Professional Documents
Culture Documents
Analysis of The Blade Boundary-Layer Flow of A Marine Propeller Using A RANS
Analysis of The Blade Boundary-Layer Flow of A Marine Propeller Using A RANS
Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng
Keywords: In this paper a comparison between RANS simulations carried out with the 𝑘 − 𝜔 SST turbulence model and
Marine propeller ̃ 𝜃 transition model, and experimental measurements for marine propeller P4119 is made. The experiments
𝛾 − R𝑒 𝑡
Blade boundary-layer flow were conducted at the David Taylor Model Basin and comprehended three-dimensional velocity components
RANS solver
measurements of the blade boundary-layer using a LDV system in uniform conditions. The present work
Transition modeling
includes an estimation of the numerical errors that occur in the simulations, analysis of the propeller blade flow,
chordwise and radial components of the boundary-layer velocity profiles and boundary-layer characteristics.
From this comparison and depending on the selected inlet turbulence quantities, we conclude that the transition
model is able to predict the extent of laminar and turbulent regions observed in the experiments. Moreover,
the predicted velocity profiles show good agreement with the experimental measurements and verified the
location of transition.
∗ Corresponding author.
E-mail address: joao.baltazar@tecnico.ulisboa.pt (J. Baltazar).
https://doi.org/10.1016/j.oceaneng.2020.107633
Received 4 February 2020; Received in revised form 6 May 2020; Accepted 8 June 2020
0029-8018/© 2020 Elsevier Ltd. All rights reserved.
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
2
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
model was originally introduced in 1994 by Menter to deal with the Table 1
strong sensitivity to the free-stream conditions of the original 𝑘 − 𝜔 Overview of the grid sizes and number of cell faces on a single blade. Maximum 𝑦+
values refer to the design condition.
turbulence model and to improve the prediction of the boundary-layer
Grid Volume Blade 𝑦+max
flow with adverse pressure gradients (Menter, 1994). In the present
work, the 2003 version of the 𝑘 − 𝜔 SST model is used (Menter et al., G1 37.6M 73.9k 0.20
G2 21.0M 42.3k 0.24
2003). The transport equations to calculate the modeled turbulence
G3 9.9M 25.6k 0.31
kinetic energy and turbulence dissipation rate are: G4 6.1M 18.6k 0.39
[ ]
𝜕(𝑉𝑗 𝑘) 𝜕 ( ) 𝜕𝑘 G5 1.9M 8.6k 0.51
=𝑃𝑘 − 𝐷𝑘 + 𝜈 + 𝜎𝑘 𝜈𝑡 , G6 0.9M 4.9k 0.66
𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗
[ ] (2)
𝜕(𝑉𝑗 𝜔) 𝜕 ( ) 𝜕𝜔 ( ) 𝜎 𝜕𝑘 𝜕𝜔
=𝑃𝜔 − 𝐷𝜔 + 𝜈 + 𝜎𝜔 𝜈𝑡 + 2 1 − 𝐹1 𝜔2 ,
𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜔 𝜕𝑥𝑗 𝜕𝑥𝑗
which only includes a production term 𝑃𝜃𝑡 . The terms 𝑃𝛾 , 𝐸𝛾 and 𝑃𝜃𝑡
with the turbulent viscosity defined as
and the constants 𝜎𝑓 and 𝜎𝜃𝑡 are given in Langtry and Menter (2009).
𝑎1 𝑘
𝜈𝑡 = ( ), (3) The coupling between the 𝑘 − 𝜔 SST turbulence model and the
max 𝑎1 𝜔, 𝑆𝐹2 ̃ 𝜃 transition model is made by an effective intermittency 𝛾eff
𝛾 − Re 𝑡
where 𝑃𝑘 and 𝐷𝑘 are the production and dissipation terms of the which is used to control the production and dissipation terms in the
turbulence kinetic energy, respectively, 𝑃𝜔 and 𝐷𝜔 the production and 𝑘 transport equation, Eq. (2). The definition of the turbulent viscosity
dissipation terms of the turbulence dissipation rate, respectively, 𝑆 is of the 𝑘 − 𝜔 SST turbulence model remains unchanged. This model is
the strain-rate magnitude, 𝐹1 and 𝐹2 auxiliary functions, and 𝑎1 , 𝜎𝑘 , 𝜎𝜔 ̃ 𝜃 −SST model, because it makes use
also sometimes known as the 𝛾 − Re
and 𝜎𝜔2 are constants of the model. The complete model and constant 𝑡
̃ 𝜃 , in addition to the 𝑘 and 𝜔 equations of
of the equations for 𝛾 and Re
values are given in Menter et al. (2003). 𝑡
The 𝛾 − Rẽ 𝜃 transition model proposed by Langtry and Menter the SST turbulence model (Menter et al., 2003). Therefore, this model
𝑡
(2009) is coupled with the 𝑘 − 𝜔 SST turbulence model of Menter et al. corresponds to a four-equation transition SST turbulence model.
(2003) for the simulation of the flow around the marine propeller at
model-scale including transition. This transition model contains two
transport equations and accounts for transition due to free-stream tur- 2.3. Flow solver
bulence intensity, pressure gradients and separation. One is a transport
equation for intermittency 𝛾:
[( ) ] The RANS equations are solved using the ReFRESCO code, devel-
𝜕(𝑉𝑗 𝛾) 𝜕 𝜈 𝜕𝛾 oped within a cooperation led by MARIN for hydrodynamic applica-
= 𝑃𝛾 − 𝐸𝛾 + 𝜈+ 𝑡 , (4)
𝜕𝑥𝑗 𝜕𝑥𝑗 𝜎𝑓 𝜕𝑥𝑗 tions (ReFRESCO, 2019). A finite-volume technique with cell-centered
where 𝑃𝛾 and 𝐸𝛾 are the production and relaminarisation terms, respec- collocated variables is used for solving the equations. A pressure-
tively. The blending between the laminar and turbulent flow regimes correction equation based on the SIMPLE algorithm is used to ensure
is made by the intermittency, which changes between zero and one, mass conservation. A segregated approach is adopted here for the
respectively. In the present formulation, the intermittency is also set solution of all transport equations. The implementation is face-based,
equal to one in the freestream instead of a small value as in the original which permits grids with elements consisting of an arbitrary number of
model (Menter et al., 2002). The second one is a transport equation for
̃ 𝜃: faces. For the convective flux terms, a second-order scheme (QUICK) is
the local transition onset momentum thickness Reynolds number Re 𝑡
[ ] applied for the momentum equations, and a first-order upwind scheme
̃ 𝜃)
𝜕(𝑉𝑗 Re ( ) 𝜕 ̃
Re is applied for the 𝑘 − 𝜔 SST turbulence model and 𝛾 − Re ̃ 𝜃 transition
𝑡 𝜕 𝜃𝑡
𝑡
= 𝑃𝜃𝑡 + 𝜎𝜃𝑡 𝜈 + 𝜈𝑡 , (5)
𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 model. In this study, ReFRESCO version 2.3 is used.
3
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
• At the inlet, the uniform velocity and the turbulence level depend-
ing on the turbulence model are prescribed, while the pressure is
4
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
Fig. 6. Convergence of 𝐾𝑇 (left) and 10𝐾𝑄 (right) with grid refinement ratio ℎ𝑖 ∕ℎ1 .
Table 2 𝜈𝑡 ∕𝜈 = 500 at the inlet boundary are specified. The choice of these inlet
Decay of turbulence quantities from inlet (𝑥∕𝑅 = 10) to the vicinity of the propeller
turbulence quantities is based on a previous study (Baltazar et al.,
(𝑥∕𝑅 = 1). Turbulence quantities at 𝑥∕𝑅 = 1 refers to section 𝑟∕𝑅 = 0.7.
2018), where the 𝛾 − Re ̃ 𝜃 transition model has been used for the flow
Model 𝑥∕𝑅 = 10 𝑥∕𝑅 = 1 𝑡
simulation on two marine propellers at model-scale and the results
𝑇𝑢 𝜈𝑡 ∕𝜈 𝑇𝑢 𝜈𝑡 ∕𝜈
have been compared with paint-tests. Table 2 presents the turbulence
̃ 𝜃
𝛾 − Re 1.20% 500.0 1.14% 151.5
̃ 𝜃
𝑡 intensity and eddy-viscosity ratio at the inlet boundary (𝑥∕𝑅 = 10) and
𝛾 − Re 1.50% 500.0 1.37% 222.5
𝑡
𝑘 − 𝜔 SST 1.00% 1.0 0.13% 0.74 plane 𝑥∕𝑅 = 1 upstream of the propeller at the radial section 𝑟∕𝑅 = 0.7.
The results show that the large value selected for the eddy-viscosity
ratio (𝜈𝑡 ∕𝜈 = 500) is crucial in the control of the turbulence decay from
the inlet boundary to the propeller plane (Spalart and Rumsey, 2007),
extrapolated from the interior assuming zero normal derivative and therefore in the prediction of transition occurrence and location.
(Neumann boundary condition). For the 𝑘 − 𝜔 SST turbulence model, standard values, i.e. 𝑇 𝑢 = 1.0%
• At the outlet, an outflow condition of zero normal derivative is and 𝜈𝑡 ∕𝜈 = 1, are assumed at the inlet boundary. In this case, and as
prescribed for all flow variables.
shown in Baltazar et al. (2018), the influence of the inlet turbulence
• At the outer boundary, a constant pressure is set, while a Neu-
quantities on the propeller simulations is negligible.
mann boundary condition is used for all other variables.
• For the propeller blades and hub, no-slip and impermeability
conditions are applied at the surface and Neumann boundary 4. Results
conditions are used for the pressure 𝑃 , intermittency 𝛾 and local
transition onset momentum thickness Reynolds number Re ̃ 𝜃 . In 4.1. General
𝑡
addition, the turbulence kinetic energy 𝑘 is set equal to zero and
𝜔 is specified at the near-wall cell center (Menter, 1994).
Results are presented for the marine propeller P4119 in open-water
In this study, the turbulence conditions. The propeller open-water characteristics are expressed by
√ level is specified at the inlet from the
turbulence intensity 𝑇 𝑢 = 100 2𝑘∕(3𝑉𝐴2 ) and eddy-viscosity ratio 𝜈𝑡 ∕𝜈, the advance coefficient 𝐽 , thrust and torque coefficients 𝐾𝑇 , 𝐾𝑄 , and
where 𝑉𝐴 is the incoming flow velocity and equal to the propeller the open-water efficiency 𝜂0 as follows:
advance speed. Due to the strong dependence of the 𝛾 − Re ̃ 𝜃 transition 𝑉𝐴 𝑇 𝑄 𝐽 𝐾𝑇
𝑡
model on the inlet turbulence quantities, 𝑇 𝑢 = 1.2% and 1.5% and 𝐽= , 𝐾𝑇 = , 𝐾𝑄 = , 𝜂0 = , (6)
𝑛𝐷 𝜌𝑛2 𝐷4 𝜌𝑛2 𝐷5 2𝜋 𝐾𝑄
5
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
Fig. 7. Convergence of 𝐶𝑝 (left) and 𝐶𝑓 (right) at 𝑠∕𝑐 = 0.2 (top) and 0.7 (bottom) on the blade suction side with grid refinement ratio ℎ𝑖 ∕ℎ1 .
where 𝑛 = 𝛺∕(2𝜋) is the rotation rate in rps, 𝐷 the propeller diameter, 𝑇 is a reference velocity defined as the undisturbed onset velocity at
the propeller thrust and 𝑄 the propeller torque. Other useful quantities the radial position 𝑟. The use of the propeller advance speed on the
are the pressure coefficient 𝐶𝑝 and the skin friction coefficient 𝐶𝑓 , definition of the skin friction coefficient allows a better identification
defined as: of the transition location over the entire propeller blade, whereas for
𝑝 − 𝑝∞ 𝜏𝑤 the pressure coefficient the analysis is carried out at different radial
𝐶𝑝 = , 𝐶𝑓 = , (7) sections and the standard definition based on the undisturbed onset
2
1∕2𝜌𝑉ref 1∕2𝜌𝑉𝐴2
velocity to those sections considered.
where 𝑝∞ is the undisturbed static pressure here defined√at the inlet The RANS simulations are compared with open-water tests and
boundary, 𝜏𝑤 is the local wall shear stress and 𝑉ref = 𝑉𝐴2 + (𝛺𝑟)2 LDV measurements of the blade boundary-layer that were performed
6
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
̃ 𝜃 transition model with 𝑇 𝑢 = 1.2% (top) and 𝑇 𝑢 = 1.5% (middle) for 𝜈𝑡 ∕𝜈 = 500,
Fig. 9. Limiting streamlines and skin friction distribution over the propeller blade using the 𝛾 − Re 𝑡
and 𝑘 − 𝜔 SST turbulence model with 𝑇 𝑢 = 1.0% and 𝜈𝑡 ∕𝜈 = 1 (bottom).
in the DTMB 24 in. water tunnel. The propeller was tested at 10 rps First, the iterative error is analyzed for the RANS simulations with
at an advance coefficient of 𝐽 = 0.833, corresponding to a Reynolds
̃ 𝜃 transition model. For the
the 𝑘 − 𝜔 SST turbulence model and 𝛾 − Re
number√ of 9.5 × 105 , based on chord-length 𝑐0.7𝑅 and inflow velocity 𝑡
𝑉0.7𝑅 = 𝑉𝐴2 + (0.7𝑅𝛺)2 at 0.7 of the propeller radius 𝑅. ̃ 𝜃 transition model, the inflow turbulence quantities are set to
𝛾 − Re 𝑡
1.5% and 500 for the 𝑇 𝑢 and 𝜈𝑡 ∕𝜈, respectively. The iterative error is
4.2. Numerical errors
estimated from the norms of the residuals of the momentum equations,
In CFD (Computational Fluid Dynamics) methods the three main
pressure correction, and transport equations of the turbulence and
contributions to the numerical error are: round-off error, iterative error
and discretization error (Oberkampf and Roy, 2010). Since double- transition models. In this work, we have adopted the 𝐿∞ norm:
precision is used in the present calculations, the round-off error is
neglected. Therefore, an analysis of the iterative and discretization
errors involved in the present computations is shown in this section. 𝐿∞ = max |res(𝜙𝑖 )|, 1 ≤ 𝑖 ≤ 𝑁cells , (8)
7
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
𝜙𝑖 = 𝜙0 + 𝛼ℎ𝑝𝑖 , (10)
8
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
where 𝑝 = 0.6. Lower numerical uncertainties are obtained with the Table 3
𝑘−𝜔 SST turbulence model. We note that large numerical uncertainties Estimated discretization error |𝜙1 − 𝜙0 |, apparent order of convergence 𝑝 and numerical
uncertainty 𝑈num for 𝐾𝑇 , 10𝐾𝑄 , 𝐶𝑝 and 𝐶𝑓 at 𝑠∕𝑐 = 0.2 and 0.7 on the blade suction
̃ 𝜃 transition
are estimated for the skin friction coefficient with the 𝛾 − Re 𝑡 side.
model. For sake of completeness, the apparent orders of convergence Model 𝑘 − 𝜔 SST ̃ 𝜃
𝛾 − Re 𝑡
and estimated numerical uncertainties are summarized in Table 3. The
𝜙 |𝜙1 − 𝜙0 | 𝑝 𝑈num |𝜙1 − 𝜙0 | 𝑝 𝑈num
discretization errors estimated from the least-square fits |𝜙1 − 𝜙0 |, are
𝐾𝑇 6.0 × 10−4 1.0 0.6% 1.2 × 10−3 1.8 1.0%
also presented in the table. In general, larger discretization errors are
10𝐾𝑄 1.5 × 10−4 1.8 0.2% 1.1 × 10−3 1.6 0.5%
estimated for the 𝛾 − Rẽ 𝜃 transition model.
𝑡 𝐶𝑝(0.2) 4.3 × 10−4 2.0 0.8% 3.1 × 10−4 1.0 3.7%
Since not only global quantities, but also local flow quantities are 𝐶𝑓(0.2) 4.3 × 10−5 1.0 3.6% 1.5 × 10−3 2.0 27.8%
analyzed and compared with the experimental results, the influence 𝐶𝑝(0.7) 5.3 × 10−5 2.0 0.6% 1.0 × 10−3 1.0 2.8%
of the grid density on the propeller forces and blade boundary-layer 𝐶𝑓(0.7) 8.8 × 10−4 – 3.6% 2.4 × 10−3 0.6 9.8%
velocity profiles are discussed. The variation of the thrust and torque
coefficients, and open-water efficiency for each grid compared to the
finest grid is listed in Table 4. Differences lower than 1% are obtained
9
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
Table 4
Variation of the force coefficients with grid density compared to the finest grid.
Model 𝑘 − 𝜔 SST ̃ 𝜃
𝛾 − Re 𝑡
Table 5
Pressure (𝑝 ) and friction (𝑓 ) contributions to the propeller thrust and torque and
open-water efficiency. Comparison with experimental data.
Model 𝑇𝑢 𝜈𝑡 ∕𝜈 𝐾𝑇𝑝 𝐾𝑇𝑓 𝐾𝑇
̃ 𝜃
𝛾 − Re 1.2% 500 0.150 −0.00227 0.148
𝑡
̃ 𝜃
𝛾 − Re 1.5% 500 0.148 −0.00313 0.145
𝑡
𝑘 − 𝜔 SST 1.0% 1 0.146 −0.00448 0.142
Exp. – – – – 0.146
Model 𝑇𝑢 𝜈𝑡 ∕𝜈 10𝐾𝑄𝑝 10𝐾𝑄𝑓 10𝐾𝑄
̃ 𝜃
𝛾 − Re 1.2% 500 0.254 0.0161 0.270
𝑡
̃ 𝜃
𝛾 − Re 1.5% 500 0.250 0.0226 0.272
𝑡
𝑘 − 𝜔 SST 1.0% 1 0.247 0.0312 0.278
Exp. – – – – 0.280
Model 𝑇𝑢 𝜈𝑡 ∕𝜈 – – 𝜂0
̃ 𝜃
𝛾 − Re 1.2% 500 – – 0.727
𝑡
̃ 𝜃
𝛾 − Re 1.5% 500 – – 0.707
𝑡
𝑘 − 𝜔 SST 1.0% 1 – – 0.676
Exp. – – – – 0.692
for the grid with 9.9 million cells (G3). Fig. 8 presents the chordwise
component 𝑉𝑠 of the blade boundary-layer velocity profile on the
suction side at 0.5 of the chord-length and 0.7 of the propeller radius.
Results of all grid sizes are compared for both models. The velocity
profile is presented along the normal direction, where 𝑦 is the wall
distance to blade surface and 𝑐 the chord-length. A minor influence
of the grid refinement level on the blade velocity profiles predicted
by the 𝑘 − 𝜔 SST turbulence model is observed. However, convergence
̃ 𝜃 transition model velocity profiles is difficult to obtain.
of the 𝛾 − Re 𝑡
The maximum differences in the velocity profile between the grids with
9.9 million cells and 37.6 million cells are of the order of 4%. The
results show that fine grids are required to capture the flow details.
Therefore, the results obtained with the finest grid (37.6 million cells)
are considered for the analysis of the blade flow and comparison with
experimental data.
In this section the propeller blade flow predictions using the tur-
Fig. 12. Predicted chordwise component 𝑉𝑠 of the flow velocity on the pressure side
bulence and transition models are examined. For the analysis of the at 𝑟∕𝑅 = 0.7 using the 𝑘 − 𝜔 SST turbulence model and 𝛾 − Re ̃ 𝜃 transition model with
𝑡
different flow solutions, the limiting streamlines and skin friction co- different inlet turbulence quantities. Comparison with experimental data (Jessup, 1989).
efficient 𝐶𝑓 on the blade surface are presented in Fig. 9. The location
where flow transition occurs is identified by the sudden rise of the skin
friction distribution and the change in the orientation of the limiting The predicted open-water characteristics, including the pressure and
streamlines towards the chordwise direction. The results obtained with friction contributions to the force coefficients, are compared with the
the 𝛾 − Rẽ 𝜃 transition model present a laminar flow region with
𝑡 experimental results in Table 5. Due to the growth in the turbulent
transition to turbulent flow at some distance from the leading-edge.
flow region, an increase in the magnitude of the friction contribution
A strong sensitivity to the inlet turbulence quantities in the prediction
is observed. At the same time, a decrease in the pressure contribution
of laminar-to-turbulent transition is observed. The domain of laminar
is observed, which may be explained by a decambering effect due to
flow reduces with the increase of the turbulence intensity at the inlet.
As expected, the 𝑘 − 𝜔 SST turbulence model leads to turbulent flow on a thicker boundary-layer. A better agreement with the experimental
the propeller blades. In this case, a short laminar flow region is present torque coefficient is obtained with the 𝑘 − 𝜔 SST turbulence model.
along the leading-edge. This comparison shows that detailed experi- For the thrust coefficient and open-water efficiency, a better agreement
mental information is crucial for the selection of the inlet turbulence ̃ 𝜃 transition model
with the experimental data is obtained for the 𝛾 − Re 𝑡
quantities when using the 𝛾 − Rẽ 𝜃 transition model. with 𝑇 𝑢 = 1.5% and 𝜈𝑡 ∕𝜈 = 500 at the inlet boundary.
𝑡
10
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
In addition to the propeller forces, the predicted blade pressure propeller frame of reference and is calculated as follows:
distribution is compared with experimental measurements. Since the [ ] [ ]
pressure distributions presented in Jessup (1989) were derived from the 𝛥𝑝𝑡 = 𝑃 + 1∕2𝜌 𝑉𝑋2 + 𝑉𝑌2 + 𝑉𝑍2 − 𝑃inlet − 1∕2𝜌 𝑉𝐴2 + (𝛺𝑟)2 , (12)
measured chordwise and radial velocity components at the boundary- with 𝑃inlet = 𝑝∞ + 2∕3𝜌𝑘inlet , which results in a negative variation of
layer edge, a similar approach is considered in this study for the the total pressure inside the boundary-layer. In the present work, the
RANS simulations. Following Jessup (1989) the pressure coefficient is boundary-layer thickness is defined by the distance normal to the wall
calculated by: 2 )
to a point where the total pressure loss coefficient 𝐶𝛥𝑝𝑡 = 𝛥𝑝𝑡 ∕(1∕2𝜌𝑉ref
( ) ( )
𝑉𝑠 2 𝑉𝑟 2 is equal to −0.01, corresponding approximately to the definition of
𝐶𝑝 = 1 − − , (11) 99.5% of the free-stream velocity at that point.
𝑉ref 𝑉ref
√ The comparison between the calculated pressure distributions with
with 𝑉ref = 𝑉𝐴2 + (𝛺𝑟)2 , where 𝑉𝑠 and 𝑉𝑟 are the chordwise and radial the turbulence and transition models and the experimental data is
velocity components in the propeller frame of reference calculated at presented at the radial sections 𝑟∕𝑅 = 0.3, 0.7 and 0.9 in Fig. 10. The
the edge of the boundary-layer 𝛿, respectively. The estimated boundary- non-dimensional chordwise location is defined as 𝑠∕𝑐. Similar results
layer thickness 𝛿 is obtained from the total pressure loss 𝛥𝑝𝑡 in the are obtained between the turbulence model and transition model. A
11
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
reasonable to good agreement between the numerical results and the 4.4. Comparison of velocity profiles
experimental data is obtained at the radial sections 𝑟∕𝑅 = 0.7 and
0.9. At the radial section 𝑟∕𝑅 = 0.3, some discrepancies between the In this section a comparison of the velocity profiles in the blade
calculations and experiments are visible in the pressure distribution. boundary-layer between the RANS simulations and the experiments
The larger disagreement observed at 𝑟∕𝑅 = 0.3 in comparison with is made. The chordwise 𝑉𝑠 and radial 𝑉𝑟 components of the flow
the other radial sections, may be due to the boundary-layer thick- velocity were measured by a LDV system comprising a one compo-
ness corrections applied by Jessup (1989) to the chordwise location nent lens system with a fiber optic probe. The measurement volume,
of the measurements. This correction is especially important towards formed by the interception of the two transmission beams, is placed
the blade root, due to the thicker blade sections. The values larger tangential to the blade surface at each measurement point. As the
than 1.0 obtained at the trailing edge indicate lack of accuracy in propeller rotates, velocity measurements along the boundary layer are
the determination of the measured chordwise positions. In addition, acquired. The chordwise and radial velocity components are measured
some differences are observed near the blade leading-edge for all radial by setting the plane that contains the transmission beams tangentially
sections. Jessup (1989) suggested that these differences may be due to or perpendicularly to the blade surface, respectively. In order to fully
errors in the geometry model. capture the large flow gradients near the blades, a measurement volume
12
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
13
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
̃ 𝜃
Fig. 16. Boundary-layer thickness 𝛿 (top), displacement thickness 𝛿 ∗ (middle), and shape factor 𝐻 (bottom) at 𝑟∕𝑅 = 0.7 using the 𝑘 − 𝜔 SST turbulence model and 𝛾 − Re 𝑡
transition model with different inlet turbulence quantities. Comparison with experimental data (Jessup, 1989).
5. Conclusions and pressure and skin friction coefficients (local quantities) has been
determined from an estimation of the discretization error based on
In this paper, a comparison between RANS calculations and ex- systematically refined grids. Low numerical uncertainties (less than
perimental data has been presented. The analysis is carried out for 2%) are obtained for the propeller force coefficients. For the local
marine propeller P4119, for which three-dimensional velocity com- quantities, small numerical uncertainties (less than 4%) are predicted
ponent measurements on the blade boundary-layer are available at with the 𝑘 − 𝜔 SST turbulence model. For the 𝛾 − Rẽ 𝜃 transition model,
𝑡
model-scale. In the model propeller, two blades had rough leading- larger uncertainties are obtained (more than 9%), especially for the skin
edges and the other blade was smooth. For the RANS calculations two friction coefficient. The numerical predictions of the velocity profiles
models are considered: the commonly-used 𝑘 − 𝜔 SST eddy-viscosity have been compared with blade boundary-layer measurements. An
turbulence model, and the 𝛾 − Re ̃ 𝜃 transition model. The influence agreement is found between the turbulent velocity profile predicted by
𝑡
of the iterative errors, discretization errors and boundary conditions the 𝑘 − 𝜔 SST turbulence model and the measured velocities from the
on the propeller flow predictions has been analyzed. The 𝛾 − Re ̃ 𝜃 ̃ 𝜃 transition model, the agreement with the
tripped blade. For the 𝛾 − Re
𝑡 𝑡
transition model does not satisfy the iterative convergence criterion. smooth blade velocity measurements is highly dependent on the inlet
Still, its influence on the propeller forces is assumed to be small turbulence quantities. In this work, from the selected inlet turbulence
due to the fast convergence of the thrust and torque coefficients. quantities, a qualitative agreement is obtained for the blade boundary-
The numerical uncertainty of the propeller forces (integral quantities) layer velocity profiles. However, this information may not be available,
14
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
Acknowledgments
The authors are indebted to José Falcão de Campos, Luís Eça and
Rui Lopes for many fruitful discussions on this topic.
References
Abdel-Maksoud, M., Heinke, H., 2002. Scale effects on ducted propellers. In:
Proceedings of the 24th Symposium on Naval Hydrodynamics. Fukuoka, Japan.
Baltazar, J., Rijpkema, D., Falcão de Campos, J.A.C., 2018. On the use of the 𝛾 − Re ̃ 𝜃
𝑡
transition model for the prediction of the propeller performance at model-scale.
Ocean Eng. 170, 6–19.
Bhattacharyya, A., Krasilnikov, V., Steen, S., 2016a. Scale effects on open water
characteristics of a controllable pitch propeller working within different duct
designs. Ocean Eng. 112, 226–242.
Bhattacharyya, A., Krasilnikov, V., Steen, S., 2016b. A CFD-based scaling approach for
ducted propellers. Ocean Eng. 123, 116–130.
Bulten, N., Nijland, M., 2011. On the development of a full-scale numerical towing
tank Reynolds scaling effects on ducted propellers and wakefields. In: Proceedings
of the Second International Symposium on Marine Propulsors. Hamburg, Germany.
Eça, L., Hoekstra, M., 2008. The numerical friction line. J. Mar. Sci. Technol. 13 (4),
328–345.
Eça, L., Hoekstra, M., 2014. A procedure for the estimation of the numerical uncertainty
of CFD calculations based on grid refinement studies. J. Comput. Phys. 262,
104–130.
Funeno, I., 2002. On viscous flow around marine propellers - hub vortex and scale
effect. J. Kansai Soc. N. A 238, 17–27.
GridPro, 2019. http://www.gridpro.com.
Hasuike, N., Okazaki, M., Okazaki, A., Fujiyama, K., 2017. Scale effects of marine
propellers in POT and self propulsion test conditions. In: Proceedings of the Fifth
International Symposium on Marine Propulsors. Espoo, Finland.
International Towing Tank Conference, 1978. Final report and recommendations of
the propulsion committee to the 15th ITTC. In: Proceedings of the 15th ITTC. the
Hague, the Netherlands.
International Towing Tank Conference, 2017. 1978 ITTC Performance Prediction
Method. Technical Report ITTC - Recommended Procedures and Guidelines
Fig. 17. Boundary-layer thickness 𝛿 (top), displacement thickness 𝛿 ∗ (middle), and 7.5-02-03-01.4, Revision 07.
shape factor 𝐻 (bottom) on the suction side at 𝑟∕𝑅 = 0.9 using the 𝑘 − 𝜔 SST Jessup, S.D., 1989. An Experimental Investigation of Viscous Aspects of Propeller Blade
̃ 𝜃 transition model with different inlet turbulence quantities.
turbulence model and 𝛾 −Re Flow (Ph.D. Thesis). The Catholic University of America.
𝑡
Comparison with experimental data (Jessup, 1989). Krasilnikov, V., Sun, J., Halse, K., 2009. CFD investigation in scale effects on propellers
with different magnitude of skew in turbulent flow. In: Proceedings of the First
International Symposium on Marine Propulsors. Trondheim, Norway.
Kuiper, G., 1981. Cavitation Inception on Ship Propeller Models (Ph.D. Thesis). Delft
̃ 𝜃 transition model
which limits the predictive capabilities of the 𝛾 − Re University of Technology.
𝑡
Langtry, R., 2006. A Correlation-Based Transition Model using Local Variables for
at model-scale, specially because the selected inlet values are not real-
Unstructured Parallelized CFD Codes (Ph.D. Thesis). University of Stuttgart.
istic from the physical point of view. Finally, boundary-layer integral Langtry, R., Menter, F., 2009. Correlation-based transition modeling for unstructured
quantities obtained from the measured and computed velocity profiles parallelized computational fluid dynamics codes. AIAA J. 47, 2894–2906.
have been analyzed. The evolution of these quantities along the chord- Menter, F., 1994. Two-equation eddy viscosity turbulence models for engineering
applications. AIAA J. 32, 1598–1605.
wise direction shows typical laminar and turbulent flow behaviors.
Menter, F., Esch, T., Kubacki, S., 2002. Transition modelling based on local variables.
The comparison between the RANS calculations and the experimental In: Proceedings of the 5th International Symposium on Engineering Turbulence
measurements shows that the considered models can be used, not only Modelling and Measurements. Mallorca, Spain.
for the prediction of integral quantities, but also for the analysis of the Menter, F., Kuntz, M., Langtry, R., 2003. Ten years of industrial experience with the
local features of the propeller blade flow. Nevertheless, the selection SST turbulence model. In: Proceedings of the Fourth International Symposium on
Turbulence, Heat and Mass Transfer, Vol. 4. pp. 625–632.
of the inlet turbulence quantities is crucial in the prediction of the Moran-Guerrero, A., Gonzalez-Gutierrez, L.M., Oliva-Remola, A., Diaz-Ojeda, H.R.,
transition location, and in the accuracy of the model-scale propeller 2018. On the influence of transition modeling and crossflow effects on open water
performance prediction. Therefore, the assessment of the predictive propeller simulations. Ocean Eng. 156, 101–119.
15
J. Baltazar et al. Ocean Engineering 211 (2020) 107633
Müller, S.-B., Abdel-Maksoud, M., Hilbert, G., 2009. Scale effects on propellers for large Spalart, P., Rumsey, C., 2007. Effective inflow conditions for turbulence models in
container vessels. In: Proceedings of the First International Symposium on Marine aerodynamic calculations. AIAA J. 45, 2544–2553.
Propulsors. Trondheim, Norway. Stanier, M., 1998. The application of RANS code to investigate propeller scale effects.
Oberkampf, W.L., Roy, C.J., 2010. Verification and Validation in Scientific Computing. In: Proceedings of the 22th ONR Symposium on Naval Hydrodynamics. Washington,
Cambridge University Press.. DC, USA.
ReFRESCO, 2019. http://www.refresco.org. Uto, S., 1993. Computation of incompressible viscous flow around a marine propeller.
Rijpkema, D., Baltazar, J., Falcão de Campos, J.A.C., 2015. Viscous flow simulations J. Soc. Nav. Archit. Japan 173, 67–75.
of propellers in different Reynolds number regimes. In: Proceedings of the Fourth Walters, D.K., Cokljat, D., 2008. A three-equation eddy-viscosity model for reynolds-
International Symposium on Marine Propulsors. Austin, Texas, USA. averaged Navier–Stokes simulations of transitional flow. ASME J. Fluids Eng. 130
Rijpkema, D., Vaz, G., 2011. Viscous flow computations on propulsors: verification, (12), 121401.1–121401.14.
validation and scale effects. In: Proceedings of the International Conference on Wang, X., Walters, K., 2012. Computational analysis of marine-propeller performance
Developments in Marine CFD. London, UK. using transition-sensitive turbulence modeling. ASME J. Fluids Eng. 134 (7),
Sánchez-Caja, A., González-Adalid, J., Pérez-Sobrino, M., Sipilä, T., 2014. Scale effects 071107.1–071107.10.
on tip loaded propeller performance using a RANSE solver. Ocean Eng. 88, Yao, H., Zhang, H., 2018. Numerical simulation of boundary-layer transition flow of a
607–617. model propeller and the full-scale propeller for studying scale effects. J. Mar. Sci.
Shin, K.W., Andersen, P., 2017. CFD analysis of scale effects on conventional and Technol. 23, 1004–1018.
tip-modified propellers. In: Proceedings of the Fifth International Symposium on
Marine Propulsors. Espoo, Finland.
16