Proceedings of the Twentieth (2010) International Offshore and Polar Engineering Conference Beijing, China, June 2025, 2010

Copyright 2010 by The International Society of Offshore and Polar Engineers (ISOPE) ISBN 978-1-880653-77-7 (Set); ISSN 1098-6189 (Set); www.isope.org

Collision Risk Analyses for Offshore Wind Energy Installations

Daniel Povel, Volker Bertram
FutureShip GmbH A GL Company Hamburg, Germany

Mathias Steck
GL Gerrad Hassan Bangalore, India

ABSTRACT
Over the period of one decade, Germanischer Lloyd (GL) has developed and applied risk analysis procedures to assess the collision risks for offshore wind energy installations (WEI). The revised GL guideline focuses on the identification and assessment of collision risks between ships and wind energy installations. Other risks (e.g. collision risks with air traffic) are discussed, but not in the same depth as ship-WEI collisions. The basic approach is a pragmatic mix of formal procedure and empirical data required for model applications. Basic assumptions were harmonized in 2004 between major providers of such risk analyses for offshore WEIs in Europe. The basic assumptions and their underlying reasoning are described. The guideline is an important contribution to ensure comparability, repeatability and transparency of such risk analysis. The guideline addresses the main factors influencing collision probability and damage extent, including risk control measures. Concepts are illustrated by applications taken from existing or planned North Sea installations. While illustrative examples are all taken from the North Sea, the procedure has been applied more widely.

The risk analysis should be made in the early project stages to be most beneficial and cost efficient. Safety starts with the conceptual design of a wind farm. Later corrections of shortcomings due to neglected systematic investigations are generally costly. The GL guideline for OWF risk analyses has the following objectives:

Common structure and similar scope, level of detail, approach and quality make submitted risk analyses comparable and reproducible. This facilitates the review for the authorities and accelerates the approval process. The OWF operator understands better what is required and analysed within a risk analysis. This promotes an early focus on safety aspects. General public and lobbies can read, understand and interpret risk analysis reports better.

BASE DATA AND ASSUMPTIONS Harmonized Assumptions

Following a request of German authorities, experts of Germanischer Lloyd harmonized basic assumptions for OWF risk analyses in cooperation with other major providers of risk analyses in 2004. In particular, there were common basic assumptions for the scenarios involving powered and drifting ships, intended to ensure comparable results by different expert consultants. The harmonised assumptions are integrated in the English GL guideline (GL, 2010).

KEY WORDS: wind energy; offshore; risk assessment; collision. INTRODUCTION

Modern safety engineering offers assorted methods to assess complex systems and their risk for man and environment. The main task is to apply these proven methods appropriately within risk analyses for offshore wind farms (OWF). Germanischer Lloyd (GL) has developed a guideline for this purpose (GL, 2010). The assessment of accidents ends with the quantification of spilled pollutants, harm to property or persons. Subsequent damage due to spilled substances can be investigated in environmental risk or impact analyses and is not subject of the GL guideline. The risk analysis is a documented approach to ensure the retraceability of lines of arguments and calculations. An OWF risk analysis should assert verifiably

Traffic and Environment

The shipping traffic in the considered area should be described in terms of line traffic and spread-out traffic. Line traffic is limited to shipping channels and traffic separation zones; the rest is denoted as spread-out traffic. Ship routes are to be displayed in suitable charts along with the OWF perimeter. A quantification of line and spread-out traffic is indispensable for the risk analysis. Local maritime authorities and/or service providers should be contacted to obtain data concerning shipping traffic. Generally data from several sources should be compared. Data sources must be documented. For the analysis, the following points should be considered:

whether a planned OWF fulfils the safety requirements, whether under some failure conditions threats to persons or environment may arise and what the associated consequences would be, how they are to be rated and how they could be managed.

For all shipping routes in the vicinity of the OWF, the traffic direction has to be considered.

745

Waypoints of shipping routes are described by their geographic coordinates. The section between two waypoints is a route segment. For each shipping route, the following items should be considered:
- Width of shipping lane (if available), - Lateral distributions of shipping traffic on individual route

planned, the effectiveness of the measures is to be discussed without detailed consideration of investment and operational costs. Risk control measures can be implemented on the ships, the WEIs, and the analyzed sea area. The possibilities to implement additional measures on the wind farm are rather limited, as identified measures are already required by many authorities. For example, to reduce the consequences of a collision, the German authorities require the construction of collision friendly foundations. The foundation shall not disrupt the shell of a single-hull tanker, drifting with 2 m/s against the construction. These requirements are already considered in the risk model. Implementation on ships only considered effective if enforced by IMO (International Maritime ). In the risk analysis, the current state-ofpractice equipment of vessels is considered. This includes the probability of self repair dependent on the elapsed time of drift and the probability of a successful emergency anchoring maneuver depending on the prevailing wind speed. The effectiveness of these measures is considered based on the harmonised assumptions. The following measures acting on the sea area have been identified: installation of Automatic Identification System (AIS) transponders for the WEI, different options of traffic control or traffic monitoring and the assistance of salvage tugs for vessels not under command. The efficiency for collision risk reduction is discussed in the remaining section. Depending on the sea area, additional risk control measures may be applied and considered in the risk assessment.

segments
- Vessel handling (e.g. rule of keeping to the right on one-

way traffic lanes in separation zones)

- Percentage of vessels outside the shipping route (if avail-

able)
- Shipping occurrence per year sorted by ship type and by

ship size
- If relevant, anchorages and roadsteads in the vicinity of the

OWF (number of anchoring manoeuvres per year)

- Types and amounts of transported hazardous goods sorted

by packed (e.g. container) and bulk goods (if available)

- If relevant, occurrence and routes of fishing vessels, sport

boats, tourist boats and navy vessels should be considered. Document if no reliable data for these ships can be obtained. In this case, a quantification of the collision risk for these vessel groups is not possible. Therefore a qualitative approach must be chosen for these vessels.
- Traffic rules for the considered area (incl. rules for trans-

Automatic Identification System (AIS)

AIS allows early recognition of collision courses with other ships or offshore installation that are also equipped with AIS. Since AIS is mandatory for ships above 300 GT in international waters, most ships in OWF areas are equipped with AIS. The crew is then warned by the AIS on board. AIS effectively reduces the collision risk of powered ships with AIS on board. As long as there are no consolidated findings from empirical data or detailed simulations, the factor of effectiveness is assumed conservatively to be 1.25.

port of hazardous goods)

Adjacent offshore installations obstruct traffic. This influences the collision and oil spill risk for the OWF under investigation. Protected and restricted zones (e.g. natural conservation areas, firing danger areas) in the vicinity of the planned OWF can influence shipping traffic in a similar manner as adjacent offshore installations. There are to be considered appropriately.

Traffic Control and Traffic Monitoring

All environmental conditions used as input data in the quantitative risk analysis should be described specifying sources for data. This includes always wind, sea states, currents and tides. Ice and visibility may be considered in addition, if appropriate or required by the authorities. Risks can be reduced by traffic control (by state authorities) or traffic monitoring (by OWF operators). The objective is to identify and warn ships with a course towards the OWF or its surrounding safety zone, and to give navigational suggestions or orders when indicated.

RISK CONTROL MEASURES Impact for drifting ships

Risk control measures are design and operative measures to reduce existing risks, by decreasing the frequency of occurrence or the severity of the consequences. In listing the risk control measures, one should

discuss how frequency of occurrence or severity of consequences are reduced, state whether the measures are
- in place - planned and expected to be implemented - planned, but implementation is doubtful by the time the

OWF will enter operation. For risk control measures, where the implementation is doubtful, reasons for doubts are to be stated and (if possible) alternatives should be given.
- not planned. For risk control measures which are not

Traffic control can identify ships drifting towards the OWF, contact them specifically and (when indicated) initiate salvage operations. AIS and radar support this process. The effectiveness of the measures depends on the required effort, training and number of involved persons. AIS surveillance will identify drifting ships with active AIS with high probability. For the North Sea, it is assumed that at least 90% of all ships send AIS data. There are several AIS and radar based methods to identify drifting ships that do not actively report their problems. AIS offers some advantages compared to radar, because ship specific data and status reports are transmitted in addition to the position, heading and velocity data based on GPS. Radar gives less information, but here the signal does not depend on the ship. The highest success rate for identification is achieved by combining AIS and radar. Software to identify drifting ships automatically is being developed. Studies of Germanischer Lloyd indicate an almost certain identification of drifting ships with AIS. The AIS may actively report the problem (e.g. not

746

under command or restricted ability to manoeuvre) and the drifting can be passively identified by simple evaluation of the difference between reported course over ground and heading) or more complex evaluation of time histories of several AIS parameters (speed over ground, course over ground, heading). If both AIS and radar are used it can be assumed that 98% of all drifting ships are identified. Communication with trained nautical staff may also reduce the collision risk. Support comes in the form of advice for specific use of remaining equipment (side thruster, rudder, anchor) to influence the drift. In addition, the psychological effect of counselling may in itself improve the situation. As this effect is difficult to quantify, an effectiveness factor of 1.1 is assumed conservatively; i.e. in approximately 10% of all identified cases communication can prevent a collision between drifting ship and OWF.

Table 1: Risk reduction due to traffic control / monitoring for powered ships Scenario assumed effectivecollision of powered ship with OWF installations ness factor without monitoring and warning messages 1 with monitoring following variant 1 4 with monitoring following variant 2 3 with monitoring following variant 3 2.5 Table 2: Risk reduction due to communicative nautical support of identified drifting ships Scenario assumed effectiveness collision of drifting ship with OWF installations factor without communicative support by trained nautical 1 staff with communicative support by trained nautical staff 1.1

Impact for powered ships

Three variants are distinguished in considering measures of traffic control / monitoring for powered ships. These variants differ in terms of equipment, approach and effectiveness.
- Variant 1: Comprehensive traffic control / monitoring

Salvage Tug Capacity

Salvage tug reduce the risk only for drifting ships. Requirements for a successful salvage operation are identification of the drifting ship and availability of the tug. These are to be estimated justifiably. Scenarios for tug operation to avoid collisions should be discussed. For each tug considered, the following should be specified:

A comprehensive monitoring consists in a permanent monitoring of the traffic by trained nautical staff using AIS and radar. Within the harmonised assumptions, an effectiveness factor of 4 for the reduction of the collision probability is assumed.
- Variant 2: Automatic monitoring with manual option

Here there is a permanent automatic evaluation of all AIS equipped vessels with an additional manual evaluation in regular intervals. If threshold parameters are exceeded (e.g. ship on collision course in a given distance) an automatic alarm is given with subsequent manual monitoring of the vessel by nautical staff. After identification of the vessel, the potential collision vessel can be warned by radio or satellite telephone. An effectiveness factor of 3.66 was found for a traffic monitoring based solely on AIS. As the AIS signals are initially evaluated automatically, a lower effectiveness should be assumed. An effectiveness factor of 3 for the reduction of the collision probability is assumed.
- Variant 3: Automatic evaluation

name of tug performance parameters (e.g. bollard pull, maximum speed), operational limits (e.g. wind force, sea state) position (coordinates following WGS 84, sea chart with marked tug port(s) and OWF) transit time from alarm to arrival at the planned OWF considering prevailing weather conditions and geography availability (in percent).

Contrary to Variant 2, the traffic is monitored permanently automatically via AIS. If threshold parameters are exceeded an automatic alarm is given. After manual verification suitable measures are taken. In addition, a text message via AIS or a distress and safety message via DSC channel 70 can be sent automatically to the identified ship. As the AIS signals are evaluated only automatically, an even lower effectiveness should be assumed. An effectiveness factor of 2.5 for the reduction of the collision probability is assumed. Tables 1 and 2 compile the effectiveness factors for the reduction of the collision probability due to traffic control / monitoring for the different variants.

The success of the salvage tug will depend on the identification of the drifter, the availability of a suitable tug, the time required to stop the drift and the prevailing weather conditions.

Identification Salvage tugs can assist drifting vessels if they are identified. The fraction of identified drifters depends on the quality of traffic control or traffic monitoring. In the risk model a conservative estimate is made that only 15 % of drifting vessels will report their status in time to the authorities. If in the sea area under consideration a traffic control or monitoring scheme is in place an event tree predicts that 88 % of all drifters will be identified, when interpreting the ships AIS signal. If in a combined evaluation the radar and AIS signals are analyzed, the detection rate of drifting vessels can be assumed to be 98 %. These are the assumed factors for the initialization of available salvage tugs in the risk analysis. Availability Apart from the alarm, a salvage attempt requires an available salvage tug that can reach the drifting vessel before it collides with a WEI. To

747

be able to stop the drift of a vessel, the tug in question must feature enough bollard pull. In the risk model the prevailing weather conditions at the time of the drift and the specifications (bollard pull, size) of pulling and pulled vessel are considered. In the German Bight, the salvage tug Oceanic is on alert position 10 nm north of Norderney. The vessel will leave this position only for refilling, maintenance and repairs and will generally be substituted by other vessels for these times. It is assumed that the vessel is available for 96 % of the time, i.e. 16 days of the year it is not available. Time for salvage The time required for a successful salvage is a sum of the times for alarming and activating of the salvage tug, the time required for the journey to the disabled vessel, the time to establish a towing connection between the ships and the required time to stabilize the vessel (Fig. 1), Table 3.
failure probability of salvage tug trials
1 0.9 0.8 0.7

Table 3: Required time of salvage Time for alarming and activating Journey time Time to establish towing connection Time to stabilise and start towing

0.5 h variable (v=14 kn) 1h ( 8 Bft. 3.5 h) 0.5 h ( 8 Bft. 1.0 h)

Tug Effectiveness Optionally, the tug effectiveness can be determined explicitly from the calculated collision probabilities for drifting ships: tug effectiveness = collision probability without tug collision probability with tug

0.6 0.5 0.4 0.3 0.2 0.1 0

The tug effectiveness can be given for the analysis of individual wind farms and for the cumulative analysis of OWFs. The tug effectiveness was determined by GL for several wind farm clusters (Dausendschn and Povel, 2008). Exemplarily the results for the wind farm cluster consisting of the areas north of Borkum are shown in Fig. 2. The simulations were considering the prevailing traffic and environmental conditions within the area. For this cluster between the two traffic separation schemes German Bight Western Approach and Terschelling German Bight two variants of traffic control or monitoring systems are considered. The first option is based on the evaluation of AIS data; in the second option the sea area is additionally monitored with radar. gives the tug effectiveness of both options for the cluster north of Borkum.

failure probability

activation

journey

towing connection time after average

stabilising

Fig. 1 Failure probability of salvage attempts. With effective decision procedures in the German Bight, the decision of alarming a salvage tug is taken in a few minutes. The considered tug Oceanic is on stand-by at sea, so that it will be ready for action in an average time of 30 min. The time required for the journey to the disabled ship depends on the distance to the ship and the speed of the tug. The speed itself depends on the prevailing weather conditions. In the model of GL the speed is estimated conservatively to 14 kn (corresponding for the `Oceanic to 9 Bft headwind and a significant wave height of 5 m). The time to establish the towing connection depends on prevailing sea conditions. Under normal weather conditions 1 hour is considered sufficient for wind forces up to 7 Bft. For higher wind forces (8 Bft), 3.5 h are assumed for stabling the connection. After establishing a connection between the vessels, the process of stabilising the vessel starts. The time depends on the performance of the tug boat and the weather conditions. Moore et al.(2001) report an average time of 15 min. to stabilise tankers at 30 kn wind speed. Towing tank simulations at the Institute for Maritime Logistics / ISSUS demonstrated that these time spans can extend to one hour. In the simulations, 30 minutes are taken for wind speeds up to 7 Bft and 1 hour for wind speeds of 8 Bft and above.

It has been observed by evaluating several wind farm clusters that the tug effectiveness increases with decreasing distance to the standby position of the vessel.The efficiency of additional local tugs close to the considered wind park cluster has been investigated by GL (Dausendschn and Povel, 2008).

cluster north of Borkum

Oceanic

Fig. 2 Traffic in vicinity of analysed cluster. (AIS data: red westbound, green eastbound)

748

Table 4: Factors of tug effectiveness.

Salvage tug 200 t bollard pull north of Norderney Identification of drifting vessels with AIS Identification of drifting vessels with AIS and radar Tug effectiveness on drifting vessels 3.7 5.2

The probability PfK,k for ships on collision course describes the percentage of ships in the sea area having a collision course with the OWF. For the computation, the model assumes that ships sail on idealised routes consisting typically of the shortest way between departure port and destination port. Due to nautical restrictions, the course needs to be changed in general at way points. A route thus consists of at least one route segment. Route segments are limited by departure port, destination port, or way points. Due to inaccuracies in setting the course and offsets due to wind, current and waves, the actual routes of ships differ more or less from the idealised route. For time and cost reasons, the ships command will try to sail the ideal, shortest route. Thus small course deviations are more probable than large deviations. For the calculation, the deviation of ships orthogonal to the ideal route is described in so-called lateral distributions (statistical distribution functions). For shipping routes without any nautical restrictions, one assumes Gaussian distributions. For prescribed lanes on traffic separation zones or buoyed waterways the maximum of the distribution function (modal value) is shifted towards the edge of the waterway, because the buoys are often used for orientation and collision regulations prescribe keeping to the right side. The probability PfK,k that a ship is on a collision course with a WEI or transformer station follows from the area under the lateral distribution hl = f(x) of route segment k, covered by the shadow area of WEI or transformer station (UW) plus 1.2 times the ship width B. For the i-th installation with diameter Di, this probability is calculated as follows:

RISK ASSESSMENT
A significant part of the risk assessment is the qualitative risk analysis. The analysis identifies the potential hazards and evaluates their associated consequences. Qualitative risk analyses generally investigate only individual events respectively failures, not event or failure combinations. In the preparation of a failure tree analysis or a Monte-Carlo simulation, the qualitative analysis is an important step, identifying the multitude of events with relevance to the risk. One distinguishes between three major types of risks: Risk for the safety of persons, risk for the environment and risk for property. Although some events may contribute to more than one of these types, a combined assessment of different risk types is difficult and should be avoided. The qualitative risk assessment is a formal, inductive and deterministic method to identify and assess possible hazards and their consequences. Structure and procedure in this approach follow in analogy to a Failure Mode and Effect Analysis (FMEA). Frequencies and consequences of hazards are estimated qualitatively. Hazards identified to imply a high risk have to be assessed quantitatively. For OWF projects in the German exclusive economic zone (EEZ) the collisions of powered and drifting ships with WEIs have been identified to have a considerable impact on the risk of the OWF.

PfK ,k = Pfk ,k ,i =
i i

k Di +1.2 B

( x) dx

(2)

RISK CALCULATION FOR COLLISIONS

The two scenarios powered ship (failure to follow proper route) and drifting ship (failure of propulsion) are considered separately.

The assumption of 1.2 times the ship width considers the possibility of an oblique collision. This may occur when the ship moves with a small drift angle to the course line (e.g. due to wind). Besides category I collisions (due to lateral offset of ships), also category II collisions (due to missing a way point) are considered following Friis Hansen (2000). The probability PfM,k that a collision course is not changed in a route segment can be determined from statistical data or using analytical methods (failure tree analyses; Bayesian networks). The first option drops out due to lack of statistical data. However, ship collisions with bridges, grounding and ship-ship collisions for various sea areas give an idea for an order of magnitude. Table 5 summarizes minimum and maximum values for PfM,k for various accident scenarios in shipping (Friis Hansen, 2000). Table 5: Bandwidths for probability PfM,k from other accident scenarios in shipping Scenario Min.value Max. value ship - ship collision (frontal collision) 4.910-5 5.1810-4 ship ship collision (side collision) 9.510-5 1.1110-4 -5 ship grounding 8.010 6.310-4 ship bridge collision 5.410-4 4.010-5

Collisions of Powered Ships with OWFs

Pedersen (1995) presented an approach to compute the collision frequencies of ships with offshore installations. The approach is based on two conditions, which both need to be met for a collision to happen: 1. 2. The ship must be on a collision course with a WEI or transformer station. The probability for this is denoted by PfK,k. The ships command corrects the course too late or not at all. The probability for this is denoted by PfM,k.

For the computation of the collision frequency, these probabilities are multiplied by the number of ship passages on a route segment nS,k:

ncoll , pow =

n
k

coll , pow , k

P
k

fM , k

PfK , k n S , k

(1) Fujii and Mizuki (1998) give for collisions of ships with fixed objects a bandwidth of 0.81210-4 < PfM,k < 5.2610-4. This interval resulted from radar observations and contains all sorts of reasons for collisions including human error. The GL guideline recommends to take PfM,k = 3.010-4. This value corresponds to the harmonised assumptions and lies in the middle of the interval given by Fujii and Mizuki (1998).

If a sea area contains several shipping routes the total collision frequency ncoll,pow of all routes is found by adding the collision frequencies ncoll,pow,k for the individual route segments. The calculation of the probabilities PfM,k and PfK,k is explained in the following.

749

Collisions of Drifting Ships with OWFs

The collision frequencies for drifting ships is calculated for each route segment as the product of the probability of loss of command, probability of collision due to drift motion, probability that efforts to prevent collision fail, and number of ships on the route segment. The basic assumption for loss of command is the failure of rudder or propulsion system. According to the harmonised assumptions, the failure rate is set to 2.510-4 per hour. This corresponds to a mean time between two failures of 4000 hours. The collision probability due to drifting is calculated in a Monte-Carlo simulation, by filtering all possible parameter combinations. The collision probability of ships with OWF installations is determined by simulating drifts in many simulation runs, prescribing for each run randomly new initial conditions. Traffic and environmental parameter are determined in each run based on their probability of occurrence at the location of the drifting vessel. This includes ship type and size, position of the vessel at start of drift, wind speed and direction and currents (tidal current as a function of time and wind current as a function of prevailing wind). Each run determines whether there is a collision risk for the given initial conditions. The collision probability is the ratio of number or simulation runs with collision risk to total number of simulation runs. Initially, risk control measures are not considered in the calculation. These are considered later in calculating the collision risk including risk control measures. In order to check whether drift direction leads to collision, the area (drift corridor) is calculated that is covered by the drifting ship as shown in Fig. 3. If an OWF installation or transformer station lies within this corridor there is a collision risk. The collision risk for drifting ships with WEIs follows from the product of collision frequencies and collision consequences (mass of oil spill).

for ship-WEI collisions and for personal safety. The risk assessment links consequences and frequencies of occurrence. The risk matrix links categorised frequencies and categorised consequences. Probability or frequency of occurrence can be determined in failure tree analysis or simulation. The collision safety can be estimated with the help of appropriate simulation software (collision analyses). Explicit finite-element analyses for dynamic simulations or equivalent state-of-the-art codes shall be used. Simple reference to already performed collision analyses of WEIs is admissible, provided that the boundary conditions in the previous analyses are at least as severe as required for the WEI now under investigation. The combination of frequency and consequence yields a risk priority number (RPN) between 1 and 7. The RPN is an indicator for the risk. RPN 1 to 3 represent low risks, acceptable to the administration. For higher RPN the authorities may require additional analyses or measures. The frequencies and consequences are quantified in the BSH standard The Frequencies are defined quantitative (Table 7); the consequences are defined qualitative (Table 8) for the ship-WEI collision, for the environment and for personal safety individually. Table 6: Risk matrix catastrophic Consequence serious considerable insignificant

4 3 2 1 extremely rare

5 4 3 2 rare

6 5 4 3 occasional

7 6 5 4 often

Frequency of occurrence

vs Driftkorridor
Table 7: Classification of frequency (after BSH, 2007)
qualitative probable often Frequency H [1/year] H > 0.1 Safety / quantitative Frequency H [1/year] H > 0.2 Environment / quantitative improbable occasional improbable rare extremely improbable extremely rare H 0.001

0.1 H > 0.01

0.01 H > 0.001

0.2 H > 0.02

0.02 H > 0.002

H 0.002

Fig. 3 Drift corridor of a drifting ship (not to scale)

ACCEPTANCE CRITERIA
The German Authorities have published the requirements for approval of a OWF in the German EEZ in the standard on engineering design of offshore wind energy installations (BSH, 2007). According to the standard a quantitative risk analysis is required. The risk assessment follows the risk matrix of Table 6. Risks are assessed for the environment,

750

Table 8: Classification of consequences (after BSH, 2007)

Consequence class insignificant considerable serious catastrophic

CONCLUSION The developed GL guideline for risk assessment for collisions of ships with offshore wind farms is based on harmonized assumptions which were agreed upon by various expert institutions. The guideline is expected to contribute towards a comparable, transparent and repeatable process for such risk assessments. Furthermore the paper gives an overview of the efficiency of risk control measures and on the acceptance criteria of the German authorities for the risk of OWFs in the German EEZ. REFERENCES

WEI operational (might require repairs)

WEI fails

WEI destroyed

nacelle or large parts thereof fall on ship

ship-WEI

significant pollution large pollution Consequences none or small pollution oil or fuel from hopper tank or double bottom tank pour into the water (double hull not penetrated) environment cargo tank leakage and spill of cargo (double hull penetrated) ship brakes apart / ship sinks

nobody injured

a few persons injured

severely injured, a few casualties

high number of casualties

BSH (2007), Konstruktive Ausfhrung von OffshoreWindenergieanlagen, Bundesamt fr Seeschifffahrt und Hydrographie, Hamburg. Dausendschn, K, Povel, D (2004), Offshore Windparks Wirksamkeit kollisionsverhindernder Manahmen - Abschlubericht Report Germanischer Lloyd. http://www.offshore-stiftung.de/PDFDokumente/NB-ER%202008.178_V1.8klein.pdf. Fujii, Y, and Mizuki, N (1998), Design of VTS Systems for Water with Bridges, Ship Collision Analysis, Balkema ISBN 90 54 10 962 9, Rotterdam. Friis Hansen, P (2000), Grounding and Collision: Basic Modelling Principles and Validation of Software for Prediction of Frequencies, ISESO Project 7 and 8, Report I107/I108.00.02.052.004. GL (2010), Guideline for Technical Risk Analyses for Offshore Wind Farms, GL Guideline, Germanischer Lloyd, Hamburg. Moore, C, Bossett, N, Michel, K, and Boniface, D (2001), An Application of Collision and Grounding Simulation to Regulatory Assessment, Int. Conf. Collision and Grounding of Ships; Copenhagen. Pedersen, PT (1995), Collision and Grounding Mechanics, Proc. WEMT `95, Copenhagen

personal safety

751