A Study on Weather Vaning
Dynamic Positioning System
J.A. Pinkster, H. Hagiwara, R. Shoji and
H. Fukuda
Report 1201-P
September 1999
Published in. Journal of Japan Institute of Navigation, VoI. 101, September 1999, PP. 83 - 93.
TU Deift
Faculty of Mechanical Engineering and Marine TechnologyShip Hydromechanics Laboratory
.
-a,'"
P7S1FThA B
JAPAN INSTITUTE OF NAVIGATION
ISSN 0388-7405,
'99-11-29 20
FPÜIi I*3'3': 9'
3-5245-?302 F.119)
THE JOURNAL OF
JAPAN INSTITUTE OF NAVIGATION
SEPTEMBER 1999
VOL.101
CONTENTS
A Positioning System
withGeostaionary Satellites-
.-Construction of Positioning
System-Chunming FAN. Akio YASUDA, Hiromurie NAMIE
and Masashi KAWAMTJRA C 1)
A Study on the Orbital Analysis of Geosynchronous Satellires- V
--The Relation between Post-fir Range Residuals and
MeteoroIoical Elements-
Masashi KAWAI ( 7)GPS Positioning Accuracy Improvement Using Multiple Reference Stations
Takeyasu SAKAT and Kazunobu KOREM1JRA
(15)
4
A Própoal of the Transponder with Frequency Shift
Manarni IDE and Shogo HAYASI-IIA ( 21)
A Statistìcal Analysis of the Relative Positions between Vessels
Tatsuto YAMADA. Alcira NAGASAWA and Asao NISI-JINO
(29)
A Study on the Relationship between Ship's Encounter
Situations and Traffic Density
Dongping ZHOU and Hayama IMAZU --- ( 37)
Automatic Course Line Setting - ifi.
-Estimatiou.from the Shape of Available Water
Area-Takahiro TANAKA and Kinzo ¡NOUE ( 47)
8. A Pilot Study for the Influence on Links of Marine Casualties' Factors
...Vusuke YAMAZAKI, Yoshio MURAYAMA and Makoto ENDO ( 55)
Statisticàl..Characteristics of 'Casualty Distance' byAnalizing Randomness in
Marine Disasters and the Application of the Measurement to the Evaluation
of Traffic Accidents Tsukasa NAGAHATA' t 65)
.10. An AizroTnùtic Control System Design of Berthing
Manoeuvring with Decoupling Control Seiji IWAMOTO ( 75)
11
A Study on Weather ii aning Dynamic Positioning System
JA. ?ICSTER, Hideki HAGIWARA.
.
Run SHOJI and 1-litoi FUKUDA 83) '9-1 i -2's 20 CiT FPOFl
I t3:.I:ii1' 03-5245-7382
F'.A Study on Weather Vaning Dynamic Positioning System
J.A. Pinkster*, Hideki Hagiwara**, Run Shoji*
and Hitoi Fukuda***Abstract
In this study, a simple dynamic positioning (DP) system called "Weather Vaning DP system' has
been developed. This DP system uses only the controllable pitch propeller of the main engine and bow thruster driven by the PID feedback plus wind feedforward control algorithm to position the reference point on the ship at the specified point on the earth. The system possesses automatic weathervaning properties, i.e., the heading of the ship is not controlled by the system hut is governed by environmental
conditions.
Onboard experiments on the weather vaning DP system were carried out in December 1998 using the
training ship Shioji Maru of the Tokyo University of Mercantile Marine. The main results of the
experiments are as follows.
In the weather vaning DP system, the stable equilibrium condition is obtained when the ships bow points in the wind (or current) direction, which makes it possible to perform accurate DF with a small thrust of the bow thruster.
The wind feedforward control is very effective to reduce lateral motions of the vessel in a strong
wind condition.
The differential feedback control is inevitable to perform stable DP.
The stability of the weather vaning DF deteriorates as the reference point is moved aft from the bow; the DF is stable when the reference point is put forward of the bow.
Experiments to pursue a specified point moving on a square and semicircles were also carried out,
and it was found that the control algorithm of the weather vaning DP was applicable to automatic
tracking.
* Non-member Delft University of Technology
Melclweg 2, 2628 CD, Delft, The Netherlands ** Member Tokyo University of Mercantile Marine
2-1-6 Etchujima, Koto-ku, Tokyo 135-8533
Student Member Postgraduate School, Tokyo University of Mercantile Marine 2-1-6 Etchujima, Koto-ku, Tokyo 135-8533
IflttO(IUCtiOfl
The term Dynamic Positioning (DP) means automatic positioning of the Reference Point (RP) on a
ship
at a Specified Point (SP) on the earth.
DP is an indispensable technique for oil drilling in submarine oil fields, laying submarine cables, and sea bottom surveys using underwater robots. In manycases, the conventional control system of DP employs bow thruster (BIT) and stern thruster (SR'), in addition to the main engine-driven controllable pitch propeller to position RP, and simultaneously carries
out heading-keeping. Such a system involves a complex algorithm, and heading-keeping often becomes impossible when there are strong winds and tidal streams.
In response, a new control system has been developed, which carries out dynamic positioning only
with CPP and BR' in the absence of heading-keeping
This control system enables dynamicpositioning of RP with a small thrust force, because the heading of a ship naturally coincides with the direction of the wind or the tidal stream. (This control system is called 'Weather Vaning DP".) By carrying out onboard experiments with the training ship "Shioji Maru" of the Tokyo University of Mercantile Marine, equipped with an RTK-GPS positioning system, the effectiveness of the newly
developed system was assessed.
For sea bottom exploitation using submarine robots and submersible vehicles, it is necessary that the surface support ship should be capable of accurately tracking the underwater vehicles. In this specific
connection, experiments were also carried out by moving SP
ata low speed and having RP
automatically track it to assess system performance.
Weather
Vaning Dynamic Positioning System
The control system of the Weather Vaning DF combines feedback control
relying on datarepresenting the displacement of RP from SP and feedforward control relying on data representing the external forces acting on the ship's hull. Calculations of displacements of RP from SP are carried out as
shown below.
Here and A1), denote the latitude and the longitude of a specified point (SP) on the earth, 4 and X, the latitude and the longitude of a reference point (RP) on the hull, and 4 and A1, the latitude and the longitude of the position where the antenna of onboard RTK-GPS receiver is located, respectively. If xi
and yi denote the longitudinal distance and the transverse distance between the position of the antenna
and RP, respectively, (4, A) can be obtained from the following formulae. (xi takes a positive value
when the antenna position is forward of RP, while yi takes a positive value when the antenna position is on the port side of RP.) (See Fig.1.)
=
- (xi cosO + yi sinO) / R (4i)
A = A1 - (Xi sinO - yi cosO) / R ()
where
O ship's heading
R,(41) : radius of meridian of the earth at latitude 4i
Rp(4i) : radius of parallel of the earth at latitude 4
-2-Longitudinal distance Ax and transverse distance A y between RP and SP can he obtained from the following formulae. (Ax assumes a positive value when SP is located foward of RP, while Ay assumes a positive value when SP is on the port side of RP.)
Ax = (A0 A) R(1)i) sinO + ((1)0 -(1) Rn1(4)i) cosO (3)
Ay = - (A0 A) R(4i) cosO + (4o -4) R(4i) sinO
(4)The thrust force F of CPP and the thrust force F of BIT required for Weather Vaning DP can be
obtained by the following formulae.
T
F=aAx+bV +CxJ AxdtR
O Fa Ay + b Vy
+f
TAy dt - R
(6) where 3.1 Outline of experiments (5) V: dAx/dt V : clAy/citT : time elapsed from commencement of DP
R: longitudinal component of external force
R transverse component of external force
External forces acting on the ships hull include wind pressure, hydrodynamic force due to tidal
streams (or ocean currents), and drifting force due to waves. F and R take positive values if they act
on the ship to move her forward, while F and R take positive values if they act on the ship to move her towards the port side. (See Fig.!.) Coefficients ax and ay are feedback gains for proportional control, and coefficients b and b are feedback gains for differential control, and coefficients cx and Cv
are feedback gains for integral control.
With this control system, the center of effort of external forces such as winds and tidal streams musi coincide with the position of BIT in a state of equilibrium. For example, we consider a case in which RP and SP are coincident, and the center of effort of external force is deviated aft from the position of
BIT. If the external force is divided into a longitudinal component and a transverse component, the longitudinal component can be set off by the thrust force of CPP, but the transverse component produces a moment around the center of gravity due to the thrust force of BIT, which is located on the foward side, equal in magnitude and opposite in direction. As a result, the bow turns in the direction in which
BIT acts. When the center of effort of external forces is at the position of BIT, the moment due to
transverse component of external force and the moment due to transverse thrust of BIT set off each
other, thus RP can be kept in the position. The center of effort of external forces coincides with the position of BITwhen the ship receives winds, tidal streams or ocean currents nearly from dead ahead, hence, RP can be retained with the bow in the wind, tidal stream or ocean current.
Dynamic positioning experiments using Shioji Maru were carried out in Tateyarna Bay during the period from December 2 to 4, 1998. The master station of RTK-GPS was located at Banda, near the SW end of Tateyama Bay, and experiments were carried out in waters approximately two miles from Banda. The positioning accuracy of RTK-GPS used was 5-20 cm CEP.
Concerning the feedback gains in formulae (5) and (6), those with minimum ship motions recorded in experiments with the Shioji Maru using varying gains in July 1998 were used. Table i shows the
feedback gains used in the experiments. It was difficult to predict the magnitude of hydrodynamic force due to tidal Stream and drifting force due to waves acting on the Shioji Maru, hence, only the wind pressure was considered as the external force in the experiment. Wind pressures R and R acting on
the Shiojï Maru were calculated using the following forniulae2:
1
R=
2 p CXAT VA2 (7)i
R=
p CYALVA2 (8) wherep : density of air (0.124 kg. sec2/m4) VA: apparent wind speed (rn/see)
AT : transverse projected area of the above-water part (79.2 m2) AL : lateral projected area of the above-water part (246.3 rn2) C : fore and aft wind force coefficient
C : lateral wind force coefficient
C and Cy were measured with a 1.085 m long model of the Shioji Maru in a wind tunnel. Fig. 2 shows the measurements of Cx and
In these experiments, the effectiveness of feedforward control and differential feedback control
against wind pressures and the effects of the position of RP upon the control stability were assessed.
During the experiments, SP was first placed 25 rn forward and 25 m on the port side of RP, and control was exercised so that RP coincides with SP. Control was continued for 15 minutes (20 minutes in some experiments), and it was then observed in the next three minutes that how the Shioji Maru drifted by
winds with no control.
During the experiments, the position of the ship was fixed by RTK-GPS at regular intervals ofone
second, and CPP and BIT were controlled according to formulae (5) and (6) respectively. In these
attempts, second-to-second displacements Ax and A y of RP from SP were used without filtering them. For speeds V and Vy of RP, apparent wind speed VA and apparent wind direction from bow (this was
used to obtain C and C)), mean values of past 10 seconds were used because their variations were
excessive.
Thrust force Fx of CPP was controlled so that ahead thrust and astern thrust were confined to within four tons and two tons respectively by restricting blade pitch angles. Thrust force Fy of B/T was also controlled so that both port and starboard thrusts were confined to within 1.8 tons by restricting blade
pitch angles.
3.2 Effectiveness of feedforwarcl control against wind pressure
-4-To assess the effectiveness of feeciforward control against wind pressure (hereinafter referred to as WFF (Wind Feed Forward) control), experiment was carried out first by combining PID (proportional, integral, differential) control and WFF control, and immediately thereafter experiment was carried out with PID control alone. The RP was placed at the how. Mean wind speeds during the experiments were 9.2 rn/s when WFF control was added, and 11.3 rn/s when PID control alone was exercised. Wind speeds in both control modes were considerably high.
DP reached a steady state approximately six minutes after starting experiments. Fig.3 (a) shows the results of plotting the Shioji Maru every 10 seconds for the nine-minute period from the sixth minute to the 15th minute from the commencement of the experiment (WFF control added), and Fig.3 (b), the results of plotting (PID control alone). It can be seen from Fig.3 that an extremely stable DP was available when WFF control was added. Conversely, in the case of PID control alone, transverse
movements of the ship were significant with resultant instability of DP.
Fig.4 (a) shows time series data of displacements A x, A y of RP from SP plotted during experiment (WFF control added), and Fig.4 (b) shows those obtained during experiment (ND control alone). It
may be seen from Fig.4 that transverse displacements A y were confined immediately to within ±5 m when WFF control was added, but they continue cycling involving transverse movements of ±15 rn or thereabouts when PID control alone was exercised.
Table 2 summarizes mean values and standard deviations of wind directions from how, wind speeds, Ax, Ay, F, and F during the nine minutes between the sixth minute and the 15th minute after starting the experiment. Data in parentheses in the table are standard deviations. The standard deviation of
transverse displacement Ay with WFF control added was 2.0 m, and it was 8.9 m with PID control
alone. It follows that standard deviations decreased to 1/4 or less upon adding WFF control. On the
basis of these experimental investigations, it can be clearly seen that WFF control, which computes wind pressures acting on the ship's hull and creates a thrust force to cancel out such wind pressures, is highly effective in damping transverse hull movements in strong winds.
Besides, standard deviation of thrust F required for B/T, when WFF control is added, is 311 kg, and that, when PID control alone is exercised, is 558 kg, thus it can also be seen that the variation of B/T thrust could be reduced to a significant extent by adding WFF control.
3.3 Effectiveness of differential feedback control
To assess the effectiveness of differential feedback control (hereinafter simply called "differential control'), experiment was conducted without applying differential control, and immediately thereafter
experiment applying differential control was carried out. In both of these experiments, WFF control was applied, where the position of RP was set at the bow. The mean wind speed during the experiments without differential control was 14.1 rn/s and that with differential control was 13.1 rn/s. These wind
speeds were in the proximity of the critical limits under which control by BIT was possible. The
significant wave height during the experiments was approximately 0.8 rn.
Fig.5(a) shows the results of plotting the Shioji Maru every 10 seconds for the nine-minute period
between the sixth minute and the 15th minute from the commencement of the experiment (without
differential control), and Fig.5(b), the results of plotting (with differential control). It can be seen from
Fig.5 that DP accuracy was degraded due to large ships motions when differential control was not
applied. Conversely, in the case to which differential control was applied, ship's motions were very
insignificant and satisfactory DP was conducted.
From Fig.5(b), it may be seen that the Shioji Maru was subjected to DP, while she was receiving winds approximately 35 degrees on the starboard bow. This can be explained in that the position of this
experiment was far off the coast and the ship assumed a state of equilibrium, while she was receiving
tidal streams flowing NE, with her stern being pushed towards the starboard side due to the drifting
pressure of the tidal streams. Fig.5(c) shows the natural drifting motions of the ship plotted every 10 seconds under differential control when control was terminated 15 minutes after starting the experiment.
The Shioji Maru assumed the position
right abeam the wind and drifted leewards at a speed of
approximately 1.3 knots.
Time series data of displacements x and y of RP from SP during the experiments are shown in
Fig.6(a) (without differential control) and Fig.6(b) (with differential control). In the case without
differential control, it
can be seen that
both A x and A y varied significantly, involving cycled overshoots of RP over SP. In the case with differential control, on the other hand, both Ax and Aywere confined to a range within ±5 m about six minutes from starting the experiment.
Fig.7 shows time series data obtained under differential control such as wind direction from how, wind speed, required CPP thrust F, actual CPP blade pitch angle, required BIT thrust F, and actual BIT blade pitch angle. (Although B/T of the Shioji Maru produces a thrust of 0 kg at the indicated blade pitch angle of O degree, the thrust of CPP produces a thrust of 0 kg at the indicated blade pitch angle of
-1.5°.) It may be seen from Fig.7 that the blade pitch angles of CPP and B/T nearly follow the thrust requirements being updated second to second, The blade pitch angle of B/T for the limit thrust value of BIT, being 1.8 tons, is 20 degrees, and the required B/T thrust in about four minutes after starting the experiment exceeded -1.8 tons, thus the blade pitch angle of BIT was restricted to _200.
Table 3 shows the mean values and the standard deviations of wind directions from the bow, wind
speeds, Ax, Ay, F and F for the nine-minute period between the sixth minute and the 15th minute
from the commencement of the experiment. In the table, values in parentheses are standard deviations. The standard deviations of Ax and Ay are 7.8 m and 16.0 m (without differential control) and 2.4 ni and 2.3 m (with differential control) ; i.e., the standard deviation of Ax decreased to little short of 1/3,
and that of A y decreased to approximately 1/7 by applying differential control. This suggests that differential feedback control is indispensable for satisfactory DP.
Furthermore, the standard deviations of fluctuating values of F and F decreased to approximately 1/3 and 1/4 respectively by applying differential control. (Note, however, that there was a considerable
length of period during which F exceeded 1.8 tons in the case without differential control, and the
actual blade pitch angle of BIT was limited to -20° throughout that period.) It therefore follows that variations of thrust of CPP and BIT can be greatly reduced by applying differential control.
3.4 Effects of the position of RP onboard the ship upon control stability
To assess the effects of the position of RP onboard the ship upon the stability of DP, experiments were carried out locating RP at the bow, midship, 1/4L aft of the bow, and 114L foward of the bow. (L
is the overall length of the Shioji Maru, 49.93 m.) BIT of the Shioji Maru is located at 0.15L aft of the
bow. In all the four experiments, PID control coupled with WFF control were applied. These
experiments were carried out on December 2; the experiment with RP located at the bow in the morning,
the other three experiments continuously in the afternoon of the same day. The mean wind speeds during the experiments were in a range between 7 and 9 rn/s.
For the period of nine minutes from the
sixthminute to the 15th minute after starting the
experiments, movements of the Shioji Mani were plotted every 10 seconds and shown in Fig.8(a) (RP at the bow), Fig.8(b) (RP amidships), Fig.8(c) (RP 1/4 L aft of the bow), and Fig.8(d) (RP 1/4 L foward of the bow) respectively. From Fig.8(b), it can be seen that when RP was placed amidships, DF was
unstable with large ship's motions. From Figs.8(c) and (d), it can be seen that when RP was placed 1/4
-6-L aft of the bow, the yawing motions were slightly greater than in the case where RP was placed 1/4 -6-L forward of the bow. The case of DP available with RP placed forward of the bow as shown in Fig.8(d)
may be utilized for a tanker whose bow is retained with a specified distance from a single-point mooring buoy and cargo oil is discharged ashore through a pipeline led from the buoy.
Generally, Weather Vaning DP controlling the transverse motions of a ship only with BIT, control
stability degrades as RP moves from the position of BIT towards the stern. When a ship turns using
BIT, the pivoting point is located slightly aft of the centre of gravity of the ships hull. When RP is
placed in the proximity of the centre of gravity, RP does not move to any appreciable extent even if BIT
is operated so that RP might come closer to SP. When RP is placed aft of the pivoting point, RP
departs from SP if BIT is operated with resultant control inability. It is, therefore, desirable to place RP
as close as possible to the proximity of the bow. When RP is placed forward of BIT, RP comes closer
to SP at a transverse speed greater than that of BIT, and hence no control problems arise.
Table 4 shows the mean values and standard deviations of wind directions from the bow, wind speeds, A x, A y, F, and F measured during the period of nine minutes between the sixth minute and
the 15th minute after starting the experiments. The values in parentheses in the table are the standard
deviations. Standard deviations of Ax and A y became greatest when RP was placed amidships. The
standard deviation of A y, in particular, had a size equal to 2.5 to S times the values in other three cases, suggesting transverse control instability. When RP was placed 1/4 L forward of the bow, the standard
deviation of Ay was so small as 1.1 ru, thus demonstrating the good transverse control performance. Concerning the standard deviations of F and F, they were greatest when RP was placed amidships.
The standard deviations of F for the alternative locations of RP, i.e. 114 L forward of the bow, 1/4 L aft of the bow, 1/2 L aft of the bow (midship), were 107 kg, 215 kg, and 460 kg in the order of description.
lt can be seen from the above that greater thrust of BIT was required for transverse control as RP moved aft.
4.
Automatic Tracking Experiments Using Shioji Maru
4.1 Outline of experiments
The automatic tracking experiments of the Shioji Maru were carried Out n Tateyama Bay during the
period from March 9 to 11, 1999. The positionìng system used was the forementioned RTK-GPS used
in the Weather Vaning DP experiments mentioned above.
As in the experiments with Weather Vaning DP, the required thrust for CPP was calculated using
formula (5), and that for BIT formula (6). However, in view of the fact that when a condition in which
RP fails to catch up with SP continues, the required thrust becomes excessively large if integral control
is exercised, hence no integral control was applied. The values shown in Table i were used as
proportional control and differential control feedback gains, and WFF control was also applied. In all
experiments, the position of RP was fixed at the bow.
Shown here are the results of the automatic tracking experiment on SP moving along a square carried
out on March 10, and those on SP moving along semicircles carried out on March 11.
4.2 Automatic tracking experiment with SP moving along a square
Automatic tracking experiments were carried out with SP moving counter-clockwise along a 200 m
square at a speed of 2 knots. SP starts from the bottom right corner, and moves N, W, S and then E to complete a square loop. Because the course alters as much as 90° at each corner, RP on the ship's hull
speed of SP was reduced progressively at a rate of 0.1 knot/sec initiating from a point 40 m before each corner until the speed dropped to 1 knot so that SP reached the corner at this speed. The mean wind direction during the experiments was 18°, and mean speed, 8.3 rn/s.
Fig.9 shows the positions of SP (marked by the circles) and the ship's hull of the Shioji Maru plotted every 30 seconds from the start of the experiment. To help view the positional relationships between
RP and SP, they are connected with straight lines in the figure. It may be seen from Fig.9 that the
distance between SP and RP increases after SP starts moving from each corner, but RP nearly catches up with SP when it advances by approximately 2/3 of the distance of each side. At each corner, the Shioji
Maru was steered to hard port by operating BIT to its full capacity of power output, while applying
crash astern, as a result, the ship smoothly follows the subsequent course without showing significant overshooting, as the figure depicts.
While tracking SP advancing true W or true E, the ship's hull is exposed to beam winds blowing
from NNE, but when the ship proceeds true W, after tracking has reached nearly a steady state, she makes a large drift retaining her heading direction approximately 15° in the wind, and when the ship
proceeds true E, after tracking has reached nearly a steady state, she makes a large drift keeping her
heading approximately 20° in the wind.
Fig.10 shows time series data for the longitudinal distance Ax between RP and SP and transverse distance Ay between RP and SP. On Ax, it may be seen from the figure that there is an approximately 20 m overshoot (minimum in the figure) approximately 20 seconds after SP passed each corner. SP
then proceeded along the subsequent side and the bow turned towards SP. As a result, Ax sharply
increased with a maximum departure of 20 m or thereabouts. A x then decreased as RP caught up with
SP.
On Ay, it sharply increased as SP proceeded along the subsequent side after passing each corner
with a maximum value of approximately 30 m. Then, A y sharply decreased as the bow turned towards
SP. During the second half of the passage when SP proceeded true W or true E (between 350 to 400
seconds and 800 to 850 seconds after starting the experiment), A yj was maintained at approximately 13
m. During this period, the Shioji Maru made a large drift with heading in the bow wind as the ship was exposed to beam winds.
In this attempt, SP which was proceeding on a square at a speed of 2 knots, was automatically
tracked satisfactorily, while SP was exposed to winds at an approximate speed of 8 rn/s. Although it
was difficult to alter course sharply by 90 degrees at each corner, overshoots were suppressed to a
relatively small range by reducing the speed of SP before it reached the corner.
4.3 Automatic tracking experiment with SP moving along semicircles
Subsequently, automatic tracking experiment was carried out on SP moving along semicircles (a reverse letter S) at a speed of 2 knots. The radius of each semicircle forming a reverse letter S was 100 rn, and SP moved from the south end to the north end of a reverse letter S oriented to NS. 1f SP was
moved at a speed of 2 knots from the starting point, RP was considered not to satisfactorily track RP,
hence, it was decided to increase the moving speed of SP from O to 2 knots at a rate of 0.1 knot/sec
immediately after starting. The mean wind direction was 32 degrees and the mean wind speed was 5.6
rn/s.
Fig.11 shows the positions of SP (marked by the circles) and the hull of the Shioji Maru plotted
every 30 seconds after starting the experiment. In the figure, SP and RP are connected with straight
lines. It can be seen from Figli that during the period in which SP moved clockwise on the first semicircle, RP tracks relatively well SP. However, during the period in which the SP moved
-8-counter-clockwise on the second semicircle, RP gradually departed towards the starboard side of SP (outside the semicircle). This may be ascribable to: centrifugal force acting upon the ship in turning
motions, while no centrifugal force is taken into account for the thrust control of BIT; and the WFF
control reducing the thrust of BIT, causing the how of the ship to turn to port, by a magnitude
corresponding to the wind pressure acting on the starboard side hull.
Fig.12 shows time series data for longitudinal distance Ax and transverse distance Ay between RP and SP. Discontinuities are found ¡n Ax and Ay curves shown in the figure in the vicinity of 320 seconds after starting the experiment, but these are considered to have resulted from a programming error in which the speed was assumed to be zero for approximately 2 seconds at the junctions of the two
semicircles. It may be seen that Ax is confined to within approximately 18 m, and RP was not retarded much behind SP. A y gradually increased its negative value after the start, reaching approximately -12
m in the vicinity of the junction of semi-circles, and then turned into positive values with a gradual
increase to reach approximately 18 m near the goal.
In this automatic tracking experiment, it can be said that SP moving at a speed of 2 knots along a course like a reverse letter S formed by two semicircies can successfully be tracked, while receiving winds at a speed of approximately 6 m/s. However, the tendency for RP to be liable to deviate out of the semicircle needs system improvement by modifying the control algorithm.
5.
Conclusions
The control methods for the Weather Vaning DP, which has newly developed, were explained and the results of dynamic positioning experiments and automatic tracking experiments carried Out with the Shioji Maru were reported above.
The experimental results for the Weather Vaning DP demonstrated the following.
The Weather Vaning DP system stabilizes when the bow of the ship is in the wind. As a result, highly accurate DP is available with a small thrust output of the bow thruster.
Feedback control that calculates the wind pressure acting on the ships hull and outputs thrust to cancel it is very effective for damping transverse motions of the ship's hull iii strong winds.
Differential feedback control is indispensable for a good DP.
As the position of RP on the hull of a ship subjected to dynamic positioning is moved from the
bow thruster position towards the stern, DP becomes unstable. When the position of RP was set amidships, the DP was very unstable.
Stable dynamic positioning is also available, even if RP is located forward of the bow.
In automatic tracking experiments relying upon Weather Vaning DP control methods, a specified point moving on a square and semicircies could be successfully tracked at an accuracy that is sufficient
for practical purposes. Throughout the experiments, RTK-GPS was fully capable of supplying
positional information at the high accuracy needed for dynamic positioning and automatic tracking. To upgrade the dynamic positioning accuracy of the Weather Vaning DP in the future, it is planned
to carry out simulations using a model of ship's motions to determine optimum feedback gains. In
addition, drifting forces of waves not considered in these experiments will be calculated, and by adding
it to the wind pressure, new feedforward control method will be developed. It is also planned for
automatic tracking to determine optimum feedback gains from simulations, and to carry out a variety of tracking experiments on specified points moving of diverse routes at various speeds.
the Tokyo University of Mercantile Marine, Mr. Tadatane Okazaki of Ship Research Institute, Ministry of Transport, Capt. Hiroaki Fukui, Master and Mr. Noriki Hirose, Chief Engineer of the Shioji Maru, a training ship of the Tokyo University of Mercantile Marine, Mr. Toshiyuki Houa, former Chief Engineer
of the Shioji Maru, and the crew members of the Shioji Maru for their valuable guidance and
cooperation for this study.
References
J. A. Pinkster: Dynamic Positioning of Large Tanker at Sea, Offshore Technology Conference
5208, pp. 459-.476, 1986.
Takeshi Kobayashi, Master's thesis at the Tokyo University of Mercantile Marine: A Study on
Wind Forces Acting on the Hull and Mooring Line Tension, pp. 11-35, 1998.
Questions and Answers
Q Junji Fukuto (Ship Research Institute)
Isn't it possible to stabilize control using rudder force when the reference point is brought amidships?
A Hitoi Fukuda
The great merit of the Weather Vaning DP is that it enables dynamic positioning with the least bow
thruster force because the bow of a ship naturally stands into the wind. To enjoy most of the benefits offered by this technique, it is necessary to locate the reference point in the vicinity of the bow. If the
reference point is positioned amidships or in its vicinity, it is considered to be difficult to bring the bow into a strong wind for stable dynamic positioning even if the rudder and the stern thruster are used.
Q Yasuo Yoshimura (Sumitomo Heavy Industries, Ltd.)
I understand that the proposed technique assumes a system leaving directional control of a ship to
external forces without providing any interventions for control. However, steering with the rudder can produce a sizeable transverse force, while the main engine is running ahead, and I think it is not wise to overlook utilizing such a force. Although there are such drawbacks as no rudder force is produced when the main engine is running astern, and control requires some degree of complexity, I strongly wish
you to assess a control system utilizing such rudder effects.
A Hitoi Fukuda
The Weather Vaning DP independently uses the main engine driven CPP and the bow thruster for
producing longitudinal thrust and transverse thrust, hence the control algorithm is very simple. Producing transverse forces with the rudder requires ahead thrust, and dynamic positioning technique incorporating
the rudder force requires to change the control algorithm to a great extent. Concerning automatic
tracking technique, I consider tracking can be eased by utilizing the transverse force generated by the rudder force, and I will study the proposal in future experiments.
Q Takashi Morimoto (Yokogawa Electronic Equipment Co., Ltd.)
I would like to ask you how to determine feedforward gains. E-lave you any assessment criteria for determining feedforward gains; for example, variance of deviation from a specified point or optimum gain method based on other criteria? There are some criteria such as LOG feedforward that models
wind drag and several others, but what did you use?
A Hjtoi Fukuda
Wind
North
Bow thruster
y
xl
y
Fig. i Weather vaning dynailiic positioning system
Specified point
X
Current
Reference point
GPS antenna
CPP
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
IWind force coefficient of Shioji Maru
-.- Cx: Fore and aft wind force coefficient
-n- Cy: Lateral wind force coefficient
0
15
30
45
60
75
90
105
120
135
150
165
180
PID + WFF control
Ref. point : Bow
3rd Dec. 1998
09:59:30
-10:08:30 JST
V V5J
V/
/
Mean wind
Dir.
: 20.9 deg.
Speed:
9.2 rn/s
Fig. 3(a) Ship's contour plots (Time interval: 10 seconds)
(PID + wind feed-forward control)
r
PID control
Ref. point : Bow
3rd Dec. 1998
10:24:02 - 10:33:02 JST
50
V5
Mean wind
Dir.
: 19.9 deg.
Speed: 11.3 rn/s
Fig. 3(b) Ship's contour plots (Time interval : 10 seconds) (PID control)
,-X lo
20
-30
30 ..- 20 (1) 10 E O>-10
20
30
o NO CONTROL 240 480 720 960 1200Time (seconds)
i 10
20
30
30 20 (J)l0
E O>-10
20
30
oi___
L MWAIIAUI
VIVAtIV IV
V
V,
240 480 720 960 1200Time (seconds)
NO CONTROL 30 20 (1, 1 e) 10 e)r
3rd Dec. 1998
14:33:38
- 14:42:38 JST
PI + WFF control
Ref. point : Bow
V
N
Mean wind
Dir.
: 22.1 deg.
5 i M
Speed: 14.1 rn/s
L rPID + WFF control
Ref. point: Bow
3rd Dec. 1998
14:54:38 - 15:03:38 JST
L
V J5i
\/
51:
VMean wind
Dir.
: 25.4 deg.
Speed: 13.1 rn/s
Fig. 5(a) Ship's contour plots (Time interval : 10 seconds) Fig. 5(b) Ship's contour plots (Time interval : 10 seconds) (PI + wind feed-forward control) (PID + wind feed-forward control)
51: V
r
PID + WFF control
Ref. point : Bow
3rd Dec. 1998
15:03:38
- 15:06:38 JST
i
5
/
Mean wind
Dir.
: 18.8 deg.
Speed: 12.2
rn/s
Fig. 5(c) Ship's contour plots (Time interval: 10 seconds)
(No control)
V
/
x 10
20
30
30 20 10 a) -4-, 0< 20
30
40
O 240 480 720 960Time (seconds)
A
AIFI1IJIV
y
x 10
20
30
30 20 C',.10
EU
>-10
20
30
O NO CONTROL 240 480 720 960Time (seconds)
Fig. 6(a) Time series data of Ax and Ay (PI + wind feed-forward control) Fig. 6(b) Time series data of Ax and Ay (PID
+ wind feed-forward control)
NO CONTROL 30 -S 20 C,, 10 Q)
E°
30 20 a 10 4-, a)E°
C o 150 a) - bu . a) loo
o E-o
-oE
C 50 '-4-o 20lo
o 2500 2000 1500 1000 500 o -500 2000 1000 o-10
-20
4) o -NO CONTROLL.-.
0 120 240 360 480 600 720 840 960 1080Time (seconds)
Fig. 7 Time series data of wind direction from bow, wind speed, required CPP thrust,
Q) . -4-) o--1000 -2000 -c
LLJ
-3000 20 C-)a
10 O O 2 (3-2
<a
4PID + WFF control
Ref. point: Bow
2nd Dec. 1998
09:03:32
- 09:12:32 JST
V V5
V/
Mean wind
Dir.
: 34.9 deg.
Speed:
8.4 rn/s
PID + WFF control
Ref. point : Midship
2nd Dec. 1998
14:10:00
-14:19:00 JST
o
V5G
V5
V/
/
Mean wind
Dir.
: 28.9 deg.
Speed:
9.0 rn/s
Fig. 8(a) Ship's contour plots (Time interval: 10 seconds) Fig. 8(b) Ship's contour plots (Time interval : 10 seconds) (Reference point : bow) (Reference point : midship)
r
PID + WFF control
Ref. point: L/4 aft of bow
2nd Dec. 1998
14:34:34
- 14:43:34 JST
50
VFig. 8(c) Ship's contour plots (Time interval: 10 seconds) (Reference point : L/4 aft of bow)
N
Mean wind
Dir.
: 63.1 deg.
Speed:
8.6 rn/s
PID + WFF control
Ref. point : L/4 forward of bow
2nd Dec. 1998
14:55:49
-15:04:49 JST
J
V -j50
V V V/
Mean wind
Dir.
: 40.9 deg.
Speed:
7.3 rn/s
Fig. 8(d) Ship's contour plots (Time interval: 10 seconds) (Reference point : L/4 forward of bow)
J
V10th Mar. 1999
50
15:35:49
-
15:50:32 JST
V
Fig. 9 Ship's contour plots (Time interval : 30 seconds)
40
30
20
30
Automatic tracking on 200 X 200 m square (left turn)
0
200
400
600
800
1000
Mean wind
Dir.
: 31.9 deg.
Speed:
5.6 rn/s
0
50M
11th Mar. 1999
14:39:32
-
14:49:55 JST
Fig. 11 Ship's contour plots (Time interval : 30 seconds)
5
10
15
Automatic tracking on semicircies (radius
100 m)
25
20
15 Ci) a)10
4) cl) E'-5
O100
200
300
400
500
600
700
Time (seconds)
Connecting point of
Goal
Table i Feed-back gains used for dynamic positioning experiments
ax
60 kg/rn
b = -1000 kg/(m/s)
Cx =
0.2 kg/(m"s)
Table 2 Mean and standard deviation of wind direction from bow, wind speed, Ax, Ay, Fx and Fy (From the 6th minute to the 15th minute after starting experiment)
PID + WFF Control PID Control
Mean(S.D.) of wind direction
from bow 5.5 (12.2) deg 5.5 (15.3) deg
Mean(S.D.) of wind speed 9.2 (1.3) rn/s 11.3 (1.3) rn/s Mean(S.D.) of Lx
0.6 (4.0)
rn 1.1 (4.0) mMean(S.D.) of Ay 0.0 (2.0) m
2.1 (8.9)
rn Mean(S.D.) of Ex 377 (225) kg 608 (289) kg Mean(S.D.) of Fy50 (311) kg
112 (558) kg
Table 3 Mean and standard deviation of wind direction from bow, wind speed, Ax, Ay, Fx and Fy (From the 6th minute to the 15th minute after starting experiment)
Pl + WFF Control
I PID ± WFF Control Mean(S.D.) of wind threcton
from bow 35.1 (20.3) deg 36.4 (7.8) deg
Mean(S.D.) of wind speed 14.1 (1.5) rn/s 13.1 (1.2) rn/s Mean(S.D.) of Ax 1.] (7.8) m
-0.3 (2.4) m
Mean(S.D.) of Ay -3.0 (16.0) m 0.3 (2.3) m Mean(S.D.) of Fx 774 (508) kg 694(171) kg Mean(S.D.) of Fy -1747 (1690) kg -1001 (381) kg
Table 4 Mean and standard deviation of wind direction from bow, wind speed, Ax, Ay, Fx and Fy (From the 6th minute to the 15th minute after starting experiment)
Position of reference point -
Bow(
Midship
[/4 aft of bow
L/4 forward of bowMean(S.D) of wind direction
from bow 15.7 (8.1)
deg
32.8 (1 9.1) deg 36.3 (9.2) deg 23.9 (8.8) degMean(S.D.) of wind speed 8.4 (1.3)
rn/s
9.0 (1.0)rn/s
8.6 (1.0)ni/s
7.3 (1.2)rn/s
Mean(S.D.) of Ax
-0.9 (4.3)
rn-0.8 (5.6)
rn-0.2 (2.8) m
-0.6 (3.4) m
Mean(S.D.) of Ay
-1.1 (2.3)
rn-5.4 (5.7) m
0.8 (1.]) m
-1.1 (1.1) m
Mean(S.D.) of Ex