Date post: | 04-Jun-2018 |
Category: |
Documents |
Upload: | luke-parry |
View: | 218 times |
Download: | 0 times |
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 1/24
Level Set Method Applied to Topology Optimization 1 / 1 3
Level Set Method
Applied to Topology Optimization
David Herrero Perez
February 2012
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 2/24
Level Set Method Applied to Topology Optimization 2 / 1 3
Introduction
Level Set Method
1 Introduction
2 The Level Set Method Applied to Topology Optimization
3 Possible works
4 References
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 3/24
Level Set Method Applied to Topology Optimization 3 / 1 3
Introduction
The Level Set Method (LSM)
The LSM is a numerical technique for trackinginterfaces and shapes.
Advantages of LSM
Numerical computations involving curves andsurfaces on a fixed Cartesian grid can beperformed without having to parameterize theseobjects, which is called the Eulerian approach [1].
The LSM makes it very easy to follow shapes that
change topology, for example:Shape splits in two.Develops holes.The reverse of previous operations.
The algorithms for processing level sets have vastparallelization potential.
The Level Set Method.
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 4/24
Level Set Method Applied to Topology Optimization 3 / 1 3
Introduction
Curve representation (2D)
The closed curve Γ is represented using an auxiliaryvariable ϕ called the level set function. Γ is representedas the zero level set of ϕ by
Γ = (x , y )| ϕ(x , y ) = 0, (1)
and the level set method manipulates Γ ”implicitly”,through the function ϕ. ϕ is assumed to take positivevalues inside the region delimited by the curve Γ andnegative values outside [2, 3].
The Level Set Method.
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 5/24
Level Set Method Applied to Topology Optimization 3 / 1 3
Introduction
The Level Set EquationWhen the curve Γ moves in the normal direction with aspeed v , then the level set function ϕ satisfies the ”levelset equation”
∂ϕ
∂ t = v |∇ϕ|, (2)
where | · | is the Euclidean norm (denoted customarilyby single bars in PDEs), and t is time. This is a partialdifferential equation, in particular a Hamilton-Jacobiequation, and can be solved numerically, for example byusing finite differences on a Cartesian grid [2, 3].
The Level Set Method.
L l S M h d A li d T l O i i i 3 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 6/24
Level Set Method Applied to Topology Optimization 3 / 1 3
Introduction
The Level Set EquationThe numerical solution of the level set equation,however, requires sophisticated techniques because:
Simple finite difference methods fail quickly.
Upwinding methods, such as the Godunov’s
scheme, fare better.
Possible troublesThe LSM does not guarantee the conservation of the volume and the shape of the level set in anadvection field that does conserve the shape andsize.
Instead, the shape of the level set may getseverely distorted and the level set may vanishover several time steps.
The Level Set Method.
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 7/24
Level Set Method Applied to Topology Optimization 4 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 8/24
Level Set Method Applied to Topology Optimization 4 / 1 3
The Level Set Method Applied to Topology Optimization
Level Set Method
1 Introduction
2 The Level Set Method Applied to Topology Optimization
3 Possible works
4 References
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 9/24
Level Set Method Applied to Topology Optimization 6 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 10/24
Level Set Method Applied to Topology Optimization 6 / 1 3
The Level Set Method Applied to Topology Optimization
Objective
The LSM is used to to find a local minimum for the optimization problem.
Boundary representation of domain ΩThe level set function is used for describing the structure that occupies some domainΩ as follows:
ϕ(x , y )
< 0 if (x , y ) ∈ Ω= 0 if (x , y ) ∈ ∂ Ω> 0 if (x , y ) /∈ ∂ Ω
(4)
where (x , y ) is any point in the design domain, and ∂ (x , y ) is the boundary of Ω.
Level Set Method Applied to Topology Optimization 7 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 11/24
Level Set Method Applied to Topology Optimization 7 / 1 3
The Level Set Method Applied to Topology Optimization
Evolution equation
The following evolution equation is used to update the level-set function and hencethe structure:
∂ϕ
∂ t = v |∇ϕ| − wg (5)
t represents time.
v (x , y ) and g (x , y ) are scalar fields over the design domain Ω.
w is a positive parameter which determines the influence of the term involving g .
Scalar fieldsThe field v determines geometric motion of the boundary of the structure. It ischosen based on the shape derivative of the optimization objective.
The term involving g is a forcing term which determines the nucleation of newholes within the structure. It is chosen based on the topological derivative of theoptimization objective.
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 12/24
Level Set Method Applied to Topology Optimization 7 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 13/24
pp p gy p /
The Level Set Method Applied to Topology Optimization
Evolution equation
The following evolution equation is used to update the level-set function and hencethe structure:
∂ϕ
∂ t = v |∇ϕ| − wg (5)
t represents time.
v (x , y ) and g (x , y ) are scalar fields over the design domain Ω.
w is a positive parameter which determines the influence of the term involving g .
Nucleation problem
However the standard evolution equation has the major drawback that new voidregions cannot be nucleated within the structure.
Hence, the additional forcing term involving g is usually added to ensure thatnew holes can nucleate within the structure during the optimization process.
Level Set Method Applied to Topology Optimization 8 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 14/24
The Level Set Method Applied to Topology Optimization
Level Set FunctionThe level-set function can be discretized with grid-points centered on theelements of the mesh.
If c e represents the position of the center of the element e , then the discretizedlevel-set function ϕ satisfies:
ϕ(c e )
< 0 if x e = 1
= 0 if x e = 0 (6)
The discrete level-set function can then be updated to find a new structure bysolving (5) numerically.
LSF InitializationThe level-set function ϕ should be initialized.
When the forcing term involving g is added, such an initialization is not critical,and a signed distance function is enough to address the topology optimizationproblem.
Level Set Method Applied to Topology Optimization 8 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 15/24
The Level Set Method Applied to Topology Optimization
Level Set FunctionThe level-set function can be discretized with grid-points centered on theelements of the mesh.
If c e represents the position of the center of the element e , then the discretizedlevel-set function ϕ satisfies:
ϕ(c e )
< 0 if x e = 1
= 0 if x e = 0 (6)
The discrete level-set function can then be updated to find a new structure bysolving (5) numerically.
Solving LSM numericallyAn upwind finite difference scheme is used so that the evolution equation can beaccurately solved.
The time step for the finite difference scheme is chosen to satisfy the
Courant-Friedrichs-Lewy (CFL) stability condition: ∆t ≤ h
max |v |, where h is the
minimum distance between adjacent grid-points in the spacial discretization.
Level Set Method Applied to Topology Optimization 9 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 16/24
The Level Set Method Applied to Topology Optimization
Scalar fields (v and g )
The scalar fields are typically chosen based on the shape and topological sensitivitiesof the optimization objective, respectively.
Volume constraint
To satisfy the volume constraint, they are chosen using the shape and topologicalsensitivities of the Lagrangian:
L = c (x ) + λk (V (x ) − V req ) + 1
2Λk [V (x ) − V req ]
2 (7)
where λk
and Λk
are parameters which change with each iteration k of theoptimization algorithm. They are updated using the scheme:
λk +1 = λk + 1
Λk (V (x ) − V req ), Λk +1 = αΛk (8)
where α ∈ (0, 1) is a fixed parameter.
Level Set Method Applied to Topology Optimization 9 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 17/24
The Level Set Method Applied to Topology Optimization
Scalar fields (v and g )
The scalar fields are typically chosen based on the shape and topological sensitivitiesof the optimization objective, respectively.
Normal velocity v
This velocity is chosen as a descent direction for the Lagrangian L using itsshape derivative.
In the case of traction-free boundary conditions on the moving boundary, theshape sensitivity of the compliance objective c(x) is the negative of the strainenergy density:
∂ c
∂ Ω
|e = −u T e k e u e (9)
and the shape sensitivity of the volume V (x ) is
∂ V
∂ Ω|e = 1 (10)
Level Set Method Applied to Topology Optimization 9 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 18/24
The Level Set Method Applied to Topology Optimization
Scalar fields (v and g )
The scalar fields are typically chosen based on the shape and topological sensitivitiesof the optimization objective, respectively.
Normal velocity v
Using these shape sensitivities, the normal velocity v within element e atiteration k of the algorithm is:
v |e = − ∂ L
∂ Ω|e = u T
e k e u e − λk − 1
Λk (V (x ) − V req ) (11)
Level Set Method Applied to Topology Optimization 9 / 1 3
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 19/24
The Level Set Method Applied to Topology Optimization
Scalar fields (v and g )
The scalar fields are typically chosen based on the shape and topological sensitivitiesof the optimization objective, respectively.
Forcing term g
The the forcing term g can be taken as g = −sign(ϕ)δT L, where δT L is thetopological sensitivity of the Lagrangian L.
For compliance minimization, nucleating solid areas within the void regions of the design is pointless because such solid regions will not take any load.
Therefore holes should only be nucleated within the solid structure and
g
δT L if ϕ < 0
0 if ϕ ≥ 0 (12)
Level Set Method Applied to Topology Optimization 9 / 1 3
Th L l S M h d A li d T l O i i i
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 20/24
The Level Set Method Applied to Topology Optimization
Scalar fields (v and g )
The scalar fields are typically chosen based on the shape and topological sensitivitiesof the optimization objective, respectively.
Topological sensitivity
The topological sensitivity of the compliance objective function in two dimensions withtraction-free boundary conditions on the nucleated hole and the unit ball as the modelhole can be expressed as:
δT c |e = −π(λ + 2µ)
2µ(λ + µ)(4µu T
e k e u e + (λ − µ)u T e (k Tr )e u e ) (13)
u T e (k Tr )e u e is the finite element approximation to the product tr (σ) tr (ε),
where σ is the stress tensor and ε is the strain tensor.
λ and µ are the Lame constants for the solid material.
Level Set Method Applied to Topology Optimization 10/13
P ibl k
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 21/24
Possible works
Level Set Method
1 Introduction
2 The Level Set Method Applied to Topology Optimization
3 Possible works
4 References
Level Set Method Applied to Topology Optimization 11/13
Possible works
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 22/24
Possible works
Possibilities of future works using LSM
Parallel computing for accelerating the optimization based on such a method
using:Distributed CPU approach (communications based on sockets, MPI,event-based blackboards, ...).GPU computing.
Apply the LSM to problems that require the tracking of a wave front.
Level Set Method Applied to Topology Optimization 12/13
References
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 23/24
References
Level Set Method
1 Introduction
2 The Level Set Method Applied to Topology Optimization
3 Possible works
4 References
Level Set Method Applied to Topology Optimization 13/13
References
8/13/2019 Level Set Slides
http://slidepdf.com/reader/full/level-set-slides 24/24
References
References
Osher, S., Sethian, J.A.Fronts propagating with curvature-dependent speed: Algorithms based on
Hamilton-Jacobi formulations
J. Comput. Phys. 79: 12–49, 1988.
Osher, Stanley J., Fedkiw, Ronald P.Level Set Methods and Dynamic Implicit Surfaces
Springer-Verlag, 2002.
Sethian, James A.Level Set Methods and Fast Marching Methods : Evolving Interfaces in
Computational Geometry, Fluid Mechanics, Computer Vision, and Materials
Science Cambridge University Press, 1999.
Enright, D.; Fedkiw, R. P.; Ferziger, J. H.; Mitchell, I.A hybrid particle level set method for improved interface capturing
J. Comput. Phys. 183: 83–116., 2002.