Rapid increase in solar installations potentially overloading the grid
Recently the Australian media reported on problems encountered with solar panel installations affecting the electric power grids of local supply authorities.
An article appearing in The Australian on 13 October last year, titled “Solar panels overloading electricity grid” claimed that voltage rises caused by solar installations feeding into the grid could damage computers and other household appliances.
According to the article, at least one supply authority has warned the New South Wales government that “in areas with a high concentration of solar cells, voltage levels can rise and this could have consequences for appliances and equipment in customers’ homes”. So the question is: can this be correct?
The increase in grid-connected solar installations certainly has potential voltage regulation problems. ‘Potential’ is an important qualifier, as in many instances there aren’t voltage problems. However, depending on the location within a typical suburb and the incidence of highly variable loads such as air conditioners, voltage swings outside those permitted by supply companies and authorities might occur. There is also the matter of voltage swing limits imposed by the inverters, which, when exceeded, causes them to cease feeding the grid. Interestingly the use of battery energy storage for grid-connected systems is one way of dealing with ‘over voltage’ situations.
The opportunity for voltage swing is provided by high impedance supply lines to domestic and shop premises. Larger industrial loads are generally supplied by low impedance (stiff) networks. Voltage swing ‘works both ways’. A high impedance line will cause voltage at the load end to drop with respect to its supply transformer, and vice-versa an inverter supplying such a line with power flow into the grid can produce a rise in voltage.
The important features to consider relate to the details of the distribution scheme. The reticulation scheme is a radial scheme. The feeder line, drawn in heavy black, is connected to a local pole transformer that serves the houses in a street section.
We need to consider impedance levels in some detail. There is the impedance level of the individual drops from the secondary of the transformer to the consumers. The other impedance level is that of the primary winding and its MV feeder to the switchyard.
It is a reasonable supposition that the MV feeder is a very low impedance line, or in other words, part of a stiff network. That being so, the opportunity for occasioning a voltage rise as a result of feed-in from solar systems is very limited. If we examine the equivalent circuit of a typical power transformer, we see that it can be looked at from two aspects: the circuit viewed from the LV side or from the MV side.
Under normal operation conditions – and we’ll define this as an aggregate consumer load being larger than the aggregate solar power being generated – there is no excessive voltage issue. In fact the overall system is working to the advantage of both consumers and the power company. The inverters essentially operate at unity power factor and supply part of the real power of the load, leaving the supply of the same reactive current as if there were no solar inverters in operation, and the balance of the real power needs. Admittedly this means that the power factor at the transformer declines, but this is not considered to be a troublesome issue although it may cause the power factor to drop below the minimum allows according to the Australian Standard for grid-connected inverters (AS 4777).
As to voltage issues, as already mentioned, voltage can rise at the inverter end, resulting in a lift of voltage along the lines to consumers. However solar inverters are required to operate under strict voltage limits, requiring switch off at a low limit of 200V and a high limit of 270V for single phase, and 350V and 470V respectively for three-phase inverters. The deployment of increasingly higher rated PV systems, i.e. from 1.5kW to 5kW increases the chance of voltage rise under conditions of high irradiance. However, in many if not most practical situations, there will be sufficient aggregate load to absorb the power generated by the solar panels.
Reverse power flow and stability
The claim of overloading, as made in The Australian newspaper, is not without substance. There is the possibility, as indicated earlier, that a local distribution transformer is back-fed. In other words there is aggregate power flow toward the grid. In an identical manner to that applying to the consumers and their aggregate solar generation capacity neighbouring transformers that are part of the MV feeder can absorb the excess power. Undoubtedly this is a complicated situation, and generally power distribution companies do not want reverse power flow as it can contribute to network instability. One of the problems that crops up is that of setting tap changers on MV distribution transformers to compensate for line voltage drops by boosting voltage. With the injection of solar-derived power, intolerable voltage rise in the MV network is then encountered.
Power companies are not too keen to spend money in investment in protection equipment, which would not be necessary if there were no ‘distributed generators’ (DG) as part of the overall system. Solar inverters are DGs, and it has been a long established practice to protect networks against reverse power flow that might occur in large local generation plant, for example, as found in a paper mill.
Although its name implies otherwise, the network protector does not actually protect the (secondary) network cable from failure. The network protector does, however, protect the stability and dependency of the secondary grid by preventing current to travel away from the customer and towards the primary feeders. If there is a fault on the primary feeder, the substation circuit-breaker is meant to open, disconnecting the primary feeder from one side. The issue is that this primary cable is also connected to a network transformer. This network transformer is also connected to the secondary grid and it therefore necessary to open up the circuits in the secondary to prevent hazardous situations arising for service personnel.
The whole subject of limiting the effect of inverters on the grid is ideally addressed by a sophisticated communications network with appropriate power flow sensors installed in the LV side of the network. However it is the sort of capital investment that would be difficult to justify while the solar PV industry is in a state of flux – to wit, the less than consistent political will displayed by Australian governments at State and Commonwealth level.
At the present, some supply authorities are limiting the maximum power of PV systems, and that is certainly one way of solving the possibility of net reverse power flow. Other ways include putting the onus on PV system performance control. Such control can be part of the anti-islanding controls of an inverter. Anti-islanding is there to protect against failure of the grid. The chance that an inverter ‘bank’ might back feed the LV transformer must be avoided since dangerous MV voltage will be present in a part of the grid assumed to be disconnected.
Most inverters detect the islanding condition by looking for some combination of the following:
- A sudden change in system frequency.
- A sudden change in voltage magnitude.
- A sudden change in the df/dt (rate of change of frequency).
- A sudden increase in active output power (kW) well beyond the expected ‘normal’ level.
- A sudden change in reactive output power (kVAR) well beyond an expected ‘normal’ level. In respect of kVAR, the reactive component, this is supplied from the grid to the consumer. The inverter does not supply this as it operates essentially at unity power factor. Thus a failure of the grid will cause this component to fall away.
As already mentioned, inverters not only need to operate within restricted voltage limits but also in frequency limits being 45Hz at the low end, and 55Hz maximum. Thus inverters that monitor the voltage and current output in terms of frequency and phase, are capable of providing sophisticated anti-islanding protection and help in guarding against excessive reverse power flow. Basically the maths supporting the protection algorithm is:
v =Vmsin2pft and i = Imsin(2pft + q), where f and q are frequency and phase difference respectively. The rates of change in these quantities will indicate faults in the network and the rate of change in power flow will flag the possibility of excessive reverse power flow.
Power quality problems
In practice it will be difficult to determine where this has been occurring in Australia although there can be little doubt that in areas in Adelaide, and probably elsewhere it has presented itself. Unoccupied premises during times of intense insolation will be primary candidates for excessive reverse flow. The operation of air conditioners, on the other hand will draw off power, and the two events may well coincide.
Can solar PV systems cause the failure of consumer appliances? Although it has been stated that this can occur, the chances are remote. In as much as the grid voltage can drop through high loading, the PV inverters will have to follow if they are connected to low impedance lines. A drop to the lower permissible limit of 200V could have some effect on loads such as pump motors. A split-phase 240V induction motor would experience a reduction in torque equal to (200/240)2, which is an effective torque of 70% compared to that full line voltage. Depending on the torque requirements of the mechanical load, this could cause the motor to stall and burn up, if not protected by a thermal relay. Over voltage has also been mentioned as a possible cause of failure of IT equipment but regulated power supplies will generally be able to handle limited over voltage.
Harmonic contributions from PV inverters are not usually a problem; the more so that many if not most inverters use high switching frequency pulse width modulated (PWM) signal formation. PWM can provide a very low fundamental distortion sine wave. The use of filters including an interposing transformer will help assure compliance to Australian Standard requirements limiting the third, fifth, seventh and ninth to no more than 4% each. Higher orders, as a result of the high switching frequency, typically 4kHz and higher can give rise to these. Potentially these can interfere with power line communication equipment. The increasing switch to transformerless inverters by virtue of cost and weight advantages sharpens the necessity to contain harmonic current generation.
In conclusion there seems to an increasingly powerful case for aggregating solar PV systems into large systems integrated with the MV reticulation system, rather than on individual domestic dwelling roofs, and providing electricity companies with the RECs now handed to individual owners of solar PV systems.
The advantage would be a better-engineered system offering more effective protection. However we need to bear in mind that the small domestic PV systems are not likely to cause reverse power flow problems, except in relatively rare instances. There is, however, the matter of particular idiosyncrasies of inverters to switch off when maximum power flow is being exceeded, when it would seem more sensible to reduce the power to the maximum threshold. The use of boost stages can also provide increased operational hours as insolation reduces.
Power distribution authorities do have a role if the penetration of domestic solar PV systems continues at rates established in the last years. One technique, proposed by Professor Leith Elder from the University of Wollongong is the injection of reactive VARs by means of a static compensator in the MV network. Leith predicts that voltage compensation will become increasingly necessary.