By Alan Dick

Implementation of the Measuring Instruments Directive (MID) in the UK has led to changes in the way in which electricity meters for domestic customers are deemed ‘fit for purpose’ in providing confidence for consumer protection purposes. Previously meters had to be type approved by Ofgem (the UK national regulatory body) and individually certified – initially verified under regulatory surveillance.

At the time of granting approval, a fixed certification period was allocated to that particular meter type, ranging from 10 to 20 years depending on assessment during the approval process. This effectively set the period for which a meter could remain in service without any need to inspect or test, at the end of which it had to be removed and replaced with another meter having valid certification. Over the last seven years some nationally organised testing has been carried out of meters first coming off service at the end of their certification period, as a result of which some periods have been adjusted – extended by five years or, in a few cases, reduced by five years. The former allows that meters of that type not yet removed can therefore remain on circuit until the end of the extended period.

UK regulations promulgated in 2006 implementing the MID deem that a MID-compliant meter satisfies the needs in existing legislation for approval and certification, in meeting the conformity assessment procedures (of which B+D is equivalent to the previous UK processes). The assessment of conformity is carried out by any EU notified body (NB) and there is no process for the NB to set any ‘certification period’. However, the MID requires that the manufacturer provides an estimate of the meter’s durability, being the period over which it maintains the stability of its metrological characteristics. For the UK, where the legal in-service accuracy limits are set equivalent to the maximum permissible errors of the MID, this durability is equivalent to the in-service life. However, can this estimate, which may or may not be subject to scrutiny by the NB, be relied upon without further investigation?

In the UK’s competitive supply market it is the supplier who is responsible for providing the meter and who is therefore responsible for demonstrating to the national metrology body (now taken over from Ofgem by NWML) that his meters on circuit comply with legislative requirements relating to fitness for purpose, especially accuracy. Suppliers can actually be fined if one of their meters on circuit is found to be outside of prescribed in-service accuracy limits. Suppliers discharge their responsibilities through agents – meter services providers (themselves in competition with other MSPs) – who provide and maintain metering assets on their behalf. So it is the MSP who must ensure continuing accuracy of meters as part of his contract with the supplier. How can he do this?

The old certification arrangements relied on the predictable degradation characteristics (accuracy drift) of an electromechanical meter, i.e. the meter was calibrated to tighter limits than the in-service requirements and the certification period was set so that it was removed before it was expected to drift outside the in-service limits. Periodic testing of accuracy gave the drift profile so as to be able to do this.

However, the advent of electronic (static) meters – in commercial use in the domestic market in the UK since the mid-1980s – caused a problem which utilities all over the world are now facing. Static meters are not thought to drift, at least not in a necessarily predictable way, and periodic testing of accuracy therefore gives no clues as to the ‘life’ of the meter, which is governed predominantly by partial or total failure of components. In other words, this is more a reliability than an accuracy problem. Prior to the MID this was addressed by Ofgem – which had to assess the length of the certification period – through use of a predictive model as described in IEC TR 62059-41: 2006 (Reliability prediction).

The UK has an Industry Metering Advisory Group (IMAG) which reports to and advises an executive comprising representatives of Ofgem, the market adminstrator (ELEXON) and relevant government departments (DBERR and DEFRA). IMAG set up a sub-group to recommend a process by which MSPs could demonstrate that MID-compliant meters in service were fit for purpose.

Despite the known limitations of accuracy testing of what were expected to be mainly static meters, it was decided that a testing regime was the way forward. The sub-group took account of similar work in other forums, for instance the Australian/New Zealand standard AS/NZS 128413:2002 and, particularly, a draft document entitled ‘Extension of the period of validity of the verification of utility meters on the basis of sampling inspections’, being produced by TC4/SC4 of OIML. However the first major decision was to depart from the latter in choosing to pursue sampling by variables rather than attributes, the reason being that testing by variables requires a smaller number of meters per sample.

Previous testing of meters in the UK had been on meters removed from circuit due to the certification period having expired; MID meters had no such period and very little ‘churn’ was expected, so meters would have to be deliberately removed, with consequent inconvenience to customers and curtailing of full expected asset life – important in a competitive environment. (‘Churn’ is where meters come off circuit for various reasons such as a change of tariff or vacation of premises. It should be noted that the UK’s competitive market arrangements do not require the meter to be changed when a customer changes supplier, although some are).

The decision to sample by variables meant that much of the process would comply with ISO 3951:1989 which, for instance, indicates appropriate sample sizes according to the population under test. Other key decisions made included:-

  • Meters would not be required to be tested until eight years after installation, and thereafter every five years until such time as the assessment criteria indicated that they were no longer fit
  • Each meter type need not be tested every year. A ‘superpopulation’ could be formed of up to five years’ production, providing all meters were homogeneous (e.g. no production changes). Tests carried out at eight years and at subsequent five-yearly intervals would apply to that whole super-population
  • Rules would apply for deciding whether a meter collected to form part of a sample should be tested. Some meters might be discarded as unsuitable for testing (e.g. tampered with, badly damaged or subject to an accuracy dispute) or excluded from testing but this fact recorded (e.g. faulty display, advance under no load)
  • Accuracy would be tested at three load points – 1 amp, 20 amps and maximum rated current. For success a meter would have to pass at all three (i.e. there would be no weighting)
  • Simple outlier criteria were set based on sample size, rather than have complicated tests (such as Grubb’s Test)
  • Pass criteria was set at an AQL of 5. Samples not passing would be subject to a fallback criteria of 10. Passing this fallback would mean an annual number of meters removed as they come up to the age of the first year. Failing this fallback would mean immediate withdrawal of all meters in the population
  • Normality (normal distribution of test data) would be assumed and not tested for.

With regard to super-populations, the regulator had some concerns with this concept and safeguards had to be built in. For instance, the sample size is based on the entire super-population (this will be known by the time of testing at eight years) rather than any one year’s population to avoid unrepresentative results if the first year’s installation was small compared to later years. Whereas the concept significantly reduces the amount of testing needed, it increases risk to the MSP that the entire super-population might need to be removed, rather than that of one year.

With regard to reporting of excluded meters, although reports would not form part of the pass/fail criteria they could indicate a potential reliability problem which might need to be investigated.

With regard to the AQL, there was much discussion about what this should be. Although 5 is not a recommended value in ISO 3951 it was selected as a ‘round figure’ of acceptable meter population accuracy, based on a guideline produced by WELMEC as to what level of consumer protection is reasonable, the main consideration being to achieve a fair balance between what was reasonable and the costs and inconvenience of the sampling required. Currently a report has been presented to the IMAG executive for endorsement with a recommendation for some new work on governance and organisational aspects.

It remains to be seen how the process will be taken up by industry; for instance MSPs could apply it to their individual asset mixes, but there could be benefits in some central body managing the process nationally, so that collection of required sample numbers could be shared amongst more than one MSP, lessening the customer and asset impacts referred to above. However, this may be difficult where MSPs are in competition – they will need to be reassured that each party involved is pulling its weight!

In summary, the process gives UK MSPs a way to demonstrate that they are meeting their supplier’s legal obligations as regards in-service accuracy of their meter populations through use of a methodology backed up by internationally accepted statistical methods and industryagreed criteria.

Ofgem – Office of Gas and Electricity Markets, the UK energy regulator
DBERR – Department for Business, Enterprise and Regulatory Reform
DEFRA – Department for Environment, Food and Rural Affairs
NWML – National Weights and Measures Limited, the UK national legal metrology body
OIML – Organisation de Mètrologie Lègale, the international legal metrology body
WELMEC – Western European Legal Metrology Co-operation, a European metrology advisory body
ISO – International Standards Organisation, producing international standards
IEC – International Electro-technical Committee, producing international electro-technical standards