As long as you don't want to develop a perpetual-motion machine, there is almost nothing that is technically impossible today. Limits come from two interrelated factors: can you afford to do it? And can you do it in the time available? Since time so often equates to money, this is really one question rather than two.
And it is a question that engineer Whitbybird is keen to tackle. The practice has grown rapidly during the past few years, an indication that it knows not only how to find interesting solutions to difficult questions but also that it understands how to make money. Merely working on award-winning projects is no guarantee of success - the work can be out of all proportion to the fees involved.
The practices that win the prizes can find themselves struggling financially or yearning for some bread-and-butter work to get back on an even keel.
So members of the practice are understandably excited to have found a technique that they believe will allow them to be more imaginative on the difficult projects, and to explore ideas and complexity on the more modest projects in a way that was previously only available on the blockbusters.
The technique comes courtesy of two software specialists with their origins not in construction but in product design. Ian Dulley and James Turner have together formed Helium 3 and are using their knowledge to address problems within architecture in ways that have not been done before.
Whitbybird co-founder Mark Whitby is enthusiastic. 'This is beginning to allow us to develop smaller-scale solutions that were previously only available on a larger scale, ' he says. 'And in a manner that can be predicted, and that we can analyse.' Mike Hitchens of Whitbybird stumbled across the technique by accident. The engineer was working with architect Design Engine on a project that involved a small-scale timber gridshell, and Design Engine was working with Helium 3. 'They had a way of solving three-dimensional problems that blew me away, ' says Hitchens.
Underpinning the technique is a method of modelling the surface of a structure and, from this, generating a geometry that can then be modified easily without the need to go back to basics. This means that the design team can generate the model relatively early in the process, and can make numerous changes without having to go back to the beginning and do lots of calculations all over again. It works by using spatial geometry to link all the components in a building intelligently. From there you make your adjustments to the shape, and the drawing automatically regenerates.
Only at the very end do you have to print out your drawings.
On this original gridshell, the engineer was able to develop a standard node with Helium 3. This node, which allowed variations of angle in connection, could then remain, however the geometry of the gridshell varied.
The team had created a reproducible, manufacturable component that allowed constantly varying geometry in the structure itself - exactly the combination of repetition and flexibility that is needed if affordable, imaginative design is to be a reality.
For Whitbybird this was a great discovery because it knew how much time working on non-standard structures commonly involved. 'We can do complex 3D design, ' says Hitchens, 'but we are always looking for ways to do things better. It is really exciting when something new comes along.' Recent challenges included the transfer structure on the BBC building with MacCormac Jamieson Pritchard which, says Hitchens, 'took a lot of grunt', and the much-criticised design that Foster did for Selfridges, which seems to have quietly died. This was also a hugely complex structure that had to be modelled with lasers. The Helium 3 solution, in contrast, generates relatively small files that can be re-run easily.
So when the practice was appointed as the structural engineer on the competition-winning design by Snøhetta/Spence for Turner Contemporary, it turned gratefully to the boffins at Helium 3 for assistance. It knew that it had taken on a really difficult project and that it needed all the help it could get.
Turner Contemporary is a £15 million gallery in Margate, birthplace of JMW Turner. It is intended both as a gallery for work by Turner and other artists of his time and ours, and as a catalyst for regeneration of the rundown seaside town.
Snøhetta/Spence's solution will certainly make a splash. The gallery sits in the sea itself, at a very exposed point on England's east coast, and is linked to a smaller entrance building on the Grade II-listed pier by a 12m footbridge. A rounded, organic form, the gallery existed originally only as a handcrafted sanded timber model.
Whitbybird, appointed after the architect won the competition with a different engineer, faced a number of unusual challenges. It had to ensure that the building not only stood up and kept the rain out - the usual constraints - but also that it kept out the sea and that neither all nor part of it would be washed away in storms.
One of the first moves was to persuade the architect to change from its original choice of materials to steel, so that shipbuilding techniques could be used. After all, ship designers and builders know how to deal with exactly the conditions that Turner Contemporary will encounter in the worst of the winter storms. Like a steel ship permanently at anchor, the gallery will consist of two steel welded skins, separated by curved steel ribs.
They designed a set of stepped concrete-piled foundations that will sit comfortably beneath the stepped underside of the building, which will be floated in from the shipyard, rotated, and bolted into place on its concrete seat during the six hours of low tide.
So, problem solved. Well, not quite.
Just a few questions remained. How to describe the structure so that the fabricators could assemble it? Indeed, how to describe the structure at all?
And, of equal importance, how to describe it so that the architect could change and develop it? After all, this was a competition-winning design, and we all know that such designs are far from complete. They may have a brilliant vision, but start working for real with the nitty-gritty of the project and there is bound to be a need for several modifications.
Originally, the architect had developed a shrink-wrapped computer model of its initial physical model in FormZ. This was a very large model, hungry for computer power and, therefore, discouraging experimentation and change. Whitbybird then tried modelling it in Bentley TriForma but found that it was, according to Hitchens, 'not intelligent', and that making changes took too long.
Then Helium 3 rode to the rescue - or at least was called in. It simplified the original FormZ model by remodelling it, first tracing the original FormZ model with a new intelligent curved profile. From this it was possible to define a standard curved rib, forming elliptical hoops, which changes automatically as the external shape changes.
This meant both architect and engineer were able to make changes, refining the shape of the window, reflecting the internal requirements, and making it more streamlined and wave-resistant. Changes happened rapidly and the results could then be exported for analysis. 'You can very easily convert the information into TriForma, ' says Hitchens. 'Then the normal process takes over and you churn out the 2D drawings.' Hitchens found the process to be much quicker and far less demanding of staff. Whereas in the past an operation may have required the attention of four engineers, now it can be completed with just one engineering technician.
This process is also valuable because it can generate DWG and DGN drawings in a manner that the fabricators can understand easily.
Turner Contemporary is due to open in 2007. When it does, some of the refinement at least will be thanks to the work with Helium 3.
Whitbybird is not just a successful engineer; it is also fiercely independent and proud of its own skills, so that when, for example, there was a rash of pedestrian bridges with architects in the driving seat of the design, Whitbybird remained aloof from these collaborations. Instead it insisted that an engineering practice could design such projects on its own (albeit with the occasional architect in its employ) and proved it by going on to win many projects and awards.
So this degree of reliance on an external supplier like Helium 3 seems surprising. Why doesn't the engineer just want to absorb the technology and run with it itself? First, explains Hitchens, because it is just so good.
'It's not that Helium 3 has come up with something that nobody else had considered - just that it has got it right.
'We have been talking to Bentley about parametric modelling for two years, ' explained Hitchens. 'These guys from another industry have got it and have used it. We are in our infancy.' And learning is not a rapid process. 'We have found that taking on the software and learning it is easy, ' says Hitchens, 'but learning how to use it on real projects takes about 18 months.' By which time, doubtless, Helium 3 will have advanced its technology further, and will be as indispensable to the engineer then as it is today.