Following on from our previous article which gave a short overview of regenerative drive operation, here we look at some of the special factors which need to be considered with regenerative operation.
These are the regulations imposed by power companies to ensure that embedded generation does not disrupt the safety or reliability of the electricity supply. The rest of this section covers the most important topics controlled by regulations. If the regen system is to be used as an intentional generator then it will have to comply with the regulations in force at the place of use. If not, then the topics covered should still be considered whenever the current rating of the regen system is a considerable part of the local power system rating, to ensure that in the event of a fault or malfunction no damage will be caused to other connected equipment.
A local generator contributes to the current into an electrical fault (short circuit), and might affect the safety or required rating of electrical switchgear. In any proposed generator installation the added fault current has to be calculated. However the Regen inverter has a negligible impact, because the electronic over-current protection interrupts excessive current much more quickly than a circuit breaker or fuse can act. The peak short circuit current for the present generation of CT inverters is 260% of rated current, and once the inverter trips the current decays to zero in less than 4 ms (depending on the value of the a.c. choke). The peak current for protection rating is taken over at least one half-cycle of the supply frequency. These factors together make the fault current contribution negligible.
If the power grid becomes disconnected from an installation where a generator is operating then there is a possibility that a power “island” might occur where the local generation unintentionally keeps the local loads energised. This is not very likely, because without a purpose-designed governor to regulate the frequency there is nothing to ensure the supply and demand are balanced. Normally the frequency quickly moves out of the working range and the system trips. Also, there is no control of voltage or reactive power. However if an island did occur then there would be a safety risk to power workers, and a risk of damage to local equipment in the island if the voltage or frequency moved beyond their safe range.
If a Regen system has a source of energy (e.g. an engine, battery or other energy store) so that island operation is possible, then protection should be provided against a damaging island occurring. The frequency range of the regen drive should be restricted in the parameter settings to a safe range, and an overvoltage relay should be incorporated to trip the inverter if the voltage becomes excessive.
For intentional generators there are standards for islanding protection (sometimes referred to as “Loss of Mains” protection (LOM). Some of these require the inverter to operate a special algorithm to detect the island condition, which is available as a standard facility in the Unidrive M drive. Some of them require an independent approved protection relay.
Intentional Island, backup generation
Please note that the regen inverter cannot be used as a stand-alone generator, for example as a back-up supply in case of mains supply loss. It can only be used in conjunction with an existing supply to which it synchronises itself.
Intentional generators may be required to continue in operation during a power grid disturbance. The most common situation is where a fault (short circuit) occurs somewhere in the power network which results in a sudden voltage dip at the generator terminals. This might be balanced or unbalanced between the three phases. During ridethrough, it might not be possible to continue generating the rated power if the voltage is too low, but reactive current is required to support the voltage and assist the grid in recovering control once the faulty circuit has been disconnected by automatic protection devices.
A simple AFE is quite sensitive to voltage disturbances because its operation relies on an accurate balance between the mains voltage waveform and that generated inside the inverter. Unless it has ridethrough capability it is more inclined to cause nuisance trips than a simple rectifier. The Unidrive M has a selectable ridethrough capability which meets major national standard requirements such as the BDEW guidelines for generators connected to MV networks.
It is important to appreciate that in normal regenerative operation the Regen inverter adjusts its power output to the AC power system so as to regulate the DC bus voltage to the desired value. During a voltage disturbance it can no longer generate its full rated power, so it may be unable to continue regulating the DC voltage. The power source then has to take over this role. If it does not do this then there may be an over-voltage trip if the incoming power exceeds the outgoing. Alternatively a braking resistor can be connected to absorb excess power.
The balance of supply and demand in the AC power grid is achieved by regulating the frequency. An intentional generator may be required to help in this by responding to external power commands, or to implement a control function of power against frequency. This can be programmed as an application in the drive.
As discussed in the first regen blog, the regen inverter generates negligible levels of true harmonic current, i.e. at integer multiples of the AC supply frequency. It interacts with existing harmonics on the supply, and also it generates PWM modulation products. These are at high frequencies which for many years were considered to be beyond the range of harmonics, which were generally considered to end at order 40. However more recent technical standards and instruments have begun to consider harmonics up to order 100.
For example, take a system operating at a nominal line frequency of 60 Hz and a switching frequency of 3 kHz. The main switching-related frequencies present will be 2880 Hz and 3120 Hz. These are 48 and 52 times the supply frequency. However the two frequencies are not commensurable quantities, or in other words they are not phase-locked. If the line frequency were 60.1 Hz then the equivalent product frequencies would be 2879.8 Hz and 3120.2 Hz. When a harmonic analysis instrument is connected in such a system it will probably indicate these as the 48th and 52nd harmonics, if it has the standard 5 Hz bandwidth, or it might indicate an inability to synchronise the data.
If the switching frequency had been 4 kHz then the main frequencies present would be 3880 Hz and 4120 Hz which are not harmonic frequencies. They would be indicated as “interharmonics” by an analyser with interharmonic facility, or they might be ignored by a basic harmonic analyser with a normal 5 Hz bandwidth.
The harmonics and interharmonics discussed above are all 3-phase sets with either positive or negative phase sequence. These means that unlike the high-frequency common-mode “noise” voltages they are passed through transformers and can cause interference beyond a site supply transformer. The switching-frequency filter is necessary to reduce their magnitudes to acceptable values.
When a regen system is connected to a DC source or load, thought must be given to the control of the DC voltage. In a regenerating drive system, the machine drive becomes effectively a constant-power source, and the regen inverter adjusts its power export to balance the incoming power at the desired DC voltage. Other systems may have quite different characteristics. For example, in a PV inverter the DC voltage and current are controlled by the PV array voltage/current curve for the given insolation and temperature. The inverter does not naturally “know” which voltage to choose, so the regen inverter DC voltage reference has to be adapted by the MPPT algorithm to find the optimum power point.
The DC supply in a regen system has an unusual common-mode voltage, that is the average voltage between its poles and ground. A full voltage waveform analysis is quite complex, but referring to the simplified schematic in Figure 1 of the first blog, you can deduce that when one of the input inverter transistor pairs changes its state, the tendency is for the voltage of the DC bus circuit to change by a step equal to V_DC, with respect to ground. In fact the step is restricted by the voltage division around the input choke to 1/3 V_DC. This step happens whenever a phase switches, i.e. six times in each PWM switching cycle.
This means that when the AC supply is a conventional mains LV supply with neutral connected to ground, the DC bus carries a high common-mode voltage which is a complex PWM pattern with fast-rising edges, containing a wide spectrum of frequencies. Some of the effects of this are given in the following list:
For special applications where these effects are unacceptable, one solution is to use an isolating transformer at the input so that the AC supply is isolated from ground. It is then possible to operate with one pole of the DC bus connected to ground directly, so there is no common-mode voltage. Or it can be connected through capacitors to ground or an RFI filter as required, to reduce the high-frequency common-mode noise which is most likely to cause interference. This is used for example in photovoltaic inverters, and in systems where the DC power has to be distributed to multiple loads.
The switching –frequency filter has been discussed above in connection with avoiding interference to other equipment connected to the same supply circuit. The filter must also be considered in terms of its effect on the inverter control system.
In a 3 kHz inverter the filter turnover frequency is about 800 Hz, so that it gives useful attenuation at 2900 Hz. The turnover is somewhat affected by the supply impedance, which is an unknown for these unusual frequencies. This means that the current loop gain in the inverter must not be set too high otherwise stability around 800 Hz becomes marginal and the system becomes sensitive to disturbance and liable to malfunction. For the majority of conventional drive applications, that is where the regen system is one of many loads in an industrial LV distribution network, there is sufficient natural damping that no special requirements arise. The default values are usually effective.
Where one or more regen systems are supplied from a dedicated supply, with little else connected, it is possible for the current loops to be underdamped. This can be indentified easily using an oscilloscope to view the line current waveforms, since bursts of oscillation (“ringing”) occur with a period of about 800 Hz, often at 6 points in each mains cycle. In this situation the stability can be recovered by reducing the P term in the current control loops. It might also be necessary to reduce the voltage loop gains to avoid an underdamped voltage control caused by the slower current loop. If the application is highly dynamic so that these lower gains are not acceptable, then an alternative method for improving the damping is required. There are two options:
Both options have been used effectively. The drawback of option 1. is that several capacitors may be required, occupying space and also causing a high standing reactive current which might then have to be offset by the inverter reactive current control function. The drawback of option 2. is that the resistors cause some standing power loss, and also they have to be protected from overloading in the event of abnormal harmonics in the supply, making the option rather complex.