Abstract
This paper describes a full-scale experimental investigation into the effects of box gutter geometry on the open channel flow conditions above siphonic roofwater outlets. In particular, the effects of channel width (300, 400, 480 and 600 mm) and length (14.86 and 32.00 m) were investigated through measurements of flow rate, water depth and longitudinal velocity in the box gutter. The experimental results showed that for the same outlet flow rate, the depth of water in the gutter varied by up to 211% for the two different gutter lengths tested. Generally, the greatest water depths for the different flow rates were recorded in the 400-mm wide gutter and the lowest water depths were recorded in the 300-mm wide gutter. It was also found that the maximum flow rate through the single 110-mm diameter outlet varied depending on the width of the gutter.
Keywords
Introduction
An important functional requirement of many water sensitive urban design (WSUD) systems is to harvest and reuse runoff collected from roofs, roads and other urban catchment surfaces. This often becomes a challenge in highly developed residential, commercial and industrial areas due to space constraints caused by the generally larger building sizes. However, there is a relatively new method for harvesting water from large roof surfaces that can be used as part of a WSUD system. Siphonic roofwater harvesting systems utilize the height of the building to develop sub-atmospheric pressures in the system pipework. This then forces the gutter water into the siphonic pipework through the special outlets and this causes very high flow velocities and flow rates in the pipe system. Because the system flows under pressure, it is possible for the majority of the pipework to be laid horizontally rather than vertically and there is also a greatly reduced number of downpipes required to drain the system – quite often only one or two. Consequently, only one or two rainwater tanks are required for storing the harvested water and this can then be reused later within the WSUD development. In addition, these tanks, which are often modular in design, can be installed inside the building in any under-utilised space such as under a stairwell or in the basement (Figure 1). Therefore siphonic roofwater systems and their associated storage and reuse systems offer a new and efficient way to implement WSUD in highly urbanized environments.
Modular, interior rainwater tank situated under a mezzanine floor.
The main components of a typical siphonic roof drainage system are:
Siphonic outlets Box gutters Siphonic pipe system In-ground drainage system
The depth of water in the open channel box gutters above the outlets of siphonic drainage systems is an important design variable and the system design process is highly dependent on accurate estimation of these water depths. The depth of water also plays a significant role in successful system operation. Not only can high water depths considerably increase the risk of building flood damage through gutter overtopping but a certain minimum depth of water above the outlets is necessary to assist in developing the sub-atmospheric pressures required within the pipework. A minimum depth of water above the outlet is also required to prevent air from entering the system, which can diminish or break the siphon effect. It is therefore imperative in the design of siphonic roof drainage systems to be able to accurately estimate the open channel water depths during all phases of operation.
Lucke and Beecham1 conducted an experimental investigation into how negative system pressures, the degree of aeration and gutter water levels are affected by the number of outlets in a siphonic roof drainage system. The experimental results showed that system pressures decreased if outlets were blocked. It was also shown that the depth of water above an outlet was strongly influenced by the negative pressure acting at the outlet. As the suction effect at the outlet increased, due to lower system pressures, more water and air was drawn into the outlets and this resulted in reduced system efficiency and a corresponding increase in overall gutter water depth. The results also showed that there is often built-in redundancy in multiple outlet siphonic systems experiencing lower-intensity rainfall events, and even if one or two outlets were blocked, the system would still operate satisfactorily. In an extension to this work, Lucke and Beecham2 conducted an experimental investigation into air entrainment in siphonic roof drainage systems. By injecting known quantities of air into a 4.7-m high experimental test rig, the effect of aeration on flow velocity and flow capacity in the siphonic system was determined. A preliminary relationship between air/water content and a decrease in system capacity was formulated. This indicated that system capacity can decrease by up to 40% for air/water ratios approaching 45%. At typical operational air/water ratios of approximately 10%, system capacity decreased by 16% from the assumed no-air-design situation. In addition, average velocities in the pipework decreased by 13% for air/water ratios of 35%. Friction and form losses were also measured and compared with theoretical values.
Siphonic outlet blockage has also been investigated by Arthur and Wright 3 . They stated a key benefit of a siphonic system is the ability to redistribute flows between outlets if one becomes blocked. Their experimental data indicated that where an outlet was blocked prior to the initiation of a simulated rainfall event, the system acted as a single outlet siphonic system. Although flows were reduced, the pressures developed in the system were lower than the fully primed capacity. They concluded that blocking single outlets in multi-outlet systems can lead to reduced operating pressures and that if a system was designed to operate at very low pressures, a complete blockage of an outlet may result in the onset of cavitation and/or failure of the system.
May and Escarameia 4 tested a number of commercially available siphonic outlets and found that the design of the baffle plate, which is typically included in siphonic outlet designs to reduce air ingress, had very little effect on the capacities of siphonic outlets installed in gutters. The experimental set-up used by May and Escarameia 4 was relatively small-scale, which may explain why they did not observe a significant influence due to the presence of baffle plates. Baker 5 investigated how outlet geometry impacts on siphonic system performance using an experimental apparatus that included an adjustable baffle plate and water depth recorders. The research concluded that the gutter water depth is directly proportional to the height of the baffle plate.
Bramhall and Saul 6 investigated the performance of siphonic outlets in relation to their position within the gutter. They found that the minimum gutter water level above the outlets was generally achieved when the spacing of the outlets was equidistant within the gutters, i.e. when the length of gutter flow on either side of the outlet was equal. Their results indicated that the water level required above the outlets to achieve maximum capacity was 45 mm. The maximum flow rate they achieved through an outlet was approximately 12.0 L/s. Bramhall and Saul 6 also described a draw-down of the water level at the outlet due to suction effects.
Results of research conducted by Wright et al. 7 indicated that the water levels required above their outlets, in order to ensure limited aeration and maximum outlet capacity, were between 90 mm and 130 mm. The maximum flow rate through their outlets however was only 5.9 L/s and 7.8 L/s, respectively.
The variation in the results of the water depths above the outlets presented by May, 8 Bramhall and Saul, 6 and Wright et al. 7 suggest that there may have been various factors that affected the water depth measurements. One possible factor may have been that the water level measurements were taken at different positions in the gutter for each of the studies. The flow rate through the outlet seems to have been another factor. As the geometries of the gutters of the experimental rigs used by May, 8 Wright et al., 7 and Bramhall and Saul 6 were all different (width × length = 350 mm × 10 m; 600 mm × 4 m and 600 mm × 35 m, respectively), comparisons between the water depths they obtained in their research results become difficult.
Bramhall and Saul 9 hypothesized that there is a linear relationship between the depth of water above an outlet and the gutter width and developed charts to predict the depth of water above an outlet in gutters of various widths. As previously discussed, it is essential in the design of siphonic roof drainage systems to accurately determine the gutter water depths during all phases of operation in order to optimise performance and minimise the risk of flooding. The research presented in this paper attempts to further examine the effects of various gutter widths and lengths on the open channel water depth. The effects of gutter geometry on flow velocities are also reported.
Experimental procedure
Gutter geometry
To examine the effects of gutter geometry on the water depth above a siphonic outlet, the gutter width and length was varied for a single siphonic outlet located in the full-scale testing rig shown in Figure 2. A single 110 mm diameter outlet was installed in the gutter at a distance of 7.43 m from the left-hand side (LHS) end and 24.53 m from the right hand side (RHS). This position was chosen for the outlet in order to observe the different effects that unequal gutter lengths have on the flow profiles. The ratio of gutter lengths on each side of the outlet therefore was approximately 1:3. The pipework beneath the outlet was all 110 mm in diameter and consisted of a 0.5-m length vertical tailpipe followed by a 3-m length of horizontal pipe before a 6-m length of vertical downpipe.
Single siphonic outlet position in gutter.
To investigate the effect of gutter width on the water surface profiles in the vicinity of the single siphonic outlet, a 5-m long set of walls, with an adjustable width, was constructed. The walls were then fitted into the 600-mm wide original gutter so that the position of the single outlet was equidistant from each wall end. The walls were also positioned so that the outlet was centrally placed between each wall face (Figure 3). A wall length of 5 m was selected after trialling gutter width reductions over various lengths. Further details of this methodology are given in Lucke.
10
Four different gutter widths (300, 400, 480 and 600 mm) were tested using the adjustable wall system. Measurements of the water depth were taken at distances of 0, 0.5, 1.0, 1.5 and 2.0 m from either side of the centre of the single outlet.
Wall positioning and measurement locations (upper = plan view, lower = side view).
The system was run for a minimum duration of 5 min before any measurements were taken for each test. This was done to allow the water profiles in the entire gutter to stabilise. Flow rates of 5, 10, 15, 20, 25, 27 and 29 L/s were tested for four different gutter widths (300, 400, 480 and 600 mm).
In addition to the water surface profile measurements, the average flow velocity in the gutter on each side of the outlet was recorded using a water current meter (type: R.OTT-C.2). The velocity measurements were taken at the same position for each measurement (middle of gutter, 1.0 m left and right from the centre of the outlet), but the submersion depth of the meter was adjusted to be as close as possible to the maximum flow velocity position as explained below. Munson et al. 11 suggest that the maximum flow velocity in an open-channel is found below the water surface at a distance of approximately one third of the overall water depth at that point. While the spatially varied flow experienced in roof gutters may vary this position slightly, the work of Khiadani 12 indicates that one third of the overall depth is still suitable.
It should be noted that square rather than circular siphonic outlets were used in this study. The square opening in the gutter sole reduces in this case to a 110-mm diameter pipe, as shown in Figures 2 and 3.
Equal gutter lengths
In order to observe the influence that the length of gutter has on the outlet performance and gutter water levels, the 600-mm wide gutter was blocked off at a distance of 7.43 m from either side of the single outlet so that the outlet was receiving flow from the same length of gutter on both sides. This gutter length (14.86 m) meant that the outlet was receiving inflow from 39 roof sheeting pans on each side. The flow rate from each pan in the roof sheeting was calibrated to determine the lateral inflow rate on each side of the outlet and these were found to be equal to each other within 3 ± 0.7%. Three different flow rates were tested with this configuration. The water surface profiles were then compared to the profiles when using the whole gutter length of 32.00 m.
Results and discussion
Geometry testing
It was found that the maximum flow rate through the single 110-mm diameter outlet varied depending on the width of the gutter. The maximum flow rate for the 400 mm and 480-mm wide gutter was 27 L/s and the maximum flow rate for the 300-mm and 600-mm wide gutter was 29 L/s. The gutter water depths on each side of the outlet for varying flow rates in the 300-mm wide gutter are shown in Figure 4. The gutter water depth results for the 400-mm, 480-mm and 600-mm wide gutters are shown in Figures 5–7, respectively.
Gutter water depths for a 300-mm wide gutter. Gutter water depths for a 400-mm wide gutter. Gutter water depths for a 480-mm wide gutter. Gutter water depths for a 600-mm wide gutter.
As the maximum flow rates through the outlet varied with gutter width, a comparison of the water levels for the highest common flow rate of 27 L/s was made and is shown in Figure 8.
Gutter water depths for each gutter width at a flow rate of 27 L/s.
For the 480-mm wide gutter, the water depth profile for 27 L/s appears to be different to those for the other flow rates. In particular, a noticeable depression in water depth occurs on the RHS approximately 1 m from the outlet. While further research would be required to investigate the specific cause of these unusual flow conditions, these results do indicate that there may be certain gutter width and length combinations for which different siphonic outlets may perform more efficiently.
The average velocities and the Froude numbers of the flow in the LHS and the RHS of the gutter at different flow rates were recorded. The results for the 300-mm, 400-mm, 480-mm and 600-mm wide gutters are shown in Figures 9–12, respectively.
Velocity and Froude numbers for a 300-mm wide gutter. Velocity and Froude numbers for a 400-mm wide gutter. Velocity and Froude numbers for a 480-mm wide gutter. Velocity and Froude numbers for a 600-mm wide gutter.
A comparison of the channel velocities in the four different gutter widths for varying flow rates is shown in Figures 13 and 14 for LHS and RHS of the gutter, respectively.
Left-hand side (LHS) channel flow velocity. Right-hand side (RHS) channel flow velocity.
It was found that the maximum flow rate through the single 110-mm diameter outlet varied depending on the width of the gutter. The maximum flow rate appeared to be dependent on the degree of aeration that occurred around the outlet due to the varying flow conditions for each gutter width. The degree of aeration may be influenced by the outlet geometry and further investigation using different shaped outlets is recommended to investigate this phenomenon.
Figures 4–7 show that the water depths above the outlet vary significantly with the different gutter channel widths tested. Generally, the greatest water depths for the different flow rates were recorded in the 400-mm wide gutter and the lowest water depths were recorded in the 300-mm wide gutter. These results contrast with those presented by Bramhall and Saul 9 who hypothesized that there is a linear relationship between the depth of water above an outlet and the gutter width. Figure 8 does not show a direct relationship between water depth and the gutter width for the geometries used in this investigation.
The results of the gutter channel flow velocity measurements in Figure 9 show that the average channel velocity in the 300-mm wide channel increased as the flow rate increased. Figure 9 also shows that the Froude number on each side of the gutter remained approximately constant for increasing flow rates. This is because the increase in flow velocity is compensated for by the increase in water depth.
The results of the gutter channel flow velocity measurements in Figure 10 show that the average channel velocity in the 400-mm wide channel increased as the flow rate increased up to 20 L/s. However, at flow rates greater than 20 L/s, the flow conditions changed and the outlet became “drowned”. The outlet drowning resulted in a rapid increase in gutter water depth and the formation of two distinct vortexes, which are shown in Figure 15.
Vortex formation in a drowned outlet in a 400-mm wide gutter. Comparison of gutter water depths in a 600-mm wide gutter at 15 L/s.
The increased gutter water depths due to the drowned outlet are reflected in the sharp decrease in the Froude numbers on both sides of the outlet as shown in Figure 10. This increase in depth corresponds to a relative reduction in outlet discharge. Wright et al. 7 also investigated the effects of different downpipe configurations on system performance and found that a drowned discharge outlet can reduce the system capacity by up to 5%. In the current investigation, the outlet drowning effect described above only occurred in the 400-mm wide gutter. This was an interesting result that suggests that there may be optimum gutter widths for which different siphonic outlets may perform more efficiently. Further research would be required to investigate this effect.
The results of the gutter channel flow velocity measurements in Figure 11 show that the average channel velocity and the Froude number in the RHS of the 480-mm wide channel both increased as the flow rate increased. In contrast, the channel velocity in the LHS did not change significantly. However, the Froude number in the LHS decreased as the flow rate increased. This was caused by a steady increase in gutter water levels as the flow rate increased.
Figure 12 shows that both the average channel velocity and the Froude number in the LHS of the 600-mm wide gutter remained approximately constant as the flow rates increased. The average channel velocity in the RHS did not significantly change. However, the Froude number in the RHS tended to decrease at flow rates greater than 20 L/s.
Figure 13 shows that the gutter channel flow velocity on the LHS of the outlet varied significantly with varying gutter widths. The channel flow velocity in the LHS of the 600-mm wide gutter was between 67% and 78% lower than the other three gutter widths. This is in contrast to the gutter channel flow velocities on the RHS of the outlet (Figure 14) where there was only a 7% to 34% variation between the four gutter widths.
Observation of the flow characteristics within the gutter showed that the volume of water flowing in the RHS of the gutter appeared to be consistently greater than the volume flowing in the LHS at all flow rates. This was to be expected as the total gutter length on the RHS was approximately three times as long as the LHS. Although small increases in water depths in the RHS with increasing flow rate were noticeable, the increase in the flow velocities is clearly evident from Figures 13 and 14.
Equal gutter lengths
Figure 16 shows the difference in gutter water depths in the 600-mm wide gutter when using the whole gutter length (32.00 m) compared to the depths when the gutter lengths were equal (39 pans each side, total gutter length = 14.86 m). The flow rate for both tests was 15 L/s. The results for the 20 L/s and 29 L/s tests are shown in Figures 17 and 18, respectively.
Comparison of gutter water depths in a 600-mm wide gutter at 20 L/s. Comparison of gutter water depths in a 600-mm wide gutter at 29 L/s.
The different water levels shown in Figures 16–18 demonstrate that the length of gutter on each side of an outlet has a significant influence on the water depth above a siphonic outlet. The experimental results showed that for the same outlet flow rate, the depth of water in the gutter was between 9% and 211% greater when the full 32.00-m long gutter with unequal lengths each side of the outlet (7.43 m LHS and 24.53 m RHS, respectively) was utilised than when the gutter lengths were equal (7.43 m each side; total length = 14.86 m).
It is clear from Figures 16–18 that the gutter water depths on both sides of the single outlet were always greater when the unequal gutter lengths (7.43 m LHS and 24.53 m RHS) were utilised than when the gutter lengths were equal (7.43 m LHS and 7.43 m RHS). This could provide some explanation for the differences in the water depth results obtained by May, 8 Bramhall and Saul 6 and Wright et al. 7 as the gutter lengths of their experimental models were all different. This study has indicated a significant influence of gutter length on water depth, but this should be examined through further research, perhaps using a wider range of gutter widths and outlet positions.
There are clearly many variables that affect the depth of water above the outlets of siphonic roof drainage systems. These include outlet flow rates, gutter widths, the length of gutter on each side of outlets and gutter flow velocities, which are in turn influenced by gutter slope and system pressures.
Conclusions
The effects that gutter geometry have on siphonic outlet performance were investigated. It was found that both the maximum flow rate through the single 110-mm diameter outlet and the gutter water depth above the outlet varied depending on the width of the gutter. However, no direct relationship between flow rate, gutter water depth and the gutter width was found for the geometries used in this investigation.
Generally, the greatest water depths for the different flow rates were found to occur in the 400-mm wide gutter and the lowest water depths were found to occur in the 300-mm wide gutter. Gutter flow velocities on either side of an outlet were also found to vary significantly with varying gutter widths. Decreasing Froude numbers accompanied by increasing water depths indicated drowning of the outlets under certain conditions. These results suggest that there may be optimum gutter widths for which different siphonic outlets may perform more efficiently. The results also suggest that the degree of aeration may likewise be influenced by the outlet geometry. However, further research would be required to investigate these relationships in more detail.
This research study has found that for the same flow rate, varying the length of gutter on either side of a siphonic outlet strongly influences the depth of water along the gutter and above the outlet. The depth of water in the gutter was found to be between 9% and 211% greater when the full 32.00 m gutter length (7.43 m LHS and 24.53 m RHS) was utilised than when the gutter lengths were equal (7.43 m each side).
These results suggest that there may be an optimum gutter width and length for which different siphonic outlets may perform more efficiently. While the experimental procedures presented in this paper have focused on steady flow situations, there is still further research required to understand the influence of unsteady flow conditions and particularly the priming process.
Footnotes
Funding
The authors would like to thank the members of Standards Australia Committee WS-014-03-02 – Siphonic Systems for their support and advice throughout this research project. The researchers are also grateful for funding provided by Syfon Systems Ltd (Melbourne) during the period 2006 to 2012.