| Description |
The purpose of this research is to develop methods and software tools to measure stress and the susceptibility of bulk electric power systems to cascading blackouts. We define some metrics to evaluate stress, compute those using real data from Western Electricity Coordinating Council (WECC) and relate them to cascading blackouts. The cases analyzed include five for 2016, one from 2012, plus the September 8, 2011 pre-blackout state. Our work analyzes properties of a novel network based on line outage distribution factors (LODF or DFAX) instead of the familiar network based on the Y-bus matrix. The traditional Y-bus matrix only reflects the physical connections between the buses. The Ybus network is excellent for studying how power flows through a network, but for contagion to cascading, the issue is how failures can propagate in a network, and Y-bus is not very helpful. The first chapter of this report contains a review of the literature related to blackout, in addition, our motivations behind this study are discussed. Definitions of DFAX and metrics that measure stress are presented using an example of a 13-bus system. In the second chapter, these metrics were computed using two software tools, PowerWorld and a program written using MATLAB for some areas, mostly the southwestern part of the Western Interconnection (WI).We analyze five seasonal cases for 2016, with very different levels of load. PowerWorld was used to read the WECC data and to calculate the DFAX matrix. This matrix along with some other data from PowerWorld was processed using programs written in MATLAB to get metrics showing characteristics of the cascading failure system. We compare WECC results to results from earlier studies of other systems. In the third chapter, the preblackout state of the southwestern part of the WI before the September 08, 2011 cascading blackout is analyzed. The system stress for this case is compared to the five cases of 2016 and to a peak summer case for 2012. The metrics proposed in Chapter 2 were computed for this case.We also sought a tipping point between a stressed and a not-stressed system, with unexpected results. We also analyze "hot spots" with stunning results. Monitoring these metrics, Area 1 was identified as highly vulnerable, and outages of any of a single series of four lines as critical contingencies. This series includes the line whose failure initiated the blackout and a second outage of the same series which would have had exactly the same effect. Had these results been available on September 8, the substation work which triggered the blackout absolutely would not have been approved without prior action to reduce the vulnerability of Area 1. Chapter 4 (plus a copy of a computer program) contains the effect of errors in the network model on the values of the DFAXes, computed by a standard power flow program. Then the effects of inaccurate DFAXes on the stress metrics are examined. Finally, we added a chapter with our overall conclusion to this study. A description the software developed and submitted to WECC is added in the appendix. |
