This paper introduces the concept of building ecologies as a framework to expand considerations of the temporal and relational aspects of architecture by highlighting the ongoing processes that constitute a state of continuous construction. Drawing influence from conservation, Science and Technology Studies, and sustainability studies, building ecologies develop the web of relational context from the world by building sensitivities to care, adaptation, and repair. These sensitivities render visible the often-hidden values, processes, and frictions influencing buildings, and knowledge of them can then be used to inform architectural production. The paper presents a case study examining how care, adaptation, and repair can become integrated into design and fabrication practices, enabling a rethinking of reparatory action and lifespan in architecture. Through building ecologies and to practices of care, adaptation, and repair, a holistic conception of sustainability is proposed which intersects with regenerative design by working to both improve the quality of relations and to incorporate a situation-specific and ever-evolving temporal responsiveness that can be applied to any architectural project.

Bio-based materials are emerging a key alternative for carbon-intensive construction materials, which are becoming increasingly ill-suited to society’s ambitions to decarbonize construction. Beyond already established low carbon biomaterials such as timber, locally sourced second and third generation waste and side-stream biomass is recognized as having significant potential to be used in architectural elements. However, the use of these materials requires processing routes based in contemporary design and fabrication technologies. Additive manufacturing is one such route. This paper outlines an approach for graded 3D printing with a biopolymer-composite that incorporates local waste and agricultural side stream residues. In this paper, we describe the underlying motivations, challenges and opportunities for multi-material grading, and detail the development stages of a custom-built robotic tool and feeding system that enables continuous grading of two or three material inputs. Further, we explain the complications involved with material viscosity changes and the correlation to the design of the printing toolpath. We demonstrate the use of the custom-built robotic tool and feeding system on a set of architectural-scale weather screen panels, using different biopolymer mixtures and grading strategies.

This paper details the development of innovative grading techniques for 3D-printed biopolymer composites that utilize locally sourced, cellulose-based fibre streams to produce architectural-scale components. It examines the design considerations, methodologies, and fabrication strategies that are necessitated by the utilisation of biopolymers for architectural applications, and which underlie key processes of designing for and with variable materials. The presented research interrogates the methodological challenges of formulating new approaches that actively engage architects and designers with the ecological implications of their design choices. It outlines new methods for material grading that enable targeted compositional variation through three interlinked contributions: a gradable recipe, a design-interfaced specification process for grading, and an infrastructure for large-scale 3D printing of biopolymer composites. The paper presents the Rhizaerial demonstrator as an implementation of these contributions. Rhizaerial is a full-scale interior ceiling vault system, whose curved components are printed as a 3D porous lattice structure that creates an interplay of light, visual transparency, and colour, while maintaining structural integrity. We detail the gradable biopolymer composite recipe, and the residual and regenerative material streams it combines. We outline the implicit modelling pipeline, which includes methods for locally specifying lattice structures for 3D printing, as well as assigning continuous grading specifications to print paths. Finally, we describe the fabrication infrastructure and tooling for robotic printing of large-scale graded biopolymer composites.

Reclaimed timber often exhibits warp and deformation that compromise conventional CAD-driven joinery workflows. Existing scan–register–cut routines correct global pose but leave the joint geometry fixed, forcing nominal interfaces onto non-nominal material and producing misfits in form-fitting connections. These workflows also depend on time-consuming full registration and manual CAD re-modeling, where each correction propagates across the assembly—an impractical step in real fabrication environments. This paper introduces Fit by Feedback, a robot-centric adaptive fabrication framework that integrates sensing, geometric analysis, and machining in a continuous loop. The system performs fast, minimal scanning, completing full acquisition in under 10 s, derives a scan-informed CAM model, and morphs joint operations directly to the measured geometry of each beam and its neighbors. Rather than editing CAD models, the workflow adapts the toolpaths themselves, allowing each connection to conform to the as-found material. Experiments show millimeter-level accuracy in groove following, pocket compatibility, and joint engagement across moderately warped beams. The results demonstrate that closed-loop toolpath adaptation can restore joint precision and assembly fit without heavy registration, re-modeling, or reconditioning, offering a practical pathway for integrating warped reclaimed timber into robotic fabrication.

Reliable evaluation of mechanical performance is a central constraint for structural reuse of reclaimed timber. This study provides a controlled comparative assessment of non-destructive indicators for piece-wise characterisation of reclaimed Norway spruce elements intended for reuse as glulam lamellas. Visual assessment, longitudinal dynamic excitation, and CT-derived modelling indicators were evaluated against global and zone-wise flatwise bending tests on 56 reclaimed beams from a single material stream with known species, cross-section, and initial strength class. Dynamic excitation provided the most reliable single predictor of global bending stiffness at specimen level. Visual assessment showed limited discriminatory power for this comparatively homogeneous material. CT-derived indicators based on density and orientation fields reproduced principal intra-member stiffness variations associated with knots and fibre disturbance and showed predictive capability of similar order to established mechanical non-destructive indicators. Systematic differences between methods were primarily attributable to moisture-state uncertainty, modelling assumptions, and the global nature of the mechanical reference tests. Strength-related results, based on a limited destructive subset, are interpreted as exploratory. Within these bounds, the study establishes a comparative reference dataset for multi-modal evaluation of reclaimed structural timber and demonstrates how conventional and volumetric indicators complement each other in resolving global stiffness and intra-member variability relevant to allocation in engineered wood products.

This research presents a reinforcement learning-based framework for adaptive robotic 3D printing of biopolymer composites, naturally derived materials with highly variable and unpredictable behavior. Current additive manufacturing approaches address this by modifying material recipes or constraining geometries, but rarely by integrating the material’s behavior into the control process itself. The proposed system introduces a simulation-based, material-aware reinforcement learning environment that trains an agent to adaptively control print parameters in response to changing material and geometric conditions. Within a simplified physics model, the agent learns to balance deposition velocity, nozzle tilt, and local positioning to minimize sagging, compression, and collapse. Using proximal policy optimization, the agent refines its policy through trial and error, optimizing for structural stability, geometric accuracy, and printing efficiency. The trained policy is then transferred to a robotic printing setup, demonstrating emergent strategies such as adaptive bead width and localized tilting to stabilize challenging geometries. Results show that the system successfully learns to mitigate common failure modes in biopolymer extrusion, advancing robotic additive manufacturing toward process-aware autonomy. This research reframes robotic 3D printing from static toolpath execution to a dynamic, material-aware process.

Life-cycle assessment (LCA) stands as a vital tool in gauging the environmental impacts of building endeavors. Extending LCA to emerging research practices like advanced digital manufacturing of bio-based materials becomes pivotal for refining materials, appraising outcomes, and steering architecture toward sustainable development and circularity goals. To outline the main obstacles and to provide a potential methodology, the chapter presents two cases of application of LCA to digital fabrication with bio-based materials in experimental research practice. The application is framed within the ISO and EU standards for LCA and is tested through an ex-post “cradle to construction” analysis of two European Research Council (ERC) funded projects developed by the Center for Information Technology in Architecture (CITA) at the Royal Danish Academy. Specifically, a product LCA is performed for bio-polymeric composited 3D robotic fabrication using a novel collagen-based 3D print material, and a comparative LCA is carried out for Glulam manufacturing optimization connecting data from the timbers source in the forest and sawmill with its design and fabrication. In both cases, the prototypes assembly and exhibition are covered by the analysis. The unavailability of data, difficulties in standard protocols adaptation, and material and energy flows tracing in the research process emerge as the main barriers and contribute to aggravate the analysis’s uncertainty. The chapter shows how to manage such uncertainties via sensitivity analysis to evaluate design options according to different impact scenarios. The knowledge established and the methodology outlined in this research could be useful for researchers, designers and industry in the implementation of sustainable digital fabrication processes and new construction materials.

The reuse and recycling of timber are crucial for a circular economy, but barriers like insufficient information and concerns about material quality hinder their industrial-scale implementation. Uncertainty about mechanical properties often leads to downcycling and CO2 release. Circular practices involve cascading reuse of timber, but downcycling occurs at each step, leading to waste. To maintain carbon storage, reclaimed timber should be used with maximum integrity, like in load-bearing beams. Non-destructive assessment methods for reclaimed timber face challenges due to variations in origin, age, and wear conditions. X-ray computed tomography (CT) scanning in conjunction with computational mechanics provides a means to structurally assess wood based on its internal density distribution. In this paper, a modelling pipeline is proposed using CT-based finite element analysis to assess the quality of reclaimed timber elements. The pipeline is part of an ongoing investigation where timber stiffness and strength are evaluated both destructively and non-destructively using various measurement modalities. Accurate non-destructive assessment of the mechanical properties of reclaimed timber could optimize its use and enable repeated reuse. In subsequent research, the pipeline will be validated and simplified to aim for practical application.

The shift towards regenerative architecture necessitates a fundamental transformation in material selection, moving away from fossil-based synthetic fibers toward bio-based alternatives. This abstract presents a project that investigates a method for using natural plant-based fibers in architectural applications. Knitted membranes offer a promising material platform for adaptive material distribution due to their additive composition and capacity for localized reinforcement (Sinke, Tamke, and Ramsgaard Thomsen 2023). The research presents a series of material tests conducted at both the yarn and textile levels, acknowledging that the architecture of knitted structures affects the structural performance of the textiles. Various knit structures are evaluated, and material testing allows for the grading of knitted structures based on their strength capacities. By leveraging the data from material tests, digital simulation and computational design methods, this study explores a novel approach in which yarn properties and knitted structures are analyzed and strategically assigned within a tensile membrane structure. This introduces the multi-dimensional grading, where the surface is graded both with yarns but also knitted structures. A design of graded knitted fabric allocates the yarns and structures of higher strength capacities with the surface regions experiencing greater stress and structural demand, while the yarns and knit structures of lower strength are allocated to areas with minimal load-bearing requirements and lighter structural utilization. This computationally driven material distribution enables performance optimization while retaining the sustainability benefits of natural fibers.

The article asks how additive manufacturing for the circular bioeconomy can create the foundation for rethinking the architectural axioms of permanence and durability, instead moving us toward a new ideal of renewability and repair. It presents a case study into additive manufacturing for repair through the 3D printing of biopolymer composites. This case study connects machine vision-based surveying of damaged panels with repair through conformal 3D printing. This deployment of bio-based materials aims to enable additive manufacturing as a method for disrupting the sharp delineation between fabrication and repair leading to new practices of continual construction. With point of departure in our bespoke systems for 3D printing and unique biopolymer composites, we examine how their particular material characteristics allow for material adhesion and buildup and how novel methods for iterative 3D printing can support design integrated strategies of repair. As part of this process, we include the sociotechnological dimension, as human-in-the-loop decision-making becomes part of the material surveying regimes necessary for damage detection. The article demonstrates processes of repair through three repair actions that address different kinds of damage.

Drawing inspiration from 'broken world thinking,' aligning repair with innovation, this paper presents the development of an open and flexible decision-making framework for integrating repair into building practice. It introduces a case study examining 3D-printed biopolymer composite panels, which exhibit highly transformative behavior that catalyzes iterative acts of repair. The paper outlines the data streams and thresholds connecting human and automated processes of monitoring and diagnosis, leading to an adaptable repair repertoire. This study examines two cycles of exposure and repair, maturing the decision-making tree and formalizing the information flow across repair stages. This approach establishes procedural workflows in highly dynamic material systems. The iterative repair process through additive manufacturing enables continuity in tool deployment, aligning initial design practices with repair practices.

The shift towards bio-economies of architectural fabrication necessitates particular consideration of how characterizing aspects of bio-materials such as heterogeneity and anisotropy impact the designed performance of architectural elements. These aspects must be integrated into the modelling and representation of architecture and must be instrumentalised for digital design and fabrication processes. The use of timber in construction is challenging due to its complex material behaviours, and therefore robust methods for predicting and modelling these are crucial for exploiting the timber resource more effectively. In light of this, we focus on developing a holistic and integrated digital modelling approach for glue-laminated timber construction elements that connects the digital design model to the specific material resource and incorporates its material complexity into design simulation workflows. We question the role of the lamella in the glulam blank and trace its agency through a series of four disparate projects, from an initially silent and generic constituent of the blank to a key, operative actor in the design of performance-graded timber products through the added specificity of material composition and a liberation of its form. The first project develops a modelling approach that connects material specification and lamella sizing to free-form timber element geometries. It uses principles of industrial glulam production and knowledge of the anisotropic nature of timber to speculate on new forms of glulam blanks that could arise from a deeper engagement with the glue-lamination process. The second project looks further back in the timber value chain at the processing of the specific forest resource into tailored building elements, attaching a material specificity to the lamella. The third project expands the modelling and prototyping of non-standard glulam blanks into considerations of timber waste streams and aesthetics. The final project aims to further speciate the lamella by tailoring its form as well as its material composition to respond to simulated performance demands. Through these projects we outline the development of a novel digital framework that begins as a modelling approach for free-form glulam beams and grows to accommodate the mapping and allocation of specific, heterogeneous input material. As the digital framework matures, increasingly detailed and interlinked digital simulations of mechanical performance are integrated.

This paper introduces an innovative method for tracking and analysing the three-dimensional behaviour of materials outdoors, focusing on bio-based building elements. The constraints of current methods limit monitoring material changes over time to 2D snapshots of an initial and final state. We propose a system utilizing low-cost components like depth cameras and IoT sensors that allow for frequent 3D scans over long periods and overcome challenges such as weather conditions and IR interference. Our method includes a novel automated analysis system, which converts 3D scan data into voxel models for comprehensive evaluation. Our analysis can include different parts of the tracked properties, such as colour, depth and position, which allows us in this way to track changes on overall, region, and feature levels. This approach opens avenues for better understanding material behaviour and its implications for future design, construction and maintenance practices.

By mapping material properties that influence mechanical performance across the timber resource at a high spatial resolution and matching them to digitally simulated performance demands in designed glulam elements (Fig. 1), workflows such as the RawLam Experiments hint towards the ability to use material stocks usually unemployed in glulam manufacturing. However, the broader implications of this highly tailored and selective process remain to be quantitatively evaluated, examining the increase in timber resource used for construction and addressing how waste generation in the RawLam process is measured against established glulam fabrication methods. The paper presents possible scenarios for an industrial scale-up of the process and an evaluation of the co-product cascade associated to each scenario. Findings on the cascade variations are then leveraged to put in place a new version of the RawLam allocation procedure. This paper reveals the key differences between the standard sawmilling and glulam production process and the RawLam method in terms of the specific production steps involved and the resultant impacts on co-product ratios. This enables an evaluation of the perceived benefits of the RawLam approach to design modelling glulam blanks against broader metrics of economic value and carbon sequestration. As part of a larger effort to integrate a wider segment of the timber value chain into architectural design processes, this work demonstrates how industry-relevant indicators such as the value impacts of the co-product cascade can help steer the design modelling of engineered timber elements.