Correlation of diverse measurement records for the diagnosis of the 2020 Mumbai blackout

A blackout generally involves a cascade of events that may involve many components in a system. These events are captured by measurement devices like Phasor Measurement Units, Disturbance Recorders, Relay Fault Recorders, and Sequence of Event Recorders. These recorders have diverse characteristics, may not all be time-synchronized, record measurements at different locations, and may belong to several different utilities. Therefore collation and reconciliation of the records are necessary for an accurate diagnosis. This involves significant manual effort and domain expertise. Using the 2020 Mumbai blackout example, we extract, analyze and co-simulate some characteristic features in the diverse records, leading us to the final diagnosis. The preprocessing steps that allow easy correlation between recordings from dissimilar devices, feature extraction methods, and the steps in the diagnosis that are readily translatable to algorithms are identified so that these can potentially be incorporated into a semi-automatic diagnostic tool.