Solar Photovoltaic Inverters and Ride-Through Capability

In study after study, we (Pterra) are encountering this seemingly mounting issue of ride-through capability in solar photovoltaic (“PV”) inverters. For now, the matter is isolated to frequency ride-through in small grids such as those that may be found in the Hawaiian islands. However, there is potential for this to be critical in even larger systems as the number of inverter-based PV arrays interconnecting to existing distribution and transmission systems increases.

The crux of the present issue relates to the frequency ride-through settings and capability for commercial PV inverters. The industry standard that addresses frequency settings for solar PV is IEEE standard 1547, Interconnecting Distributed Resources With Electric Power Systems, which specifies that (for 60 Hz systems):

  • For inverters < 30 kW, trip at frequency < 59.3 Hz clearing in 0.16 sec
  • For inverters > 30 kW, trip at frequencies 57-59.8 Hz clearing in 0.16 to 300 sec (adjustable setpoint), or at frequency < 57 Hz clearing in 016 sec

Certain commercial PV inverters are equipped with extended frequency ride-through capability that will not trip until frequency drops below 55 Hz.

The operating range for a typical 60-Hz small grid is within 0.2-0.3 Hz during normal operations, with no contingencies. This gives a lower frequency range between 59.7 to 59.8 Hz. In this operating range, inverter-based PVs are expected to remain online. However, on contingencies, frequency excursions can dip to much lower values. If the frequency dips below 59.3 Hz for more than a second or two, the smaller PVs may trip offline. The effect of PV generation trips is to push frequency down even further. Synchronous or isochronous machines will generally act to recover the frequency to within the normal range subject to how much PV trips off, and how much load the interconnected system’s rotating machines have to compensate for. For severe contingencies, the frequency may drop below 57 Hz which could lead to all the IEEE-1547 compliant PVs to trip offline.

(Note 1: This is assuming such PVs have been pre-set to trip at 57 Hz. This may not always be the case. If so, the PVs may trip off at even higher frequencies.)

As before, any loss of PV generation causes the rotating machines to work harder to recover frequency.

(Note 2: Some systems may have underfrequency load shedding in the range of 58 Hz and lower, and the actuation of this system may relieve the underfrequency condition and allow system frequency to recover. However, there is the possibility that the load shedding system may shed PV that is in the same circuit as load being shed, negating some of the benefits of the load shedding scheme. This brings up the issue that load schemes need to be reviewed when solar PV or any other form of generating resource is added to a distribution circuit.)

In any case, dynamic simulations can be run to test the potential for a runaway condition, where underfrequency is uncontrolled leading to eventual collapse of the interconnection. If conditions for collapse are confirmed, the solution alternatives may include:

  • Requiring new PV to have the extended frequency ride-through capability, keeping the PV online at frequencies as low as 55 Hz. This ensures that some PV generation remains online to support frequency recovery. The amount of PV that should have this feature will vary depending on the specific system and needs to be determined by simulations.
  • Requiring additional backup capability in the form of batteries. The function of the battery, in this scenario, is to pump in watts, when frequency is low. This is not a normal operating mode for backup batteries, so it is necessary to ensure that the batteries to be used have this capability and are set accordingly.

In conclusion, there is growing concern that the lack of frequency ride-through capability in commercially available PV inverters may lead to potential system collapse issues in small power grids. The approach is to calibrate the magnitude of the underfrequency issue by conducting dynamic simulations. Based on the baseline, potential resolutions can be identified such as requiring specific frequency setpoints for existing and proposed PV installations, extended frequency ride-through capability or adding batteries.