by Elif Erdine and Alexandros Kallegias

17 May 2017
Architectural Association
Bedford Square, London & Hooke Park, Dorset


Weave.X is the recipient of Architizer A+Award Popular Choice Winner in the Architecture +Technology Category

Weave.X is the final working prototype designed, developed, and fabricated during the Summer DLAB Visiting School, which took place between the AA’s campuses in Bedford Square and Hooke Park from 25 July to 12 August 2016. Twenty-one participants from eleven countries partook in the programme in order to investigate themes of generative design, material computation, and robotic fabrication technologies within the agency of concrete and robotic rod-bending protocols.


Geometrical rationalisation and the generation of steel reinforcement bars for robotic rod-bending

The Weave.X prototype, built by our students and tutors, is the Architizer A+Award Popular Choice Winner in the Architecture+Technology Category (see link at end).


Since 2014, AA Summer DLAB has addressed the use of concrete as a medium to establish ways of integrating computational design methods within various design and analysis platforms coupled with advanced techniques of fabrication.


Form-work generation process around steel reinforcement bars.

Within this framework, the academic setup is organised around the concepts of team-based work and design experimentation, which involves analysis of design problems thoroughly, evaluation of these analyses, and generation of original interpretations for potential outcomes as an integrated design unit. In 2014, real-time generative form-finding methods based on branching and bundling systems were linked to earth scaffolding, fabric formwork, and concrete materiality for the design and realisation of a concrete dome shell formed by branching elements, Callipod. In 2015, robotic milling processes were utilised towards the design and fabrication of custom-made, differentiated moulds for the realisation of a complex, doubly-curved wall element, InFlux.


Robotic rod-bending system.

Weave.X is the outcome of such ongoing innovative research. The design objectives have focused on the evaluation and interpretation of a traditional fabrication process- steel rod bending- towards its advancement within the domain of advanced computational and robotic methods.

Initial computational form-finding techniques explore the generation of a network of interwoven elements via a bundling algorithm developed in Grasshopper. The algorithm enables the user to locally differentiate the degree of connectivity between elements in discrete parts of the global configuration, a condition that can enhance structural performance.

Simultaneously, an automated fabrication process is developed, where custom shaped steel reinforcement bars are bent using a 6 axis robot(KUKA KR- 150), custom built jig and pneumatic grippers. The bending jig system comprises 3 different bending discs, with radii of 150 mm., 100mm., and 50 mm., and a pneumatic gripper for securing one end of the steel rod in position. In this setup, the rod bending process sets up a set of constraints which have direct feedback on the computational form-finding process. Geometrical outcomes from initial bundling algorithm experiments are optimised via a custom-built Python script that evaluates the curvature of each element at specified domain intervals, finds the closest curvature value in alignment with one of the radii stated above, and rebuilds the geometry such that the final output is a series of lines and arcs with variable bending angles. This geometry serves as the steel reinforcement for the concrete structure in the future stages of design and fabrication.


Steel reinforcement bars placed inside Polypropylene form-work

The research aims to develop a novel approach to traditional rob-bending strategies by the reduction of mechanical parts for controlling the bending process and hence the desired output form. The correlation of the physical parts and robotic toolpath is achieved using the aforementioned script developed in KUKA|prc in conjunction with Python scripting.


Site preparation

The toolpath integrates the necessary material considerations, including tolerances and steel rod spring-back values, with bending motion strategies through a series of mathematical calculations in line with physical bending experiments. More than 80 steel rods, each bearing a length of 1500 mm. and radius of 8 mm., have been robotically bent within a short period of time thanks to the speed, precision, and low tolerances of the robotic bending protocols.



The outcome of the initial computational phase, a series of interwoven 2-dimensional components made of lines and arcs, is given structural thickness via a interconnecting algorithm that generates a high-resolution mesh around the components.

This mesh is further optimised and triangulated in order to create developable surfaces that can be unrolled with accuracy. The diameter of each component ranges between 100 mm. and 250 mm. according to its location in the global configuration. The 3-dimensional interwoven model is then evaluated via Finite Element Analysis (FEA) with the Grasshopper add-on Karamba. The total displacement values gained from initial FEA serve as inputs for re-adjusting the parameters of the bundling and optimisation algorithm through various iterations.




Final prototype

The triangulated mesh acts as the formwork for the structure, and is fabricated from CNC-milled 3 mm. thick Polypropylene sheets. The fabricated sheets are folded back to generate the 3d components of the interwoven structure. As the mesh has been generated directly from the geometry of the steel bent rods, the reinforcement bars and form-work for the structure match seamlessly. In the concluding stages of fabrication and assembly, a special mix of concrete with fiberglass additives has been poured inside the form-work that is supported by the steel reinforcement, allowing the concrete mix to be cast and cured within several hours. Due to the surface finish of Polypropylene form-work, it has been possible to complete the de-moulding process of the structure in a short period of time.


Final prototype

The final configuration is characterized by a continuous network of concrete branches that support each other while creating an amorphous spatial enclosure. One of the questions raised during the design process has been the application of a form-work material that could aid in the fabrication of a complex geometrical configuration while maintaining self-supporting capacity. The use of Polypropylene has facilitated these objectives while also providing a reflective surface quality for concrete. Furthermore, the incorporation of robotic bending parameters as the initial has contributed to seamlessness between design and fabrication phases, moving away from a direct design-to-production approach.


The ongoing research intends to incorporate simple mechanical tools and cost-effective fabrication methods with the complexity embedded in generative form-finding processes, geometrical rationalization, and robotic tool-path creation that integrates material constraints. The key objective is to illustrate the architectural possibilities of using concrete in a non-conventional way by creating strong associations between computational design methodologies and robotic fabrication processes.



Technical Details:

21 days :: Duration of the Summer DLAB programme

7 days :: Design development, fabrication, assembly

21 students :: 11 nationalities

96 hours :: Fabrication, assembly, dis-assembly time

120 m :: Steel robotically-bent reinforcement rods

1.5 m3 :: High-strength concrete

30 m2 :: 3 mm. thick Polypropylene form-work



Programme Directors: Elif Erdine, Alexandros Kallegias

Tutors: Alexandros Kallegias, Elif Erdine, Angel Fernando Lara Moreira, Necdet Yagiz Ozkan, Suzan Ucmaklioglu

Research Collaborator: Alican Sungur

Robotics Collaborator: Pradeep Devadass

Students: Artemis Psaltoglou, Anna Rizou, Irini Sapka, Stelios Andreou, Alexandra Marantidou, Melike Culcuoglu, Deniz Ipek Ayasli, Isui Rodriguez, Roger Flores, Reese Lewis, Shang-Fang Yu, Anthony Ip, Mauricio Velarde, Kentaro Fujimoto, Josue Davila, Daniela Orellana, Erik Hoffmann, Zheng Luo, Jeffrey Novak, Veronica Ruiz, Justine Poulin.


AA Summer DLAB 2017 will take place from 24 July to 11 August 2017.


For more information and applications:

AA Summer DLab Site
AA Summer DLab Application
AA Summer DLab on AA Conversations
WEAVE X: Architizer A+Award Popular Choice Winner in the Architecture+Technology Category