Despite the significant improvements over the past 15 years in understanding the economic management of real losses on distribution systems, little information is available on the true extent of transmission mains leakage, and how best to quantify and manage it. For decades, it has been common knowledge among water professionals that surveying transmission mains for unreported bursts (also referred to as leaks) using conventional technology won?t turn up many.? But as to the question of why not, there is no such consensus.
The simplest explanation has always been that if we aren?t finding them, they must simply not be there. The bursts that do form on transmission mains tend to be extremely large, often between 5 and 10 m3/hour, and it might be assumed that bursts of this size will nearly always surface immediately. Another explanation has persisted, however, that unreported bursts do exist on large diameter transmission mains, but aren?t detectable with previously available technology.
The suspicion that transmission mains do sustain significant unreported leakage has remained strong enough to spur the growth of a niche market for transmission main leak detection technologies.? Over the past decade, several such technologies have been developed, and have now been used extensively. The results of these surveys allow new insights into the frequency of unreported transmission main bursts and allows us to settle the question once and for all of whether they can sustain significant levels of real losses.
Survey Methods
Transmission mains present difficulties to the conventional acoustic correlation approach to leak detection. Sound waves attenuate more quickly as diameters increase, meaning that the larger the pipe, the closer together the access points need to be for these methods to be effective. Conversely, transmission mains tend to have far fewer service connections, valves or hydrants available for use as connection points; a connection point every 1,000 meters is quite typical on transmission mains, as compared to perhaps every 50 meters on distribution mains. To make acoustic leak detection work in these environments, either the space between listening points must be reduced to mere meters, or maximum effective separation of listening points must be increased to kilometers.
Over the past decade, both resolutions to this problem have in fact been brought to market. The first approach is to bring the acoustic sensor to the source of the sound, by running tethered or free-swimming acoustic sensors through the pipeline.? This offers unmatched sensitivity to small bursts, however is quite labour and cost intensive (typically around $25,000 per km). The second is to develop better sensors, algorithms and methods to allow those sounds that do remain after long distances to be detected and isolated, which has resulted in the development of transmission main correlators. These devices offer a less expensive approach to transmission main leak detection (around $10,000 per km), but at the cost of limited sensitivity.
Both approaches have gained significant traction in the industry, and both have their merits in certain circumstances, with both the particular goals of the utility and the nature of the network playing into the considerations.
Frequency of Unreported Bursts
The following analysis uses information and figures provided by Pure Technologies Inc., and WRc plc. The results of over 3,000 km of international inline survey data were considered:
?Inline surveys avoid the acoustic attenuation problem completely, so the data should provide a sound picture of the frequency of detectable unreported bursts. The picture it paints, however, is not a pretty one. The surveys revealed a range of 22 to 166 unreported bursts per 100 km, with an average of 92 per 100 km.
Interestingly, irrespective of differences in pipe material and geographic location, older groups of pipe seemed to have higher rates of unreported bursts. A plot of the average age for the various pipe groupings versus the unreported burst prevalence, as shown in Figure 1, confirms that this is a relatively consistent trend. The strong correlation of unreported bursts located with pipe age raises an intriguing possibility ? unreported bursts in the mains surveyed are in fact forming at a relatively consistent pace of around 1.6 bursts per 100 km per year (1 per 100 miles/year), and accumulating as a backlog over the decades.
The figure of 1.6 unreported bursts per 100 km per year can be compared with the default used for rate of accumulation of unreported bursts for distribution mains in the equation for Unavoidable Annual Real Losses (UARL), using the Bursts and Background Estimate (BABE) methodology which lies at the root of the modern water audit approach. This expected rate of 0.65 bursts per 100 km per year at 50 meters pressure represents 5 percent of the assumed burst frequency of 13 per 100 km/year for well-maintained distribution mains in good condition, and excludes leaks less than 500 litres/hour.
Correcting for the impact of pressures on burst frequency to a more typical 80 meters for transmission mains brings the UARL prediction up to around 1 unreported burst per 100 km per year (shown in blue on Figure 1). This suggests that not only do unreported bursts occur at a similar rate on transmission mains as on distribution mains, but also that the majority of transmission mains bursts, if not reported immediately, will in fact never be reported, continuing to flow for the lifetime of the pipe. This is not to say that none will ever be reported.? Bursts of this size are capable of washing out tremendous amounts of soil, creating large underground voids.
Regarding pipe material, steel and concrete mains had fewer than 50 per 100 km surveyed, and cast iron 166 per 100 km surveyed. A trend was visible on diameters with smaller pipes, generally showing higher incidence of unreported bursts, but with the trend reversing below 600 mm. The lower unreported burst incidence for diameters below 600 mm may be an indication that conventional technologies have been detecting them more effectively, leaving fewer remaining for inline surveys to find. The extent to which these figures were influenced by region, pipe material and diameter (or years elapsed since any previous checks), could not be identified due to such data being generally unavailable.
Using Data for Improved Management
The analysis of unreported transmission mains burst frequency (1.6 per 100 km/year) and average flow rate (7.7 m3/hr) allows an average rate of rise (RR) of real losses due to unreported bursts on the transmission mains covered by these data to be assessed as 0.016 x 7.7 = 0.12 m3/hour per km of mains, per year of pipe age (or per year since previous intervention). For example, 80-year-old transmission mains that have never been checked could accumulate around 10 m3/km/hour of leakage from unreported bursts.
Along with the value of water, the rate of rise is a critical factor required to calculate the Economic Intervention Frequency for transmission mains. The Economic Intervention Frequency is the frequency with which a distribution system or transmission system should be proactively surveyed to minimize the total combined costs of both the lost water and the cost of the surveys. The cost of intervention on transmission mains can indeed be quite high. The technology and expertise required to conduct these surveys is far more involved than surveying distribution networks. One would therefore expect that the Economic Intervention Frequency (EIF) for transmission mains will be at longer intervals than the one to two years typical of distribution networks with unreported bursts on both mains and services, and typical survey costs of $100 to $300 per km of mains.
Because the EIF is calculated using the square root of the rate of rise, cost of intervention, and variable cost of water, the calculation is not particularly sensitive to variation in any of these values. However, whilst cost of intervention (CI) and variable cost of water (CV) are not too difficult to assess, rate of rise for individual lengths of transmission mains can be expected to vary widely, particularly for small transmission main systems.
Transmission mains do indeed accumulate unreported bursts at rates similar to distribution mains. Interventions on transmission mains are rather more expensive than on distribution networks, and ought to be undertaken less frequently ? perhaps every 5 to 10 years rather than every 1 to 2 years.? However, most transmission main networks have never been properly surveyed, and have likely accumulated a massive backlog of unreported bursts. Simply put, bursts will not heal themselves!
The study has also highlighted the importance of the routine recording age of mains (or date since mains were last checked) and operating pressure, for inclusion in analysis of results and ongoing research by the authors to improve of targeting of transmission mains for both initial and successive interventions.
Kevin Laven is the executive vice president of Echologics Engineering, based in Mississauga, Ontario, Canada. Allan Lambert is managing director of Water Loss Research & Analysis Ltd.
