The design and construction of the 35 metre high rose window to the west transept (Passion Façade) in a little over twelve months is a significant example of “lean construction”. The processes involved include traditional stone masonry, and semi-automated construction methods. At a time when digital tectonic studies enter the spotlight, this project serves as a useful beacon for blending traditional craft with numerically driven efficiencies to their mutual advantage.
The Rose Window (Rossasa in Catalan) is an eight metre wide by thirty-five metre high assembly of 15 openings, culminating in a huge single elliptical opening that sits between the two towers of the west transept and throws west light into the central crossing.
As there was no surviving restored plaster model from Gaudí’s time for this element, the starting point for detailed design was provided by 1:25 scale plaster studies of this window carried out during the 1970s by Puig I Boada and Bonet Gari, two of Gaudi’s followers. The proportional system of the East Rose Window in the Nativity Façade, built during Gaudi’s lifetime and the lineage of 1:10 scale models for the lower and upper side aisle windows also informed the interpretation.
THREE DIMENSIONAL DIGITISING
Puig I Boada and Bonet Gari’s 1970s model was digitised, using a pointer on a 3D arm that locates points in Cartesian space. These points were converted to points for representation in a corresponding virtual coordinate system in a 3D modelling program. This digital facsimile of the plaster model was used as the base for creating a precisely controlled flexible model using the technique known as parametric design or associative geometry.
THE PARAMETRIC MODEL
This computer modelling process involved the Boolean subtraction of a series of solids, primarily hyperboloids of revolution, from the notional solid of the wall. Each subtracted or sculpted form is described or defined in the digital history by a geometrical definition with editable parameters, and there are sequential geometrical relationships between each element, also governed by parameters.
This approach utilised high-end computer software traditionally targeted at ship, automobile and aircraft design. Such software results in a model that can subsequently be altered, subtly modifying the position or size of an opening or the proportional system governing the relationship of the openings.
Making a single change to the geometry of one opening will impact on all the neighbouring openings and this way of modelling permits the whole 3D model to update without any need for erasure or remodelling.
For the Rose Window there are 3,800 events in the parametric modelling history. So for any characteristic of the model to be changed there are 3,800 events that have to be checked and updated before the geometry of the model can be regenerated. The model was developed over 6 weeks in Barcelona and completed in Australia.
MARK BURRY 