Statistics of Collision, Grounding and Contact Accidents of Passenger and Container Ships

Transcription

1 Statistics of Collision, Grounding and Contact Accidents of Passenger and Container Ships A. Pagiaziti 1), E. Maliaga 1), E. Eliopoulou 1), G. Zaraphonitis 1) and R. Hamann 2) 1) National Technical University of Athens, Greece, 2) DNV-GL, Germany, Abstract The main objective of this paper is to present the results of the analysis of collected data from accidents to passenger ships and container ships. The accident types that have been considered are collision, grounding and contact. The developed databases are aimed to provide the necessary data to serve the following two distinct objectives: The development of a quantitative risk model for the damage stability related risk to human life associated with collision and grounding accidents to passenger ships and the development of a probabilistic model for the damage characteristics, in the form distribution functions for the location and extend of a breach, as a result of a grounding accident. Keywords Collision, Grounding, Risk model, RoPax, Cruise ship. 1. Introduction Safety of ships against sinking/capsize in case of loss of their watertight integrity is a subject of prime importance to the international and national maritime regulatory bodies, the maritime industry and the scientific community. In 2002, IMO approved the guidelines on Formal Safety Assessment (MSC/Circ.1023, 2002) in order to support decision making by a process formalizing the assessment of new regulations using risk analysis and cost benefit assessment. Meanwhile the guidelines on Formal Safety Assessment (FSA) has been further updated (MSC-MEPC.2/Circ. 12, 2013). Based on this procedure, a series of FSA studies have been elaborated within the EU research project SAFEDOR for the most important ship types including ship types in focus of this investigation (. MSC 83/INF.8, 2007, MSC 83/INF.2, 2008, MSC 83/INF.3, 2008). The new SOLAS damage stability regulation, based on a probabilistic model was developed in the same period and came into force in In the years that followed the introduction of SOLAS 2009, the assessment of the impact of the new regulation on ship s safety and the investigation of possible improvements continued to be a research topic of the highest priority, especially for passenger ships (EU funded FP7 project GOALDS, EMSA I, EMSA II). So far, with the exception of the High Speed Craft Code, damage stability regulations are focusing mainly on collision and even if the impact of grounding accidents on the safety of ships has been the subject of extensive research for several years (Vanem and Skjong, 2004, Kehren and Krüger, 2007, Cerup- Simonsen et al. 2009, Sormunen 2014) a related stability criterion found no consideration in SOLAS. A research project was assigned recently by EMSA (EMSA III) to a research group coordinated by DNV GL, aiming to assess the risk level of passenger ships related to damage stability. One of the main objectives of this study is to evaluate the risk from grounding accidents, with emphasis on side-raking damages (in view of the accident of the cruise ship Costa Concordia in January 2102) and to propose possible amendments to the regulatory framework. The basic prerequisite of EMSA III is to collect, investigate and analyze data from accidents to passenger ships, and the development of a database, in order to provide a sound basis for the elaboration of the above objectives. More specifically, the developed database should provide the necessary data to serve the following two distinct objectives: The development of a probabilistic model based on accident statistics for the damage characteristics, in the form of distribution functions for the location and extend of a breach, as a result of a grounding accident. The update/development of risk models with respect to collision and grounding accidents to passenger ships. Although in principle, the employed procedure could address also the environmental risk; this work is currently limited to the risk to human life. Both serious and non-serious accidents are recorded in the database. However, only serious accidents are considered for the development of the probabilistic model for the damage characteristics and the risk model. In order to meet the requirements of the first objective, it is necessary to collect and analyze quantitative data (measurements) for the breach dimensions (length, width, penetration) and its location along the ship s hull. Fortunately, major accidents to passenger ships are relatively rare events; however, a probabilistic approach to damage stability requires the availability of an adequate data sample. Therefore, it was decided to extend 1

2 the scope of work, in order to include the development of an additional database with data from accidents to container ships, considering certain similarities in the damage characteristics between these two ship types. This is a procedure that was adopted also in the GOALDS project (Papanikolaou et al. 2013), where the considered ship types were divided into two main categories, i.e. full ships and non-full ships. The analysis of data from grounding accidents, carried out within GOALDS, verified the initial assumptions with respect to the statistical distribution of damage characteristics of these two ship categories, revealing a common behavior of the statistical values of the grounding damage characteristics of passenger ships and container ships (non-full ships) on one hand, and tankers and bulk carriers (full ships) on the other hand. The elaboration of the second objective (i.e. the development of risk models for collision and grounding accidents) is generally based on qualitative information, such as area of operation at the time of the accident, operational conditions (e.g. powered or drift grounding), occurrence of a hull breach, water ingress etc. Since this type of data is much easier to find than the quantitative data, it was possible to exclude data from container ships from this part of the work and to develop the risk models based on only data coming from accidents to passenger ships. The accident types considered included collision (CN), grounding (GR) and contact (CT) 1. Collision accidents were included because one of the objectives of the EMSA III study is to revise and update the corresponding risk model developed in GOALDS, considering additional information from recent accidents. Contact accidents were included because they are associated with hull breaches at the side of the ship, which are of particular interest for the present study. Two already existing ship accidents databases, developed by the Ship Design Laboratory of NTUA in the framework of the EU project GOALDS (passenger ships) and of the bilateral research project CONTIOPT (container ships) carried out by Germanischer Lloyd SE and NTUA (Eliopoulou et al. 2013, Hamann et al. 2013) have been used as the starting points of the present work. These databases were extended to include as many additional accidents as possible, while at the same time the already included accidents were thoroughly revisited in order to verify existing data and to supplement it with missing information, as far as possible. Relevant information was searched in Sea-web and GISIS (Global Integrated Shipping Information System), from the EMSA III partners, Flag Administrations and from the internet. In the following, the collected accidents data will be summarized in a series of tables and figures, while the de- 1 The definitions for the accident types used in this study are those given in MSC/Circ.953: Collision: striking or being struck by another ship (regardless of whether under way, anchored or moored). Stranding or grounding: being aground, or hitting/touching shore or sea bottom or underwater objects (wrecks, etc.). Contact: striking any fixed or floating object other than those included in Nos. 1 or 2. veloped/updated risk models will be presented and briefly discussed. 2. Passenger ships In total, 430 accidents to passenger ships have been identified and included in the database. The following parameters were used to filter the casualty data: Ship types: cruise, pure passenger ships, Ro-Pax and RoPax-Rail; Casualty time period: ; GT 1,000; L OA 80 m; Built 1982; Froude No. 0.5 (in order to eliminate High Speed Craft from the study). The distribution of collected data with respect to the type of accident and the four passenger ship types is summarized in Table 1. Given the operational scenario of RoPax and RoPax Rail ships, which in several cases includes many port calls per day, an increased frequency of contacts should be reasonably expected. As a matter of fact, the most common type of accidents for RoPax and RoPax Rail ships is the contact (38%), followed by collisions (32%) and groundings (30%). For cruise and pure passenger ships on the other hand, grounding is the most frequent type of accidents (41%), followed by contacts (31%) and collisions (28%). Table 1: Passenger ships database: number of reports subdivided over type of accident and type of ship Ship type CN GR CT Total RoPax RoPax-Rail Cruise Passenger Total Three operational states, associated with different sea areas, ship speed and traffic densities, were considered for the risk analysis: En route open sea: 12 miles off the coast. The ship has full operational speed. Limited waters coastal: < 12 miles off the coast, harbor, rivers/canals. Terminal areas: at anchorage or berth, in port or port approach (entrance). The distribution of reports with respect to the type of accident and the operational states at the time of the accident is summarized in Table 2 independent of ship type. More than half of the collisions (54%) occurred in terminal areas, followed by limited waters (37%) and only 10% in the open sea. Approximately 75% of groundings occurred in limited waters, only 21% in terminal areas, while there are 3 accidents (2%) reported to occur in the open sea (obviously in shallow waters but at a distance of more than 12 sea miles from the nearest shore). As expected, most contacts (79%) oc- 2

3 curred in terminal areas, a smaller percentage (18%) in limited waters and 4 of them (2%) in open sea. Table 2: Passenger ships database: type of accident and area of operation Area of operation CN GR CT Total Open Sea Limited waters Terminal areas Unknown Total The impact of collision, grounding and contact accidents to RoPax and RoPax-Rail ships only is summarized in Table 3. Corresponding results for the cruise and pure passenger ships are presented in Table 4. According to these results, 16% of collision accidents resulted in a major damage in case of RoPax and RoPax-Rail ships, and 20% in case of cruise and pure passenger ships. No total loss is reported as a result of a collision accident. In case of groundings, 43% of the accidents to RoPax and RoPax-Rail ships and 48% of the accidents to cruise and pure passenger ships resulted in major damage, while 4 ships were lost and one was broken up. Although contacts are generally assumed to be associated with minor consequences, it is interesting to note that 23% of the accidents to RoPax and RoPax Rail ships 26% of the accidents to cruise and pure passenger ships resulted in major damages 2. Table 3: Impact of the accident on RoPax and RoPax- Rail ships Accident s severity CN GR CT Total No damage sustained Minor damage Major damage Total Loss Break up Unknown Total marized in Table 5. The fatalities from grounding accidents to RoPax ships reported in this table are all related with one single accident: the Princess of the Stars lost in South China Sea in 2008 with 523 persons killed and 308 missing. The ship was caught by a typhoon and it was claimed that at some point engine troubles were reported, while subsequently it run aground. Sometime later, it developed list and capsized. The circumstances of this accident are relatively unclear and may not be suitable for drawing conclusions regarding the impact of grounding accidents on human life. From the 34 fatalities resulting from grounding accidents to cruise ships reported in Table 5, 32 are related with the accident of Costa Concordia and the remaining two with the sinking of Sea Diamond. Table 5: Impact on human life Ship type CN GR CT Total RoPax RoPax-Rail Cruise Passenger Total The location of the breach(es) along the ship s hull in case of collision accidents to RoPax and RoPax-Rail ships or cruise and passenger ships is summarized in Tables 6 and 7 respectively. Table 6: Location of breach(es) from collision accidents to RoPax & RoPax-Rail ships Breach Location Struck Striking Unclear Total Side Bottom Bow Stern Outfitting Unclear None Total Table 4: Impact of the accident on cruise and pure passenger ships Accident s severity CN GR CT Total No damage sustained Minor damage Major damage Total Loss Break up Unknown Total The impact of the accidents on human life (number of persons killed plus number of persons missing) is sum- 2 Identification of major damages is not always straightforward. As a general rule, a ship suffering a major damage would be sent for dry-docking or remain out of service for more than three days. Typical examples of major damages include flooding of the engine room or other large spaces, damages to propeller and/or rudder, large gashes or deformation of plates or staying aground. Table 7: Location of breach(es) from collision accidents to cruise & pure passenger ships Breach Location Struck Striking Unclear Total Side Bottom Bow Stern Outfitting Unclear None Total The location of the breach(es) along the hull of the ship in case of grounding or contact accidents to passenger ships is summarized in Tables 8 and 9 respectively.the annual distributions of serious and non-serious accidents to passenger ships are presented in Fig. 1 to Fig. 6. A considerable increase in the number of accidents per year is observed in these figures roughly after The same tendency has been observed also in other studies, looking to other accident types and/or other ship types. 3

4 It is quite probable therefore that this increase has to do with a change in the reporting practice. It should be also noted that, since we are looking at accidents to ships being built on or after 1982, in 1990 we are considering only a small part of the fleet at risk, with age not greater than 8 years, while in the following years we are considering a continuously increasing percentage of the fleet (e.g. in 2008 ships with age up to 26 years). This fact partly explains the small number of accidents at or shortly after year 1990, and its subsequent gradual increase, up to year 2008, but can offer no explanation for the decreasing trend in the number of accidents during the last five years. Apart from a possible congestion of sea routes due to the increased number of ships during the last decades, there is no obvious reason to believe that there is any real increase in the accidents frequency over the last 10 years, and it is therefore concluded that the recent figures are closer to reality than older data. The collected data from accidents to passenger ships were used for the development/update of the risk models for grounding and collision accidents to cruise and RoPax ships, presented bellow (sections 5 and 6). In addition, the quantitative data for the breach characteristics, along with corresponding data from accidents to containerships, were used as the basis for the development of a probabilistic model for the location and extend of breaches to the ship s hull, resulting from grounding or contact accidents to non-hull ships. Table 8: Location of breach(es) from grounding accidents per type of ship Breach RoPax & Cruise & pure Location RoPax Rail passenger ships Total Side Bottom Bow Stern Outfitting Unclear None Total Table 9: Location of breach(es) from contact accidents per type of ship Breach RoPax & Cruise & pure Location RoPax Rail passenger ships Total Side Bottom Bow Stern Outfitting Unclear None Total Fig. 1: Annual distribution of collision accidents to RoPax and RoPax-Rail ships Fig. 2: Annual distribution of collision accidents to cruise ships and pure passenger ships Fig. 3: Annual distribution of grounding accidents to RoPax and RoPax-Rail ships Fig. 4: Annual distribution of grounding accidents to cruise ships and pure passenger ships Fig. 5: Annual distribution of contact accidents to RoPax and RoPax-Rail ships 3. Container Ships In total, 866 accidents to container ships (466 collisions, 265 groundings and 135 contacts) have been identified and included in the database. The following parameters were used to filter the casualty data: Ship types: fully cellular container ships; Fig. 6: Annual distribution of contact accidents to cruise ships and pure passenger ships 4

5 Casualty time period: (October) GT 1,000 Built 1982 The distribution of data with respect to the types of accident and the area of operation at the time of the accident is summarized in Table 10. The impact of the accident is summarized in Table 11. According to the collected data, 44% of collision accidents to container ships resulted in a major damage, while no total loss is reported. In case of groundings, 41% of the accidents resulted in major damage, six ships were lost and five were broken up. In case of contacts, 53% of the accidents resulted in major damage while no total loss is reported. The impact of the accidents on human life (number of persons killed plus number of persons missing) is presented in Table 12. The location of the breach(es) along the hull of the ship in case of collision accidents or grounding and contact accidents to container ships is summarized in Table 13 and Table 14. Table 10: Container ships database: type of accident and area of operation Area of operation CN GR CT Total Open Sea Limited waters Terminal areas Unknown Total Table 11: Container ships database: impact of the accident Accident s severity CN GR CT Total No damage sustained Minor damage Major damage Total Loss Break up Unknown Total Table 12: Container ships database: impact on human life Ship type CN GR CT Total Container ships Table 13: Location of breach(es) for collision accidents to container ships (struck / striking) Breach Location Struck Striking Unclear Total Side Bottom Bow Stern Outfitting Unclear None Total The annual distributions of serious and non-serious accidents to container ships are presented in Fig. 7 to Fig. 9. Similar to passenger ship accidents a quite small number of accidents is reported before year 2000, followed by a gradual but at the same time quite significant increase during the next 10 years and then a gradual reduction until year Comments and explanations set forth in the case of passenger ships are assumed to be valid also in case of container ships. The quantitative data for the breach characteristics, with corresponding data from passenger ships, were used for the development of a probabilistic model for the location and extend of breaches to the ship s hull, resulting from grounding or contact accidents to non-hull ships. Table 14: Location of breach(es) for grounding and contact accidents to container ships Breach Location GR CT Total Side Bottom Bow Stern Outfitting Unclear None Total Fig. 7: Annual distribution of collision accidents to container ships Fig. 8: Annual distribution of grounding accidents to container ships Fig. 9: Annual distribution of contact accidents to container ships 4. Quantitative data regarding grounding/contact related hull breaches As has already been mentioned, one of the main objectives of this study is to provide a set of quantitative data (measurements) regarding the breach dimensions (length, width, penetration) and location along the ship s hull, based on which a probabilistic model for the characteristics of damages resulting from grounding or contact accidents may be developed. In this section, a 5

6 summary of the collected data is presented, but at first it is useful to describe in brief the types of breaches that have been considered. A typical damage due to grounding is usually considered to start at the bottom and extend vertically in (and possibly above) the double bottom. This is the type of damage that was investigated in detail during the GOALDS project. Within the context of the present study, however, it was considered necessary to investigate an additional type of longitudinal damage, starting from the ship s side and extending inwards with a (primarily) horizontal penetration direction. Simplified geometrical models have been introduced to describe the shape of each type of damage: Bottom damages (denoted in the following as B00 type of damage) are assumed to have an orthogonal shape, uniquely described by a set of five parameters: the starting point of the damage (i.e. the longitudinal position of its foremost point), the transversal position of the damage (the transverse distance from the ship s center plane to the center of the breach), the damage length, transversal width and penetration measured vertically from the bottom. The geometrical model introduced for side damages (denoted in the following as S00 type of damage) has a relatively more complex shape: assuming a constant depth of penetration, the hull breach is following the ship s waterline at a certain height. Based on this assumption, a side damage may be uniquely described by the following set of parameters: the starting point of the damage (i.e. the longitudinal position of its foremost point), the damage length, the height of the lower point of the damage from the base plane, the vertical extend along the side of the hull, the penetration, measured inwards from the hull side, and an indicator for the side of the damage (port/starboard). It is acknowledged of course that, in reality, grounding damages may have much more complex shapes. However, it is not possible to develop generic geometric models capable of resembling all possible damage shapes and, at the same time, being practically applicable for the elaboration of quite complex numerical tasks, such as the probabilistic damaged stability calculations. In this respect, the simplified models of the bottom and side damage described above are considered to represent a reasonable compromise between the contradicting requirements for generality and simplicity, providing a suitable basis for the development of the corresponding probabilistic models. It should be noted at this point that, in case of multiple breaches, an artificial damage envelope is used, corresponding to the bounding region enclosing all the breaches. This is the same procedure followed in GOALDS for the case of bottom damages. Data relevant to accidents of type B00 (bottom damages) have been extensively analyzed during the GOALDS project, and a probabilistic model for the damage characteristics, with the corresponding distribution functions, is readily available. However, such a probabilistic model, or even quantitative data relevant to accidents of type S00 (side damages), particularly for passenger ships, have not been published before, to the knowledge of the authors. The collection and analysis of quantitative data pertaining to side damages was one of the main reasons behind the development of the accidents databases described in this paper. Although the number of accidents in the databases is relatively large, only in a limited number of cases it was possible to retrieve quantitative information regarding the location and extent of the resulting hull breaches. The number of accidents for which it was possible to find this type of information is summarized in Table 15 (side breaches). Table 15: Collected quantitative data for the location and extent of side hull breaches Longitudinal position of forward end of damage Longitudinal extent of damage Damage penetration Vertical position of lower limit of damage Height of damage above lower limit Passenger ships Containerships Total GR CT GR CT As it may be observed in Table 15, quantitative information regarding the actual location and extent of the resulting hull breaches is rather scarce. For some of the cases included in Table 15, quantitative information in the investigation reports was missing, but it was possible to derive reasonable estimations based on other evidence (such as drawings or photographs of the breaches). The damage characteristic for which it was most difficult to find quantitative information was the transverse extent (penetration). From the five accidents included in Table 15, only in the case of one passenger ship grounding and one container ship contact it was possible to find explicit measurements of the resulting penetration. For the remaining three cases, the penetration was estimated based on other evidence. However, for 25 additional accidents (18 passenger vessels, 7 container vessels) it is reported that the penetration was small. In these cases, the breach was qualitatively described as a gash, tear, crack or minor. This fact was explored during the development of the probabilistic model for the side damage characteristics, in order to support the development of a reasonable nondimensional distribution for the damage penetration. A reasonable explanation for the lack of quantitative information on the damage depth would be that it is not possible to measure the penetration unless an inner bulkhead has been involved. Passenger ships are mainly transversely subdivided; therefore the actual size of a (small) penetration may have a minor impact on the survivability of the ship. In addition, in the absence of a longitudinal boundary, placed in a small distance from the hull, it may seem meaningless to define and measure the actual penetration (the longitudinal bulkheads limit- 6

7 ing the lower hold in the case of large RoPax ships are located at a transverse distance from the outer hull which is very far, compared with the penetration from a typical gash). In the case of the container ships, it seems that the inner hull was not affected by the accidents, which is also supporting the hypothesis of relatively shallow penetrations. 5. Grounding Risk Model Starting point for the grounding risk model is the highlevel grounding event sequence (Fig. 10) considering the following high-level events characterizing the development of the consequences: location of accident (operational area), area of contact (side or bottom damage), type of sea bed (hard/soft), possibility of breaching the hull, possibility of water ingress, possibility of the ship remaining aground after the accident, and if not, possibility of sinking. The developed grounding risk model for cruise ships is shown in Fig. 11. The same risk model for grounding accidents is used for both types of ships, RoPax and cruise ships with the following two differences: Initial accident frequencies are determined separately for each ship type. The probability of fast sinking is set equal to 18% for cruise ships and 50% for RoPax ships. Grounding Level 1 Level 2 Level 3 Level 4 Level 5 Level 6 Level 7 Area of Operation (Terminal Areas, Limited waters & Open Sea) Area of contact (Side damage, Bottom damage) Hard/Soft contact point (Hard Bottom/Structure, Soft Bottom) Hull Breach Water Ingress Staying aground Fig. 10: High-level event sequence for grounding of cruise and ROPAX ships These values for the probability of fast sinking of cruise and RoPax ships were initially so defined in the GOALDS project on the basis of sample simulations and considering relevant historical data, and are also used so far in the EMSA III project. Slow sinking is generally associated with progressive flooding, while capsizing, as a result of loss of transverse stability may take place quite fast. However, even in case of capsizing, this could also take place at a later stage, for example as a result of wave action, shift of cargo (in case of RoPax ships), or progressive flooding. What is of primary importance at this point, besides of the actual mechanism leading to the loss of the ship, is the time available for an orderly evacuation of the passengers and crew. The probabilities of fast sinking of cruise and RoPax ships proposed in GOALDS have been up to now kept unchanged also in the present study; however, Afloat Level 8 Consequences it should be noted that a 50% probability of fast sinking of RoPax ships, particularly in case of bottom damages, seems to be quite high. For the calculation of consequences in case the ship sinks or capsizes, the number of fatalities is calculated as follows: 80% of POB (Persons on Board) fatalities in case of fast sinking/capsizing in limited waters or in the open sea; 5% of POB fatalities in case of slow sinking/capsizing, or in case the accident takes place in terminal areas. Initial accident frequencies are determined considering the fleet at risk data for the period from 2000 to 2012 (only IACS classed ships) and ship losses during the same period of time due to grounding accidents, recorded in the accident data base for passenger ships. Therefore, the full set of criteria for the selection of accidents, to be used for the calculation of initial frequencies are: Ship types: cruise and pure passenger ships, OR RoPax and RoPax-Rail; Casualty time period: ; GT 1,000; L OA 80 m; Built 1982; IACS classed; Accident type: serious; Froude number 0.5. The results with respect to casualties and frequencies (casualties per ship-year) of grounding accidents for the time period from year 2000 to year 2012 are summarized in Table 16. The corresponding data for contact accidents are also reported. The fleet at risk is equal to 2763 ship-years for cruise and pure passenger ships and 5328 ship-years for RoPax and RoPax Rail. The dependent probabilities within the risk model are calculated merging the available data from both ship types (cruise and RoPax ships). To this end, casualties reported in the groundings database for passenger ships for the period from 1990 to 2012 are used, including also non-iacs ships. All accidents in the database reported as non-serious have been excluded. Table 16: Number of casualties and accident frequencies for cruise and RoPax ships Cruise ships RoPax ships Casualties Frequencies Casualties Frequencies GR E E-03 CT E E-02 The full set of criteria for the selection of accidents, to be used for the calculation of initial frequencies are: Ship types: cruise and pure passenger ships, AND RoPax and RoPax-Rail; Casualty time period: ; GT 1,000; L OA 80 m; 7

8 Built 1982; IACS and non-iacs classed ships; Accident type: serious; Froude number Collision Risk Model The collision risk model developed in EMSA III project is based on the risk model developed in GOALDS, updated with newly available information. Starting point for the risk model is the high-level collision event sequence (Fig. 12) considering the following high-level events: the ship being struck or striking (initiator), the Fig. 11: Grounding risk model for accidents to cruise ships location of the accident (operational area), the possibility of water ingress, and in case of water ingress the possibility of sinking. The developed collision risk model is shown in Fig. 13 for the example of ship type cruise. The main differences to the GOALDS collision risk models are: Merging scenarios en route and limited waters since the same dependent probabilities were used in both branches; Reduce fatality rate for sinking in terminal area to 5% in order to consider the effects of limited water depth and improved SAR capabilities; Estimate dependent probabilities for the events ini-

9 tiator, operational area and water ingress on basis of a sample received by merging the reports for cruise and RoPAX. The number of casualties for ship categories cruise and RoPax as well as related initial accident frequencies for the periods from year 1994 to year 2012 and from year 2000 to year 2012 are summarized in Table 17. Table 17: Number of casualties for cruise and RoPax ships and initial accident frequencies Time Period Time Period No of casualties Frequencies No of casualties Frequencies Cruise E E-03 RoPax E E-03 Collision Fig. 12: High-level event sequence for collision accidents 7. Conclusions Level 1 Level 2 Level 3 Level 4 Level 5 Initiator Operational Area Water Ingress Sinking Consequences While the risk from collision accidents has been the subject of extensive research and regulatory development for many years, the risk from grounding to passenger and cargo ships received until recently considerably less attention. In practice, the only regulatory requirement aiming to provide protection to conventional passenger ships (with the important exception of those complying with the High Speed Craft Code) against grounding accidents is the construction of a double bottom of ample height, extending from the collision bulkhead to the afterpeak bulkhead, as far as this is practicable and compatible with the design and proper working of the ship, assuming that this would be enough to provide protection and to ensure safety. Historical data, however, indicates that this safety precaution can quite often be insufficient, since a series of grounding accidents resulted in ship losses and a significant number of fatalities. As a matter of fact, in the case of passenger ships the impact of grounding accidents to human life seems more severe in comparision to that of collision accidents. Grounding 3 Serious cases, IACS ships at the time of incident. 4 Calculated considering IACS classed ships and the selection criteria specified: 3290 ship years for Cruise ships, 6738 for RoPax. 5 Calculated considering IACS classed ships and the selection criteria specified: 2673 ship years for Cruise ships, 5328 for RoPax. accidents are traditionaly associated with bottom damages. However, a common characteristic of a series of severe grounding accidents (the most recent of them is the accident to Costa Concordia on the 13 th of January 2012) is that the area of the hull breach is not at the bottom, where the double bottom could offer some protection, but at the ship s side. For this reason, two dinstict types of damage have been considered in the present study: Bottom damages, starting at the bottom and extending vertically in (and possibly above) the double bottom. Side damages, starting from the ship s side and extending inwards with a (primarily) horizontal penetration direction. The first type of damage has been thoroughly investicated during the GOALDS project and a probabilistic model for the damage characteristics is readily available. Such a probabilistic model, or quantitative data relevant to side damages are not available so far. The collection and analysis of quantitative data pertaining to side damages was one of the main objectives behind the development of two databases with data from collision, grounding and contact accidents to passenger and container ships. The second objective was to provide the data required to support the update/development of risk models, to be used for the quantitative evaluation of the risk to human life, associated with collision and grounding accidents to passenger ships. The collected data summarized above is used to develop a new grounding risk model for passenger ships and the update of the risk model developed in GOALDS research project. The risk models presented herein enable the quantification of risk to human life from the operation of a ship, expressed as the (annual) potential loss of life (PLL), associated with collision or grounding/contact accidents. Having defined the initial frequencies of accidents and the dependent (internal) probabilities of the various events for the various combinations of ship types and accidents, the PLL may be readily calculated from these risk models based on the number of persons onboard (POB) and the probability of sinking in case of an accident. The latter is set equal to the ship s Attained Subdivision Index, which in case of collisions is calculated according to the SOLAS 2009 regulation. For grounding accidents however, there is no such procedure defined in SOLAS 2009, or any other regulation to facilitate the calculation of the corresponding A-index. In the framework of the EMSA III study, a proposal for a new probabilistic regulation has been developed, along with the corresponding software tool, enabling the calculation of the attained subdivision index of passenger ships, in case of grounding or contact accidents resulting in hull breach and water ingress. The ultimate objective was to combine this procedure with the presented risk models to calculate the risk to human life for a series of RoPax and cruise ships, in order to quantify the risk from grounding and, if necessary, to propose a new safety standard based on the probabilistic procedure. 9

10 8. Acknowledgements This study has been partially funded by the European Maritime Safety Agency (EMSA) in the framework of project EMSA/OP/10/2013, project no. PP090623, coordinated by DNV-GL. The opinions presented herein are those of the authors. References Cerup-Simonsen, B, Törnqvist, R, Lützen, M (2009). "A simplified grounding damage prediction method and its application in modern damage stability requirements", Marine Structures 22 (2009) pp Eliopoulou E., Hamann R., Papanikolaou A. and Golyshev P (2013). "Casualty analysis of Cellular Container Ships", 5th Int. Maritime Conference on Design for Safety-4th Workshop on Risk-based Approaches in the Marine Industries (IDFS 2013), November, Shanghai China. Hamann R., Papanikolaou A., Eliopoulou E. and Golyshev P (2013). "Assessment of Safety Performance of Container Ships", 5th Int. Maritime Conference on Design for Safety-4th Workshop on Riskbased Approaches in the Marine Industries (IDFS 2013), November, Shanghai China. Kehren FI, Krüger, S (2007). "Development of a Probabilistic Methodology for Damage Stability Regarding Bottom Damages", 10 th International Symposium on Practical Design of Ships and Other Floating Structures, Houston, Texas. MSC/Circ.1023 (2002). Guidelines for Formal Safety Assessment (FSA) for use in the IMO Rule-Making Process. IMO, London (revision: MSC- MEPC.2/Cric.12). Fig. 13: Collision risk model for accidents to cruise ships MSC-MEPC.2/Circ. 12, 2013: Revised Guidelines For Formal Safety Assessment (FSA) For Use In The IMO Rule-Making Process. IMO London. MSC 83/INF.8 (2007). FSA container vessels Details of the Formal Safety Assessment. Danish Submission to IMO Marine Safety Committee, London. MSC 83/INF.2 (2008). FSA Cruise ships - Details of the SAFEDOR Formal Safety Assessment. Marine Saftey Committee, International Maritime Organisation, London. MSC 83/INF.3 (2008). FSA RoPax ships - Details of the SAFEDOR Formal Safety Assessment. Marine Saftey Committee, International Maritime Organisation, London. Papanikolaou, A, Hamann, R, Lee, B-S, Mains, C, Olufsen, O, Vassalos, D, Zaraphonitis, G (2013). "GOALDS - Goal Based Damage Ship Stability & Safety Standards for Passenger Ships", Journal of Accident Analysis and Prevention, Elsevier, /j.aap Sormunen, OV (2014). "Ship Grounding Damage Estimation Using Statistical Models", Probabilistic Safety Assessment and Management PSAM 12, June 2014, Honolulu, Hawaii. Vanem, E, Skjong, R (2004). "Collision and grounding of passenger ships risk assessment and emergency evacuations", In proceedings from International Congress on Collision and Grounding of Ships, ICCGS 2004, Izu, Japan, October 2004.