optimization of sailplane wing parameter
DESCRIPTION
Engineering Optimization Process of Sailplane wing parameter to obtain minimum rate of descent with given constraint and boundary.Optimization are using Genetic Algorithm, Simulated Annealing and fmincon solver of Matlab 2013aKeyword: Sailplane, Glider, Optimization, GA,TRANSCRIPT
MAJOR ASSIGNMENT REPORT
OPTIMIZATION OF SAILPLANE WINGS TO FIT INDONESIA’S
ATMOSPHERE PROPERTIES
Submitted as Requirement for Completing Engineering Optimization Course
Hendi Aji Pratama
23613300
Advisor:
Dr. Rais Zain
AERONAUTICS AND ASTRONAUTICS ENGINEERING
FACULTY OF MECHANICAL AND AEROSPACE ENGINEERING
INSTITUTE TECHNOLOGY OF BANDUNG
2014
CHAPTER 1
INTRODUCTION
1.1 Background
One of Indonesia Aerosport activity is gliding or soaring with a sailplane. The fact that
Indonesia only has one type of sailplane which is Schweizer SGS 1-26 that are
manufactured at 1950 make the achievement of its community is very limited. The
main problem of SGS 1-26 is not because this sailplane is bad, but it is not quite
compatible with Indonesia climate and atmospheric condition. SGS 1-26 may perform
greatly in another side of the world, but in Indonesia, it’s just not good enough.
This is the list the specification and performance of the SGS 1-26:
The rate of descent of SGS 1-26 may still climb the thermal updraft of Indonesia,but it
is not low enough. Thermal updraft only occurs in a few minutes, so climbing the
thermal as fast and as high as possible is quite a mandatory. SGS 1-26 performance
cannot help the sailplane pilot to reach the peak of Indonesia’s thermal. Based on this
problem, a sailplane need to be designed, to fit the atmosphere and thermal updraft
properties of Indonesia.
Figure 1.1 Schweizer SGS 1-26
The design need to have Rate of Descent / Sink as low as possible, so it can climb
thermal effectively. The Glide Ratio needs to as high as possible so it can fly further.
The requirement of the design is able to climb Indonesia’s weak and narrow thermal .
Indonesia’s thermal mostly have 200 – 300 meter diameter and 5000 m in height. The
strength of the thermal is varied from 0.5-3.0 m/s. That makes the glider needs to
have rate of sink less than 0.5 m/s , just to barely climb the thermal.
To answer the requirement, optimum design need to be found. The optimization are
targeted to find the optimum physical properties that related to the requirements.
Properties those related to rate of sink and glide ratio are wing Aspect Ratio, wing
area,lift coefficient CL,drag coefficient CD,and weight of the aircraft. To simplify the
problem, the parameter those will be optimized are wing area, aspect ratio, and
weight.
1.2 Purpose and Goals
The purpose of this project is to perform an optimization to sailplane design
parameter that will result in good performance of the sailplane,which is indicated by
minimum value of Rate of Descent. The goal of this project is to achive optimum design
parameter and minimum value of Rate of Descent that way lower than Indonesia’s
thermal strength.
CHAPTER 2
PROBLEM FORMULATION
2.1 Objective Function and Optimization Parameter
The objective of this optimization is to found the minimum value of Sinking Speed /
Rate of Descent which can formulated as following equation:
𝑉𝑠𝑖𝑛𝑘 =𝐶𝐷
𝐶𝐿
√2𝑊
𝜌𝑆𝐶𝐿𝑐𝑜𝑠∅ … (1)
The fact that to achieve rate of sink, The value of W must be greater than L, it can
become a constraint:
𝐿 < 𝑊 … (2)
To simplify the optimization, constraint (2) can be turn into some equation that help
completing the objective function:
𝐿 + 𝛿 = 𝑊 . . (3)
And then using the equation of lift:
𝐿 =1
2𝜌𝑣2𝑆𝐶𝐿 … (4)
Equation (3) become:
𝐶𝐿 =2(𝑊 − 𝛿)
𝜌𝑉2𝑆 … (5)
Another equation that may helping this optimization is the drag coefficient equation:
𝐶𝐷 = 𝐶𝐷0 +𝐶𝐿
2
𝜋 𝐴 𝑒… (6)
In which, the value of CL can be calculated using equation (5). The value of can be
calculated with the function of A as shown in this equation:
𝑒 = 1.78(1 − 0.045 𝐴0.68) − 0.64) … (7)
To find the value of V in equation (5), equation of turning radius as shown below,will
be also put into the objective function:
𝑅 =𝑉2
𝑔 𝑡𝑎𝑛𝜑… (8)
The Objective function then can be completed using substitution of equation (5) ,
(6), (7) and (8) into equation 1. From that substitution, then using assumption:
- Air density, ρ = 1.225 kg / m3
- Drag Coefficient, CD0 = 0.0125
- Banking Angle, Φ = 30⁰
The equation then left 5 variables that will used as the optimization parameter:
𝑋 = [ 𝑥1 𝑥2 𝑥3 𝑥4 𝑥5] = [𝑆 𝐴 𝑊 𝑅 𝛿]
2.2 Goal and Constraint
The main goal of this optimization is parameter of glider’s wing that proven can be
used to climb a thermal updraft that has vertical speed of 0.5 m/s - 3.0 m/s. Then the
optimization is success if at least:
0.5 ≤ 𝐹(𝑥) ≤ 3.0
To narrow the optimization, these are the constraint that will be used in this
optimization:
- Glide Ratio: 10 ≤ 𝐶𝐿/𝐶𝐷 ≤ 150
- Wing Area : 10 ≤ S ≤ 30
- Aspect Ratio: 10 ≤ A ≤ 25
- Glider’s weight, 220 kg ≤ W ≤ 350 kg
- Turn Radius, 30 m ≤ R ≤ 150 m
- −1000 ≤ 𝛿 ≤ −0.01
CHAPTER 3
OPTIMIZATION
3.1 Optimization Process
Optimization process are done using 3 solver that available in Matlab R2013a: Genetic
Algorithm, Simulated Annealing, and Constrained Minimization (fmincon). All of the
solver will have 3 initial point, which are:
- Outside Boundary : X = [100 100 10000 400 -7000]
- Inside Boundary 1 : X = [20 20 3000 100 -500]
- Inside Boundary 2 : X = [15 11 2500 50 -10]
Total there will be 5 computation that will be done in this optimization. Genetic
Algorithm and fmincon will be executed twice,once using Glide Ratio constraint (see
section 2), and another one will be without Glide Ratio constraint. Simulated
Annealing will executed without Glide Ratio Constraint, because currently, Matlab’s
simulated annealing only support bound constrainted optimization. These are the
optimization step of the 3 solver:
3.1.1 Constrained Optimization (fmincon)
The computation using fmincon are using following code:
a. With Glide Ratio constraint:
function [history,searchdir] = runfmincon
history.x = [];
history.fval = [];
searchdir = [];
UB=[30 25 3500 150 -0.01];
LB=[10 10 2200 30 -1000];
x0 = [100 100 10000 400 -7000];
options =
optimoptions(@fmincon,'OutputFcn',@outfun,...
'Display','iter','Algorithm','active-set');
xsol =
fmincon(@obfun,x0,[],[],[],[],LB,UB,@confun,options);
function stop = outfun(x,optimValues,state)
stop = false;
switch state
case 'init'
hold on
case 'iter'
history.fval = [history.fval;
optimValues.fval];
history.x = [history.x; x];
searchdir = [searchdir;...
optimValues.searchdirection'];
iter=optimValues.iteration+1;
n(iter)=iter;
x1(iter)=x(1);
x2(iter)=x(2);
x3(iter)=x(3)/100;
x4(iter)=x(4);
x5(iter)=x(5)/100;
fx(iter)=optimValues.fval;
title('Sequence of Points Computed by
fmincon');
case 'done'
hold off
otherwise
end
end
function f = obfun(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
f=(CD/CL)*sqrt(2*x(3)/(1.061*x(1)*CL));
end
function [c, ceq] = confun(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
c =[10-CL/CD;
CL/CD-150];
ceq = [];
end
[ax,h1,h2]=plotyy(n,[x1' x2' x3' x4'
x5'],n,fx,'plot','plot');
legend([h1;h2],'x1 - S(m2)','x2 - AR','x3 - W(x 100
N)','x4 - R(m)','x5 - delta (x 100 N)','fx -
Vsink(m/s)','Location','southwest','Orientation','ver
tical');
end
b. Without glide ratio constraint
%=========================
% FMINCON
% Without Constraints
%=========================
function [history,searchdir] = runfmincon
history.x = [];
history.fval = [];
searchdir = [];
%BOUNDS
UB=[30 25 3500 150 -0.01];
LB=[10 10 2200 30 -1000];
%INITIAL VALUES
x0 = [15 11 2500 50 -10];
options =
optimoptions(@fmincon,'OutputFcn',@outfun,...
'Display','iter','Algorithm','active-set');
[x,fval]=
fmincon(@obfun,x0,[],[],[],[],LB,UB,[],options)
function stop = outfun(x,optimValues,state)
stop = false;
switch state
case 'init'
hold on
case 'iter'
history.fval = [history.fval;
optimValues.fval];
history.x = [history.x; x];
searchdir = [searchdir;...
optimValues.searchdirection'];
iter=optimValues.iteration+1;
n(iter)=iter;
x1(iter)=x(1);
x2(iter)=x(2);
x3(iter)=x(3)/100;
x4(iter)=x(4);
x5(iter)=x(5)/100;
fx(iter)=optimValues.fval;
title('Sequence of Points Computed by
fmincon');
case 'done'
hold off
otherwise
end
end
%OBJECTIVE FUNCTION
function f = obfun(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
f=(CD/CL)*sqrt(2*x(3)/(1.061*x(1)*CL));
end
%CONSTRAINT FUNCTION
function [c, ceq] = confun(x)
% Nonlinear inequality constraints
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
c =[10-CL/CD;
CL/CD-150];
% Nonlinear equality constraints
ceq = [];
end
subplot (3,2,1)
plot(n,x1)
title('Wing Area - x1 (m2)')
ylabel('x1 (m2)')
subplot (3,2,2)
plot(n,x2)
title('Aspect Ratio - x2 ')
ylabel('x2')
subplot (3,2,3)
plot(n,x3)
title('Weight - x3 (N)')
ylabel('x3 (N)')
subplot (3,2,4)
plot(n,x4)
title('Turn Radius - x4 (m)')
ylabel('x4 (m)')
subplot (3,2,5)
plot(n,x5)
title('Delta [W-L]- x5 (N)')
xlabel ('Number of iteration')
ylabel('x5 (N)')
subplot (3,2,6)
plot(n,fx)
title('Rate of Descent - f(x) (m/s)')
xlabel ('Number of iteration')
ylabel('f(x) (m/s)')
ha=axes('Position',[0 0 1 1],'Xlim',[0 1],'Ylim',[0
1],'Box','off','Visible','off','Units','normalized','
clipping','off');
text(0.5,1,'\bf Fmincon Result (WITHOUT Glide Ratio
Constraint, Initial point inside boundary - 2
)','HorizontalAlignment','center','VerticalAlignment'
,'top')
end
c.
The output of the solver and the plot of parameter change in every iteration,at every
cases can be seen in Appendix A at the back of this report.
3.1.2 Genetic Algorithm
The computation using Genetic Algorithm are using following code:
a. With Glide Ratio constraint:
%==================================
% GENETIC ALGORITHM
% (with glide ratio constraint)
%==================================
function [history,searchdir] = ga_ma2
ObjectiveFunction = @ma_objfun_ga;
nvars = 5; % Number of variables
% BOUND
UB=[30 25 4000 150 -0.01];
LB=[10 10 2200 30 -1000];
% INITIAL VALUES
X0=[15 11 2500 50 -10];
ConstraintFunction = @ma_constraint;
opt = gaoptimset('OutputFcns',@ga_outfun, ...
'Display','iter','Generations',50,'TolFun',1E-
20,'TolCon',1E-
20,'Mutation',@mutationadaptfeasible,'InitialPopulati
on',X0);
[x,fval,exitflag] =
ga(ObjectiveFunction,nvars,[],[],[],[],LB,UB, ...
ConstraintFunction,opt)
function [state, options,optchanged] =
ga_outfun(options,state,flag)
optchanged = false;
switch flag
case 'init'
hold on
case 'iter'
iter=state.Generation;
n(iter)=iter;
x1(iter)=state.Population(1,1);
x2(iter)=state.Population(1,2);
x3(iter)=state.Population(1,3);
x4(iter)=state.Population(1,4);
x5(iter)=state.Population(1,5);
fx(iter)=state.Best(iter);
case 'done'
hold off
end
end
subplot (3,2,1)
plot(n,x1)
title('Wing Area - x1 (m2)')
ylabel('x1 (m2)')
ylim([25 32])
subplot (3,2,2)
plot(n,x2)
title('Aspect Ratio - x2 ')
ylabel('x2')
subplot (3,2,3)
plot(n,x3)
title('Weight - x3 (N)')
ylabel('x3 (N)')
subplot (3,2,4)
plot(n,x4)
title('Turn Radius - x4 (m)')
ylabel('x4 (m)')
subplot (3,2,5)
plot(n,x5)
title('Delta [W-L]- x5 (N)')
xlabel ('Number of generation')
ylabel('x5 (N)')
subplot (3,2,6)
plot(n,fx)
title('Rate of Descent - f(x) (m/s)')
xlabel ('Number of generation')
ylabel('f(x) (m/s)')
ha=axes('Position',[0 0 1 1],'Xlim',[0 1],'Ylim',[0
1],'Box','off','Visible','off','Units','normalized','
clipping','off');
text(0.5,1,'\bf Genetic Algorithm Result (WITH Glide
Ratio Constraint, Initial point inside boundary - 2
)','HorizontalAlignment','center','VerticalAlignment'
,'top')
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1))
e=(1.78*(1-0.045*(x(2)^0.68))-0.64)
CD=0.0125+(CL^2/(3.14*x(2)*e))
CL_CD=CL/CD
V=sqrt(x(4)*6.3)
%OBJECTIVE/FITNESS FUNCTION
function f=ma_objfun_ga(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
f=(CD/CL)*sqrt(2*x(3)/(1.061*x(1)*CL));
end
%CONSTRAINTS FUNCTION
function [c, ceq] = ma_constraint(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
c =[10-CL/CD;
CL/CD-150];
ceq = [];
end
end
b. Without Glide Ratio constraint:
%================================== % GENETIC ALGORITHM % (without glide ratio constraint) %==================================
function [history,searchdir] = ga_ma2 ObjectiveFunction = @ma_objfun_ga; nvars = 5; % Number of variables % BOUND UB=[30 25 4000 150 -0.01]; LB=[10 10 2200 30 -1000];
% INITIAL VALUES X0=[15 11 2500 50 -10]; ConstraintFunction = @ma_constraint; opt = gaoptimset('OutputFcns',@ga_outfun, ... 'Display','iter','Generations',50,'TolFun',1E-
20,'TolCon',1E-
20,'Mutation',@mutationadaptfeasible,'InitialPopulation',X
0);
[x,fval,exitflag] =
ga(ObjectiveFunction,nvars,[],[],[],[],LB,UB, ... [],opt)
function [state, options,optchanged] =
ga_outfun(options,state,flag) optchanged = false; switch flag case 'init' hold on case 'iter'
iter=state.Generation; n(iter)=iter;
x1(iter)=state.Population(1,1); x2(iter)=state.Population(1,2); x3(iter)=state.Population(1,3); x4(iter)=state.Population(1,4); x5(iter)=state.Population(1,5); fx(iter)=state.Best(iter);
case 'done' hold off end end
subplot (3,2,1) plot(n,x1) title('Wing Area - x1 (m2)') ylabel('x1 (m2)') ylim([25 32])
subplot (3,2,2) plot(n,x2) title('Aspect Ratio - x2 ') ylabel('x2')
subplot (3,2,3) plot(n,x3) title('Weight - x3 (N)') ylabel('x3 (N)')
subplot (3,2,4) plot(n,x4) title('Turn Radius - x4 (m)') ylabel('x4 (m)')
subplot (3,2,5) plot(n,x5) title('Delta [W-L]- x5 (N)') xlabel ('Number of generation') ylabel('x5 (N)')
subplot (3,2,6) plot(n,fx) title('Rate of Descent - f(x) (m/s)') xlabel ('Number of generation') ylabel('f(x) (m/s)')
ha=axes('Position',[0 0 1 1],'Xlim',[0 1],'Ylim',[0
1],'Box','off','Visible','off','Units','normalized','clipp
ing','off'); text(0.5,1,'\bf Genetic Algorithm Result (WITHOUT Glide
Ratio Constraint, Initial point inside boundary - 2
)','HorizontalAlignment','center','VerticalAlignment','top
')
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1)) e=(1.78*(1-0.045*(x(2)^0.68))-0.64) CD=0.0125+(CL^2/(3.14*x(2)*e)) CL_CD=CL/CD V=sqrt(x(4)*6.3)
%OBJECTIVE/FITNESS FUNCTION function f=ma_objfun_ga(x) CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1)); e=(1.78*(1-0.045*x(2)^0.68)-0.64); CD=0.0125+(CL^2/(3.14*x(2)*e)); f=(CD/CL)*sqrt(2*x(3)/(1.061*x(1)*CL)); end
%CONSTRAINTS FUNCTION function [c, ceq] = ma_constraint(x) CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1)); e=(1.78*(1-0.045*x(2)^0.68)-0.64); CD=0.0125+(CL^2/(3.14*x(2)*e)); c =[10-CL/CD; CL/CD-150]; ceq = []; end end
The output of the solver and the plot of parameter change in every generation, at
every cases, can be seen in Appendix A at the back of this report.
3.1.3 Simulated Annealing
The computation using Simulated Annealing are using following code:
%==========================
% SIMULATED ANNEALING
%==========================
function [history,searchdir] = simulated_annealing
Objfun=@ma_objfunc_sa;
%INITIAL VALUES
x0=[15 11 2500 50 -10];
%BOUNDS
UB=[30 25 4000 150 -0.01];
LB=[10 10 2200 30 -1000];
options = saoptimset('OutputFcns',@sa_output);
[x,fval,exitflag,output]=simulannealbnd(Objfun,x0,LB,UB,op
tions)
function [stop,options,optchanged] =
sa_output(options,optimvalues,flag)
stop = false;
optchanged = false;
switch flag
case 'init'
hold on;
case 'iter'
iter=optimvalues.iteration;
n(iter)=iter;
x1(iter)=optimvalues.bestx(1);
x2(iter)=optimvalues.bestx(2);
x3(iter)=optimvalues.bestx(3)/100;
x4(iter)=optimvalues.bestx(4);
x5(iter)=optimvalues.bestx(5)/100;
fx(iter)=optimvalues.bestfval;
case 'done'
hold off;
end
end
subplot (3,2,1)
plot(n,x1)
title('Wing Area - x1 (m2)')
ylabel('x1 (m2)')
ylim([5 35])
subplot (3,2,2)
plot(n,x2)
title('Aspect Ratio - x2 ')
ylabel('x2')
subplot (3,2,3)
plot(n,x3)
title('Weight - x3 (N)')
ylabel('x3 (N)')
subplot (3,2,4)
plot(n,x4)
title('Turn Radius - x4 (m)')
ylabel('x4 (m)')
subplot (3,2,5)
plot(n,x5)
title('Delta [W-L]- x5 (N)')
xlabel ('Number of iteration')
ylabel('x5 (N)')
subplot (3,2,6)
plot(n,fx)
title('Rate of Descent - f(x) (m/s)')
xlabel ('Number of iteration')
ylabel('f(x) (m/s)')
ha=axes('Position',[0 0 1 1],'Xlim',[0 1],'Ylim',[0
1],'Box','off','Visible','off','Units','normalized','clipp
ing','off');
text(0.5,1,'\bf Simulated Annealing Result (WITHOUT Glide
Ratio Constraint, Initial point inside boundary - 2
)','HorizontalAlignment','center','VerticalAlignment','top
')
%OBJECTIVE/FITNESS FUNCTION
function f=ma_objfunc_sa(x)
CL= 2*(x(3)-x(5))/(6.93*x(4)*x(1));
e=(1.78*(1-0.045*x(2)^0.68)-0.64);
CD=0.0125+(CL^2/(3.14*x(2)*e));
f=(CD/CL)*sqrt(2*x(3)/(1.061*x(1)*CL));
end
end
The output of the solver and the plot of parameter change in every generation, at
every cases, can be seen in Appendix A at the back of this report.
CHAPTER 4
ANALYSIS AND CONCLUSION
4.1 Solver Analysis & Conclusion
Table 4.1 Result Summary
Table 4.1 above shows the summary of the result of the 3 solvers. Genetic Algorithm and
Simulated Annealing show unconsistent result from every execution. It mostly caused by the
concept of these optimization is creating random number and calculate the fitness/objective
function until the optimum result is found. However, from the iteration plot, GA result is more
stable when SA result is more likely to oscilate randomly.
Fmincon yield consistent result , event in constraint-unconstrainted case. But the result is still
not as good as GA. With different initial point it seems that fmincon end up in different
minima. It should be because the solver found the local minima and assume it as the global
minima,so the algortihm stopped.
Different initial point also affect the result of GA and SA, but it’s unclear if it’s because
of one of the initial point closer to the global minima, or because of the random
algorithm of SA and FA . But from the plot of the parameter change in every iteration,
initial point inside the constraint are more likely to make some parameter reach
convergence faster. This may be the proof that the initial point is closer to the global
minima.
4.2 Optimization Conclusion
The optimum parameters obtained from this optimization, taken from the GA best
result are:
Wing Area : 30 m2
Aspect Ratio : 23
Weight : 2279 N
Turn Radius : 30 m
That will yield optimum Rate of Descent 0.51 m/s. However, the set of result of the
optimization are only the ideal numbers. In practical, it’s very difficult to design a
sailplane with those set of parameter. For example with Aspect Ratio of 23 the wing
and the aircraft will more likely need tougher structure and become very heavy. It may
be practically unfeasible to have the aircraft weighted 2279 N.
To understand the behavior of every parameter in the optimization, variations of
boundary / side constraints are applied. Each time, one parameter are varied when
the others still the same as the constraints explained at the early part of this report.
Figure 4.1 Aspect Ratio vs Rate Of Descent
0
10
20
30
40
50
60
70
0.4 0.45 0.5 0.55
Asp
ect
Rat
io
Rate Of Descent (m/s)
Aspect RatioBounds
OptimumAspect Ratio
Table 4.1 shows the behavior of rate of descent based on aspect ratio. Rate of descent
tend to become lower when Aspect Ratio is bigger. However, at some point of
increasing Aspect Ratio, the rate of descent are not decreasing anymore, and the
optimum aspect ratio chosen by solver are not the upper bound of the aspect ratio.
Figure 4.2 Wing Area vs Rate of Descent
Table 4.2 shows the behavior of rate of descent based on wing area. Rate of descent
tend to become lower when wing area is bigger. However, at some point of increasing
wing area, the rate of descent are not decreasing significantly anymore, and the
optimum aspect ratio chosen by solver are not the upper bound of the aspect ratio.
Figure 4.3 Weight vs Rate Of Descent
0
500
1000
1500
2000
2500
3000
0.44 0.46 0.48 0.5 0.52 0.54 0.56
Wei
ght
(N)
Rate Of Descent (m/s)
Weight LowerBounds
Optimum Weight
0
5
10
15
20
25
30
0.51 0.52 0.53 0.54 0.55 0.56 0.57
Win
g A
rea
(m2
)
Rate Of Descent (m/s)
Wing Area Bounds
Optimum Wing Area
Table 4.3 shows the behavior of rate of descent based on weight. Rate of descent tend
to become lower at lower weight. However, based on the result trend of the
optimization the weight seems not significantly affect the optimization. The optimum
value barely moving close to upper or lower bound. It may be caused by that weight
are not the critical parameter that define rate of descent.
Figure 4.4 Turn Radius vs Rate Of Descent
Table 4.4 shows the behavior of rate of descent based on turn radius. Rate of descent
tend to become lower at lower turn radius. The optimum value of turn radius is more
likely very close to the lower bound.
0
10
20
30
40
50
60
0.48 0.5 0.52 0.54 0.56 0.58 0.6
Turn
Rad
ius
(m)
Rate Of Descent (m/s)
Turn Radius Lower Bound
Optimum Turn Radius
CHAPTER 5
RECOMMENDATION
Based on the result and conclusion of this optimization project, future works that can be done
related to this optimization project should discover and use another solver. Matlab still have
limitation that need to be fixed. Trial using another solver or maybe the improved version of
Matlab solver needs to be done to validate which are the real best solver for this
optimization.
Optimization topic in this project also still need to be explored deeper. In this project, basicly
the parameter that are used are only four, Where the rest of parameter related to rate of
descent are simplified using assumed value. The case in this project is also still the simplified
version, because the optimization are only centered in sailplane wing. In the real case, rate of
descent should be affected not only by the wing, but also by the fuselage and the tail. In future
works, many parameters are need to be added, such as wing twist, taper, incidence, fuselage
drag, tail drag, etc. to find the real optimum design of sailplane that have minimum rate of
descent.
The algorithm applied to the solver also need many improvements. One necessary
improvement is control to the constraints, so the combination optimum value that come out
from the solver are really make sense in the real world. So the optimization result might
applied to the the real sailplane design project.
REFERENCE
[1] FAA. Glider Flying Handbook, Skyhorse Publishing,Washington DC.July 2007
[2] Thomas,Fred. Fundamentals of Sailplane Design. College Park Press, Maryland.
1999
[3] Handojo,Vega H. 2013. Sailplane Performance Estimation. ITB Internship Report,
Bandung
[4] Rujgrok, G.J.J. 1990. Element of Airplane Performance. Delft University Press,
Netherland
[5] Pratama, Hendi A. 2014. Definition of Glider Design Basic Parameter Based on
Thermal Characteristic of Indonesia. Design Project 1 Report. Bandung
APPENDIX A
TABLE A.1 Result of Constrained Minimization (fmincon)
With glide ratio constraint.initial point outside boundary x0 = [100 100 10000 400 -7000]
Without glide ratio constraint.initial point outside boundary x0 = [100 100 10000 400 -7000]
With glide ratio constraint.initial point inside boundary - 1 x0 = [20 20 3000 100 -500]
Without glide ratio constraint.initial point inside boundary - 1 x0 = [20 20 3000 100 -500]
With glide ratio constraint.initial point inside boundary -2 X0 = [15 11 2500 50 -10];
With glide ratio constraint.initial point inside boundary -2 X0 = [15 11 2500 50 -10];
Figure A.1 Plot of Parameter Changes in Every Iteration (fmincon, With Glide Ratio, Initial Point Outside Boundary)
0 5 10 15 20 25 3010
15
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 25 3018
20
22
24
26Aspect Ratio - x2
x2
0 5 10 15 20 25 3034.985
34.99
34.995
35
35.005Weight - x3 (N)
x3 (
N)
0 5 10 15 20 25 300
50
100
150Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25 30-10
-10
-10Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 25 300.5
1
1.5Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Fmincon Result (With Glide Ratio Constraint, Initial point outside boundary)
Figure A.2 Plot of Parameter Changes in Every Iteration (fmincon, With Glide Ratio, Initial Point Inside Boundary - 1)
0 5 10 15 20 2510
15
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 2520
22
24
26Aspect Ratio - x2
x2
0 5 10 15 20 2529.996
29.998
30
30.002
30.004Weight - x3 (N)
x3 (
N)
0 5 10 15 20 2520
40
60
80
100Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25-5.03
-5.02
-5.01
-5Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 250.4
0.6
0.8
1Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Fmincon Result (With Glide Ratio Constraint, Initial point inside boundary - 1)
Figure A.3 Plot of Parameter Changes in Every Iteration (fmincon, With Glide Ratio, Initial Point Inside Boundary -2)
0 5 10 15 20 2515
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 2510
15
20
25Aspect Ratio - x2
x2
0 5 10 15 20 2524.999
24.9995
25
25.0005Weight - x3 (N)
x3 (
N)
0 5 10 15 20 2530
35
40
45
50Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25-0.104
-0.102
-0.1
-0.098Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 250.4
0.6
0.8
1Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Fmincon Result (With Glide Ratio Constraint, Initial point inside boundary - 2)
Figure A.4 Plot of Parameter Changes in Every Iteration (fmincon, Without Glide Ratio, Initial Point Outside Boundary )
0 5 10 15 20 25 3010
15
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 25 3018
20
22
24
26Aspect Ratio - x2
x2
0 5 10 15 20 25 3034.985
34.99
34.995
35
35.005Weight - x3 (N)
x3 (
N)
0 5 10 15 20 25 300
50
100
150Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25 30-10
-10
-10Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 25 300.5
1
1.5Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Fmincon Result (WITHOUT Glide Ratio Constraint, Initial point ouside boundary )
Figure A.5 Plot of Parameter Changes in Every Iteration (fmincon, Without Glide Ratio, Initial Point Inside Boundary - 1)
0 5 10 15 20 2510
15
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 2520
22
24
26Aspect Ratio - x2
x2
0 5 10 15 20 2529.996
29.998
30
30.002
30.004Weight - x3 (N)
x3 (
N)
0 5 10 15 20 2520
40
60
80
100Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25-5.03
-5.02
-5.01
-5Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 250.4
0.6
0.8
1Rate of Descent - f(x) (m/s)
Number of iterationf(
x)
(m/s
)
Fmincon Result (With Glide Ratio Constraint, Initial point inside boundary - 1)
Figure A.6 Plot of Parameter Changes in Every Iteration (fmincon, Without Glide Ratio, Initial Point Inside Boundary - 2)
0 5 10 15 20 2515
20
25
30Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 2510
15
20
25Aspect Ratio - x2
x2
0 5 10 15 20 2524.999
24.9995
25
25.0005Weight - x3 (N)
x3 (
N)
0 5 10 15 20 2530
35
40
45
50Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25-0.104
-0.102
-0.1
-0.098Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 5 10 15 20 250.4
0.6
0.8
1Rate of Descent - f(x) (m/s)
Number of iterationf(
x)
(m/s
)
Fmincon Result (WITHOUT Glide Ratio Constraint, Initial point inside boundary - 2 )
TABLE A.2 Result of Genetic Algorithm
With glide ratio constrain, initial point outside boundary x0 = [100 100 10000 400 -7000]
Withoulide ratio constraint.initial point outside boundary x0 = [100 100 10000 400 -7000]
With glide ratio constraint.initial point inside boundary – 1 x0 = [20 20 3000 100 -500]
Without glide ratio constraint.initial point inside boundary - 1 x0 = [20 20 3000 100 -500]
With glide ratio constraint.initial point inside boundary -2 X0 = [15 11 2500 50 -10];
With glide ratio constraint.initial point inside boundary -2 X0 = [15 11 2500 50 -10];
Figure A.7 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, With Glide Ratio, Initial Point Outside Boundary)
0 2 4 6 8 10 12 14 16 18 20
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 2 4 6 8 10 12 14 16 18 2022.9
23
23.1
23.2
23.3Aspect Ratio - x2
x2
0 2 4 6 8 10 12 14 16 18 202220
2240
2260
2280Weight - x3 (N)
x3 (
N)
0 2 4 6 8 10 12 14 16 18 2030
40
50
60Turn Radius - x4 (m)
x4 (
m)
0 2 4 6 8 10 12 14 16 18 20-700
-680
-660
-640Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 2 4 6 8 10 12 14 16 18 200.5
0.55
0.6
0.65
0.7Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITH Glide Ratio Constraint, Initial point outside boundary )
Figure A.8 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, With Glide Ratio, Initial Point Inside Boundary - 1)
0 2 4 6 8 10 12 14 16 18 20
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 2 4 6 8 10 12 14 16 18 2023.1
23.2
23.3
23.4
23.5Aspect Ratio - x2
x2
0 2 4 6 8 10 12 14 16 18 202400
2420
2440
2460Weight - x3 (N)
x3 (
N)
0 2 4 6 8 10 12 14 16 18 2030
32
34
36Turn Radius - x4 (m)
x4 (
m)
0 2 4 6 8 10 12 14 16 18 20-520
-510
-500
-490Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 2 4 6 8 10 12 14 16 18 200.52
0.53
0.54
0.55Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITH Glide Ratio Constraint, Initial point inside boundary - 1 )
Figure A.9 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, With Glide Ratio, Initial Point Inside Boundary - 2)
0 2 4 6 8 10 12 14 16 18 20
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 2 4 6 8 10 12 14 16 18 2023.12
23.14
23.16
23.18
23.2Aspect Ratio - x2
x2
0 2 4 6 8 10 12 14 16 18 202260
2280
2300
2320Weight - x3 (N)
x3 (
N)
0 2 4 6 8 10 12 14 16 18 2030
30.05
30.1
30.15Turn Radius - x4 (m)
x4 (
m)
0 2 4 6 8 10 12 14 16 18 20-500
-490
-480
-470Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 2 4 6 8 10 12 14 16 18 200.514
0.516
0.518
0.52
0.522Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITH Glide Ratio Constraint, Initial point inside boundary - 2 )
Figure A.10 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, Without Glide Ratio, Initial Point Outside Boundary)
0 5 10 15 20 25 30 35 40 45 50
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 25 30 35 40 45 5022.5
23
23.5
24
24.5Aspect Ratio - x2
x2
0 5 10 15 20 25 30 35 40 45 502800
2900
3000
3100
3200Weight - x3 (N)
x3 (
N)
0 5 10 15 20 25 30 35 40 45 5030
35
40
45Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25 30 35 40 45 50-800
-600
-400
-200Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 5 10 15 20 25 30 35 40 45 50
0.58
0.6
0.62Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITHOUT Glide Ratio Constraint, Initial point outside boundary )
Figure A.11 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, Without Glide Ratio, Initial Point Inside Boundary - 1)
0 5 10 15 20 25 30 35 40 45 50
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 25 30 35 40 45 5015
20
25Aspect Ratio - x2
x2
0 5 10 15 20 25 30 35 40 45 502000
2500
3000
3500Weight - x3 (N)
x3 (
N)
0 5 10 15 20 25 30 35 40 45 5030
35
40
45Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25 30 35 40 45 50-900
-800
-700
-600
-500Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 5 10 15 20 25 30 35 40 45 500.5
0.55
0.6
0.65
0.7Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITHOUT Glide Ratio Constraint, Initial point inside boundary - 1 )
Figure A.12 Plot of Parameter Changes in Every Iteration (Genetic Algorithm, Without Glide Ratio, Initial Point Inside Boundary - 2)
15 20 25 30 35 40 45 50
26
28
30
32Wing Area - x1 (m2)
x1 (
m2)
0 5 10 15 20 25 30 35 40 45 5021
22
23
24
25Aspect Ratio - x2
x2
0 5 10 15 20 25 30 35 40 45 502200
2400
2600
2800Weight - x3 (N)
x3 (
N)
0 5 10 15 20 25 30 35 40 45 5035
40
45
50Turn Radius - x4 (m)
x4 (
m)
0 5 10 15 20 25 30 35 40 45 50-600
-400
-200
0Delta [W-L]- x5 (N)
Number of generation
x5 (
N)
0 5 10 15 20 25 30 35 40 45 500.55
0.6
0.65Rate of Descent - f(x) (m/s)
Number of generation
f(x)
(m/s
)
Genetic Algorithm Result (WITHOUT Glide Ratio Constraint, Initial point inside boundary - 2 )
TABLE A.3 Result of Simulated Annealing
Without glide ratio constraint.initial point outside boundary x0 = [100 100 10000 400 -7000]
Without glide ratio constraint.initial point inside boundary - 1
x0 = [20 20 3000 100 -500]
With glide ratio constraint.initial point inside boundary -2
X0 = [15 11 2500 50 -10];
Figure A.13 Plot of Parameter Changes in Every Iteration (Simulated Annealing, Without Glide Ratio, Initial Point Outside Boundary)
0 1000 2000 3000 4000 5000 6000 7000
10
20
30
Wing Area - x1 (m2)
x1 (
m2)
0 1000 2000 3000 4000 5000 6000 700010
15
20
25Aspect Ratio - x2
x2
0 1000 2000 3000 4000 5000 6000 700034
36
38
40Weight - x3 (N)
x3 (
N)
0 1000 2000 3000 4000 5000 6000 70000
50
100
150Turn Radius - x4 (m)
x4 (
m)
0 1000 2000 3000 4000 5000 6000 7000-10
-9
-8
-7
-6Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 1000 2000 3000 4000 5000 6000 7000
0.8
1
1.2
1.4Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Simulated Annealing Result (WITHOUT Glide Ratio Constraint, Initial point outside boundary )
Figure A.14 Plot of Parameter Changes in Every Iteration (Simulated Annealing, Without Glide Ratio, Initial Point Inside Boundary - 1)
0 1000 2000 3000 4000 5000 6000
10
20
30
Wing Area - x1 (m2)
x1 (
m2)
0 1000 2000 3000 4000 5000 600010
15
20
25Aspect Ratio - x2
x2
0 1000 2000 3000 4000 5000 600024
26
28
30
32Weight - x3 (N)
x3 (
N)
0 1000 2000 3000 4000 5000 600020
40
60
80
100Turn Radius - x4 (m)
x4 (
m)
0 1000 2000 3000 4000 5000 6000-7
-6
-5
-4
-3Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 1000 2000 3000 4000 5000 60000.4
0.6
0.8
1Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Simulated Annealing Result (WITHOUT Glide Ratio Constraint, Initial point inside boundary - 1 )
Figure A.15 Plot of Parameter Changes in Every Iteration (Simulated Annealing, Without Glide Ratio, Initial Point Inside Boundary – 2)
0 500 1000 1500 2000 2500 3000
10
20
30
Wing Area - x1 (m2)x1 (
m2)
0 500 1000 1500 2000 2500 300010
15
20
25Aspect Ratio - x2
x2
0 500 1000 1500 2000 2500 300024.5
25
25.5Weight - x3 (N)
x3 (
N)
0 500 1000 1500 2000 2500 300030
40
50
60Turn Radius - x4 (m)
x4 (
m)
0 500 1000 1500 2000 2500 3000-1
-0.8
-0.6
-0.4
-0.2Delta [W-L]- x5 (N)
Number of iteration
x5 (
N)
0 500 1000 1500 2000 2500 30000.5
0.6
0.7
0.8
0.9Rate of Descent - f(x) (m/s)
Number of iteration
f(x)
(m/s
)
Simulated Annealing Result (WITHOUT Glide Ratio Constraint, Initial point inside boundary - 2 )