Siphonic roof drainage
For many years Geberit International has felt that the way forward has often been to examine previous research and development for knowledge that can be applied to new, sophisticated and exciting products.
Although not a product in the usual sense, the well-established principle of siphonic roof drainage offers an example of this ‘test and re-test’ approach.
When the system was developed a couple of decades ago the big question was whether it would work under extreme rain run-off. Time has shown the answer to be a resounding yes.
Most in the industry did not question whether the system should be examined under minimal run-off – if it could handle peak loads then everything else would be all right.
However, Geberit International engineers have conducted a research program to examine the system’s low run-off performance, and they presented a paper to the CIB W062 2006 Symposium held in Taipei, Taiwan.
Building services and plumbing design personnel are well aware of the many difficulties in achieving a safe, efficient and economical roof drainage system.
A system with very large gutters and down-pipes may be the answer in many instances but is unacceptable in many modern buildings due to aesthetics, space or cost restraints.
As industrial, commercial and supermarket projects continue to grow in scope, alternative solutions are being applied.
About 20 years ago Geberit International was at the forefront of developing a siphonic roof drainage system relying on a designed head pressure partly dependent on the use of relatively small drain lines. Early systems also used the roof design to create a substantial head and volume of water over the surface outlet point. Another feature related to the size of the vertical drain conveying water from the under-roof space to the stormwater drains. Pipelines in the under-roof space were intentionally installed at a flat grade.
These features separately and in combination resulted in reduced space requirements and removed rainwater from roof to underground drains safely efficiently and economically. The greater the rainfall the more effectively the system worked.
Under minimal rainfall the system did not reach its full capability until the roof outlet and horizontal drain lines were filled in order to achieve a siphoning condition. This low flow was evident most of the time, as heavy downpours are the exception rather than the rule. The part-load condition led to air suction and mixing, which created flow activity characterized by a multi-phase phenomenon.
The Geberit researchers’ contention is that the behavior of the water-air mixture should be considered when designing a system, particularly where an accurate estimate of working pressures under part-load conditions is required.
Additionally, study of the mixed flow is important when there is a need to provide for dynamic and acoustic characteristics.
The research was based on a tower with a single outlet fitted to the top level, utilizing pipes of three sizes and two vertical drop lengths. Differing rainfall intensities were simulated with a variable-speed pump. The pipeline configuration replicated as near as possible a typical installation.
The pipe sizes were DN56, DN63 and DN75. The vertical drop lengths were 10m and 5m. Flow rates in the pipelines, expressed in liters per second, were: 2.6, 6.1, 9.0, 11.9 and 13.5. Each pipe size was tested on both vertical drops.
The installation was fitted with a siphonic roof drain (SRD) outlet connected to a short length of vertical pipe which, by use of a 90° bend, connected to a short section of clear tube of the desired diameter. Flow in this section of pipe was observed and filmed using a high-speed camera (640 frames per second).
Transducers were fitted to the vertical line just below the SRD and results relayed to a computer via a pressure sensor. Transducers were connected in another seven locations along the horizontal and vertical drop pipes, including one as close to the base line as possible. Special equipment was obtained, as the task required a high degree of accuracy.
It is relatively straightforward to determine the flow rate when the system is filled with water. This becomes far more difficult when the system is partly filled, and the situation is exacerbated when large volumes of air are trapped.
The flow rate can be determined by tracking fine air bubbles recorded on film or by seeding solid particles with a density similar to that of water. Information relayed to the computer readily shows the water velocity.
Similarly, the pressure at any point in the system can be determined. Readings may range from negative pressure immediately below the connection point to the roof outlet which reduces as the flow traverses the horizontal line then changes dramatically at the bend at the top of the vertical pipe.
Changes in the volume of trapped air can create a temporary block in the flow. Air bubbles can have a substantial effect on the system to the point where they cause excessive dynamic pressure vibration in the piping. This can cause foaming, and the influence of these multi-phase characteristics are evident downstream.
Air effects are readily observed with flows up to about 7L/s. Above this level air bubbles reduce until they disappear at 13L/s, when the SRD is fully primed.
This indicates that various water/air mix velocities should be determined and taken into account when evaluating an SRD system’s performance.
The research paper examines effective flow velocity (Ve) and compares it with the idealised full-flow line. Interestingly, the research determined that the value of Ve was greater for the experimental installation than the idealised full-flow line until the point where the system was fully primed. The researchers attribute this entirely to the effect of trapped air on water velocity.
The value of Ve reaches its maximum at a flow rate much lower than the design limits of the system. This finding is contrary to the previously held prognosis of the idealised full-flow calculation. Additionally, the maximum level of Ve is retained even if the flow rate is increased until the priming condition is attained. The contention is that this is directly related to the formation of an homogeneous bubble distribution in the water, instead of the two-layer configuration of air and water.
The research outlines some associated findings after examining the relationship between the effective velocity maximum (Qs) and the primed flow condition (Qp). The observation has been made throughout the measured cases that Qs/Qp has almost a constant value. In the cases of this study this value was 0.7, regardless of the size of the drain line.
The research summation confirms that the aim was to examine flow conditions – from the multi-phase air/water stage, occurring under part load, to the single-phase water flow under primed conditions – in greater detail than had previously been established. Finally, the paper indicates:
“The flow characteristics under part-load and primed conditions are determined to be qualitatively similar irrespective of piping dimension. One of the important results of this study is the observation of a flow regime where the effective flow velocity in the piping remains constant for a relatively wide range of flow rates. It is assumed that the breaking up of large air pockets into small size dispersed bubbles and their transition to a homogeneously distributed regime is the cause of this unique flow phenomenon. It is also observed that the onset of this flow structure is well related to the ratio of flow rates corresponding to the velocity maxima and primed conditions. This ratio is found to be a constant around Qs/Qp = 0.7 for the cases of this study.”
A detailed examination of the full report may dispel a few previously held opinions as well as alert designers to the need to examine the anticipated flow, velocity and system capacity in the installation they are considering. The research has been done and presented. It is now a matter of identifying how this can be applied to future projects for the benefit of all involved.
For engineers working in North America, ASPE has issued a new Standard: ASPE 45, Plumbing Engineering Design of Siphonic Roof Drainage Systems. The new Standard is available to members for US$50 and non-members for US$75. The book is available through the ASPE website, www.aspe.org.