D. A. KELLY, L. B. JACK

Transcription

1 E2 - Steps towards a unified design methodology for rainwater drainage systems D. A. KELLY, L. B. JACK Abstract The performance of rainwater drainage systems is coming under ever increasing pressure due to factors such as urbanisation and climate change. With the impact of both expected to increase in the future, system capacity will become more and more unable to drain the anticipated increased volume of rainwater, leading to increased damage and disruption due to flooding. Resilience to these inevitable changes can only be achieved if current design methods are updated to enable a whole system design approach. This paper assesses the performance of the rainwater drainage system of a UK-based case study building through application of a holistic numerical simulation model. To establish a baseline of system performance, design rainfall events were derived from current design standards and used to drive the model. Differences between the values of event duration, return period and profile shape as recommended by the separate design standards are highlighted and discussed with the aim of determining appropriate criteria for whole system design. Keywords Property rainwater drainage system, urbanisation, climate change, design standards 1 Introduction Many areas within the UK have suffered severe flooding in recent years. Urbanisation has resulted in an increase of impermeable surfaces in towns and cities which generate increased volumes of rainwater runoff and larger peak flows than those generated by natural or pervious surfaces. Surface water flooding occurs during heavy rainfall events as surface areas become inundated with rainwater and drainage system design capacities are exceeded. It is estimated that 80,000 properties in the UK are currently at very high risk of surface water flooding causing an average of 270 million of damage each year (Evans et al. 2004). Increased urbanisation, combined with the additional pressures of climate change, which projections indicate could increase winter rainfall by up to 30% (Murphy et al., 2009), will together significantly increase surface water flood risk in the future. Evans et al. (2004) estimates that the number of properties at risk could increase to 400,000 by the 2080s, potentially causing billions of pounds of damages per year. The rainwater drainage systems that are designed to convey runoff and provide protection to properties against flooding include: (i) roof drainage systems (either conventional gravity or siphonic systems consisting of gutters, outlets and downpipes); (ii) surface area drainage (consisting of curbs, gullies, and permeable pavements); and (iii) underground drain and sewer systems (consisting of collection pipes and manholes). It is rare that these elements are designed together as a whole system. More typically, each is designed as an individual entity in accordance with separate design standards with an attempt to line-up the various outflows at the later part of the design. This disjointed method not only neglects the risk of flooding due to confluence of downstream flows, but also ignores any potential benefit of possible downstream flow attenuation. If rainfall events change as expected in response to climate change, the rainwater drainage systems that were built in the past or are currently being designed may not have adequate capacities in the future. There is, therefore, a need for an integrated design philosophy that combines the benefit of whole system flow interaction with the requirements to meet both current and future performance requirements. Such complex flow analysis is only possible through the application of numerical simulation models. In this paper the performance of the rainwater drainage system of a case study building in Edinburgh is assessed using a numerical simulation model developed previously at Heriot-Watt University. Input rainfall events are derived from two separate design standards (one concerned with the design of roof drainage systems, and the other with drain and sewer systems) in order to compare how the entire system responds to both sets of recommended design events and not just the individual system element for which they correspond. The impact of each event on the performance of the system is assessed against performance indicators to determine if the system is operating within operational limits. Different values of event duration and return period are investigated, as although current design is based on the selection of a single value of each (based on the site physicality and economic constraints), this assumes that future rainfall can be represented by guidance based on historic climate records. Analysis of future trends of extreme 308

2 SESSION E SUSTAINABLE CONSTRUCTION AND CLIMATE CHANGE 2011 Symposium CIB W062 Aveiro Portugal rainfall events found that the current 100 year event in London may become equivalent to the 25 year event by the 2080s (Sanderson, 2010). 2 Case study: Thomas Thompson House Thomas Thompson House was chosen for this case study as it is fairly typical of the out-of-town commercial developments that have become popular in the UK within the past five to ten years, Figure 1. Like many of this type of development, the building is positioned within a large catchment and has an extensive roof area which is drained by a dedicated siphonic roof drainage system. Previous research at Heriot-Watt University (Arthur and Swaffield, 2000; Arthur et al., 2005) used this same siphonic system to derive valuable empirical data used in the development of the numerical simulation model used within this paper. During the course of this current project, the system has now been monitored for over a year to collect data on how the system operates under current climate conditions. Within this time, the system was found to have failed seven times by overspilling of water from the gutter. These failures, however, have been attributed to the accumulation of debris around the gutter outlet, and not the effect of extreme rainfall events (Kelly et al., 2010). 2.1 Site description Thomas Thompson House, constructed in 1994 and located within a large industrial estate approximately 9 km west of the centre of Edinburgh, is served by separate foul and surface water drainage systems. For the purposes of this cases study, only the surface water drainage system will be investigated. The property catchment has a total area of 15,424m 2 which consists of roof areas of 2794m 2, impermeable surface areas (including roads, carparks and walkways) of 5091m 2, and permeable surface areas (including grass and planting) of 7539m 2. The breakdown of each area type, as measured from scaled drawings of the site, are listed in Table 1. Table 1: Thomas Thompson House surface areas Surface type Area (m 2 ) Percent of total area Impermeable Roof 2794 Road Carpark 1232 Walkways 1338 Permeable Grass Planting 1517 TOTAL The ratio of impermeable and permeable surfaces is almost equal with 51% of the site consisting of impermeable surfaces and 49% consisting of permeable surfaces. With over half of the property catchment covered in impermeable surfaces the amount of potential surface water run-off is significantly greater than that for the site in its natural state. Reliance is therefore focused on the adequate capacity and successful operation of the property drainage system to protect against extreme rainfall events. Fig. 1 Property boundary of Thomas Thompson House (Google Map Data 2011) 3 Numerical model Previous research at Heriot-Watt University (Arthur et al. 1999, Arthur et al and Wright et al. 2006b) developed a numerical simulation model, known as ROOFNET, which encompasses components for the simulation of both roof and local area drainage. 309

3 The roof drainage component simulates the flow of rainfall from the roof surface down to ground level. In addition to variable rainfall conditions, the model can also account for different roof geometries (i.e. area, slope and roughness), the effect of green roofs, different gutter shapes (i.e. half round, rectangular and trapezoidal), various longitudinal gutter slopes, and any selected pipe diameter. The model calculates the unsteady flow conditions within the gutter (including flow rates, water depths and velocities) as well as the flow conditions (both free surface and fullbore flow applicable to both conventional and siphonic system simulation) within the system pipework. The local area drainage component simulates the flow conditions, including any pipe surcharge, from system entry points (i.e. gullies and rainwater downpipes) to the boundary of the local underground drain system. Rainwater runoff from surrounding areas is calculated using a simple volumetric approach based on intensity of rainfall and area drained. The ROOFNET model allows the operation of the whole property rainwater drainage system to be assessed under current and future rainfall conditions. System failures (characterized by gutter overtopping, surcharging of local drainage pipework, and the subsequent flooding of surrounding areas) can be defined and quantified indicating areas of the system where intervention may be required. A more in-depth discussion of the model is provided elsewhere (Wright et al, 2006a). 3.1 Numerical model set up The ROOFNET numerical simulation model was used to investigate the response of the property drainage system to a range of different design rainfall events derived from separate design standards. Having eight individual siphonic roof drainage systems, the roof area was subdivided as appropriate to drain to a designated gutter and outlet. The gutter dimensions were set to 360mm wide and 200mm high. A Manning s roughness coefficient, n, of 0.02 was used for the gutter condition as this was found by Swaffield et al. (1999) to be representative of the hydraulic roughness in typical roof gutters. All pipework was set as 50 mm stainless steel, with a pipe roughness coefficient, k p surrounding ground surface area was divided into smaller contributing areas consisting of six permeable areas and fourteen impermeable areas. Each contributing area was connected directly to the underground drain network which was represented by 57 pipes ranging from 100 mm diameter upstream to 400 mm diameter downstream. For the drain pipes, k p was set to 1.0 mm as recommended in BS EN :2000. All system dimensions were determined from site measurement. As input to the model, each rainfall event was applied as values of intensity over time. Design rainfall intensities were derived by reference to the appropriate design standards. In the UK, roof drainage systems are designed in accordance with BS EN :2000, siphonic roof drainage systems in accordance with BS EN 8490:2007, and underground drain and sewer systems in accordance with BS EN 752: Deriving a design rainfall event from BS EN :2000 Although the roof drainage system at Thomas Thompson House is a siphonic system, BS EN 8490:2007 directs the designer to BS EN :2000 to determine an appropriate design rainfall intensity. In BS EN :2000, the design rainfall intensity depends on three factors: (i) the duration of the rainfall event; (ii) the return period of the event; and (iii) the geographical location of the property. When selecting an appropriate rainfall duration, it is recommended that this is equal to the time taken for rain falling on the uppermost part of the roof to flow down the roof and along the gutter to the outlet (known as the time of concentration). The characteristics of the roof, particularly size, pitch, material and the wetness at the start of the storm, all have an influence on this value. For design purposes, this gives the worst case scenario as the maximum rate of run-off is generated when the duration of the peak rainfall intensity is equal to the time of concentration. A duration of 2 minutes 1 is recommended, however, rainfall intensities can be estimated for durations between 1 and 10 minutes. To assess the implications of duration on system performance (which in the future may not correspond so well with the time of concentration) rainfall durations of 2 minutes, 3 minutes, 5 minutes and 10 minutes were selected., of 0.15 mm (as recommended in BS EN 8490:2007). The 1 Derived from flow calculations of typical roofs (May, 1996) and verified by field studies (Escarameia, 1998) 310

4 SESSION E SUSTAINABLE CONSTRUCTION AND CLIMATE CHANGE 2011 Symposium CIB W062 Aveiro Portugal Selection of an appropriate return period is primarily an economic decision, rather than a meteorological one. Longer return periods lead to systems with greater capacity, providing a higher standard of drainage, but at a higher cost. The Standard advises that it is impractical to design for very infrequent extreme storm, and instead system design must strike a balance between system cost and the frequency and consequence of failure. The return period is based on the selection of an accepted category of risk defined in terms of the probability of the design rainfall intensity being exceeded during the life of the property. However, as previously mentioned, values of return period cited in these current standards may have no relevance to such storms in the future. To assess the implications of different return periods on system performance, values of 50 years, 500 years and 10,000 years (regarded in the Standard as producing the maximum probable rainfall intensity) were selected. No return periods between those stated were tested as the Standard advises against this due to the inherent variability of rainfall. For example, the Standard recommends that if a return period is calculated which is greater than 50 years, then the designer should proceed up to the next return period and select a design rainfall intensity based on a 500 year return period, and so on. The resultant design event used for sizing a roof drainage system represents the most intense part of a longer storm in which the intensity varies continuously with time. The design event is provided as a maximum average rainfall intensity over the selected duration of event. For the purposes of simulation, the design event will be applied to the model as a constant rainfall intensity over a selected duration preceded and followed by negligible rainfall. As an example, Figure 2 shows the design rainfall event derived from BS EN :2000 for a 2 minute storm in Edinburgh with a return period of 50 years (referenced in the Standard as a 2min M50 event). This translates to a peak steady-state rainfall intensity of 100mm/h over the 2 minute duration. 5 Deriving a design rainfall event from BS EN 752:2008 In BS EN 752:2008, the designer is directed to the Wallingford Procedure (1981) to determine design rainfall intensities for catchments greater than 4000m 2. When using this method, the design rainfall intensity is, again, dependent upon defining the event duration (taken to be equal to the time of concentration, i.e. the shortest time in which the uppermost part of the catchment will contribute to the flow in the downstream drain), the return period, and the geographical location of the property. The Wallingford Procedure provides guidance on how to determine time of concentration for any catchment. This was calculated to be 10 minutes for Thomas Thompson House. However, to assess the implications of event duration on system performance, rainfall durations of 10 minutes, 15 minutes, 30 minutes and 60 minutes were selected. Fig. 2 Example of design rainfall derived for Thomas Thompson House from current design standards BS EN :2000 and BS EN 752:2008 Suggested return periods are given for different property types and for different levels of acceptable risk. A return period of 5 years is recommended as being appropriate for city-based commercial properties, however, to assess the implication of more extreme events on system performance, return periods of 10 years, 50 years and 100 years were also used. 311

5 The design rainfall intensity for a 10 minute storm event in Edinburgh with a return period of 5 years (referenced in the Standard as a M5-10min event) is 33mm/h. However, instead of applying this averaged intensity, the Wallingford Procedure recommends the use of the 50% summer profile as outlined in the Flood Studies Report (NERC, 1975) as the design storm profile for system design. The 50% summer profile is symmetrical and single-peaked. Its shape does not vary with event duration and it is considered invariant with location. Figure 2 shows the M5-10min event as converted to the 50% summer profile. 6 Definition of system failure To assess system performance, it was first necessary to define the conditions that would constitute a system failure condition: Gutter water depth with a gutter height of 200mm, the maximum possible depth of water in the gutter is limited to this height. Once the gutter water depth exceeds 200mm, ROOFNET calculates the volume of water overspilling from the gutter. Any gutter overspill will be regarded as a system failure condition. Surface area water depth BS EN 752:2008 provides suggested allowable depths of water for various types of paved areas. For pedestrian areas and car parks this is limited to a depth of 6mm. Any surface area water depth greater than 6mm will be regarded as a system failure condition. Drain surcharge surcharging of the drain or sewer system can potentially lead to flooding problems upstream due to increased backwater levels. Although, in reality, some minor pipe surcharging will not necessarily cause the system to fail, any surcharging will be regarded here as a system failure condition. 7 Model results Having input the details of the rainwater drainage system into the ROOFNET numerical simulation model, the performance of the whole system could be assessed in response to the range of rainfall events derived from current design standards. Table 2 provides a summary of the various system performance indicators for each rainfall event. 7.1 Comparison of rainfall events derived from BS EN :2000 and BS EN 752:2008 In line with observed records of rainfall patterns from around the world, the rainfall events derived from the two design standards show decreasing intensity with increasing duration and decreasing return period. Total rainfall, however, can be seen to increase with duration and return period despite the decrease in intensity. What is clear when comparing the design rainfall events from each standard is that those derived from BS EN :2000 are of a significantly shorter duration with a correspondingly higher intensity (based on concentration times of the roof) than those derived from BS EN 752:2008 (which are based on concentration times of the catchment). The following sections look at how the rainwater system as a whole operates in response to both sets of design rainfall events. 312

6 SESSION E SUSTAINABLE CONSTRUCTION AND CLIMATE CHANGE 2011 Symposium CIB W062 Aveiro Portugal Table 2: Model results of system performance in response to a range of design rainfall intensities derived from BS EN :2000 (upper table) and BS EN 752:2008 (lower table) 313

7 7.2 System response to events derived from BS EN :2000 Table 2 shows that the whole system operates effectively during all event durations for the 50 year return period. Capacity issues begin to occur, however, for events with a return period of 500 years where minor surcharging of the drain system occurs during all event durations and overspilling of the gutter occurs for both the 5 and 10 minute events. Unsurprisingly, these issues worsen for the 10,000 year return period events with major system failures due to significant gutter overspilling (which worsen as event duration increases) and severe drain surcharging. Added to this, both the 2 and 10 minute duration events cause flooding to the surrounding surface areas with water depths of 11.6mm and 8.58mm respectively. Interestingly, system failure was found not be problem in response to the short duration high intensity events, but instead it was the longer duration lower intensity events (which produce a resultant higher depth of rainfall) that caused incapacity problems. From these results, the system can be classified as being able to drain a 10min M50 event under current design methods. 7.3 System response to events derived from BS EN 752:2008 Looking again at Table 2, it can be seen that the whole system operates effectively for all event durations for the 5 and 10 year return periods. However, the capacity of the system is unable to cope with any event with a return period of 50 or 100 years, as characterised by overspilling of the gutter and minor surcharging of the drain system. Again, it is this lower intensity longer duration events that cause system failure. The system can be classified as being able to drain a M10-60min event. 7.4 Comparison of event profile shape In addition to the differences of event duration and return period, the rainfall events derived from the two design standards, differ further by their event profile shape. Whilst BS EN :2000 provides a maximum averaged intensity representative of the most intense part of a longer storm which, in this study, has been applied as a constant intensity over the selected duration, BS EN 752:2008 provides a time-varying centrally-peaked profile based on the 50% summer profile. To investigate the impact of event profile shape on system response, a direct comparison can be made of the results of the 10min M50 event and the M50-10min event, derived from BS EN :2000 and BS EN 752:2008, respectively. Both events represent a 10 minute rainfall event with a 50 year return period, however, the 10min M50 event has an averaged steady-state profile and the M50-10min event has a time-varying intensity profile. From Table 2, the rainfall depth and average intensity for the 10min M50 event are 9.3mm and 56mm respectively, while the values for the M50-10min event are only slightly lower at 8.7mm and 52.2mm (just 7% lower). However, the system response to each of these events is really quite different. The averaged intensity of the 10min M50 event induces steady state flow conditions and the gutter water depth reaches a maximum of 140mm, while the surface area reaches a maximum of just 0.56mm. In contrast, the time-varying intensity of the M50-10min event induces unsteady flow conditions which now mean the gutter water depth exceeds 200mm resulting in overspill, the surface area depth increases to 1.54mm, and there is minor surcharging of the drain system. Despite having slightly lower values of rainfall depth and average intensity, the time-varying profile of the M50-10min event, which most accurately represents the intensity profile of observed rainfall events, allows the true unsteady nature of normal system operation to be included as an aspect of system design assessment. In this case a rainfall event with an assumed averaged steady-state intensity profile appears to pose no risk to system operation, however, when the same event is applied as a time-varying profile, system capacity is compromised and failure occurs. 314

8 SESSION E SUSTAINABLE CONSTRUCTION AND CLIMATE CHANGE 2011 Symposium CIB W062 Aveiro Portugal 8 Conclusion Increasing factors such as urbanisation and climate change are placing ever greater pressures on the operational performance of rainwater drainage systems. The already substantial costs associated with flood damage, and the potential for these to rise significantly in the future, call for an enhanced design methodology allowing understanding of whole system flow interaction. Currently, however, system design is undertaken by considering each element of the system as an individual component, with different design guidance provided in separate design standards. With the aim of developing an integrated design philosophy necessary to meet both current and future performance requirements, this paper has assessed, using numerical simulation, the performance of the whole rainwater drainage system of a case study building in response to a range of rainfall events derived from two separate design standards. Major differences between the design rainfall events derived from each standard have been highlighted, including: (i) differences in event duration (BS EN :2000 provides rainfall intensities for durations between one minute and 10 minutes, while BS EN 752:2008 provides rainfall intensities for durations between five minutes and 24 hours); (ii) differences in return period (BS EN :2000 provides values for return periods from one year to the maximum probable event of 10,000 years, while BS EN 752:2008 provides values for return periods between one year and 100 years); (iii) difference in event profile shape (BS EN :2000 provides design intensities as an averaged peak profile, while BS EN 752:2008 recommends the use of the time-varying 50% summer profile). Results showed that from BS EN :2000 and BS EN 752:2008, the system could be classified as being capable of draining both a 10min M50 event and a M10-60min, respectively. With current levels of return period already shown to be unrepresentative of future extreme events, it is likely that the return periods of these events will decrease, leading to potential flooding issues for this building in the future. Similarly, new systems designed using these current design methods may not have adequate capacity in the future. Although, historically, the two standards were developed for the design of different elements of the rainwater drainage system, it is important, especially when considering future impacts of climate change, to establish a design solution for the entire system with a unified design philosophy. Values of event duration and return period are site and building specific, and therefore should be assessed for each individual system. However, when undertaking whole system design it will be necessary to select values for each which are applicable to the whole system and not just to the individual element. Furthermore, it has been shown that by assuming an averaged block rainfall intensity, the resultant idealised steady-state flow conditions may in fact provide lower estimates of water depths, and hence overlook potential system failure, than those calculated when applying a more realistic time-varying intensity profile. 9 References Arthur, S. and Swaffield, J.A. Onsite evaluation of an installed siphonic rainwater drainage system. CIBW62 Water Supply and Drainage for Buildings Symposium, Rio de Janeiro, Brazil, September, Arthur, S., Wright, G. and Swaffield, J.A. Operational performance of siphonic roof drainage systems. Building and Environment, 40, , Arthur, S. and Swaffield, J.A. Siphonic roof drainage system analysis utilizing unsteady flow theory. Building and Environment, 36, , Arthur, S. and Swaffield, J.A. Numerical modelling of the priming of a siphonic rainwater drainage system. Building Services Engineering Research and Technology, 20(2), 83-91, BS EN 752-2:1997 Drain and sewer systems outside buildings. BS EN 8490:2007 Guide to siphonic roof drainage systems. BS EN :2000 Gravity drainage systems inside buildings Part 3: Roof drainage, layout and calculation. Design & analysis of urban storm drainage: The Wallingford Procedure Volume I. Principles, methods & practice. National Water Council, Standing Technical Report No. 28,

9 Escarameia, M. Monitoring of roof drainage systems. HR Wallingford, Report SR 500, Evans E., Ashley R., Hall J., Penning-Rowsell E., Saul A., Sayers P., Thorne C. and Watkinson A. (2004) Foresight. Future Flooding. Scientific Summary: Volume 1 Future risks and their drivers. Office of Science and Technology, London. Kelly, D.A., Jack, LB. and Beattie, R.B. An investigation into the impact of climate change on the performance of property drainage systems. CIBW62 Water Supply and Drainage for Buildings Symposium. Sydney, Australia, November, May, R.W.P. Manual for the hydraulic design of roof drainage systems. A guide to the use of British Standard BS 6367:1983. HR Wallingford, Report SR 485, Murphy, J.M., Sexton, D.M.H., Jenkins, G.J., Boorman, P.M., Booth, B.B.B., Brown, C.C., Clark, R.T., Collins, M., Harris, G.R., Kendon, E.J., Betts, R.A., Brown, S.J., Howard, T. P., Humphrey, K. A., McCarthy, M. P., McDonald, R. E., Stephens, A., Wallace, C., Warren, R., Wilby, R., Wood, R. A. (2009), UK Climate Projections Science Report: Climate change projections. Met Office Hadley Centre, Exeter. Natural Environment Research Council, Flood Studies Report, Sanderson, M. (2010). Changes in the frequency of extreme rainfall events for selected towns and cities. The Met Office, Devon. Swaffield, J.A., Escarameia, M. And Campbell, D.P. Unsteady roof gutter flow: Development and application of simulation. Building Services Engineering Research and Technology, 20(1), 29-39, Wright, G.B., Jack, L.B. and Swaffield, J.A. Investigation and numerical modeling of roof drainage systems under extreme events. Building and Environment, 41, , 2006a. Wright, G.B., Arthur, S. and Swaffield J.A. Numerical simulation of the dynamic operation of multi-outlet siphonic roof drainage systems. Building and Environment, 41, , 2006b. 10 Presentation of Author Dr David Kelly has been a Research Associate in the Drainage Research Group at Heriot-Watt University since His research interests include the monitoring and prevention of cross-contamination from building drainage systems and the impact assessment of climate change on rainwater systems. Dr Lynne Jack is Director of Research in the School of the Built Environment and has been a member of the Drainage Research Group at Heriot-Watt University since Her Research interests include the simulation of air pressure transient propagation in building drainage ventilation systems and the assessment of property drainage system performance when subject to climate change impacts. D.A.Kelly (1), L.B.Jack (2) (1)d.a.kelly@hw.ac.uk (2) l.b.jack@hw.ac.uk (1) (2) Drainage Research Group, School of the Built Environment, Heriot-Watt University, Edinburgh, EH14 4AS 316