arup journal 1-2009 beijing stadium

7/30/2019 Arup Journal 1-2009 beijing stadium

1/52

1/THE BEIJING NATIONAL STADIUM SPECIAL ISSUE

The Arup Journal


2/52

2 The Arup Journal 1/2009

Contents 4 Introduction

Stephen Burrows

5 Competition, team, and site

Tony Choi Michael Kwok

Sports architecture

8 The architectural

design concept

J Parrish

The main roof

16 The Stadium geometry

Stephen Burrows

Martin Simpson

20 Analysis model and results

Kylie Lam Thomas Lam

24 Seismic design of the roof

Xiaonian Duan Goman Ho

28 The retractable roof design

John Lyle

The bowl

32 Layout and analysis model

Tony Choi Thomas Lam

34 Seismic design of the bowl


Specialist engineering design

36 Roof cladding and

acoustic ceiling

Tony Choi

37 Wind conditions in the

Stadium and external plaza

Alex To

38 Thermal comfort in the Stad

Rumin Yin

40 Fire engineering concepts

Mingchun Luo

41 Building services design

Lewis Shiu

44 The lighting concept designJeff Shaw

Rogier van der Heide

Construction and conclusio

48 Completing the programme


49 Constructing a stadium

50 Credits


3/52

The Arup Journal 1/

The Beijing

National Stadium

Known universally as the Birds Nest, the 91 000-sea

National Stadium was conceived and built as the

primary venue for the XXIX Olympiad, held in Beijing i

August 2008. This special edition of The Arup Journal

documents the sports architecture design and the full

engineering design by Arup over the six and a half yea

from initial concept to project delivery.


4/52


Cantilever structures or the roo will be virtually impossible to build or spans o

approximately 60m with the additional loading o the removable roo.

These principles, agreed at the beginning, were important rst steps in our de

and set in place rm oundations or what ollowed. The very rst sketch o the ro

emerged some weeks later (Fig 2): this was our starting point or the Birds Nest

design. The competition was won in April 2003 and so began the process odelivering one o the worlds greatest buildings.

But the e-mail trail doesnt tell the whole story. In Basle we worked days and n

to nd a cultural clue to the design that would win such a competition. The mode

building went on day and night too. We had un, we still tell the stories, and we

utilised Arups power wherever the skills lay to put the best people onto the proje

My recollection o the entire process, rom the initial idea o a consortium to th

integrated working o teams rom Herzog & de Meuron, CADG (China Architectur

Design & Research Group, the Local Design Institute partner) and ArupSport, wa

o a smooth and harmonious development. We had a single aim to win and w

ocused on how to achieve that. So it didnt matter that ArupSport determined th

unctional geometry, our ideas or the roo carried weight alongside those o othe

we agonised over the scale o the spans and the scale o the project, we constan

had a better idea (and some were actually quite good, though many were not),

arguments were ew, and dinners were lively aairs.

I remember, when we won, Michael Kwok calling me Steve, we won! and

a moment I had to think what he meant. Then the reality hit home, the calls bega

and the opportunity to shape a piece o history grew to enormity.

For Arup the schematic design stage was carried out in Europe. Manchester a

London were the core oces, and many people played their part. We have tried t

credit everyone who made a signicant contribution (see p50) but some have m

on to pastures new.

However, all o us have a shared experience; all o us will have watched the 20

Olympic Games with a sense o shared pride, wherever we were; all o us know o

contribution to the project and its important contribution to Arups goal to shape

better world.

This is no overstatement. The Olympic Games is a global event, the decision t

hold it in China was a pivotal political moment, and the Stadium will long remain asymbol o that decision, an important part o an important moment in history and

symbol o the power o positive thought and action by the Peoples Republic o C

I am proud o what we achieved, and I am also in awe o the skill and dedicati

our sta, o the ease with which Arup worked across geographic boundaries, o t

incredible perormances o our collaborators, and not least o the builder o this

wonderul piece o engineering architecture.

As we say in the North o England, twas a bloody great eort!

In January 2003, alongside 13 competitor rms rom

all over the world, Arup began work on the design

competition or the Beijing National Stadium.

In writing this introduction to The Arup Journal

eature on this great project, I looked back to the rst

meeting notes rom 10 January 2003, when J Parrish

and I met with the architects Herzog & de Meuron at

their oce in Basle, Switzerland. These struck a

chord with me as I recalled how we interpreted the

brie and how it would infuence our design.

To quote these notes: Bowl shape design will be

carried out essentially by ArupSport throughout the

competition works, HdeM will incorporate these and

co-ordinate with other areas o building. It is perhaps

possible or the running track to be completely

covered by the roo, ArupSport to check with IOC.

The track cannot however be only partly covered as

this will induce uneven conditions on dierent lanes.

Introduction

Stephen Burrows

2. Initial design sketch for the roof.

Structuraldepth oftrusses

Bottom mat

Top mat

1. Typical elevation of the structures exterior.


5/52

The Arup Journal 1/

NationalAquatics

Centre

NationalIndoor

Stadium

NationalStadium

Sports CentreGymnasium

EthnicCulture

Park

OlympicsVillage

ChineseScience &

TechnologyMuseum

BeijingOlympics

Park

OlympicGreen

Fencing Hall

North 4th Ring Road

The design competition

Selecting the winning scheme for the National Stadium also involved the citizens ofBeijing. All 13 competition schemes were displayed at the Beijing Exhibition Centre in

March 2003, attracting thousands of visitors, and alongside the deliberations of the

international jury panel, votes by the general public were also taken into account.

By the end of March 2003, it was announced that the Birds Nest scheme was

selected as the winner, both by the jury panel and by public voting.

However, the route from winning the design competition to winning the contract

as designer for the implementation of the project was not a simple journey.

In parallel with the design competition, the Beijing Development Planning

Commission (BDPC) called for an ownership tender for the National Stadium.

The bid winner was to join the Beijing state-owned Assets Management Corporation

(BSAM) to form the project company, which would be responsible for the investment,

construction, operation, and transfer of the project. A consortium led by CITIC was

selected as the successful bidder, and duly joined with BSAM to form the project

company National Stadium Co Ltd, which became Arups client for the project.

Negotiation of the design contract between the design consortium (Herzog &

De Meuron, Arup, and CADG) and the client started in July 2003, and the tough

commercial negotiation took more than four months to conclude with contract signed

in early December 2003. Concurrently, the schematic design was progressed at fast

pace, with the Birds Nest groundbreaking ceremony held on 24 December 2003.

The Arup team

Arups success in delivering the project was truly a result of team effort and global

collaboration, with everyone working seamlessly as one Arup. The ArupSport teams

in London and in Manchester, the teams in Beijing, Hong Kong and Shenzhen, and

the London Advanced Technology and Lighting groups, all gave of their very best.

Within weeks of commencing the schematic design stage, engineers from Beijing and

Hong Kong were assigned to the Manchester office to work with the team there.At the same time, another team was mobilised in the Beijing office to liaise and

co-ordinate closely with the client, with CADG, and with local authorities. During the

preliminary design stage, some UK members stayed in Beijing to work with the team

at critical stages to ensure smooth implementation. Arups ability to mobilise global

expertise and deliver locally was key to the success of the project.

Arups scope of service covered sports architecture and all engineering disciplines

including structural, mechanical, electrical, public health, wind, fire, and seismic

engineering, environmental and microclimate studies, acoustics, and lighting design.

Arup global expertise was deployed to achieve a world-class, state-of-the-art design.

The firm was responsible for schematic design and preliminary design for the above

scope, whilst CADG was responsible for construction documentation.

Site profile

The National Stadium is located in the southern part of the Olympic Green, which was

masterplanned by Sasaki Associates and covers an area of 1135ha on the north side

of Beijing, close to the citys central axis (Fig 1). The Stadium is the centrepiece venue

of the Olympic Green, on an irregular quadrangle approximately 20.4ha in extent

(Fig 2). The terrain is relatively flat, with ground elevations ranging from 42m to 47m,

highest at the south-west corner and lowest at the north-east corner. The position

was chosen so that there would be a gradual rise in level from the city roads in the

north-east, forming a gentle slope up to the Stadium plinth, about 5.3m higher.

The plinth connects to the main concourse, level 1 of the Stadium.

Competition, team, and site


2. Key plan.

3. Artists impression of Olympic site.

2. Key plan.3. Site plan.

National Stadium

AiOther venues

1. The Olympic Green relative to the city of Beijing.


6/52


3. The Stadium site.


7/52

The Arup Journal 1/

Sports architectur


8/52


Introduction

At the time the architectural competition or the

Beijing National Stadium was announced, Herzog &

de Meuron and ArupSport (Arups multidisciplinary

practice specialising in sports architecture) were

already working together on the Allianz Arena in

Munich1. This successul creative partnership was

based on a shared desire to innovate: Herzog & de

Meuron in creating unique buildings with strong local

cultural resonances, and Arup in designing stadiums

that perorm ever better or spectators, athletes, and

operators. As already noted, or the Beijing

competition the two practices joined orces with one

o the leading Chinese Design Institutes, CADG.

Within this integrated team, the architects at

ArupSport were responsible in particular or the bowl,

the concourses, and the spectator acilities, which

together dened the orm o the Stadium. They also

produced an initial optimised structural proposal or

the roo and envelope, which Herzog & de Meuron

then developed. CADG provided vital local expertise

during the competition and scheme design, and then

took the baton or the nal stages o the project,

liaising with the local authorities, producing

construction inormation and monitoring the workson site. Backed by Arups engineering expertise,

the competition team was able to submit a highly

developed, ully realisable architectural concept.

As a result, despite some signicant changes to the

brie, the orm o the built Stadium is very close to

the original winning design.

The architectural

design concept

I was delighted that the competition areas did

everything that was set out for them to do. The path

from drop-off for athletes to the warm-up area, with

access to the Technical Information Centre for Team

staff along the route; the fact that there were separate

corridors to make sure that athletes making their way

from the Call Room to the track could do so securely

and without being disturbed by other athletes or

coaches, without minimising space for others prepari

themselves, and the space provided for athletes and

staff to move around, made it the ideal stadium for th

Paralympics Games. The fact that all of these spaces

were absolutely accessible for athletes and staff usinwheelchairs made it a delight to use. Added to this, th

fact that spectator areas provided enough good acce

for those using wheelchairs was superb.

Chris Cohen: Chairman o IPC Athletics.

The brie called or a landmark building that would be the main venue or track an

eld events during the 2008 Beijing Olympics, with a subsequent working lie o 1

years. Ater the Games, it would become an important venue or both athletics an

soccer. The Stadium was to have a capacity o 100 000 during the Games, and

80 000 seats in legacy mode. (The client subsequently decided to reduce the Olycapacity to 91 000.) There was no dened legacy business plan, and so the desig

team tried to make the Stadium as fexible and adaptable as possible. There is

potential, or example, to add a hotel or box holders within the main envelope.

Originally the Stadium was to have a retractable roo (Fig 1). This was particula

challenging in structural terms as the building also had to have the resilience to

withstand a major earthquake. Late in the programme, the client omitted this

requirement rom the brie as part o the general review o the Olympic venues,

beore work started on site.

J Parrish

1. Original design with retractable roof.

2.


9/52

The Arup Journal 1/

Bowl design involves a skilul balancing o several

key criteria. Most importantly, spectators want to

be as close as possible to the action and to have

good view o the eld, while the stadium develope

needs to accommodate a certain number o seats

within a dened budget.

These requirements oten confict. For example

more space between rows creates better sightline

but draws spectators urther away rom the eld a

results in a larger stadium with increased construc

costs. Even a tiny adjustment to the conguration

the seats can have a huge impact on the overall

design and cost o the building. To nd the optimu

solution, it is essential to set priorities.

The bowl

The architects ambition was to create not only an instantly recognisable symbol o

Chinas cultural, sporting, and economic renaissance, but also the most exciting

stadium in Olympic history. Every Games has its own thrilling I was there moments,

when athletes perorm miracles and new records are set. The team wanted to create

a stadium that would harness and ampliy this excitement in the way the worlds

best-loved soccer venues do.

Like most modern stadia, the Birds Nest was designed inside out, beginning

with the bowl the competitive eld and the seating stands around it (Fig 4). This is

because the orm o the bowl and the distribution o seating types largely determine

all other aspects o a stadium, including the shape and structure o the roo, the levelsand locations o the concourses and premium acilities, and the amount o natural

light and ventilation reaching the playing area. The team worked closely with the

international Olympic and local organising committees to streamline and rationalise

the on-eld acilities. The result is a more compact bowl with less distance between

the spectators and the track.

3.An I was there moment.

4. Like most modern stadia, the Birds Nest was designed inside out, beginning with the bowl.


10/52


This complex process has been transormed in recent years by parametric

relationship modelling. Using powerul computer sotware, designers can quickly

generate the initial orm o a stadium within defned parameters such as geometri

constraints, environmental actors, and the limitations o construction materials.

Having produced the initial concept, the architect can rapidly explore and test

options by adjusting variables such as the height o a row o seats. For the Nation

Stadium, ArupSport used its own specialist parametric modelling sotware to dev

a bowl geometry optimised or Olympic athletics that would also work well or soc

in legacy mode. The team produced 33 versions o the design to fne-tune the o

o the bowl (Fig 5).

The team decided that this landmark Stadium should have the same distinctiv

external orm in both Olympic and legacy modes, and so the temporary additiona

seating needed to be accommodated within the main envelope. The temporary s

which are mainly to the rear o the top tier (Fig 6), have the least-avourable views

the Stadium and are located in zones that can be converted to other revenue-

generating uses.

Creating a stadium that will be both an athletics and a soccer venue is always

a challenge. Athletics felds are bigger than ootball pitches, which means that

spectators in the stands are urther away rom the action. Consequently, people i

upper tiers may not be able to see the ball on the pitch, and the atmosphere wh

is so important to a soccer crowd may be seriously diluted. One solution to thisproblem is to add a moveable lower seating tier or soccer matches, but the brie

the National Stadium did not allow or this. Instead, the team opted or a cantileve

middle tier, with the ront 15 rows o seating extending over the lower tier (Fig 7).

This brings spectators in the middle and upper levels closer to the action and pro

a quality o view equivalent to that in a stadium with a moveable tier. The colour o

seats ranges rom red in the lower tier to white at the top, helping to make the

Stadium look ull, even when some places are empty (Fig 8).

5. Parametric design of built version 33.

8. The colour of the seats randomly merge from red to white.

a)

7. Fifteen-row cantilever of middle tier over lower tier.

6. Initial seating capacity of 100 000.

b)


11/52

The Arup Journal 1/

The team members had to design a stadium that conormed to rigorous local seismic

codes, while providing a structure stable enough to support a moving roo. To meet

these two key elements o the brie, they decided at an early stage to keep the bowl

structurally separate rom the aade/roo structure. The bowl consists o six

structurally-independent segments with 200mm wide movement joints between them.

The continuously-curved orm o the seating tiers provides better viewing standards

or all spectators with lateral views as well as an enhanced C value (the quality o a

spectators view over the row in ront) or VIP and premium seats (Fig 9).

The elliptical orm o the bowl, the depth o its structure, the acoustic refectivity o

its envelope, and a special lining below the ETFE (ethyltetrafuoroethylene) roo

membranes, all give the Stadium an outstanding acoustic quality (Fig 10). During the

Olympics, many visitors were surprised and delighted by the atmosphere o intense

excitement and drama.

The faade/roof structure

While Arup was working on the bowl, Herzog & de Meuron began gathering ideas or

the external orm o the Stadium. The team members knew that to win this prestigious

architectural competition, they would need to come up with an inimitable design that

would refect both Chinas rich cultural heritage and its 21st century technological

prowess. The distinctive roo structure does just that. Its appearance, inspired by

local crackle-glazed pottery and veined scholar stones, dees structural logic.It is an amazing display o architectural, engineering and construction innovation.

Local people aectionately nicknamed the Stadium the Birds Nest while the initial

competition entries were on display in Beijing.

The roo structure spans a 313m x 266m space, closely enveloping the bowl and

concourses to orm both aade and roo. The aade incorporates the Stadiums

main staircases. The result is a compact and sinuous external orm uninterrupted by

masts, arches, or stair cores. While the aade is open, a roo covering made o

single-layer ETFE membranes stretched between the steelwork sections protects the

spectators rom wind and rain (Fig 11).

9. The continuously-curved orm o the seating tiers provides better viewing standards or all spectators .

10. The acoustic refectivity o its envelope and lining belothe roo membranes, all give the Stadium an outstandiacoustic quality.

11. The ETFE membrane.


12/52


12. Sections through the bowl.Top: north-south.Below: east-west.

Level -1 and mezzanine levels Level 0 Level 1 Leve13. Successive levels ofthe Stadium.


13/52

The Arup Journal 1/2

Level 4 Level 5Level 3 Levels 6 & 7 Key

Spectator

FOP and warm-up

Event manager

BroadcastMedia

Olympic amily

Sponsors

Venue operation

Special events

Security

Urban domain

Others

Athletes and team


14/52


Reference(1) BURROWS, S, et al. The Allianz Arena: A new ootball

stadium or Munich, Germany. The Arup Journal,41(1),

pp24-31, 1/2006.

We gratefully acknowledge the assistance of Felicity Parsons,

independent architectural writer based in London, in preparing

this article.

The bowl and external orm o the Stadium were

developed in parallel, with Herzog & de Meuron

working on the aade and roo while Arup defned

the size o the bowl and proposed an optimised roo

structure. The team agreed at an early stage to work

with 24 nodes or the primary roo structure support,

and Arup very quickly defned the top and bottom

roo planes required or the most efcient structure.

This provided Herzog & de Meuron with an envelope

orm that did not change signifcantly, even in the

projects fnal construction design stage.

The seemingly accidental arrangement o steel

members that orms the envelope makes it almost

impossible to distinguish between the primary

structural elements supporting the roo, the

secondary staircase structures, and the tertiary

elements that add to the random eect.

Each o the aades steel members retains a

1.2m wide external profle as it twists and bends to

ollow the saddle-shaped geometry o the Stadium.

The steel structure is painted light grey, contrasting

with the red-painted external concrete wall o thebowl, which is clearly visible through the aade.

This creates a variety o impressive eects,

particularly when lit at night.

Conclusion

With the lavish opening and closing ceremonies,

the thrill o broken records, and the tragedy o

shattered dreams, an Olympic Games is nothing i

not theatrical. The architectural team wanted the

audience to eel part o the Olympic spectacle rom

the moment o arrival. To enhance the sense o

drama, the team decided to leave the aade unclad,

allowing the staircases that orm part o the roo

structure to remain open. Weaving past each other

and oering clear views into every passing zone, they

ensure visitors have an unusual degree o interaction

with the building. The result is arguably one o the

worlds most exciting architectural experiences.

Importantly, the Stadium is also one o the most

comortable, usable and high-perormance sports

venues in the world. Arup has received an

unprecedented number o glowing testimonials rom

athletes (both Olympic and Paralympic), spectators,

the media, the organisers, and the operators.

Everyone loves the Birds Nest.

14. The structure, painting, and lighting create an impressive effect, especially at night.

15. The open structure offers clear v iews both beyond the building and into the zones, ensuhigh degree of interactivity within the Stadium.


15/52

The Arup Journal 1/

The main roo


16/52


Introduction

The overall shape and orm o the National Stadium directly responded to tworequirements o the initial project brie it had to have a moving roo, and it shoul

be designed to withstand seismic events twice the magnitude o the 1976 Great

Tangshan earthquake that killed more than a quarter o a million people in Beijing

This would not be the frst stadium with a moving roo to be constructed in a

seismic zone, nor would it be the frst or Arup (the frm engineered the 45 000-se

Toyota Stadium, Japan, and the 42 000-capacity Miller Park baseball stadium in

Milwaukee, USA1). It would, however, be the largest, with an initial capacity o ov

100 000 spectators.

In addition to the requirements o the brie, the team rom ArupSport and

Herzog & de Meuron also wanted to reduce the Stadiums visual mass and avoid

such structural solutions as masts and arches. So instead, the team opted to wra

the roo structure closely to the geometric constraints o the seating bowl and the

concourses (Figs 1, 2).

Having adopted a philosophy or the buildings orm, the next task was to crea

a structural solution that conormed to the requirements o brie, location, and

aesthetics. The answer lay in separating the roo structure rom the bowl structur

The ormer could be a complete entity with no movement joints, providing a stab

platorm or the moving roo and thereby greatly simpliying the mechanisation.

The bowl structure could also be simplifed, as there would be no signifcant inter

with the roo. The resulting bowl structure was ultimately realised as six complete

separate buildings each with its own stability system, and 200mm movement join

between each building.

Original inspiration

Though the Beijing National Stadium is oten reerred to as the Birds Nest, the

original inspiration was rom a combination o local Chinese art orms - the crackl

glazed pottery that is local to Beijing (Fig 3), and the heavily veined Chinese schostones2. However, when the artist Ai WeiWei3 frst saw the proposal he quickly d

a bird in a tree. The panelised approach gave way to infnite lines o structure and

name Birds Nest quickly became synonymous with the project.

The challenge or the team was to create a loadpath that was sympathetic to

the architectural intent but also robust enough to deal with both the vertical loads

resulting rom the large spans and the horizontal loads rom seismic events.

The solution was a system in which successive layers o structure are superimpo

This gives the appearance o a chaotic geometry (Figs 4, 5), but has the underlyi

logic required to resist loading.

Centreline geometry defnition

Most the geometry can be assigned to three categories:

Primary:Thiscomprisedthespacetrusslinesandthemainstructuralsystem

Secondary:Thiswasusedtobreakupthepanelsizecreatedbythemainstructural system to acilitate the cladding system panels.

Stairs:Theaccessstairstothetoptierofthebowlwereintegratedintothew

supporting the roo structure.

Firstly, the envelope was defned to wrap as closely as possible to the seating bo

taking the orm o an ellipse on plan with sloping walls and a torus orming the ro

suraces (Figs 6a-d).

The Stadium geometry

Stephen Burrows Martin Simpson

1. Optimum seating bowl confguration or Olympic mode.

2. The resulting enclosed volumethat would orm the roo surace.

3. Beijing crackle glazed pottery:the original inspiration or theStadium roo.

4 (above); 5 (below). The appearance o chaos.


17/52

The Arup Journal 1/

The geometry or the primary elements orms a relationship between the supporting

points at ground level and the size and shape o the opening roo position (Figs 7a-b).

Initially, this opening was defned as small as possible to keep the moving roo

efcient. When eventually the moving roo was removed rom the design, the size o

the opening could become much bigger and relate more to the seating bowl.

The primary geometry was then developed into a 3-D portalised space truss,

enabling the roo to ollow closely the architectural orm o the bowl and concourse

structure, while rising to 60m and spanning the required 313m x 266m (Fig 8).

The secondary geometry, subdividing the primary elements, was only located in

the outer layer o the aade. This geometry was related back to the primary roo grid

on plan, but then adjusted using the centre point to create a rotated plane instead o

a vertical plane (Fig 9). This plane was then struck through the outer surace to create

the actual secondary geometry used to defne the centre lines o the elements.

The fnal elements contributing to the overall geometry ormed the perimeter stairs

(Fig 10). These elements were defned initially by the requirements o the stairs in

terms o number o risers beore a balcony, length o balcony, and the overall pitch.

The defnition lines were then allowed to become continuous and run over the roo

surace to join the aade on the opposite side.

Though some scripting was required to create the initial geometry, the fnal

geometry required much manual intervention in moving elements and tweaking the

angles. In many ways the project is sculptural, and achieving the fnal eect relied on

a very close working relationship between engineer and architect.

9. Defnition o the secondary geometry.

10. (a) & (b) Stair element geometry.

8. The primary geometry as 3-D portalised space truss.

6. (a) Elliptical plan o bowl; (b) sloping sides; (c) roo ormed rom a toroid patch;(d) part o torus surace.

7. (a) & (b) Primary element geometry around opening position.

a)

c)

c)

a)

b)

d)

d)

b)

a)

a)

b)

b)


18/52


Twisted elements

O all the geometrical conditions within the Stadium, perhaps the most challenging

rom the abrication viewpoint was the requirement to use a continuous box-prole

over the whole aade.

This box section was dened using a control surace that was part o the structure

envelope. The outer fange o the box always remains parallel to the control surace,

resulting in a twisting, curving box section that changes as the element progresses

along the surace o the structure. This twisting orm is most pronounced at the eaves

o the structure or the low-angle elements such as the stair lines (Fig 11).

Luckily these are usually very lightly loaded.

The way the geometry was dened resulted in even the most twisted element

being ormed rom developable suraces. This meant that the individual suraces

orming the box sections could be fattened out and cut rom a fat steel plate and

then rolled to orm the abricated box section (Fig 12). This investigation was crucial to

proving that, though complex, the structure could actually be built.

Use of virtual prototyping

The use o CAD sotware was critical to success o the National Stadium, and the

platorm adopted was CATIA by Dassault Systme. It is used extensively in the

automotive and aerospace industries, and at the time was the only sotware that

could handle the complex suraces and geometry requirements o the elements.

A

Referencesurface

Centre line

of beamA

C

A

InsideOutside

C

AC

A

C

Referencesurface

Relative position of beam surface

is the same as reference surface

11. Detail o curved element at the eaves.

12. Twisted element, showing the our suraces fattened

13. Stadium models in CATIA, showing the roo (a) closeand (b) open.

10. Curved elements at the eaves.

a)

b)


19/52

The Arup Journal 1/

CATIAs ability to deal with a vast number o components allowed the whole Stadiu

to be assembled in a single environment (Fig 13). The model contained all the

structural elements, including the perimeter stairs, and the interactions between all

components were also managed in the same environment. This approach is called

virtual prototyping4 as all elements can be assembled and tested in a virtual

environment beore commitment to building the physical reality.

CATIA is a parametric component-based modelling package. The advantage o

using parametric sotware is signifcant when dealing with design that is required to

be adjustable and continually changing like the Stadium. The basic premise is that

instead o assigning rigid values to geometry such as length, angle, depth, etc, the

can be assigned parameters that can be adjusted later. Because the sotware is al

associative, relationships can be set between geometries that allow changes in

parameters to be propagated through the model and downstream implications o

changes assessed.

A simple example is the geometry o the stair line, which was controlled by an

angle at the level 5 landing. This angle changed the geometry o the stair so that al

the treads and landing could be hidden behind the supporting structure. However,

though the stairs terminated at the top level, they ormed part o a continuous line t

was rom fve separate parts but maintained tangency between each line (Fig 14).

Using a component modelling system also allowed multiple design scenarios to

investigated and then deployed throughout the structure. Even though the controll

geometry was dierent at each location, with the Stadium only having two-oldrotational symmetry, the details that components shared were generally part o a

amily. The advanced replication acilities with CATIA allowed these amily details to

propagated throughout the model even i the local geometry conditions were dier

Physical prototypes

At each stage o the project, the design team had to satisy itsel and the client tha

the structure was buildable. Early prototypes were constructed rom card, oamboa

or 3-D wax printers (Fig 15a). Herzog & de Meuron also built a ull-scale oam-boar

model to illustrate the scale o the elements being considered (Fig 15b).

Beore the end o the preliminary design, one o the steel abricators bidding or

project also completed a ull-scale mock-up o one the nodes rom 40mm steel pla

(Fig 16). This exercise showed the whole team that this was a realistic design that

could be abricated in time or the Olympics.

Final geometry

The original geometry changed late in the design process due to the omission o th

moving roo, due to the client needing to reduce the resources and overall cost o t

Games. It should be noted that the actual cost o the Stadium itsel was comparing

well to its original estimate, but the overall budget or the Games had to be cut.

However, due to the advanced sotware technique developed by the team in te

both o geometry and also analysis, design, and optimisation, the project was able

be completed on time with only a small delay in the construction programme.

References

(1) CHAN, C, et al. Miller Park. The Arup Journal, 37(1), pp24-33, 1/2002.

(2) http://en.wikipedia.org/wiki/Chinese_scholar%27s_rocks

(3) http://en.wikipedia.org/wiki/Ai_Weiwei(4) BAILEY, P, et al. The Virtual Building. The Arup Journal,43(2), pp15-25, 2/2008.

14. Tangency was maintained between the fve sets ogeometry or the stair line by adjusting only one parameteror the angle o the original stair line.

15. (a) Small-scale card prototype;(b) Full-scale oam board prototype.

16. Full-scale steel prototype.

a)

b)


20/52


Original roof analysis model and results

The main roo comprises interconnected 12m deep plane trusses, orming a thre

dimensional truss network structural system. A 3-D structural analysis model was

to carry out static and dynamic analysis o the roo structure, with the complete

primary and secondary steelwork structure modelled as a skeletal space rame in

ArupsGSA sotware (Fig 1).

The analysis model was created using beam elements. The roo is supported

24 column truss structures, each comprising two inclined truss elements and one

vertical diamond-shaped element. Fig 2 shows the truss member arrangement o

column head. At their lower portions, the three column truss elements are very clo

to each other, and detailed so that all three members merge to orm a single large

steel element. The column truss structures were assumed to be fxed to the pilec

with the oundation spring stiness estimated based on the pile load test results.

The retractable roo included in the original design was attached in its closed

position to the top o this ull model to allow dynamic analysis with the correct ma

distribution. Springs with dierent restraints were used to model the bearings and

bogies that would support the retractable roo (Fig 3).

Two basic GSA analyses o the roo were perormed, and the results verifed o

SAP2000 analyses.

Analysis and prototype testing

Kylie Lam Thomas Lam

1. Main components o the original roo design: The retractable section (top right); the main steel trusses supporting the roo, the aade and the retractable section (ar lesecondary members as bracing elements to the main trusses, orming the Stadium geometry (centre). The complete structural model is at the bottom right.

2. Column head, showing principal elements.

3. Interace between retractable and fxed roos, andmodelling o the support points restraint.

Table 1. Limiting element utilisation ratio.

Element type

Primary structure:columns

Primary structure:main truss

Secondaries

Static

80%

80%

90%

80%

80%

80%

90%

90%

100%

Seismiclevel 1

Seismiclevel 2

Seismiclevel 3

100% for slender sectioand 110% for others

100% for slender sectioand 110% for others

Not limited: member desassessed by non-linear ana

Retractableroof

Primary trusssuporting theretractableroof

Spring elementto simulate thesupport condition

Link element tosimulate thebogies


21/52

The Arup Journal 1/

A static analysis under various combinations o dead,

live, wind, snow, temperature, and seismic loading

was carried out. The eects o pattern loading due to

snow driting and the eects o dierent positions o

the retractable roo were evaluated separately.

Dynamic analysis established the undamental

requencies o vibration and mode shapes, and a

modal analysis was also undertaken on the ull

3-D analysis model.

Detailed seismic analyses were also perormed

to study the structural behaviour under a level 2

earthquake. In addition, the rare level 3 earthquake

was studied to ensure that the roo would not

collapse under this condition.

Member design check criteria and force/capacity

utilisation ratio

On the roo truss member design, design check

criteria and limit o orce/capacity utilisation ratios o

members were set up or dierent types o element in

terms o their unction and importance to the whole

structural system (Table 1).Fig 4 shows the members utilisation ratio o the

main truss under static load combinations.

Redesign of Stadium roof

Ater the rst preliminary design submission, the

Stadium roo was redesigned to meet the reduced

budget. The major changes included removal o the

retractable roo and enlargement o the roo opening.

Fig 5 shows the roo plan, including the retractable

roo, at the early stage, and Fig 6 shows the nal

stage o the preliminary design. Fig 7 shows the

evolution o the arrangement o the main trusses

during the roo redesign.

It was essential to maintain the Stadiums

architectural design principle that secondary

members should be indistinguishable in size rom

primary members. To save costs, however, the sizes

o some 1.2m x 1.2m box sections were revised.

For example, the cross-section o some top chord

truss members, invisible rom the plaza level, was

reduced to 1.0m square. The aade element section

size, however, was kept at 1.2m x 1.2m.

Construction stage analyses

Staged analyses o the xed roo were perormed

in conjunction with the assumed construction

sequence. The true refection o constructionsequence to analysis is important or a long-span

stadium structure, in which the lock-in stress eect

on secondary members is corrected and prevented i

the analysis is carried out as a unied whole.

4. Members utilisation ratio o main truss understatic load combinations.

5. Preliminary roo design, March 2004.

6. Preliminary roo design: redesign in November 2004.

7. Evolution o the roredesign, rom

preliminary desigstructural concepthe unifcationscheme.

1.0

0.8

0.6

0.4

0


22/52


The construction stage analyses that reected th

actual erection sequence included 78 installation

support points or alsework or the roo structur

erection. The key installation sequence is illustrat

in Figs 8a-g.

Based on the loading stage o the structure,

key construction phases were determined or the

static construction stage analysis, as ollows:

Phase1:Construct24columns,faade

secondary structure, ring trusses in the midd

and the primary truss (with temporary suppor

Phase2:Removethetemporarysupportafte

assembly o primary trusses in sections

(completion o the main structure).

Phase3:Constructsecondarystructureonth

top surace and acade stairs.

Phase4:Installthepipelinesforcladding

structure, catwalks, light fttings, and drainpip

Finite element analysis at nodes

Forthecurvedandtwistedmembersoftheroof

the connection nodes where many members metogether, fnite element analysis was used to stud

the stress distribution. Assuming the material is i

elastic stage, the results o the calculations were

expressedinthevonMisesstressdiagram.

Based on the analysis results, the member an

connection node design were optimised. The iss

stressconcentrationcanbeimprovedbymeans

local member thickening and adjusting the locat

stieners. Fig 9 shows the fnite element analysis

theelbowtrussattheeave.

Prototype testing

To ensure the saety o the design, prototype tes

werecarriedoutasverication.A1:2.5scaleelb

truss and a twisted thinned wall box section wer

testedattheBeijingTsingHuaUniversity(Figs1

11),whilst1:2.5scaleprototypesofthedouble

K-node o primary truss and column top, where

members merge at the node, were tested at the

ShanghaiTongjieUniversity(Figs12,13).

8. Key installation sequence for steel structure: (a) Column bases; (b) Columns and faadesecondary structure; (c) Primary truss and inner ring truss lifted panel by panel and jointed athigh level; (d) Removal of temporary support; (e) Secondary structure of the top surface;(f) Construction of facade stairs; (g) Completion of installation.

9. Finite element analysis at the elbow truss at the eave.

a)

e)

c)

b)

f)

g)

d)

Maximumnode

Minimumnode

Von Mises(max (Z1Z2

> 3.3

< 3.3

< 2.7

< 2.2

< 1.6

< 1.1

< 5.8

< 3.3


23/52

The Arup Journal 1/

11. A twisted and bent member at a top roundcorner connecting the top chord truss elementand the raking outer column (elbow truss).

12. Double K-node of a primary truss.

13. Column truss connection.

14. Intersection of inner side of spokes memberswith diamond-shaped inner column.

15.

10. Twisted thin-walled box secondary memberbeing tested at Beijing Tsing Hua University.


24/52


The challenge

The unique structural orm, the architectural constraints, and the clients and the Arupteams desire to reduce the steel tonnage, all posed great challenges to the seismic

design o the main roo o the Birds Nest.

The very long 313m span caused the seismic design to be signifcantly dierent in

several ways rom that o typical tall buildings. Seismic design measures that usually

achieve the collapse prevention perormance objective or tall buildings under the level

3 earthquake, or instance limiting inter-storey drits and detailing or ductility, were

insufcient or the Birds Nest roo structure. It could have collapsed straight

downwards without lateral sway, due to damage to its gravity orce-resisting system

rom vertical earthquake ground shaking alone.

The long span also causes the strength capacity o the primary truss members to

be taken up primarily by gravity loads. The box section top chord and diagonal

members in the primary trusses are subject to high axial compression orces under

gravity loads, and will sustain damage and degrade in strength due to global as well

as local buckling. They may not retain sufcient strength to prevent collapse when

damaged by a level 3 earthquake.

Ductile detailing measures or the bracing members in special concentrically

braced rames were thus insufcient to prevent collapse o the roo, because the

bracing members in special concentrically braced rames o normal buildings are not

part o the gravity orce-resisting system. It was necessary to limit the post-buckling

axial shortening o the top chords and the compression diagonals o the primary

trusses, thereby limiting degradation o compressive strength. This, however, was

beyond the conceptual ramework o the conventional code prescriptive seismic

design methodology.

A critical architectural constraint was the uniorm 1.2m x 1.2m cross-section o the

box section truss members. This is central to the architectural language o the Birds

Nest a seemingly arbitrary pattern that leaves spectators wondering which

members are primary structures and which are secondary. To meet the limiting plate

Seismic design of the roof


width-to-thickness ratio (b/t) o 16 (16:1) and to

achieve the seismically compact sections require

GB50011-2001, the minimum thickness needed

be approximately 70mm. Such a plate thickness

would result in the use o unacceptably large

amounts o steel, and lead to very high structura

sel-weight. This would urther increase the gravi

load on the structure as well as stiening it, leadto even higher seismic orces. In addition to bein

uneconomical, using thicker steel plates would

also have been less eective in achieving the

collapse prevention perormance objective or a

level 3 earthquake.

From the structural design point o view, an

eective and cost-efcient solution to reducing s

tonnage and thereby gravity loads and meet th

ductile detailing requirement o b/t 16 would

substantially reduce the outer dimension o the b

section members in both primary and secondary

trusses. It would be ar easier to achieve seismic

compact sections with much thinner plates, but

to the architectural constraints, this option was r

out in the early stages.

The behaviour o box section members with t

walls beyond the elastic limit is governed by thei

post-buckling behaviour. The Arup team investig

the eectiveness o welding longitudinal stiener

transverse diaphragms to the box section walls o

improving the ductility capacity o these member

Nonlinear fnite element simulations o the post-

buckling behaviour o a typical member with a ra

o stiener sizes and a range o diaphragm dista

showed that, while the stieners and diaphragm

eective in postponing local buckling o the walls

and thereby increasing member axial compressiv

1. The Stadium illuminated at night against its Beijing backdrop.


25/52

The Arup Journal 1/

strength, their eect on improving post-buckling

ductility is negligible, because the stieners

themselves buckle in the post-buckling range o

response (Fig 2). This set o nonlinear nite elementsimulation results convinced the Arup team early in

the project to abandon the option o seeking ductility

so as to meet code prescriptive rule, and instead to

adopt an alternative seismic design methodology.

The Arup solution: performance-based

seismic design

Having examined several options, the Arup team

adopted the perormance-based seismic design and

analysis approach or the roo structure. This is not

only the most technically rigorous, but also leads to

the most cost-ecient design. To achieve the

collapse prevention perormance objective or a level

3 earthquake, Arup established the ollowing

perormance targets or the structural members:

Primarytrussmembersshallremainelasticor

nearly elastic.

Secondarytrussmembersarepermittedto

sustain severe damage.

Arup used its own Oasys LS-DYNA nonlinear nite

element analysis sotware to demonstrate how the

collapse prevention perormance objective could be

achieved. The nonlinear response history analysis

captures the time histories o orces and

deormations in every primary and secondary truss

member in the inelastic range when subjected to

triaxial earthquake acceleration time histories,

representing the ground shaking rom a level 3

earthquake. A total o three sets o strong motion

records were used to represent the level 3

earthquake ground motion input.

The plate thickness o the box-section primary

truss members was determined by the need to

remain elastic or nearly elastic when subjected to the

level 3 earthquake, without meeting the b/t 16

requirement or ductile detailing. As a result o this

Beijing is in an area o moderately high

seismicity. The regions most recent

destructive event, the 1976 magnitude 7.8

Great Tangshan earthquake, had its epicentre

some 150km south-east o Beijing, which

suered severe and widespread structural

damage. Ocial gures indicate that in total

some 250 000 died as a result o theearthquake, and in Beijing itsel many were

orced to live in temporary housing or years

ater. The Beijing municipality implemented an

extensive programme to retrot surviving

buildings, and some o the multi-storey

masonry residences strengthened by

reinorced concrete rames can still be easily

identied in the newly-emerging CBD around

Arups Beijing oce.

The 1990 edition o the Chinese earthquake

intensity zonation map1 divides the country

into ve seismic zones, varying rom V (low) to

IX (high). Beijing is assigned to intensity zone

VIII. According to the 2001 edition o the

Chinese seismic ground motion parameter

zonation maps2, the peak ground acceleration

corresponding to 10% o probability o

exceedance in 50 years is 0.2g. The level oprobability o exceedance adopted or

drawing up these maps is consistent with

those in the 1997 edition o the Uniform

building code3 in the US and Eurocode 84 in

the EU. Compared to the seismic zone map

o the USA published in the 1997 UBC, the

seismicity o Beijing is equivalent to zone 2B

a level lower than that o Caliornia and

comparable to most parts o Washington,

Oregon, and Nevada.

Performance objectives required by the

Chinese seismic design code for buildings

The 1989 edition o the Chinese seismic

design code or buildings, GBJ 11-895,

established the ramework or seismic

perormance objectives o buildings in China.

The ollowing three levels o perormance haveto be achieved:

(1) no structural damage and limiting

non-structural damage in small but

requent earthquakes (50-year

return period)

(2) repairable damage when subjected to

an intermediate earthquake (500-year

return period)

(3) collapse prevention when subjected to

a large but rare earthquake (2500-year

return period).

The intermediate earthquake (level 2)

corresponds to ground motion intensity value

as shown in the Chinese seismicity zoning

maps. The small but requent earthquake

(level 1) is a once-in-a-lietime event or thedesign working lie o a building. The rare

earthquake (level 3) has a very low probability

o being exceeded during a buildings design

working lie.

The current Chinese seismic design code,

GB50011-20016, urther developed this

conceptual ramework and design/analysis

methodology by introducing modern,

non-linear response history analysis and

non-linear static pushover analysis methods

to quantitatively veriy satisaction o the

collapse prevention perormance requirement

under the level 3 earthquake.

For buildings within the limitations and scope

o applicability oGB50011-2001, a dual-leve

seismic design approach is prescribed. Both

the level 1 and level 3 perormance objectivesare required to be veried explicitly: strength

design and limiting inter-storey drit under the

level 1 earthquake, and checking and limiting

inter-storey drit and inelastic deormation

o members under the level 3 earthquake.

In addition, detailing measures or ductility are

prescribed or various seismic load-resisting

systems in various seismic zones.

The acceptable limits on inter-storey drit

under the level 1 earthquake are very

restrictive, refecting the intent o

GB50011-2001 to limit non-structural

damage. For instance, the limits on drit ratios

in reinorced concrete moment-resisting rame

systems and moment rame/shear wall

systems are 1/550 and 1/800, respectively.

The restrictive drit limits prescribed in

GB50011-2001 oten result in stierstructures compared to similar structures in

comparable seismic zones but designed to

other codes.

The level 2 earthquake perormance objective

is deemed to have been achieved by

GB50011 - 2001 i the design has satised

the level 1 and level 3 perormance

requirements and those or ductile detailing.

Regional seismicity

design approach, the lowest plate thickness o the primary truss box-section

members was reduced to 8mm, with the highest being 100mm. Most members ha

a plate thickness


26/52


AXIAL DEFORMATION (m)

FEMA 356 backbone curve

Non-linear finite element simulation

0.060.050.040.030.020.010

AXIALFORC

E(MN)

8

6

5

4

3

2

1

0

-1

4. Damage states of (a) primary truss members and (b) primaryand secondary truss members.

5. The varying plate thicknesses of the box section members are entirely concealed.

3. Post-buckling axial force/axial deformation relationship of atypical primary truss member.

The Arup teams computer simulation suggested that the box section members

possess, to some extent, higher strength and deormation capacities, but the gre

curve was adopted so as to be conservative in the global structures nonlinear

response history analysis.

Initial nonlinear computer simulations indicated that, in some analysis cases,

collapse may occur when subjected to the strong ground shaking o the level 3

earthquake. Arup examined the collapse process in these computer runs and

identifed the critical primary truss members that needed to be strengthened.

Ater a ew iterations, the collapse prevention perormance objective was achieve

in all analysis cases.In the damage states o the roo truss members (Fig 4), most primary membe

remained elastic (green), but some sustained moderate damage (blue: the immed

occupancy damage state), entering slightly into the post-buckling range o respo

Only a ew reached the signifcant damage state (yellow: the lie saety damage st

responding well into the post-buckling range o response but without reaching th

point at which strength starts to degrade.

On the other hand, as the perormance objective had intended, many second

truss members were damaged severely (red: the collapse prevention perormanc

objective), exhibiting signifcant strength degradation.

Elastic Immediate occupancy

Life safe Collapse prevention


27/52

The Arup Journal 1/

The expert panel review process for approval

The importance o the National Stadium project meant that, besides the normal

approval procedure, the Beijing Municipal government set up an expert panel

committee to review the structural design, a process similar to that in Japan.

In both countries, expert panel review and approval oten requires explicit verifcatio

o perormance under all three earthquake levels, and nonlinear response history

analysis is required to demonstrate that the collapse prevention perormance

objective under the level 3 earthquake has been achieved.In May 2004, the expert panel met or two days in Beijing to review the prelimina

design o all disciplines or the Birds Nest. The panel included several chie

structural engineers o local architectural design institutes, as well as members o t

China Academy o Engineering who are recognised experts in long-span roo

structures. At the end o the rigorous review meeting, Arups structural preliminary

design passed the review and was endorsed by the panel or approval.

Added value

Arups perormance-based seismic design is not only innovative and rigorous,

but also cost-efcient, creating exceptional value or the client. The innovative conc

o nearly elastic design subjected to the level 3 earthquake, assisted by the

perormance-based seismic design and analysis methodology using state-o-the-a

nonlinear numerical simulation technology, not only convincingly demonstrated

achievement o the collapse prevention perormance objective, but also resulted in

very signifcant reduction in the quantity o steel used. The plate thickness o most

1.2m x 1.2m box-section roo members is substantially lower than the 70mm requ

by the ductile detailing rules specifed in many international seismic design codes,

instance American Institute o Steel Constructions Seismic Provisions or Structura

Steel Buildings10, or achieving seismically compact (equivalent to class 1 plastic in

terms o Eurocode 37) sections.

Figs 6 and 7 illustrate the distributions o plate thickness o the chord members

the primary trusses. Only two groups o top chord members and our groups o

bottom chord members reach or exceed 70mm plate thickness.

References

(1) CHINA EARTHQUAKE ADMINISTRATION. Earthquake intensity zonation map o China. State Coo the Peoples Republic o China, 1990.

(2) CHINA EARTHQUAKE ADMINISTRATION. GB18306-2001: Seismic ground motion parameter

zonation map o China. General Administration o Quality Supervision, Inspection and Quarantine o

Peoples Republic o China, 2001.

(3) INTERNATIONAL CONFERENCE OF BUILDING OFFICIALS. 1997 Uniorm building code. Volume

Structural engineering design provisions. UBC, 1997.

(4) EUROPEAN COMMITTEE FOR STANDARDIZATION, EN 1998-1: 2004/Eurocode 8: Design o

structures or earthquake resistance Part 1: general rules, seismic actions and rules or buildings,

December 2004.

(5) MINISTRY OF CONSTRUCTION. National Standard of the Peoples Republic of China GBJ 11-89

Code or seismic design o buildings. The Ministry, 1989.

(6) MINISTRY OF CONSTRUCTION. National Standard of the Peoples Republic of China

GB50011-2001. Code or seismic design o buildings. The Ministry, 20 July 2001. Actualised: 1 Jan

2002.

(7) EUROPEAN COMMITTEE FOR STANDARDIZATION. EN 1993-1-1: 2005/Eurocode 3. Design o

structures. Part 1-1: general rules and rules or buildings. EC, May 2005.

(8) FEDERAL EMERGENCY MANAGEMENT AGENCY. FEMA 356. Prestandard and commentary o

seismic rehabilitation o buildings. American Society o Civil Engineers, 2000.

(9) MINISTRY OF CONSTRUCTION. National Standard of the Peoples Republic of China GB50017

2003. Code or design o steel structures. The Ministry, 25 April 2003. Actualised: 1 December 20

(10) AMERICAN INSTITUTE OF STEEL CONSTRUCTION.AISC/ANSI 34105. Seismic provisions o

structural steel buildings. AISC, March 2005.

20 x 20

20 x 25

25 x 25

25 x 30

30 x 30

30 x 36

30 x 42

36 x 36

36 x 42

42 x 42

42 x 50

50 x 60

60 x 60

20 x 20

20 x 25

25 x 25

20 x 30

25 x 30

20 x 36

30 x 30

25 x 36

30 x 36

36 x 36

30 x 42

36 x 42

42 x 42

42 x 50

50 x 60

25 x 30

20 x 36

30 x 30

25 x 36

30 x 36

25 x 42

30 x 42

36 x 42

30 x 50

50 x 50

50 x 60

50 x 70

70 x 70

80 x 80

90 x 90

20 x 20

20 x 25

25 x 25

20 x 30

25 x 30

30 x 30

30 x 36

36 x 36

42 x 42

42 x 50

50 x 50

60 x 60

70 x 70

70 x 80

6. Plate thickness distribution of (a) 1.2m deep x 1.2m wide,and (b) 1.0m deep x 1.2m wide top chord members of theprimary truss (all thicknesses in mm).

7. Plate thickness distribution of (a) 0.8m deep x 1.2m wide,and (b) 1.2m deep x 1.2m wide bottom chord members ofthe primary truss (all thicknesses in mm).

a)

a)

b)

b)


28/52


The retractable roof design

John Lyle

2. The retractable roo in open position.

1. The fnal design allowed a larger opening above the pitch and a reduction in the amount o steel used in the fxed roo.

Design concept

Arups brie or the retractable roo covered the

development o a perormance specication

alongside the structural, mechanisation, and con

system scheme design to demonstrate easibility

The original competition entry comprised two

large retractable roo panels that split at the halw

line and parked at the ends over the xed roo w

open. Further development o this concept led to

retractable roo structure that refected the seem

irregular Birds Nest structure o the xed roo.

Retractable roos and the systems required to

move them need rom the start to be considered

holistically with the xed structure. The sheer size

weight o what is being moved means that its

Background

Any account o the development o Beijing National Stadium would be incomplete

without some reerence to the retractable roo. Its design dominated much o the

Stadiums early development beore it was nally omitted as a cost-saving measure in

June 2004, due both to the rising cost o steel and political pressures to keep the

Olympic budget under control.

When planning Olympiads, the use o the stadium ater the Games has become a

major part o the sustainability and economic discussions - Olympic venues are oten

noted more or their poor utilisation ollowing the Games than their long-term

contributions to regenerate or add new acilities to host cities. The Beijing Organising

Committee (BOCOG) intended to resolve these issues by including a retractable roo

that could transorm the Stadium into a large indoor arena and thereore extend the

range o events that could be held throughout the year. This did not happen. However,

removing the retractable roo rom the design allowed a larger opening above the

pitch and a reduction in the amount o steel used in the xed roo, and in hindsight,

the iconic architecture around the Beijing Olympic Park and the overall success o the

National Stadium (even without its retractable jewel) justies the decision.

Arup took the design o the retractable roo rom its early concept up to a airly

advanced scheme design stage. All this work, including discussions with specialist

contractors and initial meetings with the expert panel review team in Beijing, was

completed beore the decision was taken to cancel the retractable roo.


29/52

The Arup Journal 1/

behaviour infuences the perormance o the other components, and vice versa.

Arups concept, thereore, needed to address the compatibility o movements

between the xed and the movable structures induced by the latter as well as

imposed loads (such as snow, wind and seismic), thermal movements, and

construction tolerances.

Fabrication and erection issues also had to be considered rom the outset.The overall erection strategy adopted by Arup was to maximise preabrication and

minimise in situ assembly undertaken 70m-80m above ground. The scheme reduced

construction and commissioning time by using ground level-based assembly

methods, allowing near-nished components to be craned onto the xed roo.

This approach was combined with an o-site test and development programme

to eliminate any development during nal installation, as part o the overall risk

reduction process.

Preliminary design

A retractable roo design that met both the architectural ambitions and was

mechanically reliable was the obvious goal, and these targets became the key drivers.

Retractable roof structure

The retractable roo structure geometry comprised two halves, each spanning 75mand 70m long. At the back edge o each hal (ie the ends urthest rom the opening),

the perimeter ollowed the same curve (in plan) as the xed roo perimeter so that

back edge o the retractable roo would merge with the xed roo when in the open

position. At the ront o each hal, the edge was a more complex curve: when the two

halves moved rom open to closed, they would orm the distinctive yin-yang shape

at the halway line (Fig 3).

The adopted design split each hal-roo into ve dierent triangular panels so that

each hal o the roo would move as a train o connected panels (Fig 3). This approach

would reduce the loads in both retractable and xed structure considerably.

The triangular panels consisted o primary and secondary steel trusses, the ormer

(maximum 8.2m deep) spanning between bogie support points and carrying the load

across the main span. The secondary members, spanning between the primary

trusses, would act both as lateral restraint or the primaries and as a method o

transerring vertical loads back to the main spans.

Separating the roo into discrete panels had signicant benets:

Supportingthethreecornersofeachtr iangularpanelmeantthatthesupports

were always in contact with the main roo. This statically determinate condition

allowed the support conditions to be simplied.

Theseparatepanelsalsoallowedtheretractablerooftoarticulate,meaningthat

the xed roo did not need to conorm to strict displacement criteria; vertical

movements in it would be easily accommodated.

Separatingtheroofintosmallerpanelsmeantthat itcouldbebuiltontheground

and lited in, reducing the amount o in situ construction.

The layout o the primary and secondary trusses w

co-ordinated with the xed roo geometry to reduc

the visual density o steelwork when seen rom ab

during TV coverage o major events. In the open

position, the secondary structural members in the

retractable roo aligned directly above the steelwoin the xed roo. When closed, the retractable roo

primary members aligned with the xed roo

members to provide visual continuity (Fig 4).

Structural analysis

A 3-D structural model o the panels was construc

and analysed using the Oasys program GSA to

assess static and dynamic load cases on all ve

panels and to check compliance with the Chinese

steel code.

Imposed loads were similar to those used or th

xed roo, with the ollowing additions:

Seismic: A rst pass seismic analysis was

perormed using a code-based spectra and dynam

response analysis. Because o the complexity in th

load paths, this was later developed into a combin

xed and retractable non-linear seismic model usi

LS/DYNA non-linear nite element analysis sotwa

3. (a) Retractable roo truss structure; (b) detail.

4. Continuity between fxed and retractable roo.

a) b)


30/52


Racking loads: Two additional static loads were reviewed or out-o-tolerance

positions during movement (100mm longitudinal racking load and a 200mm verti

dierential movement within a panel.)

Mechanisation system

This comprised the bogies and drive components needed to move the retractabl

roo. While there is no universally preerred approach or retractable roo bogies a

drive systems, the mechanisation design strove or several objectives in pursuit o

reliability and cost-eectiveness. The key eature connecting these objectives wa

mechanical simplicity.

Bogie design

Each bogie, typically weighing about 3 tonnes, would support the corners o the

triangular roo panels. At the interace between the bogie and panels, proprietary

spherical thrust and sliding bearings would accommodate the movements and ca

the lateral loads induced by the drive system and inclined tracks.

The bogies also had to provide stability in an extreme seismic event, and addi

restraint was provided by sliding restraints transerring loads onto the xed roo

structure. These tie-downs also transerred any uplit loads induced by wind.

Drive system

The gradient o the curved track on the xed roo (10 at its steepest) meant that

powered railway-type bogie system could not be driven reliably without a rack-anpinion drive or winch-driven system. While there was sucient space within the b

to package the ormer, the design progressed using a wire rope (cable) winch sys

as this was the most cost-eective option.

The reeving arrangement chosen conveniently houses the winches within the

retractable roo, reducing the amount o exposed equipment on the xed roo.

Mounting the haul ropes, drums and winches on the bogies also reduced the ove

length o steel cables required and improved positional control. The cable would

move relative to the xed roo, so additional sheave rollers on the roo or return pu

would not be needed (Fig 4).

Based on the scheme selected, either hydraulic motor drives or three-phase

electric induction motor systems (around 150kW) could be used to move the roo

Control system

An automatic system was selected to control the movement o the roo, with onlyminimal operator intervention. A sel-equalising drive system would ensure that th

roo moved without skewing on the rails. Accurate positional control would minim

position errors caused by tolerances, structural defection, wind, or lack o

synchronisation between motor drives on each side, and i errors did occur,

they could be corrected quickly.

Electrically-controlled ail-sae brakes were included in the design to eliminat

risk o control system ailures. Arup also completed an initial FMEA (ailure modes

eects analysis) or the retractable roo to evaluate the system-wide risks or

potentially catastrophic events such as cable ailures.

Retractable roo perormance specifcation

A signicant reason or undertaking the retractable roo scheme design was to

develop a robust perormance specication, which as a result not only developed

basic unctional requirements such as opening and closing speeds, design lie,

operating wind, and temperature envelopes, but also allowed relevant structural

interace loads, defections, and tolerances to be described. Other details, such a

drainage and sealing and control and maintenance requirements, were also ident

in the specications.

The combination o the reerence design and perormance specication allowe

competitive tenders to be obtained or the mechanisation systems as part o a

retractable roo procurement process that was based on a properly integrated de

8. Plan of retractable roof showing positions of bogies.

Closing seal to otherhal o retractable roo Anchor block

Track

Rollerbearing unit

Slidingconnection

Pinnedconnection

Pinnedconnection

Tip lock device

Brakingunit

Runway beam - topchord o 12m truss

Anchor block

Bogie withcable drum

Cable

Passive bogie

Passive bogie

Bogie withcable drum

Front oretractable roo

Rear oretractable roo

6. Bogie in place.

7. Detail of bogie.


31/52

The Arup Journal 1/

The bow


32/52


Geometry and profle

The plan geometry comprises a radial grid that denes the rames and a aceted grid

dening the circumerence (Fig 3). The east and west seating radius varies rom about

270m to 320m, whilst the north and south seating radius is between 60m and 110m.

The nominal spacing o the radial grid is 7.5m, tapering towards the pitch. To suit the

roos overall saddle shape, the number o storeys varies at dierent places, rom a

maximum o seven (51m tall) on the east/west centreline to ve (45m tall) on the

north/south centreline (Figs 4, 5).

Foundations

All vertical loadbearing elements are supported on reinorced concrete pile

caps supported by cast in situ concrete bored piles with a diameter between

800mm and 1000mm, ounded in the cobble/gravel stratum layer, about 38m below

existing ground. A plinth with a one-storey basement surrounds the concrete bowl

area, resting on a shallow pad oundation on the natural subgrade at about 8.5mbelow existing ground.

Superstructure

The bowl is split into six segments (Fig 3) with 200mm wide movement joints between

them. Each segment orms an independent structure with its own stability system

provided by column-beam rame action and the concrete staircase and lit cores.

The six segments are between 120m and 150m long. The movement joints that

separate them are continuous through every foor o the bowl, including the terracing,

but are not required at basement level. The lower ground level is o 500mm thick fat

slab construction, acting as a foor diaphragm to tie together the oundations.

The upper foors are generally 175-225mm thick reinorced concrete slabs

spanning between 600mm x 1000mm deep primary beams at about 7.5m centres on

the radial gridlines that dene the rames. The slab thickness changes due to the

increasing span caused by the tapering o these gridlines.

Layout and analysis model

For the middle and upper tiers, the terracing is

ormed rom precast L-shaped units spanning

between the primary rames, and supported on

inclined tribune beams (Fig 6). For the middle tie

the tribune beams are 1000mm x 1000mm deep

on the upper tier, due to increased spans, their d

increases to 1.2m.

The columns are generally located on every ra

grid line. Under the lower tier they are all vertical,or the middle and upper tiers, the ront column i

inclined towards the pitch in the radial plane to

reduce the cantilever length o the tribune beams

At the back, the columns are inclined both radia

and circumerentially. Inclining the columns is a

eature o the architectural design, bringing the

designedly chaotic aade member arrangeme

into the concourse area (Fig 7).

Tony Choi Thomas Lam

1. Inclined tribune beams to support precast units that form the terracing.

North/southsegments

East/westsegments

3. Grid system and movement joint disposition.

2. Columns are inclined both radially and circumferentiathe back of the bowl.


33/52

The Arup Journal 1/

Ring beam

Tribune beam

Tribune beam

Ring beam

Tribune beam

Tribune beam

7. Inclined columns in concourse area.

5. East/west segment.

6. (a) Section through north/south segment;(b) Section through east/west segment.

(a)

(a)

(c)

(e)

(g)

(b)

(d

(f)

(b)

8. Modelling analysis of the structure.

4. North/south segment.


34/52


Compared with that o the roo, the seismic design o the bowl structure o theBirds Nest was more straightorward, within the limits and scope o the seismic

code GB50011-2001. Apart rom being supported on a single continuous pile

oundation system, the two structures are completely separated rom each other,

and as already noted, the bowl is divided into six independent structures by

movement/seismic joints, wide enough to accommodate both thermal expansion

seismic moments (Fig 1). In dividing the bowl, the symmetrical plan layout adopte

had the eect o reducing the number o dierent bowl structures to two: the eas

west bowls and the north\south bowls, respectively approximately 150m and 120

long. The east\west bowl structure has six to seven storeys with a maximum

structural height o 51m; the north\south bowl has ve to six storeys and a struct

height o 45m on the lowest point on the north/south centreline.

In each independent bowl structure, the two lit cores eccentrically located tow

the back o the structure orm two structural shear wall cores, resisting both grav

orces and most o the lateral orces delivered to them by the foor diaphragms (F

The moment-resisting rames primarily support gravity loads and, together with th

cores and diaphragms, orm a combined reinorced concrete moment rame\she

wall lateral orce resisting system.

As required by GB50011-2001, a dual-level seismic design approach was ado

or the bowl structure. Moment rames and core walls are sized and proportioned

that member strength capacities equal or exceed member orce demands, and

inter-storey drit ratios are limited to 1/800 when subjected to a level 1 earthquak

Arup was responsible or the bowl structures up to scheme design level, and

subsequently assisted the Local Design Institute CADG on the preliminary design

and construction drawings.

Seismic design of the bowl


2. Structural system, showing lift cores.

1. Movement/seismic joint between segments of thebowl structure.


35/52

The Arup Journal 1/

Specialist engineering desig


36/52


The roof comprises two membrane layers. The outer is a single-layer transparent

ETFE (ethyltetrafluoroethylene) stretching membrane system (Fig 2), which functio

as weatherproof protection to the spectator stands. The inner and ceiling membr

is a single-layer translucent PTFE (polytetrafluoroethylene) membrane system (Fig

which serves as the acoustic ceiling and provides shade for the spectators.

The separation between the membranes is approximately 13m (Fig 4).

Because of the interwoven truss structure, the shapes of the roof segments ar

entirely irregular, varying between triangular and octagonal. There are around 100

ETFE panels on the roof, ranging in size from 1m2 to 230m2. Altogether, the ETFE

panels total some 38 000m2. The ETFE membrane is stressed over a subframe oarches in tubular steel supported on the structural gutter elements, welded to the

top chord (Fig 1).

The approximately 800 PTFE panels for the acoustic ceiling range from 5m2 to

250m2, and total about 53 000m2. The PTFE acoustic ceiling membrane system

is stretched to the tube subframe structure suspended from the underside of

the roof truss.

Arups scope on the roof cladding and acoustic ceiling was to design for the

loading effects onto the supporting roof structure.

Tony Choi

Roof cladding and

acoustic ceiling

Shade for spectators

Minimal TV shadow

Matt ETFE film

Protection fromcold air

Warm air releasedthrough openings

Outer ETFE membrane

Inner PTFE membrane

PTFmem

Acousticabsoption

100%

Daylighttransmission

93.0%

55.8%

18.4%

TransparentETFE film

Outer ETFE membrane

Stretching wireshemed with ETFE film

ETFE film

ETFE film

Weldedgutter

Weldedgutter

Steel pipearches

Steel pipearches

Steel platebrackets

Steel platebrackets

Stretching wires for acousticceiling suspension poles

Watertightaluminium strip

Watertightaluminium strip

Security system along all beams

1. Details of fixing for ETFE cladding.

4. Section through Stadium showing locations of ETFE and PTFE cladding.

3. The innermembrane is a single-layer translucent PTFE membrane, which serves as the acoceiling and provides shade for the spectators.

2. Outer ETFE cladding.


37/52

The Arup Journal 1/

A combined boundary layer wind tunnel and numerical modelling study was

carried out to assess spectator comort levels or the Stadium, with wind tunnel

measurements being made or the external plaza surrounding it, and or the

concourses and key seating areas within. These wind speeds were used in assessing

pedestrian saety and comort in and around the Stadium (Figs 1, 2). Wind conditions

in the Stadium and external plaza are generally suitable to strolling or or short periods

o standing or sitting. No areas would be uncomortable or strolling, which was

entirely acceptable or the intended usage.

The International Association o Athletics Federations (IAAF) competition rules

stipulate that, or all athletics records up to and including 200m, the long jump and

the triple jump, inormation concerning wind speed must be available. I the wind

velocity behind the athlete in the direction o running averages more than 2m/sec, the

record will not be accepted. Measurements were thereore also made o wind speeds

around the tracks, and the results presented in terms o percentage o the time that

mean and gust wind speeds would exceed around 2m/sec on the track, notably or

the sprint and horizontal jumps area (Table 1). The results showed wind conditions in

the athletic arena during the summer months to be, on average, very benign.

In addition, wind speed measurements were made over the tured areas o the feld

so as to develop appropriate turfng strategies or the Stadium (Fig 3). An important

aspect o tur health and growth is air movement. Assuming a reasonable criterion or

acceptable ventilation to be 1m-2m/sec, Fig 3 shows that the south-west and

north-west corner zones are better ventilated than other areas o the feld (north is at

the right). In addition, the tur ventilation data are combined by the tur consultant with

assessments o sunlight patterns and daily temperatures and humidity to determine

how well tur grass will thrive under the given combined conditions.

Alex To

Wind conditions in the Stadium

and external plaza

3. Wind speed contour over the turfed area.

2. Wind comfort range in the external plaza.

1. External plaza.

Table 1. Exceedances of 2m/sec tailwind for various evenduring summer.

EventsAmount of time ta

exceeds 2m/

100m/200m sprints 5.44%

100m/110m hurdles 4.80%

Long and triple jumps (north to south) 0.00%

Long and triple jumps (south to north) 7.07%


38/52


The Beijing Olymp

arup journal 1-2009 beijing stadium

Documents