MiCOM Pxxx (PR) 20 PRP Notes

PRP NOTES

CHAPTER 20

Pxxx/EN PR/C22 Page (PR) 20-1

(PR) 20 PRP Notes MiCOM Pxxx

Date (month/year): 03/2014

Hardware suffix: J, K, L & M
Software version: 04 and later

Page (PR) 20-2 Pxxx/EN PR/C22

Contents (PR) 20 PRP Notes

CONTENTS

Page (PR) 20-

1 Parallel Redundancy Protocol (PRP) Notes 5

1.1 Introduction to PRP 5
1.2 Protocols 5
1.3 PRP Summary (IEC 62439-3 Clause 4) 5
1.4 Example of a PRP Network 6
1.5 Structure of a DAN 7
1.6 Communication between SANs and DANs 8
1.7 PRP Technical Data 10
2 PRP and MiCOM Functions 11
2.1 MiCOM Products and PRP 11
2.2 MiCOM S1 Studio Software 11
2.3 MiCOM Relay Configuration 11
2.4 Hardware Changes 11
2.5 PRP Parameters 12
2.6 Product Requirements 13
2.6.1 Abbreviations and Acronyms 14

TABLES

Page (PR) 20-

Table 1 - MiCOM model numbers for PRP options 11
Table 2 - PRP parameter values (for PRP Protocol Version 1) 12

FIGURES

Page (PR) 20-

Figure 1 - PRP Redundancy Network 6
Figure 2 - Communication between two DANs 7
Figure 3 - Frames without and with RCT and padding 9

Pxxx/EN PR/C22 Page (PR) 20-3

(PR) 20 PRP Notes Figures

Notes:

Page (PR) 20-4 Pxxx/EN PR/C22

Parallel Redundancy Protocol (PRP) Notes (PR) 20 PRP Notes

1 PARALLEL REDUNDANCY PROTOCOL (PRP) NOTES

1.1 Introduction to PRP

This section gives an introduction to the Parallel Redundancy Protocol (PRP); and how it
is implemented on MiCOM-based products manufactured by Schneider Electric.

1.2 Protocols
Industrial real-time Ethernets typically need much better levels of availability and
uninterrupted operation than normal office-type Ethernet solutions. For power networks,
even a short loss of connectivity may result in a significant loss of functionality or
impaired safety. To recover from a network failure, various redundancy schemes have
been considered, including: Rapid Spanning Tree Protocol (RSTP), Media Redundancy
Protocol (MRP) and Parallel Redundancy Protocol (PRP). The key properties of these are
as follows:
RSTP this uses mesh-based topologies and computes a tree, based on path
costs and priorities. In case of network failure, a typical reset time for
RSTP-based system is normally a few seconds.
MRP this uses ring-based topologies. In case of network failure, the network is
broken into two separate lines, which are reconnected by de-blocking the
previously blocked part. The guaranteed reset time for MRP protocol-
based systems is typically around 100ms.
PRP this does not change the active topology as it uses two independent
networks. Each message is replicated and sent over both networks. The
first network node to receive it acts on it, with all later copies of the
message being discarded. Importantly, these details are controlled by the
low-level PRP layer of the network architecture, with the two networks
being hidden from the higher level layers. Consequently, PRP-based
networks are continuously available.
Power networks need to be able to respond to problems very quickly (typically in less
than 10ms), and PRP is available protocol which is quick enough to achieve this. The
PRP protocol used in the MiCOM relay/IEDs is defined in the IEC62439-3 (2012)
standard and is configured using the existing redundant Ethernet card(s).

1.3 PRP Summary (IEC 62439-3 Clause 4)

A summary of the main PRP features is given below:
x Ethernet redundancy method independent of any industrial Ethernet protocol or
topology (tree, ring or mesh)
x Seamless switchover and recovery in case of failure, which supports real-time
communication
x Supervises redundancy continuously for better management of network devices
x Suitable for hot swap - 24 hour/365 day operation in substations
x Allows the mixing of devices with single and double network attached nodes on the
same Local Area Network (LAN)
x Allows laptops and workstations to be connected to the network with standard
Ethernet adapters (on double or single attached nodes)
x Particularly suited for substation automation, high-speed drives and transportation

Pxxx/EN PR/C22 Page (PR) 20-5

(PR) 20 PRP Notes Parallel Redundancy Protocol (PRP) Notes

1.4 Example of a PRP Network

Essentially a PRP network is a pair of similar Local Area Networks (LANs) which can be
any topology (tree, ring or mesh). An example of a PRP network is shown in Figure 1:

DAN
PRP Redundancy Network (D3)

Power
SAN Supply
(A)
(A1)

DAN
(D1) DAN
(D2) SAN
(B2)

LAN_A
SAN
(B1)

RedBox
Power
Supply
VDAN (B)

VDAN
LAN_B
VDAN P0299ENa

Figure 1 - PRP Redundancy Network

Figure 1 shows two similar Local Area Networks (LANs) which have various Nodes in
common. The key features of these networks include:
x With the exception of a RedBox (see below), no direct cable connections can be
made between the two LANs.
x Each of these LANs can have one or more Single Attached Nodes (SANs). These
are normally non-critical devices that are attached only to a single network. SANs
can talk to one another, but only if they are on the same LAN.
x Matched pairs of devices which are critical to the operation of the overall scheme
are connected one to each network at single Dual Attached Nodes (DANs).
x To be sure that network messages (also known as frames) are transferred correctly
to each device at the DAN, each device must have the same Media Access Control
(MAC) code and Internet Protocol (IP) address. This will also mean that TCP/IP
traffic will automatically communicate with both of the paired devices, so it will be
unaware of any two-layer redundancy or frame duplication issues.
x A Redundancy Box (RedBox) is used when a single interface node has to be
connected to both networks. The RedBox can talk to all other nodes. So far as
other nodes are concerned, the RedBox behaves like a DAN, so a RedBox is also
called a Virtual Doubly Attached Node (VDAN). The RedBox must have its own
unique IP address.
x Transmission delays can also be different between related Nodes of the two LANs.
x Each LAN (i.e. LAN_A and LAN_B) must be powered from a different power
source and must be failure independent.
The two LANs can differ in terms of performance and topology. The redundant Ethernet
interface can be made using an optical fiber connection with an LC or ST
connector type or with RJ45 copper connector type. There is no need for an optical
interface away from the relay.

Page (PR) 20-6 Pxxx/EN PR/C22

Parallel Redundancy Protocol (PRP) Notes (PR) 20 PRP Notes

1.5 Structure of a DAN

Each DAN has two ports that operate in parallel. They are attached to the upper layers of
the communications stack through the Link Redundancy Entity (LRE) as in Figure 2:

Dual Attached Node (DAN) Structure of a

Dual Attached
Sending Node Receiving Node Node
(DAN)
Upper Layers Upper Layers

Link Redundancy Link Redundancy

Entity (LRE) Entity (LRE)

Tx Rx Tx Rx

Port A Port B Port A Port B

LAN_A
LAN_B P0298ENa

Figure 2 - Communication between two DANs

The LRE has two main tasks:
x handling message frames and
x management of redundancy
When an upper layer sends a frame to the LRE, the LRE replicates the frame and sends
it through both its ports at nearly the same time. The two frames move through the two
LANs with slightly different delays, ideally arriving at the destination node within a small
time window.
When receiving frames, the LRE forwards the first frame it received to its upper layers
and then discards the duplicate.
As both DAN nodes have the same MAC and IP addresses, this makes redundancy
transparent to the upper layers. This allows the Address Resolution Protocol (ARP) to
work in the same way as with a SAN. Accordingly, to the upper layers of a DAN, the LRE
layer shows the same interface as the network adapter of a non-redundant adapter.
To manage redundancy, the LRE:
x Adds a 32-bit Redundancy Check Tag (RCT) to each frame it sends and
x Removes the RCT from each frame it receives

Pxxx/EN PR/C22 Page (PR) 20-7

(PR) 20 PRP Notes Parallel Redundancy Protocol (PRP) Notes

1.6 Communication between SANs and DANs

A SAN can be connected to any LAN and can communicate with any other SAN or DAN
on the same LAN. However, a SAN which connected to one LAN can not communicate
directly to a SAN which is connected to the other LAN.
A DAN is connected to both LANs and can communicate with the RedBox or any other
DANs or any SANs on either network. For communication purposes, a DAN views the
RedBox as a VDAN.
When a SAN generates a basic frame, it sends the frame only onto the LAN to which it is
connected.
Originating at the SAN, a typical frame contains these parameters:
x dest_addr Destination Address
x src_addr Source Address
x type Type
x data
x fcs Frame Check Sequence (i.e. extra checksum characters added to
allow error detection and correction)
The frame from the SAN is then received by the DAN; which sends the frame to its upper
layers, which act accordingly.
When a DAN generates a frame, it needs to send the frame onto both of the LANs to
which it is connected. When it does this, it extends the frame by adding the 32-bit
Redundancy Control Trailer (RCT) into the frame.
The RCT consists of these parameters:
x 16-bit Sequence Number
x 4-bit LAN identifier, 1010 (0xA) for LAN_A and 1011 (0xB) for LAN_B
x 12-bit frame size
x PRP suffix

Note The Sequence number is a measure of the number of messages which

have been sent since the last system reset. Each time the link layer sends a
frame to a particular destination the sender increases the sequence number
corresponding to that destination and sends the (nearly) identical frames
over both LANs.

Accordingly, originating at the DAN, a typical frame then contains these parameters:
x dest_addr Destination Address
x src_addr Source Address
x type Type
x lsdu Link Service Data Unit
x Padding if needed
x RCT data:
ƒ 16-bit sequence number:
4-bit LAN identifier

Page (PR) 20-8 Pxxx/EN PR/C22

Parallel Redundancy Protocol (PRP) Notes (PR) 20 PRP Notes

12-bit frame size

ƒ 16-bit PRP suffix (0X88 0XFB)
x fcs Frame Check Sequence

LSDU The Link Service Data Unit (LSDU) data allows PRP frames to be
distinguished from none-PRP frames.

Padding After the LSDU data, there may be some data padding. This is added to
frames which would otherwise be too short for conventional network traffic
(minimum frame size is 64 octets).

Size The frame size will vary depending on the contents of the frame and how it
has been tagged by the various SANs and DANs. In VLANs, frame tags
may be added or removed during transit through a switch. To make the
length field independent of tagging, only the LSDU and the RCT are
considered in the size.

Figure 3 shows the frame types with different types of data.

Figure 3 - Frames without and with RCT and padding

The key points about these differing frame structures is that:
x SANs do not implement any redundancy features, so they generate basic frames
which SANs and DANs can understand.
x SANs can still understand the frames that come from DANs, as SANs ignore the
RCT components in frames which come from DANs (a SAN can not distinguish the
RCT from the IEEE802.3 padding)
x If a DAN receives a frame which does not include the RCT component, it sends a
single copy of the frame to its upper layers.
x If a DAN receives a frame which does include the RCT component, it does not
send a duplicate copy of the frame to its upper layers.
x If a DANP can not identify that the remote Node is a DAN, it inserts no RCT.

Rejection Algorithm
A Nodes table maintains a list of all the Nodes (SANs and DANs) on each network. This
table is used by the rejection algorithm to assess the possibility of frames being
duplicates and for network integrity monitoring. The table maintains a list of all the
sequence numbers so it can keep track of the frames which have been sent.
For every node on BOTH networks, the table also maintains:
x The expected next sequence number (so it can establish whether messages are
not being received or are being received in an incorrect order)
x Message counts to show how many messages have been received

Pxxx/EN PR/C22 Page (PR) 20-9

(PR) 20 PRP Notes Parallel Redundancy Protocol (PRP) Notes

x Error counts to show whether frames were received out-of-order or were not
received or were sent/received on the incorrect LAN
x A time field to keep track of when a frame was last received from the node
x An indicator to track whether the node is a SAN or not
The receiver scans the frames starting from the end and uses the algorithm to analyse
the frames it has already received and the various data included in the current frame. The
algorithm then combines the frame data with the information in the Nodes table to decide
whether the frame is the first one it has received, or whether it is a candidate for rejection.

1.7 PRP Technical Data

x One VLAN tag supported.
x 128 VDANs supported.
x Up to 100Mbit/s full duplex Ethernet.
x Dynamic frame memory allocation (page manager).
x Configurable duplicate detection.
x Wishbone interface for configuration and status registers.
x CPU port interface - Ethernet or Wishbone.
x Support for link-local protocols - CPU may send to specific ports only - CPU knows
receive port.
x Configurable frame memory and queue length.
x Duplicate detection with configurable size and aging time.
x MAC address filtering (8 filter masks for interlink, 6 for CPU).
x Support for interfaces with or without Ethernet preamble.

Page (PR) 20-10 Pxxx/EN PR/C22

PRP and MiCOM Functions (PR) 20 PRP Notes

2 PRP AND MICOM FUNCTIONS

2.1 MiCOM Products and PRP

The PRP functions being introduced as part of the overall MiCOM product range provide
additional functionality, which is backwards compatible with existing Schneider Electric
MiCOM equipment. This means that existing MiCOM relays/IEDS can be used on
networks which use PRP functions, with no changes being made to those relays/IEDs.
The new MiCOM products that use the PRP, will interrogate other equipment to
determine the equipment model number, and then use the model number to decide (at
runtime), whether that particular item of equipment can support PRP or not.
The model number with the follow Ethernet board assembled (Digit 7 Hardware option is
N, P, Q or R) provided the possibility for PRP function support. This is shown in Table 1:
Hardware Option Type Example Model Number
“N” at digit number 7 Modulated PRP board P64521NA6M0048M
“P” at digit number 7 Unmodulated PRP board P64521PA6M0048M
2LC + 1 RJ45 ports redundant
“Q” at digit number 7 P64521QA6M0B08M
Ethernet board
3 RJ45 ports redundant
“R” at digit number 7 P64521RA6M0B08M
Ethernet board
Table 1 - MiCOM model numbers for PRP options
The MiCOM relay/IED firmware has been modified to allow the PRP options to be
accepted for the power-up tests in addition to the implementation of the supervision frame
transmission.

2.2 MiCOM S1 Studio Software

The addition of the PRP feature has no impact of the MiCOM S1 Studio support files so
there is no need to upgrade any MiCOM S1 Studio software.

2.3 MiCOM Relay Configuration

There is no need to change the configuration of any relay (as relays which include
support for PRP will be able to recognize other devices which support PRP).

2.4 Hardware Changes

The PRP protocol is implemented using the existing dual redundant Ethernet card as a
starting point. The PRP Frame management is achieved by re-programming the Field-
Programmable Gate Array (FPGA).
The low-level management of the redundant frames is performed within the FPGA; this
being defined as the Link Redundancy Entity (LRE). This will involve the addition of the
Redundancy Check Tag (RCT) to a frame to be transmitted; this identifies the LAN and
the sequence number of the message over the two networks. The FPGA is also
responsible for the stripping of the RCT from received frames and discarding the
duplicated messages such that only a single application frame is received by the Ethernet
processor.
The LRE functionality of the supervision frame transmission is performed by the Ethernet
processor card.

Pxxx/EN PR/C22 Page (PR) 20-11

(PR) 20 PRP Notes PRP and MiCOM Functions

2.5 PRP Parameters

The Redundant Ethernet standard (IEC 62439-3:2012/FDIS) defines several parameters
for the PRP protocol; these being fixed at a default value within this release. The
following values are set:
Parameter Value Description
Supervision Frame Target MAC Address for multicast supervision
01-15-4E-00-01-00
Multicast Address frame
Life Check Interval 2 seconds Period between transmission of supervision frames
This is normal PRP mode, Duplicate address will
PRP Mode Duplicate Discard
not be supported.
Node Forget Time 60 s
Duration that the received message Sequence
Entry Forget Time 400 ms
number will be held to discard a duplicate message.
Duration following reboot for which no PRP frames
Node Reboot Interval 500ms
should be transmitted.
Table 2 - PRP parameter values (for PRP Protocol Version 1)

Page (PR) 20-12 Pxxx/EN PR/C22

PRP and MiCOM Functions (PR) 20 PRP Notes

2.6 Product Requirements

Here is a list of the main Product Requirements for MiCOM products which support PRP:
x The MiCOM relay/IED provides two redundant Ethernet ports using PRP.
x The MiCOM relay/IED must be connected to the redundant Ethernet network as a
Double Attached Node (DAN) using PRP (DAN using PRP is known as DANP)
x The redundant Ethernet interface can be made using an RJ45 or an optical fibre
connection with an LC or ST connector type.
x The management of the PRP redundancy is transparent to the application data
provided via the Ethernet interface.
x The PRP option is available with any of the existing protocol options via the
Ethernet Interface (IEC61850)
x Loss of one of the LAN connections to the device does not cause any loss or
degradation to the Application data over the Ethernet interface.
x The link fail alarm within the MiCOM relay/IED firmware can’t be detected the
failure of the link on ST interface and can be detected for LC and RJ45 interfaces.
x The MiCOM relay/IED supports the transmission of the PRP Supervision frame at a
fixed time period (LifeCheckInterval) of 2s (+/- 100ms)
x Each supervision frame includes a sequence number as defined in the IEC 62439-
3:2012/FDIS specification. This will be incremented for each supervision message
and the value will start from zero following a system restart.
x The MiCOM relay/IED will not process received supervision frames to provide
supervision of the redundant network.
x The MiCOM relay/IED does not provide for the PRP management to be configured
(via either the MiCOM relay/IED HMI or the Ethernet interface). Accordingly, the
default values (as defined within this document) are used for all PRP parameters.
x The performance of the Ethernet Interface is not degraded by using the PRP
interface.
x We are stating that the PRP interface of the Px40 relay will support up to 128
connected DANP nodes.

Pxxx/EN PR/C22 Page (PR) 20-13

(PR) 20 PRP Notes PRP and MiCOM Functions

2.6.1 Abbreviations and Acronyms

Abbreviations / Acronyms Meaning
CRC Cyclic Redundancy Check
DAN Dual Attached Nodes
DANP Double Attached Node implementing PRP
FPGA Field-Programmable Gate Array
HMI Human Machine Interface
IED Intelligent Electronic Devices
IP Internet Protocol
LAN Local Area Network
LRE Link Redundancy Entity
MAC Media Access Control
MRP Media Redundancy Protocol
PRP Parallel Redundancy Protocol
RCT Redundancy Check Tag
RedBox Redundancy Box
RSTP Rapid Spanning Tree Protocol
SAN Singly Attached Node
TCP Transmission Control Protocol
VDAN Virtual Doubly Attached Node

Page (PR) 20-14 Pxxx/EN PR/C22