Euler Method Simulation - Enhanced Algorithm

June 14, 2022, 5:26 p.m.
cthiels

June 17 2022

For the first part of this internship we compared the difference between Euler's method of integration with Euler-Cromer integration. It was observed that when using Euler's method in a Simple Harmonic Oscillator (SHO) the method would tend to increase the amount of energy in the system which is incorrect. 

 The energy issue is solved by using the Euler-Cromer method. 

As shown on the figures, The Euler-Cromer method holds together much better and energy is conserved.

The code used is as follows.

close all
clear all
clc

% This script compares Euler's method of numerical integration to the analytical method.
% Euler's method
g = 9.81; %acceleration of gravity [m/s^2].
m = 1; %mass of the object [kg].
k = 1; %spring constant [N/m].
y0 = 5; %initial position in relation to equilibrium [m].
v0 = 0; %initial velocity [m/s].
dt = 0.01; %time interval between steps [s].
n = 10000; %number of steps.
t(1) = 0; %initial time condition.
y(1) = y0; %initial position condition.
v(1) = v0; %initial velocity condition.
pe(1) = .5 * k * y0^2;
ke(1) = .5 * m * v0^2;
e(1) = pe(1) + ke(1);
for i = 2:n
  t(i) = t(i-1) + dt; % adds time interval to previous time and saves the value in a vector.
  v(i) = v(i-1) - (k./m) .* y(i-1) .* dt; %Calculates new velocity and saves the value in a vector.
  y(i) = y(i-1) + v(i-1) * dt; %Calculates new position and saves the value in a vector.
  pe(i) = .5 * k * y(i)^2;
  ke(i) = .5 * m * v(i)^2;
  e(i) = pe(i) + ke(i);
end
figure 1
plot(t,v,'r',t,y,'b')
title("Euler's method of numerical integration")
legend('Velocity [m/s]','position [m]')
%----------------------------------------%
% Analytical method
v_exact = -y0 * sqrt(k ./ m) .* sin(sqrt(k ./ m) .* t);
y_exact = y0 * cos(sqrt(k ./ m) .* t);
pe_exact = .5 .* k .* y_exact .^ 2;
ke_exact = .5 .* m .* v_exact .^ 2;
e_exact = pe_exact + ke_exact;
figure 2 
plot(t,v_exact,'r',t,y_exact,'b',t,v,'r--',t,y,'b--',t,e,'g',t,e_exact,'y')
title('Euler vs exact')
legend('v exact','y exact','v Euler','y Euler','Energy Euler','Exact Energy')

close all
clear all
clc

% This script compares the Euler-cromer method of numerical integration to the analytical method.
% Euler-cromer method
g = 9.81; %acceleration of gravity [m/s^2].
m = 1; %mass of the object [kg].
k = 1; %spring constant [N/m].
y0 = 5; %initial position in relation to equilibrium [m].
v0 = 0; %initial velocity [m/s].
dt = 0.01; %time interval between steps [s].
n = 10000; %number of steps.
t(1) = 0; %initial time condition.
y(1) = y0; %initial position condition.
v(1) = v0; %initial velocity condition.
pe(1) = .5 * k * y0^2; %initial spring potential energy.
ke(1) = .5 * m * v0^2; %initial object kinetic energy.
e(1) = pe(1) + ke(1); %initial total energy.
F(1) = -k * y0; %initial force on the object.

for i = 2:n
  t(i) = t(i-1) + dt; % adds time interval to previous time and saves the value in a vector.
  F(i) = -k .* y(i-1); % calculates the force on the object.
  a(i) = F(i) / m; % calculates the accelerationon of the object.
  v(i) = v(i-1) + a(i) * dt; % calculatesthe velocity of the object.
  y(i) = y(i-1) + v(i) * dt; %calculates the position of the object.
  pe(i) = .5 * k * y(i)^2; % calculates object potential energy.
  ke(i) = .5 * m * v(i)^2; % calculates object kinetic energy.
  e(i) = pe(i) + ke(i); %calculates object total energy.
end

%----------------------------------------%
% Analytical method
v_exact = -y0 * sqrt(k ./ m) .* sin(sqrt(k ./ m) .* t);
y_exact = y0 * cos(sqrt(k ./ m) .* t);
pe_exact = .5 .* k .* y_exact .^ 2;
ke_exact = .5 .* m .* v_exact .^ 2;
e_exact = pe_exact + ke_exact;

plot(t,y_exact,'b',t,y,'r--',t,e,'g',t,e_exact,'y')
title('Euler vs exact')
legend('y exact','y Euler','Energy Euler','Exact Energy')


June 24 2022

During this week we analyzed the Euler-Cromer method of numeric integration in a rigid pendulum. We also added resistance and oscillatory forces into our system.

As you can see the graphs are a lot more chaotic.

The code used is as follows:

clc
clear all
close all

g = 9.81;
L = 2;
m = 5;
b = 3;
I = (m * L^2) / 3;
Tau = 10;
w = 2.5;

omega(1) = 0;
theta(1) = 120 * pi / 180;
alpha(1) =  ((3 * g) / (2 * L))^2 * sin(theta(1));
nsteps = 10000;
dt = .001;
t(1) = 0;
for n = 2:nsteps 
  t(n) = t(n-1) + dt;
  alpha(n) = -((3 * g) / (2 * L))^2 * sin(theta(n-1)) - (b*omega(n-1))/I + Tau * cos(w*t(n));
  omega(n) = omega(n-1) + alpha(n) * dt;
  theta(n) = theta(n-1) + omega(n) * dt;
end

plot(t,theta)
title('Euler-Cromer')
figure 2
plot(theta, omega)
title('Euler-Cromer')


July 1 2022

This week we studied the relationship between a cubic oscillator’s period and its amplitude. We calculated that the period of the oscillator is proportional to the inverse of the amplitude. 

 

We also worked with animation and modeling techniques in Matlab. 

This code calculates and animates a rigid pendulum.

clc

clear all

close all

%%% This code models and animates the motion of a rigid pendulum.

g = 9.81; %gravity

L = 2; % length

m = 5; % mass

b = 0; % resistance

I = (m * L^2) / 3; %moment of inirtia

Tau = 0; % added force

w = 2.5; % added force interval

omega(1) = 0; %initialize omega (initial velocity)

theta(1) = 30 * pi / 180; % initialize theta (initial angle)

alpha(1) = ((3 * g) / (2 * L))^2 * sin(theta(1)); %initialize alpha (initial acceleration)

nsteps = 1000; % sets number of steps

dt = .01; % sets time step

t(1) = 0; %initialize time

% Euler-Cromer method

for n = 2:nsteps

t(n) = t(n-1) + dt;

alpha(n) = -((3 * g) / (2 * L))^2 * sin(theta(n-1)) - (b*omega(n-1))/I + Tau * cos(w*t(n));

omega(n) = omega(n-1) + alpha(n) * dt;

theta(n) = theta(n-1) + omega(n) * dt;

end

plot(t,theta)

title('Euler-Cromer')

figure

plot(theta, omega)

title('Euler-Cromer')

X = -.5 * L .* sin(theta); % locates center of mass in the x direction

Y = -.5 * L .* cos(theta); % locates center of mass in the y direction

figure

% animation for the pendulum

for z = 1:length(theta)

plot ([0,2*X(z)],[0,2*Y(z)],'linewidth',2)

axis([-2,2,-2,0])

hold on

plot(X(z),Y(z),'ro')

pause(.01)

%storing frames

K(z) = getframe(gcf);

hold off

end

videofile = VideoWriter('pendulum.avi','Uncompressed AVI')

open(videofile)

writeVideo(videofile,K)

close(videofile)

I am also working on a simulation of the ISS orbiting Earth, but I have run into an issue when calculating the angle and I have not been able to figure it out yet. Here is my code so far. This is still very rough.

clc

close all

clear all

mass_Earth = 5.972e24;

mass_object = 20000;

x(1) = 0;

y(1) = 6786100;

alt(1) = sqrt(x(1)^2 + y(1)^2);

vx(1) = 7778;

vy(1) = 0;

ax(1) = 0;

ay(1) = 0;

nsteps = 100;

dt = 60;

t(1) = 0;

theta(1) = 0;

for n = 2:nsteps

f(n) = -6.67428e-11 * (mass_Earth * mass_object) / alt(n-1)^2;

fx(n) = f(n) * sin(theta(n-1));

fy(n) = f(n) * cos(theta(n-1));

ax(n) = fx(n)/mass_object;

ay(n) = fy(n)/mass_object;

vx(n) = vx(n-1) + ax(n) * dt;

vy(n) = vy(n-1) + ay(n) * dt;

x(n) = x(n-1) + vx(n) * dt;

y(n) = y(n-1) + vy(n) * dt;

alt(n) = sqrt(x(n)^2 + y(n)^2);

v(n) = sqrt(vx(n)^2 + vy(n)^2);

theta(n) = theta(n-1) + (v(n)/(2 * pi * alt(n)))*dt;

plot(x(n),y(n),'rx','LineWidth',10)

axis([-7000000,7000000,-7000000,7000000])

pause(.1)

t(n) = t(n-1)+dt;

end

 

July 15 2022

This week we worked on waves as a function of displacement and time. We were able to come up with a function to predict the future state of a wave. 

future(n) = r2(current(n+1)+current(n-1))+2(1-r2)current(n)-past(n)

We applied this function to many different initial conditions including: fixed point “plucked” string, oscillating ends, and “sliding” end string.

Videos to demonstrate our solutions are located on YouTube at: https://www.youtube.com/watch?v=OmheQyhl__w https://www.youtube.com/watch?v=TBb7mGHyR8k https://youtu.be/hs6mkq1AJvE