Parametric modelling freed up the design development of Foster + Partners’ Hydro, writes Felix Mara
As I sit on the mezzanine of Foster + Partners’ London office, slides documenting the evolution of the practice’s The Hydro flash across the screen. Set in Glasgow’s former docks, it is now at fit-out stage and will open open later this year as a multipurpose performance and sporting venue ahead of the 2014 Commonwealth games.
‘Because of its non-repetitive, offset-cone geometry, the only way it could be costed and built involved constructing an intelligent parametric model, shared with other project team members,’ says partner Ben Scott. ‘But there’s no such thing as a completely automated parametric model: there are always breaks in the system.’
Theoretically, it might be possible to build a full-blown detailed parametric model of what will be Scotland’s largest arena, but it would be incredibly unwieldy. It’s much easier to isolate variables for analysis, as engineers do. It would also be prudent to have roadblocks, in case it got out of control.
After hammering out the initial concept, which involved cutting the steeply raked bowl diagonally, concentrating seating at the front and optimising public areas, Fosters used Bentley’s GenerativeComponents associative parametric modelling engine to explore the tilted double-layered diagrid spaceframe roof and evaluate design development options.
‘Computational design allows you to set up series of rules in a geometric or script environment and ask “what if?”,’ says Bentley product marketing manager David Huie. Scott’s architect colleague, David Gillespie, adds: ‘We developed a series of Geometry Method Statement (GMS) drawings in GenerativeComponents, documenting the geometrical relationships and properties of essential components, which together form a 3D setting-out model.
‘It was important to codify these so relationships between building components could be understood by the whole project team and we developed a GMS roadmap, a higher-level symbolic diagram showing relationships between certain elements and the impact of changes to specific components’ setting out.’
Certain relationships, those between the column and roof set-out and the roof space frame, for example, were specified by co-ordinates rather than geometric dependencies.
In these cases, the proviso was that if these one of these changed, co-ordinates should be regenerated. Spreadsheets and 2D drawings exported from this model were integral to the project team workflow.
The roadmap was crucial where packages interfaced or involved contractor design. GMS objects were identified by camelCase. For example, the eaves detail interface involves design responsibilities as well as a complex set of components. By defining essential 3D setting-out zones, it was possible to establish interfacing components’ design intent through their relationship.
This was vital in specialist construction, such as the roof’s acoustic build-up. The team worked to nominal build-ups, and the specification allowed input from the contractor’s supply chain to optimise value and performance, potentially allowing zones to be variable to ensure certain elements aligned at their interfaces. Flexible parametric setting out of these elements was essential. Where there were variations in geometry, relationships between components could be understood and described in detail.
Because the bank around the arena undulates to suit internal requirements such as loading bay heights, Fosters devised a control system to safeguard consistency with the design logic. A law curve controlling the bank’s height at specific gridlines was used to explore options, enabling Fosters to identify heights and relationships at critical points. Detailed parametric modelling also enabled geometric variations to be monitored, allowing the cost implications to be more easily explored, and the model facilitated on-site setting out and construction. It enabled Fosters to develop detailed formwork patterns and alignment and set out the straight precast nosing units forming the bank’s perimeter edge as a close approximation to a B-spline curve.
The cladding and perimeter roof lights are a natural extension of the bank geometry. The twisting surface of flat glass panels was developed through parametric cladding modelling with bank geometry input, facilitating exploration of alternative ways of using them to form complex surfaces.
‘In the final design, a panel popping technique was used to flatten panels locally, enabling us to control the geometry and ensure resulting interfaces were buildable,’ says Scott. ‘As with all elements, these relationships were codified within the GMS.’ This enabled the contractor to develop the model and interpolate relationships, leading to a prototype and on-site installation using computer-controlled manufacture from the model.
The in-situ fin walls involved complex co-ordination and geometrical problems, as their size and shape was driven by circular setting out grids with different loci, forming a crescent configuration which interfaced with nearly every package and building element. Here digital design was used at all stages, through to CNC fabrication. Fosters used parametric modelling for reiterative checks and it encouraged fluid design development and optioneering, avoiding a premature design freeze during the lengthy gestation. The GMS has communicated and recorded the arena’s complex network of design and set-out principles. ‘It’s about relationships,’ says Gillespie. ‘And, as with all relationships, the more you have, the more complex it gets.’
Start on site 2011
Client Scottish Exhibition Centre
Structural engineer Arup
Quantity surveyor Gardiner & Theobald
M&E engineer Arup
Budget £18 million
Software used Bentley GenerativeComponents, Architecture, MicroStation, Navigator and Microsoft Excel
Design thaw: Hydro by Foster + Partners