Greenland National Gallery

Using digital intelligence to identify lines of principle stress in a structure.

For the competition-winning design of the new Greenland National Gallery the RCD group developed a reinforced concrete slab solution that looked original and striking while also being more energy efficient than a conventional reinforced concrete flat slab. The engineering design was based on extensive research into mapping stress flows through flat plate structures, using advanced digital tools. The concept uses new research into the actual stress flows through a flat plate structure and also revisits work from the early 1950’s by the great Pier Luigi Nervi.

Pier Luigi Nervi famously expressed the lines of principle stress in the reinforced concrete floor slabs of the Gatti Wool Factory. When he analysed the stress vector field, he used an intuitive approach based on trial and error.

The engineering team wrote a computer script to map stress flow pathways accurately and comprehensively. By mapping the principle stress vector field of a flat concrete plate with support locations based on the positions of columns in the building frame they were able to design a precisely calibrated slab, where slab thickness varies according to the structural requirement. When the stress flow paths are expressed in reinforced concrete they create elegant, curved, ribbed structures.

D_pod pavilion

Computational analysis optimises uniquely curved form

D-Pod is a multi-use temporary grid shell structure. The architect’s original design (created using parametric software) posited each member as both curving and twisting. However, elements are often easier to fabricate if they are curved in one direction only. Connection detailing is also easier to standardise and therefore less costly if the curvature occurs only along one plane.

The engineering team created a digital tool to reveal the lines of principal curvature in real time to the designer. The architect was able to assess the curve network aesthetically before deciding on a final surface form.
Applying this curvature network constraint early made it easier to remove the twist effect and simplify connections once the design was finalised.

To make the shape of the grid shell more structurally efficient, the engineers morphed the underlying surface into a new position using a self-written script that integrated with the parametric software. Applying the principal curvature tool from the previous exercise to each new surface ensured the resulting structure was buildable.

Tallinn Town Hall roof

Volume packing and repulsion driven algorithm.

The new town hall for Tallinn, the capital of Estonia, is a structure composed of 13 intersection boxes, the largest of which is a 60m tall tower comprising the main council
chamber and 37m high lightweight glazed façade. A deep roof structure incorporating a staircase to take visitors to a viewing arcHitect platform was required and as a result the
RCD team developed a tool for finding the most efficient way of fitting structure to the irregular volume of the roof void.

As well as providing support against wind and snow loading, the roof serves to prop the high side walls and huge glazed façade allowing lateral forces to be transferred to the stability core at the rear of the structure.
The tool that was developed used a routine that served to release a specified number of nodes into the volume. Each node is then given a simulated electrical charge so that they repel each other. When the nodes find an equilibrium position they are all an equal distance apart and members are drawn between adjacent points. As such a triangulated structure is derived where all members are of equal length. Engineering judgment was then used to refine the structure in areas of high or low structural density.

Tallinn Town Hall

Algorithm mimics natural selection to evolve optimal structure.

The new town hall for Tallinn, the capital of Estonia, is a structure composed of 13 intersecting boxes, each of which cantilevers a considerable distance from inset columns at ground floor level. The side walls of each box are blank and therefore provided the opportunity to hide the arcHitect supporting cantilevering trussed frames.

The RCD group used the theory of genetic design to evolve and engineering solution that went beyond the traditional structural typology of a truss to deliver a more
optimised result.

The genetic algorithm was originally developed by John Holland in the 1960s and is a computer simulation of Darwinian evolution.

The engineers’ genetic algorithm solver initiated a population of possible truss arrangements which were assessed against a performance-related fitness criterion – in this case the deflection of the trussed frame. The resulting solution, which could not have been intuited using guesswork or traditional engineering assumptions, is the best optimised solution that effectively reduces deflection.


Automatic re-arrangement of the Tallinn facade trusses using a genetic algorithm