measuring flows in partially-filled pipes in siphonic … yan qu.pdf · measuring flows in...

ORIGINAL ARTICLE

MAPAN - Journal of Metrology Society of India, Vol. 26, No. 4, 2011; pp. 315-327

Measuring Flows in Partially-Filled Pipes in Siphonic Roof Drainage Systems

* 1YING YAN QU , TERRY LUCKE and SIMON BEECHAMSchool of Natural and Built Environments, University of South Australia, Mawson Lakes Campus,

Mawson Lakes, Australia, 50951School of Science and Education, University of Sunshine Coast, Queensland, Sippy Downs Campus,

Maroochydore DC, Queensland 4558*e-mail: [email protected]

[Received: 16.12.2010 ; Revised: 15.08.2011 ; Accepted: 16.08.2011]

AbstractWhile a variety of flow measurement devices are available to measure the flow of water through closed pipe systems, these devices generally only function correctly when the pipes are completely full of water. Accurate measurement of water flows in partially-filled pipes is extremely difficult. In siphonic drainage systems, this problem is further compounded by the unsteady flow conditions that occur in the pipework during the priming process. This has been a major obstacle to understanding the performance of these systems in practice. In order to accurately model the priming process in multi-outlet siphonic roof drainage systems, a method of estimating the instantaneous flows through the partially-filled individual pipes needs to be developed. This paper describes an experimental method of determining flows in partially-filled pipes using a propeller-type current meter to measure flow velocity and a pressure transducer to measure water depth and a modified version of the continuity equation. A computational model is presented which estimates the unsteady flows passing through partially-filled pipework. Overall, the experimental results are promising and correspond well with the model. The results of this study will ultimately be used to develop an unsteady flow model of the priming process in multi-outlet siphonic roof drainage systems.

315

1. Introduction design capacity. Through the use of specially designed gutter outlets, air is purged from the

Siphonic roof drainage is a highly efficient type pipework and the pipes quickly fill with water. This of drainage system that is particularly suitable for process is generally referred to as priming. Once large buildings and other structures that are taller primed, the pipes then experience sub-atmospheric than approximately four metres in height. These pressure and the driving head is effectively the systems were first developed in the 1970s by Ebeling difference in level between the water in the gutter and Sommerhein in Scandinavia [1]. Unlike and the discharge point, which is usually near traditional drainage systems, the pipework of ground level. This causes significant increases in siphonic systems are designed to flow full at their both the flow velocity and volumetric flow

compared to traditional system [2]. © Metrology Society of India, All rights reserved 2011.

Siphonic systems are universally designed University, and extensive testing of our own, the using peak flows rather than full hydrographs. This system described in the paper was developed means that the performance of the system up to and specifically for measuring these very unusual flow after peak flow conditions are attained are conditions. unknown. This has long been a major issue in the roof drainage industry and several research groups Due to the unusual flow conditions, the existing have attempted to address this problem (notably at methods of open channel flow measurement are not Heriot-Watt University in Scotland and at the adequate for measuring the flow conditions that University of South Australia in Australia). More occur in part-filled pipes in siphonic roof systems, particularly, siphonic outlets are balanced only for particularly when the system is approaching the peak flow conditions and there is no guarantee that primed state. In such situations the relative pressure they remain so for lower discharge rates. This means in certain parts of the pipe system can become that designers could face conditions of localised negative while other parts are still experiencing ponding or even gutter flooding. It could also mean part-full pipe flow conditions. Thus the choice of that sub-optimised outlets could suck air thereby apparatus for such systems not only depends on the prematurely breaking the siphon. local flow conditions but also on the flow conditions

within the entire system.The theory and hydraulic performance of

siphonic systems experiencing pipe-full, steady, Siphonic roof drainage systems are designed to flow conditions is well understood and has been operate with full-pipe flow. The design procedure studied extensively [1, 3-6]. However, the unsteady for siphonic roof drainage systems therefore focuses flow conditions that occur in siphonic systems on single-phase water flow. For the maximum during the priming process, when the pipes are only capacity design of a siphonic system this procedure partially-filled, are still largely unknown. In order to is sufficient [1, 9]. However, as previously accurately model the priming process in multi- mentioned, a method of estimating the outlet siphonic roof drainage systems, a method of instantaneous flows that occur in the partially-filled estimating the instantaneous flows in the individual individual pipes during the priming stage is also pipes needs to be developed for conditions when required to fully understand the complete siphonic those pipes are flowing part-full. drainage process. Because of the large size of

siphonic roof drainage systems it is not practical to The most appropriate choice of a velocity or implement expensive techniques such as laser

flow meter to measure liquid flows in open-channels Doppler velocimetry.depends on the purpose of the measurement and the accuracy required. Properly designed flow As part of this study, a current meter and a measurement systems need to be compatible with pressure transducer were installed inside a 150mm the process or fluid they are measuring [7, 8]. They diameter acrylic pipe that was subjected to a range of must also be capable of producing the accuracy and flows and flow conditions. The instantaneous repeatability that are most appropriate for the readings from the current meter and pressure particular application. As with any flow transducer during this flow testing stage were measurement application, it is necessary to consider observed and recorded. The two instruments were the operating constraints in order to be able to select then calibrated against known flows to produce the most appropriate method to use. In the case of calibration curves. This paper presents results of an measuring open channel flows in siphonic systems, experimental method to estimate flows in partially-it is necessary to develop a system that can operate in filled pipes using instantaneous readings from a small diameter part-filled pipes that convey current meter and a pressure transducer in unsteady, highly aerated flows. After extensive conjunction with the equations of a set of calibration searching, consultation with international experts curves. such as Professor Adrian Saul of Sheffield

316

Ying Yan Qu, Terry Lucke and Simon Beecham

2. Siphonic Roof Drainage Systems system. Once each tail pipe begins to prime, the horizontal collector pipe will begin to fill at an

2.1 Performance of Siphonic System increased rate. Flow will be moving from sub-critical to supercritical conditions as the tail pipes prime. In

A typical layout of a siphonic system is shown in free surface flow, supercritical conditions occurs Fig. 1 with four outlets connected by tailpipes to a when the flow velocity is larger than the wave horizontal collector pipe and discharging to ground velocity, while subcritical conditions occur when the level via a single vertical downpipe. The outlets in a wave velocity is larger than the flow velocity. siphonic system are specially designed to restrict the Transitional flow occurs between these two states. entry of air and smooth the flow into the tail pipes. When full bore flow conditions reach the vertical The pipe which connects the roof outlet to the section of the downpipe, the mass of water collecting horizontal collector pipe is known as the “tail pipe”. in the vertical pipe causes the depressurisation of the The tailpipe plays a vital role in the priming process; system. The system is then fully primed [3]. if the tail pipe itself does not prime the system driving head will be limited to the flow depth in the 2.2 Current Research Investigation gutter and the operating capacity will be substantially curtailed. The Bernoulli equation in conjunction with the

Colebrook-White equation is often recommended as Priming of the tailpipe also increases the flow the basis to predict the flow capacity of a siphonic

into the essentially empty downstream pipe. It is system but it does not follow that the system will also the location that the proposed measuring automatically be able to self-prime and develop a method will be adopted in the multi-outlet siphonic satisfactory siphonic action.

Fig. 1. Typical layout of a siphonic system


317

For a siphonic system to prime, flow velocities in (i) the design flows are factored up by 10%; the pipes during the filling stage need to be high

(ii) the minimum velocity in tailpipes and enough to create turbulence and cause local sealing

horizontal pipe (longer than 1m) should not be of section of pipe. If the flow velocities are too low, it

less than 1.0 m/s under design conditions; is possible that a system may never prime or that it will not be able to respond quickly enough to deal (iii) the minimum allowable velocity in vertical with high-intensity storms of short duration. downpipes should not be less than 2.2 m/s Application of suitable minimum flow velocities in under design conditions; design will help ensure that siphonic systems are

(iv) the calculated filling time of a system should not capable of priming. Recent work by Arthur and

be greater than 60 seconds, unless suitable Swaffield [4] had been undertaken to develop a

measures are taken to safely store the excess numerical flow model of siphonic systems that has

runoff occurring during the design storm the capability to simulate the filling process, but this

without flooding or leakage; andis currently a research tool and is not used for routine design of systems. (v) pressures in siphonic systems under design

conditions should be lower than -7.8 m water May [10] investigated safety factors for siphonic head below atmospheric pressure.

systems and recommended that the systems should be designed by ensuring that: These are design criteria for siphonic roof

drainage systems during priming. However, how

Fig. 2. Siphonic system tailpipe

318


the system performs before priming is still largely uncertainty about the exact hydraulic design criteria unknown. that should be applied to ensure satisfactory

operation of siphonic systems. It is the purpose of Despite the data and experience that have been this research investigation to better understand the

accumulated, it remains the case that few direct filling process up to priming. As mentioned in measurements have been made on large siphonic Section 2.1, the proposed method will be installed in installations where the flow conditions approach horizontal tail pipes within the siphonic system (Fig. some of the limits beyond which performance 2) as tail pipes play a vital role in the priming would be unsatisfactory. Experience gained from process. If the flow is measured in this way in each the large number of installations existing around the tailpipe it is a simple exercise to combine the flows to world can help in identifying satisfactory design predict the total discharge in the downpipe. criteria, but it is not possible to determine how many

3. Experimental Set-upof the systems have actually experienced flow rates at or near the design capacity and how many of them

One of the major challenges in estimating flows would then have operated close to the limiting flow in partially-filled pipes is to accurately determine conditions. Like conventional roof drainage the flow velocity, particularly at low flow velocities. systems, some siphonic systems have failed in use Low flow velocities are usually encountered in and have not provided the expected degree of siphonic system pipes during the initial stages of a protection against flooding, but it has not always rainfall event before priming. The average velocity been possible to determine with certainty the in these early stages is often below 0.1 m/s. This reasons for failure. requires a measurement system that can operate accurately at these very low flow velocities. In It follows that there is currently some


Fig. 3. Calibration testing experimental apparatus

319

addition, a measurement system must also be able to pressure transducer was developed.operate in the corresponding low flow depths that also occur in the initial stages of a rainfall event. A variable slope testing rig was constructed for

the experimental work presented in this paper. The Various methods and hydrometric instruments rig consisted of a 4400 mm long, 150 mm diameter

are available to measure flows in open-channels. acrylic pipe, an OTT C2 current meter, a high-One of the most reliable measuring devices for flow accuracy pressure transducer (BCM 430S) and an 80 velocity measurements in natural and artificial mm diameter (ABB MagMaster) electromagnetic channels is the propeller type current meter. This flow meter (EFM). The diameter of the propeller of produces reliable measurement results in both high the OTT C2 meter is 50 mm and it also has a 50 mm and low velocity flow and depth situations. In this pitch. The propeller follows a cosine law with an study we used an OTT C2 current meter, which is accuracy of +/- 1 % of the measured value. The based on the simple principle that the rotational pressure range of the BCM 430S pressure sensor is

2from -3000 to 7000 N/m (gauge), and the combined speed of the propeller is proportional to the local

error is +/-0.5 %. The 80 mm EFM came with a flow velocity [11]. The rotational speed is

factory calibration of +/-0.5 % and the measurement determined by the number of revolutions per second

accuracy is consistently to 0.15 % according to the made by the propeller and this is then shown on a

manufacturer [13].numerical display. In this study, a flow measurement method using a combination of a

A schematic diagram of the test rig is shown in propeller type current meter and a piezoelectric

320


Fig. 4. Current meter and pressure transducer installed in a 150 mm diameter test pipe

Fig. 3. The instrument readings from the current slope is steep enough, super-critical flow conditions meter, transducer and EFM were recorded by a will develop producing a correspondingly shallow DT82 Data logger that was connected to a personal flow depth. Conversely, if the pipe slope is very computer. The EFM was located at the upstream end mild, deeper, sub-critical flow conditions will of the pipe (Fig. 3) between two gate valves in a develop for the same flow. It was therefore section of 80 mm diameter pipe. The high accuracy necessary to calibrate the test rig under an extensive of the EFM for pipe-full flow conditions was verified range of discharge and depth conditions (Fig. 3). separately. The apparatus was therefore designed so that the flow conditions at the EFM were always In order to calibrate the test rig, the flow pipe-full. Water was supplied from a reticulated behaviour in the acrylic pipe over a range of known supply connected to an underground reservoir. The flows was observed and recorded. The flows

3calibrated in the test rig were all between 0.003 m /s gate valves shown in Fig. 3 were used to control the 3 3flow through the system. An 80 mm to 150 mm and 0.015 m /s, increasing in 0.001 m /s increments.

reducer was used downstream from valve 2 to These calibration flows for each test were verified connect the 150 mm diameter acrylic test pipe. This using the EFM. ensured the development of open-channel flow conditions in the acrylic test pipe. A specially A variety of steady flow regimes was generated

3designed adjustable weir was installed at the for each test flow between 0.003 m /s and 0.015 3downstream discharge point of the test pipe. This m /s. A range of water depths between empty and

was used to control the depth of water in the test full flow were produced in the pipe during the pipe. calibration process to represent all probable steady

flow conditions in the pipe for each flow. These The OTT C2 current meter, with a 50 mm individual steady flows included both sub-critical

diameter anodized aluminium propeller fitted, was and super-critical flow conditions to mimic the flow installed inside the acrylic pipe. This was held in conditions in siphonic roof systems. To produce sub-place by a stainless steel rod as shown in Fig. 4. The critical and super-critical flow conditions for each height of the propeller was set at 1 mm above the single flow, the test rig was first set at a steep slope to invert of the acrylic pipe to ensure the current meter produce super-critical flow, and then the slope was would rotate even in the smallest flows but would gradually changed and the weir height was adjusted not be affected by sediments in the flow. Although until sub-critical flow conditions were achieved. the highest flow velocities in open-channels are Flow conditions that produced hydraulic jumps generally found at a position of approximately one above the transducer were avoided.third of the water level depth below the water surface, in this study, only the velocity of the flow in A minimum of eight different water depths the vicinity of the current meter was required. The above the transducer were tested for each flow. The pressure transducer was held in place via a threaded average number of revolutions made by the current boss that was fitted to the underside of the 150 mm meter propeller for each of the different water diameter pipe, approximately 20 mm upstream depths for each test flow was recorded. Each test was from the tip of the propeller (Fig. 4). The pressure repeated three times to ensure accuracy and transducer converted the pressure to an electrical repeatability. The data for each flow was plotted and signal (using piezoelectric theory) and was a modified log function trendline for the line of best calibrated to the depth of the water in the pipe [12]. fit for the measured data was added. The correlation

between the data and the trendline was generally 24. Methodology very close, with the lowest R value being 0.83. The

equations for the trendlines for each data set were For a constant flow in a partially-filled pipe, then used to develop a set of calibration curves for

there can be many different combinations of flow this test rig (Fig. 5).velocities and water depths. For example, if the pipe


321

The algorithm for flow estimation was derived the two trendline equations to determine the flow. from the modified continuity equation. This For example, for the black point in Figure 5, the modified continuity equation included two model will first locate the two closest trendlines and variables, namely the current meter generated interpolate between the two trendlines to determine revolution number and the converted water level the flow. In this case, it will be approximately 0.0035

3m /s. from the pressure transducer. However, this equation is not the traditional continuity equation

In order to test the accuracy of the computer but instead uses the calibration curves shown in Fig.

model, a range of flows and flow conditions was 5 to estimate instantaneous flow rates based on these

again produced through the test rig. The estimated two variables. The calibration curves shown in Fig. 5

flows generated by the computer model were then were then used to develop a numerical computer

compared to the actual flows measured using the model that estimates instantaneous flows based on

EFM. The results of this comparison are presented in the current meter and the pressure transducer

the following section.output. For each one second time interval, the model first counts the number of revolutions completed by

5. Resultsthe current meter and establishes where that number lies on the X-axis of the calibration curve

A comparison between the flows estimated by chart (Fig. 5). The model then examines the water the computer model and the flows measured using level corresponding to the transducer reading in the EFM was undertaken using separate order to determine between which two trendlines experimental tests to the ones described for the that value lies on the Y-axis. The model finally uses a calibration procedure. Figure 6 shows both the weighted average method to interpolate between

322


3Fig. 5. Calibration curves for flows from 0.003 to 0.015 m /s

predicted and the measured flows over a 30 minute data sets was undertaken using a paired t-test 3period. The initial flow was 0.003 m /s and this was analysis. The null hypothesis was that the Measured

3 Q = Predicted Q and the test significance level was increased by 0.001 m /s intervals every 5 minutes up 3 set at 5%. The analysis results gave a P value of 0.876 to a maximum flow of 0.015 m /s. The flow was then

3 3 which means there was no significant difference decreased from 0.015 m /s to 0.003 m /s over a between the flows. period of approximately 4 minutes.

6. Discussion A safety factor for siphonic systems as mentioned in Section 2.2 should be provided by

Figure 6 shows that the main variation between increasing the design rates of runoff by 10%; these the measured and predicted flow results is when the increased rates should also be used for sizing gutters flows are either very low or when they are drained by siphonic systems to allow for possible approaching the maximum flows for the test pipe. partial clogging of outlets. The minimum velocity in These variations were possibly due to the difficulty tailpipes and horizontal pipe (longer than 1m) in obtaining a constant water level reading at low should not be less than 1.0 m/s under design flows attributable to the oscillation effects or caused conditions. The minimum allowable velocity in by the turbulent flow conditions experienced at high vertical downpipes should not be less than 2.2 m/s flows. Figure 6 shows that the measured and under design conditions. The calculated filling time predicted flow results were more closely matched of a system should not be greater than 60 seconds, for the intermediate flows. This was due to the more unless suitable measures are taken to safely store the stable flow conditions at these flows. The variation excess runoff occurring during the design storm between the results may also mean that more without flooding or leakage. Pressures in siphonic variables need to be included in the computer systems under design conditions should be lower model. This will be examined in a further research than -7.8 m water head below atmospheric pressure. study.

The initial results of this flow measurement To quantify the variation between the measured technique used in this study suggests that the

and estimated flows, a statistical analysis of the two accuracy of the computer model developed in this


Fig. 6. Comparison of measured and estimated flows

323

study would satisfactorily estimate the flows when flows in an experimental, multi-outlet, siphonic the siphonic pipework is flowing part-full in the drainage system during all stages of its operation. experimental test rig. Providing the flow measuring devices proposed in

this paper are installed in the pipework of each of the 6.1. Disturbance Testing outlets in the experimental multi-outlet siphonic

system, then the disturbance effects caused by the The experimental measuring method presented devices will be balanced and the hydraulic

in this paper is intended to estimate flows in interaction between the outlets will not be affected. partially-filled pipes using a current meter and a pressure transducer. The method has been shown to Although the disturbance effects of having flow produce satisfactory results. However, it is clear that measuring devices installed in the pipework will not any measuring device that intrudes into the fluid influence the current experimental results, it was will cause some degree of disturbance of the flow. decided to quantify these effects for future reference While this disturbance will probably not have any and testing. A new disturbance test rig was significant effect on the overall flow, it could constructed for this purpose. The configuration for possibly cause a backing-up effect on the upstream this new disturbance rig was similar to the one flow conditions resulting in an increase in the water shown in Fig. 3. However, the 150 mm diameter pipe depth. However, for the experiments described in was replaced with an 82 mm diameter pipe to more this paper, this backwater effect was not found to closely represent the smaller diameter pipes that are cause any significant detrimental effects on the flow commonly used in multi-outlet siphonic drainage measurement technique. The overall long-term aim systems. The current meter was installed inside the of the research presented in this paper is to develop a 82mm diameter pipe for the disturbance testing. suitable method of estimating the instantaneous

Fig. 7. Schematic of disturbance test rig

324


Two additional piezoelectric pressure In order to measure the disturbance effects that transducers were installed in the pipe in order to the current meter had on the flow conditions, a slot measure the disturbance effects from the current was cut in the top of the pipe. This allowed the meter meter on both the upstream and the downstream to be lifted in and out of the flow, as shown in Fig. 8. water levels. One of the additional pressure For each of the nine steady flows tested, the transducers was installed one metre upstream from procedure was as follows:the position of the current meter and the original transducer (midstream). The other additional 1. The three pressure transducer readings were transducer was installed one metre downstream first recorded when the pipe was empty to from the midstream position. The positions of the identify one boundary condition (water level = 0 three transducers are shown in Fig. 7. mm);

2. The pipe was then completely filled with water A total of nine different steady flow and the three transducer readings were

configurations were tested in the disturbance test recorded to identify the other boundary rig. The nine configurations comprised a condition (water level = 82 mm); combination of three different water depths and 3. The required flow was set up in the pipe without three different flow velocities. The three flow depths the current meter installed; tested were 33%, 66% and 100% respectively, of the 4. The three transducer readings (and water levels) pipe's 82 mm internal diameter. The three flow were recorded for this flow (without the current velocities were approximately classified as slow, meter inserted); medium and fast. The slow flow was sub-critical, the 5. Without changing any flow conditions, the fast flow was super-critical and the medium flow current meter was then inserted into the flow was transitional. through the slot in the top of the pipe;


Fig. 8. Disturbance test rig midstream flow conditions

325

7. Conclusion6. The pressure transducer readings (and water levels) were again recorded for this flow (with

The purpose of this experiment is to develop a the current meter inserted); andmeasurement technique that is able to determine 7. The test was repeated for the next flow.flow rates in siphonic roof drainage systems. The results show that this method is suitable for The three pressure transducer readings (and predicting flow rates in full-scale siphonic roof corresponding water levels above each transducer) systems. This measuring method will be used later were recorded by a data logger and transferred to a in a new multi-outlet system. The predicted flow PC for each of the nine steady flows tested. The rates from this multi-outlet system will then be water level measurements above each transducer compared with the output from a numerical model with and without the current meter installed were for calibration and verification purposes. then compared. The differences in water level are

shown in Table 1. In order to accurately model the priming

process in multi-outlet siphonic roof drainage The results in Table 1 show that generally, the systems, a method of estimating the instantaneous largest differences in the water levels occurred flows through the partially-filled individual pipes above the midstream transducer. This was to be needs to be developed. This paper presents the expected as the propeller of the current meter was results of an experimental investigation to only 20 mm upstream of the midstream transducer. determine the flows in partially-filled pipes using a There was no apparent consistent trend in any of the propeller-type current meter to measure flow water level changes. It is believed that this was due velocity, a pressure transducer to measure water to the wave patterns that developed within the depth and a modified version of the continuity pipework for some flows.equation.

While water level differences of 14% may A numerical model was developed that appear relatively large, in reality the effects of the

estimates instantaneous flows based on the current current meter on the overall flow were negligible. meter and the pressure transducer outputs. In order However, as previously mentioned, as long as the to test the accuracy of the numerical model, a current meter is installed in the pipework for all comparison between the flows estimated by the testing, its effects do not significantly influence the model and the flows measured using an EFM was measured flows.

326


Table 1Water levels differences above each transducer (%)

Transducer position

Water level Downstream Midstream Upstream

Slow Flow 33% 0% 0% 3%

66% 3% 6% 0%

100% 15% 1% 0%

Medium Flow 33% 0% 10% 3%

66% 0% 7% 6%

100% 8% 11% -1%

Fast Flow 33% 0% 14% 5%

66% 0% 14% 5%

100% 11% 14% -6%

and Technology, 23 (2002) 127–141.undertaken. A statistical analysis of the two data sets showed that there were no statistically significant [4] S. Arthur and J.A. Swaffield, Siphonic Roof differences between the flows, at the 5% level. These Drinage System Analysis Utilising experimental results show that the computer model Unsteady Flow Theory, Building and developed in this study can be utilised to Environment, 36 (2001) 939-948.satisfactorily estimate partially-filled flows in the

[5] T. Lucke and S. Beecham, Cavitation, experimental test rig.

Aeration and Negative Pressures in

Siphonic Roof Drainage Systems, Building Although the disturbance effects of the current

Services Engineering Research and meter in the pipework did not influence the current

Technology, 30 (2009) 103-119.experimental results, it was decided to quantify [6] T. Lucke and S. Beecham, Capacity Loss of these effects for future reference and testing. The

Siphonic Roof Drainage Systems Due to disturbance testing results showed that the largest Aeration, Journal of Building Research and difference in the water levels above the transducers

Information, 38 (2010) 206-217.was 14% and this occurred above the midstream transducer. It has been shown that the proposed [7] G.F. Hewitt, Measurement of Two Phase flow measurement method is suitable for use in Flow Parameters, Academic Press London more detailed investigations of multi-outlet (1978), 294. siphonic systems, providing the current meters are

[8] P.P. Kremlevsky, Flowrate Measurement in installed on all outlets.

Multiphase Flows, Begell House USA, (1999), 232.

This method will be further developed in future [9] S. Arthur and J. Swaffield, Siphonic Roof research investigations in order to accurately model

Drainage: Current Understanding, Urban the priming process in siphonic drainage process.Water, 3 (2001), 43-52.

References [10] R. May, Design Criteria for Siphonic Roof Drainage Systems, in Report SR 6542004:

[1] R. May, Design of Conventional and Wallingford, England (2004).Siphonic Roof Drainage Systems, in

[11] C. Rohwer, The Rating and Use of Current Proceedings of Public Health Services in

Meters, in Technical Bulletin 3, Colorado Buildings – Water Supply, Quality and

Agricultural College, USA (1933).Drainage, IWEM Conference, London [12] R.J. Goldstein, ed., Fluid Mechanics (1995).

Measurements, Hemisphere Publishing [2] S. Arthur, G.B. Wright and J.A. Swaffield, corporation: New York (1983), 701.Operational Performance of Siphonic Roof

[13] ABB, MagMaster Electromagnet ic D r a i n a g e S y s t e m s , B u i l d i n g a n d Flowmeter. 2011 Available from: Environment, 40 (2005) 788-796.http://www.abb.com/product/seitp330/[3] G.B. Wright, J.A. Swaffield and S. Arthur, 05d2d070fe77f6f0c1256d32002bffcd.aspx?pThe Performance Characteristics of Multi-roductLanguage=us&country=AU, Outlet Siphonic Rainwater Systems, [accessed: 10th July,2011]Building Services Engineering Research


327

measuring flows in partially-filled pipes in siphonic … yan qu.pdf · measuring flows in...

Documents