Procedia Computer Science

Volume 51, 2015, Pages 2397-2405

CrossMark

ICCS 2015 International Conference On Computational Science

3D simulation of ship motions to support the planning of rescue operations on damaged ships

J.M. Varela1*, J.M. Rodrigues1^ and C. Guedes Soares1§

1 Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Técnico, Universidade de Lisboa, Lisbon, PORTUGAL §Also Research Center for High Performance Computing, ITMO University, Saint-Petersburg, Russia *varela@centec.tecnico.ulisboa.pt, fmiguel.rodrigues@centec.tecnico.ulisboa.pt, §c.guedes.soares@centec.tecnico.ulisboa.pt

Abstract

The paper describes a software system to simulate the ship motions in a crisis situation. The scenario consists of a damaged ship subjected to wave excitation forces generated by a random sea state. The simulation is displayed in an interactive Virtual Environment allowing the visualization of the ship motions. The numerical simulation of the sea surface and ship motions requires intensive computation to maintain the real-time or even the fast-forward simulations, which are the only ones of interest for these situations. Dedicated tools to analyse the ship behaviour in time are also described. The system can be useful to evaluate the responses of the ship to the current sea state, namely the amplitude, variations and tendencies of ship motions, and help the planning and coordination of rescue operations.

Keywords: Simulation system, ship motions simulation, intensive computation, real-time simulation

1 Introduction

The planning and management of rescue operations in the maritime environment is a vital task in post-accident situations. Typical rescue operations are firefighting, passenger evacuation from ungoverned ships during fire or flooding situations, and the towing of damaged ships to ports. The first two are obviously more urgent, however, the towing may also be considered an operation that must be performed rapidly before the ship status or environmental conditions degrade even more. Due to the adverse nature of the maritime environment, these operations always involve a certain degree of risk both for the rescued and for the rescuer. Moreover, when there are human lives in danger, the pressure for immediate rescuing increases the probability of new accidents. From Papanikolaou et al. (2015), it can be seen that many serious accidents still occur even when considering only the last 20 years. The majority of accidents lie on the category of hull or machinery damage, which implies the

Selection and peer-review under responsibility of the Scientific Programme Committee of ICCS 2015 2 3 97 © The Authors. Published by Elsevier B.V.

doi:10.1016/j.procs.2015.05.416

risk of flooding and consequent towing (if the ship is not lost). Also, high overall frequency of accidents occur for ships carrying passengers, namely for cruise ships and therefore passenger evacuation is a top priority operation.

When rescue operations are performed in open waters, the sea state is a critical factor. For passenger evacuation, the amplitude of ship motions influences the moving capability of the passengers and may determine the use of aerial or maritime rescue equipment. In the IMO-MSC/Circ. 909 and IMO-MSC/Circ. 1033, it is stated that ship motions' effects on passengers' behaviour should be accounted for by a safety factor, however, their influence on the final value of this factor is not clear. The progressive flooding simulation provides an estimation of the remaining time to perform the rescue. For towing, it is important to evaluate the ship motions along the planned route. Therefore, knowing in advance how the damaged ship will behave in waves is very useful to plan the operations and coordinate emergency teams.

Computer simulations for these objectives are a well-recognized and efficient tool to analyse this type of scenarios. The increased power of modern simulation technologies allows the creation of complex systems that combine simulation models of different areas such as seakeeping, manoeuvrability, progressive flooding or fire propagation into interactive emergency training simulators.

A computer system to simulate and analyse the motions of damaged ships in well-defined sea states has been under development (Varela and Guedes Soares 2007, 2014, 2015). Simulation-based modules are developed, implemented and integrated into a system which allows the 3D visualization of the ship motions in real-time and in fast-forward simulations.

The use of 3D visualization and Virtual Reality techniques in computer simulations and expert systems to support emergency situations on-board ships dates back to the nineties with the research work in the Naval Research Laboratory (Tate, 1991; Tate et al., 1997).

Some research work has been published recently, regarding the development of computer simulation systems that use Virtual Reality techniques to support training and planning in emergency situations. In Baldauf et al. (2012), simulation-based models have been integrated into training units and courses' programs to create a simulation laboratory with combined ship handling, safety and security test facilities. The simulator enables officers and crew to use safety equipment and available emergency systems while moving around inside the vessel. Varela et al. (2014) have integrated a progressive flooding algorithm into a Virtual Environment, creating a Decision Support Tool for ship flooding emergency response. The Virtual Environment is used to setup, control and visualize the simulation properly. Briano and Caballini (2011) describe a simulator for training logistic operators in ports, e.g. crane operators and truck drivers. In their work, Virtual Reality simulation models devoted to train Straddle Carrier, Quay Crane and Mobile Harbour Crane operators are implemented and emergency situations such as fire or falling of containers are also considered in the simulation.

In section 2, the system design and software architecture is depicted, including the simulation modules, the workflow and dataflow of the simulation, and the main functionalities of the system. The next three sections describe the main simulation modules, their underlying mathematical models and the approach taken for the numerical simulations. Finally, conclusions are taken in section 6.

2 System design

The system was developed using the object-oriented programming paradigm in conjunction with a modular approach to the system components. It is composed by four main modules: the Data Input, the Sea Surface, the Ship and the Graphics Engine modules. The Ship module is then composed by the Ship Motions and Flooding simulation modules as presented in Figure 1.

Figure 1 - The system is composed by three simulation modules, one data input module and a 3D Graphics Engine

The sea state is given in the form of a directional wave spectrum in frequency domain. Different approaches to estimate the sea state include methods based on the stationary ship motions measurements (Pascoal and Guedes Soares, 2008), on the marine radar imaging of the sea surface (Nieto Borge and Guedes Soares 2000) or on wave-induced buoy displacements (Jessen and Herbers, 2012). The Sea Surface module is then responsible for computing the free surface elevation according to the defined sea state. The other input is the ship damage. For this case, hull damages, namely holes that lead to the flooding of hull compartments are provided as well as the water level at compartments already flooded (Flood Condition in Figure 1).

The Ship module contains a set of pre-calculated transfer functions in the frequency domain for different ship speeds, headings, encounter frequencies, flooding situations. Given the wave phases and amplitudes in time, the ship position is computed in time domain based on the transfer functions and on the current ship status. The Flooding module adds the damage stability component into the system. From the current ship status, hull damage and flood condition, the flooding algorithm computes a ship flooded status that is added as a new cargo condition to the calculations of the ship motions on the next cycle. In this case, the floodwater is treated as an added cargo into the ship compartments.

The simulation workflow includes three main phases: the scenario setup, the numerical simulation and the analysis of the results, with consequent planning and management of rescue operations. In more urgent situations, such as when passenger evacuation is required, the setup of the simulation scenario must be done as soon as the alarm is given. The information required is the sea state and the hull damages, namely holes that lead to the flooding of hull compartments.

The Graphics API represents the ship status and the sea surface elevation given by the Sea Surface and Ship Motions modules respectively.

3 Sea surface simulation

The surface elevation at each location is given by the superposition of a large number of sinusoids with different amplitudes, frequencies and directions of propagation. This generates an irregular random sea surface based on very small steepness sinusoids from which the induced ship motions can be computed and non-linear terms can be neglected for calm and moderate sea states.

The sea state is defined in the frequency domain by a directional spectrum. The module responsible for the generation of the sea state and simulation of the sea surface receives the spectral function directly from the wave measurement devices.

The discretization of the spectrum is based on the method developed by Varela and Guedes Soares (2014), which ensures that the sea-state displayed by the real-time numerical simulations is very close to the sea state defined by the original spectrum.

A coordinate transformation is applied to the spectrum function to convert it from the frequency-angle coordinate space to the wave vector space. This is achieved by preserving integral equality between both spaces and according to the substitution rule (Fréchot, 2007). The directional spectrum function in the wave number space in Cartesian coordinates is given by the following expression:

e{K , K )=¿-Jfs M) «

where S (0,0) is the spectrum in the frequency-direction of propagation coordinate space. The discretization is uniformly applied in this coordinate space and depends of the simulation grid. The discretized components (wave systems) for a simulation grid with N x M points are obtained from the following expression:

^ 2m 2rnn^

where kn, m is the wave vector of the component (n, m) in the simulation grid, and n and m are integers with bounds -N/2 < n < N/2 and -M/2 < m < M/2. The wave vector defines the direction of propagation of the wave system and the wave number of each component is given by the length of the wave vector. The dispersion relation for deep waters, co1 = gk , is applied.

Numerical simulation of the sea surface in real time with a relatively large number of components is still only possible with FFTs. The uniform discretization of the spectrum in the wavenumber space allows the use of the IDFFT algorithm to compute the sum of the sinusoids at each grid point. The complex FFT based representation of the wave height field at the simulation grid point pnm is given by the sum of sinusoids with complex amplitudes:

,t) = ZZ~(kn,m (3)

where the vectors are evaluated at the (n,m) point of the simulation grid and h{jknm,t) are the height amplitude Fourier components and are given by the following:

H (t,t)= )t+*(i ® (4)

where a{kk) is the amplitude of the wave system derived from the spectrum function and ^(k) is the random phase term.

From the statistical properties of the wave spectrum, the following expression is derived for the wave amplitude:

a(k ,k )2 = 2f f S(k ,k )dk dk « 2S(k ,k W Ak (6)

V n' m/ I , I , V n m/ n m \ n m/ n m V /

•»Akn •'Akm

where Akn = Akm for uniform discretization of the directional spectrum.

4 Ship motions simulation

In general, the numerical seakeeping problem is solved following the procedure: 1) representation of the natural seaway as superimposition of many regular (harmonic) waves; 2) the individual reactions of each mode of the floating body to these harmonic waves are determined; 3) all reactions are superimposed to get the behaviour of the body in waves.

In this process it is implicit that the body reacts to waves independently, which means that the reactions will be summed linearly the same way one sums various waves linearly, resulting in a linear dependence on the wave height. In reality, this is valid for small amplitude (linear) waves, with height to length ratio equal or less of 1/20; higher amplitude waves may also be considered taking into account these prepositions, however acknowledging its limitations. If these simplifications are not made, the computations become considerably more expensive. So the nonlinear approach is mostly used only to solve particularly complex problems related, for instance, to extreme motions, such as capsizing investigations - when considering nonlinearities, the time domain approach is the tool of choice. In general, simple methods like the ones described in Lewis (1990) and Faltinsen (1993) suffice for obtaining global properties such as ship motions and accelerations. The method in Salvesen et al. (1970), commonly known as Strip Theory is also often used to compute the transfer functions for each condition, wave heading and frequency. The method of Santos and Guedes Soares (2008, 2009) is based on the strip theory and it accounts for the transient flooding and for the dynamics of the flooded ship. It is used to pre-compute the transfer functions for the numerical simulations of the ship motions. A set of configurations is made up constituting the database to be queried by the system, regarding the real time wave characteristics.

Transfer functions are pre-computed for specific ship speeds, headings and water levels in the flooded tanks. In the numerical simulation, these values are evaluated on a per cycle basis. Motion amplitudes and phases are then computed by linear interpolation between the transfer functions that are closer to the evaluated values. This results in a simulation where the transfer functions are updated continuously when the ship speed, heading or flood conditions changes. Transfer functions are computed for the full range of encounter angles and wave frequencies defined in the wave spectrum. Figure 2 represents a schematic view of the transfer functions that must be pre-computed.

Considering the example of Figure 2, the number of transfer functions, computed and stored in the database, is given by the following expression:

N = 6 • km ■ t ■ z ■ p (7)

where m is the number of tanks that can be flooded, k is the number of water levels considered for each tank, t is the number of forward ship speeds, z the encounter angles and p the wave frequencies. In this case, the transfer functions are computed for the 6 motions in the six degrees of freedom. From (7) it can be seen that if flooding situations of more than one tank are considered, the number of transfer functions and consequently the size of the database increases substantially. Transfer functions are stored in linked lists ordered by increasing values of water levels, forward speeds and headings. In order to increase the performance of the lookup procedure in the database during the real time simulation, the system stores the position in the list of the previous transfer functions that were used for the interpolations. As the ship speed, heading and water levels in tanks

suffer small changes between two consecutive cycles, the transfer functions that will be used in the next cycle will most probably be the same, the next or the previous element in the list. Therefore, the transfer functions used in the previous cycle will be the starting point for looking up the new ones in the next cycle.

Figure 2 - Ship transfer functions are computed for different flooding conditions, forward speeds, ship headings and wave frequencies.

5 Flooding simulation

A progressive flooding algorithm has been developed with a quasi-static approach, considering still water. The problem of a damaged vessel in waves is, thus, divided into to three fundamental problems: 1) determination of the amount of water inside each flooded compartment; 2) calculation of the average position of the ship; 3) the behaviour of the ship, considering this quantity due to waves. The progressive flooding algorithm is responsible for the first two items.

Wave induced flooding and outflow are not accounted, yet such an approach allows the formulation of a unified scheme were time domain results for flooding progression are coupled with frequency domain predictions of the behaviour of the vessel in waves. This motion will depend on the wave system, but also on the flooding water additional mass and the inertial aspects resulting from the oscillation of the free surface inside each flooded compartment. In this work, the first two are accounted for.

A literature review and discussion on the reasoning that supports the validity of the aforementioned approaches, relating to the flooding algorithm, may be found in Varela et al. (2014). Nevertheless, for the sake of completeness, its main characteristics are herein listed:

■ potential flow is assumed

■ Bernoulli equation used for flow calculations

■ zero net flow based nonlinear equations used for accounting with full compartments

■ free surface in flooded compartments remains horizontal

■ exact calculation of pressures and forces, within an adaptive quad-tree meshing scheme for considering interfaces, developed by Rodrigues and Guedes Soares (2014)

■ quasi-static motion calculation

■ RK4 solving scheme for motion solution at each call

Recent implementations of this algorithm may be found in Rodrigues et al. (2015), where a set of 90 probabilistically distributed damage configurations has been considered and applied to a shuttle tanker, so as to perform the loads assessment of the damage structure. In Figure 3, the progressive flooding regarding one of these damage configurations is depicted. Worth of notice is the considerable leaking of the cargo tanks into the neighbouring ballast tanks, leading to a much lesser final listing of the vessel.

Figure 3 - On the left the position of the ship with the levels on each compartment are shown; the meshing detail in way of the damage is shown on the right. The white polygons are the intersections of a box shaped box with the structure, defining the borders of the flooding openings.

6 Conclusions

A computer system which simulates the ship motions of damaged ships in irregular waves has been presented. The main purpose of the system is to evaluate the ship motions on a crisis situation, namely to have an indication of the amplitude and acceleration values at various zones of the ship. In conjunction with other information, the results of the simulation can help rescue teams to better determine the safer zones to evacuate passengers and the most adequate equipment to use.

The methodology used for the sea surface simulation is based on the wave spectrum estimated for the casualty scenario. Therefore, the realism of the simulation depends on the accuracy of the wave estimation. This is an important factor to have into consideration, because rescue teams that intend to use the system, must also have a reliable way of estimating the sea state. The simulation of the ship

motions is based on pre-calculated transfer functions, which are stored in a database and accessed in real time during the simulation. This requires that the database must exist a priori and be available to the rescue teams when the casualty occurs. Although with some limitations, mainly related with the linear nature of the method used to compute the transfer functions, the strip theory method used in this work presents very good results.

The modular approach adopted for the system allows changing the computation methods in a simulation module without affecting the others or the overall workflow of the system, as long as interfaces are maintained. Therefore, new methodologies for improving the accuracy of the ship motions are already planned and will be implemented in a near future.

Acknowledgments

The first author was funded by the Portuguese Foundation for Science and Technology (FCT -Fundagao para a Ciencia e Tecnologia) under its annual funding to the Centre for Marine Technology and Ocean Engineering (CENTEC).The second author was funded by the Portuguese Foundation for Science and Technology (FCT), under the grant nr. SFRH/BD/64242/2009.

References

Baldauf, M., Schroder-Hinrichs, J., Benedict, K., Tuschling, G. 2012. Simulation-Based Team Training for Maritime Safety and Security, Journal of Maritime Research, 9(3): 3-10.

Briano, E., Caballini, C. 2012. Simulation as a support tool for training logistic operators. Recent Researches in Engineering Education and Software Engineering, Rudas, Zaharim, Sopian and Strouhal (Eds), WSEAS Press, Cambridge, UK, ISBN: 978-1-61804-070-1, pp. 188-193.

Frechot, J. 2007. Realistic Simulation of Ocean Surface Using Wave Spectra, Journal of Virtual Reality and broadcasting, 4(11).

Faltinsen O.M., 1993. Sea loads on ships and offshore structures, Cambridge University Press

International Maritime Organization (IMO), MSC Circular n° MSC/Circ. 909, Interim Guidelines for a Simplified Evacuation Analysis on Ro-Ro Passenger Ships, (4 June 1999).

International Maritime Organization (IMO), MSC Circular n° MSC/Circ. 1033, Interim Guidelines for Evacuation Analysis For New and Existing Passenger Ships, (6 June 2002).

Lewis, E.V., 1990. Principles of Naval Architecture. The Society of Naval Architects and Marine Engineers, Volume III.

Nieto Borge, J. C. and Guedes Soares, C. 2000. Analysis of Directional Wave Fields Using X-Band Navigation Radar. Coastal Engineering. 40(4):375-391.

Papanikolaou, A., Bitha, K., Eliopoulou, E., Ventikos, N. 2015. Statistical analysis of ship accidents that occurred in the period 1990-2012 and assessment of safety level of ship types. Maritime Technology and Engineering, Guedes Soares & Santos (Eds), Taylor & Francis Group, London, ISBN: 978-1-138-02727-5, pp. 227-233.

Pascoal, R., Guedes Soares, C. 2008. Non-parametric wave spectral estimation using vessel motions, Applied Ocean Research, 30(1): 46-53.

Rodrigues, J.M., Guedes Soares, C. 2014. Exact pressure integrations on submerged bodies in waves using a quadtree adaptive mesh algorithm, International Journal for Numerical Methods in Fluids, 76(10): 632-652.

Rodrigues, J.M., Teixeira, A.P., Guedes Soares, C., 2015. Assessment of still water bending moments for damaged hull girders. Maritime technology and Engineering, Guedes Soares & Santos (Eds), Taylor & Francis Group, London, ISBN: 978-1-138-02727-5, pp. 331-340.

Salvesen, N., Tuck., E.O.,Faltinsen, O., 1970. Ship motions and sea loads, Transactions, The Society of Naval Architects and Marine Engineers;

Santos, T. A. and Guedes Soares, C. 2008; Study of Damaged Ship Motions Taking Into Account Floodwater Dynamics. Journal of Marine Science and Technology. 13:291-307.

Santos, T. A. and Guedes Soares, C. 2009; Numerical Assessment of Factors Affecting the Survivability of Damaged Ro-Ro Ships in Waves. Ocean Engineering. 36:797-809.

Tate, D. 1991. A Graphical User Interface Design for Shipboard Damage Control, NRL Report 9355, 26 Aug. 1991.

Tate, D., Sibert, L. and King, T. 1997, Using VEs to Train Firefighters, IEEE Computer Graphics and Applications, 17(6): 23-29.

Varela, J. M. and Guedes Soares, C. 2007. A Virtual Environment for Decision Support in Ship Damage Control. IEEE Computer Graphics & Applications, 27(4):58-69.

Varela, J.M. and Guedes Soares, C. 2014. Ring Discretization of the Wave Spectrum for the Sea Surface Simulation, IEEE Computer Graphics & Applications, 34(2): 58-71.

Varela, J.M., Rodrigues. J.M., Guedes Soares, C. 2014. On-board Decision Support System for Ship Flooding Emergency, Procedia Computer Science, 29: 1688-1700.

Varela, J. M. and Guedes Soares, C. 2015. Interactive 3D Desktop Ship Simulator for Offloading Manoeuvres. Applied Ocean Research . http://dx.doi.org/10.W16/jMpor.2015.0L013