SAND2017-2492J
Additive manufacturing: Toward holistic design Bradley H. Jared1, Miguel A. Aguilo, Lauren L. Beghini, Brad L. Boyce, Brett W. Clark, Adam Cook, Bryan J. Kaehr, Joshua Robbins Sandia National Laboratories PO Box 5800 Albuquerque, NM, 87185 Abstract Additive manufacturing offers unprecedented opportunities to design complex structures optimized for performance envelopes inaccessible under conventional manufacturing constraints. Additive processes also promote realization of engineered materials with microstructures and properties that are impossible via traditional synthesis techniques. Enthused by these capabilities, optimization design tools have experienced a recent revival. The current capabilities of additive processes and optimization tools are summarized briefly, while an emerging opportunity is discussed to achieve a holistic design paradigm whereby computational tools are integrated with stochastic process and material awareness to enable the concurrent optimization of design topologies, material constructs and fabrication processes. Keywords Additive manufacturing, solid freeform processes, simulation, modeling, analytical methods Motivation Additive manufacturing (AM) is leading a renaissance in global manufacturing and product development spurred in part by claims that “complexity is free” [1,2]. AM offers prospects to design complex geometries that can be optimized for performance gains inaccessible under conventional manufacturing constraints. AM further introduces the potential to generate complex, engineered materials with composition gradients, microstructures and properties that are impossible via traditional synthesis techniques. In a similar, synergistic trajectory, topology optimization (TO) is receiving growing attention and use by engineers who seek advanced design tools to leverage the full capabilities of AM materials and processes. Seminal work from both fields emerged in the late 1980s [3,4,5], but research and development activities remained in isolation with earliest references to integration occurring near the turn of the millennium [6,7]. Recent activity is reversing this trend as researchers, engineers and even consumers across a range of applications and disciplines are coupling TO with AM to satisfy growing appetites for robust products, design freedom and high yield production. However, before complexity becomes truly free, or at least significantly cheaper, capability gaps and challenges in materials, processes and optimization tools must be addressed. The discussion that follows will identify key challenges and discuss intersections across the additive and optimization communities where
1
Corresponding author at: Sandia National Laboratories, PO Box 5800, MS 0958, Albuquerque, NM, 81785, USA E-mail address:
[email protected]
1
research and technology maturation is necessary to realize a holistic design paradigm that intimately couples design optimization with process and material capabilities. State-of-the-Art Additive Processes and Materials At their core, additive processes generate material and geometry concurrently as material is deposited, typically in a layerwise fashion, to fabricate three-dimensional parts from solid model representations [8]. As a result, AM introduces the unique possibility to generate and locally control geometry and material at every volume element, i.e. “voxel”, in a part. Complex freeform geometries, internal and reentrant features, architectured materials, and multi-functional, multimaterial parts all become realizable and enter the design space within the additive paradigm. ASTM currently recognizes seven additive process categories; vat photo-polymerization, material extrusion, material jetting, binder jetting, powder bed fusion, directed energy deposition, and sheet lamination [8]. Since process descriptions and details are available through a multitude of sources [9,10,11], general capabilities relative to geometric and material complexity are highlighted. Geometric complexity is inherently available in every additive technique. Polymer based methods are considered the most mature and capable as they represent the earliest additive processes [3,4]. Vat photo-polymerization, i.e. stereolithography, and material jetting leverage photopolymer material systems, enable almost arbitrary geometry forms, and provide the best surface finish, part accuracy and feature resolution among processes [4,12,13]. Material extrusion and powder bed fusion, i.e. selective laser sintering, process a much wider range of thermoplastic materials, including those with fillers [14,15], but are restricted relative to overhang geometries, surface finish and feature resolution [13,16]. Powder bed techniques, i.e. powder bed fusion and binder jetting, are most commonly used to fabricate complex metal part geometries [17,18]. A range of metal alloys are printable as feature resolution is sub-1mm, overhang slopes are limited to roughly 45°, and form accuracy and surface finish are competitive with castings. Directed energy deposition processes a similar range of alloys, but provides higher deposition rates for larger parts, courser features and rougher surfaces that restrict its use for complex structures [17]. Sheet lamination is available for ultrasonically weldable materials where layer thickness and in-process machining capabilities limit part geometry [19]. Material jetting represents an exciting, alternative technology with great promise for geometric complexity. It is traditionally associated with low melting temperature metals such as solders [20], but represents an active research area for a larger alloy range with recent promises for commercial equipment [21,22,23]. While additive processes for ceramics lag polymers and metals, binder jetting has been successful for rapid prototyping due to its ability to produce complex, full color geometries, albeit with poor surface finish and weak materials [24,25]. More recently accessible and useful materials include sand [26], glass and tungsten carbide [18]. Material extrusion is a common process route for ceramics due to its scalability and compatibility with traditional processing routes. Geometry, however, is limited by nozzle shape and feedstock rheology [27,28]. Overhang features are generally inaccessible and surface morphology is dominated by extrusion patterns. Vat photo-polymerization represents a new
2
technology for ceramics that is advancing rapidly and shows great potential for complex parts with feature resolutions below 100µm [29,30]. While geometric freedom is a classic motivation for additively manufactured components, emerging applications are increasingly driven by an unprecedented ability to access material complexity that is now available in three forms. Architected materials, ex. metamaterials or lattices, leverage the geometrical complexity and scaling afforded by AM to create sub-part scale structures that produce effective material properties distinct from fully dense monoliths. Architected materials are primarily of single material constructions and have been used for negative stiffness [31], light-weighting [32], and flexible electronics [33]. Fig. 1 shows one such example, a lattice structure designed to solve the 3D Mitchell beam problem [34]. The loading configuration consists of a vertical load on the center of the right face and a fixed displacement on the left face. Homogenized elastic constants are computed for a 20% dense octahedral lattice unit cell and then used in computation of the optimized topology. A conformal hexahedral mesh is computed and applied to the resultant topology. The octahedral unit structure is then mapped into each hexahedron to arrive at the final part geometry. The structure provides a stiffness that is roughly 70% greater than a fully dense part with identical mass. A 38mm tall version of the structure was fabricated in 316L stainless steel using laser powder bed fusion, while multiphoton lithography was used to print a 500µm tall version of the part in IP-S photosensitive resist. Microstructure control is common in any material formation process, but only additive techniques enable control at discrete voxels through a part volume. Process maps have quantified ranges for microstructure control in additive processes for over a decade [35,36]. Researchers have also used electron beam melting to change the crystallographic texture of grains in Inconel 718 to produce coarse columnar grains, epitaxial deposits and fully equiaxed grains through precise control of process parameters [37], Fig. 2. Such work highlights the ability to generate local microstructures and hence material properties by controlled manipulation of process inputs to accommodate the complex stress state in a part. Multi-material parts contain diverse material compositions within a single geometry, often with multi-functional capability. Polymer processes are most mature as locally varying material properties [38,39,40], color [38,39,40,41,42,43] and material gradients [8,14,15] have been demonstrated using commercial machines [38,39,40,41]. Directed energy deposition provides access to gradient metals through OEM systems [44,45], while powder bed techniques have been demonstrated but not commercialized for multi-material parts [46,47,48]. Sheet lamination provides access to gradient materials and the unique integration of embedded electronics and sensors [19]. Techniques for printing across material types (i.e. polymer, metal, ceramic) are limited, but extremely compelling and are poised to enable an even wider design space. Direct write material extrusion has generated parts with embedded electrical capabilities [49,50], while material jetting has produced opto-electronic devices with printed optics and drop-in electrical components [12]. Design Optimization Generally, the topology optimization problem involves finding the spatial distribution of material attributes that optimizes a performance objective for a design domain and set of requirements. In its simplest form, it is an iterative procedure. First, the response to a candidate design is computed as fields satisfying the state equations are determined for the current design and then used as the basis for evaluation of the performance objective and its sensitivity to change. Field 3
solutions can be found using well-established analysis codes such as Abaqus™ [51], Nastran™ [52], Sierra Mechanics [53], or Albany [54]. Design changes are then determined as the objective, sensitivity, and other information from the performance calculation are passed to an optimization engine that updates the design while enforcing constraints and performance requirements. Basic iteration continues until convergence to an optimal design with respect to functional requirements is met or iteration limits are exceeded. Optimization based design has become an increasingly active and diverse field of research as multiple techniques and tools are now available [55,56,57]. Research codes are readily accessible, but provide limited capabilities and are not properly supported to address user needs [58,59]. Commercial software is more user friendly and can deliver size, shape, bead, topography, topometry and freeform optimization methodologies to complement topology based calculations [60,61,62]. Minimization of compliance is a common structural problem given a fixed mass budget, but tools exist to minimize weight, stress or strain, to achieve a desired frequency response, or to optimize fluid channel flow [60,61,62]. Available boundary constraints include displacements, velocities, accelerations, forces, moments, body loads, contact pressure, center of gravity, temperature, heat flux and transient thermal loads [60,61,62,63]. Design solutions most commonly utilize linear isotropic properties, but orthotropic and anisotropic materials can be specified [62,64]. An important element in any design methodology is its consideration of process constraints. Current commercial software includes some additive process constraints such as minimum feature size [65,66,67] and surface roughness [68]. Overhang constraints have been implemented by researchers for 2D topologies [69], and self-supporting 3D structures have been designed which perform similarly to those designed without manufacturing constraints [70]. Methods are also being developed to account for manufacturing variability [71, 72], build orientation material anisotropies [73] and design-for-AM principles [74]. While no commercial software includes a multi-material capability, research in multi-materials has been active for over two decades [75,76]. In designs with distinct phases [77,78,79], intermediate phases are penalized to prevent mixed volumes. For designs that allow mixed regions, the constitutive response is approximated with mixture rules such as Hashin-Shtrikman bounds or homogenization-based techniques [80,81]. Architected materials, i.e. lattices, have been recently incorporated into tools [34,82], motivated by widespread interest in light-weighting. Technology Gaps and Research Needs The intersection of additive processes and design optimization has introduced revolutionary capabilities for design, product development and manufacturing. “Complexity is free” has been a common mantra; but its promise can currently only be realized where part and/or material requirements are minimal and consequences for failure can be ignored or easily overcome. Anyone involved in the qualification or certification of additive processes or materials will quickly acknowledge that complexity is currently not free, even with recent advances in rapid qualification for AM [83]. Optimization codes have, until recently, resided predominantly in the domain of highly advanced users, i.e. computer scientists, mathematicians and academics. Access to engineers has only occurred in the past few years with the release of commercial software packages. To realize the potential of a more holistic approach, to increase design freedom, and to reduce uncertainties in design solutions and materials; research and development
4
must be performed to make AM processes and materials more capable and deterministic and design tools more expedient. The extensive freedom afforded by AM produces a multi-dimensional design space that extends across process, material and geometry with numerous inputs and constraints. The complexity of this space, however, signifies the necessity for a computational framework to optimize both process and design domains. Fig. 3 depicts a holistic design paradigm whereby additive processes, additive materials, design optimization, and process and material models work in concert to produce an optimal design that meets requirements with margins. In a full embodiment, each element of the paradigm provides input to and receives feedback from the other three elements. Design optimization, for example, would be informed by process capabilities, material distributions and models to generate a design solution which informs the final design analysis, material specifications and process definition. The current state-of-the-art does not approach this framework as material and process knowledge remains incomplete, computational and validation costs remain extensive, and attempts at integration remain immature. Each step to connect elements, ex. incorporating material distributions into optimization toolsets or improvements in process models, will bring significant and immediate benefits to the design space, to the process and material space, to applications and to end users. Additive Processes and Materials Additive processes provide numerous advantages over conventional fabrication techniques, but manufacturing constraints persist. Surface finish, form accuracy and/or process throughput are widely acknowledged challenges whose importance depends on the process, the material, the competing conventional manufacturing technique and the application. Complex geometrical freedom is also never truly arbitrary as boundaries exist relative to minimum feature resolution, maximum part size, down-facing slopes and overhang spans. Some of these limits are related to available hardware, ex. part size, while others are controlled by process physics and material behavior, ex. overhangs. Residual stress, part distortion, shrinkage and/or swelling are also common problems that impact final part geometries. While polymer parts are routinely utilized in their as-printed form, metal and ceramic parts typically require post processing to achieve acceptable surface finish, form accuracy and material properties. A relatively untapped benefit of AM is its potential to control material at the voxel level. This access presents risk and uncertainty, however, as defects can subsequently be introduced at similar scales. Since material and geometry are formed concurrently, traditional techniques to certify and qualify material feedstocks and final geometry independently are no long viable. New methods for material assurance must be developed whereby material performance deterministically meets requirements with acceptable margins. On-going work to understand material properties and process physics must continue since process-structure-propertyperformance (PSPP) relationships spanning material and time scales form the fundamental framework necessary to understand, optimize and control processes. Commercially available AM processes predominantly operate open loop with limited process monitoring or control preventing a priori knowledge of as-printed material performance. In-situ sensing techniques must be implemented to detect, quantify, control and/or correct material defects during fabrication. Such capabilities are preferred to post-process inspection techniques which are 5
expensive, time-consuming, inaccurate and in some cases, impossible to implement. An extension of this work is to implement predictive process controls whereby PSPP based process models, advanced path planning and feedforward controls are used to optimize and assure material performance in context with part geometry, material and performance requirements. Process optimization must incorporate knowledge of how process settings translate into heterogeneous material structure (porosity content, grain size, crystallographic texture, residual stress, etc.) and ultimately into stochastic material properties. Recent work by the authors has shown that structural properties of additive metals can exhibit significant variability when sampled across statistically relevant data sets. Fig. 4 shows a novel high-throughput test array with 120 tensile dogbones that can be tested in a few hours, not days [84]. Fig. 4 compares cumulative probability distributions for strain at failure of wrought 17-4PH with two 17-4PH arrays fabricated by separate vendors using laser powder bed fusion. The additive material clearly underperforms, a trend repeated for yield and ultimate strength, which has been attributed to the presence of lack-of-fusion voids and a microstructure distinctive from wrought material [84]. While defects should not be presumed for every additive material, the work highlights the importance of process controls to insure final material behavior and the need to properly capture and evaluate material distributions to establish accurate design thresholds. The work also highlights a growing need to leverage rapid, high-throughput characterization techniques of material structures and properties to inform holistic, combinatorial process optimization. For processes providing material complexity, the limitations and challenges discussed for part geometry and material performance also apply. While voxel level control is an enticing capability, its practicality and scale is limited to sizes on the order of the material interaction volume, e.g. the melt pool for a laser based process or the droplet size in material jetting. Materials are inherently bounded by microstructures, phases and composition as undesirable material properties, such as weak intermetallics or large residual stresses, are to be avoided in any process. Printing across material classes is further constrained by material and process interactions at interfaces. New, hybrid processes and integrated machine platforms combining multiple additive, and even subtractive, techniques are necessary to access highly integrated, multi-material constructions. Design Optimization Just as developments are necessary to improve the determinism, reliability and flexibility of additive processes, advancements are necessary for optimization toolsets to address real world design problems. Improving the fidelity, efficiency and accuracy while simultaneously increasing the size and complexity of currently available design solutions requires significant computational expense. Traditional high-performing simulation codes are architected for analysis workflows, while commercial optimization algorithms are not suited for massively parallel calculations. These inherent weakness makes the development of novel optimization algorithms difficult within existing optimization engines. One approach to overcoming computational cost is to run problems in parallel on high performance computing (HPC) platforms. Topology optimization capabilities have been demonstrated using an open-source parallel toolkit leveraging multi-grid techniques to run problems approaching 100s of millions of degrees of freedom [85]. The authors have performed TO design problems effectively on 1000s of 6
processors on HPC machines using Sandia’s “PLATO” code [34]. Typical users, however, will not have access to HPC platforms and many optimization packages currently don’t scale to 1000s of processors. With the understanding that “using bigger machines” will not always be the answer, research is on-going to improve algorithm performance and reduce computational expense. Impressive gains have been shown using a “Pareto” based approach that leverages assembly-free solver techniques and GPU processing [86,87]. The authors are also investigating Reduced Order Models (ROM) and automatic mesh pruning and refinement to reduce the computational costs at each iteration and are encouraged with results to date. To facilitate the adoption of AM and enable prompt dissemination of state-of-the-art optimization algorithms, a standard design platform optimized for modern multicore computer architectures is needed. Algorithms in commercial tools do not utilize state-of-the art technologies and may rely on inconsistent formulations and heuristics. A standard platform would enable interoperability between different modeling approaches, analysis solvers, and optimization algorithms to encourage collaborative research and to provide a framework for comparing new approaches and algorithms against standard benchmarks [55]. The design platform would give researchers access to state-of-the art optimization technologies to provide the foundational framework necessary for rapid testing and validation of novel ideas and algorithms. The design platform would also bridge the gap between academic research and production software to maximize impact and return on investments. Finally, the design platform would serve as an efficient delivery mechanism for the latest optimization technologies that have been peer-reviewed and validated for reliable and rapid distribution to end users. One approach to address the stochastic nature of additive materials is to improve process determinism. An alternate, complementary approach is to account for uncertainties in the design process and generate solutions that insure performance requirements are met and margins are quantified. Uncertainties abound in any design scenario and include sources from requirements, boundary conditions and environments; not just process capabilities, feedstocks and final material properties. Advanced uncertainty quantification (UQ) methodologies have been available for years [88,89,90], but even basic capabilities for quantification and propagation in design remain unavailable to end users. Since existing tools do not propagate uncertainties through their analyses, there is no guarantee that their design solutions are globally optimized or robust to failure. While holding great potential and value for product performance and qualification, design under uncertainty is a significant challenge due to the computational resources necessary to create high-fidelity calculations. This computational toll limits the design iterations available to explore designs robust to failure. To make design under uncertainty feasible for complex engineering applications, critical algorithmic issues must be solved. First, reliable optimization algorithms must leverage the simulations performed during statistical analyses to minimize the compute time for large-scale calculations. Novel sampling algorithms are also necessary to reduce sample sizes required to accurately quantify and propagate multiple sources of uncertainty. Finally, algorithms must efficiently utilize all available computing resources to increase performance, speed and accuracy. Any designer must recognize and accommodate fabrication processes. Unfortunately, current optimization tools are essentially process agnostic and oblivious to process phenomena which impact manufacturability, geometry, material properties and final performance. As a result, it is 7
common for TO geometries to require excessive support materials, to be difficult to post process, to be impossible to print or to have features with excessive residual stress and deformation. Incorporation of even heuristically based process constraints within optimization codes could reduce fabrication and design iterations. Limitations associated with feature size are relatively straightforward and have been discussed. Researchers have shown capabilities to generated geometries that consider maximum down-facing slopes and overhang distances to generate selfsupporting design solutions [69,70]. Structures and their performance vary widely with part orientation and have revealed challenges in obtaining globally optimized design solutions. The work is extremely encouraging, however, and illustrates a potential to guide optimal part orientation in printing, reduce or eliminate support structures and minimize post-processing. Similarly, multi-material design solutions have been demonstrated but not propagated into the hands of designers. Gradient materials, for example, have been available for over a decade [91], but their use has been minimal [92] since design tools are unavailable and intuition is limited in its ability to guide designers. Just as the emergence of geometrical complexity via AM has motivated the wider use of optimization tools, introducing material complexity capabilities into design tools will introduce unique and compelling design solutions that motivate further use and advancement of multi-material processes and next generation engineered materials. Integrated Computational Materials Engineering (ICME) Current optimization tools rely on a heuristic paradigm to represent bulk material properties and anisotropies which assume invariance in the manufacturing process. Additive material properties, however, can experience significant local variations, whether controlled or stochastic, based on changes in part geometry [93] and process inputs. Such variations must be considered in design solutions to insure accuracy and robustness. The integration of empirical material variations and distributions using UQ methodologies represents an important first step. While process-structureproperties-performance relationships encompass numerous facets, initial work to integrate them into optimization design tools must start relatively simply due to computational costs. Empirically derived relationships that capture material property trends relative to feature size or build orientation are obvious starting points. Multi-dimensional optimization of process parameters relative to relevant material properties is another significant opportunity [94,95]. Combining optimization with high-throughput characterization techniques discussed previously would have significant near term impact to improve the throughput, efficiency and accuracy of material and process development which are currently empirically based and typically involve large factorial design of experiments. A farther reaching and more ambitious vision would be the implement of an Integrated Computational Materials Engineering (ICME) paradigm [96,97,10] wherein computational tools leverage constitutive process-structure-property-performance relationships to inform design decisions. Within this framework, optimization toolsets could be used to define optimal manufacturing processes, to predict resultant material structures and properties, and to design part topologies and material composition optimized against performance metrics. Such metrics could be encompassing and include not only design requirements such as strength or weight, but process considerations such as residual stress, distortion, cost and through-put.
8
Multiscale computational models that link atomic-scale unit processes through mesoscale phenomena up to macroscale behavior represent an important and powerful component of the ICME paradigm [98]. Researchers across the additive community are developing these models at scales ranging from feedstocks [99] to parts [100] with the intention of quantifying and exploring multi-physics interactions, optimizing process parameters and predicting material heterogeneities [101], part deformation and residual stress [100,102]. Commercial software [103,104] is now available that provides predictive capabilities for residual stress and part distortion for laser powder bed fusion. Such information can be used to inform support structure design and perform process optimization without expensive, iterative part builds. While the complexity of the underlying physics driving additive processes varies across categories and materials, phenomenological models must capture the relevant physics at time and length scales appropriate for accurate prediction of material performance. The complexity of such full scale models will preclude their use in optimization design tools for the foreseeable future due to their extensive computational costs. Instead, reduced order material and process model surrogates must be developed for use during design calculations that have been informed by the full scale models and empirical datasets. Conclusions The potential for additive processes and optimization tools to initiate and sustain disruptive advancements across a range of applications and industries can be far-reaching. Access to geometrical complexity has already impacted product development through design freedom and performance inaccessible through conventional designs and fabrication processes. Extending material complexity across material classes and introducing voxel level control would allow even more radical changes to product design and manufacturing. Imagine printing not just “the box” of a monolithic mechanical structure, but “the box and everything inside it” of a complex, multifunctional, multi-material device. While such constructs will undoubtedly rely initially on heuristic-based design processes; robust, process-aware optimization design tools are ultimately necessary to account for uncertainties in materials, processes, and requirements. As such, there is the opportunity to exceed the practices of conventional manufacturing; to not only optimize the topology during the design phase but to simultaneously optimize process conditions so as to achieve an optimized material structure tailored to satisfy the local requirements at every voxel element [105]. While numerous challenges and opportunities exist before realizing this future vision; concerted, collaborative research and development efforts across additive and optimization communities will be crucial to sustain momentum and guide progress. Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC0494AL85000. This document has been reviewed and approved for unclassified, unlimited release under SAND2017-xxxxA. References [1] T. Friedman, When Complexity is Free, The New York Times, 2013. [2] D. Hartmann, Complexity is Free, Lulu Publishing Services, 2015. 9
[3] C. Hull, Apparatus for production of three-dimensional objects by stereolithography, in: U.S.P. Office (Ed.) 1986. [4] J.J. Beaman, C.R. Deckard, Selective laser sintering with assisted powder handling, Google Patents, 1990. [5] M.P. Bendsøe, N. Kikuchi, Generating optimal topologies in structural design using a homogenization method, Computer methods in applied mechanics and engineering 71(2) (1988) 197-224. [6] K.-H. Chang, P.-S. Tang, Integration of design and manufacturing for structural shape optimization, Advances in Engineering Software 32(7) (2001) 555-567. [7] D. Rosen, Design for additive manufacturing: Past, present, and future directions, Journal of Mechanical Design 136(9) (2014) 090301. [8] ASTM, Standard Terminology for Additive Manufacturing Technologies, 2013. [9] G.N. Levy, R. Schindel, J.P. Kruth, Rapid manufacturing and rapid tooling with layer manufacturing (LM) technologies, state of the art and future perspectives, Cirp Ann-Manuf Techn 52(2) (2003) 589-609. [10] NIST, Measurement roadmap for Metal-Based AM, (2013). [11] W. Gao, Y. Zhang, D. Ramanujan, K. Ramani, Y. Chen, C.B. Williams, C.C. Wang, Y.C. Shin, S. Zhang, P.D. Zavattieri, The status, challenges, and future of additive manufacturing in engineering, Computer-Aided Design 69 (2015) 65-89. [12] K. Willis, E. Brockmeyer, S. Hudson, I. Poupyrev, Printed optics: 3D printing of embedded optical elements for interactive devices, Proceedings of the 25th annual ACM symposium on User interface software and technology, ACM, 2012, pp. 589-598. [13] M. Vaezi, H. Seitz, S. Yang, A review on 3D micro-additive manufacturing technologies, The International Journal of Advanced Manufacturing Technology 67(5-8) (2013) 1721-1754. [14] Stratasys Fused Deposition Modeling Materials. https://www.stratasysdirect.com/materials/fused-deposition-modeling. [15] 3D Systems Production Printers. http://www.3dsystems.com/3dprinters/production/overview. [16] P.-H. Lee, H. Chung, S.W. Lee, J. Yoo, J. Ko, Review: Dimensional Accuracy in Additive Manufacturing Processes, ASME 2014 International Manufacturing Science and Engineering Conference collocated with the JSME 2014 International Conference on Materials and Processing and the 42nd North American Manufacturing Research Conference, ASME, 2014, pp. V001T04A045-V001T04A045. [17] D.D. Gu, W. Meiners, K. Wissenbach, R. Poprawe, Laser additive manufacturing of metallic components: materials, processes and mechanisms, International Materials Reviews 57(3) (2012) 133-164. [18] ExOne, What is Binder Jetting”. http://www.exone.com/Resources/TechnologyOverview/What-is-Binder-Jetting. [19] Fabrisonic, Solid-state welding at low temperature. http://fabrisonic.com/uam-overview. [20] MicroFab. www.microfab.com. [21] Demson, The world’s first multi-metal printer. https://www.demcon.nl/en/showcase/theworlds-first-multi-metal-printer. [22] HP Multi Jet Fusion™. http://www8.hp.com/us/en/printers/3d-printers.html. [23] XJet. www.xjet3d.com. [24] E. Sachs, J. Haggerty, M. Cima, P. Williams, Three-dimensional printing techniques, in: U.S.P. Office (Ed.) 1993. 10
[25] Z Corporation. https://en.wikipedia.org/wiki/Z_Corporation. [26] VoxelJet Materials for 3D printers. http://www.voxeljet.de/en/systems/material. [27] J. Cesarano, P. Calvert, Robocasting": direct fabrication of ceramics and composites, Ceram. Trans 83 (1997) 241. [28] Robocasting, 3-D printing of ceramics. www.robocasting.net. [29] R. Felzmann, S. Gruber, G. Mitteramskogler, P. Tesavibul, A.R. Boccaccini, R. Liska, J. Stampfl, Lithography-Based Additive Manufacturing of Cellular Ceramic Structures, Advanced Engineering Materials 14(12) (2012) 1052-1058. [30] A.D. Lantada, A. de Blas Romero, M. Schwentenwein, C. Jellinek, J. Homa, Lithographybased ceramic manufacture (LCM) of auxetic structures: present capabilities and challenges, Smart Materials and Structures 25(5) (2016) 054015. [31] E.B. Duoss, T.H. Weisgraber, K. Hearon, C. Zhu, W. Small, T.R. Metz, J.J. Vericella, H.D. Barth, J.D. Kuntz, R.S. Maxwell, Three-dimensional printing of elastomeric, cellular architectures with negative stiffness, Advanced Functional Materials 24(31) (2014) 4905-4913. [32] X. Zheng, H. Lee, T.H. Weisgraber, M. Shusteff, J. DeOtte, E.B. Duoss, J.D. Kuntz, M.M. Biener, Q. Ge, J.A. Jackson, Ultralight, ultrastiff mechanical metamaterials, Science 344(6190) (2014) 1373-1377. [33] J.R. Raney, J.A. Lewis, Printing mesoscale architectures, MRS Bulletin 40(11) (2015) 943950. [34] J. Robbins, S.J. Owen, B.W. Clark, T.E. Voth, An efficient and scalable approach for generating topologically optimized cellular structures for additive manufacturing, Additive Manufacturing 12, Part B (2016) 296-304. [35] A. Vasinonta, J.L. Beuth, M. Griffith, Process maps for predicting residual stress and melt pool size in the laser-based fabrication of thin-walled structures, Journal of Manufacturing Science and Engineering 129(1) (2007) 101-109. [36] J. Beuth, N. Klingbeil, The role of process variables in laser-based direct metal solid freeform fabrication, JOM 53(9) (2001) 36-39. [37] R. Dehoff, M. Kirka, W. Sames, H. Bilheux, A. Tremsin, L. Lowe, S. Babu, Site specific control of crystallographic grain orientation through electron beam additive manufacturing, Materials Science and Technology 31(8) (2015) 931-938. [38] Stratasys PolyJet Technology. http://www.stratasys.com/3d-printers/technologies/polyjettechnology. [39] 3D Systems MultiJet Printing Process. http://www.3dsystems.com/media/3d-printingprocess-mjp. [40] HP Multi Jet Fusion™ Technology. HP technical white paper. [41] Makerbot Replicator 2X. https://store.makerbot.com/replicator2x. [42] R.B. Wicker, E.W. MacDonald, Multi-material, multi-technology stereolithography, Virtual and Physical Prototyping 7(3) (2012) 181-194. [43] M. Baumers, C. Tuck, P. Dickens, R. Hague, How can material jetting systems be upgraded for more efficient multi-material additive manufacturing, Proceedings of the Solid Freeform Fabrication (SFF) Symposium. Texas: The University of Texas at Austin, 2014. [44] Sciaky, Electron Beam Additive Manufacturing (EBAM™). http://www.sciaky.com/additive-manufacturing/electron-beam-additive-manufacturingtechnology. [45] Optomec LENS MR-7 Systems. http://www.optomec.com/3d-printed-metals/lensprinters/metal-research-and-development-3d-printer. 11
[46] D. Godlinski, S. Morvan, Steel parts with tailored material gradients by 3D-printing using nano-particulate ink, Materials Science Forum, Trans Tech Publ, 2005, pp. 679-684. [47] M. Ott, M. Zaeh, Multi-material processing in additive manufacturing, Proceedings of the Solid Freeform Fabrication Symposium, 2010, pp. 195-203. [48] L. Jepson, J. Beaman, D. Bourell, K. Wood, SLS processing of functionally gradient materials, Proceedings of the Solid Freeform Fabrication Symposium, 1997, pp. 67-80. [49] Voxel8, The world’s first 3D electronics printer. http://www.voxel8.co. [50] F. Medina, A. Lopes, A. Inamdar, R. Hennessey, J. Palmer, B. Chavez, D. Davis, P. Gallegos, R. Wicker, Hybrid manufacturing: integrating direct-write and stereolithography, Proceedings of the 2005 Solid Freeform Fabrication. Journal of Manufacturing Science and Engineering (2005). [51] Dassault Systems Abaqus Unified FEA. http://www.3ds.com/productsservices/simulia/products/abaqus/. [52] MSC Nastran Multidisciplinary Structural Analysis. http://www.mscsoftware.com/product/msc-nastran. [53] Sandia National Laboratories Sierra, https://sierradist.sandia.gov. [54] Sandia National Laboratories Albany Multiphysics Code. http://gahansen.github.io/Albany/. [55] O. Sigmund, K. Maute, Topology optimization approaches, Structural and Multidisciplinary Optimization 48(6) (2013) 1031-1055. [56] N.P. van Dijk, K. Maute, M. Langelaar, F. Van Keulen, Level-set methods for structural topology optimization: a review, Structural and Multidisciplinary Optimization 48(3) (2013) 437-472. [57] J.D. Deaton, R.V. Grandhi, A survey of structural and multidisciplinary continuum topology optimization: post 2000, Structural and Multidisciplinary Optimization 49(1) (2014) 1-38. [58] RMIT BES0 Topology Optimization Software. http://www.rmit.edu.au/research/researchinstitutes-centres-and-groups/research-centres/centre-for-innovative-structures-andmaterials/software. [59] Technical University of Denmark TopOpt Research Group. http://www.topopt.dtu.dk/?q=node/898/. [60] Simulia Tosca Structure. Dassault Systemes technical white paper. [61] MSC Nastran™ Optimization. MSC Software technical white paper. [62] Genesis Structural Design Optimization Software. Vanderplaats Research and Development, Inc. technical white paper [63] Altair Inspire SolidThinking. www.solidthinking.com/Default.aspx. [64] Simulia Tosca Fluid. Dassault Systemes technical white paper. [65] M. Zhou, Y. Shyy, H. Thomas, Checkerboard and minimum member size control in topology optimization, Structural and Multidisciplinary Optimization 21(2) (2001) 152-158. [66] J.K. Guest, J.H. Prévost, T. Belytschko, Achieving minimum length scale in topology optimization using nodal design variables and projection functions, International journal for numerical methods in engineering 61(2) (2004) 238-254. [67] S.R.M.d. Almeida, G.H. Paulino, E.C.N. Silva, A simple and effective inverse projection scheme for void distribution control in topology optimization, Structural and Multidisciplinary Optimization 39(4) (2009) 359-371. [68] O. Sigmund, Morphology-based black and white filters for topology optimization, Structural and Multidisciplinary Optimization 33(4-5) (2007) 401-424.
12
[69] A.T. Gaynor, J.K. Guest, Topology optimization considering overhang constraints: Eliminating sacrificial support material in additive manufacturing through design, Structural and Multidisciplinary Optimization 54(5) (2016) 1157-1172. [70] M. Langelaar, Topology Optimization of 3D Self-Supporting Structures for Additive Manufacturing, Additive Manufacturing (2016). [71] O. Sigmund, Manufacturing tolerant topology optimization, Acta Mechanica Sinica 25(2) (2009) 227-239. [72] S. Chen, W. Chen, S. Lee, Level set based robust shape and topology optimization under random field uncertainties, Structural and Multidisciplinary Optimization 41(4) (2010) 507-524. [73] R. Hoglund, D.E. Smith, Non-isotropic material distribution topology optimization for fused deposition modeling products, Proceeding of the 2015 Solid Freeform Fabrication Symposium, Austin, TX, pg, 2015. [74] S. Yang, Y.F. Zhao, Additive manufacturing-enabled design theory and methodology: a critical review, The International Journal of Advanced Manufacturing Technology 80(1-4) (2015) 327-342. [75] J. Thomsen, Topology optimization of structures composed of one or two materials, Structural optimization 5(1-2) (1992) 108-115. [76] A.T. Gaynor, N.A. Meisel, C.B. Williams, J.K. Guest, Multiple-material topology optimization of compliant mechanisms created via PolyJet three-dimensional printing, Journal of Manufacturing Science and Engineering 136(6) (2014) 061015. [77] O. Sigmund, S. Torquato, Design of materials with extreme thermal expansion using a three-phase topology optimization method, Smart Structures and Materials' 97, International Society for Optics and Photonics, 1997, pp. 52-60. [78] O. Sigmund, S. Torquato, Design of smart composite materials using topology optimization, Smart Materials and Structures 8(3) (1999) 365. [79] A.M. Mirzendehdel, K. Suresh, A Pareto-Optimal Approach to Multimaterial Topology Optimization, Journal of Mechanical Design 137(10) (2015) 101701. [80] F.V. Stump, E.C. Silva, G.H. Paulino, Optimization of material distribution in functionally graded structures with stress constraints, Communications in numerical methods in engineering 23(6) (2007) 535-551. [81] A. Ramani, Multi-material topology optimization with strength constraints, Structural and Multidisciplinary Optimization 43(5) (2011) 597-615. [82] Altair Design and Optimization of Lattice Structures for 3D Printing. http://insider.altairhyperworks.com/design-and-optimization-of-lattice-structures-for-3d-printingusing-altair-optistruct/. [83] A. Peralta, M. Enright, M. Megahed, J. Gong, M. Roybal, J. Craig, Towards rapid qualification of powder-bed laser additively manufactured parts, Integrating Materials and Manufacturing Innovation 5(1) (2016) 1. [84] B.C. Salzbrenner, J.M. Rodelas, J.D. Madison, B.H. Jared, L.P. Swiler, Y.L. Shen, B.L. Boyce, High-throughput stochastic tensile performance of additively manufactured stainless steel, Journal of Materials Processing Technology 241 (2017) 1-12. [85] N. Aage, E. Andreassen, B.S. Lazarov, Topology optimization using PETSc: An easy-touse, fully parallel, open source topology optimization framework, Structural and Multidisciplinary Optimization 51(3) (2015) 565-572. [86] P. Yadav, K. Suresh, Assembly-Free Large-Scale Modal Analysis on the GPU, Journal of Computing and Information Science in Engineering 13(1) (2013). 13
[87] K. Suresh, Efficient generation of large-scale pareto-optimal topologies, Structural and Multidisciplinary Optimization 47(1) (2013) 49-61. [88] I. Babuška, F. Nobile, R. Tempone, A stochastic collocation method for elliptic partial differential equations with random input data, SIAM Journal on Numerical Analysis 45(3) (2007) 1005-1034. [89] I. Babuska, R. Tempone, G.E. Zouraris, Galerkin finite element approximations of stochastic elliptic partial differential equations, SIAM Journal on Numerical Analysis 42(2) (2004) 800-825. [90] R.G. Ghanem, P.D. Spanos, Stochastic finite elements: a spectral approach, Courier Corporation, 2003. [91] M.T. Ensz, M.L. Griffith, D.E. Reckaway, Critical issues for functionally graded material deposition by laser engineered net shaping (LENS), Proceedings of the 2002 MPIF Laser Metal Deposition Conference, San Antonio: TX, 2002. [92] D.C. Hofmann, J.P.C. Borgonia, R.P. Dillon, E.J. Suh, P.B. Gardner, Applications for Gradient Metal Alloys Fabricated Using Additive Manufacturing, NASA Tech Briefs, 2013, pp. 54-56. [93] B. Brown, W. Everhart, J. Dinardo, Characterization of bulk to thin wall mechanical response transition in powder bed AM, Rapid Prototyping Journal 22(5) (2016) 801-809. [94] C. Kamath, B. El-dasher, G.F. Gallegos, W.E. King, A. Sisto, Density of additivelymanufactured, 316L SS parts using laser powder-bed fusion at powers up to 400 W, The International Journal of Advanced Manufacturing Technology 74(1-4) (2014) 65-78. [95] C. Kamath, Data mining and statistical inference in selective laser melting, The International Journal of Advanced Manufacturing Technology 86(5-8) (2016) 1659-1677. [96] N.R.C.C.o.I.C.M. Engineering, Integrated computational materials engineering: a transformational discipline for improved competitiveness and national security, National Academies Press, 2008. [97] J. Allison, D. Backman, L. Christodoulou, Integrated computational materials engineering: a new paradigm for the global materials profession, JOM 58(11) (2006) 25-27. [98] M.F. Horstemeyer, Integrated Computational Materials Engineering (ICME) for metals: using multiscale modeling to invigorate engineering design with science, John Wiley & Sons, 2012. [99] W. King, A. Anderson, R. Ferencz, N. Hodge, C. Kamath, S. Khairallah, Overview of modelling and simulation of metal powder bed fusion process at Lawrence Livermore National Laboratory, Materials Science and Technology 31(8) (2015) 957-968. [100] N. Hodge, R. Ferencz, R. Vignes, Experimental Comparison of Residual Stresses for a Thermomechanical Model for the Simulation of Selective Laser Melting, Additive Manufacturing (2016). [101] T.M. Rodgers, J.D. Madison, V. Tikare, M.C. Maguire, Predicting mesoscale microstructural evolution in electron beam welding, JOM 68(5) (2016) 1419-1426. [102] M. Francois, A. Sun, W. King, N. Henson, D. Tourret, C. Bronkhorst, N. Carlson, C. Newman, T. Haut, J. Bakosi, Modeling of additive manufacturing processes for metals: Challenges and opportunities, Current Opinion in Solid State and Materials Science (2017). [103] 3DSim exaSIM. http://3dsim.com/product/exasim. [104] Netfabb simulation. www.netfabb.com/capabilities/simulation. [105] J. Zowe, M. Kočvara, M.P. Bendsøe, Free material optimization via mathematical programming, Mathematical programming 79(1-3) (1997) 445-466. 14
Fig 1. A 3D Mitchell structure (left) designed with tetrahedron lattices [34] that was fabricated using laser powder bed fusion (center) and multi-photon lithography (right).
Fig 2. Control of grain texture in AM Inconel 718 via electron-beam melting [37].
Fig 3. Key elements of a holistic design paradigm for the optimized design and fabrication of additively manufactured parts that incorporates process and material awareness.
15
Wrought
Fig 4. A 50x50mm 17-4PH high-throughput tensile array (left) and cumulative probability distributions for strain at failure (right) for laser powder bed fusion 17-4PH compared to wrought 17-4PH [84].
16
