Sustainable Energy, Grids and Networks 15 (2018) 26–33

Contents lists available at ScienceDirect

Sustainable Energy, Grids and Networks

journal homepage: www.elsevier.com/locate/segan

IEC 61850-based adaptive protection system for the MV distribution

smart grid✩
A. Alvarez de Sotomayor b , D. Della Giustina a , G. Massa a, *, A. Dedè a , F. Ramos b ,
A. Barbato a
a
Unareti SpA, Via Lamarmora 230, 25124 Brescia, Italy
b
Schneider-Electric, C/ Charles Darwin s/n, 41092 Sevilla, Spain

article info a b s t r a c t
Article history: The installation of protection systems on the Medium Voltage (MV) network is probably the most
Received 15 December 2016 pragmatic topic of the smart grid in those contexts where the regulation of distribution systems includes
Received in revised form 2 August 2017 output-based incentives to Distribution System Operators (DSOs) related to the quality of service. One of
Accepted 6 September 2017
the main obstacles to the diffusion of such a solution is related to the MV network configuration – e.g.
Available online 10 October 2017
the relation between the feeding primary substation and fed secondary substations – which is altered
Keywords: for various reasons, requiring different settings for the protection devices placed along the network. The
IEC 61850 present work describes an innovative approach to the dynamic reconfiguration of protection devices used
Distribution network automation to implement advanced fault location, isolation and service restoration solutions.
Fault localization The proposed solution, realized during the activities of the FP7 European Project IDE4L, has been im-
Smart grid plemented based on the IEC 61850 standard adoption, in order to assure the interoperability, modularity
and scalability of the proposed approach.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction In the context of DAT, self-healing strategies, also known as

FLISR (Fault Location, Isolation and Service Restoration), play an
DISTRIBUTION Networks (DNs) are facing several challenges in important role to significantly improve network reliability statis-
recent years. The deepening penetration of distributed generation tics, increase the energy supplied and provide a high quality of
systems, as well as the higher quality of service required by na- service [6–10]. All these advantages entail a series of economic
tional and international standards, are asking for an improvement benefits by increasing utility revenues coming from augmented
of the DN management [1,2], particularly under faulted condi- energy delivery and/or regulatory authorities incentives, as well
tions [3]. Within this field, the maturity of Distribution Automation as reducing customer costs related to outages. Thus, the research
Technologies (DATs) technologies and the availability of a variety on FLISR solutions is coming at the forefront of interest of power
system technologies related research. In recent years, a large vari-
of communication technologies, gives to utilities the opportunity
ety of approaches based on various automation architectures have
to considerably reduce non-scheduled service interruptions [4].
been proposed to face the issues related to DAT. Many of them rely
Furthermore, such a DAT based approach can be taken into account
on centralized FLISR configurations, where the main logic is placed
because of the cost reduction of Medium Voltage (MV) breakers
in a central server deployed as part of the DN Management Sys-
capable to interrupt short circuit currents. Indeed, in past years,
tem (DMS) [9]. An intermediate approach, generally called semi-
only on load breakers, capable to interrupt current values up to the centralized [11], involves both the Remote Terminal Units (RTUs)
nominal one were used [5]. and the control center in the decision making process.
Other proposals are based on decentralized architectures, with
✩ The research leading to these results has received funding from the European the main logic placed at the substation and/or pole/pad mounted
Union Seventh Framework Program FP7-SMARTCITIES-2013 under grant agree- cabinet levels [11]. In general, semi-decentralized and decentral-
ment 608860 IDE4L — Ideal Grid for All. ized approaches reduce the computational and communication ef-
*Corresponding author. fort at the control center from allowing to speed up the self healing
E-mail addresses: [email protected]
processes at a local level. A pure decentralized solution is reported
(A. Alvarez de Sotomayor), [email protected] (D. Della Giustina),
[email protected] (G. Massa), [email protected] (A. Dedè),
in [12]. The paper presents a logic coordination of protection de-
[email protected] (F. Ramos), [email protected] vices in order to selectively isolate the fault – also called Logic
(A. Barbato). Selectivity (LS) – and describes the impact on the quality of service

https://doi.org/10.1016/j.segan.2017.09.003
2352-4677/© 2017 Elsevier Ltd. All rights reserved.
 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33 27

network to perform protection actions. Indeed, the protection set-

Nomenclature tings may not be adequate for the new operating conditions once
a network reconfiguration occurs or a Distributed Energy Resource
ALSM Faulty MV section monitoring (61850 logical (DER) is added to the network [15]. Furthermore, the relation
node) between a given IED and the upstream/downstream ones may
CDC Common Data Class change due to various network management actions. Thus, future
CID Configured IED Description protection devices and communication standards must foresee the
CLSF Faulty MV section management (61850 logical integration of mechanisms for remote and automatic adjustments
node) of operational parameters, allowing the proper operation of the
DA Data Attribute distribution automation system without interruptions [16].
DAT Distribution Automation Technology Concerning the field of DA, the FP7 European Project IDE4L has
DMS Distribution Network Management System studied, developed and tested a distributed protection system for
DN Distribution Network the MV network, capable to adapt itself in accordance with the
DO Data Object evolution of the power system. This approach is strongly focused
DSO Distribution System Operator on the use of the IEC 61850 standard. More in details: (i) protection
FLISR Fault Location, Isolation and Service Restoration function parameters – as well as the hierarchy among protection
FP7 Framework Program Seven devices – are mapped in the IEC 61850 data model of the devices;
GOOSE Generic Object Oriented Substation Event (ii) those values can be changed on-line by using the MMS protocol,
HV High Voltage e.g. one of the protocol proposed by the IEC 61850 for monitoring
IDE4L Ideal Grid for All and control applications. Such an approach allows to implement a
IEC International Electrotechnical Commission fully decentralized FLISR solution, in order to reduce the outages
IED Intelligent Electronic Device duration and to adapt the settings of the various IEDs to the DN
LAN Local Area Network needs.
LN Logical Node The paper reports how this idea has been put into practice by
LS Logic Selectivity IDE4L. In particular, Section 2 describes the relevant aspects of IEC
LV Low Voltage 61850 concerning the proposed application. Section 3 points out
MMS Manufacturing Message Specification the overall structure of the proposed architecture while Section
MV Medium Voltage 4 depicts the implemented solution. Finally, relevant results and
PS‘ Primary Substation conclusions are presented in Section 5.
PSR Protection System Reconfiguration
PTOC Protection Time Over Current (61850 logical node) 2. Review of the IEC 61850 standard
PTOV Protection Over Voltage (61850 logical node)
RDIR Protection Directional element (61850 logical The IEC 61850 standard is today one of the most com-
node) mon standard for the power system automation [17]. Its main
RREC Protection Autoreclosing (61850 logical node) scope is to guarantee the interoperability among automation de-
RTU Remote Terminal Unit vices from different vendors. To achieve this goal, the standard:
SAIDI System Average Interruption Duration Index (i) introduces an abstract model describing the information which
SAIFI System Average Interruption Frequency Index can be exchanged among the various devices; (ii) introduces a
SAU Substation Automation Unit set of services, e.g. the actions that can be performed based on
SS Secondary Substation this information; (iii) proposes some protocols to implement the
TSO Transmission System Operator information exchange.

2.1. The data model

indicator defined by a European Authority. This approach is based
The abstract model of a device is obtained composing basic
on installing circuit breakers along the MV feeder [9]. Thus, such
entities called Logical Nodes (LNs). Each LN describes a specific
an approach allows the installation of protection systems on the
element in the automation system. Several groups of LNs have
MV network to become one of the most cost-effective smart grid
been defined by the Standard. ‘P—’ and ‘R—’ groups are those ones
solutions in those contexts where the regulation of Distribution
System Operators (DSOs) includes output-based incentives. clustering LNs for protection functions and their related purposes.
Another pillar is related to the high technology readiness of Each LN is a complex structure containing several attributes
Intelligent Electronic Devices (IEDs), necessary to implement such called Data Objects (DOs). Each one of them has a specific class
a DA solution, relying on the experience already made on protec- called Common Data Class (CDC). For example, the PTOC (over-
tion systems for high-to-medium voltage substations. Nonethe- current protection) includes the DO called ‘Str’ corresponding to
less, several practical issues have still to be solved before moving the peak-up of the protection function, containing a general notifi-
from the pilot phase to a real world deployment. One of these cation that the fault has been detected together with its timestamp
issues concerns the performing communication media necessary t and a further attribute – called quality q – indicating if the
to implement LS based solutions [13]. In [14], a comparison among notification is trustworthy or not. The items general, t and q are
different technologies is reported, highlighting pros and cons for called Data Attributes (DAs) and are the leaves of the IEC 61850
each one of them. A second issue is represented by the flexibility tree data model [16].
of the MV network, changing its configuration – i.e. the relation
between the feeding primary substation and the fed MV network 2.2. Abstract services
portion – for several operation and maintenance reasons: to mit-
igate congestions and reduce losses; to prepare the network for Once introduced the data model proposed by the IEC 61850, it
maintenance; to restore the service after a fault, etc. A change is important to discuss two other topics related to the Standard:
on any one of these conditions can cause problems related to the (i) the set of services acting on such an information; (ii) some
settings deployed for the various IEDs placed along the distribution protocols needed to implement the information exchange.
28 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33

Among the most common services, the Report service is used to 3.1. Protection functions
retrieve information from a device with measurement capabilities
to a data collector. The service is implemented according to a Since the ’70s, specific digital protection devices have been
client/server model, periodically or event-based. The counterpart applied in High Voltage (HV) and MV networks. However, the usage
of the Report is the Read service, in which the client asks to the of these devices has been limited until now only to primary sub-
server to release the data once needed. stations or high voltage transmission systems. With the changes
Another important group of services is the one to implement introduced by the European Directive concerning the unbundling
control actions (e.g. a client changes the status of a breaker by in the energy supply chain [21], the primary substations manage-
acting on the server connected to the physical element) and to ment has been shared between Transmission System Operators
update configuration parameters. There are four different control (TSOs) and DSOs, so that the DSO domain starts from the HV/MV
models. Each one of them combines in a different way the various transformer. In the following, only the DSO MV side is consid-
control-related services (namely Select, Cancel, Operate, etc.) to ered, showing how the most used protection functions designed
achieve a different behavior. for standard primary substations can be applied also to the MV
These models can be used also to change such a parameter value distribution grid. A list of the most common protection functions
in order to update IEDs configurations. is reported in Table 1.
The last service discussed in the paper is the GOOSE (Generic 67N is mainly used to detect phase-to-ground faults both in
Object-Oriented Substation Event), which is needed to exchange distribution systems working with isolated or compensated neu-
time-critical information such as the ones related to protection tral configurations. In the case of a phase-to-ground fault, the fault
functions or errors and alarms among the various IEDs at the current is mainly generated by the capacitive component of the
substation level. Data in the GOOSE message is organized in a Data- MV cable. Thus, the zero-sequence current and the voltage have
Set structure as defined by the IEC 61850 data model [18]. a phase delay in a specific range. This characteristic is used by the
67N protection function, which measures the angle between the
2.3. Services concrete implementation zero-sequence current and voltage, in order to detect the fault.
50P is mainly used to detect a phase-to-phase short circuit,
Each of those services is described in an abstract way by the generating extremely high magnitude currents, while 51P is used
standard. However, it also proposes a possible implementation of to detect overload situations maintained for a predefined time
those services by mapping them in a specific protocol. Even if new interval.
protocol mappings have been recently proposed [19] and other After opening the breaker, the automatic reclosing (79) is often
ones are under analysis to be included within the standard, the used to restore the service in case of temporary faults. Commonly,
most common way to implement such a service is to consider the this function works well in case of overhead lines or if the cus-
MMS and the GOOSE over Ethernet protocols [20]. While the first tomers connected to the MV DN are capable to manage the faults
one is an application protocol based on the TCP/IP (ISO-OSI layer 3) in their internal grid.
and one-to-one communication, the second one is mapped directly Protection devices installed along the MV feeders can be also
on the Ethernet level (ISO-OSI layer 2 as a multicast protocol based used as monitoring systems. In particular, current, voltage and
on a publisher/subscriber model. The GOOSE over Ethernet has in power measurements can be stored locally in each substation,
general a simpler packet structure with a small overhead, meaning, where a dedicated automation unit is present.
in general, a higher information exchange speed. For those reasons,
the MMS is a good candidate to implement the Report and the 3.2. Mapping to the IEC 61850 logical nodes
SetValues services, while the GOOSE over Ethernet protocol has to
be used for FLISR logic selectivity among peers IED. The first step for the real implementation of the proposed solu-
tion is to assure that all the parameters pertaining the protection
2.4. Network requirements features could be accessible and then reconfigurable. This is done
by mapping all those parameters in the IEC 61850 data model.
The introduction concerning the protocols required to imple- The data model is specified in the so-called CID file (Configured
ment the standard based solution allows setting a first constraint IED Description). An IEC 61850 fully compliant device should be
on the underlying communication network. Indeed, if the DAT configured only by using this file in accordance with its interfaces.
includes the coordination of protection systems to implement a Nevertheless, up to now, IEDs from many vendors can be config-
distributed protection function, this imply that the GOOSE over ured only in terms of measurements and commands, while the
Ethernet protocol has to be used, requiring a large broadcast do- settings of the various protection functions can be changed only by
main. Indeed, a GOOSE message has to reach all the IEDs connected means of proprietary software tools by specific vendors. It is worth
to the same physical (or virtual) network, with a big usage of to remark that the most of those parameters – those one related
bandwidth due to a high number of small packets circulating in to protection function settings – were already defined by the first
the communication infrastructure. edition of the standard IEC 61850 part 7–4. For example:
Another relevant feature of this communication network is
related to the traffic prioritization, which could assist the delivery • time over-current protections (ANSI 50P/51P) have to be
of critical packets on time, even when there is high volume of directly mapped into a PTOC LN;
traffic on such a channel. Indeed, different services defined by • directional over-current protections (ANSI 67N) have to be
the IEC 61850 may require different performance indexes to the mapped on three LNs: (i) a PTOC to define the zero-sequence
communication infrastructure. Some common requirements are over-current part; (ii) a PTOV to define the zero sequence
the dependability of the network, the latency, the throughput and overvoltage part; (iii) a RDIR to specify the angular sector;
the synchronization accuracy. • reclosing is directly mapped into a RREC LN.
3. Architecture for an adaptive protection system The idea behind the Protection System Reconfiguration (PSR)
is to update IED settings after the fault is cleared and the service
The architectural structure of the proposed adaptive protection restored in order to allow the network reconfiguration also in
system is presented in the following subsections, starting from the presence of DERs, mapped with appropriate LNs in the second
definition of the protection functions to be used. version of the standard. Thus, the second step to implement such a
 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33 29

Table 1
Some common protection functions used on the MV network with their parameters.
Protection function Parameters
67N - ground directional overcurrent Trip delay Td
Zero-sequence current threshold I0t
Zero-sequence voltage threshold V0t
Angle 1 Φ 1
Angle 2 Φ 2
50P – phase instantaneous overcurrent Threshold current It
Trip Delay Td
51P – phase inverse time overcurrent Threshold current It
Time characteristic curve
Curve multiplier k
79 – reclosing Number of reclose cycles (NR)
First reclosing time (RT1 )
Second reclosing time (RT2 )

solution is to make the publisher/subscriber configuration dynam-

ically modifiable as well. This topic is not covered by the standard,
even in the second edition, too. Nevertheless, a complementary
standard [22] – currently in draft version – proposes new LNs to
support the integration of fault passage indicators in the frame-
work of the IEC 61850. Among them, two LNs, provided with some
extensions, were selected for this purpose

• CLSF – Logic selectivity (LS) of faulty MV section manage-

ment. This LN models the role of the local IED controller
in the logical selectivity designed within IDE4L project, in-
cluding information that will be published to be handled
by subscribed peers. The reference document [22] contem-
plated the use of this standard to model logic selectivity
schemes among Fault Passage Indicators (FPI). However, the
IDE4L project contemplates also Logic Selectivity Schemes
among Breaker Controllers to allow applicability in distri-
bution lines where secondary substations are equipped with Fig. 1. Simplified block diagram of a protection device and its IEC 61850 interfaces.
protection functions and circuit breakers. Gray boxes are not in scope of the paper.
• ALSM – Logic selectivity of faulty MV section monitoring.
This LN models the subscription to the information pub-
lished by remote peers related to fault events in order to 4.1. Logic selectivity
block the local operation of the controller over the switch
or breaker to be tripped. With the purpose of enable/disable
The LS application represents the base to carry out the proposed
the block message associated to an IED, its corresponding
adaptive solution, implementing the coordination among the var-
instance of ALSM is provided with a data object called
ious IEDs according to the following principle. When a fault occurs
Downstream IED. When set to 1, peer IED GOOSE message
is configured as a block message. in a portion of the network, the nearest IED isolates it in order to
reduce the number of customers affected by the fault as much as
The whole representation of the proposed IED data model is possible. This IED informs the upstream IEDs that it is clearing the
sketched in Fig. 1, while all the details about protection settings fault by sending them a block message.
and coordination mapped are reported in Table 2. In the example reported in Fig. 2, a fault occurs downstream
Information provided by remote peers in ALSMx is used by the to IED3 . A directional protection function is used to identify its
internal logic when the blocking is enabled for that IED. Thus, presence.
it is possible to block the opening of the switch or the breaker
commanded by protection functions. • IED1 , IED2 and IED3 detect the fault and publish a block mes-
In a radial or open loop grid, blocking should be enabled only for sage on the communication bus using the GOOSE protocol,
IEDs of the same type located upstream. If there is a reconfiguration waiting for incoming messages from other IEDs;
on the network which cause inverse power flow, blocking schema • IED4 does not detect any fault because the fault current
should be adapted to the new configuration to contemplate the presents an opposite direction compared to the one of the
IEDs which are now located downstream. In addition, protection IED4 settings.
functions related settings, such as the reference angle for direc- • IED5 does not detect any fault or current because the cir-
tional protection or current threshold, may need to be adapted to cuit breaker is opened, thus secondary substation is discon-
the new situation. nected. IED5 only detects a drop of voltage caused by the
fault.
4. Demo description • IED6 and IED7 do not detect the fault since the MV network
is open at the secondary substation where IED5 is installed;
The present Section describes the LS and the PSR general ap- some IEDs – according to the block messages that have
proach, with a particular focus on the technical implementation of been published and the configuration reported in Table 3 –
the solution developed a part of the IDE4L project. subscribe the message:
30 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33

Table 2
Mapping of protection functions and logic selectivity coordination upon the IEC 61850 data model.
Functions/ Signals Logical node Data object Attribute Attribute
Logic name Type Name Type Name Type
Curve type: fixed PTOC TmACrv CURVE setCharact ENUMERATED
Zero-sequence current PTOC StrVal ASG setMag AnalogueValue f FLOAT32
threshold
Settings Trip delay PTOC OpDlTmms ING setVal INT32
Directional mode PTOC DirMod ENG setVal ENUMERATED
67N
Zero-sequence voltage PTOV StrVal ASG setMag AnalogueValue f FLOAT32
threshold
Reference angle (Φ 2) RDIR ChrAng ASG setMag AnalogueValue f FLOAT32
Maximum deviation RDIR MaxFwdAng ASG setMag AnalogueValue f FLOAT32
(Φ 2 + Φ 1/2)
Over current PTOC Op ACT general BOOLEAN
Provided
indication
Data
Zero-sequence over PTOV Str ACD neut BOOLEAN
voltage indication
Direction RDIR Dir ACD dirGeneral ENUMERATED
Curve type: fixed PTOC TmACrv CURVE setCharact ENUMERATED
Settings Phase current PTOC StrVal ASG setMag AnalogueValue f FLOAT32
50P/51P
threshold
Trip delay PTOC OpDlTmms ING setVal INT32
Provided Over current PTOC Op ACT phsA BOOLEAN
Data Indication phsB
phsC
Reclosing time 1st RREC Rec1Tmms1 ING setVal INT32
Reclosing time 2nd RREC Rec1Tmms2 ING setVal INT32
79 Settings Maximum number of RREC MaxCyc ING setVal INT32
reclose cycles
Used max number of RREC UseCyc ING setVal INT32
reclose cycles
Circuit breaker RREC OpCls ACT General BOOLEAN
operate enabling
Logic se- Settings Peer IED Locking ALSM DownstreamIED SPG setVal BOOLEAN
lectivity Enabled
(Remote) Control MV Section Fault ALSM MVScFltDecInh SPC stVal BOOLEAN
Detection Inhibition
(Lock from peer IED)
Logic se- Settings Timer(s) for the CLSF TmLogCrd1 ING setVal INT32
lectivity coordination of the
(Local) IED within the logical
selectivity
Provided MV Section Inhibition CLSF MVPredInh SPS stVal BOOLEAN
Data (Lock) to predecessor
controllers

◦ IED3 is configured not to subscribe any block message, the FLISR solution to load each file in their corresponding IED. The
◦ IED2 is configured to subscribe the block message from proposed solution allows the exchange of MMS messages with IEDs
IED3 , to assign new values only to the operational settings that require
◦ IED1 is configured to subscribe the block messages changes after the network reconfigurations.
from IED2 and IED3 ; In addition, new GOOSE communication schemes need further
and more complex changes in CID files, not only settings values.
• since IED3 do not subscribe any block message, it trips its
With the addition of a setting to enable or disable blocking mes-
breaker extinguishing the fault.
sages from peer IEDs, edition of GOOSE subscription in CID files
In case fault passage indicators are present in some secondary is no longer required. Logic Selectivity LN have been proposed, in-
substations, a second step could be introduced to further reduce cluding this new setting to support remote changes of GOOSE com-
the area affected by a fault operating their commutation after the munication schema without interrupting FLISR operation. These
fault isolation.1 LN is neither included in the current version of the standard.
Two main items of information needed to determine the new
4.2. Protection system reconfiguration configuration are:

Concerning the PSR, the protection system has to be reconfig- • the network configuration, e.g. the hierarchy among the
ured in order to be ready to operate when a further fault occurs, feeding PS and fed SSs. According to this hierarchy, the
even if the network is not in the standard configuration. subscription of GOOSE messages has to be enabled/disabled
Current version of IEC 61850 standard does not support remote for any IED. As an example, the mapping shown in Table 3 –
changes of settings. Any change should be performed by loading after the reconfiguration of the network, has to be updated
a new CID file in each IED. In the case of the paper example, this as reported in Table 4, in order to allow the reconfiguration
process involves to edit seven CID files and stop the operation of of the network;
• the network topology e. (in terms of cables, transformers,
1 Taking into consideration that not all the secondary substations are equipped loads, generations, etc.) is needed to calculate protection
with automation systems, the third and last step is always done manually by field setting, e.g. phase-current limits, zero-sequence current
crews. limits, etc.
 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33 31

Table 3
Publisher/subscriber mapping in the standard configuration reported in Fig. 2(a).
IED1 IED2 IED3 IED4 IED5 IED6 IED7
IED1
IED2 X
IED3 X X
IED4 X X X
IED5 X X X X
IED6 X
IED7

Table 4
Publisher/subscriber mapping in the standard configuration reported in Fig. 2(b).
IED1 IED2 IED3 IED4 IED5 IED6 IED7
IED1
IED2 X
IED3 X X X X
IED4 - - - X X X
IED5 X X
IED6 X
IED7

Fig. 3. Simplified architecture for the protection system configuration.

to give the reader useful insights concerning the global automation

implementations, the high-level architecture of the proposed solu-
tion is described in next subsection.

4.3. High level architecture

The peer-to-peer LS is performed by the IEDs located in the

PS and in SSs. The protection configuration update is achieved by
adding a hierarchical control architecture where a further IED –
called Substation Automation Unit (SAU) [24] – is in charge of
supervising the MV network where the LS takes place. This unit
is used to monitor and control, through a set of algorithms, the
network connected to the substation. Each SAU can communicate
with the upper level exchanging commands and measures. It is able
to collect data from the IEDs installed in its concerning area, storing
measures and alarms in a local database. The communication with
the IEDs is realized using the IEC 61850 MMS. Data stored in the
local database is used by a set of algorithms, which provide control
and forecast functionalities defining new set points for the IEDs
and for the related SAUs. Commands and set points are required
and set in each device using the same field interface used to
acquired measures and alarms. Within the SAU, a specific module
has been inserted to defined the new values for the protection
(a) Before the fault. (b) After the fault.
devices thresholds, in order to obtain a configuration of the pro-
tection architecture compatible with the current condition and
configuration of the MV distribution grid. This module is sketched
Fig. 2. Simplified scheme of a primary substation with two MV feeders equipped
with IEDs and breakers: (a) before the fault the open point is between IED5 and
in Fig. 3.
IED4; (b) after the fault location isolation and the service restoration the open point
is between IED3 and IED 4. 5. Results and conclusions

The FLISR scheme and the protection system reconfiguration

Publisher/Subscriber mappings shown in Tables 3 and 4 are described in Sections 3 and 4 have been fully applied to a real op-
defined in accordance with the open point location and the power erating environment in Brescia (North of Italy). The MV network –
flow that reach each IED in stable conditions. shown in Fig. 4 – is composed of: 1 MV/MV substation, 3 MV lines,
The configuration and the topology analysis has to be per- 40 MV/LV substations and 9 MV customers.
formed by specific algorithms made available on simulation and The 3 lines form two rings, managed radially. 2 MV lines will
analysis tools such as [23] which, however, are not in the scope of be fully automated with monitoring, control, protection and sim-
the present paper, which focus is on the technology feasibility of ulation systems, while the third line will be involved mainly in
the proposed adaptive reconfiguration approach. Anyway, in order simulation and for the LV field trial.
32 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33

Table 5
Improvement in the SAIFI/SAIDI indexes as a function of the fault location. Baseline case (BL) versus smart grid (SG) case.
Node Cust. BL SG Fault location SAIFI SAIFI SAIFI SAIDI SAIDI SAIDI
breakers breakers BL (%) SG (%) KPI (%) BS (min) SG (min) KPI (%)
PSLN01 x x
SS0545 102 x 100% 100,00% 0,00% 86,44 51,7 40,19%
SS1464 4 x 100% 100,00% 0,00% 86,44 51,7 40,19%
SS1073 22 x x 100% 100,00% 0,00% 86,44 51,7 40,19%
SS1341 1 x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS1006 18 x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS0468 15 x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS0827 39 x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS1354 4 x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS1340 38 x x 100% 51,15% 48,85% 86,44 26,44 69,41%
SS0603 19 x 100% 7,25% 92,75% 86,44 3,75 95,66%

PSLN02 x x
SS1217 9 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS1023 187 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS1024 176 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS0338 419 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS0145 354 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS1512 94 x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS0378 63 x x 100% 100,00% 0,00% 86,44 76,15 11,91%
SS0605 84 x 100% 18,11% 81,89% 86,44 9,36 89,17%
SS1438 70 x 100% 18,11% 81,89% 86,44 9,36 89,17%
SS0117 134 x 100% 18,11% 81,89% 86,44 9,36 89,17%

used in Local Area Networks (LANs) like those present within a

primary substation. Main features are: (i) a limited number of
nodes (no more than few hundreds in the worst case); (ii) a well
confined geographical area. An host connected to the L2 network
sends/receives messages to/from all the other hosts connected on
the same network. This concept is therefore compliant with the
publisher/subscriber approach described in Section 4.1.
MMS – on the other hand – is an application protocol based on
the TCP/IP. In a IP network (also called Layer 3 network): (i) each
host has an IP address; (ii) in case the network is split in several
subnetworks, there are some special elements called routers which
are in charge to receive packets from a subnetwork and forward
them to the others. L3 networks are a more suitable option when
the number of nodes is higher, as in the case of automation sys-
tems, where IEDs are geographically distributed over a city area
covering some of the SSs which are two orders of magnitude more
than PSs.
To enable the MV automation described in the paper, a mixed
Layer 2/3 approach has been used in order to support both the MMS
and the GOOSE:

• Each SS has been connected by using a mix of technologies,

such as the Fiber Optics, the Broadband Power Line over MV
cables and the Wi-Fi (this network is an extension of the one
implemented for the FP7 European Project INTEGRIS).
• Each SS has been equipped with a switch/router device
configured in such a way that allows the GOOSE to travel
though the communication network as it traveled in a LAN,
while the MMS is based on a standard L3 network where
each SS/PS has its own subnetwork and a router to connect
Fig. 4. Distribution test network.
it to other substations.

5.1. KPIs of the logic selectivity

The following paragraph discusses some technical details about
the implementation of the physical communication infrastructure The description of the KPIs implemented and tested to show
to clarify how the field demo has been implemented. As explained the effectiveness of the proposed approach are defined as in the
in previous sections, there are two main protocol used for imple- following (calculated on a monthly base):
menting the logic selectivity and the adaptive reconfiguration of
the protection system. The former one is based on GOOSE messages • SAIFI (System Average Interruption Frequency Index) – This
exchange, while the latter uses the MMS. KPI estimates the average number of interrupted customers.
With reference to the OSI protocol stack, the GOOSE is a Layer 2 The baseline scenario is considered as the normal function-
protocol. From a practical perspective, L2 protocols are commonly ing of the demo site, before implementing the FLISR with the
 A. Alvarez de Sotomayor et al. / Sustainable Energy, Grids and Networks 15 (2018) 26–33 33

logic selectivity. The Smart Grid (SG) scenario is considered [5] Y. Chollot, J. Wild, T. Berry, A. Jourdan, J. Houss, Decentralized Self Healing
after the deployment of the IDE4L protection devices which Solution Tested in the Framework of GreenLys Smart Grid Project, in: Proc.
IEEE PowerTech Conference, 2013, pp. 1–7.
implement FLIST with the logic selectivity.
[6] P. Balakrishna, K. Rajagopal, K.S. Swarup, Application benefits of distribu-
• SAIDI (System Average Interruption Duration Index) – This tion automation and AMI systems convergence methodology for distribution
KPI estimates the average interruption duration. The base- power restoration analysis, Sustainable Energy, Grids Netw. 2 (2015) 15–22.
line scenario is considered as the normal functioning of the [7] A. Gauci, Smart Grid Fault Location, Isolation, and Service Restoration (FLISR)
demo site, before implementing the FLISR with the logic se- Solutions to Manage Operational and Capital Expenditures, pp. 1–6. Availble
on: http://www.schneider-electric.ca/documents/solutions/FLISRpdf.
lectivity. The SG scenario is considered after the deployment [8] R. Uluski, B. Wojszczyk, H. Tram, Distribution Automation - Smart Feeders in
of the IDE4L protection devices, which implement FLISR a Smart Grid World, Utility University Course (UU304), DistribuTECH, 2009.
with the logic selectivityTable 5. [9] T. Bensley, C. Grommesh y P. Stemborg, Implementing new configurable self-
healing smart grid technology with an existing distribution management
As it is possible to note from Table 5, the improvement of the system (DMS), Cooper Power Syst. (2011) 1–8.
SAIFI indicator is 31.58% on average, while the average improve- [10] J. Romero Aguero, Applying self-healing schemes to modern power distribu-
tion systems, in: Proc. IEEE Power and Energy Society General Meeting, 2012,
ment of the SAIDI indicator is 49.18%. The data included in Table 5 pp. 1–4.
were calculated assuming an average time for manual operations [11] R.W. Uluski, Using Distribution Automation for a Self-Healing Grid, IEEE, 2012,
equal to 20 min. pp. 1–5.
It is worth to point out that these results were calculated under [12] D. Della Giustina, A. Dedè, G. Invernizzi, D. Pozo Valle, F. Franzoni, Smart grid
automation based on IEC 61850: An experimental characterization, IEEE Trans.
the assumption that the network is always in the standard con-
Instrum. Meas. 64 (8) (2015) 2055–2063.
figuration. The main advantage of the proposed system compared [13] N. Kashyap, C.W. Yang, S. Sierla, P.G. Flikkema, Automated fault location and
to a standard automation approach is represented by the dynamic isolation in distribution grids with distributed control and unreliable commu-
reconfiguration of the system, especially in the case of a network nication, IEEE Trans. Ind. Electron. 62 (4) (2015) 2612–2619.
reconfiguration caused by a fault. Indeed, in this case, a standard [14] D. Della Giustina, S. Rinaldi, Hybrid communication network for the smart grid:
Validation of a field test experience, IEEE Trans. Power Deliv. 30 (6) (2015)
automation approach cannot work immediately because of the 2492–2500.
required reconfiguration realized sending the CID files to the IEDs. [15] L. Huchel, H.H. Zeineldin, Planning the coordination of directional overcurrent
Thus, in case of absence of an adaptive reconfiguration approach, relays for distribution systems considering DG, IEEE Trans. Smart Grid, 7(3)
the percentages shown in the table decrease every time a change 1642–1649.
[16] D. Della Giustina, A. Dedè, A. Alvarez de Sotomayor, F. Ramos, Toward an
in the network configuration occurs, proving the effectiveness of a
adaptive protection system for the distribution grid by using the IEC 61850, in:
PSR. IEEE International Conference on Industrial Technology, 2015, pp. 2374-2378.
[17] A. Dedè, d. Della Giustina, G. Massa, L. Cremaschini, Toward a new standard
5.2. Conclusions for secondary substations: the viewpoint of a distribution utility, IEEE Trans.
Power Deliv. (2016) 1–10.
[18] P. Parikh, I. Voloh, M. Mahony, Fault location, isolation, and service restoration
Considering the need of a more efficient way to reconfigure the (FLISR) technique using IEC 61850 GOOSE, in: Proc. Power and Energy Society
protection system when a change occurs on the power grid con- General Meeting, 2013, pp. 1–6.
figuration, the paper has proposed a possible IEC61850 mapping [19] IEC 61850-8-2, Communication networks and systems for power utility au-
for some functions commonly used for the MV level. Requirements tomation - Part 8-2: Specific Communication Service Mapping (SCSM) –
Mapping to Extensible Messaging Presence Protocol (XMPP).
and configurations of such a solution have been described and
[20] IEC 61850-8-1, Communication networks and systems for power utility au-
some relevant KPIs have been defined and calculated to show the tomation - Part 8-1: Specific Communication Service Mapping (SCSM) –
effectiveness of such an adaptive solution. Mappings to MMS (ISO/IEC9506-1 and ISO/IEC 9506-2) - Ed.2.
[21] Directive 96/92/EC of the European Parliament and of the Council of 19 De-
References cember 1996 concerning common rules for the internal market in electricity.
[22] IEC TR 62689-100:2016. Current and voltage sensors or detectors, to be used
for fault passage indication purposes - Part 100: Requirements and proposals
[1] N. Kanwar, N. Gupta, K.R. Niazi, A. Swarnkar, An integrated approach for
for the IEC 61850 series data model extensions to support fault passage
distributed resource allocation and network reconfiguration considering load
indicators applications.
diversity among customers, Sustainable Energy, Grids Netw. 7 (2016) 37–46.
[23] D. Pala, C. Tornelli, G. Proserpio, An adaptive, agent-based protection scheme
[2] G. Massa, G. Gross, V. Galdi, A. Piccolo, Dispersed voltage control in microgrids,
for radial distribution networks based on IEC 61850 and IEC 61499, in: Proc.
IEEE Trans. Power Syst. 31 (5) (2016) 3950–3960.
CIRED Workshop on Integration of Renewables into the Distribution Grid,
[3] H. Zhan, C. Wang, Y. Wang, X. Yang, X. Zhang, C. Wu, Y. Chen, Relay protection
2012, pp. 1–4.
coordination integrated with optimal placement and sizing of distributed
[24] S. Lu, S. Repo, D. Della Giustina, Standard-based secondary substation
generation in distribution networks, IEEE Trans. Smart Grid 7 (1) (2016) 55–65.
automationUnit–the ICT perspective, in: Proc. 5th IEEE PES Innovative Smart
[4] V.C. Nikolaidis, E. Papanikolaou, A.S. Safigianni, A communication-assisted
Grid Technologies, ISGT Europe, 2014, pp. 1–6.
overcurrent protection scheme for radial distribution systems with distributed
generation, IEEE Trans. Smart Grid 7 (1) (2016) 114–123.