D
ESIGN OF MICROSTRUCTURES USING STRESS
–
BASED
TOPOLOGY OPTIMIZATION
Maxime Collet
1, Matteo Bruggi
2, Lise Noël
1, Simon Bauduin
1, Pierre Duysinx
11
University of Liège, Aerospace and Mechanical Engineering Department, Liège, Belgium
2Politecnico di Milano, Civil and Environmental Engineering Department,Milan, Italy
M
OTIVATION
Microstructural design (MD) using topology optimization along with ho-mogenization to compute equivalent elastic properties is used to conceive optimal microstructures addressing specific needs:
max K s.t. V ≤ 0, 5: Koptimized = 0, 6734
Accounting for the stress level within the optimized layouts ensures the
structural integrity of the microstructure.
S
ENSITIVITY ANALYSIS OF THE STRESSES
Sensitivity along with the qp-relaxation (Bruggi,2008)a: ∂σV M,e ∂xk = δek(p − q)x p−q−1 e σV M,e + ∂σV M,e ∂xk x p−q e
Adjoint approach (active set : Na selected constraints < N total constraints)
∂σV M,e ∂xk = λ T ∂K ∂xk u − λ T pxp−1k fe,0i
with the adjoint problem and force vector:
Kλ = − h uTe M0eue−1/2 M0eue iT fe,0i = n1 X i=1 n2 X j=1 wiwjBeT H0eεi0detJ
aBruggi M (2008) On an alternative approach to stress constraints relaxation in topology
optimization.Struct Multidisc Optim 36:125-141
E
XAMPLE
1: C
LASSICAL
B
ULK MAXIMIZATION
The maximization of the bulk modulus under stress constraints with σy0 = 1, 4N/m2
max K s.t. V ≤ 0.5 AND σ ≤ σy0: Koptimized = 0, 6719
• Similar topology compared to the classical solution but with a
controlled stress level ( σσ0
y ≤ 1)
• More square shaped compared to the classical solution • Lower optimized bulk modulus
E
XAMPLE
2: Bronzi di Riace
SEISMIC INSULATOR
Bronzi di Riace statue with its marble seismic insulator:
Currently: Four marble balls used to isolate the statues in case of earth-quakes. Design requirements are high bulk and low shear (under pre-scribed value). A topology optimization can be formulated:
max
0<xmin≤xe≤1 K
s.t. K(x)ui = f (x)i; i = 1, 2
G ≤ G∗
hσV M,e(x)i ≤ σy0
• Two load cases !
• Starting point: Hole with
void at the center of the RVE
• σy0 = 1.8N/m2 and G ≤ 10−3N/m2
• No stress constraints : Koptimized = 0, 3665
• With stress constraints : Koptimized = 0, 4376
• Control of the local stress constraints ( σσ0
y ≤ 1)
• Change in the topology of the optimized layouts • Existence of local optima !
C
ONCLUSION AND
P
ERSPECTIVES
• Control of the stress level introduced in a topology optimization
framework for microstructural design (sensitivity analysis)
• The optimization might be steered toward better optimum =⇒
Presence of local minima
• Speed up the computation ! CPU requirement is still to high ! • Extension to 3D manufacturable microstructures
A
CKNOWLEDGEMENTS
The first author, Maxime Collet, would like to acknowledge the Belgian National Fund for Scientific research (FRIA) for its financial support. Part of this work was done when the first author was spending a doctoral stay at Politecnico di Milano