The new engineering

A discussion with Arup’s Tristram Carfrae

In 2008, I met Tristram Carfrae at Arup’s Sydney office, which sits in a couple of storeys of a fairly anonymous block on Kent Street. The anonymity is typical for Sydney’s ultra-business-like CBD, but so is the casually wonderful aspect — in this case, overlooking the vast Barangaroo development site on the side of the city’s CBD, and the splintered skein of deep harbour out towards Balmain, criss-crossed with ferries and functioning as the occasional water-berth of the ocean liners, great white slabs of floating architecture that add a temporary residential function to the port, before moving on east to Hawaii or north to Hong Kong.

We sit in the reception and talk for a couple of hours, over a few plastic cups of instant coffee, and just behind a model of the National Aquatic Centre, the building designed by Arup, led by Carfrae, and Australian architects PTW with Chinese partners CCDI, and that opened in Beijing just over a month ago. Known colloquially as the ‘Water Cube’, it’s being seen as a new masterpiece of structural engineering, already enjoying a flurry of publicity that will only increase in fervour when the Games start this winter. But Carfrae’s career extends well beyond the Cube, and he’s generally thought to be one of the leading structural engineers of his time.

We talked about the Cube, and the changing practice of structural engineering that informs such projects, and how this is radically changing architecture, building and cities. We chewed over multidisciplinary working and design processes in general, as well as techniques new and old. I’ve organised this piece into loose themes, as the conversation jumped around a little.

On engineers as collaborators

Tristram Carfrae started by suggesting that, during the ’90s, the structural engineering profession felt that it had become commoditised to some extent, via the widespread implementation of CAD and simple off-the-shelf analytical software on the one hand, and with the architect often being seen as “the sole author” of buildings on the other. However, Carfrae now perceives a swing back towards the structural engineer, through the potential in advancing and extending building information modelling (BIM) and allied techniques, as part of the collaborative design process.

Even though he’s clearly excited by this, Carfrae wonders if there’s even a risk that engineers become too dominant for a while — and as a result that “the spiritual side of life” or the wider social context, say, which traditionally many engineers haven’t dwelt on (with honourable exceptions, not least from Arup’s ranks) — could get momentarily lost as a result. We’ll return to this, but Carfrae sees that the most straightforward way around this potential problem is always to be found through multidisciplinary collaboration — “to keep talking”, as he puts it — allied to a shift in the outlook of the engineer.

Carfrae believes specialists must focus on moving towards each other, each attempting to “care about the whole thing” in terms of the impact of the project at hand, talking and collaborating across all disciplines. Carfrae reflects on his frequent collaborations with Philip Cox, and also those of Renzo Piano and Peter Rice and the collaboration he observed between the two. Carfrae recalls the great engineer Peter Rice accepting the Gold Medal for Architecture from the RIBA, whilst politely bristling at the suggested notion that he was essentially working as an architect — he stated clearly he wasn’t, that he was an engineer, and that that was his value in the partnership with Piano. Carfrae also sees immense value in the relationship between engineer and architect. Given what he describes as the stereotypical engineer’s methodical, grounded approach based around generating numerous possibilities for structures that will stand up, the stereotypical architect can bring a comfort of dealing with non-structural problems, a different aesthetic sensibility, and the ability to snatch a solution out of thin air. Stereotypes hide a far more fluid relationship, but for Carfrae, the roles of architect and engineer are best articulated as a collaboration between equals, with the same going for the numerous other disciplines involved in contemporary building.

So multidisciplinary work can only be predicated on the basis of mutual respect for the disciplines involved. But going beyond this, Carfrae suggests that if specialists plough a sole furrow only within their own specialism, you’re likely to end up with bad work.

Drawing from some experience with Arup Associates, in they had total control of most aspects of the project and in which every discipline was seen as equally relevant, he’d noticed that without a creative tension, hierarchy or very clear direction, people could get quite reticent about putting ideas forward. Under these conditions, you get a kind of rudderless stasis, without the spark of inspiration that comes from extending beyond the limits of their discipline, and from the crossover into another.

So instead he advocates that the acoustician, for instance, should work beyond the limits of acoustics, the structural engineer likewise. In fact, each discipline has to move some way towards each other. Aligned around this overarching sense of the project, the structural engineer is to contribute structures with best looking detail they can; while the architect has to draw something that can stand up. Both disciplines move towards each other in this way, raising their game, testing their ability and creating a tighter coupling for better communication.

It clearly also takes careful, sensitive management too — which he self-deprecatingly describes as “prodding”. Equally, there’s certainly a role for providing a clear and engaging vision — he hesitates over the word “visionary” — that enables the multiple disciplines to corral their efforts around a big idea. This unified sense of looking at “the whole thing”, allied to that of “the one idea”, provides a fulcrum for aligning various disciplines. With the Beijing National Aquatics Centre — aka ‘Water Cube’ — things aligned very neatly around a clear idea. The requirements for acoustic transparency, and to fill the rectangular footprint provided, plus the inevitable need for novelty that a competition tends to demand — these all led to the solution of an ETFE-based ‘soap-bubble’ cube structure, demonstrated through rapid prototyping. Having arrived at that, most other things followed, including the wider historical perspective drawn from the arrival of the neighbouring ‘bird’s nest’ National Stadium structure of Herzog & de Meuron/Arup, in which various local cultural references (square/circle, red/blue, male/female) fell neatly into place. That context all arrived quickly in one meeting with partners (Arup, PTW, CCDI), shortly after the ‘birds nest’ scheme had been announced. The structural strength of the core idea itself enabled effective collaboration.

On judgement, algorithms and rapid prototyping

Several times Carfrae refers to the importance of ‘judgement’, and of engendering this within the engineer. Not least in the intensive, immersive, collaborative process described above. But also in the use of software-based optimisation tools, in terms of articulating the space algorithms explore in the first place, and then secondly in a kind of call-and-response relationship as the algorithms are running through possibilities. In response to fears that this kind of ‘algorithmic architecture’ will marginalise engineers and architects, Carfrae states that this kind of approach is only really “optimising the last 10% of a problem.” The software has to be described and tuned with a particular strategy or problem in mind, and that comes from the designer, not the software. In the Water Cube, the Chinese partners CCDI wrote software in collaboration with Carfrae’s Arup team, the engineers adding structural analysis, sizing sequences, rules, connection nodes and so on, effectively re-writing the software to performance test structures in real time. In another part of the project, using rapid prototyping techniques to fabricate models, the software would occasionally build ‘incorrect’ structural elements — Carfrae describes some “false scaffolding” that emerged on one of the models output by the system — this again takes human interpretation and intervention to tweak the rule-sets and processes.

Other projects Carfrae has worked on in this way include the Melbourne Rectangular Pitch Stadium and 111 Eagle Street, Brisbane (both with Cox) and King Street Wharf in Sydney (with Fitzpatrick Partners for Multiplex). All these projects vary in process slightly, but all share the principle of using algorithmic optimisation to assess the performance aspects of various structures, with human intervention in the selection of successful patterns and outcomes.

The rapid prototyping aspect of this work also exemplifies judgement and collaboration. The technique is drawn from manufacturing industry — particularly the automobile industry — and is proving increasingly fundamental to structural engineering. For the Water Cube, it meant that the team could both immerse themselves in working solely with digital 3D models, with all the benefits of pace, fluidity and automated structural optimisation that can bring, and still have the benefit of producing physical models. These models, essentially ‘printed out’ on-demand in epoxy, provide a tangible means for exercising judgement — see the “false scaffolding” problem — and also meant that the team could present competition judges with “a kit of models”, essential to conveying the ideas behind the design and providing a platform upon which the judges could exercise their judgement (favourably, in this case.)

Sidenote: The model then persists throughout the process, ultimately delivered to contractors who can derive the entire steelwork package from the model, element by element. (The Travellers, sculptures on a bridge in Melbourne, are another case study with complex steel structures manufactured and assembled directly from the Arup 3D model. Increasingly, components will arrive on-site with data indicating their exact coordinates within the overall structure — a jigsaw puzzle that virtually assembles itself. Linked to the Building Information Model, and via RFID or equivalent to ongoing pervasive systems, it’s also a jigsaw puzzle that reports its state, behaviour, and information for disassembly and recycling, when necessary.)

Rapid prototyped model of Water Cube

You only make tools to do the job” is how he describes this sense that the software tools are written with a strategy in mind. Yet Arup are clearly making tools here, as well as outcomes, and Carfrae sees this as a natural evolution of the work. Arup have always made tools (in the widest sense), of course, but if there seems to be an increase in quantity and sophistication it is always as part of a problem-solving exercise. He admits there have been a couple of projects — or aspects of projects — where they’ve deliberately bent the solution in a certain direction in order to create and test a new tool. With the design and build of a down-pipe, for example, a tool used and evaluated was of more strategic importance to the company than the delivery of the down-pipe itself.

Tools also benefit from the collaboration described above. For the 111 Eagle Street project in Brisbane, Arup’s development of modelling software is based on algorithms that simulate the growth of plants and their constant search for sunlight. This technique can be placed squarely within the theme of ‘biomimicry’ and thus necessitates collaboration between biologists, architects, software engineers and structural engineers. In this particular case, the software generated numerous successful patterns for a perimeter frame — successful, that is, in terms of load-bearing vertically and laterally — from which the final selection is made visually. So judgement, sometimes subjective, is still the ultimate arbiter within this software-led process.

On the new engineer

This tool-making capacity, and the increased importance of a fluid and responsive judgement on the part of the engineer, could be seen as part of a new skillset required for the contemporary engineer. Carfrae outlines the facets of this new kind of engineer; one “amenable to process”, who is “happy wallowing in complexity; can appraise vast quantities of data; will make decisions using judgement (without having an absolute answer)”. He says that in some senses “we’re changing the kind of engineer we have” to enable this new hybrid.

Carfrae describes how, historically, structural engineers were essentially trained to be numerate; that it was a form of applied mathematics above all. With previous tools, engineers were licensed to use them only when they’ve demonstrated an understanding their innards in detail. With the increased complexity of today’s tools, that’s generally not possible and certainly not scalable. Equally, using contemporary modelling software, engineers can explore situations that they may not have been able to conceive — put simply, situations and processes that can’t be drawn.

Sidenote: Carfrae notes that this work is sometimes difficult to perform in collaboration; that the nature of exploratory code-based exploration can be difficult to share across a team. The role of knowledge management, in an ‘at-hand’ everyday sense, seems ever more important to this new engineering.

So now engineers, and Arup, have to be comfortable with “using stuff you don’t understand” on structures and processes that you can’t draw; a comfort with informational processes upon which they deploy critical judgements. As he puts it, this means being “more judgemental, less absolute”, and being particularly aware “to criticise the judgement they just made”.

By constantly looking at these judgements on process from different perspectives, informed through collaboration, you should end up with an outcome that is richly imaginative and holistic, yet grounded in testing and modelling.

These perspectives should also encompass the aesthetic, social, cultural, economic and political dimensions of the work, as well as the technical, again best derived through collaboration. Carfrae suggests, and is happy to describe himself in this way, that the engineer’s mindset is not naturally attuned to being interested in what’s going on in the world around them. They instead tend to focus on the problem at hand, and work within their field of expertise in order to solve it. Each new project is like being presented with a jigsaw puzzle, and the last thing they want to see is the picture on the box. Ignoring precedent occasionally — or reinventing the same wheel ten times, as Carfrae puts it — the engineer tries to work out the problem from first principles, often their own first principles, and Carfrae certainly sees that it’s the organisation’s role to interject with prompts, suggestions, collaboration and best practice from elsewhere.

It means Arup are instead cultivating engineers that are, on the one hand, comfortable with data, but on the other, aware that they are ultimately “constructing a piece of urban fabric”, as Carfrae puts it. It’s an intensification of both sides of the work.

Sidenote: Once you see information as a material itself, it appears a logical extension for a company used to working with materials. So alongside glass, steel or ETFE, say, information becomes a malleable fabric in both the design process and the outcome of that design process. So it’s a new skillset but also a natural progression for Arup and similar companies in this field.

On skillsets and tools

Carfrae refers to the ‘T-shaped person’ idea (see IDEO amongst others) as well as drawing several times from knowledge of Lockheed Martin’s ‘Skunk Works’ R&D programme. The T-shape suggests a depth of skill in one particular discipline — the stalk of the T — and yet also a broad, empathetic understanding of other disciplines and perspectives — the crossbar of the T. According to Carfrae, the traditional engineer was trained with a “deep vein of skill”, and so would represent a T with a long stalk, and not necessarily a broad crossbar. Carfrae thinks we might have to re-balance this slightly; though not necessarily at all levels. He notes the career framework suggested by Ben Rich (Skunk Works ex-boss): that each subsequent decade of your career sees a shift from learning, to being skilled at the component level, to managing airframes, to trying to fix the whole world! (The other aspect of the Skunk Works project that resonates with Carfrae is the method of deep immersion in a problem, through exploring all possible sources and references … and then deliberately consciously discarding it, leaving it to subconsciously bubble-up in a collaborative design process.)

A deep understanding of tools is still required, however. In fact, it’s almost inevitable that an engineer will develop a strong affinity for a few discrete tools. Carfrae notes he’s still using the same tools he used 20 years ago — in the form of GSS Relax, software he helped write. He knows this tool inside it, exactly what it can do. Equally Bentley’s MicroStation is another long-standing tool of choice. But there’s a tension here, which he recognises, in that a level of virtuosity with a particular tool or technique could also lead to skills and understanding being “frozen in aspic”. How to shift tools and techniques from one state to another will be interesting, given the rapidity of developments in this field. Again, though, the slightly broader, collaborative perspective of the new engineer may help.

At a larger scale, Carfrae feels that the ability to shift engineering work in general from one state to another is equally complex and fluid, and sometimes highly unpredictable. Trends come and go, just as with other pursuits. For instance, much work with stadia over the years enables him to see particular engineering trends there — currently clad arena, perhaps after Herzog & de Meuron’s Allianz Arena. These leaps to a different form of practice can occur almost randomly. He spent a year working with Philip Cox on stadia and convention centres in China, in 2002, developing a scheme every week. Perhaps understandably, at the end of the year he felt that each scheme had become a derivative of the others and he was stuck in a rut. But with his next project (with PTW), the Water Cube, came completely out of ‘left field’ and also came together in a week. Whereas on another project with Cox, a bridge in Singapore, it took the best part of two years to slowly evolve a design they could be proud of. One project takes a week and seems derivative; another takes a week and seems completely fresh; and another takes two years to evolve. And there seems little or no method that can be applied to account for these leaps in development; leaps that shift engineering from one state to another, or close off one trend and start a new one.

Carfrae likens each a trend, or stage of development, as a structure knitted together from the collective practice of several key firms. So Arup’s work constitutes a line of development over some years, intersecting with that of Cox, or Foster & Partners, or Renzo Piano, say. Together, these lines knit together to form a fabric representing a state of development.

Development shifting from one state to another

Shifting it to an entirely new plane would seem to be a process based around chance as much as development in materials and software or through inspiration, though the aforementioned process of immersion clearly helps. Rather than finding this lack of method daunting, Carfrae clearly finds this a healthy state of affairs and constantly engaging.

On cross-fertilising knowledge

That week of development on the Water Cube was actually based around some intensive web-based research by Carfrae, following up his own hunches on the ‘soap-bubble’ structure and using Google to ferret out previous research and development. A week later, most of the sources from around 200 years worth of scientific exploration were at hand, all pulled from the web, along with key contacts to talk to. An intensive process of software modelling and rapid prototyping ensued. Carfrae believes that this rich resource of the internet is something his company could tap into more, and contribute more to, as part of their business. He talks of trying to turn their intranet inside out, and sees this as another natural evolution of Arup’s role.

During the ’90s, that era of the architect as perceived sole author, he believes that Arup played a vital, often un-credited role in terms of cross-fertilisation — knitting together that fabric described above, by working across numerous projects, sharing best practice, and communicating technical and business breakthroughs. So it’s natural that Arup would now begin to explore what it can do in terms of informational innovation, using the web internally and externally. Issues of risk could be side-stepped to some degree, as Arup can have a certain confidence about their practise, derived from scale, reputation and ability. This approach fits in well with the current developments in ‘open innovation’ that are sweeping through many related business sectors. So Arup’s role as “cross-fertiliser”, as Carfrae describes it, looks set to continue. Some business arrangements don’t allow this, depending on the partners, but Carfrae is heartened by the agreements around the Water Cube project, where as long as each partner credits one another, they’re free to say what they like about it. This has led to a project which is well represented online, both during construction and after, providing a wealth of information to be taken into future problem-solving, by Arup and others.

On the model as platform

The new knowledge-sharing platform for Arup is not necessarily a wiki, email-based forum or series of special interest groups — though they have those too, of course — but actually the advanced forms of building information model (BIM) they’re creating. These models comprise a platform for an ever-increasing range of informational overlays on buildings, both during the design process and in the operational phase. It provides an independent model that can comprise almost all different disciplines. For example, working on a potential major renovation of the main Sydney Opera House theatre, the 3D structural model was linked to the facilities management (FM) databases, meaning the model became a geometric organising tool for the 26 disparate FM databases.

3D structural model of Sydney Opera House

It’s clear that this principle can be extended, both in terms of the number of overlays but also the scale of the model, zooming in to a particular component, and out to city-scale. On the former, it means Arup can explore the effects and behaviour of, for example, smoke, egress, acoustics, lighting, traffic, even microbes, allied to all the structural performance aspects. He stresses the importance of “tuning” the model, as a real-time system, and describes particularly fascinating developments using gaming engines such as Quest 3D. Working with New Zealand-based developers Massive, Arup have taken their agent-based crowd simulation system used in films and games and added behaviour to simulate crowd movement in emergencies to create their new system Mass Motion.

Other human behaviour in spaces can be added with increasing accuracy, and then exported to Quest 3D such that the results of the simulations are displayed in an environment that can be explored interactively. Carfrae is also quick to acknowledge the limitations of models too, again responding that models are useful as part of a collaborative, critical process — as a tool within that, but not an end in itself, despite the creative and business potential.

Despite that, he’s excited by the possibilities and suggests that “it’s really hard to decide how deep this well is”, as this platform emerges as the overarching organising mechanism for all existing and future information around buildings and cities. (Key strategic questions as to ‘who owns the model’ are still being explored, as there are few clearly agreed views as to who this should be, who has liability for operating it, whether it’s leased, and so on.)

Within the lifecycle of the building, it becomes invaluable as part of the design process, can be used for maintenance and operations, and also provides instructions for the eventual deconstruction or adaptation and recycling of materials. It may even change some of the value chain around building, as BIM provides a way of articulating the ongoing development of the building after initial build. Traditionally, the costs are weighted towards the design and construction, and away from the ongoing operations. As the life of the building becomes more perceptible and more easily engaged with, the value of an ongoing contract could increase, right through to the newly efficient and sustainable recycling process. With a perceived shift away from large, open, adaptable spaces (Lloyd’s, Pompidou for instance) towards more tightly integrated, highly bespoke buildings, with their corresponding reduction in capital cost, the potential benefit of this lifecycle, including deconstruction and recycling, may be significant in both business and environmental terms.

On measuring success and sustainability

Carfrae brings in the issue of measurement again here. While we might look at the Centre Pompidou and see its high operating costs over time — the cost of placing services on the exterior, in order to free up the internal spaces — Carfrae believes that a mundane measurement of their financial or environment cost is misplaced in both cases. A richer sense of what, say, the Centre Pompidou has brought to Paris, given its exceptionally high usage by citizens and tourists, would massively outweigh these perceived costs. As a spiritual, cultural and social enlivenment of the centre of Paris, it can barely be equalled. In a different way, the Sydney Opera House represents a highly bespoke building that, in some terms, doesn’t always function well. Yet its value to the city can barely be over-stated.

So a different response to measuring success is required, incorporating a far more sophisticated view of sustainability in particular. Sensing the current mood, Carfrae seems disappointed by what he calls the “false starting point that ‘no development is truly sustainable’”. It doesn’t account for what the building is truly worth, in terms of social development, say, or what its beauty might be worth. Particular buildings might be knocked for the amount of steel consumed but with that limited viewpoint most sculpture on the planet might be condemned in the same way. (Carfrae mentions the Green Star rating tool as a good development in terms of a more sophisticated measurement, as it begins to incorporate health and well-being measures alongside energy consumption.)

I offer up Bill Addis’s statement that there may be two types of building emerging. Addis describes “one type like the automobile: highly engineered, lightweight, mass-produced, cheap, and unable to be adapted for other uses. The other, designed to be capable of adaptation for unimagined, even unimaginable uses, will — like the castles, cathedrals, civic buildings, and skyscrapers of eras past — be built to last, creating in turn a living legacy for engineers of the future.” (Bill Addis, Building: 3000 years of design, engineering and construction, Phaidon 2007). Carfrae instead finds a third kind of building here. He suggests that castles and cathedrals aren’t that adaptable at all, but are instead such “beautiful, strong, poetic or lyrical” buildings that people adapt to them, and want to make a use for them. So while the efficiency of the tightly-integrated and recyclable building derived from the adapted techniques drawn from the automobile and product design industries may be more effective than the open, adaptable space, neither take into the account this richer understanding of beauty or intangible value.

He describes a new JetBlue terminal for JFK that Arup are working on, in which they have to work around Eero Saarinen’s iconic TWA terminal. Essentially defunct, the value of the latter is still so significant that it can’t — and shouldn’t — be destroyed. This is not a conservation issue though, but one of the wider value of good building. Simplistic measurement underestimates this, whereas a wider perspective forces engineers to “raise their game”.

Carfrae believes that creating sustainable buildings — by greater efficiency “at all stages from construction through adaptation and reconfiguration to demolition and recycling” — must surely be a goal when we consume more than the planet can provide. This is partly enabled by BIM and allied techniques as never before. But this richer understanding of the true value of buildings must also be understood. These intangible aspects are incredibly difficult to measure or contain, and a process aligned too closely around modelling may miss them. This is akin to trying to measure a city — measuring traffic flow, pollution, and buildings, say, won’t come close to measuring the value of, or feeling of, a city.

Sidenote: Carfrae mentions the work of John Frazer at QUT. He’s developed city models, based around variables across only 3 layers — physical, social, economic — which can accurately model the development of urban form e.g. rolling a model of Gröningen forward, from 1500, you can end up with a good approximation of the contemporary Gröningen. I’ve suggested here several times that creating City Informational Models — aggregated Building Information Models, perhaps — that provide a platform for measuring previously intangible behavioural information could provide increasingly rich ways of measuring the value or feel of a city — these are not predictive as such, but instead based around the actual performance of buildings within the city.

The new structural engineering

The particular inspiration point in the development of the Beijing National Aquatics Centre project was derived from building on previous best practice — reversing the traditional engineer mindset to reinvent the wheel — and used the open knowledge-sharing environment of the web, rather than, say, Arup’s own internal knowledge management systems, instead immersed in a world of disparate and wide-ranging data sources. Further, it was tested using algorithmic optimisation allied to informed personal judgements, using purpose-built tools, as well as rapid prototyping and building information modelling. Collaboration brings in further judgement points, derived from aesthetic, social, environmental and even historical cultural influence. In turn, the ‘Water Cube’ provides data, knowledge and inspiration for future projects.

As such, this rapid, iterative, collaborative, open and outward-facing development exemplifies the new structural engineering Tristram Carfrae is so excited by.


First published on March 31st 2008 at cityofsound.com, where you’ll find more of this kind of thing.