Research Review 2010
Transcription
Research Review 2010
Research Review 2010 A Research review was published in 2010 Contents 05 05 Global Global research research 06 06 Arup’s Arup research researchbusiness business 10 10 Climate Climatechange changeextremes: extremes;the the 26 38Monitoring geothermal piles combined combinedeffects effectsofofstorm stormtides tides and andcatchment catchmentflooding floodingininCairns Cairns at Keble College, Oxford 30 42Sustainable and quake resistant priorities priorities 14 14 Geomechanical Geomechanicalcharacterisation characterisationofof 07 07 Meet Meet the the team team 08 08 Regional Regional research research champions champions Manhattan Manhattanschist, schist,aastudy studyofofthe the77 line lineextension extensioncavern cavern façade for existing buildings 34 46Measuring change of coastal defence structures using advanced 3D laser mapping techniques 18 12 An Aninvestigation investigationofoffire fireload load density densityfor forresidential residentialbuildings buildings and in Hong offices Kong in Hong Kong 38 50Digital infrastructure and changing practices in engineering design 22 12 Retrofitting Retrofittingprivate privatehomes homesat scale: investigating at scale; investigating the business thecase business case 26Transient thermal comfort 16 Transient in thermal comfortstations modelling underground 42 54Beasties in the creative workplace 46 58Neighbourhood Pedestrian Analysis Tool (NPAT) modelling in underground stations 30Singapore buried channel research 18Singapore buried channel research 34The Life Cycle Tower: a high-rise 22 Life construction Cycle Tower; a high-rise inThe timber in timber construction 50 62Human induced vibrations on footbridges 56 66 Contact The importance information of research at Arup Prof. Jeremy Watson, heads the research strategy and research business for Arup. His background is in research and technical management in both industry and academe. Jeremy is a chartered engineer, and fellow of the Institute of Engineering and Technology. He is also a visiting professor at the Universities of Southampton and Sussex, a board member of the UK Technology Strategy Board, and Chief Scientific Advisor for the Department of Communities and Local Government. 4 Global research I am delighted to introduce Arup’s Research Review 2010, showcasing research undertaken with collaborators around the world. Arup seeks to deliver research outputs comparable with world-class universities. I continue to be impressed and excited by the quality and innovation of our client and internal development projects. Research is a key contributor to Arup’s success; we deliver new, validated thinking in support of our clients’ projects and aspirations. We maximise value by matching business requirements with developments from the academic sector. Arup experts work in collaboration with the best private and public sector partners. An internal research investment fund supports staff time and provides studentships and other contributions to university collaborators. Research has always differentiated Arup’s work and we encourage it globally. Coordination has been enhanced by the appointment of five Regional Research Champions, responsible for local research strategy and external collaboration. Arup University Arup’s research culture has been developed with the recent launch of the Arup University, providing opportunities for staff to train at Professional, Masters and Doctoral levels. The Doctoral programme is administered by the Research team and delivered in partnership with University College London. Candidates can gain an accredited Doctoral degree based on work-based research training. Research Roadmap Aimed at Arup’s research partners, including funding councils, and informed by client-facing workshops, a new Arup publication offers better coverage and detail than before. The preferred tool for capturing Arup’s research priorities, Roadmapping links global drivers with business opportunities to deliver research agendas. We also offer a customised service to clients developing research strategies, and can broker and manage external research providers. Knowledge is shared through a Research Wiki. Connecting our offices and networks, this promotes collaboration to support research objectives and our clients’ requirements. Collaboration Many Arup staff members are renowned in their fields, publishing in peer-reviewed journals. However while we have much expertise in house, our research offerings are strengthened by strategic alliances and partnerships. When working with individual universities and companies we align our thinking with national research priorities and leverage research investments through national funding programmes. Links with UK research councils are a strong asset. We have a Strategic Partnership agreement with Engineering and Physical Sciences Research Council (EPSRC), which can support mutually developed research programmes. Arup is currently using this agreement to develop research at the Thames Gateway Institute for Sustainability into Sustainable Urban Infrastructures. Arup acts for the UK in an EU programme on Energy Efficient Buildings, providing expertise as a board member of a not-for-profit company which will administer up to €2bn Public Private Partnership research funds over 10 years. We provide the UK’s national contact point, helping to shape E2B calls and promoting UK industry collaboration. Such partnerships and research initiatives are exciting, innovative and great to be involved with. As well as allowing us to grow knowledge 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 our 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. Prof. Jeremy Watson Director of Global Research 5 Arup research business priorities Multidisciplinary projects require specialist project management skills, providing a high level of coordination of a range of expert inputs. We can help clients understand their research needs, develop a strategy, and then develop and manage the research programme and deliver multidisciplinary research projects through offering a variety of services. Strategic Roadmapping is a facilitated workshop process which helps clients to address strategic issues facing their business and produce a multilevel plan which has buy-in from across their organisation. The process involves assessing key drivers, identifying the business opportunities that arise from these drivers, and reviewing the resources and processes that are needed to meet these opportunities; including research, technical investments and training. The workshop is usually supported and informed by a desk study which reviews current and emerging issues in the field. The Research Business Team’s role is facilitation of the workshops, coordination of expert inputs to desk studies, and final reporting. 6 Research Programme Management can provide high-level coordination of research programmes, including theme definition via stakeholder consultation, calls for proposals, peer review with external expert panels, liaison with research councils and fund administration. We are increasingly interested in doing this with industry consortia. Sandpit workshops are intense three to five day workshops used to shape research programmes around a particular topic and develop and fund project proposals. They involve delegates in an ‘immersion’ process, learning about the problem space, creating ideas for research solutions, and identifying and developing project ideas. The Research Business Team can run all elements of this process, from assisting the client in managing delegate applications, identifying stakeholders and mentors for the sandpit, facilitating the event itself, and helping manage the resulting research projects. Meet the team Dr Marta Fernandez, Research Relationships Manager Dr Jennifer Schooling, Research Business Manager Geraldine Ralph, Research Events and Projects Coordinator Marta 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 Intellectual Property. She represents the company on the operational group of the E2B. A chemical engineer by training, her 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. Marta is an honorary lecturer at University College London (UCL). Jennifer manages multidisciplinary research services, enabling clients and staff to access the many skills within Arup, and establishing successful collaborations with external agencies. In previous roles, she managed engineering research and development projects for both academic and commercial applications. Jennifer has also handled new product development in the semiconductor equipment industries, managing new product introductions from concept design to final launch. Geraldine is responsible for all areas of research communications including publications, disseminating research activities and leveraging Arup’s research network and partnerships through events. She also provides project support to the doctoral module programme and CASE awards. Geraldine is a member of the Chartered Institute of Public Relations. Dr Rick Wheal, Research Associate Nausicaa Voukalis, Research Associate Jackie Young, Personal Assistant Rick supports the team with UK representation on the Energy Efficient Buildings programme in the EU. He also has a client-facing role dealing with the commercialisation of research and realising benefits throughout the supply chain. Rick has a background in academe and also building consultancy services, with a strong emphasis on sustainability and architectural design. Nausicaa is a specialist in the firm to realise the value of its Intellectual Property, and is responsible for any paralegal issues required for commercial licensing and collaboration agreements with Universities. A chartered architect with an MBA, Nausicaa’s previous roles also include working on both the design and construction aspects of projects, as well as the management of variable schemes, ranging from small residential designs to large urban planning projects. Jackie provides full secretarial support to Jeremy Watson and admin support to the Research Team. She is often the point of contact for both internal and external research equiries. 7 Regional research champions Tim Keer | Americas Region Richard Hough | Australasia Region Dr Ricky Tsui | East Asia Region Tim Keer is a Principal in Arup’s New York Office and is the Research Champion for the Americas Region. His current responsibilities are focused on operational improvement. He has recently launched the Arup Americas Project Management Academy and is leading various initiatives to strengthen the region’s performance. His technical background is in the analysis of automotive structures and Tim has a particular interest in the development and application of techniques for non-linear analysis and design. Richard Hough is a Principal in the Sydney Buildings Group, and Australasia Region D&T Leader. He is also chair of the Regional Investment Coordination Group which oversees locally-funded investment projects. This role fits well with his regional research champion position, and also with his chairing of the Regional DTX, which promotes R&D and innovation in the regional offices and practices. Ricky is responsible for strategic planning on research, drives Research and Development activities in the East Asia Region and establishes links to external partners. In previous roles, he obtained and conducted over 20 collaborative R&D and technology dissemination projects under Government funding. He has also extensive machinery design experience and won several awards. Dr Mikkel Kragh | Europe Region Dr Gavin Davies | UK-MEA Region Mikkel leads the Technology subsector in the Consulting Practice in Arup’s Milan Office. He is the Research Champion for the Europe Region and he was recently appointed regional Building Physics Skills Network leader for Europe. Mikkel is Senior Visiting Research Fellow at the University of Bath where he is involved in façade engineering and building physics research. He actively promotes integrated design as the Chairman of the Society of Façade Engineering. Gavin is the Design and Technical leader and also UK-MEA Research Champion. His background is as an applied mathematician and engineer. Gavin now leads Arup’s Environmental Physics team in London with particular interest in building physics, microclimate design, fluid dynamics consultancy, climate change adaptation and commercial research and development. 8 Americas Region Australasia Region East Asia Region Europe Region UK-MEA Region Arup Partnerships Boston Chicago Houston Los Angeles New Jersey New York San Francisco Seattle Toronto Adelaide Auckland Brisbane Cairns Melbourne Perth Singapore Sydney Bangkok Beijing Guangzhou Ho Chi Minh City Hong Kong Hyderabad Macau Manila Mumbai Seoul Shanghai Shenzhen Tianjin Tokyo Wuhan Amsterdam Ankara Belgrade Berlin Bucharest Düsseldorf Frankfurt Istanbul Kraków Madrid Milan Moscow Rome St Petersburg Warsaw Wroclaw Abu Dhabi Belfast Bristol Cape Town Cardiff Doha Dubai Dundee Durban Edinburgh Gaborone Glasgow Johannesburg Leeds Liverpool London Manchester Newcastle Nottingham Port Louis Sheffield Solihull Southampton Tshwane Winchester Wrexham Abuja Brunei Bulawayo Cork Dublin Galway Harare Kota Kinabalu Kuala Lumpur Lagos Limerick Penang * The Arup Partnerships is a partnership comprising a number of independent yet inter-related practices, of which Arup Group Ltd is the largest. The Arup Partnerships controls the use of the Arup name, co-ordinates the activities of all the Arup practices, and fosters collaborative working. * 9 Climate change extremes: the combined effects of storm tides and catchment flooding in Cairns Authors: Sam Koci, Tania Cobham, Ragini Prasad 10 Tropical cyclones are responsible for significant coastal flooding due to the high intensity rainfalls and storm tides. This problem increases in small urbanised coastal catchments and is predicted to rise with the effect of climate change. Elevation (m AHD) > 10 4 – 10 2–4 1–2 <1 Fig 1. Digital elevation model of Cairns CBD and environs The influence of storm tidal ocean conditions on flooding in a coastal catchment within the environs of the Cairns Central Business District (CBD) has been investigated for this study. Hydrodynamic flood modelling of both freshwater discharge and storm tidal processes in the catchment was conducted using the one and two dimensional hydrodynamic modelling engine, TUFLOW. The study found both the magnitude of storm tidal ocean conditions and the relative phasing between freshwater discharge and storm tidal conditions, to significantly influence flood behaviour in the catchment. The results have significant implications for flood prediction and the implementation of flood warning systems in low-lying coastal areas. A series of flood maps and comparative flood extent maps have been developed to illustrate the flood behaviour and the relative influence of storm tidal ocean conditions. In order to assess the risks associated with the cooccurrence of these events, an understanding of the likelihood of such cooccurrence is required. Due to the partially dependent relationship between cyclone induced storm surges and freshwater flooding, estimating the likelihood of their cooccurrence requires a complex joint probability assessment. To overcome this problem, most flood analyses use a single, generally conservative tidal signal to reflect tailwater conditions for the modelling of any given design event. An assumption commonly made for coastal catchments is that peak storm tidal levels coincide with the onset of peak rainfall over the catchment. The Cairns area consists of a number of catchments that include natural and modified creeks, channels and piped stormwater drainage systems. The region is bound to the west by steeply rising mountain ranges leaving a narrow coastal plain of, on average, less than 10km width. 4 0000 Tropical cyclones can produce coastal flooding by generating extreme rainfalls and associated freshwater runoff, as well as the elevated coastal water levels generated by storm tides. In cases in which extreme storm tidal conditions coincide with immense freshwater discharge, the extent of resulting inundation can vastly exceed that of either of the independent processes. This problem is particularly significant in small, urbanised catchments, where high intensity rainfalls can produce peak discharges, surcharge and flooding in creeks and stormwater drainage systems, in relatively short time periods. Elevation (m AHD) 1 5000 2 0000 3 5000 Introduction Coastal flooding is becoming a key issue for coastal cities, particularly those along the far north Queensland coastline. The Queensland Government’s prediction for climate change in the region states an increased risk from extreme events due to increased storm tide events. These include sea level rise, increased cyclone intensity and frequency, and a shift southwards in cyclone tracks. The severe coastal flooding often caused by tropical cyclones, presents a significant risk hazard for many coastal communities. 0 5000 Abstract Due to the high intensity rainfalls and storm tides they can produce, tropical cyclones are often responsible for significant coastal flooding. Particularly in small, urbanised coastal catchments, surcharge and flooding in creeks and storm water drainage systems can be significantly exacerbated by the elevated oceanic tailwater conditions associated with coinciding storm tides. With climate change forecasting more frequent cyclones and rising ocean water levels, it is widely expected that the problems associated with cyclone driven coastal flooding will increase. Fig 2. Aerial photography and digital elevation model of study area Many parts of the coastal plain in the Cairns region are relatively low-lying and flat, with the majority of the CBD and surrounding areas below 2-4m relative to the Australian Height Datum (AHD) see Fig 1. During tropical cyclones, the various drainage paths are subject to surcharge and flooding from freshwater discharge, as well as significant influence from elevated storm tidal conditions. The flooding caused by urban drainage surcharge is a major problem in Cairns City, posing significant risk to human life as well as an estimated 17,000 properties. Given the extreme future weather and ocean conditions predicted as a result of climate change, there is a clear need to further develop our understanding of the combined effects of storm tides, fluvial flooding and the implications for coastal communities. A better understanding of the interaction between these processes will ensure that flood warning systems are set up to better respond to predicted meteorological conditions, and more accurate flood prediction models are used in decision making and infrastructure design. 11 Increased flood extent due to increased tailwater conditions 10yr ARI rainfall event 20yr ARI rainfall event 50yr ARI rainfall event 100yr ARI rainfall event Tailwater boundary conditions Static mean sea level (MSL) signal, at 0m AHD Inflow discharge (m3 S1) Inflow boundary conditions 12 Standard astronomical tide for the region with a range of ±0.6m AHD MHWS tide superimposed with a storm surge, producing a peak storm tidal level of 1.78m AHD (50yr ARI storm tide) MHWS tide superimposed with a storm surge, producing a peak storm tidal level of 2.4m AHD (100yr ARI storm tide) Table 1. Hydraulic boundary conditions 8 6 4 2 0 3 Tailwater ocean level (m AHD) Mean high water spring (MHWS) tide with peak level of 0.93m AHD 10 2.5 2 1.5 1 0.5 0 0.5 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Fig 3. Relative phasing between inflow tailwater boundary conditions The primary objectives of this study were to: • characterise the flood behaviour in the coastal catchment • analyse the influence of tidal and storm tidal ocean conditions on the flood behaviour • analyse the sensitivity of the flood behaviour to the relative phasing between freshwater discharge and tidal and storm tidal processes Methodology For this study a hydrodynamic flood model of a coastal catchment within the Cairns CBD environs was developed, calibrated and simulated under a range of hydraulic conditions. This was used to investigate the influence of tidal and storm tidal processes on the flood behaviour. The key objective of this study was to understand the likely influence of extreme tidal conditions on the flood behaviour. It was therefore decided to apply the proposed methodology to one suitable catchment within the Cairns CBD rather than the entire CBD. The Fearnley Drain catchment was chosen as being suitable for the study, because of its poor height relief relative to the ocean and the fact that there is little hydrologic interaction between it and the surrounding catchments. The catchment has an area of approximately 410ha, most of which sits below around 3m AHD. It includes a mixture of residential and industrial land use. The majority of stormwater drainage from the catchment is provided by a single branch man-made channel, which likely replaced one or more natural tidal channels when the land was initially reclaimed. Flap gates have been installed at one point in the channel to prevent saltwater intrusion. 12 Fig 4a. Flooding in catchment during cyclone in January 2009 Fig 4b. Storm tidal conditions during cyclone in January 2009 A dynamically linked 1D/2D TUFLOW model was developed for the Fearnley Drain catchment in order to accurately represent both the narrow Fearnley Drain channel system, and the broader floodplain. The terrain characteristics of the floodplain were represented in the 2D domain, developed from a high resolution digital elevation model and aerial photography, Fig 2. The 1D domain was developed to model the Fearnley Drain channel system which, due to its narrow width and the existence of various hydraulic structures, could not be accurately represented in the 2D domain. In the absence of historic flood data for the catchment, the model was calibrated against the results of a previous flood study of the region. Hydraulic boundary conditions A series of simulations were conducted in which these hydraulic boundary conditions were altered to vary the magnitude of the freshwater inflow and tidal tailwater conditions, and the relative phasing between the two processes. Each combination of inflow and tailwater conditions were modelled with various different relative phasings, and the results of the simulations were analysed. Fig 3. shows the set of relative phasings simulated for the 100yr ARI freshwater discharge, and 2.4m storm tide event. A similar set were simulated for all other combinations of inflow and tailwater boundary conditions. A range of hydraulic boundary conditions were applied to the model to simulate both freshwater inflows and tidal tailwater conditions in the catchment. The boundary conditions used were derived from the WBM (2001) study and are described in Table 1. It is noted that the 1.78m and 2.4m AHD storm tidal signals developed in WBM (2001) were based on research by James Cook University and the Beach Protection Authority. These tidal signals correspond to predicted storm tide levels of 50yr and 100yr average recurrence intervals (ARI) respectively, with consideration of sea level rise and other predicted climate change conditions including increased cyclone frequency and intensity. Results and discussion Consistent with historic records and recent observation of flood events see Fig 4a. and Fig 4b. the simulated model results, see Fig 4c. found the low-lying Fearnley Drain catchment to be significantly affected by flooding due to both freshwater discharge and tidal processes. Particularly in the downstream, tidal regions of the catchment, elevated ocean water levels caused by extreme tidal conditions were found to have a major influence on the magnitude and extent of flooding. The flood maps produced from the study illustrate this flood behaviour. Tidal and storm tidal ocean conditions were found to have a relatively significant influence in Inundation area (ba) 180 20yr ARI 50yr ARI 100yr ARI 160 140 120 100 80 60 40 100yr ARI (2.4m AHD) storm tide 0% 0% 17% 117% 20yr ARI 0% 0% 6% 65% 50yr ARI 0% 0% 4% 39% 100yr ARI 0% 0% 4% 36% Influence of tailwater conditions on peak flood levels in catchment - 100yr ARI freshwater discharge event 3 2.5 2 1.5 1 Mean sea level Standard tide MHWS tide 50Yyr ARI storm tide (1.78m) 100yr ARI storm tide (2.4m) 0.5 0 20 0 50yr ARI (1.78m AHD) storm tide 10yr ARI Peak flood level in channel (m AHD) 220 Base flow 10yr ARI MHWS tide Table 2. Increased flood extent due to increased tailwater conditions Influence of tailwater conditions on extent of unundation 200 Standard tide Sensitivity to relative phasing - 10yr ARI/ 2.4m Storm tide 2.450 Peak flood level (m AHD) Fig 4c. Simulated flood results in catchment % increase in flood extent from MSL tailwater conditions 2.400 2.350 2.300 2.250 2.200 2.150 Point 1 Point 2 Point 3 2.100 Point 4 Point 5 Point 6 2.050 Static MSL Standard MHWS 1.78m tide tide Storm tide Tailwater condition 2.4m Storm tide 0 1000 2000 3000 4000 Upstream distance from coast (m) 1 2 3 4 5 6 7 8 9 10 Time at which peak tidal level phased (hours from start of storm) 11 Fig 5. Influence of tailwater conditions on extent of inundation Fig 6. Peak flood levels produced by 10yr ARI fluvial flood event under various tailwater conditions Fig 7. Sensitivity to relative phasing between 10yr ARI freshwater discharge and 2.4m storm tide the tidal regions of the catchment and limited influence in the non-tidal regions. The significant change in flood behaviour beyond the flap gate (located approximately 1.4km along the Fearnley Drain channel from the coast) is illustrated in Fig 6. as the peak flood levels produced by the different tailwater conditions remain relatively constant beyond this distance. representative points in the catchment, produced by variation in the relative phasing between the 10yr ARI freshwater discharge and 2.4m storm tide, are illustrated in Fig 7. The figure shows a variation in peak flood levels of up to 0.2m at one point in the catchment. As demonstrated by the results of this study, the factors underlying these assumptions can have a significant influence on the flood behaviour. When adopting these assumptions it is therefore essential that the sensitivity of the flood behaviour to these factors is appropriately assessed. Throughout the tidal regions large variations in peak flood levels and the extent of inundation were produced by different tidal and storm tidal conditions, for all freshwater discharge events. For the 100yr ARI freshwater discharge event, peak flood levels in the tidal region produced by the 100yr ARI (2.4m AHD) storm tide were more than 0.5m greater than those produced under normal tidal conditions, see Fig 5. Fig 6. and Table 2. This shows the increase in flood extent due to changed tidal conditions. This demonstrates that tidal conditions up to a certain level (MHWS) have minimal impact on flood extents. These results also demonstrate that there is a threshold limit above which flood extents significantly increase. Table 2. shows that this occurs between MHWS and the 50yr ARI (1.78m AHD) storm tide. This is important when determining flood warning thresholds in the development of flood warning systems. The sensitivity of the flood behaviour to the relative phasing between hydraulic processes, was found to depend mainly on the magnitude of the tidal conditions. The relative phasing with the smaller tidal events (standard and MHWS tides) had very little influence on the peak flood levels recorded at any point in the catchment, whereas the relative phasing with the larger storm tide events had a relatively significant influence on peak flood levels, particularly in the downstream regions of the catchment. As an example, the variation in peak flood levels at six In general, the influence of relative phasing between the freshwater discharge and tidal processes was found to diminish with further distance from the coast as well as increased magnitude of freshwater discharge and in cases in which tidal processes completely dominate the flood behaviour. Conclusion and next steps Using the Fearnley drain catchment in Cairns as a case study, this investigated the influence of storm tidal ocean conditions on the flood behaviour in coastal catchments. The study found both the magnitude of storm tides, and the relative phasing at which they coincide with freshwater flooding, to have quite a significant influence on the flood behaviour in the catchment. Although the extent of this influence will depend on a large number of site-specific conditions, it is likely that this result would also be observed in other similar coastal catchments. In coastal areas like Cairns, where tropical cyclones frequently produce immense rainfall and storm tide events, an understanding of the interaction between tidal and catchment driven flooding is critical in predicting the flood behaviour and implementing flood warning systems and mitigation strategies. In many flood studies of coastal catchments, accurate estimation of the joint probability of cooccurrence of significant rainfall and storm tide events is not feasible and a number of assumptions are made in adopting the hydraulic conditions for given design events. 12 Acknowledgements This project was conducted at James Cook University, Townsville, under the supervision of Professor John Patterson. We would also like to acknowledge the contributions to the project made by the following: Bill Syme, BMT WBM for providing the TUFLOW license, as well as substantial project advice and assistance; Tim Smith, Nola Strawbridge, Rudd Rankine Cairns Regional Council for providing a wide range of data; Tony Martin Queensland Department of Main Roads for providing the digital elevation model data and aerial photography used in the project. References Baddiley P., The flood risk in Cairns, Natural Hazards 30, Kluwer Academic Publishers, 2003, 155-164. McInnes K. L., HubbertG.D., Abbs D. J. & Oliver S. E. A numerical modelling study of coastal flooding-Meteorology and Atmospheric Physics, Meteorol. Atmos. Phys. 80, Springer-Verlag, 2002, 217-233. WBM 2001 Cairns CBD and Environs Drainage Management Plan, Cairns Regional Council. Queensland Government 2009, Climate change in the Far North Queensland Region. 13 Climate change extremes: the combined effects of storm tides and catchment flooding in Cairns Increased flood extent due to increased tailwater conditions Freshwater discharge event Geomechanical characterisation of Manhattan schist, a study of the 7 line extension cavern Author: Seth Pollak 14 New York intro text City struggles to cope with growing demands placed on an aging metro system. The 7 line extension, the first major system upgrade in decades, is currently under construction and will be the first to be completed when it opens to the west side of Manhattan in 2013. Fig 1. Station cavern top heading excavated to full span at contact zone revealing granitic rock to the left and Manhattan schist on the right Abstract Arup was engaged by S3 II Tunnel constructors to provide initial support design for two Tunnel Boring Machine (TBM) starter chambers, three shafts, five cross passages, and the 22m wide x 300m long main station cavern. Design challenges included low cover and proximity to sensitive structures, among others. On site, Arup carried out verification of the design through geologic mapping, rock mass classification, detailed joint characterisation, and monitoring of instrumentation. As this is the first shallow cavern to be constructed in Manhattan in 40 years, there was little precedent for how the rock mass would respond to the construction sequence and the adopted support scheme of rock bolts and shotcrete. A study was undertaken to verify assumptions made during design regarding ground relaxation, impacts of junction construction and joint properties. A site specific rock mass classification correlation has been derived and a statistical database of joint properties compiled. A methodology for junction design in Manhattan Schist is proposed, based on measured ground movements. The observations and measurements made during construction of this study will benefit the design of five similar caverns slated for future construction in Manhattan. Introduction The 7 line extension project is a 2.4km long two-track subway expansion extending service from the existing Times Square Station at 41st Street and 7th Avenue out to the west side of Manhattan and terminating at a new station to be located at 34th Street and 11th Avenue. The centerpiece of the study and main focus of this research is the 300m long-mined station cavern at 34th Street, shown in plan on Fig 1. The station consists of a two level public area (200m long) with interlocking caverns on either end which facilitate track crossovers. Challenges associated with construction of the cavern include: • urban setting • less than 1 span rock cover (type 14m for 21m span) • close proximity to active rail lines and historic buildings (minimum 8m) • lack of precedent with regards to cavern construction experience in NYC workforce These factors, coupled with adverse geology, made excavation of this station anything but routine. Further complicating the construction was the need to form six perpendicular junctions which would be driven and left as stub tunnels for future entrance connections or utility adits. The cavern top heading cross section is shown in Fig 1. The full section has an excavated span of 21m and height of 16.5m. The adopted construction sequence included a staggered, multiple drift top heading (3 drifts x 50m 2 each) excavated through the full length of the main cavern, followed by benching and interlock cavern excavation, also using the multiple drift approach. Methodology Geologic setting Two different rock types are present along the cavern alignment. A central intrusion of granitoid rock, described as euhedral, crystalline, acidic, and mica deficient is present over roughly 50% of the excavation. There exists a central depression within this rock mass which is infilled with a “mica schist” often refered to as “Manhattan schist” and minor pegmatite. The contact between the granitic rock and the schist is generally intact to moderately weathered. The southern limb of the intrusion is located at the cavern/south interlock transition, where rock grades back into mica schist. The northern limb is characterised by a faulted contact between the granitic rock and mica schist. This contact was found to be approximately 1m thick and contain decomposed rock and breccia in a matrix of green, low plasticity clay. Following the contact, the schist is characterised as faulted and Fig 2. Faulted contact zone between granitic rock (bottom) and schist sheared with sub-vertical to vertical foliation fractures and seams. The fault system is a series of sub-parallel en echelon features striking obliquely across the excavation trend. These structures are discrete and bounded by higher quality schist, similar to that found in the southern end of the cavern. The total length of cavern that was excavated through this poor quality zone was 100m. In terms of rock mass behaviour, two distinct forms of Manhattan schist have been observed during excavation. The first type is what could be termed a “classic” schist: lack of blockiness, a dominant fabric orientation where foliation is very close and discrete foliation joints are difficult to discern. This fabric is present in rock of the TBM starter tunnels, located at the south end of the study. Conversely, the behaviour of the schist in the cavern is dominated by intersecting discontinuities, containing both discernable foliation joints and wider spaced sub-vertical cross foliation joints. This leads to what is termed “blocky” ground, one in which the Geological Strength Index (GSI) has been applied to derive geotechnical design parameters. The intact rock strength is also higher, resulting in schist whose foliation fabric does not necessarily control rock mass behaviour. Jointing in the granitic rock is generally orthogonal with the occasional sub-vertical joint cross cutting through the mass. Horizontal joints are typically open and clay filled (up to 15mm), and moderately to highly continuous with measured 15 RMR (89) 90 Rock mass classification correlation Granite 70 Log. (Schist) y = 6.5808Ln(x) + 42.698 R 2 = 0.3426 60 Schist 40 30 20 1 Q (Barton, 1974) 10 Results and discussion Rock mass characterisation During the design stage, the engineer is forced to make many assumptions regarding the conditions of a rock mass, sometimes only with 50mm diameter rock core and lab test data. This is especially true in urban construction where rock outcrops are rare and those that do exist have been physically weathered and/or chemically altered for decades. If rock mass classification is being done from rock core, the parameters contained within the Rock Tunnelling Quality Index (Q system) are typically easier to determine than those associated with the Rock Mass Rating system (RMR). Both of which are types of empirical classification systems. Several correlations exist between the two systems, but the most accurate correlation will be one derived from site specific data. During geological mapping of the 7 Line study, over 250 such classifications have been carried out by a limited number of engineers and geologists which minimises variability. Fig 3. shows the following site specific correlations have been derived for granitic rock and Manhattan schist. A case can be made that these correlations are valid based on the fact that the trend line for all the data is approximately equal to the well known correlation given by Bieniawski (1989) which is commonly applied in rock mass RMR = 91n Q + 44 (1) classification: It is also a difficult, if not impossible task to accurately characterise large scale joint properties from borings. If outcrops are unavailable for scanline observation, an educated guess must be made for these parameters, which in most cases leads to a conservative design. However, properties that can only be quantified by mapping, such as joint persistence, large scale waviness, and roughness, are all key input parameters into discontinuum modelling and key block analyses. For example, there will be a drastic difference in modelling results (support requirements) between a model with planar, continuous joints versus one with low persistence joints, which can only be mobilised by shearing through the intact rock. Likewise, if the amplitude of large scale waviness of the joint is great, dilation is inhibited and the joint can only mobilise if the intact rock is sheared through. * 100 21m 16 All the properties measured above can be used to improve the joint shear strength models used in discontinuum numerical models. Ground relaxation and interaction of a multiple drift excavation sequence The concept of stress redistribution, along with subsequent deformations caused by formation of a plastic zone, ahead of an advancing tunnel face is well documented since being introduced by Panet and Guenot (1982). Since then, numerous papers have been authored on the empirical and analytical shape of the longitudinal displacement profile for an advancing circular tunnel at a constant rate, ie TBM tunnelling. These profiles are then applied in 2D numerical models to account for the 3D face effects through methods such as convergenceconfinement. In this method, stress tractions of equal and opposite magnitude to the radial initial stress (o) are applied along the tunnel periphery and then systematically reduced by a factor, λ whose value is between 0 and 1 prior to the support installation stage: (2) where σ r is the ‘relaxed’ radial stress. The “relaxation factor”, λ, is defined by the ratio of the amount of radial deformation that takes place in the ground prior to arrival of the face (μr(o)) to the total amount of radial deformation that takes place at infinity behind the face (μ r (∞)). λ = μ r (o) / μ r (∞) Borehole extensionmeter Fig 4. Plan of cavern: top heading construction sequence and instrumentation arrangement Therefore, improved quantification of joint parameters during the design stage can lead to an optimised, cost effective support scheme through validated numerical models. In order to verify the design, joints were characterised whenever possible by observation and profile gauge. Large scale waviness was calculated by taking the ratio of maximum amplitude to wavelength over the visible trace length. The Joint Roughness Coefficient (JRC), as determined from profile gauge measurement, was used to estimate Joint Roughness (JR) from tables prepared by Barton (1987). Over 120 joints comprise the database. Table 2. gives statistics for the primary schist joint set. σr = (1- λ) σo * 9m Fig 3. Site specific rock mass classification correlation for 7 line study trace lengths of 10-15m. These joints also produce the majority of water inflow into the cavern with flows of up to 10 l/min observed. * D = 7m 9m Log. (Granite) y = 9.4359Ln(x) + 48.136 R 2 = 0.4958 Log. (All) y = 8.6272Ln(x) + 44.258 R 2 = 0.4013 50 10 0.1 * Plan of a cavern 80 (3) For TBM tunnelling, which takes place at a more or less constant rate, a smooth longitudinal deformation curve is assumed as the stress redistribution reaches a quasi steady state during excavation. This is a valid assumption for continuous excavation, but not for cyclic drill and blast tunnelling. This aspect was explored by observing ground relaxation in the three top heading drifts. Drift No. of Ave. first measurements response [D = 7 m] Ave. relaxation Ratio [λ ± σ] 1 5 1.2D 0.70 ± 0.28 2 2 1.0D 0.71 ± 0.17 3 4 0.7D 0.66 ± 0.07 Table 1. Summary of drift relaxation in cavern top heading The method of excavation is directly responsible for the stress path that the ground is subjected to. The stress path in turn will determine the extent of the plastic zone around the face and hence the deformation. For the 7 Line study, the situation is further complicated by having a staggered, multiple drift construction sequence in which the plastic yield zone ahead of and around the drifts will interact with each other and cause further deformations. Multipoint extensometer arrays installed from the surface in advance of construction were monitored by a real time data acquisition system so the movements could be correlated with the blasting cycle. The general layout is shown in Fig 4. By monitoring the response of multipoint borehole extensometers in front of the three advancing drift faces, a comparison can be made to the typical assumption of λ = 0.3, which is commonly used for TBM tunnel modelling. For the central heading, the relaxation factor was calculated by taking the ratio of movement occurring in the instrument up to the point of face arrival to the subsequent movement recorded up to the point when Drift 2 arrives at the instrument. This value is not quite “pure” relaxation in the sense that bolting is typically completed 1D behind the face, resulting in slightly higher λ values (ie less movement measured due to support installation). Nevertheless, the values obtained give a good indication of the range and magnitude of relaxation that could be applied in a staged numerical model. Similar observations were made for Drifts 2 and 3. A summary of the findings is presented in Table 1. The average distance of the instrument from tunnel face at first response was also observed for each drift to give an indication of the extent of the plastic zone ahead of the face. The general trend of the data suggests that the proportion of movement that occurs ahead of the face is over half of the final value in each of the drifts, ie greater than 50% relaxation. This is because the sudden strain release <1 1-3 3-6 6-9 >9 0 27 59 14 0 % per total surveyed 2-4 4-6 6-8 8-10 10-12 % 6 28 25 22 14 6 Jr 0.5 1 1.5 2 3 4 JRC % Nature of infill % Large scale waviness % >12 0 3 75 11 11 0 0-2 2-4 4-6 6-8 8-10 >10 3 3 % i° 17 31 28 19 None surface stained non-cohesive clay 96 0 4 0 None 0.01 0.01-0.02 >0.02 42 0 0 58 Run 2D numerical model of main cavern with standard support and record roof deformation Reduce stiffness of intact rock or rock mass (depending on model type) until 15% increase in roof deformation is produced Check capacity of standard support Table 2. Summary of Manhattan schist foliation joint characteristics Adit Span ratio [DAdit/ DCavern] Rock mass classification [Q/RMR89] Cavern Adit E1 0.6 1.3 / 47 1.3 / 44 E2 0.6 1.0 / 39 T3 0.6 1.0 / 43 Δδm/δmo [%] Δδa/δao [%] Fail N/A 172 2.7 / 49 8 280 0.7 / 46 13 93* Table 3. Rock mass behaviour in junction areas of 7 line main cavern. | m (o): main cavern crown extensometer movement prior to adit | a (o): main cavern adit side extensometer movement prior to adit | *Spiles used over adit caused by blasting produces more plastic damage in the rock mass than gradual strain relief, as is the case when using mechanical excavation methods. Applied in a model, this would lead to larger deformations but smaller load on the support. The difference in increasing the bolt pattern spacing by even 0.3m could have significant cost savings over the length of a 300m cavern. Rock mass behaviour at junction locations Junction design is commonly carried out in several ways. The first is an empirical approach using the Q system (namely Jn x 3 where Jn = joint set number). The increase in joint number is to account for the addition of a third dimension, formed by the intersection, along which the potential for kinematic wedge failure is increased. The second way is to utilise a structural beam-spring model to design the thickness of the shotcrete. This requires an estimate of rock load on the lining and does not account for any rock-structure interaction (ie no arching effects). Shotcrete capacity is usually designed to keep combinations of moment and thrust within the elastic envelope, neglecting the post cracking benefits of steel fibers. Both methods are typically conservative. Complex 3D models can be useful, but are time consuming and sometimes difficult to interpret. Another question is how far to extend the additional reinforcement around either side of the penetration. One adit diameter is a typical rule of thumb value used during design. Construction of junctions gave the opportunity to study how the rock mass behaved by observing ground movements recorded by the extensometers. In particular, three penetrations formed perpendicular to the main cavern were studied. Direct comparison between the junctions was possible as they were all of the same size and rock mass classification of the cavern roof revealed nearly identical rock mass quality. Note that the typical magnitude of cavern crown deformation recorded was 10-15mm. Table 3. summarises the results. The effects of spiling above the T3 junction breakout are evident in reducing the deformation by approximately 50% compared to the other two junctions. The adits were blasted only after the main cavern top heading had been fully excavated and supported. Roof movement in the cavern of less than 15% additional strain (compared to cavern movement prior to junction excavation) was observed due to E2 and T3 junction construction, both of which were in Manhattan schist. The E1 extensometer was damaged during blasting, so no reading was possible. The extent of the plastic zone around the junctions was smaller than assumed during design. Extensometers located 4.5m offset from the edge of the adit showed no response during excavation. This corresponds to a zone of influence of less than 0.5D adit either side of the penetration. In addition, extensometers located on the far side of the cavern (opposite the adit) showed no response to adit construction. Install heavier support and re-run model OK Increase thickness of shotcrete 0.5D around adit for potential wedge failure and adopt Fig 5. Methodology proposed Ground behaviour in the vicinity of junction penetrations was also studied which showed that an additional strain of less than 15% in the cavern crown was recorded following junction excavations. The extent of influence into the rock mass due to the junction was limited to 0.5Dadit around the penetration and to cavern centerline. Based on the data, a new methodology for designing perpendicular junctions in Manhattan schist has been proposed. With several caverns in the planning and design stages in similar geological conditions, the information gained during construction should go a long way in helping Arup develop support designs which are cost effective and tailored to the unique combination of rock mass and in situ stress conditions found beneath the streets of New York. Acknowledgements Based on the results presented above, a new design methodology is proposed for perpendicular junctions formed in Manhattan caverns is shown in Fig 5. This method is relatively quick and produces a support design that is based on site specific data. The increase in shotcrete thickness can be restricted to local containment of wedges around the penetration without needing to be extended across the entire cavern span, keeping the design cost effective. Conclusion and next steps The successful construction of the 7 Line study has provided a unique opportunity to study critical assumptions and carry out rock mass behaviour observations for the purpose of refining shallow cavern design methodology in Manhattan schist. Joint characterisation has quantified difficult to define properties such as large scale waviness, persistence, and roughness for each rock and joint type. The effects of multiple drift excavation on ground relaxation were studied via multipoint borehole extensometers which revealed that in all cases, more than half of the total recorded strain occurred ahead of the drift faces. This is an amended version of the paper ‘Geomechanical Characterisation of Manhattan Schist – A Study of the 7 Line Extension Cavern’ presented at the 2010 International Tunnelling Association World Congress, Vancouver. Thanks to Dr. Chris Snee of SneeGeoconsult, Brian Balukonis of GZA Geoenvironmental, Inc., S3 II Tunnel Constructors and New York City Transit Metropolitan Transportation Authority Capital Construction. References Barton, N., Lien, R., Lunde, J., Engineering classification of rock masses for the design of tunnel support, Springer-Verlag, Rock Mechanics 6, 1974, 189-236. Barton, N., Predicting the behaviour of underground openings in rock. Maunel Rocha Memorial Lecture, Lisbon. Oslo: Norwegian Geotechnical Institute. 1987. Bieniawski, Z.T., Engineering Rock Mass Classification, John Wiley and Sons,1989. Panet, M., Guenot, A., Analysis of convergence behind the face of a tunnel. In: Tunnelling ’82. IMM, London, 1982. 17 Geomechanical chatacterisation of Manhattan schist; a study of the 7 line extension cavern Persistence of joints (m) An investigation of fire load density for residential buildings and offices in Hong Kong Authors: Mingchun Luo, M Liu, SM Lo and KK Yuen 18 The average fire load in Hong Kong is higher than in Europe. The implication is that local information is important for determining the fire load density and a reliable fire safety engineering study should be supported by local data. Abstract Performance-based fire safety engineering design is now widely adopted. The potential and strength of a performance-based fire safety engineering approach becomes an important tool for specialist consultants. However, one major deficiency is that it requires appropriate design fires to determine the fire scenarios as well as the probability of fire occurrence. A design fire can be characterised by the heat release rate which may be inferred by the fire load density of the space concerned. Accordingly, a comprehensive study on the fire load density is required to be carried out in order to support the development of performance-based fire safety engineering design. Planned field surveys will be conducted in selected typical buildings. As the physical dimensions and available combustible materials can be roughly estimated, the approximate weight of those combustible materials can also be estimated and, by assuming the type of materials, a rough range of the fire load density can then be established. This study presents the results of the surveys for residential buildings and also presents the summary of the survey results for offices in Hong Kong. Data collected For information only Factors affecting fire loads in residential buildings 1 Series number 2 Address Dependent variable in fire load calculation Other information Variables in electrical equipment 1 Residents’ gender 1 Types of equipment 3 Number of elderly 2 Number of equipment 4 Number of children 3 Volume occupied 5 Number of students 4 Weight 6 Number of partition walls Shown in part I GENERAL INFORMATION 2 Number of residents Variables in Furniture 1 Types of furniture 2 Number of furniture 3 Volume occupied 4 Weight 5 Material used Variables in living rooms/bedrooms Variables in bathrooms Variables in kitchens Shown in part II Shown in part II Shown in part II SURVEY OF COMBUSTIBLE MATERIALS – LIVING ROOMS/BEDROOMS SURVEY OF COMBUSTIBLE MATERIALS – BATHROOMS SURVEY OF COMBUSTIBLE MATERIALS – KITCHENS Fig 1. Structure for the survey form Introduction In Hong Kong and many large cities in China, rapid urbanisation has caused the construction of high-rise buildings. In view of the rapid growth of population and commercial activities in the urban areas, the construction of high-rise buildings will continue. Regarding fire safety protection design of such complex high-rise building, the codecompliant or prescriptive approach may not be the only way to assure the fire safety level. Performance based fire safety engineering design is now widely adopted. The potential and strength of performance-based fire safety engineering approach manifests itself to become an important tool in specialist consultantancy. However, one major deficiency is that it requires the appropriate design fires to determine the fire scenarios as well as the probability of fire occurrence. A design fire can be characterised by the heat release rate, which may be inferred by the fire load density of the space concerned. Accordingly, a comprehensive study on the fire load density is required to be carried out to support the development of performance-based fire safety engineering design. Planned field surveys will be conducted in selected typical buildings. As the physical dimensions and available combustible materials can be roughly estimated, the approximate weight of those combustible materials can also be estimated and, by assuming the type of materials, a rough range of fire load density can then be established. This study presents some results of the surveys. Methodology In this project, the procedure for conducting the study was as follows: • specify the type of residential buildings to be investigated • specify the number of samples to be carried out • define the scope of combustible materials to be counted eg papers, books, computer equipment, furniture, partitions and miscellaneous • observe the physical dimensions and characteristics of the various content items to determine the weight of the counted items. This can be done by estimating the dimensions and other information from the photographs taken on site. Weight of materials other than wood were converted to an equivalent weight of wood using a multiplying factor based on the ratio of the heats of combustion • data collected by the survey were analysed using statistical measures (eg t-distribution to summarise the results for the fuel load density in various buildings 19 A Ra es's equation te = te FLD Af AT Av h C 0.067 Af FLD (AT Av h) 0.5 (1) = the equivalent duration of fire D = fire load density = floor area = total area of wall = surface area of ventilation = height of the opening QL&B = ∑ 0.3VLC ρLC H LC + ∑ VLF ρLF HLF + ∑m LE HLE (2) FLDL&B = (3) ∑0.3VLC ρLC H LC +VLF ρLF HLF + ∑m LEHLE A L&B FLDW = (5) ∑0.3VWC ρWC HWC + ∑VWF ρWF H WF + ∑mWEHWE AW Kitch en QK = ∑0.3VKC ρKC HKC + ∑VKF ρKF H KF + ∑m KE H KE (6) FLDW = (7) FLDW / FLDK AW / AK VWC /VKC ρWC / ρKC HWC / HKC VWF /VKF ρWF / ρKF HWF / HKF mWE / mKE HWE / HKE Where HLF m LE HLE (4) QW / QK Living room and bedroom Q L&B FLDL&B AL&B VLC ρLC HLC VLF ρLF QW = ∑0.3VWC ρWC HWC + ∑VWF ρWF H WF + ∑m WE H WE ∑0.3VKC ρKC HKC + ∑VKF ρKF H KF + ∑mKE HKE AK Where The average design fires can be computed by fire load density and equivalent duration of fire B Wash room = fire load of living room/bedroom, in MJ = fire load density of living room/bedroom, in MJ/m2 = total area of living room/bedroom, in m2 = fire load of washroom/kitchen, in MJ = fire load density of washroom/kitchen, in MJ/m2 = total area of washroom/kitchen, in m2 = volume of furniture which is a container, in m3 = corresponding density of furniture which is a container, in kg/m3 = corresponding calorific values of furniture which is a container, in MJ/kg = volume of furniture in living room/bedroom, in MJ/m2 = corresponding density of furniture, in kg/m3 = corresponding calorific values of furniture, in MJ/kg = mass of electrical equipment in living room/bedroom, in kg = corresponding calorific values of electrical equipment, in MJ/kg For the fire load density corresponding to a domestic flat, it is calculated as the ratio of the total quantity of the fuel commodities to the floor area of the space. The equations are derived as follows: = volume of furniture which is a container, in m3 = corresponding density of furniture which is a container, in kg/m3 = corresponding calorific values of furniture which is a container, in MJ/kg = volume of furniture in living room/bedroom, in MJ/m 2 = corresponding density of furniture, in kg/m3 = corresponding calorific values of furniture, in MJ/kg = mass of electrical equipment in living room/bedroom, in kg = corresponding calorific values of electrical equipment, in MJ/kg E Total fire load d ensity QT = QL&B + QW + QK (8) Where QT FLDT FLDT = QL&B + QW + QK AL&B + AW + AK (9) = total fire load, in MJ; = total fire load density, in MJ/m2; Fig 2. List of equations for calculating fire load density • with the estimated fire load density from survey, the equivalent duration of a fire can then be determined by Raes’s equation, see Fig 2. (1) • the average design fires can then be computed by fire load density and equivalent duration of fire Data collection To determine the fire load density of a domestic building, a simple approach proposed by Kumar and Rao, in which the weight of an object is related to its visual physical characteristics was employed. The approach is better than the direct weighing method as it is convenient, acceptable to residents, and also time-saving. The fire load survey was conducted between October 2005 to March 2006 covering a total floor area of about 2,787m², including 50 flats in 13 domestic buildings with the height up to thirty-six storeys. The flats and buildings were selected randomly due to the availability of residents. In this study, fire load survey for six types of offices were also conducted. The structure of the survey form is constructed and shown in Fig 1. The overall procedures for the collection of data are summarised in Fig 3. To improve the practicability of the survey, the following assumptions have been made and summarised as follows: • the fire load in basements (if any) and staircases are not included in the present 20 investigation as a basement is normally not provided for residential buildings and staircases are assumed to be a protected zone in which combustibles are limited • for paper and books, the mass is not dependent on their various compactions • the fire loads do not vary for items produced on the basis of new technologies and new materials, hence the data used for calculation is consistent for similar types of materials • for containers, it is assumed that only 30% of total volume is made of the corresponding material and 70% of area is space • the fire load contributed by doors, windows, and ventilators in between two rooms is considered as a fire load for one room only • the dimensions of commonly-used furniture and electrical equipment are the same • there is a uniform distribution of combustibles in the building • all combustibles will be involved in a fire • the combustion will be complete and the rate of heat production will be the same for all combustibles • it is assumed that the volume of clothing occupied 50% of the containers Determination of fire load density For containers, such as boxes, drawers etc, it is assumed that only 30% of total volume is made of the corresponding material, while 70% of the area is empty space. Based on this assumption, two equations are derived for determining the fire load of a living room and bedroom, see Fig 2. (2) and (3). Similarly, the equations for the washroom and kitchen can be seen in Fig 2. (4)(5) and (6)(7). Results and discussion The survey results for domestic buildings are presented in Fig 4. and Fig 5., and show that all the residential flats in Hong Kong have a fire load density higher than 1000 MJ/m2; for over 20% of flats, the fire load density is over 1,500 MJ/m2. The total fire load density can be calculated from the equations seen in Fig 2. (8) and (9). Table 1. summarises the results of a fire load survey for six type of offices. The fire load densities vary from 546 MJ/m 2 to 1,408 MJ/m 2. The weighted average fire load density for these six types of offices is over 800 MJ/m 2. To compare the results with other countries, the fire load density of a Europe Survey has been extracted from British Standard, PD7974-1:2003 and presented in Table 2. The average fire load densities of dwellings and offices are 780 MJ/m 2 and 420 MJ/m 2 respectively. These values are significantly lower than the survey results in this study. This is due to the fact that the average floor area per person occupied in Hong Kong is much less than that in the Western Countries because of the limited land available. From Raes’s Equation, Fig 3. (1), the equivalent duration of fire (t e) is in direct proportion to the floor area (A f) and the fire load density (FLD). With a higher fire load density, the equivalent duration of a fire will be longer. Survey on current public housing structure 50 nos. of domestic flats in public person and properties Office 1 Office 2 Office 3 Office 4 Office 5 Office 6 Area (m2) 330 230 280 820 750 580 Fire Load Density (kJ/m2) 784 1408 546 658 857 756 Table 1. Summary of fire load density of office buildings Preliminary analysis from the collected datum in previous stages • Overall floor area, floor area of living room, washroom and kitchen • Number of bedrooms, number of residents, number of elderly, number of students and number of children Fireload density * Occupancy Fractile (MJ/m 2) ** Average (MJ/m 2) 80% 90% 95% Dwelling 780 870 920 970 • Density of common combustible materials Hospital 230 250 440 520 • Calorific values of common combustible materials Hospital storage 2000 3000 3700 4400 Hotel bedroom 310 400 460 510 Data processing in deriving the fire load density of each flat Offices 420 570 670 760 • The volume of a furniture is calculated by multiplying the measured dimensions Shops 600 900 1100 1300 • Measured dimensions and material type of furniture • Mass of furniture = volume x density Manufacturing 300 470 590 720 • Heat release from a combustible furniture = calorific value x mass of a furniture Manufacturing & storage *** 1180 1180 2240 2690 Libraries 1500 2250 2550 – Schools 285 360 410 450 Note 2: T he values given in this table included only the variable fire loads (ie, building contents). If significant quantities of combustible materials are used in the building construction, this should be added to the variable fire load to give the total fire load • Conclusion and next steps Results of a survey for estimating the fire load density of residential flats and offices have been presented in this study. With reference to the figures given in a British Standard, the average fire load density of a dwelling is 780 MJ/m 2 and for offices 420 MJ/m 2. These are far below the survey results obtained in Hong Kong. This may indicate that the average floor space occupied by the people in Hong Kong is less than that in the UK and the corresponding fire load density may then be higher. The implication is that in each country, local information is important for determining the fire load density and that a reliable fire safety engineering study should be supported by local data. Table 2. has listed the fire load densities of a range of classifications used in the UK. It is expected that the fire load density will be different in Hong Kong. This study will be extended to other types of occupancies, particular retail shops. Table 2. Fire load densities (extracted from PD 7974-1:2003) Cumulative Frequency Distribution - Whole Flat Units In addition the survey of fire load density of various buildings in Hong Kong, the heat release rate (the design fire size) of these buildings can be estimated more accurately. The survey data provides strong support to the performancebased design in Asia and a good reference to other regions. The study will be further extended to conduct fire risk analysis for various buildings by combination of the likelihoods and the consequences. The consequences of the fire, and the resulting loss of property and life, is related to the severity of the fire. Fire risk assessments are one of the major aspects in Performance-Based Fire Safety Engineering approach. Based on the determination of the appropriate design fires, it allows the evaluation of fire risk level as well as assessment of fire protection requirement. 1901-2000 1801-1900 Fire Load Density (MJ/m2) Fire Load Density (MJ/m2) Fig 4. Fire load density histogram, whole flat units 1701-1800 1901-2000 1801-1900 1701-1800 1601-1700 1501-1600 0 1601-1700 2 1501-1600 4 1401-1500 6 1301-1400 8 1201-1300 10 100 90 80 70 60 50 40 30 20 10 0 1001-1100 Cumulative Fequency (%) 12 1101-1200 Fire Load Density Histogram – Whole Flat Unit 1401-1500 It is well known that either the fire load (the amount of combustible materials) or ventilation (the amount of oxygen supply) can be a control factor to the heat release rate (design fire size). For a high fire load space (a residential apartment or an office), the maximum design fire size will be determined by ventilation. Under fire conditions, ventilation will mainly come from the windows after the glass is dislodged and the doors are open. In general, the area of glazing windows in Hong Kong is greater than that in the UK. Compared with the UK situation, a fire in Hong Kong could last longer and be more severe as indicated from the results of this survey. • Derived from surveys: see CIB W14 Workshop Report,1985 * The 80% fractile is the value that is not exceeded in 80% of the rooms or occupancies ** Storage of combustible materials at less than 150 kg/m 2 *** 1001-1100 Fig 3. Flow chart for data collection • N0. of Flats Determine the fire load density at various type buildings 1301-1400 Analysis of data and discussion • Determine the correlation of between the parameters (ie. type of occupants, number of occupants) fire load density Note 1: L imits. The fire load densities given in this table assume perfect combustion, but in real fires, the heat of combustion is usually considerably less. 1201-1300 • Fire load density = total heat release / overall floor area 1101-1200 • Total heat release in a flat = adding up the hear release from all combustible furniture Fig 5. Cumulative frequency, whole flat units Acknowledgements The work described in this study was the result of a collaboration with the Department of Building and Construction and City University of Hong Kong. References Raes H., The Influence of a Building’s Construction and Fire Load on the Intensity and Duration of a Fire, Fire Prevention Science and Technology, Vol.16: 1977, 4-16. Kumar S., Rao C.V.S.K., Fire loads in office buildings, Journal of Structural Engineering-ASCE, Vol.123, Issue 3: 1997, 365-368. British Standard Institute, PD7974-1: 2003: The Application of Fire Safety Engineering Principles to Fire Safety Design of Buildings, British Standard Institute, BSI, 2003. 21 An investigation of fire load density for residential buildings and offices in Hong Kong Data collection Retrofitting private homes at scale: investigating the business case Authors: Ann Cousins, Matt Gitsham 22 The household sector represents 27% of total UK emissions and achieving substantial cuts here is imperative. Whilst there has been some retrofitting activity in social housing, action has been very slow moving in the privately owned sector, which accounts for 93% of Bristol City housing. Abstract This research looked at the opportunities and barriers, in terms of finance and energy efficiency, for the refurbishment of private housing in large-scale contracts. The study looked at energy savings and costs for a range of retrofit measures, such as insulation or new windows, as applied to 12 different housing types (defined by age and size), typical to the Bristol area. The main findings were that there are significant financial savings to be made in contracts of more than 50 houses, as large contractors become interested. Further savings can be made where houses are all within walking distance of each other, rather than spread across a city, as the costs of management and logistics increase as homes are further apart. Achieving an A-Rated home is difficult with the existing stock; indeed none of the packages of measures studied achieved this. Larger, older houses present the greatest potential for carbon savings, as they have higher CO 2 emissions prior to any retrofit. Retrofitting existing homes will contribute significantly to achieving the UK’s carbon emission targets. Introduction There are an increasing number of Government targets and wider initiatives for new homes producing low carbon homes that enable sustainable lifestyles. This study aimed to identify the opportunities and barriers, in terms of finance and energy efficiency, for the refurbishment of private housing in large-scale contracts. The study used the west of England as a case study and took a systems-based approach to analyse the scale of savings available. Rather than creating a consumer-facing tool, the primary audience for this work would be investors, contractors and local authorities, who want to consider the economics for large refurbishment contracts, similar to those carried out in social housing. Methodology Categorising the housing stock Data provided by the Centre for Sustainable Energy (CSE) presents the housing stock in the West of England broken down by local authority area, age of property (presented in approximately 30 year ranges), built form (detached, semi, terraced etc), number of bedrooms, and access to gas. This data showed 237 housing types in the area of the Bristol City Region. A final list of 12 predominant housing types were chosen to be examined representing approximately 55% of the west of England housing stock. Determining the base case A review of evidence was done to determine whether it would be possible to identify the current state of the housing stock in the west of England. It was decided that for completeness, we should define the base case houses as being completely unimproved (excluding the addition of central heating), with no insulation or any other additional measures. To identify further details relating to the standard construction of an “average” house, reduced data Standard Assessment Procedure (SAP) was used to produce a base case SAP rating for each housing type. Package One Energy Saving Lighting Draught proofing Roof insulation Cavity wall insulation Package Two New windows New boiler and controls Package Three Internal wall insulation Floor Insulation Package Four External wall insulation Floor Insulation Solar thermal Solar PV Table 1. Measures contained within each package Retrofit measures An initial list of potential housing retrofit measures to enable householders to live more sustainable lifestyles was drawn up. This included measures to contribute to CO 2 reduction, climate change adaptation, wellbeing and other sustainability measures. In order to prioritise this long list, it was decided to concentrate on carbon reduction measures. To enable the production of a meaningful model, we chose areas that could be measured using SAP. We excluded wind energy, as the effectiveness of this technology varies significantly depending on location. Further work was done to define each measure (eg material used and thickness of insulation) and relevance to each housing type. SAP calculations SAP is a logarithmic scale; adding different retrofit measures to a base case house has differing results depending on the combination of measures applied at any one time. It was therefore decided the most effective way of providing the results required to fulfil the aims and objectives of the study, was to define four packages. Each was designed to efficiently combine measures, as appropriate, and to become progressively more expensive (with package four being the most expensive). Table 1. shows the measures contained within each package. 23 Base house Package one SAP EPC CO2 (kg/yr) band 3-bed semi, 1930 to 1980 Package two CO2 CO2 SAP EPC (kg/yr) saved band (kg/yr) Package three CO2 CO2 SAP EPC (kg/yr) saved band (kg/yr) Package four CO2 CO2 SAP EPC (kg/yr) saved band (kg/yr) CO2 CO2 SAP EPC (kg/yr) saved band (kg/yr) 8916 35 F 5244 3672 59 D 2627 6289 79 C 2433 6483 80 C 2219 6697 82 B 3-bed semi, 1900-1930 11895 27 F 9361 2534 39 E 5467 6428 63 D 2818 9077 80 C 2253 9642 84 B 3-bed Semi, Victorian 15549 17 G 11680 3869 31 F 7864 7685 49 E 3472 12077 76 C 2790 12759 81 B 3-bed Semi, Georgian 13051 18 G 10601 2450 28 F 6925 6126 48 E 3084 9967 76 C 2473 10578 81 B 2-bed terrace, 1900-1930 6993 28 F 4978 2015 44 E 2762 4231 68 D 1566 5427 81 B 1133 5860 87 B 2-bed terrace, Victorian 7280 35 F 5239 2041 51 E 3098 4182 70 C 1763 5517 82 B 1323 5957 87 B 2-bed terrace, Georgian 7817 31 F 5626 2191 47 E 3425 4392 67 D 1828 5989 81 B 1325 6492 87 B 4-bed detached, post 1980 6923 46 E 5036 1887 61 D 3567 3356 71 C 2922 4001 76 C 2386 4537 81 B 2-bed flats, 1945-1980 (top) 5785 1 G 3190 2595 27 F 1459 4326 78 C 1363 4422 79 C 919 4866 87 B 2-bed flats 1945-1980 (middle) 4640 23 F 3756 884 34 F 1730 2910 81 B 1377 3263 84 B - - - - 3-bed detached, 1945-1980 9434 27 F 5638 3796 52 E 3375 6059 70 C 2844 6590 75 C 2386 7048 79 C 3-bed terrace, 1945-1980 7895 39 E 5133 2762 59 D 2133 5762 82 B 2005 5890 83 B 1545 6350 87 B Table 2. CO 2 savings and SAP ratings Each package is applied in order ie you cannot install package two until you have installed package one, but packages three and four are mutually exclusive. A choice must be made between internal and external wall installation. Average breakdown of costs 1% 9% Results and discussion Costs and savings There are savings to be made through a whole house approach rather than by individual measures. ROK’s cost analysis suggests that an average of over 4% can be saved by installing measures in packages. These savings are made predominantly due to the fact that more than one measure can be carried out by the same trade at the same time, cutting down on overhead costs. There may also be savings in preparation time and plant (eg scaffolding could be used for more than one measure). Our benchmarking exercise demonstrated that there are significant variations in costs of single measures to single dwellings. A large scale approach would remove this variability and give the consumer some consistency in what they could expect to be charged. There are significant savings to be made by retrofitting enough houses for a large contractor to be interested in the study. The exact number 24 Carbon emmisions in the home 28% Labour Heating homes Plant 57% Materials Overhead & Profit Costs Cost information was gathered predominantly from ROK, which provides building repairs and refurbishment services and a small local specialist retrofit firm Footprint Building. This was benchmarked against prices available from the Department for Energy and Climate Change (DECC), the Energy Savings Trust (EST), the Centre for Sustainable Energy (CSE), and World Wildlife Fund (WWF). The local builders costs will be used to compare the costs of retrofitting individual homes (1-19), with the costs from ROK of retrofitting a large number of homes (20+) as one contract. Maintenance costs were excluded from our analysis. 3% 15% 25% Heating water Appliances & lights Cooking 62% Fig 1. Average breakdown of costs Fig 2. Carbon emissions in the home of homes at which a large contractor might be interested is not an exact science, but following discussions with ROK and Footprint, we have assumed that this would be at around 20 homes. Further savings will occur when a large contractor is able to get further bulk savings. This is assumed to be for 50 houses. they are not familiar with and working with an unknown team, and the cost will reflect this risk for them. As the market changes, and installing these more “advanced” measures becomes more mainstream, it is likely that the cost of large contractors will come down to reflect this change. In order to achieve the best financial savings, retrofit at scale will have to happen in clustered groups. Cost savings from carrying out measures at scale are likely to be cancelled out by the extra transport, management and logistics costs of spreading houses too thinly. For the majority of retrofit measures, the most significant costs are associated with the materials. As can be seen in Fig 1. across all of the measures studied, except the renewable options, over 60% of the cost is associated with materials. Whilst ROK’s buying power is already likely to mean that many of these materials are already at a significantly reduced cost, it is possible that with more widespread take up, the market will become more competitive and prices will come down. At present, there are a small minority of measures that are cheaper from a small specialist builder. The most obvious of these are solar PV and solar thermal, which are currently cheaper through a small specialist contractor. However, mainstream contractors currently have limited experience in renewable technology. They will be installing systems that SAP and CO2 savings Table 2. outlines the SAP and CO 2 results for each of the four packages. As can be seen, the SAP improvements and CO 2 savings are less significant in packages three and four. This is not necessarily a reflection that the additional measures in packages three and four are less effective, but rather that the energy savings available become smaller as you progress through the refurbishment options. The Government sets out that SAP ratings are based on the energy costs associated with space heating, water heating, ventilation and lighting. Carbon savings can be achieved without any saving in the cost of energy, by the installation of renewable energy technologies. This can been seen in the example of the three-bed semi-detached property constructed between1930-1980, which only moves from a SAP rating of 79 to 82 between packages two and four, but saves 408kg CO 2/year. The base case flats are assumed to be fitted with electric heating, so the significant impact for them, particularly in terms of SAP rating is when the new A-rated gas combi-condensing boiler is added in package two. Improvement results per house # Houses Intended refits packages Houses dispersed/ clustered 3-bed semi, 1930 to 1980 50 Clustered Existing Existing Existing Install 80 82 B B 14,630 214 68.36 3-bed semi, 1900 – 1930 50 Dispersed Existing Existing Existing Install 80 84 B B 24,289 565 42.99 3-bed Semi, Victorian 50 17 31 G F 1,461 3869 0.38 3-bed Semi, Georgian 50 Dispersed 18 28 G F 1,332 2450 0.54 2-bed terrace, 1900 – 1930 15 Clustered Existing Existing Existing Install 81 87 B B 19,264 433 44.49 2-bed terrace, Victorian 15 Dispersed Existing Existing Existing Install 82 87 B B 21,431 440 48.71 2-bed terrace, Georgian 15 Clustered Install 31 47 F E 1,334 2191 0.61 4-bed detached, post 1980 15 Dispersed Install 46 61 E D 2,610 1887 1.38 New windows/Boiler and controls Internal wall insulation/Floor Insulation External wall insulation/Solar thermal/Solar PV Clustered One Two Three Four Install Install SAP before (/100) SAP after (/100) SAP rating before SAP rating after Cost/ house (£) CO2 saving (Kg/yr) £/Kg CO2 2-bed flats, 1945–1980 (converted) 1 Clustered Existing Install 27 78 F C 5,720 1731 3.30 2-bed flats 1945–1980 (purpose built) 1 Dispersed Existing Install 34 81 F B 5,272 2026 2.60 3-bed detached, 1945 – 1980 1 27 52 F E 2,446 3796 0.64 CO2 saved (kg/yr with package 4) % CO2 saved Clustered Install Fig 3. Refit Packages and improvement results It is interesting to note that the properties that attain the best SAP rating, after implementing all four packages, are the smaller, terraced houses and flats. They will have the smallest external wall area, and so lose least heat in this way. It also reflects the fact that smaller homes cost less to run, as there is less space to provide heat and light to. As can be seen from the results in Table 2. it is not easy to achieve an A-rated home, which requires a SAP rating of 93. This would require improvements in u-values (a measure of thermal resistance), or an increase in renewable technologies, either at the individual home or community scale. Either of which will add significant extra costs to the retrofit. What contribution can be made to the UK carbon reduction targets? The UK Climate Change Bill requires that the UK reduces its greenhouse gas (GHG) emissions by 80% by 2050. The Low Carbon Transition Plan sets out the emissions reduction required from each sector and states that our homes will need to be near zero carbon by 2050. One of the aims of retrofitting existing housing has to be to help reach this target. The solutions proposed as part of this study only consider single house solutions; any locally distributed energy was outside of our scope, but may help in reaching a zero carbon housing target by 2050. Table 3. sets out the percentage CO 2 reduction that can be obtained from applying all four packages of retrofit measures to each of the housing types. The average percentage reduction obtained is 78.69%. The 4-bed detached house, post 1980 will have had higher energy efficiency rating in its base case, which explains the lower percentage saving of just 65.5%. The middle-floor, purpose built flats could not install package four as it is assumed that they did not own the roof space, so could not install solar panels. This explains the lower percentage of carbon savings at just 70.3%. This highlights the need to look beyond single house measures in the long term. It is worth noting that whilst the CO 2 savings represent accurate savings from our base case houses, the base case houses represent minimum likely standards, rather than the average of the current housing stock in Bristol. It is also important to understand the elements that make up a home’s CO 2 emissions, these are set out in Fig 2. Emissions associated with appliances and cooking are not included in our calculations. There are clearly opportunities to reduce emissions associated with these areas by fitting homes with A+ rated appliances. Installing packages one and two on approximately 55% of all homes in the west of England (429,105 homes) results in an average CO 2 saving of 62.9% per house in the first year, at an average cost of £0.84 per kg of CO 2 saved. These figures are estimated average costs of carbon savings and the base case homes do not represent the accurate current condition of all homes in the Bristol City Region, but the minimum possible standards. However, this approximate approach is useful to indicate potential contributions available for reducing the 28.6% of emissions that currently come from the housing sector. It is estimated that the west of England’s total carbon emissions could be reduced by 9% by implementing such a programme and that emissions from housing could be reduced by up to 30%. Conclusion and Next steps The study resulted in the production of a model to assess SAP improvement, carbon reduction and costs of retrofitting each of the house types considered, which is shown in Fig 3. In conclusion, the principal benefits of retrofitting private homes at scale include: • consistency in price for individual homeowners • reduced costs (of up to £12,000) if a large contractor is involved (20 houses or more) • further cost reductions (up to an additional £1,500) for over 50 homes • providing a significant contribution to the UK’s target to reduce carbon emissions by 80% by 2050 The results of this research will feed into Forum for the Future’s Refit West Study, which is aiming to retrofit 1000 private homes in the west of England by 2011. The model produced features the predominant housing types in the west of England, but could easily be adapted to other areas of the country. (Beds/type) 3/semi 1930-1980 6,697 75.1 3/semi 1900-1930 9,642 81.1 3/semi Victorian 12,759 82.1 3/semi Georgian 10,578 81.1 3/terrace 1900-1930 5,860 83.8 3/terrace Victorian 5,957 81.8 3/terrace Georgian 6,492 83.1 4/detached post 1980 4,537 65.5 2/flats 1945-1980 (top) 4,866 84.1 2/flats 1945-1980 (middle) 3,263 70.3 3/bed detached 1945-1980 7,048 74.7 3/bed terrace 1945-1980 6,350 80.4 Average: 78.6 Table 3. Carbon savings for each housing type Acknowledgements This research formed part of a Forum for the Future Engineers for the 21st Century study; a collaboration between Arup, Forum for the Future, ROK and inputs from the Centre for Sustainable Energy (CSE) and Footprint Building. References Defra, The Government’s Standard Assessment Procedure for Energy Rating of Dwellings, BRE on behalf of Defra, 2005, 3. Boardman B., University of Oxford Environmental Change Institute, Home Truths: A Low Carbon Strategy to Reduce UK Housing Emissions By 80%, ECI Research Report 34, University of Oxford, November 2007. CAG Consultants and Energy Action Scotland for WWF, Carbon Countdown for Homes: How to make Scotland’s existing homes low carbon, WWF, September 2008. Centre for Sustainable Energy, Association for the Conservation of Energy and Moore R. for WWF, How Low: Achieving optimal carbon savings from the UK’s existing housing stock, WWF, March 2008 Energy Saving Trust, Roadmap to 60%: eco-refurbishment of 1960s flats, Energy Saving Trust August 2008. Department of Energy and Climate Change, Heat and Energy Saving Strategy Consultation, HM Government, February 2009. Department of Energy and Climate Change, The UK Low Carbon Transition Plan: National strategy for climate and energy, HM Government, July 2009. Michael Dyson Associates Ltd for Bristol City Council, Report of Bristol Private Sector House Condition Survey, Bristol City Council, July 2008. 25 Retrofitting private homes at scale: investigating the business case Draught proofing/Roof insulation/ Cavity wall insulation Transient thermal comfort modelling in underground stations Authors: Davar Abi-Zadeh, Mohammad Tabarra 26 Concept of local and overall sensation and comfort in the UCB thermal comfort model Physiology Tskin Local sensation vs. local skin temperature Whole body colder Local thermal comfort model Very comfortable 5 Comfortable -10 -5 0 5 10 Local thermal sensation Comfortt Local sensation The highly transient nature of underground environments leads to difficulty in using traditional comfort indices to properly address comfort conditions. This has significant impact on the comfort and well-being of passengers Local thermal comfort Slightly comfortable Slightly uncomfortable Uncomfortable Whole body warmer -5 T Local Very uncomfortable -T Local,set Local thermal sensation Overall thermal sensation 1 dominant segment 0.8 Abstract Underground stations or subway environments are highly transient and non-uniform. This is largely due to the massive exchanges in air volume which occur due to the train piston effect. In addition, when considered from the point of view of the passenger and not just the environment, the conditions that will be experienced throughout a whole journey on an underground system vary greatly due to passenger movement between different areas, eg external, station mezzanine, platform, train. This article presents a new thermal comfort modelling approach for subways and stations, which can evaluate local and whole body thermal sensation and comfort under transient conditions. The model provides a useful tool for the evaluation of alternative low-energy design and operating strategy and mitigation measurement for operating scenarios. Introduction Underground systems represent a significant challenge when it comes to passenger comfort. In design, analysis and modelling terms, underground systems represent highly transient environments with a great number of variables all having a significant impact on comfort, eg: • variation in temperature, air flows and relative humidity between the outdoor environment, the station concourse, the platform and the train • variations in passengers’ metabolic levels as they change between running, walking, standing and sitting • clothing variations between passengers There are a number of traditional thermal comfort indices in current usage, such as Predicted Mean Vote (PMV), Predicted Percent of Dissatisfied (PPD) and Effective Temperature (ET). In addition, the relative warmth index (RWI) is recommended by the United States Department of Transportation as a thermal index for underground station temperature control. Most of these indices were developed in the 1970s, at which time heat transfer models of the human body were not so well developed. asymetrically weighed segment 0.6 0.4 non-dominant 0.2 0 -6 back face e Overall thermal d hand sensation -4 -2 0 S Local 2 4 -S Mean 6 8 Fig 1. Concept of local and overall sensation and comfort in the UCB thermal comfort model They are generally derived from quasi-steadystate conditions, such as a typical office environment and even when applied with significant experience, cannot fully incorporate all the factors affecting thermal comfort in a non-uniform, transient environment or situations as experienced by underground passengers. This can lead to unsuitable designs, as engineers may mistakenly attempt to recreate, for an underground environment, the level of comfort required in a typical office, when much less stringent conditions may actually be required. This presents a wasted opportunity to utilise appropriate design in order to minimise energy usage in underground stations and other transient situations. Methodology UCB thermal comfort model It is clear that there is a need for better tools to model such environments. The University of California, Berkeley (UCB) has over the last 12 years, developed one such tool. Their UCB Thermal Comfort Model software is designed to predict thermal comfort under both steady-state and transient conditions. The model predicts thermal comfort for 16 body parts. Each part is divided into core, muscle, fat, and skin layers. An underlying model of the blood circulation system simulates the heat exchanges between the tissue layers, such as muscle and skin. Local sensation is calculated based on the local skin temperature of the body part and the mean skin temperature, which represents the whole-body thermal state. In transient conditions, the derivatives of local skin and core temperatures are included. Local comfort is calculated based on local and overall sensations. The local sensation and comfort levels are integrated to get an overall sensation and an overall comfort. The underground station model In this collaborative research between the University of California, Berkeley, and Arup, we used the software to develop a model of geometry and environmental conditions in an underground system. The principles can also be applied to transit facilities, such as railroad stations, multi-modal transit centres, and airports. Fig 1. shows the customised graphic user interface of the Arup underground simulation. Software outputs While the software can provide many different data, such as skin temperature, core temperature, sweating or shivering rate, PMV, PPD etc, the two key outputs are the thermal comfort and thermal sensation. These figures provide the indices which can be used for comparison of different environmental conditions and the impact they have on comfort. The scales are shown in Fig 2. How these scales are used must be considered carefully depending on the application. There is a difference between a comfortable condition and one which is still ‘acceptable’ to the user. This is particularly affected by user expectation. Studies in the offices show that people consider a slightly uncomfortable condition acceptable. The comfort value of minus 1 was recommended as the threshold for acceptability for an underground station environment in this study, because people’s expectations are slightly lower in underground stations than in buildings. Modelling scenarios Typical modelling scenarios were developed for passengers passing through an underground station. Each scenario is composed of a series of transitory “phases”. Each phase defines its specific environmental and personal clothing and metabolic conditions. 27 4.0 Very comfortable 4.0 Very hot 4 Mezzanine 2.0 Hot 3 2.0 Warm 2 1.0 Slightly warm 3.0 Comfortable 1.0 Slightly comfortable 0 0 Neutral Slightly uncomfortable -1.0 Outdoor Platform In train Outdoor 32º C/50%, Mezannine 34º C/50%, Platform 36º C/50%, 150/100 W/m2 dir/dif solar 1 Comfort 3.0 Outdoor 32º C/50%, Mezannine 34º C/50%, Platform 36º C/50%, 700/100 W/m2 dir/dif solar Stairs Outdoor 32º C/50%, Mezannine 34º C/50%, Platform 36º C/50%, 500/100 W/m2 dir/dif solar 0 -1 -1.0 Slightly cool Outdoor 29º C/50%, Mezannine 31º C/50%, Platform 33º C/50%, 500/100 W/m2 dir/dif solar -2 -2.0 Uncomfortable -3.0 -4.0 Very uncomfortable -2.0 Cool -3.0 Cold -4.0 Very cold Fig 2. Thermal comfort and thermal sensation scales -3 -4 0 5 10 15 Time (minutes) 20 25 30 Outdoor 29º C/50%, Mezannine 31º C/50%, Platform 33º C/50%, 150/100 W/m2 dir/dif solar Fig 3: Effect of external conditions on underground station thermal comfort The following inputs are defined for each transitory phase: • surrounding space geometry – choose from a predefined geometry, modify existing geometry, or build a new customised geometry from scratch • environmental conditions: defines air temperatures, air velocities, relative humidity, solar radiation, azimuth and altitude, surface temperatures etc • passenger metabolic level and clothing: define metabolic rates relating to different levels of activity: walking, standing, sitting etc,and clothing, whether summer or winter. Work is in progress to update the clothing editor with comprehensive customisable options. A number of phases were combined to create a complete transient scenario. Two scenarios were simulated; normal mode and congested mode. Normal mode models the movements of a passenger and conditions for an underground train in normal operations, ie moving from one station to another without stopping in tunnels for extended periods. The conditions are illustrated in Fig 4. Congested mode examines the thermal stress of passengers when conditions start to worsen due to the train being stalled in a tunnel between stations. The simulation looks at the time taken before a person’s physiological state reaches a danger threshold. Results and discussion Response to environmental transition The transient nature of the model allows certain conditions to be simulated, the effects of which cannot be assessed using steady-state models. Of particular note is the effect of transition between one environment and another. In this case, several conditions were simulated to see whether there is a ‘carry-over’ effect as the passenger moves from varying external ambient conditions into the station. A number of different conditions were considered, varying ambient temperature and solar radiation. Two conditions for ambient temperature were used: 29°C and 32°C. 28 Outdoor Walking (32°C, 50% RH, 10 min) Mezzanine (34°C, 50% RH, 1 min) Platform (conditions vary, 5 min) Stairs (34°C, 50% RH, 1 min) Train air conditioned (25.5°C, 55%, 10 min) Train arrival 24 sec, step into it Fig 4. Simulation scenarios for a typical normal mode operation Corresponding platform temperatures of 33°C and 36°C were defined based on experience. Fig 3. shows the results of varying these conditions. It shows that in general, solar load experienced while outside does not affect the comfort on the station platform. In fact the residual discomfort caused by the external solar radiation gradually disappears as the passenger wallks through the mezzanine/ stairs down to the platform area, with the comfort difference finally reducing to 0.5 scale units. This shows the residual memory of discomfort is short-lived and justifies the transient modelling approach. Such a method allows a comparison of comfort conditions that can be achieved using different cooling mechanisms by considering the integral effects of the whole scenario with multiple phases. Testing platform cooling solutions: Platform air conditioning Simulations were carried out to compare traditional cooling of the platform air with local cooling by radiant panels. In the first set of simulations, external ambient conditions and conditions in the stair and mezzanine were kept constant. Two platform temperatures were simulated 27.8°C and 29.4°C. Results are shown in Fig 5. They indicate that, while a lower air temperature results in a more comfortable condition after five minutes waiting on the platform, the difference is not great (0.5 scale units) and the conditions only just meet the comfort criterion of minus 1 that is considered acceptable. The increase is small considering the volume of treated air required to cool the whole platform by 1.6°C. This indicates that investigation of other cooling methods may result in more or equally favourable solutions. The sharp drop in comfort at 17 minutes is due to the arrival of the train pushing warm tunnel air into the platform. Testing platform cooling solutions: Radiant panels Radiant cooling has been successfully integrated in large public buildings and presents a viable cooling solution for underground stations with low latent load. In this study, thermal comfort was simulated for cooled floor, ceiling and vertical panels. The cooled panels are all at 20ºC, a value that would not cause condensation in most situations. Results are shown in Fig 5., Fig 6., Fig 7., Fig 8. and Fig 9. At the end of the period on the platform, the best comfort results are achieved with cooled vertical panels. In all cases, the cooled panels achieved a comfort condition above the required value of minus 1. Vertical cooled panels achieved results almost identical to the 27.8°C platform air condition, and a cooled floor panel achieved results only slightly below this. A cooled ceiling provided the least comfortable condition, but is still above the required criterion. Mezzanine 1.5 Outdoor 1.5 Stairs Platform In train 1.0 Platform 0 -0.5 -1.0 Threshold for thermal environmental acceptability 0 -0.5 -1.0 Piston effect 5 29.4°C 10 15 Time (minutes) 20 25 Ttunnel, ºC 0 In train 5 10 15 Time (minutes) 20 25 0 -1.0 Piston effect -2.0 30 0 29.4°C platform RHtrain, max, % 10 15 Time (minutes) 20 1.5 Time to RHtrain, max,min Mezzanine 1.0 36 10 100 15 36, with ventilation 36 7 71 12 31, no ventilation 34.6 21 100 15 31, with ventilation 34.2 17 62 16 27.8°C platform Outdoor Stairs Platform In train 0.5 0 -0.5 -1.0 Piston effect -1.5 Table 1. Time taken to reach stable temperature and humidity conditions when a train is stalled in a tunnel At the four different conditions, the train was modelled to see how long it took for the temperature to stabilise. The changes in temperature and humidity were then used to model the response of passenger core body temperature. The conditions are shown in Table 1. The difference in relative humidity between ventilated and unventilated is quite significant. Without ventilation, the car reaches saturation (100% RH) in 15 minutes. With ventilation, the air does not reach saturation, with a maximum of 71% at 36°C and 62% at 31°C. Maintaining conditions at a low RH will be key in reducing the thermal stress of passengers, therefore it is crucial that trains are ventilated in the congested mode. The response in passenger core temperature in the worst case (36°C/no ventilation) is presented in Fig 9. for two different metabolic rates, representing seated passengers and standing passengers. For seated passengers, the time to reach the threshold core body temperature is approximately 40 minutes. With the higher metabolic rate of a standing passenger, the threshold is reached after only 33 minutes. Conclusion and next steps Several scenarios have been simulated using a transient approach to thermal comfort, to determine how thermal comfort in an underground system is affected by various external and station conditions. Although the effects of ambient conditions may carry over for a short period when the passenger is in the mezzanine/stair area of the station, there is little to no carry over effect on comfort once the platform is reached. Lowering the platform air temperature from 29.4ºC to 27.8ºC improves comfort 0.5 scale units. This is not a great effect when considering the energy required to treat the air in the platform space. It indicates potential for low-energy strategies like spot-cooling to be used to cool the platform air directly in the occupied zone. Radiant cooling strategies, such as cooled floor, ceiling and vertical panels can potentially provide comfort equal to lowering the platform air temperature as with conventional HVAC. In an underground platform, 20ºC radiant panels can have the same comfort effect as reducing the air temperature by 1.6ºC. The comfort with vertical panels is slightly better than the cooled floor, which is in turn slightly better than the cooled ceiling. During the congested mode, train ventilation is very important to reduce interior humidity, which has a significant effect on passenger heat stress. In the case where 36°C and no ventilation were modelled, passengers reached the safety threshold for thermal stress after approximately 33 minutes while standing, or 40 minutes while seated. In extreme congested cases passengers should try to reduce their metabolic level, by staying calm and seated. Overall, this research demonstrates that, with the right tools, a new approach can be taken to thermal comfort by modelling transient effects and considering multiphased scenarios, which can be used to investigate and demonstrate alternative cooling solutions. -2.0 0 5 10 29.4 ºC platform 15 Time (minutes) 20 25 30 50 60 27.8 ºC platform 29.4 ºC platform, 20 ºC vertical panels Fig 8. Effect of cooled wall panel on comfort 39.0 Stand Core temperature (ºC) Four conditions were simulated for tunnel air temperatures of 31°C and 36°C, and either mechanically ventilating the train car at 0.26m³/s or no mechanical ventilation. The geometry of the train follows the London Underground’s Central Line stock, with 34 seated and 97 standing passengers in a car. 30 Fig 7. Effect of cooled ceiling panel on comfort 36, no ventilation Congested mode Passenger heat stress was analysed in the congested mode to determine how long passengers can tolerate the elevated temperatures before their core temperature begins to reach unsafe levels. A core temperature of 37.5°C is used as a threshold, set slightly lower than that used for US Army soldiers (38°C) to represent the average citizen. 25 29.4°C platform, 20°C ceiling panel Fig 6. Effect of cooled floor panel on comfort Time (min) to reach Ttrain, max, min 5 29.4°C platform 27.8°C platform 29.4°C platform, 20°C floor panel 27.8°C Ttrain, max, ºC Platform -0.5 30 Fig 5. Effect of platform air temperature on comfort Stairs -1.5 Thermal comfort 0 -2.0 Outdoor 0.5 -1.5 -1.5 Mezzanine 1.0 In train Thermal comfort Thermal comfort Thermal comfort Outdoor 1.5 Stairs 0.5 0.5 -2.0 Mezzanine 1.0 Seat quietly 38.5 38.0 37.5 37.0 36.5 0 10 20 30 Time (minutes) 40 Fig 9. Time for the core temperature to reach 37.5°C for seated and sitting metabolic rates Acknowledgements The authors wish to acknowledge technical input from Dr H Zhang of University of California, Berkeley. References United State Department of Transportation, 1976, Subway Environmental Design Handbook, Vol I, Principles and Applications. 1976, 2-4. Zhang,H., DongEun K., Arens E., Buchberger E., Bauman F., and Huizenga C, Comfort, Perceived Air Quality, and Work Performance in a Low-Power Task-Ambient Conditioning System. Center for Built Environment, Internal Report, 2008. Wang D., Zhang H., Arens E, Huizenga C., Time-series observations of upper-extremity skin temperature and corresponding thermal sensations. Building and Environment, Vol. 42, No. 12, 2007, 3933-3943. Davar Abi-Zadeh, Nigel Casey, Stefan Sadokierski, Barry Hodges, 2003, King’s Cross Underground Station Redevelopment: Analysis of Air Temperature, Velocity, Pressure and Quality Conditions, 11th International Symposium on Aerodynamics and Ventilation of Vehicle Tunnels, UK, 2003. Tabarra M., and Guan Y.,Temperature Monitoring of New York Subway System: Stations and Trains, 12th International Symposium on Aerodynamics and Ventilation of Vehicle Tunnels, Portoroz, BHRG, Slovenia, 2006. 29 Transient thermal comfort modelling in underground stations 2.0 Singapore buried channel research Authors: Timothy Mote, Anthony Bowden, and Michael McGowan 30 104˚0’0”E 1˚20’0”N Ancient river channels lie buried beneath the superficial geology of Singapore. These channels have a significant negative impact on construction work, particularly where they go undetected in the early stages of construction projects. 103˚50’0”E 1˚20’0”N 103˚40’0”E Meters 5,000 0 5,000 10,000 Study Area By using mapped occurrences of palaeochannel segments, the systems are projected back up gradient and down gradient to reveal their full extent. A key step in the interpretation is understanding a transition from bedrock controlled valleys into meandering river systems based on subsurface geology resolved from major infrastructure projects with large spatial footprints. A Geographic Information System (GIS) was used to integrate over 1200 borings that Arup has compiled from numerous jobs in Singapore. Interpretation on the location and extent of the palaeochannels was made and modelled in 3D to visualise their projection in relation to planned infrastructure. Knowledge of the location of these channels and where they can be expected to be found is of great value to further work Arup carries out in Singapore. The knowledge can provide warning to both the firm and our clients of potentially difficult and hazardous ground conditions. Study area Kallang Formation Bukit timah granite Palaeozoic volcanics Kallang formation Huat Choe Formation Gombak norite Old alluvium Jurong Formation Sajahat formation Tekong formation Jurong formation Dyke Rock 1˚10’0”N Huat choe formation Old Alluvium 1˚10’0”N Abstract In Singapore, buried meandering river channels (palaeochannels) are obscured by present day surficial cover making prediction of their extent difficult. The location of the palaeochannels beneath the superficial geology of Singapore was mapped within the context of the geologic history and subsurface investigations of the area. Prediction of yet unidentified channels was made based on a development of a geological model. Dyke rock 103˚40’0”E 103˚50’0”E 104˚0’0”E Fig 1. Study area and geologic map Introduction The Central Business District (CBD) of Singapore is founded on flat lying terrestrial and marine sediments at the southern end of Singapore Island. This CBD is located across the opening of a number of filled northwest to southeast trending valleys. In the past, tectonic activity and variable sea-level conditions both enhanced the development of the river channels and subsequently filled the channels with soft unconsolidated marine and alluvial sediments. The filled channels or palaeochannels have been shown to reach a thickness of 50m below the present ground surface. These palaeochannels follow the direction of existing rivers, with the most extensive being the Kallang Basin, extending for a considerable area around the Kallang River. Detailed knowledge of the location and extent of these palaeochannels after they emerge from the bedrock controlled valleys and flow towards the coast is limited, as the palaeo-fluvial systems are now filled with younger sediments. These palaeochannels have a significant negative impact on construction work, particularly where they go undetected in the early stages of projects. The channels are often filled with marine clay and viewed as a challenge for foundation and underground construction. Over the last couple of decades, significant underground infrastructure projects have been developed in Singapore, including the Mass Rapid Transit (MRT) network. Subsurface investigations from these and other projects have identified palaeochannels generally coincident with the current rivers, as expected. A number of investigations have also encountered palaeochannels in unanticipated areas away from the current rivers. These investigations provide key insights to the location and extent of the palaeochannels, but they are limited within the project’s footprint and do not characterise their greater location and extents across the CBD. To characterise the potential existence and occurrence of palaeochannels under the CBD, published and unpublished subsurface data are modelled in the context of current understanding of the regional geologic and geomorphologic history. Geologic setting The study area is centred on the CBD of Singapore Fig 1. Singapore is an island city-state located at the southern tip of the Malay Peninsula. The centre of Singapore Island is built on the Bukit Timah granite extending 8km north to south and 7km east to west. The Bukit Timah granite forms Bukit Timah Hill, the topographic high (166m) region north and northwest of the CBD. This topographic high creates the headwaters of the river valleys that flow to the CBD. The CBD is built over thick layers of marine clay and other young sediments of both terrestrial and marine origin comprising the Kallang Formation. These sediments are present in low-lying coastal areas, particularly where reclamation has been carried out and in-filled. The Kallang covers most of the CBD Fig1., the immediate offshore zone and the deeply incised river valleys which penetrate to the centre of Singapore Island. The formation has the widest distribution of all Quaternary (last 2 million years) sediments on Singapore, blanketing most of the areas below 5m above sea level and also forming valley fill at higher elevations. It consists of late Pleistocene (approximately 2 million years ago) 31 Controlled bedrock to meandering river transition MRT Reclamation Kallang Confirmed occurence 500 0 500 Not encountered 1,000 Meters Fig 2a. Plan view of confirmed occurrences of palaeochannel systems to Holocene (younger than 10,000 years) and recent deposits which are of marine, alluvial, littoral and estuarine origin. Within the Kallang the most important unit is the marine member locally called the marine clay. It occurs over a quarter of Singapore island but does not outcrop. Its thickness is extremely variable with a thickness of at least 55m. This unit is geotechnically problematic with in situ moisture content close to the liquid limit. Development of palaeochannels During the last 65 million years (Tertiary sub-era) block faulting and movement along pre-existing faults created large river valleys draining southeast. The upper reaches of these older fluvial systems were structurally controlled and followed a regional structural trend, northwest to southeast. As the rivers approached the continental margin they emerged from the bedrock controlled valleys, transitioning into a braided fluvial system depositing their sediment to form the Old Alluvium formation with a thickness up to 149m. Tectonic activity in the late Tertiary consisting of block faulting and warping of the Older Alluvium, created a northwest trending tectonic depression bounded on both sides by rising granite hills. These tectonic controls combined with a low sea-level and enabled a new fluvial system to scour the Old Alluvium. After the cessation of the tectonic activity in the Pleistocene the river valleys were filled in. This in-filling resulted from major changes in sea level which occurred during late glacial and post glacial times. As sea levels rose after the final glaciations the valleys were flooded and filled with marine sediments. This superimposed a Kallang-aged fluvial system onto an older Old Alluvium-aged system. Sediments of the Kallang continue to be deposited in the present day. 32 Fig 2b. Oblique view of confirmed occurrences of palaeochannels Methodology Data compilation This study compiled subsurface data from sources including geotechnical investigations for some major infrastructure projects in Singapore in published reports as well as unpublished geotechnical engineering reports. To support 3D modelling and interpretation the data was stored in a GIS. A total of 1245 boring logs were examined: of those, 656 recorded the thickness of Kallang. Many of the historic boring logs were lacking this key information. When the logged borings were lacking coordinate data, location maps of the logs were georeferenced and locations were digitised. Critical to the development of the 3D model is the elevation of the base of the Kallang. This is traditionally calculated from the elevation of the boring top and the depth to the base of the Kallang. Where borings reported soil descriptions and did not identify geologic units, an interpretation was made to pick the base of the Kallang. This was generally based on the bottom of the marine clay or a transition from markedly softer to harder material. A digital version of the geology of Singapore as mapped at 1:25,000 was integrated into the GIS to provide geologic context. A regional digital elevation model (Aster source with 30m resolution) was added to the system and processed to categorise slope highlighting topographic control and a potential indication of shallow bedrock. All pertinent data sets were integrated into a 3D geologic model to allow for interpretation of the extent of the palaeochannels in a geomorphologic context. Palaeochannel modelling interpretation By using both factual data and a conceptual understanding of the geomorphological environment in which these palaeochannel systems were developed, their present location beneath the Kallang can be interpreted. The interpretation is based on: • location of known palaeochannels • location of areas where palaeochannels are confirmed not to be present • orientations of current river systems • general hydrologic gradient towards the coast • geomorphological understanding of fluvial system formed prior to deposition of the Kallang • rivers are bedrock controlled in the west of the study area through Bukit Timah and Jurong • rivers are meandering east of the study area in Old Alluvium • regional structural trend of northwest to southeast • current topographic relief as indicators of bedrock highs The existence of palaeochannels within the bedrock controlled valleys trending northwest to southeast towards the CBD is well established. The current rivers, after emerging from these valleys, generally follow this orientation as well. There is a secondary conjugate northeast to southwest structure shown by smaller valleys and segments of current drainages, specifically reflected in the Kallang River system. A key control on the interpretation of the general trend of the palaeochannels follows these primary and secondary orientations. Singapore buried channel research Bedrock controlled channels Meandering channels (inferred) cut into Old alluvium 500 0 500 1.000 Metres Study area Kallang formation Study area High Reclamation Old alluvium Reclamation Moderate Controlled bedrock to meandering river transition Jurong formation MRT Low Confirmed occurence Not present Bukit timah granite Fig 3. Plan view showing transition from bedrock controlled channels into meandering channels The confirmed locations of palaeochannels from subsurface investigations are used to project the palaeochannels across the study area following hydrologic gradients down towards the coast and up toward the source valleys. Where subsurface investigations have shown that palaeochannels are not present, zones lacking palaeochannels in hydrologic shadows are developed. When the palaeochannels emerge from the bedrock controlled valleys they transition into meandering systems. To the west of this line, palaeochannels are inferred to be structurally controlled and relatively straight, while to the east of the line meanders are be expected. The curvature in the meanders, hence uncertainty in their location is anticipated to increase towards the east. The interpretation was further enhanced by using the current topography to represent indicators of a more resistant structure at depth and a less likely location of a palaeochannel. Results and discussion Using subsurface investigations to initiate the interpretation, the occurrence of palaeochannel systems are mapped emerging from the upland valleys into the coastal plain Fig 2a. and Fig 2b. Palaeochannel likelihood The likely occurrence or absence of palaeochannel systems across the study area is interpreted from the known occurrences of palaeochannels and an understanding of the geomorphological context. Where a known palaeochannel is shown to occur it can be connected back up to source zones following hydrologic gradients and continued down gradient to the coast. Where previous subsurface investigation has shown palaeochannels are not present, it defines zones where palaeochannels can be expected to not be encountered. 500 0 500 1.000 Metres Fig 4. Plan view of palaeochannel likelihood mapping Zones are developed and classified by a high, moderate or low likelihood of palaeochannel occurrence see Fig 4. High likelihood that a palaeochannel system will be encountered in the subsurface: these areas are generally down hydrologic gradients from a mapped palaeochannel and have an orientation that fits in the context of the regional geomorphological trends. Moderate areas where evidence that a palaeochannel will either occur or not occur is lacking. Often these are areas with limited subsurface investigation and as such regional trends can not be conclusive. combination of the subsurface investigation data in a geomorphological context allows for development of zones of potential occurrence of palaeochannels. Further refinements could be made to this interpretation by the addition of further subsurface investigation data, particularly in areas not covered. Of particular value, would be data from other major infrastructure projects and offshore seismic reflection data which may be available for the various reclamation areas near the CBD. Acknowledgements Low likelihood that a palaeochannel would be encountered in the subsurface: these areas are in the hydrologic shadows or down gradient from areas mapped where palaeochannels are not encountered. Conclusion and next steps The likely occurrence of palaeochannels across the CBD is interpreted using existing subsurface investigations and an understanding of the palaeo-geomorphological setting. By using mapped occurrences of palaeochannels defined from subsurface investigations, the systems can be tracked back up gradient or down gradient to the coast following regional structural trends. A key step in the interpretation is an understanding that the palaeochannels follow bedrock controlled valleys from their source in the uplands until they emerge and transition into meandering river systems. Where palaeochannels are shown not to be present, zones of low palaeochannel occurrence are defined based on the hydrologic shadows that these zones create. A version of this research was presented at the Underground Singapore 2009 Conference. The authors would like to acknowledge the many people who helped complete this research: Kathryn Nation, Ian Darlington, and Matthew Uidam in Sydney; Heng Kok Hui and Chris Deakin in Singapore; and Raymond Koo, Vicki Lau and Jack Pappin in Hong Kong. References Bird M I., Chang C. H., Shirlaw J.N., Tan T.S., The T.S., Proc. of the Underground Singapore Conference 2003. Chiam S. L., Wong K. S., Tan T. S., Ni Q., Khoo K. S., and Chu J., Proc. of the Underground Singapore Conference 2003. Defence Science and Technology Agency, Geologic of Singapore, 2nd Edition. Defence Science and Technology Agency, SIC/2-5/GG, ISO, 2009. Shirlaw J. N., Broome P. B., Chandrasegaran S., Daley J., Orihara K., Raju G.V.R., Tang S.K., Wong I.H., Wong K.S., Yu K., Proc. of the Underground Singapore Conference 2003. Sharam J.S., Chu J., and Zhao J., Tunnelling and Underground Space Technology, Vol. 14, no. 4, 1999. In certain areas, the lack of dense spatial data precludes detailed delineation of individual palaeochannels across the CBD, but a 33 The Life Cycle Tower: a high-rise in timber construction Authors: Jan Wurm, Tim Göckel, Martin Unger 34 There has been increasing interest in high-rise timber construction due to the growing awareness of economical and sustainable architecture. This has lead to an increase in multistorey timber constructions across Europe. Abstract Timber is carbon neutral and 100% renewable but its use was originally associated with either low-budget housing or low-rise eco-homes. The objective of this study is to demonstrate the feasibility of a commercial timber high-rise construction of 20 storeys in a densely populated urban context at detail design level. This study will focus on issues related to the structural design and fire engineering. The layout of the building with a footprint of approximately 45m x 30m and a 1.35m grid reflects the requirement of a mixed-use development for office and residential units. The structural layout in the currently preferred scheme features a central core of Glue Laminated Timber (Glulam) and Glulam Perimeter Columns. Timber-Concrete Composite Decks provide transfer of horizontal loads. For fire protection, a number of structural, technical and design measures have been developed to compensate for the combustibility of the material. Rhomberg is going to build a three storey prototype of the system on their premises mid-2010, while investigating sites to transform the design into the first built timber high-rise. This showcase study will provide a great opportunity to lift Arup’s design and engineering skills in timber construction to a new level. Fig 1. Principal layout of the life cycle tower with 20 storeys (building notched open at corner for better visualisation) Fig 2. Model of the timber core where the edges of interfacing slabs are tied into the model. The structure of the first three retail floors (shown in blue) is modelled in reinforced concrete. Methodology In 2000 the prototype of a six-storey timber frame was built in the UK as the first of its kind in the world: the so called TF 2000 project. Extensive research investigating stability, robustness and fire safety was conducted. Since 2000, about 100 platform frame buildings between five and seven storeys have been built in the UK. The current climax of this development is the recently completed nine-storey panelised timber frame building in Murray Grove, a few miles away from the City of London. A similar development can be seen in Europe. In Switzerland revised codes allow the construction of up to six storeys. In Germany, in 2008, where the use of timber is usually restricted to buildings below 13m, a seven-storey residential development in Berlin has been approved and built as a Glulam Skeleton Structure. Wiehag GmbH, the specialist Austrian timber contractor, and architect Prof. Hermann Kaufmann complete the team. The funding granted by Arup’s Design and Technical Executive (DTX) was essential to enable the team to carry out the required Research & Development Work in connection with this study. An Austrian consortium of specialists, under the lead of architectural firm Schluder, initiated a first feasibility study investigating concepts for timber structures above eight-storeys (eight+). This initiative was supported by the programme ‘The House of the Future’ by the Austrian Research Fund, Österreichische. Forschungsförderungsgesellschaft mbH (FFG). Rhomberg Bau GmbH, a developer based in Bregenz, who investigated the commercial aspects of high-rise timber construction during this feasibility phase, set up the new consortium ‘Life Cycle Tower’ (LCT) involving Arup as multidisciplinary consultant to focus on the realisation of a pilot project, securing additional funds from the FFG. Arup was commissioned as consulting engineer for structures, building services, façade engineering, fire engineering, building physics and materials consulting. The objective of this study is to take the detail design for a timber frame high rise to 80% completion, setting the base for a commercially viable study to follow. The main qualitative design drivers are: • high degree of offsite prefabrication and reduced work on site, aiming to minimise construction programme • highly flexible structure and layout to accommodate residential, office and hotel use • energy efficiency and design for the minimisation of carbon footprint • no compromise on any performance criteria in comparison to conventional structure Key challenges are: • compliance with fire performance • compliance with acoustic performance • cost An agreed outline specification defining the performance criteria forms the basis of the highly integrated design process. The criteria are based on the more conservative among European Standards or the National Standards of the key market places of Austria, Switzerland and Germany. 35 Fig 3. One floor showing the central core with services shafts along the short sides. Currently the wall thickness of the Glulam core is 360mm Client Workshops for coordination and the setting of objectives of work stages are held every six to eight weeks. In between, research and development is carried out as desk studies inside the study teams of involved disciplines, and alternatives and options are developed and compared within the set criteria. Work is periodically reviewed by internal workshops involving specialists from other Arup offices, such as timber specialist Andrew Lawrence from the Technical Development and Support Group in London. In addition, workshops with external reviewers are also held to discuss the most challenging issues. The building As the prototype design for the Life Cycle Tower needs to address all relevant issues and restrictions of a life study, the client selected an urban development site in Austria. The compact footprint measuring 27m x 43m is orientated along westeast axis. The layout and plan are based on a 1.35m grid in order to accommodate both office and residential use. The central core measures 8.10m x 18.90m. Fig 4. Current FEM model with all structural elements on the left and core only on the right Fig 5. The floors are built with composite decks with installations running between timber beams. The partition walls connect to the concrete deck providing obstruction to horizontal fire spread The first three storeys contain retail areas and are built as a conventional reinforced concrete construction on a structural grid of 8.10m. The floor-to-floor height of the remaining 17 storeys in timber construction is 3.50m. The columns The columns along the perimeter are located at 2.70m centres and are 480mm wide and 250mm deep at its maximum. The specified strength category is GL24h following EN 14080. Structure Early studies in conjunction with the results of the feasibility study ‘eight +’ showed that a cross laminated timber frame, as realised in the Murray Grove Project, would not be feasible for a highrise structure. This is mainly due to the structural restrictions at the interfaces. The most feasible solution is a shell of skeleton timber frame using Glulam components stabilised by a central core. The core Two concepts for the core are currently being investigated. A structural timber core of vertical Glulam beams connected by steel plates and rods would require a thickness of 300mm. In order to minimise joints, the maximum available length of Glulam beams of 35m would need to be considered. This solution is technically feasible and cost effective. For comparison we are pursuing in parallel a conventional concrete core as a back-up option and for even taller structures a braced façade. As such, the primary components for vertical load transfer are the perimeter columns together with the central core. Slabs and the vertical walls of the core provide stability and the transfer of lateral loads. The applied loads based on Eurocode EC1 include life load of 3.20kN/m² for rental floor areas and 5kN/m² for the corridors. The superimposed load of the structure was calculated to be 1.50kN/m². The static wind load increases from 0.66kN/m² on the bottom of the building up to 1.44kN/m² on the top. In the following section the intermediate results of our study are summarised. 36 The slab A solid timber construction for the slabs would meet neither the acoustic or the thermal performance requirements due to its lack of mass. As an in situ concrete slab construction would counteract the primary advantage of the fast and “dry” timber construction, prefabricated composite decks have been selected. The decks span generally over 8.10m although an increased span of 9.45m is currently being analysed. The prefabricated decks are 2.70m wide and correspond to the grid of the primary columns. The build-up of 180mm Glulam beams with 180mm concrete decking has been established following a parametric study also taking the frequency and sensitiveness to oscillation into account. To provide shear stiffness across the floor slab, the edges of the decks are monolithically joined with in situ concrete. Fire In order to gain building permission it has to be demonstrated to the authorities that the timber design of the Life Cycle Tower does not impose any higher risks to the inhabitants compared to a conventional building. The building addresses all fire regulations except the target that no combustible material should be used for the primary structure. Therefore a number of additional measures compensating for the increased permanent fire load are considered: • floors have full sprinklers installed • prevention of smoke and fire spread using cavity free building elements for floors and walls • design of structural components addresses reduction of section by 0.7mm/min • fire spread between units across the floor is obstructed by partitions walls supported directly on concrete The life cycle tower: a high-rise in timber construction Void filled with insulation material and encapsulated encased with plaster fireboards Composite reinforced concrete-timber slab Installation area above suspended ceiling Fig 6. Longitudinal section through composite deck in the zone of escape corridor Timber core wall encapsulated with gypsum plaster board Composite reinforced concrete-timber slab Installation area above suspended ceiling Corridor wall Fig 8. Example for a timber concrete composite decking system Corridor soffit Fig 7. Cross section through composite deck in the zone of escape corridor Timber GL24h Concrete C30/37 Steel Section 480mm x 250mm d=300mm d=220mm, t=16mm Density 380kg/m³ 2400kg/m³ 7850kg/m³ Embodied energy Co-efficient 4,6MJ/kg 1,3MJ/kg 32MJ/kg Weight, l=3,5m 160kg 600kg 280kg Embodied energy per element 740MJ 780MJ 9000MJ Table 1. Comparison of embodied energy of structural column in timber, concrete and steel • vertical fire spread between floors is obstructed by continuous concrete decking (timber columns are discontinuous) • systematic prefabricated construction is ensuring control over all details and preventing uncontrollable solutions onsite Results and discussion Comparison embodied energy A conventional construction with the same layout and loads would have led to a column of 300mm diameter (C30/37) in concrete respectively to a CHS of 220mm diameter and a 16mm plate thickness in steel. Taking into account the specific density of the materials and their embodied energy coefficients, the timber column and concrete column feature a similar quantity of embodied energy while the steel column requires more than 12 times the energy see Table 1. With approximately 40 columns per floor and 17 floors, a total volume of around 300m³ of timber will be used. Over 260 tons of CO 2 will be stored; it has been calculated that with the timber structure an equivalent of approximately 1000 ton of CO 2 will be emitted during manufacturing, construction and service of 50 years – this is 10 times less than with a conventional structure. Conclusion and next steps The structural design of the Life Cycle Tower with 20 storeys as a timber skeleton frame of Glulam elements in connection with a composite slab system has been developed to detail design level. The design of the central core in timber construction is technically feasible, and cost effective in comparison to a conventional concrete core. At present, the detail design for building services and façade engineering is being reviewed and a detailed cost analysis is carried out by our client. We anticipate finalising the design by the end of March 2010. In order to build a prototype building of three storeys on the premises of Rhomberg, a third application for funding was submitted to FFG at the end of 2009. The prototype study will allow the detail design and the final specification to be completed. In parallel we will continue the dialogue with external experts and the local building authorities aiming to get agreement to our approach to the fire protection strategy while Rhomberg is investigating suitable sites for realising the Life Cycle Tower. The foundation for the first timber highrise has been built. Fig 9. Visualisation of the currently envisaged façade concept incorporating PV-modules and “green walls” Acknowledgements We would like to thank Rhomberg Bau GmbH for founding and managing the “Life Cycle Tower” Consortium and the “Österreichische Forschungsförderungsgesellschaft mbH (FFG). In addition we would like to acknowledge research student Aliénor Dahmen for her very valuable contribution. References Schluder Architektur ZT GmbH, Final Report of the R&D project 8+, Vienna 2008. Hein C., Göckel T., VDI Conf. on Building with Innovative Materials (VDI No. 2084), Germany 2009. Braune A., Benter M., CO 2 – Check Lifecycle Tower, Conf. PE International, Germany 2010. 37 Monitoring geothermal piles at Keble College, Oxford Authors: Duncan Nicholson, David Whitaker, Natasha Kefford 38 Geothermal pile systems are quite a new technology in the UK. There is a lack of longterm data on the sustainability and performance of these systems. Abstract This study describes the monitoring of a geothermal pile system installed below the Sloane Robinson Building, Keble College, Oxford. The ground loops were installed in both the secant wall and the internal load bearing piles, and connected to a heat pump. The design provides 45kW of heating and cooling capacity with annual loads of 74MWh and 55MWh respectively. Fig 1. Completed Sloane Robinson building This collaborative research project has been undertaken with funding from the South East England Development Agency (SEEDA). The objective was to gather and interpret data from the Building Management System (BMS) to assess the long term performance of the building and compare it with the design predictions and specifications. Introduction Between 2002 and 2006 Arup lead a UK Department of Trade and Industry (DTI) project into the ground storage of building energy. This work highlighted the lack of case histories detailing the performance of ground energy systems in England. To address this, an agreement was reached between Arup, (SEEDA), and Cementation Foundations who were members of the DTI research project. This was to undertake geothermal pile monitoring at the Sloane Robinson building, Keble College, Oxford. The results from the first annual report indicate that the building heating cycle has performed as designed and the geothermal pile system has not exceeded the stipulated temperature limits. Geothermal pile system The geothermal pile system at Keble College was completed and commissioned in 2001, and comprised: However, the building cooling cycle differs from that predicted in the design. The geothermal pile system has been asked to deliver less of the daily peak summer load than designed due to the installation of additional cooling systems. The geothermal pile system has, however, delivered the higher than expected cooling loads during spring and autumn. For the period analysed (March 2007 to March 2008), the geothermal pile system appears to have caused the temperature of the ground to increase by 2°C. If this trend continues, there may be a drop in efficiency during cooling as the system will no longer be able to operate in direct cooling mode, ie the ground temperature will no longer be less than 19°C. • 15 load bearing piles, dia. 750mm, length 12.5m • 14 load bearing piles, dia. 600mm, length 7.5m • 6 1 secant wall piles, dia. 450mm, embedment length 5m, (see image left). Within these piles, a total of 41 ground loops were installed. Each ground loop comprised 150m of 20mm diameter plastic tubing connected to a series of geothermal piles. The system uses a heat pump to heat and cool the Sloane Robinson building. The design is based on a peak heating or cooling capacity of 45kW, and annual heating and cooling loads of 74MWh and 55MWh respectively. About 80% of the cooling load is supplied by direct cooling. Although the building loads are unbalanced, ie the heating load exceeds the cooling load, the actual loads placed on the ground are approximately in balance depending on the actual annual climatic conditions. The heat pump system was designed to function with ground loop temperatures of 27°C (max summer) and 1°C (max winter). Fig 2. Geothermal pile reinforcement cage In Fig 2. this shows an geothermal pile reinforcement cage. The geothermal pile system was designed by Enercret and installed by Cementation. The completed building is shown in Fig 1. The research project The objective of the research project is to assess the performance of the geothermal pile system at Keble, using data gathered from the Building Management System (BMS). The proposal of work developed by SEEDA and Arup in 2007 was to: • collect and review data collected by the BMS • analyse the data to assess the performance of the geothermal pile system • specify additional BMS instrumentation or BMS upgrades, if required • carry out further monitoring, up to a period of three years • annual reporting 39 Methodology Temperature data for 2007/2008 There are 52 monitoring points which feed data to the BMS, with data saved from February 2007 onwards. The data used in this project are: 26 21 • fluid temperature in the pipes returning from the ground 16 • heat pump flow rate (ground loop side) • heat pump flow rate (building side) • building external air temperature Temperature (˚C) • fluid temperature in pipes leaving the heat pump 11 6 An initial assessment of the operation of the geothermal pile system, between March 2007 and March 2008, was made by comparing the temperatures of the circulating fluid as it flows to and returns from the ground, and the average daily outside air temperature. Return from the ground Average daily outside air (II) T his is calculated by the formula Q (energy) = flow x delta T x specific heat capacity of water, using the specific heat capacity of water (4.182kJ/kg°C) (III) U sing the heat pump coefficient of performance (COP) of 4.4 (IV) A s the monitoring period is a quarter of an hour the total needs to be divided by four to calculate equivalent kWh For periods of direct cooling the heat pump is switched off. Therefore its influence is excluded from the calculation. To check if the system is performing as planned, annual energy abstracted from and rejected to the ground, were summed to estimate the annual balance. The return temperatures from the ground over one year were graphed. Results and discussion Temperature trends The temperature data in Fig 3. presents February 2007 to February 2008 and shows: • periods during which heating predominates, when the temperature of the fluid returned from the ground exceeds that sent to the ground • periods during which cooling predominates, when the temperature of the fluid returned from the ground is lower than that sent to the ground 40 Feb 08 Jan 08 16 Geo heat exchange flow Geo heat exchange return Outside air 20 14 19 Mar 17 Mar 15 Mar 13 Mar 11 Mar 09 Mar 27 Feb 29 Dec 24 Dec 0 19 Dec -10 14 Dec 4 09 Dec -5 07 Mar 6 05 Mar 0 8 03 Mar 5 10 01 Mar Temperature (˚C) Temperature (˚C) Geo heat exchange flow Geo heat exchange return Outside air 12 10 04 Dec (I) D ifference between temperature to and from the heat pump. A negative temperature signifies that the system is in heating mode Dec 07 Intermittent heating mode Continuous heating mode 25 15 3.58 l/s -2.3°C (I) 34.4kW (II) 44.5kW (III) 11.1kWh (IV) Nov 07 Fig 3. Temperature data for 2007/2008 As an example, heating during March 1st, 2007, 5am: Flow rate: Delta T: Energy from the ground: Energy to building: Equivalent kWh: Oct 07 Sep 07 Aug 07 Fig 4: Temperature data for 2007/2008. Jul 07 Jun 07 May 07 Apr 07 Mar 07 -4 Feb 07 As a check on both the original design and the way in which the building is operated, the design parameters for the system have been compared with BMS data. Flow to the ground 1 Fig 4. Continuous heating mode Fig 5. Intermittent heating mode • brief spikes in the temperature profile coinciding with periods during which the geothermal pile system is not being operated By contrast, Fig 5. shows the temperatures from 1st to 18th March during which the system was providing heat intermittently. Again, the outside air temperature was below 10°C for most of the period. However, there must have been solar gain and/or internal gains to the building because the system was not providing continuous heating, and the ground temperature was stable (approximately 8°C). Building Loads: heating mode A comparison of the design and operational heating energy was performed, Table 1. The data shows an extremely good match between the design and actual heating loads. This is evidence that both the building and the heat pump are functioning as designed. The system was designed to operate most efficiently with minimum average winter inflow/ outflow temperatures greater than 1°C. The actual minimum ground temperature (7.36°C) was much higher than expected, which implies that the heat pump is operating very efficiently in the heating cycle.The heating profile of the system can be further split into: • continuous building heating mode, Fig 4. • intermittent building heating mode, Fig 5. When the outside air temperature is consistently below 10°C, and there is no solar gain to the building, the geothermal pile system is constantly providing heating to the building. During this period, the temperature rejection to the ground is >2°C lower than the return flow from the ground, Fig 4. These conditions occurred between 6th and 26th December when ground return temperature dropped from 13.8°C to 8.8°C. The system is designed to maintain a temperature difference between flow to and from the ground of 2°C. There are also intervals during which the temperature of the water returned to the ground loop was higher than that of the ground. This could have been due to either the system providing cooling or the system being off and the temperature of the water next to the sensor rising towards room temperature. The latter is more plausible. Building loads: cooling mode A comparison of the design and actual cooling loads was performed, see Table 2. The monthly cooling design loads exceed and show a different monthly distribution to the recorded figures. The figures differ as follows: • the peak cooling design loads (July, August) are much higher than the recorded loads • the building requires cooling in more months than predicted When the system was originally installed, overheating in the summer was reported, and additional cooling systems were subsequently installed. In peak summer periods, the additional Ground return temperature 30 28 30 26 Design cooling energy (kWh/ month) Recorded cooling energy (kWh/ month 2007) 10,285 10,422 98 1,707 6,545 3,788 April 851 3,234 May 1,870 2,335 May 3,925 5,043 October 2,550 4,378 June 10,566 9,530 November 8,585 9,006 July 20,902 8,842 December 13,600 13,194 August 15,505 7,856 Total 43,435 43,123 September 2,747 5,968 589 3,177 0 2,368 March Table 1. Design parameters verses calculated delivery from the heat pump (heating mode). October cooling systems are operating and there is therefore a smaller cooling demand placed on the geothermal pile system. The additional months of cooling (November and December) reflect either a different usage of the building than originally planned, or a warmer winter than expected. December At the start of the cooling period the temperature differential between system inflow and outflow (ΔT) is approximately 2.5°C, and the ground loop temperature is 15°C. Therefore the system is operating efficiently. The temperature of the ground loop increases during the cooling period and ΔT becomes <2.5°C. To maintain the same level of cooling the flow rates of the system have risen and the efficiency will have dropped. This trend continues throughout the summer, Fig 7. and (ΔT) starts at 2.5°C and ends at <1°C. Significant temperature spikes represent periods where the system is not operational and November 12/04/08 22/02/08 03/01/08 14/11/07 25/09/07 06/08/07 Fig 8. Ground return temperature April The cooling provided by the geothermal pile system operates most efficiently at the beginning of the summer. Fig 8. shows a period of continuous cooling in early summer. A 10 17/06/07 Sep 07 Aug 07 Jul 07 Jun 07 May 07 Apr 07 27 Jun 25 Jun 23 Jun 21 Jun 17 Jun 15 Jun 13 Jun 11 Jun 09 Jun 07 Jun 19 Jun Recorded heating energy (kWh/ month), 2007 The geothermal pile system was designed to provide direct cooling when ground supply temperature is <19°C. This was designed to meet up to 80% of the building cooling demand. The operational data shows that 77% of cooling is provided directly. However, in the summer, the geothermal pile system is no longer meeting the total cooling demands for the building. Therefore, this is not the equivalent of 80% of the building cooling demands. B Temperature Fig 7. Summer cooling, rise in ground return temperature, reduction in delta T The maximum average design summer system temperature was 27°C, whereas that recorded was 25.88°C. Thus the geothermal pile cooling system is performing within system limits. 15 0 6 Total 20 5 Flow to the ground Return from the ground 8 Fig 6: Continuous direct cooling mode March 14 10 0 Design heating energy (kWh/ month) 16 12 Geo heat exchange flow Geo heat exchange return Outside air 5 18 128/04/07 10 20 09/03/07 15 22 18/01/07 20 25 24 Temperature (Degrees ˚C) Fluid Temperature (˚C) Temperature (˚C) 25 0 1,219 55,183 48,944 Table 2: Design parameters verses calculated delivery from the heat pump (cooling mode) the temperature of the feed pipes reaches room temperature. At the end of the summer the temperature from the ground approaches 19°C (the design limit for direct cooling). Ground temperature The annual energy balance between the system and the ground can be calculated by summing the energy abstracted and rejected to the ground. Between March 2007 to March 2008, the system rejected approximately 17,000kWh to the ground. It could thus be inferred that the ground temperature will have risen over this period. The return temperature from the ground rises in the summer and falls in the winter, Fig 8. as would be expected. However, this data suggests that over a period of one year, the temperature of the ground has risen by approximately 2°C (from 11°C (A) at the start of the monitoring period to 13°C (B) at the end). Without additional data it is not possible to say whether this is a long term trend or the result of an unseasonably warm year. However, the temperature data coupled with the fact that the system rejected 17,000kWh to the ground during this period implies that the system may be causing the increase in ground temperature. Conclusions and next steps The BMS at Sloane Robinson building, Keble College, Oxford has provided reliable monitoring data from March 2007 to March 2008. Longer term records could not be retrieved. Reviewing BMS records provides a simple method of assessing the geothermal pile system. The geothermal pile system started operation in 2001. The mean ground temperature between March 2007 and March 2008 is similar to the expected ambient temperature of 12°C. This implies a reasonable balance between annual heating and cooling loads. The return temperatures remained within the temperature limits of 27°C and 1°C. This indicates that the heat exchange is large enough to handle the loads sustainably. The building’s heating energy demand from the BMS data is similar to the design predictions. Between March 2007 and March 2008, the geothermal pile system appears to have caused the ground temperature to increase by 2°C. This figure compares the difference between the average ground return temperatures in February 2007 and February 2008. However this appears to be a function of the climate for that particular year. The building’s cooling cycle has not performed as designed. The system has met less of the peak summer loads than expected due to the installation of additional cooling systems. Conversely, the system has met the higher than expected cooling loads during spring and autumn. Recommendations The current monitoring research project should be continued. The ground return temperatures should be monitored to understand whether there is a steady annual increase in temperature. The building should be analysed to understand why the cooling loads are different to those that were expected. The supplementary cooling system should be assessed. An electricity meter should be installed on the heat pump to better understand the performance of the heat pump and the running costs of the system. Acknowledgements We would like to thank SEEDA, Keble College, University of Oxford, Cementation Foundations Contractor (contruction contractor), Pillinger (monitoring/maintenance contractor). Ryan Law Geothermal Engineering Ltd References Cementation Foundations Skanska, Geothermal Piles used at Keble College, Oxford. The Structural Engineer, 2004, 19. 41 Monitoring geothermal piles at Keble College, Oxford Summer cooling, rise in ground return temperature, reduction in delta T Continuous direct cooling mode Sustainable and quake resistant façade for existing buildings Authors: Susumu Matsunobu, Yutaka Misawa, Piet Lycke 42 (A) Integrated façade Configuration Abstract This research challenges the customary reinforcement methods which only focus on the structural performance. Instead it proposes an integrated-façade-system which not only consists of the necessary seismic structural retrofit, but also significantly improves the environmental and architectural quality of the existing building. A Buckling Restrained Brace (BRB) to increase the seismic resistance was developed with minimum dimensions and in such a way that it can be perfectly integrated and harmonised with a façade louver system. The louver system makes it possible to control and optimise the heat load and the daylight entering the building at the perimeter. As big earthquakes are mostly accompanied by fire, additional research has been done on the behaviour of the fire and its flames in this particular case. Furthermore, deploying the system on the outside of the building also ensures continuity in its usage. Initially the system has been singled out to be used for the main gathering points in the cities’ disaster evacuation plans such as in schools. However due to its flexibility it could easily be extended to other facilities in the future such as hospitals. Introduction Major cities in Japan have had a long standing reputation for wanting to have the newest and most modern buildings. This has been intrinsically linked to repeatedly demolishing and replacing the existing infrastructures with elaborate new ones. This short life span of buildings in Japan is remarkable and in the current climate of sustainable design, promoting the continued usage of buildings through improvement of their functionality is key. This will be an important factor in the process of decreasing the usage of resource materials and reducing the emission of CO 2, both associated with the construction process and the operation of the building. Meanwhile, Japan’s history has been characterised by several severe earthquakes. The most recent one in Kobe in 1995 (magnitude 7.3 on the moment magnitude scale) caused 6,434 deaths and 43,792 injuries. It was demonstrated that buildings from before the Louvers, Double skin etc + = Structure bracing member Structure Integrated façade Integrated façade on building + Plasticity effect = Seismic load Collapse Seismic load (B) Integrated façade: Structural function Seismic load The severe earthquake in Kobe, Japan, 1995, demonstrated that the city’s infrastructure, designed before the change of Japan’s building regulations in 1981, failed to secure the safety of its inhabitants. This study focuses on how to reinforce existing structures sustainably. Deformation Deformation Energy Absorption No damage to main structure Reduce maximum deformation Deformation (C) Integrated façade: Enviornmental Function Solar radiation (Summer) Exhaust heat Solar radiation (Winter) Louver Structural bracing member Solar radiation (Summer) Solar radiation (Winter) External louver Glass External wall Structural bracing member Reflection Fresh air Fig 1. Functions of the integrated-façade-system change of Japan’s building regulations in 1981 were unable to resist a major quake. Considering the fact that about 50% of the current building infrastructure in Japan dates from before 1981, a significant amount of buildings are in need of a seismic resistance structural retrofit to meet the current regulations and to reduce the amount of casualties in the event of a major disaster. In this way, it is clear that the renovation of buildings will become influential over the current scrap-and-build approach. This research is challenging the widely spread retrofit methods which focus purely on the structural aspect and proposes an integrated-façade-system. This integrates façade engineering techniques in the fields of structural engineering, environmental and architectural design and adds extra value to the reinforcement of the existing buildings. The functions aimed for in this integratedfaçade-system are shown in Fig 1. The structural function seeks to reduce the maximum response displacement and to improve the load-bearing capacity through energy absorption by a bracing system. The environmental function seeks to reduce the seasonal energy input through the addition of control mechanisms that block direct sunlight during summer and effectively use the light in winter. As the deployment of this integrated-façade-system on the outside of the building ensures continuity of its use, the system can be designed in such a way that it is also visually aesthetic. The implementation of the system therefore demonstrates potential. Methodology Evaluation axes for integrated-façade-systems In order to judge the effectiveness of the proposed integrated-façade-system, evaluation axes were established in accordance to the fundamentally required functions mentioned above: ‘Design’-axis, ‘Shelter function’-axis and ‘Environment’-axis. More detailed requirements concerning the individual axes have been defined as shown in Fig 2. As the louvers become a not merely functional element in the integrated-façade-system, but also can satisfy aesthetic demands, a ‘cost’-axis has been introduced, intersecting with each axis. Studied integrated-façade-systems The variety of integrated-façade-systems is dependant on the diversity in louver design, the structural bracing member and their possible combinations. An overview of existing louver types has been created based on a bibliographic survey. Apart from the existing types, a new bound louver type has been proposed, Fig 3. This type has openings of varying size and shows, with its unique design, almost unlimited possibilities for the future. As structural bracing member, a Buckling Restrained Brace (BRB) is used to increase the seismic resistance. The BRB is a patented product by Arup Japan and Kanagawa University and was developed with minimum dimensions and in such a way that it can be perfectly integrated and harmonised with the louver system. Considering one floor, one span and only one possible intermediate node, some of the possible BRB arrangements are shown in Fig 4. Expanding the method to multiple floors 43 Construction level Integrated façade system Evaluation axis level Design axis Shelter func. axis Environment axis Design element level Design Response control Daylight factor Compatibility Load bearing cap. Daylight use Creativity Rigidity Maintenance Fig 3. Louver types and bound louver type Cost Fig 2. Evaluation axes of the integrated-façade-system and eliminating the options which have no clear structural benefit leaves us with five considerable arrangements. When combining louvers and BRB, it is seen that only in the case of diagonal, oblique lattice or bound louvers the BRB can be integrated into the louvers. In the other cases the systems are separated. Focusing on the combined systems (diagonal, oblique lattice and bound louvers), one separated system (horizontal louvers) and excluding the combination of diamond BRB arrangement with bound louvers, with results as shown in Fig 5. Design axis Using the earlier defined evaluation criteria for the design axis – compatibility with existing structure, aesthetic appeal and creativity – a computer graphics questionnaire, was held amongst 50 randomly selected architecture students to evaluate the considered louver-BRB designs. Construction 1 scored highest in compatibility, while construction 8 had the best aesthetic appeal and constructions 11, 12 and 13 were seen as most creative. Construction 1, 2 and 3 scored notably low in creativity as they are the least innovative. Shelter function axis To evaluate the structural axis, the BRB arrangements are applied to a hypothetical four-storey building from 1971. This is before the revision of Japan’s reinforced concrete calculation standard. A push-over analysis is carried out, taking the presence or absence of partition walls and plasticity or non-plasticity of the beams as its parameters and shows that the structure has the lowest strength index when no partition walls and the plasticity of the beams were considered. Based on this result, the required quantity of reinforcement is defined to comply with the current regulations, this is for a seismic index Is=0.7 when the toughness index is at a maximum of F=1.05 (inter-storey drift approximately 1/250). When reinforcing the structure by means of H-sections (250x250x9x14mm) used in the strong axis direction, satisfying results are reached for X, V, N or diamond BRB arrangements. However, in the case of diagonal or bound arrangement, the requirements are not met. When replacing the H-sections by rectangular steel sections of 250x250x12mm, 350x250x11mm, 450x250x12mm and 500x250x14mm (which respectively approximately have the same, two, four, and six times the geometrical moment of inertia of the original H-section), an increase in 44 Fig 4. BRB arrangement methods 1 2 3 4 5,6,7 8 9 10 11,12 ,13 Fig 5. Computer graphics of the considered louver-BRB arrangements and achievement of the required load-bearing capacity and rigidity can be seen as the steel section enlarges. mock-up test with two types of diagonal louvers with a different pitch was done to confirm its efficiency: a rectangular section (short side facing the outside) and a square section (corner facing the outside) as shown in Table 2. Finally, so as to judge its influence, a dynamic response analysis of the structure with the applied reinforcement is carried out. The cases with H-sections or rectangular sections 250x250x12mm or 500x250x14mm reinforcements were considered and El Centro-NS, Taft-EW and Hachinohe-NS seismic waves, were introduced. Over a two-day period and at an interval of one hour from 09:00 to 15:00, two types of measurements were taken: the illuminance was registered by an illuminometer and a photograph was taken using a fish-eye lens. The result in Fig 6. shows that the inter-storey drift is smaller than 1/250 for the cases where the BRB is arranged in X,V,N and diamond shapes. In the case of diagonal arrangement with H-sections the inter-storey drift exceeds 1/250, however, when rectangular sections are applied, values smaller than 1/250 become possible as the section size increases. We can conclude that the installation of the louvers has no negative influence on the amount of daylight in the interior. On the contrary, the louvers are diffusing the daylight which improves the daylight levels further away from the opening. As the louvers are controlling the amount of direct sunlight entering at the perimeter, visual discomfort of the building users will also be reduced. Environment axis The intergrated-façade-system has different levels of enviornmental functionality. A first objective is to reduce the heat load introduced in the interior by reflecting the solar radiation. According to the orientation and the solar altitude in a particular situation, optimal heat shielding can be obtained by adjusting the shape and pitch of the louvers. Cost axis The cost is evaluated on the quantity of components used in the BRB-louver arrangements and is for one floor span. At the same time, louvers can also serve to control the amount of daylight entering the perimeter. As in many cases the installation of louvers is rather seen to have a negative effect on the brightness of the space and the view to the exterior, a full-scale (one floor, one span) First, the amount of steel in the columns and beams according to the BRB arrangements is considered. For X, V, N or diamond shape, the steel quantity is equal, but when rectangular steel sections are used (inter-storey drift smaller than 1/250) it increases considerably. As the manufacturing and transportation costs rise together with the rise in material quantity, it could be assumed that the cost when using rectangular sections is higher. However, this difference needs to be handled with care as the study does not Taft EW wave 5 1/150 1/250 1/150 No reinforcement 2 4 RF floor 3 3 2 0 10 20 KEY: X shape 30 V shape 40 N shape 1/150 No reinforcement 4 RF floor RF floor 1/250 No reinforcement 4 1 Hachinohe NS wave 5 1/250 1 50mm Diamond 3 2 0 10 Diagonal (H-250) 20 30 Diagonal ( -250) 40 1 50mm 0 10 20 30 40 50mm Diagonal ( -500) Fig 6. Dynamic response analysis results 15 10 Environment axis Cost axis Does not apply 4500 BRB yield not at the same time Not sufficient Rigidity All BRB yield at the same time Daylight Factor Average value of daylight factor Minimum value of daylight factor is under 10% at 450mm is under 10% at 450mm Does not apply Daylight use Average value of daylight factor at 2250mm increases when installation of the louvers Does not apply Minimum daylight factor at 2250mm increases when installation of the louvers Maintenance Linked with the louver area of the cost axis 3/ (steel quantity of each studied construction/smallest steel quantity of all studied constructions) Louver surface area 3/ (Louver surface area of each studied construction/smallest louver surface area of all studied constructions) Table 1. Rating standards for the considered BRB-louver arrangements Results and discussion Based on the results of the investigations of design, shelter function, environment and cost axes, a quantitative evaluation of the considered BRB louver arrangements is attempted. Rating standards, as shown in Table 1. are defined for each criterion and comparatively stable performances are discovered in the evaluation of the studied arrangements. Whilst the option with horizontal louvers gains the highest evaluation, all considered arrangements are verified as being realistic and potentially applicable options. Additional research: fire safety Severe earthquakes are usually accompanied with the outbreak of fire. As the integratedfaçade-system is seen to be applied to major gathering points in the cities’ disaster evacuation plans in the first instance, it is important to know the behaviour of the fire to ensure the safety of the building’s users and to prevent the fire from spreading to upper floors and adjacent buildings. No research had been done on the influence of external louvers on the way the fire spreads. and a full scale mock-up test was completed to ensure safety of the proposed system. b) Cvalue type Max./min. Average value Direct daylight factor 15 10 5 0 900 0 2700 Measurement point (mm) 4500 Does not yield Quantity of steel include the amount of steel required for (transverse) web stiffeners in the case of H-sections; this becomes more critical as the load-bearing capacity of the structure increases. Next, the total surface area of the louvers in one floor, one span is considered as an increase in surface area is connected to an increase in maintenance cost in addition to the material cost. It could be assessed that this value is the smallest for diagonal louvers. 20 The experiment was done for several different distances between the louvers and the opening. The behaviour of the ejected plume from the opening was verified through temperature and radiated heat measurement around the opening and the louvers and through study of the shape of the plume. As neither the installation of the louvers nor their distance to the opening had an effect on the temperature on the central axis of the opening, it is asserted that the ejected plume separates two main directions due to their presence. Furthermore, it can be considered that, after ejecting from the opening, the plume will rise in accordance with the inclination of the diagonal louvers. At last, the thermo-camera confirmed that the shape of the plume is depending on the distance of the louvers to the opening and that the point of that shape-transformation is at about 600mm-800mm from the opening. Conclusion and next steps The undertaken research on the proposed integrated-façade-system shows that it is more than a static unique idea which consists of combining a necessary seismic retrofit of existing buildings with measures to significantly improve its environmental and architectural quality. On the contrary, it is a dynamic, flexible system which is full of possibilities and has the potential to be applied to a wide range of buildings and to anticipate on what the future, with ever growing requirements towards energy conservation and sustainability, will bring. 100 mm 250 mm 100 mm BRB: Buckling restrained brace 100 mm 5 Shelter Response Max. response; Interstorey drift Max. response; Interstorey 0 0 drift 1/150 or less Control 1/250 or less function. 900 2700 4500 900 2700 0 0 axis Load bearingpointSufficient load bearing Measurement (mm) Measurement point (mm) Capacity reinforcement a) R type 350 mm Creativity 2 20 C type 25 Max./min. value Average value 1 Direct daylight factor 100 mm Questionnaire Compatibility R type Score 350 mm 5 Design 25 Daylight factor (%) Exterior 10 Max./min. value Average value 3 Direct daylight factor Louver section Daylight factor (%) Design 15 axis Without louver Evaluation criteria Daylight factor (%) 25 Evaluation 20 axis BRB: Bucklingrestrained brace Table 2. Mock-up used for illuminance measurements Acknowledgements This research was carried out using a Ministry of Land, Infrastructure, Transport and Tourism Grant; We would like to thank Prof. Iwata and Prof. Iwamoto of Kanagawa University. References Hikone Shigeru, Misawa Yutaka, Makoto Nakamura, Iwamoto Shizuo, Iwata Mamoru: Lighting environment of diagonally arranged louver on integrated façade system, Journal of Architectural Institute of Japan, No.644, 2009, 1187-1193. Makoto Nakamura, Hikone Shigeru, Misawa Yutaka, Iwamoto Shizuo, Iwata Mamoru: The integrated façade consists of louvers and buckling-restrained-braces as a building system, Journal of Architectural Institute of Japan, No.647, 2010, 121-129. Misawa Yutaka, Hikone Shigeru, Aburano Kenji, Omiya Yoshifumi, Iwamoto Shizuo, Iwata Mamoru: Fire experiment on diagonally arranged external louver for integrated façade system, Journal of Architectural Institute of Japan, 2010. Fire and Disaster Management Agency: Great Hanshin Awaji Earthquake 2006. Ministry of Land, Infrastructure, Transport and Tourism. Kaneki Yohei, Hikone Shigeru, Yamashita Tetsuo, Iwata Mamoru: Seismic Strengthening by the buckling restrained braces arranged diagonally, Journal of structural and construction engineering, 73 (634), 2008, 2215-2222. Japan Construction Disaster Prevention Society: 2001 Revised Edition Seismic Retrofit Design Guidelines and Explanation for Existing Reinforced Concrete Construction Buildings, 2001. The Building Centre of Japan: Structure design guideline for high-rise building, 2002. 45 Sustainable and quake resistant façade for existing buildings EL centro NS wave 5 Measuring change of coastal defence structures using advanced 3D laser mapping techniques Authors: Ilse Steyl, Dean Crowley, Patrick Kuhn and Simon Bray 46 Main Rivers The challenge to effectively manage coastal assets to encompass changing use, sea level rise, coastal squeeze and development pressures, can be great. Monitoring these assets is central to this management process. Using advanced technology could assist in time and cost savings. Study Area Urban Areas Groynes Lymington New Milton Highcliffe Groynes Bournemouth nt ole S he Milford on Sea T Keyhaven Christchurch Bay Abstract Technological evolution and the increased development of remote sensing technology provide an opportunity to rapidly acquire data and accurately measure changes over wide spatial areas. The monitoring of ecologically and economically valuable coastlines and inter-tidal zones is difficult, mainly due to poor access, large extent and tidal restrictions. This can result in reactive rather than proactive management. In this study the use of terrestrial Light Detection and Ranging (LIDAR) equipment, as a cost effective and accurate tool for monitoring coastal defence structures and measuring change, is discussed. The survey location was at Highcliffe in Christchurch Bay along the south coast of Britain, where rock armour groynes have been constructed to protect the coast against erosion. Recorded information highlighting the impacts of erosion in the area has been recorded since the 1700s. LIDAR surveys of three of the ten rock armour groynes were undertaken on three occasions during low tide: 22 October 2008, 19 February 2009 and 9 September 2009. Data were processed using remote sensing and Geographic Information System (GIS) software. Areas of erosion and deposition between the three survey dates were calculated. The research demonstrated the value of using terrestrial laser scanning to collect detailed data for small area monitoring of coastal defences. The relative ease and speed of collecting data can allow for frequent return visits to monitor structures and their integrity. Totland 0 0.5 1 2 3 Isle of Wight 4 kilometres Fig 1. Study Area, Christchurch Bay, Hampshire Introduction Economically and ecologically valuable coastlines and intertidal zones are difficult to monitor. This is mainly due to poor access, large extent and tidal restrictions. As a result, management can be piecemeal and reactive rather than proactive. It is a dynamic challenge to effectively manage coastal assets to encompass changing use, sea level rise, coastal squeeze and development pressures. Access to cutting edge techniques enabling improved understanding of coastal processes and mechanisms can be very valuable. Typically topographical and ecological survey methods have depended on field observations, or more recently airborne remote-sensing techniques. Technological evolution has led to remote survey systems becoming faster and more affordable. This, along with the compact nature of modern systems, has paved the way for the further development of Terrestrial LIDAR Scanning (TLS) possibilities. TLS allows rapid data acquisition and precise measurements over a wide spatial scale at relatively low costs. TLS has been used in the urban environment for examining building dimensions or deformation monitoring of constructions for example. Its use has been extended to the natural environment and is now also commonly used in surveys as diverse as archaeology, landslide dynamics, monitoring of volcanic activity and fluvial morphodynamics. However, TLS has not yet been extensively used in coastal surveying, although Nagihara et al (2004) employed it for mapping sand dune morphology, whilst Rosser et al. (2005) used the technology to monitor cliff erosion. Shoreline Management Plans and coastal strategy studies have consistently identified the need for development of coastal management. The use of LIDAR in monitoring programmes is listed as one of the objectives of the Strategic Coastal Monitoring Programme for the South East Region. This research was divided into two separate phases, each focusing on a distinct environment, measuring specific facets. The first phase of the research examines: • The use of terrestrial LIDAR as a cost effective and accurate tool for monitoring the change of coastal defence structures: analysing design and monitoring to enhance proactive management • Using terrestrial LIDAR to map intertidal algal community distribution Monitoring coastal defence structures The focus of this study is on the mapping of coastal defence structures and change measurement. The survey location is at Highcliffe on Christchurch Bay along the south coast of Britain, Fig 1. A number of rock armour groynes have been constructed to mitigate erosion along the coastline. The geology of the area is dominated by the Barton beds and the overlying Headon Hill Formation (of the Solent Group), all of which form part of the Hampshire Basin and deposited between approximately 42.1-35.4 million years ago. The whole of the cliff face fronting Christchurch Bay is a Site of Special Scientific Interest (SSSI). The cliff face exposes the only complete late-middle to early-late Eocene sequence in the world. Records of erosion at Highcliffe date from the early 1700s. Coastal defences in the form of better drainage were constructed during the mid-1800s, but these were eventually destroyed. Further work was undertaken in the beginning of the 1900s and by the 1960s a cliff stabilisation scheme was completed as well as timber groynes constructed to enhance the protection of the coast by absorbing wave energy. By 1991, the timber groynes were replaced with alternative long and short rock armour groynes, see Fig 2. 47 Fig 2a. Rock armour groynes at Highcliffe Fig 2b. Rock armour groynes at Highcliffe Fig 4a. Example of groyne after cleaning Fig 4b. Example of groyne after cleaning Stabilisation work along the cliff included improved drainage and planting of salt tolerant grass mixes. of the registrations during post-processing, while providing the added benefit of shifting the scans to their correct world coordinates. The survey control was undertaken in conjunction with a qualified surveyor, who surveyed strategically placed reflectors, at the same time as they were captured by the laser scanner. Due to time constraints, it was not feasible to utilise survey control for all scans, so it was limited to use on the scan positions on the cliff above the groynes, as these areas were not impacted by the incoming tide. Methodology Data collection and processing Three of the long rock armour groynes labelled H2, H4 and H6, were surveyed on three separate occasions during low tide: 22 October 2008, 19 February 2009 and 9 September 2009. Data collection was undertaken using a Riegl LMS Z420i terrestrial laser scanner, Fig 3. This has a range of up to a 1000m with a level of accuracy of 100mm. Properties of scattered light are measured to find the range to a distant target/surface. A set of three dimensional coordinates is generated, normally referred to as a ‘point cloud’. Mounted on top of the laser scanner is a Nikon D100 digital SLR camera, allowing digital photos to be captured whilst scanning, provide colour data for each point. At each groyne, scanning was undertaken from at least three different vantage points. One at beach level east of the groyne, another at beach level west of the groyne and one or more on the cliff looking south onto the groyne. During some return surveys it was necessary to scan the side of the groyne from more than one location due to the shifting of shingle on the beach, creating shadow zones. This caused gaps in the data collection that needed to be filled. At each of these scan positions, two single scans were performed; one a 360° overview scan at low resolution and the other a fine scan (high resolution) focusing on the groyne. The 360° scans were required mainly for the purposes of registration for the post-processing stage, while the fine scans provided the detailed point data on the relevant areas of interest. Survey control was undertaken simultaneously with the laser scanning on the first and third visits, in an endeavour to maximise the accuracy 48 The survey scans were translated, rotated and registered to the controls scans using I-Site v3.1.1 software. Once all the scans were registered together, a quality check was undertaken to ensure no errors were made in the registration and each scan fitted seamlessly into place. Data were then cleaned manually in I-Site to remove any stray points such as people, vehicles, or rain drops, reflections from water and glass. This included any other reflective surface and distant points not relevant to the study such as the upper cliffs and groynes further along the beach. At this point the data was filtered removing any points nearer than 25mm to any other point. This effectively thinned the point surface near the high density ‘point circle’ immediately around the scan position, resulting in a more manageable file size, Fig 4. Data was then exported as a text file (with the associated x, y, z coordinates for each point) and imported into ArcGIS 9.3.1. A surface model was created, which represents the survey area as a continuous raster surface (in this case each cell was 25cm x 25cm). The raster surface is generated through interpolation of the point data. Interpolation predicts the values for cells from a limited number of sample points. The method used to generate the surface for the groynes is called natural neighbour interpolation. This method finds the closest subset of input samples to a query point and applies weights to them based on proportionate areas (rather than distances) in order to interpolate a value. Fig 3. Riegl LMS Z420i Three surfaces were created representing the data for each visit. To determine the areas of change where erosion and deposition occurred between the three time periods, geoprocessing was undertaken using tools within the 3D Analyst™ extension within ArcGIS. The surface was generated through The 3D Analyst™ and allows effective visualisation and analysis of surface data through the application of a cut-and-fill procedure, in which the elevation of a surface is modified by the removal or addition of surface material. The surface of a specific location at two different time periods are used to identify where surface material has been removed, added or where no change has occurred. Volumes and areas demonstrating the change between each surface was generated. Fig 6. is a representation of the changes that occurred between the three survey dates. A comparison was made between the first and second surveys, the second and third surveys and lastly between the first and third surveys. The changes in elevation of surfaces can be visualised by generating a profile of a specified area. Fig 5. shows the changes in profile over time for groyne H2. Results and discussion The preliminary analysis showed clearly which areas are prone to erosion and deposition. In all three surveys, the area with the highest build-up of shingle was to the west of groyne number H2, whilst the areas around the base of the groynes were more prone to erosion. It is also interesting to note that the area to the east of groyne number H4 is more prone to erosion. The highest build-up of material took place to the west of groyne number H2 during the winter storms between the first and second visit. This can clearly be seen in Fig 6. which shows the profile of groyne H2. The area between H4 and H6 are more prone to erosion. Profile of groyne H2 – third visit 3.0 3.0 3.0 2.5 2.5 2.5 2.0 2.0 2.0 1.5 1.0 1.5 1.0 0.5 0.5 0 0 0 10 20 30 Distance (m) 40 Elevation 3.5 50 1.5 1.0 0.5 0 0 10 20 30 Distance (m) 40 50 0 10 20 30 Distance (m) 40 50 Fig 5. Changes in profile between visits for groyne H2 0 25 50 100 150 200 N Deposition Erosion Fig 6a. Erosion and accretion between first and second survey Acknowledgements This research was undertaken in colaboration with University of Southampton (Dr Simon Bray). We would like to thank Steve Woollard and Mike Hinton from Christchurch Borough Council and Dr Dafydd Lloyd Jones (EMU Ltd). References Dale W., Archaeology of West Hants: A Natural History of Bournemouth and District, including Archaeology, Topography, Municipal Government, Climate, Education, Fauna, Flora and Geology. Bournemouth Natural Science Society, Bournemouth. Morris, D (ed), 1914. Fig 6b. Erosion and accretion between second and third survey Harland WB., Cox AV., Llewellyn PG., Pickton CAG., & Walters R., A Geologis Time Scale. Cambridge University Press, New York, 1982. Hinton MT., The causes, effects and mitigation strategies relating to coastal landslides at Highcliffe and Naish Farm on the Dorset – Hampshire border, Christchurch, Dorset County Council, 2007. Lichti DD., Gordon SJ., Stewart MP., Ground-Based Laser Scanners: Operation, Systems and Applications, Geomatica, 2002. Fig 6c. Erosion and accretion between first and third survey Fig 5: representation of the changes that occurred between the three survey dates. Fig 6. Erosion and accretion between first and third survey During the survey period, repairs were needed to groyne H6, which confirms the results of the analysis. The deposition of material to the west of H2 was also very significant and more detailed analysis of the volumes will be undertaken. The survey data collected for the three groynes took approximately five to six hours on each of the three survey days. The survey control data was collected over a similar time frame by a separate survey team. The cleaning and processing of the data for each survey was undertaken over a period of two person days, with an additional day for survey control to be referenced and checked. The rapid acquisition of the data and relative ease of providing the raw data in a format to allow for the generation of terrain models, geostatistical analysis and undertake measurements, allows for cost savings and the storage of data in an accessible format. Monitoring of the coastal defences can therefore be undertaken in a structured way, allowing for rapid analysis. Conclusion and next steps The research demonstrated the value of using terrestrial laser scanning to collect high detailed data for small area monitoring of coastal defences. The relative ease of collecting the data and speed at which it takes place can allow for frequent return visits to monitor structures and their integrity. The information can be used for routine monitoring of coastal defences and could highlight where repair work is needed before structures deteriorate or collapse. The data collected for this study will be further analysed in conjunction with data from other sources to give a better understanding of the vertical and horizontal changes along the groynes. More detailed analysis of the volumes and areas of the deposited and eroded areas will be undertaken. This will be done through comparison between different interpolation techniques, for example kriging. Maas HG., & Vosselman G.,Two algorithms for extracting building models from raw laser altimetry data, ISPRS Journal of Photogrammetry & Remote Sensing 54, 1999, 153-163. Milan DJ., Heritage GL., & Hetherington D., Application of a 3D laser scanner in the assessment of erosion and deposition volumes and channel change in a proglacial river. Earth surface Processes & Landforms. Vol. 32, 2007, 1657-1674. Mouginis-Mark PJ., & Garbeil H., Quality of TOPSAR topographic data for volcanology studies at Kilauea Volcano, Hawaii: An assessment using airborne lidar data. Remote Sensing of Environment. Vol. 96 (2), 2005, 149-164. Nagihara S., Mulligan KR., & Xiong W., Use of a three-dimensional laser scanner to digitally capture the topography of sand dunes in high spatial resolution. Earth Surface Processes and Landforms, Vol. 29, 2004, 391-398. Rosser NJ., Petley DN., Lim M.,Dunning SA., & Allison RJ., Terrestrial laser scanning for monitoring the process of hard rock costal cliff erosion. Quaterly Journal of Engineering Geology & Hydrogeology. Vol. 38, 2005, 363-375. Rowlands A., & Sarris A., Detection of exposed and subsurface archaeological remains using multi-sensor remote sensing. Journal of Archaeological Science. Vol. 34 (5), 2007, 795-803. Schulz W.H., Landslide susceptibility revealed by LIDAR imagery and historical records, Seattle, Washington. Engineering Geology. Vol. 89 (1-2), 2006, 67-87. Sibson R., A brief description of natural neighbour interpolation In: Interpolating multivariate data, John Wiley & Sons, New York, 1981. Worthing Borough Council: Strategic Coastal Monitoring in the South East. 49 Measuring change of coastal defence structures using advanced 3D laser mapping techniques Profile of groyne H2 – second visit 3.5 Elevation Elevation Profile of groyne H2 – first visit 3.5 Digital infrastructure and changing practices in engineering design Authors: Jennifer Whyte, Sunila Lobo, Mark Neller, Sarah Bowden 50 New text intro skills are needed to compete, as integrated software solutions provide a digital infrastructure for projects. This changes the practice of information management and engineering design on next generation projects. Abstract Integrated software is becoming used as a digital infrastructure for the delivery of large building and infrastructure projects. Its introduction fundamentally changes engineering design as project stakeholders work together through shared sets of technologies. In this research, the organisation and management of design through integrated software was investigated across three Arup projects: Motorway 6 (M6) Toll, Channel Tunnel Rail Link (CTRL) and SAFElink Motorway. On each of these projects design teams have been at the forefront of developing new ways of using the digital infrastructure for delivery and of adding value to the client in data handover. The research highlights how strong in-house technology capabilities are important to Arup in responding to diverse client requirements and system configurations on projects and in contributing to innovation on these projects. It identifies new skills that are needed within the industry as integrated software solutions provide a digital infrastructure for project delivery and draws out lessons for information management on the next generation of projects. Introduction Capabilities in using and developing integrated software are increasingly central to the delivery of major building and infrastructure projects. Such tools open up possibilities to improve design processes by collating and making available data for better decision-making. Their introduction is part of a broad move towards advanced manufacturing in which the opportunities for technological innovation shift. In construction, integrated software involves a centralised repository of data; standard methods for using that data and the data itself, which often takes the form of a 3D Computer Aided Design (CAD) model, with associated attributes. Studies into the use of such digital representations have described them as propagating waves of innovations across project supply chains. Fig 1. The three projects: M6 Toll; SAFElink; and the Channel Tunnel Rail Link (CTRL) The changes brought about by integrated software are having profound impacts on the way that design and construction work is organised. On one of the projects studied in this research, rainfall data provided an example of how the digital infrastructure changes the visibility of information across firm boundaries. There was an issue with rain on the construction site. It had been raining heavily and given the geology of the area this was turning ponds orange in a neighbouring golf club that was about to host a major competition. To design temporary holding ponds, an engineer seconded from one part of the project to another needed to know how much it rained. However, as the license for this data had been agreed at firm level then the information could not be released. To fix the problem the rainfall had to be recalculated. While firm-level ownership of information is normal practice, the integration of the design data through an integrated software solution strengthened the sense of the project being at the relevant organisational level for such a purchase. Access to a project-wide dataset made this project-level more visible in engineers’ day-to-day work. Hence, in the research community, attention is shifting from a focus on the technical challenges of integrated software to organisational questions. Return on investment from the use of building information models is beginning to be analysed, with 70% of owners reporting positive returns in a recent survey. At a project level, integrated software solutions have become a strategic rather than purely operational matter and by focusing on strategic decisions about IT use on projects the research community is responding to and supporting this shift. At an industry level, the growing maturity of the technologies for supporting integrated working has lead to initiatives that bring researchers and practitioners together. This supports standardisation of information management processes, both in the UK and internationally. As the example of rainfall data suggests, by increasing the available information, new software has both intended and emergent properties. They get used alongside older work practices and recombined in new forms of practice. While digital technologies are transforming the delivery of both buildings and infrastructure, most research has paid attention to their use in building design. Here, researchers detail how ‘an accurate virtual model of a building is constructed digitally’, providing a building information model. Less attention has been paid to the benefits and challenges of implementing such an approach in the infrastructure sector. Yet there are some particular challenges in infrastructure, which involve global supply chains, multiple stakeholders and significant organisational complexity in project delivery. 51 The aim of this study is to develop a practical understanding of how digital modelling, collaboration and project management tools support the delivery of global infrastructure projects. Hence the research seeks to understand how engineers and other stakeholders use information on projects by looking at their every day practices, rather than the formally documented processes. The focus is on uncovering the challenges and opportunities that engineers face and understanding how their practices are institutionalised in business and regulatory environments. The purpose is to inform the next generation of engineering and design projects. Methodology The research is part of a trajectory of work on technologies and practices at the University of Reading that has drawn lessons from large UK projects such as Heathrow Terminal 5 (T5), Crossrail, and the London 2012 Olympics, and which involved a range of leading industrial partners. The academic research team worked with Arup to analyse technologies and practices on three Arup projects: Motorway 6 (M6) Toll, Channel Tunnel Rail Link (CTRL), and SAFElink, see Fig 1. They looked for lessons that could be transferred to ongoing work and disseminated more widely to improve UK construction. Data on these projects, and the information management processes and tools used, was collected using three set-up meetings, thirty six semi-structured interviews, non-participant observation, and access to internal documents, databases and information. The research was designed as an embedded case-study, studying three projects within the same firm, with data collection in 2008-9. Analysis is ongoing through a process that iterates between the data and the literature to develop a theoretical contribution. Preliminary results were discussed in feedback meetings with each of the project teams and with Arup Directors. Results and discussion By studying three projects that represent milestones in Arup’s innovative use of information management on projects, this research charts the growing use of these technologies across the infrastructure division. Though remembered as all being equally large projects within Arup, the projects are of different orders of magnitude, scale and complexity, as shown in Table 2. The M6 Toll near Birmingham is the first toll road built in the UK under the private finance initiative; CTRL, now known as High Speed 1 in its operational phase, is a high-speed railway link between London and the Channel Tunnel in the UK; and SAFElink is a motorway link between Brisbane and Ipswich in Australia. CTRL is an order of magnitude larger than the other two projects. Its timescale for its delivery overlapped with that of M6 Toll and staff that worked on the early stages of M6 Toll then spent time at CTRL before returning to the former project. On the other side of the globe, SAFElink is a more recent project that is pioneering new technologies and processes. 52 Knowledge and standards Best practice, Regulations, Specifications CAD Drawings, Calcs, Engineering Briefing Functional req, Estimates, Conditions, Requirements Project Data Demolition Refurbishment Rebuild, Demolition, Restoration Facilities Managment Letting, Sale, Operations Maintenance, Guaranties Construction Management Scheduling, Logistics, 4D Procurement Product Database, Price Database VRML & Simultions Visualisations, 3D Models, Light & Sound, Life Cycle Specifications Spec Sheets, Classifications, Standards, Estimates Fig 2. Diagram showing significant areas of construction projects,which benefited from using central source project data All the projects were important in the development of in-house capabilities for the use of integrated software solutions. Design teams gained experience developing and using integrated processes and tools. The projects involved coalitions of design and engineering firms working together to deliver large scale infrastructure, with the work divided up across the many offices, disciplines, teams and firms involved. On one project, for example, twenty offices were involved in the delivery of project deadlines at one stage. Here the study observed how the integrated software solutions become vital in the coordination of design work and acted as a digital infrastructure for project delivery. The main innovations were the use of a collaboration extranet software, Integration, as a collaboration tool for project wide document dissemination on M6 Toll, the implementation of IT tools and processes across CTRL and their stability for the duration of the project and work using 3D modelling on SAFElink. The three projects had features that led to different strategic decisions about IT use outlined below. M6 Toll The two design firms involved were based in home offices rather than being colocated. Arup introduced Integration to exchange documents and data across the project. This software facilitated this method of working in an efficient and effective manner, enabling the sharing of published documents. The main role of this digital repository was to make visible knowledge about the audit trail of design deliverables and where the project was in the review process. It managed the workflow approval process, it overcame the version control problem, enabled digital global information sharing and preserved final design knowledge. CTRL The colocation of a project team away from the firms’ offices meant it was possible to have all designers using the same software tools and to hold software stable over the limited duration of the project, rather than upgrading the software alongside corporate IT systems. The project was highly successful in implementing a robust IT strategy and processes for shared use of data. Holding software stable simplified matters for the project itself since it did not have to deal with the costs and disruption of upgrades, but it did mean that Arup staff returning to their offices were unfamiliar with new technology implemented on the corporate network while they had been on the project. SAFElink A small and more recent project, but the largest project run out of the Brisbane office. The Arup design contract was ‘novated’ to the contractor, who was unable to provide the IT infrastructure to support the wide range of software tools required by the design team. Arup therefore used the corporate systems and these were augmented by a project extranet collaboration tool (Incite) to support data exchange between several Arup offices and the contractor. Processes and data management procedures were developed to support the 3D CAD modelling tool deployed to efficiently facilitate multilocation and multidisciplinary design teams and the complex approvals process. As a result of the success of this approach Arup is now looking to disseminate it more widely across the firm. The management of information through the whole life-cycle of the built environment is one motivator for using integrated software solutions. The ability to add value to the client in data-handover has steadily improved across these projects. On M6 Toll, at the end of the project documents were taken from the extranet CTRL SAFElink Teams left to use own design tools. Integration used as a collaboration tool to exchange documents and data. Software was updated through the project. Because of an overlap in timescales and staff there was a transfer of learning to M6 Toll from CTRL Client organisation selected the best tool for the job. Limited duration of the project meant little incentive to keep up with latest software releases, so software versions were fixed from 1996 for the project duration Arup used own IT system, a 3D CAD modelling environment and a project extranet (Incite) to facilitate efficient iterative design, review and approvals. Software was updated through the project Technology through life Final design documents were handed over in paper and on CD to the client. Design and other documents were handed over to the client in an electronic format. Arup will hand over 3D models and a structural model for bridges to the client. Project features influencing IM strategy Two design teams, geographically separate and from two different organisations, operating on independent IT networks. Arup developed the Integration solution to provide an IT infrastructure. Teams were colocated and used client’s IT network and software. Brisbane was the lead Arup office, with design undertaken globally across a range of Arup offices. Significant multilocation interdisciplinary review and complex contractor and client approvals. Not all parties on the Arup IT network. Table 1. Preliminary analysis on IT use on the projects and handed over to the client in electronic and paper forms. On CTRL the documents were handed over to the client in an electronic format, while on SAFElink Arup will hand over 3D models and a structural model for the bridges. This is due to Arup’s growing capabilities and also the maturity of clients being able to manage advanced data-sets. Though there was a significant investment in training on each of these projects, there were difficulties in interfacing with the wider business and regulatory environment. In the approval process the terminology used in the system was found to be extremely important in enabling a digital review of the design rather than a paper based review. For example, on one project it became important to enable the client design advisor to “receive” rather than “approve” documents so that they could engage with the system. This small change in terminology enabled the team to use the system. It then became a record of decisions that were made as to reject a drawing, the reviewer had to input the rationale for his or her decision, making these judgments more widely visible to the appropriate parties within the project. These findings are summarised in Table 2. The strong similarities between the three cases are that each project made use of advances in IT tools and that this use has been influential across Arup. On each of these projects, design teams have been at the forefront of developing new ways of using the digital infrastructure for delivery and of adding value to the client in data handover. The many areas of project delivery in which these technologies become significant is summarised in the Arup diagram show in Fig 2. Conclusion and next steps Integrated software is becoming used as a digital infrastructure for large building and infrastructure projects. The research indicates the increasing sophistication of this digital infrastructure being used to integrate work across the many offices, disciplines, teams and firms involved in project delivery. While most studies have focused on its role in large building projects, here the focus has been on the role of integrated software in three large infrastructure projects. Information management becomes a core skill area for infrastructure projects. The introduction of integrated software as a digital infrastructure for delivery is changing practices in engineering design, making information more broadly available across projects and increasing the distribution of design work. For clients and for design, engineering and construction firms, new skills are needed to compete on the next Type of project M6 Toll CTRL SAFElink First toll road built in the UK under a private finance initiative Railway line built between London and the Channel Tunnel, Folkstone, UK Motorway link between Brisbane and Ipswich, Australia Overall cost approx. £500m approx. £6bn approx. £400m Project dates 1986-2003 1990-2007 2006-2010 Main innovation Use of Integration as a collaboration tool for project wide document dissemination. IT tools and processes implemented and fixed for the duration of the project Work undertaken using 3D modelling and this was provided as a deliverable to the client Table 2. Summary of project characteristics generation of projects. These include 3D modelling skills; wider software skills; standardised approaches to information management and a host of other skills in developing and implementing new processes for the coordination of ongoing design data. Arup is developing the strong in-house technology capabilities needed to add value through, and interact with, different information management approaches. Although there are industry-level initiatives to create standards across projects, each major project is currently organised differently. Hence, for Arup, understanding the capabilities of digital technologies and how to exploit them effectively are essential to delivering design excellence efficiently to the benefit of the client. It is these technology skills that have enabled Arup to be central to the innovations on the projects studied, responding to diverse requirements and different system configurations on projects. There are many remaining research questions about how such systems are changing practices and the emerging modes of design in the digital economy. Two areas in which further research is needed are the handover of information and links between the design and operation of infrastructure, and the different types of technology strategies and their implications for different types of projects. New practices are being developed on projects across the globe. At the University of Reading, the future plan involves developing a repository of international best practice through a new exploration group, the Design Innovation Research Centre. This project is a first step in this wider programme of research. Within Arup, the knowledge gained from this study has been used to assist in the production of an Information Management Toolkit and associated training course. These are intended to help equip project managers with the knowledge and information they need to make the right Information Management decisions to support the requirements of their commissions. Acknowledgements This research was written in collaboration with the University of Reading (Jennifer Whyte, Sunila Lobo).The authors acknowledge their funder, the Engineering and Physical Sciences Research Council (EPSRC) through the project ‘Infrastructure through Life: Technology use in Global Projects’, part of the Innovative Construction Research Centre (ICRC) at the University of Reading. References Avanti (2006) Project Information Management: a Standard Method & Procedure. London, Avanti Toolkit 2 Version 2.0. Boland, R.J., Lyytinen, K. & Yoo, Y. 2007, Wakes of Innovation in Project Networks: The Case of Digital 3-D Representations in Architecture, Engineering, and Construction. Organisation Science 18 (4): 631-647. Eastman, C., Teicholz, P., Sacks, R. & Liston, K. 2008, BIM Handbook: A guide to building information modeling for owners, managers, designers, engineers and contractors, Wiley. ICE IS Panel 2008, Briefing: Knowledge and information management for major projects Proceedings of the ICE - Management, Procurement and Law 161 (1): 9-16. Harty, C. & Whyte, J. 2010, Emerging hybrid practices in construction design work: the role of mixed media, Journal of Construction Engineering and Management, 136: forthcoming. Kallinikos, J. 2005, The consequences of information. London, Blackwell. McGraw Hill, 2009, The Business Value of BIM: Getting Building Information Modeling to the Bottom Line, SmartMarket Report, New York, McGraw Hill Construction. Pavitt, K. 2002, System integrators as postindustrial firms. First draft of What are advances in knowledge doing to the large industrial firm in the new economy (Welcome Lecture) DRUID Summer Conference on Industrial Dynamics of the New and Old Economy – who embraces whom? Shen, G., Brandon, P. & Baldwin, A., Eds. 2009, Collaborative Construction Information Management. London, Spon Research. Whyte, J. & Levitt, R. 2010, Information Management and the Management of Projects, In Oxford Handbook on the Management of Projects (P. Morris, J. Pinto and J. Söderlund eds) Oxford, Oxford University Press. 53 Digital infrastructure and changing practices in engineering design M6 Toll Strategic decisions about IT use Beasties in the creative workplace Authors: Duncan Wilson, Julie McCann, Asher Hoskins, Chris Roadknight, Jane Tateson 54 Wireless Sensor Networks are a key technology in the emerging ‘internet of things’. By monitoring office activity using such a device as the ‘Beastie’, we can gain a better understanding of workplace performance. Abstract This paper presents an evaluation of an installation of the Beastie wireless sensor network to monitor a creative workplace, in this case the University of the Arts Innovation Centre in London. The sensor network was tasked with passively monitoring of the environment and space usage taking into account environmental conditions and activity. The paper focuses on one of the many platforms used in the trial, the Beastie, using an architectural description language implementation called Tesserae, to measure meeting room and social space usage. This data was correlated using feature mapping at node level on tiny 8-bit devices, and state changes were propagated up the network to the database. Some monitoring results are provided and the performance of the algorithms on the small devices discussed. Fig 1. Beastie board Introduction Many approaches have been used to measure the usefulness of a work space, such as integrated workplace performance, Post Occupancy Evaluations (POE) and research into comfort. POE, for example, is predicated upon there being a relationship between people’s performance and the environment within which they act. To obtain insight into how well a space works, correlations with the ambient environment usually involve taking temperature readings, etc. from the Building Management Systems (BMS) or sets of wired sensors placed in the building. This solution is costly in terms of human resources and technologies and moreover, due to the offline nature of the system, the turnaround time of results is quite slow. Our method extended beyond the current practice of getting a general feel of the inhabitants’ experience in a one-off questionnaire; we attempted to examine the relationship between their subjective experiences and environmental conditions over time. To this end, Wireless Sensor Networks (WSNs) captured tangible environmental factors such as light levels, heat levels, and noise levels and correlated these to the workforce’s reports on intangible factors such as perceptions of personal energy levels, well-being, stress, and feelings of connectivity. The purpose of the project was to investigate how pervasive computing can be used to understand the creative workplace. Methodology The results presented are based on a trial carried out at the University of the Arts, Central Saint Martins’ (CSM’s) Innovations Centre, London, in Spring 2007 which ran for just under a month. The experiment was part of a two-year Technology Strategy Board (TSB) funded project called ‘Bop’ and used a series of three distinct metrics covering: • behaviours, ie the extent to which certain activities were in evidence • feelings, ie how individual occupants felt about how well their needs or wants were met The main requirements for the devices used for this project were as follows: • small, low-cost sensor nodes that could be easily configured and deployed en masse • programming environment and language(s) that would allow non-sensor systems programmers to deploy their code easily • nodes that would accurately and efficiently provide data 24 hours a day, 7 days a week, to a database repository • nodes that could withstand the wear-and-tear of a creative office given that they would be retrofitted and, in some cases, free-standing devices The highly heterogeneous architecture used to achieve this consisted of four different hardware platforms using many different systems and languages. This study focuses on one platform, the Beastie, which used an architectural description language implementation called Tesserae to measure meeting room and social space activity. WSN Architecture Four different WSN platforms were deployed in the whole project to facilitate the variety of sensors used – Crossbow motes, Arduinos, Bengt Polls and Beasties. Data fusion for all the devices took place at a centralised MySQL server where each data logger would push data in order to provide a pseudo-synchronous view of the WSNs. To achieve the link between the base stations and the centralised database, bespoke software was developed to forward data to the database. • activities, ie the extent to which certain kinds of activities were being done 55 Light and proximity levels and their effect on state (sensor value vs time) 300 Ranger3 250 Light State 8 Bit Reading 200 150 100 Fig 4. Beastie node counting sensor 50 0 0 500 1000 1500 2000 2500 3000 3500 4000 Data Sequence Fig 2. Light and proximity levels and their effect on state (sensor value vs time) State changes in a space due to part-time work (sensor vs time) 30 Fig 5. Beastie node Light and proximity levels and their effect on state (sensor value vs time 25 300 8 Bit Reading State 20 Ranger3 250 15 Light State 200 10 150 5 0 100 State 0 5000 10000 15000 20000 25000 30000 Sequence 50 Fig 3. State changes in a space due to part-time work (sensor valve vs time) State changes in a meeting space (sensor vs time) 0 30 0 500 1000 1500 2000 2500 3000 Fig 6. Beastie node in corridor 3500 4000 State meeting space activity detection system, were A web-based Representational State Transfer Data Sequence programmed using Tesserae. This allowed the (REST) 25 query interface was used to interact development of component-based systems to with the database and perform queries on provide better software engineering for WSNs. filtered data based on location variables. 20 Tesserae is highly lightweight and encourages Via the Bricks architecture, a user (or an the programmer to produce better-engineered application) can seamlessly query multiple 15 solutions by separating out the implementation sensors that are part of the same location or of a component from its composition within the the same location subgroup. Bricks will 10 architecture; the behaviour of components is automatically process the values from all the defined through the use of interfaces. sensors that logged data, hiding the complexity 5 of the underlying WSNs and providing a State Measuring room states on the Beastie platform persistent interface for the end-user. Each Beastie was placed in a social meeting 0 0 5000 10000 15000 25000 30000 space and 20000 contained three sensors–temperature, Beastie hardware and software light level and multiecho ultrasound. The latter The Beasties are simple, wireless, microcontrollerSequence produced a ping signal which recorded the first devices designed for prototyping wireless sensor 16 echoes. Through initial calibration of each node setups. Each Beastie had an 8-bit micro space, we found that the first, third and eighth controller, low-power digital radio (433MHz), a echo readings were adequate to represent a power supply and an expansion bus connector. distance covering activity within near, middle The Beasties communicated using an autonomic and far-off ranges, respectively. These data network, which was self-configuring, selfwere combined with the temperature and optimising (device usage was automatically kept light readings and placed into a feature map at an optimum level) and self-repairing (the which was processed and stored on each network adapted to take account of new devices individual Beastie. as they were added to or removed from the system). Fig 1. shows the Beastie platform. These data derived a number of states for each Applications, both ambient sensing and the space. For example, dark and empty, bright and empty, hot bright and someone there, bright and empty, etc. Using the meeting schedule for the 56 more formal spaces allowed us to interpret the results from the feature map. This in turn allowed us to follow state changes in space usage such as night-time, cleaner entering a space and vacuuming, space in use, not in use, etc. and night-time again. Fitting the SOM to the Beastie The ranger algorithm used on the Beasties in the CSM trial was a modified self-organising map similar to a Kohonen network. An initially random set of points was placed in an n-dimensional space, which was moved towards each sensor reading as it was plotted into the space. The map used in this trial had five 8-bit dimensions, one for each of the sensor readings taken at one time, and 32 points or ‘states’. The map was stored in electrically erasable programmable read only memory so that it would survive battery changes. Every minute, the Beastie would ‘wake up’ and take light, temperature and three ultrasound measurements (the first, third and eighth echoes from a single ping) to produce five variables to be fed into the map. Some modifications were made to the algorithm both to enable its deployment on low-powered devices but also to avoid excessive tuning and overtraining. As we were expecting to see multiple states, our self-organising map 0 0 5000 10000 15000 20000 25000 30000 Beasties in the creative workplace Sequence State changes in a meeting space (sensor vs time) 30 25 State 20 15 10 5 State 0 0 5000 10000 15000 20000 25000 30000 Sequence Fig 7. State changes in a meeting space (sensor vs time) algorithm was designed to not converge to a single state or even to an optimal number of states, but to make use of the 32 initial states for the period of the trial. Thus only the single point in the map nearest the sensor reading, ie the ‘winner’, was moved. For simplicity, the winner was always moved an absolute distance closer to the reading rather than a percentage closer. Results and discussion Over the four-week trial, more than 1m sensor readings were gathered, processed in situ and wirelessly forwarded to the base station. An example of a daily trace of data is shown in Fig 2. For clarity, only two of the five sensor readings are shown along with state changes. The response times of the ultrasound and light levels appear to have an important effect, as would be expected. Although day and night are expressed as states there also appear to be substrates within these, which could be attributed to out-of-hours working and gatherings of people. Looking over a period of four weeks, the changes in states offer a good indication of changes in usage of various work spaces. Fig 3. shows the part-time weekly work practice of a member of staff and Fig 7. shows how a meeting space was used differently over time. Further analysis of data-to-state relationships is beyond the scope of this paper but it is apparent that the devices deployed were effective in gathering and analysing a large quantity of diverse data. Evaluation and requirements The main adaptations that had to be made to the algorithms were to restrict them to integer maths and simple functions (no square root for calculating distances between points, for example) and to limit the amount of memory they used (the feature map had to fit in the 1K RAM of the Beastie). Although the devices worked well to identify a range of different activity levels within a space, there is evidence that within the crowded office environment, the sensitivity was limited to picking up coarse changes. Improvement may have been possible by further hand-tuning the placement of the device and the echoes used as Fig 8. Ultrasound used to analyse occupancy sensor inputs. However, in the spirit of a deployment that is self-optimising, it would be better to implement an algorithm to enable the devices to choose the echoes themselves, based on the maximum variability of readings, during a training period. Greater sophistication could also be achieved by subtracting persistent environmental reflections from the data points recorded, so that only changes in the environment are registered. In the open spaces of the foyer and corridor, the background-reflected sound signals had a simple profile, so changes to that profile, due to human activity, stood out clearly. Within the crowded office environment, however, complexreflected sound signals were generated because of office clutter, which masked changes caused by physical activity. This did not give erroneous results, it merely restricted the sensitivity of the device, resulting in a more limited number of states being observed. Conclusion and next steps Two common themes ran through the experiences obtained by the developers on the project: • the hardware was fiddly and required more than a basic knowledge of electronics to configure • the software was immature and not suitable for a non-computing programmer (even programmers had problems deploying the systems) These observations reflect the fact that the main component developers in the Bop project are not computer scientists/programmers but technologists, designers and system integrators. This is compounded by the fact that unlike general server or PC-based technologies, WSN technologies are relatively immature, stemming from university and corporate research laboratories and have been designed for experimental rather than mass-market usage. database technologies, to allow the less experienced developer to easily be creative. However, this does highlight a situation that the WSN community has to face and is attempting to address. Whilst the initial trend in selling the advantages of WSN focused on ease of deployment (ARC report a 10% cost of a wired design) the current focus of research has been driven by the desire to use WSN to support energy saving activity. Consequently, there is a body of research focused on systems that trade-off energy consumption against user preferences and an increasing trend for Green IT surveys. Acknowledgements This paper has been adapted from a longer article published with Julie McCann and Asher Hoskins (Imperial College) and Chris Roadknight and Jane Tateson (BT) in Intelligent Buildings International 1, 2009, 222-229. This work was supported in by the UK Technology Strategy Board. References Bordass, W., Leaman, A. and Eley, J., A Guide to Feedback and Post-Occupancy Evaluation, York, Usable Buildings Trust, 2006. Kohonen, T., Self-Organizing Maps, 3rd ed, Berlin, Springer, 2000 McCann, J.A., Huebscher, M. and Hoskins, A., Context as autonomic intelligence in a ubiquitous computing environment, International Journal of Internet Protocol Technology (IJIPT) special edition on Autonomic Computing, Inderscience 2007 2 (1), 2001, 30-39 Qiao, B., Liu, K. and Guy, C., A multi-agent system for building control, in Proceedings of the IEEE/WIC/ACM International Conference on Intelligent Agent Technology (IAT’06). Hong Kong, December 2006, 653-659. Roberts, S., An integrated framework to improve the workplace, Facilities Management January, 2006, 10-12. Yet many have been marketed en masse, even though they do not have the development tool-sets, robustness, ease-of-use, interface and integration, and general maturity, as do general 57 Neighbourhood Pedestrian Analysis Tool (NPAT) Authors: Varanesh Singh, Eric Rivers, Carla Jaynes 58 There is a need to perform pedestrian analysis at a broad level with the goal of identifying and prioritising pedestrian improvements. At present there are limited solutions by which to collect and analyse this data. Abstract This research looks at the development of a Neighbourhood-level Pedestrian Analysis Tool (NPAT) that addresses practitioners’ concerns and provides agencies a means by which to analyse a large scale pedestrian environment in a user-friendly, graphical environment while leveraging their existing data collection methods and information sets. Using the Geographic Information Systems (GIS) platform as the basis for the NPAT, the design team identified key inputs and parameters that must be incorporated into the tool. With this information in hand, the design team evaluated two methods to perform the analysis: a polygon method where pedestrian space is represented as shapes, and a link/node method where pedestrian space is represented as a series of links and nodes, similar to typical vehicular capacity analyses. The design team selected the polygon method because it allows for easier analysis and better represents pedestrian movement and results. This method was then applied to a small test case with favourable results. The design team also coordinated with the New York City Department of Transportation and received a favourable response including the potential for future collaboration. The research shows that there is a feasible method to perform the broad-brush analysis of pedestrian spaces. There are some areas requiring further development which will be addressed in future phases of research and design. Fig 1. The pedestrian environment Introduction Many cities face the challenge of providing adequate pedestrian amenities for its residents and visitors among the other competing requirements within the public right of way. With congestion becoming more common for all modes of travel, cities need a way to understand pedestrian-specific congestion on a large scale in order to balance and prioritise pedestrian space improvements and make the best use of limited funds and resources. Currently, the accepted criterion used by agencies to evaluate performance of pedestrian space is Level of Service (LOS). Existing methodologies and standards include those established by the Highway Capacity Manual (HCM) which is based upon Fruin’s thresholds for density, flow and delay. The processes within the HCM require analysis at the intersection level, which can be onerous in terms of inputs and time if analysis is sought at the corridor or neighbourhood level. Acknowledging the numerous additional factors that contribute to a pedestrian environment, cities and agencies are starting to look towards a different type of measurement, Quality of Service (QOS). QOS measures the quality of the pedestrian experience as people move through a space, including characteristics of the built and natural environments and factors such as safety, comfort of design, and land use. Despite abundant existing data, there are currently limited means to perform pedestrian analysis at a broad-brush level. Large scale micro-simulation models, require extensive data and can be expensive to build and calibrate. While less expensive spreadsheet analysis can be performed in a shorter time frame, this type of analysis is often piecemeal and does not provide graphical representation of the results. Given these limitations, there is a need for a tool that can economically and holistically perform pedestrian analysis. This research looks at the development of a Neighbourhood-level Pedestrian Analysis Tool (NPAT) that addresses the concerns noted above. it also provides agencies and practitioners a means by which to analyse a large scale pedestrian environment in a user-friendly, graphical environment while leveraging their existing data collection methods and information sets. This tool will ultimately allow practitioners to pinpoint and prioritise those sidewalks or crosswalks within a neighbourhood in need of attention and ensure that walking remains the most attractive mode of transportation. Methodology The research presented below represents the first of a three phase process. The methodology of phase 1 was to evaluate various technical processes and select a preferred means for the NPAT. The resulting technical process was then applied to a case study to determine if the process and outputs were adequate and warranted further action. Future phases will include the automation and refinement of many of the processes identified in this phase. Prior to conducting the core research, the design team identified parameters which the NPAT must incorporate, the first parameter being the platform for analysis. The design team selected Geographic Information Systems (GIS) for this phase of research because of its accessibility for major cities, as well as the fact that most geographic data is stored in GIS. The final parameters identified are key inputs and outputs which are described in detail below. Inputs The required inputs relate to pedestrian space and its characteristics as well as the characteristics of the pedestrians and how they move through the space. 59 A D B E C F Fig 2. Visual representation of pedestrian walkway LOS (adapted from HCM) Defining the walkable limits of a space is important in establishing the area of analysis. For this tool, four distinct walkable regions for pedestrian analysis have been identified as important for determining volume, speed, and density. These regions contain unique characteristics that require they are treated differently. They are summarised below: • sidewalks are defined as any space that is primarily pedestrian and lacks modal conflicts • crosswalks consist of road space allocated for pedestrians crossing from one corner to another • corners consist of the portion of the sidewalk allocated to crossing behaviours such as queuing, changing directions or waiting • areas of mixing occur where perpendicular flows of traffic meet. These include areas such as building entrances or subway exits The pedestrian space for analysis must also take into account obstacles such as lamp poles, fire hydrants, newspaper boxes, trees etc which affect the way in which pedestrians utilise a space. For the purposes of analysis, obstacles are typically accounted for by subtracting out the obstacle’s area plus a standard buffer distance around the obstacle. Beyond the geometry of the space, it is also important to consider the following inputs: • character of the pedestrian space: this is most critical when conducting a QOS analysis; characteristics of the built environment such as the type of adjacent roadway, adjacent land use, and presence of street furnishings all affect the QOS • aspects of the natural environment such as wind, shade and sun effects are also important in evaluating the space • characteristics of the pedestrians: this is an important component as it impacts both LOS and QOS calculations; these characteristics include age, walking speed, trip purpose and relevant cultural or demographic information 60 • pedestrian demand data: this is the most critical piece of information and dictates how many pedestrians move through the space within a given period; this is typically collected in 15 minute intervals, but the intervals can range between 5 minutes and 1 hour Outputs The outputs are the quantitative results that will be used by practitioners to evaluate the performance of the pedestrian space. LOS is the primary output of evaluation for this phase of the research and consists of a range of pedestrian densities and volumes categorised into a scale of A through F as demonstrated in Fig 2. Other outputs could include conflict analysis and QOS results. While QOS outputs are still being developed and refined by the pedestrian planning community, they can easily be incorporated into the NPAT at a later date. Results and discussion With the key inputs and outputs defined, the design team evaluated methods for achieving the analysis within the GIS environment. Two separate processes were evaluated: one process using polygons and another using a link/node system. Polygons would represent the actual boundaries of the space and would visually represent the walkable area including any space removed to account for obstacles. A link and node system would represent the spaces schematically. A node would represent a corner or area of mixing and each link would represent the sidewalk or crosswalk. While the link/node system typically used in vehicular analysis, the design team deemed the polygon method to be the more suitable process because pedestrian LOS, delay and quality is so dependent on the physical definition of the space rather than the number of lanes. The resulting polygon process was broken down into four modules to collect the inputs and produce outputs. The details of each module are described below. Process of defining pedestrian space Curbline + building footprint Divide polygon Building area Building area Sidewalk Divided sidewalk Fig 3. Process of defining pedestrian space Walkable space module The module for identifying the walkable space for pedestrians creates a new GIS shapefile of the area between building lines and curblines using the “Intersect” function. This newly created shapefile then takes into account obstacles by creating a buffer around each obstacle and subtracting it out of the walkable area using the same “Intersect” function. The design team found that there are limitations to this type of buffer analysis in heavily used pedestrian areas, including the potential to arrive at a negative amount of pedestrian space. Furthermore, in congested situations, pedestrians can accept lower buffer distances. These challenges will be reviewed further in the next phase of work. Polygons representing crosswalks are also created as part of the pedestrian space boundary. These must be manually drawn unless cities have this information stored in graphical format. For the purposes of this research, crosswalks will be addressed in detail in the next phase of work. The resulting walkable space is then further subdivided into smaller analytical areas that allow for more detailed pedestrian analysis. The analysis regions are determined based on a critical width of each of the polygons in the pedestrian space boundary. The resulting shapefile consists of manageable areas that can be categorised as sidewalks, crosswalks, corners or areas of mixing. The process of producing this shapefile involves creating a new line file that specifies where the pedestrian space boundary polygon will be cut. Based on these lines, the polygons are subdivided using the “split polygons by polylines” feature in the topology tool. Fig 3. shows the process of defining pedestrian space. et 101 Ch u rc hS t re 402 lto nS t re dw 405 404 501 Thru (Dest1) To (Dest2) Demand 104 103 102 53 102 103 104 78 103 501 404 105 404 501 103 211 103 502 201 175 201 502 103 93 403 404 - 257 404 405 - 75 103 104 Br oa 102 403 et ay Fu From (Orig) Neighbourhood pedestrian analysis Tool (NPAT) 401 502 504 LOS A 301 507 201 B D E F Fig 4. Graphical display of results Characteristics module This module allows the user to incorporate inputs that pertain to characteristics of space and pedestrians. The majority of these inputs will be used to produce a QOS evaluation. But some of the information, such as those related to pedestrian speed and trip purpose can be used to adjust the LOS outputs. Information on the characteristics of the space and pedestrians are based on geographic conditions and can be spatially joined to each pedestrian analysis region. For example, pedestrian speed is related to the land use and will tend to be higher in business districts. Demand module Pedestrian demand data input into the tool will be stored in tabular format and then spatially joined to the shapefiles representing the available pedestrian space. Using a path system, the pedestrian demand data is linked to polygons based on the start polygon, through polygon, and end polygon of each movement. One of the benefits of using the path system is that it stores data in a format that can later be used to estimate an Origin-Destination (O-D) matrix for a given area. This O-D matrix can be used to get a better understanding of travel patterns, help identify deficiencies in a network and provide a key input for micro-simulation models. The ability to estimate an O-D matrix requires a sufficient amount of data and an algorithm that can process the data intelligently. As a result, this process has been identified as a longer term goal. Results module Outputs will be generated through scripts which perform the LOS calculations and then associate the final result with the corresponding shape. The shape can then be symbolised or colourcoded to graphically present findings based on established performance measures. For the first phase of the research, the results are calculated separately and then joined to the shapefiles. Future phases of the research will include processes to automate this step of the analysis. Fig 5. Demand module interface Case test and follow-up A case test of the described process was performed on a small area of Lower Manhattan in New York City. This location was selected because of Arup’s recent work with New York City Department of Transportation (NYCDOT) in this area and the availability of existing data. The case test resulted in a colour-coordinated map of polygons shaded to correspond to LOS values. The results generally corresponded to the LOS observed on site. Fig 4. shows a graphical representation of the results. The results of the case study showed that significant manual work is required to generate the pedestrian space. Shapefiles have to be adjusted and field verified before adequate pedestrian space shapefiles can be created. Future phases of the model must address ways to improve this process. Arup met with staff of the NYCDOT to present the idea of the research. Based on the information discussed, NYCDOT is interested in the potential of the NPAT and supports Arup’s pursuit in further developing the tool. There is interest in using the city as wider-area case test, to be included in phase 2 of the research study. NYCDOT and Arup have agreed to meet and further discuss the potential of this tool. NYCDOT have also recommended a larger and diverse neighbourhood for use to be included in phase 2 of the research. They have also agreed to provide data and other resources to assist in this wider-area case study. Conclusions and next steps The research demonstrated the feasibility of creating an affordable tool to effectively analyse pedestrian activity at a mesoscopic scale. However, further work is required to automate several key tasks and to incorporate additional evaluation measures. In addition to developing a proof of concept, the team has identified a potential partner in the NYCDOT who has the need and interest for such a tool. Further refinement of the NPAT and collaboration with NYCDOT could yield a valuable analytical tool for agencies in North America and around the world. A goal of future phases is to determine methods of automating a majority of the processes required for the analysis. The automation will reduce the amount of manual input, thus making the process more user-friendly and providing an overall more efficient and effective tool. Future phases will involve developing more rigorous methods to validate the results to existing conditions, a process which is important in determining the legitimacy and strength of the tool. Further coordination with NYCDOT is also identified including a wider area test study in a larger more diverse area of New York City. Acknowledgements We would like to thank New York City Department of Transportation for providing input and feedback before and during the research. References City of New York, City Environmental Quality Review Technical Manual, 1 st ed., City of New York, 2001 Fruin, J.J., Pedestrian Planning and Design, 2 nd ed., Elevator World, Inc., 1987 San Francisco Department of Public Health, Pedestrian Environmental Quality Index, version 1.1, Available online [http://www.sfphes.org/ HIA_Tools/PEQI_Methods_2008.pdf], 2008 Transportation Research Board National Research Council, Highway Capacity Manual, 4 th ed., National Academy of Sciences, 2000 61 Human induced vibrations on footbridges Authors: Iemke Roos, Peter Burnton 62 The past few decades has seen demand for better quality pedestrian and cycleway facilities. Together with developments in materials, this has led to the design of longer, more complex and slender footbridges that can be more sensitive to dynamic forces. Abstract Considerable public money is spent on footbridges and the bridges are expected to offer comfortable passage to the public. These bridges can be more sensitive to dynamic forces brought on by pedestrians, resulting in vibrations of the bridge deck and possibly affecting the overall ‘comfort factor’ of the bridge. Older, simplified design rules, often based around the movement of a single pedestrian, are considered to be no longer adequate. This study explores the natural frequency, damping, pedestrian load models and the public human response to footbridges. The objective is to compare several load models described in current codes of practice (British, European and Australian) intended for practical engineering application. To validate the load models, the computer generated responses are compared to the real behaviour of two bridges, the Goodwill Bridge and the Milton Road Bridge, both located in Brisbane and designed by Arup. This article focuses on vibrations in the vertical plane, although horizontal or Syncronous Lateral Excitation (SLE) effects were also considered. Proposal annex C UK national annex Australian standard Single pedestrians (walking) 3 3 3 roup of Pedestrians G (walking) 3 3 7 Takes into account: Joggers 7 3 7 Crowd 3 3 7 Non-moving harmonic Moving and non-moving harmonic Moving harmonic Loading Time Until Steady State Depending on velocity Depending on velocity Load Frequency Natural Frequency Natural Frequency Between 1.75Hz and 2.5Hz Load model characteristics: Load Dynamic load factor dependent on: Group size 3 3 Not mentioned Natural Frequency 3 3 Not mentioned Degree of synchronisation between pedestrians 3 3 Not mentioned Application conditions: Vertical fv < 5 Hz fv < 8Hz 1.5Hz < fv < 3.5Hz Table 1. Comparison of codes Introduction Anyone who has walked over a bridge has probably felt or seen small movements of the deck, moving up and down, caused by traffic, pedestrians or even wind. These vibrations are usually small and only perceptible with a static reference point or when standing still on the bridge. The magnitude of these movements depends on many factors: length of the bridge, stiffness of the bridge, load type, load magnitude, load position, sensitivity of the observer and many more. To fully understand the response of the bridge it is essential to model the loads correctly. Pedestrian loads are difficult to model because of the unrelated variables such as: weight of the pedestrian, walk velocity, number of pedestrians, distribution of the pedestrians over the bridge, etc. The centre of gravity of the human body is located at about 55% of its height and makes a sinusoidal motion during walking, both in vertical and horizontal directions. The force thus has three components: a vertical, a longitudinal and a lateral component. The vertical component is up to 40% of the body weight. The other components are considerably smaller. Walking, running or jumping each produce a different loading curve over time and different frequencies for the load pulse. The vertical force component during walking shows a characteristic double hump, which is the result of the impact of the heel on the ground followed by the push-off force. The force magnitude tends to increase with increasing step frequency. The footfall force envelopes overlap as in walking both feet are briefly on the ground at the same time, Fig 2. Both feet can be off the ground at the same moment when running. Synchronisation of pedestrians is more likely to occur at higher pedestrian densities, when people are not able to walk freely. At a density of 1.0 person per m² the freedom of movement is greatly inhibited. When the density reaches about 1.5 persons per m², walking becomes difficult and pedestrians are dependant on the pace and direction of other bridge users. The pedestrian velocity decreases as the density increases and consequently the dynamic forces on the bridge decreases. Synchronisation between runners is less likely to occur, as the velocity is quite high and thus the density is lower. Typically people do not adjust their stride to the vertical movement of a bridge and hence synchronisation between pedestrians and the natural movement of the bridge structure is not significant for the assessment of vertical vibration. Perception of vibrations by pedestrian bridge users is subjective and psychology is an important variable. Each person considers an uncomfortable vibration differently, depending on the environment, activities around them, type of bridge, what they are doing, cultural influences, age, etc. This study has used Goodwill Bridge and Molton Road Bridge, Brisbane to validate the load behaviour model. 63 Methodology Goodwill Bridge The Goodwill Bridge, Fig 1. is a bridge for pedestrians and cyclists that spans over the Brisbane River. It links the southern part of the Central Business District (CBD) of Brisbane with South Bank which offers many public attractions as well as a railway station and bus service. It was opened in 2001 and is used by approximately 40,000 people per week. The 450m long footbridge has three distinct parts: the Rampart on the South Bank riverside, the Arch as the main span over the river and the steel girder spans on the CBD approaches. The Arch spans 102m and provides 13m navigations clearance above the tidal water level. Users are a mix of commuters and recreational users. Under normal circumstances, small vibrations can be perceived when standing still on the main span of the bridge. These vibrations are not usually felt when walking on the bridge. Larger vibrations in the main span have been noticed when groups of joggers use the bridge. These vibrations can be felt by people standing, walking or jogging, but have not been reported as uncomfortable. The vibrations at Goodwill Bridge had been measured at three strategic locations on the deck during a community marathon event. Measurements were also taken during normal use of the bridge. This data was made available for this study. Milton Road Bridge The Milton Road Bridge, Fig 2. was built to link the 52,500 seat Suncorp Stadium with the Milton Train Station. The steel truss bridge is largely used by people going to, or coming from, the stadium in large groups. The 86m long and 8.45m wide twin span bridge spans over Milton Road with a 6.5m vertical clearance. Vibrations have been noticed near the midspan of the longer span of the bridge by people standing still but have not been reported as uncomfortable. Vibrations are best perceived when one or two pedestrians are crossing the bridge and at a location where an adjacent tree provides a static reference relative to the bridge movement. This bridge is essentially used by large crowds moving between the stadium and the train station. Therefore, people do not usually stop on the bridge. The codes The load models considered for this study have been selected from the European, British and Australian codes of practice. Proposal Annex C for Eurocode 1 In 2001 a Proposal Annex was issued for Eurocode. This Annex has not been officially approved but issued as guidance for designers. Annex C gives guidance on determination of the natural frequencies, structural damping and dynamic load models. British National Annex for Eurocode 1 of EN 1991-2 The aim of the UK National Annex is to provide sufficient guidance to account for the effects of vibration of complicated structures and those in 64 CBD Brisbane river South Bank The Arch The Pier Goodwill Bridge Direction water flow The Rampart Fig 1. View of Goodwill Bridge, Brisbane Single pedestrian Group of pedestrians Joggers Crowd Proposal Annex C UK National Annex Australian Standard a max = 0.184m/s² a max = 0.015m/s² a max = 0.015m/s² u max = 1.25mm u max = 0.09mm u max = 0.09mm Milton Road Bridge a max = 0.360m/s² a max = 0.110m/s² a max = 0.005m/s² u max = 0.52mm u max = 0.16mm u max = 0.02mm Goodwill Bridge a max = 0.552m/s² a max = 0.027m/s² u max = 3.76mm u max = 0.16mm Milton Road Bridge a max = 0.257m/s² a max = 0.216m/s² Goodwill Bridge u max = 0.37mm u max = 0.32mm Goodwill Bridge - a max = 0.182m/s² - u max = 0.70mm Milton Road Bridge - a max = 2.202m/s² Goodwill Bridge Milton Road Bridge - u max = 0.29mm a max = 4.116m/s² a max = 2.087m/s² u max = 28.16mm u max = 14.16mm a max = 1.520m/s² a max = 0.456m/s² u max = 3.68mm u max = 0.66mm Table 2. Values for maximum acceleration and displacement for each bridge and pedestrian load type (a max = maximum acceleration, u max = maximum displacement) sensitive locations, without imposing undue conservatism that might constrain designers in achieving an economic solution. Australian Standard AS5100.2 AS 5100-2004 is the Australian Standard for Bridge Design. Clause 12.4 of Part 2 deals with vibration of pedestrian bridges. This clause is similar to that in earlier Australian Bridge Design Codes and is representative of earlier Codes in the UK where the structural response to a single pedestrian is used for the compliance criteria. Table 1. provides a comparison of key aspects of the approach in each Code. In addition there is variation in the specific loading and acceptable comfort criteria in each Code. Results and discussion The Goodwill Bridge and the Milton Road Bridge were analysed according to the three Codes discussed above. The results of these analyses are then compared to the measured and reported performance of the two bridges. The maximum values from the analyses are summarised in Table 2. The table contains both the maximum acceleration a max and the maximum displacement u max for each bridge and pedestrian load type. The results satisfy the compliance requirements stated in the Codes, except those generated with crowd load cases. Both Annex C and UK Annex appear to overstate the influence of crowds compared to the observed experience at the two bridges. This over estimation could be explained by the fact that the pedestrian density is not taken into account in the amplitude of the dynamic forces of pedestrians. Pedestrians tend to walk slower in high density situations and as a consequence produce smaller dynamic forces. Crowd loads are also applied as a point rather than a patch load which will increase the analysis result. The degree of pedestrian synchronisation may also be overstated for the case of the two bridges. Human induced vibrations on footbridges To Milton train station Vibrations Load Load View along the bridge (looking to the Stadium) Walking Running Ti m e Le To Suncorp Stadium Ri ft gh fo tf Ti ot oo m Le e Ri ft gh fo tf t ot oo t Fig 3. Patterns of running and walking forces Acknowledgements This study is entirely based on the M.Sc. Thesis undertaken by Iemke Roos, Delft University of Technology while he was embedded in the Arup Bridge Design team in Brisbane. We thank Iemke and the Academic Staff in the Faculty of Civil Engineering and Geosciences at Delft University of Technology. Connection trusses to the main chords Mid-support columns View under the bridge Fig 2. The Milton Road Bridge Proposal Annex C calculations produce accelerations that are too large compared to those observed on the two bridges. However UK National Annex and Australian Standard (in the case of a single pedestrian), generate accelerations and displacements that are slightly lower than observed with the exception of the load case representing a crowd. Results generated with the UK National Annex are smaller than those calculated with Proposal Annex C. This is mainly because moving loads are used in the UK Annex meaning that the load is not located at the point of maximum influence for more than a single stride. Moving harmonic loads appear to best represent pedestrian loads however they are more complex to both model and interpret. The static load alone does not explain the magnitude of the differences between the two codes. The dynamic load factor is thought to be a further contributor to this difference. When considering the load cases representing a single pedestrian and a group of pedestrians on the Milton Road Bridge, the responses calculated with Proposal Annex C for a single pedestrian are three times higher than the ones calculated with the UK National Annex. However, there is nearly no difference in the case of a group of pedestrians. The Milton Road Bridge has a relative high natural frequency which influences the dynamic load factor of Proposal Annex C considerably. The dynamic load factor in the UK National Annex is more complicated. Different parameters are used to assess the dynamic load factor such as the natural frequency and degree of synchronisation. The number of pedestrians has a considerable influence and is the key reason why the dynamic load factor is higher for a group of pedestrians compared to a single pedestrian. The acceptance criteria for bridge vibration given in the Codes do not vary with differing bridge form, use, or location. The acceptance criteria are different in each of the Codes but do not vary for the different bridges when any one Code is considered. One can foresee a situation where the lack of flexibility in the acceptance criteria could have a significant impact on the design outcome. Conclusions and next steps Of the three codes considered, the UK Annex offers the best representation of pedestrian induced vertical vibration. Loading for most load types are possibly a little understated. The representation of crowd load effects is significantly overstated in both codes considered. Moving loads offer a better load representation than static loads but are more complex to analyse and interpret. The Annex C approach overstates vertical vibration effects for all pedestrian load types given in the code. The Australian code only considers a single pedestrian and may prove to be inadequate for some bridges. References H. Bachmann, W.J. Ammann, F. Deischl, J. Eisenmann, I. Floegl, G.H. Hirsch, G.K. Klein, G.J. Lande, O. Mahrenholtz, H.G. Natke, H. Nussbaumer, A.J. Pretlove, J.H. Rainer, E. Saemann, L. Steinbeisser, Vibration Problems in Structures: Practical Guidelines, 1995 D.R. Leonard, Human Tolerance Levels for bridge Vibrations, Ministry of Transport RRl Report No. 34, Road Research Laboratory, Harmondsworth, 1966 J.W. Smith, The vibration of Highway Bridges and the effects on human comfort, Ph.D. Thesis, University of Bristol, September 1969 C. Barker, S. DeNeumann, D. MacKenzie, R. Ko, Footbridge Vibration Limits – Part 1: Pedestrian Input, Footbridge 2005 International Conference D. MacKenzie, C. Barker, N. McFadyen, B. Allison, Footbridge Vibration Limits – Part 2: Pedestrian Input, Footbridge 2005 International Conference C. Barker, D. MacKenzie, Design Methodology for Pedestrian induced Footbridge Vibrations, Footbridge 2008 International Conference fib Bulletin No. 32, Guidelines for the design of footbridges, 2005 J. Blanchard, B.L. Davies, J.W. Smith, Design Criteria and Analysis for Dynamic Loading of Footbridges, Symposium on Dynamic Behaviour of Bridges, 1977 J. Blaauwendraad, CT2022 Dynamica van Systemen, TU Delft, 2006 A. Romeijn, CT5125 Steel Bridges, part 1, TU Delft, 2006 The next steps in this study could include the consideration of flexibility in the acceptance criteria given in the Codes; improved understanding of crowd loading within the context of bridge vibration analysis; and improvement to the way moving dynamic loads are modelled in structural analysis software. 65 The importance of research at Arup The Arup Research Review is one of our more recent publications. It presents some of the work which has been conducted by our Arup colleagues at the cutting edge of our industry over the past year. It captures one of the most important principles of Arup’s ethos, that being we bring something special to what we do, wherever possible. Client value can be added in many ways, one of the most fundamental contributions from our firm has been in the area of ‘breaking the mould’. We like to make things possible that have previously been considered impossible. The Research Review demonstrates that this tradition is still alive and well in Arup today. Not all research ends up with immediate application in areas that bring direct benefit to our clients. That is the nature of cutting edge investigation. However, high quality research combined with creative thinking and innovative design, means that unconventional solutions to the problems brought to us by our clients are always a possibility. That, in turn, means the possibility of step-change and consequent advantage, is always present. This in a nutshell, is the bedrock of all that we do. There are many examples we could cite since the firm was founded, ranging from the creative use of concrete, through geotechnics, fire engineering and advanced computer simulation. All these contributions were made possible because the firm has employed members of staff capable of working in the commercial environment at the leading edge of research and development. 66 John Miles Head of Energy, Resources and Industry Market Research Review | May 2010 Editors: Geraldine Ralph, Anna Goswell Designer: Terry Nicholls Published by:Arup, 13 Fitzroy Street, London, WIT 4BQ Printed by: Fulmar Colour ulmar print using alcohol-free printing F methods and vegetable-based inks were used throughout. Solvents and waste used in the printing process are recycled. Fulmar Colour Printing is a CarbonNeutral Company, and is certified to ISO 14001, and is a FSC and PEFC accredited company. ISBN number: 978-0-9516602-8-7 Research Team Regional Research Champions Arup Prof. Jeremy Watson Americas 13 Fitzroy Street Director, Global Research Tim Keer London W1T 4BQ T +44 (0)20 7755 2235 E [email protected] E [email protected] Dr Marta Fernandez Richard Hough T +44 (0)20 7636 1531 F +44 (0)20 7580 3924 W www.arup.com Research Relationships Manager E [email protected] T +44 (0)20 7755 5105 E [email protected] East Asia Dr Jennifer Schooling Australasia Dr Ricky Tsui E [email protected] Research Business Manager Europe T +44 (0)20 7755 2912 E [email protected] Dr Mikkel Kragh Geraldine Ralph UK-MEA Gavin Davies Project Coordinator T +44 (0)20 7755 6433 E [email protected] E [email protected] E [email protected]