PRP and HSR eliminate recovery time, but they do not cost or get maintained the same way. HSR wins in small, simple substations, where a ring without switches is enough. PRP wins when the network is large, traffic must be limited with VLANs or work permits isolate equipment. The decision is made on total cost, not only installation cost.
Two protocols, one goal
Short answer. PRP and HSR (IEC 62439-3) achieve zero-time recovery because they send every frame over two paths at once. The difference lies in the network: PRP uses two independent networks with ordinary switches; HSR uses a ring of devices without switches. From that come their costs, their size limits and their behavior during maintenance.
How they work inside (RCT trailer, HSR tag, supervision frames) is explained in PRP and HSR. Here they are compared in order to decide.
First: is a high-availability network needed?
Reconfiguring a network with STP takes minutes, with RSTP seconds and with proprietary RSTP variants hundreds of milliseconds. That is acceptable for telemetry and manual control. It is not acceptable for a blocking or trip GOOSE, which is repeated at growing intervals up to the heartbeat and can lose all its useful copies during reconfiguration, nor for sampled values, which a relay discards if they arrive late.
| Application | Is a ring with RSTP enough? |
|---|---|
| Traditional SCADA: reports (buffered and unbuffered) and MMS controls | Yes |
| SCADA with automation schemes | Depends on whether the schemes use critical messages |
| SCADA plus blocking, permissive or trip GOOSE | No: PRP or HSR |
| Process bus (sampled values and control GOOSE) | No: PRP or HSR |
If the project already planned two networks for redundancy, PRP adds no new network: it only requires that the devices support it, or a RedBox for older single-port devices.
Comparison criteria
| Criterion | PRP | HSR |
|---|---|---|
| Network components | Two networks with standard switches | Only HSR devices and fiber; RedBox for single-port devices |
| Paths per message | Four (two directions in each network) | Two (one direction around the ring each way) |
| Traffic control (VLAN, multicast filtering) | Yes, with usual techniques | Not in a simple ring: every frame travels around the ring |
| Growth | Expanded by changing switches | More devices or more bandwidth require several rings joined with RedBox or QuadBox |
| Single-port devices | Connected to one LAN as SAN | Require a RedBox |
| Isolating a device for testing | The other network keeps communication | Breaks the ring |
The cost of HSR looks lower because there are no switches. That advantage shrinks as the network grows: rings are joined with RedBoxes or QuadBoxes, which cost the same as a switch, and they require design, configuration and commissioning of the mesh.
The forgotten factor: the work permit
In an HSR ring, taking a device out of service opens the ring. If the device must keep communicating with its neighbors, cables must be reconfigured: several are disconnected, jumpers are installed and everything is undone at the end, in a defined order and with staff who can handle fiber. Each cable change briefly interrupts other devices, and afterwards it must be verified that the network communicates properly. With PRP, switching off a device does not affect the others because the second network remains complete.
That is why the honest comparison is one of total cost: initial investment plus maintenance and testing over the life of the substation.
What suits each case
- Small distribution substation, with few devices close together. HSR is adequate: there are no switches or panels, and bandwidth is not a concern. If the whole substation can be switched off during a work permit, or only one device is isolated at a time and the risk is an unwanted busbar trip, HSR is enough. Many utilities, in fact, connect everything in a star to a single switch for simplicity.
- Transmission substation, with many protection zones and redundant relays. PRP is better: two identical networks, traffic control by VLAN, bandwidth expansion by changing switches and maintenance without touching the cabling. A large HSR would require several rings joined with QuadBoxes and does not allow GOOSE filtering.
- Process bus in a breaker-and-a-half arrangement. Almost every signal is critical. PRP is the most appropriate: it allows VLANs to distribute bandwidth and connect several zones. In a published case study, each process unit consumed about 5.5 Mbit/s and a single 100 Mbit/s network was at its limit once GOOSE was added, so several HSR rings would be required.
- Process bus in a line bay. Both work and neither is ideal. The most balanced option is an HSR ring only for the bay’s process devices, joined with two RedBoxes to the substation’s two PRP networks. Each bay has its own ring and data is shared over PRP.
- Does the network carry critical GOOSE or sampled values?
- No: A ring with RSTP is enough
- Yes: Is the network small, with devices close together?
- No: PRP
- Yes: Is everything switched off, or one device isolated at a time?
- Yes: HSR is adequate
- No: PRP
How bandwidth is measured
Before choosing, calculate the real traffic: GOOSE, sampled values and frame duplication. Our free process bus network tool helps estimate how much the duplication adds to each link; for flow details see SV and GOOSE bandwidth.
PRP and HSR with Tiza Group
Are you going to design or expand a substation network and do not know whether PRP, HSR or a mix suits it? At Tiza Group we assess traffic and the maintenance plan with our IEC 61850 engineering and test switching between paths in the commissioning tests. The five substation types of IEC 61850-1 and how the bus grows with size are explained in Substation Types D1 to T2 in IEC 61850-1.
Related guides
- PRP and HSR (IEC 62439-3)
- PRP and HSR RedBox
- Process bus
- Sampled Values and GOOSE bandwidth
- GOOSE: messages between relays
- Traffic storm in the substation network
- Substation Types D1 to T2 in IEC 61850-1
References
- IEC 62439-3: high-availability industrial automation networks, PRP and HSR clauses.
- R. Hunt and B. Popescu, “Comparison of PRP and HSR Networks for Protection and Control Applications”, Western Protective Relay Conference, 2015.
- IEEE 802.1D (STP) and IEEE 802.1w (RSTP).
Frequently asked questions
Is PRP or HSR always needed in an IEC 61850 substation?
No. If the network only carries SCADA monitoring and control, a ring with RSTP is usually enough. PRP or HSR are justified when the network carries trip or blocking GOOSE, or sampled values, because those messages do not tolerate network reconfiguration.
Why does HSR complicate testing of a device?
Because there are no switches: every device is part of the ring. Isolating a device for testing breaks the ring, and with two devices isolated some others may lose communication until a temporary fiber bypass is made.
Can PRP and HSR be combined?
Yes. An HSR ring per bay with its process bus devices, joined by two RedBoxes to two substation-wide PRP networks, is a hybrid solution that limits traffic and makes isolation for testing easier.