A

research review

# | Research Year Book

research review:Published 2008

research review:

Introduction: Jeremy Watson 2

Meet the team 3

SuperLight Car (SLC) 4

Sustainable Bridge Structures: 8 Support for Offshore Wind Farms and Energy Storage

Embodied Energy as an Environmental Impact 12 Indicator for Geotechnical Infrastructures

High-Resolution 3D Measurement and 3D 16 Computational Fluid Dynamics Modelling of Reservoir Spillways

Water Quality Modelling 20

Responsiveness To Market (RTM) Building System 24 for the Residential High-Rise Building Design Process

Innovative Approaches to Pedestrian 28 Planning at Toronto’s Union Station

Air Ventilation Assessment by External Wind 32 Modelling Using Different Turbulence Models

Innovative Façade Technology Based on Pultruded 36 Fibre-Reinforced Polymer Composites

Using Fire Dynamics Simulator (FDS) 40 to Predict and Model Fire Propagation

Behaviour of Tunnel Rock 44 Subject to Fire Loading

Full Scale Fire Tests in Tunnels and the use of Advanced 48 CFD modelling to Predict Fire and Smoke Behavior

Climate Scenarios for Urban Design: 52 A Case Study for the London Urban Heat Island

The importance of research to Arup 56

contents

research review | 1

It gives me a great deal of pleasure to introduce the Arup research review. This publication offers a series of case studies demonstrating examples of the research work we have completed, many in collaboration with other organisations and research institutes. As Director of Global Research at Arup I am increasingly excited and inspired by the breadth, depth and quality of the research projects we undertake, both for our clients and to develop our internal capabilities.

Research at Arup

Research has a strong history at Arup and we strive to deliver new thinking and research excellence in support of our clients’ projects and aspirations. Our strategy seeks to maximise opportunities for the future by identifying the needs of our clients and businesses, as well as monitoring and supporting new developments in the academic sector. This ensures that our practices and groups remain at the forefront of innovation and delivery. At Arup we like to view our research capabilities as filling the gaps between current best practice and the needs and aspirations of future business.

Research has always underpinned Arup’s work and it is increasingly important that it forms an integral part of the projects we deliver. In 2006, focus was enhanced by the establishment of a formal Research function, working closely with our Design and Technical Executive (DTX). We have Research ‘Champions’ across the world in our Australasia, East Asia, Americas and Europe regions, each of whom works alongside the regional Design and Technology leader. This has allowed us to position strategic research as a separately funded area, enabling external collaborative research around the world, and leading to knowledge creation, in addition to offering our research capabilities as a service stream to our clients. Our DTX leaders, with their teams, identify business opportunities and define what needs to be researched. Our Champions then liaise with regional Arup teams and their university contacts and other institutions to execute regionally-steered research work

Research Priorities

Our research prioritisation processes have been reviewed to ensure they are clear and effective. Key programmes and themes are defined through exploratory and delivery phases. The exploratory phase is undertaken through a roadmapping exercise which links global drivers with Arup business opportunities to deliver a research agenda which addresses the future needs of our business. Additionally we also offer customised strategic Roadmapping workshops to our clients to support their research agendas, and look to act as a broker for external research providers,

integrating internal and university partner capabilities to address clients’ research needs. To facilitate knowledge sharing and create an interactive collective work environment, we have designed an innovative approach using a Wiki. This connects our offices and our existing networks, enabling rapid sharing of proposed research programmes and research results across the globe to support research objectives and our clients’ requirements.

Collaboration

Our areas of expertise are wide and our breadth and depth of knowledge and skills impressive, with many staff members being respected experts in their fields, publishing articles in peer-reviewed journals. However, while we have much expertise in house, our research offerings are greatly strengthened by strategic alliances and partnerships. Our collaborative work with universities is world renowned and clients are seeing growing benefits from this joint approach. We nurture our partnerships with organisations and businesses who contribute to and support our research initiatives, and many of our experts contribute to universities around the world as visiting professors, as well as continuing to be independent advisers to industry. Arup staff find it very stimulating and enjoyable to work in a company culture which encourages this.

As well as the work with individual organisations we also ensure that our thinking influences and aligns with national research priorities, and we seek to leverage company investments in research through national funding programmes.

Arup is proud of its long-term relationship with the Engineering and Physical Sciences Research Council (EPSRC) in the UK. This relationship recently grew with the establishment of a Strategic Partnership, the first of its kind in our sector. Together with the EPRSC we define research programme themes for UK universities and agree funding for chosen research areas. Project proposals are invited by calls from EPSRC, with winners selected by their peer review process. Projects are jointly funded, with the UK Government providing 50% of the necessary finance through the Council and the remainder coming from Arup together with private sector partner organisations. Arup is currently forming a consortium to jointly raise the additional 50% of funding required. This group will work with an elite network of Universities to undertake research identified by the partners. We recently secured an investment of £1.5M from the EPSRC on behalf of participating universities, to facilitate research networking in

international Eco-City issues, the focus for this being the Dongtan Eco-City near Shanghai. With RCUK and the British Embassy, Arup led successful discussions with universities and funding agencies in China, working on a high-level Memorandum of Understanding for Sino-British collaborative research. This is an exemplar for our credibility and thought-leadership in partnerships with universities internationally, to secure joint funding and define the scope of the projects which arise from international collaboration. Working with four Universities on Eco-City research themes, around 50 research projects are being identified over the course of this year which focus on four key areas. University College London is managing two network themes on City Planning and Engineering Infrastructure, while Imperial College is leading a theme on Management and Execution of Eco-City projects. Southampton is co-ordinating the complete process. Tongji University, Shanghai is a key partner in all of these Networks.

Arup is also leading the way for the UK on behalf of the public and private sectors, in an EU Joint Technical Initiative (JTI) on Energy Efficient Buildings. Arup has provided the technical expertise and understanding to join the preliminary team in Europe which seeks to secure funding from the Commission for this €1.5 – 2bn initiative. Work on the JTI will form a central part of Arup Research’s vision and focus over the next 18 months.

Such partnerships and research initiatives are innovative and great to be involved in. As well as allowing us to grow our own research and demonstrate excellence in many technical areas, they ensure we develop relationships with the best in industry and academia to provide research that adds value for our clients.

The case studies included in this research review go some way to demonstrating this expertise but they are a small selection chosen from the many projects available.

I hope that you find this research review interesting and inspirational, and that you will want to collaborate with us.

Jeremy Watson jeremy.watson@arup.com www.arup.com/researchinarup

Author: Jeremy Watson, Director, Global Research, Arup

Introduction

2 | research review

Prof. Jeremy Watson,

Director, Global Research, is responsible for Arup’s research strategy and its research business. He has a background in research and technical management roles in industry and academe, including service with the previous UK Department for Trade and Industry and EPSRC. A chartered engineer and fellow of the Institute of Engineering and Technology, Jeremy is a visiting professor at the Universities of Southampton and Sussex. He has recently been appointed a board member of the UK Technology Strategy Board, funding university industry collaboration.

Dr Jennifer Schooling,

Research Business Manager, manages Arup’s multidisciplinary research service, enabling clients and staff to access the many skills within Arup, and establishing successful collaborations with external agencies. In previous roles, Jennifer managed engineering research and development projects for both academic and commercial applications. She has also handled new product development in the semiconductor equipment industries, managing new product introductions from concept design to product launch.

Dr Marta Fernandez,

Research Relationships Manager, focuses on relationships between Arup’s internal network and research partners externally, as well as supporting our efforts to realise the value of the firm’s IP. A chemical engineer by training, Marta’s previous roles included commercialising early-stage technology in renewable energy start-ups, and forging links between industry and academe in the engineering, energy and environmental sectors.

Jackie Young,

PA to Jeremy Watson, provides general administration support for the research team.

With more than 20 years’ service at Arup, Jackie has worked in many groups including Finance, Building Engineering and the Group Board.

Meet the team

research review | 3

4 | research review

SuperLight Car (SLC)

Introduction

The challenge of climate change has added increased urgency to the need to reduce our disturbance of the environment. The Stern report states that “The scientific evidence is now overwhelming: climate change is a serious global threat, and it demands an urgent global response.” Stern concludes that the developed world needs to reduce total emissions by 60-80% by 2050.

In Europe, passenger cars contribute approximately 12% of man-made CO2. Thus the reduction of passenger car emissions has become a major priority, and has been the subject of continuing discussions between the European Commission and vehicle manufacturers.

Reducing vehicle mass is an effective way of reducing vehicle emissions. A 10kg mass saving can reduce CO2 emissions by about 0.8g/km. There is therefore a strong driver to bring forward research on vehicle body mass reductions.

The SuperLight Car programme (SLC) was conceived by a consortium of 38 European organisations. The consortium comprises 7 car manufacturers (Volkswagen – project coordinator, Volvo Technology, Fiat, Opel, Renault, DaimlerChrysler and Porsche), 13 academic or research centres, 13 automotive component or service suppliers, and 5 small/medium enterprises. SLC is a €20m research programme, and is part-funded by the European Commission as part of the Sustainable Surface Transport priority of the

Sixth Framework Agreement. It is scheduled to run from 2005-2009.

The objective of SLC is to develop a lightweight, multi-material concept body suitable for mass production. The target is a 30% body mass reduction, using the already efficient VW Golf 5 body as a benchmark, and with a maximum cost penalty per vehicle of €5 per kg saved. In addition, the performance of the benchmark body in stiffness, strength, durability and crash is to be maintained.

The multi-material solution allows optimum use of the properties of different materials in appropriate locations of the body. This approach differs from many previous research programmes that have concentrated on the use of one material (eg ULSAB-AVC and NSB in steel, ULCEV and FLOAT in aluminium, and Hypercar in composites).

A mass saving of 30% of the Golf body equates to 85kg, but the saving also allows secondary mass savings of around 40kg by downsizing other vehicle systems (eg engine, chassis and brakes).

The ultimate potential emissions saving in Europe can thus be calculated as follows.

• 125kgvehiclemassreductionleadsto10g/kmreduction in CO2 emissions

• AnnualdistancecoveredbyanaverageEuropean car is 15,000km

• AnnualreductioninmassofCO2 per car becomes 150kg

• 225mpassengercarsonEuropean(EU26)roads in 2006

• TotalannualEuropeancarCO2 reduction is 33.8m Tonnes CO2

Note that with the average European car being 8 years old, the benefits will be delayed in coming to fruition.

Methods

Overview of SLC project

A conventional vehicle body programme can rely on experience gained over several decades of manufacture. The use of advanced materials and a multi-material solution raises questions that are outside the normal comfort zone relating to all aspects of the life cycle; material selection and performance, forming of components, methods of joining, corrosion and durability, recycling and, of course, cost.

It was therefore necessary to support the engineering design with data from research into the new materials and processes. The project was devised to encompass these requirements and was subdivided into 5 subprojects covering the following topics.

• SP1Vehicledesignandengineering

• SP2Formingandjoiningtechnologies

• SP3Designandsimulationtoolsand other enablers (life cycle analysis, recycling and costing)

• SP4Developmentofdemonstratorcomponents and systems

• SP5Trainingandmanagementactivities

Arup’s role was primarily within SP1, and the majority of this paper is related to the SP1 activities. It is however useful to understand the relationships between SP1 and the other design subprojects, SP2 and SP3.

SP1 and SP2 have worked closely to develop the body structure, with SP1 proposing component material preferences, and SP2 solving the related manufacturing and assembly issues.

In addition, SP2 have been studying new technologies to offer improved solutions for the SP1 design team to consider for use. For example, forming of magnesium, long fibre thermoplastic (LFT), ultra high strength steels, and aluminium alloys.

The deliverables from SP2 have been a material database, design guidelines, and manufacturing feasibility evaluation on sub-assemblies. Although primarily aimed at meeting SLC requirements, they are also enablers for future multi-material vehicle body design.

Abstract

The SuperLight Car (SLC) programme was conceived by a consortium of 38 organisations across Europe to develop cost-effective lightweight vehicle body concepts suitable for high volume manufacture. The target of the 4-year programme, part-funded by EC, was to reduce the body mass of a car by 30%, using a multi-material approach. It was a requirement of SLC that the final design matched the reference VW Golf 5 vehicle body in crash, modal and durability performance.

The multi-material solution has required feasibility support from experts in forming and joining technologies, backed by testing for strength and durability of joints. Bespoke tools for estimation of manufacturing costs and environmental impact have been used to ensure a practical and sustainable solution.

Arup has worked within the core engineering design team to create the design concepts, and has had particular responsibility for the prediction of vehicle performance using computer simulations.

Now in its last year, SLC has been shown to meet its core targets, and work continues to manufacture demonstration bodies to aid in the dissemination of the knowledge to the wider automotive engineering community.

Author: Neil Butcher

research review | 5

SP3 has provided detailed data to assist in the SP1 concept development.

• Materialtestinghasbeenperformedandtheresults assimilated into material property cards for the finite element computer simulations.

• Testingofjoiningmethodsandjointconfigurations has been performed to provide failure criteria for durability simulation, and data for the stiffness and crash simulations in SP1.

• Costingmodelshavebeencreatedtoallowfastcomparisons between the benchmark vehicle and the alternative concepts.

• Disassemblyandrecyclingmethodshave been developed for the new materials and joining techniques.

• Objectivecomparisonsofcostandenvironmentalimpact between design options have been performed within a specially developed software suite using a life cycle approach.

SP1 responsibility has been the development of the CAD design concept, together with computer-based performance simulation. The design has followed the package constraints of the Golf to ensure that the design is practical, and to ensure a fair comparison with the benchmark.

Arup’s role has been to work as a team member for the concept design, and it has also been responsible for the computer simulation work to ensure that the stiffness and crash performance are not compromised.

The SP1 body concept development team comprised six car makers (VW, Fiat, Opel, Volvo Technology, Porsche) plus “neutral” engineering experts, Arup and IKA (an Institute for applied automotive research based in Germany). To ensure close integration, the team met regularly in Aachen throughout the design phase.

The timing for the concept development is shown above. Initially three separate concepts were developed. ULBC (Universal Light Body Concept) was a low-risk solution based on known

technologies that could be utilised in a near-future programme. SLBC (Super Light Body Concept) was based on developing technologies and was a higher risk vision for the longer term. The Steel Intensive Body Concept (SIBC) offered a lower cost alternative using the latest steel materials. In 2007, the three concepts were merged to ensure the best solution for each area of the body.

There were three major design loops in 2007. Each loop incorporated finite element analysis (FEA) of the vehicle for crash and stiffness as the guide for structural modifications. A final design loop was performed in early 2008.

Finite Element Analysis

Arup performed the FEA primarily at their Midlands Campus office in UK. However, significant assistance was also provided by Arup staff in Nagoya, Japan and Detroit, USA. The ability to perform the work at different locations aided the programme timing, especially in the short duration design loops.

The automobile industry’s favoured FEA software for modal frequency analysis is usually MSC/Nastran, and for crash prediction, LS-DYNA. The input models for the two packages are different. For SLC, it was decided to minimise design loop timing and costs by running the modal analyses using the implicit analysis option in LS-DYNA. This had the advantage that only one FEA model was meshed and assembled, and also allowed the use of the Arup-developed OASYS software environment for automatic model-checking and results post-processing. The FEA process, from meshing to report, was successfully reduced to four weeks using this approach.

Each design iteration was analysed for four loading cases, and the results compared to the benchmark structure:

• Modalfrequencywithtargetsforglobal body torsion and bending frequencies

• FrontalimpactusingtheEuroNCAPtestprocedure

- 40% offset impact into a deformable barrier at 64kph.

• SideimpactusingthetwoEuroNCAP test procedures

- 50kph side impact by a deformable barrier mounted on a 950kg trolley

- 29kph side impact of the vehicle into a rigid 254mm diameter pole

• RearimpactusingtheUStestprocedure for FMVSS 301

- Rear impact by an 1814kg rigid barrier at 48kph

Results and discussion

As stated above, the three body concepts (Steel Intensive, Universal Light and Super Light) were developed in parallel prior to merging to become the final SLC concept. A summary of the final three concepts is given below.

Steel Intensive Concept

The SIBC succeeded in reducing the body mass by 55kg (20%) at an additional parts cost of 0.19 €/kg saved. This was achieved through extensive use of high strength steels. There was also wide use of tailor-welded blanks. The structural performance was comparable to the benchmark body.

Universal Light Body Concept

The ULBC concept achieved 82kg (29%) body mass reduction. The additional parts cost was 2.69 €/kg saved. Steel was used in the main load paths, and aluminium for structurally less critical panels eg the roof and the front end other than the main front longitudinal rails. The design almost met the weight target and was within the cost target. The combination of materials led to complex technology requirements for the panel joints.

SuperLight Car (SLC)

Figure 1. Concept development timeline Figure 2. Arup’s OASYS LS-DYNA Environment

Feb 05 – Jan 07 Feb – Jul 07

Loop 1 Loop 2

Concept optimisation

Loop 3

Jul – Nov 07 Nov – Dec 07 Jan – Feb 08

Detailing

Steel intensive 2,5€/kg

ULBC 5€/kg

SLBC 10€/kg Final SLCconcept

objective:> 30% < 5€/kg

SLC ConceptPRIMER

LS-DYNA

Analysis Process

Model preparationfor each loadcase

Explicit or Implicitsolution in LS-DYNA

Automatic reportgeneration in Oasys Reporter

REPORTER

T/HIS

D3PLOT

6 | research review

Super Light Body Concept

With this more revolutionary design, a weight saving of 114kg (41%) was achieved, but at an increased cost of 7.61 €/kg saved. SLBC included a sheet magnesium roof and a cast magnesium front suspension tower.

Final SLC Body Concept

Work in 2007 concentrated on combining the best features from the three concepts, and then improving the combination in a total of four design loops. The concept has achieved 101kg (35%) mass reduction at a cost of around 7.80 €/kg saved.

The design is shown opposite and retains the magnesium roof and front suspension tower from SLBC, the fibre-reinforced plastic rear floor, and aluminium outer side panels with high strength steel inner panels.

The structural performance of the design has been analysed in LS-DYNA for the four performance targets described above; modal frequencies, and frontal, side and rear impacts.

Modal performance was comparable with the baseline VW Golf structure.

Views of the final deformed shapes in impact are shown below.

The impact reports were generated automatically from the LS-DYNA output using Arup’s OASYS Reporter software. Reporter was used to extract the important criteria from the data:

• Frontalimpact–accelerationandpassengercompartment intrusion data

• Sideimpact–peakintrusionvelocityanddisplacements, intrusion vertical profile

• Rearimpact–peakintrusions,fuel system integrity

The SLC concept has been created with excellent support from the other technical sub-projects within SLC. The concept is thus robust in terms of its manufacturability for high volume vehicles.

In addition, the life cycle environmental and costing models created for this programme have ensured that the design is both affordable and sustainable.

One body structure plus two front end assemblies are to be manufactured in a simulated production environment. These will be used for dissemination of the learning to parties outside the SLC consortium.

Conclusion

The SuperLight Car programme will successfully meet its target of 30% body mass reduction. This will be achieved with a predicted life cost increase of 7.80 €/kg saved.

Aluminium sheetAluminium cast

Aluminium extrusionSteel

Hot-formed steelMagnesium sheet

Magnesium diecastingGlasfibre thermoplastic

Figure 3. Final SLC Body Concept

Figure 4. Post-impact deformed shapes

References

EU Economic Report. ACEA. Feb 2008

Explanatory Memorandum on the Contexts of the Proposal to “Set Emission Performance Standards for New Passenger Cars as part of the Community’s Integrated Approach to Reduce CO2 Emissions from Light-duty Vehicles”. EC publication 2007/0297 (COD) dated 19th December 2007.

Motor Vehicles in Use. – European Automobile Industry Report. ACEA. May 2007.

The Stern Review on the Economics of Climate Change. HM Treasury, UK. 2006.

Acknowledgments

SLC Consortium – www.superlightcar.com

LS-DYNA is a registered trademark of Livermore Software Technology Corporation

The multi-material approach of SLC has required research and testing of novel joining methods that are applicable to high volume manufacture.

The costs and sustainability of the designs have been monitored using a life cycle approach.

The programme is scheduled for completion in early 2009 and the gained knowledge will be fully disseminated.

SLC will provide practical examples of lightweight vehicle body design that can be utilised in new vehicle programmes. Full utilisation of the lessons could lead to reductions of 6% in European car transport emissions.

EuroNCAP Frontal Offset

EuroNCAP Pole Impact FMVSS 301 Rear Impact

EuroNCAP Side Impact

research review | 7

8 | research review

Author: Carlos Wong

Introduction

It is clear that many countries have elevated the development of wind power to the national energy strategy level. On 9 December 2007, the UK Business Secretary suggested that the installation of up to 7,000 turbines in territorial waters would be sufficient to power all UK homes by the year 2020. In 2006, China implemented the National Regenerated Energy Law and the Finance Department published a government guidance to boost both the wind power generating industry and the wind power implementation strategy.

Wind Farms

There are many factors affecting cost effectiveness of a wind farm; these include annual wind characteristics, the cost structure of electricity and the availability of land. However, the single most important issue is that of connection to the grid. Large wind farms generate power at a similar magnitude to power stations, so it is likely that the output will have to be exported to the grid. Additionally, scaling factors often favour a large farm over a small one.

In the past few years, a number of wind farm projects have commenced in China. Almost all of these have been inland. One exception is the Donghai Wind Farm in Shanghai. This is an offshore wind farm along the Donghai

Bridge, consisting of 50 turbines in three rows at distances of 600 – 2200m from the bridge, with a total installed capacity of 100MW.

Construction cost at the reporting date is RMB 2.112bn or RMB 21.12m per MW, which is about double that of a corresponding land based wind farm.

Table 1 summarises the pros and cons of land based and offshore wind farms. From the table it is clear that the only barrier to offshore wind farms is their high construction and maintenance costs in comparison to inland farms. In view of huge untapped offshore wind power in domestic European seas, the development of offshore wind power has been written into the Berlin Declaration as policy for development of offshore wind power amongst the EU member states.

Review of Sea-Crossing Bridges

Sea-crossing bridges such as the proposed Hong Kong–Zhuhai–Macau Bridge, the Shanghai Donghai Crossing, the Hangzhou Bay Bridge, Qingdao Bay Bridge, the Sunshine Bridge in the US and the Oresund Bridge between Denmark and Sweden are all over 20km in length. Each uses either long span sections or submerged tunnels to provide one or two navigation channels. Approach viaducts have movement joints every

few hundred meters, and a span in the range of 50-80m.

All the fixed crossings currently being built or planned in China consist of a dual carriageway with a least 3 lanes plus a hard shoulder in each direction, and a deck width of 15.25-16m. The approach viaduct has movement joints every 500-700m, and so can be considered as a series of modules between the joints. One or several piers are assigned to take the horizontal longitudinal loads due to wind or seismic activity. These piers provide the ideal locationfor a turbine support.

To avoid power losses, the turbines must be widely spaced in the prevailing wind direction, with a minimum distance of seven times the blade diameter between upwind and downwind turbines.

Offshore Wind Farm on a sea-crossing bridge

The proposed scheme consists of a 70m rotor diameter turbine on each bridge module, which will typically deliver 1.5-2.0MW, with spacing of 500-700m between turbines. See Figure 1.

The turbine is supported on a column set at a distance 60m away from the side of the bridge. The column has its own foundation but is linked to the bridge, forming a rigid frame, and is stabilised with cables. The turbine base is at the level of the elevated deck, thus giving an extra 30m of elevation over typical offshore turbines. Greater wind speeds are found at these higher elevations.

Highlights of the scheme are:

1. The integrated foundation gives additional strength against horizontal loads such as wind and earthquake loads. Substantial savings on installation costs can be achieved if it is constructed at the same time as the bridge.

2. The bridge structure can be used for access during turbine installation and maintenance, avoiding the costs of a marine operation.

3. Electricity cables can be laid inside or alongside the bridge, avoiding the expense of marine cables.

4. A weak point can be built into the turbine column, so that failure causes collapse into the sea rather than onto the bridge.

5. The turbine support also functions as ship impact protection for the bridge.

An alternative would be to use smaller turbine so that it fits in a span, with a rotor diameter of around 10m. In this arrangement, a turbine is fitted into every pier, offset from the bridge structure by 10m. Support can be provided with a simple cantilever, with no separate foundation. However, since the wind energy captured is proportional to the turbine sweeping area, the use of a large number of smaller turbines would not necessarily always match the capacity of a single large turbine.

Abstract

Global warming has prompted the development of sustainable energy at a faster pace than has ever been seen in human history. Wind energy has been singled out as one of the most promising renewable energy candidates for the near future, and farms with multiple turbines converting wind energy into electricity are in high demand.

This work is the results of continued development of earlier research. It contains two separate innovations:

1. Integration of the bridge structure with wind turbines, forming both a fixed access to the turbine and platform for laying the electricity cable, resulting in an economically viable offshore wind farm.

2. An Innovative Mass Energy Storage Bridge, using Compressed Air Energy Storage (CAES) technology to store large scale energy in the bridge structure in the form of compressed air. This allows regulation of the intermittent output of a renewable energy source, producing a smooth output. Additionally, surplus grid energy can be stored until a demand for it arises.

CAES technology could also be used to store energy within city structures such as the viaducts and flyovers of city ring roads. Depending on the electricity market and the policies that govern the percentage of supply surplus over demand, potential savings in national energy bills, and associated carbon emissions, may amount to between 5% and 10%, a figure that will be of interest to any government.

Sustainable Bridge Structures: Support for Offshore Wind Farms and Energy Storage

research review | 9

Problems with conventional wind turbines

• Themainproblemwithanywindfarmisthatthenature of the output is intermittent, due to the irregularity of wind speeds. In addition, its output load cannot be scheduled to meet peak demand.

• Typicalturbinearedesignedforathresholdwind speed of 3-4 m/s, and a cut-off speed of around 12m/s. Above the cut-off speed, the turbine is designed to shed excessive wind energy. Potential power output is therefore sacrificed to maintain a steady load.

• Theuseofexpensiveelectronicdevicesisnecessary, to ensure the quality of the supply to the grid.

• Theintermittentnatureofwindenergymeansthat it cannot be allowed to dominate grid supply. The maximum levels of 20% of the total demand limit the use of wind as an independent power supply.

Problems with conventional power supply

A margin of supply over predicted demand has to be maintained for grid stability. In the intensively competitive electricity market of North America, this margin is set at minimum of 7% [Current Energy]. In a closed market like Hong Kong, where the return is calculated as percentage of the assets, there is no incentive to aim for a tight margin.

The variation in a typical day’s supply and demand curve for electricity creates a problem: since the generated electricity cannot be stored, it is wasted if it is not consumed. Mass energy storage would solve this problem, allowing surplus energy in the grid to be stored and then released back into the grid when demand outstrips supply.

Mass energy storage

Currently, there are only two Compressed Air Energy Storage (CAES) plants in operation: the Huntorf plant in Germany (1978), and the McIntosh plant in Alabama, USA (1991) [DOE, Compressed at Energy Storage]. These are used to store surplus energy, returning it to the grid during periods of peak demand. When grid demand is low, electric motors are used to drive compressors, which pressurise air to 65-80 bar and store it in sealed underground natural caverns. At the same time, heat is extracted from the air.

When demand is greater than supply, the compressed air is used to generate electricity. It is extracted from the cavern, and preheated in the recuperator, which reuses the heat energy extracted in the compression cycle. This is then mixed with a small quantity of oil or gas, and burned in a combustor. The hot gas from the combustor expands in the turbine to drive power generators. In a conventional gas turbine, 2/3 of the energy input is used in the compressor. Therefore, in production mode, a CEAS plant can save 2/3 of the energy.

Figure 1. Image of the wind farm on the bridge

A report on the economic impact of CAES on wind power in three US states concludes that it “significantly improves the delivery profile of renewable energy to grid” [Septimus van der Linden, 2006].

The use of bridge structures as mass energy storage

The proposed installation of a wind farm on a sea-crossing bridge has the benefit of integrating the two independent structures into one single entity and thus achieving economic saving. The use of the CAES technology to store energy in the bridge body itself can provide further benefits, as it can be used to smooth out the peaks and troughs of demand.

Studies of state-wide energy supply in the USA suggest that if suitable caverns cannot be identified for energy storage, buried steel pipes could be used as an alternative. In this proposal, incorporating storage pipes into the structure of the bridge elements delivers two benefits from a single element.

With high storage pressures of 40-80bar, only steel pipes or carbon fibre bags can be considered. Pressure changes due to heat lost or gained to the atmosphere must also be included in the design air pressure load.

Two case studies of potential pipe configurations have been developed to demonstrate the potential for compressed air storage in bridge structures.

Both are composed of steel pipes acting as longitudinal beams, in composite action with a concrete deck with an 80m+80m span.

Factors Land Based Wind Farm Offshore Wind Farm

Land Large land area required with typical turbine grid at 500-700m spacing. The land value will be affected as people are reluctant to live close to a wind farm.

No land required. There will be some effects on local fishing industry.

Wind Lower wind speed. Higher turbulence due to uneven terrain.

Higher wind speed. Open sea conditions produce lower turbulence.

Power output Lower due to lower wind speed and higher turbulence.

Higher due to higher wind speed and lower turbulence.

Noise More noise due to higher turbulence causing vibration of parts.

Less noise due to lower turbulence.

Community objection Complaints from people living in the proximity of a wind turbine often relating to noise and shadow flicker.

There are no permanent residents living in the proximity of the wind turbine.

Birds Equal chances of bird collision, mitigation measures possible.

Equal chances of bird collision, mitigation measures possible.

Construction cost Lower, construction on land. Higher, special construction plants are needed, especially heavy lifting vessels.

Maintenance cost Lower, site is accessible by vehicle and cable can be laid underground or via pylons

Higher, site is only accessible by special marine craft. Expensive marine cable is necessary.

Service life Shorter, about 20 years due to high turbulence causing vibration, leading to early fatigue failure.

Longer, about 25 years due to lower turbulence leading to longer fatigue life.

Environment Turbine occupies arable land. Turbine and marine cable may affect marine life during construction stage.

Table 1. Comparison between land based and offshore wind farms

Sustainable Bridge Structures: Support for Offshore Wind Farms and Energy Storage

10 | research review

Analysis of the case studies shows the advantage of this proposal. The hoop stress induced by the compressed air will be double the magnitude of the induced longitudinal tensile stress. As the hoop stress approaches yield stress at the Ultimate Limit State, the longitudinal stress will still only be half way to yield. Since the pipe thickness must be sized to withstand the air pressure, the available tensile force in the longitudinal direction will be sufficient to support the live and dead bridge loads.

In case study 1 (shown in Figure 2), using 32mm thick pipe shows satisfactory results under loads of internal pressure 80 bars and 45 units HA+HB loads. The von Misses stress at the joint and at critical locations is less than the codified allowable stress.

Case study 2 is shown in Figure 3. A large (2.1m) diameter pipe replaces the two 1.5m pipes in Case study 1. Two webs are attached to the sides of the pipe, which act as the bottom flange of the deck. Pipe thickness is 36mm for a simply supported span of 80m.

Comparison of the two case studies is shown in Table 2. The self weight per square meter for case 2 is close to the normal plate girder solution, suggesting only a marginal increase in cost over a standard option. The deck for case 1 is about twice the weight, but it also has about twice the storage capacity. The most economical solution will balance these two factors.

Costing the proposed scheme for the HKZMB bridge

The combined scheme was costed for the proposed 32km Hong Kong–Zhuhai–Macau Bridge. The proposals include the installation of 53 2MW wind turbines giving an installed capacity of 106MW, with storage in modules between the movement joints (500-700m apart). Indicative financial results are summarised in Table 2. The energy saved or reused in today’s Hong Kong tariff is in the range of US$95 to 162m, depending how many compressed air pipes are installed.

Conclusion

Incorporating Wind Farms into Sea-crossing Bridges

The integration of bridge structure and wind turbine boosts the economic returns of a wind farm scheme. The bridge provides a number of

Item Case Study 1 Case Study 2 Remarks

Main parts 10x1.5m diameter Pipes of t=32mm thick with 2 in a stack-up position

3x2.1m diameter Pipes of t=36mm thick as bottom flange of the deck

Two carriageways of 15.5m wide

Storage capacity at 80 bars

78kWh/m 46kWh/m Per m run along deck

Self weight of main parts

714kg/m2 358kg/m2 Based on a bridge width of 31m

Air Volume 777,544m³ 451,196m³ Based on 22km length of dual carriageway

Stored Energy 1728MWh 1016MWh Compressed to 80 bar

Hour of energy produced by a 103MW wind farm

54 hours 32 hours Based on a capacity factor 0.3 for the wind farm

Financial benefit of the stored energy if the stored-release cycle is 12 hours

1,261,440MWh

US$162m p.a.

741,680MWh

US$95m p.a.

Based on a electricity tariff of Hong Kong US$0. 128/kWh

Figure 2. Case Study 1-Medium diameter pipes

benefits to the wind farm, including; foundation support, fixed access for construction and maintenance crews and cables, and elevation of the turbine. The technology is conventional; the innovation lies in the combination of the two elements, and involves structural design of both the bridge and the turbine tower.

Energy Storage Bridges

The use of compressed air to store energy is not a new technology. The innovation is in the use of an existing structure as storage medium, and in the use of large-scale energy storage to smooth out variations in both demand, and in the intermittent supply from renewable sources such as a wind farm. This has the potential to reduce both energy wastage and greenhouse gas emission.

The case studies for the Energy Storage Deck show a modest increase in steel weight over a traditional steel plate girder deck. This will lead to economic benefits for the investor, who could repay his investment using the profits from saving surplus energy. It will also be welcomed by environmentalists, and by politicians who wish to show support for environment issues. The energy storage concept should be prioritized in any major city planning, particularly in this energy hungry era dominated by environmental issues.

Table 2. Comparison of Case studies 1 and 2

References

Berlin Declaration, third European Policy Workshop on Offshore Wind Power Development, Federal Ministry for the Environment, Nature Conservation and Nuclear Safety, Feb 2007

Carlos Wong, David Xiong, Livia Wong, “Investigation into the viability of an offshore wind farm along the proposed Hong Kong – Zhuhai – Macau Bridge”, 3rd Symposium on Sustainable Development of Guangdong, Hong Kong and Macau, 23-24 Nov, 2006, Macau

Carlos Wong, David Xiong, Livia Wong, “Utilizing sea-crossing bridges for wind farm erection and its body for wind energy storage”, Offshore Wind Farm Technology Workshop, China Wind Energy Association (CWEA), May 2007, Beijing

Current Energy: www.currentenergy.lbl.gov/

“Delivering Offshore Wind Power in Europe – Policy Recommendation for Large Scale Development of Offshore Wind Power in Europe by 2020”, European Wind Energy Association, 2007

DOE, Compressed at Energy Storage: www.eere.energy.gov/de/cs_energy_storage.htm/#compressed_air.html

“Iowa Identifies Site for Underground Compressed Air Energy Storage Facility”, Energy Efficiency and Renewable Energy, U.S. Department of Energy

Ridge Energy Storage and Grid Services L.P., “The Economic Impact of CAES on Wind in TX, OK, and NM”, Texas State Energy Conservation Office, 27 June 2005

Septimus van der Linden, “Bulk energy storage potential in the USA, current developments and future prospects”, ScienceDirect, Energy 31 (2006) 3446-3457

United Nations Climate Change Conference in Bali, 2007

Wind Turbines Expansion, December 2007

www.envir.cn/an/20060314/]

Figure 3. Case Study 2: Large diameter pipes

research review | 11

12 | research review

Author: Nick O’Riordan, Chris Chau, Kenichi Soga, Duncan Nicholson and Toru Inui

Introduction

There are concerns regarding the rate at which the human population is extracting resources from the Earth, and emitting pollution and wastes to the environment. This has raised issues of sustainability and efficiency in many sectors. This includes the construction industry, which currently accounts for about 50% of all global resource use. An active field of research has been the study of the amount of energy embodied in residential buildings.

Further research by Chau et al consisted of a study of the embodied energy in several typical retaining walls options for a highway widening project, based on a hypothetical London geotechnical profile. The current study extends that work, by calculating the embodied energy of alternative basement perimeter wall designs and anchoring systems for two existing sites in London. On completion, this study will allow comparisons to be made of the relative environmental performance of some commonly available retaining wall options.

The techniques described below have been applied to a range of infrastructure construction activities, for example tunnelling, railway trackbed, and embankments on soft ground. The bibliography contains the technical reference to these applications.

Embodied energy background

Embodied energy is defined as the total energy (in Joules) that can be attributed to the use of an item or component. For the construction industry, embodied energy includes the energy used in extraction of the raw materials from the earth; the processing of that raw material into finished products; transportation to suppliers and then on

The EEI values used in this study are derived from research by Kiani by simply discarding values which are more than two standard deviations from the mean. From the remaining values, the mean values of each material were used for the calculation; the maximum and minimum values were used for a sensitivity analysis. Table 1 lists the mean and the range of EEI values adopted for relevant materials and fuels.

Calculation Methodology

The first step in evaluating the embodied energy of a construction component is to identify all relevant processes for each stage of the calculation. Figure 1 shows an example flowchart. In this calculation, three types of process are shown, covering materials, installation and transportation energies. Of these, the material energy calculation involves finding the total volume of each material used, calculating its weight, and multiplying this by its EEI value. The transportation energy is that which is required to move all equipment and materials. This is calculated using the litres of fuel consumed by the vehicles multiplied by the respective EEI value for the fuel. The installation energy is calculated by multiplying the amount of fuel and electricity used by the machinery with its EEI value; this stage includes any temporary work required. All three values of the material, transportation and installation energy are then summed to give the total embodied energy. Note that excavation energy was not required in this study but would normally be included.

Abstract

The aim of this study was to quantify the environmental impacts of the basement walls of two commercial buildings sites in London. Four different retaining wall options (based on steel and concrete systems) and a number of anchoring options were designed for each of the sites. The embodied energy and CO2 emissions for each of the alternative were estimated and compared. Results show that there are notable differences in embodied energy between different wall designs.

The results also showed that the use of recycled steel over virgin steel has the potential to significantly reduce the overall embodied energy of the designed wall. In comparison, the difference in the embodied energy of various anchor designs is relatively insignificant, and therefore the practicality of design for a specific site should be the determining factor for selecting the types of anchors to be used.

Comparisons with car emissions and household energy consumption demonstrate that the contribution of construction elements to overall carbon emissions is significant.

Embodied Energy as an Environmental Impact Indicator for Geotechnical Infrastructures

Figure 1. Processes flowchart of EE

Table 1. Embodied Energy Intensity (EEI) for some materials

Materials EEI [MJ/kg] or [MJ/l]

Max Mean Min

Steel (Virgin) 60 38.1 20

Steel (Recycled) 18 11.1 9

Concrete 2 1.8 1.5

Mortar Grout 4 3 2.5

Fuel 41.2 36 35.4

Resources Extraction

Manufacturing

Resources Extraction

ManufacturingTemporary Works

Machinery Assembly

Material Energy

Installation Energy

Excavation

Landfill

Excavation Energy

Excavation

Demolish

Reuse, Recycle/Dispose

Resuming Energy (excluded in study)

Concrete Steel

Transportation energy

to site; the construction process; the demolition and recycling; and the construction and maintenance of any associated temporary works.

Quantifying embodied energy is important because it encompasses associated environmental impacts such as resource depletion and greenhouse gas emission. Research into the relationship between embodied energy and carbon dioxide (the main contributory gas to the greenhouse effect) shows a high correlation; every GJ of embodied energy produces 0.098 tonnes of carbon dioxide . Therefore, although there are no direct environmental impacts associated with embodied energy, the link to carbon dioxide suggests a context for interpreting embodied energy data.

Embodied energy intensity

For this study, the calculation of embodied energy involves the use of published Embodied Energy Intensity (EEI) values. These represent the amount of embodied energy required to produce 1kg of construction material from the point of resource extraction to the end product. Units used for EEI are MJ/kg for solids and MJ/l for liquids.

There has been research into EEI values since 1979 from both the public and private domains. However, certain materials, including steel, exhibit a wide range of values. The variation can be attributed to the different types of steel assessed, assumptions made, and study boundaries drawn during EEI evaluations.

research review | 13

Sheet pile AZ18 Secant pile Steel tubular piles Combi wall

Toe Level -12m -12m -12m -12m

Volume of materials/m

Steel: 0.57m3 Steel: 0.16m3 Concrete: 17.9m3

Steel: 0.42m3 Steel: 0.62m3

1 Level anchors 971.4kN @ +1m - - -

2 Level anchors 384kN @ +3m 320kN @ -3m

470kN @ +1m 320kN @ -2.5m

420kN @ 2.25m 605kN @ -3m

348kN @ 2.25m 640kN @ -3m

Site 1 was chosen to investigate a generic section. It is far from other underground structures and there is sufficient room for anchors. Accordingly, four standard retaining wall options were considered for this site: sheet pile, secant pile, steel tubular piles and contiguous walls. A two-level anchorage design was considered for each option. Additionally, the sheet pile option was used as an example to further investigate the embodied energy of the six anchoring systems most commonly used in industry: three standard sizing of anchors (0.12m, 0.15m and 0.20m in diameter), arranged in either one or two levels. Appropriate lengths and the number of steel bars for each anchoring system were designed, and the resulting embodied energies calculated.

Site 2 has a location which is close to the river, with an ageing canal wall that requires either strengthening or replacement. Three propped options and a cantilever option were considered for this location. Excavations for the propped options were completed by tying the props across to the existing canal wall using a sheet pile or two

options for diaphragm walls all with their toe levels at around 12m. For the cantilevered option considered, the toe level was at approximately 18m resulting in a 23m wall.

Results

Tables 2 and 3 show the configurations and sizes of the wall designs, the volume of materials used and design forces for anchors for Site 1 and 2 respectively.

Figure 2 and Figure 3 show the overall embodied energy of the wall designs described in Table 2 and Table 3 respectively. These values are given in terms of Joules per meter run, for ease of comparison.

The total energy consumed in a meter run of wall on average is approximately 100GJ/m. This is around 1.6 times the annual average UK household energy consumption from 2005. Comparisons within the same site show that the maximum difference in embodied energy between

Sites

Two sites, situated in central London close to the River Thames, were used for this study. Chau et al (2007) provide a full description of the ground conditions and design parameters for these sites .

The proposed building at Site 1 is 40 storeys, 150m in height, with a three level basement at -6mOD. Ground level is at +5m, resulting in an expected 11m dig, with the toe of the retaining wall at approximately 13m. This site has a layer of made ground and terrace gravel overlaying the Lambeth clay and sand, and an underlying layer of Thanet sand.

At Site 2, the proposed development involves six commercial buildings varying between 6 and 50 storeys in height, and seven residential buildings varying between 30 and 50 storeys. This study assumes an average building height of 40 storeys, to allow comparison with Site 1. The total excavation depth is about 6m.

Design Specifications

The retaining walls for basement construction have been designed according to BS 8002 (1994). It has been assumed that the walls will be left in place at the end of their 120 year design life, that corrosion has been taken into consideration, and that no maintenance will be required during the service life. For UK designs the serviceability requirements are based on lateral wall deflection of less than 50mm during any point of the construction. For the exposed section of retaining wall, the specification includes water tightness according to ICE guidelines (1996), allowing damp conditions but no running water.

The retaining walls were designed using FREW software, developed by OASYS. This study considered the wall deflections under serviceability limit state (SLS) conditions. Partial factors were applied to soil parameters to assess the walls for ultimate limit state (ULS) conditions. The different wall and support systems were analysed under both conditions, in order to derive the most onerous. Corrosion allowances were made by increasing the steel wall thickness.

Types of retaining walls considered in this study

This study considered a basement perimeter wall designed for the car park of a commercial building. Although in practice extra layers are sometimes added to give an aesthetic finish, for the purposes of this study these have been excluded. As with all large basement projects, different design options are required around the perimeter of the wall. This is due to the varying profiles, surrounding structures and water conditions.

Table 2. Site 1 pile design configuration, materials and required anchor forces

Table 3. Site 2 pile design configuration, materials used and required anchor forces

Figure 2. EE of Site 1 basement wall designs (per metre run)

Figure 3. EE of Site 2 basement wall designs (per metre run)

Embodied Energy as an Environmental Impact Indicator for Geotechnical Infrastructures

Sheet pile AZ34 Propped Diaphragm 1 Propped Diaphragm 2 Cantilever Diaphragm

Toe Level (length) -12m (17.4m) -12m (17.4m) -12m (17.4m) -18m (23.4m)

Sizing (width) AZ34 800mm 1000mm 1500mm

Volume of materials/m

Steel: 0.42m3 Steel: 0.13m3 Concrete: 13.8m3

Steel: 0.13m3 Concrete: 17.3m3

Steel: 0.30m3 Concrete: 34.8m3

Strut force 239kN @ +1m 239kN @ +1m 260kN @ +1m -

450

400

350

300

250

200

150

100

50

0Sheet Pile Propped D1 Propped D2 D/Cantilever

Em

bo

die

d E

nerg

y [G

J/m

]Virgin SteelRecycled SteelConcreteTransportInstallation1 Row of Anchors

350

300

250

200

150

100

50

0Sheet Pile Secant Pile TubularPile Contiguous

Em

bo

die

d E

nerg

y [G

J/m

]

Virgin SteelRecycled SteelConcreteTransportInstallation1 Row of Anchors

14 | research review

the most energy-consuming and energy-efficient walls is approximately 250GJ/m. Given that an average 250m² commercial building would have an perimeter wall of approximately 200m in length, this difference in embodied energy would amount to 50TJ, or 785 annual household equivalents.

This shows that careful choice of retaining wall designs and materials can contribute significantly to reductions in the overall environmental impacts of a development.

Secondly, material energy is the greatest contributor to the overall embodied energy values in all cases; the proportion of energy is much greater than that of the transportation and installation energy combined. This result is consistent with that found by Chau et al in the design of retaining walls. This shows that the choice of material is important, and that where possible, recycled steel should always be used due to the large reduction in embodied energy in comparison to virgin steel. However, when purchasing steel, clients do not always have a choice between the purchase of virgin and recycled steel. This presents a difficulty in drawing conclusive remarks about the embodied energy of the walls; therefore for completeness values for both virgin and recycled steels are presented in the results.

Thirdly, comparing across designs built from the same steel, results from Site 1 suggests that steel based designs such as sheet pile, steel tubular piles and combi-walls built purely from recycled steel embody significantly less energy than other retaining wall options.

This result is reversed when virgin steel is used. This again shows that the proportion of recycled steel in the selected source of steel is very important in establishing the embodied energy and carbon footprint of the various designs.

Finally, combined results from the two sites show that the cantilever diaphragm wall system embodies far more energy than any of the propped systems. This is because a cantilever system will always require foundations at a lower level, resulting in the use of significantly more materials and therefore embodied energy.

Figure 4 shows the comparison of embodied energy for the anchorage options embodied energy comparison of the different anchoring systems using the same sheet pile wall from Site 1.

Comparatively, designs with two rows of anchors rather than one row consume less energy. This is because the required anchoring force for a one row design is larger than the sum of the required forces from the two rows of anchors. Therefore, single row anchors must be longer, resulting in more use of materials. However, the results show that on average, the anchoring systems consume approximately 25% of the total energy. In comparison to the overall magnitude of embedded energy, the difference between the one or two rows of anchors is relatively small. Therefore, this choice should be based on the practicability of the situation rather than the environmental impacts

Carbon Emissions

In a separate study, the CO2 emissions of all wall options included in this study were evaluated, using a similar methodology to that used for embodied energy. The main difference is that instead of EEI, CO2 emission factors published by Architectural Institute of Japan and Japan Society for Civil Engineering were used. The CO2 emissions in general are strongly correlated to the embodied energy values, with the average for all the walls being approximately 10 to 15tCO2/m.

According to the UK Car Fuel Data Organisation, an average 2.0L engine family car emits approximately 200gCO2/km. Therefore, emissions due to a typical wall such as those investigated in this study would be equivalent to running a family car for 50,000-75,000km.

Conclusion

The study was focussed on the design of various retaining walls for large basement construction at two riverside sites in London under a chosen set of design criteria. Their embodied energies were computed to assess their relative environmental impacts.

Results show that recycled steel wall systems generally consume less embodied energy and emit less CO2 than equivalent concrete wall systems. Comparing across materials, there is significant difference between designs built with virgin steel and with recycled steel. The difference in anchor designs is generally insignificant in terms of differences in embodied energy.

Comparison with other carbon dioxide sources, such as car use and household energy consumption, shows that the embodied energy in construction components makes a significant contribution to overall carbon emissions.

Bibliography

Architectural Institute of Japan (2003): Guideline on the lifecycle assessment of buildings, 2nd edition, Architectural Institute of Japan

BS 8002 (1994): Code of practice for earth retaining structures, British Standards Institution.

Chau et al (2007): see acknowledgments below

Chau, C, Nicholson, D, and Soga, K (2006): ‘Comparison of Embodied Energy of Four Different Retaining Wall Systems’, Reuse of Foundation for Urban Sites, Proceedings of International Conference, A.D. Butcher, J.J.M Powell, H.D. Skinner (eds) HIS Press, EP73, pp277-285.

CSIRO (2007) www.cmmt.csiro.au

Edwards, B. and Hyett, P. (2002). Rough Guide to Sustainability, RIBA Publications

Institution of Civil Engineers (1996): Specification for Pile and Embedded Retaining Walls, Thomas Telford.

Japan Society of Civil Engineers (1997): Report on lifecycle assessment of environmental impact, JSCE Committee on LCA of Environmental Impact, (in Japanese).

Kiani, M. (2006): The Whole Life Environmental Impact of Glass within Glazed Commercial Building Envelopes, PhD Thesis, University of Brighton.

Lawson (1996) ‘Building materials, energy and the environment: towards ecologically sustainable development’, RAIA, Canberra.

National Statistics and Defra (2006): The Environment in Your Pocket 2006

O’Riordan NJ (2006) Comparisons of ballasted track and slab track for high speed railways: predictions, measurements and the use of embodied energy calculations to inform the choice of trackform. Conference on high speed railways. FEUP, Porto.

O’Riordan NJ (2007) Technical sustainability of construction on soft ground. Conference on Soft Ground Engineering, Portlaise, Ireland. Engineers Ireland (IEI).

UK Car fuel Data Organisation (2007) http://www.vcacarfueldata.org.uk/

Workman R and Soga K(2004) Embodied energy of tunnel construction. CUED Technical report. Cambridge

Figure 4. Site 1 Sheet Pile EE with various anchoring systems

Acknowledgments

This is an amended version of the paper ‘Embodied energy as an enviornmental impact indicator for Geotechnical Infrastuctures’ by Chau, Soga, Nicholson O’Riordan and Inui, presented at ASCE Geocongress 2007, New Orleans. Chris Chau is a CASE research student, jointly funded by Arup and EPSRC, working under the academic supervision of Prof Kenichi Soga, Cambridge University.

1 Anchor – 0.12m

1 Anchor – 0.16m

1 Anchor – 0.20m

2 Anchor – 0.12m

2 Anchor – 0.16m

2 Anchor – 0.20m

Embodied Energy (GJ)

0 10 20 30 40 50 60 70 80

EE of Wall

Material of Anchor

Transport of Anchor

Installation of Anchor

research review | 15

16 | research review

Authors: David Hetherington, Patrick Kuhn, Mutlu Ucuncu and Matt Simpson

Introduction

Water engineers often require estimates of various fluid flow parameters (eg velocity, shear stress, water surface elevation) on and around infrastructure components for various reasons. Historically, these estimates have been made using physical models which as well as being expensive and time-consuming to create, are inherently problematic due to fluid behaviour varying disproportionately with changing scale.

Computational Fluid Dynamics (CFD) software packages provide a means of mathematically modelling complex flows within and around physical objects and scenes. Such approaches allow fluid motion to be modelled for various mediums including water, ice, air and heat. The accuracy of model results is highly-influenced by how well the physical environment is represented digitally within the software environment (the surface boundary conditions).

Conventionally, the spatial data incorporated into CFD models are based upon more traditional survey methods including Electronic Distance Measurement (EDM) theodolite, Global Positioning (GPS), and photogrammetric survey systems. All of these measurement techniques have their own advantages and disadvantages in terms of point accuracy, achievable resolution and measurement range. Often a trade-off exists between accuracy (allowing precision and high-resolution) and range/coverage area. Object or scene misrepresentation can also occur due to “shadowed” or unmeasured areas which result in interpolation errors, which can mask small-scale structural variation.

Data quality should be considered based upon how well point cloud data, and potentially the resultant Digital Elevation Model (DEM), represents the actual physical characteristics of a feature or environment. Cooper (1998) explains how it is inappropriate to consider DEMs, and thus the data sets that form them, in terms of their “accuracy”, and that it is better to use surface quality as a measure of appropriateness and representivity. Cooper (1998) continues by describing how terrain surface quality is controlled by data precision, data reliability and the accuracy of individual points. This quality/representitivity is dictated both by the nature of the data that are acquired, which depends upon the selected measurement method, and by how the data are subsequently processed.

In short, the level of data quality is entirely dictated by “what is good enough” for any given project and it is the best-case results (resolution, point accuracy, coverage) for a specific type of measurement technique which will dictate its potential range of appropriate applications (Hetherington, 2008). Medium-long range Terrestrial Lidar (light detection and ranging) Scanning (TLS) techniques now exist which allow objects and scenes to be measured and represented to high degrees of quality. Resultant data sets can be obtained which are of sufficient high-resolution to represent the most intricate scales of topographic variation, over relatively large ranges (up to 1km in a single scan), whilst minimising shadow affects by acquiring scan data from multiple locations and perspectives.

A series of generic best-practice recommendations can be followed when scanning linear hydraulic features, such as river channels, which can ensure that data quality is maximised (after Heritage and Hetherington 2005, 2007), and the best results are often achieved in complex environments when TLS surveys are morphology-focused and driven (Hetherington et al, 2007). The raw point-cloud data collected by TLS survey can be difficult to process and handle due to the large file sizes which can contain huge volumes of data (up to 100m coordinate points in some cases). This has previously been an obstacle in the widespread use of this technique.

TLS data acquisition techniques have the potential to provide CFD models with data that are of sufficient high-quality as to not be a significant source of error in flow prediction. Before this is possible, the raw point cloud scan data must first be conditioned to make it appropriate for the model framework. This is a difficult process in extremely complex environments (such as natural river beds) as data have to be organised into an “airtight” framework without losing the detail that is represented within the TLS data. Conventionally, if model realism is to be maximised after calibration, then detail that is poorly represented spatially in CFD models would be accounted for during the roughness parameterisation of the model framework. Theoretically, if spatial measurement and structural representation is detailed enough then small-scale roughness elements, and their affects on flow, will be represented in the CFD model as appose to being accounted for theoretically within a roughness value.

This situation would positively impact on the modelling process by reducing the potential impact of subjective roughness parameterisation on modelling results. Available software capability and hardware processing power have been long-standing restricting factors in CFD modelling,

High-Resolution 3D Measurement and 3D Computational Fluid Dynamics Modelling of Reservoir Spillways

Figure 1. The Riegl LMSZ420i terrestrial laser scanner

Abstract

Computational Fluid Dynamics (CFD) software packages provide a means of mathematically modelling complex flows within and around physical objects and scenes. The accuracy of model results is highly-influenced by how well the physical environment is represented digitally within the software environment (the surface boundary conditions). Terrestrial Laser Scanning (TLS) data acquisition techniques have the potential to provide CFD models with data that are of sufficient high-quality as to not be a significant source of error in flow prediction. Available software capability and hardware processing power have been long-standing restricting factors in CFD modelling, which have meant that channels and structures often have to be under-represented in order for simulations to run successfully. Recent developments have meant that larger, higher-quality digital frameworks can be incorporated into CFD models, which in turn allow better quality CFD models to be developed. The main aim of this work was to develop a methodology by which scenes, objects and structures could be measured using TLS, represented in a spatial model and then used as the spatial basis for stable CFD modelling. This would allow a series of repeatable stages to be created which would allow future projects requiring detailed estimates of flow to be conducted in an efficient and consistent way using state-of-the-art techniques throughout.

research review | 17

which have meant that channels and structures often have to be under-represented in order for simulations to run successfully.

Recent developments have meant that larger, higher-quality digital frameworks can be incorporated into CFD models, which in turn allow better quality CFD models to be developed.

The main aim of this work was to develop a methodology by which scenes, objects and structures could be measured using TLS, represented in a spatial model and then used as the spatial basis for stable CFD modelling. This would allow a series of repeatable stages to be created which would allow future projects requiring detailed estimates of flow to be conducted in an efficient and consistent way using state-of-the-art techniques throughout.

Study Site

The Ulley Reservoir is situated in Ulley Country Park, South Yorkshire (NGR: SK 453 877). It is 9km to the east of Sheffield and 5.5km to the south-east of Rotherham. In June 2007 damage was caused to the reservoir embankment during a storm event which resulted in the M1 motorway being closed for reasons relating to public safety. The Spillway that has been focused upon during this study is the main spillway for this reservoir.

Materials and Methods

A terrestrial lidar survey was conducted over one day in February 2008 using a Reigl LMS-Z420i: with integrated digital camera (see Figure 1). The survey was conducted in a morphology-focused manner and scans were taken from 40 different positions in order to ensure that surface shadowing did not occur. Each scan contained approximately 1 million measurement points. After scanning the individual scan data had to be registered into one common coordinate system, this was conducted using the I-Site software package using iterative pattern-matching techniques (as shown in figures 2a and b). Once registered, the scan model data then had to be semi-manually filtered (again using I-Site) in order to remove any unnecessary data (eg points in the distance and local vegetation).

The final filtered point-cloud scan data is shown in Figure 2c. A final filtered point-cloud processing step was to filter the data so that flat surfaces were not over-represented, and complex (topographically-variable) surfaces retained their structure digitally. This was done in an automated fashion in I-Site after setting a local topographic tolerance threshold, which removed topographically similar local points.

The processed point-cloud model was then imported in to the Rhino (version 4) software package using the Pointtools4Rhino transfer module so that it could be converted into a connected network mesh model. In Rhino the pre-filtered point cloud model was manually “traced” using 3D snap lines which represented the structure of the spillway (steps, curves, walls etc). Snap lines were extrapolated to a corner where data had been masked in the field by in-channel leaf litter. Structurally complex areas were represented using more intricate snap lines in relation to flatter more uniform areas. In order for this framework model to be compatiable with successful CFD modelling it

had to be ensured that it was airtight, with no overlap of multiple layering. The final Rhino model (as shown in Figure 3) was then exported for use in the CFD aspect of the work.

The CFD modelling aspect of the work was conducted in the Star-CCM+ version 2.10.017 software environment. Star-CCM+ uses the well

Figure 2a. Upstream view of unfiltered point cloud Figure 2b. Downstream view of unfiltered point cloud

Figure 2c. Downstream view of filtered point cloud

Figure 3. The final reservoir spillway mesh model as produced in the Rhino software package.

Figure 4. A section of the spillway as represented in StarCCM+, the grey mesh represents the surface boundary and the brown mesh represents fluid domain (both air and water).

High-Resolution 3D Measurement and 3D Computational Fluid Dynamics Modelling of Reservoir Spillways

18 | research review

References

Cooper M.A.R, 1998, Datums, Coordinates and Differences, In: Landform Monitoring, Modelling and Analysis, Lane S.N, Richards K.S and Chandler J.H (eds), Wiley, Chichester, 21-36.

Heritage G.L and Hetherington D. (2005) The use of high-resolution field laser scanning in mapping surface topography in fluvial systems, Proceedings of symposium S1 (sediment budgets) held during the Seventh IAHS Scientific Assembly at Foz do Iguaçu, Brazil, April 2005. IAHS Publ. 291,

Heritage G.L and Hetherington D, (2007) Towards a protocol for laser scanning in fluvial geomorphology, Earth Surface Processes and Landforms, 32 (1), 66-74

Hetherington D, German S.E, Utteridge M, Cannon D, Chisholm N and Tegzes T, 2007, Accurately representing a complex estuarine environment using terrestrial lidar, RSPsoc Annual Conference Proceedings, 11-14 September 2007, Newcastle Upon Tyne, UK.

Hetherington D, in press for 2008, General data issues, data quality and protocols, In Large A.R.G, Heritage G.L and Charlton M.E (Eds), Laser Scanning for the Environmental Sciences, Blackwells Publishers.

Figure 5. CFD Simulation progression (moving forwards in time a to d) on the main Ulley Dam spillway (at the inlet to the structure).

This is a data inter-operability issue which may impact on projects to differing extents depending on the structure that is to be modelled. Once commenced the CFD simulation took about 1 day in real time to complete. It is important to note that in future, run times will depend on the complexity of the physics (ie more advanced turbulence models, or simulations including particles, will increase the run time). The sequence of images shown in Figure 5 illustrate how the flow structure developed during the CFD simulation of the main spillway at the Ulley Dam.

Conclusions

A methodology has been developed which enables Arup to rapidly capture the geometry of complex hydraulic structures in a high-quality, digital format that can be imported into a CFD environment. This allows the hydraulic performance of such structures to be analysed in detail, and potentially in a more flexible and cost-effective manner than can be achieved using physical modelling approaches. In future, it is anticipated that the entire process of developing a CFD model using TLS data could take approximately 10 days in less complex cases. However, further work is required in order to assess and quantify more fully, the improvements that the use of TLS-acquired data can bring to the CFD modelling process. In particular, the use of TLS will potentially reduce the time spent calibrating models in order to represent channel roughness.

known Volume of Fluid (VOF) approach with High Resolution Interface Capturing Scheme (HRIC). The 3D CAD Rhino model was imported in a STL format in readiness for meshing. In order to ensure that detail was not lost, a model mesh spacing equivalent to the smallest levels of structural variation in the Rhino model was used in the CFD software. Prism layers were also defined so that the boundary layer was modelled appropriately, which would in turn contribute to a stable modelling schedule. A hexahedral mesh shape was used as this type of mesh is successfully used in a wide range of free-surface flow simulations in the literature. The “sky” boundary was defined to represent the boundaries of the CFD domain. This boundary was far away from the area of interest and was meshed at a lower degree of detail in this instance to reduce the computational cost (see Figure 4)

Pressure boundaries were defined at water inlet and outlet specifying the water height and corresponding total pressure at the inlet and static pressure at the outlet. This approach was deemed satisfactory in this instance as no secondary analysis of flow results was going to be carried out as part of this project. The channel walls and bed were defined as “wall with no slip” conditions, while the air/structure surface interface layer was defined as “wall with free slip” conditions. The sky boundary was defined as a pressure boundary which allowed air to enter and leave the domain (ie a theoretical unbounded system). This is a correct way to model, however this could have proven to be a source of model instability. In this instance a free-slip wall condition was used in the simulations since the air flow through the sky boundary would not have a mathematical affect on the free-surface behaviour. A RANS model, k-epsilon turbulence model with wall functions was used to predict turbulence patterns and losses. The selected time step (0.125 seconds) was kept small to maximise the likelihood of simulation stability, whilst being large enough for reasonable run times. It should be noted that for detailed resolution of the free-surface motion, smaller time step sizes are required. The total simulated-time duration was 90 seconds.

Results and Discussion

The process described in this paper should be transferable to many other scenarios when predictions of complex flows are required. Major challenges were faced during this work when trying to mesh the spillway channel appropriately in StarCCM+. This was partly due to repairs that needed to be done to the model framework after meshes were created to different degrees of detail. Lessons have been learnt from this process meaning that this will be a more efficient process in future. Although the digital framework data from Rhino were very clean, after exporting it to STL format, further correcting was required (especially to the curved surfaces).

A

Iteration = 400Timestep = 20Solution = 5(s)

B

Iteration = 600Timestep = 30Solution = 7.5(s)

C

Iteration = 800Timestep = 40Solution = 10(s)

D

Iteration = 1000Timestep = 50Solution = 12.5(s)

research review | 19

20 | research review

Author: Lei Yang

Introduction

Arup Water has multi-disciplinary design capabilities in water, wastewater and stormwater practices. This is underpinned by recruiting academics from world class research centres in universities, which promotes excellence in sustainable water resources management. This is true in their water quality work where Arup Water uses state-of-the-art numerical modelling tools in their approach to water quality assessment and forecasting.

The Usk modelling was undertaken using MIKE software (flexible mesh system) with an integrated 1-D and 2-D modelling process encompassing: i) the sediment transport processes to provide predictions of suspended sediment concentrations ii) the dissolved oxygen transformation processes to trace dissolved oxygen concentrations by modelling the changes from sediment oxygen consumptions. The Usk modelling has successfully applied the concepts of sediment-linked water quality modelling developed from earlier research work by Dr Lei Yang working in Cardiff University. The fine grid modelling domain is shown in Figures 1 and 2. The hydrodynamic character (velocity distributions) around the proposed piers is shown in Figure 3.

Method

Dissolved Oxygen (DO) concentration in the aquatic environment is affected by environmental factors such as the photosynthesis of algae and phytoplankton, the respiration of water plants, animals and bacteria, the oxygen consumptions from nitrification, from COD/BOD degradation and from the sediment oxygen consumptions.

The variation of DO will be dominated by one of these factors depending on the specific local environmental circumstances.

The mass transfer of dissolved oxygen in the Usk estuary follows the general advective-diffusion equation (well-known as ADE to water quality modellers) which gives mathematical descriptions based on i) advection term indicating the mass transportation with the advection of flow, ii) diffusion term describing the mass movements produced along horizontal and vertical dispersion/diffusions by eddies, and iii) reaction term which incorporates all physical, chemical and biological reactions between DO and the other environmental factors.

The River Usk is a tidal river with a Spring range of up to 10m which provides natural re-aeration to keep the river DO concentration generally above 9.0mg/l in winter. However high turbidity from the tides is a characteristic of the river with the potential to deplete oxygen levels when high suspended solids occur. The aim of the DO modelling was to investigate the risk of oxygen sag (or DO depletion) from sediment oxygen consumptions.

Accordingly, the DO modelling comprises two major processes, i) the re-aeration process, ii) the sediment oxygen consumption processes.

The re-aeration process modelled the oxygen exchange from the air to the water driven by the temperature and by salinity dependent saturation oxygen levels. The differential equation for re-aeration is:

where:DO = dissolved oxygen concentration in

a water column, mg/l

kr = re-aeration coefficient, day 1

DOs = saturation concentration of DO, mg/l

A time-varying re-aeration coefficient, which is a function of flow velocity and water depth, was used in this study. The time and grid varying saturation oxygen concentration was determined based upon an empirical equation developed by Weiss R. (1970).

Abstract

Recently, Arup Water carried out an environmental impact modelling study as part of the new M4 project on behalf of the Welsh Assembly Government. A major component was to investigate the water quality in the River Usk.

The River Usk is part of a system containing the world’s second largest tidal range which has a complex hydrodynamic character which in turn makes for challenging water quality modelling.

The Usk modelling developed from a desk study of existing research literature, to field surveys of hydrodynamics and relevant water quality parameters to numerical modelling. There was regular collaboration between Arup Cardiff office and Professor Binliang Lin from the Hydro-environmental Research Centre, Cardiff University to capture the latest research understanding of complex numerical schemes.

187400

187200

187000

186800

186600

186400

186200

186000

185800

185600

185400

185200

185000

184800

184600

184400

184200

184000

183800

183600

Geographic view

332000 332500 333000

North Boundary

South Boundary

Land Boundary

KeyBoundaries

Figure 2. Fine grids around proposed piers

Figure 3. Velocity distributions around proposed piers

Figure 1. 2D modelling domain with flexible mesh

185750

185650

185550

185450

185350

185250

185150

185050

184950

331900 332000 332100 332200 332300 332400 332500 332600

Spring tide conciding mean annual fluvial flow

Water Quality Modelling of Dissolved Oxygen in River Usk

dDOdt

= kr (DOS – DO) ------ (1)

research review | 21

In this study, the sediment oxygen consumption was considered to be caused primarily by the degradation of organic material attached to the sediments, but not originating from pollution sources ie BOD from combined sewer overflow outfalls. However, future BOD changes produced by wastewater disposal would lead to different sediment consumption, which was beyond the scope of the current study.

The processes of sediment oxygen consumption will depend on DO concentration, sediment concentration and the water temperature. It includes two parts, i) bed sediment oxygen consumption, ii) suspended sediment oxygen consumption. A set of advanced governing equations were used to represent these processes.

These differential equations were inputted into MIKE numerical engine through the MIKE Ecolab template, DO modelling was performed by linking the MIKE hydrodynamic module and sediment transport modelling results together with the Ecolab module. MIKE flexible mesh software uses the advanced numerical solution by applying

a cell-centred finite volume method based on an unstructured grid system with either triangular or quadrilateral elements. Within the MIKE FM hydrodynamic model an approximate Riemann solver (Roe’s scheme, see Roe, 1981) was used to calculate the convective fluxes at the interface of the cells. Second-order spatial accuracy is achieved by employing a linear gradient-reconstruction technique. The average gradients were estimated using the approach by Jawahar and Kamath (2000). To avoid numerical oscillations a second order TVD slope limiter (Van Leer limiter, see Hirch, 1990 and Darwish, 2003) was adopted.

Results and Discussion

One of the study objectives of this stage was to characterize a concept model describing the kinetics of oxygen consumptions in the River Usk where measured field data was used to determine the apparent oxygen consumption coefficient.

Figure 6. The observation stations along the main flow flume

Figure 5. Predicted dissolved oxygen concentrations along main flume in lower reach of River Usk

Main flume ouput points

Geographic view

Point 5Point 6

Point 7

Point 8

Point 9Point 10

Point 4Point 3

Above 17.616.0 - 17.612.5 - 15.010.0 - 12.57.5 - 10.05.0 - 7.52.5 - 5.00.0 - 2.5-2.5 - 0.0Below -2.5Undefined value

Bathymetry

187400

187200

187000

186800

186600

186400

186200

186000

185800

185600

185400

185200

185000

184800

184600

184400

184200

184000

183800

183600332000 332500 333000

Point 2

Point 1

20

18

16

14

12

10

8

6

4

2

000:00 00:00 00:00 00:00 00:00 12-17 12-18 12-19 12-202007-12-16

20

18

16

14

12

10

8

6

4

2

000:00 00:00 00:00 00:00 00:00 12-17 12-18 12-19 12-202007-12-16

12.0

11.0

10.0

9.0

8.0

7.0

6.0

5.0

4.000:00 00:00 00:00 00:00 00:00 12-17 12-18 12-19 12-202007-12-16

npWQ1: DO, predicted [mg/l]

npWQ1: DO, measured [mg/l]

npWQ2: DO, predicted [mg/l]

npWQ2: DO, measured [mg/l]

Point 1: DO, dissolved oxygen [mg/l]Point 3: DO, dissolved oxygen [mg/l]Point 5: DO, dissolved oxygen [mg/l]Point 7: DO, dissolved oxygen [mg/l]Point 9: DO, dissolved oxygen [mg/l]

Initial results indicated a good agreement between the field data and the model predictions. Some of the calibrated DO results are given in Figure 4, regarding to two site surveys measured during a neap tide in December 2007.

Literature shows natural bed sediment oxygen demand is typically in the range 0.2 – 1.5gO2/m2/day. However as an urban river which generally receives effluents from combined sewer overflows, the sediment-attached nutrient content could increase its oxygen demand by up to 5.0gO2/m2/day for the sediment consumption.

The Usk DO calibration was started from a point of 0.5gO2/m2/day and found the calibrated sediment oxygen consumption coefficient to be 1.2 – 2.5gO2/m2/day. Based on an oxygen demand of 2.5gO2/m2/day, the predicted DO concentrations along the river reach are plotted in Figure 5.

The modelling results in Figure 5 indicate i) the DO concentration in lower Usk is generally between 9.5 – 11.0mg/l, which is about 90% oxygen saturation. This is a good modelling result confirming the river’s natural re-aeration process functions efficiently within this brackish water environment; ii) the DO concentration at point 1, which is about 3km upstream of point 9 is on average 0.3mg/l higher than the location of point 9 (see the locations in Figure 6). Downstream, the lowest predicted DO value during neap tide condition is 9.0mg/l.

This confirmed that the landward river had higher DO concentrations than the seaward estuarine water. The Environment Agency (EA) website gives an averaged DO measurement for 2004 to 2006 in the inland river Usk to be 99.35% oxygen saturation. Our modelling results agree with the EA observations.

Water Quality Modelling of Dissolved Oxygen in River Usk

Figure 4. Time series dissolved oxygen concentrations at measured sites

22 | research review

The predicted DO results within the study area are given in Figures 7-10. These show the DO distributions at different surface water levels during a neap tidal cycle.

Whilst DO modelling is not a new area of work, until now much research has focused on existing literature. The task of modelling has often been hampered by a lack of measured historical data coupled with the lack of suitable tools to handle the natural complexity of the local water environment.

The study of water quality, including modelling, remains challenging and continues to demand advanced research and state-of-the-art computational technologies.

The future is for more challenging environmental water management accompanying the climate change which will have bearing on water industry practice. Arup, with its research capabilities in water quality studies, is there to face the challenges.

Collaborators

Arup has established close collaboration with the School of Engineering at Cardiff University in the field of hydro-environmental modelling. An agreement has been recently reached between Arup and the University for the Appointment of Binliang Lin as an Arup Professor in Hydro-environmental Engineering. Professor Lin’s main research interest is in hydrodynamic and water quality modelling, including: flood risk assessment, sediment and heavy metals transport, bathing water quality and marine energy. He has developed hydrodynamic and water quality models for river, estuarine and coastal waters and published 120 papers in journals and international conference proceedings. He has been involved in modelling water quality processes in many estuaries

References

Banks, R.B., Herrera, F.F. Effect of wind and rain on surface re-aeration. J. Env. Eng. Div., ASCE 103, June 1977, pp 489-504.

Darwish, M. S. and Moukalled, F. 2003. TVD schemes for unstructured grids, Int. J. of Heat and Mass Thansfor, 46, pp 599-611.

DHI, MIKE by DHI, WQ templates, Scientific Description, Release 2008.

DHI, MIKE by DHI, MIKE 21 and MIKE 3 Flow Model FM, Hydrodynamic and Transport module, Scientific Mannual, 2008.

Hirsch, C. 1990. Numerical computation of Internal and external Flows, Volume 2: Computional methods for Inviscid and Viscous Flows, Wiley.

Jawahar P. and Kamath, H. 2000. A high-resolusion procedure for Euler and Navier-Stokes computations on unstructured grids. Journal Comp. Physics, Vol. 164, pp165-203.

Jorgensen, S.E. 1979. Handbook of Environmental Data and Ecological Parameters. National society for Ecological Modelling. ISBN 87 87257 16 5.

Kasper, H.F. Denitrification in marine sediments: measurement of capacity and estimate of in situ rate. Appl. Env. Microbiol., Vol.43, No.3, 1982, pp522-527.

Malmgren-Hansen, A., Mortensen, P., Moller, B. Modelling of oxygen depletion in coastal waters. Wat.Sci.Tech., Vol. 17, 1984, pp 967-978.

Roe, P.L. 1981. Approximate Riemann solvers, parameter vectors, and difference schemes. Journal of Comp. Physics. Vol. 43, pp357-372.

Weiss, R. 1970. “The solubility of nitrogen, oxygen, and argon in water and seawater”. Deep-Sea Res. 17: 721-35.

Figure 7. Dissolved oxygen distributions with velocity rectors at mid-ebb, neap tide

Figure 10. Dissolved oxygen distributions with velocity rectors at high water, neap tide

Figure 8. Dissolved oxygen distributions with velocity rectors at low water, neap tide

Figure 9. Dissolved oxygen distributions with velocity rectors at mid-flood, neap tide

Acknowledgments

The author would like to thank the Welsh Assembly Government for their support in carrying out this work. The author also thanks Professor Binliang Lin of Cardiff University who reviewed the overall methodology, Titan. Environmental Surveys Ltd who carried out extensive surveys of the River Usk and also acknowledges the support of colleagues at Arup including David Evans, Jo Atkinson, Pete Wells, Tim Wilkinson, Rosemary Cripps, Sarah Williams and Dan Smith.

and coastal basins in the UK, including: the Severn, Humber, Mersey, Ribble and Thames Estuaries and Carmarthen, Swansea, Cardiff, Morecambe and Irvine Bays. He has also been involved in modelling the morphological process in the Yellow River and Bohai Bay in China.

-187400 - -187200 - -187000 - -186800 - -186600 - -186400 - -186200 - -186000 - -185800 - -185600 - -185400 - -185200 - -185000 - -184800 - -184600 - -184400 - -184200 - -184000 - -183800 - -183600 -

332000 332500 333000

Above 12.011.5 - 12.012.5 - 11.010.0 - 10.59.5 - 10.09.0 - 9.58.5 - 9.08.0 - 8.57.5 - 8.07.0 - 7.56.5 - 7.06.0 - 6.55.5 - 6.05.0 - 5.5Below 5.0Undefined value

DOs, dissolved oxygenconcerntrati [mg/l]

1 m/sGeographic view

-187400 - -187200 - -187000 - -186800 - -186600 - -186400 - -186200 - -186000 - -185800 - -185600 - -185400 - -185200 - -185000 - -184800 - -184600 - -184400 - -184200 - -184000 - -183800 - -183600 -

332000 332500 333000

Above 12.011.5 - 12.012.5 - 11.010.0 - 10.59.5 - 10.09.0 - 9.58.5 - 9.08.0 - 8.57.5 - 8.07.0 - 7.56.5 - 7.06.0 - 6.55.5 - 6.05.0 - 5.5Below 5.0Undefined value

DOs, dissolved oxygenconcerntrati [mg/l]

1 m/s

-187400 - -187200 - -187000 - -186800 - -186600 - -186400 - -186200 - -186000 - -185800 - -185600 - -185400 - -185200 - -185000 - -184800 - -184600 - -184400 - -184200 - -184000 - -183800 - -183600 -

332000 332500 333000

Above 12.011.5 - 12.012.5 - 11.010.0 - 10.59.5 - 10.09.0 - 9.58.5 - 9.08.0 - 8.57.5 - 8.07.0 - 7.56.5 - 7.06.0 - 6.55.5 - 6.05.0 - 5.5Below 5.0Undefined value

DOs, dissolved oxygenconcerntrati [mg/l]

1 m/s

-187400 - -187200 - -187000 - -186800 - -186600 - -186400 - -186200 - -186000 - -185800 - -185600 - -185400 - -185200 - -185000 - -184800 - -184600 - -184400 - -184200 - -184000 - -183800 - -183600 -

332000 332500 333000

Above 12.011.5 - 12.012.5 - 11.010.0 - 10.59.5 - 10.09.0 - 9.58.5 - 9.08.0 - 8.57.5 - 8.07.0 - 7.56.5 - 7.06.0 - 6.55.5 - 6.05.0 - 5.5Below 5.0Undefined value

DOs, dissolved oxygenconcerntrati [mg/l]

1 m/s

Water Quality Modelling of Dissolved Oxygen in River Usk

research review | 23

24 | research review

Author: Derek So

The Idea

Since 2002, we had experienced the efficiency of prefabricated construction. In the Fu Tei residential building project, we designed various structural and non-structural, linear and volumetric elements to allow fast, high-quality and sustainable building construction. We learnt that in order to improve the productivity and product quality, part of the construction processes should be industrialized. That means, some of the building elements should be manufactured in a systematic production chain, under a well-monitored quality control system and follow standardized engineering methodology.

Being the top professional engineer in the industry, we are not satisfied with building standardized/typical houses; we want to produce customized, high-ended, luxury houses that can add value to our product, our client, as well as our company. We believed that once we have designed hundreds of standardized building modules, we can form thousands of unique building design. Each homeowner can select from a huge database with different pre-defined building modules to design his customized dream house.

With this belief in mind, we began to study practical engineering solution on how these pre-defined building modules can be produced, replaced, interchanged, modified and fixed together to provide flexibility in the floor plan layout.

At the same time, we promoted this innovative insight with our business partners and try to make it real.

The Team

Shui On Land, one of the major developers in Mainland China, shares our vision in mass customized housing and commits us to implement this concept in their high-rise luxury apartments. We teamed up with a product manufacturing specialist, Advanced Manufacturing Institute (AMI), of the Hong Kong University of Science and Technology, and a IT specialist with well experienced in architectural design and construction management, Tecton Limited. This objective was to develop a holistic approach in the whole supply chain of residential building development with the adaptation of mass customisation concept that can achieve Responsiveness To the Market (RTM) and to provide renewable, efficient, adaptive and liveable housing to the consumers.

Methodology

The Fundamental Theory

Mass customization is the use of computer-aided manufacturing systems which can involve the consumer in the design process in order to produce customized solution. Rooting from the mass production concept, in our mass customization system, numbers of standard building modules are developed with the flexibility for individual customization. These standard modules, which have interchangeable properties, allow the manufacturer to control the unit cost and therfore the production cost. With a great variety of options, the customer product is no longer uniform, but is innovative and quality-oriented. This mass customization business model has been well established in the computer and vehicle industries, and now, we want to bring it to a new realm, the high-rise residential building market.

The foundation of the RTM approach is the product structure, that is, considering residential units as products. Their commonality is extracted using the Product Family Architecture (PFA) approach for developing building modules. The underpinning design logics are deducted for configuration, and individual variant products can then be configured for individual customer to form a home which is structurally stable, durable, aesthetically pleasing and flexible to the change in homeowner’s lifestyle.

The backbone of the RTM is a computer-aided manufacturing configurator contributed by the real estate developer, architect, engineer, contractor, manufacturer and the consumer. It is a database in Building Information Model (BIM) with sophisticated kits of building modules and hardware for developing a residential building. It is also a parametric design engine for analysis; pre-defined design logics based on real estate developers’ requirements and statutory design criteria/rules is formulated in the engine to govern the design.

Each module/hardware object in the database contains certain parameters for design and rule-checking purposes. During the selection of interchangeable objects, the designer/customer can receive immediate response from the parametric design engine to ensure that the design is feasible and complies with relevant statutory rules.

Abstract

Due to the rapid growth of economy in the East Asia, urban population has risen dramatically in the past few years. Chinese is seeing a great movement of population from the rural interior to cities in the East. It is estimated that more than 55% of Chinese population will live in cities by 2020, 60% by 2030. Some 300m to 400m new inhabitants will leave the countryside to settle in cities in the next 20 years. More than 15bn square metres of housing will be constructed to accommodate these new urban dwellers. The scale of new house building is significant. Therefore there is an urgent need to reformulate the current construction practice in order to boost up the productivity. Besides, as the society is wealthier, customers require a higher standard of living; the housing market should therefore quickly respond to homeowner’s diversified expectations. However, the property development is a long process. The volatile change in the market place during the long property development cycle is one of major factors deciding a project win or loss. Therefore, speed to adapt to the market changes has been recognized as a critical success factor.

Moreover, in this sustainability era, both the government and the homeowner require high energy efficient and low-carbon homes. Engineer should design advanced construction material that not only provide structural strength, but also is environmental friendly in the manufacturing and construction process, and provides high heat and sound insulation performance.

To tackle these problems, Arup takes a non-conventional engineering approach.

Figure 1. Mass customisation concept for sustainable housing

Figure 2. RTM Building system

Standardisation(Mass Production)

Customisation

Quality Sustainable Housing

RTMBuilding

SpeedMarket

QualityConstruction

Value

Responsiveness To Market (RTM) Building System for the Residential High-Rise Building Design Process

research review | 25

The RTM system can save lots of time. Compared to traditional development processes where once a design is changed, the project team has to check all the design criteria again manually. Thus, the system allows the developer to make a timely response to the housing market in a systematic manner, whilst meeting the aggressive cost target and quality standards without compromising rules and regulations.

The Research Project

RTM Building System is an integrated building platform (including materials, products and processes) that addresses three conventionally independent performance indices simultaneously, namely Speed, Quality, and Value. It renders the enhancement of design through technical thought, industrial production methods and practical construction methods. It allows Shui On to provide

Figure 3. Customized modules for high-rise building

customized products and service information directly to the consumer at the point of decision. In view of the project complexity, the project was divided into phases. Phase 1 was a research phase, in which we explored industrial techniques and developed an overall product family strategy for Shui On. We built up a framework of the RTM system for precast façade construction design. The scope of the project also covered advanced material technology to improve thermal and acoustic performance and buildability of façade. Simulation and sensitivity analysis tool was also proposed as the decision support systems to facilitate project managers in assessing different design scenarios. Phase 1 was successfully completed in March 2008. The details are described in later paragraphs. At the time of this progress review report, we have just committed to Shui On for the Phase II RTM project.

Results and Discussion

The Phase I RTM

The objective of Phase 1 RTM system was to develop the skeleton of a user-friendly interface that provides a common platform for all stakeholders (the developer, the architect, the design engineer, the project manager, the contractor, and the homeowner) to exchange ideas, and communicate the design. To add value to the BIM, architectural and structural design rules, as well as the construction considerations, were formulated in the system to ensure that all customized product fulfilled the statutory standards and took into account the major construction constrains. State-of-the-art technique that used to be applied in aeroplane manufacturing process, was used allowing the RTM system to also cater for preconstruction management, construction cycle optimization, and supply chain management. The RTM system aims to cover the overall performance, time to market consideration, economical evaluation and total life cycle cost analysis of a development project. Pre-cast façade construction was chosen to verify the application of RTM concept in property development project in this phase of research.

Construction project normally consists of three basic flows (design process, material process and work process). For most participating organizations, these processes repeat from project to project with moderate variations. In Phase I RTM, we aimed to streamline these three basic flows. The RTM system consists of three major components, namely the Building Block Family Platform, the Parametric Design Engine, and the Decision Support Simulation System.

1. RTM Building Block Family Platform consists of a set of Product Family Libraries, Process Family Libraries and Material Libraries. The Product Library consists of edge library, window library, finishing library etc, and these libraries form the model server for the RTM Parametric Design Engine (as described in Point 2 below). Each library contains various design options. These options are façades’ commonalities in existing Shui On’s building blocks, and also include Arup’s innovative design. For instance, we proposed a design of façade connector which allows fast and accurate connection between precast façade and cast in-situ concrete, and provides with sufficient strength and flexibility to cater for seismic loading.

2. The Process Library contains information about the construction productivity and construction procedure provided by Shui On. The information was formulated in the Decision Support Simulation System (as described in Point 3 below) for the process simulation and sensitivity analysis and for decision making facilitation. RTM allowed the designer to choose from a variety of different construction material. Useful information such as material

Figure 4. Example for product family for high-rise building

Interchangable options

Option 1

Option 2

Option 3

Option 4

“Make to stock” prefab componentKitchen Model x-xx

CatalogueOEMS(s)

Tailor-madeOEMS(s)

Tailor-madeOEMS(s)

Different ProductPlatforms

Example of Mass Customized Modules for High-rise Building Design

“Make to order” prefab componentbay-window Model xx-xx-xx

Responsiveness To Market (RTM) Building System for the Residential High-Rise Building Design Process

26 | research review

weight and thermal conductivity were stored in the Material Library. In Phase I of RTM, we introduced an advanced material design of precast façade to improve the thermal and sound insulation, as well as to provide a aesthetically pleasing surface of building; a 200mm thick façade containing a exterior layer of flexcrete (cementitious material consisting special fibres) supporting lightweight cellular concrete. Structural steel frame made up by standardized steel members forms the structure of the façade. The RTM design system optimized the steel frame member size based on the façade size and weight in order to provide the most cost-effective design to the client. The material cost of Arup’s advanced façade design was found compatible with the commonly used sandwich façade (XPS layer filled in-between two layers of concrete panel), while the overall weight of the façade was significantly reduced. As the lifting capacity is always a fixed constrain in a precast construction process, reducing the weight of façade provides greater flexibility in controlling the façade/room size. The bigger the façade implies fewer number of façades; thus, can shorten the construction cycle and the overall project duration.

3. The RTM Parametric Design Engine is a parametric-rule based configurator which embeds the design rules, client’s requirements, product fabrication rules and the construction process rules to provide a convenient design platform for the designer to develop different façade solutions following given design criteria. In Phase I RTM, we successfully implemented thermal, daylight and ventilation rules as per the Chinese statutory codes. All structural criteria for façade design (steel frame design, seismic load, façade connection detailing etc) were catered in the design. Moreover, façade dimensions and weight controlled by the transportation and erection considerations were formulated in the system. During the design stage, different libraries were called out

from the model server for designer’s selection. The designer could check the compliance of rules throughout the façade design process. RTM is a web-based interface, rooted in Revit Architecture 2008.The RTM system also allows a real-time 3-D walkthrough of the precast product. Every design option is represented by a Menu Item Master (MIM) code. The finished precast façade panel can be expressed by a series of MIM code, based on which a shop drawing is generated automatically with all the construction details specified. RTM system can also generate design report showing the design parameters and the rule checking result to facilitate the statutory submission process.

4. Apart from the design function, RTM Decision Support Simulation System can assist Project Managers to perform sensitivity analysis and decision simulation analysis and assess different design options.

In this phase, DSM was proposed as the Decision Making Tool. DSM contains a list of all constituent activities and the corresponding information exchange and dependency patterns. That is, the information flows among different tasks. According to the working sequence, construction productivity given by Shui On, the DSM was able to analyse the time implication caused by each proposed design change.

Phase 1 RTM also provided a platform for steel mould fabrication management. Because the cost of a precast façade mould is high in contrast to the cost of the façade itself; it is crucial for the project manager to estimate the optimal number of steel moulds at the beginning of the project that will not disturb the on site construction. This management tool can:

• optimizenumberofsteelmouldforeachtypeof façade given the production lead time and construction schedule, reducing the asset tied-up with inventory and promoting Just-In-Time methodologies:

Figure 6. Sample for composite material for facade

• estimatethetotalcostformouldfabrication

• provideadetailedproduction,planningandcontrol schedule, and inventory analysis on site andinmanufacturingforeachtypeoffaçade•enable impact estimation in case of moulds failures or facades failures during the production and installation, and enable the user to find extra moulds

Conclusion

We believe that customized, high-quality, sustainable residential building is the trend in the market. RTM Building System will totally shift the existing business model to industrial development production. Future homebuyers will be able to customize the volume, exterior and interior design arrangements in their home according to their own needs, desires, lifestyle and expectations.

Acknowledgments

Louis Wong, Managing Director – Project Management, Shui On Land

Clement C.C. Lau, Director of Project planning and Design, Shui On Land

Hui Shing Sun, Executive Director, Shui On Land

Frankie Lai, General Manager – QS & Procurement, Shui On Land,

Lam Hiu Fai, Senior Manager – Technical Research and Development, Shui On Land

Calvin Wong, C.E.O, Tecton Limited

Toste Wallmark, System Architect, Tecton Limited

Professor Mitchell Tseng, Head of Advanced Manufacturing Institute, UST

Anita Siu, Engineer, Ove Arup & Partners Hong Kong Ltd

References

Karhu, Vesa “Product Model Based Design of Precast Facades” Royal Institute of technology, 1997

RTM

Phase 1 RTM OverallSystem Architecture

Shop Drawing

Opening, Window, Edge, Connection, Finishing, etc

Product Library

Statutory, Architectural, Structural, Others

RulesMaterialMaterial Library

Building BlockFamily Platform Building Block

Family Platform

Construction Productivity.Construction Procedure

Process LibraryDecision support

Simulation System

Decision supportMaking Tool

Design Engine

GANTT Chart

Figure 5. RTM overall system architecture

research review | 27

28 | research review

Author: Erin Morrow

Union Station, Toronto

Union Station functions as Toronto’s primary transportation hub for commuters on rail, bus, subway and light rail services. Every year, the station serves more passengers than all three terminals at Pearson International Airport, Canada’s busiest airport. As of 2006, the station handled:

• 49mtransitrailandbuspassengers

• 20msubwaypassengers

• 2.3mrailpassengers

These numbers are expected to double over the next 20 years.

Enhancement of this historic facility to accommodate increased usage and improve pedestrian connections is a priority for Toronto. Revitalization will continue to be coordinated with the station operators, their expansion programs, and future population and employment increases in the downtown area.

Policy Background

Toronto City Council are developing a Future Concept Plan for the revitalisation of Union Station and its vicinity. Given that all transit passengers are ultimately pedestrians, planning principles for the Future Concept Plan need to ensure from the outset that current and future pedestrian circulation is maintained or improved.

The Future Concept Plan is being developed within the context of a number of important policy documents:

• TheTorontoOfficialPlanand PedestrianCharter: These high level policy documents provide general guidance on the development of the station. They consider the refurbishment of the station; enhancement of its passenger handling capacity; a programme of street improvements to improve safety for pedestrians and cyclists; and the creation of an urban environment and infrastructure that ensures safe, comfortable, attractive and convenient pedestrian conditions.

• PedestrianPolicy,Principles,andObjectives: This document outlines the existing situation and forms a background to the Union Station Master Plan.

• TheUnionStationMasterPlan: This is a bold, visionary roadmap for the restoration, revitalisation and operation of the Union Station complex. It was adopted in 2004, and brings together a cohesive policy framework for pedestrian planning. It provides specific principles and objectives for the station building and for the immediate areas surrounding the station.

• TheUnionStationDistrictPlan: This was adopted in 2006, and sets out a number of guiding principles for the public realm that are intended to enhance and improve the pedestrian environment. It aims to integrate heritage and transport requirements into the fabric of the downtown area, by improving pedestrian connectivity and enhancing pedestrian crossings.

The primary objectives of the Phase 1 study this study were to:

• Developaquantitativepictureofpeakpedestrian flow conditions in and around Union Station, for both current and future forecast conditions, and;

• Identifypotentialconstraintsonpedestrianflows associated with the Union Station Concept Plan.

• Generatesufficientdatatosupportmoredetailed flow analysis in Phase 2.

• Supportdecisionmaking,designeffortsandoperational planning for the Future Concept Plan.

The study was completed for two principal time frames: base conditions (2003) and projected conditions (2021). A third 2011 condition was also documented on the basis of a straight line interpolation between 2003 and 2021. These time frames correspond to forecast horizons for the various transportation providers, and to the City of Toronto’s employment and residential population data.

The analysis focuses on pedestrian movement volumes for the morning peak, the afternoon peak, and special events at adjoining venues. It highlighted a number of opportunities and constraints, which helped to define the scope and focus of the Phase 2 Internal Pedestrian Circulation study.

Abstract

Union Station is the Greater Toronto Area’s most important transportation hub, and its revitalization is intended to improve the delivery of local, regional and national rail passenger services. This process is being facilitated by the City of Toronto, which is working with other interested parties to coordinate transportation and pedestrian planning initiatives. These will respond to anticipated increases in transit passenger numbers over the next 20-30 years, and address the constraints that this growth will place upon the existing infrastructure.

Arup has been working with the City of Toronto to develop advanced analysis and planning models based on new micro-simulation technology. These new tools have been developed by Arup to address shortcomings in existing computational techniques and have been instrumental in predicting future activity at Union Station and communicating findings to diverse stakeholders.

The City of Toronto has created several high-level policy documents aimed at improving the quality of the Station and its surrounding environment. The Union Station Master Plan and Union Station District Plan both advocate strong pedestrian connections leading to, through, and from the Station. They define parameters for specific studies that have been undertaken to help improve the overall pedestrian amenities within and in the vicinity of the Station.

Innovative Approaches to Pedestrian Planning at Toronto’s Union Station

Figure 1. 3D Stimulation enviroment

© A

rup

research review | 29

Union Station is an environment with complex route choices and very high volumes of pedestrian traffic, particularly during the morning rush hour. The challenge was to develop an approach to modelling the pedestrian traffic at the station in a way that would allow accurate calibration of the model inputs, without overly prescribing the detailed behaviour of pedestrians within the station.

To achieve this, Arup applied MassMotion (a leading edge ‘agent based’ simulation method), a full 3D model of the environment, and John Fruin’s industry standard planning and design guidelines for pedestrian behaviour. It was also necessary to improve the resolution of the information produced during Phase 1. A series of statistical methods, cross-checks, and operator-approved assumptions were used to produce a minute-by-minute breakdown pedestrian movement, from entry into the study area to final exit.

The primary differentiator between MassMotion and to other pedestrian simulation tools is that it actually models pedestrian behaviour rather than testing a user’s preconceptions about pedestrian behaviour. For example, if a room has doors on all four sides, other pedestrian simulation tools require the user to input what percentage of the room’s population will use each door. In a MassMotion simulation, each agent decides which door to use based on what it knows about the distance to its goal from each door and how long the queue is for each door.

There are two noteworthy advantages to such a system. The first is that in MassMotion an agent only needs to be assigned an end goal to navigate through an environment. In contrast, other models require the user to input percentage splits at all potential branching points and for all sub groups within a simulated population. Therefore a MassMotion model requires far less time to set up or modify, and scales much more efficiently as the complexity of the simulation environment increases.

The second advantage is that MassMotion actually predicts how rational pedestrians will navigate through an environment and how they will respond dynamically to constantly evolving situations. Other tools (as a result of the static route assignment approach) are not able to predict how pedestrian congestion and route choice will evolve throughout the course of simulation. For these

reasons, MassMotion was considered an ideal choice to model a future Union Station that had both a reconfigured layout and doubling of current pedestrian volumes.

During the calibration phase of the simulations a phenomena was observed which has had a significant impact on how the agent behaviours were designed. The original assumption about commuter behaviour was that they are focused on selecting the most efficient route to their destinations. In reality this is only partially true, as even expert users of a facility will not have perfect awareness of distance and congestion on any given route. Furthermore individual assessments of route cost may include preferences for factors that are not captured within the range of variables in the software. As a result, agents were permitted to select from a range of available routes that were within a specified deviation from the most efficient route. This had the effect of distributing the simulated pedestrian flows more smoothly throughout the model and also resulted in statistical results that were significantly better aligned to survey data. The results suggest that the model provides a very clear indication of how the station will be used, and that the behavioural profile could be applied to future scenarios with confidence.

After extensive simulation of the proposed Future Concept Plan in its current configuration, Arup concluded that the proposed layout of the station in the Plan is appropriate from a pedestrian flow perspective will also support the pedestrian volumes estimated for 2021.

Opportunities were revealed for:

• Providingadditionallinkagestothestreet and pedestrian networks to ensure appropriate levels of pedestrian capacity, and to further disperse pedestrian movement related to the Station.

• FurtherplanningandreviewoftheUnionStation Front Street Plaza and external areas, in the context of the Station Master Plan.

• EnhancingtheporosityoftheStationtominimize bottlenecks and pedestrian congestion.

Constraints were revealed as follows:

• Therewillcontinuetobeaveryheavyorientation of pedestrian flow between the Station and the downtown core, resulting in increasing congestion on the existing network. This is likely to require the introduction of several mitigating measures to improve future grade pedestrian conditions.

• TheentryandexitpointstotheStationrepresent the most critical points of potential congestion. Methods of metering flows further up stream may need to be considered.

Phase 2-The Union Station Internal Pedestrian Circulation Study

The intention of the Phase 2 study was to provide greater insight into the existing and future operation of Union Station from a pedestrian flow perspective, and to refine concepts for the layout of retail, commercial and transit-related components within Union Station. This work was coordinated with other transportation planning elements (for example, loading, servicing, taxi stands) and other initiatives in the immediate area, and was intended to answer four fundamental questions:

• HowwilltheproposedUnionStation Future Concept Plan impact on pedestrian flow patterns?

• IstheFutureConceptPlanappropriatefrom a pedestrian flow perspective?

• Wherearetheinternalandexternalcongestionpoints, and what conditions cause congestion?

• Wherearetheareasofflexibilitythatcouldprovide opportunities for other revitalization initiatives?

Phase 2 focused on three distinct configurations for the station and surrounding environment, the Current Configuration (2003), the Future Concept Plan (2021), and the Future Concept Plan (2021) with a new Northwest pedestrian connection. All configurations were simulated with the pedestrian volume forecasts for the 15 minute morning peak.

Innovative Approaches to Pedestrian Planning at Toronto’s Union Station

Figuire 2. VIA Rail Concourse connection ramp Figure 3. Passenger ticketing in Great Hall

Figure 4. Detail of historically listed station building

© A

rup

© A

rup

© A

rup

30 | research review

While there are some internal areas which experience high densities of commuters during the morning peak, the flow of traffic is consistent and does not degrade to conditions of static congestion. The external pedestrian routes at Union Station provide adequate capacity for demand at all areas except for the sidewalks and crosswalks at corners of Front Street and Bay Streets, and the corner of Front Street and York Street.

An additional benefit of the modelling was that it allowed Arup to identify a number of underutilized areas. Further testing of platform staircase locations, retail and layouts may reveal the potential for significant gains in terms of balancing passenger flow patterns and improving the user experience.

Next Steps

The City of Toronto is currently considering options for Union Station Revitalization, including opportunities to reconfigure the internal Station layout and connections in a way that supports pedestrian movement and compliments the highest and best use of space within the building. Within this context, the City will continue to study pedestrian movements within and beyond the Station to increase its effectiveness as a major transportation hub.

As the next step in the initiative (Phase 3), a simulation is required for the commuter weekday afternoon peak 15-minute period.

The model simulation and analysis envisioned in Phase 3 will address the following questions:

• Howwilltheproposedlayoutplanimpactpedestrian flow and queuing/waiting patterns, particularly in the future pedestrian concourses?

• Whataretheopportunitiesforpedestrianflowand station uses (for example, the expansion of retail outlets in some areas?

• Wherearetheinternalandexternalcongestionpoints, what conditions may be causing congestion, and what measures may be required to alleviate unacceptable levels of congestion?

The work plan envisioned for the next phase of pedestrian planning will apply the MassMotion model and City database for Union Station to create a simulation of current conditions, and forecast conditions (2021) for existing and future networks. The model will confirm anticipated levels of service for all key locations (ie doorways, stairs, ramps, escalators, corridors, etc.) within the Station and adjacent pedestrian areas, and provide visual pedestrian flow simulations within the three-dimensional environment.

Conclusions

The process of policy development and confirmation involved a wide range of interests from the station operators, decision makers and the public. The importance of pedestrian activities at Union Station and its environs, and the need to plan for and accommodate growth, are clearly articulated in City policy and programs.

From a modelling perspective, the Phase 2 study was an exceptional test for the MassMotion toolset. The very high volumes of pedestrians being simulated, in combination with the complex layout of the station facility, demonstrated the usefulness of an agent based approach to pedestrian simulation and analysis.

During the study MassMotion has proved itself as an invaluable tool. It allows visualization of pedestrian flows, and identifies areas where further design refinements are required, particularly where pedestrian levels of service need improvement.

Arup has continued to develop MassMotion by improving the software based on lessons learned during studies of Union Station. User interface improvements have significantly reduced setup time and user input errors, while the way-finding algorithms have been expanded to incorporate signage and signals. In the future, it will be possible to build on the Phase 2 work and consider other station configurations and traffic conditions.

As the cost of developing pedestrian modelling applications decreases, or as applications become more scalable and user friendly in nature, pedestrian modelling and simulation work is likely to become commonplace. In this context, an integrated planning approach that merges technology and the human experience will become increasingly important.

References

Arup Canada (2005). Union Station Pedestrian Movement Study-Phase 1

Arup Canada (2006). Union Station Pedestrian Movement Study-Phase 2

City of Toronto (2002). Official Plan

City of Toronto (2002). Pedestrian Charter

City of Toronto (2004). Union Station Pedestrian Planning Principles and Objectives

City of Toronto (2004b). Union Station Master Plan

City of Toronto (2006). Union Station District Plan

Fruin, J. ([1971] 1987). Pedestrian Planning and Design, Revised Edition. Mobile, AL: Elevator World

Acknowledgments

Arup wishes to acknowledge and thank Tim Laspa and Jeff Bateman of the City of Toronto Planning Department for their co-authorship of the original version of this paper which was presented at the Walk 21 conference in 2007.

Figure 5. Simulated morning peak activity in 2021

© A

rup

research review | 31

32 | research review

Author: Raymond Yau, Sui-Hang Yan, Rumin Yin

Introduction

Excessive urbanization makes some metropolis like New York, Tokyo, Shanghai, Hong Kong etc. paid for a heavy cost on poor ventilated environment. Air pollutants, heat island effect, lack of natural ventilation, these sunk costs of high-density living are such that they have an adverse effect on living standard. Good planning and building designs are critically important. The unique urban fabric of Hong Kong creates a special ventilation environment. Its pattern of streets, building heights, open spaces, density, features, landscape and so on determine the environmental quality (natural air ventilation, solar radiation, daylight, air temperature, etc) both within buildings and outside. Some of the more important views of built environment of Hong Kong could be summarized as follows:

• Alackofwellconsiderednetworkofbreezeways and air paths towards the prevailing wind.

• Tallandbulkybuildingscloselypacked together forming undesirable windbreaks to the urban fabric behind.

• Uniformbuildingheightsresultinginwindskimming over the top and not being rerouted into the urban fabric.

• Tight,narrowstreetsnotaligningwiththeprevailing wind, and with very tall buildings on both sides, resulting in very deep urban canyons.

• Alackofgeneralurbanpermeability:fewopenspaces, no or minimal gaps between buildings or within large and continuous buildings, and excessive podium structure reducing the air volume at pedestrian level.

• Largebuildinglotswithinsufficientair spaces, and with building on them not generally designed for wind permeability and forming wind barriers.

• Projectionsfrombuildingandobstruction on narrow streets further intrude into the breezeways and air paths.

• Agenerallackofgreenery,shadingand soft landscape in the urban area.

• Wall-likebuildingsatwaterfrontprevent wind penetrating to hinterland.

With the intention to improve the wind environment of Hong Kong, Planning Department of Hong Kong SAR Government has issued a guideline on building ventilation performance, namely Air Ventilation Assessment system to provide a good practice on building design.

This system is aimed to provide effective air flow in the external macro built-up environment which would not lead to adverse or restricted conditions to cause human discomfort or be unfavorable for the predominant land use activities. The primary objective of the assignment is to explore the feasibility of establishing some protocols to assess the effects of major planning and development proposals on external air movement for achievement of enhanced macro wind environment. The objectives would involve examination of the methodology, standards, scope and mechanism for possible application in assessing the external air movement impact of planning and development proposals; and examination of the practicality, prima-facie reliability, cost-effectiveness and limitations of the assessment protocols to support their feasibility for general application in future planning and development proposals if considered necessary. It is not the purpose of this assignment to address air ventilation performance of individual building designs or indoor space designs, nor the air quality impact assessment for air pollutants control that are respectively governed by the Buildings Ordinance and Environmental Impact Assessment Ordinance.

Air Ventilation Assessment System

AVA system proposed an indicator, Wind Velocity Ratio (VR), which is defined as pedestrian wind velocity over wind velocity at the top of wind boundary layer (usually at 600m or above city centre). VR indicates how much wind availability of a location could be experienced and enjoyed by pedestrians on ground taking into account the surrounding buildings, topography and the proposed development. The site wind availability (incoming wind condition from surrounding) is defined by the Meso-scale mathematics model (eg MM5) or large-scale wind tunnel test. In general, the characteristic of the site wind availability data are reported in 16 directions with respective wind frequency. Test points, where VRs are reported, are scattered over the proposed site area and its surrounding to identify the wind performance under pre- and post development. Based on the VR of the test points, the resultant wind environment of the project can be assessed.

Computational Fluid Dynamics is a recommended tool for AVA study. The model area is one of the prime concerns under AVA system; the area under consideration is classified as project area, assessment area and the surrounding area. The project area is defined by the project site boundaries and includes all open areas within the project that pedestrians are likely to access. The assessment area of the project includes the project’s surrounding up to a perpendicular distance H (H = tallest building on site) from the project boundary. The test points within the assessment area indicate the wind impact of the proposed site to its surrounding. And the surrounding area is defined as 2H area from the project site boundary. This surrounding area is found important as it can moderate the

Abstract

Hong Kong is one of the densest populated cities in the world. High-density living has the advantages of efficient land use, public transportation as well as the closer proximity of daily amenities. However, this planning style has its disadvantage on the benefits of natural environment – wind ventilation. The wind performance at pedestrian level in Hong Kong is more and more concerned by the public due to more and more wall-type building cluster built in Hong Kong that block the wind penetration. Recently, a guideline on building ventilation performance, namely Air Ventilation Assessment (AVA) system has been issued to provide a good practice on building design. Computational Fluid Dynamics (CFD), as one of the accepted assessment tools, has been applied for several Air Ventilation Assessment studies. However, there is always debate on the accuracy of the simulation result. Conventional RANS turbulence model may not be accurate enough for complex flow but more accurate Large Eddy Simulation (LES) is impracticable for practical applications. Combining time-efficient of RANS model and accuracy of LES model, hybrid model – Detached Eddy Simulation (DES) receives concern in modern flow modeling industry. This study was focused on comparing the turbulence model performance of RANS type models, including Spalart-Allmaras model (SA), Renormalized Group k-e (RNG) and V2F model (V2F) and DES model. The simulation results were verified with the full-scale wind measurement of Shijuku Area by Architectural Institute of Japan (AIJ) Group. External wind modelling under environmental wind engineering approach will be discussed in detail. It was found that the DES turbulence model could provide generally better results than other RANS turbulence models, especially at congested locations. However, it was also found the DES model is more sensible to grid resolution.

Air Ventilation Assessment by External Wind Modelling Using Different Turbulence Models

research review | 33

incoming wind to give a much appropriate incoming wind profile and turbulent structure rather than a prescribed wind profile.However, CFD still receives challenge on modeling accuracy precisely, but it is still well accepted that it could capture general wind flow behavior. One of the modeling challenges on CFD modeling is selection of turbulence model. Conventional RANS turbulence model may not accurate enough for complex flow but more accurate Large Eddy Simulation (LES) is impracticable for practical applications. Combining time-efficient of RANS model and accuracy of LES model, hybrid model/Detached Eddy Simulation (DES) receives concern in modern modeling industry. In this study, it is tried to compare the turbulence model performance of RANS type models, including Spalart-Allmaras model (SA), Renormalized Group k-e (RNG), V2F model (V2F) and DES model.

Methodology

The simulation results would be verified with the full-scale wind measurement of Shijuku Area by Architectural Institute of Japan (AIJ) Group. The field measurement was undertaken in 1977, over 30 measurement points data point were available (Figure. 1). The area of concern is about 1km x 1km. The wind velocity was measured by three-cup anemometers and the measurement heights are different according to its locations which are 3-9m above the ground surface. The measured velocity was normalized by the wind speed at reference point. These points were at the top of Shinjuku Mitsui Building (237m) for the NE-N-NW wind directions, and at the top of the KDD Building (187m) for the other wind directions. The site consists of high density and low rise zone at the northern and eastern part high-rise zone at the centre, and some open spaces are found at the south-western part of the site. Also some wind corridors can be clearly identified at the site and therefore examination on effectiveness of wind corridor can be held on same simulation.

It gives a benefit to investigate how these highrise and low-rise buildings interacted in terms of air ventilation.The STAR-CD Version 3.26

Computational Fluid Dynamics Program was employed in this study to calculate the velocity distribution within the flow domain of the CFD model. STAR-CD is one of the leading multi-purpose thermofluid analysis codes for engineering and construction industry. It is widely favored by engineers requiring a robust and efficient software tool, capable of modeling fluid flow, heat transfer, mass transfer and chemical reaction. To represent the physical phenomena resulting from the wind, the aforementioned turbulence model employed while the fluid is specified as incompressible. In present study, the high order scheme Monotone Advection and Reconstruction Scheme (MARS) scheme is applied to discretize the momentum and turbulence equation. And the temporal discretisation was handled by fully implicit scheme. Then the PISO algorithm was employed the tackle pressure-velocity linked equations CFD simulations were conducted for 16 wind directions. The CFD model was built-up according to the CAD data found in data published by AIJ Group The computational domain size was 3km x 3km which is triple size of the building coverage in CAD to ensure elimination of edge effect. The domain consist 4,000,000 cells and cells were refined at the buildings area to enhance the modeling accuracy. Four turbulence models were chosen in this study which are Spalart-Allmaras model (SA), standard k-model (KE) and V2F model (V2F) and DES model.

Result and discussion

Figure 2 shows some extracted point results under 16 wind directions. The leftmost graphs (point 6 and 17) show the points results which are relatively accurate, and the rightmost graphs (point 11 and 28) show those are inaccurate. For the measurement point result point 2, 6, 15 and 17, it shows that CFD could give a good prediction on general pattern over 16 wind

Figure 1. Measurement points

Figure 2. Result on comparison for 16 wind directions

Air Ventilation Assessment by External Wind Modelling Using Different Turbulence Models

directions if the point is located at the centre of domain and surrounded by certain layers of buildings. In contrast, point 11 and 28 could not give a good wind prediction as they are located at the edge of the building model. By this observation, it can conclude that a “wind pre-conditioning” layers of buildings is important for providing an accurate simulation result. The VR of the measurement is easily been over-predicted/under-predicted even a pretreated wind profile is plugged into the simulation. This finding is coherent to the philosophy of the 2R surrounding area as stipulated at the AVA system. ie there would have 1H area for the “wind pre-conditioning” layers of buildings on ensuring modeling accuracy.

Windward condition

Figure 3 demonstrates some point results under windward condition. The prediction for all turbulence models could give relatively good results except point 32. It may be due to the aforementioned reason that there has not enough preconditioning zone for the incoming wind for giving a right incoming wind profile. Also, all models tend to over-predict the VR for the points where the prevailing wind is coming from an open space (Point 7, 8 and 9), and under-predict the incoming flow for wind coming from a wind channel (Point 20). In general, DES model could give better result among the others while comparing with measurement result, and the second accurate model would be V2F model.

Leeward condition

Figure 4 indicates some point results under leeward condition. All RANS models give a fairly good result in predictions but DES outperforms the others obviously, especially for point 21, 22 and 26. In Kataoka et al’s urban wind study, it

NNNW

NW

WNW

SW

SSWS

SSE

SE

ESE

E

ENE

NE

NNE

WSW

W

0.6 0.5 0.4 0.3 0.2 0.1

NNNW

NW

WNW

SW

SSWS

SSE

SE

ESE

E

ENE

NE

NNE

WSW

W

0.7 0.6 0.5 0.4 0.3 0.2 0.1

NNNW

NW

WNW

SW

SSWS

SSE

SE

ESE

E

ENE

NE

NNE

WSW

W

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

NNNW

NW

WNW

SW

SSWS

SSE

SE

ESE

E

ENE

NE

NNE

WSW

W

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

KeyPoint 11Point 6

Point 17 Point 28

EXPSARNGV2FDES

34 | research review

compared RANS model and LES model performance for the wake field behind building, it was found that wake zone produced by RANS was larger than LES and larger velocity defect was found. Current finding is coincident with Kataoka etal’s observation that RANS would underestimate the wind velocity at the wake zone.Here, DES could give similar result as LES did at the wake zone since DES was run as a LES model at this far wall region. For point 16, it is located at the gap building two high-rise building, all turbulence models could give good prediction at this location even the measurement point is located behind the building. Aforementioned, RANS model trend to underestimated the wind velocity at wake zone, however, in this case, the wake zone is destroyed by the second building right behind the first building. Vortex shedding created by first building may not used to entered to this wake zone as the vortex was deflect by the second building. Hence, RANS model may able to give a good estimation in wind velocity there. There would be no great difference between turbulence models in velocity prediction.

Open Space

Figure 5 demonstrates some point results under open space condition. All turbulence models fail to give an accurate result at the open space area. DES model gives a comparatively worst result among all models. It may be the reason that the grid size at the open space is relatively large which makes the small turbulent is filtered out in the DES model. According to Bunge et al, DES length scale is determined by the minimal of RANS length scale and grid scale times a model constant. Which implies that grid should be fine enough to switch on the LES sub-grid model, if not, it would be functioned as a RANS model. From current study results, it can be observed that predictions by DES are similar to those for RANS model since the grid size is not fine enough at open space region.

Congested Space

The incoming wind condition for congested space measurement point is shown in Figure 6. As the measurement point is located deep in the building cluster, the flow behavior for these measurement locations is similar to the flow over a street canyon. On the extensive review on CFD modeling of street canyons flow by Li et al, it was summarized that RANS type model could reproduce the general flow pattern but under estimated the velocity within the cavity. Similar conclusion was also found in some numerical studies (Walton et al and Sahm et al). On current simulation results, it is also observed that RANS model trend to under-predict the flow velocity in this region. The under-estimation problem would worse if the canyon depth becomes deeper, ie point 10. DES could give a relative better results but it is still under-estimation the wind velocity within the cavity.

Figure 4. Leeward condition comparison

Figure 6. Congested space conditionFigure 5. Open space condition

Figure 3. Windward condition comparison

References

A. Walton, A.Y.S. Cheng and W.C. Yeung. Large-eddy simulation of pollution dispersion in an urban street canyon–part I: comparison with field data. Atmospheric Environment. Vol. 36 (2002) 3601-3613

Architectural Institute of Japan. Guidebook for practical applications of CFD to pedestrian wind environment around buildings. Architectural Institute of Japan, Japan, 2007

CD adapco Group. STAR-CD – Methodology. CD adapco Group.UK, 2003

H. Kataoka, T. Tamura, Y Okuda and M. Ohashi. Numerical evaluation of the wake field behind high-rise building by RANS and LES. 12th International Conference on Wind Engineering. Australia, 2007.

P. Sahm, P. Louka,M. Ketzel, E. Guilloteau and J. Sini. Numerical and experimental modeling of pollutant dispersion in a street canyon. Journal of Wind Engineering and Industrial Aerodynamics. Vol. 90 (2002) 321-339

Planning Department of HKSAR. Feasibility study for establishment of air ventilation assessment system. Hong Kong, 2005

U. Bung, C. Mockett and F. Thiele. Guidelines for implementing Detached-Eddy Simulation using different models. Aerospace science and Technology. Vol. 11, (2007) 376-385

X.X. Li, C.H. Liu, D.Y.C. Leung and K.M. Lam, Recent progress in CFD modeling of wind field and pollutant transport in street canyons. Atmospheric environment. Vol. 40 (2006) 5640-5658

Acknowledgments

Prof. Mochida, Prof. Yoshie, Dr. Kataoka, and Prof. Tominaga of AIJ CFD Group for providing the experimental data and invaluable comments. Arup Group Ltd’s Building Sector Board and Design and Technology Executive for providing the joint funding to this enabling technology research project.

Collaboration

Professor Akashi Mochida - Tohoku University, Sendai, Japan. Professor Ryuichiro Yoshie – Tokyo Polytechnic University, Tokyo, Japan. Professor Yoshihide Tominaga - Niigata Institute of Technology, Niigata, Japan. Dr Hiroto Kataoka – Technical Research Institute, Obayashi Corp. Tokyo, Japan

Conclusion

This paper introduces the Air Ventilation Assessment (AVA) system of Hong Kong, which providing a constructive guideline to the building industry on evaluating the “air-right” building design. As Computational Fluid Dynamics is the recommended assessment tool for this system, prediction accuracy of this tool is one of the prime concerns of assessor. This study presents some modeling concern on urban wind modeling and a comparative study on difference turbulence model including Spalart-Allmaras model (SA), standard k-model (KE) and V2F model (V2F) and DES model.

The simulation results suggested that the “pre-conditioning” layer of building is important for providing a good incoming wind condition. The measurement location should be located well within the buildup area in order to capture a well-conditioned incoming wind. This philosophy is already reflected in the 2H area of the AVA system

RANS models are the industrial standards in engineering practices due to its simplicity and relativity low computational demand. However, this type of model has its limitations on prediction accuracy. Implementation of DES model is relatively complex and the computational cost may be 10 times of RANS one. The simulation results shown that DES model outperforms the others at both windward, leeward and congested locations although DES model may not give a good prediction at open space due to grid resolution issue. DES model is recommended for studying external wind modeling.

0.70

0.60

0.50

0.40

0.30

0.20

0.10

0.007,WSW 8,SSW 9,WSW 20,ESE 32,W

VR

Point

EXPSARNGV2FDES

Windward0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.002,NE 16,W 21,SSE 22,SSE 26,E

VR

Point

EXPSARNGV2FDES

Leeward

0.60

0.50

0.40

0.30

0.20

0.10

0.0029,NE 31,W 13,NE

VR

Point

EXPSARNGV2FDES

Open Area0.30

0.25

0.20

0.15

0.10

0.05

0.003,NE 5,W 10,NE

VR

Point

EXPSARNGV2FDES

Congested

research review | 35

36 | research review

Author: Mikkel Kragh

Introduction

This paper introduces a commercial development project, which in a number of ways challenges conventional thinking and modes of working in the construction industry. In essence, the project deals with prefabrication and industrialised architectural technology in a time where construction needs to focus on the agendas of climate change and sustainability.

The project originated under the Building Lab DK programme, which, over a limited period of time, facilitates innovation in the construction sector. Within this context, it is key that the project explores ways of working across traditional boundaries between disciplines and commercial players.

The project focuses on integration at several levels – Integration of physical systems and façade components as well as integration of the design, procurement and production processes.

After an exploratory pre-project stage, a consortium has been set up with founding partners spanning across business advice, design and engineering, design and manufacturing, materials and components supply, installation and architectural design. The result is a unique combination of highly specialised know-how and differentiated commercial drivers, which poses both challenges and truly unusual opportunities. The aim is to draw upon existing technologies and develop novel applications. The short (22 months) programme and the constituent partners will not allow for fundamental research to be carried out as part of the project.

Abstract

Sustainability is increasingly on the global agenda. In the construction industry, standards are imposed by legislation such as the EU Energy Performance of Buildings Directive. At times the requirements conflict with architectural trends of curtain walling and transparency. The paper outlines some of the limitations of current façade technology and identifies scope and opportunities for development.

A consortium has launched a development project, exploring innovative use of known technology. The aim is to facilitate high quality, high performance architectural solutions in a cost effective way. And high performance is meant to include sustainability in this context. Novel applications of composite materials and emerging technologies are rendered feasible by a blurring of traditional component boundaries and a rethink of industrialised manufacturing.

The emphasis is on integration of design and procurement processes as well as integration of components and subsystems within a modular product architecture. The paper introduces the aspects of integration, identifies some of the key challenges and unfolds a scenario for a novel approach to economically and environmentally sustainable building envelope systems.

High performance facade systems in a changing world

The building envelope has a major impact on both the indoor environment and the energy consumption of any modern building.

The performance of commercially available curtain walling systems is enhanced incrementally through design optimisation and development of glazing technology. In terms of thermal performance, the principal limiting factor is the framing, which is typically based on use of thermally broken aluminium extrusions. The combination of aluminium frames and the glazing edge conditions leads to linear thermal losses and relatively high thermal transmittance (U-value). The benefit of high performance insulation is somewhat limited by the performance of the framing and it can therefore be hard to justify the associated costs.

There is a need for façade systems, which respond to the building energy regulations as well as prepare for significantly more stringent requirements in the future. The technology must be environmentally responsible as planning and legislation are increasingly used as instruments to drive a more sustainable development. Visionary clients will often rank sustainability very highly from the outset, while comparatively more conservative clients will need to address the issues in order to comply with building regulations and/or get planning permission.

Sustainability is a complex topic, which lies beyond the scope of this introductory paper. It is, however, important to note that high performance should not be achieved at the expense of architectural expression. Rather, high performance façade systems should ideally facilitate interesting and exciting architecture as part of the sustainability agenda. As someone once said: “All good architecture should also be sustainable design, whereas sustainable design is not necessarily good architecture”.

A systemic and integrated solution

The aim of the project is to develop a commercial framework and set a new benchmark in terms of façade performance and procurement. The brief is both ambitious and challenging:

• Anintegratedfaçadesystem,utilising ‘new’ materials

• Markedperformanceadvantagesoverconventional solutions

• Highdegreeofintegrationofstandard and bespoke sub-systems

• Industrialisedmanufacturing

• Facilitationofmasscustomisation

• Abusinessmodelforcommercialisation of systems and processes

In addition to these material and performance attributes, the system needs to respond to variables associated with architectural degrees of freedom and still be commercially viable.

The project explores the use of fibre reinforced polymers (FRP) in construction, initially with particular emphasis on pultruded FRP, which has had a substantial uptake in applications such as bridges, off-shore and transportation, where its attributes are key to its success.

While most of the challenges the consortium has set itself can be said to be generic, and many of them are already dealt with in currently available solutions, it is the opportunities offered by the properties of the FRP material, which have led to the initial brief.

The decision to design an entirely new curtain walling system brings about interesting opportunities in terms of optimisation.Figure 1. Aerogel - example of high performance

translucent insulation material

© C

abot

Cor

pora

tion

Innovative Façade Technology Based on Pultruded Fibre-Reinforced Polymer Composites

research review | 37

In a truly integrated way, the system can be prepared to receive a range of products, including emerging technologies, with interfaces that are well-defined from the very early stages.

Materials/Methods

Curtain walling

Curtain walling is a form of vertical building enclosure, which supports no load other than its own weight, that of ancillary components and the environmental forces which act upon it (CWCT, 2000).

Unitised curtain walling is the preferred method of construction in high rise buildings as installation can be carried out from within the building, without the need for external access and scaffolding. Moreover, prefabrication in a controlled (factory) environment facilitates quality assurance.

Unitised systems comprise narrow, storey-height units of steel or aluminium framework, glazing and panels pre-assembled under controlled, factory conditions. Mechanical handling is required to position, align and fix units unto pre-positioned brackets attached to the concrete floor slab or the structural frame. Unitised systems are more complex in terms of framing systems, have higher direct costs and are less common than stick systems (based on profiles and panels assembled on site). The smaller number of site sealed joints in the unitised curtain walling simplifies and hastens enclosure of the building, requires fewer site staff and can make such systems cost effective. The reduced number of site-made joints compared with stick systems, generally leads to reduction in air and water leakage resulting from poor installation (CWCT, 2000).

Fibre-reinforced polymer (FRP) composites

In general engineering terms, a composite is a combination of two or more materials used together for any reason. This paper is only concerned with fibre-reinforced polymer composites.

Fibre-reinforced polymer composites is a general term used to describe a wide range of products made up of a combination of fibres in a matrix material. These materials are used extensively, particularly in the marine, aerospace and wind turbine industries, where their high strength to weight ratios and good performance in harsh environments mean that they are the best choice (Cripps, A. 2002).

In the construction industry fibre-reinforced polymer composites are widely used in applications such as cladding, pipes, for repair and in strengthening work. Construction makes up around 30 percent of the total market for FRP composites, second only to the automotive sector. However, there are many situations where they are not used. This may be because alternative and better understood materials are able to meet the requirements of the project, for significantly lower initial costs. There are other conditions however, where the best solution would be the use of FRP composites, but they are still not being used (Cripps, A. 2002).

Pultruded FRP

With a view to maximising the industrialisation of the manufacturing processes and develop a modular product architecture the project explores the use of pultruded FRP profiles. In the pultrusion process, tightly packed tows of fibres, impregnated with catalysed resin, are pulled through a shaped die to form highly aligned, continuous sections of simple or complex geometry. Curing of the resin may be achieved either by heating the die itself or by the use of dielectric heating (Cripps, A. 2002).

Solid and hollow sections may be produced by this process, and because of the high fibre content (70 percent by volume is achievable) and the high degree of the fibre alignment resulting from the tensile force used to pull the fibre bundle through the die, extremely good mechanical properties can be obtained (the highest achievable in any variety of composite). Off-axis fibres may

also be introduced into the structure if required (Cripps, A. 2002). Typical applications of pultruded shapes are concrete reinforcing bars and pre-stressing tendons, I-beams and other sections, roof trusses, space frames, walkways, shear stiffeners, electrification gantries, racking, etc. These racking systems are produced by joining pultruded sections together in the same manner as for timber or steel frameworks (Cripps, A. 2002).

In many current structural applications the FRP material is typically replacing steel or timber and the design of the connections is similar to conventional steel or timber designs. This approach to some extent disregards the unique (and variable) properties of the FRP material. In contrast, the novel facade concept is being developed with a focus on the intrinsic properties of the material and the opportunities offered in terms of detailing and bonding of interfaces and joints.

FRP material properties

One of the advantages of FRP is that it is possible to modify the characteristics of the composite by varying the matrix, the type and content of fibres.

Table 1 on the following page sets out typical properties for the most common fibres and resins, and demonstrate how strength and stiffness are bound with other factors such as cost and weight (Cripps, A. 2002).

A full description of the material properties lies beyond the scope of this paper. The specific composition of the FRP composite for the façade application is being determined by careful consideration of the manufacturing (pultrusion) process, required façade performance (fire, strength, thermal insulation, etc) and cost. The initial choice of matrix aims to reduce the risk involved in the early stages of development, while future stages will explore alternative matrix designs.

The ability to manufacture large hollow sections, potentialy without significant thermal bridging,

Figure 2. Installation of unitised curtain wall Figure 4. Example of structural FRP profile Figure 5. Pultrusion plant. Glass fibres entering and finished pultrusions leaving the die

Figure 3. Diagrammatic illustration of pultrusion process

Resin injection

ReinforcementHeating and curing Pulling devices Saw

Ventilation

© P

erm

aste

elis

a

© F

iber

line

© F

iber

line

© F

iber

line

Innovative Façade Technology Based on Pultruded Fibre-Reinforced Polymer Composites

38 | research review

Fibre Tensile strength (GPa) Young’s Modulus (GPa) Density (10³kg/m³) Cost (£/kg)

Aramid 3.15-3.60 58-130 1.39-1.47 20

Carbon 2.10-5.5 200-500 1.74-2.20 10-200

Glass 2.4-3.5 72-87 2.46-2.58 2.5

Resin Tensile strength (MPa) Young’s Modulus (GPa) Density (10³kg/m³) Cost (£/kg)

Polyester 50-75 3.1-4.6 1.11-1.25 ~ 2.5

Epoxy 60-85 2.6-3.8 1.11-1.20 ~ 5-10

Phenolic 60-80 3.0-4.0 1.00-1.25 ~ 2

impacts on the way the system is conceived and detailed. The fact that the structural material has a low thermal conductivity means that the overall system depth can be minimised, which in turn leads to a compact (or slim) facade system.

Results and discussion

An innovative approach to development work

An important aspect of the project is the close collaboration between leading industry partners and non-manufacturing partners with a range of highly specialised skills and a global presence.

The collaboration will only be successful as long as the commercial framework, or the business model, allows for all of the partners to benefit, albeit in very different ways. In addition to the measurable benefits of return on investment, the cross-company collaboration allows each of the partners to develop and optimise solutions beyond their own core business. This is potentially a very important aspect, as it is exactly the facilitation of holistic thinking that can lead to truly innovative products and solutions.

Less tangible benefits to the partners result from the generous access to technical expertise and general networking opportunities and a number of collaborations are already taking place. Architects within the consortium will challenge the direction of the technical development and demand aesthetic and functional flexibility. While the architects are influencing the brief, they will also gain an understanding of the potential of the system, which will inform their thinking on their projects. Ultimately, the architectural partners identify the first projects, which are best suited for this innovative technology. It is easily a win-win situation.

There is no formal obligation by the supporting institutions to disseminate the results or generate income. There is no intention to carry fundamental research as part of the project. The sole focus is on the development of a commercially viable product and a business model, which will successfully attract funding for a second stage of the venture.

The project has identified numerous areas of research, which are being discussed with research centres and academia in parallel with product development and commercialisation work.

Moreover, the consortium partners are reaching out to a series of possible networking partners as

the project progresses. These partners are in a position to inform the development work, thereby facilitating integration of proprietary products. Conversely, learning about the integrated design concept enables network partners to develop and adapt their products to seamlessly integrate with the system. The network partners are materials suppliers, glass processors, and suppliers of systems such as heating, ventilation, photovoltaics, lighting etc.

The first concept

A first concept has been developed. The concept is aimed at maximum utilisation of the intrinsic properties of the composite material within the context of curtain walling:

• Lowthermalconductivity

• LargepultrudedFRPsections

• Compact(slim)system

• Structurallybondedconnections

• Lightweight

• Limitednumberofparts

• Appearance(potentiallytranslucent)

The current concept addresses the fabrication and assembly processes and actively aims to minimise the number of parts and the need for machining. The result is a potentially highly rationalised manufacturing process and reduced risk in terms of workmanship.

The architects within the consortium have been part of the process and have reacted very positively to the opportunities. Rather than attempting to replicate solutions based on conventional technology, the new generation of systems potentially offers a new architectural language. This perspective have been embraced by the architects, who have responded with studies of new interpretations of the building envelope.

An important attribute of the development version of the facade system is its compactness. The reduction in facade depth as compared with conventional products can translate into maximisation of lettable floor area for a given building outline. The value proposition to the investors consequently not only covers high performance and low carbon design, but also return on investment in a direct and measurable way.

Conclusion

In parallel with the development work, a more or less formal network is being be established in order to both let the industry know about the initiative and invite manufacturers of materials and emerging technologies to develop compatible and complementary components and solutions.

The ability to respond to the different architectural requirements on a project-to-project basis will drive the development of more versions of the system, until one can finally speak of a ‘family of systems’ or even a catalogue of configurable solutions.

The success of the concept depends entirely on the way it is used on real buildings and pilot projects will prove instrumental to getting the technology ‘off the ground’. The project team is already testing the concepts against live projects as a way of ensuring that the solutions are both buildable and relevant.

In the first instance, the main objective is to develop a sound concept, which is thoroughly put through the tests needed to be deemed truly fit for purpose as well as commercially viable.

A network is beginning to form and a host of emerging technologies will eventually more or less seamlessly integrate with the system. The aim is a highly modular set of solutions, where high performance is not penalised, but rather encouraged through integration and maximised utilisation of the core of the system – the materials and the design.

References

Cripps, A. 2002. Fibre-reinforced polymer composites in construction. CIRIA C564.

CWCT. 2000. Curtain Wall Types. Technical Note No. 14. Centre for Window and Cladding Technology.

Acknowledgments

The project entitled The Integrated Building Envelope is supported by the Danish Realdania foundation (www.realdania.dk) under the Building Lab DK programme (www.buildinglab.dk). For further information visit the project website: www.integratedbuildingenvelope.com.

Collaboration

Permasteelisa Group: www.permasteelisa.comFiberline Composites: www.fiberline.comArt Andersen: www.art-andersen.dkCabot: www.cabot-corp.com3XN: www.3xn.com Make: www.makearchitects.com

Figure 6. Studies of façade panels Table 1. Typical properties for the most common fibres and resins

© M

ake

Arc

hite

cts

research review | 39

40 | research review

Author: Andrew Coles and Armin Wolski

Real-Scale Rail car Mokup Fire Test

The mockup fire test configuration is shown in Figure 1a (pre-test) and Figure 1b (post-test). It consists of seats, wall linings, ceiling linings, and carpet inserted in a standard sized (8ft by 8ft by 12ft) ISO room calorimeter. Two transverse double seats are installed against a wall of the burn room to mimic the standard seat arrangement in the rail car. A single seat is installed in front of the two double seats. The seats consist of a foam/fabric cushion and seat shrouds. The burn room is lined with gypsum board. A false ceiling of three phenolic composite panels and one gypsum board panel is installed above the seating at a height of 2m from the floor to mimic the as-built ceiling height. The rear wall is lined with one phenolic panel extending downward from the false ceiling.

A 1.1m by 2.3m section of carpet is placed in the vicinity of the seats and is fixed to the floor to prevent curling of the edges. The ignition burner (a 0.3m by 0.6m rectangular propane sand burner) is placed between two transverse seats located approximately 0.05m from the wall.

Instrumentation includes a single thermocouple rake consisting of 24 thermocouples installed vertically at a spacing of 0.1m; additional thermocouples with different bead diameters are installed at five locations to allow for radiation correction of the thermocouple temperatures. Eight thin skin calorimeters are installed at floor level near the seats to measure heat flux.

The ignition burner’s peak HRR is 500kW, roughly approximating a flammable liquids spill of 4 litres (1 gallon).

This “extreme” ignition source, unlikely to be present in an actual car fire, was selected because most of the interior lining materials are fire retardant and it was felt that a small (trash bag sized) fire would not cause significant flame propagation. Using a large ignition source strength ensures that fire spread occurs after a minimal incubation period, ultimately requiring shorter simulation times in the modeling phase of the project. Figure 2a shows the heat release rate of the sand burner (full HRR reached after 75 s) and the total heat release rate measured by oxygen consumption calorimetry. Figure 2b shows the temperature measured 0.5m below the ceiling at the door, and the heat flux measured by thin skin calorimeter at the floor. Additional experimental data are presented later in the paper where they are compared with the model calculations.

A peak net heat release rate of 1.4MW (1.9MW total HRR) occurs approximately 110s into the test. Temperatures near the ceiling approach 730ºC, and heat flux levels at the floor approach 30kW/m2. These temperatures and heat fluxes exceed the threshold rule of thumb for onset of flashover (heat flux to the floor of 20 – 25kW/m2 and upper layer temperature rise of 500 – 600ºC) yet flashover did not occur as evidenced by the unburned seats that can be seen in Figure 1b. Flashover may not have occurred due to the relatively small section of carpet installed at the floor and the relatively small combustible wall lining area. At the time of peak heat release rate, one double seat was completely burning. It is also likely that the phenolic panels contributed significantly to the heat release rate, but smoke obscured the visual record. Post-test inspection of the burn damage (see Figure 1b) showed that the phenolic wall panels became detached from the walls.

Abstract

The most promising long-term prospect for modeling flame spread and fire growth at building scales is the coupling of condensed phase fuel generation models to computational fluid dynamics (CFD) models that simulate the gas-phase fluid mechanics, combustion, and heat transfer aspects of a fire. The primary advantage of this approach is its flexibility, and it has been suggested that this type of fire growth modeling will become an ‘invaluable tool for researchers and engineers’ due to this flexibility. With a coupled pyrolysis/CFD fire model, it should be possible to consider complex geometries and ignition scenarios, evaluate the impact of design changes on expected fire behavior, and assist in forensic fire reconstruction.

To date, there have been few rigorous attempts to validate CFD-based fire growth models, and flame spread prediction remains largely a research area. Most fire model validation work has involved ‘gas burner’ type problems where the movement of heat and smoke from a fire having a predetermined heat release rate (HRR) is predicted and compared to experimental measurements, eg the US Nuclear Regulatory Commission’s reports. Typically, a fire growth model is evaluated by comparing its prediction of large-scale behavior to experimental HRR measurements, thermocouple temperatures, or pyrolysis front position. The overall predictive capabilities of a fire growth model depends on decomposition, heat and mass transfer physics in the pyrolysis model and treatment of gas-phase fluid mechanics, turbulence, combustion chemistry, and convective/radiative heat transfer in the CFD model.

In this research, Fire Dynamics Simulator (FDS) Version 5 is used to simulate fire growth in a full scale rail car mockup. Model calculations of heat release rate, temperatures, and heat flux levels are compared to analogous experimental data. The rail car mockup consists of actual seat, carpet, wall, and ceiling lining materials removed from a rail car inserted in a standard-sized fire test compartment and arranged to simulate the as-built configuration. The material properties required for the FDS 5.0 pyrolysis model are estimated from bench-scale Cone Calorimeter test data using an automated optimization algorithm. In standalone simulations, the pyrolysis behavior of these “complex” real-world materials (which contain fire retardants and char heavily) can be simulated reasonably well with the FDS pyrolysis model.

Using Fire Dynamics Simulator (FDS) to Predict and Model Fire Propagation

Figure 1a. Mockup fire test configuration. Pre-test photograph;

Figure 1b. Mockup fire test configuration. Post-test photograph.

research review | 41

Using Fire Dynamics Simulator (FDS) to Predict and Model Fire Propagation

Solid-Phase material property estimation

One of the most challenging aspects of fire growth modeling is characterizing solid materials or assemblies in terms of the material properties that control their overall reaction to fire. For the present application where it is desired to simulate fire development in a compartment fire, this means quantifying each material in terms of the input parameters needed by the FDS 5.0 pyrolysis submodel. While based on a sound physical and chemical treatment of solid-phase pyrolysis as it is presently understood, material property estimation for the FDS 5.0 pyrolysis model is onerous. Each condensed-phase species (ie virgin wood, char, ash, etc.) must be characterized in terms of its bulk density, thermal properties (thermal conductivity and specific heat capacity, both of which are usually temperature-dependent), emissivity, and in-depth radiation absorption coefficient. Similarly, each condensed-phase reaction must be quantified through specification of its “kinetic triplet” (pre-exponential factor, activation energy, reaction order), heat of reaction, and the reactant/product species. For a simple charring material with temperature-invariant thermal properties that degrades by a single-step reaction, this amounts to 12 parameters that must be specified.

In the present work, an automated computer program based on genetic algorithm optimization is used to estimate the required material properties for the four materials used in the real-scale rail car mockup from Cone Calorimeter experiments. For each material, multiple Cone Calorimeter tests are conducted at irradiance levels between 15kW/m2 and 80kW/m2. In addition to the quantities normally measured in Cone Calorimeter tests (mass loss rate, heat release rate, etc.) measurements of surface temperature and back-face temperature are made and used in the optimization process. For modeling purposes, the flame heat flux is estimated at 30kW/m2 based on the work of Rhodes and Hopkins.

Figure 3 shows a comparison of the measured and modeled mass loss rate for each of the four materials. Although surface temperature and back face temperature measurements are also used in the optimization process, for clarity of presentation only mass loss rates are shown. It can be seen that the pyrolysis model reproduces the major features of the mass loss rate curves but certainly does not capture every detail.

Comparison of real-scale fire growth calculations to experimental data

An FDS model of the experimental geometry shown in Figure 1 is assembled using cubic cells 5 cm on edge. A side view of the FDS representation of the experiment is shown in Figure 4 (rotated 90º from the “head on” view shown in Figure 1).

2000

1500

1000

500

00 50 100 150 200 250 300 350 400

Time (s)

Hea

t re

leas

e ra

te (

kW)

Total HRRBurner HRR

800

700

600

500

400

300

200

100

0

25

20

15

10

5

00 50 100 150 200 250 300 350 400

Time (s)

Hea

t re

leas

e ra

te (

kW)

Hea

t rf

lux

rate

(kW

/m2 )

Temperature 0.5 m below ceiling

Heat flux at floor (TSC 3)

16

14

12

10

8

6

4

2

00 20 40 60 80 100 120 140 160 180 200

Time (s)

Mas

s lo

ss r

ate

(g/m

2 -s)

ExperimentFDS

45

40

35

30

25

20

15

10

5

00 50 100 150 200 250 300

Time (s)

Mas

s lo

ss r

ate

(g/m

2 -s)

ExperimentFDS

16

14

12

10

8

6

4

2

00 50 100 150 200 250 300 350

Time (s)

Mas

s lo

ss r

ate

(g/m

2 -s)

ExperimentFDS

16

14

12

10

8

6

4

2

00 50 100 150 200 250 300 350 400

Time (s)

Mas

s lo

ss r

ate

(g/m

2 -s)

ExperimentFDS

Each combustible solid surface is assigned material properties estimated above by genetic algorithm optimization and handbook values for gypsum wallboard are used. Since FDS can accommodate only a single gas-phase combustion reaction, its properties are selected to represent a mixture of propane and the combustible solid materials in the mockup. Apart from the reaction and material properties, all default FDS values are used.

Figure 5 compares the measured and modeled heat release rate curves. The overall shapes of the curves match well, but the peak heat release rate is over predicted by 15% and the modeled peak occurs 45s later than the experimental peak. The temperature and heat flux calculations (Figures 6 and 7) trend with the calculated HRR behavior. That is, the peak modeled temperatures and heat flux levels agree well with the analogous peak experimental quantities, but the modeled peak temperatures and heat flux levels occur later than seen experimentally. Figures 6 and 7 show a slight bias toward underprediction of peak temperatures and heat flux levels.

Discussion

If a fire growth model gives sensible predictions of fire development in a rail car mockup fire test, it is reasonable to extend that model to predict the expected fire development in fire scenario involving a full rail car. While beyond the scope of the present paper, such predictions should be considered engineering estimates subject to considerable uncertainty bars rather than absolute predictions.

In addition to making engineering estimates of full-scale fire development in an as-built rail car configuration, one of the biggest promises of this type of fire growth modeling is that it allows the designer to answer “what if” questions.

Figure 2a. Heat release rate Figure 3a. Phenolic liner

Figure 2b. Temperature and heat flux Figure 3b. Seat cushion

Figure 3c. Seat shroud

Figure 3d. Carpet

Figure 4. FDS representation of mockup fire test.

42 | research review

References

Grosshandler, W., Bryner, N., Madrzykowski, D., and Kuntz, K., (2005) Report of the Technical Investigation of the Station Nightclub Fire, NIST NCSTAR 2: Vol. I. Building and Fire Research Laboratory, National Institute of Standards and Technology.

Hopkins, D., (1995) Prediction the Ignition Time and Burning Rate of Thermoplastics in the Cone Calorimeter. NIST GCR-95-667, Building and Fire Research Laboratory, National Institute of Standards and Technology.

ISO 5660-1, 2002, Reaction-to-fire Tests Heat release, Smoke Production and Mass Loss Rate Part 1: Heat Release Rate (Cone Calorimeter Method)

Karlsson, B., North, G., and Gojkovic, D., (2002) Using Results from Performance-based Test Methods for Material Flammability in Fire Safety Engineering Design. Journal of Fire Protection Engineering 12: 93–108, 2002.

Lautenberger, C., Rein, G., and Fernandez-Pello, A.C., (2006) The Application of a Genetic Algorithm to Estimate Material Properties for Fire Modeling from Bench-Scale Fire Test Data. Fire Safety Journal 41: 204-214.

Lautenberger, C., (2007) A Generalized Pyrolysis Model for Combustible Solids”, Ph.D. Dissertation, Department of Mechanical Engineering, University of California, Berkeley. Available electronically at http://me.berkeley.edu/~clauten/lautenberger_dissertation.pdf.

McGrattan, K., Hostikka, S., Floyd, J., Baum, H., and Rehm, R., (2007) Fire Dynamics Simulator (Version 5) Technical Reference Guide, NIST Special Publication 1018-5. Building and Fire Research Laboratory, National Institute of Standards and Technology.

Rein, G., Lautenberger, C., Fernandez-Pello, A.C., Torero, J.L., and Urban, D.L., (2006) Application of Genetic Algorithms and Thermogravimetry to Determine the Kinetics of Polyurethane Foam in Smoldering Combustion. Combustion and Flame 146: 95-108.

Rhodes, B.T., (1994) Burning Rate and Flame Heat Flux for PMMA in the Cone Calorimeter. NIST GCR-95-664, Building and Fire Research Laboratory, National Institute of Standards and Technology.

US Nuclear Regulatory Commission, (2006) Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications, Volume 1: Main Report, Office of Nuclear Regulatory Research, Rockville, MD.

For example, expected fire development could be assessed for several different wall lining materials, allowing the designer to select a material that balances fire performance with other considerations that must be contemplated in the design of rail cars such as cost, durability, ease of maintenance, acoustic damping properties, etc. Additionally, fire development from several different initiating fires can be investigated.

As an example, the model is used here to predict the expected fire development from a trash bag fire, often used in rail industry fire tests. In this example the trash bag fire is assumed to reach a peak HRR of 290kW after approximately 2 minutes. The heat release rate of the assumed trash bag fire and the calculated total heat release rate (including the contribution from the rail car) are shown in Figure 8. The model predicts that only localized burning occurs, consistent with actual fire tests (not reported here). A peak net heat release rate of 100kW occurs around 120s. In comparison, the peak net heat release rate is 1,700kW with the 500kW ignition source (see Figure 5). This modeling suggests that while extreme ignition sources (flammable liquids spills akin to malicious arson) may cause fire spread beyond the area of the initiating fire, “nuisance” vandalism fires such as burning trash should cause only localized burning.

Concluding remarks and future work

The modeling results shown here for fire spread in a real-scale rail vehicle mockup indicate that the peak heat release rate is well-predicted, but it occurs slightly later in the model than in the experiment. The modeling results show that temperatures and heat flux levels are well-predicted when the heat release rate is well-predicted.

Fire development predictions are strongly sensitive to the specified material properties. Consequently, the material property estimation process (here, accomplished by genetic algorithm optimization) is of critical importance for predicting fire development. There are no widely accepted, standardized methods for determining all of the material properties required for fire modeling, and additional research in this area is strongly encouraged.

With careful model verification/validation and calibration, use of FDS 5.0 to make engineering estimates of mockup-scale fire development may be viable, even for non-simple geometries such as rail cars. Although FDS-based fire development predictions are potentially very useful in design applications, it is prudent to view such calculations as engineering estimates rather than absolute predictions.

The accuracy of “blind” fire growth predictions (comparing model predictions to actual-scale fire testing without prior knowledge of the test data) remains to be demonstrated with FDS 5.0, so actual-scale fire testing remains an integral part of the model calibration process. Additional research to assess the capabilities of FDS for predicting fire development for other scenarios is currently being studied.

Related to the fire performance of the rail car investigated in this paper, it is unlikely that a nuisance arson fire, such as a trash bag fire, would lead to fire spread beyond the area of origin. However, extreme ignition sources (flammable liquids spills akin to malicious arson) may cause fire spread beyond the area of the initiating fire. The magnitude and growth rate of such fires could potentially be investigated with FDS, subject to the caveats stated above.

2500

2000

1500

1000

500

00 50 100 150 200 250 300 350 400

Time (s)H

eat

rele

ase

rate

s (k

W)

ExperimentFDS calculation

450

400

350

300

250

200

150

100

50

00 50 100 150 200 250 300 350 400

Time (s)

Hea

t re

leas

e ra

te (

kW)

Total heat release rateInitiating fire (trash bag)

Figure 5. Comparison of measured and modeled heat release rate.

700

600

500

400

300

200

100

00 50 100 150 200 250 300 350 400

Time (s)

Tem

per

atur

e (º

c)

0.5m-FDS 0.5m-Exp.1.5m-FDS 1.5m-Exp.

Figure 6. Comparison of measured and modeled gas temperatures.

25

20

15

10

5

00 50 100 150 200 250 300 350 400

Time (s)

Hea

t flu

x (k

W/m

2 )

#3 - FDS #3 - Exp.#1 - FDS #1 - Exp.

Figure 7. Comparison of measured and modeled heat flux levels.

Figure 8. Calculated heat release rate from trash bag initiating fire.

Acknowledgments

NicholasDembsey, Associate Professor, Worcester Polytechnic Institute, Department of Fire Protection Engineering

research review | 43

44 | research review

Author: David Moorehead

Introduction

The stability of underground tunnels during and after a fire has become an issue due to a number of recent incidents:

• OnMarch24,1999,afirestartedintheMontBlanc Tunnel connecting France and Italy. It burned for 2 days, and over 40 people died.

• DuringtheEnglishChannelTunnelFire,the air temperature reached more than 1000°C, heating the concrete to around 1300°C.

• OnOctober24th2001,afireoccurredintheGotthard Alpine Tunnel. The threat of cave-in along the tunnel hampered the effort of the rescuers to reach the accident site.

One aspect of the issue that needs to be addressed is a clearer understanding of the properties and characteristics of tunnel materials under fire situations. While research has been carried out on the responses of concrete and

related materials to fire, there is little information on behaviour of other material such as natural rock supporting structure in the tunnel.

This project was initiated in April 2003 to study the effects of fire on the physical, chemical and mineralogical properties of Sydney sandstone. This material was chosen as the focal point of the work because there are currently a number of major tunnelling projects in Sydney, Australia. It is hoped that with appropriate data, effective fire protection can be provided to different sections of the Sydney tunnel structures in the most cost effective way.

The experimental part of this work was carried out at the University of New South Wales School (UNSW) of Materials Science and Engineering under a collaboration arrangement between David Moorehead of Arup, Charles Sorrel and George Yang of UNSW.

Sydney Sandstones

Sandstones around Sydney are medium- to coarse-grained and contain about 80% quartz and about 20% clay. Most of the city lies on Hawkesbury sandstone and the Narrabeen Group sandstone, which lies directly below the Hawkesbury group. Sydney sandstone is found in almost horizontal layers ranging in thickness from centimetres to several metres. The layering usually consists of beds of coarse/fine/pebbly sands. The common types of clay minerals in Sydney sandstone are kaolinite and sericite/illite groups, as well as the less stable swelling clays of illite/smectite. Siderite (a carbonate mineral) and other iron bearing minerals such as limonite, goethite and hematite are also present in Sydney sandstone at notable percentages.

A small number of studies have been carried out on the effects of fire on sandstone. These are summarised below:

• ResearchbyHajpal(2002)investigatedtheeffects of fires caused by natural disasters on historical monuments. The study recorded the changes in the physical, chemical and mineralogy of various sandstones sourced in Germany and Hungary.

Abstract

A number of recent incidents involving fires in tunnels have indicated the need to investigate the properties and characteristics of tunnel materials under fire situations. Most studies so far have focussed on concrete and related materials rather than natural materials such as the supporting rock. If the behaviour of different rock types under fire can be better understood, there may be the potential to reduce the use of expensive concrete linings.

In this project, core samples of Sydney sandstone of varying diameters were subjected to a controlled 2-hour fire regime, with thermocouple sensors recording the temperature at varying distances from the heat source.

Examination of the results indicated that the temperatures recorded in the sample cores were progressively lower as the diameter of the core increased from 50 to 100mm and as the distance from the heated surface increased. Inspection also showed that all samples had experienced cracking at 15-22mm from the heated surface. A thin section made for microscopic examination of one of the 50mm diameter core samples showed quartz grains, cavities and cracking in the microstructure.

A Point Load Testing device was used to evaluate the influence of hydrocarbon fire regimes on strength of the sandstone cores. These results showed that the effect of the 2 hour hydrocarbon fire regime was most severe on the 50mm diameter cores. Where the maximum temperatures exceed 300°C, there was a 20% loss of strength, and if the maximum temperature exceeds 500°C, a 70% loss. There was no loss in strength where the maximum temperature does not exceed 250°C.

The results suggest that a heated sandstone surface does not spall due to a build up of water vapour pressure, as the water vapour pressure resulting from the decomposition of the clay minerals and the residual moisture content of the stone can escape freely through permeability paths opened up by the decomposing minerals. This is in contrast to the explosive spalling noted with dense concrete where water vapour pressure at 300°C can reach 8.5MPa and this pressure can exceed the tensile strength of the concrete.

A scale up of the testing to an unrestrained sandstone slab 1.1 x 1.1m x 200mm thick element was tested by the CSIRO. This test showed that the unrestrained block failed. It is believed this was due to a tensile force that developed close to the heated surface.

Further work will involve the testing of samples of sandstone or other sedimentary rock while the samples are restrained to closer simulate the conditions of the rock in a tunnel.

Behaviour of Tunnel Rock Subject to Fire Loading

Figure 1. Aftermath of fire in the Mont Blanc Tunnel

research review | 45

• Inanearlierstudy,Hajpal(1998)foundthatsandstones of different cement types may show different fire resistance. The comparison of results provided useful information in the choice of the replacement or repair materials for historical monuments.

• inapaperentitled“TheEffectofFireDamageon Natural Stonework in Buildings” 3, Chakiabarichas review the earlier work on the effects of fire on stone masonry.

While these works were of interest they did not relate to or simulate the effects of a hydrocarbon fire directed to one surface of a sandstone unit or indeed to Sydney sandstone.

For the purpose of the engineering design for fire protection to a sandstone tunnel, we needed specific information that could only be obtained by simulating a hydrocarbon fire directed at one surface of a Sydney sandstone element.

Materials and experimental set up

The sandstone used in the experiments was obtained from the New South Wales Department of Commerce. The core samples were taken from a site at Pyrmont. They were taken at right angles to the bedding plane, and were described as “1st grade stone”.

The experimental set up was capable of accommodating sandstone cores up to 100mm diameter. It consisted of an LPG burner, ceramic insulation blankets and ceramic fibreboards.

Prior to fire testing, the stone samples were conditioned indoors at 23°C for several weeks to obtain consistent moisture content. The moisture content was measured at 0.4% by determination of weight loss at 105°C.

Three samples for each core size were used, with of 50, 75 and 100mm diameters and approximately 250mm in length. Thermocouples (Type K) were inserted into 5mm holes drilled to a depth of 15mm into the cylinder surface, and spaced at approximately 20mm intervals along the length of the sample, with an additional thermocouple placed at the centre of the heated surface. The square sawn face of the core was presented to the heat source.

The Incorporation of a stainless steel mesh at the exposed surface of the core caused excessive heat loss, making it difficult to match the temperature-time curve at the sample surface to the desired hydrocarbon fire curve. The mesh was abandoned in the set-up for the bulk of the experimental work.

Several experiments were also carried out on spare cores to verify the heating conditions were repeatable and that the temperature time curves measured on the surface of the cylinder closely matched to that shown in the Time-temperature curve in Figure 2.

Results and Discussion

The results of the temperature measurements made on the three cores from each of the 50, 75 and 100mm diameter cores are shown in Figure 3, 4 and 5.

The results indicated that the temperatures recorded were progressively lower as the diameter of the core increased from 50 to 100mm and as the distance from the heated surface increased.

Inspection of the cores after 2 hours of firing showed that they had all cracked parallel to the heated surface at a distance of about 15-20mm from the heated surface.

The cracking of the 50mm diameter cores and the 100mm diameter cores appeared at very similar depth from the heated surface. The heating regimes as measured did not affect the surface cracking of the samples.

It is also suspected that the cracking occurred during heating rather than cooling since the cooling rate (as with the heating rate) would be very different for different core sizes.

The microstructure of one of the 50mm cores was examined using a stereomicroscope. Additionally, a cross-sectional sample (thin section) of this core near the heated surface was also taken for microscopic examination. A microscopic examination of this section show the quartz grains, which are up to 1mm in size, and the cracks left in the microstructure after the 2-hour hydrocarbon-fire heating regime.

Figure 3. Temperature vs. time for 50mm diameter core samples

Figure 2. Hydrocarbon fire temperature-time curve

Figure 4. Temperature vs. time for 75mm diameter sandstone cores.

Figure 5. Temperature vs. time for 100mm diameter sandstone cores

Behaviour of Tunnel Rock Subject to Fire Loading

1200

1000

800

600

400

200

00 50 100 150

Time (min)

Tem

per

atur

e (º

C)

0 mm

22 mm

43 mm

63 mm82 mm103 mm123 mm

75 mm Diameter, No.1

1200

1000

800

600

400

200

00 50 100 150

Time (min)

Tem

per

atur

e (º

C)

0 mm

22 mm

43 mm63 mm82 mm

123 mm100 mm Diameter, No.1

103 mm

1200

1000

800

600

400

200

00 50 100 150

Time (min)

Tem

per

atur

e (º

C)

0 mm

22 mm

43 mm

63 mm

82 mm103 mm123 mm

50 mm Diameter, No.1

1400

1200

1000

800

600

400

200

00 20 40 60 80 100 120 140

Time (min)

Tem

per

atur

e (º

C)

867928

1078 10801068

1014

Hydrocarbon

The results suggest that there is a difference in the sandstone microstructure across the cracking, particularly in the features of the quartz grains and the apparent visual change in the clay binder.

A Point Load Testing device was used to evaluate the influence of hydrocarbon fire regimes on strength of the sandstone cores. The procedure used was similar to that described in Australian Standard AS 4133.4.14 diameter cores. These results suggested that the sandstone cores had similar properties.

The following guidelines are given in the Australian Standard AS 17265

Based on these guidelines, the sandstone used in this work would be classified as High-strength material. The implication is that a protective/support lining may not be required for this stone, particularly in wall situations.

The results of point load testing after the heat treatment have been summarised in Table 1, 2 and 3. In all cases, conservative approximations have been used.

These results show that the effect of the fire testing regimes was more severe on the 50mm cores in comparison with the larger cores. It is suggested the results for the 50mm core should be taken as a guide, since the temperature distribution across this core would have been more uniform.

Therefore, a general approximation of the effect of 2-hour regime of hydrocarbon fire on Sydney sandstone is follows:

46 | research review

Maximum temperature reached

Point load strength index

> 600 °C ~ 0 (due to delamination)

> 500 °C ~ 0.6MPa (70% reduction)

> 400 °C ~ 1.0MPa (50% reduction)

> 300 °C ~ 1.7MPa (20% reduction)

<250 °C ~ 2.1MPa (no apparent reduction)

Maximum temperature reached

Point load strength index

> 600 °C ~ 0 (due to delamination)

> 500 °C ~ 1.2MPa (65% reduction)

> 400 °C ~ 1.44MPa (25% reduction)

> 300 °C ~ 1.92MPa (no apparent reduction)

<250 °C ~ 1.92MPa (no apparent reduction)

Maximum temperature reached

Point load strength index

> 600 °C ~ 0 (due to delamination)

> 500 °C ~ 0.87 MPa (55% reduction)

> 400 °C ~ 1.25 MPa (35% reduction)

> 300 °C ~ 1.81 MPa (10% reduction)

<250 °C ~ 1.95MPa (no apparent reduction)

XRD traces. This suggests that experimental methods used in preparation of samples for XRD work may need to be refined in future work. The kaolinite mineral was characterised by major peaks at 7.15, 3.58 and 2.34Å and quartz by major peaks at (among others) 3.342, 4.257 and 1.8179Å.

While the XRD data were not ideal for detailed analysis, it is evident that the kaolinite in the sandstone at depths of up to 40mm from heat surface underwent dehydroxylation. This is based on the replacement of the peak at 7.15Å with a broad hump of XRD traces in samples taken from 50mm diameter core.

It is known that kaolinite undergoes a hydroxylation process at about 550°C, and that this process can start at temperatures as low as 450°C. Temperature-time curves measured at 43mm from the heat source in the 50mm diameter core samples suggested that the 550°C had been reached for a reasonable length of time after 2 hours of hydrocarbon fire testing. The dehydroxylation of kaolinite is accompanied by shrinkage and formation of amorphous metakaolinite.

Besides the observed dehydroxylation of kaolinite, the – conversion of quartz (~575 °C) can also be expected. Other possible effects in the zone near the heated surface include the melting of minor clay minerals such as feldspar (~920°C); the decomposition of iron-bearing compounds; and formation of mullite (~ 950°C).

The general indication is that serious strength loss of Sydney sandstone in fire situation appears to be associated with the extent of thermal degradation of the clay binder, in this case the dehydroxylation of kaolinite. The cracking would have been caused by densification of the surface layer, similar to that occurring in industrial firing of mixtures of quartz and other clays (for example kaolinite).

Upscale Fire Testing

A further development of this work was carried out in 2006 was to determine the effect of fire on a larger scale sample. To this end a sandstone block from the same source 1.1 x 1.1m x 200mm thick was tested at the CSIRO to the same hydrocarbon fire profile.

Results of this testing are recorded in an internal report dated the 7 February 2006. This work was jointly funded by The Road and Traffic Authority and Arup. This work showed the failure of the unrestrained block a few minutes after its exposure to the fire. It is believed that the failure was caused by the thermal expansion of the heated surface relative to the cooler upper surface. The resulting tensile stresses exceeded the tensile strength of the block. The failure resulted in a crack developing right through the block and its rapid disintegration. This failure mode was not anticipated although a finite element analysis subsequently gave some insights into possible stress conditions near the exposed surface in an unrestrained block. It should also be noted that apart from the possible saving in the cost of fire protection for tunnel this work may

lead to additional safety issues related to the explosive nature of concrete exposed to fire?

Conclusions and steps

The information gathered during the core sample project is useful for a number of reasons;

i) Data has been gathered on the heat flow from the surface of Sydney sandstone that has been heated to about 1050°C for a period of two hours.

ii) A good estimate has been provided of the physical, chemical and mineralogical changes that take place as the distance from the heated surface increases.

iii) Confidence has been gained that a heated sandstone surface does not spall due to a build up of water vapour pressure, as the water resulting from the decomposition of the clay minerals and the residual moisture content of the stone can escape freely. This is in contrast to the spalling noted with dense concrete.

iv) A useful test rig for rock has been established at the UNSW Department of Materials Science.

v) A scale up of the testing to a large unrestrained sandstone block was inconclusive as an attempt to extrapolate to real tunnel conditions in that the rock mass lacked restraining stresses.

The next step proposed will be to carry out further testing on samples that are restrained during the heating regime. The work planned could involve testing at the Institute for Materials Research and Testing Leipzig GmbH. Contact with Dr Frank Dehn of this institute has confirmed that they have equipment capable of carrying out this fire testing while the test sample is subjected to various conditions of restraint.

The information gained will give us more confidence in designing the appropriate level of fire protection needed in sandstone tunnels and reduce the risk of lining over design.

• Lossof20mmsurfacelayer

• 70%reductionofstrengthinlayerwheremaximum temperature exceeds 500°C

• 50%reductionofstrengthinlayerwheremaximum temperature exceeds 400°C

• 20%reductionofstrengthinlayerwheremaximum temperature exceeds 300°C

• Nostrengthreductioninlayerwheremaximumtemperature does not exceed 250°C

It must be stressed that the above suggestions are only applicable to a 2-hour hydrocarbon fire regime. Different criteria would need to be established for other hydrocarbon fire durations. This is due to the temperature-time effect, which cannot be fully covered by using maximum temperature as an indicator. The above guidelines are also limited to unprotected Sydney sandstone. Other situations, such as that of sandstone lined with a shotcrete layer, are not covered in this work.

Limited work using the X-ray diffraction method (XRD) (Cu Kα radiation) was used on samples taken from different distances from the heated surfaces of 50mm and 100mm cores. The two main minerals found in the samples were quartz and kaolinite. Other expected minerals including iron-bearing minerals were not detectable in the

References

Australian Standard AS 4133.4.1 Methods of testing rocks for engineering purposes – Method 4.1: Rock strength Tests – Determination of point load strength index 5i Australian Standard AS 1726 (1993) Geotechnical site investigation, Table A8 – Strength of Rock Material.

Chakrabarti, B T Yates, A Lewry-Construction and Building Materials, 1996, Elsevier Effect of fire damage on natural stonework in buildings.

Hajpal, M Changes in sandstones of historical monuments exposed to fire or high temperature. Fire Technology 38: Issue 4, 2002 October, 373-378.

Hajpal, M and Torok, A Petro physical and Mineralogical studies of burnt sandstones 2nd Int. PhD Symposium in Civil engineering, 1998, Budapest.

Table 1. Fire testing of 50mm core samples

Table 2. Fire testing of 75mm core samples

Table 3. Fire testing of 100mm core samples

Acknowledgments

University of New South Wales, CSIRO, New South Wales Department of Commerce and more recently the Leipzig Institute for Materials Research and Testing.

George Yang (UNSW) for carrying out much of the laboratory work.

research review | 47

48 | research review

Author: Gabriele Vigne and Jimmy Jönsson

Introduction

Arup Fire in Madrid formed part of an ECTP research project in Spain; The Multidimensional City, Subproject SP-6, Task T362: “Extinguishing systems and ventilation control in case of fire”. The global aim of the project is to look into more extensive use of underground spaces. The specific goal of the T362 task is to look into factors that have a significant influence on fires in underground spaces (specifically extinguishing systems and ventilation), and how these factors could be simulated and quantified in a reliable way.

Arup Fire’s specific role was to investigate how computer simulations could be used to predict the effects of fire in underground spaces taking into account different ventilation regimes and water based extinguishing systems on the fire environment. Arup Fire is part of a team (consultants, research institutes, testing facilities, etc) that are investigating the topic by means of large scale fire tests and computer simulations. The main aim is to determine how effective CFD programs are at reliably predicting the interaction

mentioned earlier and identify where improvements need to be made if there are shortcomings.

The work will result in datasets that can be used in collaboration with research centres to validate the CFD programmes and to further enhance their capability.

Overview

Two different series of fire tests were performed in two different tunnels. The first series of tests was done in the January 2007 in the TST Tunnel (San Pedro de Anes, Spain) and the second in March 2008 in La Mina Escuela (Ponferrada, Spain), (See Fundación Santa Barbara).

Both tunnels were equipped with the relevant instrumentation necessary to be able to post process the data produced during the tests.

The tunnels are quite different in nature and due to this different fire sizes and ventilation set-ups were used in each tunnel.

Test Setup

Because of a lack of rigor observed in the instrumentation set-up in the January 2007 tests, Arup Fire ensured that they were closely involved in the test setup for all the subsequent tests carried out in La Mina Escuela. This included advice on the instrumentation, the development of tests runs (fire sizes, ventilation set-ups, etc.)

Arup Fire also had significant input regarding how to measure and calculate the heat release rate (HRR) during the test.

Instrumentation

Both tunnels were instrumented with thermocouples and bi-directional probes. Some thermocouples were placed in the open inside the tunnel to measure gas temperatures, while other thermocouples were placed behind metal sheets in such a way that heat flux measurements could be obtained. The bi-directional probes were used to measure air velocity along the tunnel.

Abstract

Arup Fire in Madrid took part in an European Construction Technology Platform (ECTP) research project in Spain. The goal of the subproject that Arup Fire was involved in was to look into factors that have a significant influence on fires in underground spaces and how those could be modelled in a reliable way.

Full scale test fires were undertaken in tunnels in Spain, each one being repeated at least once to verify results. The first series of tests were done in January 2007 and the second in March 2008.

Research into the modelling of watermist was also performed as part of the project.

Advanced Computational Fluid Dynamic (CFD) models were developed to simulate the test fires and compare results with those of the real fire tests.

The research showed that the CFD program used was capable of producing appropriate and reliable results for fires in a tunnel.

Up to this point the main benefits gained from the research project by Arup Fire are the following:

• Abetterunderstandingofhowtoconstructtunnelmodelsandsimulate fires in tunnels, using advanced CFD models.

• Valuableexperienceregardingtheplanning/designoftestsetups and test runs.

• Wehavedevelopedatechnicalknowledgethatenablesustobeintheforefront of what can be done with CFD programs. This will give us a business advantage over our competitors.

• AlthoughtheworkwasdirectedbytheMadridoffice,anumberofpeoplefrom other offices were also involved, generating a great synergy and knowledge exchange.

Figure 1. TST Tunnel

Figure 2. La mina Escuela Tunnel

Figure 3. TST tunnel-15MW fire

Full Scale Fire Tests in Tunnels and the use of Advanced CFD modelling to Predict Fire and Smoke Behaviour

research review | 49

Experimental Fires

Due to the difference in shape and size of the tunnels different fire sizes were used for the test fires.

In the San Pedro de Anes test tunnel (TST) fires of up to 20MW were tested, in the La Mina Escuela Tunnel significantly smaller fire sizes, up to 2MW, were used.

Water Mist

This part of the investigation project was carried out in conjunction with Marioff who provided technical data of their watermist system and also provided results from the full scale fire tests they commissioned in the San Pedro Tunnel (Leon, Spain) extinguishing a HGV fire with the Hi-Fog Water mist system.

Water mist is a fine water spray consisting of a wide range of droplet sizes, many of which are true mist particles and some of which are considerably larger. The nozzles produce sprays that have a high fraction of very fine droplets, significantly smaller than what is typical for standard sprinklers or water spray nozzles (see Figure 4).

The aim of this study was to investigate the features of the water mist systems and the possibility to model it accurately with FDS ie how effective FDS simulations are at reliably predicting the behaviour of water mist and the interaction of the droplets with the fire.

The Models

The fire modelling of the tunnel utilised the Computational Fluid Dynamics (CFD) program called Fire Dynamics Simulator (FDS).

FDS is a software package developed by the National Institute of Standards and Technology (NIST) of the USA. FDS numerically solves a “form of the Navier-Stokes equations appropriate for low-speed, thermally-driven flow with an emphasis on smoke and heat transport from fires”. Version 5.1.4 of FDS has been used for La Mina’s test whilst version 4.0.7 was used for the TST tests.

For further information on the background and modelling equations of FDS, refer to the FDS User Guide and Technical Guide. These can be downloaded from the FDS website.

Models geometry

It was necessary to construct advanced computer models to simulate the fire tests.

For the second model (La Mina) one of the most challenging goals was to model the rough rock surface of the tunnel. This was achieved with several measurements of the tunnel and by capturing the vast experience Arup Fire has to offer in the creation of complex models.

Both models have their peculiarities and a particular attention had to be given to enable us to model the tunnels as close to reality as possible.

The Water Mist model

Water Mist is not covered in the FDS software as a standard feature. The only way to model Water Mist to date is to work iteratively by adjusting the different sprinkler parameters.

The suppression feature that is implemented in FDS is in fact based on a normal sprinkler system. Water particles are introduced in the models using a Lagrangian approach.

There are three important FDS inputs that govern the behaviour of the watermist. These three parameters have lots of sub-variables that have to be defined. The aim of this analysis is to use all the parameters and sub-variables available without leaving any variable to default values (if no sub-variables are defined the FDS software assigns a default value to each variable).

Results and Discussions

Both models showed that FDS is capable of producing reliable results for non-sprinklered fires in tunnels.

FDS is not yet totally capable of producing accurate results for water mist systems, but we have identified practical ways in which the software could be further developed to address this shortcoming.

This research programme did not cover the real fire tests and simulations of fire controlled with standard sprinkler systems, as these would not normally be used in tunnels.

During the validation work it was seen that a well defined fire was essential to be able to properly validate the model. The reliability of the simulation results was heavily dependent on the quality of the output data from the instrumentation and on the accuracy of the constructed model.

The second series of tests (Tunnel La Mina Escuela) permitted us to accurately predict the Heat Release Rate (HRR) produced by the fire, eliminating a major source of uncertainty that influenced the reliability in the case of the simulations for the TST tunnel.

Main conclusions

FDS accurately predicted the temperature trend measured by the thermocouples during the real fire tests.

The behaviour of the smoke in the tests and in the simulations was very similar. Back-layering occurred in both tunnels and FDS was able to predict it well.

Figure 4. Water Mist

Figure 5. FDS models

© M

ario

ff

Shaft

Fire seat

Thermocouples

Full Scale Fire Tests in Tunnels and the use of Advanced CFD modelling to Predict Fire and Smoke Behaviour

50 | research review

Figure 9. FDS, water particles

References

European Construction Technology Platform, www.ectp.org

FDS, Fire Dynamic Simulator, http://fire.nist.gov/fds

Fundación Santa Barbara, www.fsbarbara.com

NIST, National Institute of Standard and Technology, USA, www.nist.gov

Tunnel Safety Testing S.A., www.tunneltest.com

VTT, Technical Research Centre of Finland, www.vtt.fi

Acknowledgments

Marioff, Finland

Fundación Santa Barbara, Spain

Aitemin, Spain

NIST, USA

Arup Fire-Manchester

Arup Fire-London

Arup Fire-Leeds

Figure 6. Results graph

Figure 8. FDS visual results (smoke)Figure 7. FDS visual results (temperature)

Outcome of research project

The research project is still ongoing and we are involved in the planning of future tests etc. Until now the main benefits gained from the research project by Arup Fire are the following:

FDS–Constructionofmodels Arup Fire has gained a lot of experience in building large and complex models. A lot was learned about how to divide a large space into different meshes, to take obstructions into account and to include complex ventilation systems.

In recent tunnel projects done by Arup Fire some of the tunnel models had a length of nearly 1.5km, they included on and off ramps, vehicle objects and complicated ventilation systems. The results from the research project have been of great benefit for these projects.

FDS–Simulations Different techniques for easily and quickly comparing large amounts of simulations data (output from FDS) with test data were developed during the research project. The influence of different kind of objects on air movement within a tunnel was investigated and important lessons were learned. The importance of dividing large volumes into different types of meshes was easily seen while simulating the large models. The influence from grid size and mesh size on simulation results was noted, something which needs to be investigated further.

The experience gained from the research project so far has helped us to run models more efficiently.

Fullscalefiretesting Arup Fire took part in many full and medium scale tests, which has given us valuable experience regarding test set ups etc.

For the latter tests (Mina Escuela) Arup Fire developed the whole test setup strategy including the schedule for the fire tests that were conducted. This is very valuable for future fire tests, whatever the type of tests to be conducted.

For the tests in the “La Mina Escuela” tunnel the ventilation conditions had a strong influence on the flame shape and the Heat Release Rate (HRR). The first models that were used overestimated the real HRR. The subsequent models that were carried out later, using the mass loss obtained from the load cells placed below the pool fire, showed a much better correlation with the real test results. For the validation process it could be seen how important it was to create subsequent models.

The experiments demonstrated that the burning rate of a pool fire is strongly dependent on the longitudinal ventilation velocity used. A reduction factor of approximately 1.5 and up to a factor of 2 in the associated burning rate was observed between the low-velocity and high-velocity transverse ventilation scenarios.

The work done indicated that FDS can sufficiently account for turbulent fluctuations in large-scale tunnel fires due to rough walls and obstructions, by using specific parameters that represent large deviations in the geometry.

The modelling of Water Mist in FDS is still an open issue and the current capabilities to model the mist spray within FDS are insufficient. Additional research is needed to develop methods to measure and define the initial spray characteristics. Further research is needed to refine the interactions with the sprays with FDS’s flow model. More work is needed on investigating this subject, with international collaboration between institutions such as NIST, VTT and a number of Universities.

Businessbenefit The research program has given us technical knowledge that enables us to be in the forefront of what can be done with FDS. We have developed our contacts with research facilities and clients. This will give us a business advantage over our competitors.

-30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120Distance from fire (m)

Tem

per

atur

e (º

C)

800

700

600

500

400

300

200

100

01:30Test TCFDS TCFDS Gas

Ceiling Centreline Temperature

-20 -15 -10 -5 0 5 10 15 20Downward distance from fire (m)

Tem

per

atur

e (º

C)

800

700

600

500

400

300

200

100

FDS HRRFDS HRR TcFDS PyrolFDS Pyrol TcTest 10

Temperature along tunnel at 925s

4.50

4.05

3.60

3.15

2.70

2.25

1.80

1.35

0.90

0.45

0.00

Smoke view 5.1.5

Water Particles

research review | 51

52 | research review

Author: Jake Hacker

Introduction

The Urban Heat Island (UHI) is a term used to describe the tendency for temperatures in urban areas to be elevated above those in the rural surroundings. The origins of this effect are the manner in which the urban land surface, which is relatively dry, lacking in vegetation, and highly heterogeneous and irregular, affects the storage and release of heat, principally from the sun but also anthropogenic heat inputs – waste heat from buildings and transport. Substantial UHIs have been observed in cities around the world, including London.

London has a city population of around 8m and a metropolitan area population of 12-14m people, making it one of the world’s ‘megacities’. The modern form of the London UHI and its spatial distribution has been revealed through a programme of monitoring across the city, analysis of meteorological station data, and satellite observations. The UHI in London is mainly a night time effect, with minimum temperatures in the city being on average around 3–4°C degrees higher than in the rural surroundings, and broadly speaking shows an intensification towards the centre of the city (Figure 1). The UHI is typically largest under warm summer weather, on clear, still nights, for which instantaneous values up to 9°C have been recorded.

Within the general spatial distribution of the UHI there are also significant local variations according to local land-use characteristics, with more densely built up areas showing stronger UHI intensities. Figure 2b shows the variation of the maximum urban heat island intensity with the percentage of “continuous urban” land-use within a 1km radius centred on the temperature measurement site. There is an increase in the maximum heat island intensity with urban land-use, from 4°C at 30% continuous urban to 6°C at 70% continuous urban.

Abstract

Temperatures in urban areas are typically higher than in the rural surroundings, particularly at night, a phenomenon called the Urban Heat Island (UHI). Heat islands exacerbate the impacts of heatwaves by affecting thermal comfort, human health and the ability to passively night-cool buildings. Urban heat islands are produced because cities have a climatology that is distinct from their rural surroundings, because of the different way the urban land surface stores and releases heat, and also because of the concentration of anthropogenic heat sources – buildings, traffic and other transport and infrastructure.

In recent years there has been growing interest in the planning community in using urban design to reduce the summertime UHI of cities. The objective is to reduce the need for cooling in buildings and infrastructure and thereby assist efforts to reduce energy consumption and carbon emissions. It can also help improve ameniety, by increasing the thermal comfort of outdoor spaces and making natural ventilation of buildings more feasible. Management of the UHI can also help deal with the impacts of climate change. Since climate change is expected to increase temperatures in most parts of the world, reducing the magnitude of UHIs can help to locally offset some of the projected warming. The measures that urban designers can take to reduce the intensity of the UHI include the use of shading, greenspace, water and building massing. In order to understand the relative benefit of these approaches, it is necessary to better understand, quantitatively, the contribution of the various elements to UHI generation.

Here, two models are used to examine how different meteorological and land-use factors contribute to the urban heat island and what the future impact of climate change on the UHI might be. The ultimate aim is to provide urban designers and policy makers with tools that can be used to produce climate scenarios for urban areas, to assess design options that can have a beneficial affect on the climate of cities and to assist the development of climate change adaptation policies.

Climate Scenarios for Urban Design: A Case Study for the London Urban Heat Island

Figure 1. The spatial pattern of the UHI across London averaged over urban heat island events during August 2003, with Borough boundaries superimposed. The scale shows the number of occasions the temperature exceeded 19ºC for more than 48 hours.

Figure 2a. Land-use types from the Centre for Ecology and Hydrology land use dataset: white is ‘continuous urban’, red is ‘suburban’, green is ‘vegetated’, blue is ‘water’; the cross hair is centred on the British museum, the small circles indicate the positions of the temperature sensors, and the larger shaded circle indicates a circle of radius 1km. Figure 2b shows the UHI at these locations.

© G

LA (2

005)

© C

rack

(200

3)

A10

5

0

-5

-10

-15

10

10

1010

15

15

15

15

-15 -10 -5 0 5 10 15

No

rthi

ng,

mile

s

Easting, miles

26

24

22

20

18

16

14

12

10

8

6

4

2

0

Nu

mb

er

of

Oc

ca

sio

ns

research review | 53

25

20

15

10

5

0Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month

0.6

0.5

0.4

0.3

0.2

0.1

0-3 -2 -1 0 1 2 3 4 5 6 7

Heat Island intensity

Ave

rag

e n

um

be

r o

f d

ays

UHIMax >3CUHIMax >4CUHIMax >5CUHIMax >6CUHIMax >7C

Pe

rce

nta

ge

ho

urs

DataModel

Figure 3 shows the number of occurrences of UHImax values of given magnitude, and demonstrates that LHR experiences a significant UHI, which is largest in summer.

Further analysis shows that the UHI varies during the course of the day when averaged across a number of months. This diurnal cycle shows a characteristic shape. During the day the UHI is nearly constant, with a mean value of about 1°C. Following sunset the heat island intensity builds to a maximum and then reduces following sunrise. Based on these observations, the night time UHI amplitude, A, can be approximated by a sinusoidal form:

where tu and td are sun-up and sun-down times and t is the hour of the day, ΔTd is the constant urban heat island intensity during the day and ΔTmax is the maximum night time urban heat island.

The London UHI tends to be larger during periods of calm, cloudless conditions. The inverse relationship between UHImax and wind speed and cloud cover during conditions conducive to UHI development can be described by an exponential decay. Although there is some scatter in observed results, a simple empirical model can be constructed to provide a predictive model of the UHI, as follows:

where ΔT is the heat island intensity as a function of hour of the day, t; A(t) is the amplitude function given by equation (1); the two exponential decay terms represent the dependency on wind-speed, U, and cloud cover, C, with respective decay rates Ue and Ce; and the last term T’ is a stochastic term which imparts a degree of randomness into the prediction of the model, to mimic the scatter shown in the data.

Figure 4 shows the distribution of Heathrow’s heat island calculated from the statistical model using hourly observed data for wind speed and cloud

cover and compared with the observed values of UHI for 1996. The parameters used for this analysis in equations 1 and 2 were: Δtd = 1°C, ΔTmax = 6°C, Ue = 4 knots, Ce = 4 oktas, and T’ was chosen from a Gaussian distribution with zero mean and standard deviation of 0.25°C.

The statistical model is in reasonable agreement with the measurements, particularly for heat island events stronger than 1°C. The poorer agreement between the model and the measurements for UHI<1°C is because the model is designed to capture the strong events: weaker temperature differences are treated with the random temperature fluctuation.

The urban heat island as a dynamic phenomenon

The empirical model developed in the previous section provides a useful framework for examining how different factors influence the statistical characteristics of the UHI at a given location. It may also be used as a predictive model for the statistical characteristics of the UHI for given meteorological conditions.

Examination of heat island events in London indicates that the summertime UHI builds over the course of a period of warm weather, typically reaching a maximum on or around the hottest day of the warm spell. Analysis of the development of UHI events at the London Weather Centre (situated at the centre of the city, close to the British Museum) relative to Beaufort Park as the rural reference shows that in both cases the nights with strong UHI occur during the period where maximum daily temperatures are rising.

A dynamic thermal model of the UHI

In order to investigate the causative processes governing the heat island, a dynamic thermal model which represents essential aspects of the underlying physics has been developed at Reading University. The model is a 1-dimensional column model that solves for the vertical heat balance above an area of land surface, which can be

Figure 3. The Heathrow – Beaufort Park heat island: Average number of days of occurrence in each month of the heat island. Data from 1993-1999.

Figure 4. Distribution of Heathrow’s heat island: values calculated with the statistical model and compared with measurements for 1996

Climate Scenarios for Urban Design: A Case Study for the London Urban Heat Island

Equation 1

∆Td + (∆Tmax – ∆Td) sin ({ (

π t – tdtu – td

A(t) =

Equation 2

∆T(t) = A(t)e – U(t)/Ue – C(t)/Ce +T’

Figure 2b Maximum UHI recorded at the stations shown in 2a) as a function of the percentage of continuous urban land use in the 1km areas about each site; blue symbols are stations on the outer ring of stations and red are stations on the inner ring shown in 2a). [From Crack 2003]

0 1 2 3 4 5 6 7

Maximum heat Island Intensity (ºC)

% C

ont

ino

us u

rban

100

90

80

70

60

50

40

30

20

10

0

Inner

Outer

The understanding that has been built up of the London UHI is that it is produced by the different urban land surface types altering the vertical heat exchange with the atmosphere. Lateral heat exchange across the heterogeneous land surface also causes a ‘smearing’ out of the UHI, leading to the type of distribution shown in Figure 1. This is why the UHI tends to be most intense during periods of low wind speed. At the present time, it is thought that anthropogenic heat inputs (from transport and buildings) play a relatively minor part in the London UHI, although these effects are important for the UHI of cities with areas of intensive use of air conditioning, such as Tokyo, and even in London may be significant locally.

The urban heat island as a statistical phenomenon

To investigate further the characteristics of the London UHI, hourly temperature observations from an urban and a rural site have been examined for the period 1993–98. The urban site was London Heathrow Airport (LHR), and the rural site was Beaufort Park, 50km west of central London. The maximum UHI (UHImax) at LHR is taken to be the maximum daily difference in temperature between LHR and Beaufort Park.

B

54 | research review

References

Belcher, S.E., Hacker, J.N., Powell, D.S. (2005) Constructing design weather data for future climates. Building Services Engineering Research and Technology Vol 26(1) pp. 49–61.

Crack, R. (2003) Parameters Controlling the Spatial Distribution and Temporal Development of London’s Heat Island. MSc Thesis. University of Reading

GLA (2006) London’s Urban Heat Island: A summary for decision makers. Greater London Authority.

Graves, H, Watkins, R, Westbury, P and Littlefair, P. (2001) Cooling buildings in London: Overcoming the heat island. Building Research Establishment, Garston, UK.

Harman, I.N. (2003) The energy balance of urban areas. PhD Thesis, University of Reading

Harman, I.N. and Belcher, S.E. (2007) The surface energy balance and boundary layer over urban street canyons. Quart J Roy Meteorol Soc 132, 2749-2768.

Oke, TR (1987) Boundary layer climates. Routledge

Acknowledgments

Prof. Stephen Belcher, (University of Reading, Department of Meteorology) Mel Allwood (Arup) Alex Nickson (Greater London Authority)

Figure 5. Modelling results from the dynamic thermal models effect of building height (building spacing and width are kept fixed at 20m)

written mathematically as:

where Q* is the net radiation, which forces the heat balance, composed of the energy gained from the sun and loss due to long-wave radiative cooling; QF is the anthropogenic heat source eg from transport, domestic and industrial fuel combustion; QH is the outgoing sensible heat flux; QE is the outgoing latent heat flux; ΔQS is the change in heat storage; and ΔQA is the heat transfer by advection.

The dynamic model solves the heat balance equation at a fine temporal resolution according to the changing heat inputs (eg from solar radiation during the day) to provide a prediction for the land surface temperature and the temperature in the atmosphere above. Here only the vertical heat balance is modelled, corresponding in simple terms to situations existing in windless conditions.

Two versions of the model have been developed: the Surface Energy Balance–Boundary Layer model (SEBBL), and the Urban Boundary Layer energy balance model (UrbanBL). The formulation of the two models is similar but the essential difference is that SEBBL models a flat surface whereas UrbanBL models a corrugated land surface composed of repeating street canyons and roofs. Both models essentially have three components: a diffusion model for the transport of heat in the substrate; the flow of heat, momentum and long wave radiation in the atmospheric boundary layer; and the budget of heat at the land surface. In UrbanBL, each of the four facets of the land surface has a separate surface energy balance that takes into account the incoming radiation, and the penetration of heat into the substrate. Each of the four urban facets can have different material properties and the height, width and spacing of the buildings can be varied. In both models, initial vertical profiles of temperature in the atmosphere and substrate are specified. A geostrophic wind speed and the latitude of the site form the other inputs.

Modelling results

A number of ‘experiments’ were carried out by Arup, with the model running in all cases with parameters for LHR during August for a period of three days.

Figure 5 shows an example from a model experiment using the UrbanBL model to examine the role of building height. As the building height is increased, the ‘thermal mass’ of the land surface increases, but other effects also come into play: an increase in the effective surface roughness and the reduction of the ‘sky view’ through which the street and walls cool by long-wave radiation to the sky. As the building height is increased the near-surface air temperature shows a reduced diurnal temperature range, cooling much less at night than the flat surface. The overall effect is qualitatively very similar to the observed nature of the UHI: an elevation of night-time temperature and a slight suppression of daytime temperatures.

Similar experiments, using both models, have been used to analyse the roles of surface moisture, surface reflectance (albedo) and aerodynamic roughness. The results of these experiments also point to a potentially significant role of these factors in generating aspects of the UHI, particularly the generally drier nature of the urban land surface and its greater aerodynamic roughness. Currently we are developing these models further so that more detail regarding the urban land surface at the building scale can be built in and the effect of different design options investigated.

Climate change and the UHI

A set of regional climate change models has been developed by the UK government for climate change adaptation studies called the UKCIP02 scenarios. These are presented on a 50km × 50km grid. In common with most climate change models, the underpinning model uses a single (rural) land surface type for all land areas.

For this work, changes in London’s climate have been obtained by taking an average of the four relevant model grid squares. The main changes projected for summer include:

• increasedsummertemperaturesofupto7°C(High emissions, 2080s) and increased diurnal range (around 1°C);

• smallchangesinwindspeed(<10%);

• moderatechangesinsolarirradiance(<20%);

• adecreaseinrelativehumidityandanincreasein specific humidity;

• adecreaseinsummerprecipitation(<54%).

In order to assess how climate change might affect the London UHI, the statistical model developed above has been applied to data for Heathrow airport, adjusted (‘morphed’) to modify hourly weather data consistently across a set of variables

35

30

25

20

15

10

50.5 1 1.5 2 2.5 3 3.5Te

mp

erat

ure

(atm

osp

here

firs

t m

od

el le

vel)

(ºC

)

Time (days)

Urban land surface effect of building height (heavyweight facades and roofs)

Field-dryho - 5.0mho - 10.0mho - 20.0m

using timeseries adjustment. The results for the 2080s under the High scenario (the most extreme case in terms of projected changes in UKCIP02) indicate only modest changes in the magnitude of the UHI under climate change. However, it is possible that dynamic changes not captured by the statistical models will lead to changes in the UHI. Further research by the University of Reading is applying meteorological forecast models to examine these and other issues relating to the London UHI. This work is being done under the EPSRC funded LUCID project, on which Arup are also project partners.

Conclusions

Dynamic thermal modelling of urban temperatures under different land use scenarios qualitatively produces observed features of the London UHI, principally the elevation of night-time temperatures and the surpression of diurnal temperature range. This model also indicates a high degree of sensitivity to various urban morphology factors. These results indicate that the observed UHI across London may be interpreted as being due to the amalgamation of numerous ‘local UHIs’, produced by particular types of local land use and building morphology.

The existence of the UHI and its complexity across the heterogeneous urban land surface present challenges to both researchers investigating its cause and to urban designers looking to create beneficial modifications. Increasing urbanisation trends mean that understanding and modelling of urban climate is likely to be an important and developing aspect of managing the impacts of climate change.

Equation 3

Q*+ QF = QH + QE + ∆QS + ∆QA ,

research review | 55

Author: Michael Willford, Arup Fellow

AsanArupFellowandleaderofArup’sAdvancedTechnologyandResearchpractice,MichaelWillfordhasbeeninvolvedinresearchprojectsatArupformanyyears,includinghisworkonhumaninducedvibrationandondampingsystemsforstructures.InthisarticleMichaeloutlineshisviewsontheimportanceofresearchtokeepArupattheforefrontofbusinessanddeliverbestvaluetoourclients.

We are in business, and I believe that our research programme should be informed by the potential for business opportunity. One of the hallmarks of Arup is our attitude to innovation as a generator of value for our clients: research and development has to be an integral part of this.

Research becomes commercially useful when it enables us to deliver one of two things to our clients: either something of value to their business that could not have been achieved before, or something that was available before but which we can now offer at lower cost or with better performance. Ultimately clients will always want ‘more for less’ and our research and development capability can enable us to deliver this tangible value to them. At the same time this maintains Arup’s position as a leader in its fields, and the market advantage that follows from that.

Whilst significant investment may be required to move us to a new position of expertise, very often we conduct research on a modest scale to improve our service incrementally - continually working to keep Arup at the forefront. We do this by self-investing in the areas where we see the most benefit for our business and our clients. Ideas and influences for research are many, but often research that delivers the best value is generated either to address the stated needs of our clients, or from exploiting a new idea or an idea prevalent in one industry but not in another; technical transfer if you like.

For example, ten years ago we conducted novel research into better prediction of vibrations due to footfalls in laboratories and other sensitive facilities. This was driven by the needs of clients on a number of buildings under design at the time, and they benefited directly by getting better performance with lower cost and lower risk. This work put us in the forefront of understanding and technical expertise in this area, and publishing the results led to its adoption in a number of industry standards.

Validation is very important to us. Often there is inertia to the adoption of new concepts and methods in the engineering and construction industries. There is a perception that bringing in new things increases risk: can it be built, will it perform as predicted, will it be accepted by the regulators and authorities, will it actually lead to delay and increased cost? Validation of our research with measurements means that we can demonstrate that our methods are reliable thereby reducing the perceived risk.

As well as conducting our own internal research we remain close to academic and research institutions around the world. Our awareness of the developments in universities, and our constant interaction with academics helps us to deliver more innovative projects than our competitors.

The belief that Arup can shape a better world continues to drive everything that we do. We genuinely believe in pushing the state of design and construction practice forward. We continue to fight for change and improvement, not simply repeating what has always been done. We want to be, and are, seen as different.

This attitude allows us to recruit some of the best technical people and innovative thinkers in the industry. We can talk with authority and influence to our clients, at conferences and in academia. This commitment to research and fresh thinking whilst remaining at the practical end of projects means we can lead with relevance. This is one of the qualities that make Arup unique.

The importance of research to Arup

56 | research review

MichaelWillford

michael.willford@arup.com www.arup.com/advancedtechnologyandresearch

Arup has doubled in size over the last 10 years. This makes the sharing of research priorities, monitoring client needs and research projects challenging. The many research projects Arup funds need to be put in the context of the firm’s needs as a whole.

Arup uses roadmapping techniques to inform our research planning process. Business opportunities are plotted on a timeline against a context of market trends, global drivers of change and competitor activity

Roadmaps provide a strategic focus for our research. They help us to

•Informprioritiesforresearchfunding

•Revealnewmarketopportunities

•Identifyclientneedsandbusinessdirections

•LeverageArupresourceinvestment

•Plancollaborativepartnerengagement

The following research priorities chart is based on our corporate roadmap and shows some of our short to medium term research priorities across the firm. Research in these topics will ensure that we are ready for future markets and can capitalise on opportunities to improve our business performance.

The drivers in the chart have been categorised according to the work conducted by our Foresight team on Drivers of Change. The relevant themes for each driver have been highlighted in the chart, as well as the research priorities and the examples of business opportunities.

The research priorities chart gives a picture of how Arup as a whole is responding to these key world trends and the resulting opportunities. It helps to communicate the focus of our research investment and highlight areas of current or potential external collaborative research.

Arup Research Priorities

research review | 57

Research Team

Prof. Jeremy Watson

Director, Global Research

T +44 (0)20 7755 4210

E jeremy.watson@arup.com

Dr Jennifer Schooling

Research Business Manager

T +44 (0)20 7755 2912

E jennifer.schooling@arup.com

Dr Marta Fernandez

Research Relationships Manager

T +44 (0)20 7755 5105

E marta.fernandez@arup.com

Arup Research Wiki:

http://oasys.intranet.arup.com/

ArupResearch/index.php/Main_Page

Investment in Arup:

http://oldcorporate.intranet.arup.

com/fx/investinarup/

DTX Region Leaders

Australasia

Richard Hough

T +61 (0)2 9320 9321

E richard.hough@arup.com.au

East Asia

Jack Pappin

T +852 2268 3437

E jack.pappin@arup.com

Europe

Richard Terry

T +44 (0)20 7755 3925

E richard.terry@arup.com

Americas

Fiona Cousins

T +1 212 229 1057

E richard.terry@arup.com

DTX Sector Leaders

Buildings

Tristram Carfrae

T +61 (0)2 9320 9477

E tristram.carfrae@arup.com.au

Consulting

Andrew Hall

T +44 (0)20 7755 3025

E andrew-j.hall@arup.com

Infrastructure

Peter Chamley

T +1 212 510 2660

E peter.chamley@arup.com

Arup

13 Fitzroy Street

London W1T 4BQ

T +44 (0)20 7636 1531

F +44 (0)20 7580 3924

W www.arup.com

Research Team

Prof. Jeremy Watson

Director, Global Research

T +44 (0)20 7755 4210

E jeremy.watson@arup.com

Dr Jennifer Schooling

Research Business Manager

T +44 (0)20 7755 2912

E jennifer.schooling@arup.com

Dr Marta Fernandez

Research Relationships Manager

T +44 (0)20 7755 5105

E marta.fernandez@arup.com

Arup Research Wiki:

http://oasys.intranet.arup.com/

ArupResearch/index.php/Main_Page

Investment in Arup:

http://oldcorporate.intranet.arup.

com/fx/investinarup/

DTX Region Leaders

Australasia

Richard Hough

E richard.hough@arup.com.au

East Asia

Jack Pappin

E jack.pappin@arup.com

Europe

Richard Terry

E richard.terry@arup.com

Americas

Fiona Cousins

E fiona.cousins@arup.com

Arup

13 Fitzroy Street

London W1T 4BQ

T +44 (0)20 7636 1531

F +44 (0)20 7580 3924

W www.arup.com

A