The Digital Workflows and Computational Design Panel
Ashley Kacha, the Institution's new Senior Engineer for Digital Workflows and Computational Design, outlines the work of the Institution's Panel and reports on his trip to Digital Construction Week 2017.
As structural engineering moves further into the digital realm, the Institution is making considerable effort to provide the support members need through its Digital Workflows and Computational Design Panel. Formed at the end of 2016, the Panel is a group of practicing engineers, academics and architects with interest and experience in the use of computational methods for design. The Panel meets once a quarter, with smaller meetings scattered throughout the year.
The Panel’s work ranges from its contributions to the March 2016 Digital Design issue of The Structural Engineer, to talks by Panel members Tom Webster, Tim Lucas and Stephen Melville, and has been very well received so far, with much more activity planned for 2018.
As Senior Engineer for Digital Workflows and Computational Design it's my role to support, compile and articulate the Panel's work to members and the wider construction industry. This October I attended Digital Construction Week to explore trends on the horizon for structural engineering.
Digital Construction Week
Digital Construction Week is a yearly event with topics and audiences varying considerably, with advances in design software, construction robotics, drone survey equipment, and augmented reality showcased between ‘theatres’ hosting a variety of case studies and demonstrations. These all sit in the shadow of the main conference programme which is similarly varied in content, but closer in style to presentations we’re accustomed to seeing in keynote speeches from the giants of Silicon Valley. The following are just a couple of the key themes with implications for structural engineering.
The use of artificial intelligence and machine learning was a particularly popular theme at this year’s conference. This is unsurprising, especially given that the recent surge of research into automatous robotics has sparked renewed interest in machines that can learn from data instead of simply following preprogramed instructions.
Carolina Bartram's talk ‘In the Year 2525 if Man is Still Alive’ and Giulio Antonutto's talk ‘Will AI redesign the designer?’ both emphasised the benefits in using AI to explore new forms and to generate options that may have never otherwise been considered.
Carolina’s talk related to the role of ‘design thinking’ in engineering. An amalgam of engineer and architect, the way Carolina works echoes the likes of renowned designers such as Frei Otto and Pier Luigi Nervi. Communication is key, and Carolina’s passion for design was evident in her presentation of the benefits AI could bring to the framework which engineers use to think and create.
Giulio began with the history of computing and AI, referencing the Turing test and the iconic chess match between IBM’s Deep Blue and Garry Kasparov. The talk then speculated on real world applications of AI. Examples of how feeding AI with classical music teaches the software to compose its own, indicated the basis upon which scheme designs could be generated in the future. Giulio’s presentation also emphasised that understanding the processes behind machine learning will serve the design engineer in years to come. He introduced the concept of data bias in the learning process and that producing what amounts to ‘learned behaviour’ would undermine the key benefits behind utilising AI in design.
Mike Haley (Autodesk Research) tackled the issues of simulation and ‘Big Data’ for teaching machines in his talk ‘The Future Will See You Now; AI for Design & Build’. He argued that machines undertaking ‘self-driven design’ run the risk of creating black boxes that human minds would never be able to comprehend.
The philosophy behind the application of AI is that it need not be a challenge to more traditional methods. AI can instead augment current design processes, populating the design space with options based on the fulfilment of design objectives. This implies that in contrast to the concept of machines running amok, the engineer will become a guiding hand with complete control of the final outcomes. The engineer will have the power to intervene in the generation of design options – choosing acceptable levels of trade-off between different design objectives.
Above: Radar charts for objective fulfilment
The ‘Industry 4.0 Theatre’ hosted presentations on the topics of digital fabrication and computational design.
Panel member Edoardo Tibuzzi (AKTII) demonstrated how computational methods had enabled his projects to become easier to deliver. With emphasis on the 2016 Serpentine Pavilion with Bjarke Ingels, Eduardo demonstrated how rapid delivery of feedback on the architectural form was achieved with the use of computational methods. Enhanced interoperability between software packages via detailed global and local finite element models connected through a web-based ‘middle man’ allowed stability checks and design of individual glass fibre reinforced polymer (GFRP) modules as the architectural design developed. The Panel is keen to communicate the benefits of developing such workflows along with guidance on how to do so within the coming months.
Taking cues from the automotive industry; a number of speakers had demonstrated the use of robotic manufacture. Indeed, the sources for many of these robots are refurbished factory assembly machines adopted and repurposed by students from the Bartlett School of Architecture, MX3D (of steel bridge printing fame) and the aptly named HAL robotics for use in construction. We’re likely to see much more of this to come as research and testing is undertaken on projects currently in production.
The popularity of Digital Construction Week is a clear indicator that the construction industry is waking up to the potential innovation new technologies offer.
Anthony Buckley-Thorp, Panel member and managing partner at FLUX, presented with Balfour Beatty about using hackathons to quickly solve problems and implement solutions within days, rather than months or years. It seems that our industry’s resistance to change is beginning to soften. However, it should be said that a healthy scepticism is still very much alive. There is little doubt that critical thinking and caution will be indispensable in a future where our design tools may develop their own methods to solve problems.
In the coming months, the Institution aims to promote computational methods and demonstrate the value offered by defining digital workflows within the design office. The latest developments in the industry are often viewed with a level of cynicism by some, but careful and well managed adoption of computational methods offers considerable benefits for structural engineering as a profession.
From the Panel:
Read The Structural Engineer Digital Design Issue (March 2016)
Panel member Tom Webster discusses Awareness of Scripting (Webinar)
Panel member Tim Lucas discusses Digital Fabrication (Webinar)
Panel member Stephen Melville discusses The success and potential failure of engineering computational design (Webinar)
Ashley is a structural engineer based in London. A University of Bristol graduate, he worked for Fluid Structures on a wide variety of projects from scheme design to construction. Prior to joining the Institution earlier this year, Ashley had been developing the computational design team within the practice.