Jafm 2207 1426 1 2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Applied Fluid Mechanics, Vol. x, No. x, pp. x-x, 200x.

Available online at www.jafmonline.net, ISSN 1735-3572, EISSN 1735-3645.


https://doi.org/10.47176/jafm.xx.xx.xxxxx

Shape Optimization of an Asymmetric Airfoil for low


Wind speed region having Adjoint based optimization
technique
Pranoti Shinde 1†, S. S. Ohol 2, Vipin K. Tripathi 3
1 Research Scholar, Mechanical Engineering Department, College of Engineering Pune, Pune, Maharashtra,
411005, India
2 Associate Professor, Mechanical Engineering Department, College of Engineering Pune, Pune,

Maharashtra, 411005, India


3 Professor, Mechanical Engineering Department, National Institute of Technology, Agartala, Tripura,

799046, India

†Corresponding Author Email: sp18.mech@coep.ac.in

ABSTRACT

The power generation from the renewable energy source is extensively increasing and shifting towards
decreasing the large area to implant wind turbines. For that reason, a considerable amount of study is going on
to increase the power extraction from the wind turbines for low wind speed regions. The Wind turbine blades
play a significant major role in utilizing the maximum amount of energy from the wind. The wind turbine
blade's aerodynamic characteristics depend on the airfoil shape. The shape optimization of an Asymmetric
S2027 airfoil for a low wind speed region is studied using the Adjoint-based optimization technique. The most
important objectives of the study are to maximize the coefficient of lift, coefficient of drag minimization, and
maximization of lift to drag ratio. The optimization is based on the adjoint method for Re no. variation in the
range of 2×105 to 5×105 and angle of attack variation from 0° to 12°. A two-dimensional RANS CFD model is
created initially for all operating parameters and deployed for optimization. Experimental validation has been
carried out for aerodynamic performance. About 16 shapes are obtained for each optimization function. The
aerodynamic performance at different operating conditions of each optimized shape is found. Different airfoil
shapes with a specific form of a mean line, leading and trailing edge, and width arrangement is obtained. About
2-30% reduction in coefficient drag, 2-35% increment in coefficient of lift, and up to 40% increment in lift/drag
ratio is achieved.

Keywords: Airfoil shape optimization; Wind turbine blade; XFOIL; ANSYS Fluent; Discrete adjoint
Technique; Wind tunnel Experimentation

NOMENCLATURE

C is camber Vd design variable


c chord length Vff flow field variable
Cd coefficient of drag 𝑣𝑡 eddy viscosity
Cl coefficient of lift xi, xj spatial coordinates
Cmax Maximum chamber of an airfoil ɛ rate of dissipation of turbulent kinetic energy
D Drag force α Angle of attack
g computational grid 𝛽*, 𝛽1 , 𝛽2 , 𝜎𝑘1 , 𝜎𝑘2 , 𝜎𝜔1 , 𝜎𝜔2 , 𝛾1 , 𝛾2 Equation
k turbulence kinetic energy constant
L lift force 𝜕
Eulerian time derivative
R flow field 𝜕𝑡
Re Reynolds number ρ density of the fluid
t thickness of the airfoil ω specific rate of dissipation
Tmax Maximum thickness of an airfoil 𝜇 viscosity
𝑢𝑖 the time derivative of velocity within a flow 𝜇𝑡 viscosity in the RANS region
Vc lagrangian multiplier vector 𝜏𝑖𝑗 stress tensor
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

1. INTRODUCTION The demand for efficient systems for generation of


energy from alternate sources is increasing rapidly
(Kang & Park, 2013). In
In wind turbines, the effectiveness continuously optimized shapes are analysed using X-foil for
increases with improvement in airfoil design. The evaluation of the performance of each optimized
airfoil shape optimization for improvement in wind airfoil with a varying range of input parameters. For
turbine blades aerodynamics for higher energy the solving of both flow and adjoint equations,
production and reduction in the cost of energy has streamline-upwind and pressure-stabilized
been carried out by various researchers (Chen & Petrov/Galerkin (SUPG and PSPG) stabilization
Wang, 2018; Fuglsang & Bak, 2004; Q. Wang et al., methods are applied (Srinath & Mittal, 2010;
2015) Tezduyar et al., 1992).
The requirement to decrease the while devoted to the
growth series, from beginning till the concluding The research work shown in this paper is basically
product at the factory outlet, has commanded the depending on the urban weather condition—the data
arrival of various optimization methods. Various collected on the circumambient condition to better
researchers have optimized airfoils based on understand working situations. The novelty of the
different objective functions and constraints having research work is the method used to reduce the
different tools and methods of optimization (Sharma computational time and efficiency of results. The
et al., 2020). Aerodynamic shape optimization along discrete adjoint solver technique for shape
with computational fluid dynamics is in optimization of an airfoil is implemented. The
consideration for a long time. modified shapes of the airfoil could work in urban
areas' weather conditions according to the
CST parameterization (Liang & Li, 2018), requirement. The research is the first step toward
parameterization (Della Vecchia et al., 2014; shifting the use of extensive field areas to tight areas
Fuglsang & Bak, 2004), genetic algorithm (Bedon et without wastage of land to generate power from the
al., 2016; Liang & Li, 2018; Mukesh et al., 2014; wind turbine.
Ram et al., 2019), adjoint method (Diwakar et al.,
2010; Srinath & Mittal, 2010), pressure load inverse 2. DISCRETE ADJOINT BASED
method (Henriques et al., 2009), and Neural network
OPTIMIZATION
(Mortazavi et al., 2015) are the methods adopted by
the various researchers. Since the airfoil shape
optimization elaborates numerous design The discrete adjoint approach is used for the
parameters, the gradient-based technique is utilized aerodynamic optimization of an airfoil. It is used to
with the purpose of attaining the optimum shape solve the flow equations around the surface of the
regarding the objective functions. The gradient- airfoil. It can be obtained by using the set of discrete
based optimization method is determined over field equations with the direct application of control
distinct computational methods like the finite theory (Nadarajah & Jameson, 2000).
difference and adjoint methods. Adjoint-based The wind turbine works for an extensive range of
optimization needs less time for computation to Reynolds numbers. The performance of wind
obtain the cost function gradient than the finite turbines varies as the range of Reynolds number
difference method as it can Fig. out the gradient in a changes due to the fluctuations in turbulence,
single cycle while not depending on design variables vibrational forces, airfoil surface roughness, etc.,
(Carpentieri et al., 2007; K. Wang et al., 2019). (Chen & Wang, 2018). In the present analysis, the
Adjoint-based optimization is the tool considered adjoint approach along with CFD simulation has
more efficient while handling the number of been applied for the optimized shape. Fig. 1 shows
variables (Gomes & Palacios, 2020; Othmer, 2014; the procedure for the present analysis and validation
Schramm et al., 2018). The adjoint-based approach adopted for the shape optimization of an
optimization can be employed for both shape and airfoil to reduce the computational time with respect
topology optimization. In a shape optimization to the airfoil shape optimization.
process having an adjoint method, the computational
field varies at each optimization cycle via re-meshing
(Karpouzas & De Villiers, 2014).
The objective of the current analysis is to optimize
the profile of the S2027 airfoil, with an adjoint
method for Re number variation in the range of
2×105-5×105 and angle of attack variation from 0°
to 12°. For the optimization, coefficient of lift and
lift/drag proportion maximization and drag
minimization are considered as the objective
function. Airfoil optimization is for wind turbine
application thus, the flow is considered time-
dependent and the real-valued functions are based on
over period averaged aerodynamic factors. The

2
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

Fig. 1. Shape Optimization strategy enclosed domain are about 20, 16, and 30 times
respectively, far away from the airfoil to neglect the
2.1. Numerical Aerodynamic Simulation boundary effect on the airfoil. Grid generated about
2.1.1. Governing Equations 1.14×105 and nodes on 280 airfoil surfaces. The y+
value was kept near 0.6 for a better degree of fineness
With the advanced computation facility, the Navier- in the wall region.
Stokes equation can accurately define the fluid
motion without any subdivision based on viscosity,
compressibility, and speed of flows. The Navier-
Stokes equations are the governing equations for
CFD simulations. For turbulency in fluid flows, the
equations are averaged, leading to Reynolds
averaged Navier-Stokes equations (RANS) (Halila et
al., 2020; Malone et al., 1991). Modelling of
Reynolds stress tensor is essential for the
mathematical explanation of RANS equations. It
mathematically could be written as:
Original k-ꞷ model:
(a)
𝜕 𝜕 𝜕𝑢𝑖 𝜕
+ 𝑢𝑖 (𝜌𝑘) = 𝜏𝑖𝑗 − 𝛽 ∗ 𝜌𝜔𝑘 + [(𝜇 +
𝜕𝑡 𝜕𝑥𝑖 𝜕𝑥𝑗 𝜕𝑥𝑗
𝜕𝑘
𝜎𝑘1 𝜇𝑡 ) ] (1)
𝜕𝑥𝑗
𝜕 𝜕 𝛾1 𝜕𝑢𝑖 𝜕
+ 𝑢𝑖 (𝜌𝜔) = 𝜏 − 𝛽1 𝜌𝜔2 + [(𝜇 +
𝜕𝑡 𝜕𝑥𝑖 𝑣𝑡 𝑖𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗
𝜕𝜔
𝜎𝜔1 𝜇𝑡 ) ] (2)
𝜕𝑥𝑗
Transformed k-ɛ model:
𝜕 𝜕 𝜕𝑢𝑖 𝜕
+ 𝑢𝑖 (𝜌𝑘) = 𝜏𝑖𝑗 − 𝛽 ∗ 𝜌𝜔𝑘 + [(𝜇 +
𝜕𝑡 𝜕𝑥𝑖 𝜕𝑥𝑗 𝜕𝑥𝑗
𝜕𝑘 (b)
𝜎𝑘2 𝜇𝑡 ) ] (3) Fig. 2. Generated grid system (a) Refined
𝜕𝑥𝑗
𝜕 𝜕 𝛾2 𝜕𝑢𝑖 𝜕 boundary of airfoil and (b) Meshed model
+ 𝑢𝑖 (𝜌𝜔) = 𝜏 − 𝛽2 𝜌𝜔 2 + [(𝜇 +
𝜕𝑡 𝜕𝑥𝑖 𝑣𝑡 𝑖𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 of C-sectional 2 D computational domain
𝜕𝜔 1 𝜕𝑘 𝜕𝑘
𝜎𝜔2 𝜇𝑡 ) ] + 2𝜌𝜎𝜔2 (4)
𝜕𝑥𝑗 𝜔 𝜕𝑥𝑗 𝜕𝑥𝑗 1.6

1.4
Coeffecient of Lift (Cl)

The analysis is based on time-dependent, two- 1.2


dimensional RANS equations. Menter developed the
Shear Stress Transport model (Menter, 1994), which 1
can be combined with closed-wall and far-field 0.8 No of grids 112000
regions. For modelling of turbulent flow, the two- No of grids 113000
0.6
equation eddy viscosity (SST k-omega) model can be No of grids 114000

applied, having higher accuracy and efficiency as it 0.4 No of grids 115000


No of grids 116000
syndicates the recompenses of both standard k- 0.2
omega and k-epsilon model by associated functions. -3 2 7 12
Angle of Attack (α)
Second-order upwind the discretization along with
pressure velocity couples SIMPLEC method is (a)
adopted for the analysis. For the simulation
procedure, finite volume, education licensed CFD 0.03
Coeffecient of Discharge (Cd)

software ANSYS is used.


0.025

2.1.2. Grid generation and dependence 0.02

Quadratic elements are used for grid generation. Fig. 0.015


2 shows the 2-dimensional generated grid system. 0.01
No of grids 112000
No of grids 113000
The grid size is more refined with matched No of grids 114000
convergence criterion at different flow situations 0.005 No of grids 115000
No of grids 116000
near the airfoil, while refinement decreases as 0
distance increases from the airfoil. C-shape, 2D -3 2 7 12
Angle of Attack (α)
computational domain has been generated as it is
adopted by various researchers (He et al., 2016; (b)
Mortazavi et al., 2015). The initial grid generation Fig. 3. Coefficient (a) lift and (b) drag
has been carried out for the S2027 airfoil with Re variation concerning angle of attack with
number 2×105 and angle of attack from 0-12°. The varied number of grids
inlet, top and bottom, and the outlet boundary of the

3
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

For analysis of grid dependence, both lift and drag side 300 x 300 x 1000
are the two variables (objective function) that are glass length (in millimeter)
considered. The variation of both the variables for material:
the S2027 airfoil concerning the angle of attack Size:
(design variable) with variations from 0 to 12° is Fan blade 5 blades – Aluminium
found at a different number of grids. It has been die cast
found that at the number of grids more than Air Velocity (test 2-30 m/sec
1.14×105, both the variables have the same variation section)
with respect to the angle of attack; it can be seen in Contraction ratio 9:1
Fig. 3, this is because of the independence of grid at Lift/drag force sensors 0-20kg beam type
and above 1.14×105. Thus, for the present analysis, load cell
1.14×105 grids are found to be more convenient and Multi-tube manometer height -300mm
adopted for further analysis. Anemometer Velocity range – 0-
display 30m/s
2.1.3. Boundary Conditions Digital
Regarding the boundary conditions, the inlet, upper Stain gauge balance two channels
and lower borders are established as the velocity inlet capacity a) Force – 0-
boundary condition. The outlet boundary is 20kg for lift
established by means of pressure outlet boundary b) Force – 0-
conditions. The outlet is at constant pressure with 20 kg for
atmospheric pressure value. Airfoil outer geometry drag
is established as a no-slip wall state. At airfoil
boundaries, the stream velocity is zero. The residuals An anemometer of digital type having range of
limit of convergence is considered as 10-6. The wind velocity of about 0-30 in meter per second, two
1
inlet velocities are varied in such a way that the Re channel 3 force indicator of digital type and
2
number varied from 2×105 to 5×105, having inlet channel strain gauge balance around two are in the
velocities of 3.3m/s, 4.95 m/s, 6.6m/s, and 8.25m/s, system for measurement. Based on the
respectively. The air fluctuation from mean manufacturers data, about 2.8% error in measuring
percentage and turbulent viscosity ratio is considered the velocity, can be observed.
as 6% and 9%, correspondingly.
Airfoil Manufacturing
2.2. Validation of Aerodynamic
characteristics analysis results The most effective way of manufacturing the airfoil
for wind tunnel testing is an additive manufacturing
Before going for the optimization process, the CFD technique. The precision in designing the
results are required to be validated. For validation, complicated asymmetric airfoil shape can be
both X foil and experimental analysis are carried out, achieved by using this technique. The airfoil
and results are compared for base model of the S2027 manufacturing machine is shown in Fig. 4. It is also
airfoil. The computational fluid dynamics method is an inexpensive way of developing any model for
performed in ANSYS Fluent for precision in the testing purposes. Post-processing for a smoother
result and reliable performance of the software. The surface can be done for accurate results while testing
validation of CFD results is important to scrutinize in the wind tunnel.
the boundary conditions and variables carefully.

2.2.1 Experimental Analysis


For experimental analysis, wind tunnel testing has
been carried out which is an open type wind tunnel.
The main duct having cross section of 300 mm×300
mm and length of 1000 mm. The tunnel having a
blower from which air speed can obtain in the range
of 2 to 30 m/sec. For measurement of lift and drag
forces 0-20 kg beam type load cell are used. Two
pitot tubes (5/16’’and a micro-0-20mm pitot tube) Fig. 4. Airfoil Manufacturing machine
and 0-50 mm WC inclined manometer are installed
in the system for setting and correction of flow
velocities as mentioned in the table 1.

Table 1 Wind Tunnel Specifications

Design Open type wind


testing tunnel
working area for Coating of powder
testing: with mild steel
Fig. 5. Pressure points to measure pressure
main Perspex sheet of about distribution on airfoil lower and upper
duct: 8 mm in thickness curved surface

4
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

Fig. 6. Supporting parts of airfoil

The fabricated airfoil in Fig. 5 has 11 pressure points


to study the pressure distribution over the curved
upper and lower surface of the airfoil. The pressure
points are further extruded for the attachment to read
the results. The dimension of the airfoil considered is
150mm chord length with a span length of about
150mm. Acrylonitrile butadiene styrene Spool of
radius 2.85mm is used as a material for the
development of an airfoil model. The supporting
parts can be seen in Fig. 6 attached to the airfoil to (b)
support the airfoil in the sections to test the Fig. 7. CFD, experimental and XFOIL
aerodynamic values. results validation for Coefficient of (a) lift
(b) drag regarding Re number and angle of
2.2.2 Environment for investigation in attack
XFOIL
The aerodynamic performance of the initial S2027 When the comparison is made in the results for both
airfoil is tested in the XFOIL tool at different Re lift and drag, it can be observed that the forces are in
number, i.e., 2×105, 3×105, 4×105, and 5×105. close proximity for the lower value of angle of attack
About 240-panel nodes are used. NCrit, i.e., free for lift and in the vicinity for the higher value of
flow turbulence level, is considered the same as CFD angle of attack for drag force. The effects of lift and
analysis. Viscous acceleration is zero. drag force for all three methods are within the
acceptable limit. Thus, this model setup of CFD
2.2.3 Performance analysis analysis can be used for further aerodynamic
analysis.
For validation, both experimental and XFOIL results
are compared for S2027 airfoil for coefficient of lift 3. SHAPE OPTIMIZATION
and drag at different Re number and angle of attack.
A minor difference can be obtained from XFOIL,
For Optimization, a deterministic, gradient-based
and experimental results can be seen in Fig. 7. In Fig.
non-linear constraint-based shape optimization
7(a), the lift force increases in the same particular
method is adopted. The Optimization is based on
behavior. The drag force in Fig. 7(b) also shows a
three different objective functions, i.e., maximization
similar response while increasing the angle of attack.
of lift (at constant drag), drag minimization (at
constant lift), and Maximization of lift/drag ratio.
Concerning the design approach, Lagrange Function,
L, can be described for the objective function and
adjoint variables as

L (Vd , Vff , g, Vc ) = f(Vd , Vff , g) + VCT R(Vd , Vff , g) (5)

The objective function is represented by, while R, are


the Lagrangian multiplier vector, flow field, design
variable, flow field variables, and computational
grid, respectively (M Schramm). The discrete adjoint
formulation can be given by

dL ∂f ∂g T ∂f ∂V T ∂f
=( +[ ] ) + [ ff] ( +
dVd ∂D ∂Vd ∂g ∂D ∂Vff
∂R T ∂R T ∂g T ∂R T
[ ] Vc ) + [ ] + ([ ] + [ ] Vc ) (6)
∂Vff ∂D ∂D ∂g

As the Lagrangian multiplier vector is basically


uninformed, thus the multiplied terms can be
(a) neglected. The term remains

5
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

After iterations, as the convergence takes place, the


∂R T ∂f rate of changes of the objective function selected
[ ] Vc = − (7)
∂Vff ∂Vff previously regarding each point position on the
airfoil surface is there with variable sensitivity for
The above equation represents the discrete adjoint particular boundary condition situations are
equation. The objective functions to optimize the obtained. After the computation of adjoint shape
shape of an airfoil is to modification, ANSYS Fluent provides the
Minimize / Maximize Fi =fi (α, Re), (i = 1, 2, 3) opportunity for both simultaneous applications of
Subject to: adjoint predictions with volume mesh Morpher
t ≤ Tmax optimizer having a smooth surface with numeral
C ≤ Cmax control points using Bernstein polynomials. S2027
airfoil is considered as baseline geometry. For the
Re(L) ≤ Reynolds no. (Re) ≤ Re(U) analysis, an enclosed control volume for airfoil was
α(L) ≤ Angle of attack(α) ≤ α(U) initially defined. A network of control points is
placed over the meshed surface for mesh morphing
to bring out the smooth changes of airfoil shape
Where chord length (c)=constant, F is defined as the
based on Bernstein polynomials. About 25 control
objective functions, the superscript for the variables
points along the chord length and 10 perpendiculars
L and U denotes the upper and lower limit of the
of the chord length are considered. The morphing
range of design variables. ‘i’ shows the no. of
box touches the airfoil boundaries with a complete
objective functions. The objective functions F1, F2,
enclosure.
and F3 stand for Minimization of coefficient of drag,
Maximization of coefficient of lift, and
Maximization of coefficient of lift to drag ratio, 4. RESULTS OF OPTIMIZED PROFILES
respectively. Tmax is a Maximum value of thickness
which is 15% at 30% c and Cmax is the Maximum For the optimization process, three different
chamber which is 2.5% at 39% c of an airfoil. The objectives are considered. The optimization has been
lower limit of the Reynolds no. and angle of attack is carried out based on different Re number and angle
2 x 105 and 0˚ with the increment of 1 x 10 5 and 4˚, of attack. The Re number varied from 2×105 to
respectively. The upper limit for both the design 5×105, and the angle of attack varied from 0-12°.
variable is 5 x 105 and 12˚, respectively. After the adjoint optimization process, airfoil
performance for varied series of variables with
Continuous and discrete method are the two methods different angle of attack has been obtained using
for the adjoint solver. In the continuous method, XFOIL and CFD simulation process.
unique relations for partial derivatives are linearized,
then adjoints are formed, mathematically discreted,
4.1 Minimization of coefficient of drag
and explained. While in the case of others, the non-
linear stream is linearized completely, having a In the optimization process, we found the optimal
linear numerical arrangement that is required to shape considering the minimization of coefficient of
answer for the adjoint result. For the present analysis, drag at different Re number and angle of attack. Fig.
ANSYS Fluent (academic 19.3 R) is used in which 8 shows the different airfoil profiles gotten with the
discrete adjoint solver is employed as it proposed objective function of minimization of coefficient of
robustness and consistency and evades the foisting of drag for different Re number and angle of attack. The
adjoint environment compulsory for continuous shape changes happen mainly on the leading-edge
method (Petrone et al., 2014). thickness and at the maximum thickness surface of
the airfoil.

Fig. 8. Airfoil profiles with the objective function of minimization of coefficient of drag at
various Re number and angle of attack

6
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

Fig. 9. Variation of lift coefficient at different Reynolds no. of the optimized airfoil for
an objective function, viz. Minimization of coefficient of drag
The graphs in Fig. 9 show fluctuations from almost 2-30% reduction in coefficient drag can be
maximum to minimum value for the coefficient of obtained for all the optimized airfoil shapes.
drag for each optimized airfoil obtained from a
selected range of angle of attack and Reynolds 4.2 Maximization of coefficient of lift
number. As the objective function demands
minimization of the function value, all the optimized This section discusses the optimal shapes obtained
airfoils, as shown in Fig. 8, manifest the minimum for one of the three objective functions, i.e.,
drag value for the span of 0˚ to 12˚ angle of attack. maximization of the coefficient of lift. The new
The observations are made based on the variation of airfoil profiles acquired with the objective function,
drag coefficient for minimum and maximum values maximization of coefficient of lift at various Re
from 0.000315 to 0.057949, respectively, in the numbers, and angle of attack are depicted in Fig. 10.
graphs of Fig. 9 for different angles of attack at The airfoil thickness decreases with a lower Re
specific Reynolds numbers. A study of the number, while changes in the leading edge are
aerodynamic behavior of these airfoils has also been massive for higher Re number considerations.
carried out for the coefficient of lift and L/D ratio
variation for chosen design parameters. Overall,

Fig. 10. Airfoil profiles with the objective function of maximization of coefficient of lift at
different Re numbers and angle of attack

7
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

Fig. 11. Airfoil profiles with the objective function of maximization of coefficient of lift at
different Re number and angle of attack

Fig. 11 shows the variation in coefficient of lift of The third objective function, lift/drag ratio, has been
respective airfoil for a wide range of angle of attack considered for optimization. Fig. 12 shows the
at a particular Re number. In Fig. 11, the lift optimal airfoil shapes obtained to maximize lift to
coefficient shows fluctuations in the merits of its drag at different Re no. and angle of attack. The
objective function. The maximum value of optimized airfoil thickness does not seem to be more affected
shapes of airfoils is in the range of 1.5 to 2. This while the leading-edge shape shows different shapes
range shows the slightest deviations in the value. The with higher Re number, and it is more when angle of
implemented optimization process grants an attack is also higher.
increment of 2 to 35% in Lift coefficient.

4.3 Maximization of lift/drag ratio

Fig. 12.Airfoil profiles with the objective function of maximization of lift/drag ratio for
various Re number and angle of attack

8
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

Fig. 13. Variation of lift coefficient at different Reynolds no. of the optimized airfoil for an
objective function viz. Maximization of L/D ratio

Fig. 13 shows the L/D ratio variation of respective maximize the coefficient of lift & L/D ratio and to
airfoils. The application of the optimization process minimize the drag coefficient. The new series of
shows up to a 40% lift/drag ratio increment. airfoils contains a total 48 number of profiles (refer
Furthermore, A study for CL, CD, and L/D ratio on fig 8, 10, and 12). Each profile is novel from one
each selected objective function's optimized shapes another. Furthermore, each profile is numerically
of airfoils is performed to understand the overall investigated. The study of aerodynamic
aerodynamic behavior of all the optimized airfoils. It characteristics of a new airfoil series is also essential
has been observed that all the optimized airfoil for all the objectives (refer to fig 9, 11, and 13).
shows satisfactory and acceptable nature for other 1. The numerical analysis method shows good
objectives indeed. Improvement in the aerodynamic agreement with the Xfoil and experimental
characteristics in comparison to base airfoil is method results. A difference of up to only 5 %
perceived. is observed in all the methods applied for
Numerical Analysis and Experimental
5. DISCUSSION AND CONCLUSION validation (refer to fig. 7). The difference in
the data is because of the external factors and
This paper presents the aerodynamic characteristics slight error in the equipment used for the
of an asymmetric S2027 airfoil for low wind speed experimental analysis.
region targeting the urban areas. The Aerodynamic 2. The new series of 16 airfoils (refer to fig 8) for
behaviour of the any airfoil is highly dependent on drag coefficient minimization shows a
the main two design parameters. The parameters are
tremendous reduction in drag coefficient
Reynolds number and angle of attack. The selection
compared to the S2027 airfoil. The objective to
of Reynolds number ranging from 2 x 10 5 to 5 x 105
is accordance with the wind speed in urban area. reduce in drag coefficient is about 2 to 30 %
Numerical Investigation on asymmetric S2027 achieved (refer to fig 9).
airfoil is performed for a span of angle of attack from 3. Improved aerodynamic behavior for one of the
0˚ to 12˚. The results of numerical investigation are three objectives, i.e., maximization of lift
compared with XFOIL and experimental results. The coefficient, is accomplished. The lift
Experiments are performed in wind tunnel testing coefficient's merit increment is up to 2 to 35%
equipment on an asymmetric S2027 airfoil for experienced (refer to fig 11). The increment in
Numerical results validation. the lift coefficient is appreciated for the new
An adjoint-based optimization approach is followed series of 16 airfoils (refer to fig 10).
for producing a series of new airfoil profiles. CFD 4. Aerodynamic Investigation shows good
RANS-based model was created to solve both adjoint statistics for 16 profiles (refer to fig 12) of new
and fluid flow problems and validated using airfoil series for maximization of the Lift-to-
experimental and Xfoil results for primary S2027 Drag ratio. The statistics for the objective, i.e.,
airfoil. The objectives for enhancing the Maximization of L/D ratio, exhibit up to 40%
aerodynamic characteristics of an airfoil are to

9
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

better performance than the S2027 airfoil Henriques, J. C. C., Marques da Silva, F.,
(refer to fig 13). Estanqueiro, A. I., & Gato, L. M. C. (2009).
Design of a new urban wind turbine airfoil
The performance of the S2027 airfoil is enhanced via using a pressure-load inverse method.
shape optimization, and the aerodynamic Renewable Energy, 34(12), 2728–2734.
performance of the new airfoil series is also https://doi.org/10.1016/j.renene.2009.05.011
magnificent. The enhanced aerodynamic Kang, T. J., & Park, W. G. (2013). Numerical
characteristics are due to the changes in the shape of investigation of active control for an S809
an airfoil. The shape changes in the airfoil are at wind turbine airfoil. International Journal of
maximum thickness section, and the leading edges Precision Engineering and Manufacturing,
are observed with smooth surfaces. Overall, an 14(6), 1037–1041.
airfoil shape is introduced for every combination of https://doi.org/10.1007/s12541-013-0139-2
design parameters, which is further aerodynamically Karpouzas, G. K., & De Villiers, E. (2014). Level-
studied and shows improved aerodynamic set based topology optimization using the
performance at low wind speed regions. continuous adjoint method. OPT-i 2014 - 1st
International Conference on Engineering
REFERENCES and Applied Sciences Optimization,
Proceedings, June, 56–74.
Bedon, G., De Betta, S., & Benini, E. (2016). Liang, C., & Li, H. (2018). Effects of optimized
Performance-optimized airfoil for Darrieus airfoil on vertical axis wind turbine
wind turbines. Renewable Energy, 94, 328– aerodynamic performance. Journal of the
340. Brazilian Society of Mechanical Sciences and
https://doi.org/10.1016/j.renene.2016.03.071 Engineering, 40(2).
Carpentieri, G., Koren, B., & van Tooren, M. J. L. https://doi.org/10.1007/s40430-017-0926-2
(2007). Adjoint-based aerodynamic shape Nadarajah, S. K., & Jameson, A. (2000). A
optimization on unstructured meshes. comparison of the continuous and discrete
Journal of Computational Physics, 224(1), adjoint approach to automatic
267–287. aerodynamic optimization. Aerospace
https://doi.org/10.1016/j.jcp.2007.02.011 Science Meeting and Exhibit.
Chen, J., & Wang, Q. (Eds.). (2018). Wind Turbine https://doi.org/10.2514/6.2000-667
Airfoils and Blades. Science Press Ltd. and
Walter de Gruyter GmbH, Malone, J. B., Narramore, J. C., & Sankar, L. N.
Beijing/Berlin/Boston. (1991). Airfoil design method using the
Della Vecchia, P., Daniele, E., & D’Amato, E. Navier-Stokes equations. Journal of Aircraft,
(2014). An airfoil shape optimization 28(3), 216–224.
technique coupling PARSEC https://doi.org/10.2514/3.46015
parameterization and evolutionary algorithm. Menter, F. R. (1994). Two-equation eddy-viscosity
Aerospace Science and Technology, 32(1), turbulence models for engineering
103–110. applications. AIAA Journal, 32(8), 1598–
https://doi.org/10.1016/j.ast.2013.11.006 1605. https://doi.org/10.2514/3.12149
Diwakar, A., Srinath, D. N., & Mittal, S. (2010). Mortazavi, S. M., Soltani, M. R., & Motieyan, H.
Aerodynamic shape optimization of airfoils (2015). A Pareto optimal multi-objective
in unsteady flow. CMES - Computer optimization for a horizontal axis wind
Modeling in Engineering and Sciences, turbine blade airfoil sections utilizing exergy
69(1), 61–89. analysis and neural networks. Journal of
Fuglsang, P., & Bak, C. (2004). Development of the Wind Engineering and Industrial
Risø wind turbine airfoils. Wind Energy, Aerodynamics, 136, 62–72.
7(2), 145–162. https://doi.org/10.1016/j.jweia.2014.10.009
https://doi.org/10.1002/we.117 Mukesh, R., Lingadurai, K., & Selvakumar, U.
Gomes, P., & Palacios, R. (2020). Aerodynamic (2014). Airfoil shape optimization using non-
driven multidisciplinary topology traditional optimization technique and its
optimization of compliant airfoils. AIAA validation. Journal of King Saud University -
Scitech 2020 Forum, 1 PartF, 2117–2130. Engineering Sciences, 26(2), 191–197.
https://doi.org/10.2514/6.2020-0894 https://doi.org/10.1016/j.jksues.2013.04.003
Halila, G. L. O., Martins, J. R. R. A., & Fidkowski, Othmer, C. (2014). Adjoint methods for car
K. J. (2020). Adjoint-based aerodynamic aerodynamics. Journal of Mathematics in
shape optimization including transition to Industry, 4(1), 1–23.
turbulence effects. Aerospace Science and https://doi.org/10.1186/2190-5983-4-6
Technology, 107, 106243. Petrone, G., Hill, D. C., & Biancolini, M. E. (2014).
https://doi.org/10.1016/j.ast.2020.106243 Track by track robust optimization of a F1
He, X., Wang, J., Yang, M., Ma, D., Yan, C., & front wing using adjoint solutions and radial
Liu, P. (2016). Numerical simulation of basis functions. 32nd AIAA Applied
Gurney flap on SFYT15thick airfoil. Aerodynamics Conference, June, 1–9.
Theoretical and Applied Mechanics Letters, https://doi.org/10.2514/6.2014-3174
6(6), 286–292. Ram, K. R., Lal, S. P., & Ahmed, M. R. (2019).
https://doi.org/10.1016/j.taml.2016.09.002 Design and optimization of airfoils and a

10
Pranoti Shinde et al. / JAFM, Vol. x, No. x, pp. x-x, 200x.

20 kW wind turbine using multi-objective


genetic algorithm and HARP_Opt code.
Renewable Energy, 56–67.
https://doi.org/10.1016/j.renene.2018.08.040
Schramm, M., Stoevesandt, B., & Peinke, J. (2018).
Optimization of airfoils using the adjoint
approach and the influence of adjoint
turbulent viscosity. Computation, 6(1), 1–23.
https://doi.org/10.3390/computation6010005
Sharma, P., Gupta, B., Pandey, M., Sharma, A. K.,
& Nareliya Mishra, R. (2020). Recent
advancements in optimization methods for
wind turbine airfoil design: A review.
Materials Today: Proceedings, 47(xxxx),
6556–6563.
https://doi.org/10.1016/j.matpr.2021.02.231
Srinath, D. N., & Mittal, S. (2010). Optimal
aerodynamic design of airfoils in unsteady
viscous flows. Computer Methods in Applied
Mechanics and Engineering, 199(29–32),
1976–1991.
https://doi.org/10.1016/j.cma.2010.02.016
Tezduyar, T. E., Mittal, S., Ray, S. E., & Shih, R.
(1992). Incompressible flow computations
with stabilized bilinear and linear equal-
order-interpolation velocity-pressure
elements. Computer Methods in Applied
Mechanics and Engineering, 95(2), 221–242.
https://doi.org/10.1016/0045-7825(92)90141-
6
Wang, K., Yu, S., Wang, Z., Feng, R., & Liu, T.
(2019). Adjoint-based airfoil optimization
with adaptive isogeometric discontinuous
Galerkin method. Computer Methods in
Applied Mechanics and Engineering, 344,
602–625.
https://doi.org/10.1016/j.cma.2018.10.033
Wang, Q., Wang, J., Chen, J., Luo, S., & Sun, J.
(2015). Aerodynamic shape optimized design
for wind turbine blade using new airfoil
series. Journal of Mechanical Science and
Technology, 29(7), 2871–2882.
https://doi.org/10.1007/s12206-015-0616-x

11

You might also like