Access the full text.
Sign up today, get DeepDyve free for 14 days.
S Polzer, TC Gasser (2015)
Biomechanical rupture risk assessment of abdominal aortic aneurysms based on a novel probabilistic rupture risk indexJournal of The Royal Society Interface., 12
S De Putter, B Wolters, M Rutten, M Breeuwer, F Gerritsen, F Van de Vosse (2007)
Patient-specific initial wall stress in abdominal aortic aneurysms with a backward incremental methodJournal of biomechanics., 40
LC Lee, L Ge, Z Zhang, M Pease, SD Nikolic, R Mishra (2014)
Patient-specific finite element modeling of the Cardiokinetix Parachute® device: effects on left ventricular wall stress and functionMedical & biological engineering & computing., 52
L Bruder, J Pelisek, HH Eckstein, MW Gee (2020)
Biomechanical rupture risk assessment of abdominal aortic aneurysms using clinical data: A patient-specific, probabilistic framework and comparative case-control studyPloS one., 15
C Reeps, A Maier, J Pelisek, F Härtl, V Grabher-Meier, W Wall (2013)
Measuring and modeling patient-specific distributions of material properties in abdominal aortic aneurysm wallBiomechanics and modeling in mechanobiology., 12
C Farhat, P Avery, T Chapman, J Cortial (2014)
Dimensional reduction of nonlinear finite element dynamic models with finite rotations and energy-based mesh sampling and weighting for computational efficiencyInternational Journal for Numerical Methods in Engineering., 98
J Biehler, S Kehl, MW Gee, F Schmies, J Pelisek, A Maier (2017)
Probabilistic noninvasive prediction of wall properties of abdominal aortic aneurysms using Bayesian regressionBiomechanics and modeling in mechanobiology., 16
A Hemmler, B Lutz, C Reeps, G Kalender, MW Gee (2018)
A methodology for in silico endovascular repair of abdominal aortic aneurysmsBiomechanics and modeling in mechanobiology., 17
S Ockert, D Boeckler, J Allenberg, H Schumacher (2007)
Rupturiertes abdominelles aortenaneurysmaGefaesschirurgie., 12
ME Johnson, LM Moore, D Ylvisaker (1990)
Minimax and maximin distance designsJournal of statistical planning and inference., 26
D Perrin, P Badel, L Orgeas, C Geindreau, S rolland du Roscoat, JN Albertini (2016)
Patient-specific simulation of endovascular repair surgery with tortuous aneurysms requiring flexible stent-graftsJournal of the mechanical behavior of biomedical materials., 63
S Sankaran, AL Marsden (2011)
A stochastic collocation method for uncertainty quantification and propagation in cardiovascular simulationsJournal of biomechanical engineering., 133
J Lu, X Zhou, ML Raghavan (2007)
Inverse elastostatic stress analysis in pre-deformed biological structures: demonstration using abdominal aortic aneurysmsJournal of biomechanics., 40
GH Chang, CM Schirmer, Y Modarres-Sadeghi (2017)
A reduced-order model for wall shear stress in abdominal aortic aneurysms by proper orthogonal decompositionJournal of biomechanics., 54
MW Gee, C Förster, W Wall (2010)
A computational strategy for prestressing patient-specific biomechanical problems under finite deformationInternational Journal for Numerical Methods in Biomedical Engineering., 26
B Haasdonk, M Ohlberger (2008)
Reduced basis method for finite volume approximations of parametrized linear evolution equationsESAIM: Mathematical Modelling and Numerical Analysis., 42
Neto E de Souza, D Perić, M Dutko, D Owen (1996)
Design of simple low order finite elements for large strain analysis of nearly incompressible solidsInternational Journal of Solids and Structures., 33
P Virtanen, R Gommers, TE Oliphant (2020)
SciPy 1.0: Fundamental Algorithms for Scientific Computing in PythonNature Methods., 17
TA Davis (2004)
Algorithm 832: UMFPACK V4. 3–an unsymmetric-pattern multifrontal methodACM Transactions on Mathematical Software (TOMS), 30
M Gee, C Reeps, H Eckstein, W Wall (2009)
Prestressing in finite deformation abdominal aortic aneurysm simulationJournal of biomechanics., 42
T Lieu, C Farhat, M Lesoinne (2006)
Reduced-order fluid/structure modeling of a complete aircraft configurationComputer methods in applied mechanics and engineering., 195
S Kalra, M Nabi (2019)
TPWL Simulation of Large Nonlinear Circuits Using Subspace Angle Based Adaptive SamplingIEEE Transactions on Circuits and Systems II: Express Briefs., 67
C Farhat, T Chapman, P Avery (2015)
Stability and accuracy properties of the energy-conserving sampling and weighting (ECSW) method for the hyper reduction of nonlinear finite element dynamic modelsInt J Numer Methods Eng., 102
DA Vorp (2007)
Biomechanics of abdominal aortic aneurysmJournal of biomechanics., 40
H Liu, JR Hervas, YS Ong, J Cai, Y Wang (2018)
An adaptive RBF-HDMR modeling approach under limited computational budgetStructural and Multidisciplinary Optimization., 57
JS Hesthaven, B Stamm, S Zhang (2014)
Efficient greedy algorithms for high-dimensional parameter spaces with applications to empirical interpolation and reduced basis methodsESAIM: Mathematical Modelling and Numerical Analysis., 48
J Biehler, MW Gee, WA Wall (2015)
Towards efficient uncertainty quantification in complex and large-scale biomechanical problems based on a Bayesian multi-fidelity schemeBiomechanics and modeling in mechanobiology., 14
K Carlberg, R Tuminaro, P Boggs (2015)
Preserving Lagrangian structure in nonlinear model reduction with application to structural dynamicsSIAM Journal on Scientific Computing., 37
Y Maday, B Stamm (2013)
Locally adaptive greedy approximations for anisotropic parameter reduced basis spacesSIAM Journal on Scientific Computing., 35
J Biehler, W Wall (2018)
The impact of personalized probabilistic wall thickness models on peak wall stress in abdominal aortic aneurysmsInternational journal for numerical methods in biomedical engineering., 34
L Pronzato (2017)
Minimax and maximin space-filling designs: some properties and methods for constructionJournal de la Societe Francaise de Statistique., 158
S Quicken, WP Donders, EM van Disseldorp, K Gashi, BM Mees, FN van de Vosse (2016)
Application of an adaptive polynomial chaos expansion on computationally expensive three-dimensional cardiovascular models for uncertainty quantification and sensitivity analysisJournal of biomechanical engineering., 138
JA Niestrawska, C Viertler, P Regitnig, TU Cohnert, G Sommer, GA Holzapfel (2016)
Microstructure and mechanics of healthy and aneurysmatic abdominal aortas: experimental analysis and modellingJournal of The Royal Society Interface., 13
M Raghavan, DA Vorp (2000)
Toward a biomechanical tool to evaluate rupture potential of abdominal aortic aneurysm: identification of a finite strain constitutive model and evaluation of its applicabilityJournal of biomechanics., 33
SS Garud, IA Karimi, M Kraft (2017)
Design of computer experiments: A reviewComputers & Chemical Engineering., 106
A Quarteroni, G Rozza, A Manzoni (2011)
Certified reduced basis approximation for parametrized partial differential equations and applicationsJournal of Mathematics in Industry., 1
ML Raghavan, MM Hanaoka, JA Kratzberg, Higuchi M de Lourdes, ES Da Silva (2011)
Biomechanical failure properties and microstructural content of ruptured and unruptured abdominal aortic aneurysmsJournal of biomechanics., 44
K Crombecq, E Laermans, T Dhaene (2011)
Efficient space-filling and non-collapsing sequential design strategies for simulation-based modelingEuropean Journal of Operational Research., 214
J Jiang, Y Chen, A Narayan (2017)
Offline-enhanced reduced basis method through adaptive construction of the surrogate training setJournal of Scientific Computing., 73
T Akman (2015)
Local improvements to reduced-order approximations of optimal control problems governed by diffusion-convection-reaction equationComputers & Mathematics with Applications., 70
S Doll, K Schweizerhof (2000)
On the development of volumetric strain energy functionsJ Appl Mech., 67
TC Gasser, G Görgülü, M Folkesson, J Swedenborg (2008)
Failure properties of intraluminal thrombus in abdominal aortic aneurysm under static and pulsating mechanical loadsJournal of vascular surgery., 48
G Martufi, M Lindquist Liljeqvist, N Sakalihasan, G Panuccio, R Hultgren, J Roy (2016)
Local diameter, wall stress, and thrombus thickness influence the local growth of abdominal aortic aneurysmsJournal of Endovascular Therapy., 23
K Carlberg, M Barone, H Antil (2017)
Galerkin v. least-squares Petrov-Galerkin projection in nonlinear model reductionJournal of Computational Physics., 330
A Hemmler, B Lutz, G Kalender, C Reeps, MW Gee (2019)
Patient-specific in silico endovascular repair of abdominal aortic aneurysms: application and validationBiomechanics and modeling in mechanobiology., 18
T Lieu, C Farhat (2007)
Adaptation of aeroelastic reduced-order models and application to an F-16 configurationAIAA journal., 45
gee@tum.de Mechanics & High Performance This work proposes a framework for projection-based model order reduction (MOR) of Computing Group, Technical computational models aiming at a mechanical analysis of abdominal aortic aneurysms University of Munich, Parkring 35, 85748 Garching b. München, (AAAs). The underlying full-order model (FOM) is patient-specific, stationary and Germany nonlinear. The quantities of interest are the von Mises stress and the von Mises strain field in the AAA wall, which result from loading the structure to the level of diastolic blood pressure at a fixed, imaged geometry (prestressing stage) and subsequent loading to the level of systolic blood pressure with associated deformation of the structure (deformation stage). Prestressing is performed with the modified updated Lagrangian formulation (MULF) approach. The proposed framework aims at a reduction of the computational cost in a many-query context resulting from model uncertainties in two material and one geometric parameter. We apply projection-based MOR to the MULF prestressing stage, which has not been presented to date. Additionally, we propose a reduced-order basis construction technique combining the concept of subspace angles and greedy maximin distance sampling. To further achieve computational speedup, the reduced-order model (ROM) is equipped with the energy-conserving mesh sampling and weighting hyper reduction method. Accuracy of the ROM is numerically tested in terms of the quantities of interest within given bounds of the parameter domain and performance of the proposed ROM in the many-query context is demonstrated by comparing ROM and FOM statistics built from Monte Carlo sampling for three different patient-specific AAAs. Keywords: Abdominal aortic aneurysm, Nonlinear model order reduction, Prestressing, Finite element method Introduction The potential of computational analysis to support clinical decision making is of great value for both physicians and patients. In particular the possibility to gain spatially and temporally resolved information on the patient-specific pathology at minimal intervention with the patient’s body is driving this field of research. The human cardiovascular system is a specific example for the application of computa- tional models [1] for risk assessment [2,3], planing of medical intervention and assessment © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. 0123456789().,–: volV Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 2 of 31 of its effect [4–6] or general understanding of disease progression. More specifically, the pathology under consideration in this work is the abdominal aortic aneurysm (AAA). An AAA corresponds to a dilatation of the aorta, which shows degraded mechanical properties in the widened segment [7] and is prone to rupture with highly-probable lethal outcome [8]. Given that aortic wall degradation and rupture is related to material fail- ure, mechanical analysis of AAAs has been used for understanding and quantifying the development and progression of the disease [3,9–11]. The AAA finite element models in this work are patient-specific, large-scale, stationary as well as materially and geometrically nonlinear. AAA geometries are extracted from medical screening images following the protocol in [12]. Given that imaged geometries are under blood pressure, an accurate computational model needs to impose a meaningful stress state, keeping the imaged configuration fixed. This is achieved in a modified updated Lagrangian formulation (MULF) [13,14] prestressing stage, wherein a physiological stress state is imprinted for an assumed diastolic blood pressure load. The vessel is subsequently deformed under further loading up to an assumed systolic blood pressure. A common factor in most if not all works related to accurate state-of-the-art com- putational analysis of AAAs is a lack of knowledge on essential parameters related to mathematical modeling. This lack of knowledge results from the high inter- and intra- patient variety of AAA properties [15,16] and the limited accessibility to patient-specific data, given that the object of interest is located within the human body. From a computa- tional perspective, this lack of knowledge typically results in the application of statistical methods, which attempt to propagate uncertainty through the computational model and involve sampling. Since the computational full-order models (FOMs) under consideration are nonlinear and large-scale, sampling with a high number of model evaluations quickly becomes too expensive to be practical in terms of computing power. A well known approach to overcome the burden of impracticable requirements on com- puting power is projection-based model order reduction (MOR), which typically includes the following steps [17]. In a computationally expensive offline stage, the FOM is evaluated and a low-dimensional subspace is extracted from resulting solution snapshots in terms of the column span of an orthogonal matrix (the so-called reduced-order basis (ROB)). The ROB in turn is used to diminish the number of model degrees of freedom (DOFs) (also referred to as dimension or order in the current context). If constructed accurately, the reduced-order model (ROM) can replace the FOM in the given context of application. The objective of the current work is to: 1. present a framework for the construction of a dimensionally reduced model (DROM) as well as a both dimensionally reduced and hyper reduced model (DHROM) for prestressed AAAs applying the Galerkin projection [18] for dimen- sional reduction and the energy-conserving mesh sampling and weighting (ECSW) method [19,20] for hyper reduction. The AAA models are parametrized in two material (low-strain range and high-strain range stiffness) and one geometric (AAA wall thickness) parameter. 2. demonstrate the applicability of both ROMs for assessment of the von Mises stress and the von Mises strain field in the aortic wall within bounds for the model parametrization. These quantities of interest are relevant in AAA rupture strati- Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 3 of 31 fication and are therefore of essential importance for the progression of the disease [3,9,10]. 3. demonstrate the robustness of the presented framework by investigating three patient-specific computational examples which differ in geometry, parameter domain bounds as well as the number of DOFs. Several techniques in the realm of the mechanical analysis of aneurysms have been proposed in the past to overcome the burden of limiting computing power. One example is the application of a computationally cheap intermediate mapping, which is utilized for sampling in place of the FOM. An example can be found in [3], wherein the authors fitted an inverse power-law function to represent the relation between aneurysm wall thickness and peak wall stress. In [21] a polynomial chaos expansion is built in order to investigate aneurysm wall stress assuming uncertainty in two material parameters, the wall thickness as well as the arterial pressure. A stochastic collocation method can be found in [22], wherein the authors interpolate the Navier-Stokes flow solution in order to evaluate the mean shear stress over the vessel wall. Alternatively, the application of a cheap and possibly inaccurate model in terms of a multi-fidelity approach is presented in [10,23]. Therein, the cheap model is not supposed to replace the high-fidelity model, instead it rather serves as a means to decrease the number of high-fidelity model evaluations by providing additional information. A similar stochastic structure of the high-fidelity and the low-fidelity model is a prerequisite. Also the applicability of projection-based MOR for computational feasibility of large- scale aneurysm models has been demonstrated in the past. In [24], the authors address variable inflow angles and build a ROM for AAA hemodynamics. In [25], a ROM for the prediction of periodic regime hemodynamics of a cerebral aneurysm is derived. We motivate the application of projection-based MOR for the following reasons. A surrogate model constructed by projection-based MOR will recover the FOM, if the degree of reduction is reversed. In this sense, projection-based MOR is consistent with FOM physics and contrasts the idea of an intermediate mapping as described above, given that such a mapping only exploits local FOM physics by sampling. The mentioned multi-fidelity approach incorporates the contribution of inaccurate information to specific quantities of interest. As opposed to projection-based MOR, a surrogate model producing high- dimensional information and being able to serve as inexpensive FOM replacement is not created. To the authors knowledge, no parametrized projection-based ROM has been presented for prestressed, large-scale, patient-specific and nonlinear solid mechanics AAA models to date. In particular, application of projection-based MOR to a MULF prestressing stage is a challenging task, which is investigated in this work. This involves a mathematical refor- mulation of our MULF prestressing stage, given that the original formulation accumulates an imprinted deformation gradient instead of computing displacement modes and there- fore is not suitable for snapshot collection. Additionally, a sampling strategy combining greedy maximin distance sampling on parameter space subdomains and the concept of subspace angles is presented for snapshot collection and subsequent construction of the ROB. The remainder of this paper is organized as follows. We first introduce the patient- specific AAA computational model. Special interest in view of projection-based MOR is Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 4 of 31 devoted to the prestressing stage. Next, we present the DROM as well as the DHROM, assuming a given ROB and ECSW displacement modes and continue by describing the approach for construction of the ROB and ECSW displacement modes. For this purpose, a greedy maximin distance sampling approach and a stopping criterion based on subspace angles is applied. Finally, we present numerical experiments on three patient-specific AAA models and conclude. Methods The framework presented in this work includes a large-scale finite element model for AAA simulation, a projection-based MOR process and a sampling strategy for the construction of the ROM. These building blocks are described in the following. Computational modeling of abdominal aortic aneurysms In this section, we introduce the computational model in terms of its governing equations. We differentiate between the prestressing stage and the deformation stage, which, when combined, yield a mechanical state of the aortic segment under systolic blood pressure. Particular focus is placed on the prestressing stage, given that special treatment is required for the purpose of snapshot collection. Patient-specific computational model Our computational model consists of an aortic segment, which fully includes the AAA as well as short segments of the iliac arteries, see [12] for a detailed description of the work- flow from imaging to finite element simulation. The aortic vessel is treated as an elastic solid consisting of an intraluminal thrombus (ILT) and the aortic wall. Pressure is exerted on the luminal (i.e. inner) surface of the ILT and the aneurysm is loaded to an assumed systolic blood pressure, which is the mechanical state of interest. The proximal and distal end surfaces of the model are constrained by a zero-displacement Dirichlet condition for vessel fixation. Figure 1 exhibits an example of a patient-specific computational domain. Model equations The governing equations read ∇· P = 0 in (1) P · N = T on (2) p,0 u = 0 on (3) with −T T = T(u,p) =−pJ(u)F (u) · N . (4) The weak form of the governing equations is given by the principle of virtual work (PVW) δW = δW − δW = P : ∇δu dV − T · δu dA = 0 ∀δu. int ext (5) 0 p,0 δW, δW and δW denote the total, internal and external virtual work, P denotes the int ext first Piola-Kirchhoff stress tensor, N is the outward normal vector in the reference config- uration and denotes the reference configuration pressure load surface (i.e. the luminal p,0 Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 5 of 31 Fig. 1 Example of a patient-specific computational domain (left) and cut through the computational domain (right) exposing the separation of into the vessel wall and the intraluminal thrombus (ILT). 0 p denotes the pressure load boundary condition surface, denotes Dirichlet boundary condition surfaces ILT surface). We emphasize that the traction boundary condition T depends on the dis- ∂u placement field u,see Eq.(4). Therein F(u) = I + is the deformation gradient with ∂X respect to the reference configuration, X ∈ denotes reference configuration material coordinates, J(u) is the deformation gradient determinant and p is the pressure. We make use of hyperelastic constitutive relations P = (6) ∂F introducing the strain-energy function and apply an isochoric-volumetric split for ILT as well as the vessel wall strain-energy ILT ILT ¯ ¯ ¯ ¯ (I , I ,J) = (I , I ) + (J), (7) ILT 1 2 1 2 iso vol wall wall ¯ ¯ (I ,J) = (I ) + (J), (8) 1 1 wall iso vol wherein ¯ ¯ I = tr(C), (9) ¯ ¯ ¯ I = (tr(C) − tr(C )) (10) are the first and second principal invariant of the modified right Cauchy Green tensor T − C = F F with F = J F. In more detail, we model the isochoric strain-energy iso iso iso contribution of the ILT as given in [12,26] ILT 2 ¯ ¯ ¯ ¯ (I , I ) = c(I − 2I − 3) (11) 1 2 2 iso 1 and the isochoric strain-energy contribution of the vessel wall as given in [9,12] wall 2 ¯ ¯ ¯ (I ) = α(I − 3) + β(I − 3) . (12) 1 1 1 iso Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 6 of 31 The parameter c is a stiffness parameter of the ILT, while α (referred to as α-stiffness in the following) and β (referred to as β-stiffness in the following) can be interpreted as low- strain range and high-strain range stiffness of the vessel wall, respectively. The volumetric wall ILT parts , of the strain-energies are chosen as given in [12,27] vol vol x 2 (J) = (J − 2ln(J) − 1), (13) vol wall ILT with x ∈{ILT, wall} and κ , κ being sufficiently large to reflect almost incompressible material behavior. MULF prestressing AAA geometries obtained from computed tomography imaging are exerted to blood pressure. From a continuum mechanics perspective, this corresponds to a non stress- free reference configuration [14,28,29]. Our simulations are therefore divided in two stages: The prestressing stage, which aims at imprinting a physiological stress-state into the imaged (i.e. fixed) geometric configuration at assumed diastolic blood pressure, is performed first. At second, the vessel is loaded to an assumed systolic blood pressure at evolving geometry in the deformation stage. We apply the Modified Updated Lagrangian Formulation (MULF) [14] prestressing approach in the first stage. MULF is an efficient prestressing method which especially was validated for the simulation of AAAs [10,12,13,30]. In the MULF prestressing approach an imprinted prestress deformation gradient F is built up incrementally with boundary conditions evaluated at the imaged configuration. Snapshot collection as required for data-driven construction of a ROB (cf. section “Con- struction of reduced-order model components”) is not possible for MULF prestressing, given that displacement modes are not generated. To overcome this problem, we present a reformulation of MULF prestressing, shifting the wanted quantity from the prestress deformation gradient F to a virtual prestress displacement field u . p p For consistency, we briefly review the original MULF prestressing formulation from a continuum mechanics perspective (details on implementation in the realm of the finite element method can be found in [14]) and state the mentioned reformulation in direct comparison with the original. As starting point we recall the following kinematic relations. Given a virtual displace- ˜ ˜ ment field u˜, from a virtual configuration X to the current configuration x,a ∂x displacement field u from X to , a deformation gradient F = and a virtual ∂X ∂X deformation gradient F = , we state ∂X x = X + u˜ = X + u, (14) ∂u ∂(X + u) ∂(X + u˜) F = I + = = ∂X ∂X ∂X ˜ ˜ ∂(X + u˜) ∂X ∂u˜ = = I + · F. (15) ˜ ˜ ∂X ∂X ∂X As a result, the identical first Piola-Kirchhoff stress field P can be expressed as (F), (16) P = P P = P (u), (17) u Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 7 of 31 P = P (˜u, F), (18) u,F defining P : F → (F), (19) ∂F ∂ ∂u P : u → I + , (20) ∂F ∂X ∂ ∂u˜ ˜ ˜ P :(u˜, F) → I + · F . (21) u,F ∂F ∂X Applying the introduced notation into the PVW, we review the original MULF pre- stressing and subsequent deformation stage as In prestressing stage, find F such that : P (0, F ): ∇δu dV − T(0,p ) · δu dA = 0 ∀δu, (22) u,F p dia 0 p,0 In deformation stage, find u (with given F ) such that : d p P (u , F ): ∇δu dV − T(u ,p ) · δu dA = 0 ∀δu. (23) u,F p sys d d 0 p,0 Equation (22) implicitly defines the prestress deformation gradient F , which is eval- uated applying an assumed diastolic blood pressure load T(0,p ) at the known imaged dia geometry. Equation (23) utilizes the precomputed deformation gradient F in order to evaluate the deformation stage displacement field u applying an assumed systolic blood pressure load T(u ,p ) at the deformed geometry. d sys Recalling Eqs. (20)and (21), we can equivalently state the prestressing and deformation stage PVW as In prestressing stage, find u such that : P (u ): ∇δu dV − T(0,p ) · δu dA = 0 ∀δu, (24) u p dia 0 p,0 In deformation stage, find u (with given u ) such that : d p P (u + u ): ∇δu dV − T(u ,p ) · δu dA = 0 ∀δu. (25) u d p d sys 0 p,0 A comparison of (22), (23)with (24), (25) reveals the following. Instead of seeking a prestress deformation gradient F , we solve for a virtual prestress displacement field u p p fulfilling the PVW at a diastolic blood pressure load of the imaged geometry T(0,p ). u dia is then used in the deformation stage to account for the stress in the imaged configuration at a systolic blood pressure load of the deformed configuration T(u ,p ). d sys We emphasize that the reformulation from (22), (23)to(24), (25) corresponds to a mathematical transformation of variables, physics remains unchanged. We also empha- size that both formulations are a well-posed approximation to the ill-posed inverse design problem as further detailed in [14]. From the perspective of projection-based MOR how- ever, formulation (24), (25) enables a collection of virtual prestress displacement mode snapshots, an essential step in the data-driven construction of the ROB. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 8 of 31 Finite element discretization Applying the usual finite element discretization to the PVW for the MULF prestressing and deformation stage gives (e) In prestressing stage, find u = u such that : e∈E (e) (e) P (u ): ∇δu dV (e) e∈E (e) (e) − T(0,p ) · δu dA = 0 ∀δu , (26) dia (e) e∈F p,0 (e) (e) In deformation stage, find u = u (with given u ) such that : d p e∈E (e) (e) (e) P (u + u ): ∇δu dV (e) e∈E (e) (e) (e) − T(u ,p ) · δu dA = 0 ∀δu , (27) sys (e) e∈F p,0 (e) (e) (e) (e) (e) (e) wherein u = d , δu = δd are the continuous element-wise displacement field and weighting function, which are interpolated by finite element shape functions (e) contained in and the element-wise displacement and weighting degree of freedom (e) (e) (DOF) vectors d , δd , respectively. Furthermore, we introduced the computational domain mesh element set E as well as the set F of elements loaded by the pressure load boundary condition. Given element-wise internal and external force vectors such that (e) (e) (e) (e) (e) (e) (e) (e) f (d ) · δd = P ( d ): ∇( δd ) dV ∀δd , u (28) int (e) (e) (e) (e) (e) (e) (e) (e) (e) f (d ,p) · δd = T( d ,p) · ( δd ) dA ∀δd , (29) ext (e) p,0 Eqs. (26)and (27) in assembled form read In prestressing stage, find d such that : f (d ) · δd − f (0,p ) · δd = 0 ∀δd (30) p dia int ext ⇒ f (d ) − f (0,p ) = 0, (31) p dia int ext In deformation stage, find d (with given d ) such that : f (d + d ) · δd − f (d ,p ) · δd = 0 ∀δd (32) p sys int d ext d ⇒ f (d + d ) − f (d ,p ) = 0. (33) p sys int d ext d (e) (e) Thereby the global internal force vector f = L f , global external force vector int e∈E int (e) (e) (e) (e) f = L f , global displacement DOF vector d = L d as well as global ext e∈F ext e∈E (e) (e) weighting DOF vector δd = L δd result from an assembly of the corresponding e∈E (e) element-wise vectors, while L is the usual finite element assembly operator towards the global system. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 9 of 31 Summarizing, we denote the high-fidelity finite element model residual as N N ⎪ R → R for prestressing stage: r : , (34) d → f (d ) − f (0,p ) ⎪ p p dia int ext for deformation stage (with given d ): d → f (d + d ) − f (d ,p ) d d p d sys int ext wherein the deformation stage only can be evaluated after the prestressing stage, which yields the virtual prestress displacement field d as a solution. The nonlinear finite element system of equations in residual form reads r(d) = 0 (35) and is solved applying Newton-Raphson iterations. Reduction of the full-order model In this section we briefly review the well known Galerkin projection, which yields a dimen- sionally reduced computational model. For nonlinear problems, the Galerkin projection is usually not sufficient to gain substantial computational speedup, given that the full- order residual still needs to be assembled. For this reason, we additionally review the energy-conserving mesh sampling and weighting [19,20] hyper reduction method, which approximates the full-order residual with only a small subset of assembled mesh elements. Galerkin projection on linear subspaces The Galerkin projection has proven its applicability in structural mechanics problems [19,31,32]. Assuming a given orthogonal ROB V (its construction will be discussed in section “Construction of reduced-order model components”) N ×n T V ∈ V :={W ∈ R : W W = I }, (36) N,n the dimensionally reduced model (DROM) retrieved from the Galerkin projection reads V r(V d) = 0, (37) with d ∈ R assuming n N. The argument of the residual is restricted to the column span of the ROB V d ∈ span(V ), which corresponds to a reduction of the number of DOFs. Consistently, the number of equations is reduced by multiplication with the transposed ROB V r(V d) ∈ R . As a result, application of the Newton-Raphson iteration scheme leads to i i+1 i T T ˆ ˆ ˆ V J (V d )V d =−V r(V d ) (38) i+1 i i+1 ˆ ˆ ˆ d = d + d , (39) wherein J is the residual Jacobian with respect to the displacement field d. Equations (38), (39) reveal that only low-dimensional linear systems of equations have to be solved. Hyper reduction of internal force contribution The Galerkin Projection (37) leads to a dimensionally reduced model, however the FOM ˆ ˆ residual r(V d) still needs to be evaluated together with its Jacobian J (V d) throughout r Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 10 of 31 Newton-Raphson iterations (38). Especially assembly of the internal force component of the residual (cf. Eq. (34)) is time consuming, given that every element of the computational mesh needs to be evaluated. To reduce the cost of evaluation and assembly of the residual and its Jacobian, we apply the energy-conserving mesh sampling and weighting (ECSW) hyper reduction scheme [19,20] and give a brief review in the remainder of this section for completeness and adaption to the current context of application. N N The idea is to replace the internal force vector f ∈ R with a surrogate f ∈ R , int int which will result in an accurate approximation after projection, that is T T ˜ ˆ ˆ V f (V d) ≈ V f (V d). (40) int int f is retrieved by a weighted assembly of a small mesh element subset and is derived int from the requirement of an accurate approximation of the internal virtual work, which can be written as δW (d, δd) = δd f (d) (41) int int or for a dimensionally reduced model d, δd ∈ span(V ) ˆ ˆ ˆ ˆ δW (d, δd) = δd V f (V d). (42) int int Applying a sum over all element internal force contributions we can rewrite (e) T (e) (e)T ˆ ˆ ˆ ˆ δW (d, δd) = δd V L f (L V d) (43) int int e∈E (e)T using L to extract DOFs of element e from a vector with global DOF numbering into a smaller vector with element DOF numbering. We now seek for an approximation W (d, δd) of the internal virtual work such that int ˜ ˆ ˆ ˆ ˆ ˆ ˆ δW (d, δd) ≈ δW (d, δd) ∀δd ∈ R (44) int int with (e) T (e) (e) (e)T ˜ ˆ ˆ ˆ ˆ δW (d, δd) = w δd V L f (L V d). int int (45) e∈E In contrast to (43), (45) only contains a summation over a reduced element set E. Addi- (e) tional non-negative element weights w ∈ R are introduced for approximation (44)to become feasible with a small cardinality of the reduced element set. A remaining question is the actual choice of elements in E as well as the element weights (e) w . For this reason, (44) is turned into an optimization problem by restriction to a finite set of displacement modes S S ={V d : d ∈ S} (46) with m m m S ={d , d + d , ... , d , d + d } (47) p(0) p(0) d(0) p( −1) p( −1) d( −1) 2 2 2 being a set of known displacement modes (referred to as ECSW displacement modes here, the actual selection of modes will be discussed in section “Construction of reduced-order model components”). Note that S consists of m (even number) modes corresponding to virtual prestress displacement modes d and the sum of virtual prestress displacement p(i) and deformation stage displacement modes d + d with i ∈{0, ... , − 1}. p(i) d(i) 2 Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 11 of 31 ˆ ˆ In its unassembled shape, the restriction of Eq. (44)to ∀d ∈ S reads T T (e) (e) (e) T (e) (e)T T (e) (e)T ˆ ˆ ˆ ˆ w δd V L f (L V d) ≈ δd V L f (L V d) int int ˜ e∈E (48) e∈E ˆ ˆ ˆ ∀d ∈ S, ∀δd ∈ R . In order to keep the cardinality of the reduced element set E low, approximation (48)hasto be accurate with a minimal number of non-zero weights. The corresponding optimization problem is minimize w |E| w∈R (49) subject to Aw − b ≤ ε b 2 h 2 min(w) ≥ 0. The zero-norm (•) counts the number of non-zero entries and is used as the objec- tive function applied to the vector of element weights w, which is constrained to have non-negative values expressed by its minimum entry min(w) being non-negative. This constraint is required in order to ensure a positive semi-definite Jacobian of the inter- nal force vector [19]. The other constraint is a fulfillment of Eq. (48) up to the relative tolerance ε . Consequently, ⎡ ⎤ a ... a (0,0) (0,|E|−1) ⎢ ⎥ . . . n∗m×|E| ⎢ ⎥ . . . A = ∈ R (50) . . ⎣ ⎦ a ... a (m−1,0) (m−1,|E|−1) with vector valued entries (j) T (j) (j)T (51) a = V L f (L V d ) (i,j) int and ⎡ ⎤ ⎢ ⎥ . n∗m ⎢ ⎥ b = . ∈ R (52) ⎣ . ⎦ m−1 with vector valued entries b = a (53) (i,e) e∈E ˆ ˆ wherein i ∈{0, ... ,m − 1},j ∈{0, ... , |E|− 1} and d are vectors from the set S. Optimization problem (49) can be approximately solved with a sparse non-negative least-squares solver. Thereby sparse refers to the solution vector w, in the sense that the number of non-zero entries is kept minimal. For details on the iterative solver the reader is referred to [19]. An (approximate) solution to (49) returns the element weights as well as the reduced element set by an extraction of elements with non-zero weights. As a consequence, the hyper reduced internal force vector and its Jacobian read (e) (e) (e) (e)T f (d) = w L f (L d), (54) int int e∈E (e) (e) (e) (e)T (e)T J (d) = w L J (L d)L , (55) int int e∈E Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 12 of 31 (e) (e) ∂f int with the element stiffness J = . int (e) ∂d We denote the dimensionally reduced as well as hyper reduced model (DHROM) as V r˜(V d) = 0 (56) and state the corresponding Newton-Raphson iterations i i+1 i T T ˜ ˆ ˆ ˆ V J (V d )V d =−V r˜(V d ) (57) i+1 i i+1 ˆ ˆ ˆ d = d + d , (58) ˜ ˜ wherein r˜ is a residual approximation using f and J is the corresponding hyper reduced int residual Jacobian. Construction of reduced-order model components In the given many-query context, the residual (34) depends on a modifiable set of model parameters. Introducing a parameter vector μ ∈ P = [lb ;ub ] × ... × [lb ;ub ] ⊂ R (59) 0 0 n −1 n −1 P P with [lb ;ub ] μ being the lower and upper bounds for parameter μ , we extend the i i i i notation of Eq. (35)to r(d(μ); μ) = 0, (60) and attempt to find a ROB V andasetofECSWmodes S such that the resulting DROM as well as DHROM will accurately approximate the FOM solution for all μ ∈ P. A prerequisite for an accurate ROM is that the FOM solution d(μ) can be accurately represented within the column span of the ROB d(μ) − VV d(μ) d(μ) . (61) This motivates a data-driven approach for construction of ROBs by a collection of FOM solution snapshots at different parametric configurations and subsequent orthogonaliza- tion with or without data compression [33]. In case of the presented stationary AAA computational model two snapshots per para- metric configuration (virtual prestress displacement and deformation stage displacement) are retrieved and organized in the so-called snapshot matrix S = [d (μ ), d (μ ), ... , d (μ m ), d (μ m )] (62) p d p d 0 0 −1 −1 2 2 N ×m with S ∈ R . Two data-driven approaches for the construction of the ROB have gained special inter- est in projection-based MOR. Proper orthogonal decomposition (POD) can be used to N ×m orthogonalize S yielding a ROB V ∈ R such that [17] pod 2 2 n nT T S − V V S = min S − WW S , (63) pod pod F W ∈V F N,n wherein V corresponds to a selection of the first n columns of V with n ≤ m and pod pod (•) is the Frobenius norm. Consequently, POD is used whenever solution snapshots can be accurately represented by a low-dimensional subspace. The second approach for construction of the ROB are greedy methods [34]. The idea herein is to successively build the ROB by evaluating selected configurations within the Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 13 of 31 parameter domain and enrich the span of the ROB by the span of the newly computed snapshots. Based on an (hopefully inexpensive and sharp) a posteriori error estimator, greedy methods attempt to find solution snapshots which are represented worst by the ROB constructed up to this point. However, even this local optimization problem quickly becomes too expensive. Several approaches [35–37] have been presented to date to over- come this computational bottleneck. In order to avoid evaluating a posteriori error estimates in every greedy iteration, we apply a selection of solution snapshots from a greedy maximin distance sampling together with a stopping criterion based on subspace angles and an exclusion of subdomains. We dedicate section “Maximin distance design” and “Subspace angles for the comparison of subspaces” to the notion of maximin distance design and subspace angles, respectively. Section “A greedy maximin distance sampling approach for the construction of solution subspaces” introduces the actual sampling algorithm. Maximin distance design Space-filling designs is a topic from design of experiments. Maximin distance (MMD) is introduced in [38] as a criterion which can be used to rate the space-filling property of a design or to construct space-filling designs by optimization of that criterion. A greedy version with reduced computational complexity is presented in [39] under the name “Coffee-House Design” and a recent review on maximin distance sampling can be found in [40]. In contrast to a globally optimal MMD design, the greedy MMD design can be evaluated iteratively. We apply the following terminology. A point is a specific instance of the parameter vec- tor μ, also referred to as parametric configuration. Points are distributed by the sampling algorithm in the parameter domain. A sample corresponds to FOM solution snapshots at a given point. Algorithm 1 depicts the steps for the selection of a greedy MMD point. Given an input grid ⊂ P as subset of the parameter space and a set of previously chosen points ⊂ P, i c the next point μ ∈ is chosen such that the minimal distance to a neighboring point −1 p ∈ is maximized in a reference hypercube. Thereby χ and χ map from physical domain to reference hypercube and vice versa, respectively. Algorithm 1 MaxiMinPoint( , ) (select a greedy MMD point) i c Input: input grid ⊂ P, previously chosen points ⊂ P i c Output: selected grid point μ ˜ ˜ 1: = χ( ), = χ( ) transform grids to reference hypercube i i c c 2: μ = arg max min q − p˜ get next point in reference hypercube ˜ ˜ q∈ p˜ ∈ i c −1 3: return χ (μ) return point in physical domain The steps for a greedy MMD design on a training grid ⊂ P are depicted in Algo- rithm 2.Figure 2 illustrates a greedy MMD design, wherein the first parametric configu- ration was chosen at random. The idea of MMD sampling for the purpose of surrogate modeling in general is dis- cussed broadly in literature [41–44]. Specific applications for instance can be found in [45], wherein the authors use the notion of MMD in their algorithm to sample cut lines Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 14 of 31 Algorithm 2 MaxiMinDesign( , μ,n ) (construct greedy MMD design) t μ Input: training grid ⊂ P, starting point μ ∈ , number of points to select n t t μ Output: chosen points 1: ={μ} 2: for i ∈ (1, ... ,n − 1) do 3: μ = MaxiMinPoint( , ) t c 4: ← ∪ μ c c 5: end for 6: return Fig. 2 Greedy MMD design for a 2D parameter space with n points. The first point is chosen at random and planes of the parameter domain in order to construct a radial-basis-function approx- imation surrogate. In [46], MMD sampling is used to distribute points in Voronoi cells for multifidelity radial-basis-function metamodeling. Subspace angles for the comparison of subspaces Subspace angles (or principal angles) are a concept from matrix computations [47]. Given N ×n N ×m two matrices Y ∈ R , Z ∈ R with n ≤ m, subspace angles can be defined recur- sively as the minimum value θ = min arccos(y z)with k ∈{0, ... ,n − 1}, (64) ⊥ ⊥ y∈Y ,z∈Z k k while the corresponding principal vectors follow from the minimization arguments y , z = arg min arccos(y z)with k ∈{0, ... ,n − 1}. k k (65) ⊥ ⊥ y∈Y ,z∈Z k k The sets Y ={y : y ∈ span(Y ), y = 1} Z ={z : z ∈ span(Z), z = 1} (66) for k = 0 Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 15 of 31 Fig. 3 Geometrical interpretation of subspace angles between 2D subspaces of a 3D space. The subspaces 3×2 are defined by the column span of Y , Z ∈ R in the current illustration. θ and θ are the two subspace 0 1 angles, y , y ∈ span(Y)and z , z ∈ span(Z) are the corresponding principal vectors 0 1 0 1 and ⊥ T Y ={y : y ∈ span(Y ), y = 1, y y = 0} k j ⊥ T (67) Z ={z : z ∈ span(Z), z = 1, z z = 0} for j ∈{0, ... ,k − 1},k ∈{1 ... n − 1} depend on the minimization arguments y and z of the previous k iterations. A maximum subspace angle of 0 indicates that span(Y ) ⊆ span(Z), while a maximum subspace angle of 90 indicates that there is at least one direction in span(Y ) which is orthogonal to span(Z). More general, the maximum subspace angle can be interpreted as a distance measure from span(Y)tospan(Z). In the following we will refer to the maximum subspace angle as the subspace angle distance (SAD). Figure 3 illustrates subspace angles for 2D subspaces embedded in a 3D space. Algorithm 3 [47] states the computation of subspace angles applying a singular value decomposition. Algorithm 3 SSA(Y , Z) (computation of subspace angles) N ×n N ×m Input: Y ∈ R , Z ∈ R with n ≤ m Output: subspace angles α 1: Y = Q R , Z = Q R perform thin QR factorization [47] Y Z Y Z T T 2: Q Q = Udiag(σ)Q perform thin singular value decomposition [47] 3: α = arccos(σ) transform to angle 4: return α In projection-based MOR, subspace angles have been used for the purpose of interpo- lation and sampling. In [48–50], the authors present and apply a subspace angle interpo- lation of ROBs for flow problems. In [51], subspace angle interpolation is performed with respect to the diffusion coefficient for a Diffusion-Convection-Reaction problem. The application of subspace angles as a stopping criterion for sampling has been presented in [52–54]. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 16 of 31 A greedy maximin distance sampling approach for the construction of solution subspaces We expand the greedy MMD design with a stopping criterion based on the SAD and introduce adaptivity to the sampling by a division of the parameter domain into subdo- mains. Those subdomains are subsequently excluded from sampling and the algorithm stops, when all subdomains have been excluded. Algorithm 4 SDMaxiMinSampling( , μ, α ) (greedy MMD sampling on subdomains) sd m Input: subdomain set ={ ... , }, starting point μ ∈ , threshold SAD α sd sd,0 sd,n −1 sd,0 sd Output: chosen grid points ={μ , ... , μ },ROB V,snapshotmatrix S 0 n −1 1: s(μ) = QR thin QR factorization [47] of initial snapshot matrix 2: V = Q 3: S = s(μ) 4: ={μ} 5: = ( , ... , , ) define subdomain tuple for iteration sd,1 sd,n −1 sd,0 sd 6: while True do 7: for ∈ do iterate over subdomains sd,i 8: μ = MaxiMinPoint( , ) sd,i c 9: ← ∪ μ c c 10: α = SSA(V , s(μ)) 11: α = max(α) 12: if α< α then in case of small SAD 13: ← \ exclude subdomain from sampling sd,i 14: end if 15: [V , s(μ)] = QR add new modes and perform thin QR factorization 16: V = Q 17: S ← [S, s(μ)] 18: end for 19: if =∅ then if no subdomain left 20: break stop algorithm 21: end if 22: end while 23: return , V , S Algorithm 4 exposes the individual steps. After having calculated an initial ROB (line 2:) from the local snapshot matrix s at the initial parametric configuration μ (line 1:), the algorithm iterates over a (predefined) tuple (line 5:) of subdomains ⊂ for sd,i i ∈{0, ... n − 1} (n consequently is the number of subdomains), wherein ∩ i=j sd sd sd,i n −1 sd =∅ for i, j ∈{0, ... ,n − 1} and ∪ = . In every iteration, a parametric sd,j sd sd,i i=0 configuration is chosen within by a greedy MMD step (line 8:). Note that in line 8: sd,i distances to all previously selected points are taken into account, although the new point is selected exclusively from . Subdomains are excluded from sampling (line sd,i 13:) depending on the threshold α (line 12:). The algorithm stops, if all subdomains have been excluded (line 19:,20:). On output, Algorithm 4 returns a set of selected grid points , a global orthogonal ROB V as well as a globally collected snapshot matrix S. We use the ROB for dimensional reduction by the Galerkin projection (37), while the snapshot matrix is used to compute the set S (47). Consequently, the number of ECSW displacement modes |S| coincides with the dimension of the subspace span(V ). Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 17 of 31 Note that the ROB enrichment strategy of Algorithm 4 (line 15:,16:) extends the span of the ROB by the span of the current snapshot matrix s(μ) in every greedy maximin iteration. As a consequence, we can state that span(V ) ⊆ span(V ) for j ≥ i, wherein V i j i and V denotes the ROB after greedy maximin iteration i and j, respectively. Following the terminology in [43], we classify the proposed approach as global, sequen- tial / adaptive and fine-grained. Additionally, the MMD criterion as well as initial itera- tions over subdomains introduce the property of domain exploration, while subsequent sampling of a subset of subdomains amounts to local exploitation. In more detail, by introducing subdomains the sampling algorithm can evaluate more samples in specific parameter domain regions as compared to others, if this need is iden- tified by the subspace angle criterion. As a result, the purely explorative greedy maximin sampling receives a feedback from the parameter domain and sampling is refined. Refer to section “Application of greedy maximin distance sampling” for a demonstration in the given context of prestressed and parametrized AAAs. A stopping criterion for sampling based on subspace angles already has been presented in [52–54]. In [52,53] the authors present adaptive sampling of a linear time-invariant state-space system, while in [54] adaptive selection of linearization points in trajectory piecewise linear approximation is in the focus of interest. In contrast to [52–54]the approach in this work aims at the construction of a global ROB instead of interpolation between parametric configurations. Additionally, the notion of MMD yields to finely granular sampling applicable to parameter domains with multiple dimensions. Results and discussion The proposed framework is applied to three patient-specific computational examples of AAAs. The ROB is constructed by greedy subdomain MMD sampling with 8 subdomains following Algorithm 4. The accuracy of the resulting DROMs and DHROMs is evaluated in terms of the quantities of interest (von Mises stress field and von Mises strain field in the aortic vessel wall) and wall clock timings are reported. The choice of von Mises type quantities of interest is based on preceding numerical studies on AAAs with emphasis on solid mechanics and rupture risk (e.g. as presented in [9,55,56]). Nonetheless, other quantities of interest could have been selected, given that AAA pathological progression is still subject to research. Finally, accuracy of the proposed MOR framework in a statistical sense is demonstrated by comparing maximum von Mises stress and von Mises strain probability distributions gained from FOM and DHROM sampling. Patient-specific computational models Figures 4, 5 and 6 visualize the computational mesh, a cut through the computational domain depicting a separation between the ILT and the aortic wall and exemplary von Mises stress distributions for patient 1, patient 2 and patient 3, respectively. The ILT is discretized using linear tetrahedral and pyramid elements, wherein pyramids are introduced to connect the ILT to the aortic wall, which is discretized using linear hexahedral elements with F-bar element technology [57]. Table 1 depicts information on the model discretization. Referring to section “Computational modeling of abdominal aortic aneurysms”, we quantify boundary conditions by a diastolic blood pressure of p = 87 mmHg (11.6kPa) dia Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 18 of 31 Fig. 4 Patient 1 mesh (a), cut exposing the ILT (b), exemplary von Mises stress distribution (c) Fig. 5 Patient 2 mesh (a), cut exposing the ILT (b), exemplary von Mises stress distribution (c) Table 1 Number of degrees of freedom N and number of elements N for patient-specific computational models Patient 1 Patient 2 Patient 3 N [-] 109587 189504 479487 N [-] 140019 149499 776106 e Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 19 of 31 Fig. 6 Patient 3 mesh (a), cut exposing the ILT (b), exemplary von Mises stress distribution (c) Table 2 Patient-specific bounds for parameter domain Patient 1 Patient 2 Patient 3 [α ; α ] [kPa] [28.23; 345.22] [18.15; 344.07] [26.46; 503.16] l u [β ; β ] [kPa] [541.46; 6164.14] [543.15; 9686.14] [450.97; 7986.08] l u [t ; t ] [mm] [1.09; 2.66] [0.94; 2.46] [1.03; 2.73] l u and a systolic blood pressure of p = 121 mmHg (16.1 kPa). The ILT stiffness c is sys interpolated linearly from a luminal stiffness of c = 2.62 kPa to a medial stiffness of c = 1.98 kPa and from the medial stiffness to an abluminal stiffness of c = 1.73 kPa [26]. Together with the aortic wall thickness t the model parametrization is given as ⎡ ⎤ ⎢ ⎥ μ = β ∈ P = [α ; α ] × [β ; β ] × [t ; t ] ⊂ R , (68) ⎣ ⎦ u u u l l l with the subscripts l and u denoting the lower and upper bound. Table 2 exhibits parameter domain lower and upper bounds for the three models. The bounds were computed from patient-specific Log-normal probability distributions from [58] for each entry of the parameter vector μ. In more detail, the parameter domain bounds are chosen as (69) (γ , γ ) = (Q (0.025; μ , σ ),Q (0.975; μ , σ )) for γ ∈{α, β,t} l u log γ γ log γ γ with −1 Q (p; μ , σ ) = exp(μ + 2σ erf (2p − 1)) (70) log γ γ γ γ being the p-percentile value for a Log-normal distribution with expectation μ and stan- dard deviation σ . erf denotes the error function. Consequently, the range within the chosen parameter domain bounds covers 95% of realizations of μ. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 20 of 31 We perform 15 equally spaced load steps for the prestressing stage and 10 equally spaced load steps for the deformation stage. Multiple thousands of simulations were performed and postprocessed for the results presented in the following sections. Individual unconverged simulations were dropped from analysis. For linear systems of equations arising in FOM simulations, we use an iterative, parallel GMRES solver with algebraic multigrid preconditioning implemented in Trilinos [59]. For the ROM linear systems of equations we apply a direct solver [60], given that arising linear systems have less than 100 unknowns. Application of greedy maximin distance sampling In our first numerical experiment, we create a one-shot (i. e. no adaptation) design dis- tributing 200 points in the parameter domain (Algorithm 4 with ={ }, α = 0.0 sd sd,0 m and stopping at | |= 200). If a simulation fails to converge, a neighboring point is taken in the set of selected points instead and the FOM is recomputed. The parameter domain grid is created from all combinations of 100 equidistantly placed points in each direction of the parameter domain axes. The initial point is chosen as the “minimum-value” point μ = [α , β ,t ] (see Table 2) for each patient-specific example. l l l As a result, the greedy MMD design returns identical points (except for few individually shifted points due to convergence failure) in the reference cube for all three computa- tional examples. The corresponding MMD in the reference cube is depicted in Fig. 7d. Simultaneously, the SAD (Algorithm 4 (line 11:)) is depicted in Fig. 7a–c, wherein we highlight SADs corresponding to (−−−)-octant configurations of the parameter domain (i.e. α< α ,β< β ,t < t ) in blue and SADs corresponding to (+++)-octant con- m m m figurations of the parameter domain (i.e. α> α ,β> β ,t > t ) in dark red (given m m m 1 1 1 α = (α + α ), β = (β + β )and t = (t + t ) as the axes mid values). m l u m l u m l u 2 2 2 The resulting distribution of SADs is affected by two contributions. The first contribu- tion is the MMD for a newly set point. As one can observe from Fig. 7d, the MMD strongly decreases initially, while a stagnation occurs with an increasing number of samples. This behavior also reflects in the SAD, which shows a pronounced decay in the beginning and increased scattering with ongoing stagnation of the MMD. The second contribution are different sensitivities of FOM snapshots with respect to the parameter domain. For instance, (+++)-octant value parametrizations (dark red points) yield lower subspace angles than (−−−)-octant value parametrizations (blue points), such that the (+++)- octant of the parameter domain can be said to show lower sensitivity in solution snapshots. For practical reasons (see numerical examples presented next), we are only interested in the region of pronounced decay of the SAD. As a consequence, distance in the reference parameter space is a suitable and efficient sampling criterion for the problem at hand. We include adaptivity to the greedy MMD design by introducing subdomains as pre- sented in section “A greedy maximin distance sampling approach for the construction of solution subspaces”. In more detail, we create eight equally shaped subdomains by splitting each parameter domain axis in 2 intervals and run Algorithm 4 with α = 0.1, ={ ... , } and the initial configuration μ = [α , β ,t ] .Figure 8 depicts sd sd,0 sd,7 l l l the decay of SADs for each patient-specific computational model. As one can observe, sampling runs until the last sample in each subdomain yields a SAD below α . m Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 21 of 31 Fig. 7 a–c Decay of the SAD. Dark red points correspond to parametric configurations in the (+++)-octant of the parameter domain, while blue points correspond to parametric configurations in the (−−−)-octant of the parameter domain. d Decay of MMD with every newly set point Table 3 Number of points distributed in the individual subdomains. The corresponding decay of subspace angles is depicted in Fig. 8, the point distributions in Fig. 9 Subdomain01234567 Patient1 44443332 Patient2 65443533 Patient3 64443322 Table 3 depicts the number of distributed points in the individual subdomains. Note the differences in the point distributions, especially prominent for patient 3. Figure 9 depicts the selected parametric configurations in the 3D parameter domain. As one can see, the most sensitive subdomain 0 (compare Table 3) corresponds to the low-stiffness and thin vessel wall range of the parameter domain. This is plausible from a physical perspective, given that soft and thin-walled tissue will deform more than stiff and thick-walled tissue. Patient-specific reduced-order models We compute ROBs from greedy subdomain MMD sampling with parametric configura- tions as depicted in Fig. 9. A Galerkin projection (37) yields the patient-specific DROMs. Hyper reduction is achieved via ECSW (cf. section “Hyper reduction of internal force −4 contribution”) parallelized on 4 processors with global tolerance set to ε = 10 ,see Eq. (49). The parallelization is implemented in terms of a domain decomposition as described in [20]. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 22 of 31 Fig. 8 Decay of SAD throughout parameter domain sampling. The parameter domain is subdivided into eight subdomains. The stopping criterion is a SAD below 0.1 in each subdomain. Corresponding point distributions in the parameter domain are depicted in Fig. 9 Fig. 9 Parametric configurations selected throughout greedy subdomain maximin distance sampling. The parameter domain is subdivided into eight subdomains Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 23 of 31 Fig. 10 Reduced meshes gained from ECSW. Only the colored mesh elements are evaluated and assembled in patient-specific DHROMs Table 4 Number of degrees of freedom n and number of elements n of constructed DHROMs Patient 1 Patient 2 Patient 3 n [-] 54 66 56 n [-] 1994 3193 2975 Figure 10 illustrates selected mesh elements, while Table 4 shows the number of DOFs and selected mesh elements. Inspecting Fig. 10, note the increasingly accurate sampling in the neighborhood of vessel fixation (proximal and distal vessel ending) and in regions with increased curvature of the vessel wall (compare with Figs. 5 and 6). Accuracy of the reduced-order model We evaluate accuracy in terms of the relative error in the quantities of interest (von Mises stress field σ and von Mises strain field e in the aortic wall). The relative error is vM vM given as x˜ − x RE(˜x, x) = , (71) wherein x ∈{σ , e } corresponds to FOM quantities and x˜ ∈{σ˜ , e˜ } corresponds vM vM vM vM to ROM approximations. A validation grid with 1000 points in the parameter domain is used. The grid results from all combinations of 10 equidistantly placed points in each direction of the parameter domain axes. Note that the resulting grid corresponds to a full factorial design being created independently of the points used for the construction of the ROB by greedy subdomain MMD sampling. Figures 11 and 12 depict the corresponding errors. The majority (> 98%) of relative errors are below 1%, while individual runs show a relative error above 1%. We conclude, that DROM as well as DHROM are accurate models for the von Mises stress and von Mises strain field in the aortic wall in a statistical sense. Individual simulations might show increased relative errors, caution is required when applying the ROMs for the prediction of point estimates. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 24 of 31 Fig. 11 Relative l2-error of the DROM in term of the von Mises stress and von Mises strain fields in the aortic wall. The test grid corresponds to a full factorial and equidistant 10 × 10 × 10 grid in the parameter domain. The results state individual simulations showing errors above 1%, while the majority (> 98%) of simulations shows errors below 1% We investigate the influence of α and ε , i.e. the maximum subspace angle defining m h the stopping criterion in Algorithm 4 and the relative tolerance for the ECSW algorithm introduced in Eq. (49). On the introduced validation grid with 1000 points, we evaluate n −1 sim RE = RE (˜x, x) (72) x i sim i=0 as the mean value of all relative errors given the number of performed simulations n , sim x ∈{σ ,e } denoting von Mises stress or von Mises strain as the quantity of interest vM vM and RE (˜x, x) being the relative error (71)ofsample i. i Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 25 of 31 Fig. 12 Relative l2-error of the DHROM in terms of the von Mises stress and von Mises strain fields in the aortic wall. The test grid corresponds to a full factorial and equidistant 10 × 10 × 10 grid in the parameter domain. The results state individual simulations showing errors above 1%, while the majority (> 98%) of simulations shows errors below 1% Figure 13 depicts results for patient 1 as exemplary model. In more detail, Fig. 13a corresponds to DROM results with ROBs created by Algorithm 4 at different SADs, while Fig. 13b corresponds to DHROM results with ECSW meshes at different tolerances and an unchanged ROB at α = 0.1. As can be observed from Fig. 13a, lower values for α m m yield larger ROBs, while at the same time the mean relative error decreases indicating a more accurate DROM as expected. Figure 13b illustrates that stricter ECSW tolerances ε lead to an increased number of selected mesh elements with decreasing mean relative error indicating a more accurate DHROM. We conclude, that both α and ε strongly m h influence ROM accuracy. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 26 of 31 Fig. 13 Mean relative l2-error for DROM over varying threshold SAD α (left plot) and mean relative l2-error for DHROM over varying ECSW tolerance ε (right plot) are depicted in blue (curves and y-axes). Curves and y-axes in orange depict the corresponding number of ROB modes n (left) and number of selected ECSW elements n (right) vMmax vMmax Table 5 Mean values μ and standard deviations σ with x ∈{stress, strain} for maximum x x von Mises stress and maximum von Mises strain in the aortic wall computed from Monte Carlo sampling with 10000 identical (per patient) samples. The corresponding parametric configurations are drawn from patient-specific probability distributions Value Patient 1 Patient 2 Patient 3 vMmax MC FOM μ [kPa] 205.57 276.84 302.49 stress vMmax σ [kPa] 44.950 65.926 71.761 stress vMmax μ [−] 0.17320 0.19404 0.19070 strain vMmax σ [−] 0.038272 0.052792 0.049465 strain vMmax MC DHROM μ [kPa] 205.80 277.13 301.85 stress vMmax σ [kPa] 45.0518 65.9210 71.8432 stress vMmax μ [−] 0.17361 0.19413 0.18946 strain vMmax σ [−] 0.038483 0.052680 0.049128 strain Monte Carlo sampling on the reduced-order model To demonstrate applicability of the constructed ROMs for approximation of probability distributions in the quantities of interest, we compare the 99.9 percentile aortic wall von Mises stress (referred to as maximum von Mises stress in the following) and the 99.9 percentile aortic wall von Mises strain (referred to as maximum von Mises strain in the following) probability distributions retrieved from Monte Carlo sampling of the FOM and Monte Carlo sampling of the DHROM. Both models are evaluated on 10000 identical (per patient) parametric configurations drawn from the corresponding patient-specific Log-normal probability distributions. Table 5 depicts mean and standard deviation of the quantities of interest. As one can see, FOM and DHROM results are very close, relative errors are < 1%. Figure 14 depicts kernel-density-estimated probability distributions gained from FOM and DHROM sam- ples. We apply the Gaussian kernel-density-estimator scipy.stats.gaussian_kde available in the SciPy [61] (version 1.3.0) ecosystem of the Python programming language. The plots show negligible differences between probability distributions gained from FOM and DHROM sampling. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 27 of 31 Fig. 14 Kernel-density-estimated (Gaussian kernel) probability distributions based on FOM and DHROM evaluation of 10,000 identical (per patient) parametric configurations. The quantities of interest are the maximum von Mises stress as well as the maximum von Mises strain in the aortic wall Timing We report wall clock timings of the patient-specific computational models as well as corresponding speedups in Table 6. All simulations in this section were performed on a workstation with Intel Xeon W-2133 (3.60GHz) processors. The values in Table 6 are mean values corresponding to seven simulations (per patient) evaluated at face mid-points as well as the mid-point of the patient-specific parametric domains, compare with Table 2 for domain lower and upper bounds. As the reader can observe, only slight speedup can be achieved with DROM models, given that the full-order residual as well as its Jacobian need to be evaluated. A rather Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 28 of 31 Table 6 Timing for patient-specific models. All computations were performed on 4 cores (Intel Xeon W-2133 (3.60GHz)). The reported timings are mean values of seven simulations (per patient). Speedup is computed using FOM timing as a reference Model Computing time [s] Speedup [-] Patient 1 FOM 199.5 1.0 DROM 128.1 1.6 DHROM 13.5 14.8 Patient 2 FOM 261.1 1.0 DROM 175.9 1.5 DHROM 20.9 12.5 Patient 3 FOM 1126.1 1.0 DROM 618.5 1.8 DHROM 31.8 35.5 Table 7 Timing for offline stage steps of patient-specific models performed on 4 cores (Intel Xeon W-2133 (3.60GHz)). Timing is given as multiple of a single FOM evaluation time, which in turn is estimated from the mean of seven simulations per patient Reduction step Multiple of FOM evaluation time [-] Patient 1 construction ROB 29.2 ECSW 5.4 Patient 2 construction ROB 38.4 ECSW 38.0 Patient 3 construction ROB 31.0 ECSW 16.7 substantial speedup can be achieved by DHROM models, recalling that only a small portion of the computational mesh is evaluated and assembled. Table 7 depicts offline stage timings, subdivided into the ROB construction stage by greedy subdomain MMD sampling (as described in section “Application of greedy max- imin distance sampling”) and the hyper-reduction by ECSW (as described in “Patient- specific reduced-order models”). For details on theory please refer to section “A greedy maximin distance sampling approach for the construction of solution subspaces” and section “Hyper reduction of internal force contribution”. Conclusions We presented a framework for projection-based MOR of patient-specific AAA models. A dimensionally reduced model was built by a Galerkin projection on a low-dimensional subspace and a dimensionally reduced as well as hyper reduced model was built by the Galerkin projection and energy-conserving mesh sampling and weighting. Specific atten- tion was dedicated to the MULF prestressing stage, given that the original MULF aims at a calculation of an imprinted deformation gradient and therefore is not suited for snapshot collection. A sampling algorithm relying on the maximin distance criterion was presented for the construction of low-dimensional solution subspaces. Therein, a stopping crite- rion based on subspace angles and an exclusion of subdomains was applied. Finally, three patient-specific computational examples with different complexities were demonstrated. The proposed sampling algorithm led to comparable results in terms of reduced-order basis size as well as number of sampled mesh elements for all three patient-specific compu- tational models. Subsequent experiments on ROM accuracy revealed relative von Mises Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 29 of 31 stress and von Mises strain field errors below 1% for more than 98% of all simulations onavalidationgrid. We conclude that theproposedMOR frameworkisrobustacross patient-specific AAA geometries and parameter domains. Direct Monte Carlo sampling on the dimensionally reduced as well as hyper reduced model was performed calculating the maximum von Mises stress and maximum von Mises strain in the vessel wall. Comparison with the corresponding FOM reference solution revealed a very good match between FOM and ROM kernel-density-estimated probability distributions for the maximum von Mises stress and the maximum von Mises strain in the aortic wall. The proposed sampling algorithm led to appropriate dimensionally reduced (and hyper reduced) models as a conclusion from numerical experiments. However, a certification in terms of an upper bound for the error in the quantities of interest was not presented in this work. Furthermore, motivated by practical reasoning, this paper investigated numerical experiments on parametrizations in terms of two material and one geometric parameter. Deviations from this setup need further validation. We did not include calcification and did not distinguish between healthy and aneurysmatic sections of the simulated vessel in terms of material behavior. These are two example features that would yield a more realistic full-order model. Abbreviations MOR: Model order reduction; AAA: Abdominal aortic aneurysm; FOM: Full-oder model; ROM: Reduced-order model; MULF: Modified updated Lagrangian formulation; ROB: Reduced-order basis; DOF: Degree of freedom; DROM: Dimensionally reduced model; DHROM: Dimensionally reduced and hyper reduced model; ECSW: Energy-conserving mesh sampling and weighting; ILT: Intraluminal thrombus; PVW: Principle of virtual work; POD: Proper orthogonal decomposition; MMD: Maximin distance; SAD: Subspace angle distance. Acknowledgements The authors gratefully acknowledge support and funding by the Leibniz Rechenzentrum München (LRZ) of the Bavarian Academy of Sciences under the contract pn34cu. Authors’ contributions AS developed the idea, performed the software implementation, conducted numerical experiments and wrote the manuscript. MWG fine-tuned the research idea, suggested numerical experiments and revised the paper. All authors read and approved the final manuscript. Funding Open Access funding enabled and organized by Projekt DEAL. Funding was provided by Bayerische Akademie der Wissenschaften (DE), LRZ (Grant No. pn34cu). Availability of data and materials All data referring to the provided examples is contained in the manuscript. Declarations Competing interests The authors declare that they have no competing interests. Received: 19 January 2021 Accepted: 31 July 2021 References 1. Formaggia L, Quarteroni A, Veneziani A. Cardiovascular Mathematics: Modeling and simulation of the circulatory system, vol. 1. : Springer Science & Business Media; 2010. 2. Leach JR, Kao E, Zhu C, Saloner D, Hope MD. On the Relative Impact of Intraluminal Thrombus Heterogeneity on Abdominal Aortic Aneurysm Mechanics. Journal of Biomechanical Engineering. 2019;141(11). 3. Polzer S, Gasser TC. Biomechanical rupture risk assessment of abdominal aortic aneurysms based on a novel proba- bilistic rupture risk index. Journal of The Royal Society Interface. 2015;12(113):20150852. 4. Hemmler A, Lutz B, Kalender G, Reeps C, Gee MW. Patient-specific in silico endovascular repair of abdominal aortic aneurysms: application and validation. Biomechanics and modeling in mechanobiology. 2019;18(4):983–1004. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 30 of 31 5. Perrin D, Badel P, Orgeas L, Geindreau C, rolland du Roscoat S, Albertini JN, et al. Patient-specific simulation of endovas- cular repair surgery with tortuous aneurysms requiring flexible stent-grafts. Journal of the mechanical behavior of biomedical materials. 2016;63:86–99. 6. Lee LC, Ge L, Zhang Z, Pease M, Nikolic SD, Mishra R, et al. Patient-specific finite element modeling of the Cardiokinetix Parachute® device: effects on left ventricular wall stress and function. Medical & biological engineering & computing. 2014;52(6):557–66. 7. Niestrawska JA, Viertler C, Regitnig P, Cohnert TU, Sommer G, Holzapfel GA. Microstructure and mechanics of healthy and aneurysmatic abdominal aortas: experimental analysis and modelling. Journal of The Royal Society Interface. 2016;13(124):20160620. 8. Ockert S, Boeckler D, Allenberg J, Schumacher H. Rupturiertes abdominelles aortenaneurysma. Gefaesschirurgie. 2007;12(5):379–91. 9. Raghavan M, Vorp DA. Toward a biomechanical tool to evaluate rupture potential of abdominal aortic aneurysm: identification of a finite strain constitutive model and evaluation of its applicability. Journal of biomechanics. 2000;33(4):475–82. 10. Biehler J, Gee MW, Wall WA. Towards efficient uncertainty quantification in complex and large-scale biome- chanical problems based on a Bayesian multi-fidelity scheme. Biomechanics and modeling in mechanobiology. 2015;14(3):489–513. 11. Vorp DA. Biomechanics of abdominal aortic aneurysm. Journal of biomechanics. 2007;40(9):1887–902. 12. Maier A. Computational modeling of rupture risk in abdominal aortic aneurysms. Technische Universität München. 13. Gee M, Reeps C, Eckstein H, Wall W. Prestressing in finite deformation abdominal aortic aneurysm simulation. Journal of biomechanics. 2009;42(11):1732–9. 14. Gee MW, Förster C, Wall W. A computational strategy for prestressing patient-specific biomechanical problems under finite deformation. International Journal for Numerical Methods in Biomedical Engineering. 2010;26(1):52–72. 15. Reeps C, Maier A, Pelisek J, Härtl F, Grabher-Meier V, Wall W, et al. Measuring and modeling patient-specific distri- butions of material properties in abdominal aortic aneurysm wall. Biomechanics and modeling in mechanobiology. 2013;12(4):717–33. 16. Raghavan ML, Hanaoka MM, Kratzberg JA, de Lourdes Higuchi M, Da Silva ES. Biomechanical failure properties and microstructural content of ruptured and unruptured abdominal aortic aneurysms. Journal of biomechanics. 2011;44(13):2501–7. 17. Quarteroni A, Manzoni A, Negri F. Reduced basis methods for partial differential equations: an introduction, vol. 92. : Springer; 2015. 18. Carlberg K, Barone M, Antil H. Galerkin v. least-squares Petrov-Galerkin projection in nonlinear model reduction. Journal of Computational Physics. 2017;330:693–734. 19. Farhat C, Avery P, Chapman T, Cortial J. Dimensional reduction of nonlinear finite element dynamic models with finite rotations and energy-based mesh sampling and weighting for computational efficiency. International Journal for Numerical Methods in Engineering. 2014;98(9):625–62. 20. Farhat C, Chapman T, Avery P. Stability and accuracy properties of the energy-conserving sampling and weighting (ECSW) method for the hyper reduction of nonlinear finite element dynamic models. Int J Numer Methods Eng. 2015;102:1077–110. 21. Quicken S, Donders WP, van Disseldorp EM, Gashi K, Mees BM, van de Vosse FN, et al. Application of an adaptive polynomial chaos expansion on computationally expensive three-dimensional cardiovascular models for uncertainty quantification and sensitivity analysis. Journal of biomechanical engineering. 2016;138(12):121010. 22. Sankaran S, Marsden AL. A stochastic collocation method for uncertainty quantification and propagation in cardio- vascular simulations. Journal of biomechanical engineering. 2011;133(3):031001. 23. Biehler J, Wall W. The impact of personalized probabilistic wall thickness models on peak wall stress in abdominal aortic aneurysms. International journal for numerical methods in biomedical engineering. 2018;34(2):e2922. 24. Chang GH, Schirmer CM, Modarres-Sadeghi Y. A reduced-order model for wall shear stress in abdominal aortic aneurysms by proper orthogonal decomposition. Journal of biomechanics. 2017;54:33–43. 25. Negri F. Efficient Reduction Techniques for the Simulation and Optimization of Parametrized Systems. Ecole Poly- technique Fédérale de Lausanne. 2015. 26. Gasser TC, Görgülü G, Folkesson M, Swedenborg J. Failure properties of intraluminal thrombus in abdominal aortic aneurysm under static and pulsating mechanical loads. Journal of vascular surgery. 2008;48(1):179–88. 27. Doll S, Schweizerhof K. On the development of volumetric strain energy functions. J Appl Mech. 2000;67(1):17–21. 28. De Putter S, Wolters B, Rutten M, Breeuwer M, Gerritsen F, Van de Vosse F. Patient-specific initial wall stress in abdominal aortic aneurysms with a backward incremental method. Journal of biomechanics. 2007;40(5):1081–90. 29. Lu J, Zhou X, Raghavan ML. Inverse elastostatic stress analysis in pre-deformed biological structures: demonstration using abdominal aortic aneurysms. Journal of biomechanics. 2007;40(3):693–6. 30. Hemmler A, Lutz B, Reeps C, Kalender G, Gee MW. A methodology for in silico endovascular repair of abdominal aortic aneurysms. Biomechanics and modeling in mechanobiology. 2018;17(4):1139–64. 31. Rutzmoser J. Model Order Reduction for Nonlinear Structural Dynamics. Technische Universität München. 2018. 32. Carlberg K, Tuminaro R, Boggs P. Preserving Lagrangian structure in nonlinear model reduction with application to structural dynamics. SIAM Journal on Scientific Computing. 2015;37(2):B153–84. 33. Haasdonk B, Ohlberger M. Reduced basis method for finite volume approximations of parametrized linear evolution equations. ESAIM: Mathematical Modelling and Numerical Analysis. 2008;42(2):277–302. 34. Quarteroni A, Rozza G, Manzoni A. Certified reduced basis approximation for parametrized partial differential equa- tions and applications. Journal of Mathematics in Industry. 2011;1(1):3. 35. Hesthaven JS, Stamm B, Zhang S. Efficient greedy algorithms for high-dimensional parameter spaces with applications to empirical interpolation and reduced basis methods. ESAIM: Mathematical Modelling and Numerical Analysis. 2014;48(1):259–83. Schein and Gee Adv. Model. and Simul. in Eng. Sci. (2021) 8:18 Page 31 of 31 36. Maday Y, Stamm B. Locally adaptive greedy approximations for anisotropic parameter reduced basis spaces. SIAM Journal on Scientific Computing. 2013;35(6):A2417–41. 37. Jiang J, Chen Y, Narayan A. Offline-enhanced reduced basis method through adaptive construction of the surrogate training set. Journal of Scientific Computing. 2017;73(2–3):853–75. 38. Johnson ME, Moore LM, Ylvisaker D. Minimax and maximin distance designs. Journal of statistical planning and inference. 1990;26(2):131–48. 39. Müller WG. Coffee-house designs. In: Atkinson A, Bogacka B, Zhigljavsky AA, editors. Optimum design 2000. : Springer; 2001. p. 241–8. 40. Pronzato L. Minimax and maximin space-filling designs: some properties and methods for construction. Journal de la Societe Francaise de Statistique. 2017;158(1):7–36. 41. Yondo R, Bobrowski K, Andrés E, Valero E. A review of surrogate modeling techniques for aerodynamic analysis and optimization: current limitations and future challenges in industry. In: Advances in Evolutionary and Deterministic Methods for Design, Optimization and Control in Engineering and Sciences. Springer; 2019. p. 19–33. 42. Van Der Herten J, Van Steenkiste T, Couckuyt I, Dhaene T. Surrogate Modelling with Sequential Design for Expensive Simulation Applications. Computer Simulation. 2017;p.;173. 43. Crombecq K, Laermans E, Dhaene T. Efficient space-filling and non-collapsing sequential design strategies for simulation-based modeling. European Journal of Operational Research. 2011;214(3):683–96. 44. Garud SS, Karimi IA, Kraft M. Design of computer experiments: A review. Computers & Chemical Engineering. 2017;106:71–95. 45. Liu H, Hervas JR, Ong YS, Cai J, Wang Y. An adaptive RBF-HDMR modeling approach under limited computational budget. Structural and Multidisciplinary Optimization. 2018;57(3):1233–50. 46. Cai X, Qiu H, Gao L, Wei L, Shao X. Adaptive radial-basis-function-based multifidelity metamodeling for expensive black-box problems. AIAA Journal. 2017;p. 2424–2436. 47. Golub GH, Van Loan CF. Matrix computations. 3rd ed. : Johns Hopkins University Press; 1996. 48. Lieu T, Lesoinne M. Parameter adaptation of reduced order models for three-dimensional flutter analysis. In: 42nd AIAA Aerospace Sciences Meeting and Exhibit;2004. 49. Lieu T, Farhat C, Lesoinne M. Reduced-order fluid/structure modeling of a complete aircraft configuration. Computer methods in applied mechanics and engineering. 2006;195(41–43):5730–42. 50. Lieu T, Farhat C. Adaptation of aeroelastic reduced-order models and application to an F-16 configuration. AIAA journal. 2007;45(6):1244–57. 51. Akman T. Local improvements to reduced-order approximations of optimal control problems governed by diffusion- convection-reaction equation. Computers & Mathematics with Applications. 2015;70(2):104–31. 52. Bazaz MA, Nahve S, Nabi M, Janardhanan S, Rehman M. Adaptive parameter space sampling in matrix interpolatory pMOR. In: 2015 International Conference on Recent Developments in Control, Automation and Power Engineering (RDCAPE). IEEE; 2015. p. 83–9. 53. Varona MC, Lohmann B, Nabi M. Automatic adaptive sampling in parametric model order reduction by matrix interpolation. In: 2017 IEEE International Conference on Advanced Intelligent Mechatronics (AIM). IEEE; 2017. p. 472–7. 54. Kalra S, Nabi M. TPWL Simulation of Large Nonlinear Circuits Using Subspace Angle Based Adaptive Sampling. IEEE Transactions on Circuits and Systems II: Express Briefs. 2019;67(3):575–9. 55. Martufi G, Lindquist Liljeqvist M, Sakalihasan N, Panuccio G, Hultgren R, Roy J, et al. Local diameter, wall stress, and thrombus thickness influence the local growth of abdominal aortic aneurysms. Journal of Endovascular Therapy. 2016;23(6):957–66. 56. Bruder L, Pelisek J, Eckstein HH, Gee MW. Biomechanical rupture risk assessment of abdominal aortic aneurysms using clinical data: A patient-specific, probabilistic framework and comparative case-control study. PloS one. 2020;15(11):e0242097. 57. de Souza Neto E, Peric´ D, Dutko M, Owen D. Design of simple low order finite elements for large strain analysis of nearly incompressible solids. International Journal of Solids and Structures. 1996;33(20–22):3277–96. 58. Biehler J, Kehl S, Gee MW, Schmies F, Pelisek J, Maier A, et al. Probabilistic noninvasive prediction of wall prop- erties of abdominal aortic aneurysms using Bayesian regression. Biomechanics and modeling in mechanobiology. 2017;16(1):45–61. 59. Heroux MA, Willenbring JM. Trilinos users guide. Sandia National Laboratories. 2003. 60. Davis TA. Algorithm 832: UMFPACK V4. 3–an unsymmetric-pattern multifrontal method. ACM Transactions on Math- ematical Software (TOMS). 2004;30(2):196–9. 61. Virtanen P, Gommers R, Oliphant TE, et al. SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods. 2020;17:261–72. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
"Advanced Modeling and Simulation in Engineering Sciences" – Springer Journals
Published: Sep 4, 2021
Keywords: Abdominal aortic aneurysm; Nonlinear model order reduction; Prestressing; Finite element method
You can share this free article with as many people as you like with the url below! We hope you enjoy this feature!
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.