Document [original]

This version is available at https://doi.org/10.14279/depositonce-7044.

right to use is granted. This document is intended solely for

personal, non-commercial use.

Müller, S.; Schreiter, K.; Luckner, R.; Manzey, D. (2017): Manual Flying and Energy Awareness. Aviation

Psychology and Applied Human Factors, 7(1), 18–27. https://doi.org/10.1027/2192-0923/a000111.

This version of the article may not completely replicate the final version published in Aviation Psychology

and Applied Human Factors. It is not the version of record and is therefore not suitable for citation.

Müller, Simon; Schreiter, Karolin; Luckner, Robert; Manzey, Dietrich

Manual flyin

and ener

y awareness

Beneficial effects of energy displays combined with a new approach of augmented

thrust control

Accepted manuscript (Postprint)Dokumententyp |

Manual Flying and Energy Awareness

Beneଏcial Effects of Energy Displays Combined With a New

Approach of Augmented Thrust Control

Simon Müller,

Karolin Schreiter,

Robert Luckner,

and Dietrich Manzey

Manually flying an aircraft can be understood as a complex

task of managing physical energies (e.g., Filippone, 2012).

In order to accomplish certain objectives like climbing,

descending, or level flight, the pilot must take care to

provide the aircraft with proper total energy in terms of

thrust or drag. Then, the available energy must be

distributed across potential energy, representing the altitude

of the aircraft, and kinetic energy, representing its speed.

This is usually achieved by elevator deflections via control

inputs, that is, yoke or sidestick. However, finding

appropriate thrust settings and maintaining a proper

awareness of the energies needed for a target flight state

are not always an easy task.

The complexity of energy management in manual flight

can best be illustrated by the control of potential and kinetic

energy during approach and landing. The main goal during

this flight phase is to fully reduce the total energy of the

aircraft until it eventually touches down and stops on the

runway. Several ways to reduce energy are available and

may be used for this purpose, including lowering the thrust

setting and/or inducing additional drag by extending speed

brakes, flaps, or landing gears. However, during approach

and landing this task is complicated by the fact that the full

reduction of energy needs to be accomplished until

touchdown at a certain range on a runway. At the same

time, steady flight states are repeatedly required throughout

the approach. To ensure this, pilots need to continuously

gauge both the thrust as well as the aircraft’s pitch angle

by commanding elevator deflections to control the relative

reduction of potential and kinetic energy in an appropriate

way. However, the specific settings needed in a given flight

situation directly depend on factors such as the actual

aircraft mass, configuration of the high lift system, airspeed,

and altitude. Thus, the proper thrust and pitch angle

settings required for target flight states can change

considerably during the flight. Because no pilot can ever

memorize all proper pitch angle and thrust settings for a

given aircraft and all possible constellations of flight

parameters, pilots usually base their control inputs on a

subset of crucial combinations that they know.

Subsequently, they make adjustments to obtain the proper

settings based on a trial-and-error principle. This approach

is called pitch-and-power flying and belongs to the basic

flying skills of pilots. It generally works well in routine

situations of manual flight. But, it can become extremely

cognitively demanding, for example, in situations

requesting complex changes of energy, in phases of high

workload, and in safety-critical conditions.

Department of Work, Engineering and Organizational Psychology, Technische Universität Berlin, Berlin, Germany

Department of Flight Mechanics, Flight Control and Aeroelasticity, Technische Universität Berlin, Berlin, Germany

Using pitch and power as cue values to control the flight

state has evolved historically. But this lacks precise infor-

mation of the physical energies and does not seem to

ensure pilots have sufficient energy awareness, which can

even compromise flight safety. This is suggested by several

recent incidents and accidents in civil aviation showing that

even highly trained pilots can lack essential energy

awareness in energy-critical phases of a flight (Dutch Safety

Board, 2010; National Transportation Safety Board, 2013).

It often results from insufficient monitoring and considera-

tion of all crucial flight parameters in high-workload phases

of a flight, but can also be due to an inappropriate mental

processing and understanding of the relevant flight

parameters with respect to the current flight state of the

plane. One way out of this problem is to keep the pilot

out of the loop of direct control and let the aircraft always

fly and land automatically. However, this is in direct

contrast to the current request posed by the Federal

Aviation Administration (2013;2016), to even increase

the manual flying in commercial aviation in order to avoid

issues of de-skilling of pilots. Moreover, several of the most

recent issues of energy awareness and proper energy

management occurred when pilots were forced to take over

control from the autopilot, that is, in situations where the

automation failed for some reason.

Thus, instead of only relying on cockpit automation,

measures are needed that better support pilots in finding

proper thrust settings and unload them from the typical

demands of pitch-and-power flying when flying manually.

This is the main objective of a new system referred to as

nxControl that has been proposed recently based on a

new concept of thrust control and the provision of

augmented energy information to the pilot (Müller,

Schreiter, Manzey, & Luckner, 2016). Specifically, it con-

sists of two components. The first component includes a

primary flight display (PFD) supplemented by additional

energy indicators that inform the pilot about the change

of the aircraft’s energy in a direct and salient way. The sec-

ond one includes an assistive demand controller that

transfers the concepts of fly-by-wire flight control laws for

an aircraft’s attitude also to the control of thrust. However,

the bandwidth of thrust control is an order of magnitude

lower than for attitude control.

The idea of augmented cues providing information about

the current energy state of an aircraft is not new. In fact,

various ideas and possible implementations of what is

referred to as energy displays have already been proposed

(e.g., Amelink, Mulder, van Paassen, & Flach, 2005;

Catton, Starr, Noyes, Fisher, & Tim, 2007; Lambregts,

Rademarker, & Theunissen, 2008). The common element

of most of these concepts includes an augmented presenta-

tion of important energy status information as integral part

of the PFD, mostly represented by indicators of the total

energy angle (TEA, also known as potential flight path

angle) and the flight path angle (FPA; Amelink et al.,

2005; Lambregts et al., 2008). The TEA provides informa-

tion about the current rate of change of total energy.

The FPA provides information about the current rate of

change in potential energy. The implementation of these

indicators, in the format chosen for the nxControl system,

is shown in Figure 1based on the example of a current

Airbus A320 PFD (Müller, Manzey, Schreiter, & Luckner,

2015). The TEA is represented by a horizontal line; the

FPA is shown as a circle with a dot in its center. According

to Airbus color standards green color was chosen for both

symbols. The main advantage of this display is that the

dynamic spatial interaction of both the TEA and FPA pro-

vides a direct and integrated picture of the control inputs’

consequences on kinetic, potential, and total energy.

At the same time, it informs whether the aircraft acceler-

ates or decelerates and climbs or descends at a glance in

the following way (see Figure 2): While the relative position

of the FPA in relation to the artificial horizon line directly

indicates if the aircraft is descending or climbing, the rela-

tive position of the TEA and FPA, that is, whether the TEA

is above or below the FPA, shows if it is gaining or losing

speed. If both symbols converge, it indicates that the air-

craft is moving with constant speed. Thus, the specific spa-

tial relation of the TEA, FPA, and artificial horizon line

always provides information about the current energy state

in what has been referred to as an emergent feature (Bennet

& Flach, 2011). The use of the emergent feature in display

design has been proposed in particular to reduce the

Figure 1. Energy-enhanced primary flight display with FPA and TEA in

the center of attitude indicator. FPA = flight path angle; TEA = total

energy angle.

attentional demands when processing complex information

derived from different parameters (Wickens, Hollands,

Banbury, & Parasuraman, 2013). Integrating this informa-

tion in the PFD also serves the so-called proximity compat-

ibility principle introduced by Wickens and Carswell (1995).

That is, all primary flight parameters needed to aviate

safely are presented along with information about the

relative distribution of kinetic and potential energy in close

spatial proximity.

It is expected that providing such augmented energy

status information results in maintaining better energy

awareness during manual flying, even when overall

workload is high and flight patterns require complex energy

adjustments. The first evidence for these proposed

beneficial effects of energy status displays is provided by

simulator studies suggesting that providing energy status

information indeed improves the energy awareness of

pilots and might reduce the cognitive burden of classical

pitch-and-power flying (Catton et al., 2007; Müller et al.,

2015; van den Hoven, de Jong, Borst, Mulder, &

van Paassen, 2010).

The second element of the nxControl system consists of

an assistive flight controller for manual control of thrust

and speed brakes (Müller et al., 2016). This controller,

called nxController, directly uses the TEA as a com-

mand value to control the rate of change in total energy.

For a given TEA setting, the controller executes the

designated change in total energy by adjusting the needed

thrust and speed brakes settings. At the same time, the

controller compensates for the changes in thrust or drag

induced by the aircraft configuration changes or

atmospheric disturbances. In this way, a control concept

is realized that directly corresponds to the format of

augmented energy information in the energy displays

referred to earlier. Beyond that, the general approach of a

manual control augmentation is also compatible with the

flight control laws implemented for sidesticks and pedals

in current aircraft. Inputs to the controller are provided

by a new nxLever, which replaces the conventional thrust

lever. Combined with the new controller, a new nxStatus

display is added to the cockpit’s engine instrumentation,

which provides feedback about the control inputs together

with information about the current performance limitations

of the aircraft. A detailed description of the nxController,

the nxLever, and the nxStatus display, which is beyond the

scope of this paper, can be found in the publication by

Müller et al. (2016).

Recently, the first prototype of the complete nxControl

system was developed and implemented in a fixed-base

research simulator. In the first suitability study, 11 pilots

had to perform four basic air work flight tasks with the

new system. The results provided evidence of its feasibility.

Specifically, it was shown that the participating pilots were

able to perform the air work tasks with sufficient precision

and less workload in terms of thrust lever movements after

only a short familiarization with the new system.

The current study directly capitalizes on this research

and intends to evaluate the consequences of human

performance with the new system in a complex and

demanding flight scenario. Experienced airline pilots with

Airbus-type ratings had to manually fly a complex approach

pattern with high demands on energy management in a

fixed-base flight simulator. The pilots’performance,

workload, and situation awareness (SA) during the

approach were compared in conditions where pilots were

and were not supported by different components of the

new system. In order to separate possible effects of the

energy display and the new controller, three different

conditions were compared, including manual raw-data

flying, manual flying with the enriched energyPFD, and

manual flying with the complete nxControl system.

We hypothesized that flying with the nxControl

system would lead to beneficial effects on performance,

workload, and SA compared with conventional raw-data

flying. Specifically, we expected that flying with the support

of the nxControl system (a) would enable a higher precision

in controlling airspeed and vertical flight path, (b) would

lower the cognitive and physical effort involved in proper

thrust control, and (c) would enhance the situation

awareness of pilots with respect to energy-relevant flight

parameters. It was further expected that the beneficial

effects of the complete nxControl system, that is, the

augmented display of energy status information in the

PFD combined with the new concept of thrust control,

would be greater than when adding only energy status

information to the PFD.

Figure 2. Relationship between total energy angle, flight path

angle, and artificial horizon. “nxControl instead of pitch-and-

power: A concept for enhanced manual flight control” by S. Müller,

K. Schreiter, D. Manzey, & R. Luckner (2016). CEAS

Aeronautical Journal, 7, 110.

Method

Participants

In all, 24 licensed commercial airline pilots (all male,

10 captains, 14 first officers) were recruited as participants.

Their average age was 40 years (SD =12.6years) and

ranged from 24 to 63 years. The pilots had flight experience

of between 600 and 25,000 flight hours (M=8,505 hr,

SD =7,422 hr). All pilots possessed an Airbus type-rating

(20 A320,3A330/A340,1A380). Therefore, they were

familiar with the Airbus displays, sidestick, and its control

laws. The pilots were offered an overall expense allowance

for participation and access.

Apparatus

The study took place in a fixed-base flight simulator

(Simulator for Educational Projects and Highly Innovative

Research, SEPHIR) of the Chair of Flight Mechanics, Flight

Control and Aeroelasticity of the Technische Universität

Berlin. The simulator was configured as a simulation of

the VFW 614-ATD and contained sidesticks with a manual

flight control system. The flight characteristics, handling,

and the cockpit layout of the simulation used were

comparable to those of an Airbus A320. The outside view

was simulated by a collimated, high-quality visual system.

Primary flight displays, navigation displays, and engine

displays were presented on 1000 displays with a resolution

of 1,280 "1,024 pixels. For this study, the usual thrust

levers were replaced with a newly developed nxLever that

is compatible to both executing thrust control in a conven-

tional way (i.e., control of N1) as well as controlling thrust

according to the nxControl concept assisted by the

nxController.

Task

The flight task included a considerably complex flight

scenario with respect to energy management. The partici-

pants had to fly the required navigation performance

(RNP) approach pattern to Salzburg LOWS runway 33 from

south-east (Austro Control GmbH, 2014). In order to

increase the demands on precise flight path control, the

required performance was adjusted from RNP 0.3to RNP

0.1. This results in tolerances of ±100 ft vertically and

±0.1nautical mile horizontally. Additionally, there were

steady 15-knot crosswinds from 057"(minimal turbulences)

that sometimes acted as tailwind owing to the required

heading changes in the middle of the approach. The flight

scenario was designed for the use of speed brakes to some

extent. The visibility range of the simulated outside view

was reduced by fog, thus the runway was not visible until

reaching 1,240 ft above ground which is 100 ft above

minimum decision height.

For purposes of experimental control, the pilots were

obliged to follow a predefined approach procedure. This

should ensure that the data of all participants are

comparable, thus reducing influences and inflation of

error variance due to different approach strategies.

The prescribed procedure precisely defined when to

configure flaps or landing gear by reference to the dis-

tance to next waypoint. In addition, required changes in

airspeed were predefined as well. Thereby, the sequence

was aligned to standard procedures in civil aviation, but

altered in some critical points to enhance the observability

of how well the participants would cope with energy

adjustments.

Design

Three experimental conditions were compared in a within-

subject study design. In the first condition (conventional),

the simulator was configured for conventional, manual

raw-data flight without any enhanced energy information

and no assistance of thrust control. This condition served

as a control condition for assessing effects of the new

nxControl components. The second condition (energyPFD)

involved the conventional concept of thrust control but

combined it with the energyPFD, that is, a PFD enriched

by the presentation of the TEA and FPA (see Figure 1).

Finally, the third condition (nxControl) consisted of all

component of the nxControl system (energyPFD, the inter-

nal nxController for thrust and speed brakes control, and

the nxStatus display). The sequence of experimental

conditions was counterbalanced across participants.

Procedure

The experimental procedure was split into two parts.

The first part served as an accommodation and training

phase. It involved the pilot’s familiarization with the

simulator, the enhanced energy information in the PFD,

and the nxControl system with nxStatus display. The famil-

iarization included a standardized briefing and demonstra-

tions of the meaning and functionality of the displays and

systems. Afterward, the participants were trained on the

basis of four standard flight tasks, such as air work and

straight-in instrument landing system approaches.

This familiarization was followed by an introduction and

training of the RNP Salzburg approach and the predefined

approach procedure. The training included one approach

with conventional simulator configuration. During this

approach, reminders with respect to the procedure and

feedback were provided by the experimenter. Overall, the

accommodation and training phase took about 1.5hr, and

was followed by a short break of approximately 15 min.

The experimental data were collected in the second part

of the experiment. This part was divided into three

experimental blocks. Each block represented one condition

of this study. At the beginning of each block, the pilots first

performed a short practice flight to get accustomed to the

specific simulator configuration. Afterward, the participants

completed the RNP approach twice in a row with the

respective simulator configuration. During all flights, the

pilots were instructed to maintain the given flight path,

airspeeds, and configurations as precisely as possible.

They performed the approaches as pilot flying from the left

seat. The experimenter served as pilot monitoring from the

right seat and executed the ordered configuration changes

of the aircraft, or selected the requested flight parameters,

for example, altitude, speed, and heading. In addition,

the experimenter mentioned whether flight parameters

exceeded tolerances and read back the common callouts,

too. After the three experimental blocks, a debriefing

interview took place, in order to get additional informa-

tion about the subjective assessment of the pilots concern-

ing the different components of the nxControl system.

Dependent Measures

Performance

In order to assess how precisely the participants followed

the prescribed approach pattern, deviations from the given

reference pattern across time were examined by means of

the root mean square error (RMSE) with reference to the

prescribed altitude, airspeed, and lateral flight path targets.

The precision to match the given target parameters of

altitude and airspeed was particularly used for assessing

the pilot’s energy management performance. Both of these

parameters are directly affected by the total energy and its

distribution in manual flight. The lateral deviation from the

requested flight path was additionally assessed in order to

explore potential indirect impacts of the enhanced energy

information or the nxControl system on flight path control.

Workload

Workload was assessed both subjectively and objectively.

As an objective measure of workload involved in energy

management, the input activity at the thrust and speed

brake lever or nxLever was assessed, respectively.

This input directly reflects the demands on a pilot in terms

of the number of necessary thrust adjustments, that is,

indicates how often the pilot has to invest cognitive and

physical effort to re-assess and adjust the energy state of

the aircraft. A lever movement was counted if a change

in lever position was greater than 0.2cm (approximately

0.5% of the entire lever range) in a time span of 2s.

The accumulated time span in which a lever movement

was detected was related to the overall duration of the flight

scenario. The higher the percentage, the more movement

on the lever was required to complete the flight task, and

thus, the higher the workload involved in energy

management.

The NASA-TLX (Hart & Staveland, 1988) was used to

assess the subjectively experienced workload. Pilots provide

their ratings on the six subscales after each single approach,

that is, twice per condition. Owing to time constraints and

methodological considerations, the NASA-TXL was used

without the weighting procedure (Byers, Bittner, & Hill,

1989).

Situation Awareness

An adjusted form of the Situation Awareness Global

Assessment Technique (SAGAT) was used to objectively

assess the pilot’s SA while flying in each experimental

condition (Endsley, 1988). In each experimental block, the

simulation was suddenly frozen at one of three possible

predefined points of the first approach. At the same time,

the flight displays were blanked. The participants were

then asked to recall the following eight flight parameters:

indicated airspeed, lateral RNP deviation, vertical RNP

deviation, barometric altitude, vertical speed, pitch, fan

rotation speed N1, and heading. If the answers matched

the actual parameter within a given range (see Table 1), it

was counted as correct. The sum of all correct answers

was then taken as an indicator of the level of SA varying

between 0and 8. The specific point where the SAGAT

assessment took place during the approach was counter-

balanced across the experimental conditions.

In addition, a subjective SA assessment was performed

by means of the SA-SWORD. This technique is an SA

adaption of the Subjective Workload Dominance (SWORD)

tool described by Vidulich, Ward, and Schueren (1991).

The participants were required to judge which simulator

configuration supported their SA better in three pair-wise

comparisons. The judgments derived from the pairwise

comparison were then checked for consistency and

transformed into SA rating scores for each of the three

configurations according to the procedure described by

Vidulich et al. (1991). These scores describe the extent of

subjectively perceived support of SA by the different

configurations on a common scale.

Data Analysis

All statistical analyses, except analyses of SA assessments,

were performed based on aggregated data from both

approaches per experimental block. Differences between

the three experimental conditions were statistically

analyzed by means of a one-factorial analysis of variance

(ANOVA) with repeated measures. In the case of violations

of the assumption of sphericity, a correction of degrees of

freedom according to Huynh-Feldt was applied. Specific

post hoc comparisons between means were conducted

according to the Dunn-Šidák procedure for multiple

comparisons.

Results

Performance

Airspeed

The means of the RMSE of airspeed for the three experi-

mental conditions are shown in Figure 3. As is evident,

the pilots were better able to maintain the preset airspeeds

in the condition energyPFD (M=2.8knots) and nxControl

(M=2.8knots), compared with the conventional condition

(M=3.5knots). In the ANOVA this was reflected in a

significant main effect, F(1.58,36.35) = 12.02,p< .001,

= .343. Pairwise post hoc comparisons of the three

means revealed the differences between energyPFD and

conventional condition, p< .001, as well as between

nxControl and conventional condition, p= .005, as signifi-

cant. No significant effect, however, emerged between

energyPFD and nxControl, p= .945.

Altitude

Figure 4shows the mean RMSE of altitude for each

experimental condition. In contrast to our assumptions,

the average error in altitude was higher in the nxControl

condition compared with the other two conditions (conven-

tional: M=31.2ft; energyPFD: M=32.0ft). However, the

differences were only small and did not become

statistically significant, F(1.51,34.79) = 1.98,p= .163,

= .079.

Lateral Flight Path Deviation

The mean lateral flight path deviations when flying with

the three different configurations are shown in Figure 5.

The mean deviations were the same for the conventional

and energyPFD condition (0.030 nautical miles) and only

slightly higher when flying with nxControl (0.035 nautical

miles), with no statistically significant differences between

conditions, F(2,46) = 2.00,p= .147,η

= .080.

Workload

Lever Activity

Figure 6shows a graph of the lever activity assessment.

The white bars mark the means of lever activity across

the three experimental conditions. In the conventional

and energyPFD conditions, they represent movements of

the thrust lever, in the nxControl condition, it shows the

movements of the nxLever.

The ANOVA comparing these conditions revealed that

the pilots were able to perform the given flight task with

lower lever activity in condition nxControl (M=16.0%)

as compared with both energyPFD (M=24.4%) and

conventional (M=23.4%), F(1.35,31.09) = 42.87,

Figure 3. Mean RMSE of airspeed with standard error of each

configuration. PFD = primary flight display.

Figure 4. Mean RMSE of altitude with standard error of each

configuration. PFD = primary flight display.

Table 1.Tolerances for SAGAT evaluation

Flight parameter Tolerance

Indicated airspeed ± 2.5 knots

Lateral RNP deviation ± 0.25 dots

Vertical RNP deviation ± 0.25 dots

Altitude ± 50 ft

Vertical speed ± 50 ft/min

Pitch ± 1"

Engine speed N1 ± 2.5%

Heading ± 2.5"

Notes. This table shows the tolerances used to evaluate each flight

parameter in the SAGAT questioning. If the participant’s answers matched

the actual parameter within the tolerance, it was counted as correct.

The range of tolerance was derived from usual flight tolerances and read-

off accuracy. RNP = required navigation performance. SAGAT = Situation

Awareness Global Assessment Technique.

p< .001,η

= .651. Pairwise post hoc contrasts (Šidák)

revealed that significantly less lever activity was needed

in the condition nxControl compared with both other

conditions (both p< .001), while no such effect emerged

between conventional and energyPFD, p= .299.

The gray bars in Figure 6represent the activity at the

speed brake lever. Obviously, there are only data available

in those conditions where speed brakes had to be operated

by means of a separate input device, that is, conventional

and energyPFD. As can be seen, no differences emerged

between these two conditions with respect to this variable

(M=1.9% for both conditions). In condition nxControl,

however, the speed brakes were automatically controlled

by the nxController when activated by the pilot. Thus, no

separate control inputs for speed brakes were necessary.

NASA-TLX

The data of the overall NASA-Task Load Index (TLX) score

for condition energyPFD (M=41.9) were slightly lower

than in the conventional (M=44.0) and nxControl

(M=44.2) conditions. No significant difference emerged

between all three conditions, F(2,46) = 0.327,p= .723,

= .014. However, by looking at the specific workload

dimensions, differences emerged with respect to the sub-

jective assessment of physical demands (Figure 7). That

is, the subjective ratings of the pilots’physical demand were

about nine points lower for the conditions energyPFD

(M=36.8) and nxControl (M=36.2) than the conventional

condition (M=45.2), F(2,46) = 4.45,p= .017,η

= .162.

Pair-wise comparisons of the three means revealed a

significant effect between the conventional and energyPFD

conditions, p= .029. The difference between conventional

and nxControl just missed reaching the usual level of

significance, p= .062.

Situation Awareness

SAGAT

The mean numbers of correct answers (out of eight) in

the SAGAT were 3.1for the conventional condition, 2.9

for the energyPFD condition, and 2.5for the nxControl

condition. No significant effects emerged between the three

conditions, F(2,46) = 1.307,p= .280,η

= .054.

SA-SWORD

The subjective ratings of pilots’SA as assessed by the

SA-SWORD questionnaire are depicted in Figure 8. Before

analyzing the data, the pair-wise ratings needed to be

checked for consistency according to Crawford and

Williams (1985). If the consistency measure exceeds

=0.56, the rating is discarded (Budescu, Zwick, &

Rapoport, 1986). Thus, 11 ratings were excluded from the

analysis.

SA ratings were higher in the conditions energyPFD

(M= .61) and nxControl (M= .61) than the conventional

condition (M= .17). The ANOVA proved this effect to be

significant, F(1.16,13.92) = 7.06,p= .016,η

= .370.

The post hoc analysis revealed significant effects between

Figure 7. Mean TLX values of dimension physical demand with

standard error of each configuration. PFD = primary flight display.

TLX = Task Load Index.

Figure 6. Means of lever activity with standard error of each

configuration. White bars represent values for thrust lever or nxLever.

Gray bars represent values for speed brake lever. PFD = primary flight

display.

Figure 5. Mean RMSE of lateral flight path deviation with standard

error of each configuration. NM = nautical miles. PFD = primary flight

display.

the conventional and nxControl conditions, p= .001, as

well as between the conventional and energyPFD

condition, p= .006. No significant effect was observable

between energyPFD and nxControl, p=1.00.

Discussion

The present study investigated the human performance

consequences of an energy-enhanced flight display and a

newly proposed concept of total energy-related thrust

controller on pilots’performance, workload, and SA.

It was expected that the new system elements would

increase performance and overall SA while at the same

time reducing the pilots’workload. The results provide

support for most of these assumptions.

Considering the effects on the different aspects of pilots’

flying performance, especially regarding the results for

airspeed, altitude, and flight path deviations, it is evident

that the assumed performance benefits are particularly

observable in the better matching of the airspeed require-

ments. However, no comparable effect was found for the

altitude measure. With respect to matching the altitude

requirements, the pilots achieved similar precision in all

experimental conditions. This contrasts our expectations

and is somewhat surprising, given that the vertical speed

of an aircraft is directly affected by the management of

energies in manual flight. Obviously, neither the provision

of augmented energy information in the PFD nor the

additional provision of the new assistive system for thrust

control enabled the pilots to improve the precision of their

vertical flight path control. One straightforward explanation

of this lack of effect lies in the already very good

performance of altitude or vertical speed control in the

conventional condition. Thus, the lack of effects might

not reflect a lack of effectiveness of the new system

components but instead be related to a floor effect that

has hindered finding additional improvements in

performance. Two factors might have contributed to this

effect. Firstly, maintaining the requested altitude within

the tolerances of RNP 0.1might not have been challenging

enough to produce distinguishable results. However, RNP

0.1requirements currently represent one of the most

challenging requirements with respect to flight path

precision and it seems highly implausible that they were

too easy for our pilots. More likely, the already very good

performance in the conventional condition might be

explained by the control laws of the sidestick that, in

accordance with the Airbus philosophy, were designed to

hold the selected flight path. Thus, changes in thrust or

drag cause changes in speed rather than changes in

altitude. This might have biased the effects in our study

with respect to airspeed rather than altitude. Neither effect

was found for the precision of lateral flight path. However,

controlling the lateral flight task is not directly connected

to issues of energy management. In the current study

this aspect of performance was simply assessed as a type

of control variable to check for general changes in

flight precision due to unspecific effects not directly

related to the particular differences between experimental

conditions.

Workload was assessed as lever activity and in terms of

the subjective measure NASA-TLX. With respect to the

effort needed for proper thrust control, that is, lever activity,

the results suggest that nxControl can reduce the effort of

pilots during energy management, but only with the whole

system enabled, comprising the complete user interface as

well as the nxController. These results validate outcomes of

an earlier study that assessed nxControl in air work flight

tasks and straight-in approaches, but originated only from

a small amount of data (Müller et al., 2015,2016). On a

subjective level, this effect did not result in a perceived

reduction of overall workload. No difference was found in

the mean TLX ratings for the three conditions. Only a

reduction of the experienced physical load was found

which, however, emerged in both the energyPFD and

nxControl conditions. It might be that the general demands

of the complex approach dominated the subjective work-

load and masked any smaller differences between the

experimental conditions.

The results of SA assessment were somewhat inconsis-

tent. On the one hand, SAGAT did not show any significant

differences between the SA assessed across conditions. It is

worth mentioning that the SA score was generally very low

in all experimental conditions. This might be due to a too

narrowly chosen tolerance range that defines an answer

as correct, or that in this specific flight situation some of

the flight parameters are not crucial even in raw-data flight.

On the other hand, the subjective rating of the SA-SWORD

showed improvements of SA for the conditions energyPFD

Figure 8. Mean SA-SWORD ratings with standard error of each

configuration. PFD = primary flight display. SA-SWORD = Situation

Awareness-Subjective Workload Dominance.

and nxControl. This proves that the pilots appreciated the

additional information as an important improvement of

their situation assessment. Yet, it cannot be excluded on

the basis of the current data that this result might also

reflect the consequence of pilots thinking that provid-

ing more information in principle implies by definition a

better SA.

Overall, the results of the present study confirm the

results of our last study that showed beneficial human

performance effects associated with augmented energy

information and a new logic of thrust control (Müller

et al., 2016). It is striking that these effects again emerged

even though the participating pilots only had limited

practice with the new system components and their

subsequent performance was comparable to a condition

(conventional raw-data flying) in which they had much

more training. This suggests that the new system

components are easy to understand and to apply. It also

suggests that the findings of the present study are some-

what conservative and even more beneficial effects might

be expected if pilots gain more experience with the new

components.

The results also provide additional insights on the

specific effects of energy-enhanced displays and the new

concept of thrust control. Most of the beneficial perfor-

mance effects, with the exception of the effects on lever

activity, were already found in the energyPFD condition

where the pilots used the conventional concept of thrust

control but were supported by an energy display added to

the PFD. This provides direct empirical evidence for the

presumed advantages of providing augmented energy

information to pilots (e.g., Amelink et al., 2005; Lambregts

et al., 2008) and supports our expectation that providing

this information in the form of an emergent figure has

made it easier for the pilots to maintain a proper

distribution of energy and target flight states. It further

complements previous findings indicating that energy

displays can improve a pilot’s energy management

reflected in better precision of flying (van den Hoven

et al., 2010). Specifically, the current study proves this

effect is valid also in the context of a complex flight

scenario when energy information is added to the standard

head-down instrumentation. However, the beneficial

effects of a more precise adjustment of total energy

reflected in a reduced number of lever movements were

only gained by using the complete nxControl system,

comprising the energy-enhanced PFD, nxController,

nxLever, and nxStatus display. Since each adjustment of

thrust can be considered as an output of a new cognitive

assessment of the current pitch-and-power relationship,

the finding of reduced lever movements confirms our

expectation that the proposed nxControl approach indeed

can unload the pilot from the demands of pitch-and-power

flying to some extent. At the same time it also can reduce

the physical load involved in lever activity to control thrust.

In conclusion, the results of the current study suggest

that the proposed approach of nxControl as an alternative

to conventional thrust control can effectively and better

support manual flying, which need to be maintained and

applied in order to maintain pilots’competency to always

take over manual control of their aircraft. Providing such

support will further enable pilots to fly manually with

sufficient precision even in future aviation with presumably

more complex and demanding flight trajectories.

Acknowledgments

This work is funded by the DFG (German Research

Foundation) under Contract LU 1397/3-1, MA 3759/3-1.

The authors thank all pilots who participated in the exper-

iments and Vereinigung Cockpit e.V. that supported the

experiment by distributing the invitation to their members.

Further thanks are due to B. Boche, D. Bieniek, A. Kaden,

I. Karakaya, and M. Schaumburg for their support at

different phases of this research.

References

Amelink, M. H., Mulder, M., van Paassen, M. M., & Flach, J. (2005).

Theoretical foundations for a total energy-based perspec-

tive flight-path display. The International Journal of Aviation

Psychology, 15(3), 205–231. doi: 10.1207/s15327108ijap1503

Austro Control GmbH. (2014). Instrument approach chart –ICAO

(RNAV (RNP) Z RWY 33). Retrieved from http://eaip.austrocontrol.

at/lo/150430/Charts/LOWS/LO_AD_2_LOWS_24-6-6_en.pdf

Bennet, K. B., & Flach, J. M. (2011). Display and interface design:

Subtle science, exact art. Boca Raton, FL: CRC Press.

Budescu, D. V., Zwick, R., & Rapoport, A. (1986). A comparison of

the eigenvalue method and the geometric mean procedure for

ratio scaling. Applied Psychological Measurement, 10(1), 69–78.

doi: 10.1177/014662168601000106

Byers, J. C., Bittner, A. C., & Hill, S. G. (1989). Traditional and

raw task load index (TLX) correlations: Are paired comparisons

necessary? In A. Mital (Ed.), Advances in industrial ergonomics

and safety (pp. 481–485). Cincinnatio, OH: Taylor & Francis.

Catton, L., Starr, A., Noyes, J. M., Fisher, A., & Tim, P. (2007).

Designing energy display formats for civil aircraft: Reply to

Amelink, Mulder, van Paassen, and Flach. The International

Journal of Aviation Psychology, 17(1), 31–40.

Crawford, G., & Williams, C. (1985). The analysis of subjective

judgment matrices. Tech. rep. Santa Monica, CA: United States

Air Force.

Dutch Safety Board. (2010). Crashed during approach, Boeing

737-800, near Amsterdam Schiphol Airport, 25 February 2009.

Final report of Dutch Safety Board.

Endsley, M. R. (1988). Situation awareness global assessment

technique (SAGAT). In Proceedings of the IEEE National

Aerospace and Electronics Conference (pp. 789–795).

New York, NY: IEEE.

Federal Aviation Administration. (2013). Safety alert for operators:

Manual flight operations (SAFO 13002). Washington, DC: US

Department of Transportation.

Federal Aviation Administration. (2016). Enhanced FAA oversight

could reduce hazards associated with increased use of flight

deck automation (Report No. AV-2016-013). Washington, DC:

US Department of Transportation.

Filippone, A. (2012). Advanced aircraft flight performance.

Cambridge, UK: Cambridge University Press.

Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX

(Task Load Index): Results of empirical and theoretical

research. In P. A. Hancock & N. Meshkati (Eds.), Human mental

workload (pp. 139–183). Amsterdam, The Netherlands: North

Holland Press.

Lambregts, T., Rademarker, R., & Theunissen, E. (2008). A new

ecological primary flight display concept. In Digital Avionics

Systems Conference (pp. 4.A.1–1–4.A.1–20). St. Paul, MN: IEEE.

doi: 10.1109/DASC.2008.4702820

Müller, S., Manzey, D., Schreiter, K., & Luckner, R. (2015).

Implementing energy status in head down cockpit displays:

Impact of augmented energy Information on pilot’s perfor-

mance. In Proceedings of the Human Factors and Ergonomics

Society 59th Annual Meeting (pp. 926–930). Santa Monica, CA:

HFES. doi: 10.1177/1541931215591270

Müller, S., Schreiter, K., Manzey, D., & Luckner, R. (2016).

nxControl instead of pitch-and-power: A concept for enhanced

manual flight control. CEAS Aeronautical Journal, 7(1),

107–119. doi: 10.1007/s13272-016-0182-7

National Transportation Safety Board (Producer). (2013).

Chairman Hersman’s final media briefing on Asiana flight 214

crash July 11, 2013 [Video file]. Available from http://www.

youtube.com/watch?v=I1GopE_siVY

van den Hoven, M., de Jong, P. M., Borst, C., Mulder, M., &

van Paassen, M. M. (2010). Investigation of energy manage-

ment during approach–Evaluating the total energy-based

perspective flight-path display. In AIAA Guidance, Navigation,

and Control Conference (pp. 1–21). Reston, VA: AIAA.

Vidulich, M. A., Ward, F. G., & Schueren, J. (1991). Using the

Subjective Workload Dominance (SWORD) technique for pro-

jective workload assessment. Human Factors, 33(6), 677–691.

Wickens, C. D., & Carswell, C. (1995). The proximity compatibility

principle: Its psychological foundation and relevance to display

design. Human Factors, 37(3), 473–494.

Wickens, C. D., Hollands, J. G., Banbury, S., & Parasuraman, R.

(2013). Engineering psychology and human performance

(4th ed.). Boston, MA: Pearson.