Australian Renewable Generation Harmonic Assessment Review - Part 3 Site Assessment

In this article, I’m going to talk about the process of assessing harmonic emissions by site measurement, during and after commissioning of a generator. You might also be interested to look back at Part 1 covering allocation of harmonic limits and Part 2 related to desktop harmonic studies.

So, you’ve completed desktop studies and possibly installed some beefy harmonic filters to squash any pesky harmonic emissions. The transformers are humming away nicely and you’re exporting green megawatts into the network. It’s time to take a few voltage measurements and we’ll know exactly how much harmonic voltage distortion the generator is emitting. Then we can forget about harmonics for the next 20 years of operation. If only it were that simple!

Emissions Calculations

The standard R2 commissioning assessment procedure used in the NEM to assess harmonics is to take background harmonic voltage measurements before connection of the generator over a one-week to 30-day period. IEC 61000.4.30 measurement techniques are applied to extract 3-second and 10-minute average values. 95th and 99th percentile values over the measurement period are extracted according to IEC 61000.3.6 requirements to provide background measurements defining the “pre-connection” or “background” harmonic distortion.

Note that these measurements are done with the Connection Point circuit breaker open to completely disconnect the generating system from the network. This is because the generating system impedance may result in amplification or attenuation of background harmonic voltage distortion, and hence a harmonic contribution from the generating system, even if the inverters are not injecting harmonic currents.

After the plant has been commissioned, the process is repeated to measure the post-connection harmonic distortion. Then the pre-connection emission values are subtracted from the post-connection distortion values to calculate the emissions. Frequently these post-connection measurements are captured several weeks or months after the pre-connection assessment.

The commonly applied approach is referred to as Emission assessment based on summation laws in CIGRE C4.109 Technical Brochure 468 (TB 468). TB 468 highlights the advantages and disadvantages of this and other methods. A plethora of methods are proposed and evaluated in the literature but there is little consensus on a reliable or preferred approach.

But I’m going to stick my neck out here and say that the Emission assessment based on summation laws process is not just flawed, it’s plain wrong. The method relies on the following assumptions:

1)????? the network topology and load conditions are identical for both measurement periods, and

2)????? the background voltage distortion remains constant.

Hence, this method of assessment is only strictly valid in short duration measurements (seconds to hours) where measurements with the generating system connected and disconnected are taken within a short time frame (referred to as the One-shot harmonic emission evaluation in TB 468). The assumptions are unlikely to hold true when the two measurement periods are separated by any significant time due to changes in network configurations, generator dispatch, load profiles, and changes in harmonic emissions from other network users – for example, seasonal changes due to agricultural activities or use of different heating and cooling loads.

In areas with high concentration of renewable generators, emissions from neighbouring plant may differ in different periods and new plants may even be commissioned during the period between the two measurement periods. In the case of future Renewable Energy Zones (REZs), network infrastructure may be rapidly evolving, and new generators commissioned simultaneously which would further exacerbate the problem.

Considering the definition of harmonic contribution in IEC 61000.3.6, calculation of the emission contribution of an installation requires phase angle information, as it is the vector subtraction of post- and pre-connection voltage distortion, as illustrated in the following figure (emission set to zero if |Vh(post?connection|<|Vh(pre?connection)|):

Angle information is not easily captured in long-duration averaged measurements, due to the difficulty in defining an “average angle” especially when the angle varies significantly throughout the measurement period.

In the absence of angle measurement data, suitable alpha coefficients must be selected for reverse application of the general summation law to estimate the angle difference. The standard IEC 61000 alpha summation calculation coefficients are often applied in reverse application of the general summation law. This increases the calculated emissions at higher harmonic orders and reduced them at lower orders. Although the standard alpha coefficients may be reasonable for application in simulation studies involving multiple diverse harmonic sources when phase angle data is not available, I do not believe they are necessarily valid for calculation of harmonic emissions from a measurement data at a specific site.

Similar to harmonic modelling methods, the Emission assessment based on summation laws method may identify major non-compliances, but cannot provide precise assessment of compliance against emissions limits. So why are we using this method? This is a somewhat speculative view but I believe that this standard approach is applied because it is simple and standardised. Confirming compliance of the measurement data and methodology is trivial and not easily subject to challenge. Other methods typically require more data conditioning, interpretation and engineering judgement. Everybody wants an engineer to solve their compliance issues but the application of an engineer’s judgement in the demonstration or evaluation of compliance is less palatable. There may also have been inadequate evaluation of alternative options at the time the process was initiated, resulting in precedent being established and inertia perpetuating it.

Various alternative methods of assessing emissions at the connection point rely on continuous simultaneous measurements of harmonic voltage and current at the connection point. Some challenges relevant to various methods are reliance on reasonable estimates of the network harmonic impedance at the connection point, requirement for harmonic current and voltage measurements, and requirement for harmonic phase angle measurement. Methods may also require careful data conditioning, sorting and interpretation of measurement data and knowledge of network and generator operating conditions.

Estimating the network impedance is particularly challenging. I highlighted the inaccuracy of using wide-area network models to derive the impedance polygons in Part 2. Invasive and non-invasive measurement methods exist to measure the network impedance but these would provide a snapshot view only. Methods of continuous online monitoring of network impedance have also been proposed in the literature. The method outlined in the paper presented by ONE at CIGRE seeks to eliminate the network impedance and background distortion using parameter estimation techniques, relying on significant variations in both network impedance and background distortion during the measurement period. This seems to hold promise but still has its own limitations.

I won’t go further into different methods here or recommend a preferred option. There may not even be a one size fits all appropriate to all sites and applications. Suffice to say, I believe that there are methodologies that will produce more accurate estimates of the harmonic emissions from a generator than the methodology currently applied. Unfortunately, there is no definitive industry standard prescribing an assessment calculation method (or methods). I would say that robust engagement is required, involving experts from across the industry, to agree on the best approach.

Measurement Uncertainty

But suppose we found a better method, capable of perfectly calculating the emissions from the generator based on the measured harmonic distortion. We need to consider whether our measurement apparatus is adequate to accurately measure the harmonic distortion at the Connection Point to within the required tolerance for a reliable assessment of compliance, especially considering low emission limits of 0.1% of rated voltage. Unfortunately, I’ve got more bad news for you here… The measurement equipment and its site application is generally inadequate to provide the required accuracy.

The primary components of the power quality metering system are the instrument transformers and the meter. A typical metering arrangement to measure harmonic emissions at the connection point of a generator is show in the figure below.

Power quality meters must be certified according to IEC 61000.30.4 Class A which specifies a minimum requirement of 0.1 % of declared input voltage (Unom) over the range of 10 % to 150 % of Unom.

Instrument transformers are the interface between the high voltage conductors carrying the electrical power and the sensitive electronic meters. As the name implies, they are instruments for transforming high voltages and currents to low values – typically 110V for voltage and 1A or 5A for current. Hence, the meter is not sensing the actual voltage or current but a transformed version of it. This introduces an additional source of error as the transformation is not perfect.

Only voltage measurements are required for the Emission assessment based on summation laws approach but power quality meters are used for other purposes, including energy metering, and usually also measure current inputs. As mentioned earlier, current measurements are required for most alternative methods of compliance assessment so I will consider both voltage and current transformers here.

The standard requirement is for both voltage and current transformers to be specified as Class 0.2 instruments, meaning that they have a guaranteed accuracy of 0.2% of rated voltage and current respectively. So, considering the metering system in the figure, the guaranteed accuracy or measurement uncertainty, affected by the voltage transformer and meter is approximately 0.3% of the rated voltage, and similar for current inputs.

But the 0.2% accuracy is only required between 40% and 120% or rated voltage and 20% and 120% of rated current. Outside this range, it is not defined and may be significantly worse. Class 0.2S CT’s include an additional requirement for 0.75% accuracy at 1% rated current. Considering harmonic limits to typically be in the fractions of a percent range, the measurement uncertainty could become quite large. Tolerances on phase angle measurements also start to become large at low voltage and current.

But it gets worse because those accuracy limits are only defined at the fundamental frequency (50?Hz). Instrument transformers have a frequency-dependent transfer characteristic over the harmonic frequency range from 50?Hz to 2500?Hz. Capacitive voltage transformers used for extra high voltage applications (typically greater than 132?kV) have very significant non-linearities and need on-site calibration to correct these errors. But electromagnetic CTs and VTs also exhibit non-linearities which increases measurement uncertainty.

This is not to say that all of these inaccuracies will be additive in the real world – they are worst-case and some of these would probably cancel others. The point to highlight is that measuring compliance with harmonic voltage limits as low as 0.1%, specified with a precision of two decimal places (0.01%) is impossible. Measurements will be indicative, not definitive. I did attend an Engineers Australia webinar where Dr Ilya Budovsky from the National Measurement Institute described advanced calibration techniques could be used to achieve accuracy of 0.1%. But he noted that accuracy at lower voltages and currents is easier to achieve and that of VTs and CTs dominate the measurement uncertainty for higher voltage and current applications.

The instrument transformers and meters applied for power quality metering are aligned with international practice and significant improvement in performance may not be technically or economically viable. Hence, we have to accept living with a measure of uncertainty. The important point to take away is that there is a significant degree of uncertainty in any measurements taken for compliance evaluation and this should be considered when interpreting the results.

Conclusion

The use of inaccurate methods of emission calculation based on data with high uncertainty relative to the desired precision results in a calculation of harmonic emissions that could best be described as indicative. It is surely not reliable to a precision of 0.01%, and probably not even close to 0.1%. It is important to recognise this level of inaccuracy and uncertainty in our calculation of harmonic emissions and consider this in the evaluation of the data.

Reasonable assessment of the harmonic emissions from a generator at a snapshot in time is plausible but long-term statistically valid assessment is more challenging.

I do not have data on demonstrated non-compliances for existing generators, how many generators have been assessed as non-compliant, what the degree or severity of non-compliance was or what course of actions were agreed to rectify the problem. But I believe that some degree of pragmatism has been applied by NSPs such that a generator would not be penalised for small exceedances assessed during the R2 commissioning process. Time would be provided to mitigate any exceedances that are not threatening to result in exceedance of the planning levels at the connection point.

As with all aspects of harmonic management, measurement is a complex and challenging problem. I'm hoping the ARENA-funded University of Wollongong research project will make some good recommendations to improve upon the status quo.

In Part 4, I plan to wrap up this series by putting down some final thoughts and mentioning a few emerging challenges.