An official website of the United States government
Abstract The development of high‐performance elastomers for additive manufacturing requires overcoming complex property trade‐offs that challenge conventional material discovery pipelines. Here, a human‐in‐the‐loop reinforcement learning (RL) approach is used to discover polyurethane elastomers that overcome pervasive stress–strain property tradeoffs. Starting with a diverse training set of 92 formulations, a coupled multi‐component reward system was identified that guides RL agents toward materials with both high strength and extensibility. Through three rounds of iterative optimization combining RL predictions with human chemical intuition, we identified elastomers with more than double the average toughness compared to the initial training set. The final exploitation round, aided by solubility prescreening, predicted twelve materials exhibiting both high strength (>10 MPa) and high strain at break (>200%). Analysis of the high‐performing materials revealed structure‐property insights, including the benefits of high molar mass urethane oligomers, a high density of urethane functional groups, and incorporation of rigid low molecular weight diols and unsymmetric diisocyanates. These findings demonstrate that machine‐guided, human‐augmented design is a powerful strategy for accelerating polymer discovery in applications where data is scarce and expensive to acquire, with broad applicability to multi‐objective materials optimization.
more »
« less
Accelerated discovery of 3D printing materials using data-driven multiobjective optimization
Additive manufacturing has become one of the forefront technologies in fabrication, enabling products impossible to manufacture before. Although many materials exist for additive manufacturing, most suffer from performance trade-offs. Current materials are designed with inefficient human-driven intuition-based methods, leaving them short of optimal solutions. We propose a machine learning approach to accelerating the discovery of additive manufacturing materials with optimal trade-offs in mechanical performance. A multiobjective optimization algorithm automatically guides the experimental design by proposing how to mix primary formulations to create better performing materials. The algorithm is coupled with a semiautonomous fabrication platform to substantially reduce the number of performed experiments and overall time to solution. Without prior knowledge of the primary formulations, the proposed methodology autonomously uncovers 12 optimal formulations and enlarges the discovered performance space 288 times after only 30 experimental iterations. This methodology could be easily generalized to other material design systems and enable automated discovery.
more »
« less
- Award ID(s):
- 1955697
- PAR ID:
- 10374342
- Date Published:
- Journal Name:
- Science Advances
- Volume:
- 7
- Issue:
- 42
- ISSN:
- 2375-2548
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Anisotropy in additive manufacturing (AM), particularly in the material extrusion process, plays a crucial role in determining the actual structural performance, including the stiffness and strength of the printed parts. Unless accounted for, anisotropy can compromise the objective performance of topology-optimized structures and allow premature failures for stress-sensitive design domains. This study harnesses process-induced anisotropy in material extrusion-based 3D printing to design and fabricate stiff, strong, and lightweight structures using a two-step framework. First, an AM-oriented anisotropic strength-based topology optimization formulation optimizes the structural geometry and infill orientations, while assuming both anisotropic (i.e., transversely isotropic) and isotropic infill types as candidate material phases. The dissimilar stiffness and strength interpolation schemes in the formulation allow for the optimized allocation of anisotropic and isotropic material phases in the design domain while satisfying their respective Tsai–Wu and von Mises stress constraints. Second, a suitable fabrication methodology realizes anisotropic and isotropic material phases with appropriate infill density, controlled print path (i.e., infill directions), and strong interfaces of dissimilar material phases. Experimental investigations show up to 37% improved stiffness and 100% improved strength per mass for the optimized and fabricated structures. The anisotropic strength-based optimization improves load-carrying capacity by simultaneous infill alignment along the stress paths and topological adaptation in response to high stress concentration. The adopted interface fabrication methodology strengthens comparatively weaker anisotropic joints with minimal additional material usage and multi-axial infill patterns. Furthermore, numerically predicted failure locations agree with experimental observations. The demonstrated framework is general and can potentially be adopted for other additive manufacturing processes that exhibit anisotropy, such as fiber composites.more » « less
-
Cellular materials widely exist in natural biologic systems such as honeycombs, bones, and woods. With advances in additive manufacturing, research on cellular metamaterials is emerging due to their unique mechanical performance. However, the design of on-demand cellular metamaterials usually requires solving a challenging inverse design problem for exploring complex structure–property relations of microstructured representative volume elements (RVEs) in the design domain. Here, we propose an experience-free and systematic methodology for exploring a parametrized system for microstructures of cellular mechanical metamaterials using a multiobjective genetic algorithm (GA). Globally, by considering the importance of the initial population selection for a population-based heuristic optimization method, we study the impact of the populations initialized by the different sampling methods on the optimal solutions. Locally, we develop our method by using a micro-GA with a new searching strategy, which requires the standard genetic algorithm to be conditionally run for a sufficient number of times with a small population size during the global searching process. We have applied our method to explore optimal solutions for applications mapped on two different parameter spaces of the cellular mechanical metamaterials with periodic and nonperiodic RVEs effectively and accurately.more » « less
-
Additive manufacturing (AM) enables the fabrication of complex, highly customized geometries. However, the design and fabrication of structures with advanced functionalities, such as multistability and fail-safe mechanism, remain challenging due to the significant time and costs required for high-fidelity simulations and iterative prototyping. In this study, we investigate the application of Bayesian Optimization (BO), an advanced machine learning framework, to accelerate the discovery of optimal AM compatible designs with such advanced properties. BO uses a probabilistic surrogate to strategically balances the exploration of design space with few test designs and the exploitation of design space near current best performing designs, thereby reducing the number of design simulations needed. While existing studies have demonstrated the potential of BO in AM, most have focused on static or simple designs. Here, we target multistable structures that can reconfigure among multiple stable states in response to external conditions. Since mechanical performance (e.g., strength) is configuration-dependent, our goal is to identify high performing designs while ensuring that strength in all stable configurations exceeds a prescribed threshold for structural robustness.more » « less
-
Additive manufacturing (AM) has impacted the manufacturing of complex three-dimensional objects in multiple materials for a wide array of applications. However, additive manufacturing, as an upcoming field, lacks automated and specific design rules for different AM processes. Moreover, the selection of specific AM processes for different geometries requires expert knowledge, which is difficult to replicate. An automated and data-driven system is needed that can capture the AM expert knowledge base and apply it to 3D-printed parts to avoid manufacturability issues. This research aims to develop a data-driven system for AM process selection within the design for additive manufacturing (DFAM) framework for Industry 4.0. A Genetic and Evolutionary Feature Weighting technique was optimized using 3D CAD data as an input to identify the optimal AM technique based on several requirements and constraints. A two-stage model was developed wherein the stage 1 model displayed average accuracies of 70% and the stage 2 model showed higher average accuracies of up to 97.33% based on quantitative feature labeling and augmentation of the datasets. The steady-state genetic algorithm (SSGA) was determined to be the most effective algorithm after benchmarking against estimation of distribution algorithm (EDA) and particle swarm optimization (PSO) algorithms, respectively. The output of this system leads to the identification of optimal AM processes for manufacturing 3D objects. This paper presents an automated design for an additive manufacturing system that is accurate and can be extended to other 3D-printing processes.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1126/sciadv.abf7435
