Metal additive manufacturing (AM), once confined to rapid prototyping, is now a cornerstone of modern defense strategies, particularly for sustaining aging military platforms with complex, low-volume geometry-rich spare parts. Traditional supply chains, however, are increasingly ill-equipped to meet these demands. Obsolete tooling, vanishing suppliers, and fragile logistics present significant vulnerabilities. The ability to print mission-critical metal spares near the point of need offers a compelling solution, but a persistent question looms large for defense ministries: can parts printed in the field be trusted without extensive, centralized laboratory testing?
To address this critical gap, the European Defence Fund (EDF) has supported the DISCMAM project (Digital Supply Chain for On-Site Maintenance by Additive Manufacturing). This initiative is pioneering a transformative approach, moving beyond conventional physical testing to embrace digital qualification. By leveraging in-situ monitoring and advanced digital twins, DISCMAM delivers real-time quality assurance for laser powder bed fusion (LPBF) processes. This groundbreaking method combines functional acceptance criteria, artificial intelligence-driven defect detection, and statistical modeling to classify printed parts into graded acceptance levels immediately after production. These advancements are already being integrated into Fieldmade’s deployable NOMAD03 microfactory, signaling a tangible shift towards operationalizing this technology.
The Paradigm Shift: Digital Evidence Replaces Physical Testing
The core objective of the DISCMAM project is not to eliminate physical testing entirely, but to replace the resource-intensive, time-consuming, and often destructive traditional methods with a robust digital qualification framework. Instead of relying on techniques such as tensile testing, micrographic porosity analysis, or comprehensive metrology for every printed part, DISCMAM capitalizes on the digital footprint generated during the AM build process. By meticulously collecting in-situ data from each layer as it is printed, the project constructs a detailed digital record of the manufacturing process.
This revolutionary approach hinges on three fundamental elements:
-
Operationally Meaningful Acceptance Levels: Recognizing that not all parts or missions require the same level of assurance, DISCMAM has developed a tiered system. This framework moves away from a rigid pass/fail against strict Original Equipment Manufacturer (OEM) specifications, which is often impractical in theater. Instead, it defines graded acceptance levels that reflect the expected service life of a part and potential mitigations in its use. Crucially, this system does not depend on measuring absolute material characteristics in the field, which is frequently infeasible. Instead, it utilizes relative quality intelligence to build confidence in the manufactured components.
-
Correlated In-Situ Metrics: A sophisticated monitoring system is essential to detect and quantify process anomalies. DISCMAM’s approach focuses on metrics such as streaking, spatter, pinning, and warpage. These events have been statistically linked to common defects like porosity, dimensional deviations, and undesirable surface properties. By precisely measuring and correlating these in-situ indicators, the system provides early warnings of potential quality issues.
-
Fast-Response Digital Twin: The third critical component is a high-speed digital twin. This sophisticated model acts as a virtual replica of the printing process. It ingests the in-situ metrics captured during the build, alongside the process parameters, to generate an evidence-based classification of the part’s quality immediately upon completion of the printing cycle. This allows for rapid decision-making regarding the part’s suitability for deployment.
The DISCMAM project has successfully developed and validated all three of these elements using defense-relevant parts on an industrial LPBF platform. The Danish Technological Institute (DTI), in collaboration with Euler3D and Fieldmade, has designed and piloted a digital AM supply chain tailored for on-site maintenance. Their efforts have focused on two specific aluminum use cases, provided by the Belgian Ministry of Defence (MoD). Both components were manufactured using LPBF technology with AlSi10Mg powder on a Nikon SLM Solutions 280 machine at DTI.
Figure 1: The Unimog Compressor Cover – A Critical Component for Field Reliability
The first use case involved the 3D printing of a compressor cover for a Mercedes Unimog heavy truck. This specific component, part of the U1350L air-compressor system, is designed to enclose and seal a pressurized volume. Its functionality relies heavily on a gasketed downskin face, which is susceptible to failure through gasket leakage or mechanical cracking. The complexity of its geometry and the critical nature of its sealing function made it an ideal candidate to test the DISCMAM framework’s ability to ensure quality for vital spare parts.
Figure 2: Unimog Fuel Filter Housing – Ensuring Fuel System Integrity
The second use case focused on a 3D printed fuel filter housing for the same Unimog vehicle. This component plays a crucial role in the fuel system, holding filters under slight vacuum upstream of the fuel pump. Imperfect sealing faces or threaded interfaces can lead to vacuum leaks, while structural failures can arise from mishandling or vibration. These parts exemplify the challenges faced in defense AM: intricate geometries, critical mating surfaces, and well-defined failure modes directly related to porosity, dimensional accuracy, and surface texture.
Tailored Acceptance: Urgency and Function Dictate Quality Standards
Building upon the technical and operational requirements of these use-case parts, DISCMAM has translated them into a set of functional acceptance criteria for LPBF aluminum spares. This framework establishes four distinct acceptance levels, catering to varying degrees of urgency and criticality:
- Level 1: Permanent Replacement: Parts meeting this highest standard are deemed suitable for immediate, long-term replacement of the original component, with no functional limitations.
- Level 2: Temporary Replacement: Components at this level can serve as a reliable substitute for a defined period, potentially with minor operational considerations.
- Level 3: Emergency Use: This designation applies to parts intended for critical, short-term operational needs, where functionality is paramount but performance may be temporarily compromised.
- Level 4: Scrap: Parts falling into this category do not meet the minimum quality requirements and are deemed unfit for use.
Quantitative ranges have been proposed for each level and use case, encompassing parameters such as hardness and ultimate tensile strength (UTS) bands, acceptable roughness limits for both functional and non-functional downskins, and thread torque windows. While establishing absolute thresholds remains a recognized challenge for both Ministries of Defence and OEMs due to a scarcity of legacy data, the DISCMAM framework offers a pragmatic solution. Instead of relying on un-validatable absolute boundaries, it employs relative digital evidence to build operational confidence. A critical aspect of this approach is that any digital intelligence indicating poor quality or process anomalies serves as an immediate disqualifier, even for Level 3 emergency use, as such deviations signal a compromised production integrity.
These graded acceptance levels are the bedrock of DISCMAM’s LPBF quality assurance strategy. The in-situ monitoring system and the digital twin do more than simply flag defects; they indicate the likely acceptance level a part will fall into, thereby providing clear guidance on how and for how long it can be safely utilized.
From Defect Maps to Quantifiable Quality Insights
During LPBF builds, Euler3D’s sophisticated system captures layer-by-layer images of the powder bed and the exposed surface. Utilizing a combination of classical image processing techniques and deep neural networks, this technology quantifies key defect modes. These include streaking (often caused by recoater blade damage), spatter accumulation, pinning (where powder adheres to the melt pool), warpage or obstructions, smoke events, and localized powder loss.
This capability was deployed and rigorously validated at DTI during the production of the Unimog compressor cover and fuel filter housing using a Nikon SLM Solutions 280 machine. For each printed part, defect frequency plots were generated, illustrating defect distribution across the build height. Spatial heatmaps revealed areas of concentrated defects across the build plate. Integrated metrics, such as total streaking volume, spatter count, and overall defect area, were also compiled.
These in-situ data were then meticulously compared with post-build analyses. This included high-resolution computed tomography (CT) scans to measure porosity and geometric deviations (root mean square error and total deviation against the CAD model), downskin roughness measurements, Profilometry-based Indentation Plastometry (PIP) for local mechanical properties, and hardness testing.
The correlations observed were compelling:
- Streaking and Spatter as Indicators of Porosity: Higher levels of streaking and spatter, as detected by Euler3D, were consistently associated with increased CT-measured porosity and a higher number of indications in functionally critical regions. This strongly confirmed that recoater-induced disturbances are reliable proxies for lack of fusion and internal pores.
- Recoater Damage and Geometric Deviation: Specimens exhibiting dense streaking signatures, deliberately induced by damaging the recoater lip, also showed elevated geometric deviations when compared to their CAD models via CT scans. This highlights the direct impact of recoater integrity on dimensional accuracy.
- Surface Roughness and Streaking: Downskin roughness was found to be significantly influenced by streaking patterns. Continuous linear disturbances translated into extended variations in height, while localized pinning events contributed to sharp, protruding surface features.
- Density and Mechanical Performance: Samples achieving relative densities at or above approximately 99.9% – a result of optimized energy density and gas flow during printing – exhibited higher yield strength, UTS, and hardness. Conversely, parts with slightly lower densities and greater in-situ defect activity demonstrated reduced mechanical performance.
Collectively, these findings provided robust evidence that Euler3D’s defect maps serve as reliable early indicators of porosity, geometric inaccuracies, and surface conditions that directly impact functional acceptance.
DISCMAM then encapsulated these intricate correlations into a fast-response digital twin for LPBF quality classification. This culminated in the development of a digital "go/no-go" framework, illustrated in Figure 3. For each build, one or more test coupons are printed alongside the primary target part under identical conditions. Euler3D and DTI collect defect metrics from these coupons and compare them against a reference population of "known good" parts – components previously validated through extensive CT scanning and mechanical testing.
Figure 3: DISCMAM Go/No-Go Framework for Acceptance Level Classification
This framework visualizes the acceptance level of test specimens from three distinct build batches. The left plot uses marker colors to differentiate between these builds, while the right plot uses color to indicate the known quality of the test specimens. New parts are evaluated by comparing their defect metrics to those of the reference parts. The statistical distance from the relative distribution of known good parts is then used to assign the newly printed component to one of the four functional acceptance levels established in agreement with the Belgian MoD: Level 1 (permanent replacement), Level 2 (temporary replacement), Level 3 (emergency use), or Level 4 (scrap). Concurrently, layer-by-layer status indicators flag "potential" or "critical" layers, signaling areas that may require focused review.
During trials conducted at DTI, including builds of the Unimog compressor cover with accompanying coupons, this framework consistently ranked the coupons within the "permanent acceptance region" when manufactured under DISCMAM’s optimized parameters. This enabled the main compressor cover part to be classified as Level 1 without the need for additional CT scanning, while still retaining CT as a validation option. In essence, Euler3D’s in-situ defect detection, coupled with its proven correlation to porosity and geometric deviation, closes the loop on quality assurance. It replaces conventional physical inspection with a robust stream of digital evidence, transforming raw defect maps into a practical, data-driven tool for classifying parts into four distinct acceptance levels. This empowers faster, more confident decisions regarding the deployment of field-printed metal spares.
Demonstrating the Solution: The AM Village Workshop in Albacete
To accelerate knowledge transfer and provide a tangible demonstration of the DISCMAM solution, the project consortium organized a joint workshop at the AM Village 2026 in Albacete, Spain, in March. The event, hosted by the Spanish Air Force and organized by the European Defence Agency (EDA), brought together a diverse group of industry, academia, and defense stakeholders, including representatives from Ministries of Defence and Maintenance and Logistic units from various military forces.
During the workshop, DISCMAM partners – DTI, Fieldmade, and Lortek – showcased the critical importance of quality assurance for on-site, deployed AM operations. Participants witnessed firsthand how essential it is to move beyond visual inspection alone, which is incapable of identifying internal quality flaws. Instead, they observed the identification of a defective part using rapid, digital quality assurance methodologies integrated within Fieldmade’s Nomad03 containerized solution. This highly effective demonstration garnered significant recognition, with Fieldmade AS receiving the "Best Performance" award at the event. This award underscored the profound value of DISCMAM’s digital solutions in enabling high-quality on-site spare parts manufacturing within portable and containerized environments.
Attendees observed live LPBF builds monitored by Euler3D’s system, gaining insight into how the digital twin classifies parts into the four acceptance levels immediately after production. The workshop provided an invaluable opportunity for participants to discuss how this framework could be integrated into their own qualification schemes, standards, and operational procedures. By grounding these discussions in concrete use cases and real test data, the DISCMAM consortium is actively working to transition from abstract promises to practical, readily adoptable solutions for the defense sector.
About the Authors:
Daniel Ferreira
Daniel Ferreira is a technical consultant in the Center for Industrial 3D Printing at the Danish Technological Institute (DTI). He holds a degree in materials engineering from the University of Aveiro and a PhD in additive manufacturing of metal alloys. Dr. Ferreira’s current role is focused on driving the adoption of metal AM within the defense sector, developing novel methodologies, and advancing the digitalization and certification of powder-based AM processes.
Juan Carlos Pereira
Juan Carlos Pereira is the principal research scientist and leader of the Large-Format Metal Additive Manufacturing and Repair Processes Unit at Lortek Technological Research Center. He possesses a degree in mechanical engineering from UC Venezuela, a Master’s in mechanical engineering from USB Venezuela, and a PhD in industrial engineering and production from ITM-UPV Spain. He is an accomplished researcher with over 35 refereed publications and currently leads numerous European, national, and regional R&D projects.
Marta Muñiz Mangas
Marta Muñiz Mangas is a researcher at the Large-Format Metal Additive Manufacturing and Repair Processes Unit at Lortek Technological Research Center, with a specialization in metal AM utilizing Directed Energy Deposition (DED) technologies. She holds a degree in chemical engineering and a Master’s in materials science and technology from the University of Oviedo, along with a PhD in materials engineering from the University of Sheffield/British Steel. Her expertise includes the spark butt welding of carbide-free bainitic steels for railway applications.
Tobias Rönneberg
Dr. Tobias Rönneberg serves as the head of quality at Fieldmade, where he spearheads initiatives to ensure the quality and performance of AM processes, particularly in remote, field-based applications. He earned his PhD in mechanical engineering from Imperial College London, with his doctoral research concentrating on defect generation mechanisms in LPBF and their impact on material properties. At Fieldmade, he has been instrumental in integrating innovative quality assurance solutions into the company’s Nomad mobile manufacturing systems.
Hildur Einarsdóttir
Dr. Hildur Einarsdóttir is the Chief R&D Officer and co-founder of Euler3D. She holds a Master’s in electrical engineering and a PhD in applied mathematics and image analysis from the Technical University of Denmark (DTU). In her current capacity, Dr. Einarsdóttir leads Euler3D’s R&D efforts, focusing on real-time, camera-based defect detection and the development of AI and image analysis methods that transform layer imagery into actionable quality insights for production AM.
