Earthquake Engineering by the beach - Capri, 2-4 July 2009

Advanced Computational Tools for

Structural Mechanics and
Earthquake Engineering

Alessandro Reali
Dipartimento di Meccanica Strutturale, Università degli Studi di Pavia, Italy
European Centre for Training and Research in Earthquake Engineering (EUCENTRE), Pavia, Italy
Istituto di Matematica Applicata e Tecnologie Informatiche (IMATI), CNR, Pavia, Italy
Centro di Simulazione Numerica Avanzata (CeSNA), IUSS, Pavia, Italy

alessandro.reali@unipv.it http://www-1.unipv.it/alereali

Reasearch Group Webpage:

http://www-1.unipv.it/dms/compmech
 Earthquake Engineering by the beach - Capri, 2-4 July 2009

Acknowledgements

in alphabetical order:

Jamal Arghavani (Sharif University of Technology);

Domenico Asprone (University of Naples “Federico II”);
Gabriele Attanasi (ROSE School, Pavia);
Ferdinando Auricchio (University of Pavia);
Giuseppe Balduzzi (University of Pavia);
Lourenco Beirao da Veiga (University of Milan);
Annalisa Buffa (IMATI/CNR, Pavia);
Thomas JR Hughes (University of Texas at Austin);
Carlo Lovadina (University of Pavia);
Gaetano Manfredi (University of Naples “Federico II”);
Andrea Prota (University of Naples “Federico II”);
Giancarlo Sangalli (University of Pavia);
Ulisse Stefanelli (IMATI/CNR, Pavia);

as well as the whole research groups led by Prof. Auricchio

and Prof. Hughes.
 Presentation outline

  Isogeometric Analysis
  Shape Memory Alloy Modeling and Application
  Other Interesting Research Works in Progress
•  Meshless Methods
•  Beam Finite Elements
  Conclusions
 Presentation outline

  Isogeometric Analysis
  Shape Memory Alloy Modeling and Application
  Other Interesting Research Works in Progress
•  Meshless Methods
•  Beam Finite Elements
  Conclusions
 FEM analysis and CAD geometry

CAD (1970’s - 1980’s) – Engineering Design Process:

  engineering designs are encapsulated in CAD systems;
  CAD geometry is exact;
  more than 100,000 analyses of CAD designs are performed
in engineering offices throughout the world every day
FEM (1950’s - 1960’s) – Engineering Analysis Process:

  CAD geometry is replaced by FEM geometry (“mesh”);

  mesh generation accounts for more than 80% of overall
analysis time and is the major bottleneck;
  mesh refinement requires interaction with CAD geometry;
  the mesh is an approximate geometry
IDEA: Isogeometric Analysis
  analysis framework based on functions – e.g., NURBS –
capable of representing exact geometry;
  one, and only one, geometrical description;
  mesh refinement vastly simplified
 NURBS

Main properties of Non-Uniform Rational B-Splines:

 basis functions from an open knot vector constitute a partition of unity;

 the support of each basis function is compact;
 basis functions are point-wise non-negative;
 basis functions possess high continuity;
 NURBS enjoy the affine covariance property (i.e., NURBS affine
transformation is obtained from control points affine transformation);
 knot insertion and degree elevation are simple operations

References:

 Rogers. An Introduction to NURBS with Historical Perspective. Academic

Press, San Diego, CA, 2001
 Piegl and Tiller. The NURBS Book (Monographs in Visual Communication),
2nd ed. Springer-Verlag, New York, 1997.
 Isogeometric Analysis

Isogeometric Analysis:
exact geometry alternative to standard FE analysis (based, e.g., on NURBS),
including FEA as a special case, but offering other possibilities:
  precise and efficient geometric modeling
  simplified mesh refinement
  superior approximation properties
  integration of design and analysis
 Isogeometric analysis

Main features:
 geometry defined by control points (associated with basis functions)
 isoparametric concept invoked: unknown variables (dof’s or control variables)
represented in terms of the basis functions defining the geometry
 three refinement strategies available:
-“h-refinement” (by knot insertions)
-“p-refinement” (by degree elevation)
-“k-refinement” (sort of high-order/high-continuity h-refinement)
 array assembly strategy same as in FEM; Dirichlet b.c. applied to control
variables, Neumann b.c. satisfied naturally, as in FEM
 structural analysis: all rigid body motions and constant strain states
represented exactly (i.e., standard “patch tests” passed)
Reference:
 Hughes, Cottrell, and Bazilevs. Isogeometric analysis: CAD, finite elements,
NURBS, exact geometry, and mesh refinement. Computer Methods in
Applied Mechanics and Engineering, 194:4135–4195, 2005.
 Structural vibrations

Isogeometric analysis for structural vibrations:

main advantages with respect to standard FEM

  basis function high continuity;

  basis function point-wise non-negativity, which implies
element-wise non-negativity of the consistent mass matrix;
  exact geometry

Focus on spectrum analysis, i.e.

analysis of the error for the natural
frequencies in structural vibration
problems
 Structural vibrations

Longitudinal vibrations of a fixed-fixed rod

Numerical results:
discrete spectra normalized
to the exact solution
(quadratic NURBS versus
quadratic FE)

6th eigenmode 9th eigenmode 12th eigenmode 15th eigenmode 18th eigenmode

NURBS (top) and FEM (bottom) [21 c.p.] vs. exact (red-dashed)
 Structural vibrations

Longitudinal vibrations of a fixed-fixed rod

high order NURBS versus high order FE
(i.e.: k-method versus p-method)

NURBS spectra show a nice convergence to the exact solution increasing the
order p, while higher-order FEM have no approximability for higher modes
 Structural vibrations

2D problem (membrane)

Transversal vibrations of a
clamped elastic membrane:
NURBS normalized discrete
spectrum

high order NURBS vs high order FE

(i.e.: k-method vs p-method)
 Structural vibrations

Fourth order problems (i.e., beams and plates)

1D: Transversal vibrations of a

simply-supported Euler-Bernoulli
beam: NURBS normalized
discrete spectrum

2D: Transversal vibrations of a

simply-supported Kirchhoff plate:
NURBS normalized discrete
spectrum
 Structural vibrations

Vibrations of a clamped thin circular plate

using 3D solid elements

exact solution:
where:
and: 8-element mesh

first three frequencies obtained using first three eigenmodes

different order NURBS basis functions (p = 4, q = 5, r = 2)
 Structural vibrations

Study of the NASA Aluminum Testbed Cylinder (ATC) structure

Experimental modal results available

for the whole structure as well as for
single components

NURBS (as-drawn) model of the

whole structure (frame + skin and
frame only):

Model details:
 Structural vibrations

Study of the NASA Aluminum Testbed Cylinder (ATC) structure

Stringer frequencies and Main rib frequencies and Whole structure frequencies and
first three x-z modes first three out-of-plane modes first Rayleigh and Love modes
 Structural vibrations

References:

 AR. An isogeometric analysis approach for the study of structural vibrations.

Journal of Earthquake Engineering, 10(s.i.1):1–30, 2006.

 Cottrell, AR, Bazilevs, and Hughes. Isogeometric analysis of structural

vibrations. Computer Methods in Applied Mechanics and Engineering,
195:5257–5296, 2006.

 Hughes, AR, and Sangalli. Duality and Unified Analysis of Discrete

Approximations in Structural Dynamics and Wave Propagation: Comparison
of p-method Finite Elements with k-method NURBS. Computer Methods in
Applied Mechanics and Engineering, 197:4104–4124, 2008.
 Wave propagation

Dispersion analysis
Focus on dispersion analysis,
i.e. analysis of the error for the
wave-number in (time-harmonic)
wave propagation problems
described by the Helmholtz
equation
Duality principle

Spectrum analysis is equivalent to

dispersion analysis in the regime
where kh is real

Linear case:

Resolution limit:
 Wave propagation

Stopping bands
(bands with complex discrete wave-number,
present before the resolution limit)

p-method (FEM, p=2): acoustical and optical

branches separated by a stopping band
In general (p > 2): p branches separated by
p-1 stopping bands

k-method (NURBS, p=2): one single branch

with no stopping bands
In general (p > 2): again, one single branch
with no stopping bands
 Wave propagation

Stopping bands
(bands with complex discrete wave-number,
present before the resolution limit)

Stopping bands

No stopping bands
 Wave propagation

1D (steady-state) problem: frequency

response spectra for p = 2

computed at x = L/10, L/2, and 9L/10

 Wave propagation

1D (steady-state) problem: frequency

response spectra for p = 3

computed at x = L/10, L/2, and 9L/10

 Wave propagation

1D (steady-state) problem: b.v.p. solution for p = 2

NOTE: k = 33 is within the p-method stopping band → unphysical attenuation

 Wave propagation

1D (steady-state) problem: b.v.p. solution for p = 3

NOTE: k = 31.5 is within the 1st p-method stopping band but no evident attenuation is observed
since the imaginary part is very small; k = 31.5 is however close to a resonance peak not well
reproduced by p-method
k = 71 is within the 2nd p-method stopping band → unphysical attenuation
 Wave propagation

Reference:

 Hughes, AR, and Sangalli. Duality and Unified Analysis of Discrete

Approximations in Structural Dynamics and Wave Propagation: Comparison
of p-method Finite Elements with k-method NURBS. Computer Methods in
Applied Mechanics and Engineering, 197:4104–4124 2008.
 Summary and future developments

Isogeometric analysis for structural vibrations and wave propagation:

main advantages with respect to p-FEM
  no optical branches
  better approximation properties on a per-d.o.f. basis
  exact geometry
main drawback with respect to p-FEM
  high number of (gaussian) quadrature points
solution: new quadrature rules taking into account inter-element regularity
 Hughes, AR, and Sangalli. Efficient Quadrature for NURBS-based
Isogeometric Analysis. Computer Methods in Applied Mechanics and
Engineering, doi:10.1016/j.cma.2008.12.004
main future development
  isogeometric collocation techniques for structural dynamics
(work in progress…)
[Preliminary results presented by Tom Hughes during the opening plenary
lecture of the COMPDYN 2009 Conference (Rodhes, June 2009)]
 Current isogeometric projects

  Isogeometric Analysis for Naval Ship Structures

Coordinator: Tom Hughes
Funded by: Office of Naval Research
Duration: January 2006 - December 2009

  GeoPDEs - Innovative compatible discretization techniques

for Partial Differential Equations
Coordinator: Annalisa Buffa
Funded by: FP7 Ideas ERC Starting Grant
Duration: July 2008 - June 2012
 Presentation outline

  Isogeometric Analysis
  Shape Memory Alloy Modeling and Application
  Other Interesting Research Works in Progress
•  Meshless Methods
•  Beam Finite Elements
  Conclusions
 Shape memory alloy modeling

SMAs: materials with an intrinsic capability of recovering their initial

“shape” even after undergoing severe deformations
From a macroscopic point of view they show the following two effects
not available, in general, in traditional materials:

Superelastic
Shape-memory
effect
effect
(mechanical
(thermal recovery)
recovery)

“Background” SMA models:

Souza et al. (1998);
Auricchio&Petrini (2004)
 Shape memory alloy modeling

Experimental tests show that, besides modeling superelastic and shape-

memory effects, other secondary effects should be taken into account. In
particular, in some cases it is important to:

 include and control cyclic

permanent inelasticity and
degradation
 include and control tension/
compression asymmetries
  in the yield stress level
  in loop length
  in loop width

 include a dependence of the

elastic properties on the phase
transformation level
 Shape memory alloy modeling
Therefore, enhanced SMA models able to include such secondary effects
have been designed, mathematically analyzed and numerically tested:

a)  Model including permanent inelasticity and degradation effects:

[based on a new (extra) tensorial variable describing permanent inelasticity]

References:
 Auricchio and AR. A Phenomenological One-dimensional Model Describing
Stress-induced Solid Phase Transformation with Permanent Inelasticity,
Mechanics of Advanced Materials and Structures, 14:43-55, 2007.
 Auricchio, AR, Stefanelli. A Three-dimensional Model Describing Stress-
induced Solid Phase Transformation with Permanent Inelasticity,
International Journal of Plasticity, 23:207-226, 2007.
 Shape memory alloy modeling
Therefore, enhanced SMA models able to include such secondary effects
have been designed, mathematically analyzed and numerically tested:

b)  Model including tension/compression asymmetry and elastic properties

depending on the phase transformation level:

Reference:
 Auricchio, AR, Stefanelli. A macroscopic 1D model for shape memory alloys
including asymmetric behaviors and transformation-dependent elastic
properties, Computer Methods in Applied Mechanics and Engineering,
23:207-226, 2009.
 SMA modeling validation

Collaboration on SMA modeling and validation with SAES Getters, an Italian

company recently become the world-leading SMA producer:
 SMA modeling validation

Collaboration on SMA modeling and validation with SAES Getters, an Italian

company recently become the world-leading SMA producer:
 SMA modeling validation

Within the collaboration with SAES Getters:

development of a reliable parameter identification
technique given two strain-temperature curves at
two different constant stresses (typical curves
characterizing SMAs for actuators)

Reference:
 Auricchio, Coda, AR, Urbano. SMA numerical modeling versus experimental
results: parameter identification and model prediction capabilities, Journal of
Materials Engineering and Performance, doi:10.1007s11665-009-9409-7,
2009.
 Design and FE simulation of biomedical devices

Many devices, in particular for biomedical applications (e.g., stents, spinal

spacers, micro-actuators, etc.), have been simulated using SMA constituive
models implemented within commercial FE codes (e.g., Abaqus).
Such a tool allows also the development of new design strategies.

References:
 Auricchio and AR, Shape Memory Alloys: material modeling and device finite
element simulations, Materials Science Forum, 583:257-275, 2009
+ references therein reported
 SMA applications in Earthquake Engineering

Our research group is active in SMA applications for Earthquake Engineering

[cf. also recent collaborations with the group of Prof. R. DesRoches]
Recently the focus is on SMA Isolation Devices [work developed within G.
Attanasi’s PhD thesis at ROSE School]
Comparison of a SMA isolation system with a Lead Rubber Bearing (LRB):

LRB – SMA system comparison:

  same secant stiffness to the
design displacement
(equivalence)
  different theoretical hysteretic
dissipation
 SMA applications in Earthquake Engineering
Summary of time-history analysis results:

 Superelastic hysteresis eliminates the residual displacements

 Overall flag-shape isolation system behavior close to elastoplastic response in:
controlling maximum resisting force; limiting maximum displacement demand;
dissipating the absorbed energy
 Results valid for many values of dissipation capability and rigid&flexible
superstructure condition
 SMA applications in Earthquake Engineering

References:
 Attanasi, Auricchio, Fenves. Feasibility Assessment of an Innovative Isolation
Bearing System with Shape Memory Alloys, Journal of Earthquake
Engineering, 13(s.i.1):18-39, 2009.
 Attanasi, Auricchio, Fenves. “Feasibility Investigation of Superelastic Effect
Devices for Seismic Isolation Applications Journal of Materials Engineering
and Performance, doi:10.1007/s11665-009-9372-3, 2009.

Work in progress:
Design and development of a real device
able to provide the horizontal displacement
– base shear relation like the one assumed:

Main issues are related to:

  provide enough displacement capacity
  validate the dynamic behavior of the material
  investigate on the multidirectional response of the device
 Current SMA projects

  SMARTeR: Shape Memory Alloys to Regulate Transient

Responses in civil engineering
Coordinator: Michel Fremond
Funded by: ESF Eurocores Programme
Duration: December 2006 - December 2009

  BioSMA - Mathematics for Shape Memory Technologies in

Biomechanics
Coordinator: Ulisse Stefanelli
Funded by: FP7 Ideas ERC Starting Grant
Duration: September 2008 - October 2013
 Presentation outline

  Isogeometric Analysis
  Shape Memory Alloy Modeling and Application
  Other Interesting Research Works in Progress
•  Meshless Methods
•  Beam Finite Elements
  Conclusions
 Work in progress: meshless methods

Besides the aforementioned isogeometric collocation techniques, we are

currently focusing on evolutions of Smoothed Particle Hydrodynamics (SPH).

Aim: reliable simulation of fast dynamics, impact problems and blasts.

Tool: particle methods (particularly suitable for explicit solvers)

But: classical SPH-based methods show many problems in approximating

Mechanics of Solids equations [lack of or low order h-convergence; bad or
tricky imposition of boundary conditions; etc.]

Idea: going back to the basis of SPH [on simple examples!] removing
unnecessary hypotheses and approximations [many!], focusing on obtaining
higher orders of convergence.
Reference:
 Asprone, Auricchio, Manfredi, Prota, AR, Sangalli. SPH methods for a 1D
elastic model problem: error analysis and development of a second-order
accurate formulation, submitted, 2009.
[also higher dimension static results are now available; next step: dynamics]
 Work in progress: beam finite elements

We also deal with the study and development of advanced finite element
beam formulations.

In this context, we recently devoted a good deal of attention to the study of

geometrically nonlinear beam elements.

Reference:
 Auricchio, Carotenuto, AR. On the geometrically exact beam model: a
consistent, effective and simple derivation from three-dimensional finite
elasticity. International Journal of Solids and Structures, 45:4766-4781,
2008.

Currently the focus is on the development of advanced theories for

homogeneous and multi-layered beams, for which some good preliminary
results have been obtained [work developed within G. Balduzzi’s research
period to be completed within his (future) PhD thesis].
 Presentation outline

  Isogeometric Analysis
  Shape Memory Alloy Modeling and Application
  Other Interesting Research Works in Progress
•  Meshless Methods
•  Beam Finite Elements
  Conclusions
 Conclusions

Idea: introducing some Computational Mechanics topics we

are currently working on, i.e.
  Isogeometric analysis
  Shape memory alloys
  Meshless methods
  Advanced beam models
Aim: collaborating with interested structural engineers!

Final message: before running an analysis, always keep in mind

the two Basic Principles of (good) Engineering [Prof. MP Collins]:
  You can’t push on a rope!
[never ask a model to do something that it simply can’t do]
  To find an answer, you must know the answer!
[never use numerical analysis as a “black-box” but, if you have
to, use your engineering judgment before relying on the results]
 Conclusions

Thank you all for your attention!

Thanks to Iunio and Paolo for their kind invitation

(despite I am a Computational Mechanics guy…)