ARCHEFROCEEDINGS
v. :,L.ttpoot,
Technische Hoges600i
DeNt
VOLUME 2
P1975-7
Volume 2
THE SYMPOSIUM WILL BE HELD IN THE NETHERLANDS, THE HAGUE - CONGRESS CENTRE - 27-31 OCTOBER 1975
Statements and opinions expressed it the papers are those of the authors, and do not necessarily represent the views of the Royal Netherlands Navy.
The papers have been reproduced exactly as they were received from the authors.
SESSION Dl:
Chairman: R. Bernotat
Professor in Human Engineering, Research Institute for Human Engineering, Meckenheim.
Static and dynamic simulation as a working tool for Human engineering research applied to the design of ship bridges.
H. Schuffel and A. Lazet.
Ergonomics and reliability in ship handling systems -theories, models and methods.
H.O. Istance and T.B.K. Ivergard.
The plight of the operator. J. Stark and J. Forrest.
SESSION D2:
Chairman: W. Verhage
Department of Naval Architecture, Royal Netherlands Naval College.
Performance of azimuthing thrusters. I.S. Gibson.
A hydrostatic "over the stern" 3600 training propulsion unit fitted to a minehunter.
G. Gardiner and R.D. Mulholland.
About a new electronic equipment for ship's trials. T. Mozai and I. Tanaka.
SESSION El:
Chairman: J.G.C. van de Linde
Commodore R. Neth. N., Flagofficer in charge of training and education Royal Netherlands Naval College.
Maritime collision avoidance as a differential game. T. Miloh and S.D. Sharma.
The collision avoidance systems of the US Navy surface ship bridge control console.
A.D. Beary.
Development of an optimal marine collision avoidance
system.
J.A. Sorensen.
Collision region: A new mathematical concept for safe navigation.
J.S. Cardenier and L.A. Stoehr.
VOLUME 2
Page 271 2-13 see Vol.6 2-43 2-61 2-97 2-109 2-129 2-134 2-156°SESSION E2
Chairman: J. Gerritsma
Professor in naval architecture, Delft University of Technology
Feasibility study of steering and stabilising by rudder. 2-172 J.B. Carley.
Sea trials on a roll stabiliser using the ship's rudder. 2-195
W.E. Cowley and T.H. Lambert.
Roll stabilisation by rudder. 2-214
A.R.J.M. Lloyd.
Measurement of ship roll dynamics by pseudo-random binary 2-243/257 sequence techniques.
G.P. Windett and J.O. Flower.
STATIC AND DYNAMIC SIMULATION AS A WORKING TOOL FOR HUMAN ENGINEERING RESEARCH
APPLIED TO THE DESIGN OF SHIP BRIDGES
BY
Ir. H. Schuffel and A. Lazet
SUMMARY
To optimize the man-machine relationship there are two different approaches. The adaptation of man to machine and at the other hand machine to man. The first mentioned approach is dealing with training and selection, the latter is
con-sidering the adaptation of the machine to given human characteristics, generally called Human Engineering.
At the present time the research in the field of Human Engineering, in relation to maritime operations, concentrates mainly on the design of appropriate lay-out of all control elements installed on board. The working tool for this research is simulation.
Mock-ups (static simulation of the work-environment in future) can be built to study an appropriate lay-out of the workspot. If the man is to work as an operator of control elements with the machine then the dynamics of the machine are to be simulated and adapted to the limited dynamics of the human operator. The aim of this type of studies is an increasing of the effectiveness of the man/machine system and a contribution to the well-being of the individual. Applications of these techniques are discussed.
INTRODUCTION
The control of the safe conduct of vessels is still increasing in complexity. There are a number of factors which contribute to this situation. The increas-ing density of the ship traffic in confined areas, the increasincreas-ing speed of vessles, the grow of their dimensions, sea pollution, automation and the de-creasing number of crew are some of these factors.
All these factors are related to man/machine systems and, as is shown in acc -dent statistics, groundings and collisions are often due to personnel failure. Looking to these facts, one can fomulate the question, can you predict how the man/machine relationship will function in a new (man/ship) system. This means can you predict the behaviour of man in several conditions, the interaction of man and lay-out of the work environment (the bridge as well as the environment of the ship).
For these questions simulation as a working-tool for human engineering research can be considered as a towing tank for naval-architecture.
For this paper difference is made between static and dynamic simulation. Static simulation means the representation of an environment which will be built in future without dynamic aspects.
Dynamic simulation contains the static simulation with time dependent functions.
STATIC SIMULATION
A mock-up, scale 1:1, considered as a static simulation, offers an opportunity to allocate the tasks within the man/machine system.
A rough estimation of the overall dimensions of the needed room for the bridge can be based on studies of the relations of man to man, man to instrument and
Figure 1. Photograph of a full-scale mock-up of the bridge of a pilot-cutter.
instrument to instrument, although, of the old system.
For an existing system it is also possible to ask the users their opinion on their work environment.
We did so for a study on bridge lay-out of Dutch Merchant Vessels. For 115 mariners (67 officers, 48 pilots) a questionnaire was compiled. For the main
part this questionnaire contains statements made by navigating officers during interviews on a number of short voyages on board of Dutch Merchant Vessels. People could give their responses by underlining one of three possibilities
agree, no opinion, disagree (see Table I). The order of statements was at random. They belonged to one of the following topics.
some general statements perception of the environment position of equipment
bridge lay-out
functioning of instruments
reading/operating instruments and equipment comfort
lighting of bridge instruments team work
training task load task uncertainty
This statements with antropometric data (Fig. 2) data concerning the illumina-tion, data of room to walk, noise, climate, etc. are the starting points to build the mock-up. When this first state of the study, is finished, the users in future are invited to judge their environment.
This can be done also by the use of questionnaires. It is possible to compare one lay-out with another, because it is very easy to rearrange the lay-out of
Table I. Examples of a questionnaire completed by captains, first, second and
third officers from a number of Dutch Ships and a group of pilots. Answers
were
given in percentages of the group c= (captains and navigating officers) and
(pilots).
2-3
agree
opinion
noagree
dis-POSITION OF EQUISTYENT; BRIDGE LAYOUT
cpcpcp
27 An extra radar stand at the front of the wheelhouse
is always recommendable
96 982 0 2 2
28 VHF terminals should be placed on the wings of the
bridge
64 90 5 4 30 629 Rudderangle indicator and tachometers should be
easily observable from the wings of the bridge
96 90 2 2 2 230 Peloruses mounted only on the wings of the bridge
are sufficient
57 50 4 4 39 4631 It is no luxury to have two chart tables in the
wheelhouse
71 56 2 17 27 2732 The use of certain instruments often depends on
the place where they are mounted
79 96 0 2 21 233 It is not necessary for a pelorus to be mounted
in the middle of the wheelhouse
54 71 16 12 30 1734 The chart table must be mounted in a completely
separate room
9 8 2 2 89 9035 Operating consoles (steering, propelling,
communi-cation) should be placed at the front of the bridge
91 67 0 2 9 3136 A VHF telephone must be in the immediate vicinity
of the radar stand
95 100
0 0 5 037 The window
frames on the bridge should often be
narrower
84 90 4 6 12 4
PERCEPTION OF THE ENVIRONMENT
15 From the ship's a view range of 215° is certainly
sufficient
7 4 0 2 93 9416 Ship's bridges have to be built in such a way that
the range of view approacheds 3600 as much as possible
100 100
0 0 0 017 The helmsman's range of view of the environment
must be as wide as possible
75 81 0 0 25 19LIGHTING OF BRIDGE AND INSTRUMENTS
90 Red light meets the requirements best for lighting
instruments and meters having white letters on a
black background
59 56 27 31 14 12
91 Red light meets the requirements very well in
lighting the Chart table
27 23 11 25 62 5292 Orange light meets the requirements very well in
lighting the chart table
45 75 9 2146
493 Generally speaking the lighting of ship's bridges
is. still deplorable
64 69 2 2 34 2994 Coloured light does not meet the requirements for
lighting the chart table
46 10 5 25 48 6595 It is wrong when at night one has to use a pocked
lantern or lighter to read or operate instruments
96 98 0 0 4 296 It often happends that one has to use apocket
lan-tern or lighter onthebridge to illuminate something
86 96 0 2 14 2COMFORT
97 There ought to be a toilet in the near vicinity
_
Figure 2. Antripometric data of the man sitting behind a console (50th percentile).
the mock-up. We compared four lay-outs in cooperation with Dutch captains, navigating officers and pilots for a study on bridge design of Dutch merchant vessels. Two of these lay-outs are shown here (see Figure 3).
There is one very strong preference; pilots do prefer a lay-out with a free passage between the front bulkhead and the consoles. However, the individual differences of the subjects in stating their preferences are quite large which accounts for the not significant general preferences.
In other words, optimum bridge-lay-out with regard to personal preferences Ls
not very stricktly limited.
However, you should take into account three factors designing bridge lay-out, looking to these personal preferences. The first factor is connected with the total lay-out of the bridge, tidy, systematic, apparatus grouped in consoles etc. The second factor may be interpreted as ease of operation and the third as confort on the bridge.
Thiswasthe result of a factor analysis (see Table II). Subjects (26 Dutch captains and navigating officers and 21 P's) evaluated a mock-up by rating 59 polarity scales.
An other interesting question to captains and navigating officers is what they
think about the need of apparatus on the bridge.In Figure 4 these instruments are given in groups; communication, propulsion, steering, radar observation, radiodirection finder, course, sounding, position indication condition of ship
_
_
and cargo.
106 subjects were asked to give their opinion on the importance of these instruments for captains and navigators.
The results are the mean of four see areas and two visibility conditions. Making difference between the importance of the instruments used by captains and officers of the watch it appears that for all instruments the relative importance tousethemis greater for the officers (see Figure 5). There are only a few exceptions; communication, propulsion, sounding and course setting dur-ing berthdur-ing are judged as more important for the captains.
An other question is the helmsman. The helmsman is considered by the captain and navigator as somebody who has always to be supervised and only obey given orders. In the figures 6 and 7 this is illustrated. The idea goes so far that it is found really irrelevant that the helmsman can observe the environment of the ship. In our opinion the helmsman's function should only be performed in heavy traffic situations and use is to be made of an auto pilot as long as
2-4
12 3
Lay-onc I
Conn,Ns placed gainet front
balkhead of the wheelhouse
Derhoht itt h. front of
the wheelhouse.
(3)..Chert-table st starboard
Anna., chart-table wan ealizat.on
Two redertutses ,n one COM pal nxnt points to be distributt, 10 5 0 5
/
Aiiimillummmomar
OWWWWW*
AV
etv
WEIWWINEME
WW
Elenne
/
10 Lay-out 11a. not tension front bulkhead of < .05
the wheelhouse < .01
<.01
. not at the front ofthe wheel- < .01 house
<.01
chart-cable at port < .01
<.05
id. at Starboard <.01
one radar in compartment and
at starboard <.01 <.01 difference between lternetivee (p) I , 10 5 0 5 10
Fig. 3. Comparative evaluation of main differences in arrangement of consoles of lay-out I and
lay-out It by distribution of
Table II. Results of factor analysis on polarity-scales (principal components,
Favi
100% 90 80 70 60 50 40 30 20 10 p. 150 14045] EOM 112 151 2-6 )27321 131321 128 331 f253M captainEl
officer of the watchF-3711
COMM. PROP STEER. RAD.OBS. R.D.F. COURSE SOUND. POSIND COND.S TOTAL
Figure 4. Differences in relative importance of groups of instruments for two members of bridge personal as rated by three groups of subjects combined (C, N, P, F).
varimax rotation).
Item factor I factor II factor III
14. organized .57 16. systematic .53 45. app. grouped .52 49. tidy .72 58. comprehensible .64 5. easy to operate .81 8. practical .89 11. app. well-placed .69
21. not causing stress .69
26. well-arranged .79 27. easy of access .52 29. real .65 31. efficient .71 34. well-considered .79 35. handy .84
43. app. easy to reach .65
50. inf. present good .71
57. easy .77 6. good atmosphere .56 20. pleasant .70 39. warm .60 42. colourful .55 44. sociable .67 52. healthy .52 55. right proportions
53
I00% 90 80 70 F., 60 50 .140 30 20 /0 0 COX. 100% 90 80 70 60 0 0 30 20 10 0
P np, STEER. RAD.OBS ROE
Figure 6. Differences in relative
importance of groups of
instru-ments, to be used by helmsman for
optimal task performance, rated
by C/N, P and F's.
2-7
0
captains and navigating officers(C/N) pilots (F.) gg foreign subjects (F) 80 50 1007 90
II
good visibility poorvisibility 4 30 20 10 0COMM PROP STEER.
19
Figure 7. Differences in relative
im-portance of groups of instruments to
be used by helmsman, with regard to
two visibility conditions; rates C/N,
P, F combined.
fee 38 351 171 40 411 59 35 34 !37 27 21 !Iv u of 23 20 1,025 29i 143 26 231 139 29 221 3050 2,11
Figure 5. DifferenceVEZTWeen
groupi-Z1 subjects in. rating tile relative
impor-tance of groups of instruments for optimal task performance by captains and OW.
possible, depending of course, on the quality of the auto-pilot and the traffic
situation. Lavery dense traffic the helmsman should be replaced by a qualified
officer, who navigates and controls the course, next to the officer of the
watch.
17i al
141 28 42/ 610]EEI
F377iiLEI
005 ND CON0.5.< TOP, SOUND
Summarizing the data on bridge lay-out, gathered in cooperation with Dutch captains, navigating officers and pilots, with regard to tankers, containers and freighters we can divide the floor space of the bridge into four groups of
apparatus:
the conning position against the front bulkhead of the bridge comprising the steering of the ship and the control of the engine, the position for the pilot and restricted possibilities to plot the ship's position in combination with a daylight radar;
the radar compartment, not constructed against a side of the wheelhouse. It is not relevant whether this compartment is situated at starboard or port; the plotting side with chart-table and position finding apparatus. Chart-table and radar compartment must be close together;
signalization console to control and to monitor the condition of ship and cargo and a monitor position for the machinery plant.
Special attention has to be paid to the bridge-wing with regard to manoeuvering the ship in the harbour (controls and information presentation).
Pilots have a strong preference to a free passage along the front bulkhead, a daylight radar in the front at the wheelhouse at starboard and at least VHF at their disposal.
There is no significant difference within the groups of subjects about the position of the helmsman; at the front or placed backwards. If there is a helmsman, he should only control the course.
Captain and navigating officers have the sane opinion about the information and controls they need for their tasks. As most important instruments they consider apparatus for communication, propulsion and steering. Radar, peloris, souding, position indicating and the signalization for the condition of
ship
and cargo are secondary (see Figure/4).DYNAMIC SIMULATION
lot of information can be gathered from a static-mock-up, as is stated before. However, how to judge personal preference when there exists opposite idea's between two groups of personal as is illustrated before with pilots and navigating officers.
Does it influence the collision danger, ship's course? A possibility to measure the influence of bridge lay-out, ship's environment and the behaviour of people on the bridge, on the sailed course of the ship is asimulator. In that case all conditions are controlled and manoeuvres can be repeated.
For this human engineering research we evaluated in our Institute a facility to simulate the environment of a ship.
The simulator consists of four parts; the mock-up of the bridge with instru-ments, a computer to calculate the ship's moving characteristics, a scale model of the ship's environment (scale 1:100 or 1:500) and a video system to project
the image of the environment an a screen.
The simulator is situated in a building of 45 x 25 x 9 m (see Figure 8).
For the calculations of the ship's movements (two translations x, y and one rotation *) there are available a PDP-15 digital computer and an extended
analog computer 1{itachi-240.
The scale model of the environment has thedimensions of the ground surface of 20 x 25 m, so this area can cover scale 1:500 a surface of 10.000 x 12.5000 m.
In the scale model there are build dikes, leading lights, illuminated houses of the town in the background etc.
There also can move other ships in the vicinity of the by yourself steered ship.
X OCAMERA'S SCALE MODEL ,Lp COMPUTER
2-9
( . BRIDGELiar
Figure 8. Four parts of the simulator, situated in a building.
This scale model technique Wilb been succesfully applied already for the design of the harbour entrance at Hook of Holland.
The mockup itself has been discussed already. For the simulation we use a mockup with a groundsurface of 9 x 7 m.
All parts of the bridge mocup are versatile.
Use can be made of position indicating instruments, steering, direct control of the engines etc.
There is a radar simulator available also developed in the Institute. It is a relative radar display, with head up of north up stabilizing, range and bearing marker, off center etc.
A synthetic radar display with anti collision devices is under construction now.
Above the scale model a set of T.V. camera's can move.
This set is connected to a girder construction with a span of 20 m, the set can be steered in x and y direction (see Figure 9).
MOCKUP
x
Y COMPUTER RUDDER TELEGRAPH X Y 'VFigure. 10. Scheme of the partially dynamic
simulation of bridges with the use of an
environmental display.
The set of camera's can rotate around the vertical axis
(ti, -
rotation). It is possible to install 9 T.V. camera's, to reach a 3600 view angle.The video signal is projected on a screen. For each camera there is a projector. At his moment we use 3 projectors providing a viewing angle of 1200. The resolution of this T.V, system is about 6-7 arc minutes.
The man on the bridge (mock-up) can control the rudder and the revolutions of the propeller. This signal is fed into the computer. The machine calculates the displacements of the ship (this means the place of T.V. camera's) and the
an in the mock-up perceives change of course and change of speed. hen he can take action again on what he perceives (see Figure 10).
Qtth tnis set-up it is possible to change the environmental conditions systema-tically which is also the case with the lay-out of the bridge itself, the number of personal on the bridge and the degree of automation.
Their influence can be measured with the objective parameters, like sailed course, time that the ship is kept near the target, time before a failure in the console is taken away, etc. Comparing these parameters gives the indications where in specific conditions an optimum for the man/machine relationship exists.
From this type of human engineering research contributions may be expected to ship's safety and the well-being of the individual.
REFERENCES
A. Lazet, H. Schuffel, J. Moraal, U. Leebeek and H. van Dam Bridge design on dutch merchant vessels; an ergonomic study. Part I: A summary of ergonomic points of view (Dutch), 1973.
J. Moraal, H. Schuffel and A. Lazet
"Bridge design on dutch merrh nt vessels; an ergonomic study. Part II". First results of a questionnaire completed by captains, navigating officers and pilots, 1973.
A. Lazet, H. Schuffel, J. Moraal, H.J. Leebeek and H. van Dam
"Bridge design on dutch merchant vessels; an ergonomic study. Part III": Observations and preliminary recommendations.
J. Moraal, H. Sdhuffel, H.J. Leebeek and A. Lazet
"Bridge design on dutch merchant vessels; an ergonomic study. Part IV": Evaluation of standards and recommendations by means of a static mock-up,
ERGONOMICS AND RELIABILITY IN SHIP HANDLING SYSTEMS - THEORIES, MODELS AND METHODS
BY
ISTANCE, H AND IVERGARD, T
The Swedish research and consultancy firm ERGOLAB, is doing work under contract for the Swedish Ship Research Foundation on reliability of ship handling sys-tems. The study is especially concerned with the human problems involved in the control and manoeuvring of ships.
The main aim of this project is to define the areas of unreliability due to de-ficiencies in the man-machine interface. The study started with a detailed cri-tical literature review (6), and a number of visits to other research groups (8).
Following this, a functional system analysis has been carried out and the re-sults from this are given in this report. There is also a theoretical background for an analysis of unreliability in the man-machine interface. A great emphasis has also been placed on the development of an information model which could be used to study this problem. A number of data collection studies have been done including a large questionnaire study, and ship visits which include interviews and observation studies.
The information model and the methods of data collection derived from it are also discussed in this report.
1. A FUNCTIONAL ANALYSIS OF SHIP CONTROL
The objective of a functional system analysis is to obtain an analysis of a system with as few preassumptions as possible about existing technology.
The system under study is that of ship handling and ship control. The main pur-pose of this kind of system is to transfer the ship, including cargo, from one harbour to another. The purpose has, of course, to be fulfilled in such a way that certain criteria are satisfied. The main criteria will be those of safety to people, environment and cargo, and also those of economy and time. The dif-ferent laws and regulations, and internationally based technological systems
(satellites, Loran etc) also have to be considered.
In this section a control system with the previously stated purpose will be dis-cussed. First some definitions will be given and then a functional flow diagram is discussed. The following sections discuss the relevance of this diagram from the Point of view of reliability.
1.1. Some definitions
A function is defined as a performance of a system which is required to accom-plish a given objective. A function may be performed by man or machine, or a combination of both.
A functional analysis is an analysis of a system in terms of the functions re-quired to accomplish a given objective. A functional analysis allows the meth-ods of performing functions in existing systems to be evaluated in relation to the system objective. Functional analyses are also necessary for determining the optimal allocation of functions in new or modified systems.
A functional analysis of ship control allows the methods of performing the
necessary functions to be evaluated with reference to such criteria as safety, efficiency and reliability. Areas of weakness in existing systems can be iden-tified, and better methods of performing the functions and allocating the func-tions between man and machine can be suggested for new or modified systems.
1.2. Description of the functional flow diagram
The functional flow diagram (see figure 1) contains a number of blocks, eachof
which represents an individ,q1 function. Four types of functions are distin-guished as an aid to understanding the diagram, but it should be appreciated that the distinctions are fairly arbitrary. The four types of functions are:
Information input. This is where information must be obtained to perform another function. Reference input is distinguished as a special form of infor-mation input where the inforinfor-mation is available in reference form, for example books, individual experiences etc, as opposed to information which must be ob-tained at the time it is required. Some examples are given on diagram for cla-rity, but there is no intention to predetermine how the function (that is ob-taining information) should be performed.
Assessment/evaluation. This type of function involves the preliminary pro-cessing of information which is required for the performance of other functions. For example, it is necessary to establish the position of one's own ship and the position of surrounding traffic, before any collision avoidance function can be performed.
Prediction. This is a further type of preliminary information processing characterised by the fact that it involves the prediction of future states. Such
functions are essential in ship control, since the response times of ships are so great. Decisions are taken with regard to predicted future states rather than with regard to present states, for example the future relative
positions of
ship in collision
avoidance.Determinationjdecision. These functions involve the basic decisions in ship control, such as where the ship is to go and the necessary control settings re-quired to get it there. The three other types of functions can be seen as pro-viding and processing information which is used in these basic decisions.
The arrows between the blocks indicate the relations and general information flow between the functions. Feedback loops are not specifically indicated in this type of diagram.
Functions 1 and 2 in figure I are concerned with determining the route to be followed by the ship. Functions 3 and4 are concerned with assessing the state of the ship, that is, its position, speed, swing etc. This information, together with the information provided by functions
5, 6
and7,
enables the future state of the ship to be predicted (function8).
Functions 9 and 10 are concerned with assessing the state of surrounding traffic, which, with the constraints provided by functions 11, 12 and 13, enables the future state of the surrounding traffic to be predicted (function14).
Function 15 lies at the centre of the functional flow diagram. Here information about the state of one's own and surrounding ships is integrated with information about manoeuvring constraints and the rules of collision avoidance in the determination of the desired future state of the ship. The final function, number16,
involves determining the appropriate control set-ting for achieving the desired future state.1.3. Some known sources of unreliability in ship control
Detailed consideration of the functions shown in the functional flow diagram and existing methods of performing the functions will reveal the sources of un-reliability in ship control. A preliminary consideration of the problem suggests
the following examples:
3
4
Assess external disturbance
Information input: wind, speed, di-rection etc. Reference input: tide, tables etc.
Information input: landmarks, speed through water, over ground, direction etc. Reference input: "map"
Assess ship state (position, speed, course, swing etc)
Predict future state of ship (position, speed course, swing etc)
Information input: control settings Reference input: ship dynamics
Figure 1. Functional Flow Diagram of Ship Control 1 2 Reference input: destination, ship size etc. Determine Route Determine appropriate control setting to achieve desired state
Reference input: 13 collision, rules
(formal/informal) 10
Predict future state of surrounding traffic (extrapolation/plotting etc)
Assess manoeuvring con-straints and position of obstacles etc. 11 9 Information input: observation, radar etc. Assess state of surrounding traffic Information input: observation, radar etc. Reference input: charts, pilots knowledge etc. 12 15 Determine desired 14 future state of ship in relation to route, traffic, manoeuvring, con-straints, collision rules etc
There is a lack of information which can be used for ,he precise assess-ment of the ship's state. This is particularly critical during berthing and manoeuvring, where the future state of the ship must be predicted with some accuracy. The short range position finding methods (Decca navigator, radar etc)
are not accurate enough, and some types of log cannot be used at slow speed, and in any case, Only show the ship's forward velocity. New types of docking aid based on doppler radar or doppler sonar systems are being developed, but these are rarely able to provide information of sufficient reliability under all conditions. Usually the pilot and master must rely on information received directly through their own senses, but it is rather difficult to judge dis-tancies and velocities with sufficient accuracy, (the changes often being be-low the thresholds of the human sense organs) as the large number of incidents
during berthing testifies. New equipment which can reliably supply the neces-sary information is urgently required.
Knowledge of the ship dynamics is reference information required for pre-dicting the future ship state (function 8) and determining the correct control settings (function
16).
Inadequate knowledge of how the ship will react is a major reliability problem, especially during the berthing of VLCC's with their excessive time lags. Training the masters and pilots, perhaps using models or simulators, is one possible approach, which may be contrasted with the approach involving the automation of this function using a computer sytem. However, computer predictions of the ship's response require accurate programming of the dynamics, which can only be represented in a simplified form. Hence, there is a reliability in both of these approaches.The future state of surrounding traffic cannot be predicted with abso-lute accuracy (function 14), partly because ships may follow informal colli-sion rules, in addition to the formal ones. It could be of value here to be
able to Speak to Other ships to find out their intentions. Unfortunately, the use of VHF communication for this purpose is unreliable since it is not
pos-sible to precisely identify the ship one is talking to. A system showing a signal (similar to Racon) on the radar screen when a ship is transmitting, together with the directional control of VHF transmissions, would help eli-minating much of the unreliability.
The determination of the appropriate control setting (function
16)
to achieve the desired future state of the ship (function 15), is a source of un-reliability, particularly during manoeuvring. In the case of ships equipped with only a propeller and rudder, there is no one-to-one transformation be-tween control settings and ship states, when anything other than forward motion is considered. Bow and stern thrusters, and similar devices, can give direct control of the desired ship states, thus reducing the unreliability caused by the demand of frequent manoeuvring.In function 15 the information about the state of one's own ship and sur-rounding ships is integrated with information about manoeuvring constraints and the rules of collision avoidance in the determination of the desired future state of the ship. This is a very difficult decision task which today is mainly performed by the master or pilot. Computer systems which integrate col-lision avoidance systems, can be of great help. In low arousal situations
(open sea, fatigue etc) the risk of wrong decisions or omission of decision is especially high. The design of suitable warning systems is of high priority.
2. A THEORETICAL BACKGROUND
This section presents a theoretical backgroundabout man as. an information pro-cessor. A special emphasis is placed upon the reliability in this information
processing.
2.1. Introduction
Within a system, there are a series of functions which can be defined which must be fulfilled for the system to work. In a man/machine system, these func-tions can be given either to men to complete, or to machines, or to a combina-tion of the two. In the design of a system the stage is reached where the de-signer must decide how to allocate the functions between men and machines. One approach to this problem has been to devise checklists which state general types of function which men are better at performing, and those which machines are better at performing. However, this approach is not satisfactory as there are a number of factors, other than relative superiority, that have to be considered. The relative attributes of men and machines have been summarized by Jordan(9),
who says that machines can be depended to perform consistently but are totally inflexible, whereas men are flexible but cannot be relied on to perform con-sistently. Thus reliability of a system is going to be affected, if not de-termined by the combination of flexibility and consistency.
The ship has been described as one of the most complex socio-technical systems which man has devised to date. Within the total system of the ship there are
subsystems, and one of these is the ship handling system. The main criterion for assessing the reliability of a system is the probability of failure of the system or part of the system.
In investigation of weapons and aero space systems the estimation of man's con-tribution to failure is between 20% and 80% (4). Thus the human component of the system can contribute in varying degree to the failure of the system. If reliability is examined by the criterion of failure, then human failure is largely a question of human error. In a few cases, error is not the reason for failure, for example sudden death, though in the majority of cases this is so.
Attempts have been made to quantify human error in certain situations and thus put a number (or a range of numbers) on human reliability. The advantage of the prediction of errors is that it enables the reliability of the human component of the system to be quantified. The value of attempts to predict error rate is doubtful except perhaps in the simplest of tasks. This is because of the varia-tion of tasks, of human and situavaria-tions. Edwards and Lees make this reservavaria-tion when reviewing work on this subject. They approach the problem by categorizing work under different headings, for example, simple tasks, vigilance tasks, control tasks and so on. The largest degree of success has been obtained with the category of simple tasks (these include for instance assembly tasks). One approach has been to break down a task into its constituent behaviour elements and examine the probability of error associated with each of the elements.
The predicted error rate for the whole task can be obtained from the predicted error rate of its constituents. Generally this method is applicable to tasks with little scope for variation in the work. One can see that as tasks of greater complexity, and with possibilities for greater variation, are con-sidered, then the range of values obtained by this method become more and more approximate. The stage is reached when the values are so approximate as to be of no practical value whatsoever. For tasks of greater complexity a less rigid and more qualitative approach has to be adopted. Indeed, the quantitative ap-proach previously described is primarily for assessing human reliability, and no direct indication is given about increasing the reliability of the system. A more qualitative approach has more scope to examine reliability from the point of view of improving it.
The ship handling system itself is composed of a number of subsystems, for example the navigation system, the collision avoidance system. Often when there is only one officer on the bridge who makes decisions, for example, in periods of open water, then he is the human component of all the subsystems which make up the ship handling system. Thus the human role in the ship handling system as
a whole can be seen to be very complex. The role of the human component in the ship handling system can be examined in order to determine the type of investi-gation into reliability which is needed.
The key point in the system is the decision maker. This can be the officer on watch, the master or_the pilot. Sometimes a machine is used as a decision maker. Every time this person (or machine) is aware of any change in the system he makes a decision. Often the decision is that no action is required and he is
satisfied that the system will continue to perform adequately despite the change. When he monitors a change in the situation, for example the encounter of another ship, the same decision process continues. After a period of deciding no action is required on his part he may, for example, decide to alter course. There are Other men in the ship handling system but mainly their role is to aid the de-cision maker by providing him with information. The watch keeper informs him of changes in the external situation. The radio operator informs him of information obtained from sources external to the ship. The helmsman performs the function of implementing the decisions that are made, relating to the helm control
func-tions.
2.2. The causes of errors
Errors that are made by the decision maker can, on analysis, often be attributed to a number of factors and rarely is there only one factor which causes an error. There can be errors caused by the situation in which the man must work, and also errors caused by the way in which man reacts to the situation. Often factors from both areas contribute in the making of an error. This can form the basis for a very broad division in the factors causing errors. In general, this division can be summarized into firstly factors arising from man's ability to make a decision and secondly, factors arising from the working situation.
This division of errors was made by Rooke, and he divided errors into human caused errors (H.C.E.) and situation caused errors (S.C.E.). Swain, in using this, quoted work by Juran, which said on average human caused errors amounted to 20% and situation caused errors amounted to 80% (4).
Pitanova (13) quotes work by a group of Polish psychologists which presented a somewhat different ratio of the contribution of these two error types. Their work resulted from the study of the reasons for navigational accidents in the light of errors made by bridge officers. Their ratio was that 59% of the errors were a result of the personal qualities of bridge officers, while41% of errors were the result of shortcomings of the navigational information. This ratio will be discussed further in section 3.1.
There are, however, certain reservations to be made when making this distinction. Edwards and Lees
(4)
doubt to some extent the scientific basis for Rookesclas-sification of "human caused errors" and "situation caused errors". Indeed, one could argue that all errors of decision are human caused errors, and are caused by the human shortcomings to perform adequately. A stimulus could be below human
detection threshold level and not be detected by the vast majority of humans but this could be classified as a human failing. Also, as has been said, errors are usually caused by a number of factors which could come from both the areas of Rooke's classification. However, one can apply this type of distinction with
suitable reservations, when trying to categorize the factors which can cause errors. In this sense, it is important to distinguish between the different sources of factors which can cause an error to be made. Still using Rooke's terminology then "human caused errors" could result from factors that affect an individual's ability to make a decision at a given time, whereas the situation .caused errors result from factors arising from the man's working situation and not from his decision making ability. These two categories will be discussed in greater detail later (see section 2.4).
2.3. The decision making process
This description is intended to show more clearly the sites of the factors causing errors. It uses the concept of the mental model, which was developed by the Gestalt school of psychologists to account for certain discrepancies in be-haviour patterns. The mental model idea differentiates between the bebe-havioural world and the geographical world. The individual structures the situation in his mind and reacts to his own structure - Koffka, quoted in Singleton (14). If his personal structure is misleading then he will make errors. The steps where the model is used will be pointed out as they arise.
The same procedure is shown as that used in figure 1 in the functional flow
analysis. Information-. Sensor 1 Assessment Prediction 2-19
Action taken results in change of information received 1 Decision or determination 1 1 Implementation 0
-It is important to state that figure 1 is not a description of the decision making procedure, but a function flow diagram.
Fundamental to any decision is the information on which it is based. This in-formation can tell the man that a decision is required, what type of decision, and also information to use when making the decision. This information is ob-served, which can be through any of the sensory receptors. The information is ordered and interpreted into its relevant context, and is then used to update the mental model of the situation. Thus any new information will be interpreted
in the light of this model. Then the decision maker compares his model of the present situation with the desired situation and if there is no difference then the process reverts back to observation of information. If there is a signifi-cant difference then the question is asked, is action required? If the answer
is "no" and the difference between the present and desired states of the situa-tion can be tolerated without acsitua-tion, then the process reverts back to
ob-serving information. It is possible that in deciding whether action is neces-sary or not, a prediction of the future state of the system is used, to inter-prete the effect of the difference. If the difference cannot be tolerated then the next stage is the actual decision of the action to be taken. In the making of this decision, a prediction of the future states is used, and as stated earlier decisions are taken with reference to predicted future states of the system rather than with regard to present states.
In this general breakdown of decision making, further breakdown of this step is not applicable as the decision of what the action is, is dependent on the
re-quired by the decision maker.
When a decision is made which results in action being taken, the effect of the action is to change in some way the information that is received. This is the method of feedback to the decision maker on the effect of his decision, and
how they have changed the situation of the system. For an example the effect of a rudder angle order can be seen from visual observation of external cues, how the ship's bows are moving in relation to fixed reference points. Also the effect of the order can be seen from displays of information on the bridge, for example a rate of turn indicator, how fast the ship's heading changes is shown by the compass and so on. All the effects of action taken by the de-cision maker, are fed back to him through changes in the information he
re-ceives.
Finally, there are a series of external influences, or environmental factors which will affect the process. These will affect both sides of the diagram. The transfer of information from the source to the decision maker will be affected by the lighting conditions if the information is visual, and the noise levels if the information is auditory. The decision maker's own performance will be affected, for example, by the thermal environment, how hot or cold it is, and how he reacts to these conditions.
2.4. The sites of possible errors in relation to the decision making process
The two areas of error possibilities previously described will be discussed in greater detail, beginning with the "human caused error". If all the informa-tion required to make a particular decision is assumed to be satisfactory, then a number of factors can affect the ability of decision makers to make the correct decision at a given time. The state of some factors are constant and will have a constant effect on this ability, while the state of other factors can change, having varying effects on this ability.
Thus these can be considered together if one uses the ability to make the correct decision at a given time. Training and experience both affect this abil-ity. The greater the training and the more experience a man has, then itis likely that he will be better qualified to make a correct decision. Alterna-tively, his probability of making an error is likely to be lower than that of a less qualified man. It is these factors that affect performance when an officer changes ships. If he is trained and works for a long time on one par-ticular ship, and then is moved to another ship, his previous training may im-prove his performance (positive transfer of training), make no difference (no transfer), or may decrease his performance (negative transfer).
The personality of the individual and its associated factors will affect de-cision making ability. Such factors include intelligence and perceptive abili-ties. The necessity for a certain level of these qualities is fairly obvious. However, this is especially significant in the assessment of the changes of a situation, and coupled with training and experience, fundamental to the formu-lation of strategy. Another personality factor is the degree of extrovertion or introvertion. This factor could affect strategy, and affect one's attitude
to caution. In an encounter situation one may adopt a very cautious attitude and affect collision avoidance procedures very early. Alternatively a less cautious person may choose to adopt those procedures much later in the
en-counter.
The degree to which one's attention is affected by distraction is another in-dividual variable which contributes to one's ability to make a decision at a given time. This is true indeed of parts of the task with's. strong vigilance
component. This is likely to become more important in the future when it seems probably that much more of the ship handling system becomes automated. There-fore, man's role becomes more that of monitoring the system as it is
tamed by machines. There is much work on the subject of vigilance and attention, and no attempt will be made to review it here; however, reference is made to Mackworth (11). The performance of people under stressful situations varies from person to person. This is therefore another variable which will affect the likelihood of an error being made. In a certain situation, some people may re-main calm and not regard the situation as stressful or critical. Others, how-ever, may become excited and nervous. This may depend to some extent on ex-perience, to some extent it will be an individual personality variation, and it can also be affected by training.
Also there are the short-term variables which affect performance. Perhaps the most common of these is fatigue. All too common are situations when officers, especially masters, are called on to make decisions after long hours of duty on the bridge. Also, drunkenness is a factor which will affect performance to varying degrees.
Thus it can be seen that there are a number of factors and variables which will affect the decision making process, and one's ability to make the correct de-cision through this process. Also, considering the mental model concept, it is these factors which will determine how close the mental model constructed is to reality. How well the mental model is used is also dependent on those factors.
The second main group of errors are those which Hooke, quoted in Edwards and
Lees
(4)
described as the "situation caused errors". The factors which affect this category will be considered more closely. The information which the de-cision is based on is of fundamental importance. One can consider the ideal case where all possible information is available in a suitable form to the decision maker, then in this case there is no reason for error through defectivein-formation. If one then compares this ideal state with reality then it is possi-ble to list areas of information defectiveness.
It is possible that none of the information required is available on a cer-tain aspect of a situation. This is the case for example when the decision maker would like to know the intended behaviour of an encountered ship, and this ship has its V.H.F. system switched off.
There may be some information available to the decision maker but there is not enough of it. Thus the first two categories are defects of quantity of
in-formation.
There is the defect of quality of information. Although there is an ample amount of information, it may be not accurate enough to base a decision on with-out the possibility of error.
Although the quantity and quality of information may be satisfactory, the way in which it is presented to the decision maker may be unsatisfactory.
It is possible that the information is not presented clearly enough and it takes the man too long to extract the information and process it. Another aspect of this factor is if the decision maker requires two pieces of information from different sources to use together, and these sources are apart from each other. The man has to go to one display, read and remember the information and take it with him to the second display. Memory aids can be used, for example, writing down the information from the first display. One example of this situation, is when the pelorus on the bridge wing is used, the bearing is read and then often the man has to return to the chart table with this information and use it in conjunction with the information on the chart.
If the information is stored in the short term memory, and if the man has to keep it in the short term memory store for too long, then the store will decay.
The result of this is that either the man tries to remember the information, which could lead to errors, or he goes back and checks it again, leading to time delays. If the decision maker obtains information, not directly from the source, but via another man in the system, then the information may become
torted. Thus the information transmitted by one man may be different from what the receiver understands the information to be. Many factors influence the qua-lity of a communication link such as this. For example, there could be problems of language, clarity of speech and external noise interfering with the communi-cation link. The factor of noise could be noise on the bridge if the information
is transmitted directly, or noise in communication systems, such as telephones, walkie talkies etc.
A more general review of the "situation caused errors" show them to be caused by factors associated with the situation in which the human has to work. This in-cludes many factors.
A first group is the working environment, including the thermal environment, the visual environment, the acoustic environment (the physical environment) and the
chemical environment. These factors can all have significant effects on human performance, and thus on the probability of errors being made. Examples of this area of factors can include the effects of the lighting of the bridge, the light-ing of worklight-ing surfaces. The levels of noise are examples of the acoustic en-vironment. One must consider the levels and intensity of noise, the different types of noise on the bridge and their possible effects on performance. The heating and ventilation of the bridge are factors relating to human comfort and these can also affect performance. The chemical environment is perhaps not such a large factor in the case of the bridge and generally refers to pollutants and toxicants in the environment, which can affect the man and thus indirectly can affect his performance.
A second group of factors to be considered in the question of situation caused errors are those associated with the design of the work space. These factors
in-clude consideration of the operator's position, posture and reach in the design of the work space. Although these factors only indirectly affect performance and thus reliability, their importance should not be underestimated. Indeed these factors are attaining increasing importance with the development of bridge
de-signs
which
are aimed at one-man-control of the vessel from one position. Herethe navigation and control instruments are brought together into a console at which the man is seated from which he can control the ship. An example ofthis
is to be found on the Bore I, a ferry operated by Silja Lines from Stockholm to Finland. MacKay (10) advocates a similar idea which has been developed by the British Navy and compares the bridge design to a pilot's cockpit in anaircraft.
To localize the sphere of factors even further, a third group of factors are those associated with the design of the man/machine interface. These can be sum-marized as the influence on the operator and his decisions of firstly displays,
which can be thought of as the sensory input to the operator, secondly, the in-fluence of the design of controls which can affect the motor output from the operator, and thirdly of display-control compatibility.
This approach of classifying these factors from the viewpoint of the general con-sideration of situation caused errors is fully described in Chapter 2, Applied Ergonomics Handbook (1).
A fourth group of factors are those concerned with the work organization. These factors more directly influence performance and thus indirectly influence the probability of error. Nevertheless this still constitutes a contributory source of possible unreliability. Among these factors, there is length of duty. This is normally four hours for a particular watch duty on the bridge. However, it is by no means uncommon for the master to spend much longer than four hours on duty during periods of restricted water and restricted visibility. Another factor of a similar nature is that where the man in the system has other duties apart from those of ship handling, for example, the chief officer is often responsible for
supervising the loading and unloading of cargo in port. If the ship leaves port at the beginning of his normal watch period, then he begins his watch, possibly
very tired and fatigued. Recently the cruising speed of ships has increased and has brought as a result another problem of this category. Ships cruising across time zones at speeds of 30 knots, have found the time shift to have a stressing effect on the crew, The body has an internal time clock and this adjusts itself to the normal pattern of working. The general problem of matching working routines has been encountered to a large extent in airline
pilots and the problem is generally referred as that of circadian rhythms. If the change in working routine is slow enough then the body can adjust and keep up with these changes, However, now it has been found with the increased
speed of ships that these changes are too fast for the body to adjust to. This means that there is a lag between working routines and bodily routines and this shows itself as fatigue and strain. A possible indirect source of unreliability is the social relationship between the officers on the bridge. A young junior officer may well be reluctant to insist of his point of view in situations where his views and those, say of the captain, differ. This in fact was taken to be a contributing factor in an airplane accident in the U.S.A. some years ago. A junior flight officer was aware of a dangerous situ-ation approaching and after informing the pilot several times, remained silent, and the accident occurred. It is not unconceivable that a similar situation could occur on a ship's bridge.
Thus it can be seen that there are very many factors that will affect either directly or indirectly reliability in the ship handling system, and these cannot all be studied in detail.
3. MODELS AND METHODS
Following the introduction to the factors affecting reliability in the ship handling in Section 1, this section discusses and describes the approach adopted for the study. It is clear that in such a study of reliability one has
to be selective and concentrate on certain aspects of the problems.
3.1. A first methodological approach and the development of an information model
Although the use of Rooke's classification of error into "human caused errors" and "situation caused errors" was primarily intended to be illustrative, it can form a general basis on which to begin selection.
Thus, does one concentrate on the area of reliability concerned with the de-decision making process, (the human caused errors) or the errors concerned with the situation in which the human has to work (the situation caused errors)? The aim of a study such as this is not merely to review and assess the state of reliability in ship handling systems, but moreover to look at ways in which the reliability can be improved. Thus, recommendations must primarily be of practical value and not of purely academic interest. The areas of study chosen must lend themselves to these requirements - that recommenda-tions for improvements can be made, and that they be of a practical nature and can be realistically expected to be applied. This suggests concentration on the situation caused errors rather than the human caused errors. For a study of the ship handling system, it is doubtless necessary to study such a system in operation. Therefore visits to ships during voyages provide the main means of collecting data. However, this imposes the constraint that the techniques used for gathering data must not interfere with the normal running of the ship. Thus, study of the ship handling system in action is basically restricted to observation techniques.
In order to gather data on aspects of the ship handling system, then it is necessary to study the ship handling system in action. This is so that real instances of unreliability can be observed. A simulation of the system in action could not achieve this and indeed the most real situation is the
actual system itself in operation at sea, Therefore the data gathering stage of the study was done on board ship, during a number of study visits,
Another feature of carrying out the data collection onboard is that there is suspicion among the officers who are part of the ship-handling system as to the purpose of researchers working on board their ship. This is the result of a fear that criticism will be made of the way in which they work. This, in fact, has generally been the case during ship visits throughout the data col-lection period. Thus, to allay these feelings, and also from common courtesy, the purpose of the study has first to be explained in detail to the officers concerned. Again this points to a study of situation caused errors, as it is much simpler to explain that no personal criticism will result from the visit than if the study was of human errors. The good-will and cooperation of the men is essential to obtain the maximum amount of information from a ship visit. Thus from a social position, study of situation caused errors is preferable. It is most difficult in this social environment to study the man as a de-cision maker and errors associated with the dede-cision making process. It is much more suitable to study the system as a whole and man as a component part
of this system.
Borg (2), in work on flight safety, questions whether it is necessary to col-lect data on human caused errors as they are very difficult to do anyting about. He also agrees with the point previously made by saying that an attempt to collect data on human caused errors may cause suspicion and a negative attitude to such a study.
Thus, whereas a study of human caused errors could result in some improvements, for example recommendations relating to selection and training, it is likely that a greater and quicker improvement
in
reliability could result from study of situation caused errors. An improvement in reliability would only occur when the recommendations were actually implemented, and thus it is likely that shipping companies will be more attracted to applying the recommendations if the benefits of doing so are more obvious to them. Again, improvements re-sulting from reduction of situation caused errors are more likely to achievethis.
Finally, the ratio of human caused errors to situation caused errors given by Juran (20% H.C.E. and 80% S.C.E.) provide numerical weight for a study of S.C.E. A study aimed at reducing 80% of the errors is likely to improve overall re-liability much more than a study aimed at reducing 20% of the errors, The evi-dence quoted by Pitanova in this context will be discussed later.
Therefore it was decided to place the emphasis of the study on situation caused
errors.
To begin with, the approach was adopted of trying to carry out a complete logging of the functioning of the ship handling system. This was in order to obtain a series of complete records of the system in action on the different ships visited. As was said previously, one is limited to observation techniques when studying the system in action so that one does not interfere with the work of the officers. While most officers were happy to tolerate the presence
of the investigators on the bridge, some of them felt that even this was too
much an intrusion into their normal working routines. The observations were recorded by a combination of logging on paper, and also using taperecorded
commentary. It was intended to log all activities of the personnel active at the time on the bridge in the ship handling system. In addition, all the levant details of the system which would initiate courses of action were re-corded, such as the encounter of other ships, their behaviour and also constant reference to chart positions and manoeuvring restrictions. By this means it was hoped to compile a complete record of the ship handling system in operation in order that it could be analyzed later.
The analysis was to be such that causes of unreliability could be isolated and examined later in more detail from the data collected. Also, direct comparison between ships of differing size and function, and with different handling sys-tems could be made, and any differences found examined from the point of view of system reliability, In addition to this data logging, it was considered very important to record the views and subjective opinions of the men in the system, namely the bridge officers. Although there are certain limitations of sub-jective data of this kind, there is much very relevant information to be ob-tained from the opinions and experiences of those concerned with the handling of ships. Also there are factors which are important in their effects, though may not occur very often, and may well be missed if one was to rely purely on data logging. Also instances of critical situations could be discussed and the reasons for the situations becoming critical examined. To this end a
loosely structured interview was developed. The basis of the interview was a discussion of different functional groups of the equipment on the bridge. This was discussed in relation to the criteria of
contribution to safety efficiency
ease of use and comfort.
As the interview was loosely structured, digressions could be made to discuss different aspects of reliability and factors affecting these.
The areas listed for discussion during the interview were as follows:
Communication equipment and communication systems. These were dealt with in three parts
1/ within ship communication, 2/ ship/ship communication, 3/ ship/shore communication.
Equipment concerned with the control of the ship.
Radar.
Navigational aids (traditional methods, Decca systems, satellite systems, etc).
Other information which has to be monitored from the bridge, though not directly related to the ship handling system.
Collision avoidance, and critical incidents.
General discussion on automation and its effects.
A preliminary analysis was made of the data collected after the first ship visit. The analysis showed this approach to the study to be unsatisfactory from the point of view of the data collected. The primary reason was that it was not possible to record sufficient detail of information from the observa-tion studies on the bridge. If an approach using complete logging of activity is to be used, then a full, very detailed record must be made if causes of errors are to be deduced at a later date using the recorded data. A less de-tailed record is totally unsatisfactory from this point of view, as one cannot
return to recorded situations of interest and extract more detail.
In periods of high activity on the bridge the situation was obviously more difficult, and it was found on analysis that attempts to record some activities in sufficient detail resulted in other activities being missed (or not re-corded) altogether.
Also it was found to be very difficult for the
observers to maintain the high level of recording with respect to the required detail for long periods. This was not a problem of motivation on the part of the observers. It is very dif-ficult to maintain man's attention
and concentration at high levels for
tended periods of time. However, as previously stated, data of this kind which is not of sufficient detail is not worth very much at all, and is not really worth collecting.
Attempts on the second ship visit to increase the quality of data recorded by alteration of the methods adopted still failed to provide adequate detail of information. Nevertheless, the information obtained from the interviews made during these visits provided much useful information. At this stage it was decided to review the approach to the study. It was evident that more selec-tivity had to be applied and a more specific approach adopted to the study of reliability.
If one simplifies figure 1 it can be seen to be composed of four basic flows.
(1)
(2)
(3)
(4)
Information - Assessment of state - Prediction - Determination
The last three stages (2 - 4) depend on the activity of the decision maker, and errors made in these stages are going to be errors in the decision making process. The stages 1 - 2 involve the flow of information from the various
sources to the decision maker to enable him to assess the situation.
Thus, still maintaining the original decision to restrict the study to situ-ation caused errors, this flow of informsitu-ation can be studied in detail. As stated previously, the information is fundamental to decision making and errors resulting from defects in the information, and the flow of information from the source or sources of information to the decision maker, are them-selves fundamental to the reliability of the system. The possible causes of defects in information and the flow of information were discussed in general in Section 2.4. Thus by considering the information used in the system in these terms it provides an important area of reliability which is compatible for study in accordance with the restrictions previously discussed. This type of study is also simpler to carry out as regards the mentioned restrictions in carrying out the data collection on board ships. Firstly, the study can be made using a combination of direct observation techniques and interviews. These methods will be discussed in detail later. Also, the contents and object of the study can be described fairly simply to officers on board the ship visited, in a way that implies that no criticism will be made of the way they
work.
As previously discussed, there can be variations in the nature of the informa-tion itself with respect to quantity and quality. There is also the possibility of defectiveness in the flow of information from the source to the decision maker. In the functional analysis of ship control, five basic areas of in-formation are considered. This can be seen from the functional diagram
(figure 1).
These five categories are as follows;
Information concerning the state of the own ship. This corresponds to the top left hand corner of figure 1. This includes information about the position, speed, course, rate of turn, etc.
Information concerning the state of external disturbances. These are in-fluences on the ship which cannot be affected by man, but will influence the ship. These include wind, tide, currents, etc. (Bottom left of figure 1).
Information concerning the state of surrounding traffic (other ships) This information includes knowledge of the range, bearing and speed of other ships.
(Top right of figure 1).
Information concerning the state of manoeuvring restrictions (Bottom right of figure 1).
Information which will influence the ship handling system, in an indirect way. This category includes, for example, owner requirements with respect to
