Solving a set of complex trig equation

45 views (last 30 days)
Irtiza Masrur Khan
Irtiza Masrur Khan on 25 Nov 2024 at 20:15
Edited: Walter Roberson on 27 Nov 2024 at 20:09
Hello,
I have a very complex set of trig equations:
eqns1 = (-rake + (R_values(i)*(d/2))*deg2rad(skew)*tan(p) + (0.5 * chord_len(i) - x_c(1)) * sin(p) + y_u_L(1) * cos(p)) == xp(i*200);
eqns2 = R_values(i)*(d/2) * sind(skew -(180 * ((0.5 * chord_len(i) - x_c(1)) * cos(p) - y_u_L(1) * sin(p))) /(pi * R_values(i)*(d/2))) == yp(i*200);
eqns3 = R_values(i)*(d/2) * cosd(skew -(180 * ((0.5 * chord_len(i) - x_c(1)) * cos(p) - y_u_L(1) * sin(p))) /(pi * R_values(i)*(d/2))) == zp(i*200);
Attaching a picture of the equations for convineince of viewing.
Here I have named iG => rake, θs => skew, and θnt => p in my MATLAB script, The unknowns are: rake, p, and skew. All the other values are known.
I have tried using solve, vpasolve, and fsolve. solve and vpasolve generates an emptry structure for S.
Code of the loop that solves the equations:
for i = 1:9
% Calculate airfoil properties
chord_len = chord_length_at_R * d;
max_c = max_camber_at_R .* chord_len;
max_th = max_thickness_at_R * d;
cam = -cam_d * max_c(i);
th = th_d * max_th(i);
x_c = X_c * chord_len(i);
theta1 = atan2(der_y, 1);
y_u_L = cam + th .* cos(theta1);
% Define equations
eqns1 = (-rake + (R_values(i)*(d/2))*deg2rad(skew)*tan(p) + ...
(0.5 * chord_len(i) - x_c(1)) * sin(p) + y_u_L(1) * cos(p)) == xp(i*200);
eqns2 = R_values(i)*(d/2) * sind(skew - ...
(180 * ((0.5 * chord_len(i) - x_c(1)) * cos(p) - y_u_L(1) * sin(p))) / ...
(pi * R_values(i)*(d/2))) == yp(i*200);
eqns3 = R_values(i)*(d/2) * cosd(skew - ...
(180 * ((0.5 * chord_len(i) - x_c(1)) * cos(p) - y_u_L(1) * sin(p))) / ...
(pi * R_values(i)*(d/2))) == zp(i*200);
% Solve equations
try
S = vpasolve(simplify([eqns1, eqns2, eqns3]), [p, skew, rake], [pitch(i),0, skew_angle_at_R(i)]); % With initial guess
if isempty(S)
disp(['No solution for section ', num2str(i)]);
else
rake_sol = double(S.rake);
p_sol = double(S.p);
skew_sol = double(S.skew);
fprintf('Section %d: Rake = %.4f, P = %.4f, Skew = %.4f\n', i, rake_sol, p_sol, skew_sol);
end
catch ME
disp(['Error solving section ', num2str(i)]);
disp(ME.message);
end
end
And with fsolve I can only converge the first two sections.
Code of the loop that solves the equations:
for i = 1:n
% Scale airfoil data for current section
chord_len = chord_length_at_R * d;
max_c = max_camber_at_R .* chord_len;
max_th = max_thickness_at_R * d;
cam = cam_d * max_c(i);
th = th_d * max_th(i);
x_c = X_c * chord_len(i);
theta1 = atan2(der_y, 1);
y_u_L = cam + th .* cos(theta1);
% Define equations for fsolve
fun = @(vars) [
(-vars(2) + (R_values(i) * (d / 2)) * vars(3) * tan(vars(1)) + (0.5 * chord_len(i) - x_c(1)) * sin(vars(1)) + y_u_L(1) * cos(vars(1))) - xp(i * 200);
R_values(i) * (d / 2) * sind(vars(3) - (180 * ((0.5 * chord_len(i) - x_c(1)) * cos(vars(1)) - y_u_L(1) * sin(vars(1)))) / (pi * R_values(i) * (d / 2))) - yp(i * 200);
R_values(i) * (d / 2) * cosd(vars(3) - (180 * ((0.5 * chord_len(i) - x_c(1)) * cos(vars(1)) - y_u_L(1) * sin(vars(1)))) / (pi * R_values(i) * (d / 2))) - zp(i * 200)
];
% Use pitch, skew_angle_at_R, and rake as initial guesses
initial_guess = [pitch(i),0, skew_angle_at_R(i)];
% Solve using fsolve
options = optimoptions('fsolve', 'Display', 'none', 'MaxIterations', 10000, 'MaxFunctionEvaluations', 20000);
[solution, fval, exitflag] = fsolve(fun, initial_guess, options);
% Handle non-convergence
if exitflag <= 0
fprintf('fsolve did not converge for section %d. Trying alternative guesses...\n', i);
alt_guesses = [rand(1, 3); [pitch(i),0, skew_angle_at_R(i)]]; % Use random or skew-related guesses
for j = 1:size(alt_guesses, 1)
[solution, fval, exitflag] = fsolve(fun, alt_guesses(j, :), options);
if exitflag > 0
break;
end
end
end
% Store solutions
if exitflag > 0
solutions(i, :) = solution;
else
fprintf('Failed to converge for section %d even with alternative guesses.\n', i);
end
end
I would really appreacite any help on this as I've been stuck on this for some time. I know that these equations are highly non-linaer which is why MATLAB is probably having a hard time solving this, but I am not sure how else to approach this problem.
  2 Comments
Torsten
Torsten on 25 Nov 2024 at 20:23
Please supply executable code (with all parameters defined).
Irtiza Masrur Khan
Irtiza Masrur Khan on 25 Nov 2024 at 20:35
Edited: Torsten on 25 Nov 2024 at 21:41
Attached are the executable code and the excels.
clear;
clc;
% Load airfoil and coordinate data
airfoil_path = 'airfoil_data_fixed.xlsx';
coords_path = 'ORCA_Pointwise.xlsx';
% Read the airfoil data
Airfoil = xlsread(airfoil_path);
Coords = xlsread(coords_path);
X_c = Airfoil(:, 1); % Chord coordinates (normalized)
y_c = Airfoil(:, 2); % Camber distribution
der_y = Airfoil(:, 3); % Slope of camber
th_d = Airfoil(:, 4); % Thickness distribution
yp = Coords(:, 1);
zp = Coords(:, 2);
xp = Coords(:, 3);
% Normalize camber distribution
cam_d = y_c / max(y_c);
% Define polynomial functions
MaxCamber = @(x) -4448.8369*x.^12 + 30393.6831*x.^11 - 92977.6043*x.^10 + 168066.8833*x.^9 - 199490.6759*x.^8 + 163416.9800*x.^7 - 94493.8152*x.^6 + 38765.7447*x.^5 - 11176.4246*x.^4 + 2207.9559*x.^3 - 285.0846*x.^2 + 21.9443*x - 0.7490;
ChordLength = @(x) -143202.4761*x.^12 + 978274.9902*x.^11 - 2992184.0323*x.^10 + 5408923.9625*x.^9 - 6424276.8851*x.^8 + 5271614.6993*x.^7 - 3058632.5267*x.^6 + 1261908.7720*x.^5 - 366745.1423*x.^4 + 73096.4455*x.^3 - 9470.1908*x.^2 + 716.1619*x - 23.7157;
MaxThickness = @(x) -630.6311*x.^12 + 2601.5312*x.^11 - 2668.3834*x.^10 - 4627.0740*x.^9 + 16182.8469*x.^8 - 21144.9359*x.^7 + 15960.6510*x.^6 - 7582.2691*x.^5 + 2294.7299*x.^4 - 431.4940*x.^3 + 47.9318*x.^2 - 3.0647*x + 0.1614;
Pitch = @(x) 19344.5071*x.^12 - 114044.8587*x.^11 + 280789.2801*x.^10 - 357377.6146*x.^9 + 207947.2705*x.^8 + 43330.8173*x.^7 - 173099.4797*x.^6 + 143116.5772*x.^5 - 65570.9523*x.^4 + 18410.2055*x.^3 - 3128.7530*x.^2 + 294.0671*x - 10.4419;
SkewAngle = @(x) -719361.0309*x.^12 + 5176951.0162*x.^11 - 16746858.1600*x.^10 + 32161071.3897*x.^9 - 40784306.1013*x.^8 + 35930276.7571*x.^7 - 22515881.4272*x.^6 + 10096627.1844*x.^5 - 3209956.6167*x.^4 + 704272.6595*x.^3 - 100947.8092*x.^2 + 8462.8187*x - 312.8100;
% Scaling factor for the geometry
d = 1.4;
% Define the R values where we want to evaluate the polynomials
R_values = [0.1834, 0.2408, 0.2982, 0.3556, 0.413, 0.4704, 0.5278, 0.5852, 0.6426] / 0.7;
% Evaluate the polynomials at R_values
max_camber_at_R = MaxCamber(R_values);
pitch_dia_at_R = Pitch(R_values) * d * 1.4;
chord_length_at_R = ChordLength(R_values) * 0.99;
max_thickness_at_R = MaxThickness(R_values);
skew_angle_at_R = SkewAngle(R_values);
pitch = atan2(pitch_dia_at_R, (2 * pi * R_values)); % Angle in radians
chord_len = chord_length_at_R * d;
max_c = max_camber_at_R .* chord_len;
max_th = max_thickness_at_R * d;
% Number of sections (airfoils)
n = length(R_values);
% Solve for each section
solutions = zeros(n, 3);
for i = 1:n
% Scale airfoil data for current section
chord_len = chord_length_at_R * d;
max_c = max_camber_at_R .* chord_len;
max_th = max_thickness_at_R * d;
cam = cam_d * max_c(i);
th = th_d * max_th(i);
x_c = X_c * chord_len(i);
theta1 = atan2(der_y, 1);
y_u_L = cam + th .* cos(theta1);
R_values(i)^2*d^2/4-yp(i*200)^2-zp(i*200)^2
% Define equations for fsolve
fun = @(vars) [
(-vars(2) + (R_values(i) * (d / 2)) * vars(3) * tan(vars(1)) + (0.5 * chord_len(i) - x_c(1)) * sin(vars(1)) + y_u_L(1) * cos(vars(1))) - xp(i * 200);
R_values(i) * (d / 2) * sind(vars(3) - (180 * ((0.5 * chord_len(i) - x_c(1)) * cos(vars(1)) - y_u_L(1) * sin(vars(1)))) / (pi * R_values(i) * (d / 2))) - yp(i * 200);
R_values(i) * (d / 2) * cosd(vars(3) - (180 * ((0.5 * chord_len(i) - x_c(1)) * cos(vars(1)) - y_u_L(1) * sin(vars(1)))) / (pi * R_values(i) * (d / 2))) - zp(i * 200)
];
% Use pitch, skew_angle_at_R, and rake as initial guesses
initial_guess = [pitch(i), skew_angle_at_R(i), 0];
% Solve using fsolve
options = optimoptions('fsolve', 'Display', 'none', 'MaxIterations', 10000, 'MaxFunctionEvaluations', 20000);
[solution, fval, exitflag] = fsolve(fun, initial_guess, options);
% Handle non-convergence
if exitflag <= 0
fprintf('fsolve did not converge for section %d. Trying alternative guesses...\n', i);
alt_guesses = [rand(1, 3); [pitch(i), skew_angle_at_R(i), 0]]; % Use random or skew-related guesses
for j = 1:size(alt_guesses, 1)
[solution, fval, exitflag] = fsolve(fun, alt_guesses(j, :), options);
if exitflag > 0
break;
end
end
end
% Store solutions
if exitflag > 0
solutions(i, :) = solution;
else
fprintf('Failed to converge for section %d even with alternative guesses.\n', i);
end
end
ans = 8.1922e-10
ans = 4.1339e-11
ans = 0.0309
fsolve did not converge for section 3. Trying alternative guesses...
Failed to converge for section 3 even with alternative guesses.
ans = 0.0375
fsolve did not converge for section 4. Trying alternative guesses...
Failed to converge for section 4 even with alternative guesses.
ans = 0.0816
fsolve did not converge for section 5. Trying alternative guesses...
Failed to converge for section 5 even with alternative guesses.
ans = 0.0948
fsolve did not converge for section 6. Trying alternative guesses...
Failed to converge for section 6 even with alternative guesses.
ans = 0.1521
fsolve did not converge for section 7. Trying alternative guesses...
Failed to converge for section 7 even with alternative guesses.
ans = 0.1719
fsolve did not converge for section 8. Trying alternative guesses...
Failed to converge for section 8 even with alternative guesses.
ans = 0.2424
fsolve did not converge for section 9. Trying alternative guesses...
Failed to converge for section 9 even with alternative guesses.
disp(solutions);
2.1293 0.2920 0.2256 0.8014 0.1956 0.8773 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Sign in to comment.

Accepted Answer

Torsten
Torsten on 25 Nov 2024 at 20:46
Edited: Torsten on 25 Nov 2024 at 21:13
Do you see the problem ? Your R-values and your y- and z-values are not compatible (at least for n = 3).
It must hold that
R_values(i)^2*d^2/4-yp(i*200)^2-zp(i*200)^2 = 0
for all i.
  1 Comment
Irtiza Masrur Khan
Irtiza Masrur Khan on 27 Nov 2024 at 15:22
You are right. My previous code was reading the xp,yp,zp values wrong from the excel files.

Sign in to comment.

More Answers (0)

Categories

Find more on Airfoil tools in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!