Optimization of truss with regards to mass, with cross sectional areas and node position as design variables
6 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Tobias Rattfelt
am 8 Jan. 2017
Kommentiert: aws ali
am 19 Jan. 2017
This is a homework assignment so I only need a push in the right direction.
I have a 4 bar truss with 4 nodes, bar 1 and 2 are fixed in one end so there are 4 DOFs. Picture of truss
The objective function to be minimized is the mass of the truss. The design variables are the cross sectional areas of the bars and the location of node 3.
The constraints are displacement on node 4 and the stresses in the bars.
My question is, how do I start? I am somewhat familiar with optimizing functions of multivariable equations, but since the constraints are stresses and displacements I am having some troubles figuring out how to go about the problem, since I cant very well include the equations for the stresses in one constraint function?
So my idea was to use penalty functions for the stress constraints, but so far I have only used exterior penalty functions and that seem to be a very bothersome way to go about the problem.
I have the hint from our professor to use the optimization toolbox, but the fmincon that I thought of first does not seem applicable here, thankful for any suggestions!
Best regards
/T
0 Kommentare
Akzeptierte Antwort
John D'Errico
am 8 Jan. 2017
Bearbeitet: John D'Errico
am 8 Jan. 2017
Well, you did ask for only a push.
There are 4 degrees of freedom, so the 4 cross-sectional areas of the bars.
You can compute the truss mass trivially given the bar lengths, the cross sectional areas, and the density of the material.
As far as the constraints go, you apparently have a constraint on displacement at node 4. So IF you knew the areas of each bar, and the force applied, then you could compute the displacement. That is a simple function of the unknowns.
Similarly, you apparently have a constraint on the stress in any of the bars, so a maximum stress allowed. Again, if you knew the cross-sections of each bar, and the force applied, then you can compute the displacement of each node due to that force. Therefore the strain in each bar is a trivial computation, and so also the stress. So you will have 4 constraints. In fact, those constraints should be linear, if I recall my grad school Mech E days from 30+ years ago. (Ok, linear as long as F is not sufficiently high that the truss exceeds the limits of linear elasticity and undergoes buckling. That gets nasty to deal with.)
And, yes, you WILL use fmincon. In fact, as I said, if you write down the equations carefully, I think the constraints will be linear inequality constraints, although you could write them as nonlinear inequalities. The latter would be the lazy way out, but still viable. :)
3 Kommentare
aws ali
am 19 Jan. 2017
Hi friend, Have you came throug the book of (Applied Optimization with MATLAB Programming) by P. Venkataraman? If not you can go to the website of the book, they allow to download the examples of optimization (Matlab codes), which I assume they very good to start with. You need chapter 10 from that book where the examples associted with the matlab codes stated there (i.e. ch.10).
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Structural Analysis finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!