11.02.2013 Views

Composite Materials Research Progress

Composite Materials Research Progress

Composite Materials Research Progress

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

88<br />

Michaël Bruyneel<br />

In Figure 8.8 the evolution of the vertical displacement under the load is drawn with<br />

respect to the fibers orientation in the case of the homogeneous membrane (Figure 8.4, n=1).<br />

The global minimum displacement is obtained for a value of 170°. When the starting point of<br />

the optimization process of the problem (8.2) is close to 45°, 0° fibers orientation is found as<br />

a local optimum. As -45° is chosen here for the initial design (i.e. 135°), the global optimum<br />

can be reached. This illustrates the fact that a gradient based method is not able to reach the<br />

global optimum, unless the starting point is in its vicinity. In Figure 8.8, the influence of the<br />

mesh refinement on the solution is presented, as well.<br />

8.3. Multi-objective Optimization<br />

A symmetric laminate made of 4 plies and subjected to 2 in-plane load cases is considered.<br />

N 2<br />

3<br />

N 1<br />

1<br />

θ<br />

x<br />

N 6<br />

N 1<br />

Figure 8.9. Laminate subjected to in-plane loads.<br />

The applied loads and the initial configuration are reported in Table 8.2. The load case<br />

(2) is variable : the factor k takes the values 0,1,2,…,8. The extreme load cases are, on one<br />

hand (1000,0,0) and on the other hand the combination of (1000,0,0) and (0,2000,0) N/mm.<br />

Table 8.2. Definition of the problem: load case and starting point<br />

Load case (1) Load case (2) Initial orientations Initial thickness<br />

( N 1,<br />

N 2 , N6<br />

) ( N 1,<br />

N 2 , N6<br />

) θ = ( θ1,<br />

θ 2 )<br />

t = ( t1,<br />

t2<br />

)<br />

in N/mm<br />

in N/mm<br />

in degrés<br />

en mm<br />

(1000,0,0) (0, k × 250 ,0) (30,120) (1,2)<br />

The performance of three approximation schemes are compared : GCMMA, MMA and<br />

SAM. The optimization problem writes :<br />

1<br />

min max ε( ) ( j)<br />

2 j Aε<br />

j 1,2<br />

T<br />

θ, t =<br />

TW ( j)<br />

( θ i , ti<br />

) ≤1<br />

i<br />

, j = 1,<br />

2<br />

2<br />

N 2

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!