Utilizing Topology Optimization and Additive Manufacturing to Increase Accessibility of Highly Capable UAVs

ABSTRACT

Countries such as the US, EU, and Canada maintain strict weight limits on unregistered drones at 250g, significantly restricting their use in fields such as agriculture and environmental modeling. In response to these restrictions, drone designers often reduce weight at the cost of structural integrity or performance by using weaker materials that compromises operational lifespan, or lower capacity batteries and weaker motors, which further degrades usability and flight duration. This study aims to apply Topology Optimization (TO) through Finite Element Analysis (FEA) simulations to reduce the mass of a representative drone whilst maintaining or increasing its structural integrity. Furthermore, the resulting design is tailored to Additive Manufacturing (AM), whose accessibility is offset by weight and strength tradeoffs. Utilizing software, TO guided by FEA removed non-essential material, and the design was then further optimized towards AM, by implementing field driven infill. To assess aerodynamic drag, computational fluid dynamics (CFD) in SolidWorks was utilized. Results demonstrated a maximum reduction of mass by 13.1% (30.2g to 26.7g) and maximum stress by 93% (136MPa to 9.5MPa), with negligible aerodynamic penalties. These results validate the applicability of TO and FEA in leveraging the accessibility of AM towards UAV design, without prior knowledge of structural engineering. This research empowers long-term, regulation-compliant, and high-performance UAV usage being developed by small-scale designers, along with broader adoption in tangential fields using AM.

INTRODUCTION.

Additive manufacturing (AM) is a mature manufacturing process in which material, usually plastics, is added layer by layer to create an object. These objects are often made in machines called 3D printers which intake a design made in CAD (Computer Aided Design) software along with plastic filaments. In recent years, additive manufacturing has become democratized through the mass instalment of machines in public areas such as makerspaces, libraries, and schools. This leap in accessibility has benefitted small-scale makers who are now able to design and manufacture goods iteratively without the typical financial overhead from methods such as injection molding [1]. Specifically in the realm of unmanned aerial vehicles (UAVs), AM’s impact can be seen through the myriad of individually designed UAVs on communal modeling platforms such as TinkerCAD, GrabCAD, and research projects resulting in novel designs that others can also utilize. Due to AM’s recent accessibility and low cost, it has allowed these citizen scientists, hobbyists, and non-commercial users to iteratively design and test UAVs in fields that historically have been fiscally impossible, such as environmental monitoring, agriculture, or deliveries [2]. However, the accessibility of AM, especially in UAV design, comes with one significant drawback: parts created with AM often weigh far more than their counterparts due to material limitations [3]. Conversely, if the designer emphasizes the reduction of the weight of the UAV, the UAV often ends up being structurally weaker due to structurally important material removal [2]. This added weight is especially problematic for these individuals and smaller organizations, as countries such as the US, EU, and Canada set unlicensed UAVs to be below 250 grams [1]. As a result, these smaller designers are often forced to compromise facets of their ideal UAV, such as flight time, performance, or rigidity, for less mass. AM as a field, has tried to counteract this balancing act by implementing what is known as infill, which is the uniform geometric pattern inside of a hollow print. This allows for a far higher weight loss without compromising the strength of the part [2]. Despite these advances, conventional infill strategies are stress agnostic, meaning that the uniform density throughout the length of the part is not adapted to the unique stresses of the part under load. As a result, regions that experience low stress may be over-reinforced, while highly loaded regions, such as motor mounting regions, may not have sufficient internal support. This compromises the structural efficiency of the part and wastes precious mass that could be used in other facets of the drone.

Designers in these fields often use popular modeling workflows such as SolidWorks and nTop, which allow designers to create 3D models of their designs, that can be exported into printable models. These software tools often have tools such as finite element analysis (FEA), which takes in a loading condition, and the material properties of a material, and exports a visual and numerical representation of where their UAVs are under stress and the load paths that make them up [3]. FEA has seen more use in these small-scale design spaces to speed up and forgo physical stress testing and reactive redesigns. This information directly ties into topology optimization (TO), a tool largely unheard of in the space hobbyist, citizen scientists, and alike inhabit, which removes unnecessary material based on FEA stresses and load paths. The amount of material removed is determined by the parameters the user has provided, such as removing 80% of the volume of the part, whilst minimizing the deflection of the part. TO often creates novel and organic loading paths, which would be arduous with conventional designing methods and impossible to create without an extensive education in structural design [4]. Additionally, TO pairs directly into the preexisting workflow many designers already use, with the addition of software such as nTop, which takes a preexisting CAD file, and given the loading and restrained points of the body along with the conditions of the user, will output an optimized design that can be directly printed. These designs bridge the gaps many designers have when remodeling a design to conform to regulations, by preserving the most structurally significant regions of the model. Rather than forcing a perfectly uniform stress profile, TO removes material from low or non-load-bearing regions to create efficient loading paths that are designed to reach the given stiffness and strength goals [4]. Furthermore, due to the strictly computational nature of TO and FEA, users are able to forgo physical testing altogether, as they can directly create, optimize, and test their model in software. This ultimately allows these small-scale designers to use higher capacity batteries for longer flight times, or higher performance internals to increase speed or payload capacity and overall allows the benefits AM has brought to other fields to be applied to UAVs [2].

In specific, FEA models a structure by first discretizing it into a mesh of small elements and assigning the given material properties to each element. Once this mesh is constructed it then solves equilibrium equations as a series of partial differential equations to obtain node displacements and stresses, under a given displacement constraint and loading constraint. By assembling this collection of nodal stiffness contributions into a single global stiffness matrix and solving for any unknown displacements of nodes, FEA can output a 3D map of how the forces are transmitted through a part and hence were stresses would concentrate under a realistic load.

TO further builds upon these well-known FEA tools by posing the structural design process as a optimization problem with constraints in which the distribution of material within a given design domain, known as the bounding box, is adjusted to minimize or maximize objectives such as mass or max displacement. In practice, TO iteratively updates a density field based on FEA results, which typically removes material from regions with low contributions to the objective and retaining or even thickening regions along primary load paths until the resulting FEA converges to the given criteria. While FEA alone identifies where a given design is weak or strong, TO uses repeated FEA simulations to propose new geometries that redistribute the given mass constraint more efficiently, effectively turning FEA into a validation and design creation tool.

As modern tools increasingly include FEA and integrated or suggested routes for applying TO to a given model such as nTop, small scale designers can easily apply these optimization methodologies into their preexisting workflows to create high-performance parts without formal education in structural mechanics or optimization theory.

METHODS.

To best model an application of AM within small-scale UAV design, a widely used UAV model from GrabCAD was procured and the arm, which connects the main electronics hub to the motors that are held at a distance away, was isolated as the region of study (Fig. 1a). This choice reflects the realistic design starting point for small-scale designers to add further functionality, and the arm was chosen in specific, as it contributes the majority of the structural mass of an UAV.

Model Refinement and Design Domain.

The original landing leg below the main motor mount was removed in SolidWorks, so that the arm which primarily carries the thrust and maneuvering loads could be isolated, while the leg which becomes a near zero force feature during flight could be removed. Additionally, the motor mount and mounting bracket were separated and saved as independent CAD bodies, which will be later used to be enforced as passive regions that will not be modified by the TO (Fig. 1b). This will ensure mounting holes will be maintained and not deformed from the meshing process. A rectangular bounding box was then defined around the body of the arm to define the design domain for TO. The main bounding box of the drone arm is a 40 (mm) wide by 48 (mm) tall by 156mm long box with a 15-degree angled wedge 50 (mm) deep underneath the motor mount. This bounding box additionally rises above the main drone surface by 10 (mm) (Fig. 1c). This allows for smooth merging with the frame and passive regions, while also acting as an extended landing element.

Material Choice and Properties.

Polylactic acid (PLA) was selected as the material for all arm design to reflect common practices within AM and small-scale UAV manufacturing. Mechanical properties, that are necessary for accurate loading values, were taken from Wang et al., who experimentally characterized PLA under representative 3D printing settings and conditions, and these values were used as inputs for FEA and TO (Table 1) [5]. However, AM does allow for a wide spectrum of materials, and these exotic materials could be similarly utilized given known mechanical properties.

Table 1. Topology optimization parameters and mechanical properties used for the drone arm FEA and TO.
Target Volume Fraction	0.1
Objective	Minimize Structural Compliance
Filter Size	4mm
Boundary Penalty	0.8
Minimum Objective Change	0.0005
Minimum Density Change	0.01
PLA Young’s Modulus
PLA Poisson’s Ratio	0.331

Meshing and Boundary Conditions.

The combined model of the bounding box and passive mounting regions were then meshed using a standard tetrahedral meshing workflow in nTop. Mesh parameters of a 1.5 (mm) Max Edge Length, 0.5 (mm) Minimum Feature Size, 1 (mm) Tolerance, 70° feature angle, 3° span angle, and linear geometric order were chosen to produce a high-fidelity mesh at the cost of higher computational expense, as high quality meshes are paramount for accurate TO and FEA. If necessary, a small-scale designer could change these parameters within nTop based on their machine’s processing power. The 4 mounting holes on the bracket were constrained as fixed displacement nodes, and a total of 100 (N) upwards was applied to the 4 mounting holes nodes (Fig. 1b). A thrust of 100 (N) per arm approximates worst-case thrust and maneuvering conditions for a small UAV. These constraints and loads were kept constant throughout all subsequent TO and FEA simulations.

Topology Optimization Simulations.

TO was performed within nTop using a density based optimization on the meshed bounding box region with the motor mount and mounting brackets kept as passive regions with loading and displacement constraints respectively (Fig. 1b-d). Key objectives and parameters as outlined in Table 1 were set to minimize compliance under a 100N load, while being subject to a volume-fraction constraint, limiting the material left within the bounding box, and as a proxy for mass. Given that TO is an optimization problem, and specifically TO within nTop being gradient descent there inherently is a risk of converging to a local minima, or in our case, a design that fits the given volume constraint with better compliance, even though a possible design exists with the best compliance. Hence, several TO models with varied parameters, namely boundary penalties, filter size, minimum density change, and minimum objective change (Table 1). The best and final TO arm is shown within Figure 1d.

Field Driven Infill.

To address the stress-agnostic nature of conventional infill as outlined above, a field driven infill approach was taken selected to the final TO arm. A Von-Mises stress field was extracted from a FEA on the final TO arm, showing the stress on each node of the arm. Von Mises stress was used as the primary scalar stress metric for infill and later part validation because it combines the stress state which exists in many different axes into an equivalent value that is widely used to assess yielding and failure in materials under complex loading, as seen in UAV’s. This field was then inputted into a ramping function which mapped local stress values within a region of sufficient width to a gyroid cell size, producing a spatially varying infill density. This infill was then combined with a hollowed version of the TO arm, to yield the stress-aware infill (Fig. 1e).

CFD Analysis. 

To evaluate the aerodynamic performance and viability of the UAV, all arm designs were analyzed using computational fluid dynamics within SolidWorks. Each arm was placed in a 10 (m/s) airflow, representing typical small-scale drone flight speeds, and the drag was computed over a range of angles of attack from 0° to 90° in 10° increments (Fig. 1f). This methodology is representative of typical UAV flight which relies on angles of attack to create translational movement of the UAV, and hence the resulting drag values were averaged.

RESULTS.

Stress Analysis.

Under the 100N load case, a FEA simulation was taken for the original, TO, and TO with Field Driven Infill arms. The original arm contained a maximum Von Mises Stress of , the optimized arm contained , and the field-driven infill arm contained.  Additionally, the original arm held over 10 (mm) of deflection, which is tantamount to complete breakage; the optimized arm held 1.913 (mm), and the field-driven infill design had 2.78 (mm) of deflection.

Mass.

The mass of the parts was found through exporting the parts from nTop as an STL, and importing them into Cura, a popular open-source slicing platform, which converts the model into a printable code for a 3D printer. All support materials and extra materials were removed to ensure the arm alone was being weighed. The model was examined under 25% infill , except for the TO + Infill Model, using PLA plastic. It was found that the original arm weighed 30.2 (g), the optimized arm weighed 29.2 (g), and the field-driven infill design weighed 26.7 (g) (Table 2).

Table 2. Mechanical performance, mass, and average drag across Angles of Attack of the original, topology-optimized, and field-driven infill drone arms.
Design	Max von Mises Stress (Pa)	Tip Deflection (mm)	Mass (g)	Average Drag (N)
Original Arm		>10	30.2	0.201
TO Arm		1.913	29.2	0.263
TO + Field Driven Infill Arm		2.78	26.7	0.263

Drag.

From SolidWorks CFD, as described before, the drag was found from the average of 9 simulations from 0 degrees to 90 degrees, with 10 degrees of angle change per trial. The drag of the original part was 0.201 (N), whereas the optimized and field-driven infill design both had a drag force of 0.263 (N), as the exterior walls of the two models were identical (Table 2).

DISCUSSION.

In this research a model small scale drone design was procured and iterated upon using Topology Optimization and Field Driven Infill in nTop to best optimize the mass to strength ratio of the drone arm, which takes up the bulk of the mass on the drone. Furthermore, these iterations were tested through the use of SolidWorks CFD and nTop FEA. The results of this research underscore the importance of TO when paired with AM for the deployment of sub-250 g UAVs and their components. Through the applications of FEA and TO, the optimized drone arm displayed significantly improved structural strength, whilst marginally increasing the weight of the arm, which further validates the idea that TO can be applied to the preexisting workflows of small-scale designers to produce designs that rival professionally made parts.

The most notable improvements observed were in the reduction of stress and deflection of the arm. The original part exceeded stresses of 136 (MPa), and had a deflection of over 10 (mm). In the context of PLA’s brittle material properties, this stress would lead to an immediate breakage in the arm. In contrast, the topology optimized part and field-driven infill arm showed maximum stresses under 9.5 (MPa) and under 2.78 (mm) of deflection. This ensures that even in the extraneous conditions that 100 (N) may be for UAVs, the drone arm will survive. These large reductions in stress and deflection observed for TO based designs arises from the redistribution of mass and hence the novel loading paths. By thickening members along primary force paths between the motor mount and mounting bracket within the bounding box, the TO has reduced stress concentrations and smoothed stress gradients. This is evident in the visual comparison of the parts, where the original part contains sharp transitions, creating localized peaks in stress and high deformation. In contrast the smooth alignment of material along loading paths, the TO arm uses less mass to achieve greater mechanical performance. This emphasizes the democratizing power of TO, by empowering small-scale designers without formal education to create designs that would otherwise necessitate prior knowledge.

A closer examination of the internal structure of the TO with Field Driven Infill arm (Fig. 1e), clarifies how infill can further be used to contribute to stress and mass reduction. In a conventional uniform infill configuration, the internal lattice density is constant throughout the length of the part, so regions near the mounting bracket and motor mount with high stress get the same internal support as relatively low stress region along the span of the arm. In contrast, the field driven infill design has increased gyroid density, as seen in Figure 1e near the motor mount and mounting bracket, in highly loaded regions. This graded internal structure lends to the specific results and ramping functions, and could be changed to offset the impacts of lowering infill mass or vastly increase mechanical performance under a infill mass contribution limit.

Additionally, as large drone fleets expand and become commercialized, there have been concerns about the long-term strength of drones and how damaged drones may injure individuals. The substantial reduction in maximum Von Mises Stress both optimized arms suggest that these arms may have an increased margin of safety under repeated loading compared to the original geometry. While this work does not explicitly model cyclic loading experiments, lower peak stresses are generally associated with improved resistance to cracking from constant loading, similar to those UAV’s are subjected to. However, rigorous lifetime and fatigue testing must and should be confirmed in future work through simulation and physical testing. Furthermore, the deflection properties exhibited by the optimized and field-driven infill design are important given the impact excessive deflection has on motor stability, alignment, and propulsion, which can severely cut range and UAV performance.

Mass reduction is down 11.5% from the original to the field-driven infill design, without sacrificing performance. This validates the key advantage of TO, which is maintaining performance whilst minimizing mass, especially when paired with field-driven infill. Additionally, in this research, the given parameters in table 1, has led to an optimization that is far stronger than lighter than the original, but given the 250 (g) weight cap on drones, the grading method can and should be changed to incentivize lighter designs.

However, the one tradeoff that emerged was aerodynamic performance. Due to the outside of the part, where the airflow interacts, being the same for both the topology optimized and field-driven infill design, they both had the exact same 30.8% increase in drag. This is primarily due to the vertical nature of the optimized drone arm, which is now perpendicular to the airflow. However, under closer examination, the drag only increased by around 0.06 (N), which is fairly insignificant in the flight of the drone. Furthermore, the topology optimized CFD analysis, as seen in Figure 1f, showed even drag throughout the part, with air flowing smoothly across members, which helps to reduce drag, in comparison to the boxy nature of the original GrabCAD design [7]. Furthermore, drones naturally fly in varying flight angles according to their flight speed, and according to figure 2, the drag of both topology optimized arms is the lowest when the arm is close to 90 Degrees. This would mean that the optimized drone arms are actually most efficient at real-life flying conditions at speed which would best mimic the 10 (m/s) in the SolidWorks CFD and close to 90 Degree flight angle.

Most importantly, these models were made in easily accessible and user-friendly software, already apart or easily integrable into workflows used by small-scale designers. This overall forgoes any preexisting knowledge of trusses, stresses, and strains, and allows the small-scale designer to leverage preexisting AM technology to create reliable and strong UAVs, which fit under regulation.

While the study focused on a single drone arm, the methodology can be applied to a variety of designs, and UAV components, such as brackets, plates, with different materials such as carbon fiber impregnated ABS. Future research could further explore the multi-objective setups to account for drag and stresses in the original optimization. Additionally, further experimental research through flight testing and fatigue analysis, will be necessary to understand layer adhesion in AM.

Beyond improving mechanical performance under a given mass goal, the TO and FEA framework used here can also lend itself to creating more efficient and print friendly workflows. TO can be used with consideration to manufacturing constraints such as overhang angles, feature sizes, and print direction preferences such that the optimized part is not only structurally efficient, but compatible with current AM limitations. Furthermore, this research could be expanded to print time efficiency, by varying proxies of print time, such as build height, toolpath length, or even an estimated print time itself. As AM and computational algorithms continue to incorporate process-aware constraints apart from traditional objectives such as mass or compliance, this methodology has the potential to further advance the high performance and efficiency printable parts small-scale designers can create.

ACKNOWLEDGMENTS.

I would like to thank the CPML lab, Dr. Ravindra Duddu, Dr. Pamela Popp, Ashvin Oli, Duc Tien Nguyen, and the SSMV for their invaluable support and guidance throughout this project.

REFERENCES

1. S. Maricic, I. M. Haber, I. Veljovic, I. Palunko, Implementation of optimum additive technologies design for unmanned aerial vehicle take-off weight increase. Eureka Phys. Eng. 6, 50–60 (2020).
2. J. Bright, R. Suryaprakash, S. Akash, A. Giridharan, Optimization of quadcopter frame using generative design and comparison with DJI F450 drone frame. IOP Conf. Ser. Mater. Sci. Eng.1012, 012019 (2021).
3. V. Jagota, A. P. S. Sethi, K. Kumar, Finite element method: an overview. World J. Sci. Technol. 10, 1–8 (2013).
4. Y. L. Yap et al., Topology optimization and 3D printing of micro-drone: Numerical design with experimental testing. Int. J. Mech. Sci. 237, 107771 (2022).
5. X. Wang, L. Zhao, J. Y. H. Fuh, H. P. Lee, Effect of porosity on mechanical properties of 3D printed polymers: Experiments and micromechanical modeling based on X-ray computed tomography analysis. Polymers 11, 1154 (2019).
6. A. Pandey, Drone with arm. GrabCAD, https://grabcad.com/library/drone-with-arm-1 (accessed March 2025).

---

---

Posted by buchanle on Friday, May 15, 2026 in May 2026.

Tags: Additive Manufacturing, Drone, Topology Optimization