Maximizing Batch Reactor Products in matlab
4 views (last 30 days)
Show older comments
Hi..I want to maximize W_C10 which is the mass concentraion of C10 products in a batch Reactor...And what I meant by maximization is W_C10 is going to be greater than all the other W_Ci,i=4,6,..,20..Here is my constraints for T,C_TEA and C_CAT and C_M:
338.15=<T=<368.15
1=<C_TEA<=200
1=<C_M<=100
C_CAT=x*C_TEA where x=17,18,...,45
Here is my ode system to solve W_Cis..Can someone help me to do this?
clc
clear
close all
T= 3.642755276270583e2;
T_r=273.15+0;
A1=4.525e5;
E1 =3985.;
A2 = 0.5101e9;
E2 = 52670.;
A3 = 0.4949e9;
E3 = 15140.;
A4 = 465000.;
E4 = 14330.;
A5 = 1.255;
E5 = 0;
R = 1.98;
k1 = A1*exp(E1*(1/T - 1/T_r)/R);
k2 = A2*exp(E2*(1/T - 1/T_r)/R);
k3 = A3*exp(E3*(1/T - 1/T_r)/R);
k4 = A4*exp(E4*(1/T - 1/T_r)/R);
k5 = A5*exp(E5*(1/T - 1/T_r)/R);
KA = 481.2;
KB = 277.1;
KC = 6906;
KD = 40830;
KE = 59.55;
C_M= 46.946030544906776;
C_TEA=1.022918116483290e2;
C_CAT=45*C_TEA;
C_M_K=C_M/(KB*C_TEA+1);
C_CAT_K=C_CAT/(KA*C_M+1);
C_TEA_k=C_TEA/(KC*C_CAT+1)^2;
dCdt=@(t,x) [k1.*C_CAT_K.*C_M_K-k2.*x(1).*C_M-k5.*x(1);k2.*C_M.*x(1)-(k3.*x(2).*C_M)./(KD.*C_TEA+1)-k5.*x(2);(k3.*x(2).*C_M)./(KD.*C_TEA+1)-(k3.*x(3).*C_M)./(KD.*C_TEA+1)-k5.*x(3);(k3.*x(3).*C_M)./(KD.*C_TEA+1)-(k3.*x(4).*C_M)./(KD.*C_TEA+1)-k5.*x(4);(k3.*x(4).*C_M)./(KD.*C_TEA+1)-(k3.*x(5).*C_M)./(KD.*C_TEA+1)-k5.*x(5);(k3.*x(5).*C_M)./(KD.*C_TEA+1)-(k3.*x(6).*C_M)./(KD.*C_TEA+1)-k5.*x(6);(k3.*x(6).*C_M)./(KD.*C_TEA+1)-(k3.*x(7).*C_M)./(KD.*C_TEA+1)-k5.*x(7);(k3.*x(7).*C_M)./(KD.*C_TEA+1)-(k3.*x(8).*C_M)./(KD.*C_TEA+1)-k5.*x(8);(k3.*x(8).*C_M)./(KD.*C_TEA+1)-(k3.*x(9).*C_M)./(KD.*C_TEA+1)-k5.*x(9);(k3.*x(9).*C_M)./(KD.*C_TEA+1)-(k3.*x(10).*C_M)./(KD.*C_TEA+1)-k5.*x(10);(k3.*x(10).*C_M)./(KD.*C_TEA+1)-(k3.*x(11).*C_M)./(KD.*C_TEA+1)-k5.*x(11);(k4.*x(3).*C_M)./(KE.*C_TEA+1)+k5.*x(3);(k4.*x(4).*C_M)./(KE.*C_TEA+1)+k5.*x(4);(k4.*x(5).*C_M)./(KE.*C_TEA+1)+k5.*x(5);(k4.*x(6).*C_M)./(KE.*C_TEA+1)+k5.*x(6);(k4.*x(7).*C_M)./(KE.*C_TEA+1)+k5.*x(7);(k4.*x(8).*C_M)./(KE.*C_TEA+1)+k5.*x(8);(k4.*x(9).*C_M)./(KE.*C_TEA+1)+k5.*x(9);(k4.*x(10).*C_M)./(KE.*C_TEA+1)+k5.*x(10);(k4.*x(11).*C_M)./(KE.*C_TEA+1)+k5.*x(11);k5.*(x(2)+x(3)+x(4)+x(5)+x(6)+x(7)+x(8)+x(9)+x(10)+x(11))];
[t,x]=ode45(dCdt,[0,3600],[0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0]);
%Moles Disturbutions in dead chains
t1=3600;
M_total=x(t1,12)+x(t1,13)+x(t1,14)+x(t1,15)+x(t1,16)+x(t1,17)+x(t1,18)+x(t1,19)+x(t1,20);
M_C4=x(t1,12)/M_total;
M_C6=x(t1,13)/M_total;
M_C8=x(t1,14)/M_total;
M_C10=x(t1,15)/M_total;
M_C12=x(t1,16)/M_total;
M_C14=x(t1,17)/M_total;
M_C16=x(t1,18)/M_total;
M_C18=x(t1,19)/M_total;
M_C20=x(t1,20)/M_total;
T1=table(M_C4,M_C6,M_C8,M_C10,M_C12,M_C14,M_C16,M_C18,M_C20)
%Mass Disturbutions in dead chains Basis=100mole
W_total=100*M_C4*72+100*M_C6*84+100*M_C8*112+100*M_C10*140+100*M_C12*168+100*M_C14*196+100*M_C16*224+100*M_C18*252+100*M_C20*280;
W_C4=100*M_C4*72/W_total;
W_C6=100*M_C6*84/W_total;
W_C8=100*M_C8*112/W_total;
W_C10=100*M_C10*140/W_total;
W_C12=100*M_C12*168/W_total;
W_C14=100*M_C14*196/W_total;
W_C16=100*M_C16*224/W_total;
W_C18=100*M_C18*252/W_total;
W_C20=100*M_C20*280/W_total;
T2=table(W_C4,W_C6,W_C8,W_C10,W_C12,W_C14,W_C16,W_C18,W_C20)
a=[W_C4,W_C6,W_C8,W_C10,W_C12,W_C14,W_C16,W_C18,W_C20]
0 Comments
Accepted Answer
Torsten
on 2 Aug 2023
Edited: Torsten
on 2 Aug 2023
You get almost identical values for the optimal T, C_M and C_TEA for all x-values in between 17 and 45 as you can see after the first loop in the code.
By strengthening the tolerances, now W_C4 is also smaller than W_C10 in all cases.
"obj" sets the function to be maximized - in this case I chose W_C10.
"constr" defines the constraints under which the function in "obj" is maximized. In this case, I chose W_Ci - W_C10 <= 0 for i = 4,6,...,20.
The code can be rearranged to look smarter (e.g. the threefold repetition of the calculation of the M_Ci ang W_Ci should be put in its own function) - I leave these technical changes to you.
x = 17:45;
for i = 1:numel(x)
lb = [338.15;1;1];
ub = [368.15;100;200];
p0 = (lb+ub)/2;
options = optimoptions('fmincon','Display','off','Algorithm','interior-point','ConstraintTolerance',1e-16);
[p,fval,exitflag,~] = fmincon(@(p)obj(p,x(i)),p0,[],[],[],[],lb,ub,@(p)constr(p,x(i)),options);
T_opt(i) = p(1);
C_M_opt(i) = p(2);
C_TEA_opt(i) = p(3);
C_CAT_opt(i) = C_TEA_opt(i)*x(i);
value_opt(i) = -fval;
exit(i) = exitflag;
end
exit
T_opt
C_M_opt
C_TEA_opt
C_CAT_opt
value_opt
%Postprocess results
for i = 1:numel(x)
[t,x] = ode_solution([T_opt(i),C_M_opt(i),C_TEA_opt(i)],x(i));
M_total=x(end,12)+x(end,13)+x(end,14)+x(end,15)+x(end,16)+x(end,17)+x(end,18)+x(end,19)+x(end,20);
M_C4=x(end,12)/M_total;
M_C6=x(end,13)/M_total;
M_C8=x(end,14)/M_total;
M_C10=x(end,15)/M_total;
M_C12=x(end,16)/M_total;
M_C14=x(end,17)/M_total;
M_C16=x(end,18)/M_total;
M_C18=x(end,19)/M_total;
M_C20=x(end,20)/M_total;
%Mass Disturbutions in dead chains Basis=100mole
W_total=100*M_C4*72+100*M_C6*84+100*M_C8*112+100*M_C10*140+100*M_C12*168+100*M_C14*196+100*M_C16*224+100*M_C18*252+100*M_C20*280;
W_C4(i)=100*M_C4*72/W_total;
W_C6(i)=100*M_C6*84/W_total;
W_C8(i)=100*M_C8*112/W_total;
W_C10(i)=100*M_C10*140/W_total;
W_C12(i)=100*M_C12*168/W_total;
W_C14(i)=100*M_C14*196/W_total;
W_C16(i)=100*M_C16*224/W_total;
W_C18(i)=100*M_C18*252/W_total;
W_C20(i)=100*M_C20*280/W_total;
end
any(W_C4>=W_C10)
any(W_C6>=W_C10)
any(W_C8>=W_C10)
any(W_C12>=W_C10)
any(W_C14>=W_C10)
any(W_C16>=W_C10)
any(W_C18>=W_C10)
any(W_C20>=W_C10)
function value = obj(p,factor)
[t,x] = ode_solution(p,factor);
M_total=x(end,12)+x(end,13)+x(end,14)+x(end,15)+x(end,16)+x(end,17)+x(end,18)+x(end,19)+x(end,20);
M_C4=x(end,12)/M_total;
M_C6=x(end,13)/M_total;
M_C8=x(end,14)/M_total;
M_C10=x(end,15)/M_total;
M_C12=x(end,16)/M_total;
M_C14=x(end,17)/M_total;
M_C16=x(end,18)/M_total;
M_C18=x(end,19)/M_total;
M_C20=x(end,20)/M_total;
%Mass Disturbutions in dead chains Basis=100mole
W_total=100*M_C4*72+100*M_C6*84+100*M_C8*112+100*M_C10*140+100*M_C12*168+100*M_C14*196+100*M_C16*224+100*M_C18*252+100*M_C20*280;
W_C4=100*M_C4*72/W_total;
W_C6=100*M_C6*84/W_total;
W_C8=100*M_C8*112/W_total;
W_C10=100*M_C10*140/W_total;
W_C12=100*M_C12*168/W_total;
W_C14=100*M_C14*196/W_total;
W_C16=100*M_C16*224/W_total;
W_C18=100*M_C18*252/W_total;
W_C20=100*M_C20*280/W_total;
value = -W_C10;
end
function [c,ceq] = constr(p,factor)
[t,x] = ode_solution(p,factor);
M_total=x(end,12)+x(end,13)+x(end,14)+x(end,15)+x(end,16)+x(end,17)+x(end,18)+x(end,19)+x(end,20);
M_C4=x(end,12)/M_total;
M_C6=x(end,13)/M_total;
M_C8=x(end,14)/M_total;
M_C10=x(end,15)/M_total;
M_C12=x(end,16)/M_total;
M_C14=x(end,17)/M_total;
M_C16=x(end,18)/M_total;
M_C18=x(end,19)/M_total;
M_C20=x(end,20)/M_total;
%Mass Disturbutions in dead chains Basis=100mole
W_total=100*M_C4*72+100*M_C6*84+100*M_C8*112+100*M_C10*140+100*M_C12*168+100*M_C14*196+100*M_C16*224+100*M_C18*252+100*M_C20*280;
W_C4=100*M_C4*72/W_total;
W_C6=100*M_C6*84/W_total;
W_C8=100*M_C8*112/W_total;
W_C10=100*M_C10*140/W_total;
W_C12=100*M_C12*168/W_total;
W_C14=100*M_C14*196/W_total;
W_C16=100*M_C16*224/W_total;
W_C18=100*M_C18*252/W_total;
W_C20=100*M_C20*280/W_total;
c(1) = W_C4 - W_C10;
c(2) = W_C6 - W_C10;
c(3) = W_C8 - W_C10;
c(4) = W_C12 - W_C10;
c(5) = W_C14 - W_C10;
c(6) = W_C16 - W_C10;
c(7) = W_C18 - W_C10;
c(8) = W_C20 - W_C10;
ceq = [];
end
function [t,x] = ode_solution(p,factor)
T=p(1);
C_M=p(2);
C_TEA=p(3);
C_CAT = factor*C_TEA;
T_r=273.15+0;
A1=4.525e5;
E1 =3985.;
A2 = 0.5101e9;
E2 = 52670.;
A3 = 0.4949e9;
E3 = 15140.;
A4 = 465000.;
E4 = 14330.;
A5 = 1.255;
E5 = 0;
R = 1.98;
k1 = A1*exp(E1*(1/T - 1/T_r)/R);
k2 = A2*exp(E2*(1/T - 1/T_r)/R);
k3 = A3*exp(E3*(1/T - 1/T_r)/R);
k4 = A4*exp(E4*(1/T - 1/T_r)/R);
k5 = A5*exp(E5*(1/T - 1/T_r)/R);
KA = 481.2;
KB = 277.1;
KC = 6906;
KD = 40830;
KE = 59.55;
C_M_K=C_M/(KB*C_TEA+1);
C_CAT_K=C_CAT/(KA*C_M+1);
C_TEA_k=C_TEA/(KC*C_CAT+1)^2;
dCdt=@(t,x) [k1.*C_CAT_K.*C_M_K-k2.*x(1).*C_M-k5.*x(1);k2.*C_M.*x(1)-(k3.*x(2).*C_M)./(KD.*C_TEA+1)-k5.*x(2);(k3.*x(2).*C_M)./(KD.*C_TEA+1)-(k3.*x(3).*C_M)./(KD.*C_TEA+1)-k5.*x(3);(k3.*x(3).*C_M)./(KD.*C_TEA+1)-(k3.*x(4).*C_M)./(KD.*C_TEA+1)-k5.*x(4);(k3.*x(4).*C_M)./(KD.*C_TEA+1)-(k3.*x(5).*C_M)./(KD.*C_TEA+1)-k5.*x(5);(k3.*x(5).*C_M)./(KD.*C_TEA+1)-(k3.*x(6).*C_M)./(KD.*C_TEA+1)-k5.*x(6);(k3.*x(6).*C_M)./(KD.*C_TEA+1)-(k3.*x(7).*C_M)./(KD.*C_TEA+1)-k5.*x(7);(k3.*x(7).*C_M)./(KD.*C_TEA+1)-(k3.*x(8).*C_M)./(KD.*C_TEA+1)-k5.*x(8);(k3.*x(8).*C_M)./(KD.*C_TEA+1)-(k3.*x(9).*C_M)./(KD.*C_TEA+1)-k5.*x(9);(k3.*x(9).*C_M)./(KD.*C_TEA+1)-(k3.*x(10).*C_M)./(KD.*C_TEA+1)-k5.*x(10);(k3.*x(10).*C_M)./(KD.*C_TEA+1)-(k3.*x(11).*C_M)./(KD.*C_TEA+1)-k5.*x(11);(k4.*x(3).*C_M)./(KE.*C_TEA+1)+k5.*x(3);(k4.*x(4).*C_M)./(KE.*C_TEA+1)+k5.*x(4);(k4.*x(5).*C_M)./(KE.*C_TEA+1)+k5.*x(5);(k4.*x(6).*C_M)./(KE.*C_TEA+1)+k5.*x(6);(k4.*x(7).*C_M)./(KE.*C_TEA+1)+k5.*x(7);(k4.*x(8).*C_M)./(KE.*C_TEA+1)+k5.*x(8);(k4.*x(9).*C_M)./(KE.*C_TEA+1)+k5.*x(9);(k4.*x(10).*C_M)./(KE.*C_TEA+1)+k5.*x(10);(k4.*x(11).*C_M)./(KE.*C_TEA+1)+k5.*x(11);k5.*(x(2)+x(3)+x(4)+x(5)+x(6)+x(7)+x(8)+x(9)+x(10)+x(11))];
[t,x]=ode15s(dCdt,[0,3600],[0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0],odeset('AbsTol',1e-10,'RelTol',1e-10));
end
3 Comments
Torsten
on 2 Aug 2023
Edited: Torsten
on 2 Aug 2023
It's not possible. Assume you want to minimize x^2 over x > 0. Then no optimal x exists. Only if you choose the constraint as x>=0, you get x=0 as solution.
This generalizes to the domains over which optimization is performed in general. It has to be always closed (<=), not open (<0).
But I changed the checks to any(W_Ci>=W_C10) in the above code, and as you can see: all constraints are satisfied in the strict sense (W_Ci < W_C10).
More Answers (1)
Torsten
on 1 Aug 2023
Edited: Torsten
on 1 Aug 2023
max_value = -Inf;
T_array = linspace(338.15,368.15,10);
C_M_array = linspace(1,100,10);
C_TEA_array = linspace(1,200,10);
C_CAT_array = linspace(17,400,10);
for i = 1:numel(T_array)
T = T_array(i);
for j = 1:numel(C_M_array)
C_M = C_M_array(j);
for k = 1:numel(C_TEA_array)
C_TEA = C_TEA_array(k);
for l = 1:numel(C_CAT_array)
C_CAT = C_CAT_array(l);
T_r=273.15+0;
A1=4.525e5;
E1 =3985.;
A2 = 0.5101e9;
E2 = 52670.;
A3 = 0.4949e9;
E3 = 15140.;
A4 = 465000.;
E4 = 14330.;
A5 = 1.255;
E5 = 0;
R = 1.98;
k1 = A1*exp(E1*(1/T - 1/T_r)/R);
k2 = A2*exp(E2*(1/T - 1/T_r)/R);
k3 = A3*exp(E3*(1/T - 1/T_r)/R);
k4 = A4*exp(E4*(1/T - 1/T_r)/R);
k5 = A5*exp(E5*(1/T - 1/T_r)/R);
KA = 481.2;
KB = 277.1;
KC = 6906;
KD = 40830;
KE = 59.55;
C_M_K=C_M/(KB*C_TEA+1);
C_CAT_K=C_CAT/(KA*C_M+1);
C_TEA_k=C_TEA/(KC*C_CAT+1)^2;
dCdt=@(t,x) [k1.*C_CAT_K.*C_M_K-k2.*x(1).*C_M-k5.*x(1);k2.*C_M.*x(1)-(k3.*x(2).*C_M)./(KD.*C_TEA+1)-k5.*x(2);(k3.*x(2).*C_M)./(KD.*C_TEA+1)-(k3.*x(3).*C_M)./(KD.*C_TEA+1)-k5.*x(3);(k3.*x(3).*C_M)./(KD.*C_TEA+1)-(k3.*x(4).*C_M)./(KD.*C_TEA+1)-k5.*x(4);(k3.*x(4).*C_M)./(KD.*C_TEA+1)-(k3.*x(5).*C_M)./(KD.*C_TEA+1)-k5.*x(5);(k3.*x(5).*C_M)./(KD.*C_TEA+1)-(k3.*x(6).*C_M)./(KD.*C_TEA+1)-k5.*x(6);(k3.*x(6).*C_M)./(KD.*C_TEA+1)-(k3.*x(7).*C_M)./(KD.*C_TEA+1)-k5.*x(7);(k3.*x(7).*C_M)./(KD.*C_TEA+1)-(k3.*x(8).*C_M)./(KD.*C_TEA+1)-k5.*x(8);(k3.*x(8).*C_M)./(KD.*C_TEA+1)-(k3.*x(9).*C_M)./(KD.*C_TEA+1)-k5.*x(9);(k3.*x(9).*C_M)./(KD.*C_TEA+1)-(k3.*x(10).*C_M)./(KD.*C_TEA+1)-k5.*x(10);(k3.*x(10).*C_M)./(KD.*C_TEA+1)-(k3.*x(11).*C_M)./(KD.*C_TEA+1)-k5.*x(11);(k4.*x(3).*C_M)./(KE.*C_TEA+1)+k5.*x(3);(k4.*x(4).*C_M)./(KE.*C_TEA+1)+k5.*x(4);(k4.*x(5).*C_M)./(KE.*C_TEA+1)+k5.*x(5);(k4.*x(6).*C_M)./(KE.*C_TEA+1)+k5.*x(6);(k4.*x(7).*C_M)./(KE.*C_TEA+1)+k5.*x(7);(k4.*x(8).*C_M)./(KE.*C_TEA+1)+k5.*x(8);(k4.*x(9).*C_M)./(KE.*C_TEA+1)+k5.*x(9);(k4.*x(10).*C_M)./(KE.*C_TEA+1)+k5.*x(10);(k4.*x(11).*C_M)./(KE.*C_TEA+1)+k5.*x(11);k5.*(x(2)+x(3)+x(4)+x(5)+x(6)+x(7)+x(8)+x(9)+x(10)+x(11))];
[t,x]=ode15s(dCdt,[0,3600],[0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0]);
if x(end,15) > max_value
I = i;
J = j;
K = k;
L = l;
max_value = x(end,15);
end
end
end
end
end
T_array(I)
C_M_array(J)
C_TEA_array(K)
C_CAT_array(L)
max_value
13 Comments
See Also
Categories
Find more on Quadratic Programming and Cone Programming 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!

