There are two fundamentally different standards applied to the design and testing of polyphase energy meters worldwide. These are generally referred to as ANSI and IEC.
The ANSI standard originated in the USA and is now the accepted norm in North America and in some other countries where American influence on the electricity distribution system has been strong. The IEC standard originated in Europe and has been adopted in various forms in the remainder of the world.
There are fundamental differences between the two standards as applied to the design and testing of polyphase meters.
Electromechanical (Ferraris) meters contain inductive devices, which by their nature generate leakage fields, which can affect the operation of similar devices in close proximity. In a polyphase energy meter this means there will be some interaction between the coils related to the various phases. This interaction is called ‘element interference’ and is present to a greater or lesser degree in all electromechanical meter designs. By careful spacing, orientation and shielding, meter designers in the USA have reduced this effect to a minimum, but meter designers elsewhere in the world have not taken this degree of care, as it is possible to adjust out the unwanted effects of the interaction at the time of final test and calibration of the meter.
The result of this difference in approach is that an ANSI meter can be tested without any consideration being given to the phase orientation of the three phases of the supply, but an IEC meter must be tested with a supply which mimics the network conditions on which the meter will eventually be used. This means that although an ANSI meter is, to a good approximation, insensitive to the rotation direction of the supply and exhibits errors of fractions of one percent if the phase rotation is reversed, an IEC meter can exhibit errors of several percent under these conditions.
Practically this has meant that ANSI meters have historically been adjusted and tested using a series parallel connection on a single phase supply (voltage elements connected in parallel and current elements in series), whereas IEC meters have used a true three phase source, observing the correct phase rotation.
Specific concerns when testing ANSI type meters
All meter designs originating from the IEC areas are ‘true Blondel’, meaning that they comply with the requirements of Blondel’s theorem – that is, the minimum number of power measuring elements should be n-1 where n is the number of conductors in the network to ensure that the total system power, and therefore energy, measured is correct for both balanced and unbalanced systems. Some ANSI meters, however, do not comply with the requirements of this theorem. Types such as Forms 6, 8, 14 and 15, and to some extent Form 2, do not have the full complement of elements necessary to satisfy Blondel, and make assumptions about the voltage and current balance of the network in order to operate correctly in service.
Testing of these meter types requires balanced loads, a condition automatically achieved with series/parallel connection as all the elements are connected to the same source. Historically, this balance has been more difficult to achieve with a true three phase system, and for testing to IEC standards has not been necessary. The requirements of IEC 521, for example, specify that test systems which use the ‘energy comparison’ test method, and which include a class 0.05% true three phase reference meter, only require that the voltages are balanced to within 1.0% of their nominal value, and the currents within 2.0%, for meters of classes 1 and 2.
This would imply that when testing a meter at a nominal 50 amps per phase, the unbalance could be up to 1000mA, and this degree of unbalance can cause meter testing errors of the order of ±1-2% with some types of ANSI ‘non-Blondel’ meters. However, modern developments in intelligent electronic power sources such as the MTE PSU10 and PSI10 allow the three phases to be balanced to better than 0.05%, thus removing this concern and allowing true three phase sources to be used to test this type of meter. These sources also include active load control based on feedback using Fourier analysis of the output waveform, and ensure that the distortion factor is kept below 0.5%.
As electronic (solid state) meters do not fundamentally contain inductive elements exhibiting leakage effects, element interference is not normally a problem with either IEC or ANSI designs. There is, however, one important exception. If a self-contained (whole current) meter design uses internal CTs in the current measuring paths, and insufficient care is taken to minimise cross coupling, a similar effect can result. This is particularly of concern if the meter designer has used air gapped CTs to attempt to avoid errors caused by core saturation when the load contains a significant DC component. Air gaps significantly increase the leakage fields generated by these components.
Test equipment requirements
An advantage of a true three phase test bench versus a single phase test bench with series/parallel connection is that three phase and single phase test points can be performed without changing the wiring. In a single phase test bench with series/parallel connection, either the wiring must be changed manually or an expensive switching hardware must be built-in to automate these changes to test three phase and single phase test points in the same test procedure.
The test equipment requirements for the two methods are significantly different, in that only a single phase source is necessary for the ANSI method, whereas a true three phase source is required for IEC. At first sight this would seem to indicate that the capital cost of the equipment for the ANSI system would be substantially less than that for IEC.
This is not necessarily true, however, because the power requirement for testing meters by the ANSI method is three times that required for each phase of the IEC system, as each meter on test has its three elements connected to the same test supply. This means that the total power required is identical for the two systems, the difference being that the ANSI system normally requires two high power drive amplifiers (one for voltage and one for current) whereas the IEC system requires six smaller (and therefore less expensive) units.
There is no operational reason why ANSI meters cannot be tested on a three phase supply, as they are insensitive to phase rotation or position, meaning that a test system designed for IEC meters can also be used to test ANSI meters. The general worldwide availability of multi-position, true three phase, IEC-type test systems such as the Meter Test Station range from MTE in Europe makes this type of system economical for testing meters using either method. A study performed by the Jacksonville Electricity Authority and the University of Florida in the USA, and published in an IEEE transaction paper in 1993, demonstrated that although differences could be detected between the results obtained using the two methods when testing electromechanical (Ferraris) meters, these were small enough to be of no practical importance, and it is probable that the measurements obtained by true three phase testing are more representative of what would be seen in service.
Closed link testing
Self contained or ‘whole current’ IEC meters are normally tested ‘open link’. This means that the internal links between the current and voltage sensing elements are opened, giving galvanic separation of the elements. This allows multiple meters to be tested without any problems of unwanted current paths between elements. Normally ANSI meters, however, are tested ‘closed link’, with these links connected in their normal operating position.
However, testing of all types of meter ‘closed link’ is becoming the norm, even for IEC meters, as there is a possibility that if the links are opened for test they may not be closed correctly before the meter is installed. This could lead to incorrect operation, and result in large billing errors. The presence of accessible links also makes it easier for customers to tamper with the meter.
In order to test multiple self-contained polyphase meters ‘closed link’, galvanic separation between the measuring elements must be provided by external means. This is easily achieved for single phase meters by the simple addition of a multi-secondary voltage transformer to the voltage source, with each secondary providing an isolated supply to each meter, but this approach cannot be used for polyphase testing, as the three voltage sensing elements usually have a fixed common neutral connection, leading to unwanted circulating currents between phases.
It is therefore necessary to provide the required isolation between the current phases of each meter, by powering each meter via three wide range current transformers. The characteristics of these transformers have a considerable effect on the overall accuracy of the test system, and as a general rule must have an accuracy class of 0.1% or better over the required test current range, which for IEC testing will routinely be from 0.1A to 100A.
Transformers to this specification are expensive, and as three are required for each meter test position can add considerably to the cost of a test bench. In order to meet current ANSI standards, however, the test current range is not as great as that required for IEC testing. There are generally only three test points required – test current (normally 50 Amps) at power factor 1.0, the same current at P.F+0.5 lag and a low load test at 10% of test amps (normally not lower than 1.0A) at PF=1.
This allows passive transformers of sufficient accuracy (0.1% error or better) to be supplied at reasonable cost for testing of class 2 or class 1 meters over the reduced range. In order to extend this range it is necessary to use more sophisticated techniques, such as electronically compensated (zero flux) transformers, in order to maintain good accuracy at low currents. Both these solutions are available in test systems produced by MTE, allowing the same systems to be used for both standards.