Advanced Computer Simulation Model for Ship Collision Frequency Estimation

Sudari brodova pripadaju visoko rizičnim incidentnim događajima u pomorskom prometu koji mogu izazvati značajne ljudske, ekonomske i ekološke gubitke. Očekivani porast pomorskog prometa na globalnoj razini ne ide u prilog smanjenju ovog rizika, zbog čega je u posljednjem desetljeću zabilježen i porast interesa spram istraživanja u području procjene rizika sudara u pomorskom prometu. Kako je rizik ovisan o vjerojatnosti pojave incidentnog događaja i očekivane (troškovne) posljedice tog događaja, procjena vjerojatnosti i učestalosti nesreća u pomorskom prometu izrazito je važna. Analizom dosadašnjih istraživanja uočeno je da postojeći analitički modeli za procjenu učestalosti sudara brodova nisu prikladni kod modela slučajno distribuirane plovidbe brodova i za promjene smjera protezanja plovidbenog puta. Nadalje, postojeći simulacijski modeli i modeli temeljeni na obradi AIS podataka pretežno su razvijeni za specifična plovidbena područja, te izravno ovise o realnim podacima praćenja brodova. Time je otežana primjena nad ostalim plovidbenim područjima, kao i testiranje hipotetskih scenarija. U ovom istraživanju razvijen je novi računalni simulacijski model za procjenu učestalosti sudara brodova koji se sastoji od dva zasebna modula, odnosno modela za procjenu učestalosti sudara brodova kod modela slučajno distribuirane plovidbe brodova i modela za procjenu učestalosti sudara brodova kod promjene smjera protezanja plovidbenog puta. Razvijeni model integriran je u računalnu aplikaciju s korisničkim sučeljem i 2D vizualizacijom simuliranih plovidbenih situacija te koristi novu modelsku logiku i pristup za simuliranje spomenutih plovidbenih situacija. Model se može okarakterizirati kao mikrosimulacijski model stohastičke prirode temeljen na simulaciji diskretnih događaja. Validacija razvijenog modela izvršena je uz pomoć postojećih analitičko-numeričkih metoda i programskog paketa IWRAP (engl. IALA Waterway Risk Assessment Program). Također su istražene i unaprijeđene postojeće analitičko-numeričke metode za procjenu učestalosti sudara brodova korištenjem razvijenog računalnog simulacijskog modela. Točnije, rezultati računalnog simulacijskog modela za model slučajno distribuirane plovidbe brodova validirani su pomoću postojećeg matematičkog modela za procjenu učestalosti sudara brodova kod križanja kursova koji je modificiran i prilagođen za opis modela slučajno distribuirane plovidbe brodova. Prethodna istraživanja unaprijeđena su i proširena novim plovidbenim situacijama čime je omogućena potpunija procjena učestalosti sudara brodova u analiziranim plovidbenim područjima. Razvijeni računalni simulacijski model primjenjiv je u različitim plovidbenim područjima uz upotrebu realnih ili pretpostavljenih podataka o prometu brodova.Ship collisions are among the high-risk maritime incidents that can cause significant human, economic, and environmental losses. The expected increase in maritime traffic on a global scale is not conducive to reducing this risk, which is why interest in research on maritime collision risk assessment has increased over the past decade. As the risk depends on the probability of an incident and the expected (cost) consequences of the event, the assessment of the probability and frequency of maritime accidents is extremely important. As an introduction, this research explains the basic concepts of ship collision risk assessment and addresses the importance of risk assessment by evaluating the frequency of ship collisions.. Quantitative models are highlighted as the leading method for estimating the frequency of ship collisions, and the classification and review of existing quantitative models used for this purpose is presented along with a description and analysis of the basic model properties of each model. In addition, possible approaches to modeling and simulation of transportation systems are discussed, and a general categorization of models is provided, focusing on computer simulations and an object-oriented approach to modeling, which represents the basic methodological concept used in this dissertation. Based on the conducted research, it was noted that existing analytical models for ship collision frequency estimation are not suitable for random sailing directions and bend of waterway. Furthermore, the existing simulation models and AIS data-processing based models are predominantly developed for specific navigation areas and are directly dependent on real ship tracking data. This makes it difficult to apply to other navigation areas, as well as to test hypothetical scenarios. In accordance with the set goal and hypotheses in this research, a new computer simulation model for ship collision frequency estimation is developed, which consists of two separate modules, namely the ship collision frequency estimation model for the random sailing direction scenario and ship collision frequency estimation model for bend of waterway scenario. The developed model is integrated into a computer application with user interface and 2D visualization of the simulated navigation situations, uses a new modeling approach to simulate the addressed navigation situations, and can be characterized as a microsimulation model of stochastic nature based on discrete event simulation. The developed model is applicable in different navigation areas with the use of real or assumed ship traffic data. Additionally, existing analytical-numerical methods for estimating the frequency of ship collisions using a developed computer simulation model are investigated and improved. More specifically, ship collision frequency estimation model for the random sailing direction scenario is developed as a first module. The importance of considering the aforementioned navigation scenario is emphasized, and detailed model properties based on an object-oriented modeling approach are presented. The model is tested and validated by running simulations for multiple hypothetical examples of the random sailing direction scenario and by comparing the results with analytical model used for the ship collision frequency estimation in the crossing scenario. A strong linear correlation is found between the developed computer simulation model for ship collision frequency estimation in the random sailing direction scenario and the aforementioned analytical model. Using the existing analytical model for ship collision frequency estimation in the crossing scenario and regression analysis, a new modified analytical model suitable for estimating the frequency of ship collisions in the random sailing direction scenario is derived. As a second module, ship collision frequency estimation model for bend of waterway scenario is developed. An explanation of the addressed navigation scenario and the detailed modeling features is given. The model uses an analogous object-oriented modeling approach as the first module. The model is tested and validated by running simulations for multiple hypothetical examples of the bend of waterway scenario and comparing the results with IWRAP (IALA Waterway Risk Assessment Program ) software. The differences in model properties between the developed computer simulation model and the IWRAP software are discussed and the analysis of the results is presented. Validation of the developed computer simulation model is also performed on a case study, i.e., a realistic example of a band of waterway scenario, by comparing and analyzing the results of the developed computer simulation model with the results of the IWRAP software