Grid Interactive Inverter Anti-Islanding etc.

By Tim Harland, Victoria - Australia
August 1999

Some comments and suggestions regarding grid connection, which I believe are important. I would be very keen to read any responses any responses.

1) Anti-islanding by frequency drift.

There is at least one important reason why the direction of frequency drift in grid "interactive" inverters, to try to prevent islanding, should be down.

A brief explanation of this is that power authorities would generally like to see more loads with a leading (capacitive) phase angle in order to try and compensate for the majority of loads which are lagging (inductive), and it is much easier and more foolproof to achieve a to achieve a leading phase angle in an inverter with a downwards frequency drift than one with an upwards drift.

An inverter using the frequency drift approach to anti-islanding has to continually try to pull the frequency in the chosen direction, and in doing so it appears to the grid to have either a leading or lagging reactance Why is this?

Obviously a grid connected inverter cannot simply oscillate at whatever frequency or phase it feels like without regard to the grid frequency and phase. On the other hand if the inverter simply tries to feed current to the grid exactly in phase with the grid voltage and thereby tries to avoid altering or influencing the grid phase or frequency then the frequency may not drift, or only very slowly, if the grid fails.

Thus for an inverter to use the frequency drift approach to anti-islanding its output frequency must be synchronised to the grid, but if must continually try to "pull" the frequency in the chosen direction.

Obviously it cannot chose to frequency pull only when the grid is down, since it is relying on the frequency pulling to try to detect the condition of the grid. If it stops frequency pulling then its chances of detecting grid failure may be considerably reduced.

How can an inverter pull the grid frequency ? If it supplies an output current which lags the voltage it will tend to pull the frequency lower, while a leading current will tend to pull the frequency higher. Since the current out of the inverter is flowing into the grid, an inverter trying to pull the frequency down will appear to have a capacitive or leading reactance, while pulling it up will appear inductive or lagging.

2) Problems with anti-islanding.

Having many independent solar powered inverters connected to the grid would appear to offer a "golden" opportunity to achieve greater reliability of power supply than is achieved by the large utility power stations alone.

Power outages, particularly unscheduled ones, cost the community dearly in disrupted or lost work and production.

Many of the potential grid faults which could interrupt the flow of power from utility power stations to consumers could be repaired safely without requiring the affected consumers, their local reticulation lines, or any grid connected inverters connected to them, to be powered down.

Unfortunately virtually all of the current grid interactive inverter control and anti-islanding proposals (except the one in section 4) do not permit grid interactive inverters to be used in ways which would "back up", or increase the reliability of, supply.

3) Problems with existing anti-islanding proposals.

There are serious questions about the reliability of virtually all of the current anti-islanding schemes, including frequency drift. For example if the load on a frequency drift shutdown inverter has a phase angle close to the limiting phase angle of the inverter then a stable equilibrium may occur with no frequency drift, and therefore no shutdown. This scenario is actually not particularly unlikely given that the majority of load on the grid has a lagging phase angle and the proposed drift direction for invertors will result in them supplying a lagging current, and that a decrease in frequency will generally result in a decreasing impedance in a lagging load.

A grid "interactive" inverter which uses the grid as its phase and frequency reference will, via its output, alter its own reference in a way which is not completely predictable, and cannot be fully compensated for. This may to introduce a risk of instability and oscillation.

Further problems with existing proposals are that they impose significant performance penalties on inverters and unnecessary compromises on overall grid operation.

4) A better method of inverter control and anti-islanding, with other benefits as well.

I believe there is a method of grid interactive inverter control and anti-islanding which is far more robust and reliable than most, if not all, of the current proposals, avoids the frequency tracking, potential instability, and drift problems mentioned earlier, allows the power authority more flexibility and control of grid operation, imposes few if any performance compromises, has no scalability limits, would be relatively cheap, would facilitate greater reliability of supply, and offers several other potential benefits as well.

I propose that a low bit rate signal modulated on a set of low frequency carriers be injected by the power authority into the grid at certain places, and that grid connected inverters be equipped to receive this signal and to shut down or isolate their output from the grid if the signal falls below a certain threshold, or carries a shutdown command, or is corrupted for more than a short period.

The information carried by this signal should include a signal identification keyword and a phase synchronisation reference, and an error detection word, and perhaps also time and date, commands to reduce or increase power output or to retard or advance phase angle, and tariff information.

With signal injectors at suitable places in the grid this method would be highly fail safe since almost any fault which will cause the grid to fail in an area will also automatically cut off the signal to that same area, since it would be carried by the same wires, transformers, etc. as the power, and will thereby cause all affected inverters to shut down.

In any case the signal injectors would be under the control of the power authority, which could switch any one or more of them off at any time it chooses such as during repair of power distribution faults.

The method would be economical since the receiver circuit for the signal could be fairly simple and only a small proportion of the cost of an inverter, and the signal injectors, although more expensive, would only be required in small numbers, since each one could serve a large number of premises and inverters.

The signal would not cause any interference since the frequencies used are all well below the radio bands and at a much lower power than the mains fundamental and at a similar or lower level than typical existing mains noise, and would be thoroughly filtered out by the power supplies of electronic equipment.

Another important benefit is that the signal could carry a mains phase (and frequency) reference. Currently it seems that grid interactive inverters generally use the mains waveform (the 50 or 60 Hz fundamental) to control the phase of their power output. A major problem with this approach is that since their output is feeding the mains, they are interfering with and altering the very thing they use as a reference, in a way which is not fully predictable or practical to compensate for. This problem is likely to get worse, and probably unmanageable, unless a better approach, such as the one I propose, is adopted.

There need not be any concern that noise or interference might "look like" or be mistaken for the correct signal, since the probability of this occurring, even transiently, could easily be kept to vanishingly small levels by a suitable, and quite practical, length of signal identification keyword, and by use of an error detection word.

The use of a power output control command field in the signal would allow the power authority much better control of inverters during power disruptions and restorations and minimise surges and transients.

5) Suggested signal format for improved inverter control, anti-islanding, etc.

In order to minimise the number of signal injectors required, the signal carrier frequencies should be chosen so that they can propagate long distances over the grid without excessive attenuation, without causing radio interference, and without being excessively subject to interference from external sources.

The data rate required is only a few tens of bits per second.

These requirements are probably best satisfied by a signal with carriers of a few hundred Hertz.

Since most of the noise on the grid is harmonics of the fundamental, I propose to minimise interference with the signal by using carrier frequencies midway between the first few harmonics. With a 50 Hz fundamental these are 125, 175, 225, 275, ..., 625, 675 Hz. Some higher carriers in the same sequence might be used for some optional or less important information. These carriers could easily and efficiently be filtered by the FFT (Fast Fourier Transform), preferably synchronised to the mains fundamental.

The modulation should be phase shift keying, with 180° shifts, and the modulation rate for each carrier should not be greater than about 25 Baud.

It would be desirable for the two carriers in each of the pairs formed as follows (125, 225), (175, 275), (325, 425), (375, 475), (525, 625), and (575, 675) Hz, to be modulated with the same data and be 180° out of phase with each other at the zero crossings of the (ideal or reference) mains cycle.

It would be possible, and may be desirable for reasons of redundancy and noise immunity, for a number of the carrier pairs to be modulated with the same information.

These carriers can also be used (while simultaneously carrying data) to provide an accurate and reliable phase (and frequency) reference for the mains (50 or 60 Hz) fundamental. This might best be achieved if all of the carriers have a zero crossing at the same time as one of the two zero crossings of the (ideal or reference) mains cycle.

6) And furthermore ...

I have no pecuniary (financial) interest in any particular form of inverter control or anti-islanding, I simply wish to help promote renewable power and minimise or overcome any obstacles to its widespread use.


Copyright © 1999, ECO Services International