CEC/CAISO/PG&E

GRIDiant was asked by the California Energy Commission (CEC) and the Lawrence Berkeley National Laboratory (LBNL), with the participation of the California ISO and PG&E, to perform a detailed analysis of the conditions that precipitated an emergency rolling blackout affecting 100,000 homes and businesses. At the time, spiking hot weather loads stressed a key PG&E Bay Area system bus, indicating imminent system collapse due to insufficient voltage support. Pacific Gas & Electric (PG&E) was forced to shed excess load in order to avert total voltage collapse. An official report to California Governor Gray Davis concluded that the event was related to grid instability (due to high loads and short supplies in that area), which could not be relieved given the design of the transmission system.”

GRIDiant was asked to review certain decisions made by California ISO (CAISO) and PG&E on that date using GRIDfast optimization software and to produce its own solution. Given a data set comprising 2,506 buses, 3,164 branches, 438 generators, and 1,145 loads, the primary solutions GRIDfast identified involved optimized “re-control” of many of the 412 PG&E System “control” buses that were available to the CAISO in near real-time. GRIDfast first stressed the non-optimized PG&E Bay Area System to the point of collapse by increasing loads. Then, GRIDfast produced a precise, bus-specific re-control solution. Using its bus-level resource sensitivity indices (RSIs), GRIDfast was able to identify and rank the sensitivity of each bus in the system to changes in generation and load at that location. GRIDfast determined the most vulnerable bus and the amount of load it could tolerate before collapse, and confirmed that for a non-optimized system, PG&E correctly selected the bus that needed to shed load to avoid voltage collapse.

GRIDfast demonstrated several impressive solutions for CAISO and the utility:

• GRIDfast identified the best location for monitoring system stress levels (the specific bus most sensitive to impending voltage collapse).

  
  

• GRIDfast’s re-control solution minimized P (real power) and Q (reactive power) losses, improving P and Q power flows, power quality, and voltage profile in the Bay Area System. Specifically, GRIDfast proved it could reduce P losses by 8.1 percent (75.6 MW) and Q losses by 15 percent (1,147 MVAr).

  
  

• GRIDfast confirmed that sufficient generation and other resources (including transmission and distribution resources) existed at the time to have withstood voltage collapse without the need to shed load. GRIDfast showed how, through re-controls without new resources, the system as a whole could have run at higher tolerances with improved system reliability, thus achieving greater operating efficiencies.

  
  

• GRIDfast identified a solution that could have averted the blackout crisis even after shutdown of a “dirty” 710MW generator that was put online to serve heavy load while paying heavy environmental penalties.

  
  

• GRIDfast calculated the correct ordering of loads for emergency load shedding to prevent system collapse, distinguishable from the “rolling block outages” deployed.