Reducing Carbon Emissions Through Rolling Scheduling Optimization of a City-Level Virtual Power Plant Considering External Power Dominance and Multi-Resource Flexibility Aggregation: A Case Study of Guangzhou, China

Import-dependent megacities must decarbonize while accommodating distributed photovoltaics (PV) under constrained local-generation and interchange boundaries. This study develops a full-year, hourly, 48 h rolling mixed-integer linear programming model for coordinated city-level scheduling in Guangzhou, China. The VPP-controlled portfolio includes distributed PV, grid- and user-side battery energy storage systems, demand response, and aggregated electric vehicles, whereas local generators and inter-regional imports are represented as system-operator coordination variables. After correcting the energy balances of flexible resources and recalibrating the hourly interchange boundary, the baseline scheduled-import share is 68.68%, only 0.64 percentage points above the measured schedule. Across four policy-defined PV anchors, combined annual operating-and-carbon cost decreases from CNY 75.01 to 65.50 billion, carbon intensity declines from 467.9 to 412.2 g CO2/kWh, and import share falls from 68.68% to 63.92%. A dense 0–12 GW sweep at 0.5 GW intervals reveals a continuous response: curtailment first exceeds 1 MWh at 4.0 GW but remains only 0.006%, providing no evidence of a unique breakpoint. A conventional time-of-use tariff reduces import share by 0.58 percentage points while increasing carbon intensity by 1.25 g CO2/kWh. These results are conditional on the deterministic supply construction, calibrated interchange boundary, adopted emission factors, and operating-cost scope.