Precision Time Protocol profile for power utility automation Application

Authors: Hubert Kirrmann, Solutil, Switzerland and William Dickerson, Arbiter Systems, Inc, Canada

Time Synchronization in the Grid

Electrical utilities have been one of the most demanding applications regarding time precision. Historically, the electrical grid was the primary provider of time for the synchronization of public clocks (synchronous clocks) and this remains true even today as the grid operators ensure that the number of periods at the end of the day matches the 86400 seconds times the grid frequency (this is a frequency distribution, so the clocks still need to be set). To track back disturbances, for instance to reconstruct the sequence of events that led to a blackout, a precise time-stamping of events with respect to absolute time is needed. In this application, the time distribution over the network using the SNTP protocol with an accuracy of some 10 ms was sufficient.

With the introduction of digital substations, e.g. based on IEC 61850-9-2, a precise time distribution with a better accuracy is needed. Differential protection schemes protect assets by detecting unbalances in the current flow. Current and voltage are sampled at different places at exactly the same time. This allows applying simple Kirchhoff’s law to sense abnormal situations. To this purpose, a sampling accuracy of some microseconds is needed. Since only relative time matters here, the time distribution for differential protection could be done either via a dedicated link, e.g. carrying a 1 PPS (pulse per second) signal. This works well within a substation, but for differential line protection between substations, it is impracticable to span dedicated synchronization lines. Rather, the communication network was used under the assumption that the delays were predictable and symmetrical. This was the case with the old analogue telephone lines, and also SDH/SONET networks, but this virtue went lost with the digital data communication packet switching networks, which need exact time to resynchronize the samples.

Synchronization to absolute time became indispensable to detect grid instabilities. Grid collapses announce themselves by variations in the grid frequency, so comparing the phases of voltage and current at strategic nodes of the grid allows to detect an incipient black-out and take measures.
To this effect, the phase of current is measured at strategic positions in the grid by Phasor Measurement Units (PMUs). India is deploying a wide area monitoring system with more than 5000 PMUs. Synchrophasor transmission over PSN (packet switching networks) is specified in IEC TR 61850-90-5.

The phase measurement in each PMU is time-stamped with respect to absolute time within a few microseconds. Indeed, a phase error precision of 0,1° corresponds to time error of 5 µs on the sampling. A clock accuracy of 1 µs is deemed more than sufficient for this application, since the instrument transformers and the filtering introduce a larger error.  Early PMUS were synchronized by radio receivers listening to broadcast of atomic clocks, e.g. WWV in the USA or DCF-77 in Europe. The devices in the substation were connected in star fashion to the receiver by dedicated lines, e.g. according to the IRIG-B protocol. 
Synchronizing by a GPS receiver in each device has become cost-effective, but remains technically impracticable. The reason is that utilities do not trust the GPS signal. Some GPS clocks, especially in locations with restricted sky view, have shown degraded performance depending on the satellites that are visible. The antenna must also be deployed outside of the substation if the substation is entrenched or subterranean. Especially in northern countries, the GPS signal can be interrupted by aurorae. The signal could be spoofed, presenting a cyber-attack risk.

So, the solution that imposed itself, is to use atomic clocks, e.g. rubidium clocks, in dedicated network nodes. The time is transmitted over the wide area network (WAN) that controls the grid, as explained in IEC TR 61850-90-12. Similarly, the time signal is distributed within the substation using the same
Ethernet as for the data communication, as explained in IEC TR 61850-90-4. This makes that an IED can be used as a PMU. In addition, the LAN transmission eliminates the dedicated 1 PPS or IRIG-B links within the substation.

Figure 1 shows an example of time synchronization in a substation. The primary reference comes from the wide area network (e.g. an SDH/SONET or MPLS network), where a number of atomic clocks maintains the absolute time. In case of WAN disruption, a local rubidium atomic clock maintains the time until WAN recovers.  A GPS receiver is used for consistency check and to adjust leap seconds in UTC time. This diversity is also a protection against cyber-attacks that spoof time. As help for commissioning, the master clocks can generate 1 PPS signals to check that all IEDs are correctly synchronized.
Figure 1 shows three different types of bay that need synchronization: classical bays with no process bus, bays with merging units to classical instrument transformers and bays with a process bus and distributed digital instruments.

Relion advanced protection & control.
BeijingSifang June 2016