Ship Hull Resistance Calculations Using CFD Methods by ARCHIVES Petros Voxakis Bachelor of Science in Marine Engineering Hellenic Naval Academy, 2003 MASSACHUSETTS INS OF TECHNOLOGY JUN 2 8 2012 LIBRARIES Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degrees of Naval Engineer and Master of Science in Mechanical Engineering at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY May 2012 02012 Petros Voxakis. All Rights Reserved. The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created. Signature of author I Certified by I ,jv Department of Mechanical Engineering May 23, 2012 A I Chryssostomos Chryssostomidis Doherty Professor of Ocean Science and Engineering Professor of Ocean and Mechanical Engineering Thesis Supervisor Accepted by David E. Hardt Professor of Mechanical Engineering Chairman, Departmental Committee on Graduate Students 1 E Page Intentionally Left Blank 2 Ship Hull Resistance Calculations Using CFD Methods by Petros Voxakis Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degrees of Naval Engineer and Master of Science in Mechanical Engineering ABSTRACT In past years, the computational power and run-time required by Computational Fluid Dynamics (CFD) codes restricted their use in ship design space exploration. Increases in computational power available to designers, in addition to more efficient codes, have made CFD a valuable tool for early stage ship design and trade studies. In this work an existing physical model (DTMB #5415, similar to the US Navy DDG-51 combatant) was replicated in STAR-CCM+, initially without appendages, then with the addition of the appendages. Towed resistance was calculated at various speeds. The bare hull model was unconstrained in heave and pitch, thus allowing the simulation to achieve steady dynamic attitude for each speed run. The effect of dynamic attitude on the resistance is considered to be significant and requires accurate prediction. The results were validated by comparison to available data from tow tank tests of the physical model. The results demonstrate the accuracy of the CFD package and the potential for increasing the use of CFD as an effective tool in design space exploration. This will significantly reduce the time and cost of studies that previously depended solely on physical model testing during preliminary ship design efforts. Thesis Supervisor: Chryssostomos Chryssostomidis Title: Doherty Professor of Ocean Science and Engineering Professor of Ocean and Mechanical Engineering 3 ACKNOWLEDGEMENTS I would like to express my gratitude to my thesis advisor, Professor Chryssostomos Chryssostomidis, for supporting me at a difficult point in the course of my studies at MIT and giving me the opportunity to work on a very interesting subject. To Professor Stefano Brizzolara without whose support and guidance at every step of the way this work would not have been possible. For their mentoring and support during the time of my studies I would like to thank: Captain Mark S. Welsh, USN Captain Mark Thomas, USN Commander Pete R. Small, USN I would like to thank the Hellenic Navy for giving me the great opportunity to attend MIT and for financially supporting my studies. Last, but not least, I would like to express my sincere appreciation and thanks to my family and all the people whose friendship, love, and faith in me gave me the strength to be where I am today and give me the strength to move forward. 4 TABLE OF CONTENTS Abstract .............................................................................................................................. 3 Acknowledgem ents...................................................................................................... 4 Table of Contents......................................................................................................... 5 List of Tables......................................................................................................................7 List of Figures .................................................................................................................... 8 Chapter 1 - Introduction............................................................................................. 11 Chapter 2 - Existing D ata and M ethods .................................................................... 13 2.1. The DTM B 5415 Hull - Experim ental D ata ...................................................... 13 2.2. Appendage Resistance Prediction M ethods ........................................................ 18 Chapter 3 - The CFD Solver ....................................................................................... 24 3.1. Background ........................................................................................................ 24 3.2. The Reynolds Averaged Navier-Stokes (RANS) Solver .................................... 26 3.3. The Physics M odels........................................................................................... 27 Chapter 4 - The CFD M odel..........................................................................................30 4.1. Surface M esh....................................................................................................... 30 4.2. Volum e M esh ...................................................................................................... 36 4.2.1. Volum e M esh Generation and Description.................................................. 36 4.2.2. The Prism Layers............................................................................................. 40 4.3. M esh Evaluation.................................................................................................. 42 4.4. Boundary Conditions......................................................................................... 44 4.5. The 6-DOF M odel............................................................................................. 46 Chapter 5 - The DTMB 5415 Bare Hull Model with a Free Surface ............... 47 5.1. Introduction ............................................................................................................ 47 5.2. Trim , Sinkage, and Resistance Results ............................................................... 47 5.3. W ave Pattern and W ave Profile at Fr-0.41......................................................... 53 5.4. Rem arks .................................................................................................................. 55 5 Chapter 6 - The DTMB 5415 Hull Model with Appendages .................................. 57 6.1. Introduction ........................................................................................................ 57 6.2. Simulations at M odel Scale................................................................................ 58 6.3. Simulations at Full Scale.................................................................................... 60 6.4. Other Results ...................................................................................................... 62 Chapter 7 - Conclusions / Recommendations ........................................................... 75 7.1. Conclusions ........................................................................................................ 75 7.2. Recommendations for Future Work .................................................................... 75 References ...........................................................................................................-...- 77 Appendix A - Model to Full Scale ................................................................................. 79 Appendix B - Resistance Distribution And Frictional Resistance Coefficients ........ 81 Appendix C - Time Histories of Resistance, Trim, and Sinkage.............................86 6 LIST OF TABLES Table 1 - Geometrical data and Experiment 33 particulars for DTMB model 5415 and full-scale sh ip . ................................................................................................................ 15 Table 2 - Appendage form factors (1+k2). ....................................................................... 23 Table 3 - Physics models utilized by each simulation type, with or without a free surface . ........................................................................................................................... 27 Table 4 - Details on the prismatic near wall layers generated in the simulations. ........... 41 Table 5 - Boundary conditions defined for each simulation.........................................44 Table 6 - Experimental vs. CFD resistance, trim, and sinkage data.............................49 Table 7 - Experimental vs. CFD resistance, trim, and sinkage data using a different m esh ................................................................................................................................ 49 Table 8 - Numerical vs. experimental appended hull resistance data...........................59 Table 9 - Numerical vs. experimental appended hull resistance data including the values from the full scale simulations ....................................................................................... 60 Table 10 - Full scale effective power calculations when the Reynolds number of each appendage w as accounted for.................................................................................... 61 Table 11 - Drag distribution among the hull and its appendages from the appended hull 3 simu lation s......................................................................................................................6 Table 12 - Total resistance results from the bare hull simulations. The difference between appended and bare hull drag is also given. ................................................... 64 Table 13 - Subdivision of the total resistance in frictional and pressure drag for the model scale hull with and without a free surface at Fr-0.41 ..................................... 67 Table 14 - Some appendage resistance predictions with empirical methods compared to the C FD com putations................................................................................................ 73 Table 15 - Total, frictional, and residuary resistance values from the CFD computation s .................................................................................................................. 84 Table 16 - Frictional resistance coefficient for the hull and each appendage as derived from the frictional resistance computations in the CFD simulations ............. 7 85 LIST OF FIGURES Figure 1 - Geometry and photo of model INSEAN 2340. ............................................ 14 Figure 2 - Geometry and photo of model DTMB 5415. ................................................ 14 Figure 3 - Geometry and photo of model IHR 5512. ................................................... 15 Figure 4 - Geometry and photo of model DTMB 5415 as used in the CFD simulations.. 15 Figure 5 - Photographs of the Fully Appended Stem of Model 5415-1 Representing DDG-51 Without the Stem Wedge [10].................................................................... 16 Figure 6 - STAR-CCM+ Pictures of the Fully Appended Stem of Model 5415..........17 20 Figure 7 - Bilge keel geom etry .................................................................................... Figure 8 - Strut or control surface geometry. .............................................................. 20 Figure 9 - Shaft and bracket geom etry. ....................................................................... 22 Figure 10 - DTMB 5415 hull meshed to an STL file format with a close-up tow ards the bow .............................................................................................................. 31 Figure 11 -: DTMB 5415 hull with appendages meshed to an STL file format with close-ups towards the bow and stem ........................................................... 32 Figure 12 - DTMB 5415 hull meshed to an STL file format with a close up to the bow. This is the bare hull model used in the simulations without a free surface..................33 Figure 13 - The calculation domain obtained by the subtraction of the hull from a solid blo ck............................................................................................................................... 34 Figure 14 - DTMB 5415 hull remeshed in STAR-CCM+ with a close up to the bow.... 35 Figure 15 - Remeshed DTMB 5415 hull with appendages below the waterline, as used in the simulations, and close ups towards the bow and towards the stem. ................ 35 Figure 16 - Bare hull model simulations with a free surface volume mesh dimensions.. .36 Figure 17 - Appended hull model scale simulations volume mesh dimensions.............37 Figure 18 - Examples of volume shapes used to control the mesh density. ................. Figure 19 - From top to bottom, profile views of the volume meshes generated for the simulations with free surface and free motions, the simulations without free surface with appendages and without appendages.............................................39 8 38 Figure 20 - The boundary layer of the bare hull was modeled with 5 prism layers, in the simulations with the free surface, and 8 prism layers, in the simulations without free surface. ........................................................................................... ... ----- .......... 4 1 Figure 21 - The prism layers of the model size simulations were defined relatively thicker than those of the full scale simulations .......................................................... 42 Figure 22 - The wall y+ parameter values on the DTMB 5415 hull at Fr-0.41...........43 Figure 23 - The convective Courant number parameter values on the DTMB 5415 hull at Fr-0.4 1 ................................................................................................. . ....-- 44 Figure 24 - Boundary surfaces....................................................................................... Figure 25 - Total resistance experimental data plotted with the CFD results. ............. 45 Figure 26 - Trim angle experimental data plotted with the CFD results ............ Figure 27 - Sinkage experimental data plotted with the CFD results ........................... 50 49 50 Figure 28 - Time histories of resistance, trim, and sinkage as generated in CFD code STA R -C C M ±. ................................................................................................................ 52 Figure 29 - The CFD code also calculated the frictional and pressure resistances ..... 52 Figure 30 - Experimental and CFD wave pattern at Fr-0.41 ........................................ 53 Figure 31 - Experimental and CFD wave profile along the hull at Fr-0.41.................54 Figure 32 - Bow wave from the towing tank report and from the CFD simulation at F r- 0 .2 8 ........................................................................................................................... Figure 33 - DTMB 5415 hull model with appendages .................................................. 55 57 Figure 34 - Lines of the experimental effective and frictional power at the range of 10-32 knots vs. the numerical results at three speeds................................................59 Figure 35 - Lines of the experimental residuary resistance coefficient at the range of 10-32 knots vs the numerical results at three speeds..................................................59 Figure 36 - Difference in the frictional resistance coefficient value of each appendage when its Reynolds number is calculated separately at Fr-0.41 ................................. 62 Figure 37 - Appendage total resistance given as a percentage of the bare hull total resistan ce ........................................................................................................................ Figure 38 - Percentage subdivision of the total resistance in frictional and pressure, or residuary, drag for the model and full scale hull....................................................66 9 65 Figure 39 - Percentage subdivision of the total resistance in frictional and pressure drag for the model scale hull when a free surface exists at Fr-0.41 .................................. 67 Figure 40 - Streamlines around the struts and at the rudder at Fr-0.33 ....................... 69 Figure 41 - Matching of the frictional resistance coefficient for each appendage and the bare hull with the ITTC 1957 frictional line and the Blasius solution for laminar flow 70 Figure 42 - The skin friction and pressure coefficients depicted on the appended hull m odel at Fr-0.4 1 ....................................................................................................... 72 Figure 43 - Distribution of total resistance among frictional and pressure resistance......83 Figure 44 - Time histories of resistance, trim, and sinkage from the simulations with a free surface using a newer mesh at Fr-0.41............................................................... 87 Figure 45 - Frictional, pressure, and total resistance plot from the bare hull simulations w ithout a free surface at Fr-0.41............................................................................... 88 Figure 46 - Frictional, pressure, and total resistance plots of the hull with the appendages, the bilge keel, and the rudder from the appended hull simulations w ithout a free surface at Fr-0.41............................................................................... 89 Figure 47 - Frictional, pressure, and total resistance plot of the hull with the appendages from the full scale appended hull simulations without a free surface at Fr-0.41 ..... 10 90 CHAPTER 1 INTRODUCTION Starting in the seventeenth century experimental fluid dynamics appeared in France and England. Subsequently, theoretical fluid dynamics developed. Until about 1960, fluid dynamics were only studied using an experimental or theoretical approach. The rapid development of high-speed digital computers, along with precise numerical algorithms for solving problems using these computers, has introduced an important third dimension in fluid dynamics called Computational Fluid Dynamics, commonly referred to as CFD, and revolutionized the way we study and practice fluid dynamics today. In the late 1970s supercomputers were used to solve aerodynamic problems. HiMAT (Highly Maneuverable Aircraft Technology) was an experimental NASA aircraft designed to test concepts of high maneuverability for the next generation of fighter planes. Wind tunnel tests of a preliminary design for HiMAT showed that it would have unacceptable drag at speeds around the speed of sound. Redesigning and retesting it would have cost $150,000 and delayed the project unacceptably. The wing was redesigned by a computer at a $6,000 cost [4]. While the early development of CFD was driven by the needs of the aerospace community it is now used in all disciplines where the flow of a fluid is important. Some examples are the performance improvement of cars and their engines, the examination and better understanding of the real flow behavior of liquid metal during mold filling to help design improved casting techniques, and the calculation of the flow from an air conditioner. CFD can also be applied to examine the hydrodynamics of high-speed hull forms. While a large number of theoretical and experimental investigations into the hydrodynamics of ships have been carried out there are areas that require further research. The steady free surface flow and related forces prediction by numerical calculations is one example. The prediction of the flow field for high-speed hulls is complicated by the dynamic trim and sinkage which have a remarkable effect on ship generated 11 waves. The existence of a transom stem, used on most high-speed vessels, further complicates the problem as the large low-pressure area behind it generates waves, wave-breaking, and spray. CFD techniques are especially useful in analyzing flow problems in resistance prediction where complex fluid flow is present. While towing tank tests provide better absolute accuracy, CFD techniques can give results that are comparable to the towing tank results at a smaller cost in money and time. In addition, they have the advantage of allowing modifications to hull forms to be undertaken so that a comparative study of results can be made in a relatively short time and at relatively small cost [5]. In this study CFD computations are used to predict the resistance, trim, and sinkage of a highspeed hull with transom stem DTMB 5415. The results are compared to existing experimental ones and a good agreement is found. Computations are then made to predict the resistance of the same hull adding the appendages. The resistance characteristics of each appendage and how they affect-the total resistance of the ship are examined. The surfaces of the hull and its appendages were pre-processed, prepared, and meshed in CAD Software Rhinoceros 3D. Subsequently they were imported into CFD software STARCCM+ where the simulations were generated. 12 CHAPTER 2 EXISTING DATA AND METHODS In this chapter the hull used in the simulations is described, along with the types of simulations generated, and the experimental data that were used as a benchmark to evaluate the CFD results. In the second part of the chapter there is some talk about the existing empirical methods of estimating the appendage resistance of a ship. 2.1. The DTMB 5415 Hull - Experimental Data There is-an extensive benchmark database for resistance and propulsion CFD validation. Detailed tests done to create this database were reported on by the Resistance Committee of the 22nd International Towing Tank Conference [6]. The focus is on modem hull forms. Tanker (KVLCC2), container ship (KCS), and surface combatant (DTMB 5415) hull forms were recommended for use by the Resistance Committee and were used as test cases. The results were presented at the Gothenburg 2000 Workshop on CFD for Ship Hydrodynamics [7] and subsequent workshops and conferences. They are still used. The DTMB 5415 hull form was conceived as a preliminary design for a surface combatant with a sonar dome bow and a transom stem at the David Taylor Model Basin (DTMB) by the US Navy around 1980 [8]. It was constructed at the DTMB model workshop from a blank of laminated wood using a computerized numerical-cutting machine (Figure 2). The DTMB model 5415 is the hull used for all computations reported in this work. All the benchmark experimental data used to validate the CFD results were also gathered using the same hull or an exact geosym (INSEAN 2340). The bare hull resistance, trim, and sinkage results were compared with the results from Olivieri et al. [9] a combined effort from the Istituto Nazionale per Studi ed Esperienze di Architettura Navale (INSEAN, Italian ship model basin) and the Iowa Institute of Hydraulic Research (IIHR) to present experimental towing tank results for the purpose of CFD validation. The model used in this report was INSEAN 2340 (Figure 1). The CFD resistance computation results for the appended model were 13 compared to existing towing tank results performed at the David Taylor Naval Ship Research and Development Center [10]. The model used in these tests had fixed pitch shafts and struts and was designated Model 5415-1 to distinguish it from 5415, which had controllable pitch shafts and struts. In addition, the propellers, fairwaters, and twin rudders with rudder shoes were all redesigned for the fixed pitch configuration, while the bow sonar dome, the bilge keels, and the skeg were identical in both configurations. In this report, from Borda (1984) [10], several experiments were made with different appendage configurations, varying displacements, and measured quantities. From these experiments, experiment 33 was the one that corresponds to the conditions of the simulations prepared and run in the present study. For experiment 33 the DTMB 5415 hull is fully appended but with dummy hubs in place of propellers, the position of the rudders is at 0 degrees, and the resistance characteristics, with still air drag not included, are given in terms of effective horse power converted to the full scale ship with a correlation allowance CA=0.0004. The model is ballasted to represent the ship at the design displacement. Further details of the experimental set-up and the geometric characteristics of the DTMB 5415 model are given in Table 1. A picture of the fully appended model 5415-1 stem is shown in Figure 5. If the propellers in this picture were replaced with dummy hubs it would show the appendage configuration used for experiment 33. Figure 6 shows the appended hull of DTMB 5415 as modeled and used in the relevant CFD simulations presented in this study. Figure 3 shows model DTMB 5512 which is another geosym of DTMB 5415, this time at a scale of 1/46.6, which was used by the IIHR. The DTMB 5512 model has contributed significantly in the generation of benchmark information for validation of CFD results. Figure 1: Geometry and photo of model INSEAN 2340. Figure 2: Geometry and photo of model DTMB 5415. 14 Figure 3: Geometry and photo of model IIHR 5512. Figure 4: Geometry and photo of model DTMB 5415 as used in the CFD simulations. Table 1: Geometrical data and Experiment 33 particulars for DTMB model 5415 and full-scale ship. D)escr-iption Sy m1bol Shlip Scale factor M odel 24.824 Length between perpendiculars L,, (M) 142.0 5.720 Length at water level Lw (m) 142.0 5.720 Overall length Los (m) Breadth B (m) 18.9 0.76 Draft T (m) 6.16 0.248 Trim angle (Initially) (deg) 0.0 0.0 Displacement A (t) 8636.0 0.549 Volume V (m3) 8425.4 0.549 Wetted surface Sw (m2) 2949.5 4.786 Wetted surface with appendages SWA (m2) 3208 5.205 Water (1) Temperature T (degrees Celsius) 18.9 15 (2) Fresh, Salt Water FW, SW FW SW 15 AEu:IG 0 Figure 5: Photographs of the Fully Appended Stern of Model 5415-1 Representing DDG-51 Without the Stern Wedge [10]. 16 Figure 6: 3D Model of the Fully Appended Stern of DTMB 5415 used in the CFD simulations. 17 2.2. Appendage Resistance Prediction Methods The total resistance of a ship can be physically broken down into two components: frictional resistance and pressure resistance. The frictional resistance is the sum of the tangential shear forces acting on each element of the hull surface, and is solely caused by viscosity. The pressure resistance is the sum of the pressure normal forces acting on each element of the hull surface, and is partly a result of viscous effects and hull wave making. Towing tank tests to predict the resistance of a ship are initially performed on a bare hull model of the ship. The measurements of the model are then converted to the full scale ship following an extrapolation procedure. One such common procedure (Froude's method) assumes the division of resistance into skin friction and residuary resistance, where the residuary part consists of wave making and pressure form resistance. Other drag components have to be accounted for separately. These are: (i) appendage drag, (ii) air resistance of hull and superstructure, (iii) roughness and fouling, (iv) wind and waves, and (v) service power margins [12]. The appendages, in particular, can account for a significant amount of the total ship resistance. The main appendages of a twin-screw vessel are the twin rudders, the bilge keels, the twin shafting and shaft brackets, or bossings. The following methods are usually used to estimate the appendage drag: (i) Separate towing tank tests of the model with and without appendages. The difference between the two measured resistances should give the appendage drag which can then be scaled to full size. (ii) Tests are performed on a geosym set of appended models of varying scales. A form factor (1+k) is then derived which is used to predict the appendage drag of the full scale ship. This is an expensive and time consuming approach but provides a more accurate extrapolation procedure. (iii) Tests on separate models of the appendages. In this case, with high flow speeds and a larger model of the appendage, Reynolds numbers closer to full-scale values can be achieved. The hull influence on the appendage resistance is neglected. (iv) Empirical data and equations that come from model, and limited full scale, tests. The extrapolation of model test appendage resistance to full scale is not the same as with the naked hull. Some factors to take into account that complicate the task are: 18 (i) During the model tests the appendages are tested into a much smaller Reynolds number than at full scale. This means that while the appendages, in the model tests, are intersected by a laminar flow, at full scale they are likely to come across turbulent flow. (ii) Skin friction increases as the flow turns from laminar to turbulent, while the resistance in separated flow decreases. (iii) The relative thickness of the model boundary layers, when allowing for scale, is about double that of the full scale layers for usual model / full scale sizes. As a result, velocity gradient effects are greater on the model, and while the model scale appendages may be operating fully into the boundary layers, the full scale appendages may be projecting outside the boundary layers. (iv) Each appendage, attached to the hull, runs at its own Reynolds number, thus, when measuring its drag at the model size, the procedure to scale to the full size ship will be different. Some equations and data that can provide detailed estimates of the appendage drag at the appropriate Reynolds number in the absence of hull model tests are given in [12], [13], [14], [15], [16] and presented below. (i) Bilge keels The two sources of drag for the bilge keels are skin friction due to the added water surface and interference drag at the connection with the hull. A procedure recommended by ITTC to account for bilge keel drag is to multiply the total resistance with (S + SBK)/S, where S is the wetted area of the hull and SBK is the wetted area of the bilge keels. A formula to estimate bilge keel drag given by Peck [14] and referring to Figure 7 is: DB =pSBKV2CF 2 [ 2- +Y (1 When Z is large interference drag tends to zero, and when Z tends to zero interference drag can be assumed to be equal to skin friction drag. L is the average length of the bilge keel to be used when calculating CF. 19 L Figure 7: Bilge keel geometry. (ii) Rudders, shaft brackets and stabilizer fins Here the drag can be broken down to: (a) Control surface or strut drag, Dcs (b) Palm drag, De (c) Spray drag, in the case that the rudder or strut penetrates the free surface, DSP (d) Interference drag due to the connection of the appendage with the hull, DINT- The total drag, DAY, can then be written: DAP= Dcs + Dp + DSP + DNT (2) A formula proposed by Peck [14] for the control surface drag is: t31 S CM 1 pS F X 10-1, (3) + 40 1.25 -+ Dcs =-P 2 Cf A Ca where S is the wetted area, A is the frontal area of the maximum section, t is the maximum thickness, V is the ship speed, and Cm is the mean chord length which equals (Cf + Ca) (Figure 8) and is used for the calculation of CF. Cm Ca Cf Figure 8: Strut or control surface geometry. A control surface drag formula proposed by Hoerner [13] for 2D sections is: CD=CF[1+2()+60()]1 20 (4) where c is the chord length used for the calculation of CFHoerner [13] also proposed formulas for the estimation of the spray drag, DSP, the palm drag, Dp, and the interference drag, DINT. These formulas respectively are: 1 Dsp = 0.24-p 2 t, (5) 1 Dp = 0. 7 5CDpalm h DINT = Whp 1 pV 2 , (6) t 0.00031 1 2 3 0.75-t- P2tz C (t)2 203 (7) where tw is the maximum section thickness at the water surface, hp is the height of the palm above surface, 6 is the boundary layer thickness, W is the palm frontal width, CDPaIm is 0.65 for a rectangular palm with rounded edges, t is the appendage maximum thickness at the hull, and c the appendage chord length at the hull. (iii) Shafts and bossings Propeller shafts are generally inclined to the flow. As a consequence lift and drag forces are induced on the shaft and the shaft bracket. Careful alignment of the shaft bracket strut is required to avoid cross flow. The components of the resistance in this case are: (a) The shaft drag, DsH (b) Skin friction drag of cylindrical portion, CF (c) Pressure drag of cylindrical portion, CDP (d) Forward and after cylinder ends drag, CDE According to Hoerner [13], for Reynolds number Re < 5 x 105 (based on shaft diameter), the shaft drag is given by: 1 2 2 DSH = -pLSHDSV (1.1sin'a + WcCF), (8) where LSH is the total length of shaft and bossing, Ds is the diameter of shaft and bossing, and a is the flow angle relative to the shaft axis in degrees (Figure 9). 21 V Figure 9: Shaft and bracket geometry. The equations for the cylindrical portion drags as offered by Kirkman and Kloetzi [15] are: 3 CDP = 1.sin a, Re < 1 x 10 5 (9) -0.7154logioRe + 4.677, 1 x 10 5 < Re < 5 x 105, a > CDP = CDP = # (10) (-0.7154logioRe + 4.677) [sin 3 (1.7883logioRe - 7.9415)a], 1 x 10s < Re < 5 x 10s,a < fl CDP = 0.6sin 3 (2.25a), (11) Re > 5 x 10s, 0 < a < 400 (12) CDP = 0.6, Re > 5 x 105, 400 < a < 900 (13) where Re = VDc/v, projected area (L x P = -71.54logioRe + 447.7 and the reference area is the cylinder Dc). CF = 1.327Re- 0os, Re < 5 x 10s 1 1700 (3.461logioRe - 5.6)2 Re CF =7, (14) Re > 5 x 10s (15) where Re = VLc/v, Lc = L/tana, Lc > L and the reference area is the wetted surface area 71 x length x diameter. For the drag of the cylinder ends, if applicable, we have [15]: CDE = 0.9CoS 3 a, for support cylinder with sharp edges (16) CDE = 0.01CoS 3 a, for support cylinder with faired edges (17) Some other empirical equations to estimate the drag of a wide range of appendages are given by Holtrop and Mennen [16]: 22 RAPP = 1 pVsCF(l + where Vs is the ship speed, ITTC 1957 line and SAp CF SAPP + k2)E RBT, (18) is the friction coefficient for the ship determined from the is the wetted area of the appendage(s). (1 + k2)E is the equivalent (1 + k2) value for the appendages given by: ( (1 + k2)SAPP ) SAPP The appendage resistance factors (1 + k2) as defined by Holtrop are shown in Table 2 [16]. RBT in equation (18) is there to account for bow thrusters, if fitted, and is determined by: RBT = WTPVsdTCBTO, where dT (20) is the diameter of the thruster and the coefficient CBTO lies in the range 0.003- >0.012. Table 2: Appendage form factors (1+k 2 )- Appendage type (0 + k2 ) Rudder behind skeg 1.5-2.0 Rudder behind stem 1.3-1.5 Twin-screw balanced rudders Shaft brackets 2.8 3.0 Skeg 1.5-2.0 Strut bossings Hull bossings Shafts Stabiliser fins Dome Bilge keels 3.0 2.0 2.0-4.0 2.8 27 1.4 For the current study the appendages of the twin screw DTMB 5415 model consisted of the twin rudders and rudder shoes, the twin shafting, the shafting brackets or struts, and the bilge keels. CFD simulations were run for the model with and without the appendages. The total appendage drag, as well as the drag of each appendage separately, was calculated from these simulations. As part of the analysis of the results the percentage of each of the appendage drags to the total bare hull drag was found. A comparison of the CFD findings with results derived from some of the empirical equations previously mentioned was also made. 23 CHAPTER 3 THE CFD SOLVER A brief description of the CFD method for solving a fluid flow, with regard to the current study, follows. The physics models used in the CFD simulations are also briefly described in the last part of the chapter. 3.1. Background Three fundamental principles govern the physical aspects of any fluid flow: (i) the conservation of mass, (ii) the conservation of energy, and (iii) Newton's second law. These principles can be expressed in the form of mathematical equations, which, in their most general form, are integral or partial differential equations. In many cases these equations cannot be solved analytically. Computational Fluid Dynamics (CFD) replaces the integrals or partial derivatives in these equations with appropriate discretized algebraic forms that can be solved. The outcome of the CFD solution is numbers that in some way describe the flow field at discrete points in time and/or space. In modern CFD literature the Navier-Stokes equations refer to the complete system of flow equations, which solves for not only momentum, but continuity and energy as well. The complete system of flow equations for the solution of an unsteady, three-dimensional, compressible, viscous flow is: Continuity Equation Nonconservation form + pV - V = 0 Dt Conservation form Dt (21) + V - (pV) = 0 (22) Momentum equations Nonconservation form x component p Du Dt = y component pD Dt + a x ++ p+ax ax ay + pfX (23a) y + az TY+az Oz ay 24 + pfy (23b) z component + + +Zpf (23c) 2(puV)=) + + +a23 a(PV)+ V - (pV) = + + + + + Dw = - + Conservation form + x component y component z component a(Pw + V - (pwV) = az ax at ay + pf + pf az + pf z (24a) (24b) (24c) Energy equation Nonconservation form D Dt K( a(vp) ay e + V2 2)- a + . pq T) a a aT+ aT + ay k-ayl +z(kz k ( ax) ++ + pf - V (25) )v=pi+ a[p(e+Y ]+V -p(e +Y a-(k aT+ ay \ay a(uTyx) + ay a(up) -(k aT z \az az a ax _(vyy_ ___(vxy az ay a(up) 86~x aUZ + av~)+ av~)+ ay ax az a6u~xx) a(uryx) a(wp) a(PVP+a(TX)++ ay ax az (zy + (z ax az Conservation form p ax vxy)++ ax a(vp) ay ay + + Uk a wp) + a(UTXX) + ax -az az + ax + +awTxz) a(WTZ) + awTzz) + pf -V (26) ay az The conservation form of the governing equations derives from the control volume (finite or infinitesimal) fixed in space with the fluid moving through it, while the nonconservation form corresponds to the control volume moving with the fluid such that the same fluid particles are always in the same control volume. It is worth noting that this distinction between conservation and nonconservation form was triggered by the development of CFD and the question of which form of the equations was more suitable to use for any given CFD application. The conservation form of the equations is more convenient from a numerical and computer programming perspective. If we take the Navier-Stokes equations given above and drop all the terms associated with friction and thermal conduction we would have the equations for an inviscid flow which are called the Euler equations. The same (Navier-Stokes) equations describe any fluid flow, but not all flow fields are the same. The boundary conditions, and sometimes the initial conditions, determine the particular solution of the general governing equations for a specific problem. 25 3.2. The Reynolds Averaged Navier-Stokes (RANS) Solver The Navier-Stokes equations can only be solved analytically for a very small number of cases; as a result, a numerical solution is required. A numerical solution involves discretization of the governing equations of motion. When the partial differential equations are discretized then we have what is called finite differences, while when the integral form of the equations is discretized we have finite volumes. In practice, there are a few simplifying assumptions that can be made to allow an analytical solution to be obtained or to significantly reduce the computational effort demanded by the numerical solution. Such is the case with the incompressible RANS equations. By considering the flow as incompressible, which is a good assumption for most fluid flows, the continuity and momentum equations are simplified and the solution of the energy equation is no longer required. The Reynolds averaging process represents the three velocity components as a slowly varying mean velocity with a rapidly fluctuating turbulent velocity around it. It also introduces six new terms, known as Reynolds stresses. These new terms represent the increase in effective fluid velocity due to the existence of turbulent eddies in the flow. The introduction of turbulence models serves to represent the interaction between the Reynolds stresses and the underlying mean flow and to close the system of RANS equations. In the present study the hull flow was computed using RANS equations that are becoming a standard for the numerical prediction and analysis of the viscous free surface flow around ship hulls. Continuity and momentum equations, for an incompressible flow, are expressed by: pU = -Vp + yV2U + V - TRe + SM (27) where U is the averaged velocity vector, p is the averaged pressure field, [i is the dynamic viscosity, SM is the momentum sources vector and TRe is the tensor of Reynolds stresses, computed in agreement with the k-epsilon (or k - E) turbulence model. The free surface was captured using the Volume of Fluid approach that requires the solution of another transport equation for a variable that represents the percentage of fluid for each cell: (28) vof + U - VVof = 0 The Finite Volume commercial code STAR-CCM+ [19] was used for the solution of RANS equations on trimmed unstructured meshes as presented in Chapter 4. 26 3.3. The Physics Models Table 3: Physics models utilized by each simulation type, with or without a free surface. Eulerian Multiphase (water, air) Constant Density Fluid (water) Implicit Unsteady Steady K-Epsilon Turbulence K-Epsilon Turbulence Three-Dimensional Three-Dimensional Segregated Flow Segregated Flow Reynolds-Averaged Navier-Stokes Reynolds-Averaged Navier-Stokes Two-Layer All y+ Wall Treatment Two-Layer All y+ Wall Treatment Gravity No body forces Gradient Method: Hybrid Gauss-LSQ Gradient Method: Hybrid Gauss-LSQ Volume of Fluid (VOF) No Free Surface K-Epsilon turbulence model The K-Epsilon turbulence model is a two-equation model categorized as an eddy viscosity model. Eddy viscosity models use the concept of a turbulent viscosity pt to model the Reynolds stress tensor as a function of mean flow quantities. In the K-Epsilon model additional transport equations are solved for the turbulent kinetic energy k and its dissipation rate E in order to enable the derivation of the turbulent viscosity itTwo-Layer All y+ Wall Treatment To resolve the viscous sublayer, the K-Epsilon turbulence model, with the two layer treatment, divides the computations into two layers. In the layer adjacent to the wall the turbulent viscosity pt and the turbulent kinetic energy dissipation rate E are defined as functions of wall distance. The values of E in the layer adjacent to the wall are blended smoothly with those calculated in the layer further from the wall using the transport equations. The turbulent kinetic energy k equation is solved in the entire flow. The all y+ treatment attempts to emulate both the high y+ wall treatment for coarse meshes and the low y+ treatment for fine meshes while also producing reasonable results for meshes of intermediate resolution. Segregated Flow The segregated flow model derives its name from the fact that it solves the flow equations, one for each velocity component and one for the pressure, in a segregated or uncoupled manner. The continuity and momentum equations are linked with a predictor-corrector approach. This model is most suitable for 27 constant density flows. The second order upwind convection scheme was used with this model in the present study. Implicit Unsteady The implicit unsteady model uses the implicit unsteady solver and, in STAR-CCM+, it is the only unsteady solver that can be combined with the segregated flow model. The main function of the implicit unsteady solver is to control the update of the calculation at each physical time, while it also controls the time step size. In general, the implicit unsteady solver is the alternative to the explicit unsteady solver with the choice between the two determined by the time scales of the phenomena of interest. The explicit schemes have the disadvantage of being prone to instabilities if too large a time step is employed. The simulations without the free surface do not require the usage of this model, and the flow for those simulations is modeled as steady. Volume of Fluid (VoF) As already mentioned the Volume of Fluid approach is used in combination with the RANS solver to determine the location of the free surface. In this method the location is captured implicitly by determining the boundary between water and air within the computational domain. An extra conservation variable is introduced that determines the proportion of water in the particular mesh cell with a value of one assigned for full and zero for empty. For the simulations where there is no free surface, with the only fluid being the water, this model is not selected. Three-Dimensional The space models primarily provide methods for computing and accessing mesh metrics such as cell volume and centroid, and cell and face indexes. The three-dimensional space model is selected as it is designed to work on three-dimensional meshes. Eulerian Multiphase The Eulerian multiphase model is required to create and manage the two Eulerian phases of the simulations with the free surface, where a phase represents a distinct physical substance. The two phases for these simulations are water and air, each defined to have constant density and dynamic viscosity adjusted according to the average temperature of the tow tank experiments. This model is not required for the simulations without a free surface where the only fluid is water, which is again defined to have constant density adjusted to the temperature of the experiments used to validate the CFD simulations. 28 Gravity The selection of the gravity model means the action of gravitational acceleration is included in the simulations. This model provides two effects for fluids. The reference altitude (defined by the user) is taken into account in the calculation of the pressure, and the body force due to gravity is included in the momentum equations. This model is also not necessary for the simulations that do not have a free surface. Gradient Method: Hybrid Gauss-LSQ The transport equation solution methodology requires the use of gradients. One of the ways that the gradients are used is in the computation of the values of the reconstructed field variables at the cell faces. The chosen method for this computation was the hybrid Gauss-Least Squares Method (LSQ), which is considered to be a more accurate approach for the cell gradient calculations than the GreenGauss method. 29 CHAPTER 4 THE CFD MODEL In the following chapter the simulation creation process is described starting from the remeshing of the hull surface, continuing with the "construction" of the region defining blocks and the generation of the volume mesh, and concluding with the selection of the boundary conditions and the 6-DOF model. A brief section on mesh evaluating methods is also given. 4.1. Surface Mesh The DTMB 5415 hull that was used for the work presented in this document has already been described in Chapter 2. The surface of this hull had to be meshed before it was possible to work with it in the CFD software. The pre-processing of the surface and initial meshing was done in Computer Aided Design (CAD) software Rhinoceros 3D, and subsequently the surface was imported into the CFD code STAR-CCM+. The surface processing in Rhinoceros 3D for the hull without appendages included the scaling and waterline positioning so that the size and position of the waterline respectively-matched that of the hull used for the benchmark towing tank experiments [9], [10]. Also, the midships was positioned at the origin of the axes. Finally, the hull was meshed to a stereolithography (STL) file format. STL files describe only the surface geometry of a three dimensional object without any representation of color, texture, or other common CAD model attributes. STL files contain polygon mesh objects. In particular, they describe a raw unstructured triangulated surface by the unit normal and vertices (ordered by the right-hand rule) of the triangles using a three-dimensional Cartesian coordinate system. The STL format specifies both ASCII and binary representations, but the binary representation is more compact and, for this reason, more common. A binary STL file format was used for the current work. When creating the STL files the focus was to maintain a balance between a mesh that describes the complex geometry of the hull well but is not too fine, so that size of the mesh is as compact as possible and is easier to "handle" in the CFD software. Furthermore, a surface mesh that is only as fine as necessary would help create a more efficient volume mesh, and save in computational cost later in the simulation "building" 30 process. Figure 10 shows the STL mesh file generated in Rhinoceros 3D for the simulations with a free surface, while Figures 11 and 12 show the STL meshes for the hull with and without appendages for the simulations without a free surface. The mesh is finer at the more complex areas of the hull surface in order to capture the extra details and to more accurately represent them. For the same reason the mesh of the hull with the appendages needed to be finer to capture the extra, complex surfaces added by the appendages. Furthermore, the simulations without the free surface could "afford" a better mesh refinement, leading to more precise results; the lack of free surface leads to great savings in computational "effort" required for the calculation of the free surface at each time step. The non-free surface simulations also had a smaller volume mesh domain, compared to the simulations with a free surface. Since only the part of the hull below the water surface needed to be modeled, the STL file for the appended hull was created for the part of the hull below the waterline. This method of creating the non-free surface simulations led to a lot of errors in the surface mesh at the connection between the hull below the water surface and the new surface created to cover the gap left by the removal of the upper half of the hull (the CFD code required a closed surface), and this is why, when the simulations without free surface for the non-appended hull were created, the STL mesh file used was that for the full hull. This can be seen in Figure 12 and is explained later in this chapter. Figure 10: DTMB 5415 hull meshed to an STL file format with a close-up towards the bow. 31 Figure 11: DTMB 5415 hull with appendages meshed to an STL file format with close-ups towards the bow and stern. In the stern close-up the shafts, struts, rudders, and part of the bilge keels can be seen. 32 Figure 12: DTMB 5415 hull meshed to an STL file format with a close up to the bow. This is the bare hull model used for the simulations without a free surface. One laborious part of the pre-processing for the hull with the appendages was the attachment of the appendages to the hull. This was required as the hull and its appendages were created separately. The difficulty of this procedure came from the requirements that the final surface matched the surface of the model hull used in the towing tank experiments as accurately as possible and that it was watertight. The connections had to be smooth and not cause an alteration in the dimensions of the final surface. The final surface had to be closed (or watertight) and allow for the generation of an STL file that also contained completely closed (watertight) polygon mesh objects with as few errors as possible. In STAR-CCM+ the first step was to make a diagnostic check on the mesh to assess the validity of the surface and repair any errors found. This is a necessary step before creating the simulation and, later on, remeshing the surface in STAR-CCM+ and generating the volume mesh. Some common surface mesh errors that may need to be repaired are: (i) pierced faces, which are faces intersected by one or more edges of other faces, (ii) free edges, which translate to some opening or hole in the surface, and (iii) non-manifold edges, which, in our case, translated to having some extra surfaces that were not required and could reduce the efficiency and effectiveness of the mesh. 33 The next step was to "build" a rectangular block around the hull that would later become the domain of the volume mesh representing the water and air surrounding of the hull. Then the imported hull surface was subtracted from the block with the block as the base of the subtraction. The simulation proceeded with the outcome of this subtraction, the subtracted block. This way it was assured that the volume mesh would not extend to the inside of the hull. The subtracted block for the simulations with a free surface is shown in Figure 13. In this figure the block can be imagined to be cut in two symmetric parts along the symmetry plane of the ship. In the left picture we can see one of the two symmetric parts. In the right picture there is a closer view to the area of the block where the hull has been subtracted. Figure 13: The calculation domain obtained by the subtraction of the hull from a solid block. The surface of the hull was remeshed in STAR-CCM+. This way the surface quality was improved and a more suitable mesh was created to serve as the base for the generation of the volume mesh. Ideally, the surface mesh is triangulated with near equal sized triangles. The transition between areas with smaller and areas with larger sized elements should be smooth and gradual. Figures 14 and 15 show the remeshed hull surface for the simulations with the free surface and those with appendages and no free surface. 34 Figure 14: DTMB 5415 hull remeshed in STAR-CCM+ with a close up to the bow. Figure 15: Remeshed DTMB 5415 hull with appendages below the waterline, as used in the simulations, and close ups towards the bow and towards the stern. 35 4.2. Volume Mesh 4.2.1. Volume Mesh Generation and Description After the meshed hull was imported in STAR-CCM+ a block was built around it to provide, along with the hull, the boundaries for the creation of the volume mesh. The symmetry condition allowed for the block to be built, and the calculations made, on only half the hull. For the simulations without a free surface the DTMB 5415 hull was simulated at both model size and full scale. For the full scale simulations the model scale dimensions of both the hull and the surrounding block were multiplied by the scale factor 24.824. The dimensions of the block that defines the volume mesh region, for the simulations with and without a free surface are presented in Figures 16 and 17. Figure 16: Bare hull model simulations with a free surface volume mesh dimensions. 36 Figure 17: Appended hull model scale simulations volume mesh dimensions. The dimensions of the blocks were chosen with regard to the accuracy of the results. The intension was to limit these dimensions, as much as possible, as larger dimensions translated to more computational costs. The lengthwise dimension behind the ship stem is longer than that forward from the bow to capture the waves generated by the hull. The dimensions of the block for the simulations with a free surface were initially similar in size to the ones of the simulations without a free surface but, as pressure concentrations were found at the boundaries, they were gradually increased to better represent the fluid flow and improve the accuracy of the results. After the surface mesh was created and the physics models, described in Chapter 2, were selected, the next step was to generate the volume mesh. Volumetric controls were utilized to make the mesh more efficient and more effective. A volumetric control can be used to specify the mesh density for both surface and volume type meshes during mesh generation. Volumetric controls in STAR-CCM+ work in conjunction with volume shapes. A volume shape is a closed geometric figure that can be used to specify a volumetric control for surface or volume mesh refinement or coarsening during the meshing process. Volume shapes were "built" to encompass more computationally demanding and computationally important spaces of the mesh, e.g., the space around the bow and around the stem (Figure 18), as well as the space around the free surface up to the height of the generated waves. Through the volumetric controls, those spaces, covered by the volume shapes, were specified to have a more refined mesh. The driving factor behind the mesh refinement process was to maintain a balance between getting satisfactory results and keeping the computational costs as low as possible. The template growth rate controls the stepping from one cell size to the next within the core mesh. This was set to medium which gave a minimum of two equal sized cell layers per transition. 37 Figure 19 shows profile views of the volume meshes of the three types of simulations created (with free surface, with appendages and no free surface, and with no appendages and no free surface). Looking at these pictures it is easy to notice the regions where, due to the volumetric controls, the mesh has a specific and discrete density. The volume meshes of the simulations without free surface, with and without appendages, are quite similar, but the volume mesh of the simulations with the appendages is somewhat finer in order to effectively capture the details of the appendage surfaces. Figure 18: Examples of volume shapes used to control the mesh density. 38 Figure 19: From top to bottom, profile views of the volume meshes generated for the simulations with free surface and free motions, the simulations without free surface with appendages and without appendages. In the uppermost picture the bow of the hull looks to the left while in the other two to the right. 39 One of the attributes of the finite volume method implemented in STAR-CCM+ is that, unlike the finite difference method, it can be applied to mesh cells of any arbitrary shape. It does not demand a uniform, rectangular grid for computations. In other words, there is no demand for a structured mesh. This has given rise to the use of meshes with no regularity known as unstructured meshes. The benefit of using this type of mesh is that you have a lot of flexibility in shaping the mesh cells the way you like and putting them where you want in the physical space, thus making it easier to match the mesh cells with the boundary surfaces. This last characteristic is especially useful if there are complex geometries in a simulation. For this reason the surface and volume meshes generated for the simulations described in this work were unstructured. The meshing model used to generate the volume mesh in STAR-CCM+ gave a trimmed hexahedral cell shape based core mesh. One of the desirable attributes of this meshing model is that it does curvature and proximity refinement based upon surface cell size. It utilizes a template mesh constructed from hexahedral cells from which it cuts or trims the core mesh based on the starting input surface. Areas of curvature and close proximity are refined based upon the surface cell sizes. The resulting mesh is composed predominantly of hexahedral cells with trimmed cells next to the surface. Trimmed cells are polyhedral cells that can be described as hexahedral cells with one or more corners and/or edges cut off. 4.2.2. The Prism Layers One significant part of the volume meshing process is defining prismatic near wall layers. This is possible, with the volume meshing model selected, by adding the prism meshing model as part of the volume meshing process. The prism layer mesh usually resides next to wall boundaries in the volume mesh and, thus, models the boundary layer. It is required to accurately simulate the turbulent speed profile and predict the drag. Table 4 shows the number of prism layers and their thicknesses for the different simulations created. More prism layers were utilized for the simulations without free surface (Figure 20). When disregarding the appendages, the hull in all the simulations without a free surface had the same number and thickness of prism layers. The prism layer thickness of the appendages was defined relatively smaller than that of the hull, in accordance with their decreased boundary layer thickness. The relative thickness percentages, in the last row of Table 4, represent the percent of the prism layer thickness when divided by the waterline length. These values give the size of the prism layer thickness in relation to a main dimension of the hull to show how this thickness relatively decreases for 40 the full scale simulations. The relative thickness of the model boundary layers, when allowing for scale, was estimated to be about two and a half times that of the full scale layers (Figure 21). To estimate the change in the relative thickness of the boundary layer when transitioning to the full scale, the turbulent boundary layer on a flat plate thickness formula, based on the 1/7-power approximation for the velocity distribution, was used, 8 = 0.373xRx- 1/, Rx = Ux/v , where 6 is the boundary layer thickness, x is a characteristic dimension, and Rx is the Reynolds number of the flow [25]. According to this formula the turbulent boundary layer increases in thickness with distance downstream at a rate proportional to x/. Table 4: Details on the prismatic near wall layers generated in the simulations. Yes No No No No No Yes Yes Model Model Full Scale Model Full Scale - - - Hull Appendages Hull Appendages 5 8 8 8 8 8 8 0.010 0.035 0.347 0.035 0.010 0.347 0.0993 - 0.612% 0.244% 0.612% - 0.244% - No Figure 20: The boundary layer of the bare hull was modeled with 5 prism layers, in the simulations with the free surface (left), and 8 prism layers, in the simulations without free surface (right). 41 Figure 21: The prism layers of the model size simulations (left) were defined relatively thicker than those of the full scale simulations (right). The full scale simulations were generated from the model scale simulations using the DTMB 5415 hull scale factor 24.824 for the cases without free surface. These simulations are identical to the model scale simulations with the exception of the relative thickness of the prism layers and the fact that all dimensions are scaled by 24.824. More information on these simulations, as well as all the simulations without free surface, are given in Chapter 6. 4.3. Mesh Evaluation Two parameters were used to evaluate the generated mesh in each case, the wall y+ and the convective Courant number, both scalar and dimensionless. The convective Courant number can only be used with the implicit unsteady model, thus it was only used to validate the simulations with the free surface. Resolving the boundary layer demands a high mesh resolution in the near-wall region. The normalized wall distance parameter y+ is used to verify the mesh quality near the wall and within the boundary layer. It is defined as y+ = , , where Tw is the shear stress at the wall, p is the local density, y is the normal distance of the cell centroid from the wall, and v is the local kinematic viscosity. Since the potential for errors increases with large values of y+, when using the high-y+ wall treatment, it is generally prudent to aim for y+ values between 30 and 50. Some cells will inevitably have a small value of y+. That is acceptable. In general, values of y+ below 100 are considered acceptable. The lowy+ wall treatment requires the entire mesh to be fine enough for y+ to be approximately 1 or less. In this 42 work the all-y+ wall treatment is used because it is the most general and the values of y+ are intended to be below 100. The convective Courant number = V dx' , is a means to evaluate the mesh in conjunction with the chosen time step. It depends on the velocity V, the time step dt, and the interval length dx, which, in this case, is the length of the cells. It is the ratio of the time step and the time required for a fluid particle to travel the cell length with its local speed. It is typically calculated for each cell, and it gives an indication of how fast the fluid is moving through the computational cells. A finer mesh drives the Courant number at higher values, a smaller time step drives it at lower values, and a higher velocity drives it up. Implicit solvers are usually stable at maximum values in the range 10-100 locally, but with a mean value of about 1. The Courant-Friedrich-Lewy condition states that the Courant number should be less than or equal to unity. In general, Courant numbers set to values less than 1 are expected to give models that run faster and with greater stability. Figure 22 shows the wall y+ parameter values on the ship hull, for the simulations with the free surface and those with the appendages, while Figure 23 shows the values of the convective Courant number. The wall y+ parameter receives larger values at the forward part of the bulbous bow in the appended hull simulations, but all values, for both y+ and Courant number, are within acceptable limits. woofY* mV. Wai Y+ 0.000 20.00 40.00 606W 80.O 100.00 Figure 22: The wall y+ parameter values on the DTMB 5415 hull at Fr=0.41. On top is the hull used in the simulations with the free surface. 43 0 OM 02000 O00X0X.600M Convecnve Couront Number 040YY 0.80) L.00 Figure 23: The convective Courant number parameter values on the DTMB 5415 hull at Fr=0.41. 4.4. Boundary Conditions As has already been described, the boundary conditions drive the particular solution of the general equations that govern any flow. Furthermore, any numerical solution of the governing flow equations must give a compelling numerical representation of the proper boundary conditions. The boundary conditions applied at each boundary of the two types of simulations, with and without a free surface, are summarized in Table 5, while Figure 24 shows the location of the different boundaries, listed in Table 5, in the simulations with the free surface. The boundary locations for the simulations without free surface are analogous; the hull boundary includes the appendages where those exist. Table 5: Boundary conditions defined for each simulation. Ship hull Wall (no-slip) Wall (no-slip) Ship deck Wall (no-slip) Wall (no-slip) Block symmetry plane Symmetry Plane Symmetry Plane Block side plane Velocity Inlet Symmetry Plane Block bottom plane Velocity Inlet Symmetry Plane Block top plane Velocity Inlet Symmetry Plane Block inlet plane Velocity Inlet Velocity Inlet Block outlet plane Pressure Outlet Pressure Outlet 44 Region.Side .OJgSI ,ReO.DRA egmetry Figure 24: The boundaries are surfaces that completely surround and define the region. The no-slip wall boundary condition represents the proper physical condition for a viscous flow, where the relative velocity between the boundary surface and the fluid immediatelyat the surface is assumed to be zero. If the surface is stationary with the flow moving past it as in this case, then the velocity of the flow at the surface is zero. At the inlet we prescribe a constant velocity which corresponds to the Froude number at which we run the simulation. The direction of the velocity, i.e., the direction at which the flow moves, is that of the x-axis. In other words, it moves perpendicular to the inlet boundary surface and toward the outlet. The velocity inlet boundary condition is suitable for incompressible flows. It may be used in combination with a pressure outlet boundary at the outlet, as was done with these simulations. The pressure outlet boundary is a flow outlet boundary at which the pressure is specified. The pressure was specified to be the hydrostatic pressure of the flow with the reference pressure at the free surface being the atmospheric pressure at sea level. A symmetry plane boundary condition is better used when the physical geometry of interest and the expected pattern of the flow have mirror symmetry. Thus, a surface is defined as a symmetry plane boundary if it is the imaginary plane of symmetry in a simulation that would be physically symmetrical if modeled in its entirety. The solution obtained with a symmetry plane boundary is identical to the solution that would be obtained if the mesh was mirrored about the symmetry plane but in half the domain. The simulations presented in this paper had an imaginary plane of symmetry where this type of boundary condition was appropriate and utilized. The symmetry boundary condition can also be used to model zero-shear slip walls in viscous flows. It was found to work well and was also used at the side, bottom, and top boundaries of the simulations without a free surface. For the same boundaries in the simulations with a free surface the 45 velocity inlet boundary condition was found to provide better results. The velocity was that of the flow with the same magnitude and direction as at the inlet. 4.5. The 6-DOF Model The DFBI (Dynamic Fluid Body Interaction) module in STAR-CCM+ is used to simulate the motion of a rigid body in response to pressure and shear forces exerted by the fluid, as well as any additional forces defined by the user (gravity force in the current work). The resultant force and moment acting on the body due to all influences are calculated, and the governing equations of rigid body motion are solved to find the new position of the rigid body. As in the free surface simulations the hull of the ship is modeled free in heave and pitch motions the DFBI module is activated. The hull (deck included) of the ship is defined as a 6-DOF (Degrees Of Freedom) body on which the rigid body motion equations are solved. Some properties of the 6-DOF body that are defined in the simulation are its mass, the initial position of its center of mass, the diagonal components of the moments of inertia tensor, and the release time, which is the time before calculation of the body motion begins in order to allow some time for the fluid flow to initialize. The simulations without free surface did not need to use this model, as the hull was not allowed any free motion. 46 CHAPTER 5 THE DTMB 5415 BARE HULL MODEL WITH A FREE SURFACE This chapter presents the results from the simulations with a free surface. 5.1. Introduction The first set of simulations created for this study were those of the bare DTMB 5415 model hull with a free surface and free heave and pitch motions. This set included two simulations. The only difference between them was the flow velocity, which corresponded to the Froude numbers 0.38 and 0.41. During the runs the hull trim, sinkage, and drag were recorded and evaluated against the experimental ones [9]. Other results that were evaluated were the wave pattern on the free surface and the wave profile along the hull. 5.2. Trim, Sinkage, and Resistance Results In Gothenberg in 2010 at the Workshop on Numerical Ship Hydrodynamics [20] many researchers came together and presented their results on total resistance, sinkage, and trim using various CFD codes and grid densities. At Froude number Fr-0.41 the average error for the total resistance over experimental findings was 4.316%, for the sinkage 12.294%, and for the trim 11.472%. All these errors corresponding to the use of the finer grid density from those for which results were presented. Table 6 shows the results of the computations made for the present study at Fr-0.41, Fr=0.38, Fr-0.36, and Fr-0.33 compared with the experimental (Exp.) data from Olivieri et al. (2001) [9]. At Fr-0.41, with the exception of the trim, the results compare very well with those from Gothenberg. The error percentage is given by: (Exp.-CFD)/Exp. x 100. The negative values of the trim angle correspond to the ship trimmed by the stem. With the exception of the resistance, at lower speeds the agreement of the calculations from the CFD simulations with the experimental data worsens, and, as we can see from the results in Table 6, at Fr=0.36 and Fr-0.33 the CFD values for the sinkage and, especially, the trim deviate significantly from the experimental data. 47 In an effort to improve the CFD results at a wider range of velocities a new mesh was created. The volume mesh close to the hull was refined, but the volume mesh region transverse and vertical dimensions were decreased resulting to a number of cells that was somewhat decreased compared to the original volume mesh. Simulations with this new mesh were run at Fr--0.41 and Fr--0.28. The results are given in Table 7. The resistance values at both speeds show a very good and improved, in comparison with the results from the original mesh, agreement with the experimental data. The agreement of the results for the sinkage and trim is not as good though and a significant deviation of the CFD values from the experimental data can still be observed. Figures 25-27 are plots of the towing tank experimental data for total resistance, trim angle, and sinkage together with the CFD results shown in Tables 6 and 7. They give a graphic representation of the agreement between the experimental and CFD values. A fourth order polynomial line has been fit through the experimental data points in all of these graphs. The best match between experimental and computational values is observed with the total resistance and an outlier is observed in the sinkage graph with the sinkage value at Fr--0.41 using the newer mesh. Figure 28 shows the time histories of the total resistance, trim angle, and sinkage as recorded in STAR-CCM+ at Fr-0.38 for the 84 seconds that the simulation was run. The oscillatory behavior of the graphs due to the existence of the free surface (i.e. waves) can be observed. Due to this behavior of the graphs the values in Tables 6 and 7 are averages of the oscillatory data. The resistance values given in this figure have to be multiplied by 2 as the symmetry condition was used and only half the ship was simulated as has previously been described. Further time plots of the resistance, trim, and sinkage at Fr-0.41, from the simulations with the newer mesh, and from the simulations without the free surface are given in Appendix C. The CFD code can give not only the total resistance but also the frictional and residuary resistance to which the total resistance is subdivided. Figure 29 shows all these resistances at Fr-0.38. The oscillations of the pressure resistance curve can be observed while the frictional resistance is relatively stable. The oscillations of the pressure resistance values that can be attributed to the effect of the free surface are transmitted to the values of the total resistance. 48 Table 6: Experimental vs. CFD resistance, trim, and sinkage data. knots 23.94 26.11 27.56 29.74 m/s 12.31 13.43 14.18 15.30 m/s 2.471 2.695 2.847 3.071 Fr 0.33 0.36 0.38 0.41 Exp. 0.097 0.047 -0.06 -0.421 CFD 0.070 0.068 -0.0603 -0.260 E%Exp. 27.8 -44.7 -0.50 38.2 CFD 0.014 0.0162 0.0199 0.0265 Exp. 0.0164 0.0198 0.0217 0.0269 E%Exp. 14.7 17.9 8.5 1.4 Exp. 69.2 88.2 108.9 152.6 CFD 67 81.7 100 160 E%Exp. 3.2 7.4 8.2 -4.8 CFD 44 157 E%Exp. 2.6 -2.8 Table 7: Experimental vs. CFD resistance, trim, and sinkage data using a different mesh. knots 20.31 29.74 m/s 10.45 15.30 m/s 2.097 3.071 Fr 0.28 0.41 250 Exp. 0.108 -0.421 CFD 0.07 -0.37 E%Exp. 35.2 12.1 Exp. 0.0104 0.0269 CFD 0.0085 0.038 E%Exp. 18.3 -41.3 Exp. 45.1 152.6 Total Resistance Experimental vs CFD results 4 200 150 IGO 100 n Experimental Results (INSEAN) 0 CFD results * CFD results (newer mesh) -Poly. (Experimental Results (INSEAN)) 50 0 4,) qzo 4) Q, IV 4, Model Speed (m/s) Figure 25: Total resistance experimental data from Olivieri et al. [91 plotted with the CFD results. 49 Trim Angle Experimental vs CFD results 0.2 0 il -0.2 U Experimental Results (INSEAN) -0.4 CFD Results CFD Results (newer mesh) M -0.6 Poly. (Experimental Results (INSEAN)) -0.8 -1.2 0 0.05 0.1 0.15 0.2 0.35 0.3 0.25 0.4 0.45 0.5 Froude Number (Fr) Figure 26: Trim angle experimental data from Olivieri et al. [91 plotted with the CFD results. 0.04 Sinkage Experimental vs CFD Results 0.035 0.03 mExperimental Results (INSEAN) * CFD Results 0.025 a + CFD Results (newer mesh) -Poly. a (Experimental Results (INSEAN)) 0 0.02 no 0 0.015 0 0.01 U U 0.005 0 0 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Froude Number (Fr) -0.005 Figure 27: Sinkage experimental data from Olivieri et al. [91 plotted with the CFD results. 50 0.5 Rt Plot 20 30 40 50 Time (sec) - 60 70 80 Rt Y Rotation Plot 0. 11- 0.1 0. 09 0. 05 so. 07- 0. 05 0. 04 03 \ 0.02 0.0 1 0 20 30 40 50 60 Time (sec) -Trim Angle 51 70 80 Z Translation Plot Time (sec) -Sinkage Figure 28: Time histories of resistance, trim, and sinkage as generated in CFD code STAR-CCM+ at Fr=0.38. Fridional, Pressure, and Total Resistance Plot 00 30- 20 30 40 53 Time (sec) 60 70 80 Rp -Rf-Rt Figure 29: The CFD code also calculated the frictional and pressure resistances that add up to the total resistance. 52 5.3. Wave Pattern and Wave Profile at Fr=0.41 Olivieri et al. [9] provide figures of the wave pattern and the wave profile along the hull at Fr-0.41 from their towing tank measurements. The top half of Figure 30 shows the experimental result from Olivieri et al. for the wave pattern on the free surface at Fr=0.41, while the bottom half is the wave pattern that was derived from the CFD code at the same Froude number. We notice that there is a good agreement with the CFD code capturing the general form of the wave pattern from the towing tank tests. Number 1, on Figure 28, represents the distance of one ship length. Figure 31 shows the experimental wave profile along the hull from Olivieri et al. in comparison with the CFD wave profile at Fr-0.41. Distances on both axes of the graphs are divided by the waterline length of the hull. Figure 30: Experimental (top) and CFD wave pattern at Fr=0.41. 53 AM3 0.025 402 0.015 OI -nas 1) 0.1 0.2 03 (4 V.5 Oh Q7 QA Q9 + Wave Profile DTMB 5415, Fr=0.41 (CFD) --------------------------------------- ----------------------------------- ------------------------ ------------------------------------------------ I I I I 0.2 03 (CA 0.5 xt. Figure 31: Experimental (top) and CFD wave profile along the hull at Fr=0.41. In Figure 31 the experimental wave profile curves represent towing tank tests from three institutes: DTMB (A), INSEAN (B), and IIHR (C). The wave profile from the CFD code follows the same pattern as the experimental profile, with the lowest point being at about the middle of the ship, the highest point located at the bow a little after the forward perpendicular, and the wave rising again at towards the stem but not as much as at the bow. The main difference is that the wave at the bow in the CFD code rises considerably less than in the towing tank tests, with the CFD giving a maximum value of about 0.022 and the experimental result with the lowest of the highest wave height values being 0.028. As previously mentioned the values for the wave height have been divided by the ship length at waterline. 54 In the towing tank report from Olivieri et al. [9] there is a photo of the bow wave at Fr-0.28. The visual comparison between the CFD and towing tank bow waves in Figure 32 shows a similar wave length and crest height. The light refraction at the water surface, in the experiment, creates a deformed underwater image of the ship. Figure 32: Bow wave from the towing tank report (left) and from the CFD simulation at Fr=0.28. 5.4. Remarks The goal of this study was to find a CFD simulation that would give values for all measured quantities (trim angle, sinkage, and resistance) in the 5 percent range when compared to the experimental data. A lot of different meshes, different time steps, different volume mesh domain dimensions, and even some different boundary conditions were tried. While it was possible to get two of these quantities in the 5 percent range (see results previously presented in this Chapter) there was always one value that would deviate from this range. In this study the resistance computations were the most consistently accurate even in cases where the trim and sinkage were very different from the experimental data. While the focus in the present study was at higher speeds (Fr-0.38 and 0.41) some efforts were made at lower speeds as well, e.g. Fr=O. 16 and Fr-0.28. In general the results, at the lower speeds, were not as good. One reason is that for smaller values of resistance, trim, and sinkage at lower speeds more accurate numerical predictions relative to the higher speeds are required to achieve 5 percent accuracy compared to the experimental data. Another reason is that the exact same simulations that were used for the higher speeds were also used for the lower speeds, changing only the flow velocity. This 55 study showed that in order to predict the resistance, trim, and sinkage at the lower speeds with a similar accuracy as at the higher speeds adjustments would need to be made to at least the volume mesh and maybe the time step and the wall function. When a newer mesh was created refined at the area surrounding the hull combined with the use of a smaller time step and a smaller volume mesh domain the resistance at Fr--0.28 was predicted very well. 56 CHAPTER 6 THE DTMB 5415 HULL MODEL WITH APPENDAGES The results from the simulations without a free surface are presented. The appendage resistance is analyzed at model scale and full scale simulations of the DTMB 5415 hull. 6.1. Introduction In addition to the bare hull simulations, where the numerical computations of resistance, trim, and sinkage were evaluated against experimental data [9], another set of simulations was prepared that attempted to examine the effect of the appendages on the resistance and to evaluate the numerical results with respect to available experimental data [10], as has already been seen in Chapter 2. The appended DTMB 5415 hull model presented in Figure 33 was used. Figure 33: DTMB 5415 hull model with appendages. Borda ran towing tank tests using the DTMB 5415-1 model and derived the resistance of the appended hull at various speeds. He then extrapolated the results to the full scale using the ITTC 1957 correlation frictional line in combination with a correlation allowance (CA) of 0.0004. The full scale values of the effective power, the frictional power, and the residuary resistance coefficient are the ones that he gave in his paper. For the full scale predictions the ship (at the full scale) was assumed to be operating in calm, deep, salt water at 590 F (15 C). During his experiments the model was free to trim and heave, but restrained in yaw. Instead of running the simulations with the modeled hull free to trim and heave, as we did in the first set of simulations, the hull was now fixed in space (with the flow moving around it). Only the submerged part of the hull, below the waterline, was simulated; in other words, there was no free 57 surface. The assumption made was that the appendages were sufficiently below the free surface so that their resistance was not significantly affected by the wave induced flow field. Nor was the appendage resistance significantly affected by the hull motions in heave and pitch. Thus, we could achieve simulations that would run much faster. By running simulations for the hull with and without appendages the additional resistance of the appendages could be estimated. The appendage resistance could then be added to existing data for the bare hull resistance [9] and the total appended hull resistance found. This data could now be converted to represent the full scale ship, in the same manner that Borda did, and evaluated against the experimental data [10]. In evaluating the results of this set of simulations in this way, the purpose was to see how well they could give the effect of the appendages to the bare hull resistance. If the aim was a more comprehensive CFD numerical approach, independent of experimental data, instead of using data from Olivieri et al. (2001) [9], data from numerical simulations, as those presented in Chapter 5, with free surface and the bare hull free to heave and pitch, could be used instead, or, even more comprehensively, simulations could be ran with the appended model floating on a free surface allowed to move in heave and pitch. The last approach, though, would be, computationally, the most expensive. These simulations were run at three velocities and with corresponding Froude numbers. 6.2. Simulations at Model Scale As already mentioned, the appendage resistance derived from CFD formulations was added to bare hull resistance experimental data [9]. The appended model hull resistance was found and converted to full scale using standard towing tank procedures (ITTC 1978 prediction method) similar to the procedure used by Borda. This procedure assumes the division of the total resistance to a frictional and a residuary, or pressure, part similar to what was previously described in Chapter 2, section 2.2. An example of this procedure at Fr-0.41 is given in Appendix A. Table 8 summarizes the CFD numerical results at three different speeds and gives a comparison with the experimental data. The percentage difference between the two is given by Diff = Exp-CFD X Exp 100. Figures 34 and 35 give a graphical representation of the agreement between numerical and experimental results for the effective and frictional power, and for the residuary resistance coefficient respectively. 58 Table 8: Numerical vs. experimental appended hull resistance data. (knots) 29.74 23.94 11.61 (m/s) 15.3 12.31 5.97 (m/s) 3.071 2.471 1.197 Exp. 35190 12963 1186 - 0.41 0.33 0.16 CFD 35653 12687 1190 - 5 --------------- - Diff. -1.3% 2.1% -0.35% Exp. 4.135 2.357 1.369 CFD 4.228 2.269 1.401 - re kexpenmentj - - Pf (experiment) 20 Voloc~ty [Knotj CFD 10735 5715 699 Exp. 10746 5717 703 *------------------pe-(CFD)- -- * 15 Diff. -2.25% 3.73% -2.34% Diff. 0.11% 0.04% 0.5% ------ Pf(CFD) 25 3 30 Figure 34: Lines of the experimental effective and frictional power at the range of 10-32 knots vs. the numerical results at three speeds. Curve of reitac data from~ experiment233 [Borda (1984)] vs CFD results -----------------------------------...------3-----5 --- - ---------------- Experiment ----- 15-------------- CFD 3 -- ---- 15 - 20 - -- 25 --- 30 35 VelocRy [Knot] Figure 35: Lines of the experimental residuary resistance coefficient at the range of 10-32 knots vs. the numerical results at three speeds. 59 6.3. Simulations at Full Scale In addition to the model scale, these simulations were run at the full scale. As has been discussed in Chapter 4, the procedure to get from the model scale simulations to the full scale was simply scaling the dimensions using the proper scale factor, adjusting the prism layer thickness, and using the proper full scale flow velocity. Results for the full scale simulations could then be compared to the model scale results. Some evaluation of and conclusions about their differences could also be made. Subtracting the results of the full scale CFD simulations with appendages from those without appendages we could get the full scale appendage total resistance at each speed. Using standard towing tank procedures, as described in Appendix A, the bare hull model resistance experimental data [9] were converted to the full scale. Then the full scale total resistance of the bare hull was added to the full scale appendage total resistance from the simulations and, in this way, we could get the full scale total resistance to tow the appended vessel hull in calm water. This last result was multiplied by the vessel speed; the effective power was found and compared with what was found from the model scale simulations and the existing experimental data [10]. All these effective power data are given in Table 9. Table 9: Numerical vs. experimental appended hull resistance data including the values from the full scale simulations. Vsipl V'ship (knots) 29.74 23.94 11.61 (m/s) 15.3 12.31 5.97 Vmlodel (m/s) 3.071 2.471 1.197 Fr 0.41 0.33 0.16 Exp. 35190 12963 1186 CFD (Model) 35653 12687 1190 Effective Power Pe I NN] CFD (Full scale) Diff. (Model) 34344 -1.32% 12026 2.13% 1105 -0.35% Diff. (F. Scale) 2.4% 7.2% 6.8% Looking at the data in Table 9, one observes that the results from the use of the full scale simulations do not agree as well with the experimental data as those from the model scale simulations. This is expected because the procedure to derive the effective power from the full scale simulations is less compatible with the procedure to obtain the experimental data than the procedure used for the model scale simulations. More importantly, the effective power estimated from the full scale simulations is consistently smaller in magnitude than that of both the experimental tests and the model scale simulations. This also was expected. As the ship moves through the water the Reynolds number that its hull encounters is different from the Reynolds number that its appendages encounter. This is primarily due to the fact that the characteristic length of each appendage changes and then due to the fact that the local fluid velocity at each appendage varies which can be considered a second order effect. The method used in the present study and by Borda to convert the appended hull drag to the full scale did not account for this variation 60 in the Reynolds number of each appendage; as a result, it tends to overestimate the full scale ship resistance. Using directly computed full scale appendage resistance values helps "circumvent" this overestimation to some extent, and it would be expected that smaller resistance values would be obtained. To demonstrate how this different Reynolds number that each of the appendages comes across affects the total resistance outcome when extrapolated to the full scale with the ITTC 1978 method Figure 36 shows the ITTC 1957 frictional line with five points drawn on it. Each of the points corresponds to one of the appendages or the bare hull when the characteristic length of each of them is used to calculate the Reynolds number at Fr=0.41. It can be seen that as the Reynolds number gets smaller values the frictional coefficient gets larger values. Given that the appendages operate at smaller Reynolds numbers, when the extrapolation to the full scale is made, the residuary coefficient will get a smaller value which leads to a smaller total resistance prediction at the full scale compared to using the bare hull Reynolds number for all of the appendages as well. A logarithmic scale was used on the horizontal axis of the diagram in Figure 36. Calculations of the full scale ship resistance were made with each appendage accounted for separately, with its Reynolds number adjusted to its specific characteristic length during the drag conversion from model scale. The results of these calculations are given in Table 10. As expected these values are smaller than the values given in Table 8; but they are also significantly smaller than the values taken when the full scale appendages were used. This could be explained, to some extent, by the fact that when extrapolating to the full scale the change in local velocity for each appendage was not accounted for. Table 10: Full scale effective power calculations when the Reynolds number of each appendage was accounted for. (knots) (m/s) (m/s) - CFD 29.74 23.94 11.61 15.3 12.31 5.97 3.071 2.471 1.197 0.41 0.33 0.16 31569 10954 991 61 2.OOE+01 - ITTC'57 -U-Bare Hull 1.80E+01 -*- Bilge Keels -4-Shafts 1.60E+01 -U-Struts -0- 1.40E+01 Rudders 1.20E+01 1.OOE+01 8.00E+00 '4 '4 6.OOE+00 *41 4b 4.OOE+00 2.OOE+00 O.OOE+00 1.OOE+04 1.OOE+05 1.OOE+06 1.00E+07 .OOE+08 1.OOE+09 1.OOE+10 1.00E+11 RnL Figure 36: Difference in the frictional resistance coefficient value of each appendage when its Reynolds number is calculated separately at Fr=0.41. 6.4. Other Results In what follows some further results and results analysis from this set of simulations without free surface are presented. These contain further evaluation of the simulation results, and also highlight the possibilities of using CFD for the analysis of the flow around and the drag exerted on a ship hull. All computation results are presented at both model and full ship scale for the three evaluated velocities. Table 11 shows the total (frictional and residuary) resistance results of the appended hull simulations analytically. Results are given separately for each appendage; the sum of the appendages or total appendage resistance; the hull with the appendages; and the hull without the appendages. Table 12, on the other hand, gives the bare hull total resistance results from the bare hull simulations. It also gives the result of the subtraction of the bare hull resistance, from the simulations without appendages, from 62 the appended hull resistance, from the simulations with the appendages. This is the standard way of estimating the appendage added resistance to the bare hull. Figure 37 shows the total resistance of each appendage and of all the appendages together, calculated either from the difference between appended and bare hull resistance (Table 12) or directly from the appended hull simulations (Table 11), as a percentage of the bare hull total resistance (Table 12). Figure 38 shows how the total resistance, for each model or full scale ship wetted surface component, is distributed between frictional and residuary (or pressure) resistance at 0.41 Froude number. The distribution is similar for the other Froude numbers. The hull without appendages corresponds to the "Appended Hull - Appendages" total resistance values in Table 11, while the hull with appendages corresponds to the "Appended Hull" total resistance values in Table 11. Appendix B contains the charts of the total resistance distribution for the other Froude numbers, and a table with the analytical values of the frictional and residuary components of the total resistance. It is pointed out that all these values of the resistance and resistance percentages correspond to CFD simulations without a free surface. If a free surface existed then the pressure resistance, especially for the bare hull, would be larger. The frictional resistance would not be significantly affected. To demonstrate this Figure 39 shows what percentage of the total resistance is frictional drag and what is pressure drag when values from the CFD simulations with a free surface are used at Fr-0.41. The pressure drag is now larger than the frictional drag but this is because of the large speed; at lower speeds the frictional drag is larger than the pressure drag but the pressure drag with the effect of the free surface has significantly larger values than when a free surface does not exist. Table 13 gives the values of the frictional and pressure drag from the simulations with and without a free surface at Fr-0.41 to show that the frictional drag does not change much when a free surface exists whereas the pressure drag gets a much larger value. Table 11: Drag distribution among the hull and its appendages from the appended hull simulations. Model 1.197/ UJ.1O U.545b Model 2.471 0.33 3.162 2.742 2.976 3.04 11.92 60.16 48.24 Model 3.071 0.41 4.764 4.112 4.422 4.552 17.85 90.7 72.84 Ship 5.971 0.16 8668 5898 6584 7256 28406 141394 112988 Ship 12.315 0.33 35362 23476 25928 29032 113800 564776 450978 Ship 15.3 0.41 53860 35540 39120 44040 172560 855800 683240 63 Table 12: Total resistance results from the bare hull simulations. The difference between appended and bare hull drag is also given. Total Resistanc9e (N) Scale V (m1/s) Fr- Model 1.197 0.16 12 3.08 Model 2.471 0.33 48.23 11.93 Model 3.071 0.41 72.42 18.28 Ship 5.971 0.16 105691 35702 Ship 12.315 0.33 417722 147054 Ship 15.3 0.41 631068 224732 BareC Hull 64 Apene-BHll 25% t- - - - - --- 20% 15% 10% 5% 15% 10% 5% 0#% 0% 15%____ 15% 0% 0;--1 10% 5% 0% - 2501 c-<~ 10~ 10% 5% tI 04 10/ 40% _______________ for the hull total resistance. The percentage bare the of percentage a as given Figure 37: Appendage total resistance and from the difference computations directly on the appendages from both given is full set of the appendages total resistance. between appended and bare hull 65 120% Model Fr=0.41 100% 80% 60% 40% * Pressure/Total Frictional/Total 20% 0% 0%; 120% * Full Scale Fr=0.41 100% 80% 60% 40% - " Pressure/Total " Frictional/Total 20% 0% Figure 38: Percentage subdivision of the total resistance in frictional and pressure, or residuary, drag for the model and full scale hull. 66 120% Model Fr=0.41 with Free Surface 100% i 80% 60% - Pressure/Total _---a U Frictional/Total 40% - 20% 0% Bare Hull Figure 39: Percentage subdivision of the total resistance in frictional and pressure drag for the model scale hull when a free surface exists at Fr=0.41. Fr=0.41Resistanice (N) Free Surface Frictional Pressure Total No 64.58 7.84 72.42 Yes 65.2 91.8 157 Table 13: Subdivision of the total resistance in frictional and pressure drag for the model scale hull with and without a free surface at Fr=0.41. So me comments on the above graphs and tables are: 1. The bare hull total resistance (Table 12) has, with the exception of the model scale result at Fr--O. 16, smaller values than the result of the difference between appended hull and appendages resistance in the simulations with appendages (Table 11). The difference can be attributed to interaction effects between the bare hull and the appendages. This is especially important in the full scale simulations. 2. In accordance with comment (1), the appendages total resistance is found to be greater when calculated from the difference between appended and bare hull (Table 12) rather than directly from the simulations with appendages (Table 11). Again, the difference between the two is the resistance due to interaction effects between the appendages and the bare hull. These effects become more important as the flow velocity increases. 3. The struts consistently give the smallest total resistance, while the rudders, which include the rudder shoes, give the largest. The shafting gives the second largest total resistance with the 67 exception, again, of the model scale result at Fr-0. 16 where the bilge keels give the second largest total resistance. The struts have the smaller wetted surface area whereas the shafting has the largest. The rudders are thicker than the struts. 4. The rudders and especially the struts demonstrate larger pressure resistance than frictional resistance. This could be due to not being perfectly aligned with the local flow. The thickness of the rudders could also be a factor in its increased pressure resistance. Figure 40 shows the streamlines at the struts and the rudder on a horizontal section somewhat below the water surface at Fr-0.33. From this figure it is obvious that the struts are not well aligned not only from the form of the streamlines but also from the velocity distribution around the struts which is not symmetric. Figure 42 shows that the pressure coefficient has large values at the leading edges of the rudders and especially the struts, in agreement with the large percentage of pressure resistance. 5. In contrast to the rudders and struts, the total resistance of the bilge keels is almost completely frictional resistance. The bare hull also demonstrates a far larger frictional resistance component, but as has been shown this is due to the lack of a free surface. 6. The appendage resistance, when the bare hull resistance is subtracted from the appended hull resistance, demonstrates a larger pressure resistance component. This is more obvious at the full scale; it is due to the relatively large value of the frictional resistance for the bare hull, in combination with the relatively large value of the pressure resistance for the appended hull. When the appendage resistance is estimated directly from the computations in the simulations with the appendages it is found that the frictional resistance is larger than the pressure resistance. 68 .A.4 U Figure 40: Streamlines around the struts and at the rudder at Fr=0.33. The velocity magnitude on a horizontal section of the flow is also shown. Figure 41 shows the frictional resistance coefficient of each appendage and the bare hull, as computed from the simulations, drawn against the Reynolds number and compared with the ITTC 1957 frictional line. The dotted line on the left is the Blasius solution for laminar flow. At a given velocity as the characteristic length of each appendage changes so does the Reynolds number that it encounters. The larger the velocity and the characteristic length of an appendage, the larger the Reynolds number becomes. Subsequently, the results at the Full Scale (FS) are located at the right side of the diagram. The ITTC 1957 frictional line is given by Cf = Cf = 1 328 - . 0.075 0 2 (logi 0 Rn-2) . The Blasius solution for laminar flow is To get the frictional resistance coefficient from the CFD simulations we used the computed frictional resistance Rf and the formula Cf = Rf 1 2, where p is the density of the water, Sw is the -pSwV2 wetter area and V is the velocity. The CFD results are expected to be bounded between the ITTC and Blasius lines and to follow the slope of these lines. A logarithmic scale was used on the horizontal axis of the diagram in Figure 41. The frictional resistance coefficients calculated through the CFD simulations are given in Appendix B along with the corresponding Reynolds numbers. 69 2.OOE+01 - 1.80E+01 -ITTC'57 -U-Bare Hull -- r-Bilge Keels 1.60E+01 -)*-Shafts -*K-Struts 1.40E+01 0 0 0 1.00E+01 '-I - - 1.20E+01 -- Rudders with Shoes - Bare Hull FS -Bilge Keels FS - * 44~ U -- 8.00E+00 - Shafts FS - Struts FS -Rudders 6.OOE+00 -- -- with Shoes FS Blasius 4.OOE+00 2.OE+00 O.OOE+00 1.OOE+04 1.OOE+05 1.OOE+06 1.OOE+07 1.OOE+08 ~-~-.- _ 1.OOE+09 1.OOE+10 1.OOE+11 RnL Figure 41: Matching of the frictional resistance coefficient for each appendage and the bare hull, at the corresponding Reynolds number, calculated from the CFD simulations with the ITTC 1957 frictional line and the Blasius solution for laminar flow. There is, in general, a good agreement between the slopes of the lines formed from the results of the CFD simulations at three speeds and the ITTC and Blasius lines. The agreement is better for the bare hull, the bilge keels, and the shafting, all of which have a predominant frictional resistance component. The struts and rudders, with the large pressure resistance components, are to the left of the diagram, due to their smaller characteristic lengths and corresponding Reynolds number values. The struts, for the model scale simulations, are even transcending a bit to the left side of the Blasius line. The bare hull model scale CFD results are also transcending both sides of the ITTC line. The results of the simulations at Fr-0.16, especially at the model scale, seem to be the cause of some inconsistencies for the final resistance results and analysis. As has already been seen at the model scale Fr-0.16, not only does the bare hull have a greater total resistance than the appended hull without the appendages in the simulations with the appendages and the bilge keels have a larger total resistance than the shafting, but also the results for the frictional resistance coefficient at model 70 scale Fr=0. 16 seem to be causing the slopes of the lines formed by the CFD results to deviate from the slopes of the Blasius and ITTC lines and even cross them in two instances. It may be that a finer mesh was required to more precisely capture the flow effects at this low speed or that a different wall function would be more suitable at this low Reynold number to effectively capture the effect of the boundary layer. Figure 42 below shows the skin friction and pressure coefficients drawn on the appended DTMB 5415 hull at model scale and at Fr-0.41. The areas with larger values of the coefficients (drawn with a red or closer to red color) have a greater contribution to the frictional and pressure resistance respectively. It can be seen that the pressure coefficient at the leading edges of the struts and rudders receives large values as has previously been described. Table 14 shows some values of the appendage resistance estimated with the use of empirical methods which have been described in Chapter 2. The values estimated by the CFD simulations are also given for the purpose of comparison. Appendage resistance results are given from the use of two different empirical methods, the Holtrop and Mennen method and one other as seen in Table 14. The only exception is the shafting full scale drag which was not estimated empirically with a second method since the Hoerner equation for estimating the shaft drag described in Chapter 2 is not valid for the Reynolds number at which it operates. 71 0.00000 Skin Friction Coefficient 40.000 60.000 20000 O.00rw 0 20 O0 X0 2.GO SktnFrction Coefficient 40.000 60,000 Skin Friction Coefficent 40,00 60.U 80.000 80000 100.00 10.00 800L 1X0.0 rV -0.60000 -0.36000 -0. 60X -0.60 -0.36000 Pressure Coefficient -0. 120M 0 12000 Pressure Coefficient -0.120 0. 12w 0.3600 0.36X 0.60000 0.6X0 Pressure Coefficient -0.3600 -0.12X 0 1200 0.36000 0.6000 Figure 42: The skin friction and pressure coefficients depicted on the appended hull model at Fr=0.41. 72 Table 14: Some appendage resistance predictions with empirical methods compared to the CFD computations. Type Append. Fr Appendages Meth Bilge FS Keels FS Rudders FS Shafting Struts Rudders B. Keels Struts FS Shafting FS Holtrop 0.16 3.714 28956 0.922 7186 0.813 6341 0.388 3025 1.591 12403 and 0.33 13.948 112609 3.461 27947 3.054 24659 1.457 11766 5.974 48237 Mennen 0.41 20.770 169343 5.155 42027 4.549 37083 2.17 17694 8.898 72539 Peck 0.16 - - 1.473 10780 - - - - - - 0.33 - - 5.459 41656 - - - - - - 0.41 - - 8.101 62529 - - - - - - Peck and 0.16 - - - - 1.460 11350 0.712 3640 - - Hoerner 0.33 - - - - 5.362 45004 2.436 13612 - - 0.41 - - - - 7.952 68112 3.536 20248 - - Hoerner, 0.16 - - - - - - - - 0.772 - Kirkman 0.33 - - - - - - - - 2.730 - and 0.41 - - - - - - - - 4 - 0.16 3.140 28406 0.812 6584 0.828 8668 0.71 5898 0.796 7256 0.33 11.92 113800 2.976 25928 3.162 35362 2.742 23476 3.040 29032 0.41 17.85 172560 4.422 39120 4.764 53860 4.112 35540 4.552 44040 Kloetzi CFD The appendages resistance CFD values in Table 14 are given as a result of the computations in the simulations with the appendages directly. If the difference with the bare hull simulations was used we would see significantly larger values at the full scale results. The values for the bilge keels predicted by the empirical methods are larger than those predicted by the CFD code, with the Peck method giving significantly larger values than both the CFD code and the other empirical method by Holtrop and Mennen. The predictions given by the empirical methods for the rudder drag contradict each other as one gives smaller values and the other significantly larger than the CFD values; the predictions at the model scale from the Holtrop and Mennen method are very close to the values from the CFD code. The values for the struts from the CFD code are significantly larger than both empirical methods, with the exception of the model scale results using Peck and Hoerner empirical equations which are fairly close to the CFD values. As previously described the bad alignment of the struts in the CFD simulations can explain the relatively large values of the strut resistance given by the CFD code. The shafting resistance values given by the Holtrop and Mennen method are much larger than the CFD code calculations; on the contrary other empirical method gives smaller predictions at the model scale which are quite close to the CFD values. The CFD values of the appendage drag, given in Table 14, do not capture the effect of the 73 interaction between appendage and hull due to their connection. This can explain, to some extent, the cases where the empirical methods predict larger values. From the results of the few calculations of the appendage drag that were made using empirical methods in Table 14 there is a visible large scatter between the values given by different empirical formulations as well as with those estimated by the CFD simulations. Using CFD simulations at the full scale to estimate the appendage resistance there is no need for any extrapolations, as it would be necessary if model scale simulations were used, and the effects of particular design features and surface characteristics of each appendage can better be captured compared to using general and approximate empirical formulations. As a result CFD simulations at the full scale are recommended to be the method with the best potential to accurately predict the appendage resistance of a ship. 74 CHAPTER 7 CONCLUSIONS / RECOMMENDATIONS 7.1. Conclusions The present study shows how it is possible to numerically simulate the fully turbulent free surface flow around a destroyer-like hull form. The overall accuracy of the resistance results obtained with the free running model is satisfactory for design purposes, showing an average difference of about 6% with peaks up to 8% at some speeds, while for the two runs using a finer mesh at the region close to the hull combined with a smaller volume mesh domain the difference was kept below 3%. Some major advantages over model test techniques come from the possibility to accurately estimate the added drag of the appendages in both model and full scale, overcoming the historic problem of the error associated with extrapolating the model scale measurements to full scale. Another interesting possibility offered by CFD simulations is the appendage alignment study which can be done by the designer in a more effective way in respect to model test practice. 7.2. Recommendations for Future Work This study focused on predicting the bare hull resistance of a ship's model; then simulations without a free surface were used to estimate the resistance of the appendages with a minimized computational cost. The next step would be to simulate the hull with the appendages with a free surface and free attitude. It is then possible to estimate the appended hull resistance directly from these simulations without the need to add separate calculations for the appendages and the bare hull. The results could be compared with the ones presented in this study. Future work may also focus on modeling the ship to be self-propelled. Running self-propelled simulations of the ship at the full scale predictions of the thrust deduction could be made and the effect of the propulsor on the flow and the drag of the ship could be studied. Continuation of this study is also encouraged in order to investigate 75 the effect of different mesh types and different numerical solver techniques on the accuracy of the obtained results. 76 REFERENCES [1] Rouse, Hunter, and Simon Ince: History of Hydraulics, Iowa Institute of Hydraulic Research, Ames, Iowa 1957. [2] Tokaty, G.A.: A History and Philosophy of Fluid Mechanics, G. T. Foulis, Henly-on-Thames, England, 1971. [3] Anderson, J. D., Jr.: ComputationalFluid Dynamics: The Basics with Applications, McGrawHill, New York, 1995. [4] Ceruzzi, P. E.: Beyond the Limits, The MIT Press, 1989. [5] Sahoo, P.K., Doctors L.J., and Renilson M.R.: Theoretical and experimental investigation of resistance of high-speed round-bilge hullforms, Fifth International Conference on High-Speed Sea Transportation (FAST http://academic.amc.edu.au/~psahoo/Research/FAST99 [6] '99), 1-12, 1999, paper.pdf. ITTC, 1999, "Report of the Resistance Committee," Proceedings International Towing Tank Conference, Seoul, Korea & Shanghai, China, 5-11 September. [7] G2K, 2000, http://www.iihr.uiowa.edu/gothenburg2000. [8] Stern, F., Longo, J., Penna, R., Olivieri, A., Ratcliffe, T., and Coleman H.: International Collaborationon Benchmark CFD Validation Datafor Surface Combatant DTMB Model 5415, Twenty-third Symposium on Naval Hydrodynamics, 401-420, Val de Reuil 2000. [9] Olivieri, A., Pistani, F., Avanzini, A., Stern, F., and Penna, R.: Towing tank experiments of resistance, sinkage and trim, boundary layer, wake, and free surface flow around a naval combatant INSEAN 2340 model, IIHR Technical Report No. 421, 2001. [10] Borda, G.: Resistance, Powering and Optimum Rudder Angle Experiments on a 465.9 foot (142 m) Guided Missile Destroyer (DDG-51) Represented by Model 5415-1 and Fixed Pitch Propellers4876 and 4877, David Taylor Naval Ship R&D Center, 1984. [11] I.T.T.C.: Final report of The Specialist Committee on Model Tests of High Speed Marine Vehicles. Proceedings of the 22nd InternationalTowing Tank Conference, Seoul and Shanghai, 1999. 77 [12] Molland, A. F., Turnock S. R., and Dominic A. H.: Ship Resistance and Propulsion,Cambridge University Press, New York, 2011. [13] Hoerner, S.F.: Fluid-DynamicDrag,Published by the Author, Washington, DC, 1965. [14] Peck, R.W.: The determination of appendage resistance of surface ships, AEW Technical Memorandum, 76020, 1976. [15] Kirkman, K.L. and Kloetzli, J.N.: Scaling problems of model appendages, Proceedings of the 19th ATTC, Ann Arbor, Michigan, 1981. [16] Holtrop, J. and Mennen, G.G.J.: An approximate power prediction method. International Shipbuilding Progress, Vol. 29, No. 335, July 1982, pp. 166-170. [17] Edward V Lewis. Principles of Naval Architecture (Second Revision), Volume II - Resistance, Propulsion and Vibration, SNAME (1988). [18] Ferziger, J. H., and Perid M.: ComputationalMethods for FluidDynamics, Springer, New York, 2002. [19] STAR-CCM+ v. 7.02.008 User's Manual, CD-Adapco, 2012. [20] Gothenburg, 2010, "Gothenburg 2010, A Workshop on Numerical Ship Hydrodynamics", Chalmers University of Technology, http://publications. Iib.chalmers.se/cpl/record/index.xsql?pubid= 131971. [21] Batchelor, G.K.: An Introduction to Fluid Dynamics. Cambridge University Press, Cambridge, UK, 1967. [22] Wilcox, D.C.: Turbulence Modelingfor CFD. 2nd Edition, DCW Industries, La Canada, CA, 1998. [23] Hess, J.L.: Panel methods in computationalfluiddynamics. Annual Review of Fluid Mechanics, Vol. 22, 1990, pp. 255-274. [24] Godderidge, B., Turnock, S.R., Tan, M. and Earl, C.: An investigation of multiphase CFD modelling of a lateralsloshing tank. Computers and Fluids, Vol. 38, No. 2, 2009, pp. 18 3 [25] Newman, J.: Marine Hydrodynamics. MIT Press, Cambridge, MA, 1977. 78 19 3 . APPENDIX A MODEL TO FULL SCALE The conversion of the appended model hull resistance to the full scale ship followed standard towing tank procedures as given below at Froude number 0.41. The appendages were not considered separately and were assumed to come across the same Reynolds number as the hull. model wetted surface Sm:= 5.21m2 pm 998.424- Scal:=24.824 ship length S:=3208 mz full scale ship wetted surface fresh water density and kinematic viscosity at 18.9 C 2 -6 m 1.0311910 s p 1025.87- 3 salt water density and kinematic viscosity at 150C m 2 v 1.1843110 6 s m Vm :=3.071- model speed Fn := 0.41 Froude number 5 gravitational acceleration g = 9.80665m : 2 ~= V := En.,gAW scale factor 3 m vm L:=142 m model length Lm := 5.72m m 15.299892- Rtm:=170.936 N ship speed 5 appended model total resistance 79 7 Lm Rnm:=Vm-= 1.70348x 10 vm Cfm:= model Reynolds number 0.075 20.002741 2)2 (log(Rnm) Rtm Ctm = 0.006969 model frictional resistance coefficient (ITTC 1957 line) model total resistance coefficient 0.5.pm.Vm 2Sm Crm :=Ctm - Cf m= 0.004228 Rn := V- 1.834473< 109 model residuary resistance coefficient ship Reynolds number V ship residuary resistance coefficient set equal to that of the model Crs := Crn 0.075 Cfs:= = 0.001422 ship frictional resistance coefficient (ITTC 1957 line) (log(Rn) - 2)2 Ca:=0.0004 correlation allowance coefficient Cts := Crs + Cf s+ Ca = 0.00605 Rfs:= Cfs-0.5p-V 2 . ship total resistance coefficient ship frictional resistance 5.475723< 105 N Rfst (Cfs + Ca)-0.5p-V 2 . S = 7.016477x 1 N Rrs Crs -0.5p.V2-S = 1.628636< 10 N Rts := Cts-0.5p-V2 . 2.330284< 10 N ship total frictional resistance including the correlation allowance ship residuary resistance ship total resistance Pf := Rfst-V = 1.073513< 107 W frictional power Pe :=Rts -V = 3.565309x 1 7 W effective power 80 APPENDIX B RESISTANCE DISTRIBUTION AND FRICTIONAL RESISTANCE COEFFICIENTS The charts below give a picture of the total resistance distribution between frictional resistance and pressure, or residuary, resistance for Froude numbers 0.16 and 0.33 at model and full size ship. This information is given separately for each of the appendages, for the appendages altogether, calculated from both the simulations ran with appendages and from the subtraction between the simulations with and without appendages, for the hull with appendages, and for the bare hull computed directly from the bare hull simulations and from the appended hull simulations by neglecting the computations made on the appendages (hull without appendages). Table 13 gives all these values, for the total, frictional and pressure resistance as computed from the simulations. Table 14 gives the values for the frictional resistance coefficients at the three Froude numbers at which simulations were run. 81 120% Model Fr=0.16 100% 80% 60% " Pressure/Total 40% * Frictional/Total 20% 0% VIE I I III ~, .~, I 120% Model Fr=O.33- 100% 80% 60% . Pressure/Total 40% * Frictional/Total 20% 0% 41~ 82 Full Scale Fr= 0.16 1--120% 100% 80% 60% " Pressure/Total 40% * Frictional/Total 20% 0% bol. .4 120% 100% 80% 60% Nl Full Scale Fr=0.33 I U " Pressure/Total 40% " Frictional/Total 20% 0% Figure 43: Distribution of total resistance among frictional and pressure resistance. Results presented are from the CFD simulations without free surface. 83 Table 15: Total, frictional, and residuary resistance values from the CFD computations. Total Fr=0.41 Fr=0.33 Fr=0.16 Frictional IfResiduary Total Frictional IResiduary Total Frictional j Residuary Bare Hull 12 10.707 1.293 48.23 43.055 5.175 72.42 64.58 7.84 Appended Hull 15.08 12.32 2.76 60.16 48.626 11.534 90.7 73.02 17.68 11.94 10.324 1.616 48.24 41.256 6.984 72.84 62.06 10.78 3.08 1.613 1.467 11.93 5.571 6.359 18.28 8.44 9.84 3.14 1.99 1.15 11.92 7.37 4.55 17.85 10.966 6.884 Rudders 0.828 0.372 0.456 3.162 1.386 1.776 4.764 2.066 2.698 Struts 0.71 0.202 0.508 2.742 0.754 1.988 4.112 1.12 2.992 Bilge Keels 0.812 0.8 0.012 2.976 2.93 0.046 4.422 4.35 0.072 Shafting 0.796 0.62 0.176 3.04 2.302 0.738 4.552 3.4328 1.1192 Total Frictional Residuary Total Frictional Residuary Total Frictional Residuary Bare Hull 105691 91019 14672 417722 356491 61231 631068 537081 93987 Appended Hull 141394 104996.8 36397.2 564776 411770 153006 855800 620566 235234 112988 88739 24249 450978 347950 103028 683240 524368 158872 35703 13978 21725 147054 55279 91775 224732 83485 141247 28406 16258 12148 113800 63820 49980 172560 96204 76356 Rudders 8668 3267 5401 35362 12946 22416 53860 19556 34304 Struts 5898 1503 4395 23476 5788 17688 35540 8690 26850 Bilge Keels 6584 6395 189 25928 25106 822 39120 37846 1274 Shafting 7256 5093 2163 29032 19980 9052 44040 30110 13930 App. HullAppendages App. HullBare Hull Appendages Total App. HullAppendages App. HullBare Hull Appendages Total 84 Table 16: Frictional resistance coefficient for the hull and each appendage as derived from the frictional resistance computations in the CFD simulations. The Reynolds number corresponding to the characteristic length of each appendage at the different Froude numbers is also given. Rn Cf (x1000) Rn Cf (x1000) Rn Cf (x1000) Bare Hull 6.64E+06 3.087 1.37E+07 2.913 1.70E+07 2.828 Rudders 170056 4.131 351052 3.611 436293 3.485 Struts 47128 5.037 97288 4.412 120911 4.243 Bilge Keels 2242655 3.919 4629575 3.368 5753713 3.237 Shafting 1232764 3.771 2544828 3.285 3162755 3.172 Rn Cf(x1000) Rn Cf(x1000) Rn Cf(x1000) Bare Hull 8.22E+08 1.711 1.70E+09 1.576 2.11E+09 1.538 Rudders 2.11E+07 2.366 4.34E+07 2.204 5.40E+07 2.157 Struts 5.84E+06 2.445 1.20E+07 2.213 1.50E+07 2.152 Bilge Keels 2.78E+08 2.043 5.73E+08 1.886 7.12E+08 1.841 Shafting 1.53E+08 2.02 3.15E+08 1.863 3.91E+08 1.819 85 APPENDIX C TIME HISTORIES OF RESISTANCE, TRIM, AND SINKAGE In this appendix some further resistance, trim, and sinkage plots from the simulations with and without a free surface are given. Resistance, trim, and sinkage for the simulations with a free surface are plotted against time, whereas for the simulations without a free surface, where the flow was steady, resistance is plotted against solution process iterations. The resistance values given in these plots have to be doubled to get the full ship resistance as, due to symmetry, only have of it was simulated in the CFD code. Negative values of the trim angle translate to a trim by the stem. It can be seen that the plots from the simulations without a free surface lack the oscillations that the free surface causes to the simulations with a free surface. The plots for the simulations with a free surface given in this appendix took about 48 hours to generate on 124 processors while those for the simulations without a free surface about 24 hours on 15 processors. All plots were generated by the CFD code STAR-CCM+. Resistance Plot Fr-0.41 150 ___________ - ___________ ___________ __________ _ __________ _ 130 110 - -- 90 ________ 70 0 50LL 30 ____________ - 10-10-30 -50 10 20 40 30 Time (sec) Rp--Rf 86 Rt 50 Z Translation Plot Fr-0. 41 E -0,03 -0. 04 -0 20 40 30 6 50 Time (sed -Sinkage Y Rotation Plot Fr-0.41 0 7 0. 2 0. 0 40 20 50 Time (sec) -Tnm Angle Figure 44: Time histories of resistance, trim, and sinkage from the simulations with a free surface using a newer mesh at Fr=0.41. 87 Resistance Plot at Fr=0.41 for the Bare Hull, Without a Free Surface 40 30 10 0- 1000 2000 3000 5000 4000 6000 7000 9000 B000 iteration -Rf -- Rp -Rt Figure 45: Frictional, pressure, and total resistance plot from the bare hull simulations without a free surface at Fr=0.41. Resistance Plot at Fr-0.41 for the Appended Hull, Without a Free Surface 60 50 40 30 20 10 0 -10 1C00 200 3C000 400 5000 iteration -Rf -Rp 88 -Rt 600 7C00 8030 9000 Resistance Plot at Fr-0 41 of the Bilge Keel 3 0 -1 1000 2000 3000 5000 Iteration 4000 6000 7000 8000 9000 Rf -- Rp - Rt Resistance Plot at Fr-O 41 of the Rudder 5 4-. 3- 2- -1 - 1C00 1C060070000090 2000 3000v 4000 Iteration -Rt -Rp -Rf Figure 46: Frictional, pressure, and total resistance plots of the hull with the appendages, the bilge keel, and the rudder from the appended hull simulations without a free surface at Fr=0.41. 89 Pesistance Plot at Fr-0.41 for the Hull xith the Appendages at Full Scale 500000 400000 300000- 200000 100000 1000 2000 3000 4000 iteration -Rf 5000 6000 7000 000 -Rp -Rt Figure 47: Frictional, pressure, and total resistance plot of the hull with the appendages from the full scale appended hull simulations without a free surface at Fr=0.41. 90