The social and environmental effects due to leakage are also a matter of concern. For example, up to 4 million holes are dug in the UK each year in order to install or repair buried service pipes and cables. Recently, a survey on the costs of this installation/repair work estimated that street works cost about ��7 bn in losses for the UK government income annually; ��5.5 bn are due to social costs and ��1.5 bn is due to damage [1].Acoustic techniques have been used for many years in the water industry to detect leaks [2], and more recently they have been applied to locate underground pipes [3] and blockages (sediment depositions) in pipe networks [4]. Correlation techniques have been in common use for water leak detection over the last 30 years [5].

In general, these techniques work well in metal pipes, but their effectiveness in plastic pipes is limited [6]. Thus, the specific problem of detecting leaks in plastic pipes using acoustics has recently been receiving increasing attention by the research community. There are two fundamental issues that affect leak detection in plastic pipes: the first is that there is considerably more uncertainty in the noise propagation speed for plastic pipes (which needs to be known a priori for acoustic methods to be effective); and the second, which is more important, is that leak noise does not propagate as far in plastic pipes as it does in metal pipes [7]. Hunaidi and Chu [8] have described the frequency content present in leak signals measured on a bespoke buried plastic pipe rig located in Canada. Gao et al.

[9] have also used the data collected from this rig to gain physical insight into the problems by comparing experimental results with predictions from simple models of the correlation function in plastic pipes due to leaks.Although there is a body of work in the literature on leak detection using acoustic methods in plastic water distribution pipes, for example [5�C14], apart from [11], there is no work in which there is a direct comparison between the effectiveness of correlation for leak detection using measurements of acoustic pressure, velocity or acceleration. Reference [11] describes a theoretical study on the different types of sensors and how they combine with the pipe to act as a filter of the leak noise.

The aim Batimastat of this paper is to validate these findings by carrying out an experimental study in a bespoke test rig in which simultaneous measurements using hydrophones (acoustic pressure), geophones (velocity) and accelerometers (acceleration) were made. Moreover, a quality measure for the data is proposed and tested experimentally as a metric of the prominence of the peak in the cross-correlation function related to the leak noise. Two sets of data are presented, one for a strong leak where there was good signal to noise ratio, and one for a weak leak where this was not the case.