Abstract
This document presents comprehensive historical accounts on the developments of finite element methods (FEM) since 1941, with a specific emphasis on developments related to solid mechanics. We present a historical overview beginning with the theoretical formulations and origins of the FEM, while discussing important developments that have enabled the FEM to become the numerical method of choice for so many problems rooted in solid mechanics.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
The year 2021 marks the eightieth anniversary of the invention of the finite element method (FEM), which has become the computational workhorse for engineering design analysis and scientific modeling of a wide range of physical processes, including material and structural mechanics, fluid flow and heat conduction, various biological processes for medical diagnosis and surgery planning, electromagnetics and semi-conductor circuit and chip design and analysis, additive manufacturing, and in general every conceivable problem that can be described by partial differential equations (PDEs). The FEM has fundamentally revolutionized the way we do scientific modeling and engineering design, ranging from automobiles, aircraft, marine structures, bridges, highways, and high-rise buildings. Associated with the development of FEMs has been the concurrent development of an engineering science discipline called computational mechanics, or computational science and engineering.
In this paper, we present a historical perspective on the developments of finite element methods mainly focusing on its applications and related developments in solid and structural mechanics, with limited discussions to other fields in which it has made significant impact, such as fluid mechanics, heat transfer, and fluid–structure interaction. To have a complete storyline, we divide the development of the finite element method into four time periods: I. (1941–1965) Early years of FEM; II. (1966–1991) Golden age of FEM; III. (1992–2017) Large scale, industrial applications of FEM and development of material modeling, and IV (2018–) the state-of-the-art FEM technology for the current and future eras of FEM research. Note that this paper may not strictly follow the chronological order of FEM developments, because often time these developments were interwoven across different time periods.
1 (1941–1965) The Birth of the Finite Element Method
The origin of the FEM as a numerical modeling paradigm may be traced back to the early 1940s. In 1941, A. Hrennikof, a Russian-Canadian structural engineer at the University of British Columbia, published a paper in the ASME Journal of Applied Mechanics on his membrane and plate model as a lattice framework [1]. This paper is now generally regarded as a turning point that led to the birth of the FEM. In the paper, he discretized the solution domain into a mesh of lattice structure, which was the earliest form of a mesh discretization.
On May 3rd, 1941, the same year that Hrennikoff published his paper, R. Courant of New York University delivered an invited lecture at a meeting of the American Mathematical Society held in Washington D.C. on his numerical treatment using a variational method to solve a second order PDE, which arises from Saint–Venant’s torsion problem of a cylinder. In this work, Courant systematically used the Rayleigh–Ritz method with a trial function defined on finite triangle subdomains, which is a primitive form of of the finite element method [2].
Courant’s presentation was later published as a paper in 1943 [3]. Similar works of discretization and variational formulations were also reported in the literature, including McHenry [4], Prager and Synge [5], and Synge and Rheinboldt [6]. As Ray Clough commented in his 1980 paper, “One aspect of the FEM, mathematical modeling of continua by discrete elements, can be related to work done independently in the 1940s by McHenry and Hrennikoff-formulating bar element assemblages to simulate plane stress systems. Indeed, I spent the summer of 1952 at the Boeing Airplane Company trying to adapt this procedure to the analysis of a delta airplane wing, the problem which eventually led to the FEM”.
By the early 1950s, several engineers and academics had further developed and extended these early approaches to solve real engineering problems in aeronautical and civil engineering. In parallel but with different emphases, Argyris at the Imperial College London, and M. J. Turner (1950–1956) at Boeing Company, who was later joined by R.W. Clough of UC Berkeley and H.C. Martin of Washington University, developed what we know today as the earliest form of the finite element method (1954), which was called the Matrix Stiffness Method at the time. In a paper published in 1960 [14], R.W. Clough coined the phrase Finite Element Method, and this unassuming and right-to-the-point phrase was an instant hit, bringing out the essence of the method.
With his deep insights and profound vision into Courant’s variational approach, J.H. Argyris developed the energy method for engineering structures [6,7,9], a foundational development enabling FEM for solids. While importantly, Turner, Clough, Martin, and Topp successfully developed FEM interpolants for triangular elements [10], which is suitable for structural parts with arbitrary shape. In some senses, the invention of the triangle element was a “quantum leap”, and hence for a large spectrum of the engineering community, the inception of the FEM is the publication of the landmark paper by Turner et al. [10]. The following is an excerpt from a 2014 document that celebrates the 50th anniversary of the formation of National Academy of Engineering, which is an official account of that part of FEM history: To ensure safety and avoid costly modifications after planes entered production, engineers needed a reliable method for determine in advance whether
their designs could withstand the stresses of fight. M. Jon Turner, head of Boeing’s Structural Dynamics Unit, addressed that problem in the early 1950s by bringing civil engineering professor Ray Clough of the University of California, Berkeley, and Harold Martin of the University of Washington to Boeing for summer ``faculty internships,’’ Collectively, they created a method of structural analysis that Turner applied at Boeing using computers to perform the myriad calculations need to predict real-world performance. That fruitful collaboration led to Clough’s development a few years later of what he named the finite element method (FEM). Clough formed a research group at UC Berkeley that used FEM in a host of analytical and experimental activities, from designing buildings and structures to withstand nuclear blasts or earthquakes to analyzing structural requirements for spacecraft and deep-water offshore drilling. By revolutionizing the application of computer technologies in engineering, FEM continues to help engineers design to this day all sorts of durable, cost-effective structures. Meanwhile, Turner’s efforts at Boeing contributed to the success of its renowned line of commercial jets, beginning in 1958 with the 707 and continuing in 1964 with the 727, which could land on shorter runways and serve more airports. Equipped with three fuel-efficient turbofan engines, the 727 became the workhorse of commercial aviation and helped achieve a threefold increase in U.S. passenger air traffic in the’60s.
Independently and separately, in the early 1960s, Kang Feng of the Chinese Academy of Science also proposed a discretization-based numerical method for variational principles for solving elliptic partial differential equations [11]. As Peter Lax [11, 12] commented, “Independently of parallel developments in the West, he (Feng) created a theory of the finite element method. He was instrumental in both the implementation of the method and the creation of its theoretical foundation using estimates in Sobolev spaces….”, which was one of the first convergence studies of FEMs.
During this period, several great engineering minds were focusing on developing FEMs. J.H. Argyris with his co-workers at the University of Stuttgart; R. Clough and colleagues such as E. L. Wilson and R.L. Taylor at the University of California, Berkeley; O.C. Zienkiewicz with his colleagues such as E. Hinton and B. Irons at Swansea University; P. G. Ciarlet at the University of Paris XI; R. Gallager and his group at Cornell University, R. Melosh at Philco Corporation, B. Fraeijs de Veubeke at the Université de Liège, and J. S. Przemieniecki at the Air Force Institute of Technology had made some important and significant contributions to early developments of finite element methods.
To understand what happened sixty years ago, we quote an excerpt from a FEM history paper by Clough and Wilson [13], in which they recalled:
When Clough presented the first paper using the finite element terminology in 1960 it attracted the attention of his friend, Professor O. C. Zienkiewicz, who was then on the faculty at Northwestern University. A few weeks after the presentation of the paper Zienkiewicz invited Clough to present a seminar on the finite element method to his students. Zienkiewicz was considered one of the world’s experts on the application of the finite difference method to the solution of continuum mechanics problems in Civil Engineering; therefore, Clough was prepared to debate the relative merits of the two methods. However, after a few penetrating questions about the finite element method, Zienkiewicz was almost an instant convert to the method.
Z. Bazant, then a visiting associate research engineer at UC Berkeley, recalled: …… The founding of FEM was paper in the ASCE Conf. on Electronic Computation Clough [14], which was the first to derive, by virtual work, the finite element stiffness matrix of an element of a continuum (a triangular constant strain element). ……. I recall Ray Clough showing to me his 1962 report to the U.S. Engineer District, Little Rock, Corps of Engineers on his analysis of a crack observed in Norfolk Dam [15], during my stay in Berkeley in 1969. I was mesmerized by seeing that 1962 report. It presented a 2D stress analysis of large crack observed in Norfork dam. The dam was subdivided into about 200 triangular elements and provided stress contours for a number of loading cases. …… Clough was at that time way ahead of anybody else.
To collaborate Z. Bazant’s recollection, we cite E.L. Wilson’s recounting of that part of FEM history:
In 1956, Ray, Shirley, and three small children spent a year in Norway at the Ship Research Institute in Trondheim. The engineers at the institute were calculating stresses due to ship vibrations to predict fatigue failures at the stress concentrations. This is when Ray realized his element research should be called the Finite Element Method which could solve many different types of problems in continuum mechanics. Ray realized the FEM was a direct competitor to the Finite Difference Method. At that time FDM was being used to solve many problems in continuum mechanics. His previous work at Boeing, the Direct Stiffness Method, was only used to calculate displacements not stresses.
In the fall semester of 1957, Ray returned from his sabbatical leave in Norway and immediately posted a page on the student bulletin board asking students to contact him if they were interested in conducting finite element research for the analysis of membrane, plate, shell, and solid structures. Although Ray did not have funding for finite element research, a few graduate students who had other sources of funds responded. At that time, the only digital computer in the College of Engineering was an IBM 701 that was produced in 1951 and was based on vacuum tube technology. The maximum number of linear equations that it could solve was 40. Consequently, when Ray presented his first FEM paper in September 1960, “The Finite Element Method in Plane Stress Analysis,” at the ASCE 2 nd Conference on Electronic Computation in Pittsburgh, Pennsylvania, the coarse-mesh stress-distribution obtained was not very accurate. Therefore, most of the attendees at the conference were not impressed. After the improvement of the speed and capacity of the computers on the Berkeley campus, Professor Clough’s paper was a very fine mesh analysis of an existing concrete dam. The paper was first presented in September 1962 at a NATO conference in Lisbon, Portugal. Within a few months, the paper was republished in an international Bulletin, which had a very large circulation, as “Stress Analysis of a Gravity Dam by the Finite Element Method”, (with E. Wilson), International Bulletin RILEM, No. 10, June 1963.
The Lisbon paper reported on the finite element analysis of the 250-foot-high Norfork Dam in Arkansas, which had developed a vertical crack during construction in 1942. The FEM analysis correctly predicted the location and size of the crack due to the temperature changes and produced realistic displacements and stresses within the dam and foundation for both gravity and several hydrostatic load conditions. Because of this publication, many international students and visiting scholars came to Berkeley to work with Professor Clough. Also, he freely gave the FORTRAN listing of their finite element analysis computer program to be used to evaluated displacement and stresses in other two-dimensional plane structures with different geometry, materials, and loading. Therefore, professional engineers could easily use the powerful new FEM to solve for the stress distributions in their structural engineering problems in continuum mechanics. However, he did not capitalize on his success in the development of the FEM. He returned to the task of building the earthquake engineering program at Berkeley – the task he given when he was first hired in 1949.
For their decisive contributions to the creation and developments of FEM, R. W. Clough was awarded the National Medal of Science in 1994 by the then vice-president of the United States Al Gore, while O. C. Zienkiewicz was honored as a Commander of the Order of the British Empire (CBE). Today, the consensus is that J.H. Argyris, R.W. Clough and O. C. Zienkiewicz made the most pivotal, critical, and significant contributions to the birth and early developments of finite element method following an early contribution to its mathematical foundation from R. Courant.
It is worth noting that E.L. Wilson of UC Berkeley was the first person to develop finite element open-source software. An excerpt from Clough and Wilson’s paper in 1999 stated 13: In 1958 Wilson, under the direction of Clough, initiated the development of an automated finite element program based on the rectangular plane stress finite element developed at Boeing. After several months of learning to program the IBM 701, Wilson produced a limited capacity, semiautomated program which was based on the force method. An MS research report was produced, which has long since been misplaced, with the approximate title of Computer Analysis of Plane Stress Structures. …… In 1959 the IBM 704 computer was installed on the Berkeley Campus. It had 32K of 32-bit memory and a floating-point arithmetic unit which was approximately 100 times faster than the IBM 701. This made it possible to solve practical structures using fine meshes.
It is also worth mentioning that, Oden [16], a senior structural engineer in the research and development division of General Dynamics Corporation at Fort Worth at the time, wrote a 163-page comprehensive technical report with G.C. Best, in which they developed a long list of solid and structural finite elements, including tetrahedral, hexahedral, thin plate, thick plate, plate element with stringers or stiffeners, composite sandwich plate elements, and shallow shell elements [17]. In fact, Oden and Best wrote one of the first general purpose FEM computer codes at the time. The Fortran FEM code developed by Oden and Best had an element library that includes elements for 3D elasticity, 2D plane elasticity, 3D beam and rod elements, composite layered plate and shell elements, and elements for general composite materials. Their work also included hybrid methods and stress based FEMs, which may be even earlier than those of Pian [18]. Moreover, Oden and Best’s FEM code was also able to handle FEM eigenvalue modal analysis, and numerical integration over triangle and tetrahedra elements, and it had linear system solvers for general FEM static analysis that were among the most effective at that time (see [19]). This FEM computer code was used for many years in aircraft analysis and design in the aerospace and defense industry.
It appears that, I.C. Taig [20] of English Electric Aviation first introduced the concept of the isoparametric element when he used the matrix-displacement method to conduct stress analysis of aerospace structures, which was later formally dubbed ``isoparametric element’’ by Ergatoudis et al. [21].
It should be mentioned that there are some other pioneers who made some significant contributions in the early developments of FEM, such as Levy [22], Comer [23], Langefors [24], Denke [25], Wehle and Lansing [26], Hoff et al. [27], and Archer [28], among others. These individuals came together made remarkable and historic contributions to the creation of finite element method. Among them, some notable contributions were made by J. S. Przemieniecki (Janusz Stanisław Przemieniecki), who was a Polish engineer and a professor and then dean at the Air Force Institute of Technology in Ohio in the United States from 1961 to 1995. Przemieniecki conducted a series pioneering research works on using FEMs to perform stress and buckling analyses of aerospace structures such as plates, shells, and columns (see Przemieniecki [29], Przemieniecki and Denke [30], Przemieniecki [28,29,30,31,32,33]).
In terms of worldwide research interest, by 1965, FEM research had become a highly active field, with the total number of papers published in the literature exceeding 1000. During this period, there were two seemingly unrelated events for FEM development, which significantly affected future FEM developments. These events were the discovery of mixed variational principles in elasticity. In 1950, E. Reissner [34] rediscovered E. Hellinger’s mixed variational principle from 1914 [35], in which both the displacement field and the stress field are the primary unknowns. This variational principle is called the Hellinger–Reissner variational principle. Shortly after, Hu [36] and Washizu [37] proposed a three-field mixed variational principle in elasticity, which was called the Hu-Washizu variational principle. As early as 1964, Pian [18] recognized the potential of using these variational principles to formulate Galerkin weak form-based FE formulations and proposed the assumed stress FEM. This began the use of mixed variational principles to formulate Galerkin FEMs, which was followed by the assumed strain FEM developed by J. Simo and his co-workers in the later period of FEM developments.
2 (1966–1991) The Golden Age of the Finite Element Method
The mid 1960s saw rapid developments in finite element method research and applications. As T.J.R. Hughes recalled, “I first heard the words ‘the Finite Element Method” in 1967—which changed my life. I started to read everything that was available and convinced my boss to start the Finite Element Method Development Group, which he did. Dr. Henno Allik was Group Leader, and I was the Group, then we added programmers. In one year, we had a 57,000-line code, GENSAM (General Structural Analysis and Matrix Program, or something like that). That was 1969, and the code was continually developed thereafter and may still be in development and use at GD/Electric Boat and General Atomics, originally a division of General Dynamics.”
Starting from the end of 1960s, the rigorous approximation theory that underpins the FEM started to be developed (e.g. Aubin [38], Zlamal [39], Birkhoff [40], Nitsche [41], Aziz [42], Bubuska [43], Bubuska and Aziz [44], Dupont [45], Douglas and Dupont [46]. Nitsche and Schatz [47], and Bubuska [48]). This movement was first highlighted by the proof of optimal and superconvergence of FEMs. This attracted the interest of some distinguished mathematicians all over the world, including G. Birkhoff, M.H. Schultz, R.S. Varaga, J. Bramble, M. Zlamal, J. Cea, J.P. Aubin, J. Douglas, T. Dupont, L.C. Goldstein, LR. Scott, J. Nitsche, A.H. Schatz, P.G. Cialet, G. Strang, G. Fix, JL. Lions, M. Crouzeix, P.A. Raviart, and I. Babuska, A.K. Aziz, and J.T. Oden (see Bramble, Notsche and Schatz [49], Bubuska, Oden, and Lee [50]) . Some notable results developed for the proof of FEM convergence are the Cea lemma and the Bramble-Hubert lemma. The fundamental work of Nitsche [41] on L∞ estimates for general classes of linear elliptic problems also stands out as one of the most important contributions for mathematical foundation of FEM in 1970s. It may be noted that unlike other mathematics movements, the convergence study of FEMs was an engineering-oriented movement. The mathematicians soon found that, in practice, engineers were using either non-conforming FEM interpolants or numerical quadrature that violates variational principles or the standard bilinear form in Hilbert space. G. Strang referred to these numerical techniques as “variational crimes”. To circumvent complicated convergence proofs, the early FEM patch tests were invented by B. Iron and R. Melosh, which were proven to be instrumental for ensuring convergence to the correct solution.
Following T.H.H. Pian’s invention of the assumed stress element, attention shifted to the mixed variational principle based FEM [51, 52]). In 1965, L.R. Hermann [53] proposed a mixed variational principle for incompressible solids. However, most mixed variational principles are not extreme variational principles, and thus suffer from numerical instability. In early 1970s, I. Babuska and F. Brezzi discovered their groundbreaking results, known today as the Babuska-Brezzi condition, or the LBB condition, giving tribute to O. Ladyzhenskya—a Russian mathematician, who provided the early insight of this problem. The so-called LBB condition, or the inf–sup condition, provides a sufficient condition for a saddle point problem to have a unique solution that depends continuously on the input data; thus, it provides a guideline to construct shape functions for the mixed variational principle-based FEM (see: [54]).
Entering the 1970s, FEM development began to focus on using FEM to simulate the dynamic behavior of structures, including crashworthiness in the automotive industry. Various time integration methods had been developed, including the Newmark-beta method, Wilson-theta method [55], the Hilbert-Hughes-Taylor alpha method [56], the Houbolt integration algorithm, and the explicit time integration algorithm [57].
Since the early 1970s, FEM explicit time integration methods had been used to solve various engineering problems, e.g., dynamic contact problems [58], dynamics problems of solid structure [59], fluid dynamics problems [60], and it was extensively used in large scale FEM Lagrangian hydrocodes developed in Lawrence Livermore National Laboratory [61]. However, an important development came in the late 1970s when T. Belytschko, K. C. Park, and later TJR Hughes proposed using explicit or implicit-explicit, or implicit time integration with damping control to solve nonlinear structural dynamics problems. Other notable contributors to this key development are A. Combescure, A. Gravouil among others (see e.g., [62]). It turned out that the explicit time integration was a game changer for the automotive industry, establishing FEM technology as the main tool of passenger vehicle design and crashworthiness analysis. By the end of 1980s, there were thousands of workstations running explicit time integration-based FEM codes in the three major automakers in the United States.
One of the main FEM research topics in the 1980s was using FEM techniques to solve the Navier–Stokes equation as an alternative to finite difference and finite volume methods. Starting the early 1970s, J.T. Oden and his co-workers had begun working on FEM solutions for fluid dynamics [63]. The main challenge in using FEM solving fluid dynamics problems is that the Navier–Stokes equation is not an elliptical partial different equation, and the minimization or variational principle-based Petrov–Galerkin procedure may suffer both stability as well as convergence issues. To resolve these issues, T.J.R. Hughes, and his co-workers such as A.N. Brooks and T.E. Tezduyar developed the streamline upwind/Petrov Galerkin method and later Stabilized Galerkin FEM to solve Navier–Stokes equations under various initial and boundary conditions [64]. Furthermore, Hughes and co-workers later developed space–time FEMs and variational multiscale FEMs (see Hughes and Tezdyuar [65] Brooks and Hughes [66], Mizukami and Hughes [67], and Hughes [68]). For this work, from 1986 to 1991, Hughes and his co-workers such as L.P. Franca and others wrote a ten-part series on FEM formulations for computational fluid dynamics Hughes et al. [69,70,71,71] and Shakib et al. [72]. Several years later, Hughes developed a Green function-based subgrid model acting as a stabilized finite element method [68].
Towards the mid-1980s, advanced FEM mesh generation techniques have been developed, which incorporate various solid modeling techniques by using interactive computer graphics and adaptive mesh generator [73], Bennett and Botkin [74] as well as improved quadtree approach to generate FEM meshes for complex geometric shapes and objects [75]. Today, FEM mesh generation has become an integrated part of solid modeling and engineering design, which has the capabilities of automatic node insertion and refinement Wordenweber [76] and Ho-Le [77].
Among the many advances in FEM technologies in 1980s, the most notable may belong to J. Simo at the University of California, Berkeley and later at Stanford University. Simo and Taylor [78] developed the consistent tangent operator for computational plasticity, which was a milestone after the original concept of consistent linearization proposed by Hughes and Pister [79]. Moreover, after the Hughes-Liu 3D degenerated continuum shell and beam elements and the Belytchko-Tsay single-point element, Simo and his co-workers, such as L. Vu-Quoc and D.D. Fox, developed geometrically exact beam and shell theories and their FEM formulations (see Simo and Vu-Quoc [80], Simo and Fox [81], and Simo et al. [82]. Moreover, Simo and his co-workers such as MS Rifai and F. Armero also developed various assumed strain or enhanced strain methods for mixed variational formulations (see Simo and Rifai [83] and Simo and Armero [84]). It should be noted that E. Ramm and his colleagues at the University of Stuttgart have also made significant contributions on geometrically nonlinear shell element formulations over a span of more than thirty years e.g., Andelfinger and Ramm [85] and Bischoff and Ramm [86].
Another highlight of FEM technology is the development of FEM solvers for fluid–structure interaction. From the mid-1970s to early 1990s, there was an urgent need to develop techniques to solve large-scale fluid–structure interaction problems in the aerospace and civil engineering industries. A class of FEM fluid–structure interaction solvers were developed, and some early contributors include J. Donea, A. Huerta (see [87,88,89,89]), and the Hughes-Liu-Zimmermann Arbitrary Lagrangian–Eulerian (ALE) fluid–structure FEM formulation (see [90, 91]), which describes the moving boundary problem. ALE-based FEM simulations were used due to their ability to alleviate many of the drawbacks of traditional Lagrangian-based and Eulerian-based FEM formulations.
When using the ALE technique in engineering modeling and simulations, the computational mesh inside the domains can move arbitrarily to optimize the shapes of elements, while the mesh on the boundaries and interfaces of the domains can move along with materials to precisely track the boundaries and interfaces of a multi-material system. The invention of the ALE FEM may be credited to Hirt, Amsden, and Cook [92]. C. Farhat was the first person to use a large-scale parallel ALE-FEM solver to compute fluid–structure interaction problems [93]. He and his group systematically applied FEM-based computational fluid dynamics (CFD) solvers for aircraft structure design and analysis. They developed the finite element tearing and interconnecting (FETI) method for the scalable solution of large-scale systems of equations on massively parallel processors. Today, some fundamental ALE concepts have also been applied to numerical modeling engineering fields other than FEM, such as meshfree modeling. For example, to alleviate tensile instability and distorted particle distributions in smoothed particle hydrodynamic (SPH) simulations, a so-called shifting technique has been adopted in many of today’s SPH simulations (e.g., Oger et al. [94]).
The FEM fluid–structure interaction research had a major impact on many practical applications, such as providing the foundation for the patient specific modeling of vascular disease and the FEM-based predictive medicine later developed by C.A. Taylor and T.J.R. Hughes and their co-workers in the mid-1990s (see Taylor et al. [95]). Holzapfel, Eberlein, Wriggers, and Weizsäcker also developed large strain FEM formulations for soft biological membranes (see Holzapfel et al. [96]).
Another major milestone in the development of FEMs was the invention and the development of nonlinear probabilistic or random field FEMs, which was first developed by W. K. Liu and T. Belytschko in the late 1980s. (e.g., Liu et al. [97]). By considering uncertainty in loading conditions, material behavior, geometric configuration, and support or boundary conditions, the probabilistic FEM provided a stochastic approach in computational mechanics to account for all these uncertain aspects, which could then be applied in structure reliability analysis. The random field FEM research has become crucial in civil and aerospace engineering and the field of uncertainty quantification.
In the early 1980s, M.E. Botkin at General Motors research Lab [98], and N. Kikuchi and his group at the University of Michigan developed structural shape optimization FEM for the automotive industry (see [99], [100], [101]). Other contributors include M. H. Imam from Uman Al-Qura University [102]). This preceded the seminal 1988 paper of Bendsoe and Kikuchi, who developed a homogenization approach to finding the optimal shape of a structure under prescribed loading. Later developments in topology optimization were driven by Gengdong Cheng, Martin Bendsoe, and his student, Ole Sigmund.
One of the driving forces in the FEM development during in the early decades was the safety analysis of big dams, at first, then of concrete nuclear reactor vessels for gas-cooled reactors, [103], nuclear containments, hypothetical nuclear reactor accidents [104, 105] and of tunnels and of foundations for reinforced concrete structures. To simulate concrete failure, the vertical stress drops in FEMs, and progressive softening technique were introduced already in 1968. However, the spurious mesh sensitivity and its impact on strain localization was generally overlooked until demonstrated mathematically by Bazant [105], because strain softening states of small enough test specimens in stiff enough testing frames are stable. Numerically this was demonstrated by crack band FEM calculations in Bazant and Cedolin [106], where it was also shown that spurious mesh sensitivity causing the sudden stress drop can be avoided by adjusting the material strength to ensure the correct energy release rate. Hillerborg et al. [107] avoided mesh sensitivity by using an interelement cohesive softening called the fictitious crack model. Despite the success of the early FEM calculations, the concept of progressive strain-softening damage was not generally accepted by mechanicians until its validity and limitations were demonstrated by Bazant and Belytschko [108] and Bazant and Chang [109]. They showed that the existence of elastic unloading stiffness (previously ignored) makes waves propagation in a strain-softening state possible. In 1989, Lubliner, Oliver, Oller, and Onate ([110] developed a plastic-damage theory-based FEM formulation to model concrete materials by introducing internal variables, which has the capacity of modeling concrete material degradation and cracking. Today, the multiscale based homogenization and damage analysis method is the state-of-the-art FEM modeling for concrete materials [111].
In the mid-1980s, the mesh sensitivity issue in calculating strain softening or strain localization problems in computational plasticity became a challenging topic. It was eventually accepted that the partial differential equations that are associated with the classical plasticity become ill-posed after the material passes the yield point and enters the softening stage. This especially became a dire situation when civil engineers applied FEM to solve complex structural and geotechnical engineering problems, which involved complex plastic deformations of concrete, rock, soil, clay, and granular material in general. Bazant and others realized that this is because the classical continuum plasticity theory lacks an internal length scale. To remedy this problem, starting from the middle 1980s, many efforts were devoted to establishing FEM formulations of nonlocal (Bazant et al. [112]), strain-gradient, strain-Laplacian media, micropolar or Cosserat continua, because they provided an internal length scale, allowing FEM simulations to capture, in a mesh-independent manner, the strain softening, strain localization or shear band formation. Pijaudier-Cabot and Bazant [113] developed an effective nonlocal FEM in which nonlocality is applied only to the damage strain. Other influential and representative works in this topic are from De Borst [114], Peerlings, De Borest and others [115]. Steinmann and Willam [116], Dietsche et al. [117], Steinmann [118], and Iordache and Willam [119].
In 1976, TJR Hughes, R.L. Taylor, J.L. Sackman, A. Curnier, and W. Kanoknukulchai published a paper entitled “A finite element method for a class of contact-impact problems.” [120]. This is one of the earliest FEM analyses in computational contact mechanics. It is the very first work on FEM modeling of dynamic contact and impact problems, and it plays an important role in the simulation accuracy for engineering problems involving interaction between different continuous objects. Examples include sheet metal forming, target impact and penetration, and interaction between pavement and tires. Developing accurate FEM contact algorithms has been a focal point since the 1980s. Various FEM contact algorithms have been developed, and some main contributors are N. Kikuchi, J.T. Oden, J. Simo, P, Wriggers, R.L. Taylor, P. Papadopoulos, and T.A. Laursen. FEM contact algorithm research remained an active research topic until late 1990s (see Simo et al. [121], Simo and Lausen [122], Kikuchi and Oden [123], and Papadopoulos and Taylor [124]).
In this period, one interesting emerging area was the development of FEM exterior calculus by Arnold et al. [125]. The FEM exterior calculus uses tools from differential geometry, algebraic topology, and homological algebra to develop FEM discretizations that are compatible with the underlying geometric, topological, and algebraic structures of the problems that are under consideration.
3 (1992–2017) Broad Industrial Applications and Materials Modeling
The first major event in FEM development this period was the formulation of the Zienkiewicz-Zhu error estimator [126], which was a major contribution to the mathematical approximation theory of FEMs in the 1990s. The Zienkiewicz-Zhu posteriori error estimators provide the quality control of a FEM solution with an optimal use of computational resources by refining the mesh adaptively. The idea and spirit of Zienkiewicz-Zhu was further carried out by Ainsworth and Oden (see [127, 128, 129]), and today using posteriori error estimation to improve the quality has been elevated to the height of Bayesian inference and Bayesian update. This research topic is now intimately related with what is now called validation and verification (V&V).
Since the late 1970s, Szabo [130] and Babuska [131, 132] started to develop hp versions of FEMs based on piecewise-polynomial approximations that employ elements of variable size (h) and polynomial degree (p). They discovered that the FEM converges exponentially when the mesh is refined using a suitable combination of h-refinements (dividing elements into smaller ones) and p-refinements (increasing their polynomial degree). The exponential convergence makes the method a very attractive choice compared to most other FEMs, which only converge with an algebraic rate. This work continued until the late 1990s, spearheaded by M. Ainsworth, L. Demkowicz, J.T. Oden, C.A. Duarte, O.C. Zienkiewicz, and C.E. Baumann (see Demkowicz et al. [133], Demkowicz et al. [134], Demkowicz et al. [135], Oden et al. [136], Oden et al. [137], Baumann and Oden [138]. J. Fish. [139] also proposed a s-version FEM by superposing additional mesh(es) of higher-order hierarchical elements on top of the original mesh of C0 FEM discretization, so that it increases the resolution of the FEM solution.
To solve material and structural failure problems, research work in the 1990s focused on variational principle based discretized methods to solve fracture mechanics problems or strain localization problems. In 1994, Xu and Needleman [140] developed a FEM cohesive zone model (CZM) that can simulate crack growth without remeshing, which was later further improved by M. Ortiz and his co-workers, e.g. Camacho and Ortiz [141] and Ortiz and Pandolfi [142], who later used CZM FEM to solve fragmentation and material fatigue problems [143]. It should be noted that long before the invention of cohesive zone model, Pietruszczak and Mroz [144] developed the first cohesive FEM for shear fracture in soil. Later, Bazant’s group at Northwestern University developed various interface FEMs, such as the microplane model, to study size effects of concrete and other brittle composite materials (Brocca and Bazant [145], Caner and Bazant [146]). These models became a standard tool for simulating missile impact and explosions at, e.g., ERDA Vicksburg. A microplane FEM model for fiber composites has been developed for Chrysler and Ford Co. to compare various designs of automobile crush-cans (Smilauer et al. [147]). An anisotropic poromechanical microplane model has been formulated and used for FEM analysis of hydraulic fracturing (Rahimi et al. [148]).
To alleviate mesh bias issues in modeling material fracture and damage problems, T. Belytschko and WK. Liu developed meshfree particle methods, namely the element-free Galerkin (EFG) method [149] and the reproducing kernel particle method (RKPM) [150, 151], which are based on the moving least square method and the wavelet multiresolution analysis, respectively. RKPM provides consistency and thus convergence enhancements as compared to the popular smoothed particle hydrodynamics (SPH) method. Li et al. [152, 153] successfully employed meshfree Galerkin methods to accurately simulate adiabatic shear band formation and propagation with minimum mesh adaptation. At the same time, Duarte and Oden developed the so-called hp-Cloud method, Onate et al. developed a finite point method [154], and Atluri and Zhu [155] proposed a meshless local Petrov–Galerkin (MLPG method, among many other meshfree methods. Fleming and Belytschko also showed that singularity functions could be included in the approximation functions to greatly improve simulations involving fracture mechanics (see Fleming et al. [156]).
One of the most challenging problems in the development of meshfree Galerkin methods is how to integrate the weak form, because the meshfree interpolants are highly irregular and it is difficult to make them variationally consistent. In 2001, J. S. Chen and his co-workers proposed a stabilized conforming nodal integration method for meshfree RKPM method, which is not only simple and stable, but also variationally consistent with the Galerkin weak formulation (see Chen et al. [157]).
Shortly after the meshfree method developments, I. Babuska and his co-workers developed the partition of unity finite element method (PUFEM), which was later coined generalized finite element method (GFEM) (see Melenk and Babuska [158] and Babuska and Melenk [159]). PUFEM is a powerful method because it can be used to construct FE spaces of any given regularity, which is a generalization of the h, p, and hp versions of the FEM, as well as providing the ability to embed an analytic solution into the FEM discretization instead of relying upon a generic polynomial interpolant.
A significant breakthrough in computational fracture mechanics and FEM refinement technology came in the late 1990s, when Belytschko and his co-workers, including Black, Moes, and Dolbow, developed the eXtended finite element (X-FEM) (see [160, 161], which uses various enriched discontinuous shape functions to accurately capture the morphology of a cracked body without remeshing. Because the adaptive enrichment process is governed by the crack tip energy release rate, X-FEM provides an accurate solution for linear elastic fracture mechanics (LEFM). In developing X-FEM, T. Belytschko brilliantly utilized the PUFEM concept to solve fracture mechanics problems without remeshing. Entering the new millennium, Bourdin, Francfort, Marigo developed a phase-field approach for modeling material fracture [162]. Almost simultaneously, Karma and his co-workers [163], [164]) also proposed and developed the phase-field method to solve crack growth and crack propagation problems, as the phase field method can accurately predict material damage for brittle fracture without remeshing. The main advantage of the phase field approach is that by using the Galerkin FEM to solve the continuum equations of motion as well as a phase equation, one can find the crack solution in continuum modeling without encountering stress singularity as well as remeshing, and the crack may be viewed as the sharp interface limit of the phase field solution. Some of the leading contributors for this research are Bourdin, Borden, Hughes, Kuhn, Muller, Miehe, Landis, among others (see Bourdin and Chambolle [165], Kuhn and Müller [166]. Miehe et al. [167], and Borden et al. [168], Wilson et al. [169], and Pham et al. [170]).
As mentioned before, the main reason for the huge success of FEMs is their broad applicability to engineering analysis and design across scientific disciplines. On the other hand, most mechanical engineering designs are performed by using various computer-aided design (CAD) tools, such as solid modeling. To directly blend the FEM into CAD design tools, T.J.R. Hughes, and his co-workers such as J.A. Cottrell and Y. Bazilevs [171], [172] developed the isogeometric analysis (IGA) FEM, which established the Galerkin variational weak formulation in the control mesh and uses the non-uniform rational basis spline (NURBS) functions as the FEM shape function to solve the problem at design stage. IGA-FEM method successfully integrates FEM into conventional NURBS-based CAD tools, without converting data between CAD and FEA packages in analyzing new designs during development stage.
Due to the emergence of nanotechnology, various multiscale methods have been developed to couple atomistic methods such as molecular dynamics and density functional theory (DFT) and other ab initio methods with continuum scale FEMs. The most notable contributions in this area are hand-shake method [173] quasi-continuum FEM [174], and the bridging scale method developed by Wagner and Liu [175]. In 2007, Gavini, Bhattacharya, and Ortiz developed quasi-continuum orbital-free density-functional theory (DFT) FEM for multi-million atom DFT calculations (see Gavini et al. [176]).
Broadly speaking, there has been a rapid development in computational homogenization methods since the late 1990s, which is an alternate approach to obtaining continuum-scale properties based on smaller scale microstructures. The computational homogenization method or the FEM homogenization methods for composite materials may be divided into two main categories: (1) Computational asymptotic homogenization method or multiscale computational homogenization, which is aimed for modeling composite materials with periodic microstructure. The pioneer of multiscale homogenization FEM may be credited to T.Y. Hou and his co-workers [177], and other earlier contributors are N. Kikuchi [101], Ghosh et al. [178], Fish et al. [179, 180]) and M. S. Shephard, (2) Computational micromechanics method, which is mainly aimed for composite materials with random microstructure, even though it may also be applied to materials with periodic microstructure, where the main contributors are: P. Suquet et al. at the French National Centre for Scientific Research (CNRS), Dvorak et al.[181] and J. Fish at Rensselaer Polytechnic Institute and C. Miehe and his co-workers (e.g. [182]) at the University of Stuttgart. During the same period, many novel FEM fluid–structure interaction solvers have been developed, for instance, the immersed FEM developed by Zhang et al. [183], which was motivated by the immersed boundary method pioneered by C. Peskin of Courant Institute of Mathematical Sciences, New York University.
An important FEM application area emerged with the developments of computational plasticity. The early finite element computational plasticity formulation is based on hypoelastic–plastic rate formulation. To satisfy the objectivity requirement, T.J, R. Hughes and J. Winget first proposed the so-called incremental objectivity (see Hughes and Winget [184]), which was probably in the first time the esoteric continuum mechanics theory was applied to practical FEM computational formulations and computations, and it in turn promoted the development of nonlinear continuum mechanics in 1980s and 1990s. Soon afterwards. Simo and Hughes then extended Hughes-Winget incremental objectivity algorithm to the finite deformation case in computational plasticity (see Simo and Hughes [185]). The notion of consistency between the tangent stiffness matrix and the integration algorithm employed in the solution of the incremental problem was introduced by Nagtegaal [186] and Simo and Taylor [78]. Consistent formulations have been subsequently developed for finite deformation plasticity by Simo and Oritz [187] (1988) within the framework of multiplicative decomposition of the deformation gradient and hyperelasticity. Also in 1980s, based on the Gurson model, Tvergaard and Needleman [188] developed the FEM based Gurson-Tvergaard-Needleman model, which is probably the most widely used FEM computational plasticity constitutive model used in material modeling, though recently attention has turned to developing machine-learning based or data-driven computational plasticity models e.g., F. Chinesta, et al. [189], and the unsupervised machine learning data-driven finite element methods, called as the Self-consistent Clustering Analysis (SCA) (see Z. Liu et al. [190, 191]).
An important advance of the FEM is the development of the crystal plasticity finite element method (CPFEM), which was first introduced in a landmark paper by Pierce et al. [192]. In the past almost four decades, there are numerous researchers who have made significant contributions to the subject, for example, A. Arsenlis and DM. Parks from MIT and Lawrence Livermore National Laboratory [193], [194], [195], Dawson et al. at Cornell University (Quey et al. [196]; Mathur and Dawson [197]; Raabe et al. at Max-Planck-Institute fur Eisenforschung [196,197,198,201], among others. Based on crystal slip, CPFEM can calculate dislocation, crystal orientation and other texture information to consider crystal anisotropy during computations, and it has been applied to simulate crystal plasticity deformation, surface roughness, fractures and so on. Recently, S. Li and his co-workers developed a FEM-based multiscale dislocation pattern dynamics to model crystal plasticity in single crystal [202, 203]. Yu et al., [204] reformulated the self-consistent clustering analysis (SCA) for general elasto-viscoplastic materials under finite deformation. The accuracy and efficiency for predicting overall mechanical response of polycrystalline materials are demonstrated with a comparison to traditional full-field FEMs.
In 2013, a group of Italian scientists and engineers led by L. Beirão da Veiga and F. Brezzi proposed a so-called virtual element method (VEM) (see Beirao et al. [205, 206]). The virtual element method is an extension of the conventional FEM for arbitrary element geometries. It allows the polytopal discretizations (polygons in 2-D or polyhedra in 3-D), which may be even highly irregular and non-convex element domains. The name virtual derives from the fact that knowledge of the local shape function basis is not required, and it is in fact never explicitly calculated. VEM possesses features that make it superior to the conventional FEM for some special problems such as the problems with complex geometries for which a good quality mesh is difficult to obtain, solutions that require very local refinements, and among others. In these special cases, VEM demonstrates robustness and accuracy in numerical calculations, when the mesh is distorted.
As early as 1957, R. Clough introduced the first graduate FEM course at UC-Berkeley, and since then FEM courses at both graduate and undergraduate levels have been added into engineering higher education curriculums in all the major engineering schools and Universities all over the world. As J.T. Oden recalled in his 1963 paper, “I went on to return to academia in 1964 and among my first chores was to develop a graduate course on finite element methods. At the same time, I taught mathematics and continuum mechanics, and it became clear to me that finite elements and digital computing offered hope of transforming nonlinear continuum mechanics from a qualitative and academic subject into something useful in modern scientific computing and engineering.”.
By the end of 2015, there have been more than several hundred monographs and textbooks on the FEM published in dozens of languages worldwide. An exposition of FEM mathematical theory by Strang and Fix [207] was among the earliest of those FEM books. J.T. Oden and his collaborators such as GF Carey and JN Reddy and others wrote a five-volume FEM monograph in 1980s (see [167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,213, 214]). Other influential finite element books or monographs are those by Zienkiewicz and Cheung [215], and Zienkiewicz and Taylor [216], ora later Zienkiewicz, Taylor and Zhu [217], TJR. Hughes [218], Cook et al. [219] or later Cook [220]), Bathe [221], and the Nonlinear FEM monographs by Belytschko et al. [222] and by De Borst et al. [223], among others, all have made major impacts on FEM educations and applications. Among all these FEM monographs and textbooks, the book by Zienkiewicz and Taylor [216] or Zienkiewicz et al. [217] probably have had most impacts on FEM technology popularization, which may be because Taylor wrote a FEM research computer program code named FEAP, which was placed in the appendix of that book, providing an immediate guidance and instruction on how to implement FEM in computers.
The ready availability of FEM textbooks and tutorials, along with the large FEM software market, have made the FEM accessible to users across academia and industry. The development of FEM software technology started in the early1960s. In 1963, E.L. Wilson and R. Clough developed a structural mechanics computing code called Symbolic Matrix Interpretive System, SMIS, which was initially intended to fill the gap between the matrix method and hand calculating in structure mechanics. It turns out the development of SMIS led to the birth of FEM software. Then, based on SMIS, Wilson initiated and developed a general-purpose static and dynamic Structural Analysis Program, SAP. In late 1960s and the early 1970s, K, J. Bathe developed the nonlinear FEM code ADINA based on SAP IV and NONSAP. Today, the brand name SAP2000 has become synonymous with the state-of-the-art FEM structural analysis and design methods since its introduction over 55 years ago.
At the same period, to compete with the Soviet Union’s space program, NASA developed its own FEM code called NASTRAN (NASA STRuctural ANalysis Program). The first version of NASTRAN was called COSMIC Nastran, which debuted in 1969, with a key figure in its development being R. H. MacNeal. As early as 1963, R. H. MacNeal founded the MacNeal-Schwendler Software Corporation (MSC) along with R. Schwendler. Under his leadership, MSC developed its first structural analysis software called SADSAM (Structural Analysis by Digital Simulation of Analog Methods), which showed the early form of FEM analysis software technology. In response to NASA’s request for proposals in 1965 for a general purpose structural analysis program, Dr. MacNeal contributed significantly to the early efforts of the aerospace industry by successfully simulating on-the-ground physical testing through computing to deliver the right answers and physics needed to take humans to the moon. In 1971, MSC Software released a commercial version of Nastran, named MSC Nastran.
About the same time in 1960s, J. A. Swanson worked at Westinghouse Astronuclear Laboratory in Pittsburgh, and he was responsible for stress analysis of the components in NERVA nuclear reactor rockets. While there he then developed 3D FEM model and computer codes to analyze and predict transient stresses and displacements of the reactor system. To integrate different computer codes and streamline the processing, Swanson asked his employer Westinghouse to develop a general-purpose FEM computer code, but his suggestion was rejected, and then he left the company and developed the initial ANSYS FEM code. Today, ANSYS has become one of the major FEM commercial software worldwide.
Several years later, J.O. Hallquist at Lawrence Livermore National Laboratory also developed a 3D nonlinear FEM code called DYNA3D, which was extensively used impact, dynamic contact, and failure analysis of structures, which later evolved to LS-DYNA. LS-DYNA is the major FEM software used in automobile design and crashworthiness analyses. In 2018, Livermore Software Technology was purchased by ANSYS, and LS-DYNA became a part of ANSYS as ANSYS LS-DYNA.
In the early development of nonlinear FEM software, there were two early pioneers: P. V. Marcal and D. Hibbitt, who was Marcal’s student at Brown University. In the early 1970s, Marcal founded the MARC corporation to develop the first general purpose nonlinear FEM program, which is still widely used today in industry and academia for analysis of complex structures, such as nuclear reactors, car crashworthiness and manufacturing processes. While Hibbitt together with B. Karlsson and P. Sorenson formed a FEM company called HKS in 1978, and they released a large-scale commercial FEM software called ABAQUS. One of the major features of ABAQUS is that it allows the user defined subroutines, which greatly facilitates the researchers to conduct their researches by using the standard FEM solvers with reliability and efficiency. Together with ANSYS, ABAQUS is one of the two commercial FEM software that dominate the market.
By the late 1990s to early 2000s, the FEM software industry had become a multi-billion-dollar business. There were several household FEM software company names such as ANSYS, ABAQUS, ADINA, LS-DYNA, NASTRAN, COMSOL Multiphysics, CSI, among others. Today, there are also a plethora of open-source FEM software available online, such as FreeFEM, OpenSees, Elmer, FEBio, FEniCS Project, DUNE, among some others.
4 Present and Future
The modern form of the FEM can routinely solve many large and complex industrial problems. It enables developing a fundamental understanding and allows for the predictive analysis for product design. For new scientific discoveries and engineering innovations, the development of new scientific principles often trails the pace of new inventions with the sheer volume of data that are generated across multiple spatial, temporal, and design parameter (spatial–temporal-parameter) spaces. For this reason, FE researchers are studying various forms of machine and deep learning methods, of which this class of methods covers the largest class of interpolations. According to the universal approximation theorem, a neural network (NN) can be designed and trained to approximate any given continuous function with any desired accuracy [224, 225, 226, 227] which is believed to drive new discoveries and enable future computational discretization technologies. In this context, Mechanistic Data Science (MDS) FEMs, which combine known scientific principles with newly collected data, will provide the critically needed research that can be a boon for new inventions.
Scientific and engineering problems typically fall under three categories: (1) problems with abundant data but undeveloped or unavailable scientific principles, (2) problems that have limited data and limited scientific knowledge, and (3) problems that have known scientific principles with uncertain parameters, with possible high computational load [228]. In essence, mechanistic data science (MDS) mimics the way human civilization has discovered solutions to difficult and unsolvable problems from the beginning of time. Instead of heuristics, MDS uses machine learning methods like active deep learning and hierarchical neural network(s) to process input data, extract mechanistic features, reduce dimensions, learn hidden relationships through regression and classification, and provide a knowledge database. The resulting reduced order form can be utilized for design and optimization of new scientific and engineering systems (see Liu et. al. [229]). Thus, the new focus of the FEM research has shifted towards the development of machine learning based FEMs and reduced order models.
With the recent development of machine learning and deep learning methods, solving FEM by constructing a deep neural network has become a state-of-the-art technology. Earlier research focused on building up a shallow neural network following the FEM structure to solve boundary value problems. Takeuchi and Kosugi [230] proposed a neural network representation of the FEM to solve Poisson equation problems. Yagawa and Aoki [231] replaced the FEM functional with the network energy of interconnected neural networks (NNs) to solve a heat conduction problem. Due to the limitation of computationliual power and slow convergency rate in shallow neural networks, earlier applications could only solve simple PDE problems. After the 2010s, neural networks for solving computational mechanics problems have become increasingly popular with the rapid growth of deep learning techniques and the development of more sophisticated neural network structures, such as convolutional neural networks (CNN), Generative Adversarial Networks (GAN) and residual neural networks (ResNet). For its high dimensional regression ability, some researchers, for example, Ghavamiana and Simone [232] used deep neural networks as a regression model to learn the material behavior or microstructure response. Other works focus on solving PDEs using deep learning neural networks. G. Karniadakis and his coworkers (see Raissi et al. [233, 234], Karniadakis et al. [235]) proposed a Physics-Informed Neural Networks (PINNs) to solve high dimensional PDEs in the strong form with constraints to accommodate both natural and essential boundary conditions. The idea of constructing deep neural networks following the FEM structure is investigated again with advanced neural network methodologies. Weinan E and his co-workers [236] and B. Yu [237] proposed a Deep Ritz Method for solving variational problems. Sirignano and Spiliopoulos [238] proposed the so-called Deep Galerkin Method (DGM) to solve high-dimensional PDEs. Zabaras and his co-worders proposed a CNN-based physics-constrained deep learning framework for high-dimensional surrogate modeling and uncertainty quantification (see Zhu et al. [239]). Rabczuk [240] systematically explore the potential to use NNs for computational mechanics by solving energetic format of the PDE (see Samaniego et al. [241]). Lee [242] proposed a partition of unity network for deep hp approximation of PDEs and extensively the training and initialization strategy to accelerate the convergence of the solution process (see Lee et al. [242]). The constructing of element shape function by activation functions has been studied by J. Opschoor and his coworkers (See [243, 244]).
Inspired by the universal approximation of deep neural networks (DNN), Zhang et al. [245] published the first paper on the construction of the FEM shape functions-based on the hierarchical nature of the DNN, called Hierarchical Deep-learning Neural Networks (HiDeNN). Specifically, the authors demonstrated the construction of a few classes of deep learning interpolation functions, such as the reproducing kernel particle method (RKPM), non-uniform rational B-spline (NURBS), and isogeometric analysis (IGA), among other approximation techniques. Saha et al. [228] generalized HiDeNN to a unified Artificial intelligence (AI)-framework, called HiDeNN-AI. HiDeNN-AI can assimilate many data-driven tools in an appropriate way, which provides a general approach to solve challenging science and engineering problems with little or no available physics as well as with extreme computational demand.
To reduce the FE computational cost, the so-called two-stage data-driven methods have been proposed of which during the offline stage, where a database generated by the FEM is first developed, and the final solutions are computed during the online stage. C. Farhat and his group at Stanford University have developed several dimensional reductions of nonlinear FEM dynamic models, including mesh sampling and weighting for the hyper reduction of nonlinear Petrov‐Galerkin reduced‐order models (see Farhat et al. [246], and Grimberg et al. [247]). To further reduce the FEM computational burden in multiscale analysis, Liu et al. [190] applied the unsupervised machine learning techniques, such as the k-mean clustering method to group the material points during the offline stage and obtain the final solutions by solving the reduced-order Lippmann–Schwinger micromechanics equations. This class of data-driven approaches circumvents the computational burden of the well-established FE square method, the offline-online database approach to solve the concurrent FEM problems. They named the method the Finite Element-self-consistent clustering analysis (FE-SCA) of which the computational cost of the microscale analysis is reduced tremendously in multiple orders of magnitude speed up (see Li et al. [248]). Gao et al. proposed an alternative (FE-SCAxSCA…xSCA) clustering analysis, of which the continuum FEM scale is concurrently solved with the (n-1) coupled-scale Lippmann–Schwinger micromechanics equations [249, 250]. It has been recently extended to (FE-SCA**2) by He et al. [251, 252].
Ortiz and his co-workers at Caltech developed data driven FEMs for dynamics and noisy data (see [253, 254]). Chen and his co-workers have developed a physics-constrained data-driven RKPM method based on locally convex reconstruction for noisy databases (see He and Chen [255]). S. Li and his group at UC-Berkeley utilized FEM solution generated data to develop a machine learning based inverse solution to predict pre-crash data of car collision (see Chen et al. [256]). Bessa et al. [257] proposed a data-driven framework to address the longstanding challenge of a two-scale analysis and design of materials under uncertainty applicable to problems that involve unacceptable computational expense when solved by standard FEM analysis of representative volume elements. The paper defined a framework that incorporates the SCA method to build large databases suitable for machine learning. The authors believe that this will open new avenues to finding innovative materials with new capabilities in an era of high-throughput computing (“big-data”).
Reduced order modeling has been an active research field over the last decades. Early research works focused on the proper orthogonal decomposition (POD) method (also known as Karhunen-Loève transform, or principal component analysis) with the purpose of reducing the degrees of freedom of the discretized equations. The POD based model reduction has shown great success in computational fluid dynamics, see e.g., the works of Berkooz et al. [258]. For further accelerating the simulations, K. Willcox and her coworkers (see [259, 260]) proposed a missing point estimation method, which is known later as a hyper reduction method. Other notable works related to POD and hyper reduction methods are the Gauss–Newton with approximated tensors (GNAT) method, Grassmann manifold based reduced basis adaptation, thanks to C. Farhat and his coworkers (see Carlberg and Farhat [261], Carlberg et al. [262], Grimberg et al. [247], Amsallem and Farhat [263], and Farhat et al. [246]). For solid mechanics, Ryckelynck et al. [264] proposed a hyper reduction method based on FEM for dealing with nonlinear problems. Lu et al. [265] proposed an adaptive hyper reduction for coupled thermal-fluid analysis. Another type of model reduction method, which is based on mathematics and has a rigorous error bound estimate, is called reduced basis method, as proposed by Maday and Rønquist [266]. The proper generalized decomposition (PGD) based model reduction, as an extension of POD, can be
dated back to 1980s, and it was introduced by P. Ladevèze (See Ladeveze [267], Ladeveze and Rougee [268]) under the name of radial time–space approximation. F. Chinesta et al. [269, 270] developed a PGD method to account for the parameter space, aiming at building offline computational vademecum for fast online predictions. It is noted that PGD methods are based on the idea of separation of variables and in particular a canonical tensor decompostion (TD).
Recent works have been conducted to combine deep machine learning methods with reduced order modeling methods. Zhang et al. [271] consolidated the various attributes of TD and PGD methods with HiDeNN and proposed the so-called HiDeNN-TD and HiDeNN-PGD methods. The comparison of FEM, TD/PGD, HiDeNN-TD/PGD, HiDeNN, and DNN has been conducted in terms of accuracy and speed. It is shown that the HiDeNN-TD/PGD outperforms other methods with a good balance between accuracy and speed. The proposed HiDeNN-TD/PGD method is expected to provide novel powful tools for solving large scale high dimensional problems while maintaining a high accuracy. Various applications, including multiphysics coupled additive manufacturing, multiscale composite modeling, and structural topology optimization, have been discussed in a generalized reduced order machine learning finite element framework [272]. In particular, this reduced order machine learning framework is expected to enable ultra large-scale high resolution topology design that is currently challenging for the FEM based topology optimization. In this regard, Lu et al. [273] recently developed a convolution HiDeNN-TD formulation with a built-in density filter for high resolution topology design. This convolution formulation incorporates the concept of meshfree approximation into the finite element function approximation and allows smoother solutions and automatic length-scale control in topology design. It is shown that the convolution HiDeNN-TD leads to better design with smoother and fine structures. This general convolution formulation opens new perspectives to resolve the length scale effect and can be applied to many orther problems, such as additive manufacturing and microstructure modeling.
The extra-ordinary interpolating capability of neural network has resulted in numerous research in approximating and solving the ordinary and partial differential equations e.g. Chen et al. [274]. Currently, the researchers are looking into approximating the mathematical operators directly from universal approximation of operator theory (see Chen and Chen [275]). The goal of this endeavor is to solve the integral equations such as the Lippmann–Schwinger micromechanics equations (Z. Liu et. al. [190]). A major step forward towards this target is made by Li et al. [276] in their proposed Graph Kernel Network, in which they developed neural operators for solving partial differential equations. In this work, a kernel-based graph neural network is shown to be able to mimic the Green’s function method for solving partial differential equation. One drawback of this method is the associated computational cost and storage requirement increases with the size of the problem. A more general version of approximating the Green’s function operator is proposed by Lu et al. in their DeepONet [277] and by Gin et al. in their DeepGreen methods (see [278]). These approximation works are vital in computational mechanics as these can directly solve the micromechanics equation for multiscale analysis like FE-SCA. Moreover, these networks have suggested that the analytical calculus method such as differentiation and integration, and solution of differential and integration equations can be directly expressed as an approximation of neural network. In this regard, recurrent neural networks have shown promised to be identical in structure of wave equations [279]. These researches are directed towards a future when the research and education in the science, technology, engineering, and mathematics (STEM) sector with discrete calculus will be transformed with the aid of deep learning. We are envisioning this field as deep learning discrete calculus, a new perspective in teaching calculus by the integration of calculus definition, numerical analysis, and deep learning (see Liu et al. [280]).
The development of model reduction methods meets the urgent demand in the industry for fast and nearly real time simulations of engineering problems, such as online dynamic system control, structural health monitoring, vehicle health monitoring, on-line advanced manufacturing feedback control, automated driving controls and decisions, etc. Such applications usually require an intensive interaction between sensors, control algorithms, and simulation tools. Practical optimal control may require a reliable prediction within the range of milli- or sub-milli-second. Reducing the computation cost of simulations has been one of the major motivations for developing model reduction methods. Other reduced order modeling related topics are the feature engineering and data analytics, which constitute an extensive literature in field of machine learning. Thus, the reduced order modeling and the machine learning have intrinsic connections. Developing reduced order machine learning methods may enable physics-data combined models that can overcome the current bottleneck in model reduction methods and purely data-driven machine learning approaches.
Change history
28 June 2022
A Correction to this paper has been published: https://doi.org/10.1007/s11831-022-09784-x
References
Hrennikoff A (1941) Solution of problems of elasticity by the framework method. J Appl Mech 8(4):169–175
Williamson F (1980) A historical note on the finite element method. Int J Num Meth’s Eng 15:930–934
Courant R (1943) Variational methods for the solution of problems of equilibrium and vibrations. Bull Am Math Soc 49:1–23
McHenry D (1943) A lattice analogy for the solution of stress problems. J Instit Civ Eng 21(2):59–82
Prager W, Synge JL (1947) Approximations in elasticity based on the concept of function space. Q Appl Math 5(3):241–269
Synge JL, Rheinboldt WC (1957) The hyper circle in mathematical physics. Phys Today 10(10):45
Argyris JH (1954) Energy theorems and structural analysis. Aircraft Eng 26:347–356 (383-387, 394)
Argyris JH (1955) Energy theorems and structural analysis. Aircraft Eng 27:42–58 (Feb.), 80–94 (March), 125–134 (April), 145–158 (May 1955)
Argyris JH (1965) Continua and discontinua. In: Preziemiencki et al (eds) Proceedings conference on matrix methods in structural mechanics. AFFDL-TR-66-80, (Oct. 26–28, 1965), Wright-Patterson AFB, Ohio, pp 11–190
Turner MJ, Clough RW, Martin HC, Topp LJ (1956) Stiffness and deflection analysis of complex structures. J Aeronaut Sci 23(9):805–823
Feng K (1965) Variational principle-based finite difference formulation. Appl Comput Math 2(4):238–262 (in Chinese)
Lax P (1003) Kang Feng. SIAM News 26(11), 7-8.
Clough RW, Wilson EL (1999 Aug) Early finite element research at Berkeley. In: Fifth US national conference on computational mechanics, pp 1–35
Clough RW (1960) The finite element method in plane stress analysis. In: Proceedings of 2nd ASCE conference on electronic computation. Pittsburgh PA, 8–9 Sept
Clough RW (1962) The stress distribution of Norfork Dam. California Univ Berkeley Inst of Engineering Research, California
Oden JT (1963) Analysis of thin plates supported at their corners: Prepared for Office of Engineering Research, College of Engineering. School of Civil Engineering, Oklahoma State University
Best G, Oden JT (1963) Stiffness matrices for shell-type structures, Research & Engineering Department, Research Report No. 233, General Dynamics, Fort Worth
Pian TH (1964) Derivation of element stiffness matrices by assumed stress distributions. AIAA J 2(7):1333–1336
Best, G. and Oden, J.T. (1963). Stiffness matrices for shell-type structures, Research & Engineering Department, Research Report No. 233, General Dynamics, Fort Worth
Taig IC (1961) Structural analysis by the matrix-displacement method. Engl Elect. Auiat. Rep. No. SO17
Ergatoudis I, Irons BM, Zienkiewicz OC (1968) Curved, isoparametric, “quadrilateral” elements for finite element analysis. Int J Solids Struct 4(1):31–42
Levy S (1953) Structural analysis and influence coefficients for delta wings. J Aeronaut Sci 20(7):449–454
Comer RL, Levy S (1963) Deflections of an inflated circular-cylindrical cantilever beam. AIAA J 1(7):1652–1655
Langefors B (1952) Analysis of elastic structures by matrix transformation with special regard to semimonocoque structures. J Aeronaut Sci 19(7):451–458
Denke PH (1966) Engineering aspects and mathematical formulation of the problem of a computerized aircraft structural analysis system. SAE Transactions, pp 327–343
Wehle LB Jr, Lansing W (1952) A method for reducing the analysis of complex redundant structures to a routine procedure. J Aeronaut Sci 19(10):677–684
Hoff NJ, Klein B, Libby PA (1946) numerical procedures for the calculation of stresses in monocoques: influence coefficients of curved bars for distortions in their own plane. IV (vol 4). National Advisory Committee for Aeronautics
Archer JS (1965) Consistent matrix formulations for structural analysis using finite-element techniques. AIAA J 3(10):1910–1918
Przemieniecki JS (1963) Matrix structural analysis of substructures. AIAA J 1(1):138–147
Przemieniecki JS, Denke PH (1966) Joining of complex substructures by the matrix force method. J Aircr 3(3):236–243
Przemieniecki JS (1968) Discrete-element methods for stability analysis of complex structures. Aeronaut J 72(696):1077–1086
Przemieniecki JS (1972) Matrix analysis of local instability in plates, stiffened panels, and columns. Int J Numer Meth Eng 5(2):209–216
Przemieniecki JS (1973) Finite element structural analysis of local instability. AIAA J 11(1):33–39
Reissner E (1950) On a variational theorem in elasticity. J Math Phys 29(1–4):90–95
Hellinger E, Klein F, Müller C (1914) Die allgemeinen Ansätzeder Mechanik der Kontinua. Enr Math Wti 4(602):694
Hu HC (1955) On some variational principles in the theory of elasticity and plasticity. Sci Sin 4:33–54
Washizu K (1955) On the variational principles of elasticity and plasticity. Massachusetts Inst of Tech Cambridge Aeroelastic and Structures Research Lab
Aubin JP (1967) Behavior of the error of the approximate solutions of boundary-value problems for linear elliptic operators by Galerkin’s Method and finite differences. Annali della Scuola Normale di Pisa, Series 3(21):599–637
Zlamal M (1968) On the finite element method. Numerische Math 12:394–440
Birkhoff G (1969) Piecewise bicubic interpolation and approximation in polygons. In: Shoenberg IJ (ed) approximations with special emphasis on spline functions. Academic Press, New York, pp 85–121
Nitsche J (1970) Lineare spline-funktionen und die methoden von Ritz für elliptische randwertprobleme. Arch Ration Mech Anal 36(5):348–355
Aziz AK (ed) (1972) The mathematical foundations of the finite element method with applications to partial differential equations. Academic Press, New York
Babuska I (1971) Error bounds for the finite element method. Numerische Math 16:322–333
Babushka I, Aziz AK (1972) Survey lectures on the mathematical foundation of the finite element method. In: Aziz AK (ed) The mathematical foundation of the finite element method with applications to partial differential equations. Academic Press, New York, pp 5–359
Dupont T (1973) L^2-estimates for Galerkin methods for second order hyperbolic equations. SIAM J Numer Anal 10(5):880–889
Douglas J, Dupont T (1973) Superconvergence for Galerkin methods for the two-point boundary problem via local projections. Numer Math 21(3):270–278
Nitsche JA, Schatz AH (1974) Interior estimates for Ritz-Galerkin methods. Math Comput 28(128):937–958
Babuska I (1976) Finite element methods for domains with corners. Computing 6:264–273
Bramble JH, Nitsche JA, Schatz AH (1975) Maximum-norm interior estimates for Ritz-Galerkin methods. Math Comput 29(131):677–688
Babuska I, Oden JT, Lee JK (1977) Mixed-hybrid finite element approximations of second-order elliptic boundary-value problems. Comput Methods Appl Mech Eng 11:175–206
Pian THH (1965) Element stiffness matrices for boundary compatibility and for prescribed boundary stresses. In: Proceedings of the First Conference of Matrix Methods in Structural Mechanics, pp 457–477
Pian TH, Sumihara K (1984) Rational approach for assumed stress finite elements. Int J Numer Methods Eng 20(9):1685–1695
Herrmann LR (1965) Elasticity equations for incompressible and nearly incompressible materials by a variational theorem. AIAA journal, 3(10):1896–1900
Boffi D, Brezzi F, Fortin M (2013) Mixed finite element methods and applications (vol 44). Springer, Berlin, pp xiv–685
Wilson EL, Clough RW (1962) Dynamic analysis by step-by-step matrix analysis. In: Symposium on the Use of Computers in Civil Engineering, Lisbon, Portugal
Hilber HM, Hughes TJR, Taylor RL (1977) Improved numerical dissipation for time integration algorithms in structural dynamics. Earthquake Eng Struct Dynam 5(3):283–292
Houbolt JC (1950) A recurrence matrix solution for the dynamic response of elastic aircraft. J Aeronaut Sci 17(9):540–550
Anderheggen E, Ekchian D, Heiduschke K, Bartelt P (1970) A contact algorithm for explicit dynamic FEM analysis. WIT Trans Eng Sci 1:272
Biffle JH, Key SW (1977) Finite element formulations for transient dynamic problems in solids using explicit time integration. Comput Methods Appl Mech Eng 12(3):323–336
Gresho PM, Lee RL, Sani RL, Stullich TW (1978) Time-dependent FEM solution of the incompressible Navier--Stokes equations in two-and three-dimensions (No. UCRL--81323). California University, Lawrence Livermore Lab, Livermore, California (USA)
Goudreau GL, Hallquist J (1982) Recent developments in large-scale finite element Lagrangian hydrocode technology. Comput Methods Appl Mech Eng 33(1–3):725–757
Gravouil A, Combescure A (2001) Multi-time-step explicit–implicit method for non-linear structural dynamics. Int J Numer Meth Eng 50(1):199–225
Oden JT, Wellford LC Jr (1972) Analysis of flow of viscous fluids by the finite-element method. AIAA J 10(12):1590–1599
Hughes TJR, Tezduyar T (1981) Finite elements based upon Mindlin plate theory with particular reference to the four-node bilinear isoparametric element. J Appl Mech 48:587
Hughes TJR, Liu WK (1981) Nonlinear finite element analysis of shells: Part I. Three-dimensional shells. Comput Methods Appl Mech Eng 26(3):331–362
Brooks AN, Hughes TJR (1982) Streamline upwind/Petrov-Galerkin formulations for convection dominated flows with particular emphasis on the incompressible Navier-Stokes’s equations. Comput Methods Appl Mech Eng 32(1–3):199–259
Mizukami A, Hughes TJR (1985) A Petrov-Galerkin finite element method for convection-dominated flows: an accurate upwinding technique for satisfying the maximum principle. Comput Methods Appl Mech Eng 50(2):181–193
Hughes TJR (1995) Multiscale phenomena: Green’s functions, the Dirichlet-to-Neumann formulation, subgrid scale models, bubbles, and the origins of stabilized methods. Comput Methods Appl Mech Eng 127(1–4):387–401
Hughes TJR, Mallet M, Akira M (1986) A new finite element formulation for computational fluid dynamics: II. Beyond SUPG. Comput Methods Appl Mech Eng 54(3):341–355
Hughes TJR, Franca LP, Mallet M (1986) A new finite element formulation for computational fluid dynamics: I. Symmetric forms of the compressible Euler and Navier-Stokes equations and the second law of thermodynamics. Comput Methods Appl Mech Eng 54(2):223–234
Hughes TJR, Franca LP, Balestra M (1986) A new finite element formulation for computational fluid dynamics: V. Circumventing the Babuška-Brezzi condition: a stable Petrov-Galerkin formulation of the Stokes problem accommodating equal-order interpolations. Comput Methods Appl Mech Eng 59(1):85–99
Shakib F, Hughes TJR, Johan Z (1991) A new finite element formulation for computational fluid dynamics: X. The compressible Euler and Navier-Stokes equations. Compur Methods Appl Mech Eng 89(1–3):141–219
Caendish JC, Field DA, Frey WH (1985) An approach to automatic three-dimensional finite element mesh generation. Int J Numer Meth Eng 21(2):329–347
Bennett JA, Botkin ME (1985) Structural shape optimization with geometric description and adaptive mesh refinement. AIAA J 23(3):458–464
Yerry M, Shephard M (1983) A modified quadtree approach to finite element mesh generation. IEEE Comput Graphics Appl 3(01):39–46
Wördenweber B (1984) Finite element mesh generation. Comput-Aided Des 16(5):285–291
Ho-Le K (1988) Finite element mesh generation methods: a review and classification. Comput Aided Des 20(1):27–38
Simo JC, Taylor RL (1985) Consistent tangent operators for rate-independent elastoplasticity. Comput Methods Appl Mech Eng 48(1):101–118
Hughes TJ, Pister KS (1978) Consistent linearization in mechanics of solids and structures. Comput Struct 8(3–4):391–397
Simo JC, Vu-Quoc L (1986) A three-dimensional finite-strain rod model. Part II: Computational aspects. Comput Methods Appl Mech Eng 58(1):79–116
Simo JC, Fox DD (1989) On a stress resultant geometrically exact shell model. Part I: formulation and optimal parametrization. Comput Methods Appl Mech Eng 72(3):267–304
Simo JC, Fox DD, Rifai MS (1989) On a stress resultant geometrically exact shell model. Part II: The linear theory; computational aspects. Comput Methods Appl Mech Eng 73(1):53–92
Simo JC, Rifai M (1990) A class of mixed assumed strain methods and the method of incompatible modes. Int J Numer Meth Eng 29(8):1595–1638
Simo JC, Armero F (1992) Geometrically non-linear enhanced strain mixed methods and the method of incompatible modes. Int J Numer Meth Eng 33(7):1413–1449
Andelfinger U, Ramm E (1993) EAS-elements for two-dimensional, three-dimensional, plate and shell structures and their equivalence to HR-elements. Int J Numer Meth Eng 36(8):1311–1337
Bischoff M, Ramm E (1998) Shear deformable shell elements for large strains and rotations. Int J Numer Meth Eng 40(23):4427–4449
Donea J, Giuliani S, Halleux JP (1982) An arbitrary Lagrangian-Eulerian finite element method for transient dynamic fluid-structure interactions. Comput Methods Appl Mech Eng 33(1–3):689–723
Donea J, Huerta A (2003) Finite element methods for flow problems. John Wiley & Sons, Hoboken
Donea J, Huerta A, Ponthot JP, Rodríguez‐Ferran A (2017) Arbitrary Lagrangian–E ulerian Methods.In: Encyclopedia of Computational Mechanics Second Edition, pp 1–23
Hughes TJR, Liu WK, Zimmerman TK (1981) Lagrangian-Eulerian finite element formulations for incompressible viscous flows. Comput Methods Appl Mech Eng 29:329–349
Hughes TJR, Liu WK, Zimmermann TK (1981) Lagrangian-Eulerian finite element formulation for incompressible viscous flows. Comput Methods Appl Mech Eng 29(3):329–349
Hirt CW, Amsden AA, Cook JL (1974) An arbitrary Lagrangian-Eulerian computing method for all flow speeds. J Comput Phys 14(3):227–253
Farhat C (1994) Implicit parallel processing in structural mechanics. Comp Mech Adv 2:1–124
Oger G, Marrone S, Le Touzé D, De Leffe M (2016) SPH accuracy improvement through the combination of a quasi-Lagrangian shifting transport velocity and consistent ALE formalisms. J Comput Phys 313:76–98
Taylor CA, Hughes TJR, Zarins CK (1998) Finite element modeling of blood flow in arteries. Comput Methods Appl Mech Eng 158(12):155–196
Holzapfel GA, Eberlein R, Wriggers P, Weizsäcker HW (1996) Large strain analysis of soft biological membranes: formulation and finite element analysis. Comput Methods Appl Mech Eng 132(1–2):45–61
Liu WK, Belytschko T, Mani A (1986) Random field finite elements. Int J Numer Meth Eng 23(10):1831–1845
Botkin ME (1982) Shape optimization of plate and shell structures. AIAA J 20(2):268–273
Kikuchi N, Chung KY, Torigaki T, Taylor JE (1986) Adaptive finite element methods for shape optimization of linearly elastic structures. Comput Methods Appl Mech Eng 57(1):67–89
Suzuki K, Kikuchi N (1991) A homogenization method for shape and topology optimization. Comput Methods Appl Mech Eng 93(3):291–318
Bendsøe MP, Kikuchi N (1988) Generating optimal topologies in structural design using a homogenization method. Comput Methods Appl Mech Eng 71(2):197–224
Imam MH (1982) Three-dimensional shape optimization. Int J Numer Meth Eng 18(5):661–673
Rashid YR (1968) Ultimate strength analysis of prestressed concrete pressure vessels. Nucl Eng Des 7(4):334–344
Marchertas AH, Fistedis SH, Bažant ZP, Belytschko TB (1978) Analysis and application of prestressed concrete reactor vessels for LMFBR containment. Nucl Eng Des 49(1–2):155–173
Bažant ZP (1976) Instability, ductility, and size effect in strain-softening concrete. J Eng Mech Div 102(2):331–344
Bažant ZP, Cedolin L (1980) Fracture mechanics of reinforced concrete. J Eng Mech Div 106(6):1287–1306
Hillerborg A, Modéer M, Petersson PE (1976) Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements. Cem Concr Res 6(6):773–781
Bažant ZP, Belytschko TB (1985) Wave propagation in a strain-softening bar: exact solution. J Eng Mech 111(3):381–389
Bažant ZP, Chang TP (1987) Nonlocal finite element analysis of strain-softening solids. J Eng Mech 113(1):89–105
Lubliner J, Oliver J, Oller S, Oñate E (1989) A plastic-damage model for concrete. Int J Solids Struct 25(3):299–326
Wriggers P, Moftah SO (2006) Mesoscale models for concrete: Homogenization and damage behavior. Finite Elem Anal Des 42(7):623–636
Bazant ZP, Belytschko TB, Chang TP (1984) Continuum theory for strain-softening. J Eng Mech 110(12):1666–1692
Pijaudier-Cabot G, Bažant ZP (1987) Nonlocal damage theory. J Eng Mech 113(10):1512–1533
De Borst R, Mühlhaus HB (1992) Gradient-dependent plasticity: formulation and algorithmic aspects. Int J Numer Meth Eng 35(3):521–539
Peerlings RH, de Borst R, Brekelmans WM, de Vree J (1996) Gradient enhanced damage for quasi-brittle materials. Int J Numer Meth Eng 39(19):3391–3403
Steinmann P, Willam K (1991) Localization within the framework of micropolar elasto-plasticity. In: Brüller OS, Mannl V, Najar J (eds) Advances in continuum mechanics. Springer, Berlin, Heidelberg, pp 296–313
Dietsche A, Steinmann P, Willam K (1993) Micropolar elastoplasticity and its role in localization. Int J Plast 9(7):813–831
Steinmann P (1994) A micropolar theory of finite deformation and finite rotation multiplicative elastoplasticity. Int J Solids Struct 31(8):1063–1084
Iordache MM, Willam K (1998) Localized failure analysis in elastoplastic Cosserat continua. Comput Methods Appl Mech Eng 151(3–4):559–586
Hughes TJR, Taylor RL, Sackman JL, Curnier A, Kanoknukulchai W (1976) A finite element method for a class of contact-impact problems. Comput Methods Appl Mech Eng 8(3):249–276
Simo JC, Wriggers P, Taylor RL (1985) A perturbed Lagrangian formulation for the finite element solution of contact problems. Comput Methods Appl Mech Eng 50(2):163–180
Simo JC, Laursen TA (1992) An augmented Lagrangian treatment of contact problems involving friction. Comput Struct 42(1):97–116
Kikuchi N, Oden JT (1988) Contact problems in elasticity: a study of variational inequalities and finite element methods. Society for Industrial and Applied Mathematics, Philadelphia
Papadopoulos P, Taylor RL (1992) A mixed formulation for the finite element solution of contact problems. Comput Methods Appl Mech Eng 94(3):373–389
Arnold D, Falk R, Winther R (2010) Finite element exterior calculus: from Hodge theory to numerical stability. Bull Am Math Soc 47(2):281–354
Zienkiewicz OC, Zhu JZ (1992) The superconvergent patch recovery and a posteriori error estimates. Part 1: The recovery technique. Int J Numer Methods Eng 33(7):1331–1364
Ainsworth M, Senior B (1998) An adaptive refinement strategy for hp-finite element computations. Appl Numer Math 26(1–2):165–178
Ainsworth M, Oden JT (2000) A posteriori error estimation in finite element analysis. John Wiley & Sons, Hoboken
Ainsworth M, Oden JT (1997) A posteriori error estimation in finite element analysis. Comput Methods Appl Mech Eng 142(1–2):1–88
Szabo BA, Mehta AK (1978) p-convergent finite element approximations in fracture mechanics. Int J Num Meth Engng 12:551–560
Babuska I, Szabo BA, Kat IN (1981) The p-version of the finite element method. SIAM J Numer Anl 18:515–544
Babuska I, Szabo BA, Kat IN (1982) On the rates of convergence of the finite element method. Int J Numer Meth Eng 18:323–341
Demkowicz L, Oden JT, Rachowicz W, Hardy O (1989) Toward a universal hp adaptive finite element strategy, Part 1. Constrained approximation and data structure. Comput Methods Appl Mech Eng 77(1–2):79–112
Demkowicz L, Oden JT, Rachowicz W (1990) A new finite element method for solving compressible Navier-Stokes equations based on an operator splitting method and hp adaptivity. Comput Methods Appl Mech Eng 84(3):275–326
Demkowicz L, Monk P, Vardapetyan L, Rachowicz W (2000) De Rham diagram for hp finite element spaces. Comput Math Appl 39(7–8):29–38
Oden JT, Demkowicz L, Rachowicz W, Westermann TA (1989) Toward a universal hp adaptive finite element strategy, Part 2. A posteriori error estimation. Comput Methods Appl Mech Eng 77(12):113–180
Oden JT, Duarte CAM, Zienkiewicz OC (1998) A new cloud-based hp finite element method. Comput Methods Appl Mech Eng 153(1–2):117–126
Baumann CE, Oden JT (1999) A discontinuous hp finite element method for convection: diffusion problems. Comput Methods Appl Mech Eng 175(3–4):311–341
Fish J (1992) The s-version of the finite element method. Comput Struct 43(3):539–547
Xu XP, Needleman A (1994) Numerical simulations of fast crack growth in brittle solids. J Mech Phys Solids 42(9):1397–1434
Camacho GT, Ortiz M (1996) Computational modelling of impact damage in brittle materials. Int J Solids Struct 33(20–22):2899–2938
Ortiz M, Pandolfi A (1999) Finite-deformation irreversible cohesive elements for three-dimensional crack-propagation analysis. Int J Numer Meth Eng 44(9):1267–1282
Nguyen O, Repetto EA, Ortiz M, Radovitzky RA (2001) A cohesive model of fatigue crack growth. Int J Fract 110(4):351–369
Pietruszczak ST, Mroz Z (1981) Finite element analysis of deformation of strain-softening materials. Int J Numer Meth Eng 17(3):327–334
Brocca M, Bažant ZP (2001) Size effect in concrete columns: finite-element analysis with microplane model. J Struct Eng 127(12):1382–1390
Caner FC, Bažant ZP (2009) Size effect on strength of laminate-foam sandwich plates: finite element analysis with interface fracture. Compos B Eng 40(5):337–348
Šmilauer V, Hoover CG, Bažant ZP, Caner FC, Waas AM, Shahwan KW (2011) Multiscale simulation of fracture of braided composites via repetitive unit cells. Eng Fract Mech 78(6):901–918
Rahimi-Aghdam S, Chau VT, Lee H, Nguyen H, Li W, Karra S, Rougier E, Viswanathan H, Srinivasan G, Bažant ZP (2019) Branching of hydraulic cracks enabling permeability of gas or oil shale with closed natural fractures. Proc Natl Acad Sci 116(5):1532–1537
Belytschko T, Lu YY, Gu L (1994) Element-free Galerkin methods. Int J Numer Meth Eng 37(2):229–256
Liu WK, Jun S, Zhang YF (1995) Reproducing kernel particle methods. Int J Numer Meth Fluids 20(8–9):1081–1106
Liu WK, Jun S, Li S, Adee J, Belytschko T (1995) Reproducing kernel particle methods for structural dynamics. Int J Numer Meth Eng 38(10):655–1679
Li S, Liu W-K, Qian D, Guduru DP, Rosakis AJ (2001) Dynamic shear band propagation and micro-structure of adiabatic shear band. Comput Methods Appl Mech Eng 191:73–92
Li S, Liu WK, Rosakis AJ, Belytschko T, Hao W (2002) Mesh-free Galekin simulations of dynamic shear band propagation and failure mode transition. Int J Solids Struct 39(5):1213–1240
Oñate E, Idelsohn S, Zienkiewicz OC, Taylor R (1996) A finite point method in computational mechanics. Applications to convective transport and fluid flow. Int J Numer Methods Eng 39(22):3839–3866
Atluri SN, Zhu T (1998) A new meshless local Petrov-Galerkin (MLPG) approach in computational mechanics. Comput Mech 22(2):117–127
Fleming M, Chu YA, Moran B, Belytschko T (1997) Enriched element-free galerkin methods for crack tip fields. Int J Numer Meth Eng 40:1483–1504
Chen JS, Wu CT, Yoon S, You Y (2001) A stabilized conforming nodal integration for Galerkin mesh-free methods. Int J Numer Meth Eng 50(2):435–466
Melenk JM, Babuška I (1996) The partition of unity finite element method: basic theory and applications. Comput Methods Appl Mech Eng 139(1–4):289–314
Babuška I, Melenk JM (1997) The partition of unity method. Int J Numer Meth Eng 40(4):727–758
Belytschko T, Black T (1999) Elastic crack growth in finite elements with minimal remeshing. Int J Numer Meth Eng 45(5):601–620
Moës N, Dolbow J, Belytschko T (1999) A finite element method for crack growth without remeshing. Int J Numer Meth Eng 46(1):131–150
Bourdin B, Francfort GA, Marigo J.J (2000) Numerical experiments in revisited brittle fracture. J Mech Phys Solids 48(4):797–826
Karma A, Kessler DA, Levine H (2001) Phase-field model of mode III dynamic fracture. Phys Rev Lett 87(4):045501
Karma A, Lobkovsky AE (2004) Unsteady crack motion and branching in a phase-field model of brittle fracture. Phys Rev Lett 92(24):245510
Bourdin B, Chambolle A (2006) The phase-field method in optimal design. In: IUTAM symposium on topological design optimization of structures, machines and materials. Springer, Dordrecht, pp 207–215
Kuhn C, Müller R (2010) A continuum phase field model for fracture. Eng Fract Mech 77(18):3625–3634
Miehe C, Hofacker M, Welschinger F (2010) A phase field model for rate-independent crack propagation: robust algorithmic implementation based on operator splits. Comput Methods Appl Mech Eng 199(45–48):2765–2778
Borden MJ, Verhoosel CV, Scott MA, Hughes TJR, Landis CM (2012) A phase-field description of dynamic brittle fracture. Comput Methods Appl Mech Eng 217:77–95
Wilson ZA, Borden MJ, Landis CM (2013) A phase-field model for fracture in piezoelectric ceramics. Int J Fract 183(2):135–153
Pham KH, Ravi-Chandar K, Landis CM (2017) Experimental validation of a phase-field model for fracture. Int J Fract 205(1):83–101
Hughes TJR, Cottrell JA, Bazilevs Y (2005) Isogeometric analysis: CAD, finite elements, NURBS, exact geometry and mesh refinement. Comput Methods Appl Mech Eng 194(39–41):4135–4195
Cottrell JA, Hughes TJR, Bazilevs Y (2009) Isogeometric analysis: toward integration of CAD and FEA. John Wiley & Sons, Hoboken
Abraham FF, Broughton JQ, Bernstein N, Kaxiras E, (1998) Spanning the continuum to quantum length scales in a dynamic simulation of brittle fracture. EPL (Europhysics Letters), 44(6) p.783.
Tadmor EB, Ortiz M, Phillips R (1996) Quasicontinuum analysis of defects in solids. Philos Mag A 73(6):1529–1563
Wagner GJ, Liu WK (2003) Coupling of atomistic and continuum simulations using a bridging scale decomposition. J Comput Phys 190(1):249–274
Gavini V, Bhattacharya K, Ortiz M (2007) Quasi-continuum orbital-free density-functional theory: a route to multi-million atom non-periodic DFT calculation. J Mech Phys Solids 55(4):697–718
Hou TY, Wu XH (1997) A multiscale finite element method for elliptic problems in composite materials and porous media. J Comput Phys 134(1):169–189
Ghosh S, Lee K, Moorthy S (1995) Multiple scale analysis of heterogeneous elastic structures using homogenization theory and Voronoi cell finite element method. Int J Solids Struct 32(1):27–62
Fish J, Shek K, Pandheeradi M, Shephard MS (1997) Computational plasticity for composite structures based on mathematical homogenization: theory and practice. Comput Methods Appl Mech Eng 148(1–2):53–73
Fish J, Yu Q, Shek K (1999) Computational damage mechanics for composite materials based on mathematical homogenization. Int J Numer Meth Eng 45(11):1657–1679
Dvorak GJ, Wafa AM, Bahei-El-Din YA (1994) Implementation of the transformation field analysis for inelastic composite materials. Comput Mech 14(3):201–228
Miehe C, Schröder J, Schotte J (1999) Computational homogenization analysis in finite plasticity simulation of texture development in polycrystalline materials. Comput Methods Appl Mech Eng 171(3–4):387–418
Zhang L, Gerstenberger A, Wang X, Liu WK (2004) Immersed finite element method. Comput Methods Appl Mech Eng 193(21–22):2051–2067
Hughes TJR, Winget J (1980) Finite rotation effects in numerical integration of rate constitutive equations arising in large-deformation analysis. Int J Numer Meth Eng 15(12):1862–1867
Simo JC, Hughes TJR (2006) Computational inelasticity (vol 7). Springer Science & Business Media, New York
Nagtegaal JC (1982) On the implementation of inelastic constitutive equations with special reference to large deformation problems. Comput Methods Appl Mech Eng 33(1–3):469–484
Simo JC, Ortiz M (1985) A unified approach to finite deformation elastoplastic analysis based on the use of hyperelastic constitutive equations. Comput Methods Appl Mech Eng 49(2):221–245
Tvergaard V, Needleman A (1984) Analysis of the cup-cone fracture in a round tensile bar. Acta Metall 32(1):157–169
Chinesta F, Ladeveze P, Ibanez R, Aguado JV, Abisset-Chavanne E, Cueto E (2017) Data-driven computational plasticity. Procedia Eng 207:209–214
Liu Z, Bessa MA, Liu WK (2016) Self-consistent clustering analysis: an efficient multi-scale scheme for inelastic heterogeneous materials. Comput Methods Appl Mech Eng 306:319–341
Liu Z, Fleming MA, Liu WK (2018) Microstructural material database for self-consistent clustering analysis of elastoplastic strain softening materials. Comput Methods Appl Mech Eng 330:547–577
Peirce D, Asaro RJ, Needleman A (1982) An analysis of nonuniform and localized deformation in ductile single crystals. Acta Metall 30(6):1087–1119
Arsenlis A, Parks DM (1999) Crystallographic aspects of geometrically necessary and statistically stored dislocation density. Acta Mater 47(5):1597–1611
Arsenlis A, Parks DM (2002) Modeling the evolution of crystallographic dislocation density in crystal plasticity. J Mech Phys Solids 50(9):1979–2009
Arsenlis A, Parks DM, Becker R, Bulatov VV (2004) On the evolution of crystallographic dislocation density in non-homogeneously deforming crystals. J Mech Phys Solids 52(6):1213–1246
Quey R, Dawson PR, Barbe F (2011) Large-scale 3D random polycrystals for the finite element method: generation, meshing and remeshing. Comput Methods Appl Mech Eng 200(17–20):1729–1745
Mathur KK, Dawson PR (1989) On modeling the development of crystallographic texture in bulk forming processes. Int J Plast 5(1):67–94
Ma A, Roters F, Raabe D (2006) A dislocation density based constitutive model for crystal plasticity FEM including geometrically necessary dislocations. Acta Mater 54:2169–2179
Raabe D, Sachtleber M, Zhao Z, Roters F, Zaefferer S (2001) Micromechanical and macromechanical effects in grain scale polycrystal plasticity experimentation and simulation. Acta Mater 49(17):3433–3441
Raabe D, Klose P, Engl B, Imlau K-P, Friedel F, Roters F (2002) Concepts for integrating plastic anisotropy into metal forming simulations. Adv Eng Mater 4:169–180
Roters F, Eisenlohr P, Hantcherli L, Tjahjanto DD, Bieler TR, Raabe D (2010) Overview of constitutive laws, kinematics, homogenization and multiscale methods in crystal plasticity finite-element modeling: theory, experiments, applications. Acta Mater 58(4):1152–1211
Li S, Ren B, Minaki H (2014) Multiscale crystal defect dynamics: a dual-lattice process zone model. Philos Mag 94(13):1414–1450
Lyu D, Li S (2019) A multiscale dislocation pattern dynamics: towards an atomistic-informed crystal plasticity theory. J Mech Phys Solids 122:613–632
Yu C, Kafka OL, Liu WK (2019) Self-consistent clustering analysis for multiscale modeling at finite strains. Comput Methods Appl Mech Eng 349:339–359
Beirão da Veiga L, Brezzi F, Cangiani A, Manzini G, Marini LD, Russo A (2013) Basic principles of virtual element methods. Math Models Methods Appl Sci 23(01):199–214
Beirão da Veiga L, Brezzi F, Marini LD, Russo A (2014) The hitchhiker’s guide to the virtual element method. Math Models Methods Appl Sci 24(08):1541–1573
Strang G, Fix GJ (1973) An analysis of the finite element method. Prentice Hall, Hoboken
Becker EB, Carey GF, Oden JT (1981) Finite elements: an introduction (vol 1). Prentice Hall, Hoboken
Carey GF, Oden JT (1983) Finite elements: a second course. Prentice-Hall, Hoboken
Carey GF, Oden JT (1986) Finite elements: fluid mechanics. VI, Prentice-Hall, Hoboken
Oden JT, Carey GF (1983) Finite elements, mathematical aspects, vol 4. Prentice-hall, Hoboken
Oden JT (2006) Finite elements of nonlinear continua. Courier Corporation, Chelmsford
Oden JT, Reddy JN (2012) An introduction to the mathematical theory of finite elements. Courier Corporation, Chelmsford
Carey GF, Oden JT, Saunders H (1987) Finite elements: computational aspects. J Vib Acoust 109:219
Zienkiewicz OC, Cheung YK (1967) The Finite Element Method in Structural and Continuum Mechanics. McGraw-Hill, New York
Zienkiewicz OC, Taylor RL (2005) The finite element method for solid and structural mechanics. Elsevier, Amsterdam
Zienkiewicz OC, Taylor RL, Zhu JZ (2005) The finite element method: its basis and fundamentals. Elsevier, Amsterdam
Hughes TJR (1987) The finite element method. Linear static and dynamic finite element analysis. Prentice Hall, Englewood Cliffs, p 803
Cook RD, Malkus DS, Plesha ME, Witt RJ (1989) Concepts and applications of finite element analysis. John Willey and Sons, Hoboken
Cook RD (2007) Concepts and applications of finite element analysis. John Wiley & Sons, Hoboken
Bathe KJ (2006) Finite element procedures. Klaus-Jurgen Bathe, Berlin
Belytschko T, Liu WK, Moran B, Elkhodary K (2013) Nonlinear finite elements for continua and structures. John Wiley & sons, Hoboken
De Borst R, Crisfield MA, Remmers JJ, Verhoosel CV (2012) Nonlinear finite element analysis of solids and structures. John Wiley & Sons, Hoboken
Cybenko G (1989) Approximation by superpositions of a sigmoidal function. Math Control Signals Syst 2(4):303–314
Hornik K, Stinchcombe M, White H et al (1989) Multilayer feedforward networks are universal approximators. Neural Netw 2(5):359–366
Hashash YMA, Jung S, Ghaboussi J (2004) Numerical implementation of a neural network-based material model in finite element analysis. Int J Numer Meth Eng 59(7):989–1005
Selmic RR, Lewis FL (2002) Neural-network approximation of piecewise continuous functions: application to friction compensation. IEEE Trans Neural Netw 13(3):745–751
Saha S, Gan Z, Cheng L, Gao J, Kafka OL, Xie X, Li H, Tajdari M, Kim HA, Liu WK (2021) Hierarchical Deep Learning Neural Network (HiDeNN): an artificial intelligence (AI) framework for computational science and engineering. Comput Methods Appl Mech Eng 373:113452
Liu WK, Saha S, Mojumder S, Suarez DA, Lu Y, Li H, Xie X, Gan Z, Invention ID: Disc-ID-NU2021-102-01 (Invention disclosure filed on April 21, 2021), HiDeNN: An AI Platform for Scientific and Materials Systems Innovation (Serial No. 63/177,517)
Takeuchi J, Kosugi Y (1994) Neural network representation of finite element method. Neural Netw 7(2):389–395
Yagawa G, Aoki O (1995) A neural network-based finite element method on parallel processors. In: Batra RC (ed) Contemporary research in engineering science. Springer, Berlin, Heidelberg, pp 637–653
Ghavamian F, Simone A (2019) Accelerating multiscale finite element simulations of history-dependent materials using a recurrent neural network. Comput Methods Appl Mech Eng 357:112594
Raissi M, Perdikaris P, Karniadakis GE (2017) Physics informed deep learning (part i): Data-driven solutions of nonlinear partial differential equations. arXiv:1711.10561
Raissi M, Perdikaris P, Karniadakis GE (2019) Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J Comput Phys 378:686–707
Karniadakis GE, Kevrekidis IG, Lu L, Perdikaris P, Wang S, Yang L (2021) Physics-informed machine learning. Nat Rev Phys 3(6):422–440
Han J, Jentzen A, Weinan E (2018) Solving high-dimensional partial differential equations using deep learning. Proc Natl Acad Sci 115(34):8505–8510
Yu B (2018) The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems. Commun Math Stat 6(1):1–12
Sirignano J, Spiliopoulos K (2018) DGM: a deep learning algorithm for solving partial differential equations. J Comput Phys 375:1339–1364
Zhu Y, Zabaras N, Koutsourelakis PS, Perdikaris P (2019) Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data. J Comput Phys 394:56–81
Rabczuk T (2020) An energy approach to the solution of partial differential equations in computational mechanics via machine learning: Concepts, implementation and applications. Comp Methods Appl Mech Eng 362, p.112790
Samaniego E, Anitescu C, Goswami S, Nguyen-Thanh VM, Guo H, Hamdia K, Zhuang X, Rabczuk T (2020) An energy approach to the solution of partial differential equations in computational mechanics via machine learning: concepts, implementation and applications. Comput Methods Appl Mech Eng 362:112790
Lee K, Trask NA, Patel RG, Gulian MA, Cyr EC (2021) Partition of unity networks: deep hp-approximation. arXiv:2101.11256.
He F, Wang B, Tao D (2020) Piecewise linear activations substantially shape the loss surfaces of neural networks. arXiv:2003.12236
Longo M, Opschoor JA, Disch N, Schwab C, Zech J (2022) De Rham compatible deep neural networks. arXiv:2201.05395
Zhang L, Cheng L, Li H, Gao J, Yu C, Domel R, Yang Y, Tang S, Liu WK (2021) Hierarchical deep-learning neural networks: finite elements and beyond. Comput Methods Appl Mech Eng 67(1):207–230
Farhat C, Avery P, Chapman T, Cortial J (2014) Dimensional reduction of nonlinear finite element dynamic models with finite rotations and energy-based mesh sampling and weighting for computational efficiency. Int J Numer Meth Eng 98(9):625–662
Grimberg S, Farhat C, Tezaur R, Bou-Mosleh C (2020) Mesh sampling and weighting for the hyper reduction of nonlinear Petrov-Galerkin reduced-order models with local reduced-order bases. Int J Numer Meth Eng 127(1):1846–1874
Li H, Kafka OL, Gao J, Yu C, Nie Y, Zhang L, Tajdari M, Tang S, Guo X, Li G, Tang S (2019) Clustering discretization methods for generation of material performance databases in machine learning and design optimization. Comput Mech 64(2):281–305
Gao J, Mojumder S, Zhang W, Li H, Suarez D, He C, Cao J, Liu WK (2021) Concurrent n-scale modeling for non-orthogonal woven composite. arXiv:2105.10411
Gao J (2020) Multiscale modeling of composite laminate structures: a pathway towards predictive science and engineering. PhD diss., Northwestern University
He C, Ge J, Gao J, Liu J, Chen H, Liu WK, Fang D (2021) From microscale to mesoscale: The non-linear behavior prediction of 3D braided composites based on the SCA2 concurrent multiscale simulation. Compos Sci Technol 213:108947. https://doi.org/10.1016/j.compscitech.2021.108947
He C, Ge J, Lian Y, Geng L, Chen Y, Fang D (2022) A concurrent three-scale scheme FE-SCA2 for the nonlinear mechanical behavior of braided composites. Comput Methods Appl Mech Eng 393:114827
Kirchdoerfer T, Ortiz M (2017) Data driven computing with noisy material data sets. Comput Methods Appl Mech Eng 326:622–641
Kirchdoerfer T, Ortiz M (2018) Data-driven computing in dynamics. Int J Numer Meth Eng 113:1697–1710
He Q, Chen J-S (2020) A physics-constrained data-driven approach based on locally convex reconstruction for noisy database. Comput Methods Appl Mech Eng 363:112791
Chen Q, Xie Y, Ao Y, Li T, Chen G, Ren S, Wang C, Li S (2021) A deep neural network inverse solution to recover pre-crash impact data of car collisions. Transp Res Part C 126:103009
Bessa MA, Bostanabad R, Liu Z, Hu A, Apley DW, Brinson C, Chen W, Liu WK (2017) A framework for data-driven analysis of materials under uncertainty: countering the curse of dimensionality. Comput Methods Appl Mech Eng 320:633–667
Berkooz G, Holmes P, Lumley JL (1993) The proper orthogonal decomposition in the analysis of turbulent flows. Annu Rev Fluid Mech 25(1):539–575
Willcox K, Peraire J (2002) Balanced model reduction via the proper orthogonal decomposition. AIAA J 40(11):2323–2330
Astrid P, Weiland S, Willcox K, Backx T (2008) Missing point estimation in models described by proper orthogonal decomposition. IEEE Trans Autom Control 53(10):2237–2251
Carlberg K, Farhat C (2008) A compact proper orthogonal decomposition basis for optimization-oriented reduced-order models. In: 12th AIAA/ISSMO multidisciplinary analysis and optimization conference, p 5964
Carlberg K, Farhat C, Cortial J, Amsallem D (2013) The GNAT method for nonlinear model reduction: effective implementation and application to computational fluid dynamics and turbulent flows. J Comput Phys 242(2013):623–647
Amsallem D, Farhat C (2008) Interpolation method for adapting reduced-order models and application to aeroelasticity. AIAA J 46(7):1803–1813
Ryckelynck D (2005) A priori hyper reduction method: an adaptive approach. J Comput Phys 202(1):346–366
Lu Y, Jones KK, Gan Z, Liu WK (2020) Adaptive hyper reduction for additive manufacturing thermal fluid analysis. Comput Methods Appl Mech Eng 372:113312
Maday Y, Rønquist EM (2002) A reduced-basis element method. J Sci Comput 17(1):447–459
Ladeveze P (1985) New algorithms: mechanical framework and development. Technical Report 57, LMT-Cachan (in French)
Ladeveze P, Rougee P (1985) (Visco)plasticity under cyclic loadings: properties of the homogenized cycle”. Comptes Rendus Academie des Sciences Paris 2(301):891–894 (In French)
Ammar A, Mokdad B, Chinesta F, Keunings R (2006) A new family of solvers for some classes of multidimensional partial differential equations encountered in kinetic theory modeling of complex fluids. J Non-Newton Fluid Mech 139(3):153–176
Chinesta F, Leygue A, Bordeu F, Aguado JV, Cueto E, González D, Alfaro I, Ammar A, Huerta A (2013) PGD-based computational vademecum for efficient design, optimization, and control. Arch Comput Methods Eng 20(1):31–59
Zhang L, Lu Y, Tang S, Liu WK (2022) HiDeNN-TD: Reduced-order hierarchical deep learning neural networks. Comp Methods Appl Mech Eng 389:114414
Lu Y, Li H, Saha S, Mojumder S, Amin A, Suarez D, Liu Y, Qian D, Liu WK (2021) Reduced order machine learning finite element methods: concept, implementation, and future applications. Comput Model Eng Sci 129(1):1351
Lu Y, Li H, Park C, Wagner G, Liu WK (2022) Convolution HiDeNN: formulation and application to topology optimization. Submitted
Chen RT, Rubanova Y, Bettencourt J, Duvenaud DK (2018) Neural ordinary differential equations. In; Advances in neural information processing systems, p 31
Chen T, Chen H (1995) Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems. IEEE Trans Neural Networks 6(4):911–917
Li Z, Kovachki N, Azizzadenesheli K, Liu B, Bhattacharya K, Stuart A, Anandkumar A (2020) Neural operator: graph kernel network for partial differential equations. arXiv:2003.03485
Lu L, Jin P, Pang G, Zhang Z, Karniadakis GE (2021) Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators. Nat Mach Intell 3(3):218–229
Gin CR, Shea DE, Brunton SL, Kutz JN (2021) DeepGreen: deep learning of Green’s functions for nonlinear boundary value problems. Sci Rep 11(1):1–14
Hughes TW, Williamson IA, Minkov M, Fan S (2019) Wave physics as an analog recurrent neural network. Sci Adv 5(12):eaay6946
Liu WK, Gan Z, Fleming M (2021) Mechanistic data science for STEM education and applications, 1st edn. Springer International Publishing, New York
Acknowledgements
The authors would like to acknowledge the sources of some historic facts cited in this paper, which are taken from the following finite element history articles and books: (1) Oden, J.T. (1990). Historical comments on finite elements. In A history of scientific computing, 152–166. (2) Lax, P. (1993). Feng Kang. SIAM News, 26(11). (3) Clough, R.W., and Wilson, E.L. (1999), August. Early finite element research at Berkeley. In Fifth US National Conference on Computational Mechanics, 1–35. (4) Felippa, C.A. (2004). Introduction to finite element methods. University of Colorado, 885.(5) Owen, D.R.J. and Feng, Y.T. (2012). Fifty years of finite elements—a solid mechanics perspective. Theoretical and Applied Mechanics Letters, 2(5), p.051001. (6) Stein, E. (2014). History of the finite element method–mathematics meets mechanics–part I: Engineering developments. In The History of Theoretical, Material and Computational Mechanics-Mathematics Meets Mechanics and Engineering (pp. 399–442). Springer, Berlin, Heidelberg. (7) Clough, R.W. (1980). The finite element method after twenty-five years: a personal view. Computers & Structures, 12(4), 361–370. (8) Wikipedia: Finite Element Method, https://en.wikipedia.org/wiki/Finite_element_method.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access 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, visithttp://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Liu, W.K., Li, S. & Park, H.S. Eighty Years of the Finite Element Method: Birth, Evolution, and Future. Arch Computat Methods Eng 29, 4431–4453 (2022). https://doi.org/10.1007/s11831-022-09740-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11831-022-09740-9