I just returned from the SmartGeometry 2010 Workshop, this year held in the Mediterranean coastal city of Barcelona, the capital of Catalunia, also known for its wealth of unique historical architecture.

The theme of the workshop was centred on the challenge of ‘Working prototypes’.  The focus being to develop functioning prototypes for the purpose of proving and testing conceptual designs.  By placing fabrication at centre stage it challenged participants to design, assemble and test working prototypes.

To facilitate this year’s theme, the workshop was held at the Institute for Advanced Architecture of Catalunia (IaaC) in Barcelona.  This was a great venue and in addition to the main space for the SmartGeometry workshop there was also break-out rooms and facilities onsite including the Fablab – fabrication laboratory.

This workshop attracted a unique mix of over 100 attendees from across the world of academia and professional practice for four intensive days of design and collaboration.  Most days began at about 8:30am and did not conclude until midnight when tutors and organisers began to usher attendees back to their hotels for some well-needed sleep.

SmartGeometry 2010 was organised around Clusters. Clusters are hubs of expertise comprising of people, knowledge, tools, materials and machines. The Clusters provided a setting for workshop participants to work closely together and to exchange ideas, processes and techniques for the development and testing of working prototypes.

Ten clusters were formed for SmartGeometry 2010.  These included the following:

I attended the ‘Design to Destruction’ Cluster.  The aim of this Cluster was to control/optimise a design through a recursive process of computational analysis, small-scale prototyping and physical testing.  The key outcome was to integrate this analysis into the design process using testing as a validation of the design.  Ultimately the goal was to test each final design to destruction at full scale.

All Cluster participants were required to make a CNC milled or laser cut 1.2m timber cantilever, which would later undergo a calibrated structural test; the ‘winner’ being the design with the lowest self-weight but highest loaded capacity.  Fabrication was undertaken using the fantastic array of equipment housed within the IaaC’s Fablab.

During the first day of the workshop, I developed an Evolutionary Structural Optimisation (ESO) scripting routine that links up directly to Arup’s in-house structural analysis software GSA.  The ESO process begins with a full mesh domain of 2D plate elements and through an iterative process the mesh is gradually eroded away by removing under-utilized material ultimately leading to an optimised design.  The ESO script automatically executes this process to find the optimum 2D elemental mesh idealisation of the 1.2m timber cantilever.

In the subsequent days, I fabricated several small-scale prototypes using the rapid prototyping laser cutter machines in IaaC’s Fablab.  These represented directed output from the ESO scripting routines to provide a selection of designs to choose from.

Given the focus of the Cluster was to develop a design with the lowest self-weight to loaded capacity, I aimed to erode my designs down to a void ratio of less than 70%.  With the final solution in hand this was sent to the larger laser cutter machine for full-scale test prototype production.

The final day of the workshop saw all participants with their final full-scale designs ready for testing to destruction.  These ranged from the highly aesthetic designs, often with a low void ratio, to the very aggressively optimised solutions with void ratios greater than 70%.  A custom testing rig was assembled by the tutors prior to the workshop and this was used to test the cantilever specimens using a selection of gym weights as a loading mechanism.
Upon testing my final design with a void ratio of 73% it was able to carry a total load of 170kg giving it a self-weight-to-load carrying capacity ratio of over 55 and placing overall in a respectable second place among all designs.

Overall SmartGeometry 2010 proved to be a very intense but enjoyable experience.

The use of custom tools can improve efficiency, especially when handling large structures with thousands of elements. One of the most useful tools I have recently developed is a strength based optimisation program that links with Arup’s in-house structural analysis software Oasys GSA through its API (COM interface).

The COM interface uses Visual Basic (VB) scripting and allows remote access to most of GSA’s functions, enabling the user to drive GSA externally through programs such as Excel. This means the user can automate calculation intensive processes and handle large amounts of data in an efficient manner.

While working on a multi-billion dollar project in Singapore, I was faced with the challenge of analysing and designing several large steel structures with highly complex geometric form and 5000 plus elements in each structural model. I initially developed this tool merely to deal with the sheer number of elements that were required to be analysed, designed and sized.

Later, I saw the potential for expanding its use by incorporating an optimisation routine. This was achieved by running a computer automated numerical algorithm to determine the least steel-weight of the structure while still satisfying the relevant design codes.

The optimisation algorithm will not attempt to modify the geometric form. Instead, through an iterative process, the sizes of the structural elements are continuously substituted (from a user selected pool of sections sizes), analysed, designed and resized based on the results of the analysis, until the structure with the least steel-weight is found.

It is also possible to calibrate the algorithm for different optimisation criteria. For example, an alternative criterion to the least steel-weight solution could be shallowest depth solution, least cost solution or fastest procurement solution, or a combination.

I have used this optimisation tool on two other projects. In these cases it was used for rapid design and member sizing of various scheme options and allowed the design team to quickly assess the most economical design options.

Not only was this an effective and highly efficient method of managing the data from large structural models, it also offered the opportunity to fine tune the steel weight of the structures, resulting in potential cost savings.