Textile elements have long played a pivotal role in both traditional and contemporary adaptive building components. Their inherent lightweight nature, flexibility, and capacity for shape transformation make them particularly suited for dynamic architectural applications. These properties enable textiles to modulate environmental conditions—such as solar radiation and thermal gain—by transitioning between different states, thereby offering varying degrees of solar shading and regulating the influx of light and heat into buildings.

The adaptive behavior of textiles typically operates between two primary states: relaxed and tensioned. This principle is exemplified in large-scale shading and weather protection systems, particularly in stadium architecture. Historical precedents include the adjustable solar shades used in Roman amphitheaters (Montilla, 1969), while contemporary examples range from the membrane structures of the Montreal Olympic Stadium (Lazzari et al., 2009) to modern football arenas where centrally stored textile masses are tensioned via pulley systems to cover expansive areas (Göppert et al., 2013).

Beyond iconic structures, adaptive textile systems are widely applied in urban solar screens, temporary enclosures, and everyday shading devices such as louvres and blinds. These systems span both permanent and temporary installations, and are employed in indoor and outdoor contexts, including façades and spatial partitions (Barozzi et al., 2016).

The predominant material class used in these systems is woven textiles, often derived from fossil-based polymers. Polyester and ePTFE Expanded Polytetrafluoroethylene) yarns are both favored for their durability and resistance to environmental stressors. Furthermore, composite membranes such as PVC coated/polyester or PTFE coated/glass offer effective protection against water, UV radiation, and pollutants, while an increasing innovation of top coatings and finishing layers applied over the woven fabric is recently prolonging the service life of these membranes around 25–30 years (McKeen and Ebnesajjad, 2023).

On the other hand, these materials are scrutinized due to their environmental impact. Along with their production processes, the composite chemical nature makes both recycling and disposal problematic. Also, the manufacturing process is a very specialized and complex, which limits the wide-spread application of this architectural technology. It is often resource-intensive and generates significant waste, particularly during cutting, welding and sewing stages (European Commission. Joint Research Centre, 2023; Monticelli and Zanelli, 2021). However their very low material consumption justifies makes the use of lightweight membrane structures and textiles facades, still more sustainable than other material and structural options (Procaccini, 2025), while the use of membranes for short-term and temporary installation can be generally considered unsustainable (Monticelli and Zanelli, 2021).

Knitted structures are created by programming on loop level. The way individual loops are combined determines the behaviour, appearance and expression of a textile. Approaching knitting from a textile design perspective means to technically know the mechanical characteristic of the single loops and to have the ability to anticipate the behavior of a textile generated by the combination of loops in rows and columns.

When designing fashion products, the fit of the garment to the three-dimensionality of the human body and its movement is of key concern (Wolff et al., 2023). Here, elasticity is distributed on a garment according to the structural influence it brings to the whole piece (Liu et al., 2019; 2021). Elasticity in knitted textiles depends on the combination of diverse elements like material selection, structural design, and fabrication techniques. Key assets include:

Yarn choice: incorporating elastomeric fibers such as spandex, Lycra, or rubber into the yarn blend provides the best stretch and recovery. Among natural fibers, the elasticity depends on the fiber itself (e.g., wool is more elastic than cotton, linen is the more rigid one) but also on the structure of the yarn itself, its twisting, its construction on multiple plies rather than single.

These combined elements allow designers to produce knit fabrics with tailored stretch, resilience, and functional behavior for apparel, performance textiles, and industrial applications, and demonstrates that elasticity in knitted materials is not merely an inherent property of the fiber, but an emergent characteristic arising from the interaction of yarn properties, stitch structure, and fabrication parameters (Singal et al., 2024). Among the possible structures, ribbing is one of the foundations of knitting. Rib knit fabric is renowned for its natural elasticity, which allows it to stretch and recover without losing shape (Figures 2A–C). This characteristic is primarily due to the alternating knit and purl stitches that create a textured surface, enhancing flexibility. The comfort and ease of movement provided by rib knits make them ideal for trims on garments (wrists, collars, hems, etc.), but also for the adaptive architectural structures object of this article. In these applications, the fabric’s inherent stretchiness contributes to its durability, ensuring it retains its form over time.

While conventional elastic textiles often rely on fossil-based elastomers (Bolaji and Dolez, 2024; Eichhorn et al., 2009) the natural elasticity of rib-knit structures could serve as a sustainable alternative. Capable of achieving reversible elasticity through structural design alone, rib-based knitted fabrics maintain their elasticity even when produced from mono-material natural fibers, without the need for spandex or other synthetic elastomeric components. This approach suggests that careful selection of fiber type, yarn construction, and rib geometry can replicate the stretch-and-recovery behavior traditionally achieved with synthetic elastic blends, potentially reducing the environmental footprint of elastic textiles while maintaining functional performance.

Whether it be in fashion production or in architecture, the integration of CNC (computer numerical control) knitting technology represents a key enabler for precise, repeatable production of complex rib-based structures. CNC knitting allows designers to program loop formation at a highly detailed level, controlling stitch type, loop density, and pattern variation across the textile (Tamke et al., 2020). This level of control makes it possible to explore advanced functional properties–such as localized elasticity, structural stability, and adaptive fit–that would be time-consuming with standard knitting machines, while minimizing material waste and enabling rapid prototyping.

Together with the structural behaviour, what is common in the design development of any fashion product is its aesthetic. A lot of this is made by the colour story, which with knitting can be very versatile and offers to designers a wide space for experimentation (Black, 2002; Gitelson-Kahn et al., 2023). One particularly effective method is plating, which uses two different yarns—typically one for the front and another for the back—allowing for unique colour combinations and visual effects. This technique enables the fabric to subtle shift color across knit and purl sequences, highlighting the construction and movement of the textile. Because the two yarns are juxtaposed rather than separated, the colors blend seamlessly, producing dynamic shades that can appear concealed or revealed depending on the fabric’s structure and motion.

The design research phase of the Manta Ray project was conducted primarily by architects and followed the logic of an architectural design process. Its objective was to define the overarching goals and requirements of the system, particularly in relation to solar shading, façade integration, and underlying structural conditions. A review of the state of the art in adaptive façades and textile-based systems helped contextualize the project and led to the formulation of key design criteria.

Central to the approach was the integration of bending-active elements and knitted textiles as primary strategies for actuation and surface articulation. These principles were selected for their potential to reduce mechanical complexity while enabling expressive, responsive movement. In this phase, the design process benefited from a loose connection between scales and components, allowing individual aspects to be explored independently. This separation enabled progress without being hindered by unresolved dependencies, fostering a flexible and iterative development environment.

The hands-on exploration was conducted using proxy materials, such as woven mesh textiles, in place of actual knitted fabrics (Figures 3A–F). These materials allow rapid prototyping and intuitive testing of actuation principles through small-scale hand models. The process was embedded within a design studio setting, involving architecture students in a semester-long investigation. The studio introduced the fundamentals of bending-active systems and encouraged experimentation through conceptual development, physical probes, and working prototypes. While inspiration was drawn from stretchable knit patterns, the complexity of integrating CNC-knitted textiles exceeded the scope of this phase. Collaboration with knit designers focused instead on testing feasibility through simplified samples.

The design process operated across multiple scales and levels of abstraction. Scale-less hand models were used to explore compositional strategies, while digital simulations incorporated precise geometries, environmental data, and site-specific conditions. The workflow alternated between conceptual abstraction—examining how elements could be composed on a façade—and detailed 1:1 prototyping to resolve technical challenges.

The outcomes of this phase included a set of design requirements, a scaled working prototype, and a deeper understanding of the actuation and bending-active systems. Performance testing focused on the behavior of joints, the dynamics of textile movement, and the interaction between structure and gravity. These insights informed the system’s responsiveness, speed, and shading performance. Digital tools supported the process throughout, enabling sketching, simulation, spatial arrangement, and the development of 3D-printed components. While the results remained exploratory, they established a robust foundation for the next phase.

Building on the design parameters and performance benchmarks established in the first phase, the development research phase shifted focus toward material innovation, technical integration, and fabrication feasibility. Unlike the conceptual and macro-scale orientation of the design research, this phase concentrated on detail-level problem solving, scaling up, and cross-disciplinary integration.

Here, hypotheses introduced during the design research—particularly those concerning the interaction between knitted textiles and bending-active structures—were subjected to realization and testing. The development phase required the system to function cohesively across scales, moving beyond isolated component studies to integrated assemblies (Figures 4A–E). This included the refinement of actuation mechanisms, the scaling of textile surfaces, and the transition from cut-and-sewn mesh to CNC-knitted materials.

A technically demanding aspect of this phase was the development of CNC-knitted components. Extensive sampling was necessary to translate the specifications derived from mesh-based prototypes into stitch-level fabrication logic. This involved testing yarn behavior, stitch density, and pattern geometry to ensure compatibility with the actuation system and structural performance requirements.

The development research phase also addressed the integration of mechanical and textile elements, requiring precise coordination between material properties, fabrication techniques, and spatial configuration. The goal was to validate the system’s performance not only in terms of movement and shading but also in terms of durability, manufacturability, and architectural applicability.

This phase marked a transition from speculative design to technical exploration, where the system’s feasibility was tested through full-scale components and refined prototypes.

To ground the development of the adaptive shading system based on CNC-knitted textiles, we chose a real-world architectural testbed. The choice was an existing building, reflecting the ambition to explore transformative interventions that improve the performance of current building stock. While the design approach remains applicable to new constructions, the focus on retrofit scenarios allowed the research to engage with real constraints, existing façade systems, and urgent performance challenges.

The chosen Building 11 (Figure 5) is located on the Leonardo da Vinci Campus. Designed by Vittoriano Viganò and constructed between 1970 and 1983, the four-story steel-and-glass building features a radial layout of classrooms organized around a central circulation core. This configuration ensures that most classrooms are surrounded by extensive glazing, which—while architecturally expressive—has led to persistent issues of overheating, excessive solar gain, and visual discomfort, particularly during spring and summer months (Figures 6A,B).

Given the building’s structural and technological characteristics, the most viable strategy was to introduce an external shading layer. This intervention leverages the existing 90 × 90 cm mullion grid of the façade, which provides a modular framework for mounting adaptive shading elements. The proposed system was conceived as a versatile and low-maintenance solution, capable of addressing both thermal performance and daylight modulation.

The façade system was treated as representative of typical post-war curtain wall constructions, making the findings broadly applicable to similar retrofit scenarios. This case study thus serves as a strategic platform for testing the integration of knitted materials, bending-active structures, and adaptive control systems within a real architectural context. It enables the evaluation of both technical feasibility and architectural impact, bridging the gap between material innovation and spatial performance.

The solar gain and shading simulation were run using publicly available weather data from Milano Linate Weather Station (WML: 16,080/ICAO: LIML) using Rhinoceros Grasshopper’s Ladybug and HB Radiance plugins. The model took in consideration the glazed façade facing west. Although an exact light transmission coefficient was not measured at this stage of development, an assumed value of 20% was used.

The proposed design features a modular shading array composed of diamond-shaped CNC-knitted textile screens, each capable of transitioning between two distinct states (Figures 7A,B). In their fully tensioned configuration, the knitted membranes provide effective solar shading and contribute to the visual articulation of the façade. When retracted, the modules align in a linear arrangement with the vertical elements of the existing façade system, remaining visually unobtrusive and minimizing their impact on daylight access and the architectural expression of the building envelope. The design minimizes the interface between the shading system and the façade system to a few screws, that position compact 3D printed units of the Manta system on the vertical metal elements of the existing façade (Figure 8).

The Manta system’s three-dimensional movement enables control over both azimuthal and zenithal solar radiation, offering nuanced modulation of light and heat. This is achieved while minimizing the number of mechanical components, leveraging the elasticity of the knitted membrane and the bending-active behavior of the GFRP (glass fibre reinforced polymer) structure. The synergy between textile flexibility and structural responsiveness allows for a lightweight, low-maintenance solution with high adaptability.

Each module is equipped with an individual actuation mechanism, enabling localized responsiveness to environmental conditions. Actuation is governed by sensor inputs, allowing the system to dynamically respond to variations in solar intensity, temperature, or user-defined parameters. This modular and responsive configuration supports both functional performance and architectural integration, making it suitable for retrofit applications as well as new constructions.

The system was designed with energy performance as a major consideration. The design was optimized to maximize the covered area when the manta knit elements were fully opened, as well as to enable regulation across different stages of opening. To evaluate these design objectives on the façade, we considered a scenario in which a complete set of manta knits—referred to as a “manta swarm”—was deployed (Figure 7).

Our analysis focused on the rooms located on the second floor of Building 11, as these spaces are the most critical in terms of exposure to direct incident sunlight. Given the exploratory nature of this development and to maintain computational efficiency, we assumed that the knitted material would block 100% of incoming sunlight.The results, visualised in Figures 9, 10, revealed that a swarm of fully-deployed modules can bring down direct incident radiation on the envelope from an average of 1.32 kWh/m² to 0.04 kWh/m² on a typical summer day (July 21st, 24-h simulation, cumulative incident radiation).

In terms of internal solar gains, the overall solar irradiance at floor level of the second floor rooms was brought down from an average of 12.24 kWh/m² (with hotspots of 24.98 kWh/m²) to an average of 3.21 kWh/m².

A more accurate investigation will be needed to limit the observation to cumulative irradiance over single days and single months, as during winter days the system would not be deployed as solar thermal and illuminance gains would be desirable.

Further development is necessary to analyse the knitted fabric through spectrophotometric measurements at different stages of module deployment and fabric stretch to define an accurate material model for the energy simulation.

The actuation strategy for the Manta Knit prioritizes a reduction in mechanical components. Instead of conventional hinges or joints, the system employs bending-active elements, made of GFRP (glass fibre reinforced polymer), to achieve motion and structural transformation. These rods enable the textile to spread in a manner reminiscent of a manta ray’s wing movement. The knitted textile has channels in which the rods are placed providing stable interface of the material systems (Figure 8).

The actuation framework consists of 6 mm and 3 mm diameter GFRP rods, complemented by a 30 × 3 mm GFRP stabilizing profile. These elements are secured using 3D-printed clamps, which are mounted on vertical beams—similar to those found in the ABC building and other façade systems. The use of additive manufacturing enabled rapid prototyping and iterative refinement, allowing for the integration of multiple functions into a single component. This approach facilitated precise spatial arrangements and the development of intricate joints that would be difficult to fabricate using traditional subtractive methods.

The 3D-printed components serve multiple roles:

They secure the rods and clamp the knitted textile.

While the current components are printed in PLA, this material is considered a placeholder for more durable alternatives. Nonetheless, the process demonstrates the potential of additive manufacturing for producing complex, multifunctional parts with short lead times.

The GFRP rods fulfill two primary functions:

They act as interfaces between the textile and the actuation system.

To explore actuation strategies for knitted textiles, the Manta Knit prototype employs a simple yet effective pulley system (Figure 11). The actuation cables follow the force trajectories required for movement and are routed to a centralized actuation unit. This configuration minimizes cable length and consolidates mechanical components, simplifying maintenance and enhancing accessibility.

The central actuation unit is mounted atop the central vertical beam, a location intentionally chosen to remain concealed beneath the textile surface, thereby protecting it from environmental exposure and visual distraction. The direct transmission of force from the cables to the rods is facilitated by openings integrated into the knit, allowing 3D-printed connectors to interface with the rods and secure the actuation strings.

The cable routing follows the movement paths of the textile:

Cables run to corner pulleys, which define the outer limits of the wing’s motion.

The cable controlling the manta’s tip is routed vertically to the end of the GFRP profile, then redirected beneath the profile back to the central unit.

The actuation system comprises:

Two motors, each driving a 3D-printed spool via gears and chains. The spools feature grooves dimensioned to wind the precise length of cable required for actuation.

The motors are independently controlled, enabling differentiated movement of the wings. This allows for adaptive shading and controlled transparency, such as lifting the manta’s tip to open views through the façade. The independent control also facilitates choreographed motion sequences, where the wings initiate movement followed by the tip (Figures 12A–C).

Earlier experiments with asymmetric wing movement revealed that lateral forces on the central profile caused undesirable skewing. A previous design iteration employed a round rod as a central spine to allow for such movement, but maintaining balance in the central structure proved more critical. The current configuration ensures a stable and predictable motion pattern, minimizing structural strain and preserving the integrity of the overall system.

The Manta Ray prototype was developed over a 6-month period, culminating in its exhibition at the Politecnico di Milano in September 2024 and later at Milan Design Week in spring 2025. The initial design research phase was conducted in collaboration with a dedicated group of students. Beginning from first principles, the team explored actuation strategies and developed a 1:3 scale working prototype. This early version featured a woven mesh textile surface and was manually actuated, serving as a proof of concept.

The subsequent development research phase focused on two parallel tracks: the refinement of the knitted textile system and the engineering of the bending-active structure with integrated 3D-printed joints.

The development of the knitting programs required 2 months of work, but represents a solid and reasonably adaptable foundation in case Manta production is scaled up in the future. The pieces each take 50 min to be knitted off stably, for a total of 2 h and 30 min of total knitting time. The weight of the waste produced is approximately 20 g in total. The significant amount of time for development was due to the unprecedented needs of tensile structures, not common in the usual development of knitted garments. Starting from the present iteration, future research could focus on defining a programming framework more specifically tailored to the generation of tensile structures.

For prototyping and exhibition purposes, timber frames were employed as temporary scaffolds. These were later adapted into self-supporting structures for the installation at Politecnico di Milano (Figures 22A,B). The frames were dimensioned to reflect the scale of the ABC building’s façade, though simplified through the use of half-lap timber joints. The Manta Ray system was mounted directly onto these frames using screws, while pulleys were embedded into the timber using hex nuts. This assembly strategy could alternatively be executed with 3D-printed joints, offering a more seamless simulation of a façade transformation on an existing building.

The assembly process was intentionally designed for speed and simplicity. The Manta Ray structure, including the 3D-printed joints and actuation system, was affixed to the timber façade at a limited number of predefined points. The frame itself provided precise positioning for all components, eliminating the need for on-site measurements and significantly reducing installation time.

Disassembly was equally efficient. The system’s linear, flat-packable design facilitated straightforward transport and storage. This modularity enabled a rapid reinstallation of the prototype at Milan Design Week, where it operated continuously for three weeks—10 h per day—without any mechanical failures or interruptions.

A key outcome was the demonstration that rib knit structures can achieve reversible elasticity through geometry rather than synthetic elastic yarns. This approach allowed us to use natural fibers—typically not associated with elasticity—in adaptive systems. The shift from material-based to geometry-based elasticity is significant, as it opens new pathways for sustainable design by reducing reliance on fossil-based materials.

CNC knitting enabled precise control over the knit structure, allowing elasticity to be embedded at the stitch level. This capability was validated through both small-scale samples and a full-scale prototype exhibited two times in Milan. Over the 4-week exhibition at Milan Design Week, the rib knit fabric and embedded GFRP rods showed no signs of fatigue or deformation, indicating promising durability. However, the resulting textile was relatively thick and heavy, which may limit its responsiveness and scalability. Future iterations will need to explore lighter yarns and optimized knit densities to improve performance without compromising structural integrity.

To assess the elasticity of the knitted samples, we developed a rapid, low-tech testing method. Each sample could be produced and evaluated within minutes, providing a practical tool for early-stage design decisions. While the method lacks full repeatability due to manual execution, it proved sufficient for comparative analysis during prototyping.

However, scaling up testing remains a challenge. Existing bi-axial testing machines and protocols are not well-suited to the anisotropic and highly elastic nature of knitted textiles. This points to a broader issue: current testing standards and tools are not yet adapted to the unique behaviors of textile-architectural hybrids (Christidi et al., 2025; Tamke et al., 2015). Developing new testing protocols that account for the dynamic and directional properties of knits will be essential for advancing this field.

The rib knit structure also enabled a dual-state visual effect by embedding two yarn colors. In the relaxed state, one color dominates; under tension, the second color becomes visible. This creates a dynamic surface that responds visually to movement and deformation.

In small-scale samples, the color shift was clear and striking. However, in the full-scale manta ray prototype, the effect was more subdued. The complexity of the CNC knit pattern and the shaping of the textile surface diluted the visual contrast. This highlights a trade-off between logics on knit level and visual clarity. Achieving a strong visual effect at scale may require simpler knit patterns or more strategic color placement. Additionally, the interaction of yarns within the knit structure affects color blending, necessitating careful sampling and testing during the design phase.

The integration of elastic knit with a bending-active system using GFRP rods allowed for a reduction in mechanical complexity. The movement achieved through this hybrid system would be difficult to replicate with rigid components alone. The rods showed no fatigue during the exhibition, and the overall actuation was stable and reliable.

Despite these successes, the actuation system presented some limitations. The pulley-based mechanism was easy to implement and control, the necessary motor unit became in it’s final iteration however quite bulky, compromising the goal of fully integrating the system behind the textile wings. The mechanisms exposed nature also raises concerns about long-term durability and weather resistance in real-world applications. Moreover, the complexity of the system increases with the number of motors required. Future designs should explore more compact and integrated solutions, such as single-motor systems with staggered spools or wireless. Mesh-networked control systems could simplify wiring and allow to manage multiple units simultaneously.

At the architectural scale, the manta ray prototype demonstrated how knitted textiles can enable dynamic spatial transformations. The structure transitions between two archetypal states: a taut membrane and a draped curtain. These states reflect the dual heritage of textile architecture–the tent and curtain (Ramsgaard Thomsen and Bech, 2011) — balancing functional performance with expressive potential.

The transitions between these states are not merely mechanical but choreographed spatial events. Gravity, material behavior, and timing all contribute to the experience of movement. The inclusion of GFRP rods helped stabilize and guide the folding process, ensuring smooth transitions. The speed and sequencing of wing movement and membrane lifting were also critical in shaping the spatial and environmental performance, particularly in terms of shading and light modulation.

However, current digital design tools are limited in their ability to simulate these transitional states. Most membrane design software focuses on fully tensioned geometries, neglecting the intermediate configurations that are central to the expressive and performative qualities of knitted structures. This gap in simulation capabilities presents a significant barrier to the broader adoption of knitted systems in architecture.

The research employed a two-phase research-by-design methodology, structured to integrate architectural form-finding and textile material design through iterative prototyping and evaluation. This approach was organized into a reproducible workflow comprising (1) Design Research and (2) Development Research phases, each with defined steps, criteria, and feedback loops (Table 2).

Reflecting on the implementation of this approach, it can be stated that it in our case effectively facilitated a dynamic exchange between top-down design intent and bottom-up material behavior. In our case, early and sustained interdisciplinary collaboration was crucial. Textile designers contributed deep material expertise, while architects translated spatial and performative ambitions into a coherent concept and set of specifications. The success of the project relied on mutual respect, openness, and a willingness to iterate.

Prototyping played a central role throughout the process. Physical samples, hand models, and digital simulations were employed in tandem to test hypotheses and refine designs. This iterative, evidence-based approach proved essential for addressing the complexities of textile-architectural integration. However, it also demanded significant time, resources, and expertise, underscoring the need for dedicated frameworks to support such interdisciplinary work.

The exploratory research presented here positions CNC-knitted textiles as a viable and innovative material system for the next-generation of adaptive building envelopes. By shifting from conventional woven materials to programmable knits, the project introduces a new design paradigm—one in which material behavior is not a constraint but a driver of architectural form and function.

The Manta Ray prototype illustrates the feasibility of this approach. Through the integration of knitted textiles and bending-active GFRP rods, the system achieves responsive movement with minimal mechanical complexity. The result is a lightweight shading system that aligns with demands for sustainable and adaptable architecture. The durability and reliability of the prototype during the two exhibitions indicates its potential for real-world application.

Looking ahead, several key areas for future development emerge:

Simulation and Design Tools: To move forward future projects would strongly benefit from digital tools that can simulate the initial and transitional states of knitted textiles. Especially the intermediate configurations are critical to both the performance and expression of textile-based systems, yet they remain poorly supported by current software. Here simulation environments such as Kangaroo or animation-based textile solvers can visualize such transitions, they remain of limited use for fabrication purposes. Advancing simulation environments that bridge this gap will be essential for integrating knitted systems into mainstream architectural workflows.

In conclusion, CNC-knitted textiles offer a compelling framework for rethinking adaptive architecture. Their programmable elasticity, responsiveness to environmental forces, and seamless fabrication unlock new possibilities for building skins that are not only performative and sustainable but also richly expressive. This research lays the groundwork for future explorations where material intelligence and architectural design converge.