BARCELONA ROOF
Variable Shade Proposal for a Barcelona Superblock Intersection
| period | 2024.09–2024.12 |
| location | A Superblock Intersection, Barcelona, Spain |
| type | ETSAM Taller de Arquitectura Textil · Team of Three |
| status | Unbuilt Proposal |
| role | Structural Concept · Grasshopper Definition · Kangaroo2 Form-Finding · MPanel Patterning |
Two Membranes, One Mast
This project proposes a variable shade structure that could be repeated across intersections in Barcelona’s Superblocks. Barcelona’s Superblock programme has converted vehicle-dominated streets into pedestrian-priority spaces and intersection plazas. One of these recurring intersection types became the site for this structure-led workshop proposal.
Two vertically inverted anticlastic tensile membranes meet around a single central mast. The upper membrane rotates while remaining in tension, changing the overlap between the two open-mesh layers and, in turn, the shaded area of the plaza.
Pushing Above, Pulling Below
In the upper membrane, the central mast pushes the apex upward while the weight of the perimeter ring draws the boundary downward. In reverse, the mast presses the centre of the lower membrane downward while cables connected to four supporting frames pull the lower ring upward.
The opposing reactions of the two membranes meet and transfer through a single compression mast. The four-directional structure supporting the lower ring also restrains lateral movement and torsion of the mast. The load of the upper ring acts as a design variable for adjusting the pretension of the upper membrane.
Two-way actuation cables are connected to the upper ring independently of the structural cables. Pulling one cable rotates the upper membrane, changing the overlap of the two fabric patterns and adjusting the shaded area below.
Restrained, Yet Free to Rotate
The lower ring is fixed to the four supporting frames, while the upper ring rotates within a guide formed by the lower ring. Rollers between the two rings restrain lateral displacement and twisting while reducing rotational friction.
Turnbuckles are placed at the cable ends and allow the pretension of the membrane and cables to be maintained.
Testing the Structure by Hand
The physical model used stretch fabric, thread, timber rods, and metal fasteners selected to reproduce the relative behaviour of the membrane, cables, mast, and rings. The membrane was connected to the rings following the intended assembly sequence, and the cables were tensioned to test whether the upper and lower structures could maintain equilibrium around the shared mast.
The model demonstrated the equilibrium of the two structures, rotation of the upper ring, and lateral stability of the assembly. It was not a structural safety test, but it confirmed that the force path and rotating joint developed in the digital model could be physically constructed and operated.
Form-Finding Through Physics
Three conditions guided the process: each unit and the complete membrane had to maintain an anticlastic surface, panel dimensions had to remain within a feasible fabrication range, and rotation of the upper membrane had to provide adjustable shade across the plaza.
The number and size of the units and the cable tension were parameterized in Grasshopper. Kangaroo2 then relaxed the mesh to test successive configurations. Stream Gate separated the calculation stages so unchanged upstream data did not trigger unnecessary recomputation.
A representative solar-noon condition at Barcelona’s latitude was applied to compare the shadows produced by different rotation angles. Alternatives capable of varying the shaded area from approximately 30% to 70% of the plaza were considered viable. Because the workshop focused on structural behaviour, the study remained a geometric shadow assessment and did not calculate seasonal solar exposure or thermal performance.
Preventing Localized Displacement
The first model applied equal tension at the vertices of each fabric unit. Although computationally simple, the Kangaroo2 displacement heatmap showed localized movement around particular vertices and narrow sections. The heatmap represented relative displacement under the applied tension conditions, not absolute stress values.
Instead of transferring the load directly to isolated vertices, a catenary edge cable was placed around the perimeter of each membrane unit. Lacing cords repeatedly connect the membrane edge to this cable, allowing tension to travel along the boundary. The form-finding process was repeated under this revised boundary condition, and the displacement distributions were compared.
From Curved Surface to Cutting Pattern
After establishing the basic structural geometry in Kangaroo2, material properties were introduced in MPanel to refine the final surface. The three-dimensional membrane was then flattened into two-dimensional cutting patterns. Compensation and seam allowances were applied so the panels would reach the intended geometry after tensioning.
The proposal was developed using a PTFE-coated glass-fibre open mesh with the following properties as the reference material.
| Item | Design Reference |
|---|---|
| material | PTFE-coated woven glass-fibre open mesh |
| reference grade | FIBERFLOW 30 (04646) equivalent |
| standard usable width | 3,810 mm |
| weight | 700 g/m² |
| tensile strength, warp × weft | 5,000 × 4,500 N/5 cm |
| tear strength, warp × weft | 600 × 600 N |
| light transmission | 34% at 550 nm |
| fire classification | A2-s1,d0 · EN 13501-1 |
Near-rhombic quadrilateral panels were joined in radial rows running from the centre toward the perimeter ring. The direction of the diagonal seams remains consistent within each row, while panels in adjacent rows are mirrored to reverse the seam direction. This alternating diagonal seam layout was intended to counter directional differences caused by the seams and the warp–weft anisotropy, distributing them symmetrically across the structure.
The lacing interval and the connection between the membrane and the edge cable were managed parametrically. Metal connections between the central mast, perimeter rings, and cables were resolved separately from the algorithm, distinguishing the scope of the computational model from the details requiring direct design.
Imagining Form Through Calculation
I like bridge design because it makes the flow of forces visible as form. This canopy likewise began with forces rather than a predetermined shape. It couples two membranes acting in opposite directions through a shared mast, then transforms the remaining rotational freedom into a variable field of shadow across the plaza.
The physical model confirmed the conceptual equilibrium of the upper and lower structures, rotation of the upper ring, and lateral stability. The digital model compared membrane geometries, relative displacement, and shade coverage under a representative solar-noon condition. Full-scale wind loads, structural safety, connection fatigue, and long-term durability remain subjects.