Hull Form Development for a High-Speed Trimaran Trailership

Transcription

1 Hull Form Development for a High-Speed Trimaran Trailership I. Mizine Computer Science Corporation, Advanced Marine Center, Washington, DC, USA S. Harries & M. Brenner Friendship Systems GmbH, Potsdam, Germany ABSTRACT: The paper describes the hull form development of an innovative High-Speed Trimaran Trailership (HSTT), capable of carrying about ft-trailers in the speed range from 26 to 32 knots. Calmwater performance, in particular the trade-off between cruise speed and top speed, is studied on the basis of first-principle methods, utilizing non-linear free surface potential flow and boundary layer calculations. A baseline comprising the center hull and the side hulls was created in a traditional design process. These shapes were utilized as input to a partially parametric model by which displacement distribution, bulb shape and the side hull stagger could be changed. The design space was investigated by means of various Designof-Experiments to develop an appreciation of dependencies (exploration). A multi-objective genetic algorithm was then repeatedly used to produce non-dominated solutions (exploitation). Energy consumption at two prominent speeds was simultaneously improved, while observing important constraints such as initial stability. The proposed HSTT design addresses dual use as a fast ship for military mobility and as a highspeed complement for trade along Interstate 95 at the US East Coast. 1 INTRODUCTION High-speed sea transportation usually calls for unconventional hull forms since the energy consumption of large conventional monohulls scales up too rapidly with an increase in speed. A trimaran concept was therefore studied for both commercial and military applications, introducing an innovative High-Speed Trimaran Trailership (HSTT), see Figure 1. The HSTT follows the requirements set by America's Marine Highways (AMH), an evolving US strategy, according to which commercial trailerships could be utilized to decongest traffic along the US East Coast, in particular on Interstate 95 (Boston-Miami). High-speed sea transport over more than nautical miles would need to be covered. The HSTT also supports the requirements associated with capability for military mobility in many inter and intra theater Sealift and Sea Base scenarios. Again fast ships would be required but with a range of unrefueled voyages of up to nautical miles. Consequently, a dual use concept was worked out which would hold the potential for the US armed forces to lease necessary and capable ships for their IX HSMV Naples May 2011 missions that, otherwise, would be used regularly to shift transport from road to sea. Figure 1: Hull form of the High-Speed Trimaran Trailership The paper focuses on the hydrodynamic development of the trimaran hulls on the basis of simulation-driven design, i.e., a process in which first-principle methods, here for numerical flow simulation, are intensively utilized not only to 1

2 analyze a few competing alternatives, but rather to stir the creation of many variants by searching for favorable performance (Harries 2008). The paper's layout is as follows: Principle ideas and requirements for the trimaran design are covered in section 2. The partially parametric modeling approach taken for creating variants and the flow simulation used to assess their hydrodynamic performance are outlined in section 3. Section 4 elaborates on the formal studies to understand the design space and to identify promising candidates. Section 5 presents the conclusions along with an outlook. 2 DESIGN RATIONALE The HSTT implements the type of the hull forms developed as prototypes for costal trimarans by (Vom Saal et al 2005) and Heavy Air Lift Support Ship (HALSS) by (Mizine et al 2009). The HSTT hull form has most of the displacement in the center hull, with small waterplane area (SWA) side hulls providing stability, as show in Figure 1. This ship carries 160 x 53ft trailers in eight bays or 240 x 40ft trailers in 10 bays on two decks. The upper deck has 11 rows while the second deck has 10 rows to provide room for structure (pillars) and access. All diesel engines are in the central hull with the main engine aft and the diesel-generator sets forward. The total installed power is about 66 MW, providing 54 MW at the propellers. Diesel engine exhausts are led through the crossover deck from the center hull to the stack as shown in Figure 2. aligned with a contra-rotating 14 MW Azipod, arranged to recover wake losses. Each side hull: 12 MW electric motor geared to a waterjet, powered by diesel-generator sets located in the center hull to provide good access for maintenance. There are two main reasons to select contra-rotating propulsion. Firstly, there is improved efficiency by absorbing rotational losses with the aft propeller. Secondly, contra-rotating propulsion allows the beneficial distribution of thrust load over a larger number of propeller blades in a confined space. The added total blade area then results in improved cavitation characteristics so that more power can be fed to a contra-rotating system with smaller propellers. This is especially important for a sealift asset with requirements for shallow-draft port accessibility. Without restrictions on the diameter, the benefit of a system with less noise and cavitation remains. Finally, an azimuthing electric propulsion unit yields excellent maneuvering characteristics The selection of waterjets in the side hulls is based on overall fuel efficiency in consideration of the dual use profile. When operating at speeds up to 26 kn only the center hull propulsion would be utilized. For higher speeds the side hull waterjets would have to be brought into action. Hence, the configuration avoids substantial wind milling at speeds below the cruise speed. The HSTT hull forms were developed in consideration of arrangement requirements, stability constraints and hydrodynamic optimization. A major factor in the power requirement of a trimaran is the interaction between the wave train produced by the center hull and the wave trains produced by the side hulls. Ideally, the two should counteract each other at the primary speed(s) of interest. Lessons learned from HALSS numerical analysis and model testing confirmed that wave interference and optimization are the key issues for the proper design of trimarans (Mizine at al 2008, 2009). Hence, they were intensively studied within the HSTT project. Figure 2: Main propulsion arrangement Several machinery options were evaluated, with the following combination identified as most promising: Center hull: 19.2 MW medium-speed diesel engine geared to a fixed-pitch propeller, 3 MODELING AND SIMULATION The formal process of hull form development was completely set up within the Computer Aided Engineering (CAE) software FRIENDSHIP- Framework, making use of the interfaced flow code SHIPFLOW. The FRIENDSHIP-Framework served as both the geometric modeling engine and the controller for simulation-driven design (FRIENDSHIP 2011). IX HSMV Naples May

3 3.1 Partially parametric model Since an advanced initial design had been established in Rhino prior to the systematic hull form development, it was decided to utilize the existing lines and apply partially parametric modeling for variation. Contrary to a fully parametric model in which the entire geometry is hierarchically built up from scratch, e.g. (Harries 2010), the partially parametric model takes an existing shape and only specifies the desired changes by means of parameters. The general idea is that of a chamber of mirrors in an amusement park: Distorted mirrors lead to images that are different in shape and yield more or less favorable variants. Figure 3: Example variations of center hull without skeg (top: baseline, middle: highest increase in displacement, bottom: largest inflation of bulb) side hulls were not changed in shape but shifted longitudinally and transversally while maintaining starboard-port symmetry, Figure 4. Up to seven parameters of the partially parametric model were controlled as free variables during the investigations, three of which altered the displacement volume and longitudinal center of buoyancy of the center hull, two modified the bulb's volume and profile while the last two changed the position of the side hulls. 3.2 Computational Fluid Dynamics Typical options of Computational Fluid Dynamics (CFD) for calm-water hydrodynamics currently encompass potential flow analysis of the free wave system along with thin boundary layer calculations and, alternatively, fully viscous flow analysis including the free surface. The right choice of CFD depends on the specific trade-off between accuracy, speed and usability. The trimaran hulls being rather slender and the wave making being of paramount importance at speeds of 26 kn and 32 kn (corresponding to Froude numbers of 0.32 and 0.39, respectively, at a reference length of around 180 m), a non-linear potential flow simulation with free sinkage and trim was considered to give meaningful results for the ranking of variants. Therefore, the optimizations were based on SHIPFLOW, capitalizing on the fast turn-around time of the software's zonal approach (FLOWTECH 2004, 2009). Figure 4: Example variations of side hull position (top: baseline, middle: most forward position of side hulls, bottom: most outward position of side hulls) Here a longitudinal shift of sections was applied for the trimaran's center hull, employing the Generalized Lackenby approach of swinging sections longitudinally in a highly concerted manner (Abt and Harries 2007), while a surface delta shift was used to inflate and deflate the bulb, Figure 3. The IX HSMV Naples May 2011 Figure 5: Wave generation of center hull at 26 kn (upper part) and 32 kn (lower part) Figure 5 shows the wave generation of the center hull without the side hulls at the two speeds of interest, namely 26 kn and 32 kn. The changes in wave lengths and the distinct interferences can be 3

4 nicely observed by comparing the upper and lower halves in Figure 5. The different positions and pronunciations of crests and troughs readily give rise to the assumption of favorable and unfavorable side hull positions but also directly suggest that a beneficial side hull position for one speed might be less advantageous for the other; cp. e.g. (Yang 2001), (Brizzolara et al 2005). Both the potential flow and boundary layer calculations done with SHIPFLOW are relatively fast, typically 5-10 minutes of CPU time on an average workstation. 4 FORMAL STUDIES Figure 6: Pressure distributions on center hull with and without skeg at 26 kn (side hulls omitted) 4.1 Design with and without skeg The proposed design featured a conventional propeller plus a contra-rotating POD drive in tractor mode. An important question for the hull form development therefore was whether the characteristics of the center hull's propulsion train would have a significant influence on wave making that needed to be taken into account during the optimizations. Two designs with and without skeg were considered as options. Figures 3, 4 and 6 give an impression. For clarification a first Design-of-Experiments (DoE) was conducted in which only the center hull was altered, while the side hulls were omitted. For each form variation the two alternatives, one with and the other without skeg, were concurrently analyzed at cruise speed and at top speed. Figure 6 displays the pressure distributions over the center hull for a representative variant. It can be seen that close to the skeg the pressure distribution changes, as is expected, while closer to the free surface along the hull, i.e., the regions that stimulate wave generation the most, very little differences occur. Figure 7 exemplarily depicts the overlay of waves for a variant with and its counterpart without skeg. The isolines of wave height that appear very close to each other, in particular, aft of the hull, stem from the alternative aftbodies. In the forebody there is practically no difference to be seen while the deviations in the far field are rather subtle. When plotting the wave resistance for all variants from the DoE for the designs with skeg (ordinate) vs. the designs without skeg (abscissa), Figure 8, it becomes apparent that the ranking of designs is not really affected. For the lower speed of 26 kn the variants almost line up on a straight line (blue diamonds). For the higher speed of 32 kn there are some minor oscillations for variants (red rectangles) that perform about equally well. IX HSMV Naples May 2011 Figure 7: Overlay of waves generated by center hull with and without skeg (side hulls omitted) Figure 8: Ranking of variants with regard to wave resistance for center hull with and without skeg (side hulls omitted) It was therefore concluded that wave resistance optimizations could be undertaken either with or without skeg and then reasonably generalized afterwards. It was decided to build further investigations on the baseline with skeg. Nevertheless, the correlation seen in Figure 8 gives the comforting thought of not having to repeat the 4

5 full exercise if a vessel without skeg would be pursued in the end. well as by the change in the pressure distribution on the main hull, as shown in Figure Exploration A comprehensive DoE with 250 variants was run for the trimaran configuration with the following free variables acting on the shape of the main hull and on the positions of the center hull: deltacp: change of prismatic coefficient within the interval [-0.01, ] deltaxcb: change of the longitudinal center of buoyancy [-1%, +1%] of LPP midtan: change of tangent of the shift function at the main section, [-60, + 60 ] bulbtipdz: vertical movement of the bulb tip within the interval [-1m, +0.8m] bulbfullnessfactor: increase of bulb volume via surface shift within [0, 2] sidehulldx: longitudinal position of the side hull wrt the baseline, [-10m, +20m] sidehulldy: transversal position of the side hulls wrt the baseline, [-2m, +2m] The different resistance components and some hydrostatic properties were evaluated for every variant, with the total resistance being the main objective. In terms of speed this investigation was limited to the cruise speed of 26 kn. The purpose of the DoE was to develop an understanding of the systems behavior within the design space and to identify promising regions for further exploitation (see following section). Figure 10: Performance of trimaran variants at 26 kn plotted for longitudinal (top) and transversal (bottom) side hull positions Especially the longitudinal position of the side hulls exhibits a strong linear relationship to the resistance within the range of the design variable, as depicted in Figure 10 top. At the speed of 26 kn it is clearly better to move the side hull forward, as long as the LCB position is not negatively influenced above reason. The transversal position has a smaller, but still significant influence on the resistance, see Figure 10 bottom. It is favorable to move the side hulls to the inside, closer to main hull. The deciding constraint here is the stability. Figure 9: Pressure distribution on trimaran with skeg at 26 kn The results of the DoE showed a strong dominance in the influence of the side hulls position, while the influence of the main hull parameters, being much smaller, basically vanished in the background noise. As previously discussed the effect of the side hulls on the resistance is mostly given by the favorable or unfavorable interferences in the wave pattern, as IX HSMV Naples May Exploitation Due to the experiences made in the previous design space exploration, it was decided to separately optimize for the most favorable side hull position first, taking into account the resistances at both relevant speeds. Because the relevance of every single speed was not known well enough to obtain a meaningful weighting in a single objective function, a multi-objective optimization algorithm was 5

6 selected. This would produce a Pareto set of nondominated solutions, from which a suitable compromise design can be selected afterwards. The selected algorithm was the non-dominated sorting genetic algorithm NSGA-II (Deb 2002) with 13 generations and a population size of 12 per generation. The mutation probability was set to 0.01 and the crossover probability to 0.9. The first optimization run with the full range of the side hull variables and constant center hull shape produced ca. 150 feasible variants and a distinct Pareto frontier, see Figure 11. The first conclusion that was evident from this set of designs and the additional data for the higher speed, was the opposite influence of the longitudinal side hull position on the two resistance values. Moving the side hull forward is favorable at 26 kn, but detrimental for the resistance at 32 kn. Furthermore, moving the side hull inwards has small influence at 26 kn but a more significant effect at 32 kn. An interesting feature that could be observed was a hook-like local minimum in the upper half of the Pareto set. The design located in the tip of this hook (c in Figure 11) and four other interesting designs were selected as starting points for further investigations. The other four designs were: The design (a) with the lowest power requirement at 26 kn. The max. power requirement (at 32 kn) is higher than for the baseline. The design (b) with the same max. power requirement as the baseline (the same engine could be used) but a reduced power requirement at 26 kn. The design (d) with the same power requirement as the baseline at 26 kn, but smaller max. power requirement. The design (e) with the lowest max. power requirement, but higher power requirement at 26 kn than the baseline. A local multi-objective optimization (NSGA-II) was started from each one of these designs. The design variables for the center hull were included in these optimization runs, while the range of the side hull variables was restricted to the vicinity of the respective current values. Each run produced 50 to 250 additional variants. The purpose was to find the optimal center hull for every selected position of the side hulls, allowing for some adaptation of this position. While the other runs mainly remained in the region of the Pareto set, the run started from the variant in the tip of the hook produced the most interesting results by further enhancing this local minimum. Figure 11: Effective power requirement of trimaran variants at 26 kn (horizontal axis) and 32 kn (vertical axis) Figure 12: Design space of the side hull positions. Longitudinal position on the vertical axis, transversal on the horizontal IX HSMV Naples May

7 Some more interesting information is given by inspecting the distribution of the variants in the design space of the side hull positions, see Figure 12. Obviously, the local optimizations form distinct clusters. The Pareto set is reflected by the left and upper border, with the hook local minimum in the upper left corner. All the designs above the hook in Figure 11 are on the upper border, the ones right and below in Figure 11 are on the left border. This picture seems to give a good indication for the specific shape of the Pareto frontier. As expected from the previous investigations, the final optimized design has an inward and forward position of the side hulls. The volume is slightly decreased with almost no movement of the center hull s center of buoyancy. Furthermore, the bulb is significantly fuller, with a higher bulb tip. See the details in Table 1. complete speed range, as the computed speed-power curve shows (Figure 14). Table 1: Properties of optimized design. All variable values are equal to 0 for the baseline Variable Value deltacp E-03 deltaxcb 1.921E-04 midtan deg bulbtipdz m bulbfullness sidehulldx m sidehulldy m Figure 14: Speed-power curves for baseline and optimized design. Ranges of variation for different trim and sinkage conditions indicated by arrows. Figure 13: Wave generation of trimaran at 26 kn (top) and 32 kn (bottom). Upper part of each picture: baseline, lower part: favorable design The favorable interferences lead to a greatly attenuated wave pattern with a reduced energy content, see Figure 13 for the waves at 26 kn and 32 kn. In comparison to the performance of baseline, the power requirement was reduced by 9.5% at 26 kn and by 4.2% at 32 kn. Overall, the optimized design behaves better than the baseline within the 4.4 Robustness analysis To ensure the robustness of the optimized design for off-design conditions, a sensitivity analysis with respect to different trim and sinkage situations was carried out. The drafts at the forward and aft perpendicular were set as independent design variables and an ensemble investigation was performed with variable values within the range of +/- 0.5 m compared to the design draft. Aside from the optimized design, the same investigation was done for the baseline for comparison. IX HSMV Naples May

8 The ranges of variation with respect to the required effective power can be seen in Figure 14. It is evident that the optimized design behaves in a very similar way to the baseline design. Therefore, it is assured that the optimized design is not sitting on a knife s edge optimum. It will not suddenly become much worse, if the ship is sailing at slightly other than design conditions. 5 CONCLUSIONS A new hull form was developed for a High-Speed Trimaran Trailership (HSTT), utilizing the FRIENDSHIP-Framework for simulation-driven design. A baseline design was taken and parametrically modified in order to bring about optimal performance at two important speeds, namely at 26 kn and 32 kn. The hull form of the center hull and the side hull positions were changed systematically in order to identify a Pareto-optimal solution of minimum resistance while satisfying various constraints on stability, weight and arrangement and complying with requirements on overall beam, trim, position of longitudinal center of buoyancy etc. Hence, the best combination was sought for low power, good stability and high loading efficiency with all cargo on two decks. Hull form optimization of trimarans benefits widely from powerful tools comprising (automated) hull form transformations, adequate simulation and practical optimization strategies. This set of tools is combined within the FRIENDSHIP-Framework. The practical design case of an HSTT proved its applicability. The numerical results naturally need to be further validated in the course of model tests. ACKNOWLEDGEMENT The HSTT concept evaluation and test program have been sponsored by Center for Commercial Deployment of Transportation Technologies (CCDOTT) and the Office of Naval Research (ONR). The FRIENDSHIP SYSTEMS calculations and analysis have been performed based on Cooperative Research Agreement between FRIENDSHIP SYSTEMS AND CSC Advanced Marine Center, CCDOTT and ONR. REFERENCES Abt, C.; Harries, S. (2007) Hull Variation and Improvement Using the Generalised Lackenby Method of the FRIENDSHIP- Framework. The Naval Architect, September. Brizzolara, S.; Capasso, M.; Francescutto, A. (2005) Effect of Hulls Form Variations on Hydrodynamic Performances of a Trimaran Ship For Fast Transportation, FAST Deb, K. (2002) A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II, IEEE Transactions on Evolutionary Computation, Vol. 6, No. 2, April FLOWTECH International AB (2004) XPAN / XBOUND Theory, Gothenburg, Sweden. FLOWTECH International AB (2009) SHIPFLOW 4.3 Users Manual, Gothenburg, Sweden. FRIENDSHIP SYSTEMS (2009) User Guide FRIENDSHIP- Framework, Potsdam, Germany. Harries, S. (2008) Serious Play in Ship Design, Tradition and Future of Ship Design in Berlin. Colloquium, Technical University Berlin. (Download: Harries, S. (2010) Investigating Multi-dimensional Design Spaces Using First Principle Methods, 7th International Conference on High-Performance Marine Vehicles, Melbourne, Florida, October. Mizine, I.; Karafiath, G. (2008) Wave Interference in Design of Large Trimaran Ship, International Journal of Marine Engineers, RINA, Vol. 150, Part A4, pp Mizine, I.; Karafiath, G.; Queutey, P.; Visonneau, M. (2009) Interference Phenomenon in Design of Trimaran Ship, Proceedings of the 10th International Conference on Fast Sea Transportation - FAST 2009, Athens, Greece, October Mizine, I.; Schaffer, R.; Saal, vom R.; Thorpe, R.; Starliper, M. (2009) HALSS Affordable Air Lift Platform for Navy and Humanitarian Missions, Proceedings of the RINA Warship Airlift at Seas International Conference, London, Great Britain, March. Saal, vom R; Mizine, I.; Deschamps, L.; Thorpe, R. (2005) Dual Use Short Sea Shipping Trailership / HSST-180, Marine Technology, Vol. 3, Yang, C.; Noblesse, F.; Lohner, R. (2001) Practical Hydrodynamic Optimization of a Trimaran, SNAME Annual Meeting, Orlando, October 2001, pp IX HSMV Naples May