Paper - ZERO-DELAY IS NOT THE ANSWER FOR PROTECTION SCHEMES-Published

Public Channel / PACS

Share on Social Networks

Share Link

Use permanent link to share in social media

Share with a friend

Please login to send this document by email!

Embed in your website

Select page to start with

Post comment with email address (confirmation of email is required in order to publish comment on website) or please login to post comment

8. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 [7] IEEE 802.1D™-2004, “Local and Metropolitan Area Networks—Media access control (MAC) Bridges”, IEEE, 2004 [8] IEC 61850-8-1, “Communication networks and systems for power utility automation - Part 8-1: Specific communication service mapping (SCSM) - Mappings to MMS (ISO 9506-1 and ISO 9506-2) and to ISO/IEC 8802-3”, IEC, 2011 [9] IEC 61850-9-2, “Communication networks and systems in substations - Part 9-2: Specific Communication Service Mapping (SCSM) - Sampled values over ISO/IEC 8802-3”, IEC, 2004 [10] IEEE 1588-2008, “Precision Clock Synchronization Protocol for Networked Measurement and Control Systems”, IEEE, 2008 BIOGRAPHY Daniel Espinosa. Electrical Engineer B.S. degree from Instituto Politécnico Nacional. Professor at Universidad Autónoma de México, for Electrical Science Specialization degree, covering Substation Protection and Automation Systems. Has been working at Comisión Federal de Electricidad since 1998, on protection and substation automation systems, design, specification, standardization and testing. daniel.espinosa@pwmc.mx Page 8 of 8

3. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 network conditions like week in-feed and distance to the fault. Above numbers are just references and, in some cases, just the average operation times from different fault types and power network configurations. These operation times variations should be taken in account when protection settings are coordinated. In some cases, like the ones for backup protection schemes, the longest time is used to avoid unnecessary circuit breakers tripping to clear a fault, like a Breaker Failure Protection (50BF – RBRF). FAULT CLEAR In order to clear a fault, any source should be disconnected. This means to disconnect local and remote sides of the transmission line; in a bus bar fault, all circuit breakers connected to it; and so on. This time is the desirable method, it will always be designed to affect the minimum load feed and reduce affects in components’ expected life or power network stability. When one of the primary methods to detect and clear a fault fails, meaning no just the strictly required power network elements, a backup protection should operate. In the first case, transmission media should provide the shorter time to transmit messages. In the second, you should consider communication network recovery time. LAN PERFORMANCE For LAN we should understand a Local Area Network using interconnected LAN Switches, providing a way to interchange Layer 2 frames, according with IEEE 802.2 LAN [6] . Switches interconnection, could daniel.espinosa@pwmc.mx Page 3 of 8 Figure 1 : Total fault clear time, considering system stability

5. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 time to 50 milliseconds, in order to avoid system instabilities and start backup protections like Breaker Failure (50BF – RBRF), calculus include remote backup protections. Figure 2 : Total fault clear, considering network recovery time HIGH AVAILABILITY PROTOCOLS IEC 62439-3 standard, provides several parts, one per each high availability protocols, some of them like PRP and HSR provides ZERO recovery time. PRP use two networks at the same time, in order to avoid any recovery time present when one network fails. HSR, use just one ring network and only HSR capable devices are allowed in the network; it avoids use of LAN Switches. Both protocols duplicate frame signals, PRP in two independent networks to avoid collisions and HSR in two directions, while collisions are handled by each participant node of the network. PRP can use ring topologies for each network. Then, RSTP or MRP are allowed transparently. Is the device connected to the network, which is supposed to support PRP duplicated messages. HSR don’t allow any standard IEEE 802.2 capable devices. PRP have an IEEE 802.2 capable connection but is single non-redundant connection to each network. PRP and HSR are incompatible between them. It is impossible to mix PRP capable devices in HSR networks. Non HSR capable devices (nodes) are not allowed in its ring network. MRP is considered a high availability protocol, but just LAN Switches are supposed to support it, in order to create a ring. While RSTP have no a part in IEC 62439 standard is considered a high availability protocol, with a deterministic recovery time. daniel.espinosa@pwmc.mx Page 5 of 8

1. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 INTERNATIONAL COLLOQUIUM “ BUILDING SMARTER SUBSTATIONS” Mexico City, Mexico ZERO-DELAY IS NOT THE ANSWER FOR PROTECTION SCHEMES 213 DENIEL ESPINOSA PWMC Mexico SUMMARY Protection schemes have been developed to trip in a short time. Ideally, in most cases, should be zero. This is not possible because the number of devices involved in between trip signal and fault clear. When auxiliary relays were used, sometimes to protect binary outputs of protection relays against circuit breaker’s trip coil make/break demands, trip signal should wait until these devices make to energize trip circuit. Once a trip was received, energizing circuit breaker’s trip coil, requires to open and extinguish arcs in order to clear high fault currents. This is a time to be considered when clear a fault. Then we have: Where: T clear = Total time to clear a fault P op = Protection Relay operation time Aux op = Auxiliary device/media operation/transmission time CB op = Circuit Braker Operation Time When auxiliary relays were replaced by communication protocols like GOOSE messages, Aux op is the time required to be generated, transmitted and received by publisher and subscriber. Then: Where: T trans = Total time to transmit a message for tripping daniel.espinosa@pwmc.mx Page 1 of 8

7. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 You should calculate, each time, recovery time for your network and check its effect in your system stability; primary equipment expected life time; and if you can coordinate all local and remote backup protections. When GOOSE [8] messages are used for tripping circuit breakers, you can choose from different high availability protocols providing zero recovery time, at the cost of incompatible systems. Consider the impact on backward compatibility of IED’s with existing installed base, when choosing high availability protocols; specially if your systems already use RSTP or MRP redundancy networks, with standard IEEE 802.2 [6] fail-over redundant connections. This is because your IED are incompatible with PRP and HSR. In order to avoid incompatible systems and to optimize spare parts availability, choose one and only one high availability protocol. PRP, HSR and IEEE 1588 [10] could receive updates in a backward incompatible way, if so, many systems already running with these protocols should be updated. It is mandatory to request Vendors to supply IED and Switches with support for these upgrades, in order to avoid device replacement. Adaptive algorithms should be analyzed in detail in order to take all advantages of Sample Values over Non-Zero recovery time networks. BIBLIOGRAPHY [1] IEC 62439-1, Industrial communication networks – High availability automation networks – Part 1: General concepts and calculation methods, IEC, 2010. [2] IEC 62439-2, “Industrial communication networks – High availability automation networks – Part 2: Media Redundancy Protocol (MRP)”, IEC, 2010. [3] IEC 62439-3, “Industrial communication networks – High availability automation networks – Part 3: Parallel Redundancy Protocol (PRP) and High-availability Seamless Redundancy (HSR)”, IEC, 2010 [4] Daniel . Espinosa, Rene. Aguilar, James. Ariza, Jorge Mendoza, “Testing Ethernet Network For GOOSE Tripping”, IEEE Sección México, Reunión de Verano de Potencia, 2010. [5] Daniel Espinosa, Isaac Góngora, Ricardo Velazquez, Jesús Coronel, “Desempeño de RSTP y MRP para disparo de interruptores en subestaciones eléctricas”, IEEE Sección México, Reunión de Verano de Potencia, 2014. [6] IEEE 802.2™-1998 (ISO/IEC 8802-2:1998), “Information technology — Telecommunications and information exchange between systems—Local and metropolitan area networks — Specific requirements — Part 2: Logical Link Control”, IEEE, 1998 daniel.espinosa@pwmc.mx Page 7 of 8

6. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 SAMPLED VALUES NETWORKS While sample values as defined in IEC 61850-9-2 [9] , provides a way to transmit analog values over Ethernet channels. These signals require a High quality synchronization between sampled values, in order to be used by protection or metering algorithms, due to high dependency on signal wave form and relations between different sampled values sources, like current and voltage for impedance calculations or differential protection for bus bars or lines. Is possible to use PTP, defined in IEEE 1588 standard, for cost effective distribution of synchronization signals. PTP requires support in LAN Switches in order to distribute and compensate errors, giving a high quality synchronization comparable with IRIG-B. Sample Values and PTP synchronization have issues when PRP or HSR are used, due to duplicating signals. Changes on PRP or HSR can be expected in order to support synchronization over Ethernet as PTP provides, should consider backward compatibility if so. The worse case will be to modify PTP in a backward incompatible way . ADAPTIVE ALGORITHMS ON SAMPLE VALUES NETWORK FAILS Using Non-Zero recovery time for sample values, requires changes on protection and metering algorithms, in order to:  Block protection function on sampled values lost, like when a network recovery is in progress  Re-entrant calculations for fault detection, when delayed sampled values are received after a network recovery  Automatic re-start a new calculation based on new sampled values, dropping old ones In order to recover sampled values, LAN Switches should provide large buffers to store message frames in the period of recovery. In order to reduce buffer sizes, recovery time should be as short as possible. CONCLUSIONS When GOOSE [8] messages are used for tripping, is not mandatory to have zero recovery time protocols. Non-Zero recovery time protocols are options to choose to, while the network is short enough in RSTP; but a real option when MRP is used. As for IEC 62439-1 [1] and experiences [4] [5] , you should avoid mesh networks when RSTP is used for redundancy. In order to improve network recovery time in one ring networks, is recommended to use MRP, defined in international standard IEC 62439-2 [2] . daniel.espinosa@pwmc.mx Page 6 of 8

4. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 happened in different ways, physical media (wire, fiber optic) and with a different number of connections between two or more switches. A ring interconnection can happen with one point to point closed loop connection between switches. While a mesh interconnections, means more than one interconnection points between switches, creating a number of rings; this rings could be constructed without previous plan. In order to clear a fault in the shortest time, transmission media should take and deliver messages in order of milliseconds. Fortunately, LAN Switches are able to transmit GOOSE, as defined by IEC 61850-8-1 standard, message frames in less than 1 millisecond even with high traffic on a 100MB network. This is considering no environment interference is present, like when using short UTP cables or Fiber Optic. LAN REDUNDANCY Once a network is used for tripping, is a good idea to provide a level of redundancy. Different paths to transmit signals could be a problem, when used with Ethernet. On IEEE 802.2 LAN [6] redundancy means, in ring or mesh networks, lost of messages traffic when the network is recovering from a fail. IEC 62439-1 [1] , recommends to use a single ring, with a specific configuration, to provide deterministic recovery times when RSTP is used. For redundancy RSTP was developed and enhanced [7] to provide deterministic and shorter recovery times in less than 32 milliseconds with 9 switches [4] [5] , connected in a single ring and optimized configuration. IEC 62439-2 [2] , offers an enhanced protocols other than RSTP, called Media Redundancy Protocol, for short MRP. This protocol has been optimized for one ring configurations and provides a recovery time of less than 10 milliseconds, in worse case, for up to 14 interconnected switches or 13.22 milliseconds for up to 30 switches. Because this numbers, was not possible to measure recovery time when 9 switches were used in test case in [5] . In order to offer complete redundancy, is recommended to use devices with redundant connection Ethernet ports. These redundant ports could be standard IEEE 802.2 fail-over when using RSTP or MRP. MAXIMUM RECOVERY TIME According to IEC 62439, is possible to calculate recovery time for MRP [2] and RSTP [1] , in worse case. With this in mind and taking in account typical settings for backup protections, is possible to estimate maximum recovery time for a local network, in case it is used for tripping. In Figure 2, we see an estimation using a factitious time for system stability, while typical settings and operation times for different components is used. Based in these, is possible to set maximum recovery daniel.espinosa@pwmc.mx Page 4 of 8

2. INTERNATIONAL COUNCIL ON LARGE ELECTRIC SYSTEMS STUDY COMMITTEES B3, B5 AND D2 P goose = Time required by Protection Relay to “push” a message in transmission media M trans = Time required by media to transmit a message frame to subscriber TD receive = Time required by subscribed device to recognize and validate a message, plus the time to activate binary outputs to energize trip coils. Both P goose and TD receive are difficult to measure; while M trans could be estimated by manufacturer, based on network latency and media propagation. When we combine all these we see with a level of certainly, that we have NON-ZERO delay on tripping. This is especially true, when we should consider local and remote backup protections. They have intentional delays in order to provide time enough for a normal trip of main protection schemes. On LAN Networks, based on IEEE 802 series known as Ethernet, several standard protocols have been developed to transmit frames from one end to another, allowing interoperability at this level. Layer 2 in OSI model represent this level of information interchange. High speed can be reached when Layer 2 frame is used, as for GOOSE messages have showed in practice. This is the T trans , when replacing auxiliary relays. But LAN Network, like Protection Schemes, is not perfect and tends to fail. This way we have local backup protections and, in a power system, remote located protections. In local LAN, we have IEEE 802.2d standard, optimized over time to provide deterministic recovery times as described in IEC 62439- 1 standard. One key with one technology or other, for protection scheme coordination, is deterministic operation time, conditions of fails and alternatives to correctly, as short as possible, clear faults. Recent advances in Ethernet, makes it an alternative to transport trips using IEC 61850 GOOSE protocols over IEEE 802.2 networks, because it is now possible to deterministically calculate transmission times. Like in protection schemes, LAN should provide alternative paths to transmit messages. Unfortunately, this paths could increase transmission times. As for settings coordination, this should be taken in account and should be deterministic too, and as short as possible, but again in some cases it can’t be ZERO. KEYWORDS GOOSE Sample Values Protection Schemes 802.2 802.2d non-zero trip PROTECTION RELAY OPERATION TIME Typically a high voltage protection relay could operate in less than 35 milliseconds and up to less than 16 milliseconds. This could be measured using a Test Set. Operation time of protection relays is not always the same, depending on used principle, like with Distance Protection one which is affected on pwer daniel.espinosa@pwmc.mx Page 2 of 8

Views

  • 285 Total Views
  • 250 Website Views
  • 35 Embeded Views

Actions

  • 0 Social Shares
  • 0 Likes
  • 0 Dislikes
  • 0 Comments

Share count

  • 0 Facebook
  • 0 Twitter
  • 0 LinkedIn
  • 0 Google+

Embeds 2

  • 1 librescl.pwmc.mx
  • 2 www.pwmc.mx