Detection of Wave Slamming Sites from Ship Deflections

Detection of Wave Slamming Sites from Ship Deflections H Omer a and A Bekker a Received 24 May 2016, in revised form 2 August 2016 and accepted 2 August 2016 Stern slamming was found to be particularly problematic on a South African Polar Supply and Research Vessel. It inhibits oceanographic research operations and causes violent wave activity on the aft-deck where container laboratories are mounted. A methodology is required to distinguish between bow- and stern slamming in order to study full-scale ship responses. It is hypothesized that the wave slamming impact site can be verified from the animation of acceleration time histories of purposely placed accelerometers throughout the vessel structure. Vertical vibration measurements were performed at fifteen locations on the ship to capture a global sense of the slamming responses. Operational deflection shapes were calculated from these signals to infer if bow- and stern slamming can be distinguished. Results reveal that slamming vibrations propagate throughout the structure, causing a transient oscillation, known as whipping. The visualization provided by operational deflection shape analysis confirms that the wave impact site can be identified such that bow- and stern slamming events can be distinguished. A future investigation is recommended to determine if slamming may lead to structural fatigue of the vessel. Additional keywords: Operational deflection shapes, wave slamming, whipping, ship vibration. Nomenclature F frequency [Hz] N order Subscripts c cut-off Acronyms SAAII S.A. Agulhas II ODS Operational Deflection Shape PSRV Polar Supply & Research Vessel FFT Fast Fourier Transform FRF Frequency Response Function PSD Power Spectral Density CMU Central Measurement Unit FE Finite Element ICP Integrated Circuit Piezoelectric DC Direct Current 1. Introduction 1.1 Slamming and whipping In the context of the shipping industry, slamming is described as the exposure of a vessel structure to large forces due to wave impacts for a short duration of time [1]. Slamming specifically relates to the impulsive pressure caused by water entry, whereas whipping or springing refers a. Sound & Vibration Research Group, Department of Mechanical & Mechatronic Engineering, Stellenbosch University. annieb@sun.ac.za to the global vibratory response that follows a slamming event [2]. Constantinescu et al. [3] describe slamming as a random, dynamic and non-linear event affecting the structure of the vessel, both globally and locally. The local response refers to the area of the impact site which is under severe loading and is prone to damage in case of repetitive impacts. Slamming magnitude is influenced by the relative motion of the ship and the contact angle between the wave and the free surface [4]. Depending on the location of the impact on the vessel hull, slamming is categorized as bow bottom slamming, bow flare slamming, bow stem slamming, wet deck slamming and stern slamming [5]. Particularly in the case of bow- and stern slamming, the wave impact event could be succeeded by large transient vibration responses which are defined as whipping [2]. Whipping is considered to contribute to fatigue loading in vessels [6]. Stern slamming combined with whipping induce vertical bending moments which influence the fatigue life and ultimate longitudinal strength of hull girders [2,7]. Moton [8] and Kapsenberg et al. [7] state that a flat, raised stern design likely predisposes a vessel to stern slamming problems. In container ships, such designs have resulted in slamming incidents in mild, following seas where the wave length is comparable to the extent of the ship [7]. 1.2 The S.A. Agulhas II The S.A. Agulhas II (SAAII) is a South African Polar Supply and Research Vessel (PSRV) which accommodates 50 crew and 100 passengers on annual voyages to South African research bases on Antarctica, Gough Island and Marion Island. As such, this vessel is designed to endure a demanding operational profile including rough open water and ice passage. The SAAII is classified as polar class PC-5 [9] ice rating, indicative of her design for year-round operation in medium first-year ice (0.7 m to 1.2 m thick) which may contain old ice inclusions. The key specifications of the SAAII are presented in Table 1. The photograph in figure 1 shows the raised transom design to allow for a greater deck area at the stern whilst providing a versatile space for scientific cargo storage or for the mounting of container laboratories. The raised, flat stern section enables undisturbed operation of the propulsion system by minimizing the number of ice impacts with the propellers, especially when the ship is proceeding astern [10]. A certain amount of bow slamming is expected in head seas for ships with ice-going hull shapes as a result of the increased bow flare angle required for effective ice-passage [1]. However, on the SAAII, stern slamming is particularly problematic as it inhibits oceanographic research operations and causes violent wave activity on the aft-deck where container laboratories are mounted [11]. Passengers have additionally reported that slamming disturbs their sleep, interferes with equipment use and disturbs motor tasks such as typing and writing [11]. 50

In order to investigate the potential impact of stern slamming on human factors, equipment damage and mechanical fatigue implications, a methodology is required to distinguish between bow- and stern slamming on the SAAII. To this end, full-scale acceleration measurements were performed at fifteen locations throughout the hull and super-structure. Slamming events were subsequently identified from continuously recorded data. Operational Deflection Shape (ODS) analysis was performed to investigate the propagation of operational excitation through the structure [12], including forced and resonant responses. It is hypothesized that the wave impact site can be verified from ODS animation of acceleration time histories from purposely placed accelerometers throughout the vessel structure. Table 1 Specifications of the S.A. Agulhas II Polar Supply and Research Vessel. Length, bpp Beam Draft, design Deadweight at design displacement Installed power Propulsion Speed Range Capacity 121.8 m 21.7 m 7.65 m 5000 t 4 Wärtsilä 6L32 3000 kw Diesel-electric 2 4500 kw 16 knots (30 km/h) (max) 5 knots (9.3 km/h;) in 1 m ice 15,000 nautical miles (28,000 km) at 14 knots (26 km/h) 100 passengers in 46 cabins 4000 m 3 cargo hold 500 m 3 of polar diesel and twisting of the main deck, as well as the propagation of vibration through the super-structure (Figure 2). The sensors were located on structural girders to avoid local vibration responses. Bekker [13] showed that wave slamming on the SAAII causes vertical, fore-aft and lateral excitation of the vessel structure. However, only vertical acceleration was measured in the present investigation. The reason for this is limited measuring capacity there are restrictions on the channel count as a result of limited space for new cabling in the cable trays through water-tight sections of the ship. Three LMS SCADAS mobile data acquisition units were connected via fibre-optic cables and used in a master-slave configuration to enable measurements throughout the expansive structure. The accelerometers included ICP (100 mv/g, Model: PCB333B32), DC (200 mv/g, Model: PCB3711B111OG) and seismic accelerometers (1000 mv/g, PCB393B12). A sample rate of 2048 Hz was selected and measurements were recorded continuously and saved in 5 minute data records. Figure 3 shows the placement of two sensors on structural girders on the port- and starboard side of the Bridge on Deck 9. An accelerometer was mounted towards the aft of the super-structure in the stairwell of Deck 8. This deck serves as an accommodation space for the Captain and high-ranking ship officers. It was decided to capture vibration as closely as possible to the anticipated slamming locations in the bow and stern of the ship structure and place additional sensors at regular intervals along the vessel length as best is practically possible. Two sensors were placed on structural girders in the floor of the flat transom in the steering gear room (Figure 3). Towards the bow, two sensors each were placed at the stern thrusters and to the port- and starboard sides of the Central Measurement Unit (CMU) on Deck 3. A further four sensors measured acceleration in the corners of the cargo hold towards the front of the super-structure on Deck 4. Finally, two sensors were placed in the bow cargo space to capture vibration in close proximity to the bow wave impact site. Additional to the acceleration measurements, environmental parameters such as wave height and wind speed were recorded along with operational parameters of the vessel such as heading and ship speed. z x Figure 1 A photograph showing the raised stern section of the S.A. Agulhas II. 2. Methodology 2.1 Measurement Full scale measurements were performed on the SAAII during a voyage between Cape Town and Antarctica in 2014/15. Fifteen accelerometers were mounted to the vessel to capture synchronous recordings of vertical bending Figure 2 Location of the accelerometers on the S.A. Agulhas II 51

subjected to high-pass filters to remove the rigid body motion. For the purposes of this study the ship was assumed to be rigid below 1 Hz in keeping with assumptions of Howard et al. [15]. Two high-pass filters were designed to attenuate the low frequency vibration measured by the ICP and DC accelerometers respectively. A Chebyshev high-pass filter with an order, N = 800, and a cut-off frequency Fc = 1 Hz, was used to filter the ICP data. A higher order filter was required for the DC accelerometers which can measure below 0.5 Hz. A Chebyshev high-pass filter with an order, N = 1400, and a cut-off frequency, F c = 1.6 Hz was used to filter the DC and seismic accelerometer data. It was decided to conduct transient frequency analysis on the selected case study signals in order to investigate other signal characteristics which could be used to identify slamming automatically through identification algorithms. Figure 3 Accelerometers: Bridge, Deck 8 stairwell, cargo hold and bow (Deck 4), CMU, stern thruster (Deck 3) and steering gear (Deck 2). 2.2 Slamming case studies 2.2.1 Data selection Accelerometers in the steering gear room and the bow were used to determine potential slamming events as they were mounted the closest to the likely wave impact sites. Slamming was further identified in the substantial full-scale vibration dataset by referencing the notes of slamming instances; the notes were kept by researchers who participated in the voyage. Crest factors were calculated for the five minute recordings. Data records with the highest crest factors were further investigated based on the rationale that slamming is impulsive [14]. After these files were segregated, the vibration time histories of the remaining sensors throughout the vessel were plotted to investigate if the suspected slamming events resulted in global transient whipping responses [14]. The presence of wave slamming could further be confirmed by listening to audio playbacks of the acceleration recordings, where a wave slapping sound can be heard. Subject to these criteria two case studies were selected to present the findings with regard to wave slamming and the identification of the wave slamming site. The case studies involved the selection of suspected bow-and stern slamming events. It was reasoned that a distinction between bow- and stern slamming incidents is possible by investigating the time difference between the impulses in the acceleration time histories as a result of wave impacts. The signal that peaked first would serve as an indication that the wave impacted the vessel closer to the sensor in question. 2.2.2 Signal processing The selected case studies were processed further using MATLAB software. The acceleration measurements were decimated from 2048 Hz to 512 Hz which resulted in a signal cut-off frequency of 256 Hz. Finally the signals were 2.2.3 Calculation of operational deflection shapes Operational deflection shapes (ODS) were determined using LMS Test. Lab 10A Operational Deflection Shape and Time Analysis for the case studies. An important pre-requisite for data purposed for such calculations is that the measurements are required to be synchronized to enable the visualization of the relative phases of the measured degrees of freedom [16]. As the data was synchronously measured, the present work satisfies this requirement. Time domain response animations were calculated to visualize the ODS of the bow and the stern slams. Figure 4 presents a diagram of the geometry model of the SAAII polar supply vessel for the visualization of ODS. The geometry was created by plotting the spatial distribution of the fifteen measurement sensors (described in Section 2.1). Figure 4 3. Results A diagram of the geometry model of the SAA II, used for the visualization of ODS. 3.1 Slamming case studies Suspected bow- and stern slamming events were identified in the data using the methodology defined in section 2.2. The reigning environmental parameters for the case studies are presented in Table 2. Figure 5 presents the acceleration-time plot of the first slamming case study. It can be seen that the impulse for the sensor, in the steering gear room at the stern, precedes the peak measured at the bow. It is hypothesized that the wave impacted the vessel closer to the sensor at the stern of the ship, resulting in a stern slam. The oscillation decays to prior operational levels after about twenty seconds. For the second case study, presented in Figure 6, the acceleration peak in the 52

bow exceeds that of the steering gear room sensor, implying a bow slamming event. A selection of the conditioned data for the stern slamming case study is shown in Figure 7. When considering the ship heading at 303 and a south westerly wave direction, it is expected that approaching waves would likely impact the ship on the port side. The results were extracted from the port-side sensors, and the cargo hold sensor represents the sensor closest to the superstructure (see Figure 4). Table 2 Environmental parameters for the bow- and stern slamming case studies. Parameters Stern Slam Bow Slam Swell height 8.0 m 2.5 m Swell direction SW ESE Wind speed 52 kn 30 kn Wind direction W SE Ship heading 303 o 96 o Figure 5 Case study 1: Time histories of the bow- and stern (steering gear) acceleration measurements for a suspected stern slamming event. Figure 7 A selection of acceleration time histories showing the global vessel response for the stern slamming case study. Firstly, it is noted that the response is globally distributed throughout the vessel. Furthermore the stiffening effect of the superstructure is evident when considering the smaller magnitude of the acceleration measurements in the cargo hold and bridge. The lower vibration levels in this area could additionally be explained by nodal points in the excited mode shapes. The most noteworthy observation is that the time response in the CMU peaks before the response at the steering gear room, which implies its closer proximity to the wave impact site. It makes sense that waves could propagate under the flat stern transom and impact the vessel closer to the stern thruster sensors. However, this was not anticipated in the initial hypothesis. Figure 6 Case study 2: Time histories of the bow- and stern (steering gear) acceleration measurements for a suspected bow slamming event. 53

Figure 9 Acceleration spectrogram of the bow acceleration for a suspected stern slam. Figure 8: A selection of acceleration time histories showing the global vessel response for the bow slamming case study. Figure 8 presents a selection of acceleration time histories from the port-side of the vessel for a suspected bow slam. Notice the highly impulsive peak and large magnitude of the time history of the port-side bow sensor, encircled in Figure 8(d). The bow slam excites the global structure and the response is attenuated after about 40 seconds. The acceleration levels generated by the bow slam in 2.5 m swells exceed those resulting from the stern slamming case study in 8 m swell. More detailed measurement and analysis of wave conditions is required to link wave conditions, slamming incidence and resulting acceleration magnitude. 3.2 Transient frequency analysis Spectrograms were generated from the bow and steering gear room acceleration measurements as shown in figures 9 and 10. The spectrograms were calculated using a Hanning window, 50% overlap and a block size of 4096 which resulted in a frequency resolution of 0.125 Hz for the decimated signal at 512 Hz. Figure 10 Acceleration spectrogram of the stern acceleration for a suspected stern slam. The following observations are possible by considering the spectrograms: The wave slamming impulse excites a broad band of frequencies. The transient whipping response comprises the ringing of more concentrated frequency bands at approximately 2.1 Hz in the stern and 2.1 Hz and 3.7 Hz in the bow. The harmonics at 12.5 Hz are attributed to the firing order of the diesel generators on the SAAII which are expected to persist regardless of slamming incidences. The present case study contains at least two smaller slamming events at about 10.8 s and 18.8 s. These slamming events are more identifiable in the measurement closer to the stern. Similarly, the spectrograms of the bow- and steering gear acceleration measurements (figures 11 and 12) for a suspected bow slamming event show the excitation of responses across the investigated bandwidth as well as the subsequent ringing at single frequencies. Three smaller slamming events are observed in the bow spectrogram (at 8.0 s, 14.3 s and 18.8 s) which is believed to be closer to the wave impact site. 54

much like the bow slamming responses, take approximately 15 seconds to die down completely. Figure 11 Acceleration spectrogram of the bow acceleration for a suspected bow slam. Figure 13 ODS of the vessel structure at the peak slamming acceleration (t=0.412 s) for the stern slamming case study. Figure 12 Acceleration spectrogram of the stern acceleration for a suspected bow slam. The more minor slamming incidents are not as easily identified from the stern acceleration spectrogram (Figure 12). Subsequent to the bow slam, whipping responses at 2.1 Hz and 3.7 Hz resonate through the structure in the bow. As with the stern slam, only the frequencies around 2.1 Hz continue to ring in the stern measurement 3.3 Time domain responses Figure 13 presents the relative acceleration of the measured degrees of freedom at the instant when the peak acceleration occurs in the steering gear acceleration time history. The upward acceleration of the sensors in the stern thruster room and CMU confirm their proximity to the wave impact site. It was found that animations of the acceleration deflection shapes provided a clearer result for determining the wave impact site. However, the whipping responses are not shown with sufficient clarity to enable the observation of the global bending and twisting of the structure. For this reason, displacement deflection shapes were generated. Figure 14 presents the displacement of the sensor locations relative to the starboard side steering gear sensor for the stern slamming case study. Figure 14a shows the upward displacement and twisting of the structure in the vicinity of the stern thruster sensors and the aft of the super-structure. This confirms the hypothesis that the wave impact site is in the closest proximity to the stern thruster sensors on the vessel. The response propagates forwards and results in the combined twisting and bending motion of the bow and cargo hold. Bending oscillation is observed at the stern, which Figure 14 Time domain ODS of the displacement time history for the stern slamming case study. The vertical acceleration as a result of a bow slam is presented in Figure 15. The port-side bow and cargo hold structures are the first sensors to accelerate vertically, confirming their proximity to the wave impact site. The progression of the displacement time history (Figure 16) shows the transmission of vibration throughout the vessel and the subsequent loading of the bridge and the stern thruster region (Figure 16c). The amplitude of the motion decreases until, after approximately 40 seconds, the whipping response dies out. 55

Figure 15 ODS of the vessel structure at the peak slamming acceleration (t=2.558 s) for the bow slamming case study. ODS analyses indicated that slamming causes transient oscillations which propagate through the global vessel structure and resonate for some time. In the case of the bow slamming event, the twisting and bending of the vessel continued for almost 40 seconds. It remains to be investigated if this bending and twisting may result in local damage or structural fatigue as a result of the whipping response. According to the transient frequency analysis, spectrograms prove useful in the identification of both major and minor slamming events. Slams produce broadband excitation, which appear more clearly on the spectrograms of sensors closer to the wave impact site. In terms of whipping it was observed that a single frequency, around 2.1 Hz, resonates in the stern of the vessel, whereas two frequencies resonate in the bow (2.1 Hz and 3.7 Hz). According to a FE analysis by STX Finland [17], responses at 2.1 Hz and 3.7 Hz are associated with the first and second bending modes of the SAAII. Soal et al. [18] performed an operational modal analysis on the SAAII in a harbor environment. The first two bending modes were found to be at 1.94 Hz and 3.40 Hz respectively. These frequencies are lower compared to the FE model and whipping responses presented. The differences could be attributed to variations such as fuel level, cargo load and the resulting draft of the ship which should be investigated in future work. Additionally, modal analysis calculates only the resonant response of the structure, whereas ODS determines both the forced and resonant responses [12]. It remains to create an automated algorithm to identify and segregate bow- and stern slamming and whipping responses. Such an algorithm could benefit the investigation of a statistically significant number of slamming incidences to determine the response profile of the structure as a result of slamming excitation. Figure 16 Time domain ODS of the displacement time history for a bow slamming event. 4. Discussion Time domain responses have been proven useful to determine the proximity of wave slamming impacts on the SAAII. The results indicate that the wave slamming site can be approximated by the animation of purposely placed synchronous acceleration measurements throughout a ship structure. It has been shown that the animation of the acceleration time history provides an approximate indication of the slamming impact locations, whereas the displacement response sheds light on the subsequent whipping response. Secondary smaller wave impacts could not be as easily identified as those resulting from wave impacts following normal vibration in operational conditions. Results indicate that waves propagate under the flat, raised stern of the SAAII and impact the vessel closer to the stern thrusters. As such it is suggested that future analyses include the consideration of sensors placed in the stern thruster section for stern slamming identification. 5. Conclusion Operational deflection shapes were used to visualize the dynamic response from synchronous acceleration measurements on a polar supply and research vessel during wave slamming events. The analysis revealed that the impact site (bow or stern) is exposed to impulsive loading where the wave impacts after which the excitation propagates throughout the vessel. This whipping response results in global oscillations of the ship that lasts for as long as 40 seconds. Bow- and stern slamming events produce responses which are associated with the first and second bending modes of the structure. It was shown that slamming impact sites can be approximated by the animation of purposely placed synchronous acceleration measurements throughout a ship structure. Furthermore, it was shown that major and minor slamming incidences could be identified by the broadband excitation patterns in acceleration spectrograms close to the wave impact site. It remains to identify and segregate bowand stern slamming responses in order to investigate a statistically significant sample in terms of fatigue implications. 56

Acknowledgments The authors would like to thank The Department of Environmental Affairs, South Africa and the S.A. Agulhas II represented by Captain G. Syndercombe for their collaboration in the measurements on board the SA Agulhas II. The support of the National Research Foundation and Department of Science and Technology under the South African National Antarctic Programme is gratefully acknowledged for project funding. 16. Schwarz BJ and Richardson MH, Introduction to Operating Deflection Shapes, CSI Reliability Week, 1999, 121-126. 17. (STX Finland), FINNSAP finite element analysis of the PSRV NB1369, Rauma, Finland, 2010. 18. Soal K, Bienert J and Bekker A, Operational Modal Analysis on the Polar Supply and Research Vessel the S.A. Agulhas II, International Opererational Modal Analysis Conference, 2015. References 1. Kapsenberg GK, Slamming of Ships: Where are we Now?, Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 2011, 369(1947), 2892 2919. 2. Dessi D, Whipping-based Criterion for the Identification of Slamming Events, International Journal of Naval Architecture and Ocean Engineering, 2014, 6(4), 1082 1095. 3. Constantinescu A, Alaoui AEM, Nême A, Jacques N and Rigo P, Numerical and Experimental Studies of Simple Geometries in Slamming, International Journal of Offshore and Polar Engineering, 2011, 21(3), 216 224. 4. Bertram V, Practical Ship Hydrodynamics, Butterworth- Heinemann, Amsterdam, 2012. 5. Luo H and Soares GG, Review of Model Test Techniques of Local Slamming on Ships, in Maritime Engineering and Technology, Taylor and Francis, London, 2012, 189 194. 6. Storhaug G, Which Sea States are Dimensioning for Container Vessels when Whipping is Included?, ASME 33 rd International Conference on Ocean, Offshore and Arctic Engineering, 2014. 7. Kapsenberg GK, Van t Veer AP, Hackett JP and Levadou MMD, Whipping Loads Due to Aft Body Slamming, 24 th Symposium on Naval Hydrodynamics, Fukuoka, Japan, 2002, 25-39. 8. Moton CJ, Open-water Resistance and Seakeeping Characteristics of Ships with Icebreaking Bows, United States Naval Acadamy, Annapolis, Maryland, 1991. 9. International Assosiation of Classification Societies, Requirements Concerning Polar Class, 2011. 10. Riska K, Design of Ice Breaking Ships, Encyclopedia of Life Support Systems, 1 43, 2011. 11. Omer H and Bekker A, A Study of Wave Slamming Vibrations and Analysis in the Context of Human Factors on the S.A. Agulhas II During a Voyage to the Southern Ocean, 50 th United Kingdom Group Meeting on Human Responses to Vibration, 2015. 12. Brincker R and Ventura C, Introduction to Operational Modal Analysis, Wiley, 2015. 13. Bekker A, Slamming Measurements on the S.A. Agulhas II, Internal Report, Stellenbosch University, 2013. 14. Carlton JS and Vlasic D, Ship Vibration and Noise: Some Topical Aspects, 1 st International Ship Noise and Vibration Conference, 2005. 15. Haward BM, Lewis CH and Griffin MJ, Motions and Crew Responses on an Offshore Oil Production and Storage Vessel, Applied Ergonomics, 2009, 40(5), 904 914. 57