Coastal Engineering 135 A numerical simulation of oil spill in Istanbul strait S. Can 1, S. Nishio 2 & M. Uchida 2 1 Maritime Faculty Istanbul Technical University, Turkey 2 Faculty of Maritime Sciences, Kobe University, Japan Abstract The Turkish strait is one of the most important routes of oil transportation. It connects the Black Sea and the Mediterranean, and it enables the transportation of oil produced in the north of Black Sea to Europe directly. The oil spill problem originating from tanker accidents is one of the major topics concerning the safe ship operation in Turkish straits. There have been large accidents occurring alongside a mega city, Istanbul, which has thirteen million inhabitants. A tanker accident in Istanbul strait can be a great disaster that directly affects life in the city. The present paper describes the numerical simulation of an oil spill in Istanbul strait enabling us to estimate the effect of an oil spill if a tanker accident occurs. Using the Star-CD code, a simulation was performed in the transient mode. Keywords: oil spill, CFD, simulation, Star-CD, tanker accident, safety ship operation. 1 Introduction A large amount of oil has been produced in the north of the Black Sea, and most of it is in Europe. The Turkish straits are one of the most important routes of oil transportation in the world. They connect the Black Sea to the Mediterranean, and enable transportation of oil from the Black Sea to Europe easily. The Turkish straits consist of two parts, and the northern half is called the Istanbul strait. It has lain beside a mega-city, Istanbul, where thirteen million people live. Disasters based on the tanker accidents have happened several times in this strait, and it has directly affected life in the city. One of the essential problems in analysing the tanker accidents and oil spill problem to consider the safe ship
136 Coastal Engineering operation in the Turkish straits. In this paper, the basic features of Istanbul strait are described, and a numerical simulation of oil spill originating from tanker accident is presented. The results of simulations for each different accident can be compared with the analyzed radar images. On the other hand, modeling of oil spills movement by remote sensing can be considered as the early stage of investigation [1]. However radar images for Nassia Accident could not be found. The main objective of the computer modeling is the prediction of the oil slicks behavior, which is the key element for oil pollution control in case of an accident. This provides the opportunity to take necessary immediate precautions. Once the model grids are developed, this study enables development of scenarios for different environmental conditions in Istanbul Strait (current speed and direction, boundary conditions, type and amount of spill etc.). The result of such a study can be used as a basis for a Regional Oil Spill Contingency Plan. 2 Istanbul strait 2.1 Geometrical feature and current in the strait Istanbul strait is a narrow and winding route. The total length of Istanbul strait is 32km, and the narrowest point is about 800m wide. Fig.1 shows the coastline of the Istanbul strait. The Black Sea lies on the north side of the strait, and the southern area is called the Marmara Sea. Istanbul lies on both sides of the strait, and the center of the Istanbul city is located at the south end of the strait. As the oil is produced in the northern area of Black Sea, full loaded tankers go south through the strait. RUMELIKAVAGI Figure 1: Coastline of Istanbul strait.
Coastal Engineering 137 Both sides of the strait are steep, and a complicated bottom configuration is observed. The bottom configuration contributes to making a complex flow field together with a high speed current. The current inside the strait is mostly based on the difference of water density between the Black Sea and the Mediterranean. The maximum current speed in the strait becomes over 2 kt (1 m/s). The following current acting on the full loaded tanker makes the maneuverability unsteady. As the main factor of the current is the difference of the water density, the counter current exists at the bottom. Then a shear flow would exist in the middle of the section, and there are possibilities that the counter flow and strong shear layer attacks the full loaded tanker. 2.2 Tanker accidents in Istanbul strait Disasters originating from tanker accidents occurred several times in the Istanbul strait. As the accidents occurred very close to life in the city, it requires a quick a response as possible. Table 1 shows one of the worst cases of tanker accidents, called the Nassia Accident. A full loaded tanker Nassia (100,000 DWT) collided with bulk carrier named Shipbroker. Approximately 20,000 tones of crude oil, a considerable part of Nassia s cargo - caused severe pollution and a fire, which lasted over 4 days [2]. Location Table 1: Tanker accident in Istanbul strait. NASSIA ACCIDENT Istanbul strait, Rumelikavagi Date 13th March 1994 Accident Oil spill Tanker Nassia collided with bulk carrier Shipbroker. Approximately 20,000 tones oil was spilled or burned. The accident occurred at a point called Rumelikavagi, where a sharp corner and the large bottom configuration exists which can be seen fig.1. The accident occurred very close to the city area, and fire threatened the life in the city considerably. 3 Numerical simulation Numerical simulation was carried out to simulate and estimate the behavior of the spilled oil of Nassia Accident. The geometrical data supplied by the Turkish Navy-Department of Navigation was used for the grid generation. Fig.2 shows the original data point of the bottom configuration. The geometric information was given at random points, and the bottom configuration was generated by interpolating those data. The original data consisted of three parts as shown in Fig.2, and the middle part was used for the numerical simulation of Nassia Accident. The number of cells, which are used for grid generation, is 16240.
138 Coastal Engineering Figure 2: Original geometrical data of Istanbul strait. The finite volume method was employed here to obtain the solution of the Navier-Stokes equation [3]. From the perspective of the physics modeling of the process the flow was assumed to be incompressible and turbulent. Due to the unsteady nature of processing, the simulation was performed in transient mode. The time marching method was the fully implicit method, and the solution algorithm was the PISO solver. STAR-CD, which is used as a CFD code, employs implicit methods to solve the algebraic finite-volume equations resulting from the discretisation practices. The advantages of such methods are as follows: In unsteady-flow calculations, the allowable size of computational time step is limited mainly by temporal accuracy considerations and not by the severe stability constraints characteristic of explicit approaches. In steady-state problems, implicit methods can be accelerated either by sacrificing temporal accuracy (which is usually no longer of interest) and using a larger or by the alternative means of entirely dispensing with the temporal derivatives and iterating to the steady state. STAR-CD currently incorporates three different implicit algorithms, namely: SIMPLE, PISO, SIMPISO. PISO is applicable to both transient and steady-state calculations and is particularly suitable for the former, for which it has been shown to be considerably more efficient than iterative methods [4]. As a differencing scheme MARS (Monotone Advection and Reconstruction Scheme) which is a multidimensional second-order accurate differencing scheme was used for the convection terms [4]. The oil and its transport characteristics were solved as scalar entities.
Coastal Engineering 139 Table 2: System and software. HARDWARE DELLWorkstation PSW 340 CPU Pentium 4 2.53 GHz Main Memory 2 GB HDD 160 GB SOFTWARE CFD Code Star-CD 3.15 Pre&Post-Processing ProAm INLET Figure 3: Computational grids and boundary conditions. 3.1 Numerical conditions The simulation was performed in transient mode utilizing the extended version of the PISO algorithm. The time step size was 1000 seconds and 43 time-steps are employed. All calculations were performed on the system and software, is shown in Table 2. Fig. 3 shows the computational grid for the present simulation. The coastline consists of vertical walls. The flat bottom is employed here, as the initial aim of present simulation is the surface flow. The non-slip condition was applied to the sidewalls, and the free-slip condition was used for the water surface. The symmetrical condition was applied to the bottom boundary by considering the limitation of resolution, and it means that present simulation simulates the quasi 2D-flow field.
140 Coastal Engineering As the one way current exists on the surface of Istanbul strait, the inlet condition was applied to the north section, and the free outlet was used for the southern end section. The wind effect could be a major drive force of spill oil behavior, but we concentrate on the current factor to know the basic characteristics of spilled oil motion. Then the wind effect was neglected here. Oil spill was given from small inlet at the sidewall. The inlet speed of oil, which is shown as a Velocity-Time table in Table 3, was 1/50 of main current inlet. The detail of oil spill condition in Nassia Accident is not clear, but present simulation enables us to know the basic feature of oil spill. Table 3: Oil inlet velocity and time. T(s) 0 3000 4000 36000 U(m/s) 0.01 0.01 0 0 Figure 4: Calculated oil distribution and velocity vectors at time step 1000.
Coastal Engineering 141 4 Results and discussions Transient calculation was carried out to obtain the behavior of spilled oil. The density ratio of oil/water was 0.87, and the inlet speed of oil was 1/50 of main current inlet. Fig. 4 shows the calculated spilled oil distribution and velocity field at initial stage. The oil flows in continuously at this stage as shown in Table 3, then the total amount oil in the calculation domain kept increasing. The spilled oil began to move and spread to the south along the coastline due to the constant main current on the surface. Velocity vectors show the complex flow field due to the geometric of coastline. A dead flow area is found at the left corner of the strait, but we cannot find the counter surface flow in the present simulation. (a) time-step 3000 (b) time-step 6000 Figure 5: Transient oil behavior obtained by numerical simulation.
142 Coastal Engineering (c) time-step 9000 (d) time-step 18000 Figure 5: Continued. Fig. 5 shows the transient spilled oil behavior. Fig. 5(a) show the oil distribution at time-step 3000, and Fig.5 (b), (c) shows the following oil behavior at interval 3000 and (d) shows at 18000. The spilled oil continuously moved to the south, and the front end of the oil began to separate from the coastline at time-step 3000. The spread oil arrived at the southern coastline, and wide area of coastline will was covered with spilled oil. The coastline, where the oil once arrived, has large damage of pollution. Then calculated results shows that the quick response for the preventing of pollution will be required in this case. Some part of the spilled oil flowed out to the south, but large amount of oil stayed in the left corner of the strait. As the oil was burned in the real case,
Coastal Engineering 143 quite large pollution and threat on city life could happen with this kind of disaster. The spilled oil covers the whole width of the strait. It means that all traffic through the route was suspended. Fig. 6 shows the oil distribution and the velocity field at time-step 36000. It shows the final stage of present simulation. The oil spread seems to banish from the strait. As the main current in the strait is strong in this area, most of the oil flows down south. Then the thick oil area soon disappears from the local area when the oil spill was stopped. On the other hand, the oil flow, which flows down to the south, keeps attack to coastlines. It means that quite a large area will be polluted. Then quick countermeasure for the oil spread will be required in this case, and the complex velocity distribution, shown in Fig.6, should be considered when we build up the prevention plan. Figure 6: Calculated oil distribution and velocity vectors at time step 36000. 5 Conclusion The oil spill which happened in the Istanbul strait was investigated by a numerical simulation. A tanker accident occurring at Rumelikavagi was
144 Coastal Engineering considered, and transient spilled oil behavior was calculated. The near surface flow was calculated by combining quasi-2d-flow calculation and one-way inlet condition. The results obtained show that a tanker accident happening inside the Istanbul strait could be a considerable threat to life in the city, and a quick response to prevent the oil spill will be required. To define a successful response to a major oil spill in the Istanbul strait the critical success factor approach developed by John R. Harrald in work with the U.S Coast Guard and other U.S Spill response experts [5] can be used to emergency management,. This work is supported by JICA Project on Improvement of Maritime Education in the Republic of Turkey, and the authors appreciate their support. References [1] Simulation of Tidal Current Components and Oil Spills Spreading from Radarsat, International Institute For Aerospace Survey and Earth Science (ITC), Division of Applied Geomorphological Surveys, http://www.gisdevelopment.net/aplication/nrm/ocean/ocem001pf.htm [2] Istikbal,C., Oil Transportation through the Turkish Straits: A Great Challenge to Safety, Annual Meeting of Mediterranean Oil Industries Group, Istanbul, June, 2000. http://www.turkishpilots.org/ [3] Versteeg, H.K., Malalasekera,W., An introduction to computational fluid dynamics- The finite volume method, Addison Wesley Longman Ltd : London, pp.168-190,1998. [4] Star-CD Methodology, Ver.3.10, Computational Dynamics Ltd, 1999. [5] Harrald, J.R., Critical Success Factors for Responding to a large oil spill in the Bosporus, Proc. Of the 2 nd Int. Conf. On Oil Spills in The Mediterranean and Black Sea Regions(MEDOSC), Boğaziçi Unv.,Istanbul-Turkey&, Unv. Of Manchester,UK, Istanbul, pp. 79-88, 2000.