: RISK ASSESSMENTS OF SHIP COLLISIONS IN HIGH-TRAFFIC MARINE TERRITORY - Research Paper Example

PROJECT- II REPORT IME864: Risk Analysis Fall, Wichita Wichita, KS USA PROJECT-II: RISK ASSESSMENTS OF SHIP COLLISIONS IN HIGH-TRAFFIC MARINE TERRITORY Abhishek Pratap SinghWichita State UniversityWichita, KS, USAEmail: apsingh@wichita.eduAmirmahyar AbdolsamadiWichita State UniversityWichita, KS, USAEmail: axabdolsamadi@wichita.eduMohammed AlharbiWichita State UniversityWichita, KS, USAEmail: abosurur18@hotmail.comNita YodoWichita State UniversityWichita, KS, USAEmail: nxyodo1@wichita.eduXiaolong Cui Wichita State UniversityWichita, KS, USAEmail: xxcui@wichita.

eduAbstractIn high-traffic marine territories, ship collisions have been rated as one of the marine accidents that have a high occurrence rate. Ship collisions can pose an enormously negative impact on the subsea environments, as well as the marine transportation industries in general. Two major risks that are known to be associated well with ship collision were identified as human failures (operational error) and technical failures. In order to minimize the associated risks, Bayesian Network (BN) was employed as an assessment tool to model and quantify the potential failure events that could lead to ship collision.

In this paper, the encounter types were introduced as an additional major risk that leads to ship collisions. By implementing BN in risk assessments, one could identify a set of major critical components that have a strong influence in the probability of occurrence of ship collisions. Thus, special attentions could be directed towards these critical components in order to improve the overall reliability of marine transportations system, to minimize the potential threats, and to prevent ship collisions from happening. 1. INTRODUCTION The marine traffic has been growing exponentially during the last few years due to the enhancement of marine transportations technologies.

Shipping by sea delivers around 90% of all world trade [1]. It is highly complex and it has a significant contribution to the world economic progress, because shipping by sea is a lot cheaper than shipping by air for any non-time sensitive materials. In high-traffic marine territories, such as Gulf of Finland (GoF) and Baltic Sea, the ship to ship collision is about 20% of the yearly reported marine accidents cases [2]. Other than ship to ship collisions, a ship can collide with bridges, quay, costal reefs, icebergs and random floating objects.

The well-known devastating accident of RMS Titanic colliding with icebergs in 1912 had resulted in 1514 loss of life. Although enhancement in technologies offers convenience and safety improvements, ship collisions are still happening. Exactly 100 years later after RMS Titanic incident in 2012, Costa Concordia, an Italian cruise ship, capsized and sank after striking an underwater obstruction and lost 32 lives. Other than loss of life, fuel leakage is also associated with ship collisions. Ship collisions involved a tanker that carries oil, gas or any hazardous chemical materials could lead to severe ecological disasters and a huge financial loss.

A number of efforts have been conducted to increase the safety of marine traffic in a congested marine territory. Kujala et. al. proposed AIS data utilization in the theoretical models to calculate the geometric ship collision probabilities [2]. Hänninen et al. proposed a BN of ship collision focusing of ship loss of control [3]. On another paper, Hänninen et al. investigated the role of human factors on ship collision probability in GoF [4].This paper proposed a Bayesian Network (BN) approach to assess the risk associated with ship collisions taking account into three major associated risk factors: human failures, technical failures and types of encounter.

BN was employed to study different cases related to common failures that cause ship to collide. Sensitivity analysis was further utilized to detect the critical components of ship collisions. The objective of this paper is to improve safety of marine transportation system in a highly congested marine territory by reducing the occurrence of ship collisions. The rest of this paper is organized as follows: Section 2 explains the data mining and methodology behind the proposed BN model. Case studies of different accidents scenarios are presented in Section 3.

Section 4 discusses sensitivity analysis, contribution of this paper and the limitations. Finally, conclusions are deliberated in Section 5.NomenclatureABBREVIATIONSBN = Bayesian Network GOF = Gulf of FinlandOOW = Officer of Watch2. BAYESIAN NETWORKS2.1 RISK ASSESSMENTS FOR MODEL BUILDINGHuman failure, mechanical failure and encounter types are considered as main cause of collision. Mechanical failure and human failure affects on control of ship and combination of control and encounter type can shape the whole probability of marine collision.

Human failure consists of captain mistakes such as miscommunication and loss of steering control and Officer of Watch (OOW) mistakes like Fatigue and misjudgment. Mechanical failure consists of GPS, radio, radar, and engine failure and poor maintenance. Engine problem can easily affect on steering performance, because of marine steering systems. GPS and radar are detection systems of ship and radio is communication equipment. Both these kinds of failures impact on steering, communication and detection that can be defined as significant factors affect control of a ship.

Generally, encounter has three types: crossing, overtaking and head-on. 2.2 PROPOSED BAYESIAN NETWORKAs can be seen in the figure 1, Control as inner factor and Encounter as environmental factor have been considered in marine collision, simultaneously. Steering, communication and detection failures are result of mechanical failures or human mistakes. To be more exact, captain performance has direct impact on steering, communication and detection. OOW performance can cause communication and detection problem.

Engine can only affect on steering control, radio on communication and radar and GPS on ship situation detection and even the positions of other ships around it. Poor maintenance influences on all these equipments, so it has connections to all of them in mentioned figure.Fig. 1. Proposed network for ship collisionFigure 2 illustrates the probability of collisions in different encounter types, that crossing is the most encounter situation in marine collisions with 0.46 of probability of occurrence.Fig. 2.

Probabilities of different encounter typesOverall, with the probabilities of data shown in table 1, the probability of collision in this proposed Bayesian Network has been calculated 0.09. Table 1. probability of failure for effective variables in marine collisionVariablesProbability of FailureCollision-Encounter-Control0.189 [3]Crossing0.46 [3]Overtaking0.33 [3]Head on0.2 [3]Steering0.179168 [3]Communication1.2E-4[3]Detection0.02231 [3]Captain perf.-OOW perf.0.024 [3]Engine-Radio-Radar0.00102 [3]GPS0.

0029 [3]Maintenance0.2 [3]3. CASE STUDY3.1 CASE STUDY DESCRIPTIONS In the scenarios presented below, shows the different encounters between ships and the possibility of collision during encounter under extreme circumstances where one or more components of the ship fails. The serious consequences of ship collisions necessitate the development and identification of critical components of ships to minimize damage, reduce environmental pollution, and improve safety. For this project three scenarios have been considered for the ship collision.

In the first scenario the human failure is considered and how the probability of collision of two ships is affected during an encounter. For the second scenario mechanical failure is considered during which the steering becomes dysfunctional and its affect on the marine traffic safety. The third case is an extreme scenario where both the human failure and mechanical failure has occurred at the same time. In this scenario, maintenance of the equipment failed and its affect on navigation and communication system were observed as well.

Since crossing is the most frequent situation for marine traffic in a busy area, it is considered to be taking place during the first scenario. During crossing, the encounter of the two ships may occur. An encounter may result in either a collision or near miss. During encounter the human failure occurs and the Office of Watch (OOW) fails to perform the required duties to spot the second ship or notify the control room of the vessel about the second ship. In the second scenario a mechanical failure is considered.

In this scenario the overtaking of a ship by another vessel is considered. During overtaking another ship there is a probability that it will fail to keep a safe distance to avoid the encounter. When encounter happens the probability of collision increases. During this scenario it is assumed that the steering mechanism fails. The ship is not in control but might encounter another ship ahead of it. In the third scenario, an extreme case is considered. In this scenario the most important human function performed by the Captain of the ship is considered.

The captain fails to perform duties or take wrong decisions. This combined with failure of equipment (navigation and communication) due to lack of maintenance results in a substantial increase in the probability of collision.3.2 RESULTS AND DISUCSSIONSThe figure Y below shows the Bayesian network for the first scenario in which the two ships are crossing each other and the Office of Watch (OOW) fails to perform the required duties to spot the second ship or notify the control room of the vessel about the second ship.

In this scenario when failure occurs in OOW the probability of collision increases to 15%In the second scenario as can be seen in the figure Y below of the Bayesian network. During overtaking the mechanical failure in steering occurs. In this scenario the ship is not in control and evasive maneuvers cannot be performed. This in turn, result in an increase in the probability of collision to 32%.In the figure Y below, the Bayesian network shows that such an extreme case results in the increase of probability of collision to 54%.

In this scenario, lack of proper maintenance of equipment and ship that resulted in the failure of navigation and communication systems and failure of the captain increased the probability of collision significantly. These scenarios shows the importance of maintaining the ship equipment, and proper training of the personnel. These ships are extremely large and heavy which are difficult to maneuver quickly to avoid collision. The Bayesian network scenarios just show 3 scenarios but there can be many different scenarios that can result in a collision.

The Bayesian network also helps in identifying the critical areas of the ship most important to maintain traffic safety in the ship. Maintaining safety is important in marine traffic due to the volume of goods transported and the impact on marine life due to chemical and oil spill that might result from collision.4. DUSCUSSIONS4.1 SENSITIVITY ANALYSISSensitivity analysis was implemented in this paper to study the contribution of each variable on the probability of having ships collision. Each prior node connected to the collision node was set to be failed individually (observed as true in BN, where other nodes were held as no observations).

Their individual effects on the probability of ship collision were further measured. Figure X shows a bar chart comparison of the impact of each node in the probability of ship collision Figure X: Impact of every prior variable on ship collisionBy doing sensitivity analysis, the critical components of failure scenarios could be easily identified. According to this figure X, the critical component, which is the component that contributes the most to the probability of ship collision, is the control system.

In another words, the probability of ship collision is 34% if the control system was observed to be a failure. To improve the control system, focus has to be directed towards the steering system. Under the control system node, steering system contributes 23% to ship collision if the failure of steering system was observed to be true. Although other prior nodes could be deemed to have a low impact on the occurrence of ship collision, any combination of prior nodes failures might resulted in a high probability of ship collision as illustrated before in the Extreme case study in Section 3. 4.2 CONTRIBUTIONSA collision is an event happened because at least one moving objects violently striking against another.

Before collision happens, the two objects will have a very close unavoidable encounter. Thus, the contribution of the proposed BN is taking into account the ship’s encounter types, while most of the ship collision BN only considered human failures and technical failures. As mentioned in Section 2, there are three types of encounter a ship could face: head on, overtaking, crossings, regardless the types of the second objects. The other objects could be at idle state, such as costal reefs, icebergs, bridges and an idle ship, or moving, such as floating objects or another moving ship.

Without limiting the second object to be a ship, the proposed BN have a wider range of application. It can be applied in the risk assessment for ship collisions with floating objects, bridge or quay. Although encounter types had been taken into considerations, the proposed BN could be improved by taking account into weather conditions. It can be guaranteed that bad weather conditions such as storms could significantly affect the judgments and performance of the ship captain or OOW at the lighthouse.

Detailed mechanical failures and the performance of ship crews were not considered in the proposed BN. In the case of ship to ship collisions, failure scenarios for the second ships during encounter are not considered in the proposed BN model. 5. CONCLUSIONSFollowing the increased cases of ship collisions on high traffic marine territories, there have been great concerns on the need to reduce the occurrence of these risks by assessing and detecting the possibility of their occurrences. The causes of ship collisions have been investigated and found to include majorly human failure, technical failures, and environmental conditions.

Among these, technical and human failures form the major causes of ship collisions in high traffic marine territories. This project has categorically analyzed these factors by making relevant recommendations on how to assess and minimize such collisions. The BN, Bayesian Network, analysis has been used to analyze the possibility for the occurrence of risks caused by the aforementioned factors in the Gulf of Finland. The technical factors found to bear high possibility to the occurrence of such collisions included: poor maintenance of the technical components of the ship that may lead to technical malfunctions, as well as technical failure of the communication systems of the ship.

Human factors closely associated with the possibility for the occurrence of collisions in the high seas included: poor communication strategies by the OOW, fatigued officers in charge of operating the ship, and hiring of incompetent officers onboard (Li, Yang, Cao, & Li, 2006). Risk assessment helps in prior identification of the various sources of failure that have the highest possibility of resulting into ship collisions. The process of risk assessment is a comprehensive procedure that encompasses detailed analysis of the various components of the ship as well as the human and environmental factors.

Risk assessment and prediction in high marine territories could help in minimizing the occurrence of such collisions hence help save millions of lives put in danger by such incidents. The BN model used in the assessment process in this paper integrates navigation practices in the high marine territories, marine navigation knowledge base and the use of GPS and GIS techniques in assessing and mitigating collision risks. Through the network assessment, it has been clearly shown how one can use the procedure to identify the possibility for the occurrence of such risks hence helping in their prevention upfront. 6. AcknowledgmentsI would like to acknowledge the sincere contributions of the following, who saw the success of this project.

To begin with, I would like to send my sincere gratitude to my colleagues with whom we worked hand in hand to ensure the success of the entire project. I also acknowledge the support of my lecturers, who equipped us with the necessary skills, knowledge and technical capability to successfully complete the project. References1. “15 Years of Shipping Accidents: A review of WWF Southampton Solent University”, N. Butt, D. Johnson, N. Pryce-Roberts, N. Vigar2. Hänninen, Maria, P. Kujala, and J. Ylitalo.”Estimating the number of tanker collision in the Gulf of Finland in 2015.

” International Journal on Marine Navigation and Safety of Sea Transportation. Vol 6, no 3(2012) 3. Hänninen, Maria, and P. Kujala. "Influences of variables on ship collision probability in a Bayesian belief network model." Reliability Engineering & System Safety 102 (2012): 27-40.4. Kujala, Pentti, Maria Hänninen, Tommi Arola, and Jutta Ylitalo. "Analysis of the marine traffic safety in the Gulf of Finland." Reliability Engineering & System Safety 94, no. 8 (2009): 1349-1357.5. Li, L.-N., Yang, S.-H., Cao, B.-G., & Li, Z.-F. (2006).

“A summary of studies on the automation of ship collision avoidance intelligence,” (in Chinese). Journal of Jimei University, China , 11 (2), 188-192.