Particularly within Europe, ever increasing amounts of industrial and commercial roof space are being drained using siphonic roof drainage. One installer claims it system drains over 60 million square meters (646 million square feet) of roof area. This increase in usage is largely due to the many advantages the systems have over "conventional" systems for equivalent sized roof areas.

However, notwithstanding the increasing use of these systems, there are still uncertainties regarding just how these systems operate--particularly during priming. This lack of understanding of system operation means that if a system fails, it is often difficult to appreciate why the failure has occurred. Furthermore, as siphonic systems design necessitates a higher level of expertise than is required for convention systems, the performance of these systems can be more reactive to small inaccuracies and erroneous assumptions. To resolve this, over the past five years there has been an increasing amount of independent research underway attempting to understand how siphonic systems actually perform. This article aims to collate this work, and give an overview of how siphonic rainwater systems operate.

What is Siphonic Roof Drainage?

Conventional roof drainage systems generally consist of open roof outlets connected to vertical rainwater pipes which are designed to operate at atmospheric pressure with a continuous air core. The flow capacity of a conventional system is usually determined by the size of the outlets and by the depth of water at them--normally around 100mm (3.93 inches). Additionally, any "horizontal" pipework must be sized and installed at gradients which ensure both sufficient flow capacity and self-cleansing velocities are attained. An equivalent sized siphonic system can have a significantly higher capacity as the pipework is enabled to flow full--the driving head becomes equal to the vertical height between the roof and the point of discharge.

A siphonic system draining a roof normally consists of specially designed outlets connected to a discharge point at or below ground level, via a pipe network which has been designed to flow full at pressures significantly lower that atmospheric--for specified rainfall conditions. Table 1 illustrates typical hydraulic conditions with both conventional and siphonic roof drainage systems.

The intrinsically higher flow capacity of siphonic systems leads to, for a typical installation, the following benefits:

  • Fewer roof outlets are required.

  • Smaller diameter pipework may be specified.

  • Several outlets may be connected to each vertical discharge pipe.

  • Pressurized flow means there is much more flexibility where pipe routing is concerned.

  • Less concern may be given to attaining self-cleansing velocities.

  • As the pipework is de-pressurized, most of the horizontal pipework runs just below the roof, rather than below ground--this means there is less need for groundworks.

These characteristics enable significant savings, in terms of time and money, to be made in the construction of large industrial or commercial buildings. The need for vertical rainwater pipes inside a building can be eliminated--saving approximately 0.5 m2 (5-6 ft.2) per absent downpipe. This allows greater flexibility in the use of space within open-plan buildings, and presents a means of meeting the requirements of architects for large uncluttered areas in large public structures. Fig. 1 shows an early U.K. flagship project.

Flow Conditions

The flow conditions in siphonic systems vary considerably according to the available inflow. At lower flows, the water and air in the pipes remain separate, with the water flowing along the invert of the horizontal pipes. Therefore, the system behaves in the same way as a conventional one would, with the capacity controlled by the size of the outlets and the head of water above them.

As flow into the system increases, pipes flow full at some points and part-full at others. The behavior of the air remaining in the pipes depends on the velocity and turbulence of the inflow; some air may remain above the water surface but be drawn along by the flow, or else may become entrained into the water as bubbles. If significant amounts of air continue to be drawn into the system through the outlets, perhaps due to low flow depths, it is unlikely that the pipes will be able to flow full along their whole length. The outlets used in siphonic systems are specifically designed to restrict the entry of air and smooth the flow of water into the pipes (Fig. 2 shows a typical outlet).

If a system is suitably sized, the high speed of the water removes the air present in the pipes more quickly than it can enter through the outlets. This causes the pipes to fill and flow full with a bubbly two-phase mixture. The current design methodology assumes that at some point, the depth of water at the outlets will be sufficient to cut off the supply of air completely. If this were to occur, the flow rate is effectively equal to the maximum capacity for the system as the available head between the outlets and the point of discharge is being fully used by the flow in overcoming the hydraulic resistance of the system elements--the system is now, effectively, running at capacity (i.e. any extra rainfall rate may result in overspill).

Design Considerations

Currently, siphonic roof drainage systems are designed to accommodate a specified design storm which fills, and primes, the whole system rapidly with 100% water. This assumption means that the system may be designed easily using elementary steady state hydraulic relationships. The steady flow energy equation is used almost universally (Refs. 1, 2 and 3) as the backbone of the design procedure for siphonic roof drainage systems. This design approach has been used to estimate the flow capacity and pressure distribution for a number of siphonic roof drainage test rigs (Refs. 1 and 4), and has been found to work well for the fully primed state. There are significant calculated variations in pressure throughout the system, which are dependent upon frictional losses through fittings and changes in static height. The discrepancies which exist between these results may be accounted for by variations in air content and inaccuracies in the estimation of the head loss across fittings (Refs. 1, 2 and 5).

Typically, systems are designed to operate at pressures up to 80kN/m2 (11.6 lbf/in2) below ambient atmospheric pressure (Ref. 6). However, such pressures can vary and increase due to:

  • Interaction with underground systems.

  • Sudden partial or total blockage of outlets.

  • Changes in pipework configuration during installation from that designed.

  • Volumes of air entering the system.

Most systems are designed to work slightly beyond the design condition, this extra system capacity is termed the "reserve." For any well-designed and specified system, the design storm will exceed the vast majority of the storms it is expected to meet. Where rainfall events of quite low intensity are encountered, the system will act hydraulically as a "conventional" roof drainage system. However where increasing rainfall intensities are considered, unsteady partial de-pressurization of the system will occur. Tests have shown that this de-pressurization results in substantial amounts of air being drawn into the system (Ref. 7), which can exceed the volume of water inflow. The unsteady nature of the flow regime, which has been observed to be cyclic in nature, leads to varying amounts of noise generation, and structural vibration within the system. The structural vibration, conceivably, could lead to physical failure of a component of the system. Where pressure exceed the strength of the pipewall pipe, implosion is also a possibility.

The unsteady pressure regimes which have been observed to occur within the test rig at Heriot-Watt University (Refs. 1, 6 and 7) are illustrated by the data presented in Table 2 and 3. Table 2 illustrates data where a cyclic pressure history was recorded. The frequency of the cyclic response of the system was found to be related to the rate of inflow, and the lengths of the horizontal and vertical pipework. Table 3 shows that even in instances where the rate of gutter inflow is approaching the measured capacity of the system, the ambient pressures are far from steady. It should be noted that the current design approach does not consider how these pressure fluctuations influence system performance.

Priming Action

Understanding the priming action within a siphonic test rig is of fundamental importance. If, for whatever reason, an installed system cannot prime at the design rate of inflow, the system will fail to meet its design criteria. The priming action described in this section will consider the hydraulic conditions which prevail in a siphonic system where the inflow to the roof gutter rises rapidly to equal, or exceed, the observed inflow capacity of the test rig (i.e. the design condition). The main reported work in analyzing the priming of siphonic systems has been undertaken by Heriot-Watt University using a well instrumented idealized system (Ref. 2) and by HR-Wallingford (Ref. 3).

Initially, flow depth in the gutter increases with storm intensity, resulting in free surface flow in the horizontal pipework and annular vertical stack flow. As the inflow increases further a hydraulic jump forms at the upstream end of the horizontal pipe and gradually moves downstream, eventually generating full bore flow. This traps a volume of air between the jump and the vertical discharge from the roof outlet. Full bore flow propagates along the length of the horizontal pipework downstream of the jump. While the horizontal pipes are running partially filled, flow conditions in the vertical stack are annular. When full bore flow reaches the head of the stack siphonic conditions are established due to the falling mass of water in the vertical stack causing de-pressurization in the upstream pipework and increasing the inflow to the system and full bore flow throughout the horizontal pipework. This reduces the water level in the roof gutter and, if the intensity of the storm is insufficient to maintain through flow, air enters through the roof outlet, which is swept to the head of the stack, resulting in partial re-pressurization of the entire system.

The priming of siphonic systems is described elsewhere (Refs. 1, 2 and 3), and numerical models have been developed which may be used to represent this sequence of events (Ref. 1).

Operational Considerations

A number of siphonic roof drainage systems have failed due to the building operators treating them like conventional systems as far as maintenance is concerned (Ref. 6). This section will focus on issues that are pertinent in this area, and how they should be addressed.

Siphonic outlets are typically between 50mm and 65mm (1.96 and 2.56 inches) diameter and therefore susceptible to blockage and blinding (Fig 3). Leaf guards are fitted to outlets to prevent leaves entering the system, these also very quickly blind. Maintenance of gutters is often neglected, and while on a conventional outlet this may be tolerated to some extent, where a siphonic system is concerned the consequences are considerably more serious.

Detritus such as biological growth, wind blown materials, construction waste and material dropped by birds can rapidly accumulate on a roof, and very quickly restrict the entrance of rainwater to the system. Additionally, CCTV surveys of recently completed buildings (Ref. 6) have identified a considerable amount of construction debris in roof drainage pipes including wire, disposable gloves and rolls of tape--all of which would affect the capacity of siphonic systems significantly.

Regular inspections can only be undertaken if easy access onto the roof is allowed. Access onto a roof is often a secondary consideration and often is designed for infrequent access in the form of a vertical ladder and hatch--with no safe walkway provided across a roof. Rarely is the person cleaning the roof considered, or any facility provided to remove detritus from the roof (debris is not uncommonly swept into the roof outlet). If access is difficult, this increases the possibility that maintenance will be neglected.

At hand-over, a CCTV survey should be required to prove the pipework system is clear, as far as possible, of any obstructions. Initially outlets should be inspected at least six times a year. A maintenance program should be developed based upon the results of the first year of inspections.

When considering overall system performance it is important to consider the influence of the main surface water drain on the capacity of the system. Often sewers will surcharge at a frequency of more the once in every 5 years (U.K.), whereas the roof drainage system may have a design return period of in excess of 30 years. This means that when a storm occurs which approaches the design storm, it is likely that the connecting sewer will be surcharging. It should be ensured that this is taken into account at the design stage.


As this article has illustrated, siphonic roof is now a serious competitor to "conventional" roof drainage systems. This is largely due to the benefit this system offers. Additionally, the growth in reasserting how these systems operate, has led to a higher degree of understanding as to how these systems perform. As research initiatives in this domain progress and interact this can only lead to increased confidence in these systems. To date, the following conclusion may be drawn:

  • Siphonic roof drainage represents a significant advance in technology when compared to conventional systems.

  • Siphonic roof drainage can result in savings in both cost and time during the construction phase.

  • A siphonic roof drainage research community has been established.

  • Test facilities have been established which study differing aspects of system operation.

  • The design of siphonic roof drainage systems requires a higher level of engineering understanding.

  • The existing design method is robust, but may be simplistic.

  • The priming of idealized systems is well understood.

  • Increased instances of installations mean that the longer term maintenance issues are now understood.