Use the video umpire
Flow metering is an inexact science. Analysis techniques are vital to gather the true meaning of data, and to boost the effectiveness of network monitoring, says TUV NEL's Allick McGillivray
In both the short and long term, the UK water industry needs to maintain and manage its supplies. Climate change is making supply more variable while the demand for water continues to rise. The need to understand and control the network has never been more pressing, and metering is seen as a key factor at all stages of the supply process. The Environment Agency requires closer monitoring of extraction, Ofwat requires better balancing and reduced leakage, and domestic metering is seen as a driver in managing demand.
Flow monitoring is an increasingly important part of a water company's business. It is crucial that those using the data from metering understand that it is in itself is an inexact science. And imbalances are as likely to be caused by metering issues as by operational problems. Techniques such as uncertainty analysis and data reconciliation are cost-effective methods of improving the effectiveness of network monitoring and bringing out the true meaning from conflicting data.
Perhaps the best analogy is to think of a flow meter as the referee at a sports match - the sole arbiter of fact and law but still fallible. He has to make quick decisions based on what he can see. He may be some distance from the action, and both teams are trying to influence his decision. In an effort to eliminate mistakes, sports are turning to technology, in the form of Hawkeye ball tracking and the video referee, for a second opinion. The metering equivalent is data analysis using techniques such as uncertainty analysis and data reconciliation.
Just as a referee's decision is influenced by his position relative to the action, so a flow meter's response depends on the way in which it is used. Many meters depend on assumptions about the flow profile, and distortions caused by bends, valves and other fittings upstream will invalidate the assumed profile and so affect the accuracy of the meter. Swirling flow will affect the rotor of a turbine meter and whether it causes an under- or over-reading will depend on the direction of swirl; the sensitivity of an electromagnetic flow meter depends on its orientation relative to the profile distortion and the meter may also be sensitive to water hardness.
TUV NEL recently completed a study on behalf of the DTI to assess the current use of uncertainty analysis in UK industry. This found that the technique was used in all industrial sectors but to widely varying extents. In the water industry, uncertainty analysis of flow measurement was used mainly in the production of figures for the annual water balance. The rigour with which it is applied varies widely, with some companies using only the manufacturers' accuracy claims, while others devote a lot of effort to examining meter history and location to identify the key influences.
In contrast, in the oil and gas industry uncertainty analysis is integral to the business. This is driven principally by the high value of the product and companies simply cannot afford inaccurate flow measurement. Accounting for uncertainty in flow measurement allows them to see the bigger picture - enabling them to calculate financial exposure on fields and make strategic decisions on such issues as investment in metering infrastructure.
Rigorous uncertainty analysis lets a company go beyond simply saying "our distribution input is 950 ± 30Ml/d"; it will identify the main contributors to the uncertainty, in terms both of the key meters and of the sources of it in those meters. The nature of uncertainty is such that only by addressing these aspects can the uncertainty in the system be improved. Detailed uncertainty analysis can therefore ensure that capital expenditure is targeted to where it will produce the most benefit.
As a first, and inexpensive, step to addressing the existing demand and leakage challenges, the water industry would gain real benefits from adopting the practice of the oil and gas industry and applying rigorous uncertainty analysis at the heart of their daily network monitoring procedures. In the last few years, the availability of inexpensive computing power and measurement databases has enabled the development of powerful data analysis techniques that allow networks to be monitored on a daily basis.
Data reconciliation is basically a self-consistency check designed to complement the existing metering infrastructure. A sophisticated statistical technique applies corrections to individual flow (and other) measurements to remove imbalances from nodes in the distribution network. The size of each correction is compared with the expected uncertainty of the measurement to assign a measurement index to each value. If this index exceeds a given threshold, it is likely that the meter measuring this flow is a major contributor to the imbalance in the system. Alternatively, it may indicate that there is a leak in the pipe section containing this meter. Either way, if the value of the measurement index is high, the meter should be physically inspected. Furthermore, by trending index with time, it is possible to detect meter drift or leak development before significant operational problems arise. The technique can be applied both to trunk mains and to demand management areas (DMAs) providing that there is sufficient metering to allow some data redundancy.
There is a range of other analysis techniques available that can help increase the accuracy and reliability of flow data from a distribution network. Some are fairly simple - for example averaging or smoothing algorithms. And others are more advanced, such as genetic algorithms, where a computer-based neural network learns the behaviour of a network and then highlights anomalous behaviour.
In today's business environment, data is increasingly regarded as among the company's most valuable assets. Optimising data use is an operational imperative, especially to water companies under environmental, regulatory and resource pressure. Many water companies have recently made substantial investment in new infrastructure, data control systems and general data acquisition. The use of condition monitoring to detect problems in key plant such as pumps and filters is now routine; data analysis is the metering equivalent. Failure to protect metering investment, by not complementing it with modern, cost-effective, data analysis techniques, risks increased capital and operational expenditure through poor targeting of effort. The video referee is available - use one.