Twenty-Fourth Symposium on Naval Hydrodynamics (2003) / Chapter Skim
Currently Skimming:
Prediction of High Reynolds Number Flow Around Naval Vessels
Prediction of High Reynolds Number Flow Around Naval Vessels
Pages 40-64
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 40... ...
Indeed, when hydroacoustic performance is important the design must be optimised in the correct flow field, that is a flow which is governed by the presence of the hull and its appendages, but also by the action of the propeller, and for surface ships by the presence of the free surface. The flow field around ships has been studied experimentally on model scale for a century now and engineers have found methods to make the results of experiment at model scale useful.
|
From page 41... ...
The approach followed to achieve these objectives was to proceed as follows: measure flows around selected hull forms at a range of model scale Reynolds numbers · develop instrumentation to measure flows at full scale · measure flows in the propeller region at full scale · evaluate existing computational methods for the prediction of model scale Reynolds number flows for the selected hull forms · develop enhanced methods for the prediction of full scale flows with propulsors · validate the capability by comparison with measured data Project organization This project has been conducted through the EUCLID Memorandum Of Understanding (MOW) between armament directors of the WEAO defence organization.
|
From page 42... ...
This code was selected among several potential candidates through a benchmark effort which constituted a specific task. A second track was to evaluate the potential of an "in house" code for which the source is available The different tasks associated with the CFD work for both codes were to identify turbulence model, develop a propeller actuator disk model, and validate on test cases HULL FORMS The hull forms used in this study were selected in order to provide all the flow features present on current naval ships for both surface ships and submarines.
|
From page 43... ...
on two ships, the Dutch frigate the "De Ruyter" and the NATO research vessel, the "Alliance". The measurements performed concerned essentially 2D Laser Doppler velocimetry just ahead of the propeller plane.
|
From page 44... ...
out of a single view port (figure 10~. ~ - ~ Figure 10: LDV grid on the De Ruyter < |
From page 45... ...
An uncertainty analysis was conducted to try to assess the confidence level of the measurement by taking 0.2% 1.4% Processor2 0.1% Finite size mv 0.01% Sampling 0.25% Velocity bias 1.0% Alignment Total error Positioning Direction 0.3% 1.8% 0.01 mm 1 deg Table 4: Review of errors for a mean velocity of 4 m/see Full scale LDV results The results of L.DV measurements after analysis consists in two wake maps, on for each speed and represent the best estimate of the flow field based on the two test campaign (figure 14~. The final maps are composite maps which incorporate the data acquired for the different grids.
|
From page 46... ...
~* FS Cr:~s lA: HI 2A Figure 14.b: Full scale measurements at 12 knots Figure 14: Full scale measurement results after final analysis Alliance Pitot tube location 11,12 knot measurement 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 Figure 15: Pitot tube measurements at 12 knots TESTS IN THE WATER TUNNEL Facility The tests were performed in the GTH (Grand Tunnel Hydrodynamique)
|
From page 47... ...
The EUCLID submarine was fitted in the large test section of the GTH, and LDV measurements have been performed, mostly 3D, at over 20600 points for 5 planes during this test campaign: · far upstream of the propeller, · in the middle of ship, · upstream of appendages, · upstream of propeller, · downstream of propeller. An XI n Figure 16: GTH facility ~ it' Figure 17: LDV system for the EUCLID Sub measurements Figures 18 and 19 show examples of axial wake maps in the propeller plane for the two extreme tunnel velocities.
|
From page 48... ...
and with the propeller operating to achieve the full scale thrust coefficient. The bilge keels were not fitted since they would not have been aligned with the flow in the absence of the free surface.
|
From page 49... ...
90 ~ 80 ~ 70 60 E50 40 ~ 30 ~ 20 ~ 10 ~ o 0,8 1,0 1,2 VxNref Figure 26.a: Axial velocity profiles in plane 1 - e: == O_ . ~ ~ 0,8 1,0 1,2 1,4 Vx/\/ref Figure 26.b: Axial velocity profiles in plane 5 Figure 26: Axial velocity profiles at 1.4 and 9.9 m/s The resolution of the wake maps is illustrated in figure 28 where the transverse components of the velocity vectors are plotted for all measured data points in plane P5, aft of the rudder.
|
From page 50... ...
._ hi_ _ Figure 27.c: Appended with propeller (P 3,4,5) Figure 27: Axial velocity maps at 9.9 m/s Figure 28: Transverse velocity field in plane 5 for the < |
From page 51... ...
Instrumentation Measurements were performed using a Laser Doppler Velocimeter along five transverse sections identical to the GTH tests (figure 25~. Three dimensional velocity field measurements were carried out in two separate steps by means of two different optical configurations with the laser radii coming from below and one side, in order to measure the axial-transversal (XY)
|
From page 52... ...
.. r l~i Figure 32: Definition of the grid density on the < |
From page 53... ...
Statistical analysis is performed inside each slot to obtain mean flow field and turbulence intensity information. The slotting parameter choice is rather critical for this kind of analysis [elk et al., 2000)
|
From page 54... ...
Code selection The first task has been to select of code through a benchmarking effort by computing the flow around a representative geometry The basic requirements for the codes to be tested were as follows: Standard k-e turbulence model with wall function Optional alternative turbulence models: Reynolds Stress or non-linear models, Pressure gradient wall functions, Near wall models Recommended second order accurate discretisation (alternative higher order discretisation possible) The range of grid resolutions required to test wall functions were 0.20, 0.40, and 0.8 million cells and 0.25, 0.50, and 1.0 million cells for near wall models The different CFD packages tested were CFX4, CFX5, and Fluent with different grid resolution: 500K to 2M cells, y+ from 1 to 100 .
|
From page 55... ...
00,05 -0,1 0,05 O The results of four of the best performing models are plotted in figure 41 were the same flow features as seen in the EUCLID Sub measurements. -0.1' In order to compare the results more clearly, the difference between experiments and calculations at a given radius is plotted against angular position over a 90° sector for all four turbulence models (figure 42~.
|
From page 56... ...
This numerical uncertainty analysis was conducted for 2 turbulence models: k-e and Reynolds Stress Models and for the two Reynolds numbers corresponding to the GTH tests. The results of the numerical uncertainty analysis are as follows: · grid convergence uncertainty for the unappended case is less than 1% on average with a peak of 1.5%; for appended cases the average uncertainty is 4% with a peak of 8.5%.
|
From page 57... ...
If the simulation uncertainty is broken down into its constituent parts then: a,- \~ ~ \~,,6t ~~ ~ ~ ~~ ~ ~ f ~9~ -~.r~e I: =~.r~ Am with ~.~r^~ - GAIL ,'~130: 46.a: Water tunnel data (77::~ tempt. (£~ ::~ >{ ~T>:~ ~ 1~.~-.:f~ relI.- ~ :r~rntr~ x11~ `;~:tt ~st<:h ~F~>~;I~ ~~d ;.~l.~r 46.b: Computational result Figure 46: Comparison of axial data for the propeller inflow plane at 1.4 m/s UE = UD + USN + USPD + USMA where USN is the simulation numerical uncertainty, USPD is the simulation modelling uncertainty arising from previous data and USMA is the simulation modelling uncertainty arising from modelling assumptions.
|
From page 58... ...
The resolution is based on a time marching algorithm with local time step and multi-stage Runge-kutta with residual smoothing. Turbulence models Different turbulence models have been implemented: · Algebraic: Baldwin-Lomax · 1 equation model: Spalart-Allmaras · 2 equations models: k-£ The one equation Spalart and Allmaras model was found to give good agreement with experimental data on the velocity field although the turbulence field is affected by grid density.
|
From page 59... ...
vr~t _ ta:~s _ ~ OM cel] s _ ~.~.f~llr.r IL:~ne~ :~S wish ~Lpp~ndege~ D~:~ p~rop~:ll~r: Figure 48.b: Full scale computations at 12 knots Figure 48: Axial velocity computed at full scale (QinetiQ)
|
From page 60... ...
i i __ I_ 1 _ ~ ~ _ ~ . 1 Figure 52: Flow chart for the propeller force field Figure 51: < |
From page 61... ...
CONCLUSIONS The RTP 10.12 project on high Reynolds number viscous flows, represents an extensive European effort to improve computational tools through the acquisition of a large experimental data base at model scale and at full scale. This data has been analysed to yield the best estimates of velocities fields on different hull shapes (2 surface ships and one submarine)
|
From page 62... ...
Di Felice, F 2000, "Propeller flow field analysis by means of LDV phase sampling techniques", Experiments in Fluids, Vol 28, p.
|
From page 63... ...
Conf. on Numerical Grid Generation in Computational Field Simulation.
|
From page 64... ...
We are well aware of the Athena data; however, our motivation was to use experimental technology available today, in particular LDV, to obtain more extensive and reliable data, at full scale but also at high Reynolds number at model scale. In particular, the capability offered by the large tunnel to investigate in the same conditions a significant of Reynolds number provides a challenging data base for turbulence models.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.