A probabilistic approach to determine design loads for collision between an offshore supply vessel and offshore installations
Introduction
Offshore installations require regular supplies of food, equipment, chemicals and other logistics support from the offshore supply vessel (OSV) depending on the supply requirement. While in service, these vessels pose a high collision risk to installations, either due to accidental impact during approaching or manoeuvring, or due to operational impact during transfer or berthing operations. Fig. 1 shows an example of a collision between a supply vessel and an offshore installation. Operational impact is usually characterised by frequent collisions associated with minor consequences. Fig. 2 presents the operating circumstances that supply vessels were engaged on when a collision with an installation on the United Kingdom Continental Shelf (UKCS). It can be seen that ‘cargo transfer’ operations dominate total collision accidents, although ‘approaching installation’ also marked highly, perhaps also being the precursor to cargo operations.
Among the main causes of collisions are human error, engine and equipment failure, and harsh environmental conditions, with the dynamic and unpredictable nature of weather conditions, which increase collision risk, representing the main challenge to offshore logistics operations. A retrospective of accident databases reveals that ship-offshore installation collisions happen frequently (HSE, 2000). However, they show a decreasing trend due to technological development, the enforcement of strict operational rules and regulations and improved navigator training and skills, among other factors. Despite such efforts at prevention, collisions continue to occur, with many accidents involving minor collisions going unnoticed or unreported. In addition, many ships have yet to adopt the new rules and regulations for safe maritime operations.
The design of offshore structures regarding ship collision is based on the accidental limit state (ALS), which accounts for human safety, asset losses and damage to installations or the environment (Paik, 2018; Paik and Thayamballi, 2007). The quantitative risk assessment (QRA) approach is an efficient method for evaluating collision risk and has been successfully applied for ship-ship (Youssef et al., 2016), ship-offshore wind turbine (Dai et al., 2013; Ellis et al., 2008) and ship-installation collisions (ABS, 2013; Mujeeb-Ahmed et al., 2018; Oil & Gas UK, 2010). Fig. 3 shows a general picture of the QRA framework for ship-installation collisions; the portion inside the red dotted lines indicates the main focus of the present study, i.e., the hazard identification of various collision-affecting parameters and the probabilistic selection of collision scenarios used to analyse external collision dynamics.
In ALS, a ship-installation collision scenario can be classified into external dynamics and internal mechanics (Pedersen and Zhang, 1999), and the coupled method estimates collision forces and structural damage with a high degree of precision (Tabri, 2012; Yu et al., 2018, 2016). While the external dynamics includes a global analysis of ship motions and a calculation of initial kinetic energy imparted to the installation, the internal mechanics involves structural strength analysis. Determining the initial kinetic energy of the supply vessel is important for calculating the energy levels in the jacket structure. During a collision only part of the kinetic energy is dissipated as strain energy for both the installation and vessel. The elastic energy accumulated in the jacket structure allows the vessel to bounce back due to the elastic property of the steel structure. Table 1 shows the accident data for some recent OSV collisions (Kvitrud, 2011).
The studies on the potential consequence of collisions that use numerical or model tests depend on input loading scenarios (Zhang and Pedersen, 2017). In general, two methods are visible in the literature: deterministic and probabilistic. The deterministic method consists of selecting a few unfavourable load scenarios based on historical data, expert judgements or assumptions. Often, the chosen collision scenario represents the worst-case situation. Most researchers have followed the deterministic approach (Emami Azadi, 2011; Moulas et al., 2017), but that approach is often ineffective at generalising the collision scenarios for other installations, being based limited historical data that may be either too conservative or out of date considering the large number of parameters affecting the collision. In this approach, both past collision accidents and expert judgements form the basis for the estimation of collision design loads. Various societies and regulatory committees have recommended specific collision load parameters for design (API, 2014; DNV GL, 2017a; HSE, 2000; Lloyd's Register, 2014; NORSOK, 2004); for instance, a 2 m/s collision velocity for a 5000 tonne displacement supply vessel (with 0.5 m/s and 1000 tonnes for the Gulf of Mexico). The NORSOK standard (2007) defines the limits of the vertical impact zone to be −10 m and 13 m at the lowest and highest astronomical tide, respectively. However, these limits are a function of the draft and vertical motions of the colliding vessel. Modern supply vessels are built with displacements of more than 10,000 tonnes and designed for sailing velocities greater than 2 m/s for installations situated in deep water or ultra-deep water, considering economic transportation and navigation in rough seas. Therefore, advancement in the design of the supply vessel demands a revision of the current design values for installations (Storheim and Amdahl, 2014).
In contrast, the probabilistic approach selects prospective collision scenarios by considering each random variable in the form of a probability density function. This method has been successfully applied to various offshore and marine industry accidents (Hughes et al., 2010), for instance, ship-ship collisions (Brown, 2002; Faisal et al., 2017; Kim et al., 2015; Ko et al., 2018; Youssef et al., 2017, 2014), grounding (Youssef and Paik, 2018), sloshing (Paik et al., 2015), corrosion (Mohd et al., 2014), dropped objects (Kawsar et al., 2015), ship hull loads (Chowdhury, 2007; Garrè and Rizzuto, 2012; Ivanov, 2009; Ivanov et al., 2011) and riser loads and mooring lines (Cabrera-Miranda et al., 2017; Cabrera-Miranda and Paik, 2017). In this approach, a historical database of collision load parameters becomes the basis for quantitative risk assessment. Indeed, much research can be found on the probabilistic impact scenarios of collision damage, for ship-ship collisions, that uses historical collision data to define the collision load parameters (Brown and Chen, 2002; Goerlandt et al., 2012; Lützen, 2001; Sun et al., 2017; Tagg et al., 2002). An examination of accident databases, however, shows a lack of quantitative information on parameters. Although the HSE database (2001) provides a reasonable quantitative description of accidents, some load parameters, such as the collision velocity and the angle of the colliding vessel, are not to be found. Moreover, no single database provides a full quantitative description for all load parameters. Therefore, the lack of sufficient historical collision data on loads renders this probabilistic approach difficult to apply to cases of ship-installation collisions. In addition, collisions are highly uncertain and random, thus requiring cumbersome and time-consuming numerical computations to analyse all of the possible collision scenarios. Those limitations make the probabilistic approach more desirable.
In this context, the aim of this study is to develop a probabilistic model to determine the collision scenarios for ship-installation collisions. Using the probability density functions of input parameters, a set of credible collision scenarios is generated by using statistical sampling techniques. Numerical computations are performed, and the best-fit probability distribution and the exceedance curves of load parameters are plotted, which can be used to determine the design collision load.
Section snippets
Framework of the proposed procedure
Fig. 4 presents a procedure of the probabilistic assessment of the collision design load followed in this study. First, we define the collision event using offshore jacket installation topology, OSV characteristics, vessel operational conditions and site-specific metocean data. Then, we identify key input parameters affecting the collision, namely environmental (such as wind, wave and current) and operational parameters (such as vessel speed and impact location). The probabilistic analysis is
Considered OSV and offshore platform
The viability of the proposed procedure is demonstrated by using a hypothetical model of an OSV approaching a four-legged offshore jacket platform (see Fig. 5); the particulars are given in Table 2. For simplicity, the considered vessel does not have the DP (dynamic positioning) system and the installation is assumed to be unmanned with no collision avoiding measures beyond those that include a standby vessel, a radar alarming system and a vessel traffic system.
Parameters affecting collisions
Various parameters affecting the
Probabilistic selection of collision scenarios
The LHS sampling technique was used to generate a set of 50 collision scenarios using the PDFs of input parameters (see Table A1). Fig. 11 shows a summary of the selected values of all the input parameters in the form of a scatter diagram. The diagonal elements display the histogram of each parameter and the off-diagonal elements shows the correlation between two parameters.
Table 5 shows the summary of the fifty collision scenarios based on the simulation results. It is observed that the
Concluding remarks and future work
Accurate knowledge of load scenarios is essential for the structural design of offshore installations subjected to ship collisions. The aim of this study is to develop a new probabilistic model for the estimation of collision design load parameters. The random nature of the input parameters was treated by their respective probability distribution. Based on the selected 50 prospective collision scenarios using the LHS sampling technique, numerical computations using ANSYS AQWA were undertaken to
Acknowledgements
This research was undertaken at the Korea Ship and Offshore Research Institute (International Centre for Advanced Safety Studies) at Pusan National University, which has been a Lloyd's Register Foundation (LRF) Research Centre of Excellence since 2008. The study was supported by the research grant of Pusan National University. The authors greatly acknowledge DNV GL for providing free access to WOAD database.
References (68)
Collision scenarios and probabilistic collision damage
Mar. Struct.
(2002)- et al.
On the probabilistic distribution of loads on a marine riser
Ocean Eng.
(2017) - et al.
Risk of collision between service vessels and offshore wind turbines
Reliab. Eng. Syst. Saf.
(2013) - et al.
Influence of impact scenario models on collision risk analysis
Ocean Eng.
(2012) - et al.
Assessment of dropped object risk on corroded subsea pipeline
Ocean Eng.
(2015) - et al.
Damage analysis of ship collisions with offshore wind turbine foundations
Ocean Eng.
(2017) - et al.
Probabilistic approach for collision risk analysis of powered vessel with offshore platforms
Ocean Eng.
(2018) - et al.
Advanced method for the development of an empirical model to predict time-dependent corrosion wastage
Corros. Sci.
(2012) - et al.
A time-dependent corrosion wastage model for seawater ballast tank structures of ships
Corros. Sci.
(2004) - et al.
Design of offshore structures against accidental ship collisions
Mar. Struct.
(2014)