yeyes jasnaoe.pdf

Estimation Method for Dragged Anchor Accident Frequency on Subsea Pipelines in Busy Port Areas 173

Theoretical Study on the Motion of ROV with Crawler System, Proc. MTS/IEEE Oceans Conf., 2010.

5) T. Inoue, T. Katsui, J. Tahara, K. Ito, H. Yoshida, S. Ishibashi, K. Takagi : Experimental Research on Movability Characteristic of Crawler Driven ROV, Proc. MTS/IEEE Oceans Conf., 2008.

6) T. Inoue, T. Katsui, H. Murakami, J. Tahara, T. Shinbori: Preliminary Research on the Thruster Assisted Crawler System for a Deep Sea ROV, Proc. IEEE Oceans Conf., 2009.

7) T. Inoue, T. Katsui, H. Murakami, K. Takagi: Crawler System for Deep Sea ROVs," Marine Technology Society Journal, Vol. 43, No. 5, pp 1-8. 2009.

8) T. Inoue, T. Katsui, H. Murakami, J. Tahara: Research on Stability of Crawler System for Deep Sea ROV,” Proc. ASME 28th Int. Conf. Ocean, Offshore and Arctic Engineering, 2009.

9) T. Katsui, H. Murakami, S. Kajikawa, T. Inoue: Moving Performance of Crawler Driven ROV on the Inclined Sea Bottom, Proceedings of OMAE2010, 2010.

10) T. Katsui, S. Kajikawa, T. Inoue: Numerical Investigation of Flow around a ROV with Crawler Based Driving System, Proceedings of OMAE2012, 2012.

11) T. Katsui, T. Inoue, M. Akashi: Evaluation of moving capability of crawler driven ROV considering cable tension, Proceedings of OMAE2013, 2013.

12) T. Katsui, M. Akashi, S. Kajikawa, T. Inoue: The Motion Characteristics of Crawler Driven ROV Moving over Bumps, Proceedings of OMAE2011, 2011.

13) T. Inoue, K. Takagi, T. Shiosawa: Flipper Type Crawler System for Running on the Irregular Seafloor, Proc. IEEE Oceans Conf., 2010.

14) T. Inoue, K. Takagi, T. Shiosawa: Experiments of Flipper Type Crawler System at Sea to Verify Running Performance on Sandy or Irregular Steep Terrain Seafloor, Proc. IEEE Oceans Conf., 2011.

15) T. Inoue, H. Osawa, K. Takagi, A. Asada, M. Mochizuki, T. Katsui, S. Tsukui: Hard Rock Coring at the Steep Terrain Seafloor with the Flipper Crawler ROV,” Proc. Int. Sym. on Underwater Intervention, 2012

Estimation Method for Dragged Anchor Accident Frequency on Subsea Pipelines in Busy Port Areas

1

by Yeyes Mulyadi*, **, Student Member Eiichi Kobayashi*, Member Nobukazu Wakabayashi*, Member Trika Pitana**, Member Wahyudi **

Eko Prasetyo***

Summary

Recently, many subsea pipelines have been developed in busy port areas, including industrial ports, recreational areas, fishing areas, and other port facilities. Under busy ship traffic conditions, these subsea pipelines are likely to be exposed to the risk of damage by a dragged anchor or other dropped objects. In this paper, a model is proposed to estimate the dragged anchor accident frequency on subsea pipelines based on the concept introduced by Fujii. The proposed model is approached by estimating the number of dragged anchor candidates on subsea pipelines, Na, based on an analysis of the anchor stopping distance in a critical subsea pipeline area. The causation probability Pc is estimated using a Bayesian network method that is modified from the model of Det Norske Veritas (DNV) and Hänninen. Various factors are considered to estimate Pc, including the human factor, weather factor, technical factor, and support factor. Automatic Identification System (AIS) and Geographic Information System (GIS) data are combined to estimate the lateral probability distribution of the ship traffic around a subsea pipeline area. A case study of anchors dragging on the subsea pipeline in the Madura Strait of Indonesia is performed to demonstrate the implementation of the proposed model. The proposed model is validated by comparing the results for the estimated dragged anchor accident frequency on the subsea pipeline using the proposed model with the actual accidents recorded in the Madura Strait. The results of this validation analysis show that there is reasonable agreement.

1. Introduction

The natural gas demand in the Association of Southeast Asian Nations (ASEAN) is growing rapidly, in line with the economic growth in that region. Natural gas has become a vital source of energy, especially for power generation, petrochemicals, and industrial applications. The consumption of natural gas as a primary energy resource by the major ASEAN members was around 31% in 20081).

Many natural gas sources exist in offshore areas such as the Andaman Sea, Natuna Sea, Java Sea, and Makasar Strait. In order to shorten the transportation distances and minimize the costs, this natural gas is transported by subsea pipelines. Therefore, the domestic subsea pipeline networks in some countries in the ASEAN region are increasing rapidly. Additional cross-border subsea pipelines are either under construction or being planned in the future2). By 2008, the ASEAN members had constructed less than 10 cross-border subsea gas pipelines. These cost $14.2

billion and traversed 3952 km to transport 3095 million cubic feet (mcf) of gas per day. In the next project, in Indonesia, around 3588 km of new subsea pipelines will be constructed to connect the islands of Sumatra, Java, and Kalimantan3).

The challenge facing natural gas distribution using subsea pipelines involves the operational difficulties that are associated with transportation security and environmental protection. There are several threats that can cause the failure of a subsea pipeline, such as external impacts, mechanical defects, corrosion, and natural hazards. The subsea pipelines located near busy port areas are likely to be exposed to the risk of damage by dragged anchors and other dropped objects. Such damage may result in potential risks to people, environments, and economic loss4).

A review of the Health and Safety Executive (HSE) and Pipeline and Riser Loss of Containment (PARLOC) publications shows that there were 11 recorded incidents of anchor damage for over 25,000 km of subsea pipelines in the North Sea area during 1996–20015). In the ASEAN region, in the offshore of Vietnam, there were two dragged anchor accidents on the subsea pipeline in 19946).

This study develops a model based on investigations of an actual dragged anchor accident in the Madura Strait of Indonesia. In this model, some aspects are investigated and used:

(a) An analysis of the anchor stopping distance after a ship drops an emergency anchor is considered in this proposed model.

(b) Automatic Identification System (AIS) data are used for a maritime tracking survey.

* Graduate School of Maritime Sciences, Kobe

University ** Institut Teknologi Sepuluh Nopember, Surabaya,

Indonesia *** Meteorology, Climatology, and Geophysics Institute,

Tanjung Perak, Surabaya

Estimation Method for Dragged Anchor Accident Frequency

on Subsea Pipelines in Busy Port Areas

Received 9th January 2013

　日本船舶海洋工学会論文集　第 20 号　 2014 年 12 月　 174

2

2. Literature Review

Several models have been developed by researchers to estimate the dragged anchor accident frequency on a subsea pipeline.

Braestrup7) estimated dragged anchor frequency per year by considering the traffic flow, length of critical pipeline, ship speed, and frequency of anchor drop per year. In this model, the probability of ship passing a critical pipeline area is not considered in the analysis. The critical dragging length for dropped anchor is determined by assumption of 100 m and the frequency of drops per year ( dropf ) is estimated statistic

analysis based on the observed accident data Vitali et al.8) estimated dragged anchor frequency per km-

year by considering the traffic flow, probability of anchor drop, and geometry interaction probability which dependent on ratio between the interaction area and the total area. The interaction area is a function of the ship dimensions which is calculated using pipeline diameter plus two times the ship dimension. In this model, the ship speed and the ship traffic distribution near pipeline location is not considered to estimate the geometry interaction probability. The probability of anchor drop (hazardous scenario) is estimated by using statistic analysis based on accident data, engineering judgment, and experience on previous projects.

ERM method9) estimated dragged anchor frequency per km- year by considering incident rate (incident/year/km2), probability of anchor drop, and drag distance. In this model, the ship speed and the probability of ship passing a critical pipeline area is not considered in the analysis. ERM’s model assesses the probability of anchor drop by using statistic analysis based on the accident data.

Zhang et al.10) applied a statistical method to estimate by combining the subsea pipeline failure rate (per-km-yr) and length of pipeline. In this model, ship speed and the probability of ship passing a critical pipeline area is not considered in the analysis. Zhang’s model assesses the failure rate by using statistic analysis based on the accident data.

Stefani et al.4) assesses dragged anchor frequency per year by combining the frequency of emergency anchoring per unit distance traveled (accident/length), the number of ship per year, mean anchor dragging distance. In this model, the probability of ship passing a critical pipeline area is not considered in the analysis. Stefani’s model assesses the frequency of emergency anchoring per unit distance traveled by using statistic analysis based on the accident data.

3. Modeling

3.1 Model of Dragged anchor accident frequency on Subsea

Pipeline In marine accident analyses, several researchers have

developed models to estimate the accident frequency based on Fujii’s model such as: Pedersen11) developed a model to estimate a frequency of ship-ship collision, Talavera et al.12) modified the model of traffic distribution functions in the Pedersen’s model by using Dempster–Shafer theory, Montewka et al.13) modified Pedersen’s Na model with replace the Dij with the minimum distance to collision (MDTC). The Fujii’s model has inspired in

the models because of its simplicity and robustness11),13). The Fujii’s model can be represented as follows14).

N Na Pc (1) where N is the ship accident frequency per unit time, Na is the

number of ship accident candidates per unit time, and Pc is the causational probability, which is called the probability of failing to avoid an accident. Na reflects the marine geographic condition and intensity of marine traffic. The causational probability is currently estimated in two ways: using the scenario approach or the synthesis approach15).

In this paper, a model is proposed to estimate the dragged anchor accident frequency (Nd) on a subsea pipeline based on the concept introduced by Fujii. The proposed model consists of two main parameters, namely, the number of the dragged anchor candidates on a subsea pipeline (Na) and a causation probability (Pc). Na is developed based on an anchor stopping distance analysis to determine a critical pipeline area and the use of the AIS data to analyze the ship traffic distributions around the subsea pipeline. Pc is developed using a Bayesian network model. The proposed model for the dragged anchor accident frequency on a subsea pipeline is presented as follows.

Nd Na Pc (2)

In the anchor stopping distance analysis, a ship may drop its anchor in an emergency situation to stop the ship’s movement. In this situation, the ship still has an inertia force and can undergo a deceleration on its course as a result of an anchor holding force that is insufficient to immediately hold the ship. Under this condition, the ship will stop at a specific anchor stopping distance. The anchor stopping distance is the total distance required for a ship to completely stop after its anchor has been deployed.

Figure 1 shows a ship passing through a critical subsea pipeline area and dropping an anchor. This may lead to the anchor dragging on the subsea pipeline. The anchor will stop at anchor stopping distance S. The width of the critical subsea pipeline area, D, can be estimated from analysis of the anchor stopping distance.

Fig. 1 Representation of anchor stopping distance and critical

subsea pipeline area.

3.2 Anchor Stopping Distance Analysis The anchor stopping distance analysis is approached using a

static analysis method. The anchor stopping distance is estimated by the change in the momentum equation. Several parameters is used to estimate anchor stopping distance, including the anchor holding force, ship resistance force, and ship inertia force.

Subsea pipeline

D S

VI

y

f(y)Critical area

x

Seabed

Sea surface

yx

3

The anchor and chain of the ship are designed to withstand environmental loads on the ship, such as the current load, wind load, and wave load. The anchor system is not designed to stop the ship under way on, as the momentum exceeds the limit of the anchor holding force. In the case of a ship with initial velocity V1, the change in the momentum equation of motion is presented by Eq. (3).

2 1. - . .m V m V F T (3)

where V1 is the initial velocity when the anchor is dropped on the sea floor, V2 is the final velocity (0), m is the mass displacement of the ship, F is the total resistance force and the anchor holding force, and T is the time required for the ship to stop and is expressed by Eq. (4).

1- m V

TF

(4)

The anchor stopping distance, S, is expressed by the following equation.

2 10.5S V V T

(5)

Thus, the anchor stopping distance for a ship belonging to

class i, specific initial speed ( )1 izV V , final speed 2 0V ,

passing through fairways and the subsea pipeline zone z, is expressed as follows

( ) ( )0.5iz izS V T (6)

3.3 Geometric Probability of Ship in Critical Subsea Pipeline

Area The geometric probability of ship in critical subsea pipeline

area is determined to estimate the probability of the ship passing through the critical subsea pipeline area. A ship passes a fairway in two-dimensional space, in x axis or longitudinal axis and in y axis or lateral axis. The equation of the geometric probability is approached by the lateral probability, in y axis, multiplied by the longitudinal probability, in x axis, of the ship passing through the critical subsea pipeline area, see Fig. 2. The geometric probability is expressed as follows.

( ) ( ) ( )G T Liz iz iz

P P P

(7)

where ( )Giz

P is the geometric probability of a ship class i

passing through a critical subsea pipeline area in a fairway and

subsea pipeline zone z. ( )Tiz

P is the lateral probability of a ship

class i passing through a critical pipeline area in y axis, in a

fairway and a pipeline zone z. ( )Liz

P

is the longitudinal

probability of a ship class i passing through a critical pipeline area in x axis, in a fairway and a pipeline zone z. Here, is the fairway identification number, where fairway 1 is the entering fairway and fairway 2 is the exiting fairway.

The ( )Tiz

P is calculated using the integral of a lateral

probability distribution function, ( ) ( ),zf y in the width of a

critical subsea pipeline area of ship class i in fairway and

subsea pipeline zone z, ( )izD , see Fig.2. The equation of the

lateral probability is expressed as follows.

( )

( ) ( ) ( )iz

zTiz DP f y

(8)

where ( ) ( )zf y is a lateral probability distribution function,

in y axis, of ship passing through subsea pipeline area in fairway and subsea pipeline zone z.

The function of ( )zf y is estimated based on the ship traffic

distribution using the AIS data. The width of the critical subsea pipeline area of ship class i in fairway and subsea pipeline

zone z, ( )izD , can be expressed as follows.

( ) ( )iz izD S Sin (9)

where ( )izS is the anchor stopping distance of ship class i in

fairway and subsea pipeline zone z. Here, is the

identification number of the fairway, where fairway 1 is the

entering fairway and fairway 2 is the exiting fairway.

Figure 2 shows the ships belonging to class i, initial speed( )Viz

, passing through the width of the critical subsea pipeline

areas in fairways 1, 2 and the subsea pipeline zone z.

Fig. 2 Ships in critical subsea pipeline area.

The ( )Liz

P is calculated based on the time ( )iz

t required for a

ship belonging to class i to pass a specific subsea pipeline zone z

in fairway divided by the total time ( )iT for the ship

belonging to class i to pass the total length of fairway . The specific subsea pipeline zone z is determined based on the similarity of the geometric characteristic of the pipeline layout in the fairway. Figure 3 shows ships passing parallel to the subsea pipeline, whereas Fig. 4 shows ships crossing the subsea pipeline. The ships belonging to class i pass a specific subsea pipeline zone

z, kilometer point 1 to 2 (KP1-KP2), with initial speed ( 1,2)izV ,

in fairways 1 and 2, and along the length of the critical subsea pipeline zone, . The equation of the longitudinal probability

( )Liz

P is expressed as follows.

(2)izV

(1)izV

(2)yzf

(1)yzf

(1)izS

(2)izS

(1)izD

(2)izDy

x

Subsea pipeline

Critical area

Critical area


2

2. Literature Review

Several models have been developed by researchers to estimate the dragged anchor accident frequency on a subsea pipeline.

Braestrup7) estimated dragged anchor frequency per year by considering the traffic flow, length of critical pipeline, ship speed, and frequency of anchor drop per year. In this model, the probability of ship passing a critical pipeline area is not considered in the analysis. The critical dragging length for dropped anchor is determined by assumption of 100 m and the frequency of drops per year ( dropf ) is estimated statistic

analysis based on the observed accident data Vitali et al.8) estimated dragged anchor frequency per km-

year by considering the traffic flow, probability of anchor drop, and geometry interaction probability which dependent on ratio between the interaction area and the total area. The interaction area is a function of the ship dimensions which is calculated using pipeline diameter plus two times the ship dimension. In this model, the ship speed and the ship traffic distribution near pipeline location is not considered to estimate the geometry interaction probability. The probability of anchor drop (hazardous scenario) is estimated by using statistic analysis based on accident data, engineering judgment, and experience on previous projects.

ERM method9) estimated dragged anchor frequency per km- year by considering incident rate (incident/year/km2), probability of anchor drop, and drag distance. In this model, the ship speed and the probability of ship passing a critical pipeline area is not considered in the analysis. ERM’s model assesses the probability of anchor drop by using statistic analysis based on the accident data.

Zhang et al.10) applied a statistical method to estimate by combining the subsea pipeline failure rate (per-km-yr) and length of pipeline. In this model, ship speed and the probability of ship passing a critical pipeline area is not considered in the analysis. Zhang’s model assesses the failure rate by using statistic analysis based on the accident data.

Stefani et al.4) assesses dragged anchor frequency per year by combining the frequency of emergency anchoring per unit distance traveled (accident/length), the number of ship per year, mean anchor dragging distance. In this model, the probability of ship passing a critical pipeline area is not considered in the analysis. Stefani’s model assesses the frequency of emergency anchoring per unit distance traveled by using statistic analysis based on the accident data.

3. Modeling

3.1 Model of Dragged anchor accident frequency on Subsea

Pipeline In marine accident analyses, several researchers have

developed models to estimate the accident frequency based on Fujii’s model such as: Pedersen11) developed a model to estimate a frequency of ship-ship collision, Talavera et al.12) modified the model of traffic distribution functions in the Pedersen’s model by using Dempster–Shafer theory, Montewka et al.13) modified Pedersen’s Na model with replace the Dij with the minimum distance to collision (MDTC). The Fujii’s model has inspired in

the models because of its simplicity and robustness11),13). The Fujii’s model can be represented as follows14).

N Na Pc (1) where N is the ship accident frequency per unit time, Na is the

number of ship accident candidates per unit time, and Pc is the causational probability, which is called the probability of failing to avoid an accident. Na reflects the marine geographic condition and intensity of marine traffic. The causational probability is currently estimated in two ways: using the scenario approach or the synthesis approach15).

In this paper, a model is proposed to estimate the dragged anchor accident frequency (Nd) on a subsea pipeline based on the concept introduced by Fujii. The proposed model consists of two main parameters, namely, the number of the dragged anchor candidates on a subsea pipeline (Na) and a causation probability (Pc). Na is developed based on an anchor stopping distance analysis to determine a critical pipeline area and the use of the AIS data to analyze the ship traffic distributions around the subsea pipeline. Pc is developed using a Bayesian network model. The proposed model for the dragged anchor accident frequency on a subsea pipeline is presented as follows.

Nd Na Pc (2)

In the anchor stopping distance analysis, a ship may drop its anchor in an emergency situation to stop the ship’s movement. In this situation, the ship still has an inertia force and can undergo a deceleration on its course as a result of an anchor holding force that is insufficient to immediately hold the ship. Under this condition, the ship will stop at a specific anchor stopping distance. The anchor stopping distance is the total distance required for a ship to completely stop after its anchor has been deployed.

Figure 1 shows a ship passing through a critical subsea pipeline area and dropping an anchor. This may lead to the anchor dragging on the subsea pipeline. The anchor will stop at anchor stopping distance S. The width of the critical subsea pipeline area, D, can be estimated from analysis of the anchor stopping distance.

Fig. 1 Representation of anchor stopping distance and critical

subsea pipeline area.

3.2 Anchor Stopping Distance Analysis The anchor stopping distance analysis is approached using a

static analysis method. The anchor stopping distance is estimated by the change in the momentum equation. Several parameters is used to estimate anchor stopping distance, including the anchor holding force, ship resistance force, and ship inertia force.

Subsea pipeline

D S

VI

y

f(y)Critical area

x

Seabed

Sea surface

yx

3

The anchor and chain of the ship are designed to withstand environmental loads on the ship, such as the current load, wind load, and wave load. The anchor system is not designed to stop the ship under way on, as the momentum exceeds the limit of the anchor holding force. In the case of a ship with initial velocity V1, the change in the momentum equation of motion is presented by Eq. (3).

2 1. - . .m V m V F T (3)

where V1 is the initial velocity when the anchor is dropped on the sea floor, V2 is the final velocity (0), m is the mass displacement of the ship, F is the total resistance force and the anchor holding force, and T is the time required for the ship to stop and is expressed by Eq. (4).

1- m V

TF

(4)

The anchor stopping distance, S, is expressed by the following equation.

2 10.5S V V T

(5)

Thus, the anchor stopping distance for a ship belonging to

class i, specific initial speed ( )1 izV V , final speed 2 0V ,

passing through fairways and the subsea pipeline zone z, is expressed as follows

( ) ( )0.5iz izS V T (6)

3.3 Geometric Probability of Ship in Critical Subsea Pipeline

Area The geometric probability of ship in critical subsea pipeline

area is determined to estimate the probability of the ship passing through the critical subsea pipeline area. A ship passes a fairway in two-dimensional space, in x axis or longitudinal axis and in y axis or lateral axis. The equation of the geometric probability is approached by the lateral probability, in y axis, multiplied by the longitudinal probability, in x axis, of the ship passing through the critical subsea pipeline area, see Fig. 2. The geometric probability is expressed as follows.

( ) ( ) ( )G T Liz iz iz

P P P

(7)

where ( )Giz

P is the geometric probability of a ship class i

passing through a critical subsea pipeline area in a fairway and

subsea pipeline zone z. ( )Tiz

P is the lateral probability of a ship

class i passing through a critical pipeline area in y axis, in a

fairway and a pipeline zone z. ( )Liz

P

is the longitudinal

probability of a ship class i passing through a critical pipeline area in x axis, in a fairway and a pipeline zone z. Here, is the fairway identification number, where fairway 1 is the entering fairway and fairway 2 is the exiting fairway.

The ( )Tiz

P is calculated using the integral of a lateral

probability distribution function, ( ) ( ),zf y in the width of a

critical subsea pipeline area of ship class i in fairway and

subsea pipeline zone z, ( )izD , see Fig.2. The equation of the

lateral probability is expressed as follows.

( )

( ) ( ) ( )iz

zTiz DP f y

(8)

where ( ) ( )zf y is a lateral probability distribution function,

in y axis, of ship passing through subsea pipeline area in fairway and subsea pipeline zone z.

The function of ( )zf y is estimated based on the ship traffic

distribution using the AIS data. The width of the critical subsea pipeline area of ship class i in fairway and subsea pipeline

zone z, ( )izD , can be expressed as follows.

( ) ( )iz izD S Sin (9)

where ( )izS is the anchor stopping distance of ship class i in

fairway and subsea pipeline zone z. Here, is the

identification number of the fairway, where fairway 1 is the

entering fairway and fairway 2 is the exiting fairway.

Figure 2 shows the ships belonging to class i, initial speed( )Viz

, passing through the width of the critical subsea pipeline

areas in fairways 1, 2 and the subsea pipeline zone z.

Fig. 2 Ships in critical subsea pipeline area.

The ( )Liz

P is calculated based on the time ( )iz

t required for a

ship belonging to class i to pass a specific subsea pipeline zone z

in fairway divided by the total time ( )iT for the ship

belonging to class i to pass the total length of fairway . The specific subsea pipeline zone z is determined based on the similarity of the geometric characteristic of the pipeline layout in the fairway. Figure 3 shows ships passing parallel to the subsea pipeline, whereas Fig. 4 shows ships crossing the subsea pipeline. The ships belonging to class i pass a specific subsea pipeline zone

z, kilometer point 1 to 2 (KP1-KP2), with initial speed ( 1,2)izV ,

in fairways 1 and 2, and along the length of the critical subsea pipeline zone, . The equation of the longitudinal probability

( )Liz

P is expressed as follows.

(2)izV

(1)izV

(2)yzf

(1)yzf

(1)izS

(2)izS

(1)izD

(2)izDy

x

Subsea pipeline

Critical area

Critical area


4

( )( )

( ) ( )( ) iziz

i iLiz

Lz

tP

T T

V

(10)

where ( )tiz is the time required for a ship belonging to class i

to pass fairway in subsea pipeline zone z and ( )iT is the total

time required for the ship belonging to class i to pass the total length of fairway .

Fig. 3 Ships passing parallel to subsea pipeline.

Fig. 4 Ships crossing subsea pipeline. 3.4 Model of Dragged Anchor Candidates Na

Na is defined as the number of anchors with the potential to drag on the subsea pipeline if they are dropped from ships.

In this paper, the proposed model of Na is denoted by ( )izNa .

The equation for ( )izNa is developed by multiplying the number

of ships per unit time, ( )izQ , by the geometric probability of the

ships passing through the critical pipeline area, ( )Giz

P , as follows.

( ) ( ) ( )iz iz Giz

Na Q P

( )

( )( )

( ) ( ) ( )

iz

zizz

i iz D

Q Lf y

T V

(11)

where ( )Qiz is the number of ships per unit time that belong

to ship class i in fairway and pass through subsea pipeline zone z.

Thus, the total number of candidates for anchors dragging of

ship class i on the subsea pipeline in a specific subsea pipeline zone z is expressed as follows.

(1) (2)iz

Na Na Naiz iz (12)

where (1)Naiz is the number of dragged anchor candidates that

belong to ship class i in fairway 1 and subsea pipeline zone z, and(2)Naiz is the number of dragged anchor candidates that belong to

ship class i in fairway 2 and subsea pipeline zone z.

3.5 Model of Causation Probability Pc The causation probability Pc for an emergency dropped

anchor by ship is the probability of a ship dropping an anchor in an emergency when the ship passes through the critical subsea pipeline area. In the ship accident models, Pc is currently estimated in either of the following ways: 1) the scenario approach or 2) the synthetics approach16). The Pc that is modeled using scenario approach is calculated based on the statistics from the available historical accident data. Pedersen11) argues that the Pc based on scenario approach can be calculated from accident data at various locations and transformed to consider the analyzed location. The Pc that is modeled using synthetics approach is calculated based on a specific error situation analysis for ships. Det Norske Veritas (DnV)17) and Hänninen et al.18) applied the synthetics approach using a Bayesian network to estimate Pc for ship collision.

In this study, Pc estimation is done using the synthetics approach with the use of a Bayesian network. The Bayesian model network is adapted from a fragment of the model network developed by DnV and Hänninen. According to historical data for four cases of anchors dragging on the subsea cable in the Madura Strait area of Indonesia, the anchors were dropped under emergency conditions when the ships lost control while passing through the critical subsea cable area. Various factors contributed to the ships losing control and dropping emergency anchors in the Madura Strait, including a bad weather condition, human error, poor pilot vigilance, steering failure, and poor visibility. Based on these factors, the network parameters of the model by DnV and Hänninen are modified for practical application to an anchor dragging on the subsea pipeline in the area of Indonesia. The network structure of the proposed modified model is shown in Fig. 5.

Fig. 5 Bayesian network structure of causation probability for

dropping emergency anchor.

(1)izV

(2)1iz

t

(1)iz

t , zL

1, zL

Zone Z

Zone Z+1

Critical area

Critical area

Subsea pipelineKP1 KP2 KP3

(2)1izV

yx

KP1

KP2

Zone Z

Critical area

Critical area(1)zL S

(2)zL S

Zone ZSubseapipeline

(1)izV

(2)izV

(2)iz

t

(1)iz

t

yx

Pilot support

Loss of control

Human performanceWeather

Visibility Assessment Vigilance

Action

Steering failure

Anchor dropping


4

( )( )

( ) ( )( ) iziz

i iLiz

Lz

tP

T T

V

(10)

where ( )tiz is the time required for a ship belonging to class i

to pass fairway in subsea pipeline zone z and ( )iT is the total

time required for the ship belonging to class i to pass the total length of fairway .

Fig. 3 Ships passing parallel to subsea pipeline.

Fig. 4 Ships crossing subsea pipeline. 3.4 Model of Dragged Anchor Candidates Na

Na is defined as the number of anchors with the potential to drag on the subsea pipeline if they are dropped from ships.

In this paper, the proposed model of Na is denoted by ( )izNa .

The equation for ( )izNa is developed by multiplying the number

of ships per unit time, ( )izQ , by the geometric probability of the

ships passing through the critical pipeline area, ( )Giz

P , as follows.

( ) ( ) ( )iz iz Giz

Na Q P

( )

( )( )

( ) ( ) ( )

iz

zizz

i iz D

Q Lf y

T V

(11)

where ( )Qiz is the number of ships per unit time that belong

to ship class i in fairway and pass through subsea pipeline zone z.

Thus, the total number of candidates for anchors dragging of

ship class i on the subsea pipeline in a specific subsea pipeline zone z is expressed as follows.

(1) (2)iz

Na Na Naiz iz (12)

where (1)Naiz is the number of dragged anchor candidates that

belong to ship class i in fairway 1 and subsea pipeline zone z, and(2)Naiz is the number of dragged anchor candidates that belong to

ship class i in fairway 2 and subsea pipeline zone z.

3.5 Model of Causation Probability Pc The causation probability Pc for an emergency dropped

anchor by ship is the probability of a ship dropping an anchor in an emergency when the ship passes through the critical subsea pipeline area. In the ship accident models, Pc is currently estimated in either of the following ways: 1) the scenario approach or 2) the synthetics approach16). The Pc that is modeled using scenario approach is calculated based on the statistics from the available historical accident data. Pedersen11) argues that the Pc based on scenario approach can be calculated from accident data at various locations and transformed to consider the analyzed location. The Pc that is modeled using synthetics approach is calculated based on a specific error situation analysis for ships. Det Norske Veritas (DnV)17) and Hänninen et al.18) applied the synthetics approach using a Bayesian network to estimate Pc for ship collision.

In this study, Pc estimation is done using the synthetics approach with the use of a Bayesian network. The Bayesian model network is adapted from a fragment of the model network developed by DnV and Hänninen. According to historical data for four cases of anchors dragging on the subsea cable in the Madura Strait area of Indonesia, the anchors were dropped under emergency conditions when the ships lost control while passing through the critical subsea cable area. Various factors contributed to the ships losing control and dropping emergency anchors in the Madura Strait, including a bad weather condition, human error, poor pilot vigilance, steering failure, and poor visibility. Based on these factors, the network parameters of the model by DnV and Hänninen are modified for practical application to an anchor dragging on the subsea pipeline in the area of Indonesia. The network structure of the proposed modified model is shown in Fig. 5.

Fig. 5 Bayesian network structure of causation probability for

dropping emergency anchor.

(1)izV

(2)1iz

t

(1)iz

t , zL

1, zL

Zone Z

Zone Z+1

Critical area

Critical area

Subsea pipelineKP1 KP2 KP3

(2)1izV

yx

KP1

KP2

Zone Z

Critical area

Critical area(1)zL S

(2)zL S

Zone ZSubseapipeline

(1)izV

(2)izV

(2)iz

t

(1)iz

t

yx

Pilot support

Loss of control

Human performanceWeather

Visibility Assessment Vigilance

Action

Steering failure

Anchor dropping

5

3.6 Use of AIS for Ship Traffic Analysis around Subsea Pipeline

AIS and GIS data are used to analyze the probability distribution of the ship traffic around the subsea pipeline.

A large quantity of data and some statistical tools are needed to analyze the data for the ship traffic distributions. It is more realistic to utilize the AIS data to analyze the traffic distribution rather than making an assumption based on an engineering judgment. The use of the AIS data in the model is presented in Fig. 619).

Fig. 6 Framework of ship traffic modeling using AIS data.

3.7 Anchor Holding Force and Ship Resistance Force The ship thrust after dropping an anchor is reduced by two

forces: 1) the anchor holding force and 2) ship resistance forces generated by the ship body’s interaction with the environmental load.

The anchor holding force is the sum of the anchor holding force and friction resistance of the length of chain in contact with the seabed. The anchor holding force is estimated by Eq. (13)20).

F F Fa chp

f w f L wa a c c (13)

where Fhp is the total holding force, Fa is the holding force

by the anchor, Fc is the holding force by the chain, fa is the

anchor holding factor, fc is the chain holding factor, wa is the

anchor weight, wc is the chain weight, and L is the length of the

chain. The total resistance force can be calculated by decomposing the static longitudinal force, as in Eq. (14)21).

R R R RT F w pv (14)

where RT is the total resistance, RF is the friction

resistance, Rw is the wave resistance, and Rpv is the viscous

pressure resistance.

4. Case Study for Applying Proposed Model

4.1 Applying the proposed model The proposed model is applied to a case study to estimation

the dragged anchor accident frequency on a subsea pipeline. The

successful applying the proposed model will require several tasks to be completed as shown in Fig. 7.

Fig. 7 Proposed method approach.

4.2 Location of Case Study

A subsea pipeline located in the Madura Strait, Indonesia, is chosen in this case study. The Madura Strait contains several ports, including the Tanjung Perak Port, Cement Port, Maspion Port, Gresik Port, Smelting Port, and Petrokimia Port. There are serious threats to the subsea pipeline in the Madura Strait because of the risk of a dragged anchor.

4.3 Pipeline Description

An export subsea pipeline transports gas from the Poleng Processing Platform (PPP) to the Onshore Processing Facilities (OPF) near the Tanjung Perak Port, Surabaya. The pipeline route is located in the fairway of the Madura Strait from kilometer point 35 to kilometer point 64 (KP35 to KP64), see Fig. 819). The subsea pipeline installation was completed in April 2008.

Fig. 8 Layout of subsea pipeline in Madura Strait.

AIS data

Ship variable (type of ship, MMSI, speed,

course, latitude, longitude)

Analyze ship interaction with another subject

Analyze traffic density in specific

pipeline zone

Model of ship

traffic analysis in

pipeline


6

4.4 Ship Categorization Using AIS Data The data for the ships in the Madura Strait are taken from the

AIS transmissions recorded from 2008 to 2011. The AIS receiver was installed at the Institute Technology of Sepuluh Nopember (ITS), Surabaya, Indonesia, in cooperation with Kobe University, Japan.

The existing traffic condition is not fully reflected by the AIS data because of the limitations in the requirements for AIS application on ships. Regulation 19 of SOLAS chapter V requires AIS to be fitted aboard 1) all ships of 300 ton gross and upward engaged in international voyages, 2) all ships of 500 ton gross and upward not engaged in international voyages, and 3) all passenger ships irrespective of size13). The AIS data from the ships are incomplete. Some of the missing ship data are collected from external ship databases. However, the AIS ship data are still reliable and helpful for a ship traffic analysis.

The number of ships per year in the Madura Strait area as obtained from the AIS data is 29830, 30934, 28918, and 31420 in 2008, 2009, 2010, and 2011, respectively. The ship types included in the Madura Strait traffic are presented graphically as percentages in Fig. 9.

Fig. 9 Ship population in Madura Strait.

Table 2 in Chapter 4.5 shows the estimated number of different ship types per year as obtained from the AIS data in the Madura Strait area. Bulk Carrier (Handysize), General Cargo (small and medium), and Container (small and feeder) have large numbers per day to pass through in the Madura strait.

4.5 Traffic Distribution Modeling Using AIS Data

The traffic modeling is approximated using a statistical distribution to analyze the probability of a ship passing through the critical subsea pipeline area. The lateral probability

distribution function for ships passing a fairway, ( ) ( )f yz , is

estimated using the AIS data for each specific zone. The lateral probability distribution function of ship traffic for arrival and departure from harbor in the Madura strait respectively are defined as southbound and northbound lateral traffic distribution. The modeling of the lateral probability distributions functions of ship traffic are approximated by continuous distributions based on the ship tracks in selected locations. The selection of the fit distribution models are validated by Kolmogorov-Smirnov test (see Table 1).

For the cases of crossing the subsea pipeline in zone 1 (KP35–KP36) and zone 3 (KP42–KP46), the lateral probability distribution function is not required, because all of the ships cross the subsea pipeline. This means the value of the probability of ships passing the pipeline is 1.

In zone 9 (KP64–ORF), there is no threat from a dragged anchor because the subsea pipeline was buried 19 m beneath the seabed using the horizontal directional drilling method (HDD).

Table 1 Lateral probability distribution function for ship traffic in

Madura Strait.

ZonePipeline location(KP–KP)

Fairway number

Lateral probability distribution

function

( ) ( )zf y

Parameters

1

35 to 36 (crossing pipeline), L=1 km

None required

None required

Probability of ship passing

pipeline is 1

2

36 to 42 (parallel

to pipeline), L=6 km

Fairways 2 (southbound

traffic) Lognormal

σ = 1.05654E-5

μ = 4.72485

3


None required

None required


pipeline is 1

4

46 to 50 (parallel pipeline), L=4 km

Fairway 2 (northbound

traffic) transition fairway

Log- Pearson 3

σ = 66.775=

-2.1326E-6= 4.72468

5

50 to 54 (parallel



traffic) Lognormal

σ = 1.87659E-5

μ = 4.72442

6

54 to 58 (parallel



traffic) Lognormal

σ = 2.1564E-5

μ = 4.724386

7

58 to 62 (parallel



traffic) expanded fairway

Weibull

α = 41207.8 β =

112.667

8

62 to 64 (parallel


None required

No dragged anchor hazard


pipeline is 0

9

64 to ORF

(crossing pipeline)

None required (HDD

section)



pipeline is 0

Bulk Carrier11.3%

General Cargo39.3%Container

21.4%

Passenger6.62%

Tanker16%

Suppy2.14%

Navi/Patrol0.427%

Other2.78%

6














Madura Strait.


Fairway number


function

( ) ( )zf y

Parameters

1


None required

None required


pipeline is 1

2

36 to 42 (parallel



traffic) Lognormal

σ = 1.05654E-5

μ = 4.72485

3


None required

None required


pipeline is 1

4




Log- Pearson 3

σ = 66.775=

-2.1326E-6= 4.72468

5

50 to 54 (parallel



traffic) Lognormal

σ = 1.87659E-5

μ = 4.72442

6

54 to 58 (parallel



traffic) Lognormal

σ = 2.1564E-5

μ = 4.724386

7

58 to 62 (parallel




Weibull

α = 41207.8 β =

112.667

8

62 to 64 (parallel


None required



pipeline is 0

9

64 to ORF

(crossing pipeline)

None required (HDD

section)



pipeline is 0

Bulk Carrier11.3%


21.4%

Passenger6.62%

Tanker16%

Suppy2.14%

Navi/Patrol0.427%

Other2.78%


6














Madura Strait.


Fairway number


function

( ) ( )zf y

Parameters

1


None required

None required


pipeline is 1

2

36 to 42 (parallel



traffic) Lognormal

σ = 1.05654E-5

μ = 4.72485

3


None required

None required


pipeline is 1

4




Log- Pearson 3

σ = 66.775=

-2.1326E-6= 4.72468

5

50 to 54 (parallel



traffic) Lognormal

σ = 1.87659E-5

μ = 4.72442

6

54 to 58 (parallel



traffic) Lognormal

σ = 2.1564E-5

μ = 4.724386

7

58 to 62 (parallel




Weibull

α = 41207.8 β =

112.667

8

62 to 64 (parallel


None required



pipeline is 0

9

64 to ORF

(crossing pipeline)

None required (HDD

section)



pipeline is 0

Bulk Carrier11.3%


21.4%

Passenger6.62%

Tanker16%

Suppy2.14%

Navi/Patrol0.427%

Other2.78%

7

Table 2 Estimated number of ships per year based on ship types in Madura Strait area

Ship category Number of ship per year

Bulk Carrier (Handysize) 1991 Bulk Carrier (Handymax) 1526 General Cargo (Small) 5906 General Cargo (Medium) 6305 Container (Small Feeder) 4446 Container (Feeder) 2190 Passenger (Small) 730 Passenger (Medium) 1327 Tanker (Small) 1526 Tanker (General Purpose) 1659 Tanker (Medium Range) 1792 Supply (Medium) 664 Navy/Patrol 133 Other (Medium) 863

4.6 Result and Discussion 4.6.1 Width of critical subsea pipeline area in Madura Strait

The width of the critical subsea pipeline area for ship class i, ( )izD , is estimated using Eq. (9). The angle between the ship’s

course and the subsea pipeline axis, , is assumed to be about 15°. The data acquired for the calculation consist of the ship size and type, anchor size and weight, initial ship speed (5 knots), soil type (soft clay), average current velocity (0.45 m/s), wave condition (calm), and the values of fa (3) and fc (0.6)19),22).

The width results for the critical pipeline area based on the ship type are presented in Table 3.

Table 3 Width of critical pipeline area.

Ship type and class LWL Beam

Width of critical pipeline

area

(m) (m) (m)

Bulk Carrier (Handysize) 163 27 81.25

Bulk Carrier (Handymax) 182 30 86.09 General Cargo (Small) 96 17 46.38 General Cargo (Medium) 142 22 74.91 Container (Small Feeder) 144 23 74.01 Container (Feeder) 192 30 92.41 Passenger (Small) 62 16 29.96 Passenger (Medium) 111 20 55.61 Tanker (Small) 93 16 45.20 Tanker (General Purpose) 154 25 79.06 Tanker (Medium Range) 175 27 82.40 Supply (Medium) 63 11 29.82 Navy/Patrol 88 13 37.83 Other (medium) 86 14 36.78

4.6.2 Number of dragged anchor candidates in Madura Strait The total number of candidates for an anchor dragging of ship

class i on the subsea pipeline in specific subsea pipeline zone z,Naiz , is estimated using Eq. (12). The parameters considered to

estimate Naiz include the number of ships passing the fairway,

types and sizes of these ships, fairway dimensions, ship speeds in the fairway, and ship traffic distributions in the subsea pipeline zone.

The zone identification, the pipeline location, the length of critical pipeline zone, and the lateral probability distribution function show in Table 1. The numbers of ships per year that belonging to ship class i show in Table 2. The widths of the

critical subsea pipeline area for ship class i, ( )izD , show in Table

3. The ( )izD is required to calculate the probability of a ship

passing through the critical subsea area. The average initial ship speed in the fairway is 5 knots and the total duration time for a ship passing the Madura Strait, T, is 3.2 h.

As an example, estimated probability of ship passing through

the critical subsea pipeline area by using ( ) ( )zf y and ( )izD for

location of KP50–KP54, KP54–KP58, and KP58–KP62 are presented in Table 4

Table 4 Estimated probability of ship passing through the critical

pipeline area as an example for location of KP50–KP54, KP54–KP58, and KP58–KP62

Ship category

Probability of ship passing through the critical pipeline areaKP50-KP54

KP54-KP58

KP58-KP62

Bulk Carrier (Handysize) 0.077 0.083 0.031 Bulk Carrier (Handymax) 0.079 0.084 0.033 General Cargo (Small) 0.055 0.061 0.019 General Cargo (Medium) 0.072 0.078 0.029 Container (Small Feeder) 0.081 0.087 0.028 Container (Feeder) 0.095 0.103 0.035 Passenger (Small) 0.050 0.057 0.017 Passenger (Medium) 0.062 0.068 0.023 Tanker (Small) 0.057 0.064 0.020 Tanker (General Purpose) 0.075 0.080 0.030 Tanker (Medium Range) 0.077 0.082 0.032 Supply (Medium) 0.050 0.057 0.017 Navy/Patrol 0.053 0.060 0.018 Other (Medium) 0.052 0.059 0.018

The numbers of dragged anchor candidates for ship class i,

Naiz , in locations KP50–KP54 (pipeline parallel to fairway),

KP54–KP58 (pipeline parallel to fairway near Maspion Port), and KP58–KP62 (pipeline parallel to fairway in expanded fairway) are presented in Fig. 10.


8

Fig. 10 Numbers of dragged anchor candidates as an example,

Naiz , in KP50–KP54, KP54–KP58, and KP58–KP62.

Fig.11 Map of KP50–KP54, KP54–KP58, and KP58–KP62 in Madura Strait

4.6.3 Causation probability in Madura Strait

In this study, Pc for a dragged anchor accident on the subsea pipeline in the Madura Strait is estimated using a Bayesian network, as shown in Fig. 5. Based on the Bayesian network parameters for the Madura Strait shown in Fig 5, some of the probabilities are estimated by an analysis of data from several sources, including the Tanjung Perak Meteorology, Climatology, and Geophysics Institute, Tanjung Perak Port Authority, the National Transportation Safety Committee of Indonesia, and several external databases.

The weather condition probabilities are calculated from data acquired from the Tanjung Perak Meteorology, Climatology, and Geophysics Institute collected during 2003-2012. The probability of bad weather/heavy rain is estimated to be 0.07 and the probability of good weather condition is 0.93.

The probability of visibility is a conditional probability that is estimated based on the probability of the weather conditions. Good weather is equivalent to good visibility (>1 nm). The conditional probability of visibility is 1 in good weather conditions. Bad weather/heavy rain conditions are equivalent to poor visibility (<1 nm). The conditional probability of the visibility is reduced by 25% in bad weather/heavy rain conditions17). Thus in the heavy rainy conditions, the logical conditional probability of the poor visibility is 0.25 and the conditional probability of good visibility is 0.7517).

The probability of a port pilot guiding a ship passing through the Madura Strait is determined based on a pilotage analysis in the Madura Strait by the National Transportation Safety Committee of Indonesia that was reported in 201223). The probability of a port pilot not being on board to guide the ship is estimated to be 0.4 and the probability of a port pilot on board is 0.6223).

The probability of pilot vigilance is a conditional probability based on the availability of a port pilot on board. An on-board port pilot exercising internal vigilance could help to warn the officer on watch (OOW) of dangerous situations17). The conditional probability of the vigilance is 1 if an on-board port pilot is available. The conditional probability of the vigilance is reduced by 25% if an on-board port pilot is not available. Thus in the condition of a port pilot not being on board, the logical conditional probability of the poor vigilance is 0.25 and the conditional probability of the good vigilance to be 0.75 17).

The probability of human performance is adapted from the results of a study on the human factor for ship accidents in the Madura Strait that was conducted in 201024). The probability of the poor human performance in the Madura Strait is 1.71E-4 and the probability of the good human performance is 0.99983.

To estimate the conditional probability of a ship dropping its emergency anchor after losing control, it is necessary to perform an analysis of all of the ship accidents caused by a loss of control. This ship accident analysis is carried out based on the history of ship accidents during 2005–2011 in the Madura Strait. There were 30 cases (83.3%) of ship collisions, 5 cases (13.9%) of ship groundings, and 1 case (2.8%) of an anchor dragging on the subsea cable. Referring to this data, the conditional probability of dropping an anchor after losing control is 2.8%. The probability of steering failure for ships in the Madura Strait is estimated from three steering failure cases during 2005–2010. The total ship calls in the Madura Strait during 2005–2010 were about 174.927 ships based on the AIS data and external data. From this statistical data, the probability of the steering system not function is estimated to be 3.43E-5 and the probability of steering system in function is 0.999965725).

Finally, the result of the causation probability Pc for a ship dropping its emergency anchor after losing control in the Madura Strait is estimated to be 3.6E-5 using Bayesian network model.

4.6.4 Dragged anchor accident frequency on subsea pipeline

in Madura Strait The dragged anchor accident frequency on the subsea pipeline

in the Madura Strait, Nd, is estimated using Eq. (2). The number of dragged anchor candidates, Naiz , for ship class i in subsea

pipeline zone z is estimated using Eq. (12). The value of causation probability Pc using Bayesian network model is 3.6E-5. Thus, the results of example for Nd in 2011 based on the ship categories are presented in Fig. 12 for merchant ship categories and Fig. 13 for service and navy ship categories.

In the risk analysis context, estimation of the accident frequency is one of the vital elements. The important step in risk analysis is determining a risk ranking of each threat that has been identified previously.

I II III IV V VI VII VIII IX X XI XII XIII XIVShip types

0

10

20

30

40

Num

ber o

f dr

agge

d an

chor

can

dida

tes,

Na iz,

per y

ear

I : Bulk Carrier (Handysize)II : Bulk Carrier (Handymax)III : General Cargo (Small)IV : General Cargo (Medium)V : Container (Small Feeder)VI : Container (Feeder)VII : Passenger (Small)VIII: Passenger (Medium)IX : Tanker (Small)X : Tanker (General Purpose)XI : Tanker (Medium Range)XII: Supply Vessel (Medium)XIII: Navy/PatrolXIV :Other (Medium)

KP54-58 KP58-62KP50-54

8



















0

10

20

30

40

Num

ber o

f dr

agge

d an

chor

can

dida

tes,

Na iz,

per y

ear


KP54-58 KP58-62KP50-54


8



















0

10

20

30

40

Num

ber o

f dr

agge

d an

chor

can

dida

tes,

Na iz,

per y

ear


KP54-58 KP58-62KP50-54

9

Fig. 12 Frequency of dragging anchor (Nd) on subsea pipeline in Madura Strait (merchant ship types)

Fig. 13 Frequency of dragging anchor (Nd) on subsea pipeline in Madura Strait (service and navy ship type)

The risk ranking is determined by combining frequency

ranking and consequence ranking and then comparing the result against acceptance criteria. For subsea pipeline, the risk ranking can use the risk matrix from the recommended practice of DNV RP-F10726) that consists of: 1) Not acceptable, 2) As Low As Reasonably Practicable (ALARP), and 3) Acceptable. In the risk matrix, the frequency ranking consists of five frequency rankings:

1) So low frequency ranking or level 1 (≤10-5). 2) Low frequency ranking or level 2 (10-4≥x>10-5) . 3) Medium frequency ranking or level 3 (10-3≥x>10-4). 4) Rather high frequency ranking or level 4 (10-2 ≥x>10-3). 5) High frequency ranking or level 4 (>10-2). The result of dragged anchor accident frequency for merchant

ship categories show that the General Cargo (small and medium), and Container (small feeder) have values of more than 10-3 per year. These values are within the rather high frequency ranking of level 4 (10-2≥x>10-3), see Fig. 12.

The Bulk Carrier (Handysize and Handymax), Container (Feeder), Passenger (Small and Medium), and Tanker (Small, General Purpose, and Medium Range) have values in the frequency range 10-3≥x>10-4 per year. These values are within the medium frequency ranking of level 3, see Fig. 12.

For the service and navy ship categories, the dragged anchor accident frequency results show that the Supply Vessel (Medium) and the Other Ship (Medium) have values in the low frequency

ranking of level 2 (10-4≥x>10-5) per year. The Navy/Patrol Ship has values in the so low frequency ranking of level 1 (≤10-5) per year, see Fig. 13.

4.6.5 Validation

According to the dragged anchor accident records, there were four cases of anchors dragging on a subsea cable during 1987–2011. The subsea cable was constructed in the Madura Strait in 1987. For safety of the cable, navigation buoys have been installed and navigation maps indicate that all ships are prohibited from dropping anchor in the cable area. The location of the subsea cable is close to KP64 of the subsea pipeline. The subsea gas pipeline is a new structure in the Madura Strait. This pipeline was constructed in the Madura Strait in 2008.

The values of Nd for the subsea cable are estimated using the proposed model. The Pc value is 3.6E-5 based on the previous analysis. The results of a comparison of Nd and Nactual during the specified period of years are presented in Table 5.

35-36 36-42 42-46 46-50 50-54 54-58 58-62 62-64Pipeline Location (KP-KP)

5.00x10-4

1.00x10-3

1.50x10-3

2.00x10-3

2.50x10-3

Dra

gged

anc

hor f

requ

ency

(Nd)

on th

e su

bsea

pip

elin

e pe

r yea

r

Bulk Carrier (Handysize)Bulk Carrier (Handymax)General Cargo (Small)General Cargo (Medium)Container (Small Feeder)Container (Feeder)Passenger (Small)Passenger (Medium)Tanker (Small)Tanker (General Purpose)Tanker (Medium Range)

35-36 36-42 42-46 46-50 50-54 54-58 58-62 62-64Pipeline Location (KP-KP)

0.00x100

2.00x10-4

4.00x10-4

1.00x10-4

3.00x10-4

Dra

gged

anc

hor f

requ

ency

(Nd)

on

the

subs

ea p

ipel

ine

per y

ear

Supply Vessel (Medium)Navi/Patrol ShipOther Ship (Medium)


10

Table 5 Results of comparison between Nd based on proposed model with Nactual based on accident records.

Years Nd

(proposed model)

Nactual (actual

accidents)

Ratio

�proposed modelactualaccidents�1987–1994 0.51 1 0.51 1995–2003 0.66 2 0.33 2004–2011 0.58 1 0.58

The analyses of ratio between estimated values to actual values are approached based on agreements criteria used by Fowler and Sorgård27). The criteria consist of three categories: 1) good agreement (ratio≥0.5), 2) reasonable agreement (0.5>ratio≥0.2), 3) poor agreement (ratio <0.2).

During 1987-1994 and 2004–2011, the ratio results shows good agreement (ratio value � 0.5)27). During 1995–2003, the ratio result shows reasonable agreement (0.5>ratio≥0.2). However, average of the ratio analysis during 1987-2011 is 0.47. It can be concluded that the ratio of the validation analysis show in reasonable agreement.

5. Conclusions

This paper presented a new model to estimate the dragged

anchor accident frequency on a subsea pipeline in a busy port area. The conclusions made from this study are as follows:

1. In the accidents that involved dragging an anchor on a subsea pipeline, ships that still had their initial velocity may have dropped the anchor under an emergency condition to stop their movement. In this situation, the ship and anchor will stop at a specific anchor stopping distance.

2. In this paper, a model was proposed to estimate the frequency of dragging an anchor on a subsea pipeline based on the concept introduced by Fujii and an anchor stopping distance analysis in subsea pipeline areas. The AIS data were used to analyze the ship traffic distribution.

3. In the proposed model, the number of candidates for dragging an anchor on a subsea pipeline, Na, was estimated based on specific characteristics such as the pipeline layouts, ship traffic conditions, environment conditions, ship dimensions, anchor types, and anchor dimensions. The causation probability Pc was estimated using a Bayesian network method that was modified from the model of DnV and Hanninen. Some of the factors used to estimate Pc with the Bayesian network were evaluated based on an evaluation of the dragged anchor accidents in the Madura Strait area. These factors were the human performance factor, navigational factor, weather factor, and support factor. To use the proposed model in different locations, the same factors in the Bayesian network could be modified according to the circumstances at that location.

4. A case study to demonstrate the implementation of the proposed model was performed for the subsea pipeline in the Madura Strait, Indonesia. The results for the yearly frequency of dragging an anchor on the subsea pipeline in the Madura Strait for merchant ship categories show that the General Cargo (small and medium), and Container (small feeder) were found to fall into the rather high frequency ranking of level 4 (10-2≥x>10-3) and the Bulk Carrier (Handysize and Handymax), Container (Feeder), Passenger (Small and Medium), and Tanker (Small, General

Purpose, and Medium Range) were found to fall into the medium frequency ranking of level 3 based on DNV-RP-F107. For the service and navy ship categories, the results of the dragged anchor accident frequency per year for Supply Vessel (Medium) and the Other Ship (Medium) were determined to fall into the low frequency ranking of level 2 (10-4≥x>10-5) and the Navy/Patrol Ship was determined to fall into the so low frequency ranking of level 1 (≤10-5). A mitigation plan could be developed in the future based on the results of the analysis using the proposed model. This mitigation could be performed by reducing the value of Na or Pc where the frequency is at a medium or high level.

5. A validation analysis of the proposed model was performed by comparing the dragged anchor frequency estimated using the proposed model with the actual accident record. The average ratio of the comparison was 0.47 which showed the reasonable agreement (0.5>ratio≥0.2). This method can be applied to predict the dragged anchor frequency in the future if the situation may differ from today by adjusting the data.

6. For the safety of a subsea pipeline, it is important to install navigation buoys as markers and indicate on navigation maps that all ships are prohibited from dropping anchor in the pipeline area. This proposed model could be utilized to determine an appropriate location for putting the navigation buoys and to develop the anchoring navigation map.

Acknowledgment

The authors wish to thank Professor Masao Furusho from the

Graduate School of Maritime Sciences, Kobe University, for his evaluation of the human error probability in the Bayesian network model.

References

1) Nicolas, F.: Asean Energy Cooperation: An Increasing

Daunting Challenge, IFRI Report, 2009, www.ifri.org/files/Energie/FNicolas.pdf.

2) APERC: Natural Gas Pipeline Development in Southeast Asia, Report of Asia Pacific Energy Research Centre, Japan, March 2000, www.ieej.or.jp/aperc/final/se.pdf.

3) Benjamin, K.S.: Energy Policy and Cooperation in Southeast Asia: The History, Challenges, and Implication of the Trans-ASEAN Gas Pipeline (TAGP) Network, Journal of Energy Policy, Vol.37, pp.2356-2367, 2009.

4) Stefani, V.D., Carr, P.: A Model to Estimate the Failure Rates of Offshore Pipeline, Proceedings of the 8th International Pipeline Conference, Canada, IPC2010-31230, 2010.

5) Health and Safety Executive (HSE) Publication: Guidelines for Pipeline Operator on Pipeline Anchor Hazard, HID SI 3 Gas and Pipelines Unit, Aberdeen, 2009, www.hse.gov.uk/pipelines/pipelines-anchor-hazard.pdf.

6) ERM (Environmental Resources Management) in South Soko: Quantitative Risk Assessment (QRA) for the Subsea Pipeline from South Soko to Black Point Power Station, Part 2 Section13 Annex B, 2006, www.epd.gov.hk/eia/register/report/eiareport/eia_1252006/html/eiareport/Part2/Section13/Sec2_13_AnnexB.htm.

7) Braestrup, M.: Design and Installation of Marine Pipelines, Blackwell Science Ltd, Oxford, UK, pp 66-67, 2005.


10

Table 5 Results of comparison between Nd based on proposed model with Nactual based on accident records.

Years Nd

(proposed model)

Nactual (actual

accidents)

Ratio

�proposed modelactualaccidents�1987–1994 0.51 1 0.51 1995–2003 0.66 2 0.33 2004–2011 0.58 1 0.58

The analyses of ratio between estimated values to actual values are approached based on agreements criteria used by Fowler and Sorgård27). The criteria consist of three categories: 1) good agreement (ratio≥0.5), 2) reasonable agreement (0.5>ratio≥0.2), 3) poor agreement (ratio <0.2).

During 1987-1994 and 2004–2011, the ratio results shows good agreement (ratio value � 0.5)27). During 1995–2003, the ratio result shows reasonable agreement (0.5>ratio≥0.2). However, average of the ratio analysis during 1987-2011 is 0.47. It can be concluded that the ratio of the validation analysis show in reasonable agreement.

5. Conclusions

This paper presented a new model to estimate the dragged

anchor accident frequency on a subsea pipeline in a busy port area. The conclusions made from this study are as follows:

1. In the accidents that involved dragging an anchor on a subsea pipeline, ships that still had their initial velocity may have dropped the anchor under an emergency condition to stop their movement. In this situation, the ship and anchor will stop at a specific anchor stopping distance.

2. In this paper, a model was proposed to estimate the frequency of dragging an anchor on a subsea pipeline based on the concept introduced by Fujii and an anchor stopping distance analysis in subsea pipeline areas. The AIS data were used to analyze the ship traffic distribution.

3. In the proposed model, the number of candidates for dragging an anchor on a subsea pipeline, Na, was estimated based on specific characteristics such as the pipeline layouts, ship traffic conditions, environment conditions, ship dimensions, anchor types, and anchor dimensions. The causation probability Pc was estimated using a Bayesian network method that was modified from the model of DnV and Hanninen. Some of the factors used to estimate Pc with the Bayesian network were evaluated based on an evaluation of the dragged anchor accidents in the Madura Strait area. These factors were the human performance factor, navigational factor, weather factor, and support factor. To use the proposed model in different locations, the same factors in the Bayesian network could be modified according to the circumstances at that location.

4. A case study to demonstrate the implementation of the proposed model was performed for the subsea pipeline in the Madura Strait, Indonesia. The results for the yearly frequency of dragging an anchor on the subsea pipeline in the Madura Strait for merchant ship categories show that the General Cargo (small and medium), and Container (small feeder) were found to fall into the rather high frequency ranking of level 4 (10-2≥x>10-3) and the Bulk Carrier (Handysize and Handymax), Container (Feeder), Passenger (Small and Medium), and Tanker (Small, General

Purpose, and Medium Range) were found to fall into the medium frequency ranking of level 3 based on DNV-RP-F107. For the service and navy ship categories, the results of the dragged anchor accident frequency per year for Supply Vessel (Medium) and the Other Ship (Medium) were determined to fall into the low frequency ranking of level 2 (10-4≥x>10-5) and the Navy/Patrol Ship was determined to fall into the so low frequency ranking of level 1 (≤10-5). A mitigation plan could be developed in the future based on the results of the analysis using the proposed model. This mitigation could be performed by reducing the value of Na or Pc where the frequency is at a medium or high level.

5. A validation analysis of the proposed model was performed by comparing the dragged anchor frequency estimated using the proposed model with the actual accident record. The average ratio of the comparison was 0.47 which showed the reasonable agreement (0.5>ratio≥0.2). This method can be applied to predict the dragged anchor frequency in the future if the situation may differ from today by adjusting the data.

6. For the safety of a subsea pipeline, it is important to install navigation buoys as markers and indicate on navigation maps that all ships are prohibited from dropping anchor in the pipeline area. This proposed model could be utilized to determine an appropriate location for putting the navigation buoys and to develop the anchoring navigation map.

Acknowledgment

The authors wish to thank Professor Masao Furusho from the

Graduate School of Maritime Sciences, Kobe University, for his evaluation of the human error probability in the Bayesian network model.

References

1) Nicolas, F.: Asean Energy Cooperation: An Increasing

Daunting Challenge, IFRI Report, 2009, www.ifri.org/files/Energie/FNicolas.pdf.

2) APERC: Natural Gas Pipeline Development in Southeast Asia, Report of Asia Pacific Energy Research Centre, Japan, March 2000, www.ieej.or.jp/aperc/final/se.pdf.

3) Benjamin, K.S.: Energy Policy and Cooperation in Southeast Asia: The History, Challenges, and Implication of the Trans-ASEAN Gas Pipeline (TAGP) Network, Journal of Energy Policy, Vol.37, pp.2356-2367, 2009.

4) Stefani, V.D., Carr, P.: A Model to Estimate the Failure Rates of Offshore Pipeline, Proceedings of the 8th International Pipeline Conference, Canada, IPC2010-31230, 2010.

5) Health and Safety Executive (HSE) Publication: Guidelines for Pipeline Operator on Pipeline Anchor Hazard, HID SI 3 Gas and Pipelines Unit, Aberdeen, 2009, www.hse.gov.uk/pipelines/pipelines-anchor-hazard.pdf.

6) ERM (Environmental Resources Management) in South Soko: Quantitative Risk Assessment (QRA) for the Subsea Pipeline from South Soko to Black Point Power Station, Part 2 Section13 Annex B, 2006, www.epd.gov.hk/eia/register/report/eiareport/eia_1252006/html/eiareport/Part2/Section13/Sec2_13_AnnexB.htm.

7) Braestrup, M.: Design and Installation of Marine Pipelines, Blackwell Science Ltd, Oxford, UK, pp 66-67, 2005.

11

8) Vitali, L., Candiracci, F., Crea, C., Bruschi, R., Rott, W.: Nord Stream Project – Pipeline Safety Against Ship Traffic Related Threats: Quantitative Risk Assessment Approach, Proceeding of the Twenty-second (2012) International Offshore and Polar Engineering Conference (ISOPE), Rhodes, Greece, 2012.

9) Environmental Resources Management: Quantitative Risk Assessment: Pipeline, EIA Report, Part 2, Section 13, Annex 13B, Hong Kong, pp 19-23, 2006.

10) Zhang, F.Y, Bai, Y., Badaruddin, M.F., Tuty, S.: Quantitative Risk Assessment and Leak Detection Criteria for a Subsea Oil Export Pipeline, Journal of Marine Science and Application, Vol.8, pp.168-174, 2009.

11) Pedersen, P.T.: Review and Application of Ship Collision and Grounding Analysis Procedures, Journal of Marine Structures, Vol.23, pp.241-262, 2010.

12) Talavera A, Aguasca R, Galvan B, Cacereno A.: Application of Dempster–Shafer theory for the quantification and propagation of the uncertainty caused by the use of AIS data. Reliability Engineering and System Safety, Vol.111, pp.95–105, 2013.

13) Montewka, J., Hinz, T., Kujala, P., Matusiak, J.: Probability Modeling of Vessel Collisions, Journal of Reliability Engineering and System Safety, Vol.95, pp.573-589, 2010.

14) Fujii, Y., Yamanouchi, H., Mizuki, N.: Some Factors Affecting the Frequency of Accidents in Marine Traffic. II – The Probability of Strading and III – The Effect of Darkness on the Probability of Collision and Stranding, Journal of Navigation, Vol.27, pp.239-249, 1974.

15) Goerlandt, F., Kujala, P.: Traffic Simulation Based Ship Collision Probability Modeling, Journal of Reliability Engineering and System Safety, Vol.96, pp.91-107, 2011.

16) Kujala, P., Hannienen, M., Arola, T., Ylitalo, J.: Analysis of the Marine Traffic Safety in the Gulf of Finland, Journal of Reliability Engineering and System Safety, Vol.94, pp.1349-1357, 2009.

17) Det Norske Veritas (DNV): Formal Safety Assessment – Large Passenger Ships, ANNEX II, Report No 2003-0277, 2003. http://research.dnv.com/skj/FSALPS/FSA-LPS-

NAV.htm . Accessed 20 March 2011 18) Hänninen, M., Kujala, P.: The Effects of Causation

Probability on the Ship Collision Statistic in the Gulf of Finland, Proceedings of the 8th International Navigation Symposium, Gdansk, Poland, June 2009.

19) Mulyadi, Y., Kobayashi, E., Wakabayashi, N., Zaman, M.B., Pitana, T., Wahyudi: Usage of AIS to Estimate Dragged Anchor Probability on Subsea Pipeline, Proceedings of Japan Society of Naval Architects and Ocean Engineers, Vol.12, May 2011.

20) Pitana, T., Kobayashi, E.: Dynamic System Simulation of Ship Evacuation due to Tsunami Attack Considering Traffic Density Condition, Proceedings of International Offshore and Polar Engineering Conference, Canada, pp.551-557, 2008.

21) Schneekluth, H., Bertram, V.: Ship Design for Efficiency and Economy, Butterworth-Heinemann Publishing Ltd., Second edition, 1998.

22) Sincock, P., and Sondhi, N.: Drag Anchors for Floating Systems, Repot for HSE-Offshore Technology, OTH 93 395, Health and Safety Executive (HSE) Publication, UK, 1993.

23) The National Transportation Safety Committee of Indonesia:

Investigation of ship accident in Tanjung Perak Port area: Ship accidents of KM. Alken Pesat and KM. Alpine, Report No. KNKT-12-12-04-03, Surabaya, 2012. (in Indonesia)

24) Harahap, R.G.: Study of Human Reliability for Ship Accident: Case study in Tanjung Perak Port Surabaya, Thesis, Institut Teknologi Sepuluh Nopember, Surabaya, 2010. (in Indonesia)

25) Mulyadi Y, Kobayashi E, Wakabayashi N, Pitana T, Wahyudi: Development of Ship Sinking Model over Subsea Pipeline for Madura Strait using AIS Data, WMU Journal of Maritime Affairs, 2013, DOI 10.1007/s13437-013-0049-2.

26) Recommended Practice DNV-RP-F107: Risk Assessment of Pipeline Protection, Det Norske Veritas, October 2010.

27) Fowler, T.G., Sorgard, E.: Modeling Ship Transportation Risk, Journal of Risk Analysis, Vol.20, No. 2, pp.225-244, 2000.

yeyes jasnaoe.pdf

Documents

flipper crawler rov

stability of crawler

subsea pipelines

motion of rov

busy port areas

driving system

anchor candidates

mtsieee oceans