Interface with EMS

The main problems interfacing phasor data with EMS systems are:

• The phase angle is an instantaneous value based on absolute time
• The measurement is a complex number which is unlike other measurements
• The data rate is much higher than other control center systems

Phase angle measurement is a complicated issue.  Phase angles in synchrophasors are relative to absolute time (UTC), not system time.  Unless the system frequency is exactly at nominal-which it rarely is-the measured phase angle will rotate.  Since all phasor values are referenced to the same time, they all rotate together.  However that means angle comparisons between phasors requires measurements that are made at exactly the same time.  Measurements from a single PMU are synchronized internally, so calculating power using a voltage and current from the same PMU is correct regardless of the external synchronization.  However system phase angle between different stations requires aligning samples by timetag and assuring that the PMUs have an accurate external synchronization.  All PMUs synchronize measurements to Universal Time-Coordinated (UTC).  GPS is currently the only practical method of widely disseminating UTC time with the accuracy required by phasor measurement.
The easy way to deal with the complex number issue is to present the data to other systems as a magnitude and phase angle.  Phase angles relative to the system are meaningful; absolute angles are not.  Hence, absolute phase angles need to be related to a power system reference.  The easiest way to do this is designate one bus as the "swing bus" and subtract it from every reading.  That provides a common power system reference that "rotates" with the difference between system and absolute time.  However if the "swing bus" measurement develops an error, then all of the other measurements may fail.  Two other approaches are to report measurements as the angle between two busses (bus pairs), or use a "virtual bus" that won't drop out.  In 1998, BPA implemented an interface to the traditional SCADA system using the bus pair approach.  This interface acted as a SCADA RTU and responded to data queries.  It reported bus voltage magnitudes, frequencies, and angles between selected bus pairs.  Each bus was represented in two bus pairs, so if one measurement failed, the picture could be obtained using different combinations.  Filtering and holdovers were added to smooth outliers and short dropouts.  Flags for synchronization and data errors were reported as digital status.  As the system expanded, more angles were reported.
It was soon apparent that the pair approach was cumbersome and didn't scale well-every extra bus added two measurements.  Our state estimator vendor also developed the ability to include measured voltage phase angle in the estimation process, so we wanted to report more angles.  We developed a new interface which, like StreamReader, runs on a PC and uses the same real-time data feed from the PDC.  Using the higher processing power of current PCs, we could improve the filtering and handle data impairments more easily.  The new system simplifies setup, monitoring, and modification.  This new interface also uses a virtual bus approach for phase angle.  At each sample, the angle from a few PMUs that are electrically close are averaged and filtered to form a "virtual bus" angle for reference.  This is subtracted from all angles to obtain relative angles for reporting to SCADA.  Then each angle is only reported once.  If an angle in the reference set is lost; there is only a small change in the reported angles.  The virtual reference is only used for computing relative angles, so the state estimator can establish any swing bus and calculate bus angles from that; the system angles as used are thus always exact.  
In addition to the phase angle issue, data measurement rate creates problems.  Traditional SCADA uses low bandwidth transducers and samples at a low rate.  Spikes due to switching, faults, and other transients do no appear.  Phasor measurements are generally high speed with a wide bandwidth.  Transients do appear, particularly in early PMU models that did not use much filtering.  Reporting phasor samples directly to SCADA caused false alarms when the sample happened to occur at a peak or valley of a transient.  Simple filtering of the data removed most of this problem.  The new system provides complete anti-alias filtering.
Another difference between SCADA and the phasor data system is the manner of reporting.  SCADA is queried at slow speed with multiple attempts in each sequence.  It uses lower speed communications which is usually more reliable.  PMUs push out data with no retransmit and a higher speed which is often less reliable.  There are more dropouts in phasors than SCADA.  Due to the high reliability of SCADA, dropouts are quickly reported.  The higher phasor dropout rate causes an unacceptable number of alarms in the SCADA system.  Holdovers in the interface hold the last value for a few query cycles to minimize inconsequential alarms.  Holdover has to be carefully balanced to be sure important events are reported promptly while nuisance alarms are eliminated.