Risk based method admittance policy of large ferries

Zeszyty Naukowe 20(92) 33

Scientific Journals Zeszyty Naukowe Maritime University of Szczecin Akademia Morska w Szczecinie

2010, 20(92) pp. 33–40 2010, 20(92) s. 33–40

Risk based method admittance policy of large ferries approaching to Ystad Port

Zasady wprowadzania dużych promów do portów oparte o metody szacowania ryzyka na przykładzie Portu Ystad

Lucjan Gucma

Maritime University of Szczecin, Faculty of Navigation, Institute of Marine Traffic Engineering Akademia Morska w Szczecinie, Wydział Nawigacyjny, Instytut Inżynierii Ruchu Morskiego 70-500 Szczcin, ul. Wały Chrobrego 1–2, e-mail: [email protected]

Key words: navigational risk, under keel clearance, ship accident

Abstract The paper presents method of safety water depth evaluation for ship approaching to ports. The method

utilizes two models: real time simulation model used for determination of ships speed approaching to given

port, and Monte Carlo model for determination of probability of accidental collision with the bottom.

Minimal safety depth has been calculated with use of probabilistic acceptance criteria presented in the paper.

Słowa kluczowe: ryzyko nawigacyjne, zapas wody pod stępką, wypadek statku

Abstrakt W artykule zaprezentowano metodę szacowania bezpiecznego zapasu wody pod stępką statków podchodzą-

cych do portów morskich. Metoda składa się z dwóch osobnych modeli: modelu symulacji czasu rzeczywi-

stego oraz modelu Monte Carlo do oceny zapasu wody pod stępką. Aby ocenić bezpieczeństwo, zastosowano

metody ryzyka akceptowalnego.

Introduction

Marine Traffic Engineering (MTE) research

team of Maritime University of Szczecin since 70-

-ties is engaged in research works concerned with

evaluation of navigation safety for port design and

optimization of water areas. In this paper author’s

intention is to present complex researches for de-

termination of safety waterway depth for approach

ships. It is widely known that squat of ships is most

important factor affecting under keel clearance

(UKC). Unfortunately the major factor of squat –

ship’s speed, cannot be reduced especially in diffi-

cult navigational conditions. This happens because

it is necessary to maintain high speed to keep

proper manoeuvrability. Only one way to determine

ships speed during approach especially in non exist-

ing solutions is to use simulation models (real time

simulators) where navigators can freely adjust

speed of ships to given conditions.

From the other side speed is not the only one

factor of under keel clearance (UKC). To deal with

such complex phenomenon as under keel clearance,

Monte Carlo method is most suitable in author’s

opinion and is applied in presented researches.

With use of probabilistic acceptable risk criteria

it is possible to achieve the results of presented

method: the minimal required depth for approach-

ing ships in given area.

Modernisation of Ystad Port by simulation researches

The researches described in this paper are

focused on Ystad Port modernisation which is one

of the latest research studies of MTE team. The

main aim of researches has been focused on [1]:

1. Determination of optimal parameters of:

– approach channels to reconstructed port of

Ystad with respect to shape, width and depth;

Lucjan Gucma

34 Scientific Journals 20(92)

– inner and outer port breakwaters with respect

to its shape with respect to waving in port;

– turning places with respect to its shape and

optimal depth;

– two new berthing places in inner port in

respect to its shape, length, depth, maximal

energy of ships contact, maximal speed of

ships propeller and bowthruster streams on

the bottom.

2. Determination of safety conditions of port

operation in respect to:

– admissible meteorological conditions for

given kind of ships and manoeuvres;

– other navigational conditions and limitations

like presence of other ships on berths, use of

position fixing systems on approach, naviga-

tional markings, vessel traffic service.

3. Determination of manoeuvring procedures

during berthing and unberthing for different

kind of ships and propulsion systems.

4. Determination of under keel clearance by Monte

Carlo method.

5. Determination of usage of main engine during

entrance.

6. Determination of ferry distances to the most

dangerous objects.

7. Carrying out most typical emergency runs and

describe necessary emergency action for the

captains.

Two characteristic Ro-Pax ships have been

chosen as typical for development of Ystad Port.

M/f “Wolin” is midi size ferry originally built as

train ferry. The second ship m/f “Piast” is newly

designed ferry build for Unity Line by Stocznia

Szczecińska Shipyard. Most important parameters

of ferries are presented in the table 1.

Fig. 1. General arrangement of m/f “Piast” [2]

Rys. 1. Ogólny projekt promu m/f „Piast” [2]

Fig. 2. Photo of m/f “Wolin” ferry [2]

Rys. 2. Fotografia promu m/f „Wolin” [2]

Table 1. Main parameters of “Wolin” and “Piast” as typical

ferries operated on the Baltic Sea area [2]

Tabela 1. Główne parametry typowych promów pływających

na obszarze Morza Bałtyckiego na przykładzie promów

„Wolin” i „Piast” [2]

Parameter “Piast” “Wolin”

Operator

(route)

Unity Line

(Świnoujście–Trellborg) Unity Line

Building year 1986 / 2002 rebuild expected 2009

Length –

LOA 207 m 188.9 m

Breadth 27 m 23.7 m

Draft 6.3 m 5.9 m

DWT 8000 t 5143 t

Machinery total 21.600 kW

at 500 rpm

total 13.200 kW

at 600 rpm

Propeller

2 variable pitch

propellers turning inside

at 145 rpm

2 variable pitch

propellers turning

inside at 150 rpm

Speed approx. 21 kn 18 kn

Rudder 2 70 deg. active 2 70 deg. active

Bowthrusters 2 2.300 kW 2 1100 kW

Sternthruster 1 1.200 kW 1 736 kW

Lateral wind

area approx. 3000 m2 approx. 2700 m2

The most important aim of Ystad Port moderni-

sation is to provide access to the port by ferries up

to 210 m length and enable future port development

in the future to serve ships of 240 m length [1].

Three most important changes are planned (Fig. 3):

1. Building two instead of three ferry quays in

inner port,

2. Design of new turning place in outside port,

3. Shortening of inner western breakwater,

4. Lengthening of outer breakwater to provide

shelter in western winds of turning place.

Fig. 3. Proposed changes in berths, turning place and

breakwaters arrangements in port of Ystad [2]

Rys. 3. Proponowane zmiany w nabrzeżach, obrotnicy i falo-

chronach planowane w Porcie Ystad [2]

400m

B4

B3B1

Line of previous berths

Additionalbreakwater

Breakwaterto shorten



Real time simulation methods applied

Real time simulation the interactive method with

captains and pilots engaged in ships manoeuvring

trials have been applied. This method is assumed as

most reliable and suitable in this kind of research

studies [3]. MTE research team possesses several

kinds of manoeuvring simulators: form self made

limited task with 2D display to modern commercial

full mission simulator with 3D and real control

systems. Both simulators have been applied in

presented researches.

Real time simulation method – limited task simulator

Two classes of hydrodynamic models in MTE

team own limited tasks simulators are utilized. First

class of models are used when only limited parame-

ters are known (usually when non existing ships or

general class of ships are modelled). The second

class models are used when detailed and exact

characteristics of hulls, propellers and steering

devices are known. Additionally real manoeuvring

characteristics are used for validation of models.

In present researches the second model has been

used (m/f “Wolin” exists and sea trials are avail-

able, m/f “Piast” trial parameters has been extrapo-

lated).

The model used in researches is based on

modular methodology where all influences like hull

hydrodynamic forces, propeller drag and steering

equipment forces and given external influences are

modelled as separate forces and at the end summed

as perpendicular, parallel and rotational ones.

The model is operating in the loop where the

input variables are calculated instantly (settings and

disturbances) as the forces and moments acting on

the hull and momentary accelerations are evaluated

and speeds of movement surge, sway and yaw. The

most important forces acting on the model are:

– thrust of propellers,

– side force of propellers,

– sway and resistant force of propellers,

– bow and stern thrusters forces,

– current,

– wind,

– ice effects,

– moment and force of bank effect,

– shallow water forces,

Fig. 4. The main diagram of simulation model [2]

Rys. 4. Schemat modelu symulacyjnego [2]

Lucjan Gucma


– mooring and anchor forces,

– reaction of the fenders and friction between

fender and ships hull,

– tugs forces,

– other depending of special characteristics of

power and steering ships equipment.

The functional idea of the ship manoeuvring

simulation model is presented in figure 4.

Interface of the model is typical 2D chart

interface (Fig. 5). The interface covers information

of ships state (position, course speed, yaw etc),

quay and shore line location, navigational

markings, soundings, external conditions, tug and

line control and control elements of the model. The

model is implemented in Object Pascal with use of

Delphi™ environment and Visual C™ with use of

C++ language.

Fig. 5. Interface of simulation model ferry “Wolin” turning at

outer turning place of Ystad Port (limited task simulator) [2]

Rys. 5. Interfejs modelu symulacyjnego promu „Wolin”,

manewrującego przy obrotnicy zewnętrznej Portu Ystad

(„limited task”) [2]

Real time simulation method – full mission simulator

Kongsberg Polaris™ simulator located at Ma-

rine Traffic Engineering Centre (MTEC) premises

in Maritime University of Szczecin comprises

(Fig. 6):

– one full mission navigation bridge simulator

with 270° visual projection and live marine ship

equipment (DNV class A);

– two part task navigation bridges with 120°

visual projection and mix of real and screen-

-simulated ship-like equipment including one

Voith-Schneider tug console (DNV class B);

– two desktop PC simulators with one monitor

visual projection and one monitor screen-

-simulated ship-like equipment.

All hardware and software forming the Polaris

ship manoeuvring simulator has been granted DNV

certificate for compliance or exceeding the training

regulations set forward in STCW’95 (section

A-I/12, section B-I/12, table A-II/1, table A-II/2

and table A-II/3).

In order to create own ship models a hydro-

dynamic ship-modelling tool is available. This tool

enables creating almost any ship type (controls for

at least two engines with propellers’ controls for

fixed propeller, adjustable pitch propeller and

azimuth; rudder controls adequate for various types

of conventional rudders, active rudders, Z-drive /

azimuth and thrusters) with very high fidelity

hydrodynamics in 6 DOF (surge, sway, heave, yaw,

roll, pitch). The hydrodynamics comprise all known

to state of the art external effects like squat, bank

and channel effects.

Assumptions to safety depth evaluation

The main aim of this part is to determine safety

depth of Ystad waterway. The probabilistic criteria

of acceptable risk is applied and together with

simulation results (real time ships manoeuvring)

used to obtain minimal acceptable (under given

safety conditions) depth of waterway (Fig. 7).

Fig. 6. Bridge A at MTEC (with 270 visual projection) and

captain “at work” [2]

Rys. 6. Mostek A w CIRM (z wizualizacją w zakresie 270°)

oraz jeden z kapitanów podczas pracy [2]



The minimal depth on approach to sea ports is

very important factor which influence safety of

navigation in respect to under keel clearance which

is together with horizontal area the most important

factor of navigational safety.

Following assumptions have been taken into

consideration:

1. Swedish water levels for design:

HHW +1.65

MHW +0.87

MW +0.03

MLW –0.97

LLW –1.47;

2. Water level for calculations – MLW

3. Ship lifetime – 15 years

4. Waterway lifetime – 50 years

5. Ships draught – 6.3 m

6. Ships speed – variable according to simulations

results

7. Wave – 0.4 0 m.

Fig. 7. Method of safety depth evaluation [2]

Rys. 7. Metoda określania bezpiecznej głębokości [2]

Navigational risk

The navigational risk can be defined as proba-

bility of certain losses during expected period of

time (one year / lifetime of ships or waterway):

CPR A (1)

where: PA – probability of serious grounding

accident, C – consequences of accident.

With assumption that accidents consequences

are similar (grounding of passenger ship without

losses) we can expressed risk as probability of

accident only.

Acceptable risk criterion

Probabilistic acceptance criterion is proposed in

this study. Such criteria are widely used in Marine

Traffic Engineering (Dutch, England, Denmark,

and Poland).

Monte Carlo simulations performed in stage II

of researches enabled to find probability of accident

in single passage assumed when UKC < 0 is

expressed as PUKC < 0.

Most of the accidents are not serious. Proba-

bility of serious accident can be calculated with

assumption that serious accidents are 10% of all of

total number of accidents: PSA = 0.1 (so called

Heinrich factor usual assumption in restricted water

areas validated by real accidents statistics see [4]).

Under above assumptions probability of serious

accident PAS can be calculated as:

0 UKCSAA PPP (2)

Intensity of all accidents in given time (ex. one

year) can be calculated as:

ANP (3)

where: N – ship movement intensity per 1 year.

Typical probabilistic criterion for risk of colli-

sion with the bottom is based on Poisson process

the collisions with the bottom are random with

intensity [collision / time] and expected number n

during given time t:

!n

etnP

tn

(4)

where: n – expected number of collision with

bottom in given time period, – intensity in given

time.

No accident probability in given time period t

can be calculated with assumption that n = 0 as:

tenP 0 (5)

The opposite to above safety factor can be

expressed as occurrence at least of one accident in

given time t and expressed as:

tenP 11 (6)

Typical probabilistic safety criterion is proba-

bility of no accident in given time. For example

Dutch criterion on approach to Rotterdam (with

tides consideration) is 10% probability of more then

one accident in 25 years of waterway operation

which is expressed as:

1.011 tenP (7)

(where t = 25 years) which gives t = 0.105.

Assuming that t = 25 years of operation we obtain

= 0.0042 of all accidents per year which lead to

following criterion: one accident in 238 years

period (= /1 ). The criterion comprise all accidents

so with assumption that serious accidents are 10%

of all accidents we can calculate criteria value of

yearly intensity of serious accident as:

00042.01.0 S (8)

Real

time

simu-

lations

Ship

speed

Monte

Carlo

method

Probability

of collision

with

bottom

Minimal

depth

Acceptable risk

Lucjan Gucma


Polish criterion that is used in Marine Traffic

Engineering is slightly less restrictive due to low

traffic intensity, and nature of the bottom in ports

(sand, mud). In Poland it is assumed limit accident

rate per year at the level = 0.007 of all accidents

or = 0.0007 for serious accidents (serious acci-

dent is such accident where special rescue action

should be undertaken) as criterion value (the crite-

rion is based on acceptance of one serious accident

per ships lifetime which equals 15 years, because

ships during this time are not likely to be rebuild in

opposite to waterway which during 50 years of

operation will be rebuild few times most likely).

In further step taking into consideration the

passages of ferries which are around 10 per day in

Ystad Port (N = 10365 = 3650 passages / year) it is

possible to calculate limited probability of collision

with the bottom (accident) in single passage as:

6

accept 1015.13650

0042.0

NPA

Monte Carlo method of under keel clearance on ferry approach

The stochastic model of under keel clearance

evaluation has been presented in [3]. It is based on

Monte Carlo methodology where overall ships

under keel clearance is described by following

mathematical model (Fig. 8):

NSwiSwa

TiHoi THUKC

)(

)()( 0

(9)

where: Hoi the uncertainties concerned with

depth and its determination, Ti the uncertainties

concerned with draught and its determination, Swi –

the uncertainties concerned with water level and its

determination, N navigational and manoeuvring

clearance.

Fig. 8. Concept of probabilistic under keel clearance determi-

nation [2]

Rys. 8. Metoda probabilistyczna określania rezerwy wody pod

stępką [2]

The final model takes into account depth

measurement uncertainty, uncertainty of draught

determination in port, error of squat determination,

bottom irregularity, tides and waves influence are

deciding factors for under keel clearance of ships.

Program is capable to consider above mentioned

uncertainties using distributions and their para-

meters. The following parameters are randomly

selected from their distributions:

1. Depth – hi (averaged in following sections –

100 m, 0 m, +100 m, 200 m, 300 m from the

breakwater),

2. Sounding error – iBS ,

3. Mudding component clearance – iZ

,

4. Draught determination error – iT

,

5. Ship’s heel error – iP

.

Random draught module. User’s entered

draught is corrected for draught determination error

value and ship’s heel error. Iterated draught (Ti) is

calculated as follows:

ii PTi TT (10)

where: T – ships draught [m] assumed as 6.3 m,

iT – draught determination error (assumed

as 0.05 m), iP

– ships heel error (assumed

as 3 degrees). Water level module. Water level PWi can be

automatically load from online automatic gauges if

such exists (Szczecin solution). In these researches

the level has been modelled as normal cut

distribution with parameters (0, 0.1 m).

Depth module. Depth hi has been assumed as

constant in given sections (it varies from 9 before

and near breakwater to 8.5 m inside the port see

figure).

Squat module. Squat (ship sinking due to

decrease of water pressure during movement) is

calculated in three stages. First module calculates

squat with analytical methods used to obtain

moving vessel squat (Huusk, Milword 2, Turner,

Hooft, Barrass 1, Barrass 2). Next standard errors

of each method are applied. Squat model selection

and their standard errors have been verified by

GPS-RTK experimental research [5]. As a result of

the experiment uncertainty of each model has been

assessed and each squat method assigned weight

factor. Method’s weights and statistical resampling

bootstrap method are used later on to calculate final

ship's squat.

Under keel clearance module. Under keel

clearance Zi is determined by using draught, depth,

Minimal dredged depth

Grounding probability

Water level

Draught

Depth

Squat

V

V = 0

fs(s)

fh(h)



water level and squat results which have been

calculated before. Under keel clearance is defined

as:

FWPNiiBSZii iiiOThZ (11)

where: hi – up-to-date depth in each iteration in

sections (sounding from October 2007), iZ

–

mudding component clearance (normal cut

distribution with 0 and 0.1 m), iBS – sounding

error (normal cut distribution with 0 and 0.1 m),

Ti – ships draught with its uncertainty (0, 0.05 m),

Oi – iterated squat (bootstrap model), N –

navigational clearance (0 m), iWP – height of tide

error (0 m), F – wave clearance (wave height

assumed as h = 0.4 m before breakwater, h = 0.2 m

in the breakwater and h = 0 m inside, direction from

ferry traverse).

Results

As the result of combined method Monte Carlo

with simulations results in sections of waterway

(breakwater = 0 m) we obtain parameters of distri-

butions of UKC in distance from breakwater for

real mean depth existing in Ystad Port (Fig. 9).

Fig. 9. Mean actual depth in given sections (sounding from

fall 2007) [2]

Rys. 9. Średnia aktualna głębokość w poszczególnych prze-

krojach (sondaż z jesieni 2007) [2]

Speed of approaching ships has been determined

from real time simulations (extreme conditions

E20 m/s wind) have been applied (Fig. 10). Speed

applied in Monte Carlo model has been calculated

with 95% probability level.

In next step Monte Carlo model described in

Ystad development study has been applied to

determine histograms and parameters of

distributions of UKC in function of ships position

on the waterway. Wave influence has been taken

into account.

In the further step on the basis of Monte Carlo

results the UKC on 95% and squat has been

calculated (Fig. 11). Important for the probability

calculations is mean UKC and standard deviation of

UKC presented in figure 12. Due to lack of

distribution or probabilities of given water levels

the water level assumed in this study is equal to

MLW = –0.97 m.

Fig. 10. Speed of ferry “Piast” in knots on approach with

E20 m/s wind (x = 0 outer breakwater) obtained by mean of

real time researches [2]

Rys. 10. Prędkość promu „Piast” w węzłach na podejściu

do portu przy wietrze wschodnim o prędkości 20 m/s (przyjęto

x = 0 jako główki portu) [2]

Fig. 11. UKC on 95% and 5% level of confidence of m/f

“Piast” approaching with E20 m/s wind (x = 0 outer break-

water) [2]

Rys. 11. Zapas wody pod stępką na poziomie ufności 95%

i 5% promu m/f „Piast” wchodzącego do portu przy wietrze

wschodnim o prędkości 20 m/s (przyjęto x = 0 falochron

zewnętrzny) [2]

Fig. 12. Mean UKC and standard deviation of UKC of “Piast”

ferry (T = 6.3 m) entering to Ystad Port [2]

Rys. 12. Średni UKC i odchylenie standardowe UKC promu

„Piast” (T = 6,3 m) wchodzącego do Portu Ystad [2]

7

7.25

7.5

7.75

8

8.25

8.5

8.75

9

9.25

9.5

9.75

10

-200 -100 0 100 200 300 400

actual depth

breakwater[m]

[m]

-6

-4

-2

0

2

4

6

8

10

12

14

-600 -400 -200 0 200 400 600 800

Serie1

Serie2

Serie3

Serie4

Serie5

Serie6

Serie7

Serie8

Serie9

Serie10

Serie11

Serie12

Serie13

Serie14

Serie15

Serie16

Serie17

Serie18

mean

1wej1E20_2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

-150 -100 -50 0 50 100 150 200 250 300 350

UKC_95%

squat

UKC_5%

breakwater

[m]

[m]

1wej1E20_2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

-200 -100 0 100 200 300 400

mean ukc

st.dev. ukc

breakwater[m]

[m]

Lucjan Gucma


Final calculation of required depth (H) for

distribution with parameters m and to fulfill

assumed Dutch criterion is based on following

formula:

H

AmA PxxfP 6accept),( 1015.1d)(1 (12)

The results as required depth on approach to

Ystad Port are presented in figure13.

Fig. 13. Minimal depth in Ystad Port with criterion

PA ≤ PA–accept = 1.1510–6 [2]

Rys. 13. Minimalna głębokość w Porcie Ystad z założeniem

PA ≤ PA–accept = 1.1510–6 [2]

Conclusions

Minimal depths for ferry Piast (T = 6.3 m) in

function of distance to outside breakwater heads are

presented in figure 13. Minimal safety depths to

fulfil criterion are varied from 9.2 m before

breakwater heads to 8.3 m inside the avanport.

Probabilistic safety criterion (so called Dutch

criterion used in Rotterdam Port) has been applied

(acceptable level of 0.0042 accidents per year). On

the basis of this the limited probability of hitting the

bottom accident have been evaluated and used for

minimal depth in Ystad Port.

References

1. Computer Simulation for Port Design and Safe Manoeu-

vring of ships Simulation researches of m/f “Piast” con-

ducted by means of PC Ship Manoeuvring Simulator and

Full Mission Simulator. Stage I and II. Research Work

Maritime University of Szczecin, Szczecin 2008.

2. GUCMA L.: Wytyczne do zarządzania ryzykiem morskim.

Wydawnictwo Naukowe AM. Szczecin 2009.

3. GUCMA L.: Risk Modelling of Ship Collisions Factors with

Fixed Port and Offshore Structures. Maritime University of

Szczecin, Szczecin 2005.

4. SAVENIJE R.PH.: Probabilistic Admittance Policy.PIANC

Bulletin, Bruxelles, 1996, No. 91.

5. Determination of squat of m/f “Śniadecki” by RTK method

in Świnoujście Port. MUS Research work 2006.

Others

6. ARTYSZUK J.: Towards a Scaled Manoeuvring Mathemati-

cal Model for a Ship of Arbitrary Size. Scientific Bulletin,

Maritime University of Szczecin, Szczecin 2005.

7. GUCMA L., GUCMA M., TOMCZAK A., PRZYWARTY M.:

Experimental determination of squat and trim of sea ferry

“Jan Śniadecki” on approach to Świnoujście port by means

of RTK method (in Polish) Proc. of 15 NavSup Confe-

rence, Gdynia 2006.

8. GUCMA L., JANKOWSKI S.: Method of determining probabi-

listic models of propeller streams speed at the bottom of

manoeuvring ships. Proceedings of the 9 International Sci-

entific and Technical Conference on Marine Traffic Engi-

neering, Szczecin 2001.

9. GUCMA L., SCHOENEICH M.: Probabilistic model of under

keel clearance in decision making process of port captain.

Proc of TransNav Conference, Gdynia 2007.

10. IRIBARREN J.R.: Determining the horizontal dimensions of

ship manoeuvring areas. PIANC Bulletin, Bruxelles, 1999,

No. 100.

11. VASCO COSTA F.: Berthing manoeuvres of large ships. The

Dock and Harbour Authority. 1969.

7

7.25

7.5

7.75

8

8.25

8.5

8.75

9

9.25

9.5

9.75

10

-200 -100 0 100 200 300 400

minimal depth

breakwater[m]

[m]

Documents

Risk based method admittance policy of large ferries