Reducing run time for vpasolve
Show older comments
Hello, I am a very poor coder and need to create a program that solves for the concentration of 3 different chemical species in a set of CSTR reactors either in series or parallel. The concentration depends on the residence time of the reactor. To do this I am using vpasolve to solve my set of kinetic equations, however my program takes too long to run to completetion (not sure on the exact time but I have let it run for 30+ minutes and it has yet to produce a graph. How can I reduce the time it takes to solve the set of equations? Attached is my MATLAB file and below is the text of my code:
function main
CSTRseries
CSTRparallel
end
function CSTRseries
%variables
global k1 k2 k3 k4 Cao Cro Cso
k1=.002777778; %Rate constant (mol/(L*s))
k2=.0002777778; %Rate constant (mol/(L*s))
k3=.0002777778; %Rate constant (mol/(L*s))
k4=.0002777778; %Rate constant (mol/(L*s))
Cao=5; %Initial concentration of A (mol/L)
Cro=0; %Initial concentration of R (mol/L)
Cso=0; %Initial concentration of S (mol/L)
%Solving using the numerical ode solver (numerical solution)
syms Ca Cr Cs
tau=zeros(1,3600);
C=zeros(3,3600);
for i=1:3600
tau(i)=i;
[x,y,z] = vpasolve([(Ca-Cao)==(-k1*Ca + k2*Cr)*tau(i),(Cr-Cro)==(k1*Ca + -k2*Cr + -k3*Cr + k4*Cs)*tau(i),(Cs-Cso)==(k3*Cr + -k4*Cs)*tau(i)],[Ca,Cr,Cs],[0.0000001 Inf;0.00000001 Inf;0.0000001 Inf]);
C(1,i)=x;
C(2,i)=y;
C(3,i)=z;
end
%Plot data
figure(1)
plot(tau,C)
title('CSTRs in Series')
xlabel('tau')
ylabel('Concentration')
legend('A','R','S')
end
function CSTRparallel
%variables
global k1 k2 k3 k4 Cao Cro Cso
k1=.000833333; %Rate constant (mol/(L*s))
k2=.000833333; %Rate constant (mol/(L*s))
k3=.0002777778; %Rate constant (mol/(L*s))
k4=.0002777778; %Rate constant (mol/(L*s))
Cao=5; %Initial concentration of A (mol/L)
Cro=0; %Initial concentration of R (mol/L)
Cso=0; %Initial concentration of S (mol/L)
%Solving using the numerical ode solver (numerical solution)
syms Ca Cr Cs
tau=zeros(1,3600);
C=zeros(3,3600);
for i=1:3600
[x,y,z] = vpasolve([(Ca-Cao)==(-k1*Ca + k2*Cr + -k3*Ca + k4*Cs)*tau(i),(Cr-Cro)==(k1*Ca + -k2*Cr)*tau(i),(Cs-Cso)==(k3*Ca + -k4*Cs)*tau(i)],[Ca,Cr,Cs],[0.0000001 5;0.00000001 5;0.0000001 5]);
C(1,i)=x;
C(2,i)=y;
C(3,i)=z;
end
%Plot data
figure(2)
plot(tau,C)
title('CSTRs in Parallel')
xlabel('tau')
ylabel('Concentration')
legend('A','R','S')
end
4 Comments
Ameer Hamza
on 2 Apr 2020
Edited: Ameer Hamza
on 2 Apr 2020
Is there a specific reason for not using a numeric solver, like fsolve?
Samuel Dixon
on 2 Apr 2020
Ameer Hamza
on 2 Apr 2020
Are you solving exactly same equation 3600 times?
[x,y,z] = vpasolve([(Ca-Cao)==(-k1*Ca + k2*Cr + -k3*Ca + k4*Cs)*tau(i),(Cr-Cro)==(k1*Ca + -k2*Cr)*tau(i),(Cs-Cso)==(k3*Ca + -k4*Cs)*tau(i)],[Ca,Cr,Cs],[0.0000001 5;0.00000001 5;0.0000001 5]);
tau(i) are all zeros so nothing change in each iteration.
Samuel Dixon
on 2 Apr 2020
Answers (1)
Ameer Hamza
on 2 Apr 2020
You can get some speed improvement by using the following method. I pre-solved the equation in terms of tau variable, and inside the for loop, I substituted the value of tau. I changed one of the functions and ran for tau_=1:100 (100 elements). I observed a speed improvement of about 2x. You can similarly change other function following this method.
function CSTRseries
%variables
global k1 k2 k3 k4 Cao Cro Cso
k1=.002777778; %Rate constant (mol/(L*s))
k2=.0002777778; %Rate constant (mol/(L*s))
k3=.0002777778; %Rate constant (mol/(L*s))
k4=.0002777778; %Rate constant (mol/(L*s))
Cao=5; %Initial concentration of A (mol/L)
Cro=0; %Initial concentration of R (mol/L)
Cso=0; %Initial concentration of S (mol/L)
%Solving using the numerical ode solver (numerical solution)
syms Ca Cr Cs tau
sol = solve([(Ca-Cao)==(-k1*Ca + k2*Cr)*tau, (Cr-Cro)==(k1*Ca + -k2*Cr + -k3*Cr + k4*Cs)*tau,(Cs-Cso)==(k3*Cr + -k4*Cs)*tau], [Ca,Cr,Cs]);
tau_=1:100;
C=zeros(3,100);
for i=1:100
C(1,i) = subs(sol.Ca, tau, tau_(i));
C(2,i) = subs(sol.Cr, tau, tau_(i));
C(3,i) = subs(sol.Cs, tau, tau_(i));
end
%Plot data
figure(1)
plot(tau_,C)
title('CSTRs in Series')
xlabel('tau')
ylabel('Concentration')
legend('A','R','S')
end
6 Comments
Samuel Dixon
on 2 Apr 2020
Ameer Hamza
on 2 Apr 2020
Yes, in my code, I changed the numeric array name from tau to tau_
Samuel Dixon
on 2 Apr 2020
Ameer Hamza
on 3 Apr 2020
Can you paste you new code here? I guess there was some mistake in naming the variables.
Samuel Dixon
on 3 Apr 2020
Ameer Hamza
on 3 Apr 2020
Glad to be of help.
Categories
Find more on Chemistry 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!