3.1 SOSM (Sea and Oil Slick) 3D Simulation Model

A model suite, comprising of a coastal hydrodynamic circulation, a wave generation model, and a Lagrangian (tracer) model for the transport and physicochemical evolution of an oil slick, is synthesized by the Aristotle University of Thessaloniki [16]. It follows a model that was developed in the context of an EU DG XI project, in cooperation with the UGMM Group of the Belgian Ministry of Public Health and Environment [17]. Following that, the Oil Spill Model (OSM), based on the PAR-CEL model, was coupled with the hydrodynamic Princeton Ocean Model (POM), and applied on the Navarino bay, in SW Greece, adjacent to the Ionian Sea, to monitor the pollution impact on the benthic structure and the related fisheries [18]. Parallel to that, the Poseidon pollutants transport model (PPTM) was developed to provide real-time data and forecasts for marine environmental conditions and protect the Greek seas from oil spills, as an operational management tool, which consists of a 3D floating pollutant prediction model coupled with a weather, a hydrodynamic, and a wave model [19]. PPTM was applied in four areas of strategic interest in the Greek seas.

The two basic components of the model, that is, the hydrodynamic and the oil transport, are briefly presented here, followed by the applications under the prevailing circulation conditions in a coastal basin of Greece in the NW Aegean Sea, where the second largest port of the country is located in the town of Thessaloniki, with dense maritime traffic [16].

### 3.1.1 Model construction

The model is composed of two parts: the hydrodynamic part, comprising of the circulation and the wind wave subparts, and the oil slick transport part. The first is aimed at the description of the 3D nearly horizontal flow velocity vector field due to wind and/or tide. It is written with respect to the free surface ζ(x,y,t) and the depth-varying velocity components u(x,y,z,t), v(x,y,z,t)

$$\frac{\partial \zeta}{\partial t} + \frac{\partial}{\partial \mathbf{x}} \left\{ udz + \frac{\partial}{\partial \mathbf{y}} \right\} \nu dz = \mathbf{0} \tag{1}$$

$$\frac{Du}{dt} = -\mathbf{g}\frac{\partial \zeta}{\partial \mathbf{x}} + N\_h \left(\frac{\partial^2 u}{\partial \mathbf{x}^2} + \frac{\partial^2 u}{\partial y^2}\right) + N\_v \frac{\partial^2 u}{\partial \mathbf{z}^2} = \mathbf{0} \tag{2}$$

$$\frac{Dv}{dt} = -\mathbf{g}\frac{\partial \zeta}{\partial \mathbf{y}} + N\_h \left(\frac{\partial^2 v}{\partial \mathbf{x}^2} + \frac{\partial^2 v}{\partial \mathbf{y}^2}\right) + N\_v \frac{\partial^2 v}{\partial \mathbf{z}^2} = \mathbf{0} \tag{3}$$

The horizontal eddy viscosity-diffusivity Nh is deduced from the 4/3 law with a characteristic length related to the used Dx, and the vertical eddy viscositydiffusivity Nv is controlled by the surface friction velocity, the mixing length, the wave height and period. The model is solved numerically by finite differences, on a staggered horizontal Arakawa C grid (Figure 4), and a vertical discretization, following the σ-coordinates transform, using the technique of fractional steps (i.e., resolving explicitly for the horizontal advection and implicitly for the vertical momentum diffusion) [20].

The wind waves are resolved by an empirical model WACCAS [21, 22] based on the JONSWAP spectrum. The wave height, wave period, and wave direction on the center of the mass of the slick is used for the computation of the Stokes' drift and the vertical viscosity-diffusivity. The oil slick simulation and tracking are done in Lagrangian coordinates [23, 24], by replacing the oil mass by a big number (10<sup>4</sup> ) of particles or parcels, each one of which represents a group of oil droplets of similar size and composition. The oil spill simulation by a large number of passive (not

Figure 4. Arakawa C staggered grid configuration.

#### Oil Spill Dispersion Forecasting Models DOI: http://dx.doi.org/10.5772/intechopen.81764

interacting with the existing hydrodynamics) particles, transported in a Lagrangian frame of reference, is the most effective procedure, used in most contemporary models as: (a) it can describe oil spill geometries and shapes at subgrid level, (b) the model results are not distorted by numerical errors, like numerical diffusion, and (c) the fraction of the initially introduced particles (at the spill source) reaching the coast can be considered as the probability of pollution from a minor spill [25]. According to the tracer technique, it is theoretically possible to track the motion of all the "particles" (parcels) of an oil plume as they are transported by the sea. The density, the buoyancy velocity, the volatility, and other properties are statistically distributed to the particles. Their trajectories are tracked in space and time. Each one is advected according to the local (interpolated) water velocity and it is diffused by means of random walks related to the local diffusivity induced by currents and waves.

Evaporation and emulsification are processes quantified by means of commonly used functions relating the processes to the oil properties (to the fraction density, the temperature, the vapor pressure, the wind and water velocity, etc.). The Mackay approach is used [26, 27] for the computation of the evaporated volume fraction from each oil component. Emulsification is activated for oil fraction with densities and under wave height/length values beyond critical ones. The evaporation and emulsification processes result in: (a) loss of contained oil volume in the evaporated particles and (b) creation of a mixture of water in oil and formation of a floating mousse. Another important oil weathering process accounted in this model is beaching, which is better described by the duration of trapping on the coastal boundary of a beached particle as a function of the coastal morphology (from rocky to flat sandy beach). This process gives crucial information on the oil quantities retained on the coast and the consequent environmental effects.

The spatial and temporal distributions of values like oil concentration (or volume per unit area) are deduced from the number of particles and the contained volume inside each mesh of the discretization grid used for the hydrodynamic analysis. It is worth noting that the time step Dt, used for the integration of the hydrodynamic model, differs by orders of magnitude from the one used for the tracking of the oil slick.

#### 3.1.2 Operational application

The produced operational code was updated and adapted to the environment of a coastal basin of NW Aegean Sea, namely Thermaikos bay. The commercial port of Thessaloniki hosts among others a busy oil terminal. Although the terminal operates under the strict International Maritime Organization (IMO) and European Union (EU) regulations for oil terminal operational safety, the port area is at risk for a potential oil slick accident.

The transport and fate of the slick are closely related to the hydrodynamic conditions in the bay, the bay morphology, and the oil properties. The hydrodynamic circulation in the bay is experiencing mainly N-NW winds, whereas the circulation patterns are regulated by the bay bathymetry and the coastal topography. The bay configuration and bathymetry are shown in Figure 5, along with the initial oil spill location.

The model was tested for hypothetical scenarios of oil spill under NW and NE winds of 45 knots, to assess the risk of pollution in the coastline of the touristic areas of Chalkidiki (east) and Katerini (west), respectively. The oil is presumably spilled in the port of Thessaloniki (Figure 5), as a result of a hypothetical accidental spill during a tanker operation. An oil spill of 1000 parcels, representing a total volume of 10,000 tn, is released instantaneously on the sea surface. Total simulation run

Figure 5. Thermaikos bay configuration and bathymetry, and initial oil slick conditions.

was for 72 hrs (3 days), providing output results every 3 hrs. The evolution of the slick trajectories, for the first 24 hrs and for the two wind-generated circulation scenarios, is depicted in Figures 6 and 7, respectively, documenting the difference in the coastal impact of the slick for the two cases.

In the case of strong NW wind, the touristic coast opposite to the bay is damaged within some hours after the spillage, while in the case of the NE wind, the oil slick travels along the bay toward the open sea giving time to the relevant authorities for intervention, blocking, and final cleaning. The dispersion effect is also revealed showing that the oil slick diameter is moderately increasing during this time.

Figure 6. Evolution of the oil slick under strong NW wind.

Figure 7. Evolution of the oil slick under strong NE wind.

Figure 7 depicts more clearly the spreading of the oil spill and its statistical center of gravity movement (noted by red circles), as it is advected for many kilometers in Thermaikos bay. The oil slick dispersion after 3 hrs is shown in dark blue color, whereas the spreading of the plume at the end of a day (after 24 hrs) is in light blue. The spiral movement of the oil's statistical center of gravity in Figure 6, near the shore, may be attributed to the inertial currents frequently observed in the sea, and points out that the largest part of the oil reaches the shore (beaching) within the initial time steps (after 9 hrs). The model results show that a contingency emergency plan should be, in this case, activated in less than 12 hrs, in order to avoid the social, economic, and environmental damage in the area.

### 3.2 DIAVLOS-NASOS

The universities of Thessaloniki and Athens cooperated, in the framework of the project DIAVLOS [28–30], to produce a simulation model that could trace an oil spill in case of an accident in the Alexandroupolis Gulf, which is considered of high risk, after the construction of the Burgas-Alexandroupolis oil pipeline that has been signed by the Greek, Bulgarian, and Russian governments (on March 15, 2007), and is expected to carry huge oil quantities (35–50 Mtn/yr) that will be loaded to tankers and further transported by sea to international markets.
