Numerical Modeling of Ship-Propeller Interaction under - - PowerPoint PPT Presentation
STAR Global Conference 2014 Vienna, Austria, March 17-19 Numerical Modeling of Ship-Propeller Interaction under Self-Propulsion Condition Vladimir Krasilnikov Department of Ship Technology, MARINTEK Trondheim, Norway Norsk Marinteknisk
Norsk Marinteknisk Forskningsinstitutt
Vladimir Krasilnikov Department of Ship Technology, MARINTEK Trondheim, Norway
STAR Global Conference 2014 Vienna, Austria, March 17-19
Nominal wake 1-WT
N = 0.736
Effective wake: 1-WT
E = 0.769
In order to achieve desired high propulsive efficiency and ensure favorable cavitation and acoustic characteristics of propeller, one has to design the propeller well adapted to the wake field behind ship hull. Interaction between ship hull and propeller results in effective wake field on propeller that may differ considerably from nominal wake, which is normally measured during model tests. At MARINTEK we employ a coupled viscous/potential method to extract the effective wake field and optimize propeller design, using our in-house propeller design and analysis software. In this coupled method, STAR-CCM+ performs as a viscous flow solver.
These studies are relevant to the problems of low-speed maneuvering of ships, backing and crash-back situations. Off-design propeller analysis involves extremely complex flows, where the blade back side performs as a pressure side, and the whole blade is stalled. Extended domains of separated and re- circulated flows exist, giving rise to unsteady vortex shedding. The example shown on this slide presents the comparsion between the experimental data and numerical predictions obtained with STAR-CCM+ (unsteady RANS method) for the B-series propeller operating in the entire 1st quadrant.
Interaction between propeller and duct is a crucial mechanism behind scale effect. The regions of blade tip clearance and duct T.E. are of particular importance. Ducted propeller flow is most adequately solved in the unsteady formulation, by employing the Sliding Mesh method. Scale effect depends significantly on the duct type, propeller geometry, and radial loading distribution towards blade tip, which complicates greatly the application
Within the frameworks of the ongoing R&D project “PROPSCALE” we use STAR-CCM+ to quantify scale effect on ducted propellers of different types.
The physical mechanisms associated with the formation and development of blade tip and leading edge vortices are still not investigated to a sufficient degree. In particular, unsteady phenomena, such as vortex bursting and breaking- up, represent substantial interested from the point of view of propeller noise, erosion and induced pressure impulses. In this example, we used an unsteady RANS method of STAR-CCM+ to study the behavior of the leading edge vortex that caused erosion on the blades of a pulling podded propeller operating at bollard condition.
Unsteady (time-dependent) nature of the problem due to the interaction between the rotating parts (propeller) and stationary parts (hull, appendages, rudder). Presence of free surface of unknown geometry. Flow turbulence of various scales that need appropriate modeling assumptions. Scale effects, including those related to the presence of laminar and transient flow regimes in model scale.
Approaches: 1) Iterative coupled viscous/potential method with Actuator Disk Hull – RANS, Propeller – Panel method or Lifting surface, Coupling – Actuator Disk (Circumferential-averaged volumetric momentum source model), Free surface – VOF. 2) Unsteady RANS method with simplified account for free surface effect Hull – RANS, Propeller – RANS (Sliding Mesh), Free surface – not included (symmetry plane – «double-body model»). 3) Fully unsteady RANS method with free surface Hull – RANS, Propeller – RANS (Sliding Mesh), Free surface – VOF. Software: RANS solver: STAR-CCM+ (CD-Adapco) Panel method solver: AKPA (MARINTEK, in-house propeller analysis program) Lifting surface solver: AKPD (MARINTEK, in-house propeller design program) Actuator Disk setup: ADM (MARINTEK, in-house)
Disk thickness: 0.01D. Constant loading along the disk axis. Radial distribution of elemental thrust identical to realistic propeller calculated by PM – dT(r). Equivalent, but not identical distributions
and dQ(r). Circumferential averaged distribution of momentum sources (axial and tangential).
All-component wake field . AD induced velocities are defined from an additional Open Water calculation with the AD, having the same dT(r) as the AD behind hull. Velocities are sampled at the wake control section 0.1D upstream of propeller plane.
Hull-propeller interaction
Rotation motion of propeller region Sliding interface mesh Time-accurate 1st order, t2 of propeller revolution Two-stage solution MRF+SM
Turbulence *)
SST k-, All Y+ Treatment (used routinely) Other turbulence models (RSM, DES – investigated)
Free surface *)
VOF, 2nd Order, FlatVofWaves Blended HRIC; Pure HRIC Time-accurate 1st order, t=(0.005÷0.01)×LPP/V
DOF *)
Fixed position 2DOF (free sinkage and trim)
*) Also apply with the Coupled Method
Coarse mesh: Hex trimmed mesh, (5÷7) prism layers, stretching factor (1.25÷1.35), 1.5 mio cells per half ship without appendages, 30<Y+<90.
About 30 cells in vertical direction, 3 volumetric controls. About 100 cells per wave length near ship hull, about 30 cells per wave length in the far field. Wave damping at the Inlet, Outlet and Side boundaries; Damping length is chosen so that damping begins in the refinement zone. The size of the domain in transverse direction is large enough to avoid intersection of the Kelvin’s wake with the side boundaries.
Free surface treatment
Fine mesh: Poly mesh, (10÷30) prism layers, stretching factor (1.1÷1.2), (1.3÷1.5) mio cells per blade passage, Y+<1. Coarse mesh: Tet mesh, no BL mesh, (0.3÷0.5) mio cells per blade passage, 30<Y+<250.
Main particulars Model scale Length between PP LPP, [m] 7.2786 Maximum beam at WL BWL, [m] 1.019 Depth D, [m] 0.6013 Draft T, [m] 0.3418 Wetted surface area SW, [m2] 9.4379 Block coefficient CB 0.6505 Midship section coefficent CM 0.9849 Coordinates of propeller center *) (x/LPP, y/LPP, z/LPP) (0.4825, 0.0,
*) Origin of coordinate system at CP, midship, WL; x - downstream Propeller elements Model scale Propeller diameter DP, [m] 0.25 Hub ratio dH/DP 0.18 Number of blades Z 5 Blade area ratio AE/A0 0.8 Pitch ratio P(0.7R)/D 0.9967 Sections NACA66/a=0.8 Conditions Calm water, Fixed position and Free motion Without rudder Froude number Fr=V/(g*LPP)1/2 0.26 Reynolds number Re=(V*LPP)/ν 1.4*107 Ship speed V, [m/s] 2.19663 Propeller RPS *) n, [Hz] 9.5 *) Measured during self-propulsion tests
KRISO container ship KCS KRISO Propeller KP505
Time step, interface scheme Cp Cv Ct dt=0.01 [s], pure HRIC 0.000633 0.002840 0.003473 dt=0.02 [s], pure HRIC 0.000630 0.002842 0.003472 dt=0.03 [s], pure HRIC 0.000628 0.002842 0.003470 dt=0.04 [s], pure HRIC 0.000630 0.002842 0.003472 dt=0.05 [s], pure HRIC 0.000630 0.002842 0.003472 Time step, interface scheme Cp Cv Ct dt=0.02 [s], blended HRIC 0.000711 0.002843 0.003554 dt=0.03 [s], blended HRIC 0.000726 0.002846 0.003572 dt=0.05 [s], blended HRIC 0.000756 0.002849 0.003605 with Rudder (KRISO) without Rudder (SRI) friction line Ct Cp Cf Ct Cp -residual Cf Cf0 (ITTC-57) 0.003557 0.003534 0.000689 0.002845 0.002832
Calculations with blended HRIC scheme Calculations with pure HRIC scheme Experiment
Solution appears dependent on time step due to the Courant number limits in the blended HRIC scheme Solution is independent on time step
*) SST k- turbulence model is used in this exercise
with Rudder (KRISO) without Rudder (SRI) friction line Ct Cp Cf Ct Cp -residual Cf Cf0 (ITTC-57) 0.003557 0.003534 0.000689 0.002845 0.002832
Calculations with blended HRIC scheme, Time step dt=0.02 [s]
Turbulence model Cp Cv Ct SST k-w 0.000711 0.002843 0.003554 Real k-e 0.000717 0.002853 0.003570 RSM *) 0.000697 0.002972 0.003669 SST k-w + RSM *) 0.000703 0.002971 0.003674
*) RSM Model: Linear Pressure Strain, High-Re
Turbulence model Cp Cv Ct SST k-w 0.000630 0.002842 0.003472 Real k-e 0.000636 0.002852 0.003488 RSM *) 0.000618 0.002970 0.003588
Calculations with pure HRIC scheme, Time step dt=0.02 [s]
Influence of turbulence model. Symmetry of calculated wake field: Half ship and full ship. Influence of the inclusion of a new region: Propeller block and Actuator Disk block.
Objectives of the study
Oscillatory convergence is observed for all conditions. At lower Fr, oscillations show larger amplitude, and levels of residuals are higher. The presented results are obtained with blended HRIC, dt=0.02 [s]. Calculation done with pure HRIC reveal large oscillations and become unstable at lower Fr.
Observations
Ct, S-P RPS KTB KQB Coupled method 0.003991 9.55 0.1703 0.02942 , % +0.63 +0.53 +0.18 +2.15 Unsteady RANS method 0.003907 9.53 0.1650 0.02933 , %
+0.32
+1.84 Experiment 0.003966 9.50 0.170 0.0288 Coupled method Blended HRIC, dt=0.05 [s]; SST k-; 5 iterations between the RANS and panel method solvers. Unsteady RANS method: MRF+SM; Pure HRIC and Blended HRIC *), dt=0.02 [s] at MRF stage, dt 2 at SM stage; SST k-; About 50 propeller revolutions are performed at the SM stage. Calculation results at «ship point», SFC=30.3 [N] from model tests *) Both the calculation with pure HRIC scheme and blended HRIC scheme result in very close predictions of resistance and propeller forces, since the pure HRIC scheme is used effectively at the SM stage due to small time step.
*) Results obtained with Unsteady RANS method
Pure HRIC Offers solution independent on time step, which is advantageous in self-propulsion simulations using realistic propeller. Blended HRIC Shows more stable performance in simulations involving free motion. What is the best practice for self-propulsion simulations with free motion?
A small wetted area is predicted at the transom, in the vicinity of CP. Wave elevation is over-predicted at the stern end, but agrees well with the measurements aft of the transom. The flow in this region is influenced by vortex separation that occurs at the transom, both below and above the free surface.
With standard two-equation turbulence models the computed slipstream vortices are excessively diffusive, and they dissipate too soon downstream of propulsor. Before making the final shift toward the use of LES and DES methods, one should explore the possibilities of improvement
Norsk Marinteknisk Forskningsinstitutt
Technology for a better society
www.sintef.no