Skip to main content
Reference

PRP and HSR

Parallel Redundancy Protocol and High-availability Seamless Redundancy — the IEC 62439-3 mechanisms that give an IEC 61850 process bus zero-recovery-time fault tolerance. Two physical networks (PRP) or a single ring (HSR), in both cases delivering both copies and discarding duplicates.

Also: PRP, HSR, Parallel Redundancy Protocol, High-availability Seamless Redundancy, IEC 62439-3

PRP and HSR are the two redundancy protocols defined in IEC 62439-3 that give an Ethernet network in a substation the property protection engineering actually needs: zero packet loss on a single network failure. Not “fail over in 50 ms”. Not “recover in 200 ms”. Zero. The next packet after a cable cut is on time, because both copies were already in flight.

Both are how an IEC 61850 process bus survives a fibre being knocked out by a maintenance trolley without the protection scheme noticing.

Why “zero” matters

A protection IED running a current-differential scheme on Sampled Values at 4,800 Hz consumes a sample every 208 µs. A 50 ms STP/RSTP failover loses 240 samples. That is enough to either trip the wrong feeder (false positive on a phantom unbalance) or miss a real fault. The whole substation has to ride through the network event with no observable disturbance. STP and even RSTP cannot deliver that. PRP and HSR can.

GOOSE trip messages have a similar constraint: a 3 ms publication budget cannot tolerate any reconvergence delay.

PRP — two parallel networks

The Parallel Redundancy Protocol model:

  • Every node is connected to two physically independent LANs (LAN A and LAN B).
  • Every frame is sent on both LANs simultaneously by the source’s Doubly Attached Node (DAN).
  • Every frame received is checked against a duplicate-discard table by the destination DAN; whichever copy arrives first is delivered, the other is dropped.
  • The two LANs share no infrastructure: separate switches, separate cables, often separate physical paths through the substation.

A failure in LAN A is invisible to the application: LAN B’s copy was already on the way. The protocol overhead is one extra trailer (the Redundancy Control Trailer) per Ethernet frame, four bytes.

This is the dominant pattern for protection-critical process buses.

HSR — a single ring

High-availability Seamless Redundancy uses a ring topology rather than parallel LANs:

  • Nodes are connected in a single ring.
  • Every frame is injected in both directions around the ring by the source.
  • Each node passes frames in one direction through to the next, and consumes the first copy that arrives.
  • A break anywhere in the ring is invisible: the other direction’s copy still arrives.

HSR uses less cable than PRP — one ring rather than two redundant fabrics — but every node is in the data path for every frame, which constrains scale and limits topology choices. It is more common in substation bay-controller rings and inside cabinets than as the primary process-bus architecture.

RedBox and QuadBox

A device that is not natively PRP or HSR (a Singly Attached Node, SAN) cannot be plugged into a redundant network directly. The standard answer is a RedBox — a small Ethernet device that bridges a SAN onto the redundant fabric on its behalf, doing the duplicate insertion and elimination transparently. A QuadBox does the same job between PRP and HSR domains.

This matters in retrofit because most legacy IEDs are SANs, and the gateway-and-RedBox pattern is what lets them join a PRP process bus without firmware changes.

Where it sits in the standards stack

  • IEC 62439-3 — the protocol itself.
  • IEC 61850-90-4 — engineering guidelines for using it in substation networks.
  • IEC 61850-8-1, -9-2 — GOOSE and SV, which assume the underlying network can deliver them with zero loss.
  • IEC/IEEE 61850-9-3 — PTP profile, which has to traverse the same redundant fabric without losing accuracy. Annex A of 62439-3 defines how.

A claim that a substation network is “61850-compliant” without specifying PRP or HSR is incomplete: 61850 by itself does not require redundancy, but any deployment with protection-grade SV or GOOSE is going to need it, and the network team has to pick one.

Where deployment lands

In GB transmission substations, PRP is the typical choice for new-build process bus, with HSR appearing inside cabinets or on smaller bay rings. The cost is roughly double the switch count and double the cable run versus a non-redundant network — a non-trivial uplift, but the alternative is a protection scheme that depends on STP convergence, which no protection engineer will sign off on.

Brownfield substations without process bus rarely deploy PRP/HSR; their station bus typically runs on a non-redundant LAN with the protection-critical paths still on copper hardwiring back to the relay panel. Process-bus migration is what brings PRP into the substation, and it tends to be one of the larger network-engineering line items in any 61850-class project.