Parametric Modelling & Optimisation
A building can be thought of as a collection of objects. Parametric modelling defines the design of those objects by sets of parameters.
The parameters can be properties of the objects themselves such as length, height, density; or properties of the objects’ performance, such as the amount of heat lost in winter or the distance from a post office. We use computers to calculate the properties of proposed building designs and modify the designs until we find the ones that perform the best.
Rather than directly designing objects, we design algorithms (sets of rules) that in turn design the objects. Thousands of design options can be tested by rapid, virtual prototyping to find one or more solutions that are the best fit.
HAYWARD GALLERY ROOFLIGHTSView
The proposal for the refurbishment of the Hayward Gallery includes making better use of the existing, iconic pyramid-shaped rooflights. External rooflight shading was modelled parametrically in a 3D CAD environment. We developed and programmed our own algorithms to simulate diffuse daylight and direct sunlight into the gallery spaces.
We used a genetic algorithm to optimise the shade shapes, providing the most diffuse daylight whilst adhering to conservation requirements that dictate no direct sunlight on the artwork.
The MAX:R+I team have developed a Building Forms modelling tool that can generate and optimise the position and shape of a building in relation to set environmental parameters.
In the example below, an initial building plan is drawn at the centre of the image. Using a genetic algorithm, the plan shape is modified within a defined range to maximise the view distance from all the building’s windows. The height is kept constant and the building floor area is kept to within 10% of the original.
The same tool can be used to optimise the building shape for a range of parameters such as daylight, passive solar heating, summer shading, distance to a bus stop - even the number of mature trees that can be seen from the windows.
Any shape of building in any surrounding can be considered. The Building Forms tool uses custom written modules in Grasshopper, Rhino’s logical component modeller.
A hypothetical building in a densely built-up part of London – can you see the Gherkin? – with a model of the views from through the façade. Map used under Creative Commons license courtesy of © OpenStreetMap contributors
The Building Forms tool is not limited to obvious environmental variables. It’s possible to model for any combination where data is available, for example, social interactions, psychological preferences or the probability of seeing a red sunset!
Optimising lecture hall acousticsView
As part of our work for the new teaching building at the University of Nottingham, we investigated how to achieve the best acoustic conditions in Lecture Theatres – specifically in terms of how clearly the audience can hear the speaker’s voice.
The shape of the ceiling plays an important role because it determines how sound is reflected from the speaker to different parts of the audience. It’s obvious that there must be a ceiling shape that results in the best acoustic design but finding out what it is proved to be a tricky problem.
Our approach was to use a genetic optimisation algorithm to seek out the best acoustic design. Working together with Make Architects, the derived ceiling shape has been included in the final design.