Università degli studi di Pavia

Universitàt Politècnica de Catalunya

SEED Programme
Corso di Dottorato di Ricerca in
Ingegneria Civile e Architettura
 XXXI Ciclo 

Final Report
Computational Vademecums for Lattice Materials
using algebraic PGD

Supervisors: PhD Candidate:

Prof. Ferdinando Auricchio Alberto Pedro Sibileau
Prof. Pedro Díez UIN 440403

Co-advisors:
Dr. Simone Morganti
Dr. Alberto García González

Academic year 2018/2019

1 Introduction
The present report summarizes the research activities carried out during the PhD pro-
gramme from March 2016 to February 2019. This work presents lattice material structures
whose mechanical properties can be tailored in terms of their geometrical parameters values.
We address this multidimensional problem by means of an algebraic PGD solver, that allows
us to browse the parametric mechanical response in an ecient way.
The project has been nancially supported by the European Education, Audiovisual and
Culture Executive Agency (EACEA) under the Simulation in Engineering and Entrepreneur-
ship Development Joint Doctorate (SEED) (FPA 2013-0043). In addition, the PhD thesis
would not have been completed without the supervision of Prof. Ferdinando Auricchio
1,
2
Prof. Pedro Díez , Prof. Simone Morganti
3 and Prof. Alberto García-González 2 .

2 Research activities
This dissertation is motivated by the concept of materials by design [Reis et al., 2015].
Focusing on structures, this states that the properties in a mechanical component are not
only inherited by its constituent material, but also by the shape in which this one is dis-
tributed in space. Although this notion has been developed for thousands of years in archi-
tecture, its relevance at a smaller scale has not been conceived until the arrival of additive
manufacturing technologies [Bourell, 2016].
The materials by design approach is certainly multidisciplinary, from the study of the
shape, its representation in CAD, to the nal manufacturing by 3D printing. We here
assess the rst one, by means of numerical simulations. These play an important role, by
reinforcing our physical intuition in the resolution of the following question: which is the
material structure at a small scale (meso-structure) that features some prescribed properties
at a global scale (bulk material).
We introduce the material meso-structure as a lattice model with a parametric shape.
The mechanical properties arising at the global scale are recovered by solving an equilibrium
problem. Naturally, this one acquires the parametric nature of the lattice model. The main
diculty to handle the emerging mechanical properties of the bulk parametrically, is that the
computational complexity of numerical simulations increases exponentially with the number
of parameters.
To overcome this, we resort to the Proper Generalized Decomposition (PGD), which
provides explicit parametric solutions of our equilibrium problem [Chinesta et al., 2010]. As
a major contribution, we obtain the parametric solutions of the algebraic equations arising
from dierent lattice structures using the same PGD framework [Sibileau et al., 2018]. In this
sense, the algebraic PGD works as a non-intrusive solver, which is not limited to structural
problems but in general, any discrete form of a parametrized linear PDE.
The parametric mechanical properties of 2D and 3D lattice materials are explicitly rep-
resented by the PGD solutions or computational vademecums. In particular, we reproduce
the response of orthotropic Poisson's ratios in terms of the design parameters. Extreme neg-
ative values are identied, a characteristic that is relevant regarding the outperforming of
auxetic (or negative Poisson's ratios) properties compared to conventional materials [Evans
and Alderson, 2000]. Moreover, these computational vademecums could be further exploited
to tailor the material design through multi-objective and constraint optimizations, providing
an ecient tool to browse the parametric design space [Chinesta et al., 2013].
Finally, we extend our parametric analysis using geometrically nonlinear nite elements
to compute equilibrium, and the algebraic PGD a posteriori to interpolate their response
[Díez et al., 2018]. This is achieved with very good accuracy, for engineering purposes, at

1 Dipartimento di Ingegneria Civile e Architettura (DICAr), Università degli studi di Pavia, Pavia, Italy.
2 Laboratori de Càlcul Numèric (LaCàN), E.T.S. de Ingenieros de Caminos, Canales y Puertos, Universitat
Politècnica de Catalunya (UPC), Barcelona, Spain.
3 Dipartimento di Ingegneria Industriale e dell'Informazione, Università degli studi di Pavia, Pavia, Italy.
a considerably low number of modes. The nonlinear parametric framework surely broadens
the range of applications, and we highlight this in two distinctive situations. First, we
demonstrate its capability to describe the loading magnitude as an extra parameter to the
material properties behavior. Last but not least, we show its potential to perform buckling
analysis of lattice structures, as a function of the geometric parameters and the loading
magnitude itself.

3 Concluding remarks and future developments

The algebraic PGD provides an ecient solver to a variety of parametric structural prob-
lems, some of them involving more than 568 millions of dierent geometric congurations.
Moreover, its use is not limited to structural problems but in general, any discrete form
of a parametrized linear PDE. In this sense, it works as a non-intrusive solver for systems
of parametric algebraic linear equations. However, this solver requires that the standard
nite elements equations must be built taking care of their parametric dependence in a sep-
arated manner. We believe that this part can be done systematically, following a proposed
framework with familiar procedures for nite element programmers. In addition, this would
certainly ease the path for further research on algebraic PGD.
In the cases where an analytical solution in the parameters is constructed, by means of
symbolic computations, the algebraic PGD solver accuracy is analyzed in a global parametric
norm. In all cases, we observe that every PGD mode contributes to reduce the global error,
whose maximum value results below 0.2% (average number of modes: 35). However, further
research eorts could be addressed to estimate the residual error or some quantity of interest,
which would provide a precise criterion to stop the sequential computation of modes.
The PGD vademecum of mechanical properties shows extreme values of the Poisson's
ratios in orthotropic materials. In these cases, the material actually works as a mechanical
amplier, in the transverse directions, to the applied strain. In addition, the PGD vademe-
cum presents a groundwork on top of which multiobjective and constrained optimizations
could be further exploited for the design of tailored materials. In this context, the equi-
librium solutions and its derivatives with respect to the parameters (sensitivities) could be
supplied as a trivial post-process, for the whole space of design variables.
An explicit parametric response of a stent-like structure with geometrically nonlinear
behavior is built using a least-square PGD interpolation of the nite element solutions (4550
dierent congurations). The mechanical properties and buckling eects of this device are
presented in a vademecum which is a function, not only of the geometrical parameters, but
also the loading magnitude. The PGD interpolation retrieves very good accuracy using a low
number of modes, which is promising to believe that, based on the algebraic PGD solver, an
iterative scheme could successfully build the approximation to nonlinear parametric systems
of equations on the y, and at a reduced cost. The feasibility of this concept is subjected
mainly to the ane decomposition of the residual (which now depends on the parametric
solution itself ). In addition, the separable dependence on the parameters should also be
extended to the tangent stiness matrix if a full Newton-Raphson method is sought.
Finally, we recall that in structural analysis, a standard procedure to asses buckling
without the need of nonlinear models, implies the solution of an eigenvalue problem. For this,
the structure stiness matrix is augmented with a so-called stress stiness matrix, which is
a function of the initial conguration and the applied loads. The resulting eigenvalues dene
magnication factors at which the applied loads reach a critical state (with their eigenvectors
characterizing the shape of the buckled structure). In our framework, developing an ecient
solution of the parametric eigenvalue problems would result in a highly valuable tool for the
design of lattice structures and materials.
4 Research dissemination
• A. Sibileau, A. García, F. Auricchio, S. Morganti and P. Díez. Explicit para-
metric solutions of lattice structures with Proper Generalized Decomposition (PGD).
Computational Mechanics, 2018. https://doi.org/10.1007/s00466-017-1534-9.
• Present your thesis in 4 minutes - 1
st Edition organized by Fundació Catalana
de Recerca i Innovació (FCRI). Barcelona, May 10, 2018. PGD for architectured
materials, nalist of the Universitat Politècnica de Catalunya selection stage.

• NMASE 2018 - 12
th workshop on Numerical Methods in Applied Sciences and En-
gineering. Castelldefels (Barcelona), January 24-25, 2018. Invited talk: A PGD
solver of parametric linear system of equations (invited by P. Díez).

• CoMe seminar series - LaCàN, Universitat Politècnica de Catalunya. Barcelona,

October 11, 2017. Invited seminar: Explicit parametric solutions of truss structures
using PGD. Applications to the design of architectured materials (invited by P. Díez).

5 Education activities
• Nonlinear Continuum Mechanics, March 1-May 31, 2016. Alessandro Reali, Di-
partimento di Ingegneria Civile e Architettura (DICAr), Università degli studi di
Pavia.

• Nonlinear Computational Solid & Structural Mechanics, May 16-20, 2016.

F. Brezzi, R.L. Taylor, F. Auricchio, M. Bischo, A. Reali and G. Sangalli, Palazzo
Vistarino, Pavia.

• CISM-ECCOMAS Summer School on Computational FSI, June 27-July 1,

2016. W.A. Wall, C. Farhat, R. Ohayon, S. Badia and J.F. Gerbeau, Palazzo del
Torso, Udine.

• Summer School on Discontinuous Galerkin Methods, July 11-14, 2017. B.

Cockburn, J. Peraire, A. Huerta, S. Fernández-Méndez, R. Sevilla, X. Roca and M.
Giacomini, E.T.S. de Ingenieros de Caminos, Canales y Puertos (UPC), Barcelona.

• Advanced Numerical Methods, September 19-21, 2018. F. Brezzi, C. Lovadina,

F. Auricchio, A. Reali, G. Sangalli, A. Russo, G. Vacca, F. Dassi, G. Lorenzo, G.
Scalet and M. Tani, Palazzo Vistarino, Pavia.

6 Research stay abroad

March 2017 - February 2019: mobility period at Universitat Politècnica de Catalunya
(Barcelona), according to the European Joint Doctorate Programme SEED, under the su-
pervision of Prof. Pedro Díez.
7 Seminars
• An approach for feasible Uncertainty Quantication of complex real world
problems in Biomechanics. Prof. W.A. Wall (TUM University, Germany). March
16, 2016.

• Prediction of Blood Damage Induced by Articial Rotary Pumps. Prof.

M. Behr (RWTH Aachen University, Germany). May 23, 2016.

• Handling Deforming Fluid Domains with Spline-Based Finite Element

Methods. Dr. S. Elgeti (RWTH Aachen University, Germany). May 23, 2016.

• Modeling of Fiber-reinforced Solids with Application to Soft Tissues. Prof.

G. Holzapfel (Graz University of Technology, Austria). September 6, 2016.

• InSilicoTrials.com Project. Dr. L. Emili (Promeditec Milano, Italy). October

19, 2016.

• Una visione olistica per l'Additive Manufacturing - Verso la Digital En-

terprise. Ing. E. Bergamaschi (Siemens Milano, Italy). November 7, 2016.

• Flexible FSI Framework with Applications. Prof. Y. Bazilevs (UC San Diego,
USA). December 2, 2016.

• Isogeometric collocation methods for Cosserat rods and their application

in design optimization for additive manufacturing. Dr. O. Weeger (Singapore
University of Technology and Design). December 20, 2016.

• Uncommon knowledge about Left Ventricular Assist Devices (LVADs).

Dr. S. Vandenberghe (Cardiocentro Ticino, Switzerland). January 16, 2017.

• Isogeometric phase-eld modeling of brittle fracture in thin plates and

shells. Prof. J. Kiendl (Norwegian University of Science and Technology). January
20, 2017.

• A multiphase porous media model for transport oncophysics. Prof. B.A.

Schreer (Università di Padova). February 1, 2017.

• The inuence of cell morphology on bulk acoustic wave propagation in

periodic composite materials. Prof. L. Gambarotta (Università di Genova).
February 1, 2017.

• Earth science meets technology: from paleo-subduction to material sci-

ences. Dr. M. Alvaro (Università degli studi di Pavia). February 10, 2017.

• eXtended Hybridizable Discontinuous Galerkin (X-HDG). Prof. S. Fernández-

Méndez (Universitat Politècnica de Catalunya, Barcelona). March 29, 2017.

• Random Sketching for Model Order Reduction. MSc. O. Balabanov (Uni-

versitat Politècnica de Catalunya, Barcelona). June 9, 2017.

• A Weakly Imposed Navier Slip Boundary Condition for the Stabilized

CutFEM by means of the Nitsche's Method. MSc. M. Winter (Universitat
Politècnica de Catalunya, Barcelona). June 9, 2017.

• Computational model for the chemo-mechanical mechanism in wound heal-

ing. MSc. L. Roldán (Universitat Politècnica de Catalunya, Barcelona). June 30,
2017.

• High order nite element method for Stefan problem. MSc. J. Barceló
(Universitat Politècnica de Catalunya, Barcelona). June 30, 2017.
• Numerical Modelling of the Electrophysiology of the Neuron. MSc. V.
Lang (Universitat Politècnica de Catalunya, Barcelona). June 30, 2017.

• Modeling of cells and tissues as active uid. MSc. S.C. Divi (Universitat
Politècnica de Catalunya, Barcelona). July 7, 2017.

• Evaluation of properties of a exoelectric material using nanoindentation

technique. MSc. S. Shaikh (Universitat Politècnica de Catalunya, Barcelona). July
7, 2017.

• NURBS-Enhanced Finite Element Hybridizable Discontinuous Galerkin

method with degree adaptivity for steady Stokes ow. MSc. K. Suresh
(Universitat Politècnica de Catalunya, Barcelona). July 7, 2017.

• Adsorption of curvature inducing proteins on biological membranes. Dr.

N. Walani (Universitat Politècnica de Catalunya, Barcelona). September 29, 2017.

• Stochastic analysis of uid-driven fracture. MSc. H. Garikapati (Universitat

Politècnica de Catalunya, Barcelona). October 11, 2017.

• Cellular superelasticity in three-dimensional epithelial sheets of controlled

size and shape. Dr. S. Kale (Universitat Politècnica de Catalunya, Barcelona).
October 27, 2017.

• Numerical and experimental models for research in cell and tissue mechanobi-
ology. Prof. García-Aznar (University of Zaragoza). November 15, 2017.

• Advanced numerical techniques for inverse problems in geophysics. MSc.

O. Ortega (Universitat Politècnica de Catalunya, Barcelona). December 13, 2017.

• Application of Polyconvexity and multivariable convexity of energy po-

tentials in nonlinear solid mechanics. Prof. J. Bonet (University of Greenwich).
March 2, 2018.

• A new face-centered nite volume paradigm. Dr. M. Giacomini (Universitat

Politècnica de Catalunya, Barcelona). April 25, 2018.

• Modeling and simulation of the dynamics of soft adhesion mediated by mo-

bile binders: 2D variational modeling. MSc. D. Kaurin (Universitat Politècnica
de Catalunya, Barcelona). May 9, 2018.

• A general framework for the three-dimensional simulation of uid sur-

faces. Dr. A. Torres-Sánchez (Universitat Politècnica de Catalunya, Barcelona).
May 16, 2018.

• A global view on Riemann solvers. MSc. J. Vila (Universitat Politècnica de

Catalunya, Barcelona). May 23, 2018.

• Towards a Reduced Order Modelling approach for coupled acousto-magneto-

mechanical problems with application to MRI scanners. MSc. G. Barroso
(Universitat Politècnica de Catalunya, Barcelona). June 6, 2018.

• A non-intrusive approach to the Proper Generalized Decomposition for

ow problems in OpenFOAM. Eng. V. Tsiolakis (Universitat Politècnica de
Catalunya, Barcelona). June 22, 2018.

• Reshaping of large aeronautical structural parts: a reduced order model

approach. MSc. R. Mena (Universitat Politècnica de Catalunya, Barcelona). June
27, 2018.
 • Ambrosio-Tortorelli approximations for crack propagation and image seg-
mentation modeling with anisotropic mesh adaptation. Prof. S. Perotto
(Politecnico di Milano). September 12, 2018.

• Non intrusive Stochastic Finite Element Method for crash-worthiness with

VPS/Pamcrash. MSc. M. Rocas (Universitat Politècnica de Catalunya, Barcelona).
November 14, 2018.

• Reduced Order Models for THM Coupled System in Geological Porous

Media. MSc. Y. Larion (Universitat Politècnica de Catalunya, Barcelona). Novem-
ber 30, 2018.

• An hp-nite element formulation for the simulation of magneto-mechanical

coupling in MRI scanners. MSc. M. Seoane (Universitat Politècnica de Catalunya,
Barcelona). December 19, 2018.

• Dimensionality Reduction and Manifold Learning - Towards Interpretable

Image Analysis. Dr. G. Piella (Universitat Pompeu Fabra). February 1, 2019.

References
[Bourell, 2016] Bourell, D. L. (2016). Perspectives on additive manufacturing. Annual
Review of Materials Research, 46:118.
[Chinesta et al., 2010] Chinesta, F., Ammar, A., and Cueto, E. (2010). Recent advances
and new challenges in the use of the proper generalized decomposition for solving multi-
dimensional models. Archives of Computational methods in Engineering, 17(4):327350.
[Chinesta et al., 2013] Chinesta, F., Leygue, A., Bordeu, F., Aguado, J. V., Cueto, E.,
González, D., Alfaro, I., Ammar, A., and Huerta, A. (2013). Pgd-based computational
vademecum for ecient design, optimization and control. Archives of Computational
Methods in Engineering, 20(1):3159.
[Díez et al., 2018] Díez, P., Zlotnik, S., García-González, A., and Huerta, A. (2018). Al-
gebraic pgd for tensor separation and compression: An algorithmic approach. Comptes
Rendus Mécanique, 346(7):501514.
[Evans and Alderson, 2000] Evans, K. E. and Alderson, A. (2000). Auxetic materials: func-
tional materials and structures from lateral thinking! Advanced materials, 12(9):617628.
[Reis et al., 2015] Reis, P. M., Jaeger, H. M., and van Hecke, M. (2015). Designer Matter:
A perspective. Extreme Mechanics Letters, 5:2529.
[Sibileau et al., 2018] Sibileau, A., García-González, A., Auricchio, F., Morganti, S., and
Díez, P. (2018). Explicit parametric solutions of lattice structures with proper generalized
decomposition (pgd). Computational Mechanics, pages 121.