Building Serv. Eng. Res. Technol. 26,4 (2005) pp.

337 /348

Technical note
Recent and future advances in roof drainage design and performance
S Arthur BEng (Hons) PhD and GB Wright MEng PhD School of the Built Environment, Heriot-Watt University, Edinburgh, UK

The past 10 years have witnessed significant changes in the way roof drainage systems are understood and designed. In particular, there has been a stepchange in the confidence with which siphonic roof drainage systems may be specified and expected to perform. These changes have occurred whilst urban drainage design in general has been revolutionized by wider acceptance of Sustainable Urban Drainage Systems and greater public concern regarding pluvial flooding within the context of climate change. This text considers, in detail, both how roof drainage systems are designed and how they should be expected to perform. Particular attention is drawn to weaknesses in accepted design methods. Consideration is also given to innovative roof drainage related approaches such as green roofs and rainwater harvesting. Practical application: Over the past few years there have been many changes in how roof drainage systems are specified and designed. On large buildings, technologies such as siphonic roof drainage are now commonplace and there is an ever increasing demand for green roofs to be specified due to their potential to green developments. Based on ongoing research, this paper details how these different types of roof drainage solutions can be efficiently designed and what levels of performance can be expected.

1 Introduction Over the past decade urban drainage systems have moved towards what are now commonly known as Sustainable Urban Drainage Systems (SUDS) or Best Management Practice (BMP). Fundamental to the implementation of these systems is addressing both runoff quantity and quality at a local level in a manner which may also have the potential to offer amenity benefits to stakeholders. This has

Address for correspondence: Scott Arthur, School of the Built Environment, Heriot-Watt University, Edinburgh EH14 4AS, UK. E-mail: s.arthur@hw.ac.uk
# The Chartered Institution of Building Services Engineers 2005

led to a change in the way new developments now look and interact within catchments. However, despite the availability of such tools to reduce, attenuate and treat urban runoff, substantial areas of the urban environment are still 100% impermeable and drain rapidly; namely roof surfaces. Normally, roof drainage systems do not always receive the attention they deserve in the area of design, construction and maintenance. Although the cost of a system is usually only a small proportion of a buildings total cost, it can be far outweighed by the costs of the damage and disruption resulting from a failure of the system to provide the degree of protection required.
10.1191/0143624405bt127tn

338 Recent and future advances in roof drainage There are basically two different types of roof drainage system, namely conventional and siphonic (see Figure 1). Conventional systems operate at atmospheric pressure, and the driving head is thus limited to the gutter flow depths. Consequently, conventional roof drainage systems normally require a considerable number of relatively large diameter vertical downpipes, all of which have to connect into some form of underground collection network before discharging to the surface water drain. In contrast, siphonic roof drainage systems are designed to run full-bore (turbulent gutter conditions mean that there will always be a small percentage of entrained air within the system, typically 5%), resulting in sub-atmospheric system pressures, higher driving heads and higher system flow velocities. Hence, siphonic systems normally require far fewer downpipes, and the depressurized conditions also mean that much of the collection pipework can be routed at high level, thus reducing the extent of any underground pipework. Both types of drainage system comprise three basic interacting components: . the roof surface; . the rainwater collection gutters (including outlets); . the system pipework. Each of these components has the ability to substantially alter the runoff hydrograph as it is routed through the system. This text will
Free surface outlet conditions Driving heads Annular flow in vertical downpipes Full-bore flow in all pipework Extensive underground network

focus on the role and performance of each of these components. As the principles of siphonic drainage are generally less well understood, and certainly less well documented, particular emphasis will be placed on the performance of siphonic roof drainage systems in this text. 2 Roof surface The design of the roof surface is usually within the remit of the architect rather than the drainage designer. Notionally, there are three types of roof surface: 2.1 Flat roofs Flat roofs are normally associated with domestic properties in climates with low rainfall, and with industrial buildings in developed countries. Such roofs are seldom truly flat, but simply fall below the minimum gradient associated with sloped roofs in the jurisdiction under consideration; for example, in the UK a flat roof is one where the gradient is less than 108.1 Minimum gradients are usually specified to avoid any unwanted ponding (BS 6229:2003 specifies a 1 in 80 minimum gradient), and to help prevent the development of any adverse gradient due to differential settlement.2 Although flat roofs can be problematic if not maintained properly, they are often preferred as they reduce the amount of dead space within the building and they attenuate flows more than sloped surfaces.
Submerged outlet conditions

Minimal underground network

a. Conventional system

b. Siphonic system

Figure 1 Schematics of a typical conventional and siphonic roof drainage system (at normal design condition)

S Arthur and GB Wright 2.2 Sloped roofs Most residential and many commercial properties have sloped roofs. Such roofs are generally favoured as their ability to drain naturally means that there is less risk of leakage. In temperate climates, their specification also means that snow loading is less of an issue. Once a rainfall event is underway, the rate at which the runoff flows across a roof is a function of roof slope and roughness. Where rainfall data is available, runoff rates from roof surfaces may be readily assessed using kinematic wave theory.3 2.3 Green/brown roofs (sloped or flat) Arguably the oldest type of permanent roof is a green roof. These involve the planting of roof areas to attenuate and/or dissipate rainfall, and can take the form of a roof top garden with trees and shrubs (termed intensive), or a lightweight carpet of growth media and flora (termed extensive). The latter technology is already employed widely (e.g., the Rolls-Royce plant at Goodwood, purportedly Europes largest green roof). Many of these applications tend to focus on the aesthetic benefits such systems offer to high profile developments, and are often installed in an effort to green a development and thus help secure planning consent in sensitive areas.4 However, as well as being aesthetically pleasing and hydraulically beneficial, green roofs may also offer thermal insulation,5 reduce the heat island effect (the phenomenon whereby absorption of solar radiation by urban surfaces causes a marked increase in ambient air temperature),6 provide acoustic damping and extend the service life of the roof membrane.7 10 Green roof systems are used extensively in Germany and to a lesser extent in North America, but again their specification is primarily due to a desire for a reduced aesthetic impact associated with a particular development. Germany probably has the most experience to date, a direct result of their use in the 1800s as a low fire risk alternative

339

to tarred roofs in deprived urban areas.11 Currently, German research is focused predominantly on planting issues, and there is only a limited understanding of how the systems may be used to mitigate the impact of urban runoff. One research project, which ran from 1987 to 1989 in Neubrandenburg,8 found that an installed green roof with 70 mm of substrate could reduce annual runoff from a roof by 60 80%. Work in Vancouver (Canada), based on an uncalibrated computer model, suggests that for catchments where the roof area comprises 70% of the total surface, installing an extensive system could reduce total runoff to approximately 60% over 12 months.12 The same model was also used to assess specific synthetic rainfall events; these results indicated that the catchment experience increased runoff during longer rainfall events. Neither of the above studies detail how green roofs could be expected to perform during a particular rainfall event, or where efficiencies may be gained in the design of collection pipework. Limited testing in the USA,13 where green roofs are often irrigated, has indicated that runoff can be reduced by 65% during a single event. The most authoritative design guidance for green roofs in the USA is produced by The New Jersey Department of Environmental Protection.14 This is focused on lightweight structures and gives guidance on how to ensure rapid draining where the rainfall return period exceeds 2 years. Rainfall return periods are normally set within the context of failure probability and consequence. Conventional systems are usually designed assuming 100% runoff for a 2-minute duration storm; the 2-minute duration is selected, as it is the typical time of concentration for conventional systems. Although advice is given in codes for setting higher runoff rates, there is little guidance on setting runoff rates below 100%. These observations mean that inadequacies are encountered if conventional codes are used to design green roofs:
/

340 Recent and future advances in roof drainage . Runoff coefficients should be expected to be below that used for conventional roofs; 100% is used by BS EN 12056-3:200015 and 98.7% was recorded by Pratt and Parker.16 . Peak runoff rates will be reduced; even where there is no infiltration, the surface roughness will have a significant impact. . Time of concentration would be expected to be greater than 2-min; particularly relevant when designing collection pipework for large roof areas i.e., public sector, commercial and industrial properties. . As with other elements of urban drainage design, it is not efficient for a complex system such as a green roof to be matched to a single rainfall event. It is probable that the duration of runoff hydrographs will be orders of magnitude longer than compared with conventional systems, and runoff interactions between independent rainfall events are probable; this may make a timeseries approach more appropriate. 3 Rainwater collection gutters The basic requirement for rainwater collection gutters is that they have sufficient flow capacity to accommodate flows from the design storm.17 Although it is common practice to install gutters at a slight gradient to prevent ponding, the nature of the construction industry and the process of settlement means that it is normal to assume gutters laid at slack gradients are actually flat; for example, BS EN 12056-3:2000 stipulates that gutters at gradients less than 0.3% shall be treated as being flat. In a level gutter, the water surface profile will slope towards the outlet, and it is the difference in hydrostatic pressure along the gutter that gives the incoming water the required momentum to flow towards the outlet.18 3.1 Gutter outlet depths Key to ensuring whether or not collection gutters have sufficient capacity are the conditions that occur at the gutter outlets. As well as affecting the flow rates entering the drainage system pipework, the outlet depths also affect the upstream gutter depths (via the backwater surface profile). Hence, although the depth at a gutter outlet may not cause any particular problems, the greater depths occurring at the upstream end of the gutter may result in overtopping. Extensive experimental studies in the 1980s determined that the flow conditions in the vicinity of a gutter outlet in a conventional roof drainage system could be categorized as being either weir type or orifice type, depending on the depth of water relative to the size of the outlet.19 At depths below that equivalent to half of the outlet diameter, the flow conditions are weir type and outlet conditions are calculated using an appropriate sharp-edged weir equation.18 At higher flow depths, the flow effectively chokes and the flow regime changes to orifice type, with the outlet conditions being calculated by an appropriate sharp-edged orifice equation.18 Although conventional roof drainage systems are usually designed to ensure free discharge at gutter outlets, design restrictions may mean that the outlets cannot discharge freely; in such circumstances, additional gutter capacity (storage) will normally be required to accommodate the resulting higher flow depths. In siphonic roof drainage systems, the outlets are designed to become submerged in order to allow full-bore flow conditions to develop and be sustained; if this is the case the determination of outlet depth is complicated as the gutter conditions are dependent upon the downstream conditions (within the connected pipework) as well as the gutter inflows. Recent experimental work has also indicated that conventional roof drainage systems incorporating non-standard gutter sections, whose base width and height is significantly greater than the diameter of the outlet, can result in the development of full-bore flow conditions in the vertical downpipe and ultimately siphonic action;20 for a given gutter

S Arthur and GB Wright section, the onset and extent of such conditions were observed to be dependent on the diameter of the downpipe. Similar phenomena have also been observed in standard gutter sections (semicircular and elliptical); in these cases limited siphonic action was observed to occur for only a short distance below the outlet.18 3.2 Flow division within gutters In terms of flow division between multiple outlets in a gutter under free discharge conditions, it can be seen from Figure 2a that the flow splits evenly in any given gutter section (between two outlets or between an end wall and an outlet), irrespective of whether the gutter inflow is uniform or non-uniform. Figure 2b and 2c indicate the effect of outlet placement within a gutter; evenly spaced outlets requiring far less gutter capacity (or outlets) than those placed at the gutter extremities.
Q1 (uniform)

341

Where outlets are not freely discharging, the flow division between multiple outlets in a gutter may not be as described, as the individual gutter sections may hydraulically merge to form one continuous channel and/or downstream system conditions may become significant. For example, the pipework in a siphonic system will run full-bore when operating at or near its design point, and the flow division between outlets will be dependent on the relative losses associated with each branch of the system. 3.3 Backwater profiles The water surface profile in gutters can only realistically be assessed by applying the momentum equation for channels with lateral input. In many cases, the low velocities associated with gutter flows mean that gutter friction losses are minor and may be ignored.18 If a gutter outlet is such that it allows free discharge, and frictional effects
Q3 (uniform)

Q2 (non-uniform)

Q1

0.5 Q2

0.5 Q2

0.5 Q3

a. Flow division between multiple outlets in a gutter Q (uniform)

0.25 Q

0.25 Q

0.25 Q

0.25 Q

b. Flow division between evenly spaced outlets in a gutter Q (uniform)

0.5 Q

0.5 Q

c. Flow division between outlets positioned at gutter extremities

Figure 2 Effect of outlet positioning on ow division in gutters

342 Recent and future advances in roof drainage are neglected, the backwater profile may be determined by applying Equation (1) to determine the horizontal distance (DL ) between any given upstream depth (h1) and downstream depth (h2).
h2 :25 (2) QL  0:9  QN  0:9  2:78  105  A1 E

DL 

g
h1
/




gA3  dh Q2 So  A2 mC 2

1

Q2 T

Where: QN notional gutter design capacity (l/s); AE Gutter cross-sectional area (mm2). Clearly not all gutters can be designed by application of Equation (2). For example, BS EN 12056-3:2000 contains clauses to account for many eventualities, including:
/ /

(1)

Where Q flowrate (m3/s); T surface width (m); g gravitational constant (m/s2); A  flow area (m2), So bed slope ( ); m  hydraulic mean depth (m); C Che zy coefficient ( ). Equation (1) can be modified if frictional effects are considered to be significant (very long gutter lengths or very high flow velocities), or if the gutter outlet is not freely discharging.
/ / / / / / / /

. Location of gutter on building, which may result in varying consequences of failure e.g., eaves gutter, valley gutter, parapet gutter. . Different shaped gutter sections. . Hydraulically long gutters (where frictional effects may be significant). . Gutters laid at a significant gradient. . Changes in gutter alignment (bends, offsets, etc). . Additional system elements, such as strainers or rainwater hoppers. . Restricted flow at outlets. . Gutters installed on siphonic roof drainage systems. In addition to the type of clauses listed above, BS EN 12056-3:2000 also allows designers to utilize data obtained from experimental testing of a particular arrangement. 3.5 Numerical models A number of numerical models have been developed that can accurately simulate the flow conditions in any type of gutter as a result of either steady or unsteady roof runoff. An example of this is incorporated into the ROOFNET model recently developed as part of a academic research project dealing with the effect of climate change on urban drainage.20 This model enables the user to specify data describing the relevant aspects of a particular installation, including: details of the prevailing rainfall conditions, details of the roof surfaces to be drained and details of the actual gutters. A kinematic wave model is then used to route the rainfall over the roof surfaces and into the gutters. A method of

3.4 Current design methods The foregoing discussion has highlighted the key elements that should be considered when designing a rainwater gutter. However, without recourse to some form of numerical modelling, it is just not feasible to calculate the backwater surface profiles, and hence gutter capacities, for roof drainage systems; this is particularly the case for large commercial or manufacturing developments which may incorporate many kilometres of different types of guttering. Consequently, current gutter design methods for gutters installed in conventional drainage systems are based primarily on empirical relationships19 and the assumption of free discharge at the outlet. For example, BS EN 120563:2000 specifies that the design capacity of a short, level, semicircular gutter located on the eaves of a building (with outlets capable of allowing free discharge) is given by (QL );

S Arthur and GB Wright characteristics based solution of the fundamental equations of one-dimensional flow in open channels is then used to route the runoff along the gutters to the outlets,21 at which point the flow enters the drainage system pipework. The model automatically determines the flow conditions at the gutter outlets and, in addition to dealing with free discharge cases, can also simulate the effect of restricted flow conditions and submerged outlet scenarios. Output includes depths, velocities and flow rates along the gutter, as well as the location and severity of any gutter overtopping events. At present, models such as that described are essentially research tools, in that they are normally developed and utilized by universities for specific research projects. However, it is envisaged that such models may soon come to be used as diagnostic design aids, particularly to assist code development. 4 System pipework The type and extent of pipework incorporated into a roof drainage system depends primarily on whether the system is conventional or siphonic. 4.1 Conventional rainwater systems In conventional roof drainage systems, the above ground pipework generally consists of vertical downpipes, connecting the gutter outlets to some form of underground drainage network, and offset pipes, used where the gutter overhang is significant. It should be noted that an offset pipe is defined as a pipe with an angle less than 108 to the horizontal. The capacity of the system as a whole is usually dependent upon the capacity of the gutter outlets rather than the capacity of the vertical downpipes. The flow within vertical downpipes will normally be free surface, with BS EN 120563:2000 specifying that downpipes run no more than 33% full; this effectively installs redun-

343

dant capacity within the system. If the downpipes are sufficiently long (normally greater than 5 m), annular flow conditions may occur. Similarly, the flow conditions within offset pipes will also normally be free surface, with BS EN 12056-3:2000 specifying that offsets run no more than 70% full; indicating the need to install all offsets at a gradient. The design of the pipework can either be undertaken utilizing the design tables in BS EN 120563:2000, or by applying the Wyly-Eaton equation for vertical downpipes22 and the Colebrook-White equation for offset pipes.23 4.2 Siphonic roof drainage systems In contrast to conventional systems, siphonic installations depend upon the purging of air from the system (priming) and the subsequent establishment of full-bore flow conditions within the pipework connecting the outlets in the roof gutters to the downstream surface water sewer network (at ground level). Current design practice assumes that, for a specified design storm, a siphonic system fills and primes rapidly with 100% water.24 This assumption allows siphonic systems to be designed utilizing steady state hydraulic theory. The steady flow energy equation is normally employed,21 with the elevation difference between the gutter outlets and the point of discharge being equated to the head losses in the system. Although this approach neglects the small quantities of entrained air that always enter a siphonic roof drainage system, it has been reported to yield operational characteristics similar to those observed in laboratory test rigs at the fully primed state,25,26 However, steady state design methods are not applicable when a siphonic system is exposed to a rainfall event below the design criteria, or an event with time varying rainfall intensity. In the former case, the flow may contain substantial quantities of entrained air and exhibit pulsing or cyclical phases; a result of greatly varying gutter water levels and an indication of truly unsteady, transient flow

344 Recent and future advances in roof drainage

Gutter depth (mm)

Phase 1
150 75

Phase 2

Phase 3

Gutter 1

Gutter 2

0
Pressure P1

0.6 0.0 -0.6 -1.2

Pressure (mH2O)

Depth in gutter 1 Depth in gutter 2 Pressure P1

Schematic of test rig

Air pockets leave system

0 10 20 30 40 50 Time since start of simulated rainfall event (s ) 60

-1.8

Figure 3 Priming of a laboratory siphonic drainage test rig28

conditions. Such problems are exacerbated when the system incorporates more than one outlet connected to a single downpipe (multioutlet system), as the breaking of full-bore conditions at one of the outlets (due to low gutter depths and air entry) is transmitted throughout the system and, irrespective of the gutter depths above the remaining outlet(s), results in cessation of fully siphonic conditions. As sub-design events are the norm, it is clear that current design methods may not be suitable for assessing the day-to-day performance characteristics of siphonic roof drainage systems. This is a major disadvantage, as it is during these events that the majority of operational problems tend to occur e.g., noise and vibration. Despite any demerits that current design methods may have, many thousands of systems have been installed worldwide with very few reported failures. Where failures have occurred, they have invariably been the result of one or more of the following; 1) 2) 3) 4) a lack of understanding of operational characteristics; poor material specification; installation defects; a poor maintenance programme.

undertaken to augment the understanding of siphonic roof drainage systems, and to develop numerical models for use as diagnostic design aids.27,28 The remainder of this section will present a selection of the salient points arising from this work. In contrast to the assumption made with current design methods, the priming of a typical siphonic system was actually found to be as follows (refer to Figure 3): 1) 2) Flow conditions throughout the system are initially free surface (Phase 1). Full-bore flow conditions form at some point within the horizontal pipework (Phase 1). Full-bore flow conditions propagate downstream, towards the vertical downpipe, and upstream, towards the gutter outlets (Phase 1). Full-bore flow conditions reach the vertical downpipe, the downpipe starts to fill and the system starts to depressurize (Phase 2). Once the conditions throughout the downpipe are full-bore, any remaining air pockets are purged from the system (Phase 2). Full siphonic action occurs (Phase 3), and will continue until the gutter depth(s) falls

3)

4)

5)

In response to these perceived shortcomings, a series of research projects have recently been

6)

S Arthur and GB Wright below the level at which air can enter the system. The data shown in Figure 4a illustrates the type of unsteady flow conditions that occur when a siphonic system is exposed to rainfall events below the design point, and the gutter flow depths are insufficient to sustain full siphonic action. The data shown in Figure 4b illustrates the type of unsteady flow conditions that occur when an installed siphonic system is exposed to a real rainfall event, where the rainfall intensity varies with time.
a

345

Figure 5 shows an example of the output from one of the numerical models that have recently been developed (SIPHONET). As can be seen, the model can accurately simulate the priming of a siphonic system (0s 32s) as well as steady siphonic conditions (32s 62s). This data also illustrates that the model can simulate complex operating conditions, such as the rise in system pressures when the depth in gutter 1 drops below that necessary for fullbore flow, hence allowing air to enter the system and break the siphon (at approximately 62s)
/ /

0.4

0.0 Pressure ( mH2O )

-0.4 Regime 1 (15 - 40% Qmax ) Regime 2 (40 - 60% Qmax ) Regime 3 (60 - 80% Qmax ) -1.6 0 20 40 60 Time since start of simulated rainfall event (s) 80

-0.8

-1.2

0.5 Pressure P1 0.0 Rainfall instensity

120 Rainfall intensity (mm/hour) 100 80 60 40 20 0 7000

Pressure ( mH2O )

-0.5 -1.0 -1.5 -2.0 -2.5 0 1000 2000 3000 4000 5000 Time since start of rainfall event (s) 6000

Figure 4 (a) Measured system pressures for sub-design rainfall events within a laboratory siphonic drainage test rig.28 Note that this data refers to the pressure P1, as indicated in Figure 3; (b). Sub-design rainfall events within an installed siphonic drainage system28

346 Recent and future advances in roof drainage

Gutter depth ( mm ) 160

Measured gutter 1 depth Predicted gutter 1 depth

80

0.4 Pressure ( mH2O )

-0.4

Measured pressure P1 Predicted pressure P1

-1.2

-2.0 0 25 50 75 Time since start of simulated rainfall event ( s ) 100

Figure 5 Measured and predicted system conditions within a laboratory siphonic drainage test rig: no inow into gutter 1 between 62s and 82s.28 Note that this data refers to the system shown in Figure 3

5 Conclusion The text has illustrated how roof drainage systems are a key, but often overlooked, element of urban drainage infrastructure. It has also been shown that their design is a complex process, which relies heavily on gutter outlet performance. The following conclusions may be drawn with respect to the operation of roof drainage systems: 1) Their operation is dependent on three interacting components: the roof surface, the collection gutter and collection pipework. Green or brown roofs provide an opportunity to reduce the flow from roof surfaces, improve urban aesthetics and increase urban biodiversity. Outlet conditions are key to understanding how a system will perform. Siphonic roof drainage systems present a more efficient way to drain large roof surfaces. However, this must be considered within the context of possible higher maintenance costs. The design of siphonic roof drainage systems should include an allowance for

sub-design rainfall events and operational problems e.g., blocked outlets. Although green roofs are an attractive alternative, it is probable that conventional roof surfaces will continue to dominate domestic installations. However, it is likely that green roofs will experience a step-change in acceptance by the commercial sector once more becomes known about their performance and sustainability. Similarly, the efficiencies offered by siphonic systems means that they will continue to play a significant role in draining large commercial buildings, particularly if numerical models are applied diagnostically to improve performance and reduce the occurrence of costly system failures. The biggest threat to roof drainage comes from climate change. Existing systems may not simply become more prone to flooding; changes in rainfall patterns may result in long periods of low precipitation, and self-cleansing velocities may be attained less frequently as a result. Furthermore, changes in wind patterns may also increase levels of rooftop debris, and hence necessitate enhanced maintenance programmes. As concern regarding climate change grows, and the sustainability agenda widens, it is possible that harvesting

2)

3) 4)

5)

S Arthur and GB Wright roof runoff may become more widespread. At present water consumption varies globally between 7 and 300 l/capita per day. In the UK, average consumption is 145 L/h per d, but only 1 2 l may actually be consumed by humans, whilst 30% may be used for WC flushing.29 Studies have shown that, when coupled with storage, roof rainwater harvesting has the potential to contribute substantially to domestic water usage in both developing and developed countries.30,31
/

347

References
1 BS 6229:2003. Flat roofs with continuously supported coverings , Code of practice. British Standards Publishing Limited (BSPL), 2003. 2 Simmons T. Methods for designing proper roof drainage. Professional Roong . 1994; 22 /25. 3 Singh VP. Kinematic wave modeling in water resources: Surface water hydrology. Chichester: John Wiley and Sons Inc, 1996. 4 Tarr AR. Green roof implementation: construction and contractor issues in the UK . Green Roofs for Healthy Cities, University of Shefeld, 2003. 5 Ko hler M. The multi-beneciary system of green roofs, The green roof challenge to biophilic architecture and ecology. Proceedings of the Shefeld Conference Nature Enhanced, 2004. 6 Rosenfeld AH, Akbari H, Bretz S, Fishman BL, Kurn DM, Sailor D, Taha H. Mitigation of urban heat islands: materials, utility programs, updates. Energy and Buildings 1995: 22; 255 /65. 7 Hendricks NA. Designing green roof systems: A growing interest. Professional Roong 1994; 20 /24. 8 Ko hler M, Schmidt M, Paiva VLA and Taveres S. Green roofs in temperate climates and in hot-humid tropics */far beyond aesthetics. Environmental Management and Health 2002; 13: 9 Niachou A, Papakonstantinou K, Santamouris M, Tsangrassoulis A, Mihalakakou G. Analysis of the green roof thermal properties and investigation of its energy performance. Energy and Buildings 2001; 33: 719 /29.

10 Onmura S, Matsumoto M, Hokoi S. Study on evaporative cooling effect of roof lawn gardens. Energy and Buildings 2001; 33: 653 /66. 11 Ko hler M. Green roof technology */from a re-protection system to a central instrument in sustainable urban design . Second Green Roof Conference, Portland, Oregon, 2004. 12 GVRD. Effectiveness of stormwater source control . Greater Vancouver Regional Sewerage and Drainage District, 2002. 13 PSU. Online Research http://www.rps.psu.edu/ 0105/roofs.html, 2001. 14 NJDEP. Standard for rooftop runoff management, 2000. 15 BS EN 12056-3:2000. Gravity drainage systems inside buildings. Roof drainage, layout and calculation . British Standards Publishing Limited (BSPL), 2000. 16 Pratt CJ, Parkar MA. Rainfall Loss Estimation on Experimental Surfaces. Proceedings of the Fourth International Conference on Urban Storm Drainage . International Association of Hydraulic Runoff Research, Lausanne, Switzerland, 1987. 17 Beij H. Flow in Roof Gutters, US Department of Commerce, Bureau of Standards: Research Paper RP644. Bureau of Standards Journal of Research 1934, Vol 12. 18 May RWP. Design of conventional and siphonic roof drainage systems, Public Health Services in Buildings / Water Supply, Quality and Drainage . IWEM Conference, London, 1995. 19 May RWP. Hydraulic design of roof gutters . Proceedings of the Institution of Civil Engineers. 1984; Part 2, Vol 77. 20 Blanksby J, Ashley R, Saul AJ, Cashman A, Packman J, Maksimovic C, Jack L, Wright G, Kay D. Adaptable urban drainage (AUDACIOUS), NOVATECH 2004 . 5th international conference on sustainable techniques and strategies in urban water management, 2004. 21 Escarameia M, Swafeld JA. Prototype monitoring and numerical simulation of roof drainage gutter systems . Proceedings of the 8th International Conference on Urban Storm Drainage, Sydney, Australia, 1999. 22 Wyly RS, Eaton HN. Capacities of plumbing stacks in buildings . National Bureau of Standards Building Materials and Structures, Report BMS 132, 1952.

348 Recent and future advances in roof drainage

23 Chadwick A, Morfett J. Hydraulics in Civil and environmental engineering . 4th Edition, Spon Press, 2004. 24 Arthur S, Swafeld JA. Siphonic roof drainage: The state of the art. Urban Water 2001; 3: 43 /52. 25 May RWP. Escarameia M. Performance of siphonic drainage systems for roof gutters. Report No SR 463. HR Wallingford, 1996. 26 Arthur S, Swafeld JA. Siphonic roof drainage system analysis utilising unsteady ow theory. Building and Environment 2001; 36: 939 /48. 27 Wright GB, Swafeld JA, Arthur S. Investigation into the performance characteristics of multi-outlet siphonic rainwater systems. Building Serv. Eng. Res. Technol. 2002; 23: 127 /41. 28 Swafeld JA, Wright GB, Jack LB, Arthur S. Pressure transient analysis to inform system design for building and roof drainage systems . Proceedings the practical application of surge analysis for design and operation 9th international conference, 2004. 29 Butler D, Davies J. Urban drainage , 2nd Edition, Spon Press, 2004. 30 Thomas T, Domestic water supply using rainwater harvesting. Building Research and Information 1998; 26: 94 /101. 31 Fewkes A. Modelling the performance of rainwater collection systems: towards a generalised approach. Urban Water 2000; 1: 323 /33.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.