ESTIMATION OF DRAG FORCE ON TUG HULLS AT DIFFERENT DRIFT ANGLES

Transcription

1 Proceedings of the 3rd International Symposium of Maritime Sciences Nov , 2014, Kobe, Japan ESTIMATION OF DRAG FORCE ON TUG HULLS AT DIFFERENT DRIFT ANGLES Buddhika JAYARATHNE, Dev RANMUTHUGALA, Shuhong CHAI, Jiangang FEI Australian Maritime College, University of Tasmania, Maritime Way, Newnham, Tasmania 7250, Australia Keywords: numerical analysis, drifts angle, wet transom, dry transom, interaction 1. INTRODUCTION When two vessels of significant size variation operate in close proximity, the resulting hydrodynamic interaction between the vessels can adversely affect their handling and safety, for example when tugs are involved in assisting large ships. During these manoeuvres, both the ships and tugs are exposed to dangers such as collision, grounding, girting, and run-overs. Hensen [1] investigated tug accidents over the last four decades, concluding that the interaction effects change with ship type, width of fairway, and drift angle. Although experienced tug masters know that interaction effects differ between ship types, identifying the safety envelop for each of these cases is much harder. Hensen [1] states that these forces and their effects become prominent when the vessels are significantly dissimilar in size and operate in close proximity during tight manoeuvres. Hensen et al. [2] questioned 160 tug masters concerning their awareness of interaction effects during these manoeuvres and the accuracy of ship handling simulators to reproduce these effects. Around 30% of the tug masters had faced critical situations due to unexpected ship-to-ship interaction effects in actual ship-assisted manoeuvres and around 40% emphasised the dissimilarity between the interaction effects in simulators and those experienced in real manoeuvres. Ship handling simulators use one of the following to predict interaction effects: empirical and semi-empirical methods, theoretical and numerical methods, or potential flow solvers without free-surface effects [3]. With the exception of the latter, the others require an interaction effect coefficients database to solve mathematical models fed into the simulators, with the database developed and validated by empirical and numerical techniques. Vantorre et al. [4] conducted physical model tests to determine the ship interaction effects in head-on and overtaking encounters of similar and dissimilar ships. The test results were used to create a new mathematical model to improve the quality of the interaction effects within ship manoeuvring simulators. Using experimental methods, Jong [5] studied the interaction forces and moments on ships with tugs attending at different drift angles under numerous test conditions. However, many researchers ([3], [6] & [7]) were interested in employing potential flow solvers as an alternative to the excessive work and high cost involved in other two methods. To date only the relatively simple potential double-body panel method is utilised to provide estimates of the interaction forces and moments in real time within simulators [7]. Pinkster and Bhawsinka [6] developed a computer program to estimate and validate the interaction effects in the simulator operated by Maritime Research Institute Netherland (MARIN). Their aim was to avoid using the interaction coefficient database generation by model tests, by employing the double-body potential flow method for multi-body cases involving ships and port structures. Real time interaction forces and moments were entered into the simulator through high speed computers to solve the flow equations. However, the final results were found to be highly sensitive to the initial conditions, which was tedious to setup. Sutulo et al. [3] introduced a modified potential flow code based on the classic Hess-Smith panel method and Thomson-Tait-Kirchhoff equations, with negligible free-surface effects. This code was used to augment the offline manoeuvring simulation with online predictions. They confirmed the usability of this code in manoeuvring scenarios where a number of ships moving approximately parallel to each other [3] and Sutulo et al. [7] validated their code in deep and shallow water towing tank tests for a tug operating near a larger vessel. They identified poor agreement of surge force and substantial discrepancies in sway force at small horizontal clearances between two ships. This difference was found to be more pronounced at very small horizontal clearances and nonparallel configurations, similar to those encountered during tugs assisting ships. Fonfach et al. [8] did experimental and numerical investigations to explore the contribution of various factors to interaction effects, which were not accounted for by the potential flow method. They revealed substantial influence of free-surface effects on the accuracy of predicted interaction effects. Many researchers ([9], [10], [11], [12] & [13]) have investigated the capabilities of potential flow methods to study various hull shapes, especially transom stern hulls with free surface. Pranzitelli et al. [9] studied the free-surface flow around a semi-displacing transom stern motor-yacht advancing steadily in calm water using the potential flow method and Computational Fluid Dynamics (CFD), and comparing them to Experimental Fluid Dynamics (EFD) results. It was found that the results generated from the potential flow method were substantially different because of the inability of its panels

2 to roll-down and intersect with each other during iterations. The researchers concluded that the presence of the free-surface can make more complicated discretisation, resulting in numerical problems for complex geometries, such as for transom stern hulls. The prevailing view ([3], [6] & [7]) is that using the present day computer hardware, potential flow method is the best approach to predict online interaction effects when free-surface effects are negligible. However, the free-surface has a significant impact on interaction effects and must be considered. Some potential flow codes have accuracy issues when non-streamlined geometries are solved with free-surface effects. Considering the interaction effects on a tug during ship assist, the rapid changes of tug drift angle causes a large portion of the downstream wake due to the hull to be characterised by a stagnation pressure in a similar manner to a deeply immersed wet transom, as shown in Figure 1.Thus, it is essential to select a flow solver (for online or offline predictions) that can accurately solve wet transom hull shapes. Therefore, this study aims to examine the accuracy of the drag force prediction of the commercial potential flow package, Futureship, in complicated interaction effects conditions. Flow Direction Flow Direction Standard Tug Deeply immersed Wet Transom Drifted Tug Act similar to a deeply immersed Wet Transom Figure 1: Standard and Drifted tug during ship assist FS-Flow is the module used in Futureship for Rankine-Source panel code analysis [14] and it solves the boundary value problem of potential theory including nonlinear free-surface. In the potential flow analysis, it is assumed that the fluid is inviscid and the flow is irrotational around the bodies. Hence, FS-Flow is equipped with a separate module capable of calculating the viscous resistance in terms of a friction line in combination with the wavy wetted hull surface. Therefore the dynamic forces, static forces, and viscous forces acting on the bodies are comprised in the final results though the fluid is considered as inviscid. Total resistance and its components obtained from the potential flow solver was then compared against the model test and CFD results generated by the commercial CFD code Star-CCM+ to investigate the possibility of using the software for future analysis of interaction effects. 2. NUMERICAL ANALYSIS The setup and relevant features of the two commercial software packages, FS-Flow and Star-CCM+, are discussed in this section. 2.1 Hull Form and Coordinate System The study utilised a 1:20 scaled hull model of the Australian Maritime College s (AMC) 35m training vessel TV Bluefin. The particulars of the full and model scale hulls are listed in Table 1. The two test conditions analysed to investigate the effects of transom generated complex flow regimes were: dry transom with a model draft of 0.17m; and wet transom with a model draft of 0.18m Table 1: Main Particulars of the Hull Form Main Particulars Unit Full Scale Model Scale Length Waterline, L m Wetted Surface area, S m Dry Transom Draft m Wet Transom Draft m A three-dimensional model scale hull form was developed using the commercial software Rhinoceros V5.0 and imported to the two packages. The coordinate system for the analysis is shown in Figure 2. The flow velocity vector was in the positive X direction while the horizontal plane through the origin was considered as the free surface. 2.2 Domain and Mesh in FS-Flow Flow velocities ranged from 0.34m/s to 1.04m/s in model scale, acting along the positive X direction, with the vessel allowed to trim and heave during the analysis. The free surface had a rectangular shape, with the inlet boundary at a distance of L m upstream the origin, the outlet boundary at 3L m downstream from the origin, and a total width of 1.1L m. The dimensions were selected to match those of the AMC towing tank, except for the length, which was shorten to reduce the computational effort without adversely affecting wake resolution. The mesh configuration is illustrated in Figure 2, which was developed in FS-Flow with an unfolded free surface mesh and enhanced to improve the stability of the solver. Figure 2: Coordinates and Ship Model with Free surface The mesh independence study was conducted through mesh refinements without affecting the stability of the solver. Then the drag coefficient at a forward speed of 1.04m/s was tested for dry transom condition for the models with different panel numbers to obtain an appropriate mesh. This approach provided sufficiently accurate results while maintaining low computational effort. The finest mesh investigated had 4220 panels, while a 3490 panel mesh was selected as a suitable mesh for steady-state simulations as its predictions were within 1.5% of that for the finest mesh (Figure 3).

3 % difference % difference % % % % 0.00 % 0.50 % 1.49 % Figure 3: Absolute % difference of Drag Coefficient against 3490 panel mesh 2.3 Setup and Mesh in Star-CCM+ Star-CCM+ uses a finite volume technique to solve the Reynolds Averaged Navier-Strokes (RANS) equations [15]. In order to directly compare the CFD and EFD results, the width and depth of the AMC towing tank were replicated in the numerical fluid domain, although the length was reduced to 10m to decrease the mesh load while ensuring the pressure and wake fields generated by the hull were sufficiently resolved within the domain. In addition, since the flow around the hull is symmetrical about the centerline, only starboard half of the model was used to reduce the computational effort. The vessel was fixed in all degrees of freedom, using particular trim and heave conditions obtained from the FS-Flow simulation results. The computations were performed using hexahedral trimmed mesh generated by Star-CCM+. Following a mesh independence study (Figure 4), a mesh with approximately 3.5 million cells was selected for the investigation. The near wall spacing on the vessel was taken around y + ~30 to meet the requirement stipulated for the k- SST turbulence model utilising a wall function [16], with a customised anisotropic refinement applied to the free-surface region as shown in Figure 5 to resolve the wave field around the hull. (Note: future work will investigate the use of smaller y + values with different turbulence models) % 3.17 % 1.59 % Number of Panels 0.00 % 1.06 % 0.53 % Number of Cells (millions) FS-Flow Star-CCM+ Figure 4: Grid independent study: Absolute % difference of Drag Coefficient against 3.5 Million cell mesh Figure 5: Hexahedral Mesh used in Star-CCM+ Simulations were treated as implicit unsteady for 25s total duration having 0.024s time step with 10 inner iterations. Free surface was modelled as Euler Multiphase and the volume of fluid technique and the inflow was considered as a flat wave having particular velocity. The drag force acting on the vessel was then calculated for similar speeds and drafts as done for FS-Flow. 3. EXPERIMENTAL SETUP Captive scaled model experiments were performed in AMC s 100m (length) x 3.55m (width) x 1.5m (depth) towing tank (Figure 6). The hull model, which was allowed to trim and heave, was attached underneath the carriage using one strain gauge and two Linear Voltage Displacement Transducers (LVDTs). Experiments were conducted for the two different drafts of the hull model. At the lower 0.17m draft the transom was in the dry condition, while at the higher 0.18m draft it was wet. Both conditions were tested at speeds ranging from 0.34m/s to 1.04m/s in model scale. Figure 6: Experimental testing in AMC Towing Tank; left - stern view, right bow view 4. RESULTS AND DISCUSSION 4.1 Drag Coefficient and Friction Coefficient The drag forces obtained from the numerical and experimental work were non-dimensionalised (C T ) in accordance with the ITTC 1978 [17] as given in Eq. (1). The frictional resistance coefficients (C F ) given in Eq. (2) obtained from the numerical results were compared against the ITTC correlation line given in Eq. (3) [17]. RT CT (1) 1 2 SV 2 RF C F (2) 1 2 SV ITTC correlation line (3) (log R 2) 2 10 e With the definitions given in Section 6: Nomenclature. Dry transom with a model draft of 0.17m. In this condition the transom remained dry above the waterline, giving a streamlined water-plane. The non-dimensionalised drag force results from EFD, potential flow code FS-Flow (PF), and CFD are plotted against the Length Froude number ( ) in Figure 7. The numerical and EFD results have a similar trend except at lower, where the numerical models tend to over-predict. This may be due to the non-accurate prediction of laminar to turbulent transient region on the scaled experimental model. However, the PF and CFD remain similar even at low. The maximum difference between the PF and CFD results is 15%, while the maximum between the PF and EFD results is 7.2%, except at the lowest as discussed above.

4 C T Figure 7: C T comparison for dry transom condition The results indicate that the viscous module integrated within FS-Flow has good prediction capability. In order to verify its accuracy, the frictional coefficients (C F ) obtained from the PF and CFD simulations are compared against the ITTC 1957 correlation line in Figure 8. The C F from the PF method correlates well with the ITTC line with a maximum difference of 5%, whereas the CFD values are slightly below the ITTC prediction with an average difference of 15%. A finer mesh with different turbulence models and a smaller y + may improve the CFD results. This was not carried out since the aim of the study was to investigate the accuracy of the PF solver. From the current work it is clear that the PF solver in FS-Flow is suitable to solve flow around well streamlined hull geometries. C F Error bars on EFD at 11.8% C T at Dry Transom (DT) C F at Dry Transom (DT) EFD at DT CFD at DT PF at DT ITTC at DT CFD at DT PF at DT the direction of C T changes around a of 0.2, causing the drag force on the vessel to act opposite to the flow direction. Since the total drag is made up of viscous, pressure, and wave making components, it is necessary to decompose the resistance in the different components to identify the real cause for this discrepancy. First considering the viscous drag force, a comparison was made between those obtained by the PF solver, the CFD shear force, and the ITTC 1957 correlation line, as presented in Figure 10. It is apparent that the viscous force generated by PF is in agreement with the ITTC correlation line, which is similar to the results obtained in the dry transom conditions as discussed in 5.2. C F C F at Wet Transom (WT) Figure 10: C F comparison for Wet Transom condition 4.2 Wave Pattern and Pressure Contour Since the results discrepancy was not related to the viscous effects, the residuary components were investigated next, especially as the error increased significantly with. Thus, the free surface wave patterns generated by PF and CFD at a speed of 1.04m/s were compared to identify the influence of wave making resistance. Figure 11 provides the wave patterns for wave height between ±0.03m. ITTC at WT CFD at WT PF at WT PF Figure 8: C F comparison for dry transom condition Wet transom with a model draft of 0.18m. In order to test FS-Flow s ability to solve wet transom conditions, the model was tested at the higher draft. The non dimensionalised EFD, CFD and PF drag forces in this condition are plotted against the in Figure 9. CFD C T at Wet Transom (WT) EFD at WT CFD at WT PF at WT C T Error bars on EFD at 11.8% Figure 9: C T comparison for wet transom condition It is evident that the CFD and EFD results are in good agreement throughout the range. However, the PF results, although is relatively close the the EFD at low, it significantly under predicts C T as increases. Interestingly, Figure 11: Free surface waves heights in PF & CFD It is clearly seen from this plots that the waves generated by PF are not in agreement with that obtained from EFD & CFD. Notably, the stern wave generated by PF has the highest magnitude, whereas the CFD as expected has it for the bow wave similar to the EFD as seen in Figure 6. The inaccurate wave pattern in PF will create a

5 high pressure region at the stern of the hull, which can result in a negative drag force. In order to verify this, the Dynamic Pressure Coefficient (C P ) generated by PF was examined as shown in the Figure 12. equations on transverse panels which block the flow streamlines creating breaking and spraying waves and it can solve and provide reasonable solutions for wet transom geometries if the transom mesh is removed. It is important to investigate the possibility of utilising this finding to conduct interaction effects analysis during ship handling operations. During such operations, tugs can dramatically change their drift angle to maintain the course of the ship. If the PF code is used to solve such cases, the panel generation has to be done as shown in Figure 15. Figure 12: Dynamic Pressure Coefficient generated by PF As suspected, the PF code has a high positive pressure region at the stern due to the weakness in predicting horizontal velocity component on the transom mesh, which create a zero horizontal velocity and hence negative drag force on the vessel. Since this unrealistic result occurred due to the wet transom, it was decided to check the drag force generated by the PF code without the transom mesh (Figure 13) at 0.18m draft. The results are plotted against the as shown in Figure 14. C T Figure 13: PF hull with and without transom mesh C T at Wet Transom (WT) EFD at WT CFD at WT PF at WT PF - No transom Error bars on EFD at 11.8% Figure 14: Comparison for wet transom condition, including the PF mesh without the transom 4.3 Results without Transom Mesh It is interesting to see that when the transom mesh is removed from the hull; the accuracy of the drag force predicted by PF results is appreciably improved showing good agreement with EFD and CFD results. Removing the transom mesh mitigated the error attributed to insufficient resolution of the large pressure gradient on the hull and poor numerical conditioning of the pressure integration. Thus, it is noted that the PF code is unable to solve flow Inflow Drifted Tug Tug Downstream Free Surface Mesh Tanker Figure 15: Ship Handling Operation: Panel generation in PF As illustrated in Figure 15, when the tug drift angle changes, a large portion of the downstream wake due to the tug hull is characterised by a stagnation pressure similar to a wet transom. This downstream area represents significant percentage of the total tug hull and it is not appropriate to neglect the tug downstream mesh during panel generation. Thus, the PF code FS-Flow is not a good option for complicated interaction effects analysis. The dynamic pressure prediction algorithm in FS-Flow is not capable of handling non-streamlined geometries and it needs further improvements to solve complicated interaction effects. 5. CONCLUSION In this paper the drag forces acting on a transom stern hull were investigated using EFD, CFD and PF methods, in order to identify a suitable method to determine interaction effects on a tug operating in close proximity to a large ship. The scaled model tests of the hull operating at two drafts provided wet and dry transom conditions conducted to investigate the accuracy of the PF solver FS-Flow in wet transom conditions. Although the PF solver was able satisfactorily predict the dry transom, it failed to do so for the wet transom against CFD and EFD results, especially at higher. Further investigation revealed that these discrepancies were due to the weaknesses in FS-Flow to predict the dynamic pressure in complicated flow regimes encountered in wet transom conditions. The following conclusions were drawn from the study, which may be relevant to future work. The viscous effects module integrated within FS-Flow provide accurate predictions. FS-Flow can be used to estimate the forces acting on well streamlined bodies to obtain accurate results across the length based range. If FS-Flow is used to solve drag forces on non-drifted wet transom hulls, it is necessary to omit the transom stern during panel generation. CFD is a possible alternative for EFD when non-streamlined bodies are analysed, although modelling and solution times can be excessive.

6 If PF solvers are used in complicated analysis, it is important to carry out verification and validation studies. In future work, the authors will use CFD to predict the interaction effects acting on a tug with varying drift angles, operating in close proximity to a large tanker, with validation through EFD. 6. NOMENCLATURE AMC Australian Maritime College CFD Computational Fluid Dynamics C F friction coefficient (dimensionless) C P dynamic pressure coefficient (dimensionless), C P = (P P )/q C T drag coefficient (dimensionless) DT dry transom EFD Experimental Fluid Dynamics Length Froude Number (dimensionless), = V/ g L m g gravity (9.81m/s 2 ) ITTC International Towing Tank Conference L m length of the ship model (m) P pressure (pa) PF Potential Flow code FS-Flow P freestream reference pressure (pa) q dynamic pressure, q = ρ V 2 / 2 (pa) R e Reynolds Number (dimensionless), R e = V L m /ν R F frictional resistance on ship model (N) R T total resistance on ship model (N) S wetted surface area of ship model (m 2 ) V velocity of ship model (0.34m/s to 1.04m/s) WT wet transom density of water (1000 kg/m 3 ) kinematic viscosity of water (1.00x10 6 m 2 /s) 7. ACKNOWLEDGEMENT The authors would like to acknowledge the extensive support given by Dr. Jonathan Binns and the AMC towing tank staff during the study. 8. REFERENCES 1. Hensen, H., Safe Tug Operation: Who Takes the Lead?, International Tug & OSV, 2012(July/August ) (2012), pp Hensen, H., Merkelbach, D. and Wijnen, F.V., Report on Safe Tug Procedures, (2013), Dutch Safety Board, Netherlands. 3. Sutulo, S. and Soares, C.G., Simulation of Close-Proximity Maneuvers using an Online 3D Potential Flow Method, Proc. of International Conference on Marine Simulation and Ship Maneuverability, (2009), pp Vantorre, M., Verzhbitskaya, E. and Laforce, E., Model Test Based Formulations of Ship-Ship Interaction Forces, Ship Technology Research, 49 (2002), pp Jong, J.H., Ship Assist in Fully Exposed Conditions - Joint Industry Project SAFETUG, Proc. of Tugnology '07, (2007), pp Pinkster, J.A. and Bhawsinka, K., A Real-time Simulation Technique for Ship-Ship and Ship-Port Interactions, Proc. of 28th International Workshop on Water Waves and Floating Bodies (IWWWFB 2013), (2013). 7. Sutulo, S., Soares, C.G. and Otzen, J.F., Validation of Potential-Flow Estimation of Interaction Forces Acting upon Ship Hulls in Parallel Motion, Journal of Ship Research, 56(3) (2012), pp Fonfach, J.M.A., Sutulo, S. and Soares, C.G., Numerical study of Ship to Ship Interaction Forces on the Basis of Various Flow Models, Proc. of 2nd International Conference on Ship Manoeuvring in Shallow and Confined Water: Ship to Ship Interaction, (2011), pp Pranzitelli, A., Nicola, C. and Miranda, S., Steady-state calculations of Free Surface Flow around Ship Hulls and Resistance Predictions, Proc. of High Speed Marine Vehicles (IX HSMV), (2011), pp Eliasson, S. and Olsson, D., Barge Stern Optimization: Analysis on a Straight Shaped Stern using CFD, MSc thesis, (2011), Chalmers University of Technology, Gothenburg, Sweden. 11. Mierlo, K.V., Trend Validation of SHIPFLOW based on the Bare Hull Upright Resistance of the Delft Series, MSc thesis, (2006), Delft University of Technology, Netherlands. 12. Doctors, L.J. and Beck, R.F., The Separation of the Flow Past a Transom Stern, Proc. of First International Conference on Marine Research and Transportation (ICMRT '05), (2005), pp Doctors, L.J., A Numerical Study of the Resistance of Transom Stern Monohulls, Proc. of 5th International Conference on High Performance Marine Vehicles, (2006). 14. DNV GL Maritime, User Manual of FS-Flow Version , (2014). 15. CD-Adapco, User Manual of Star CCM+ Version 08, (2014). 16. Simonsen, C.D. and R, C., RANS Simulation of the Flow around a Ship Appended with Rudder, Ice Fins and Rotating Propeller, Proc. of 11th Numerical Towing Tank Symposium (NUTTS), (2008), pp ITTC, Resistance Test, , (2011), International Towing Tank Conference.