try&erorr

i want to ask how can i make try&erorr with reasonable erorr the program is 3 tanks have a different energy .. and they connected together by pipe .. this pipe meet togethrt at point j the iteration is to get the energy at j (Ej) .. and then find the flow rate (Q) at each pipe theoretical summtion of Q at j equal 0

this is the program but it has an erorr

clear , clc 
fprintf('            \n <<< this program calculate Q in three branched pipes from three tanks >>> \n %12.5f ')

fprintf (' \n 1- Data for highest tank \n %12.5f')

% Data for highest tank
E1=input ('\n Enter tank_ energy (m) :  ');
f1=input (' Enter pipe_friction coefficient (m) :  ');
d1=input (' Enter pipe_diameter (m) :  ');
l1=input (' Enter pipe_length (m) : ');

fprintf (' \n 2- Data for intermediate tank \n %12.5f ')

% Data for intermediate tank
E2=input ('\n Enter tank_ energy (m) :  ');
f2=input (' Enter pipe_friction coefficient (m) :  ');
d2=input (' Enter pipe_diameter (m) :  ');
l2=input (' Enter pipe_length (m) : ');

fprintf (' \n 3- Data for lower tank \n %12.5f ')

% Data for lower tank
E3=input ('\n Enter tank_ energy (m) :  ');
f3=input (' Enter pipe_friction coefficient (m) :  ');
d3=input (' Enter pipe_diameter (m) :  ');
l3=input (' Enter pipe_length (m) : ');

% resistance calculation
R1=(.8*f1*l1)/(9.8*d1^5); % 1st pipe resistance
R2=(.8*f2*l2)/(9.8*d2^5); % 2nd pipe resistance
R3=(.8*f3*l3)/(9.8*d3^5); % 3rd pipe resistance

% junction's energy assumption Ej

Ej=(E1+E3)/2;   %  E1 <Ej< E3

% Q calculation

Q1=((abs(E1-Ej))/(R1))^.5;
Q2=((abs(E2-Ej))/(R2))^.5;  
Q3=((abs(E3-Ej))/(R3))^.5;

if E2<Ej 
    Qtot=Q1-Q2-Q3; 
end

% iteration

if Qtot>0
    err=1;
    iter=0;
while abs(err)>.001
    iter=iter+1;
    Ej=Ej+.1;
Q1=((abs(E1-Ej))/(R1))^.5;
Q2=((abs(E2-Ej))/(R2))^.5;  
Q3=((abs(E3-Ej))/(R3))^.5;
Qtot=Q1-Q2-Q3; 
err=Qtot;
end
end

[Jan]

Please use a proper code formatting as explained at the "Markup help" link. Currently your code is not readable.

Answer 1

[Walter Roberson]

It isn't clear to me that the required error tolerance is certain to be satisfied when Ej is incremented by a constant value. Shouldn't you be doing some zero-crossing detection to zero in on proper Ej ? You could arrange it as an fzero or fmincon optimization.