TOBACO: Topology Optimization via Band-limited Coordinate Networks for Compositionally Graded Alloys

Aaditya Chandrasekhar, Stefan Knapik, Deepak Sharma, John Reidy, Ian McCue, Jian Cao, Wei Chen · 2025

Read TOBACO: Topology Optimization via Band-limited Coordinate Networks for Compositionally Graded Alloys by Aaditya Chandrasekhar, Stefan Knapik, Deepak Sharma, John Reidy, Ian McCue, Jian Cao, Wei Chen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-10320.txt.

Introduction

Aaditya Chandrasekhar and co-authors design graded alloy parts with a topology method that respects making limits, in a 2025 paper. Graded alloys vary their mix across a part, so strength and weight can be tuned spot by spot. New printing methods can build them, but the mix must not change too fast from point to point.

Their frame describes the mix with a band limited coordinate neural network. Capping the network's bandwidth enforces the change limit by itself, with no added constraint terms. The approach also gives mesh free designs, sharp detail, and end to end gradients. Tests on stretch and heat plus stretch cases showed the method produces printable, well performing graded parts.

Catalog record

CREATORAaditya Chandrasekhar, Stefan Knapik, Deepak Sharma, John Reidy, Ian McCue, Jian Cao, Wei Chen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-10320

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

TOBACO: Topology Optimization via Band-limited Coordinate Networks for Compositionally Graded Alloys Aaditya Chandrasekhar, Stefan Knapik, Deepak Sharma, John Reidy, Ian McCue, Jian Cao, Wei Chen Abstract Compositionally Graded Alloys (CGAs) offer unprecedented design flexibility by enabling spatial variations in composition; tailoring material properties to local loading conditions. This flexibility leads to components that are stronger, lighter, and more cost-effective than traditional monolithic counterparts. The fabrication of CGAs have become increasingly feasible owing to recent advancements in additive manufacturing (AM), particularly in multi-material printing and improved precision in material deposition. However, AM of CGAs requires imposition of manufacturing constraints; in particular limits on the maximum spatial gradation of composition. This paper introduces a topology optimization (TO) based framework for designing optimized CGA components with controlled compositional gradation. In particular, we represent the constrained composition distribution using a band-limited coordinate neural network. By regulating the network’s bandwidth, we ensure implicit compliance with gradation limits, eliminating the need for explicit constraints. The proposed approach also benefits from the inherent advantages of TO using coordinate networks, including mesh independence, high-resolution design extraction, and end-to-end differentiability. The effectiveness of our framework is demonstrated through various elastic and thermo-elastic TO examples. K eywords Topology Optimization \cdot Compositionally Graded Alloys \cdot Manufacturing Constraint \cdot Neural Networks 1 Introduction Topology optimization (TO) [ 1 ] is a class of computational methods that optimizes material distribution within a given design domain, subject to specified objectives and constraints. These methods can navigate complex design landscapes that challenge conventional, intuition-based approaches. However, conventional TO focused on designing with a single material may be insufficient when multiple, conflicting functional requirements are present [ 2 ] . For instance, a turbine blade requires varied local properties such as surface oxidation resistance, tip creep resistance, and low thermal expansion at its base, while minimizing weight and reducing the amount of critical materials [ 3 ] . It is difficult for a single material to satisfy all these demands. To balance these conflicts, researchers have explored designs incorporating multiple materials [ 4 , 5 , 6 ] ; in particular compositionally graded alloys (CGAs) [ 7 ] . CGAs ( Figure 1 ) provide enhanced design flexibility by allowing material composition to vary spatially [ 8 , 9 ] . This enables the precise tailoring of material properties to local requirements, offering a way to balance conflicting design objectives [ 10 , 11 ] . However, this flexibility complicates both the design and manufacturing processes. Recent advances in Additive Manufacturing (AM) have enabled the fabrication of complex metal [ 12 , 13 ] and polymer multi-material parts, including those with graded compositions. Despite this progress, the AM of multi-material components [ 14 , 15 , 16 ] imposes its own constraints, particularly limitations on the achievable composition gradation. Figure 1: (a) Illustration of Additive Manufacturing of CGA using Directed Energy Deposition [ 17 ] . (b) CGA with the bottom layer printed with 100\% IN625, top layer with 100\% IN718 and graded in between. Image courtesy of Rujing Zha and Rowan Rolark. This paper focuses on incorporating these gradation constraints into the TO framework [ 18 ] . Such constraints arise from two primary sources. First, sharp transitions between dissimilar alloys can induce thermal stresses, leading to delamination [ 19 , 20 ] . Second, manufacturing process parameters and machine capabilities inherently limit how rapidly the composition can be varied [ 21 ] . To address these limitations, we introduce a method for imposing gradation constraints in the TO of CGAs. Our approach first formulates the gradation constraint within the Fourier space. We leverage signal processing theory [ 22 ] to demonstrate that a constraint on the maximum gradation of composition directly corresponds to imposing a band limit on its Fourier spectrum. Then, extending a recently proposed neural network (NN) based design representation [ 23 ] , we represent the compositions using a band-limited NN [ 24 ] to implicitly enforce this maximum gradation constraint. Finally, we formulate a TO framework for designing structures composed of CGAs using this band-limited representation ( Figure 2 ). While our primary focus is on elastic and thermo-elastic problems, the proposed method is broadly applicable. Figure 2: Overview of the proposed method. The inputs are (a) a domain subjected to thermo-elastic loads and (b) a material system with composition and phase dependent properties. A neural network then maps coordinates to composition and material properties (c). The weights of the network are optimized to determine optimal material distribution. The gradation of the design composition is controlled by varying the coordinate network bandwidth, \bar{B} , to produce (d) optimized designs with varying gradation limits. 2 Related Work This work contributes to the expanding research aimed at combining the design flexibility of additive manufacturing (AM) with the computational design capabilities of topology optimization (TO) to develop higher-performing components [ 25 , 26 , 27 ] . Specifically, this study focuses on enhancing the manufacturability of CGA designs optimized via TO by explicitly incorporating manufacturing constraints. Topology optimization encompasses various methods, such as density-based approaches [ 28 ] , level-set approaches [ 29 ] and geometric projections [ 30 ] . Our research utilizes the density-based approach as its foundational framework. Density-based TO discretizes the design domain into finite elements and optimizes the fictitious material density in each element, traditionally used for single-material optimization. Recently, there has been significant interest in expanding TO methodologies to handle multi-material scenarios [ 31 ] . These include designs featuring discrete domains of distinct materials [ 32 ] and functionally graded materials (FGMs) characterized by smooth transitions between materials [ 33 ] . Such methods substantially enhance the design space by enabling spatial tailoring of material properties. Functionally graded structures, including porous configurations, have found notable applications in aerospace and biomedical industries due to their superior structural performance and multiphysics capabilities [ 34 , 35 , 36 , 37 ] . A prime aerospace example is turbine blades, which face simultaneous mechanical loads and thermal gradients, thus benefiting from spatially varying properties. Recent studies demonstrated that graded lattice structures optimized via TO significantly outperform homogeneous designs, enabling denser architectures near heated regions for improved heat conduction and lighter structures elsewhere to enhance cooling efficiency [ 38 ] . Moreover, FGMs have also proven valuable in thermal barrier coatings [ 3 ] . Extensive reviews of multi-material components and FGM applications can be found in [ 39 , 40 ] . CGA, a specialized category within FGMs , are increasingly utilized in critical aerospace, nuclear, and biomedical components to meet specific local requirements, such as corrosion resistance in one region and creep resistance in another, often necessitating the combination of distinct alloys [ 41 , 42 , 43 , 44 ] . Manufacturing optimized CGA designs requires addressing specific process-related constraints. While manufacturing constraints for conventional subtractive processes, such as turning, milling, drilling, have been widely explored [ 45 ] ; additive manufacturing techniques, such as powder-bed fusion or extrusion-based 3D printing, permit greater geometric freedom but introduce new constraints [ 7 ] . Notably, overhang constraints are critical in AM, as unsupported overhangs beyond certain angles may cause fabrication failures [ 46 ] . Topology optimization methods

Read the complete text