A Technical Analysis Of Ergonomics And Human

Factors In Modern Flight Deck Design Essay, Research Paper

A Technical Analysis of Ergonomics and Human Factors in Modern Flight Deck Design

I. Introduction

Since the dawn of the aviation era, cockpit design has become

increasingly complicated owing to the advent of new technologies enabling

aircraft to fly farther and faster more efficiently than ever before. With

greater workloads imposed on pilots as fleets modernize, the reality of he or

she exceeding the workload limit has become manifest. Because of the

unpredictable nature of man, this problem is impossible to eliminate completely.

However, the instances of occurrence can be drastically reduced by examining the

nature of man, how he operates in the cockpit, and what must be done by

engineers to design a system in which man and machine are ideally interfaced.

The latter point involves an in-depth analysis of system design with an emphasis

on human factors, biomechanics, cockpit controls, and display systems. By

analyzing these components of cockpit design, and determining which variables of

each will yield the lowest errors, a system can be designed in which the

Liveware-Hardware interface can promote safety and reduce mishap frequency.

II. The History Of Human Factors in Cockpit Design

The history of cockpit design can be traced as far back as the first

balloon flights, where a barometer was used to measure altitude. The Wright

brothers incorporated a string attached to the aircraft to indicate slips and

skids (Hawkins, 241). However, the first real efforts towards human factors

implementation in cockpit design began in the early 1930’s. During this time,

the United States Postal Service began flying aircraft in all-weather missions

(Kane, 4:9). The greater reliance on instrumentation raised the question of

where to put each display and control. However, not much attention was being

focused on this area as engineers cared more about getting the instrument in the

cockpit, than about how it would interface with the pilot (Sanders & McCormick,

739).

In the mid- to late 1930’s, the development of the first gyroscopic

instruments forced engineers to make their first major human factors-related

decision. Rudimentary situation indicators raised concern about whether the

displays should reflect the view as seen from inside the cockpit, having the

horizon move behind a fixed miniature airplane, or as it would be seen from

outside the aircraft. Until the end of World War I, aircraft were manufactured

using both types of display. This caused confusion among pilots who were

familiar with one type of display and were flying an aircraft with the other.

Several safety violations were observed because of this, none of which were

fatal (Fitts, 20-21).

Shortly after World War II, aircraft cockpits were standardized to the ?

six-pack’ configuration. This was a collection of the six critical flight

instruments arranged in two rows of three directly in front of the pilot. In

clockwise order from the upper left, they were the airspeed indicator,

artificial horizon, altimeter, turn coordinator, heading indicator and vertical

speed indicator. This arrangement of instruments provided easy transition

training for pilots going from one aircraft to another. In addition, instrument

scanning was enhanced, because the instruments were strategically placed so the

pilot could reference each instrument against the artificial horizon in a hub

and spoke method (Fitts, 26-30).

Since then, the bulk of human interfacing with cockpit development has

been largely due to technological achievements. The dramatic increase in the

complexity of aircraft after the dawn of the jet age brought with it a greater

need than ever for automation that exceeded a simple autopilot. Human factors

studies in other industries, and within the military paved the way for some of

the most recent technological innovations such as the glass cockpit, Heads Up

Display (HUD), and other advanced panel displays. Although these systems are on

the cutting edge of technology, they too are susceptible to design problems,

some of which are responsible for the incidents and accidents mentioned earlier.

They will be discussed in further detail in another chapter (Hawkins, 249-54).

III. System Design

A design team should support the concept that the pilot’s interface with

the system, including task needs, decision needs, feedback requirements, and

responsibilities, must be primary considerations for defining the system’s

functions and logic, as opposed to the system concept coming first and the user

interface coming later, after the system’s functionality is fully defined.

There are numerous examples where application of human-centered design

principles and processes could be better applied to improve the design process

and final product. Although manufacturers utilize human factors specialists to

varying degrees, they are typically brought into the design effort in limited

roles or late in the process, after the operational and functional requirements

have been defined (Sanders & McCormick, 727-8). When joining the design process

late, the ability of the human factors specialist to influence the final design

and facilitate incorporation of human-centered design principles is severely

compromised. Human factors should be considered on par with other disciplines

involved in the design process.

The design process can be seen as a six-step process; determining the

objectives and performance specifications, defining the system, basic system

design, interface design, facilitator design, and testing and evaluation of the

system. This model is theoretical, and few design systems actually meet its

performance objectives. Each step directly involves input from human factors

data, and incorporates it in the design philosophy (Bailey, 192-5).

Determining the objectives and performance specifications includes

defining a fundamental purpose of the system, and evaluating what the system

must do to achieve that purpose. This also includes identifying the intended

users of the system and what skills those operators will have. Fundamentally,

this first step addresses a broad definition of what activity-based needs the

system must address. The second step, definition of the system, determines the

functions the system must do to achieve the performance specifications (unlike

the broader purpose-based evaluation in the first step). Here, the human

factors specialists will ensure that functions match the needs of the operator.

During this step, functional flow diagrams can be drafted, but the design team

must keep in mind that only general functions can be listed. More specific

system characteristics are covered in step three, basic system design (Sanders &

McCormick, 728-9).

The basic system design phase determines a number of variables, one of

which is the allocation of functions to Liveware, Hardware, and Software. A

sample allocation model considers five methods: mandatory, balance of value,

utilitarian, affective and cognitive support, and dynamic. Mandatory allocation

is the distribution of tasks based on limitations. There are some tasks which

Liveware is incapable of handling, and likewise with Hardware. Other

considerations with mandatory allocation are laws and environmental restraints.

Balance of value allocation is the theory that each task is either incapable of

being done by Liveware or Hardware, is better done by Liveware or Hardware, or

can only be done only by Liveware or Hardware. Utilitarian allocation is based

on economic restraints. With the avionics package in many commercial jets

costing as much as 15% of the overall aircraft price (Hawkins, 243), it would be

very easy for design teams to allocate as many tasks to the operator as possible.

This, in fact, was standard practice before the advent of automation as it

exists today. The antithesis to that philosophy is to automate as many tasks as

possible to relieve pressure on the pilot. Affective and cognitive support

allocation recognizes the unique need of the Liveware component and assigns

tasks to Hardware to provide as much information and decision-making support as

possible. It also takes into account limitations, such as emotions and stress

which can impede Liveware performance. Finally, dynamic allocation refers to an

operator-controlled process where the pilot can determine which functions should

be delegated to the machine, and which he or she should control at any time.

Again, this allocation model is only theoretical, and often a design process

will encompass all, or sometimes none of these philosophies (Sanders & McCormick,

730-4).

Basic system design also delegates Liveware performance requirements,

characteristics that the operator must posses for the system to meet design

specifications (such as accuracy, speed, training, proficiency). Once that is

determined, an in-depth task description and analysis is created. This phase is

essential to the human factors interface, because it analyzes the nature of the

task and breaks it down into every step necessary to complete that task. The

steps are further broken down to determine the following criteria: stimulus

required to initiate the step, decision making which must be accomplished (if

any), actions required, information needed, feedback, potential sources of error

and what needs to be done to accomplish successful step completion. Task

analysis is the foremost method of defining the Liveware-Hardware interface. It

is imperative that a cockpit be designed using a process similar to this if it

is to maintain effective communication between the operator and machine (Bailey,

202-6). It is widely accepted that the equipment determines the job. Based on

that assumption, operator participation in this design phase can greatly enhance

job enlargement and enrichment (Sanders & McCormick, 737; Hawkins, 143-4).

Interface design, the fourth process in the design model, analyzes the

interfaces between all components of the SHEL model, with an emphasis on the

human factors role in gathering and interpreting data. During this stage,

evaluations are made of suggested designs, human factors data is gathered (such

as statistical data on body dimensions), and any gathered data is applied. Any

application of data goes through a sub-process that determines the data’s

practical significance, its interface with the environment, the risks of

implementation, and any give and take involved. The last item involved in this

phase is conducting Liveware performance studies to determine the capabilities

and limitations of that component in the suggested design. The fifth step in

the design stage is facilitator design. Facilitators are basically Software

designs that enhance the Liveware-Hardware, such as operating manuals, placards,

and graphs. Finally, the last design step is to conduct testing of the proposed

design and evaluate the human factors input and interfaces between all

components involved. An application of this process to each system design will

enhance the operators ability to control the system within desired

specifications. Some of the specific design characteristics can be found in

subsequent chapters.

IV. Biomechanics

In December of 1981, a Piper Comanche aircraft temporarily lost

directional control in gusty conditions within the performance specifications of

the aircraft. The pilot later reported that with the control column full aft,

he was unable to maintain adequate aileron control because his knees were

interfering with proper control movement (NTSB database). Although this is a

small incident, it should alert engineers to a potential problem area. Probably

the most fundamental, and easiest to quantify interface in the cockpit is the

physical dimensions of the Liveware component and the Hardware designs which

must accommodate it. The comfort of the workspace has long been known to

alleviate or perpetuate fatigue over long periods of time (Hawkins, 282-3).

These facts indicate a need to discuss the factors involved in workspace design.

When designing a cockpit, the engineer should determine the physical

dimensions of the operator. Given the variable dimensions of the human body, it

is naturally impossible to design a system that will accommodate all users. An

industry standard is to use 95% of the population’s average dimensions, by

discarding the top and bottom 2.5% in any data. From this, general design can

be accomplished by incorporating the reach and strength limitations of smaller

people, and the clearance limitations of larger people. Three basic design

philosophies must be adhered to when designing around physical dimensions: reach

and clearance envelopes, user position with respect to the display area, and the

position of the body (Bailey, 273).

Other differences must be taken into account when designing a system,

such as ethnic and gender differences. It is known, for example, that women are,

on average, 7% shorter than men (Pheasant, 44). If the 95 percentile convention

is used, the question arises, on which gender do we base that? One was to speak

of the comparison is to discuss the F/M ratio, or the average female

characteristic divided by the average male characteristic. Although this ratio

doesn’t take into account the possibility of overlap (i.e., the bottom 5th

percentile of males are likely to be shorter than the top 5th percentile of

females), that is not an issue in cockpit design (Pheasant, 44). The other

variable, ethnicity must also be evaluated in system design. Some Asian races,

for example have a sitting height almost ten centimeters lower than Europeans

(Pheasant, 50). This can raise a potential problem when designing an instrument

panel, or windshield.

Some design guides have been established to help the engineer with

conceptual problems such as these, but for the most part, systems designers are

limited to data gathered from human factors research (Tillman & Tillman, 80-7).

As one story went, during the final design phase of the Boeing 777, the chairman

of United Airlines was invited to preview it. When he stood in his first class

seat, his head collided with an overhead baggage rack. Boeing officials were

apologetic, but the engineers were grinning inside. A few months later, the

launch of the first 777 in service included overhead baggage racks that were

much higher, and less likely to be involved in a collision. Unlike this

experience, designing clearances and reach envelopes for a cockpit is too

expensive to be a trial and error venture.

V. Controls

In early 1974, the NTSB released a recomendation to the FAA regarding

control inconsistencies:

“A-74-39. Amend 14 cfr 23 to include specifications for standardizing fuel

selection valve handle designs, displays, and modes of operation” (NTSB

database).

A series of safety accidents occurred during transition training of pilots

moving from the Beechcraft Bonanza and Baron aircraft when flap and gear handles

were mistakenly confused:

“As part of a recently completed special investigation, the safety board

reviewed its files for every inadvertent landing gear retraction accident

between 1975 and 1978. These accidents typically happened because the pilot was

attempting to put the flaps control up after landing, and moved the landing gear

control instead. This inadvertent movement of the landing gear control was often

attributed to the pilot’s being under stress or distracted, and being more

accustomed to flying aircraft in which these two controls were in exactly

opposite locations. Two popular light aircraft, the Beech Bonanza and Baron,

were involved in the majority of these accidents. The bonanza constituted only

about 30 percent of the active light single engine aircraft fleet retractable

landing gear, but was involved in 16 of the 24 accidents suffered by this

category of aircraft. Similarly, the baron constituted only 16 percent of the

light twin fleet, yet suffered 21 of the 39 such accidents occurring to these

aircraft” (NTSB database).

Like biomechanics, the design of controls is the study of physical relationships

within the Liveware-Hardware interface. However, control design philosophy

tends to be more subtle, and there is slightly more emphasis on psychological

components. A designer determines what kind of control to use in a system only

after the purpose of the system has been established, and what operator needs

and limitations are. In general, controls serve one of four actions:

activation, discrete setting, quantitative setting, and continuous control.

Activation controls are those that toggle a system on or off, like a light

switch. Discrete setting switches are variable position switches with three or

more options, such as a fuel selector switch with three settings. Quantitative

setting switches are usually knobs that control a system along a predefined

quantitative dimension, such as a radio tuner or volume control. Continuous

controls are controls that require constant equipment control, such as a

steering wheel. A control is a system, and therefore follows the same

guidelines for system design described above. In general, there are a few

guidelines to control design that are unique to that system. Controls should be

easily identified by color coding, labeling, size and shape coding and location

(Bailey, 258-64). When designing controls, some general principles apply.

Normal requirements for control operation should not exceed the maximum

limitations of the least capable operator. More important controls should be

given placement priority. The neutral position of the controls should

correspond with the operator’s most comfortable position, and full control

deflection should not require an extreme body position (locked legs, or arms).

The controls should be designed within the most biomechanically efficient design.

The number of controls should be kept to a minimum to reduce workload, or when

that is not possible, combining activation controls into discrete controls is

preferable. When designing a system, it should be noted that foot control is

stronger, but less accurate than hand control. Continuous control operation

should be distributed around the body, instead of focused on one particular part,

and should be kept as short as possible (Damon, 291-2). Detailed studies have

been conducted about control design, and some concerns were such things as the

ability of an operator to discern one control with another, size and spacing of

controls, and stereotypes. It was stated that even with vision available,

easily discernible controls were mistaken for another (Fitts, 898; Adams, 276).

A study by Jenkins revealed a set of control knobs that were not prone to such

error, or were less likely to yield errors (Adams, 276-7). Some of these have

been incorporated in aircraft designs as recent as the Boeing 777. Another

study, conducted by Bradley in 1969 revealed that size and spacing of knobs was

directly related to inadvertent operation. He believed that if a knob were too

large, small, far apart, or close together, the operator was prone to a greater

error yield. In the study, Bradley concluded that the optimum spacing between

half-inch knobs would be one inch between their edges. This would yield the

lowest inadvertent knob operation (Fitts, 901-2; Adams, 278). Population

stereotypes address the issue of how a control should be operated (should a

light switch be moved up, to the left, to the right, or down to turn it on?).

There are four advantages that follow a model of ideal control relationship.

They are decreased reaction time, fewer errors, better speed of knob adjustment,

and faster learning. (Van Cott & Kinkdale, 349). These operational advantages

become a great source of error to the operator unfamiliar with the aircraft and

experiencing stress. During a time of high workload, one characteristic of the

Liveware component is to revert to what was first learned (Adams, 279-80). In

the case of the Bonanza and Baron pilots, this was the case in mistaking the

gear and flap switches.

VI. Displays

In late 1986, the NTSB released the following recommendation to the FAA

based on three accidents that had occurred within the preceding two years:

“A-86-105. Issue an Air Carrier Operations Bulletin-Part 135, directing

Principal Operations Inspectors to ensure that commuter air carrier training

programs specifically emphasize the differences existing in cockpit

instrumentation and equipment in the fleet of their commuter operators and that

these training programs cover the human engineering aspects of these differences

and the human performance problems associated with these differences” (NTSB

database).

The instrumentation in a cockpit environment provides the only source of

feedback to the pilot in instrument flying conditions. Therefore, it is a very

valuable design characteristic, and special attention must be paid to optimum

engineering. There are two basic kinds of instruments that accomplish this

task: symbolic and pictorial instruments. All instruments are coded

representations of what can be found in the real world, but some are more

abstract than others. Symbolic instrumentation is usually more abstract than

pictorial (Adams, 195-6). When designing a cockpit, the first consideration

involves the choice between these two types of instruments. This decision is

based directly on the operational requirements of the system, and the purpose of

the system. Once this has been determined, the next step is to decide what sort

of data is going to be displayed by the system, and choose a specific instrument

accordingly.

Symbolic instrumentation usually displays a combination of four types of

information: quantitative, qualitative, comparison, and reading checking (Adams,

197). Quantitative instruments display the numerical value of a variable, and

is best displayed using counters, or dials with a low degree of curvature. The

preferable orientation of a straight dial would be horizontal, similar to the

heading indicator found in glass cockpits. However, conflicting research has

shown that no loss accuracy could be noted with high curvature dials (Murrell,

162). Another experiment showed that moving index displays with a fixed pointer

are more accurate than a moving pointer on a fixed index (Adams, 200-1).

Qualitative readings is the judgment of approximate values, trends, directions,

or rate of variable change. This information is displayed when a high level of

accuracy is not required for successful task completion (Adams, 197). A study

conducted by Grether and Connell in 1948 suggested that vertical straight dials

are superior to circular dials because an increase in needle deflection will

always indicate a positive change. However, conflicting arguments came from

studies conducted a few years later that stated no ambiguity will manifest

provided no control inputs are made if a circular dial is used. It has also

been suggested that moving pointers along a fixed background are superior to

fixed pointers, but the few errors in reading a directional gyro seem to

disagree with this supposition (Murrell, 163). Comparisons of two readings are

best shown on circular dials with no markings, but if they are necessary, the

markings should not be closer than 10 degrees to each other (Murrell, 163).

Check reading involves verifying if a change has occurred from the desired value

(Adams, 197). The most efficient instrumentation for this kind of task are any

with a moving pointer. However, the studies concerning this type of

informational display has only been conducted with a single instrument. It is

not known if this is the most efficient instrument type when the operator is

involved in a quick scan (Murrell, 163-4).

The pictorial instrument is most efficiently used in situation displays,

such as the attitude indicator or air traffic control radar. In one experiment,

pilots were allowed to use various kinds of situation instruments to tackle a

navigational problem. Their performance was recorded, and the procedure was

repeated using different pilots with only symbolic instruments. Interestingly,

the pilots given the pictorial instrumentation performed no navigation errors,

whereas those given the symbolic displays made errors almost ten percent of the

time (Adams, 208-209). Regardless of these results, it has long been known that

the most efficient navigational methods are accomplished by combining the

advantages of these two types of instruments.

VII. Summary

The preceding chapters illustrate design-side techniques that can be

incorporated by engineers to reduce the occurrence of mishaps due to Liveware-

Hardware interface problems. The system design model presented is ideal and

theoretical. To practice it would cost corporations much more money than they

would save if they were to use less cost-efficient methods. However, today’s

society seems to be moving towards a global concensus to take safety more

seriously, and perhaps in the future, total human factors optimization will

become manifest. The discussion of biomechanics in chapter three was purposely

broad, because it is such a wide and diverse field. The concepts touched upon

indicate the areas of concern that a designer must address before creating a

cockpit that is ergonomically friendly in the physical sense. Controls and

displays hold a little more relevance, because they are the fundamental control

and feedback devices involved in controlling the aircraft. These were discussed

in greater detail because many of those concepts never reach the conscious mind

of the operator. Although awareness of these factors is not critical to safe

aircraft operation, they do play a vital role in the subconscious mind of the

pilot during critical operational phases under high stress. Because of the

unpredictable nature of man, it would be foolish to assume a zero tolerance

environment to potential errors like these, but further investigation into the

design process, biomechanics, control and display devices may yield greater

insight as far as causal factors is concerned. Armed with this knowledge,

engineers can set out to build aircraft not only to transport people and

material, but also to save lives.