top of page

from section iii

Working and Learning


SB1-061 Study for Smithsonian Roof Foste


Chapter 1: Built Work

I found that very many architects in London were delighted that BuroHappold was just around the corner and keen to work with them. It made it easy to spend time with them, to get to know them, and to find out how we could help. I could even call in to see them daily if necessary. I loved this immediacy. With so many architects to talk to, the conversations soon became design focused, and sketching was the essential way to communicate.


A fantastic opportunity materialised when an architecture and engineering competition was announced for a new roof for the main courtyard at the heart of the Smithsonian American Art Museum in Washington, DC. As I had already spent time with the team at Foster + Partners on the Great Court Project at the British Museum, I was approached by the architects to participate in this competition with them. This encounter led me to discover that Foster + Partners were interested in telling engineering stories with every project, and this new project promised to do that again.


Smithsonian Museum, Washington, DC

The Smithsonian Foundation had already started to renovate the Old Patent Building in Central Washington, DC, and were seeking to create a great home for the National Portrait Gallery and American Art Museum. With a plan for a new roof over the central courtyard, the clients decided to find an architect and engineering team to design this part of the transformation.

We went to D.C. to meet the client and see the space. It was interesting to see how the roof could help keep direct light away from the exhibits inside while encouraging good daylight in the courtyard. 

I was keen to find out if we could do what we had done at the British Museum where we allowed the new roof to rest on the courtyard’s existing walls. It seemed much easier than adding new columns. But, the clients were not enthusiastic—a real disappointment. They were so convinced that new columns would be needed that they had already constructed foundations in eight places around the edges of the courtyard in anticipation. Incorporating the preconceived eight new support columns would be a major design constraint. One of the Smithsonian Foundation members took me down in the basement where excavations had already started. He showed me the foundations of the existing walls. Until then I hadn’t realised that the whole of Washington is on reclaimed swampland. The ground is just sand, and the walls sit directly onto it. I knew there and then that I would need to think again. My hopes for a column free roof wasn’t going to fly. I couldn’t take the new roof onto these walls; we would need columns and piled foundations!

My early instincts were to see what a hanging chain catenary model might do to inform a suitable shape for the new roof. With the eight support positions, this would lead to a central, dominant dome with two side domes. We tried a central catenary form with leaning columns surrounded by a subsidiary doughnut shell. The shell was reminiscent of the British Museum’s roof. But it didn’t look great, and no one liked it. I remember us first doodling in a coffee shop opposite the site. We came away still feeling there was something in the idea, but it wasn’t looking right. 

Back in London we came together regularly during the competition period. We didn’t have much time to invent the perfect solution. Our competitors would be trying hard to win. We explored flat roof options and structural grids that could support a glass roof. This was a very simple approach and potentially elegant. Yet, it wasn’t very efficient because there would be no shell action like we achieved for the British Museum roof. We continued, this time trying a gentle ‘blister’ with a di-grid roof structure, rather like the new Elephant House at the Copenhagen zoo. This gained a little benefit from the curvature but still had deep structural members. We considered using these members for shading, adding louvres to the outside. Then we tried orienting these to capture light in the courtyard. It was getting quite interesting despite the eight columns that were given positions without any rationale. We had plenty of ideas but nothing that really made the most of these eight column positions we needed to use. 

We wanted to create a shape that would be structurally efficient and beautiful, based on the eight support points already defined for us.

With just a week before the competition deadline, I remember very well a design review with Lord Foster at the architects’ office where I was asked, what does the engineering want to do? What shape of roof would be most efficient? Would this give us a better design? Such questions drove the design to the final form in the submission. It was a roof form with three free-form shells that focused the loads down into the eight columns. This would make sense of the column positions. In doing this we would gain a degree of shell action, enabling us to reduce the depth of the structural members. Interestingly they were very much like the ideas we sketched back in Washington but had become more elegant.

Our plan gave us a very strong story to tell the client. Unfortunately, we only had a week left to draw it up, so the whole team was hard at it—day and night. We were really pleased; we had a clear and compelling story. The architectural, structural, environmental, and acoustical designs were all as one. Everything was working together.

We won the job—because we had such a compelling design where the architecture and engineering design came together in one elegant and natural resolution.

A crucial design moment sticks in my mind, when things could have gone so very wrong. As engineers we knew we needed to make sure that the thrusts from the roof shell would not push the columns apart. As we had a two-way grillage, rather than the three-way grillage of the British Museum roof, the shell was not fully triangulated. Squares have a tendency to lozenge, meaning they could turn from square-shaped to diamond-shaped. We put in tie rods between column heads to ensure this wouldn’t happen, but at this time the rods weren’t on the architect’s renders. I had to tell Lord Foster that the extra elements would be there. I was worried how he would respond, yet I needn’t have been. I made it clear they were doing a really important structural job. He immediately accepted them as part of the engineering story that would be absorbed into the architecture. It was a great relief but also a great sign of the benefits of collaborating with architects who appreciate the value of engineering honesty and its value to the whole story.

My wife has a different memory of that night, a disco dance with the senior funder and, after a night in high heels, a cold and barefooted winter night’s walk back to the hotel. But that’s her story to tell. I was glad she enjoyed the opening night because we had cancelled our holiday in Barcelona two years prior so I could attend the original competition interview!

SB1-103 Study for Smithsonian Roof Foste


Khan Shatyr, Astana, Kazakhstan

We were asked to join Foster + Partners for the design of a giant roof to cover what was going to be an entertainment centre in Astana.

For so many years the right solutions had been shells, like those at the British Museum, Sage, Smithsonian, and the Elephant House; lattices in steel that make efficient roofs across spans of 30 to 50 metres. But this building in Astana would be much bigger, and the site didn’t need us to work with other existing buildings. We had free rein. It had to make a strong statement as a gift from the president to the people. And it was going to be at the end of the main planning axis in Astana, where the President’s Palace and the Palace of Peace and Reconciliation were located.

It seemed that we should be ambitious. We began to think about a giant, single mast cable net structure. It would be a concave cone, using a single mast with an array of radial cables coming down from the top. This was a classic form of tension structure that I was familiar  with  from observing Frei Otto’s projects, like his own Institute in Stuttgart that I had visited in the seventies. I knew that this form would be really efficient—a three-dimensional version of a suspension bridge, well known for the way they can span across vast stretches of rivers and seas. With its single mast sitting on the city’s main axis, it would look magnificent.

We could reach this conclusion within just hours of talking through the options. It was such a strong image. We had to go for it. But this was bigger than anything I had ever seen. In fact, it would be the biggest single mast tent in the world. The enormity of our task ahead began to sink in, with emerging questions: How would we erect something so massive? Was it really practical to construct such a thing in Astana? How would snow build up on it and would icicles form and pose a threat? How would the space inside the roof feel—would it be comfortable enough? 

Thinking about comfort, we sensed that the hot air from people would rise up and fall down in addition to an opposing cold draft, so it would create its own weather system inside—clouds and rain storms in the afternoon perhaps! 

Most of the material needed wouldn’t be available in Astana. How practical would it be to take all the material across Asia to reach the site? Had we come up with something so ambitious that it wouldn’t be possible to build it?

I recall the frantic thinking and sketching. How could we split this into phases of construction? How could we stabilise the cable net against uplift without adding weight to the radial cables? How could we divide the space into sections with different climates and better control? How should we communicate all these ideas to our client?

Sketches were rushed out and exchanged between the team, sometimes on the airplane travelling to and from client meetings, as well as in the meetings themselves. These meetings were followed by models made days later in the office. The engineering models were crucial to show that with a tension cable-net only a limited number of shapes were viable. And the models were crucial to explain different ways we could configure and build the tent efficiently to the client. Having to think about issues and ways to communicate everything quickly, I created a lot of doodles. It was the way to create something fresh—to take the client with us as design ideas developed.

After a few weeks developing the design I could see a problem ahead. With the vertical mast and radial cables coming down from the mast head to the ground, we needed hoop, or circumferential, cables running parallel with the ground, pulling in the radial cables. A bit like straps around a barrel, they allow you to put stress into the radial cables. Unless you pre-stress the cables you will not have a stable roof. However, we were starting to realise that these hoop cables were not very effective in the lower half of the roof, and the radial cables were not very well restrained against wind uplift. The hoop cables would be more effective if we could fix them to the ground at both ends. We tried some versions where this could happen, but it meant splitting the roof into three or four different sections, making it rather more complicated and less of a pure single form.

I was starting to think the cone tent might not be such a great idea, certainly at this scale.

But there was another possibility. If we leaned the mast to one side, the hoop cables in the lower half of the roof could be fixed to the ground at their ends. This would make them far more effective at tying down the radial cables. We explored how this would look and realised it was far more exciting to see the mast at an angle. It also made better sense of the planning of the buildings inside the tent; the mast would lean over the parts where the accommodation was highest. The highest part of the roof would be over the highest part of the buildings. So, a good engineering solution was also a good architectural solution.

The specialist, Selami, who joined us, had a rather braver view of how to erect this gigantic tripod that I hadn’t seen coming! Rather than doing it piece-by-piece like building a tower, he wanted to assemble everything on the ground. Only once the tower was all assembled would they pull it up into position. This would happen in a matter of hours. I gulped at the enormity of the task. But it was what he did, and it was probably safer, and certainly quicker, than my way. This was very dramatic to watch. I was delighted that the erection of the mast—probably my biggest worry in the design—went so smoothly. This gives another example to show how you need to think about how to build before you define what to build, and talk to the experts.

We were invited to the opening ceremony. It was a very memorable event. On his birthday, the president was giving the people his present at an impressive ceremony. The famous Kazakh horse riders were much in evidence, and their passion for fireworks was indulged to the fullest.

I remember walking up the steps and through the main entrance. What I saw was truly impressive—as I had hoped. What stays in my memory most are the faces of the families, especially of the children, as they saw it all for the first time. There were lots of wide eyes of course, but there was also an expression of wonder as they saw something strangely familiar, like a yurt, yet extraordinary in scale and dramatic appeal . . .

Khan Shatyr Astana_Credit Nigel Young_Fo
bottom of page