At the heart of the initiative lies the challenge of improving the laser powder bed fusion (LPBF) technique—a dominant method for creating complex metal components used extensively in aerospace, medical implants, and energy systems. LPBF’s precision is often compromised by microscopic powder disturbances caused by the machine’s spreading arm. These disturbances create minute streaks in the metal powder bed, resulting in flaws that can compromise the structural integrity of the final product, forcing costly reprints or total scrapping of parts.

The UNF team’s research aims to embed an intelligent, real-time quality control system within metal 3D printers. This system will continuously monitor each printing layer and detect anomalies as they happen, enabling immediate, automated corrections to mitigate the defects. By harnessing advanced sensors, machine learning algorithms, and digital twin technology, the project seeks to elevate additive manufacturing to a new standard of precision and efficiency, significantly minimizing material waste and energy consumption.

This endeavor promises substantial economic and environmental benefits by reducing the frequency of failed builds, thus lowering production costs and decreasing the environmental footprint associated with metal manufacturing. Beyond immediate industrial applications, the work is poised to strengthen the United States’ manufacturing base by enhancing production yields and facilitating sustainable manufacturing practices.

The project also emphasizes educational impact. It includes the development of new course modules and laboratory activities designed to equip students with skills at the forefront of data-driven automation in manufacturing. By integrating cutting-edge research into the curriculum, UNF aims to prepare the next generation of engineers to thrive in evolving industrial landscapes driven by automation and artificial intelligence.

In a commitment to open science and collaborative advancement, the research team plans to release publicly accessible datasets, trained machine learning models, and decision-making software platforms. The availability of these resources is expected to catalyze broader adoption across both industry and academia, accelerating innovation in metal additive manufacturing worldwide.

The student team members involved in the project are senior students in the advanced manufacturing engineering program: Maria Fernanda Ocrospoma Figueroa, Tessa Baur, and Taylor Uhruh. Notably, Baur and Ocrospoma hold leadership positions in the Society for the Advancement of Material and Process Engineering (SAMPE) Club at UNF—as president and vice president respectively—reflecting their active roles in fostering community and innovation in materials engineering.

Adding to their accolades, Maria Fernanda Ocrospoma Figueroa and Tessa Baur recently excelled in the global Additive Manufacturing Competition at SAMPE 2026 in Seattle, achieving first and second place in category B, respectively. Their achievements underscore the high caliber of talent driving this initiative.

The implications of this research extend beyond academic excellence and into the heart of industrial practice. Metal additive manufacturing stands at the frontier of producing highly complex, customized components that traditional manufacturing methods struggle to fabricate efficiently. By introducing a rapid, adaptive fault correction system into the LPBF process, the research addresses a critical bottleneck that has limited broader industrial uptake.

Furthermore, the advancements envisioned by the UNF team carry significant promise for sectors where component precision and reliability are paramount. Aerospace, for example, demands flawless parts to ensure safety and performance under extreme conditions. Medical implant manufacturing requires exacting standards to enhance biocompatibility and longevity. Energy systems benefit from durable, efficient parts that extend operational lifetimes and sustainability. The improved quality control system will help meet these stringent demands more consistently.

At the technical core, the project leverages the integration of sensor data and real-time analytics, allowing printers to become self-correcting systems. This is achieved through sophisticated modeling of the printing process and feedback loops that adjust laser parameters or recoating actions to prevent defect propagation. The creation of digital twins—virtual replicas of the physical printing environment—enables simulations that predict and prevent failures, setting a new paradigm in additive manufacturing quality assurance.

The University of North Florida is uniquely positioned to propel this research forward, combining expertise across engineering disciplines with strong community and industrial partnerships. With a robust student body exceeding 17,600 and a commitment to individualized faculty-student engagement, UNF exemplifies a modern research university dedicated to impactful innovation and workforce readiness.

This NSF award marks a significant milestone in the journey toward smarter, greener, and more reliable manufacturing technologies. It reflects an emerging consensus on the vital role of automation, data science, and digital twin frameworks in revolutionizing industrial production. As the project unfolds, it promises not only to elevate metal 3D printing but also to inspire new standards for adaptive manufacturing systems globally.

Subject of Research:
Development of real-time quality control systems in metal additive manufacturing using laser powder bed fusion.

Article Title:
University of North Florida Advances Smart Quality Assurance in Metal 3D Printing with NSF Award

News Publication Date:
Not provided.

Web References:
https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2553012
https://www.unf.edu

Image Credits:
University of North Florida

Keywords:
Metal Additive Manufacturing, Laser Powder Bed Fusion, Real-time Quality Control, Digital Twin, Advanced Manufacturing Engineering, National Science Foundation Award, 3D Printing Defect Mitigation, Sustainable Manufacturing, Automation in Manufacturing, Material Engineering Education

ULTEM 9085, known for its exceptional strength-to-weight ratio, thermal stability, and flame-retardant properties, has increasingly gained prominence as the material of choice for structurally demanding additive manufacturing tasks. However, the inherent complexity of 3D printing processes means that tweaking parameters such as layer thickness, print speed, and raster angle can lead to significant variations in mechanical performance. Therefore, this study’s comprehensive approach of coupling experimental data with sophisticated algorithms stands out as a transformative methodology in the quest for tailored material properties through additive manufacturing.

Central to the study is the recognition that traditional trial-and-error methods for parameter optimization are both time-consuming and resource-intensive. By integrating machine learning models trained on extensive experimental datasets, the researchers demonstrate an ability to predict mechanical outcomes such as tensile strength, flexural modulus, and impact resistance with remarkable accuracy. This fusion of empirical and computational techniques not only accelerates the optimization process but also unlocks deeper understanding into how specific parameter interactions govern material behavior at micro and macro scales.

One of the notable elements of this research is the careful selection of 3D printing parameters subjected to analysis. The team systematically varied layer height, infill density, raster angle, print temperature, and print speed, thereby capturing a broad spectrum of conditions typical in industrial 3D printing environments. Each configuration underwent rigorous mechanical testing—ranging from tensile to flexural and impact assessments—to map out the performance landscape of ULTEM 9085. This exhaustive characterization serves as the foundation upon which machine learning models could reliably infer novel parameter-property relationships.

The experimental framework employed meticulous specimen preparation protocols to ensure consistency and reproducibility. Samples were printed using industry-standard fused filament fabrication equipment, with stringent control over environmental factors such as ambient temperature and humidity. This approach mitigated external variables that could skew mechanical testing results, thus preserving the integrity of the dataset for subsequent algorithmic training. The commitment to stringent experimental rigor bolsters confidence in the nuanced trends revealed by the study.

In parallel, the data-intensive nature of the investigation necessitated the application of advanced machine learning techniques. The team evaluated multiple algorithmic frameworks, including random forests, support vector machines, and neural networks, to uncover the most effective for modeling complex, nonlinear relationships inherent in additive manufacturing processes. Through systematic hyperparameter tuning and cross-validation, the study identified optimal models capable of generalizing beyond training data, enabling robust performance predictions across unseen parameter combinations.

The integration of machine learning extends beyond mere prediction, illustrating a novel paradigm where algorithms actively guide experimental design. By analyzing feature importance and sensitivity metrics, the researchers elucidated the dominant printing parameters influencing each mechanical property. For example, print temperature emerged as a critical factor in achieving superior tensile strength, while raster angle significantly impacted flexural behavior. Such insights pave the way for data-informed decision-making in 3D printing, where operators can prioritize adjustments based on their material performance goals.

Another fascinating revelation from the study concerns the interplay between printing parameters and anisotropy in mechanical properties. Due to the layered nature of additive manufacturing, materials frequently exhibit directional dependencies that can undermine structural integrity under real-world loading conditions. The researchers’ analyses revealed how specific combinations of raster angle and layer height either exacerbate or mitigate these anisotropic effects in ULTEM 9085. Consequently, the findings serve as a blueprint for engineering more isotropic parts without compromising print efficiency.

The relevance of this work stretches beyond the laboratory and into industry, where the demand for lightweight, durable components is continually escalating. Aerospace manufacturers, for instance, increasingly rely on ULTEM 9085 for its compliance with stringent regulatory standards and mechanical robustness. By enabling precise tuning of printing conditions based on predictive models, the study empowers these sectors to cut development cycles and reduce material wastage while upholding critical safety margins.

Furthermore, the methodology advanced in this research reflects a broader trend towards digital manufacturing ecosystems powered by artificial intelligence. The coupling of experimental data with machine learning not only optimizes material performance but also facilitates real-time adaptive control during printing. Imagine a future where printers ingest continuous sensor feedback, adjust process parameters instantaneously, and autonomously ensure that each part meets exacting specifications—this study lays foundational groundwork towards such intelligent manufacturing systems.

Notably, the study also underscores the importance of multi-objective optimization in additive manufacturing. Mechanical property enhancement frequently requires balancing competing factors—for instance, maximizing tensile strength while minimizing print time or cost. The researchers’ framework accommodates these trade-offs by enabling tailored predictions across diverse parameter spaces, thus offering manufacturers customizable pathways aligned with specific production priorities.

In addition to improving functional properties, controlling 3D printing parameters also affects surface quality and dimensional accuracy—attributes that are crucial for customer satisfaction and component interoperability. Although not the primary focus of this work, the comprehensive dataset and machine learning models established herein provide a platform for future investigations into how mechanical and aesthetic characteristics co-evolve based on printing strategies.

This study’s implications resonate strongly with sustainability ambitions in manufacturing. By refining parameter selection through predictive analytics, the approach reduces trial runs and waste material, thus curtailing resource consumption and environmental impact. As industries move towards greener production paradigms, insights drawn from such data-driven research will be instrumental in designing leaner, more efficient additive processes.

The convergence of materials science, mechanical engineering, and machine learning epitomized by this research heralds a new era for 3D printing. By uniting human expertise with computational intelligence, it transcends the traditional boundaries between physical experimentation and digital modeling. Such synergy unlocks unprecedented opportunities to customize materials and structures with granular control, fundamentally reshaping how engineers conceive, fabricate, and deploy advanced components.

As additive manufacturing continues its ascent across industrial sectors, the knowledge generated through this study marks a critical leap forward. Not only does it demystify the complex influence of printing parameters on ULTEM 9085’s mechanical properties, but it also showcases the power of interdisciplinary approaches to tackle multifaceted problems. Looking ahead, the integration of real-time data acquisition, cloud computing, and machine learning promises to accelerate innovation cycles, leading to smarter, stronger, and more sustainable 3D printed materials.

Ultimately, the work by Hernandez and colleagues represents a compelling case study in the possibilities unlocked when experimentation meets machine learning in additive manufacturing. Their findings transcend the specifics of ULTEM 9085 to offer a blueprint for the broader materials community eager to harness AI-driven insights. In an era where customization, agility, and precision are paramount, such research charts a visionary course towards the next generation of manufacturing excellence.

Subject of Research: Influence of 3D printing parameters on the mechanical properties of ULTEM 9085 thermoplastic using a combination of experimental testing and machine learning techniques.

Article Title: Influence of 3D printing parameters on ULTEM 9085 mechanical properties using experimentation and machine learning.

Article References:
Hernandez, K.M., O’Brien, S., Bischoff, A. et al. Influence of 3D printing parameters on ULTEM 9085 mechanical properties using experimentation and machine learning. npj Adv. Manuf. 2, 41 (2025). https://doi.org/10.1038/s44334-025-00049-6

Image Credits: AI Generated