Weak-Grid Solar Plant: Voltage, Curtailment, and Frequency Support
A 400 MWp solar PV plant connects to a 220 kV substation in a region where peak demand is 950 MW and minimum daytime demand in spring is 310 MW. The point of connection has a short-circuit ratio of 2.3. Two of the three large synchronous generators that formerly provided voltage support and inertia have been retired; the remaining unit is 350 MW. In the first year, spring curtailment reached 14% of available PV energy, voltage at the POI rose to 1.09 pu around midday, and 12 inverter trips were recorded during remote faults. After the loss of the 280 MW generator, frequency fell to 49.2 Hz before primary response arrested it. The grid code requires continuous operation down to 0.2 pu for 150 ms and reactive current injection during faults, but the plant's inverters were commissioned to an older standard. The system operator is considering a synchronous condenser, a STATCOM, or a grid-forming battery at the POI, and has asked for a technically justified recommendation.
1. Quantify the midday export condition and explain the voltage rise. Why is curtailment the operator's response rather than a permanent solution?
Read the worked answer
At 310 MW minimum demand and 400 MW PV output, the substation exports about 90 MW. Export raises voltage across the network impedance approximately as ΔV/V ≈ (P·R + Q·X)/V². With SCR 2.3 the grid is weak, so even a modest export produces a large voltage rise. The PV plant was commissioned to an older grid code and operates near unity power factor, so it cannot absorb enough reactive power to hold 1.09 pu down to the required range. The 12 inverter trips during remote faults show that the plant also loses ride-through capability, which removes both active and reactive support exactly when the grid needs it. Curtailment is an operational band-aid: it reduces P and therefore voltage, but it throws away 14% of spring energy and does not fix the underlying weak-grid condition or the fault ride-through deficiency. A permanent solution needs dynamic reactive support, improved short-circuit strength, and compliant inverter controls. Simply curtailing more is not acceptable because it transfers the cost of a network limitation to the generator and does not improve frequency or fault recovery.
2. Compare the three options. Which do you recommend, and at what approximate size? Include inertia, short-circuit strength, reactive range, and cost-effectiveness.
Read the worked answer
Option A: a synchronous condenser of about 80-100 MVAr provides continuous reactive range, short-circuit current typically 3-5 times its rating, and real inertia (H ≈ 1.5-2.5 s on its own rating), so it directly addresses SCR, voltage and rate of change of frequency. Installed cost is roughly $25-50/kVAr, i.e. $2.5-5 million for 100 MVAr, plus 1-2% losses. Option B: a STATCOM of ±100 MVAr is faster and has a smaller footprint, but it contributes almost no short-circuit current or inertia; cost is roughly $100-150/kVAr, or $10-15 million. Option C: a grid-forming BESS of 50 MW/200 MWh can provide synthetic inertia, fast frequency response and voltage support, but at $250-350/kWh the energy capacity alone is $50-70 million, which is hard to justify for a voltage problem. Recommendation: install an 80-100 MVAr synchronous condenser at the POI and require the PV plant to retrofit compliant ride-through and reactive current injection. If frequency studies still show an inadequate nadir, add a separate 30-40 MW/30-40 MWh grid-forming BESS for fast frequency response, not a large 200 MWh unit. This staged approach buys short-circuit strength and inertia at lower cost and keeps the BESS sized to the frequency need.
3. Estimate the initial rate of change of frequency after the 280 MW loss and assess whether the 49.2 Hz nadir is acceptable. What fast frequency response would you procure?
Read the worked answer
Use the swing equation. If the remaining 350 MW unit has H ≈ 4 s, stored energy E = H·S ≈ 4 × 350 = 1,400 MWs. Losing 280 MW gives an initial RoCoF of about ΔP/(2E) = 280/2,800 ≈ 0.1 Hz/s; with H = 5 s it is about 0.08 Hz/s. A 49.2 Hz nadir is 0.8 Hz below nominal and is close to underfrequency load-shedding thresholds; it is not acceptable for a system that already has low inertia. The response must arrive faster than primary governors, which typically take 10-30 s. A grid-forming battery of 30-40 MW with a 500 ms response and 30-40 MWh energy would reduce the frequency drop by roughly 30-40 MW times the response time relative to the deficit, and should be procured with a performance requirement to deliver full output within 1 s and sustain it for at least 30 minutes. Also review underfrequency load-shedding settings and require the PV plant to stay online during faults; losing 400 MW of PV on a remote fault would be a second contingency.