21.3.3.1 Probability of Powered Contact: Fig. 21.2
A powered contact generally occurs when a vessel is deviating from course and heading towards a wind farm, and this incorrect action is not corrected in time. Therefore, in order to calculate the probability the powered contact, it is recommended that the user follow the geometric-causation probability model as described in Sect.
21.2.1.1.1.
Essentially, a user must first calculate the geometric probability of accident—i.e.—the probability that a vessel is
not following its course and/or is offset from the vessel way (steps 5.1.2 and 5.1.3 in Fig.
21.2). This is generally done by looking at AIS data, and determining how many times vessels deviate from their route. The AIS data can be used to generate a probability distribution, which indicates how often vessels deviate or are offset from the median line of a vessel way. If AIS data is not available, it is reasonable to assume that vessel traffic is normally distributed along the width of the vessel-way.
Calculating the probability of a vessel
not performing a corrective action, while deviating from its course (step 5.1.4 in Fig.
21.2), is slightly more challenging—particularly because this depends on both human and technical factors. A typical approach is to use a ‘causation probability’ value from literature; a more sophisticated and thorough approach is use to use risk assessment methods like Fault Trees, Event Trees and Bayesian Networks to estimate the causation probability (Friis-Hansen
2008).
Once a user obtains a geometric probability of contact for all vessels over a given time period, and an appropriate causation probability, he or she can then multiply the two values to obtain a total probability of contact for that type of vessel over the given time period.
4
A user of the framework can also multiply this total probability value by another given geometric equation to calculate the probability of actually hitting a wind turbine rather than just sailing into a wind farm (step 5.1.5 in Fig.
21.2). Equations for this purpose are also available in literature (Ellis et al.
2008b), and generally take into account various factors such as ship length, distance between turbines, and diameters of the turbine towers.
To calculate the probability of a powered contact, the researcher recommends using the model and equations developed by SSPA, as described by Ellis et al. (
2008b).
21.3.3.2 Probability of Drifting Contact: Fig. 21.3
Once the probability of powered contact has been calculated, the next step is to calculate the probability of drifting contact (Fig.
21.3). A vessel is set to be ‘drifting’ when it suffers from loss of engine power. Therefore, in order to calculate the probability of a drifting contact event, a user must first calculate the probability of a vessel-type facing an engine breakdown (5.1.10 in Fig.
21.3). This data is generally available from maritime and ship records.
The next important parameter to calculate during a drifting-contact probability assessment is the probability that a vessel will actually drift
towards an OWF (5.1.11 in Fig.
21.3). This probability is calculated by looking at wind and wave condition data, and seeing how often the wind or current flows in a direction that can carry ships towards an OWF.
Next, the user must calculate the time for which a vessel will drift—and whether or not this time is enough for a contact accident to occur. In order to do so, the users must consider the width of the traffic distribution on the route, and calculate the time it would take different vessels to reach the wind farm boundary based on their position on the route, and the wind and wave conditions. In the same step, one must also calculate the probability that emergency measures
5 will be unsuccessful or omitted in a given time period (5.1.12 in Fig.
21.3). A successful emergency measure—in time—will ensure that a certain proportion of the vessel traffic will not reach the wind farm to cause a contact incident.
Having obtained all the mentioned parameters, a user can multiply them to calculate the overall probability of a drifting contact for one vessel along an entire vessel way, over a given period of time. Similar to the powered contact procedure, the user can further multiply this product by an equation to obtain the probability of actually drifting and hitting a wind turbine, rather than just drifting into a wind farm area (5.1.13 in Fig.
21.3).
It is also important to predict the potential speed of the vessel in drift. This is vital in order to assess the consequences later. Equations from literature (Kleissen
2006; Christensen
2007; Ellis et al.
2008b) can be used to calculate the drift speed, which depends primarily on current and wind conditions.
For the drifting model, too, the researcher recommends the use of SSPA’s comprehensive model and equations, as detailed in Ellis et al. (
2008b).
To calculate the overall probability of a type of vessel suffering a contact event, a user should sum the probability of both, powered, and drifting contact events for that vessel-type.
21.3.3.3 Consequences of Contact: Fig. 21.4
After calculating the probability of contact, the next step is to calculate the consequences, using Fig.
21.4. To calculate the consequences to a given type of vessel, one should use a ‘standard’ reference vessel, as explained earlier in this section. A user should also consider the structural properties of a wind turbine (5.1.17 in Fig.
21.4).
The first step (5.1.18 in Fig.
21.4) whilst calculating the consequences is to assess a range of possible energy dissipation values. The energy dissipation values are derived from the range kinetic energy of the vessel when it collides with the turbine, and the energy that the turbine and vessel absorb. Of course, this dissipated energy depends primarily on the velocity and mass of the vessels. Since the speed and mass are assumed to be constant values for all vessels of a specific type, the other parameters that can influence the energy dissipation have to be considered—e.g.—the angle of collision between the ship and the turbine. Therefore, a user must use an equation that relates the mass, velocity
and the angle of collision, as well as the location of collision, to the kinetic energy. Such equations have been developed by Pedersen and Zhang (
1998), and by Pedersen (
2002,
2010,
2013,
2014).
Once several kinetic energy values have been obtained, the next step is to perform a detailed numerical finite element analysis (FEA) for the worst case scenario (5.1.19 in Fig.
21.4)—i.e.—the scenario with the highest energy dissipation. Although there are simpler methods to calculate consequences—semi-analytical, probabilistic, and empirical equations being quite common—such methods are geared more towards ship-ship collisions. Without using FEA, it is hard to capture the complexity of a ship-turbine collision, and thus assess the consequences in sufficient detail. A turbine may have many different forces acting upon it, from the aerodynamics of the blades, to the structural integrity of the soil. Therefore, although FEA is more resource intensive, it is the method recommended by the current researcher. Moreover, substantial contemporary literature shows increasing progress when it comes to FEA analysis of ship-turbine contacts. In particular, Biehl and Lehmann (
2006) have done significant work on describing numerical methods for ship-turbine collision analysis. Similarly, Dai et al. (
2013) describe a procedure to calculate the damage to a turbine, using FEA, in a scenario where a support vessel collides with a wind turbine. Most recently, Bela et al. (
2015) have used FEA to assess the crashworthiness of mono-piles.
It is interesting to note that all the above-mentioned papers use the FEA software LS-DYNA for their analysis. In fact, most of the FEA work on ship-turbine collision is done using the software LS-DYNA, as it has certain features (structural element types) which make it ideal for such analysis. Therefore, when it comes to this stage, the current researcher also recommends the use of LS-DYNA. In order to perform a numerical analysis, the user is required to create a 3D model for each type of vessel, and for each type of wind turbine. The models should incorporate primarily the structural properties, but it is also important to include the hydrodynamic and aerodynamic properties for accurate calculations of consequences. A contact event can then be simulated to assess the energy dissipation in more detail, and to understand the damage to both structures.
A numerical analysis obviously requires a high level of computational resources. The intensity of resources is partially reduced since the FEA is only performed for the worst case scenario for each vessel type, although it is recommended that a user perform it at several different points along the vessel, and at various locations around the turbine to ensure the validity of the results. Developing 3D models and meshing then appropriately consumes a lot of time; in order to minimize this, the author proposes that various coastal states maintain a database of standard meshed 3D models for all vessel types operating in their waters. This can greatly help to reduce the resources required for FE modelling.
A numerical FE analysis allows the user to assess the damage to both the ship and the turbine (5.1.20.1 and 5.1.20.2 in Fig.
21.4). The damage to the wind turbine can determine the state of the turbine after a contact event, and whether it will collapse or not. Such information can be used to evaluate how much economical loss will be incurred.
From the FE calculations, the damage to the ship can be generally visualised, and quantified in terms of a certain damage length, damage height and a specific penetration depth. These parameters in turn define the oil and cargo outflow from a ship, as well as the water inrush. The water inrush can then determine the stability of a ship, and how much time is available until capsize. Using all this information, one can quantify the consequences as described in Fig.
21.4. A similar procedure can be applied to damage incurred by the wind turbine.
For oil outflow, water in rush and stability calculations (5.1.21 in Fig.
21.4), there are equations and models present in literature (van de Wiel and van Dorp
2011; Wang et al.
2002; Li et al.
2012; Goerlandt and Montewka
2014) that can allow a user to model these events; such equations directly relate the extent of damage to the aforementioned parameters. The current researcher, however, uses the software HECSALV from Herbert-ABS to model these events. HECSALV is a rapid assessment tool, developed to perform rapid assessment of vessels in distress. It incorporates widely-used equations to assess several parameters which indicate the state of damage to a vessel. If the damage extent to a ship is known, HECSALV can provide oil outflow estimates, water inrush estimates, and time to capsize estimates, amongst other factors.
The output from HECSALV can be used to estimate evacuation and emergency response times, and, when combined with Data-sets 2A and 2B from Fig.
21.1 (which can indicate the potential level of emergency response in the area), one can estimate the consequences in terms such as loss of lives, amount of total oil spill, and cumulative damage to ship. Methods described by the IMO (
2008) can also be used to calculate the potential number of injuries and fatalities. The spreading of oil can be further modelled using tools like Seatrack Web—the official HELCOM oil drift forecasting system.
The consequences of an accident can also be quantified (5.1.22 in Fig.
21.4) as monetary figures, using ‘per-unit currency’ values given in literature for different types of losses and damages—e.g.—each tonne of oil spill costs approx. $60,000 in a given sea area (Christensen
2007). Such monetary values can be obtained for a variety of consequences—such as loss of one life, loss of a turbine, and damage to environment. Further expert judgements can also be used to augment these monetary values.
Alternatively consequences can be quantified into various qualitative ‘levels’, although this approach is not recommended for the current framework.
Ideally, the process described by Fig.
21.4 should be repeated twice—once for drifting vessels, and once for powered vessels. The core difference between these two assessments would be speed of the standard colliding vessel—a vessel in drift is likely to have a lower collision speed with a turbine. After quantifying the consequences, it is recommended that
only the worst-case consequence values be documented for the next stage in the framework—but if the user wishes, they can mention the worst-case consequences for both drifting and powered contact events separately.