Scribd
Upload
41 views7 pages

Automatic Differentiation in Solid Mechanics

This document discusses using automatic differentiation (AD) for solid mechanics problems. Traditional finite element methods calculate derivatives numerically, while AD calculates derivatives analytically by tracking operations during computations. AD is applied to calculate the derivative of the free energy functional instead of the residual, allowing derivatives to be computed efficiently using dual numbers. Examples shown include a non-linear truss, axisymmetric problem, plane stress, and finite strain continuum mechanics. Source code for the examples is available online.

Uploaded by

ticonzero2010
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
41 views7 pages

Automatic Differentiation in Solid Mechanics

This document discusses using automatic differentiation (AD) for solid mechanics problems. Traditional finite element methods calculate derivatives numerically, while AD calculates derivatives analytically by tracking operations during computations. AD is applied to calculate the derivative of the free energy functional instead of the residual, allowing derivatives to be computed efficiently using dual numbers. Examples shown include a non-linear truss, axisymmetric problem, plane stress, and finite strain continuum mechanics. Source code for the examples is available online.

Uploaded by

ticonzero2010
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Automatic Differentiation for solid mechanics

1 2
Andrea Vigliotti Ferdinando Auricchio

1
Italian Aerospace Research Centre
2
University of Pavia

ICTAM 2020+1, Milano, August 22-27, 2021


Automatic Differentiation for solid mechanics
traditional FE

Z  
∂Fij
δW = Pij − f0k δuk dV0 = 0
V0 ∂uk

k
NBE NBW  
X X ∂Fij
r= wlk Pij − f0 =0
∂u rk
k=1 l=1 l

N
BE X
BWNk  
∂r X ∂Pij ∂Fhk ∂Fij ∂f0
= wlk −
∂u ∂Fhk ∂u ∂u ∂u rlk
k=1 l=1
Automatic Differentiation for solid mechanics
traditional FE

Z  
∂Fij
δW = Pij − f0k δuk dV0 = 0
V0 ∂uk

k
NBE NBW  
X X ∂Fij
r= wlk Pij − f0 =0
∂u rlk
k=1 l=1

N
BE X
BW Nk " #
∂r X ∂Pij ∂Fhk ∂Fij ∂f0
= wlk −
∂u ∂Fhk ∂u ∂u ∂u
k=1 l=1 rlk
Automatic Differentiation for solid mechanics
traditional FE AD of free energy

Z   Z
∂Fij
δW = Pij − f0k δuk dV0 = 0 Ψ= (φ − b0 ) dV0 δΨ = 0
V0 ∂uk V0

k
NBE NBW  
X X ∂Fij ∂Ψ
r= wlk Pij − f0 =0 r= =0
∂u rlk ∂u
k=1 l=1

N Nk " #
∂r BE X
X BW
∂Pij ∂Fhk ∂Fij ∂f0 ∂r ∂2 Ψ
= wlk − =
∂u ∂Fhk ∂u ∂u ∂u ∂u ∂u∂u
k=1 l=1 rlk
Automatic Differentiation for solid mechanics
traditional FE AD of free energy

Z   Z
∂Fij
δW = Pij − f0k δuk dV0 = 0 Ψ= (φ − b0 ) dV0 δΨ = 0
V0 ∂uk V0

k
NBE NBW  
X X ∂Fij ∂Ψ
r= wlk Pij − f0 =0 r= =0
∂u rlk ∂u
k=1 l=1

N Nk " #
∂r BE X
X BW
∂Pij ∂Fhk ∂Fij ∂f0 ∂r ∂2 Ψ
= wlk − =
∂u ∂Fhk ∂u ∂u ∂u ∂u ∂u∂u
k=1 l=1 rlk

Forward mode AD is based on dual numbers


x ≡ x0 + xi ıi + xij ıij where ıij ≡ ıi ⊗ ıj + ıj ⊗ ıi

with the product rule ıi ıj ≡ ıij and ıij ık ≡ 0


Examples

i3 i2
i1

Large displacement non-linear truss structure Axisymmetric problem with internal constraint

L = 20 Atop f
e
d
Openings, RO = 1.50
L = 20 c
Inclusions, RI = 1.25 h
i3
b
i2 i1 i2 i1
i1 a
Abtm i3

Plane stress with internal constraints Finite strain/displacement 3D continuum

Source code for the examples is available from https://github.com/avigliotti/AD4SM.jl


A. Vigliotti and F. Auricchio, Automatic differentiation for solid mechanics, Archives of Computational Methods
in Engineering, vol. 28, no. 3, pp. 875-895, 2021.
Automatic Differentiation for solid mechanics

1 2
Andrea Vigliotti Ferdinando Auricchio

1
Italian Aerospace Research Centre
2
University of Pavia

ICTAM 2020+1, Milano, August 22-27, 2021

You might also like