Optimization: Function taking a variable value from out of the bounds

 (Files attached)

   x0 = abs([0.01, 1, 283.15, 8, 10]);
  %options = optimset('PlotFcns',@optimplotfval);
  fitnessfcn = @(x)Lobo(x(1), x(2), x(3), x(4), x(5));
  nonlcon1 = @(x)nonlcon(x(1), x(2), x(3), x(4), x(5));
  lb = abs([0.01, 1, 283.15, 8, 10]);
  ub = abs([10, 18.3, 1473.15 ,20, 50]);
  [X, fval] = fmincon(fitnessfcn, x0, [], [], [], [], lb, ub, nonlcon1);

I am getting imaginary values of intermediary equations in Lobo because d0 (diameter) is going below 0. Why is this happening? Also, the calculated value of MATLAB do not match the manually calculated value (calculator). e.g. F cannot be beyond 1 and when a calculator is applied to the same formula it comes around 0.1.

Result: F =

      1.0087

      1.0087

d0 =

   -1.1697e+003

   -1.1697e+003

h0 =

    1.0303e+014 -7.4856e+013i

    1.0303e+014 -7.4856e+013i

Tw =

    314.2630

    314.2630

Lf =

      0.3670

      0.3670

tf =

      0.0025

      0.0025

dbc =

      0.0067

      0.0067

cbt =

      1.5030

      1.5030

dtc =

      1.5097

      1.5097

ntubes =

      1.5097

      1.5097

Ar =

    5.6371e+003

    5.6371e+003

qr =

    2.3992e+017 -1.7431e+017i

    2.3992e+017 -1.7431e+017i

[Torsten]

I think you mean

F = (C - x + 2.*sinh(2./x))/(2.*pi);

instead of

F = (C - x + 2.*sinh(2./x))/2.*pi;

Best wishes

Torsten.

[Devdatt Thengdi]

Oh! Thanks. Anything about the first question?

[Devdatt Thengdi]

Also, could you tell me why the matrix dimensions do not agree here?

dbc = db./(sin(180./nb) + (pi./nb)); %Burner circle diameter

[Torsten]

Since you don't provide a mathematical problem formulation, there will be nobody in the forum who can answer questions with regards to content.

For your question above: Use the MATLAB debugger. Or print out db and nb before dbc is calculated.

Best wishes

Torsten.

[Devdatt Thengdi]

All I need to know " Why code is extending values of d0 beyond the range I have provided?" i.e. In the line:

lb = abs([0.01, 1, 283.15, 8, 10]);
  ub = abs([10, 18.3, 1473.15 ,20, 50]);
  [X, fval] = fmincon(fitnessfcn, x0, [], [], [], [], lb, ub, nonlcon1);

I have correctly bound d0 between 0.01:10. But, when you run the program:

d0 =
   -1.1697e+003
   -1.1697e+003

d0 is negative. Why is that? Is there any syntax error?

[Torsten]

And the output you list for the variables is right after you start fmincon ? Or after some iterations when h0 already became complex-valued ? If the latter is the case, this should explain why the calculation leaves senseful limits.

[Devdatt Thengdi]

It is after some iterations. And how does this explain it? I mean I have set limits on the variables. I have set a nonlcon (on qr) so, there must some set of d0, L, Tg, nb, uo which qr lies between 32000:45000. What am I missing out on here? Is the program forcing values(d0) out of the bounds because there exists no qr (in the given range) for the given range of d0? Am I correct here?

[Torsten]

This might be the reason.

I suggest you make a pattern search:

Take a certain number of values for the variables d0, L, Tg, nb and uo between their respective lower and upper limits. Then calculate qr and qout for all combinations of these values and inspect whether you get feasible points for your problem.

Best wishes

Torsten.