14 Jan 2019
In modern substations, the protection and control system depends on successful data communication between devices. This requires a highly reliable and highly available communication network and eliminates single points of failure.
The traditional network method for availability is to use a ring network, so that each network node has two paths to communicate throughout the network. Network switches are connected to create a physical ring. Networks consist of Ethernet switches that pass data using a store and forward method, forming a virtual connection between switch ports. Ethernet switching does not permit a virtual connection that forms a complete ring, so there are always specific ports on switches configured to be virtually open. These specific “open” ports therefore won’t forward any data.Mission Critical Communications
or most substation communications, such as traditional SCADA, a reconfiguration time of hundreds of milliseconds is acceptable. However, there are new requirements arising from new applications that are mission-critical. An example is a GOOSE message containing a flag used for tripping, blocking, or unblocking. This GOOSE message must be received within carefully defined time limits that may be in the millisecond range; not hundreds of milliseconds.The IEC 61850 Standard recognises this need, and specifically defines in 61850-5 the tolerated delay for application recovery and the required communication recovery times for different applications and services.
Standard Ring protocols such as RSTP have a recovery time orders of magnitude longer than IEC 61850 fast GOOSE requirements. The answer is to find a method that achieves “zero time” for recovery, which is achieved by sending data simultaneously down multiple paths. An interruption in one path has no effect on the other one.
Zero Recovery Networks
To address this need for zero recovery time networks, IEC 61850 mandates the use of the IEC 62439-3 Standard, specifically Parallel Redundancy Protocol (PRP) and High-Availability Seamless Redundancy (HSR). Both methods of network recovery provide “zero recovery time” with no packet loss. These are the only standard methods to ensure GOOSE and sample valued transmissions, without additional delays in the case of LAN defects.
The basic concept behind PRP is that a device is connected to two independent networks. Any message this device publishes is mirrored to both networks. Subscribing devices, also connected to both networks, will accept the first version of the message received, and discard the second version. If one network link fails, the mirrored message will still go through on the second network. The two networks don’t need to be identical, but they must not be connected to each other. There are many benefits to using PRP for high-availability networks. The first of these is that PRP explicitly achieves “zero” fail over time, due to the use of mirroring frames across both networks. Another advantage is that the PRP networks can use any topology: star networks, ring networks, and any other connection. Most importantly for GOOSE messaging and sampled values, it is still possible for traffic shaping through VLANs, message priority, and MAC address filtering.
High availability seamless redundancy (HSR) is a further development of the PRP approach, although HSR uses a different method to provide multiple paths for data. All devices are connected in a ring topology. Any message from the publishing device is duplicated and sent both directions around the ring. A destination device receives two identical frames on each port within a certain interval. The device uses the first frame received and discards the second frame. If a network link fails, only one frame is received, and this frame is used. Even with a large number of nodes on the network, the time difference between the reception of the two frames is negligible, so zero recovery time is achieved.Things to consider when selecting HSR/PRP
Both PRP and HSR can be applied as a solution to provide a high-availability network. It is useful to understand the benefits of each, and the strengths of each, when selecting which method to use for a specific network or application. Some things that need to be considered when selecting which protocol to use include:
- Availability - While PRP handles n+ contingencies easily, losing one device in a HSR network, the ring is no longer complete and n+1 contingency becomes a real issue
- Total cost of ownership - This cost of ownership must include both capital costs, and operating and maintenance costs.
The general criterion then, for choosing PRP or HSR for a high-availability network, is going to be size and complexity of the network. For small, simple, and/or low cost distribution substations and industrial power systems, HSR can be a good choice. PRP is a better fit in transmission or complex distribution substations, where the size of the system is liable to be large, or where strong permit to work regulatory needs exist. For the vast majority of applications, then, PRP is going to be a better choice from both of performance and from a cost perspective.
Doing a total cost of ownership analysis, including both capital expenditures and the yearly rates return of operating expenditures, is crucial in selecting which is the best method to employ for your application.SCADA with GOOSE messaging
The use of IEC 61850 in substations often combines the traditional SCADA requirements of reporting and control along with protection signaling for blocking, unblocking, and permissive signals. This protection signaling will use GOOSE messaging, which can be published over the same SCADA network. As these GOOSE messages are mission-critical, a simple ring with RSTP is not sufficient. Dual redundant LANs require redundant GOOSE messages carrying the same data, which is not an ideal situation. The solution is to use one of the high-availability networks of IEC 62439-3. Either PRP or HSR is sufficient depending on the application requirements. Legacy devices will require the use of a RedBox such as Hirschmann’s RED25.
The rise of the “fully digital substation”, and the growing use of digital communications for all data, is driving the need for high-availability networks in substations. IEC 61850 applications are now sending mission-critical data through GOOSE messages and sampled value messages. Simple ring network’s and dual LANs are not adequate for the task of high-availability in these situations.