Periodic Reporting for period 2 - ADDOPTML (ADDitively Manufactured OPTimized Structures by means of Machine Learning)
Periodo di rendicontazione: 2023-05-01 al 2025-04-30
The construction industry, while economically vital, is one of the most resource- and energy-intensive sectors in Europe. It is responsible for over 40% of total energy consumption and contributes to 25–30% of all waste generated in the EU, making it one of the most environmentally burdensome industries. Despite its size and importance, the sector remains heavily reliant on traditional, labor-intensive processes, which are ill-suited to meet the urgent demands for decarbonization, circularity, and digitalization. At the same time, Additive Manufacturing (AM)—a technology that has already transformed sectors like aerospace and automotive—remains significantly underutilized in construction. This is largely due to the highly fragmented, interdisciplinary, and bespoke nature of building design and production workflows. These workflows, often involving custom-made large-scale structures, resist standardization and automation, hindering the widespread adoption of AM and other advanced digital technologies.
Europe’s ambition to become the first climate-neutral continent by 2050, as outlined in the European Green Deal, necessitates deep structural changes in how we build. The construction sector must drastically reduce its environmental footprint, optimize material use, and shift toward circular and resilient building practices. Innovations that enhance resource efficiency, structural adaptability, and sustainability will be instrumental in reaching these climate goals. Furthermore, the world is witnessing a growing frequency of natural disasters and humanitarian crises, often requiring fast, efficient, and adaptable construction solutions for shelter and infrastructure. Conventional construction methods are often too slow, costly, or logistically constrained to respond adequately. AM offers the potential to deliver rapidly deployable, customizable structures, designed and fabricated on demand, which can play a transformative role in post-disaster reconstruction and emergency housing, especially in remote or resource-limited settings.
The overarching goal of ADDOPTML is to establish a next-generation, intelligent manufacturing paradigm for the construction industry, centered on AM and powered by machine learning, topology optimization, and generative design. This paradigm aims to accelerate innovation in the design and fabrication of high-performance, adaptable, and environmentally responsible structures.
1. To develop a comprehensive library of data-driven constitutive models for structural materials, enabling accurate and efficient simulation of AM behavior.
2. To create a high-fidelity yet computationally efficient topology optimization framework for structural components manufactured via AM.
3. To deliver a fully automated, generative design pipeline for AM structures, integrating performance-driven design and fabrication constraints.
4. To demonstrate the application of AM in rapidly deployable steel and concrete structures, tailored for post-disaster and emergency sheltering solutions.
5. To define design and fabrication protocols for AM components for space applications, extending the project’s impact beyond Earth-bound construction.
Europe’s ambition to become the first climate-neutral continent by 2050, as outlined in the European Green Deal, necessitates deep structural changes in how we build. The construction sector must drastically reduce its environmental footprint, optimize material use, and shift toward circular and resilient building practices. Innovations that enhance resource efficiency, structural adaptability, and sustainability will be instrumental in reaching these climate goals. Furthermore, the world is witnessing a growing frequency of natural disasters and humanitarian crises, often requiring fast, efficient, and adaptable construction solutions for shelter and infrastructure. Conventional construction methods are often too slow, costly, or logistically constrained to respond adequately. AM offers the potential to deliver rapidly deployable, customizable structures, designed and fabricated on demand, which can play a transformative role in post-disaster reconstruction and emergency housing, especially in remote or resource-limited settings.
The overarching goal of ADDOPTML is to establish a next-generation, intelligent manufacturing paradigm for the construction industry, centered on AM and powered by machine learning, topology optimization, and generative design. This paradigm aims to accelerate innovation in the design and fabrication of high-performance, adaptable, and environmentally responsible structures.
1. To develop a comprehensive library of data-driven constitutive models for structural materials, enabling accurate and efficient simulation of AM behavior.
2. To create a high-fidelity yet computationally efficient topology optimization framework for structural components manufactured via AM.
3. To deliver a fully automated, generative design pipeline for AM structures, integrating performance-driven design and fabrication constraints.
4. To demonstrate the application of AM in rapidly deployable steel and concrete structures, tailored for post-disaster and emergency sheltering solutions.
5. To define design and fabrication protocols for AM components for space applications, extending the project’s impact beyond Earth-bound construction.
WP1 and WP2:
A physics-informed topology optimization framework was successfully developed, integrating high-fidelity nonlinear structural response simulations. This framework was further enhanced by machine learning models derived from WP2, which were trained and validated using extensive experimental datasets at both the material and component levels.
WP3:
This platform was seamlessly embedded within a novel digital design-to-additive-manufacturing workflow specifically tailored for large-scale structures. Research was focused on deriving manufacturing protocols for optimized structural topologies, guided by collaborative efforts between structural engineers, designers, and AM specialists. Key achievements include the delivery of structural designs optimized for printability, the definition of new AM protocols, the fabrication of scaled 3D-printed demonstrators, and their experimental validation. The aim was to empower engineers with an integrated, performance-driven toolchain for designing functional, printable structures.
WP4:
Building on the tools and methodologies developed a machine learning-enhanced prototyping framework that supports the intuitive and guided transition from conceptual design to physical realization. This framework brings together interdisciplinary expertise—from architecture and structural engineering to 3D printing and computational optimization—enabling efficient early-stage design exploration for AM-based construction.
WPs 5–7:
Three innovative demonstrators were developed to validate the scientific and technological advances and to illustrate the project’s impact potential across different application domains: WP5 focused on steel-based demonstrators, addressing performance, manufacturability, and deployability under dynamic conditions. WP6 targeted concrete-based solutions, particularly emphasizing the creation of safe, inclusive living environments, challenging conventional gender and identity stereotypes while incorporating culturally sensitive design principles. WP7 explored the use of the ADDOPTML framework in the context of space applications, developing and testing optimized subcomponents for extraterrestrial structural systems.
WP8 and WP9:
Knowledge transfer, dissemination, and exploitation activities have been actively pursued under WP8, with the project team presenting results at major international conferences and workshops, and publishing in peer-reviewed journals, thereby maximizing scientific outreach and open access to findings. WP9 ensured the smooth coordination and administrative management of the project, supporting interdisciplinary collaboration and the timely delivery of milestones and deliverables.
A physics-informed topology optimization framework was successfully developed, integrating high-fidelity nonlinear structural response simulations. This framework was further enhanced by machine learning models derived from WP2, which were trained and validated using extensive experimental datasets at both the material and component levels.
WP3:
This platform was seamlessly embedded within a novel digital design-to-additive-manufacturing workflow specifically tailored for large-scale structures. Research was focused on deriving manufacturing protocols for optimized structural topologies, guided by collaborative efforts between structural engineers, designers, and AM specialists. Key achievements include the delivery of structural designs optimized for printability, the definition of new AM protocols, the fabrication of scaled 3D-printed demonstrators, and their experimental validation. The aim was to empower engineers with an integrated, performance-driven toolchain for designing functional, printable structures.
WP4:
Building on the tools and methodologies developed a machine learning-enhanced prototyping framework that supports the intuitive and guided transition from conceptual design to physical realization. This framework brings together interdisciplinary expertise—from architecture and structural engineering to 3D printing and computational optimization—enabling efficient early-stage design exploration for AM-based construction.
WPs 5–7:
Three innovative demonstrators were developed to validate the scientific and technological advances and to illustrate the project’s impact potential across different application domains: WP5 focused on steel-based demonstrators, addressing performance, manufacturability, and deployability under dynamic conditions. WP6 targeted concrete-based solutions, particularly emphasizing the creation of safe, inclusive living environments, challenging conventional gender and identity stereotypes while incorporating culturally sensitive design principles. WP7 explored the use of the ADDOPTML framework in the context of space applications, developing and testing optimized subcomponents for extraterrestrial structural systems.
WP8 and WP9:
Knowledge transfer, dissemination, and exploitation activities have been actively pursued under WP8, with the project team presenting results at major international conferences and workshops, and publishing in peer-reviewed journals, thereby maximizing scientific outreach and open access to findings. WP9 ensured the smooth coordination and administrative management of the project, supporting interdisciplinary collaboration and the timely delivery of milestones and deliverables.
ADDOPTML has advanced the integration of machine learning, topology optimization, and generative design for additive manufacturing (AM) in the construction sector. It delivered a novel digital workflow for the design and manufacturing of large-scale structural components, moving beyond conventional, fragmented approaches:
-Open-source tools for material design optimization.
-Advanced methods for data-driven structural optimization.
-Protocols and prototypes for AM-enabled structural components.
-A machine-learning-assisted design framework for early-stage prototyping.
-Demonstrators in steel, concrete, and space construction that reflect rapid deployment, sustainability, and cultural inclusivity.
By the project’s end, ADDOPTML delivered:
-A validated end-to-end digital pipeline for AM in construction.
-Software tools, datasets, and design guidelines openly accessible to stakeholders.
-A new cohort of researchers trained in AI-driven structural design and AM.
-Broad adoption of AM methods tailored to practical construction and emergency contexts.
ADDOPTML supports Europe’s Green Deal goals by enabling efficient, low-waste, and resilient construction practices. The project accelerates the shift toward digital, automated design workflows, empowering innovation among SMEs and industry. It also addresses societal challenges, providing:
-Post-disaster shelters, rapidly deployable and culturally sensitive.
-Inclusive living spaces, responsive to gender and identity needs.
-Space-ready components, contributing to off-world construction solutions.
-Open-source tools for material design optimization.
-Advanced methods for data-driven structural optimization.
-Protocols and prototypes for AM-enabled structural components.
-A machine-learning-assisted design framework for early-stage prototyping.
-Demonstrators in steel, concrete, and space construction that reflect rapid deployment, sustainability, and cultural inclusivity.
By the project’s end, ADDOPTML delivered:
-A validated end-to-end digital pipeline for AM in construction.
-Software tools, datasets, and design guidelines openly accessible to stakeholders.
-A new cohort of researchers trained in AI-driven structural design and AM.
-Broad adoption of AM methods tailored to practical construction and emergency contexts.
ADDOPTML supports Europe’s Green Deal goals by enabling efficient, low-waste, and resilient construction practices. The project accelerates the shift toward digital, automated design workflows, empowering innovation among SMEs and industry. It also addresses societal challenges, providing:
-Post-disaster shelters, rapidly deployable and culturally sensitive.
-Inclusive living spaces, responsive to gender and identity needs.
-Space-ready components, contributing to off-world construction solutions.