The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
Malaysia, September 3-4, 2012
Numerical Simulation of Propeller Hydrodynamic Performance of LNG Carrier in Open Water Using Fluent A. Maimun a, M. Nakisab, Najmic, Y. Mohamedd, Y.S.Angb, A. Priyantod, Jasward, F. Behrouzic a: Prof. Dr., b: PhD student, c: Master student, d: Senior Lecturer Mechanical Eng. Faculty, University Technology Malaysia, Johor Bahru
[email protected]
ABSTRACT Marine propeller blade geometries, especially LNG carriers, are very complicated and determine of hydrodynamic performance of these propellers using by experimental work is very expensive, time consuming and have many difficulties in calibration of marine Lab. facilities. This paper presents the assessment of the effect of turbulent model and mesh density on propeller hydrodynamic parameters. On the other hand this paper focuses on the LNG carrier Tanaga class propeller hydrodynamic performance coefficients such as Kt, Kq and η, respect to the different advance coefficient (j). Finally, the results of numerical simulation that are calculated based on RANS (Reynolds Averaged Navier Stocks) equations, are compared with existing experimental results, then they are analysed and discussed.
KEY WORDS: Numerical Simulation; LNG Carrier; Hydrodynamic Characteristics; Propeller; RANS Equation. NOMENCLATURE D: Propeller diameter, (m) Dhub: Hub diameter, (m) Z: Number of blade P/D: Pitch Ratio R: Rake of Blades n: Rate of revolutions of propeller, (rpm) Cp: Pressure coefficient p: Static pressure at point of interest p0: Reference pressure at infinity Va: Advance velocity, (m/s) J: Advance ratio KT: Thrust coefficient KQ: Torque coefficient Br: Boss ratio AE/A0= Expanded Area Ratio (EAR) ρ = Density of water η = Open water efficiency Dr= Diameter of Rotational domain
Ship Propulsion and Propeller
Lr= Length of rotational domain Lmr= Outlet length of rotational domain Ds= Diameter of stationary domain Lso= Length of outlet stationary domain Lsi= Length of inlet stationary domain (x, y, z) : Cartesian coordinate system with its origin at the centre of propeller +X, +Y, +Z : Cartesian directions in Right-Hand system Ux,Uy,Uz : Velocity components in the Cartesian coordinate system (x, y, z)
INTRODUCTION Today both potential and viscous flow CFD (Computational Fluid Dynamics) codes are extensively used for design purposes, allowing experimental tests to be performed only in the final stages of the project. With reference to marine applications, CFD simulations can be used to predict the flow around hulls, appendages, and propellers. For the case of flow around a marine propeller, the numerical predictions have been first carried out using solvers based on the Lifting-Surface-Theory. (Kerwin and Lee, 1978; Streckwall, 1986).The viscous RANS (Reynolds Averaged Navier Stocks) approach, on the other hand, was applied later (Kim and Stern, 1990) . Subsequently, due to the progress of CFD technology and the continuous increase of computer performances, RANS solvers have become even more versatile and popular (Abdel Maksoud et al., 1998; Oh and Kang, 1992; Chen and Stern, 1999; Watanabe et al., 2003; Kawamura et al.,2006) Nevertheless, a successful RANS simulation is still influenced today by a lot of factors such as CAD geometry, topology and the dimension of the computational domain, meshing strategy, physical modelling, Regarding the meshing strategy, generally speaking, simulations carried out with Hybrid unstructured meshes i.e. - tetrahedral with prisms or hexahedral layers on solid surfaces- are less accurate
407
The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
than simulations carried out with hexahedral structured grids. On the other hand, the generation of hybridunstructured meshes generally requires less effort than that needed to generate hexa-structured meshes. As a matter of fact, hybrid mesh generation is semiautomatic, while structured mesh generation is, in general, not automatic (as for as the decomposition of the domain in blocks is concerned), and consequently requires a significant amount of work. Thus, in this work we investigated, mainly for the prediction of the propulsive characteristics rather than for a detailed investigation of the flow field, if the hexa-structured meshes can guarantee such better performances in terms of accuracy, convergence and computational time than hybrid-unstructured meshes, to justify the greater effort required for their generation. The study was carried out on one propeller in full scale: the fivebladed propeller of LNG carrier Tnaga Class. Numerical results were compared with the available experimental data. On the other hand, the present study was carried out mainly to give useful indications regarding the hydrodynamic characteristics of the propeller in open water situation. On the numerical predictions of propeller performances, besides the global field values represented by the thrust, Kt, and torque, Kq coefficients, also some selected local field values were considered in order to enforce the reliability of the comparison.
Malaysia, September 3-4, 2012
A Cartesian reference system is centred in the centre of the propeller, having the x-axis aligned along the propeller axis, the y-axis pointing upward towards the free surface, and the z-axis following a right-handed system, pointing on the port side. (Fig. 1) Table 1: propeller geometric characteristics
Z D Dhub Br P/D AE/A0 R
5 7.7 m 1.28 m 0.17 0.94 0.88 15 Deg.
BOUNDRY CONDITION The numerical predictions presented in this work were carried out with the ANSYS-FLUENT 13 (which will be referred from now on as FLUENT) commercial CFD solver. It employs the node-centred finite volume method. Fig. 2 and table 2 show the scheme and dimensions of computational domain for simulation the propeller, respectively.
PROPELLER MODEL In this study two propellers in model scale, which are extensively used for the validation of CFD codes, were considered. The propeller has a diameter D of 7.7m and hub diameter Dhub of 0.17D, and is right handed when mounted on a pulling thrust. Design pitch ratio (P/D) is 0.94 and the blade ratio (EAR) is 0.88. A propeller drawing is depicted in Fig.1 and the table1 shows the geometric characteristics. Fig.2: scheme of computational domain The MFR (Multiple Frame of Reference) approach was used to numerically predict the flow around the marine propeller. Since the flow around a marine propeller working in a homogeneous- uniform flow is periodic – with respect to the blades – numerical predictions were performed considering only one blade passage.
Fig. 1: Front view of propeller
Ship Propulsion and Propeller
408
The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
Table 2: Dimensions of computational domain
Values Dr Lmr Lr Ds Lsi Lso
rotating 1.44D 1.5D 3D
stationary
10D 3D 5D
In to a rotating part, called Rotating and in to a stationary part, called Stationary. The dimensions of the domains of propeller are given in Table 2. In Table 2 the variable Lmr, visible in Fig. 2, is the axial length of outlet the Rotating part, and D is the propeller diameter. The inlet, outlet and outer boundaries of the Stationary part were placed far enough from the propeller in order to not affect the results. The distances were set through a domain independence study, carried out considering more shapes of the Stationary part, defined varying systematically Lr, Lsi, Ds. [Fig. 3]
Malaysia, September 3-4, 2012
close the momentum equation. With Boussinesq’s eddy-viscosity assumption and two transport equations for solving a turbulence velocity and turbulence time scale (turbulence modelling), RANS equations are closed. Only steady incompressible equations are solved in the present work. The shear-stress transport SST k-e model with the “transitional flow option” is employed here due to its good performance for wall-bounded boundary layer flows. FLUENT uses a cell- centred finite volume method. Absolute velocity RANS formulations are used and equations are solved in a moving reference frame for the current application. The velocity pressure coupling and the overall solution procedure are based on SIMPLE algorithm is employed to accelerate solution convergence. The second-order QUICK scheme is used for convection terms and second-order centraldifference scheme for diffusion terms. Further detail is available in (Fluent 6.2 User’s manual, 2005)
RESULTS AND DISCUSSION Figs. 3 and 4, show the distribution of pressure coefficient on face and back surface of one blade at three radial section r/R = 0.80. It is clearly that the low pressure region is happen on back surface and the high pressure region is occurred on face surface of blade.
Fig. 3: Rotational and Stationary domain
NUMERICAL METHOD In Cartesian tensor form the general RANS equation for continuity can be written as, Fig. 3: contours of pressure coefficient on blade section (1) and equations for momentum become (
)
⌈ ( ̅̅̅̅̅̅̅
)⌉ (2)
̅̅̅̅̅̅̅ are the Where δij is the Kronecker delta, unknown Reynolds stresses that have to be modelled to
Ship Propulsion and Propeller
409
The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
Malaysia, September 3-4, 2012
Fig. 4: pressure coefficient distribution on back and face surfaces of blade section Also, the Fig. 5 shows that the distribution of low pressure and high pressure area (static pressure) on back and face surface of 5 blades, respectively.
Fig. 7: Contours of pressure coefficient on back surface
Fig. 5: distribution of static pressure on face and back surfaces of 5 blades When the propeller rotate around (+x)-direction, generate thrust in (–x)-direction and high pressure occurs on face surface and low pressure on back surface of blades, respectively. [Figs.6 and 7]. Fig. 8 shows the contours of total pressure behind the propeller. Fig. 8: contours of total pressure behind the propeller Figs. 9-11, show the Contours of velocity magnitude behind the propeller, contours of velocity magnitude on face surface and velocity magnitude vectors behind the propeller, respectively. The velocity magnitude in tip of blades is higher than other region of blade surfaces. Because of the higher rotation in tips of blades, the momentum of fluid particles is very higher than near hub.
Fig. 6: contours of pressure coefficient on face surface
Fig. 9: Contours of velocity magnitude behind the propeller
Ship Propulsion and Propeller
410
The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
Malaysia, September 3-4, 2012
slightly over predicted all the way. The maximum hydrodynamic propeller efficiency will be occurred in j=0.84.
Fig. 10: contours of velocity magnitude on face surface Fig. 11: open water diagram at full scale CONCLUSION
Fig. 11: velocity magnitude vectors behind the propeller
The computational method has been developed to predict the local flow field and the hydrodynamic performance of the propellers. RANS (ReynoldsAveraged Navier-Stokes) methods are becoming a useful tool in flow analysis, engineering design and optimization. They open up new possibilities in analysis and comprehension of flow phenomena that is impossible or difficult with traditional model tests. In according to the presented computational results based on RANS equations, the following conclusions are drawn: 1-
The forces and moments acting on the propeller when operating in a uniform fluid stream, hence the propeller generate the thrust and make pressure and velocity in the fluid particles on face blades surface and behind the propeller.
2-
The fluid velocity in the tips of face blades is higher than other areas and the fluid pressure in the face blades near leading edges is higher than other areas.
3-
The maximum hydrodynamic propeller efficiency will be occurred in j=0.84.
OPEN WATER PERFORMANCE The forces and moments produced by the propeller are expressed in their most fundamental form in terms of a series of non-dimensional characteristics: these are completely general for a specific geometric configuration. The non-dimensional terms used to express the general performance characteristics are as follows: Thrust coefficient Torque coefficient Advance coefficient The results from the Numerical simulation of propeller in open water based on RANS equation at full scale compared to the model test results can be seen in Fig. 11. Various J-values are obtained by keeping a same revolutions (n=108rpm) but varying the flow speed. The trends of results with varying advance ratio are well predicted. It should be noted that KQ and Kt are
Ship Propulsion and Propeller
REFRENCES -
-
-
Kerwin, J.E., Lee, C.S., 1978 Prediction of steady and unsteady marine propeller performance by numerical lifting-surface theory. Trans SNAME 86 (4), 218–253 Kim, H.T., Stern, F., 1990. Viscous flow around a propeller-shaft configuration with infinite-pitch rectangular blades. J. Propul. 6, 434–443 Streckwall, H., 1986. A method to predict the extent of cavitation on marine propellers by lifting-surface-theory. In: International Symposium on Cavita- tion, Sendai, Japan
411
The 6th Asia-Pacific Workshop on Marine Hydrodynamics (APHydro2012)
-
-
-
-
-
-
Malaysia, September 3-4, 2012
Abdel-Maksoud, M., Menter, F., Wuttke, H., 1998. Viscous flow simulations for conventional and high-skew marine propellers. Schiffstechnik/Ship Technol. Res. 45, 64–71. Watanabe, T., Kawamura, T., Takekoshi, Y., Maeda, M., Rhee, S.H., 2003. Simulation of steady and unsteady cavitation on a marine propeller using a RANS CFD code. In: 5th International Symposium on Cavitation (CAV2003), Osaka, Japan Chen, B., Stern, F., 1999. Computational fluid dynamics of four-quadrant marine- propulsor flow. J. Ship Res. 43 (4), 218–228. Oh, K.-J., Kang, S.-H., 1992. Numerical calculation of the viscous flow around a rotating marine propeller. KSME J. 6 (2), 140–148 Kawamura, T., Takekoshi, Y., Yamaguchi, H., Minowa, T., Maeda, M., Fujii, A., Kimura, K., Taketani, T., 2006. Simulation of unsteady cavitating flow around marine propeller using a RANS CFD code. In: 6th International Symposium on Cavitation (CAV2006), Wageningen, The Netherlands. Fluent 6.2 User’s Manual,2005
Ship Propulsion and Propeller
412
