Smaller degree of uncertainty
Dr John Duffy considers the importance of accurate open-channel flow measurement and the best practice for instrumentation accuracy as identified by Pulsonic Technologies
In addition to the uncertainty issue, consent dischargers questioned how audit competency and standards can be applied to the same degree throughout the industry i.e. how can a level playing field be established? In order to address this issue the EA is now developing a Monitoring Certification scheme (MCERTS), which will enable the UKAS accredited SIRA Certification Services to certify technical experts to carry out audits to a common standard. This is a long overdue initiative and will undoubtedly promote and raise standards within the industry.
Whilst the introduction of an MCERTS scheme addresses some of the concerns of the dischargers there is still argument about the +/- 8% uncertainty figure. Can this be achieved? Many suppliers, including Pulsonic Technologies, believe that current technology can indeed meet this criterion. To achieve this figure consideration must be given to all components of a flow measurement system including:
- manufacture of primary devices (flumes and weirs) to standards,
- installation of primary device,
- installation and commissioning of secondary device,
- accuracy of secondary instrumentation.
The accuracy of a system is a figure of merit, which describes the probability that the measured value (the measurand) is correct. In open channel flow measurement the accuracy of the measurand, i.e. level, is of paramount importance because this is directly related to flow. Inaccuracies in an open channel flow system subject to BS3680 operation may arise from:
- calibration faults,
- incorrect installation,
- incorrect construction,
- supercritical flow (weirs),
- subcritical flow (flumes),
- floating debris,
- environmental conditions,
- poor computational methods and simplification of BS3680 formulae,
- inability of measurement system to respond to rapidly changing flow rates.
In open-channel measurement errors can introduce significant inaccuracies. Assuming ideal measurement conditions and ideal installation, the BS3680 standard alone predicts an error range of up to 4% for rectangular weirs, 2% for triangular weirs and up to 5% for standing wave flumes. If any of the errors mentioned above also come into play the overall error can be quite considerable.
Errors can be classified into two types, (1) systematic and (2) random errors.
These are errors which are the same in repeated measurements of the measured quantity. For example a fixed offset on the no flow distance gives rise to a fixed calibration error. Incorrect flume dimensions give rise to a repetitive error in the flow calculation. Such errors once identified can usually be eliminated or reduced. This however does require a good understanding of open channel primary devices, their installation and operation.
Systematic errors dictate a system’s accuracy.
These errors are much more problematic in that they are extremely difficult to identify. Random errors are unpredictable and arise through chance, irregular or random causes. For example the floating of debris underneath the ultrasonic sensor gives an artificially large flow: the blockage of a V-notch by debris does likewise or the surcharge of a flume. They usually can be identified by statistical methods but because of their nature they are very difficult to remove. To ensure random errors are kept to a minimum, the operational behaviour of the discharge should be understood and housekeeping tasks established to ensure random events are eradicated or kept to a minimum.
Random errors dictate a system’s precision, i.e. its freedom from random errors. Constant independent errors are additive, whilst errors depending on the measured value have a multiplying effect. In open-channel flow measurement most errors of the type mentioned above have a direct effect on the level in the channel or weir and as a result their overall effect on the total inaccuracy is thus multiplicative and can be quite considerable.
A number of the errors identified above result from incorrect installation and/or construction of the primary device. Expert knowledge of primary device construction and installation and in particular the implementation of good housekeeping will keep these errors to a minimum.
But how can this understanding of accuracy be applied to ensure the secondary instrumentation can meet the 8% uncertainty figure? Pulsonic Technologies has identified a number of technical design features, which together minimise systematic errors. This best practice ensures that such errors are either eradicated or significantly reduced such that Pulsonic Technologies instrumentation can readily measure within the 8% limit of flow accuracy. These features include:
One of the most common sources of errors in open-channel flow measurements is short-cut computational methods. It is not uncommon for manufacturers to use simplified versions of the BS3680 equations e.g. for a V-notch weir – flow = kH^2.5 – where K is a constant and H is the liquid head in the primary device.
Referral to the BS standard will show the flow calculation is far more detailed than this and the simplified equation introduces an error, which can be significant. As flow is a cumulative measurement the error in the total flow over time can be quite considerable. Manufactures must not take shortcuts and should use the full BS3680 equations in all computations.
Some manufacturers use a look-up table embedded in memory to equate level (measured by the sensor) to flow, from a primary device flow curve. Unfortunately this conversion is only as good as the point resolution of the look-up table. Many manufacturers use only a 10-point look-up table that can give large inaccuracies when the flow distance falls between two given points particularly at highly non-linear points within the flow curve. Provision of a 25-point look-up curve enables the spatial resolution to be minimised over a greater span.
Speed of response
Flow measurement can suffer from variation in the true surface level due to (a) surface turbulence and (b) sudden variation in flow, e.g. wave effects. To ensure the variations are measured and in the case of (a) the effect reduced and in the case of (b) the wave volume measured, it is necessary to fire the ultrasonic transducer rapidly in order to track the variations in level as they occur. Transducers with an extremely fast response ranging from 2Hz to 3Hz dependent on the product type are used to enable rapid tracking of flow profiles.
Ultrasonic velocity in air varies by 0.18%/˚C. As a result, accurate and high-resolution temperature compensation is essential to ensure the measured distance from the sensor to the liquid level is correct. Measuring the air temperature to 4 decimal places with a fast responding temperature sensor is essential for accurate distance measurement.
The latest products use fast 16-bit processing and floating point mathematics to generate highly accurate flow measurement. Often flow meters are used to retransmit the flow measurement on a 4-20mA transmitter. This is often performed by low-resolution digital to analogue (D/A) converters. To ensure the measurement figure calculated can be retransmitted the resolution and stability (drift) of the analogue to digital converter is critical. The use of 12-bit D/A converters with a 50ppm stability factor is critical for retransmission accuracy.
Resolution is a term which defines the ability of a system to respond and detect the smallest change in the measurand, in this case liquid level in the primary device. In open-channel flow, measurement is often over very small distances and level measurement in millimetres is required. Whilst there is no relationship between resolution and accuracy (often confused) it is essential for flow measurement that the sensor can resolve very small increments in level and hence flow because of the integral effect of time on the total flow measured over a period. Resolution is proportional to wavelength, which is in turn inversely proportional to frequency, i.e. high resolution equals high frequency. Pulsonic Technologies sensors operate at 80Khz as opposed to the industry norm of 40KHz. This means the transducers have twice the resolution and can resolve distances to less than 1mm. This is extremely important in open-channel flow measurement where the head of liquid is measured in millimetres and the accumulative effect of an error in the head measurement can give rise to a large error in totalised flow. For example a 90 degree weir operating at maximum span with an error of 1mm in its head measurement gives rise to an error of 70m³/d. Even at minimum span this error is 5m³/d or 10% of the dry weather flow (DWF) minimum limit of 50m³.
A welcome outcome of the EA’s P150 report is it has stimulated debate amongst what had become in some respects a complacent flow industry. The fact discussion has centred on the +/- 8% figure means we have accepted inaccuracy in flow measurement for too long. It is quite proper that users of flow instrumentation should question what is achievable and be provided with assurances. The P150 document has raised awareness and sent a message that accurate flow measurement is a critical component in producing data that can be defended. It is now up to the manufacturers, installers and auditors to respond to the challenge.
A measurement system is only as good as its weakest link and should any one of the processes in the chain – from the sensor to the instrument’s output – fail to be accurate, the system as a whole fails. Pulsonic Technologies uses state-of-the-art design at all points in the flow instrument and demonstrates that manufacturers can offer highly accurate open-channel flow measurement and maintain an accuracy within the +/- 8% uncertainty limit. Anglian Water has tested the products against other industry standard products and its final test report concludes: “The in-site tests confirmed the Pulsonic Aqualogger flowmeter is better than that used in industry presently. Its ability to derive gauged head and convert this to discharge is demonstrated by its consistently higher accuracy, this is particularly evident at low flows.”
Whilst the BS3680 standard was first published in 1965 and standards exist for other types of flow metering, there has never been a standard for open-channel flow metering. The MCERTS programme is an excellent start to resolving some of the issues brought up by the P150 document but perhaps an extension to include some of the best practices for flow meter design would belay the concerns of many of the consent holders about the reliability of flow measurement data