Proper Utilization of PMU Data: Towards Prevention of Wide Area Blackouts

By: Sarina Adhikari, Senior Consultant
sadhikari@enernex.com
865-218-4600 x6177

 

 

The North American Bulk Electric Systems (BES) comprises of Eastern and Western Interconnection and Electricity Reliability Council of Texas (ERCOT). These are the combination of several complicated rotating machines along with several pieces of equipment for communication, control and protection. The complexity of the North American power grid on one hand, provides very reliable power to the customers, however, on the other hand, there is an equal risk of initiating cascading outages following some major disturbances resulting in wide area blackouts. The cascading blackout of August 14, 2003 in the northeastern U.S. and southeastern Canada has influenced many studies and practices toward preventing similar events in the future.

In order to prevent the interconnected power systems from wide area blackouts, an understanding of situational awareness is important. The transmission system power flow conditions should be continuously monitored in real time in order to provide early detection of problems. System information comprising of phase angles, frequency, rate of change of angles, etc., provide very useful information about the health of the power systems. In order to enhance the monitoring of the dynamic behavior of the system, several utilities have led an effort to deploy a network of Phasor Measurement Units (PMUs) throughout the system.  PMUs on electric power systems have been compared with Magnetic Resonance Imaging (MRI) of a human body. PMUs do provide the information about system states at a sampling rate as high as 120 samples / second, able to capture oscillatory events following any severe disturbances. The technological capability of PMUs is well understood, but the industry is still working out the best ways to use these devices. An important function would be to provide early detection and protection of the power system from system disturbances.

EnerNex recently completed a Major Disturbance Mitigation Study (MDMS) project that provided appropriate ways to utilize the information relayed by PMUs with the use of an intelligent angle prediction algorithm which is able to initiate the protective actions before the system goes unstable. By assisting in the detection and actions necessary to bring the bulk system swinging oscillation centers back to the stable operating range, the algorithms in this project can be used to prevent wide area blackouts, and  also provides important functionality to PMU data.

Controlled Systems Separation (CSS) was one of the recommendations provided in the August 14, 2003 Blackout Final Report by the US-Canada Power System Outage Task Force in order to save the system from a cascading outage. During CSS, the lines, generators and loads in the system are intentionally taken out of service following some procedures in order to stop the unintentional breakup of the network due to cascading events. Then, the system is left in a more favorable state for restoration.

EnerNex completed the earlier project named Control Systems Separation Study (CSSS) in which the possibility of getting the power system back to the stable operating conditions after severe disturbances through controlled separation of large systems into two or more islands was proven in a large bulk power system network through PSS/e dynamic simulations. This MDMS project utilizes CSS along with Under Frequency Load Shedding (UFLS) and system wide generator tripping, if required, as the mitigation measure to save the system from possible instability.

This MDMS study has successfully developed an algorithm using PMU measurements that is capable of predicting evolving angular instability under extreme contingencies both internal and external to the any bulk power system. The algorithm developed during this project was based on two Kalman filters, guided by a measurement prediction algorithm based on the Taylor series expansion and finite difference method. This algorithm was validated in several stable and unstable disturbance scenarios to make sure that the algorithm is capable of providing correct predictions of impending instability for any kind of system disturbances, and also importantly that the algorithm does not give a false alarm during stable system conditions. It was shown from all the testing that the proposed algorithm has a very fast response time and is very accurate in the prediction of angular instability. The cases tested included heavy load cases with heavy transfer conditions. The two Kalman filter prediction algorithm is used to make a decision on the system instability and then initiate the mitigation measure. Mitigation actions could be controlled system separation, along with under frequency load shedding, out of step generator tripping, and then as a last resort, system wide generator tripping. The figure below shows a basic representation for overall strategy of properly utilizing PMU measurements of angular difference between two interfaces to bring the system back to stable system conditions following any severe disturbance.

Important Contributions and Findings:

The major outcomes from this project are summarized below:

  • A new fast, online and accurate angular instability detection and prediction algorithm is proposed, developed and implemented with Python in PSS/e platform in a full stability model of the bulk system with more than 60,000 buses. The algorithm is based on two Kalman filters guided by the angle measurement prediction method based on Taylor series expansion and finite difference. This algorithm is capable of accurately predicting the angular difference between any two important interfaces in any large bulk transmission system.
  • The developed algorithm is capable of converging quickly enough to initiate mitigation measures in a timely fashion to stabilize the bulk system.
  • The testing included simulated noise on the PMU measurements that is thought to be significantly greater than what is existing with the present day technology.
  • The algorithm was tested under a wide range of stable situations as well and it did not falsely predict instability in any of those tested cases.
  • The two Kalman filter based angle prediction algorithm is capable of providing the information on impending instability 12 cycles ahead in time scale. Hence, any type of mitigation action can be initiated 12 cycles ahead in case the algorithm senses any type of system instability in future.
  • The delay of 4 cycles before initiating the CSS action was considered to account for breaker operation, teleprotection and communication delays. This is included to represent the practical scenario of the system under CSS.
  • This study reduces a large amount of offline simulations to find out the exact time when the CSS should start, because the two Kalman filter prediction algorithm is capable of finding the exact time at which the separation should begin and initiate the protection measures automatically. This adaptive capability is very important.
  • Mitigation measures included controlled separation of critical interfaces upon detection of instability across those interfaces. The mitigation measures required operation of the UFLS in places where the load/generation mismatches occurred after the controlled separation.
  • This project also recommends that the UFLS scheme needs to be adaptive and should be adjusted as needed from the settings following different criteria in order to separate the large system into several stable island systems based on the system conditions following any severe disturbance.