5 JAN. 192
ARC
National Maritime Institute
A Study of the Effect of Human
Factors on the Handling of Two
Ship Models
by
I W Dand
N M I R 112
August 1981
National Maritime Institute
Feitham
Middlesex TW14 OLQ
Tel:01-977 0933 Telex:263118
Lab.
v.
Scheepshouwbntk
Technische Hogeschool
Deift
This report is Crown Copyright but may be freely reproduced for all
purposes other than advertising providing the source is acknowledged..
NATIONAL MARITIME INSTITUTE
A STUDY OF TUE EFFECT OF HUMAN FACTORS ON ThE HANDLING OF TWO SHIP MODELS
by
IWDAND
SUMMARY
An investigation intq the way in which human operators control a ship model on a straight course is described and the results presented and discussed. Some experiments were carried out in a simulated cross-wind and a comparison.is made between calculated equilibrium drift and rudder angles and those actually used by the operators. Various parameters relating to controllability, (with special reference to controllability in port. approach channels)
are presented, and overall values for navigation lane widths both with and without cross-winds are deduced.
IBthIotheek van
Mde!in Scheepsbaw- en Schepvartkunde
TeJnische HoascIoo, Ddft
DOCUMENTATIE J:
4S8
Introduction 1
Definitions '2
The Experiments 4
3.1 General 4
3.2 Experimental Arrangements 6
3.2.1 Circulating Water Channel 6
3.2.2 Control Position 6
3.23 SimulatedView
7 3.2.4 Experimental Method 7 3.3 Human Factors 8 3.3.1 Choice of Helmsman 3.3.2 Contracted Timescale 9 3.3.3 Perception of Motion 93.4 The Ship Models 10
3.5 Instrumentation Ii
3.5.1 Model-based Instrumentation 11 3.5.2 Shore-based Instrumentation 12
3.6 Data Handling and Analysis 13
Results Obtained with. No Wind-Basic Controllability 14
4.1 General 14
4.2. Effect of Ship Type 15
4.3 Effect of Depth/Draught Ratio at COnstant Speed 17
4.4 Effect of Speed at Constant Depth 17
4.5 Effect of Trim 19
4.6 Effect of Helmsman 21
5.' Controllability in a Cross-Wind 22 5.1 Calculation of Equilibrium Drift & Rudde± Angles 22
5.2 Clculation of Wind Force and Moment 23
5.3 Experimental Results 24
General Discussion
6.1 Rudder Activity and Equiiibrium'Ru4der Angles 26
6.2 Lane Widths 27 Conclusions ' '' 30 References , 31 Acknowledgements 32 Nomenclature Figures I to 19 Plates I to 4
A STUDY OF THE EFFECT OF HU}AN FACTORS ON THE HANDLING OF TWO SHIP MODELS
PROJECT 252001
1. IntroductionIn the. design of waterways for safe navigation it is important to allow for interaction between passing ships (ref. 1) and to ensure that the banks are
sufficiently remote from the course of the ship so that bank effects (ref. 2) do not unduly affect handling. The. separate effects of ship-ship interaction and bank effects can be used in conjunction with suitable safety criteria to.determine the width of a navigation channel.
But there is a further factor that. must be taken into account in determining adequate channel width. This concerns the width of waterway or 'lane' required to control the ship along a prescribed course in calm water of a given depth in a given wind and current. It might be supposed that for a prescribed straight course in water of a constant depth in .the absence of wind or current, such a lane would have a width equal to the breadth of the ship. That this is not so was acknowledged in the study described in reference 3 in which a lane width of 1.8 times the breadth of the ship was suggested to allow for the effect of 'piloted controllability' (ref. 4).
The additional width required arises from the fact that one or more human
beings are included in the control loop of a ship which is attempting to follow a given course. The inability of the human to perceive certain changes in say
the heading of the ship may allow a drift angle or rate of turn to develop which must be controlled to maintain a given course. The control method used by the human pilot and/or helmsman will often be of a sophisticated
'trial and error' type in which the rate of turn, once perceived will be
alteredby a 'trial' amount of rudder angle until the 'error' in the course
is once again perceived. The process is complicated by various time lags within the system due to which the response of the ship to the rudder and/or engines may be very sluggish. In such.a case a significant amount of anti-c.ipation is required of the human ship handler.
If a ship is moving in calm water, with no external disturbing inflUences such as wind and current, why should it initially depart from its course at all? The reasons for such departures may be
- in shallow water regular eddy-shedding from the stern of bluff-bodied ships may occur. This vortex-street can provide significant perturbations to the ship
- the helmsman may not be able to attain exactly the correct drift angle to counter screw bias in a single-screw ship. Any slight error in drift angle could cause large sway forces and yaw moments to arise.
- the steering gear may have a bias to one side, or the rudder may be slightly asymmetrical
- the ship may have some directional instability which may be larger at some water depths than others
The subjèctof piloted controllability has been explored at NNI using two
large free-running ship models in a circulating water channel. This study, which was of necessity limited in its scope, addressed the problem of a ship moving on a straight course in shallow water of constant depth controlled at constant forward speed by a helmsman only. The case of apilot/helmsman
combination in which the helmsman responds to orders given by the pilot has not been studied.
To supplement the piloted controllability experiments, some measurements were made of the performance of human operators when the ship models were subjected to steady cross-winds of various velocities. The purpose of these experiments was to allow a direct comparison between the calculated and measured drift and rudder angles for given wind strengths; this allowed an assessment of the accuracy of lane widths calculated for wind from steady-state considerations as in ref. 5.
2. Definitions
Before describing the study in detail, it is important to define some parameters and concepts used therein. These are illustrated iti Figure 1 together with the axis system used. The main parameters are:
In this report a distinction is drawn between manoeuvrability and controllability. A ship that is of such a shape and size with a power unit and control surfaces which may make it 'manoeuvrable' (at least at the model test stage) may or may not be 'controllable' for a variety of reasons. For ecaxnple the rudder
motor may respond unduly slowly, the helmsman's view may be restricted, the
response of the engines may be slow or there may be an absence of aids which help the pilot or helmsman to determine the response of the ship to his control inputs. We therefore consider the term 'controllability' to include all
aspects of ship-handling related to manoeuvring a ship along a given course, with the essential presence of the human operator in the control loop.
Lane width, w
This is defined as the width of track swept out by the ship. More precisely
it is a measure of the width of the envelope of the swept path of the ship as it attempts to follow a prescribed target course.
Course Error y
This is defined as the difference between the mean path of the centre of gravity of the ship and the 'target' course. It is a measure of the accuracy with which the ship can be positioned in a waterway for a given set of conditions.
Basic Controllability
This is a measure of both the handling characteristics of the ship and the skill of its pilot/helmsman. It is loosely defined by means of the lane width,
course error and rudder activity as the pilot attempts to maintain a course -along the 'hydraulic' centreline of a waterway in the absence of wind, current
and interaction effects from side-banks or other ships.
-Rudder 4ctivity
Rudder activity may be measured by means of the number of rudder commands given, or the zero-crossing frequency of the rudder angle trace. The zero-crossing frequency is of little use howeverif the .rudder carries some constant mean offset as may be required when handling a ship in a cross-wind. In such
cases the 'offset crossing frequency, f' is of more use as it gives a measure
of the rudder activity about a mean level. In this report f is used as a convenient comparative measure of rudder activity.
Equilibrium Rudder Angle
This is the. rudder. angle required in a steady-state situation to balance any
external forces and moments imposed on the ship. It is taken as the average value of rudder angle obtained from rudde angle time hjstOries for each run.
Equilibrium Drift A±igle
This is the drift angle required in a steady state siti:iation to balance any
external forces arid moments imposed on the ship. It is usual for non-zer equilibrium ruder and drift angles to occur simi.iltaneously.
Other Patameters
Histograms of lane width and course errors obtained during a meastreent run
can yield information regarding the course-keeping ability of the helmsman as indeed can the mean value and standard deviation of these quantities.
Other measures of course-keeping which were obtained during the investigation described below are the crossing frequency of the model's centre of gravity about the mean track and the crossing fr.equenc g, of the heading of the
model about a mean value. These measures gafe an indication of the sway and yaw oscillations resulting from a given helmsman's performatice during a run.
3. The Experiments
3.1 General
There are several methods of studyitig ship-handling in relation to waterway design all with certain advantages and disadvantages. The obvious method wOuld
entail full-scale, measurements on a ship in a waterway with pilots and helmsmen behaving in a real environment. This has the disadvantages that there is no control over the environment during the experiment, the collection of a suitably large quantity of data is both time-consuming and expensive, and some of the results obtained tend to be specific rather than general.
Control over both the environment and ship behaviour is obtained when experiments are carried out with a ship simulator. The .use.of simulators fOr such a purpose is well-documented (see for example references 6 o 8) and
has than advantages for the study Of manlmachine interface problems, Uowever
A useful discussion of the techniques mentioned above is given in ref. 9.
-5
again, collection of an adequate quantity Of data cat be time-consuming
the behaviour Of the simulator is only as good as the mathematical mode.l
used to represent the hydrodynamic behaviour of' the ship. Variations of certain parameters such as tr.im,and hydrodynamic effects suh as eddy-shedding 'i-n shallow water are not easily simulated and i-f they havenot been included in
the simulation model, they cannot be studied.
The problem of ollect-ing a sufficient quantity of data rapidly under
controlled conditions my be 'overcome by uSing large free-running manned ship models under the control of human operators. - Carrying out such an experimant
in a facility such as a irculatirig water channel has the added advantage that very long run times may be obtaine4 as the experimenter is not limited by a tank of finite length. Changes in the condition of the model such as its draught- and at-rest trim are made easily, and the characteristics of the flow around the hull are represented reasonably accurately provided the models are
large enough to avoid serious scale effects.
-The: final methO4 of studying aspects of ship-handling uses smaller free-running radio-contr011ed ship models In this case the htan operator is remote from the ,model and his.view of the model's behaviour isuSually. obtained froman elevated
position, no attempt being made to incorporate him 'within'
the eperiment, but rather to have him asà remote external observer and
controller Small-scale models have an advantage that they. can readily be
-, incorporated in phySical dels of, say, a port approach channel or other
waterway so that handling problems due to the. geometry of the waterway ay be
studied..
But 'any model study involving human operators suffers from a major disadvantage.
As 'roude scaling is generally used in such studies, the time-scale of the model is contracted by the square root of the scale factor compared to full-scale time. Therefore the model responds. ore rapdily than does its full-scale
couLterpart, and the human operator must respond in this cpntacted time-scale.
This study used a variation of the manned model technique in conjunction with the NMI large nber 2circi4ating ,wat,,er channel.. The channel, was used in
shallow water mode only althoughboth watar depth and water speed while
remaining constant for each run, could be varied so that their effect on handling coUld be studied.
The channel has a rectangular cr ss-section with a :flat bottom. For the experithénts described here no side-banks (other than the vertical sides of
the channel) were simulated. This is possible however and limited experiments have been carried out with a sipping surface-piercing bank to one side of the
ship model. ' ' . '
The shi model in a circulati-ng water channel is normally held stationary
relative to the channel structure while the water flows past. In ship-handling experiments tle mo4el is f,ite4 with ll,7-rcontrollable motor and rudder so that the experimenter can'adjust the propeller revolutions until the model is propelling itself at model seU-propulsion point into the'' orcoming flO, the resulting 'balancedt situation caus,jng the model to renlain stationary
relative to the:charinei-'structur.e.
3.2.2 Control Position
The helmsman bad control over the rudder onLy while a separate member of the experiment team controlled motor speed to keep the model in its 'stationary'
position.' ,, ' '
The helmsman's view was provided by a 'Pt monitor linked to.a CCTV camera
mounted in the thodel with its lens at tle scale height-of-eye position of the
helmsman on the ship , '.,
- The arrangements for the helmsman are shown in Plate I while in plate 2 is
shown one of the shipmodels the Su,dy on which the CCTV camera can
It is clear from plate I that the TV picture roughly represented the view of a narrow channel at night through one bridge witidow. Various unwanted
reflections from the channel structure were removed by means of black poly-thene sheet and the target course was indicated either'by a centreline marker or by pairs of 'buoys' receding into the distance. The buoys were
simulated by table-tennis balls suspended just above the water surface by black thread. The thread was invisible in the TV picture an4 the buoys were high-lighted by means of photographic lamps.
The bows of the model were visible in the lower part of the TV picture and a
steering post was provided on the forepeak of the model under test.
3.2.4 Experimental Method
Each run lasted. 6 minutes in 'model' time which corresponded to 42 or 44 minutes
full-scale depending on the ship model used.
Each opErator toOk over control of the model from his predecessor and was allowed a few minutEs to regain complete control after this hand-over, period
which was often, carried out without stopping the water flow in the channel.
Once the Operator was satisfied, the measurement period began and the operator attempte4 to maintain a course,as close as possible to the .target course, by
direct use of the rudder alone. All control was c'arried out using visual cues derived from the TV monitor alone and, although the operator could, if necessary, look down on the. model, this was not done during a measurement period.
Starting and stopping the water flow in the channel were operations requiring some care. The model itself was held at rest by guides on the instrument carriage which engaged with pairs of rollers on the model. The model was aground on the floor of the channel while at rest and, once the water flow in the channel was started, the water level above the false floor would rise to
its running level and the model would float. The model motors were then started and revolutions increased until the model self-propulsion point was
reached. The instrument carriage with its guides was then raised clear of the
and steady the model while the carriage guides were re-engaged.
Light safety lines were attached to the. thodel to restrain it i-n the vent of a major control failure.
3.3 Huan Factors
While it was not the purpose of the study to. deal in detail with. behavioural
aspects of the human helmsmen it seems appropriate to mention one or two points which may have a bearing on the results obtained.
3.3.1 Choice of Helmsmen
Several helmsmen were used in the experiments with various degrees of experience in ship-handling. They were drawn from Nff staff and were supplemented for some tests by three practising pilots. The NMI staff experience in handling ship
models varied from a negligible to an appreciable amount and two membrs of the
experiment team (operators A and B) who were experienced operators of ship
models took part in all experiments. Their results formed a basis for couiparisoi with other experiments and other operators. Operator A was approximately twice as old as operator B and whereas operator B was of a more
phlegmatic nature, operator A was less reserved. These
-differences were to become apparent in.the way they handled the models and in
the amount of learning time required in a new situation.
The three pilots had experience of ship handling from thréé distinct areas. These were
Pilot A: experience of handling sea-going ships in the confined waters of a ship canal
Pilot B: experience of handling large ships in approach channels to a port on a large river
Pilot C: experience of handling large ships in the seaward nd estuarial approaches to a large port
The pilots therefore brought to the experiments a wide range of experience and not only did they participate in the experiments, their views were also sQught on the validity of the model technique use4.
The scale of the models used was such that time was effectively 'speeded-up' by a factor of about 7 in relation to full-scale. The response of the models to a given control input was therefore seven times faster than would be the case
on the ship and as a result the helmsman had to react seven times faster than he would on the ship.
The effect of this on the measured handling behaviour of the model and its extrapolation to full scale. is not easy to assess. Studies that have been carried out on the subject are not entirely conclusive. For example in reference. 10 some doubt is cast on the usefulness of contracted timescales for
training purposes while in reference 11 it is stated that "experience and insight gained at a contracte& timescale ...can be used in real time."
In view of this apparent uncertainty as to the effect of contracted timescales on handling results, the views of the pilots were sought regarding the 'feel' of the models.. In their view the models behayed, from a handling point of
view, as they would have expected from their professional knowledge of similar ships. This, coupled with the fact that models have been used with success
in waterway design in the past, gave some confidence in the use of this experimental technique. However it was accepted that comparison with full-sëale ship or simulator results should be carried out to check the conclusions drawn from this study.
Nevertheless it was felt that the contracted timescale would' assume less
importance if the results were used in a comparative manner to assess the effect on handling (and hence lane width) of changes in the main variables of the experiment such as ship type, speed, water depth, trim, helmsman etc.
3.3.3 Perception of Motion
Twb aspects of the view in the TV monitor might have affected the perception of motion and hence the performance of the helmsmen. The first concerns the fact that the CCTV camera provides a monocular rather than a binocular view of the outside world. The. use of a TV camera rather than a han being on the model
overcomes the disadvantage that the eyes of the human being are too far apart to scale so that his binocular vision gives him a non-scale perception of depth. This is often overcome by covering one eye of the human operator in such a situation which in effect reverts to the monocular view of the TV camera.
It is not felt therefore that use of: a TV camera in this case is a disadvantage
and it is asted that the results obtained were not affected by its use.
It has been mentioned above that buoys i-n the visual scene were simulated by
table-tennis balls Supended close to the watet surface. They were therefore stationary in the visual cene and did nt move 'towards' the helmsman, who was
therefore deprived of perception of motion in depth arising from elements of the visual field flowing radially away from some focal point This particular visual cue would seem tO be of more importance in the perception of rapidly-varying motiotis (ref 12) and would therefore be essential in, say, an aircraft simulator. However it was generally felt b the pilots who took part in the tests that in a real situation little visual information regarding motion in depth is obtained from the motion of buoys or other marks in the visual scene due to the cornparatively low velocities involved. Indeed it is the. lack of
information Of this type in real life that makes the shiphandler's task more
difficult
3.4 The Ship Models
Two ship models were uSed in the study, chosen to. represent two widely differing ship types with different handling characteristics. One, model 5098, had fine lines and a twin-screw, s.ingle rudder stern, arrangement; it is shown in plate 2'
The other, model. 5171, was.typicäl of a single scre', single rudder VLCC and is
shown in plate 3.
The principal particulars of the two models areshown in tabie 1 while the body plais are shown in Figure 2.
Both models were self-propelled, powered by 1 75W printed-circuit t' electric
mOtors. - ThE. tin shafts of model 5098 were. geared together and driven by a
single motor; this arrangement ensured. synchronisation of both shafts.
It may be noted from plates 2ànd. 3 that the wheelhouse position of the twin screw model was assted to be well 'forward while that of the single-screw model
was assted to be right aft as Shown by the CCTV camera positions.
Both models were run at load draught only,. and model 5171 was also run with head
Model Number
Length between perpendiculars
Breadth, moulded B
mld
Draught T
Trims
Displacement volume at draught T V
Lpp/Bld
Lpp/Tid
Block coefficient Displacement/length ratio Number, of rudders Rudder type Rudder area/(L T)Rudder angle range Number of propellers Propeller number Number of blades Diameter, D Mean face pitch,
P/D
Blade area ratio Ship/model scale CB 1
00V/L3
P 3m
5098 5171 6.096 6.245 0.723 0.979 0.249 0.377level level, -1.447.,+I.28%
0.597 i 1.934 8.432 6.379 24.482 16.565 0.544 0.839 0.264 0.794 I I
gnomon balanced, closed stern 0.0153 0.0155 degs ±70 ±35 2
Li
13461 13480 6LH 6RB 6 0.131 0.169 m 0.149 0.169 1.136 0.697 0.922 0.592 1/44.25 1/53 'TABLE I 3.5 InstrumentationThe measuring and control instrumentation used in the experiments is con-veniently described under two headings: that installed in the'model itself
-and that based on the experiment facility.
3.5.1 MO6el-based Instrumentation
Propeller revolutions and 'rudder movement were controlled by radio using the
standard NXL 27 MHz system. This generally worked well in an -environment
that was less-than-perfect for radio control. Some radio drop-outs occurred but often this was found to be due to the high humidity affecting the radio
receiver in the model. This problem was cured by removing the receiver after running had finished and keeping it in a room at normal ambient temperature and humidity. It also was important to maintain the receiver batteries at a high state of charge during the experiments', so every
opportunity was taken to maintain the charge on these batteries throughout a day's running.
A continuous trace of the rudder angle time history was obtained on a two-channel DC pen recorder mounted in the mOdel and visible in plates 2 and 3.
For experiments in which.a cross-wind was simulated, model 5171 was fitted with two air-fans mounted on the model at bow and stern used as air thrusters. These directed their thrust in a lateral direction and the force thus arising was used to simulate the sideforce and turning moment due to a cross-wind. Wind tunnel data from reference 13 were used to deduce the magnitude of the
fan thrusts, the speeds of the fans being controlled by a radio link separate from that used to control rudder and propeller revolutions. The
appropriate control voltages at the fans, calculated from wind tunnel data and the fan calibrations, were monitored by means of digital voltmetes
carried on the model and visible from the experiment carriage.
A diagram of the model-based instrumentatioti is shown in Figure 3.
3.5.2 Shore-based Instrumentation
The television monitor and radio control console/transmitter have been mentioned above, and are shown in Plate 1. They were situated on the experiment carriage, above and to one side of the ship model. Rudder angle, propeller revolutions and the on-board pen recorder were all controlled from this central position; the wind fans .wete controlled from a second console, remote from the first so
that wind effects could be changed without the helmsman being aware of the fact. until the heading of the model changed in his TV picture.
The other major item of shore-based equipment was a photographic arrangement used to track the position of the model in the horizontal plane. For this
two targets were mounted on the model centreline a known distance apart, these being: subsequently photographed at known time intervals by a motorised
Canon Fl camera mounted on the channel structure directly above the mpdel. The targets mounted in model 5171 can be. seen clearly in Plate 3.
Unfortunately the field-of-view of the camera was limited due to structural 'members of the channel and carriage so that the long-itudial position of the
model had to be close-ly controlled to prevent the targets from disappearing from the field-of--view of the camera.
13
-The centreline of the channel was marked on each tracking photograph by means of a tensioned white cord mounted :above the model and in the centre-plane of the channel. it can also be seen in plate 3. Run numbers. for
each run were. recorded by means of a small board which appeared at the bottom of each photograph. Typical tracking photographs for each model are
shown in ilate 4.
The time lapse on the camera was set at O seconds (about 70 seconds full-scale) and 36 exposures were obtained for each run. The photographs were subsequently digitised using a CETEC pencil-follower.
A diagram of the shore-based instrumentation is shown in Fig. 3.
3.6 Data Handling and Analysis
Rudder angle time histories were digitised using a CETEC pencil-follower and analysed using a FORTRAN computer program written for the purpose. The results obtained comprised a listing of sampled rudder angles against time,
ZCr0 crossingfrequency, 'offset' crossing frequency, mean rudder angle and its standard deviation, mean absolute rudder gnge and histograms of rudder
activity.
The tracking photographs were analysed after digitisation using a second FORTRAN program. Four points from each photograph, two defining the channel centreline and one Qtl each target., were used to deduce the lateral deviation
of the midships position from the centreline target course. Drift angle was computed readily as was the sway velocity v and the angular velocity r from
v = - y(i-1)J/(2 St) ... (1)
r. (i+1) - (i-1)/(2 t) ,.. (2)
where t is the time lapse set on the tracking camera.
lane widths were calculated frOm each photograph. One, w, was simply the projected width of the yawed model. This was obtained by converting each of a table of waterplane offsets (representing the maximum local breadths of the model) to a new offset in the space axis system by means of the trans-formation
y./cos
+ . . .
... (3)
where y. is the local half-breadth at a longitudinal position x from midships.
were. calculated
and
were then sorted in ascending order. of magnitude. Thefirst and last elements'bf theresulting sorted array when added together gave the projcted width of the model at a drift angle . This lane. width was .then non-dimensionalised with respect to the moulded breadth B.
The second lane width, w1, took into account the, lateral excursion of the model from the centreline of the channel. The assumption was then made that
the maximum halfwidth of thenavigation lane .at time t was given by
(w1)
Gt + (4)
This provided an envelope of extreme lne 'widths., etical about thE channel
centreline, with the tacit assumption that an excursion to one side of the centreline was. equally likely to the other Side. Clearly, if the channel were curved and not straight as assumed throughout this report, such a
definition would not apply and an overall envelope of the swept ttacks would be àppropriäte.
The results obtained from each tracking, hptog.raph were theti listed to provide
a time history and a suary of mEans and standard deviations were giyen of
y !;
, v/B and w1/B together with maxim values of the same paraieters. The computer output contin d with the zero crossing frequencies and offset crossing frequencies of' the model CC and , and ended with histograms of the following parameters:TABLE 2
Computer plàts of histograms of thE' above paiameers for two runs are
shown
in Figure 4.4. Results Obtained with No Wind - Basic Controllability
4.1 General
-Presentation of results from studies in<'olving the performance of human
operators is not easy due to the complex response of' the human in a given contrO situation. Several alternative methods are used in reference 14 .inwhich
Parameter Class Interval From To
YGIB
'' 0.05,
.5 0.5w/B
001
' ' 1.0 1.215
-ieults obtained on research aircraft simulators are discussed In this paper
a
simple approach is adopted in which mean values of the. major paraeters are presented together with, where appropriate, the deviation of the sampled data about the mean. indicated by one standard deviation.Results are shown for operators A and.B and .for pilots A, B and C.. Although results with other operators were obtained, - thefive chosen had t.he most
experience of handling ships or models and theirperformance was therefore
considered to be the most reliable f or comparative purposes.
All basic controllability results .for these' operators are shown in Figures 5
to 7 in which the major parameters w1/B, mean rudder angle, offset rudder
crossing frequency f0, yc/B and offset yaw angle crossing frequencyg are presented.. Relevant test conditions are indicated above each run, the run
number appropriate to the test. being shown.on each experiment point.
It is seen that in general two water depth/at rest draught. (hIT) ratios were used, the relevant values being 1.04 and 1.33. Slight variations occurred when model 5171 was run with a trim by the. head or. stern in which case the smaller
hIT value was increased to allow clearance between the keel of the model a.n4 the floor of the channel. .
For each model; two speeds were used at an h/T of. 1 .33 and one speed only at an h/T of 1.04. In spite of the fact that different Froude Depth Numbers are 'shown in Figures 5 to 7, the lower speed at: h/T=1 .33 was the same as the speed
used at h/T=i.04 This pea4 correspon.ded to a full_scalevalue of 5.2 knots
formodel 5098,and5.6 knotsfor model 5171. The higher speed corresponded to.7.8 knots for model 5098 and 8.5 knots for model 5171.
4..2 ..Ef-fecof$hipype
In F.gure. 8 are plotted the mean values of w1/B (the most useful parameter for
water-way width design) against the 1u1l parameter -IOOV/L3 for. various h/T
and helmsmen..' All results were Obtained-at the lower of the two speeds used
:arid show that, the differences between the two mo4els did not give rise to
major differences in the lane width achieved. This may be because the potential handling. diff-iculties due-to the large mass of the. single screw
model were compensated by the better control achieved through the single-screw/ single rudder combination. The smaller mass and finer form of the twin screw model were not perhaps as beneficial in its handling due to the twin screw!
single rudder combination in.which the rudder does not receive the benfit
of the screw race until sufficiently large angles are attained
Neither model appeared to be directionally unstable in the experiments but it was noted that it was unwise to allow the single-screw model, to deviate far
from its course because long-period, slowly-damped oscillations occurred in its track as attempts were made to regain the target course. These oscillations could be controlled with appropriate 'check rudder' but demanded.a good deal
of anticipation on the part of the helmsn due to the slow response of the
single-screw model to control actions. Similar oscillations were more easily controlled on the-less-massiVe twin-screw model.
The different longitudinal camera positions on the two models also bad an influence on. the handling .'fèel' of the model due' to different perceptions
of the del's position dependingon whether the camera was befqre or abaft
midships. When the camera was mounted forward as in model 5098, it was difficult to determine the position Of the sterü and hence the alignment of
the model in the channel. This was of coure agravated by the inability of the he-lsmän to look astetn and- thereby to detrmine the position of the Stern.
It was therefore possible to position the model. in such a ay that the
steering post and bow image appeared to be on cOurse, while the stern was in fact to one side of the intended track. This may account for the slightly
greafer departures OfyG/B from zero for model 5098 apparent in Figures 5 to 7.
When the camera was unted right aft as in odel. 5171 ,
perception of the behaviour of the stern wa
donsiderably iroved.
This wasfacilitated by visual cues in the TV pictu±e which allowed the helsan to
perceive lateral movement of the stern when otherwise the model appeared to be steady and on course. Lateral movements of the stern were quite noticeable on this model, emphasising its behaviour in response to rudder actions in which the initial excursiOn Of the stern in response to the rudder appeared to be greater than that Of the bow. This may be dueto the fact that the !pivot point' fOr small tiansient cOurse changes remainS narer to the bow than to the
stern, even in shallow water. It- should be noted however that for steady
turning in shallow water it has béèn ShO in reference 15 that the pivot point
17
-4.3 Effect of Depth/Draught Ratio at Constant Speed
The influence of depth/draught ratio on handling as reflected in mean values of w1/B can also be seen in Figure 8 where it is seen that in general larger lane widths were obtained with both models at the shallower depth.. This is shown more clearly in Figure 9 where the results of Figure 8 are replotted. The tentative trend lines shown in Figure 9 clearly indicate the reduction in mean w1/B for all helmsmen as h/T was increased.
This result may seem to be at variance with the general result that the
directional stability of a ship improves in very shallow water (see for example ref. 15). It was in fact this very improvement in directional stability
associated with the probably large increase in lateral virtual mass and lateral drag which could have caused the results shown in Figure 9; any
deviation from course was more difficult to correct in the shallower water due to the models' preference forcontinuing on its set course. Once the helmsman
was able toget the model to turn it was then very easy to overshoot and begin
an oscillatory track. In the deeper water the comparative reduction in
directional stability caused the model to be more responsive to the helm and
apparently easier to control.
The 'wild' spot (marked * in figures 6 and 9) from operator B with model 5171 is difficult to explain. This was the first run from this operator at this depth and speed with model 5171 and it is conceivable that insufficient
'training' time was allowed before the measurement run was started. It
appeared that this operator did require a longer time to acelimatise himself to a new set of conditions compared to other helmsmen (see section 4.6 below).
4.4 Effect of Speed at Constant Depth
The effects of speed on handling at a constant depth/draught ratio of I .33 are
shown in Figure 10 and 11. Unfortunately it was not possible to obtain results with the pilots to measure this effect so that only results from operators A and B are shown.
In Figure 10 the effect of speed
n mean lane width is shown where it is seen
that in general there was a slight trendfor the mean wi/B to increase with
increase of speed.
The exception to .this was whn operator B controlled the
single screw model in which case a slight reduction in mean
lane width
occurred.
The offset crossing frequency f
of the rudder angle time, history is s1own in
igüre 11.
It is interesting to note that rudder activity did not increase
significantly with speed when either operator handled,, the twin Screw mOdel 5098,
but both operators recorded significantly increased rudder activity at the
higher speed with the single screw model 5171
This may be associated with a similar phenqmenon noted'by Bindel 'in reference 16.
In experiments with controlled models in canals described int'hisreference,
handling problems were encouitered at certain speeds in channels of a given
blockage ratio (=B.T./Ac where Ac is the ciannel cross-sectcn area).
It was
suggested that there might be a critical speed in a canal for nianoeuvrability
The blockage ratio for model 5171 at an h/T = 1.33 was 0.201 and interestingly
enough the range of critical Froude Depth Numbers (= U/
Vi)
given by Bindel for
this blockage coefficient is 0.343-0.386.
The lower value cotresponds to about
72% of the Schiff limiting speed (see ref. 17) which is close to the assumed
value for limiting speed usually used in restricted water calculations.
The
Fiöude Depth number of model 5171 at the higher speed tias 0.28 which may be
cornpa're'd with th
lower'Bindel value.
However Bindel described the problem in
té±rns 'of the significntly increased rudder angles being requiredfr control at
the, criti'al speed;.
no suchhange waS' observdwith either model 5098
-or 5171, the parameter most affected béi
tudde
activity' rather than rudder
angle.
This in turn suggests that this effect of speed is related to the
reaction time
of the helmsman under test, with problehis associated
with anticipating correctly
the
ehav'iour of the more massive and more slowly-responding odel 5171.
These pobiems would presumably also arise at full scale and show the
importance
of manoeuvring a ship at the speed which is suitable for
the prevailing water
depth and blockage;
too high a speed could clea±ly give riSe tb
handl.ng
-
'9
-4.5 Effect Of Trim
Changing from a level at-rest trim to one by the head or by the stern had an efféct on the handling of model 5171. The trims used are given in table I and
were in faët rather larger than woUld be expectd for a laden tanker. In such, a cotditiôn the 'normal trim (defined as the difference in at-rest draughts at
the perpendiculars as a. percentage of the 'length between perpendiculars) would
robably be of the order of 0.4% with trim by the head possibly attaining a figure 4s high as -1.5% while trim by the' st'éTh might attain a value of 0.6%. The ttims
used in the experiments, although exaggerated, served to indicate the effect of this parameter on handling.
Results are shown in Figures 12 and 13, values of mean
w1IB,
offset rudder' crossing frequency f0 and offset crossing frequency g of the drift angle tima history are shown in Figure 12 using results obtained' with operators Aand,
B.
In Figure 13 the effect of trim on mean rudder angle is shownusing resUlts from all operators. Also shown on this figure are trend lines joining' the pooled mean values fr each trim condition.
'It is' clear from Figure 12 that the effect of trim'on mean w1/B is' small a both depth/draught ratios, the land widths remaining substantially the same fega.rdless of trim. The standard deviations of the measured w,/B about the
mean values were also negligibly affected by trim 'at a depth/draught ratio of 1 .4, but at a depth/draught ratio 'of 1 .33 it appeared that there were rat'her
'greater dejations from the èán'or a trim by the head than for' either zero trim or a trim by the stern The relevant values are shown in Table 3
h/T Operator - Trim 70' 1.44 0.13, 0.07 0 O. 11, 0.11' 1.28, 0.08 -1.44 0 1.28
0.08,0.10
0.06, 0.12 0.11,: , , 0.25, 0.12, 0.09 0.13 0.12Standard deviations of W
L!
TABLE 3 1.04 1' .33Values of f., the offset crossing frequency of the rudder angle. trace, are apparently significantly affected by trim. At the, lower depth/draught ratio
both operators recorded most rudder activity when the model was tried by the head and least when it .was tried by the stern. This confirmed their
subjective opinions that the model was more difficultto control when tried
by the head due to its tendency to wander off course; with a trim by the stern less rudder activity was needed as the model was far less prone to stray from its target course. This would, appear .to corroborate the known
fact that a. ship becomes more directionally stable when tried by the stern
and less so when trimmed by the head.
Interestingly enough at the higher depth/draught ratio the results of. operator
B show the opposite trend to those obtained in the shallower water in that rudder activity increases with a trim, by the, stern. The reason for this is
not at present clear. .
The behaviour of the f values is reflected in the behaviour of the g values
0
..
. - .which really give a measure of the number of oscillations about a mean heading recorded ,d31ng a rin. Once again trim by the, head appears to cause more
problems, in very shallow water and less in water that is slightly deeper.
The belTaviour of, the mean rudder angle with trim as 'shown in Figure 13 is not
without interest. Firstly the values obtained with level trim suggest that the slight rudder offset, required to cpunte screw bias may be more to
star-board a.t a depth/draught ratio of. 1.33. This suggests that screw bias may vary with h/T, a claim that was alsO made in reference 17. This is
demon-strated further by the results Obtained when the model was trimmed by the stern when it is seen that large starboard rudder angles' were required to
maintain the target coursè, the largest values Occurring at the larger depth/ draught ratio. The reason for this is not clear but may 'be connected with
the combined flow into the inclined screw disc and rudder causing a net lateral fOrce at the stern. The rudder and its control gear were double-checked after these' results had been' obtained and were not found to be faulty,
so it would seem appropriate o ass'me that screw bias varies with- both depth/draught ratio and trim.
21
-4.6 Effect-bfHelmsman -.
The effect on the results of the helmsman and the tethniqués-used to contro,l themodels isnotwithout interest.
It iS alsörélevanttôthe
interpretation and application of the. results obtäined to full-scale situations.
VMoreover a comparison of? the performance of professionar pilots copared to N staff is also of interest-.
The learning ability of NMI operators A and B is apparent from the /B results in Figures 5 and 6. Runs 1, 2 and 6 for operator A and runs 4 and 8 for
operator B were obtained underidentical conditions 'of odel, sped'and depth. The characteristics of .a 'learning curve' can be seen in tbeé results as the mean w1/B values and ye/B values -gradually approach a constant value with
increasing run number (and hence increasing time and experience). Similar
behaviour is seen in the & results. Run 36 for opetatoi- B isof ihterest
m
-in this context for it was carried out after this operator had been absent from
the experiment for some time due to other coitments. Thi result shows some regression in his performance which was soon overcome and by run 39 he had regained his earlier prof.iciency. - '-
-The performance ofthe-pilotin this context makes a remarkále Comparison with
that of the N!U staff. The results shown in Figure 7 ShOtño evidence of the need for learning (the pilots had had no previous experience in handling these models) while the resultant mean w1JB were among the most Otxsistent and
smallest" values obtained throughout the experiment series. The deviation of
their w1/B and yJB values about the mean isalso témar-kab'1r-'sãl-1 and tetifiesto the excellence of their-performance. - This iS perhaps not
surprising because the pilots brought tothe experiment not only a depth of
experience in handling various types of ships but also the ability to adapt
quickly to a new set of shiphandling circtstances; this is Of cOurse an
essential ingredient of their.job.- -'
The way in whichvariOus operators used the rudder- to control the model is also of interest. Shown in Figüre 14 are rudder angle and v1IB time-histories for operators A and B, while in Figure 15 are shown similar results for two of
thepilots. It is seen from Figure 14 that while operator A tended to use a lot of rudder mOvents to keep the model on course the rudder angle trace of
operator B shows less activity. It is nOt without interest to recall the descriptions of opetatorsA and B. given.in sect-ion 3.3.1 and. relate.these to
the wayin -which the peratprs controlled the mo4el. . It is- noteworthy that
in spite of the differences noted, the mean values of the important parameter w1/B were, about the same f Or- the two.runs with operator A producing slightly
greater deviations from the mean,.due no doub.t to h-is tendency to over-control
and possibly to his having a slightly longer reaction time than the younger
operator B'. -., .
--Results in Figure '1'5-show',a broadly similar level, of rudder activity from both
pilots but the resultant w1/B traces are very much more regular than those in Figure 14-with somewhat less deviation-about the mean. The mean w1/B values
are however lower than those shown in Figure 14 due perhaps to the abilityo
the-pilots to anticipatethe behaviour o.f the model rat.her better than operators
A and B and to apply the appropriate correcting rudder more quickly and to better effect than he NMI staff. - . .
-These results suggest that different.peopleachieve the same result in
different ways and that the rudder angle time history is as much a measure oft-he ship operator as-it is.a measure of the- handling qualities of the ship
or the-effectOf its environment. -.
--.5.
Controll14ty ji-a Cross-Wind.
-. .-Handling in a wind was simulated with model-5171--,at both depth/draught ratios
using methods descrjbed'above. It was -assumed that the apparent wind blew at 90° to -the vessel throughout the runs-with wind speeds V equal to 2,-4, 6 and 8 times the speed-o'f the ship thEough the water, U.
The
primary aims of the experiments-were . T. - -- - ---To compare mean rudder angles and drift angles used-by the human helmsman with those which can be calculated for a steady-state condition.
Observe any chatacteristicsofhandling'ina cross-wid which ay -.-bereleant to the resultant: lane widths -'and equilibrium rudder angles.
- 5.. 1 Calculatioriof Equilibrium., Drift and, Rudderg-1es
-It is--usual when estimating the lane width required-in a wind to assume that the
force Y and mOment N applied by the wind to the ship can be balanced by.a -suitable drift angle
variation of and & Over the range of interest we have,when an equili-brium conditioñis reached...
=Yt.
+YT.5
..m6m
N.T=N'.f3
+N'.cS
wm
m
which give -23
-6=
(Y-m"6
..- .. (7)wheté the, primes indicate'non-dimensionalistion according to the scheme
= N =
N/PL3.U2'
etc (8)and .Y indicates partial differentiation wit-h .respect.to .,, :Y!/,and so on.
Water density is dnoted b.y.p.
-The following va1u., deduced from measurements made at NMI, were assumed
to apply to mo4el 5171:.
5.2 Calculation of Wind Force and Moment
Wind sway fOrce and yaw momént.N were-calculated'from the non-dimensional
coefficients of -reference .,13, assuming that the ship had an, 'of 33Q
and a lateral, aboveater arQa of . 2567m2 .. The :fOilbwing values: of and
due to wind were deduced and they were used both in calculating equilibrium drift and rudder angles and also to set 'the revolutions of the 'wind fans
on the modl during the experiment.
(5)
-
(6)1.3
Yr -278. 10 - -4i.io N -79.I0 =36. 10 8.9?.lO 4.4.10 -].8.]0 -I .7.1 I :v./U
w
2.0 6.0 8.0N10
-0.083 0.013 -0.332 0.052 -0.748 0.117 -1.329 0.20,9 5.3 Experimental ResultsResults from the experiments conducted with a c±ossind are shown in Figures 16 to 18. Mean values of drift angle and rudder angle obtained from results
m
m
-of, all helmsmen Cal]. NNI staff arid pilots) are compared with values calculated
from equations (6) and (7) in Figures 16 and 17. Both and are plotted
against an ordinate of V/U, the value of U remaining constant throughout the
experiments and equal to .a ship speed of 5.6 knots.
It is clear from both Figures 16 and 17 that values of attained by the hi.an àperatbrs were reasonably close to the calculated values, agreement
improving as V/U increased. Agreement betcen measured and calculated drift
angle is not good however, the poorest agreement occurring at the lower depth/draught ratio. The reasons for the discrepancies in both and
may be due to
(1) The assumed inanoeuvring coefficients Y, etc may not apply
exactly to modl 5171. No coefficient have been measured for this model but values have been obtained for similar hull forms and stern arrangements, these being used to estimate suitable values for equations (6) and (7). It is not considered therefore that the coefficients used are sen. ously in error:.
(ii) The handling techniquesused in a cross-witid may differ from that of simply finding a 'balance' condition This was almost certainly the case,
for, while the helmsmen clearly fOund it a relatively easy task to obtain
approximately the correct ru4der angle, suchwas not thecase with drift angle.
In general the helmsmen tended to over-shoot th appropriate dft ang-le an4 thn allow the bows of the model to all aay to leeward befoe bringing them
25
-This, tended. to give .a larger mean than required. .
-The accuracy needed to obtain the very small 8 values apparently required was greater than it was possible for a helmsman to obtain 'intuitively'
in the absence of, say, a heading indicator It is also much more difficult .;..to 'set. a massive,, ship or model -to a verysl drift angle than it is, to
obtain
a
correct 6 value. tt appeared that ost. operators tended to counter the wind with'rudder alone and as a result overestjniat.ed"t.herequired rudder angle, consequently altering 8.' An exception to this was piipt A whose results are high-lighted in Figures 16 and 18. He.delibetately allowed the model to move off-course up-wind towards the side Of. the channel.
so doing he obtained a side-force and, turning, moment from the vertical 'bank'
of the channel wall, which tended to counter those applied by the wind. The
resultant reduction in rudder angle is shown on Figure 16 where it may be .'.compared with results from pi-lots B and C.. It may be remernbered that. pilot A
- had experience of operating in a ship canal where clearly this technique is
appropriate. . ..:
The visual scene presented in the television mpni.tor may have been less
than
ideal
for a ship proceeding at an angle of yaw. The operators wereconstrained to look over the bow of the model at all times and fO judge whether
-they were proceeding, along te target course. .. They therefore had less. angle of
view than in a real, life situation.
It is usual to.compute the lane width required in a wi-rid from the projected
widti of, the. ship at an equilibrium drift angle 8. This gupres the possibility of the
sKip
straying from the target course with a. subsequent effet.- on laflewidth. This is illustrated in Figure 18 where values of w/B (lane widths calculated from measured 8) are, compared wih,values
of "w1/B
(lanewidth
including coirse error for all operators. It 'is clear from thi',t.hat the
lane width actually used by the helmsmen reflects the tendency to. move the
model of ,the.target course. Generally .this movement was to windward where
the result from pilot A, mentioned above is especially notable compared to pilots ,B and C due to his deliberate policy of mak,,ing. goqd us,e of-bank effects.
Other recorded movements..were to leeward of the target course due to the,
effects of the cross-wind. The correction of this was to move the model
bodily. 'to windward, possibly to overshoot the correct position only to be blown
to leeward.again. This 'trial and error' approach to course-keeping resulted in an oscillatory track with a subsequent effect on lane ,width, reflected in
It would seem therefore that while the 'uilibiium'rudde± atigleto coütier wind
effects calculated from knOwn' hrdrodynamic and aerodytiaàic feature's of th
ship gives a good representation of reality, an 'equilibri'tá 'drifTt angl,
sImilarly calculated and translated into a lane width, will give a very
misleading impression of the width of track iept out hj
hip 'iti a crosdnd.
6.
General Discussion
6.1
Rudde
Activity and Equ1ibrium Rudder-Angles
This model study of ship-handling had the primary aim of àss'essing lane widths
and rüdder activity associated with basic controllability and crosswiñds.
These parameters are of
jbr importance in assessing the safety of a channel
from a ship-handlihg point of view, for they give some idea of the allowances
"that
shoüld be made for basic ship handling.
Rudder activity and equilibrium
rudder ahgles are used in some 'design methods to assess' the hand'lin'g Of a ship
in a 'given channel (see for example refereñces-5 and 18) butit has been shown
in the model tests described above that rudder activity is sOmetimes 'a
much a.
'measure ofthe helmsman as it is of the ship or'channel.
Different helmsmen
were found- in the tests to achieve the same' ends by differetit means.
With
identical depth/draught ratios, hull form and speed5 one helmsman' might use
a great number of rudder movements with large angles whereas another might use
very few'with smaller angles', both in the end achieving the same
lane' width.
Therefore it would see,m'that a safety cziterion for navigation channels of the
type "a rudder atigle of x degrees shOuld not be exceeded for more than -y%
Ofa transit" should be used with ëaution for it tacitly-assumes that all
helmsmen control a
hip in 'a similar manner.
''he res Its des'cribed
thsection 5 siggest that Use of the equiI'ibrii.
'rudder angle concept may be successful, at least for a cross-wind.
-Thi
is'-encoüaging as it is' a useful concept, easily grasped, for the purpose
of waterway design.
Indeed, if safe upper limits on rudder' angle were in use
on'à given waterway,, plots such as tha't 'Of Figures 16' and IT could: be used to
estite the thaxiim àllowableV/Uratio for safe transit Of the waterway.
27
-- 6.2 Lane Widths
Measured lane widths for basic controila6ility have been shOwn in Figures 5 to 7 which give all results for the selected NMI staff and professional pilots who took part in the experiments. It is not without interest to ppol the repeated results shown in these figures to give pooled mean w lB values and
standard deviations fO± the various hull forms and. cbnditiOns of the experiments. The pooled meansand standard deviationi were óbtai:ned fro
E n.x/n.
1
= (10)
wliere XN ánd
'M
.ar pooled valuesx. is au individual mean value of n. measUrements
-staiidãrddeviátjon dfi.- measurements
1
It is assurned that th Bessel Correction has been applied to all results in equation (10).
The resulting pooled values for basic controllabilityaie given in:Table 4
-wh-ich include both 'learning' aid '4ild' results.
It would seem that for NNI staff ameanvalue Of wi/B of about 1.2 with a
'worst case' standard deviation of 0.2 is appropriate, the pilots showing
consistently lOwé± én and standard deviations. Rësults given in
reference 19 cOnfir the suggestion made in reference 20 that confidence
limits of are appropriate for lane width This would then give a maximum
w1/B of 1.2 + 2(0.2)= 1.6 as applicable forbasic controllability, which is less
than the value of 1.8 suggested in ref. 3.
Full scale verification of sUch a result is not easy, foi reiults such as those
given in reference 19 apply to situations in which basic- controllability is
only one factor. Such results also tend to include measurernents of various
target coures within a given channel, rather than the deviation from that
target course as. described above.
a. 'I
Model 5098 5171 Operator Trim hIT 1.04 1.33 1.04 1.33 F nh 0.27 0.24 0.33 0.18 0.16 zero 1.201(0.121) 1.110(0.082) 1.144(0.118) 1.166(0.125) 1.072(0.053) 1.184(0.111) A head -1.199(0.106) -stern -1.159(0.084) 1.189(0.092) zero 1.205(0.111) 1.103(0.068) 1.125(0.083) 1.178(0.112) 1.230(0.159) 1.154(0.117) B head -1.173(0.087) 1.193(0.199) stern -1.196(0.107) 1.187(0.087) pilots zero 1.083(0.044) -1.066(0.046) 1.065(0.037) -Note:
Figures in parentheses are standard deviations of the
pooled mean values.
Pooled Mean Values of
29
-PoOled mean aid standard deviatiohs of w1/B for all results obtained "in the cross
wind expetimentS ate shotin in Figure 191where the'folloing-points'frjterest
may'be noted:
- 'Ci)
the 'w1/B"vàlues do not approach unit
'as \T/U +0.
This wu1d not
'have been the case if lane widths based on projected beam at the appropriate
equilibrium drift angle were used.
The va.ue5 obtaied.with'I"saf appear
to 'approach a wi/B value Of' about 1.2 as V/U+ 0 ihich agrees well with the
basic'control'lability' value dedued' frOm' Table 4.
while there is a nt'able tendeticy for w1/B 'to increase zith V/U at
hIT = 1.33 such was not the case at an h/T of 1.04 when the mean. w1'/B."values
'appear to be scattered about a constant value of about 1.3.
the small w1/B values obtained by the pilots at. V/U =.6"is
noteworthy.
Fiñallr it. is worth te-iteratin.
tha'E' this study of- lane widths h'as made several
assumptions' regarding control of tle ship.
These are
(1)
Navigation was by visual .eans otly
'Control was by rudder only,' the engines
intaining constant
revolufions Throughout.
No control was exercised' through us'e of the engines.
Visibility remained constant and of a reasonably high level' throughout.
'Coirol was by a' hClmsniati otily'
the command chain in which the
pilot gives helm ordersto the helmsmn was not simulated.
Other factors may affect cantroI'labclity to a greater or lesser' degrée and
should be borne in mind when assessing lane width for controllability.
Awng
these factors,' the folloing"ay b'emeiitioned:
' - 'Response àf the 'ngines
- . :-Visibility
' ' - . ' 'Cc)'
Reliability of 'ship's' 'equimeü
'-Cd)
Response of crew to. pilot's requirements
-Ce)
Communfcationsllafiguage problems
(f)
Navigational aids.
' .From the point of view of port approach channel design', it is not- clear to
what extent such factors should be taken into account i-n the basic desigh,
of' channel width.
' ma new port' for example; item C-f) would be incorporated
in the overall design to give maximum aid and, in conjunction with operating
procedures would help
to alleviate problems from
tern (b').
Furthermore item
(e) could be-overcome
re readily by operating procedures and training
rather economi.cally than makitig some allcwance in lane. wid for poor navigation steing from this. source. Items (a), (c) and (d) all relate to the quality of the ship entering the port and could be tackled by other means, more legislative thati technical, to improve navigational safety
The overall w1/B value, forstraigI channels Obtained, in this Study may be considered-therefore asa rninimum value for basic controllability which may be used in. conjunction with information 'fr other factors such as wind, current. bank effects and the effect of other ships to deduce a suitable channel width for safe navigation. .
7. Cpnclusions
The following main conclusions have emerged from this study relating to ship handling on a straight target course:
Professional pilots who toOk, part in the experiments produced
consistently. lower lane widths'.than helmsmen drawn from NMI staff.,
Helmsmen from N staff.needed a legrning period to produce consistent results; pilots did not..
In general slightly greater lane widths were recçrded at the lower depth/draught ratio. . .
For the two ship nxdels tested, bull form had little effect on the measured lane widths..
Trim had a significant effect on handling and mean rudder angles. Trim by the stern resulted in ,a large mean rudder offset to starboard for the
single screw model.
Rudder activity is related to the control technique of the helmsman/ pilot. Different helmsmen achieve the sa.meend by different means. Therefore safety criteria based On rudder activity should be treated with caution.
Ship speed affected rudder activity with the single screw mo4el but not with the twin screw model.
Predicted equilibrium ru4der angles in a cross-wind agreed reasonably well with measured values.
Predicted equilibrium drift angles in a cross-wind were significantly less than those measured.
- Lane widths in a crqssind are not
correctly predicted from the 'projected beam! of the ship at an equilibrium drift angle.
31
-Aeronautical Journal, January 1978
8. References
1.
DANDIW
'Some Measurements of Interaction Between ShipModels Passing on Parallel Courses! NNI TM. 32, April 1979
DANDIW
'Some Measurements of Interaction Induced bySurface-Piercing and Flooded Banks' NMI
3. GARTHUNE R et al 'The Performance of Model Ships in Restricted Channels in Relation to the Design of a Ship Canal'
,DTMB Report 601, 1948
4.
CRANECL
'Proposed Procedures for Determining Ship Controllability Requirements and Capabilities' First STAR Symposium, Washington 19755' 'Port Approach Design - A Survey of Ship Behaviour Studies,. Vols I and II'
National Ports Council, London, 1975
6. KEITh V F et al 'Real Time Simulation of Tanker Operations for the Trans-Alaska Pipeline System'
Trans. SNAHE, vol. 85, 1977, p.41.9-458
7. HOOFT J P and 'Four Years Operation Experience with the Ship Control PAYNANS P J
Simulator'
First STAR Symposium, Washington 1979
8.
PAYMANSPJ
'Human Factors in Ship Handling' Second WestEuropean Conference on Marine Technology, London 1977
9.
CRANECL
'State of the Art on How to Include Human Control Into the Method of Investigation' Lecture 3, Symposium on Aspects of Navigability of Constraint Waterways,Delft, 1978
101 WAGENAAR W A and 'The Effect of Contracted Timescales in Scale Model.
NICHONJA
Manoeuvring' Institute for Perception., TNO,
Soesterberg, Netherlands, Report IZF 1968-C3, 1968 11. TRUIJENS C L,
WAGENAAR W A and
VAN WIJK
'The Effect of Contracted Tirnescales on the Learning
Ability for Manoeuvring of Large Ships! Netherlands Ship Research Centre, TNO, report 215, December 1970 12.
REGAND,
BEVERLEY K and
'The VisUal Perception of Motion in Depth'
Scientific American, volume 241, number 1, July 1979, p.122 CYNADER N
13. 'Estimation of Wind and Current Loads on VLCC's'
Oil Companies International Matine Forum, London, 1977
This investigation was carried out by the Nautics Branch of the National
Maritime Institute. The project leader was Dr I Dand and the experiments were
undertaken by Messrs GTaylo, D Hood, W Sims, CBrake and Miss S Dewar with invaluable help and advice fromMessrsP J H Tebay, R Casbin and K Davis.
I
15. HOOPT J P 'Manoeuvring Large Ships in Shallow Wãtet -Parts I and II'
Journal of Navigation, vol. 26, p.189 and 311, 1973 16. BINDEL S 'Experiments on Ship Mánoeuvrábility in Canals as
- Carried out in the Paris Model Basin'
-DT report 1461, October 1960, p.179
17. DAID I W 'Hydrodynamic Aspects of Shallow Water Collisions' Trans RINA vol. 118, p.323 1976
18. AI1MAR A A et al 'Design of Navigation Canal Cross Section and Alignment!
-PIANC 22nd International Navigational Congress, Section 2, Subject 3, Paris 1969
19. RIBADEAU-DUNAS L 'Antifer Aids to Navigation and Channel ManoeUvring -Experimental Rasults' Symposium on Aspects of
Navigability. of Constrain Waterways, Delft 1978
20. SUKSELAINEN J 'On Ship Manoeuvring and Waterway Width'
Helsinki University of Technology, Ship Hydrodynamics Lab. Report No. 8, 1975
33
-10. NomenclatUre
Ac coss-sectional area of waterway
B moulded beam of ship CB block coefficient D propeller diameter
f offset crossing frequency - rudder angle
Froude Depth Niber = V/v1
g gravitational acceleration
g offset crossing frequency - drift angle h water depth
length of ship between perpendiculars number of data points in ith experiment.
Nw' yaw moment coefficient
N/PLpp3U2
P mean face pitch of propeller
r angular velocity
T at-rest draught
U speed of model through water
V wind speed
w
--v sway velocity
w
lane widthw lane width dedUced from 'projected beam' of ship lane width including y
x mean value
pooled mean value
22
sway force coefficient Y/lPLpp U
w'
Nw sway force and yaw ment due to cross windY, N
3Y'/,
N'/aY, N
Y'/S, aN'/S
lateral deivation of model centre of gravity from target course
drift angle
equilibrium or mean drift angle
IS rudder angle
equilibrium or mean rudder angle
p water density
a stardard deviation of the mean
AXIS SYSTEM
AXIS SYSTEM AND DEFINITIONS
SWEPT TRACK OF SHIP
(EXAGGERATED)
BOUNDARY OF LANE
WIDTH. REQUIRED BY SHIP
TARGET COURSE
Vi71__II1
r&vvrviø___ 111
ILW
IIIM
INN___
II11W__ATIJA
,,
MODEL No 5098
_____
II
1127 LOAD OPAUGHT LIGHT OPAUGHTPROJECT No. 252001
MODEL No: 5171
BODY SECTIONS
FIG. 2
LOAD DRAUGHT LIGHT DRAUGHT 71125 16 Om I4 Om 2 Om LWL hOrn hO Orn Born 6 Orn 4 Orn 2Orn 16- Orn 4 -Orn 12 Orn II - Orn LWL lOOm B Orn 6 Orn '4 Orn 2 Ornz-,
0 p-n0
0
BUOYTRACKING CAMERA
RUN BOARD
TAR GETSSTEERING POST
WIND FAN
CCIV CAMERA
/
/
/
CHARTf
I\
RECORDER MOTOR RADIOCONTROL
:E'XPE'R:IMENTAL
ARRANGEMENT
- DIAGRAMMATIC
WIND FAN
CARRIAGE MOVEMENT
-n