Sie sind auf Seite 1von 3

To minimize the mass, ModelCenter provides an Optimization Tool, which enables the user to define the objective, the

constraints, design variables and finally select an optimization algorithm over 35options.

Once, the optimization method chosen, the user can precise more details about the algorithm concerned: convergence tolerance, max-iteration To choose an appropriate algorithm, we analyzed our problem. The Mass objective to minimize is a linear function but the miscellaneous constrains are nonlinear. Since, we have 66 design variables and 700 constrains we opt for an algorithm which can handle large scale optimization and globally nonlinear problem. To simplify the process, we decide to only optimize the crossbeam through its thickness in different location of the crossbeam defined in the Excel Template.

Thus, we need to define a mathematical approach that optimizes material layout with a given space, for a given set of loads and boundary conditions such that the resulting layout meets a prescribed set of performance target. The topology optimization will try to find the best design that meets the design requirements. First algorithm chosen : SLP

Sequential linear programming (SLP) algorithms have been used successfully in structural design. This class of methods is well suited for solving large nonlinear problems due to the fact that it does not require the computation of second derivatives, so the iterations are cheap. This is a first order method that requires computing the first derivatives of the function and constraints. The problem we want to solve takes the form: Minimize Subject to The method begins with an initial design already existing; this is a crossbeam with specific dimensions and thickness. The builder defined a range for the design variables and the constraints to get a workable optimum design. To begin, the SLP method linearize the objective and the constraints around the solution using the Taylor. ( ( ) ) ( ) ( ( ) ( ) ) [ ]

Then the linear programming problem is stated as:


Minimize: Subject to:

( ( )

) ( ) (

) ( )

Thus the sequential linear programming (SLP) algorithm is an iterative method that generates and solves a sequence of linear problems in the previous form. At each iteration k of the algorithm, a previously computed point is used to generate the linear programming problem. This previous formulation is solved using the Simplex method to obtain the new vector . The original constraints are evaluated at , that is, ( ) and if ( ) for all j for some pre-specified positive tolerance value the constraints with the suitable are assumed to be satisfied and the procedure is stopped with optimal solution.

Otherwise, If some constraints are violated, i.e. constraint is determined, for instance: ( )

for some j, the most violated

Which is linearized about (

as : ) ( ) ( )

Then, it is included as the inequality constraint in the previous linear programming (LP) problem. The new iteration number is set to (k+1) and the new LP problem is solved with inequality constraints. ModelCenter SLP optimization:

Das könnte Ihnen auch gefallen