeed. In fact, a permanent change in speed at 1 g flight requires a change in Angle of
Attack and therefore an equal change in pitch attitude, which must be developed at the output of the
integrator of the elevator control channel. Also, the change in angle of attack requires a change in the
elevator trim, but this re-trim is accomplished in the updated design as part of the Short Period Model
inversion, discussed below. Using this insight, the control strategy was changed from using
(   V / g ) to using V / g as the input to the elevator control channel, during operation with the thrust
ˆ
command between Tmin to Tmax. Also, a new V / g signal with the gain Ktrim is added to the input of the
elevator control channel integrator, as shown in the revised Core Controller architecture of Figure 3.1, to
retrim the pitch attitude command during speed changes.

Airplane independent design Airplane tailored design

c +_ +
+
KTI
+ _
Engine
Engine T
S Control
.
(E T )S
 N
KTP Weight

 Specific Net Thrust

+

+
2-K Command
^.
v_
+

_
K
g Pitch Attitude Command
.
(TLIM). (SPEED PR)
KEP (typically)
(TLIM). (E D )S
^. (SPEED PR)  N
Pitch
v_c +
_
_+ KEI _
Innerloop Actuator e
+

_
g S Control
.
v^_ Ktrim
g
Figure 3.1 TECS Core Controller Architecture

Since the integral of (Vˆ / g ).K trim .KEI must equal the required change in pitch attitude, the value of Ktrim
is calculated using the 1 g relationship between Angle of Attack and true airspeed. Thus, it is found that
K trim  (1 / KEI).{W/(q.S)}/(CL .Vˆtrue ) , where W = airplane weight, q = dynamic pressure, S is the


airplane wing reference area and CL = lift coefficient change per unit angle of attack change. In addition,

the proportional signal path through the gain KEP has been revised to now use  only, at all times by
selecting K=2, see next section below. These changes do not alter the energy redistribution nature of the
elevator control, but do tend to favor suppression of path control tracking errors over speed control tracking
errors in turbulence and windshear.

TECS Core – Elevator Control Channel Design

Classical approach. The intent of the Energy Control strategy is to develop a priori coordinated thrust
and elevator control commands, in order to decouple the outerloop flight path and speed command

5
1288
 FrAT3.1

responses. If this is realized, the elevator and thrust control channels can be treated as independent
decoupled SISO components of the system. So, the elevator control channel can be designed using the
Short Period airplane pitch dynamics Model Inversion, together with the simple first order representation of
flight path angle response to pitch attitude. The thrust control channel design can also be approached as a
simple “SISO” Energy Control problem.

During the early TECS development root locus analyses was used to define the Core Controller gains and
gain schedules to support flight path and speed control modes at all flight conditions. This approach
provides little or no physical insight into the reason why and how the gains need to change. A better
approach to gain insight is to use first principles of physics, including Model Inversion, feedback
concatenation/normalization, Control Bandwidth Separation or Pole Placement. These concepts are
discussed below.

Feedback Normalization/Concatenation, Pole Placement. Consider figure 3.2, representing

a plant model of 3 chained integrators, with concatenated feedback loops/gains closed around each plant
state.

Figure 3.2 Control Loop Concatenation

The Transfer Function (TF) for this system is:

x K1 K 2 K 3
 3 (3.1)
xc S  K1S  K1K 2 S  K1K 2 K3
2

Note that the TF for a similar system with any number of concatenated system states can be written
directly. The normalized loop gains relate directly to the physical properties of the controlled system, such
as natural frequency and damping. Then, if the gain associated with each feedback loop, starting from the
innermost loop, is dropped by a factor 4 or greater, the poles of the x / xc transfer function will all fall on
negative side of the real axis in a root locus plot. Alternatively, considering (3.1), it is easy to use Pole
Placement to achieve the desired dynamics and find the required gains.

Elevator control using Short Period Model Inversion. In our approach only the airplane
rotational degrees of freedom are inverted. The Short Period Model Inversion and rebuilding of the
airplane pitch dynamics is shown figure 3.3. Note that pitch attitude (  ) can be used instead of Angle of
Attack (  ) to form the new desired pitch dynamics, as long as the frequency of the new augmented Short
Period is selected to be in the frequency range where    . Figure 3.3 does not show the pitching
moment due to thrust, but this effect is included in the full design and analyses and simulations. The Model
Inversion approach used here is not more risky that a classical design approach, since the same
conservative gain and phase margins will be maintained to provide robustness against airplane model errors
and unmodeled dynamics. If the flight test results do not match the simulation results from the
generalized control system design, there are only two possible causes: errors in the design
implementation or, insufficient fidelity of the airplane and sensor models. In that case it is more
productive to correct possible design and implementation errors and, if necessary, develop higher fidelity
airplane models, rather than revert to an ad hoc “trial and error” approach. The resulting new Short Period
dynamics are represented by the TF:
 K q K
 2 (3.2)
 c S  K q S  K q K
To provide the proper flight path angle control dynamics to support closing of the outerloop mode

6
1289
 FrAT3.1

feedback, proportional  feedback and integral control signal path of  are closed around the  / c
transfer function, as shown in Figure 3.4.

Figure 3.3 Short Period Model Inversion and build up of new Short Period
Dynamics

The  c coming from the automatic, or the augmented manual mode which develops  c by integration of
the vertical control inceptor signal ( vci ) . [An alternate architecture using (  ) in the proportional signal
path has also been developed [6], but this architecture has disadvantages for the automatic control modes.]
The controller structure of figure 3.4 was specifically developed to support the manual mode bandwidth
requirement, by using pole zero cancellation to create effectively a lower order [ /  vci ] TF, as explained
in section 4, below.

Figure 3.4 TECS Core elevator control channel

7
1290
 FrAT3.1

To make this possible the [  /  ] heave dynamic have been made to appear as an explicit part of the
[ /  c ]auto TF. By selecting K EP    the TF for the automatic modes becomes (KFFI and KFFP both set
2
to zero):
    1
   .  (3.3)
  c  auto  c  [{1/ ( K q .K .K EI )}S  {1/ ( K .K EI )}S  (1/ K EI ) S  1]( 2 S  1)
3 2

Note that the variation of  2 no longer affects the system stability. The heave time constant is defined as:
C L .Vtrue W .Vtrue
   (3.4)
2 g .C L g .CL .q .S

This  2 can be readily computed/updated by the onboard flight control computer.

Normal Load Factor limiting – Revisited. Using   instead of the (   V / g ) as the input to
the elevator control channel during operations with the thrust between the lower and upper limit also solved
a Normal Load Factor (NLF) control issue. In the earlier design, during execution of simultaneous Vc / g
and c with opposite signs, the effective Normal Load Factor limit was twice the intended value, because
both Vc / g and  c contributed to the effective NLF-limit. The new control strategy, using either   or
V / g , eliminates this problem. As mentioned above, in the current design NLF control for the automatic
modes is achieved by placing an amplitude limit on the   and V / g signals. This amplitude limit is
calculated as follows. Given a NLFlimit , the normal acceleration limit is g .NLFlimit . So the desired flight
path angle rate limit becomes   NLFlimit .g / VG . According to (3.3) for a ramping c the flight path
angle response will lag the command by an amount:
   1/ KEI   (3.5)
2

Then,      . . Therefore the amplitude limit on   should be:

  limit  (NLFlimit .g / VG )(1/ KEI   ) (3.6)
2

Also, the same limit needs to be applied to the V / g signal, to achieve the same NLF control during

operations with SoECP when the thrust command is at the upper or lower limit.
When the   or the V / g signal at its limit, the feedback path to the integrator is effectively broken, so it
must be shown that the remaining elevator/thrust control configuration maintains satisfactory  and T
response dynamics.

TECS Core - Thrust Control Channel Design

The basic control decoupling requirement is met when the responses of (   V / g ) and
(   V / g ) due to a  c or a Vc / g command are identical. The equation for Trequired is:
Trequired  W .(  V / g )  Drag (3.7)
It can be shown that during automatic mode operations the drag change due to incremental NLF (which is
limited to .1 g), can be neglected. From this equation it follows that ideally, in order to maintain
V / g  0 during the execution of a  c maneuver, the following TF identity must hold:
[(T / W ) /  c ]thrust  [ /  c ]elevator (3.8)

8
1291
 FrAT3.1

Then by inference, in order to maintain   0 during execution of an acceleration command, the

following TF identity will hold too:
[(T / W ) / (Vc / g )]  [(T / W ) /  c ] (3.9)
The above decoupling requirements can be achieved by matching thrust control channel dynamics to mimic
the elevator control channel dynamics. One approach is to select KTI  K EI and KTP  K EP , and insert
a matching filter before the final Thrust Command that represents the augmented pitch attitude and heave
dynamics, as well as the inverted engine dynamics [1 / (T / Tc )] . This approach and also another more
traditional approach, matching the TF frequency/amplitude responses, were evaluated. Both approaches
produced very good control decoupling results. However, the more traditional design resulted in a simpler
design and achieved slightly better decoupling of the flight path and airspeed command responses. It also
resulted in much lower gains KTP and KTI and therefore lower control activity in turbulence. Simplicity has
its advantages.

Provisions for Thrust and Elevator Command saturation

The classical way to limit the final thrust command is to continually calculate and apply the integrator limit
by subtracting the contribution of the proportional (  Vˆ / g ) signal path from the externally provided Net
Thrust Limit (NTL). Alternatively, the integrator may be moved to end of the net thrust command signal
processing path, where its output can be simply limited to the engine’s net thrust command limit, provided
by the FADEC. In that case, a differentiator function must be placed in the proportional (  Vˆ / g )
signal path. In the later TECS designs the latter approach is used, because it is simpler.

Likewise, similar provisions must be implemented to prevent windup of the integrator in the elevator
command processing signal path.

Priority use of Elevator when the thrust command is at a limit

The change to using   instead of (   V / g ) as the input to the TECS Core elevator control channel
during MIMO control, allowed the elevator control priority logic for conditions with thrust at the upper or
lower limit to be simplified considerably. It also allows smooth, transient-free execution of simultaneous
airspeed and flight path commands for all possible combinations of amplitude and timing to be achieved.
The updated Elevator Control Priority works as follows. When the thrust command is within the linear
control range between Tmin and Tmax, Path-on-Elevator Control Priority (PoECP) is used. PoECP also
remains in effect after the thrust command reaches Tmax or Tmin when  c  .5(ˆ  Vˆ / g ) and one of the
following modes is engaged:
 the FPA mode , or
 the Altitude Acquisition/Hold mode, or
 the Glide Slope mode, or
 the Augmented Manual control mode, and the control inceptor is at neutral
 the Augmented Manual control mode, and the control inceptor is deflected and the Vmin/Vmax
envelope protection control priority is not in effect

Using PoECP during operation in the Altitude Acquisition/Hold mode or the Glide Slope control mode
when the thrust command is at the upper or lower limit is self evident. In that case the airplane will stay on
the commanded flight path and accelerate/decelerate according to the available “excess energy rate”. This
strategy assures that the Glide Slope will be captured, when the airplane is at the right position for capture,
either from below or from above the glide slope. When the thrust command reaches Tmin or Tmax, Speed-
on-Elevator Control Priority (SoECP) is invoked in the following situations:

9
1292
 FrAT3.1

 Altitude Acquisition or FPA mode engaged and Tc  Tmax and  c  K ECP .(ˆ  Vˆ / g ) , or

ˆ
Altitude Acquisition or FPA mode engaged and Tc  Tmin and  c  K ECP .(ˆ  V / g ) , or
 Augmented Manual Control mode engaged and the control inceptor is at neutral, or
 Augmented Manual Control mode engaged and the control inceptor is not at neutral and either the
Vmin control develops a more nose down command than the Manual FPA Control mode, or the
Vmax control develops a more nose up command than the Manual FPA Control mode.
Currently, K ECP  .5 has been selected. When  c  K ECP .(ˆ  Vˆ / g ) is not true (for small values of
 c ) speed commands will be executed using PoECP, e.g. when the Altitude Hold mode is engaged, or

during a shallow climb in the FPA mode. When  c  K ECP .(ˆ  Vˆ / g ) is true, then there would be little
control authority to accelerate using the thrust only, so in that case the SoECP is invoked and part of the
energy rate used for climbing is transferred to accelerate the airplane and capture the commanded speed
quicker. Examples of Energy Management cases are shown in figure 3.7. When SoECP is invoked the  
signal input to the Core Controller is replaced with the V / g signal, see figure 3.1. In that case, to allow
for execution of a simultaneous flight path command, Control Authority Allocation (CAA) is applied to the
longitudinal acceleration command (Vc / g ) . The CCA function is explained in more detail in the next
section. Then, when the thrust command computation computes a thrust rate command that drives the thrust
command out of the limit, the control priority reverts back to flight path control priority. The thrust coming
off its limit always coincides with the start of the final flight path or speed command capture phase.

Energy Management during execution of simultaneous flight path and

airspeed commands – Revisited

In the earlier TECS design the Energy Management function and associated logic was rather complex and
not without flaws. One of the reasons was that the differential energy rate error (   V / g ) was used as
the input to the elevator control channel during operations with the thrust command between Tmin and
Tmax. When during the capture of a flight path command the thrust came off its limit, the Core elevator
control channel would revert to using (   V / g ) as its input signal. This made it difficult to smoothly
capture smooth vertical path, if at that time the V / g signal was not close to zero. The revised control
priority logic discussed above, using PoECP during operations with the thrust command between Tmin and
Tmax has made it possible to greatly simplify the Control Authority Allocation (CAA) associated signal
processing. In the current design, the Core elevator control channel only uses the V / g signal as its input
when the SoECP is invoked. Therefore, in the current design, the V / g signal that is routed the Core
ˆ
elevator control channel has the CCA amplitude limit [ K em (  V / g ) ] applied to it full time. No CCA
amplitude limit is applied the V / g signal that is routed the thrust control channel, so the thrust control
channel always uses the basic (   V / g ) signal. The CCA associated logic used in the earlier design has
been eliminated.

TECS Automatic Modes Simulation Results

Simulation. A complete TECS/THCS system simulation capability was developed in MATLAB-
Simulink. The simulation includes all TECS and THCS modes and design features discussed above in this
paper, as well as a full flight regime six degrees of freedom nonlinear airplane simulation. Realistic 2nd
order actuator models including rate and position limits were included, along with a rate limited 2nd order

10
1293
 FrAT3.1

engine model. The system time responses below were generated using this simulation. No “design tuning”
is used for any of the maneuvers shown below. The airplane model represents a generic 100-125 passenger
twin turbofan engine transport airplane at 120,000 lbs.

Results. Figure 3.5 shows the airplane response to a 25 knots step command (left plot) and a 100 knots
step command (right plot) in the IAS and Altitude Hold modes.

4 8
Legend: Legend:
3.5 Altitude Hold = 10000 ft 7 Altitude Hold = 10000 ft
IAS-cmd= 250 to 275 knots AOA ~deg delta-IAS-cmd = 100 knots
AOA ~deg
3 6

2.5 5
IAS/100~knots IAS/100~knots
IAScmd/100 ~knots IAScmd/100 ~knots
2 4
Tmax/10000 ~ lbs
Tmax/10000 ~ lbs
1.5 3

1 2
Thrust-cmd/10000 ~lbs (1 engine)
0.5 1

Thrust-cmd/10000 ~lbs (1 engine)

0 0

-0.5 -1
Altitude Deviation ~ ft Altitude Deviation ~ ft
-1 -2
10 20 30 40 50 60 70 80 90 10 20 30 40 50 60 70 80 90
Time ~seconds

Figure 3.5 TECS Responses to step IAS-cmd: 25knots (left), 100 knots (right)
The responses are very smooth and the flight path coupling error due to the execution of the speed
command is very small: the temporary altitude deviation was a little over 1 ft for the 25 knots speed
increase (within the “linear” thrust operating range) and ~2 ft altitude deviation for the 100 knots speed
change, which involved a 20 second period with the thrust command at the upper (Tmax) limit.

In Figure 3.6 the system responses are shown for the IAS and Altitude Acquisition modes for step altitude
commands of 500 ft (left plot), and 5000 ft (right plot). For both cases the responses are smooth and
without an overshoot of the command. For the 500ft step command case the thrust does not reach the limit
(Tmax). For the 5000 ft step command case the thrust reaches Tmax and stays there for ~60 seconds before
the thrust is reduced smoothly during the final linear exponential capture. There are no perceptible control
transients resulting from the reversion from PoECP to SoECP and vice versa. In both cases the maximum
speed deviation is limited to ~.2 knots. Notice also that the NLF is limited to .1 and that Tmax slowly
decreases during the climb, due to the air density effect on the engine thrust.

6 6
Alt-cmd=15000 ~ft
Alt-cmd=10500 ~ft
5 5
AOA ~deg AOA ~deg
Altitude/100 (bias= -10000) ~ ft
4 4

Altitude/1000 (bias=-10000) ~ ft
Legend:
3 3
Altitude Acq: 10000 to 10500 ft Legend:
nz ~units IAS-cmd= 250 knots nz ~units Altitude Acq = 10000 to15000 ft
2 2 IAScmd=250 knots
Tmax/10000 ~ lbs
Tmax/10000 ~ lbs

1 1
Thrust-cmd/10000 ~lbs (1 engine)
0 0
Thrust-cmd/10000 ~lbs (1 engine)
IAS-error~knots
IAS-error~knots IAScmd =250 ~knots (bias=-250) IAScmd =250 ~knots (bias=-250)
-1 -1
10 20 30 40 50 60 70 80 90 20 40 60 80 100 120 140
Time ~seconds Time ~seconds

Figure 3.6 Responses to step Altitude-cmd: 5500 ft (left), 5000ft (right)

11
1294
 FrAT3.1

In Figure 3.7 left plot, the system responses are shown for the IAS and Altitude Acquisition modes for a
step Alt-cmd = 3000 ft at t= 20 seconds and a step IAS-cmd = 100 knots at t=60 seconds.
9 9
Legend: Legend:
First Altitude Acq = 10000to13000 ft First AScmd=200 to 300 knots
8 8
Then IAScmd= 200 to 300 knots AOA ~deg Then Altitude Acq = 10000 to13000 ft
AOA ~deg
7 7
IAS/20 (bias= -200) ~knots Alt Acq cmd=10000 to 13000 ~ft
6 6
IAScmd =200 to 300 ~knots
Alt Acq cmd=10000 to 13000 ~ft
5 5
IAS/20 (bias= -200) ~knots
4 4
Altitude/1000 (bias=-10000) ~ ft Altitude/1000 (bias=-10000) ~ ft
3 3
nz ~units nz ~units
2 2
Tmax/10000 ~ lbs Tmax/10000 ~ lbs
1 1

0 0
Thrust-cmd/10000 ~lbs (1 engine) Thrust-cmd/10000 ~lbs (1 engine)

-1 IAScmd =200 to 300 ~knots -1

20 40 60 80 100 120 140 160 20 40 60 80 100 120 140 160
Time ~seconds Time ~seconds

Figure 3.7 Energy Management for Altitude Acquisition and IAS commands;
Left: Alt Acq-cmd =10k to 13k ft at 20 sec; IAS-cmd = 200 to 300 knots at 60 sec;
Right: IAS-cmd = 200 to 300 knots at 20 sec; Alt Acq-cmd =10k to 13k ft at 60 sec.

The timing of the commands are chosen to demonstrate the “pilot like” Energy Management feature built
into the system. The altitude command drives the thrust command to Tmax, establishing SoECP. Then
when the IAS-cmd follows, the Control Authority Allocation (CAA) limit, K em (ˆ  Vˆ / g ) with
K em  .5 , is placed on the Vc / g signal, causing the airplane to reduce the climb rate by ~50 % to
execute the IAS-cmd. During the execution of the IAS-cmd the c drops below .5(ˆ  Vˆ / g ) , causing
the elevator control priority to revert from SoECP to PoECP, allowing the ALT-cmd to be captured. At
this point the Total Energy demand is not yet satisfied, so the thrust stays at Tmax and the excess energy
rate is now causing the acceleration to increase. Then, when the airspeed comes within the capture range,
the thrust command drops below Tmax, PoECP is reestablished and the commanded airspeed is captured
exponentially.

In Figure 3.7 right plot, the system responses are shown for the same step Alt-cmd =3000 ft and
IAS-cmd=100 knots, but order is reversed. Now the IAS-cmd causes the thrust to ramp up to Tmax, to
accelerate the airplane while maintaining PoECP, until at t=40 seconds the step Alt-cmd results in
 c  .5(ˆ  Vˆ / g ) , causing a reversal of the elevator control priority to SoECP with the CCA
acceleration limit (Vc / g )limit  .5(ˆ  Vˆ / g ) applied to the Vc / g signal that is routed to the elevator
ˆ
control channel. This causes the acceleration to drop to Vc / g  .5(ˆ  V / g ) , thereby transferring ~50 %
of the energy rate to the execution of the climb command, while maintaining Tmax. Next, the IAS-cmd is
ˆ
captured first, but since at this point still  c  .5(ˆ  V / g ) the SoECP is maintained. During the IAS-
cmd capture, the excess energy is transferred to increase the climb rate, while maintaining T= Tmax . When
the thrust command drops below Tmax , the “linear” PoECP capture of the Alt-cmd begins.

4 Flight Path Angle based Augmented Manual Control

Specific design Objectives

12
1295
 FrAT3.1

The FBW Augmented Manual Control mode provides pilot maneuvering capability in the vertical plane by
using the vertical control inceptors. A earlier stand-alone Flight Path Angle based augmented manual
control mode was developed and flight demonstrated/evaluated under the NASA TCV Program 1976-79
[3]. Key design objectives include:
 good handling qualities at all flight conditions, with precision maneuvering capability
 reduced pilot workload using “Direct FPA Control” strategy , eliminating need for continuous
compensatory pitch attitude control.
 carefree maneuvering capability to the limits of the safe airplane performance, using envelope
protection to reduce the risk of Loss of Control (LOC)
 operational consistency with automatic modes, one pilot “mental model” for all operations
 clean and simple integration of the manual and automatic modes, shared function elements
In order to achieve the last three objectives the Augmented Manual control mode is designed as a simple
augmentation to the automatic FPA control mode. Thus, the Core elevator control channel of figure 3.4
provides the basic airplane control when the pilot is not using the control inceptor. The airplane then
maintains the last pilot-established earth referenced  , regardless of changes in airspeed, airplane
configuration changes or disturbances due to turbulence and windshear. As a result this FPA Rate
Command Hold strategy largely eliminates the need for the pilot to use a Continuous Compensatory
Control Strategy. Instead the control tracking performance tends to improve when the pilot adopts an
Intermittent Maneuver Control Strategy. So, the main reasons for going to a “Direct FPA Control”
strategy are to reduce the tedious kind of workload controlling flight path perturbations, and to simplify
interception and tracking a vertical path in space. This operation can be further enhanced by proper flight
displays, e.g. a HUD or Synthetic Vision background display of the airport and runway. This makes it easy
to capture a desired Glide Path and from there on, the FPA-based Augmented manual control algorithm
will track the pilot established flight path with little or no need for pilot corrections. This capability, then
called “Velocity Vector Control”, was first developed and demonstrated by NASA under the TCV program
in the late 1980-ties [5].

Design Implementation
For the design shown in figure 3.4 the control inceptor command signal is processed in three very basic
feed forward command paths. The first signal path integrates the inceptor command to establish the
reference  c . The second and third signal paths shape the control responses of the airplane to achieve the
exact response dynamics prescribed by a specified ideal (classical) handling qualities model. Briefly, the
TECS Core elevator control channel [ /  c ]auto TF, equation (3.3), has a unity numerator and a fourth
order denominator. The feedback and feed forward gains of this Core Controller can be selected such that
the resulting augmented manual [ /  vci ] TF results in a [ /  vci ] TF that represents a specified ideal
handling qualities model, for example:
 K g (  S  1)SP
 vci 2
(4.1)
 vci S VG S 2  2 SPSP S  SP
2

Here  vci is the vertical control inceptor deflection, K vci is the vertical control inceptor gain. Since
    1
  (4.2)
 vci  vci   vci (  S  1) 2

it follows in order to achieve (4.1) , the final [ /  vci ] TF must be

 K g 2
 vci (4.3)
 vci S VG S 2  2 S   2

13
1296
 FrAT3.1

Here VG is the groundspeed. This [ /  vci ] TF can be realized by using feed the forward gains K FFP
and K FFI to create two zeros designed to cancel two poles of the [ /  c ]auto TF, equation (3.3). One of
these numerator zeros is used to cancel the  2
associated pole and the second zero is used to cancel the
first order pole that is part of the third order part of the denominator of the [ /  c ]auto TF. Thus the
“ideal” SP frequency and damping coefficient in (4.1) can be specified. For example: selecting
  SP  2 rad/sec and    SP  1 results in :
 K g ( K FFP .S 2  K FFI .S  1)
 vci (4.4)
 vci S VG (.25S 2  1S  1)( D S  1)(  S  1) 2

The second order numerator of (4.4) must cancel the two first order poles. Therefore:
( K FFP .S 2  K FFI .S  1)  ( D S  1)(  .S  1) (4.5)
2

However, to determine K FFP and K FFI ,  D must be known, or one of the feedback gains must be known.
The simplest way is to select D . Then from (4.5) it follows that K FFP   D .  and K FFI   D    .
2 2

For example for D  1 , it follows that K FFP    and K FFI  1   . Also the following identity
2 2
must hold:
(.25S 2  1S  1)( D S  1)  [{1/ ( K q K K EI )}S 3  {1/ ( K K EI )}S 2  {1/ ( K EI )}S  1] (4.6)
[ /  c ]auto TF, Equation (3.3).
The right hand part of equation (4.6) is the third order part of the original
For D  1 the gains become K q  5 (rad/sec )/(rad/sec), K  1.6 (rad/sec)/rad and K EI  .5
2

rad/rad. The gain K EI  .5 adequately supports the outerloop altitude and airspeed modes bandwidth of .1
rad/sec. The  response lag relative to  c becomes    1 second.
The above sketched approach for designing the [ /  c ]auto and [ /  vci ] TF allows a quick evaluation
of the change in the [ /  c ]auto dynamics and the gains, in particular K q and K EI , for other choices of
 D . Here it was assumed that the linear elevator control actuator transfer function will have its lowest first
order pole located at -20 rad/sec or higher. This allows for a gain K q up to 5(rad/sec2)/(rad/sec) or
somewhat higher, while still assuring that the lowest frequency pole of the actuator dynamics will not
couple with the first order pole associated with the K q control loop, to form a lowly damped oscillatory
mode. The selected gains also allows for the addition of structural mode filters, if needed. The robustness
margins can be increased further by increasing  D which lowers K q , but reduces K EI . If the “ideal
response model” is different than the one defined by equation (4.1), or if it needs to change for different
flight conditions, it is a simple matter to recalculate the corresponding gains. More details on this FPA
based Augmented Manual control mode design can be found in [3,4].

Augmented Manual Mode – Thrust Control

No changes to the basic thrust control channel are required for the Augmented manual mode, except the
feed forward gains KTFFP and KTFFI (implemented analogous to K FFP and K FFI ) can be used to
minimize speed deviations due to vertical maneuvering. However, K FFP  0 was found to relax throttle
response during vertical stick inputs, albeit at the expense of incurring a slightly larger speed error.

14
1297
 FrAT3.1

Pilot Display Requirements for manual  -control loop closure

The FPA-based augmented manual control mode was designed to meet the classical handling qualities
requirements intended for pitch attitude control, so a standard Primary Flight Display can be used for
closing the pilot control loop using pitch attitude. However, to realize “direct FPA Control” requires the
addition of the FPA information to the PFD. In the earlier NASA “Velocity Vector Control” development
program [5] it was found that    1 second, although less than for the unaugmented airplane, it is still too
large for the pilot to be able to close the loop on  directly. Therefore a quickened  display signal must
be used. The obvious candidate signal is  c . Pilot control loop closure around  c , instead of around the
actual airplane dynamics, was found to work very well, since only 90 degrees lag is incurred in this loop,
discounting the pilot’s lag. So then the pilot can use very high gain without PIO risk. To avoid displaying
both  and  c , a blended  quick signal can be used that responds like  c during maneuvering and
reverts to  when the pilot is out of the control loop. It is based on equation (4.3):
(1/  2 ) S 2  (2 /  ) S
 quick1    c (4.7)
(1/  2 ) S 2  (2 /  ) S  1
or, by defining  q  (2 /  ) a first order approximation of (4.11) becomes:
 qS
 quick 2    c (4.8)
 qS 1
Still another approach to “on demand”  quickening was proposed in [6]. It adds a pitch rate signal to  .

Augmented Manual Mode – Envelope Protection

Speed Envelope Protection. The FPA-based augmented manual control mode should normally be
operated with the autothrust engaged because of the lack of speed stability at constant throttle setting.
However the airspeed should be allowed to drift after the thrust reaches the upper or lower limit and the
pilot commands a  in excess of the airplane’s steady state performance capability, or during maneuvering
with the autothrust disengaged, as long as Vmin and Vmax protection is provided when the airplane’s
excess kinetic energy runs out. Therefore simple independent Vmin, and Vmax control functions have been
developed that work as follows. When the autothrust is engaged, the Vmin control function is armed to
allow engagement using SoECP after the thrust-command reaches Tmax and after the Vmin control
develops a pitch command that is more nose down than the pitch command developed by the manual FPA
control. Likewise, the Vmax control function is armed to allow engagement using SoECP after the thrust-
command reaches Tmin and after the Vmax control develops a pitch command that is more nose up than
the manual FPA control. Also, the Vmin target is lowered in proportion to the nose up  vci deflection,
from the command speed at zero  vci deflection to 1.05 Vstall for full nose up deflection. Likewise, the

Vmax target is increased in proportion to the nose down  vci deflection, from the command speed at zero
 vci deflection to Vmo/Mmo + XX knots for full nose down deflection. When the autothrust is
disengaged, the Vmin/Vmax envelope protection function is always armed to engage. In this case the
Vmin target is 1.2 Vstall at zero  vci deflection. The Vmin target is lowered in proportion to the nose up

 vci deflection to 1.05 Vstall for full nose up deflection. Here Vstall  Vstall1g 1/ cos  is the stall speed
for the airplane in a level coordinated banked turn. Likewise, the Vmax target is increased in proportion to
the nose down  vci deflection, from Vmo/Mmo at zero  vci deflection to Vmo/Mmo + XX knots for a
full nose down deflection.

15
1298
 FrAT3.1

Normal Load Factor control. To prevent excessive positive nz that can result in stall or undesirable
negative nz , both a bank angle command limit and a full  vci deflection nz - command limit are imposed.
The vertical maneuver authority, ( nzv ) max , is calculated according to equation (4.9) below, and the NLF-
command is scheduled so that a full vertical control inceptor deflection always commands the maximum
safe NLF. However, it was found that simple command limiting cannot prevent exceeding the NLF limits
for extreme stop to stop vertical inceptor deflections at high speed (e.g. due to PIO). Therefore a simple
innerloop feedback NLF-limit control function was also implemented. A more detailed discussion on
Envelope Protection requirements and design for automatic and augmented manual control mode can be
found in the companion paper [4].

Scheduling of the Command Gain, K vci

To minimize the possibility of overstressing or stalling the airplane, the vertical control inceptor command
gain K vci needs to be scheduled as a function of stick deflection and airspeed, such that full vertical stick
deflection commands the maximum safe Normal Load Factor ( nz ), at any speed. At speeds greater than
the Maneuver Speed nzauthority  nzstructLim . Generally at design weight, nzstructLim  2.5 . Below the
Maneuver Speed nz  nzaeroLim  V / Vstall1g . The vertical control inceptor deflection (  vci ) is
2 2
authority

normalized to +1 for full nose up deflection and -1 for full nose down deflection. For  vci  1 The
maximum available normal load factor for vertical maneuvering, (nzv ) max , is:
(nzv ) max  nzauthority  nz  nzstallMargin (4.9)

In equation (4.10) nzauthority is the lower of nzstructLim or the nzaeroLim ; nz  (1/ cos  )  1 is the
incremental load factor due to roll angle, assuming a coordinated turn; nzstallMargin is a selected safety
margin, typically equal to .1. For this study it was decided that the negative nz control authority should be
limited to ( nzv ) min  0 , rather than (nzv ) min  1 , because a capability to command (nzv ) min  0 in a
vertical maneuver gives plenty maneuver authority for a transport airplane and protects the passengers and
the airplane against possible injuries and damage. (An arrangement should be provided to change this limit
to ( nzv ) min  1 in case the airplane becomes inverted, since the airplane must remain controllable at any
attitude.) Thus, with the nzc defined at three points for  vci =1, 0, -1, the above requirements can be met
by defining the incremental normal load factor ( nzcv ) , commanded as a function of (nzv ) max and  vci ,
using the following parabolic schedule:
nzcv  [{.5( nzv ) max  1}. vci  .5( nzc ) max ]. vci  K vci . vci (4.10)
Therefore:
c  nzcv .g / VG .  K vci . vci .g / VG (4.11)
Here nzcv is the incremental load factor commanded by the vertical control inceptor deflection. It should
be noted that for this schedule the command gradient, (nzcv /  vci )  .5(nzv ) max , is a function of
vci 0
airspeed and the inceptor deflection. When combined with a passive inceptor that has a fixed force
gradient, it produces a “stick force per g” ( Fvci / nzcv ) that at high speeds decreases with increasing
deflection and this is generally regarded as unacceptable for handling qualities. Another stick command
schedule that allows for a selectable command gradient, nz /  vci ) 0  grad , can be defined using
cv vci

16
1299
 FrAT3.1

a (1-cos) function and the requirements: for  vci  1 nzcv  (nzv )max  1 ; for  vci  0 nzcv  0 ;
and for  vci  1 nzcv  1 . Then for  vci  0 the relationship becomes:
nzcv   vci .grad  {(nzv ) max  1  grad }[1  cos{ vci .( / 2)}] (4.12)
and for  vci  0 the relationship becomes:
nzcv   vci .grad  (1  grad ).[1  cos{ vci .( / 2)}] (4.13)
Although for this command the gradient at zero control effector deflection can be selected, at high speed
the Fvci / nzcv gradient still decreases with increasing deflection. So this schedule may also be unsuitable
for use with a passive control inceptor that has a fixed force/deflection gradient. Still another alternative is
to use a constant command gradient, ( nzcv /  vci ) , at any defection and airspeed. This approach would
produce a constant Fvci / nzcv , when using a passive constant force gradient control inceptor, but it also
has a number of design and handling qualities issues. These include: matching full deflection command
with maximum maneuver authority; unequal maximum positive and negative deflection; command gradient
discontinuity around zero inceptor deflection and possibly for large nose up deflection; and possible need
for flat zones where nzcv /  vci  0 , used to prevent exceeding (nzv ) max . For these reasons, the FAA
is currently sponsoring research to define design guidelines and certification requirements for passive and
active control inceptor command gain and feel force gradient.

Control Anticipation Parameter (CAP) Requirement

The CAP is defined as the ratio 
θ t 0 / Δn z . From the [  /  vci ] TF defined by (4.5), it follows that:
t 
ct 0   vci .K vci .( g / VG ).K FFP .K EI .K .K q (4.15)
For a constant stick input the final incremental load factor is:
(nzc )t   (VG / g )c  (VG / g ). vci .K vci .( g / VG ) (4.16)
Therefore CAP  ( g / VG ).K FFP .K EI .K .K q (4.17)
Thus, for an approach condition with V  215 ft/sec and    2 and a selected    1 , it follows that
2
the required value K FFP    2 , resulting in CAP = 1.2. This is well within the .28 to 3.6 CAP range
2
allowed in MIL-STD-1757A for level 1 handling qualities. With the feedback gains K EI , K and Kq
pre-determined, the only parameter that can change CAP is K FFP . However, K FFP is a critical design
parameter that cannot be varied much from the calculated value in the above analysis and still achieve
acceptable responses. Furthermore, to achieve harmony between the initial pitch acceleration and the final
nz response K FFI (here controlling the steady state nz lag relative to  vci .K vci / S must be selected
within a narrow range, to achieve the desired value of   . Another way to analyze CAP is to look the
variation of CAP as a function of D . Changing D has no effect on the final [ /  vci ] TF , nor does it
change the product K EI .K .K q , but
 K FFP varies in proportion to D , so it is possible to change CAP
value without changing the [ /  vci ] response! One more observation: for higher values of  D the basic
[ /  c ]auto TF incurs more lag, because it reduces K EI (see equation3.5), so in order to still achieve the
same final  /  vci response, the feed forward gains and K FFI increase to compensate for the increased
lag. So then the control augmentation relies more on the K FFP direct feed through signal path to the

17
1300
 FrAT3.1

elevator and less on the feedback control signal paths. Conclusion: CAP is a dubious Handling Quality
criterion that may need further updates or replacement.

FPA based Augmented Manual Control Simulation Results

The same simulation as used above for the automatic modes was used to generate the time responses below
for the FPA-based Augmented Manual control mode as defined above. Figure 4.3 (left plot) shows the
responses to  stick =.1, starting at t=20 second for a duration of 5 seconds, resulting in a change of
pitch

flight path angle of ~2 degrees. The responses are very smooth, without an overshoot of the  c , or
oscillations. Note the responses in Angle of Attack (AOA) and NLF ( nz ) are also controlled very
smoothly.

5 8
AOA ~deg Legend: Pitch Att ~deg Thrust-cmd=Tmax:
Augmented Manual Mode Transition to Speed control Priority
7
IAScmd=250 knots
4
6
AOA ~deg Flight Path Angle-Cmd ~deg
Legend:
5 Augmented Manual Mode
3
IAScmd=250 knots
Pitch Att ~deg 4

Flight Path Angle ~deg

2 3
Flight Path Angle-Cmd ~deg
nz ~units nz ~units
Tmax/10000 ~ lbs Tmax/10000 ~ lbs
2

1
Flight Path Angle ~deg 1
Thrust-cmd/10000 ~lbs (1 engine)
Thrust-cmd/10000 ~lbs (1 engine) 0
0
IAS-error ~knots
-1
Stick (pitch) ~units
IAS-error ~knots Stick (pitch) ~units
-1 -2
10 15 20 25 30 35 40 45 50 10 20 30 40 50 60 70 80 90 100
Time ~seconds Time ~seconds

Figure 4.3 FPA-based Augmented Manual Mode: responses to  vci =.1

The flight path angle response lag is as designed:    1 second. The pitch attitude shown is biased by the
amount of the trim pitch attitude before the maneuver starts, to show its lead relative to lead  c .
The pitch attitude leads the  -response by 2 and causes the attitude to drop back when the control
inceptor is released, if    1 . The IAS-error remains less than 1knot. The thrust command does not reach
2

Tmax. Figure 4.3 (right plot) shows the responses to  vci =.1, starting at t=20 second for a duration of 16
seconds, resulting in a change of flight path angle of ~6.6 degrees. During this maneuver the thrust
command increase to within a small margin of Tmax. The IAS-error reaches a maximum of ~1.5 knot and
then gradually reduces to ~zero. Then, as Tmax decreases with increasing altitude, the thrust command
reaches Tmax at t=87 seconds and this causes a reversion from PoECP to SoECP, in order to maintain the
airspeed. At that point the flight path angle will start to fall off in proportion to the thrust fall off.

Figure 4.4 (left plot) shows the responses to  vci =.1, starting at t=20 second held indefinitely. This causes
the flight path angle to rise until the Vmin control engages (using SoECP), a short time after the thrust
command reaches Tmax. At that point the flight path angle and reaches ~ 9 degrees, then starts to decline
as a result of the Vmin control priority. As discussed above, the Vmin control has been designed to mimic
speed stability, allowing a final speed deviation in proportion to the  vci deflection. So in the left plot, the
speed bleeds of 10 % of the speed margin to 1.05Vstall. In the Figure 4.4 (right plot) a full nose up
deflection is applied (  vci =1). This results in a very aggressive maneuver using all available NLF

18
1301
 FrAT3.1

authority. For this case, the thrust command very quickly reaches Tmax and Vmin control priority is
established very shortly before the NLF and the AOA reach their peak value, at ~2.2 and 13 degrees
respectively. The stall AOA is 15 degrees. The flight path angle reaches a peak value of ~22 degree. The
real peak pitch attitude reached is ~38 degrees (34 degrees as shown + 4 degrees trim value before the start
of the maneuver). The final speed settles at 1.05Vstall. The control responses and reversion to Vmin
control are very smooth and without transients.

10 35
Pitch Att ~deg Legend: Legend:
Pitch Att ~deg
Augmented Manual Mode Augmented Manual Mode
8 IAScmd=250 knots IAScmd=250 knots
30
Full NU stick at t=20 sec - held
AOA ~deg Flight Path Angle ~deg Tmax/1000 ~ lbs indefinitely
6
25
Flight Path Angle-Cmd ~deg Transition to Vmin Control Priority
(start of FPA-cmd synchronizatuin to FPA
4 and introduction speed stability)
20
T-cmd=Tmax
nz ~units
2 Flight Path Angle ~deg
15
Thrust-cmd/1000 ~lbs (1 engine)
0
Thrust-cmd/10000 ~lbs (1 engine) nz ~units
10
Tmax/10000 ~ lbs
-2 Transition to Vmin Control Priority Stick (pitch)*10 ~units
introducing speed stability held Indefinitely
Stick (pitch) =.1 ~units 5
-4 AOA ~deg
held Indefinitely
IAS-deviation ~knots
Flight Path Angle-Cmd ~deg
0
-6
Vmin=V-cmd-dStick*(V-cmd-1.05Vstall)
IAS-deviation/10 ~knots
Vmin=1.05Vstall
-8 -5
For dStick (pitch)=1 ~units

-10 -10
10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100
Time ~seconds Time ~seconds

Figure 4.4 FPA-based Augmented Manual Mode, left: responses to  vci =.1 held
indefinitely; right:  vci =1, held indefinitely

7 Additional TECS & THCS related developments

Ecological PFD. The TECS energy based control strategy and the THCS Heading control strategy can
be embedded into the Primary flight Display (PFD), to bring out control guidance cues for using manual
Thrust, Pitch and Roll control to efficiently and simultaneously capture and track airspeed, Altitude and
heading targets in an energy efficient exponential and overshoot-free manner. This enhanced Ecological
PFD concept is summarized in the companion paper [3] and described in more detail in [8].

TECS/THCS Mode Control Panel. In order for the pilot to be able to use one mental model for all
FG&C operations and to minimize effort needed for reuse, the TECS & THCS designs use one generalized
guidance and control strategy for all automatic and augmented manual control modes. For the same reason
this strategy has also been applied to the design of the FG&C Mode Control Panel (MCP) and Flight Mode
Annunciation (FMA) function on the FFD. These developments are described in the companion paper [3].

Interactive Real-Time TECS/THCS Demonstration System. An interactive Real-Time

TECS/THCS Demonstration System was develop, using The Simulink Real Time Workshop program,
including the twin engine transport airplane simulation, an interactive TECS/THCS Mode Control Panel
with integrated Controller Pilot Data link Communication functions, a joystick manual control capability
and several versions of Primary Flight Displays including a Flight Mode Annunciation Panel.

8 Conclusion
This paper describes recent design enhancements of the Total Energy Control System (TECS). TECS uses
a “pilot like” MIMO energy-based guidance and control strategy to generalize and functionally integrate all

19
1302
 FrAT3.1

automatic and augmented manual control modes for airplane control in the vertical plane. This design
strategy enables the pilot to use one mental model for all FG&C operations. It also minimizes the effort
needed for design application on various airplane programs. To limit the scope of this paper, additional
related subject matter is covered in two companion papers [3] and [4]. Companion paper [3] covers design
updates to the Total Heading Control System (THCS), the enhanced ecological PFD [also called Energy
Management PFD (EMPFD)], and the design of the Mode Control Panel and the Flight Mode Annunciation
Panel. THCS uses strategies analogous to TECS, to generalize and functionally integrate all automatic and
augmented manual lateral-direction control modes. TECS and THCS provide full 6 degrees of freedom
airplane control capability to the limits of the safe flight envelope, without allowing LOC by departure
outside the safe flight envelope. It eliminates stand alone SISO based Autothrottle, Yaw Damper/ Turn
Coordination and Trust Asymmetry Compensation systems. The EMPFD incorporates the TECS and THC
guidance strategies to enhance pilot awareness of airplane maneuver capabilities and provide guidance cues
to use the controls in an effective and energy efficient manner. A second companion paper [4] discusses
various options for designing flight envelope protection functions.

9 References

1. A.A. Lambregts: “Vertical Flight Path and speed Control Autopilot design Using Total Energy
principles”, AIAA 83-2239CP

2. A.A. Lambregts: “Automatic Flight Controls Concepts and methods”, Koninklijke Nederlandse
Vereniging voor Luchtvaart, Jaarverslag, 1996

3. A.A. Lambregts: “THCS Generalized Airplane Control System Design”, 2013 CEAS conference
on Guidance, Navigation and Control, Delft, The Netherlands

4. A.A. Lambregts: “Flight Envelope Protection Strategies for Automatic and Augmented Manual
Control”, 2013 CEAS conference on Guidance, Navigation and Control

5. [Link] and [Link]: “Development of a Control Wheel Steering Mode and Suitable
Displays that Reduce Pilot Workload and Improve Efficiency and Safety of Operation in the
Terminal Area and in Windshear”, AIAA G&C paper 79-1887

6. A.A. Lambregts: “Fundamentals of Fly-By-Wire Augmented Manual Control”, SAE 05WAC-62

7. D. Niedermeier, A.A. Lambregts: “Design of an Intuitive Flight Control System”, 2009 CEAS
Conference

8. R. Bray: “A Head-Up Display Format for Transport Aircraft

Approach and Landing”, NASA TM-81199, July 1980; NASA HUD Report 11; N80-29295

9. A.A. Lambregts, R. Rademaker, E. Theunissen: “A New Ecological Primary Flight Display

Concept”, DASC 2008

20
1303