Professional Documents
Culture Documents
Paris 2020
B3‐302
Data to Decisions: Future‐proof Integration of Substation Intelligence
G. RAJAPPAN, P. JONES D. EROL, V. GLINIEWICZ, Y. WU
Doble Engineering Company Vattenfall
USA Sweden
SUMMARY
Sensors, communications, and analytics technologies are advancing rapidly. These technologies
proffer benefits to organizations that are able to leverage them. In addition to increased
productivity, they serve as a recruiting tool for the technology‐savvy new generation of workers.
But traditional substation architectures, with a preference for hardware over software, hardwired
communication networks over software‐configurable architectures, and point‐to‐point data
integrations over a services‐centric approach, impede the adoption of such technologies.
Vattenfall, a large European utility company, along with Doble Engineering, a USA‐based provider
of sensor, data, and analytics solutions, undertook a project to develop and implement an
architecture for integrating substation data sources with Enterprise data consumers in a scalable,
seamless, and secure fashion. A key prerequisite for this architecture was that it should be
standards based; where industry standards didn’t exist or were still being developed, we adopted
a best practice‐based approach. The goal was to ensure repeatable results built on a solid
foundation and to avoid vendor lock resulting from the purchase of proprietary systems. The
standards that were leveraged in this effort include IEC TC57 Reference Architecture, IEC
Common Information Model, as described in the IEC 61970 and IEC 61968 series of standards, IEC
61850, RESTful web services, ISO 55000 for asset management, services‐oriented architecture,
and Purdue Model for Control Hierarchy. This paper discusses the challenges and the lessons
learned in this project.
KEYWORDS
Digital Substations, Common Information Model (CIM), IEC TC57 Reference Architecture,
Substation Engineering, Condition Based Maintenance, Asset Management, Asset Health Index
(AHI), Analytics, Enterprise Service Bus (ESB).
grajappan@doble.com 1
Introduction
Emerging asset management applications are data and analytics‐centric and aim to provide full
lifecycle management of assets. In order to enable this, all pertinent data need to be accurately
maintained and available on demand to various applications, each of which may be focusing on a
different asset‐related use cases. There are several use cases that pertain to substation assets,
such as condition‐based asset management, protection planning, and security
monitoring/management. The applications that support these are increasingly incorporating
sophisticated analytics, including machine‐learning algorithms, which require all available relevant
data. There is the need for an architecture and infrastructure that supports this need.
We undertook an effort to develop and implement a standards‐based architecture that would
enable current day as well as future analytic applications. Every aspect of our approach was
standards or best practices based and designed with the mindset of scalability and incremental
upgradability to enable future data needs. One manifestation of this is the preference for
software over hardware. For instance, we favored a software gateway for substation data
collection. Another manifestation is the use of emerging and popular standards. We utilized IEC
61850‐90‐2 and REST/JSON web services‐based interfaces for communication of data from the
substations to the Enterprise, wherein we used an IEC Common Information Model (CIM) adapter
to convert the data to CIM‐friendly semantics. We also adopted digital substation concepts,
including IEC 61850 substation engineering process. We utilize this even for legacy protocols – in
our case Modbus. In the following sections, we describe various aspects of our approach.
Goals and Architectural Principles
For this study, the following were the business use cases of interest:
Asset Management: Perform maintenance based on the condition of the asset and not
based on a predefined time interval. This can help reduce maintenance cost and failure
risk.
Operation: Better decision support for operations through access to the Asset Health
Index.
IT: Avoids costs linked to complex integrations where standardized information sharing is
not employed.
Data Security: Secure acquisition of data from the substation and provision to various
analytics applications in the enterprise.
Our overall approach was based on the reference architecture recommended by IEC Technical
Committee 57 (TC57), and it covered information flow across various domains, from the Process >
Field > Station > Operation > Enterprise. As the information flows from the process domain to
enterprise domain, appropriate standards are utilized, such as IEC 61850 closer to the process and
CIM closer to the enterprise. This is shown in Figure 1.
Our goal was to develop a modular and layered architecture utilizing a standards‐based approach
– in particular, the successful implementation of a Service Oriented Architecture (SOA)
integration solution leveraging industry standards (IEC 61850, IEC 61970, and IEC61968) and a
modern SOA Integration Platform, namely an Enterprise Service Bus (ESB), for supporting critical
asset‐related use cases and asset health analytics.
2
Figure 1 – CENELEC Smart Grid Reference Architecture with IEC standards overlay.
In order to achieve the business use cases in the IEC TC57 architecture and standards, the
following were the architectural guidelines incorporated:
Leverage Substation Engineering design process and artifacts, such as SCL files.
Interface structure definition based on IEC 61968‐100.
Interface payload derived from enterprise semantic model (ESM), which is based on IEC
Common Information Model (CIM).
SOA Canonical services based on industry standard model, namely CIM.
Loosely coupled and layered architecture. Avoid tight coupling between solution
components, including components from different layers.
Compatible with and according to Vattenfall’s IT zone model to ensure security.
In the following sections, we will discuss the implementation details.
Implementation Overview
The design process began with understanding the use cases and constraints. The integration
flows were then defined, and service names / functions were identified. A model was created
using the IEC CIM as the starting point for the Enterprise Semantic Model (ESM). The ESM was
extended as needed to enable the generation of canonical service interfaces (WSDL/XSD) and
database definitions (DDL). The ESB services (adaptor and canonical) were then developed on the
ESB platform. Any problems uncovered during testing were addressed at the ESM level (if
appropriate) and brought forward from there. The DBs were implemented in a similar process
and revised in a similar model‐driven manner. The SCL file was used to configure the Substation
Gateway. It was also used to “seed” the cross‐reference DB.
Figure 2 is a detailed view of the test implementation. This figure shows multiple security zones:
Zone 1 in the Station, Zone 2 in SCADA / Operations Technology (OT), and Zone 3 in the Enterprise.
The Zone 1 setup consists of a number of IEDs, a Service PC for configuring the IEDs, and a
substation data gateway. As shown in the figure, the substation gateway is a 61850 gateway that
is incorporated within a Service PC, but in subsequent implementations this has been
implemented as a separate component that supports IEC 61850 as well as legacy protocols. The
substation gateway is described in detail later in this paper.
3
Figure 2 – Illustration of the test implementation.
The Zone 2 ESB has a number of adapters, some of which are for publishing the IEC 61850 and
disturbance event data from the substation gateway, and others for publishing events from Zone
2 data sources such as the SCADA. The data published by these adapters are consumed by ESB
services in Zone 3 and forwarded appropriately with CIM semantics. The “Send Electric Network
Event” service publishes SCADA events. The “Send Disturbance Event” service publishes the
disturbance event files to a storage – the “Get Disturbance Event” service can then be used by
end‐user applications to retrieve this data. The “Get Measurement Value” and “Send
Measurement Value” services are used to provide substation measurement data to end‐user
applications.
Measurements are best utilized with complete context of what they pertain to. This needs
accurate modeling of the measurements, at the core of which is the asset‐equipment
4
measurement mapping. We implemented a reference database for this. We provide a detailed
description of this in the next section.
We implemented a number of end‐user applications, to represent what we anticipate in
production implementation. We used SoapUI, a testing tool for SOAP and REST APIs, as a proxy
for future analytics applications that have standards‐based interfaces. We used an off‐the‐shelf
asset analytics application, dobleARMS, to represent readily available vendor applications, with
interfaces that are generally not yet aligned with the standards vision pursued in our study. We
also used a Real Time Data Historian with end‐user analytic capability.
Data Modeling Considerations
Since measurement data comes from multiple sources and may be consumed by applications for
different use cases, it is essential to model the data correctly. There are three essential concepts:
Measurement, Equipment and Asset.
Measurement represents the concept of what is being observed, not the value itself. For
example, a substation may have temperature measurements and door open indications, a
transformer may have oil temperature, a bay may contain a number of power flow
measurements and a breaker may contain a switch status measurement.
Asset represents a physical and tangible component, such as a circuit breaker, with a
specific serial number, a manufacturer. Assets age and depreciate and therefore need to
be monitored and maintained.
Equipment represents a function or role in the power system and is realized by one or
several assets. Equipment is not tangible. Example of an equipment is the circuit breaker
that can be seen in SCADA – it has a position in the grid and a function to execute.
In the scope of our study, both the concepts of equipment and asset were used. In the Substation
Configuration Description (SCD) document, measurements (signals) are associated to equipment,
and not asset, as the SCD describes the station from a logical/functional point of view, and not an
asset point of view. In order to fully capture measurement relationships, we created additional
bindings to explicitly express the asset‐equipment mappings and associate measurements with
the appropriate entities. We create these bindings based on metadata such as the substation
configuration, as detailed in the IEC 61850 SCD files. This is an important data preconditioning
step, for two reasons:
1. The network role of a physical asset could change, and the data associated with it should
reflect it. For instance, consider a power transformer that, while being refurbished, is
replaced by a spare. This power transformer network role is fulfilled by two different
physical assets at different times. The measurements pertaining to the network role
needs to be associated with the correct physical asset.
2. Some use cases take an equipment centric viewpoint, while others take an asset‐centric
viewpoint. For instance, use cases such as protection planning take an equipment (i.e.,
network role)‐centric viewpoint, and use cases such as condition‐based asset
management take an asset‐centric viewpoint. These different use cases should have
access to all pertinent data immaterial of their viewpoint.
In order to associate measurement to asset, we used the fact that the asset management system
used by Vattenfall, Netbas, holds information both for Equipment and Asset. By combining the
association between measurement and equipment from the SCD and the association between
5
equipment and asset from Netbas, we are able to link measurements to assets. As illustrated in
Figure 3, Equipment is central for the association between measurement and asset as it is the link
between the two domains. It is therefore crucial to have the highest data quality for equipment
information, as missing or erroneous data for this field will lead to missing or erroneous
association between measurement and asset.
Majority of equipment‐asset data originates from the substation. This data is particularly
challenging because it originates from IEDs and other sources that are from different vendors,
have been deployed over a period of decades and consequently have diverse data
representations and protocols. Translating all this data to a common semantic and syntactic
representation is a challenge. Furthermore, it was desired to have a consistent and automation‐
friendly substation engineering process for managing the data collection.
In order to provide scalability and upgradability for future use cases, we decided to use a
software gateway in the substation. The substation gateway was intended to support the
following use cases:
1. IEC 61850 MMS reporting for real‐time data retrieval and MMS file transfer for COMTRADE
file retrieval.
2. IEC 61850 MMS Power Quality data.
3. IEC 60870‐5‐104 and Modbus interfaces to support legacy devices for Asset Management.
4. Future protocols and applications.
For this phase of the investigation, we imposed the following exclusions:
The gateway is not meant for protection and control.
The gateway is not anticipated for use with high‐rate sources like PMU or MicroPMU.
6
In the case of Power Quality monitors, real‐time Sampled Value data is not in scope, only
MMS data for Protection Planning and Asset Management applications.
The exclusions enabled us to pursue a software‐based data gateway, which over time is able to
accommodate higher rate sources as well, due to advancement in technology. The following were
the functional requirements for the gateway:
1. Translation from IEC 61850 MMS, IEC 60870‐5‐101, and Modbus to web services.
2. Buffering to overcome any communication failures.
3. This buffering can be extended with more extended storage to provide local access for
data for HMI and limited trending.
4. Web services server in the future to provide local access of data.
In addition, the following non‐functional requirements needed to be met:
1. Interface for configuration of the gateway.
2. Security features such as logging and access control.
In terms of the functional requirements, while (1) and (2), namely translation and buffering, are
the core near‐term functionalities, (3) and (4) for local data storage and provision may come to be
expected once the initial needs of enterprise data communication are satisfied.
The non‐functional requirement of an interface for gateway configuration is important. Without
some minimal configuration interface, it is inadvisable to deploy the gateway in production. Even
simple tasks such as configuring the IP address and network details of the connected IEDs may
become cumbersome without this. The non‐functional security requirements are important as
well, with organizational or regulatory requirements typically prohibiting the deployment of
applications without some essential security features. For instance, access control that requires a
username and a strong password to be entered prior to making any configuration change to the
device, and recording in event logs of any configuration change that is made.
We accomplished our goals by developing a software substation data gateway that used IEC
61850 substation engineering and data representation concepts throughout, even in the case of
legacy IEDs, such as those that use Modbus protocol. In the case of Modbus, we extended the
Substation Configuration Description (SCD) file format with “Private” elements for the Modbus
configuration. We based the SCD file extensions on the ideas in IEC 61850‐80‐5 draft. The Modbus
SCD file extensions are added to the IED section of the SCD file. Modbus devices can be assigned
to a specific logical device (LD). XML akin to Figure 4 is utilized for Modbus description in the SCD
file. Note the following elements:
"ModbusCommunication" contains the communication specific settings of a Modbus
device (like IP address, TCP port, and the Modbus slave address).
"ModbusService" section that specifies the modbus services to be used.
"ModbusMapping" section defines data points (specifies data types, addresses, and
names for references in the IEC 61850 data model).
The mapping to the IEC 61850 data model is done elsewhere in the SCD file in the LN
elements.
7
Figure 4 – Example of Modbus extensions in the SCD file.
The substation data gateway, once configured in this manner using an extended SCD file, uses IEC
61850 client and Modbus Master components to retrieve the data from the IEDs (south side) and
convert it into an intermediate data representation (IEC 61850 Gateway Data model). This
architecture is shown in Figure 5.
8
The data from station IEDs, IEC 61850‐based or legacy, are collected by a substation gateway,
which then forwards the data using a IEC 61850‐based model to the ESB, wherein it is converted
to a IEC CIM‐based data objects and provided to analytic applications.
Enterprise Data Environment
There are some essential elements for effective asset health analytics:
Leverage all the asset related‐data that is available within the organization. Doing so
provides a more complete and accurate picture of the asset fleet and their significance to
the organization. Having a good data management, and in particular measurement
management, since bulk of the most useful asset data are measured, is essential for this.
Develop a conceptual design and implementation roadmap for a fully‐fledged asset health
solution, including the automation of asset health triggers and subsequent actions such as
work order requests. These additional parts of the asset health solution can then be
implemented incrementally according to a roadmap / masterplan.
Start from organizational objectives, and design analytics platforms/programs that
contribute to the organizational objectives.
Develop strategic and tactical plans for asset management. The analytics
platforms/programs should enable implementation of these plans. For instance:
o The organization should have a strategy for asset replacement and renewal, and
the analytics should enable this decision by providing accurate metrics.
o If analytics produces an alert or if the analytic score crosses a preset threshold on
an asset, the plan should specify how one should respond to these events. If an
organization is trying to figure out a course of action after the event, it may
already be too late.
o Determine critical assets and ensure that good data is available about these critical
assets, which is essential for accurate analytic assessment. In some cases, where
good data does not exist, the organization may rectify the situation through
measures such as more thorough or frequent inspections/testing, and targeted
addition of sensors and monitors.
o Ensure that all the Subject Matter Experts (SMEs) understand and accurately
interpret what the analytic system is telling them, and ensure that other
organizational stakeholders, such as the decision‐makers for asset replacement,
understand the type of analytic input that would drive their decision.
The enterprise data environment created in this study enables complete asset visibility
throughout the asset lifecycle, by providing on‐demand access to high quality data. The ESB
provides richly linked data to subscribing applications using IEC 61968‐100 based CIM
serializations and appropriate CIM‐based payloads. For asset management applications, IEC
61968‐4 based payloads are used.
One of the applications is a data lake that brings together the substation data with other
enterprise data, both structured and unstructured. This one repository of all relevant enterprise
data makes it easy for applications to have their data needs met from a single system, rather than
have to interface with multiple system. The data lake is a fount for a new breed of ad hoc
applications. For instance, we anticipate the use of machine learning approaches to characterize
the salient properties of this data. Widgets such as one to calculate the financial risk of assets,
drawing from disparate pieces of information such as the asset age, condition, criticality ratings
associated with its location, purchase price, and insurance information, can then be put together
9
by a data analyst, and iteratively perfected. This is radically different from the current approach of
having to put together a large project, which could potentially run for years and in the end
produce unsatisfactory results.
Conclusions
We have been working on this R&D project since 2017, and we have developed and demonstrated
critical pieces of the concept in a laboratory environment. We are currently in the process of
making further improvements based on the findings from the laboratory demonstration and
performing further pilot testing, including in live substation environment. In this paper, we
described the standards‐based approach we took to achieve the objectives.
In this project, a working prototype for a substation gateway based on the 61850‐9‐2 technical
guidelines was developed. Additionally, the SCL file used for substation configuration was
extended to incorporate the condition monitoring use case and used in order to configure the
substation gateway prototype. We were able to demonstrate that the concepts highlighted in the
TC57 reference architecture are compatible with current 61850 substation engineering process.
For instance, the substation engineering process could be extended to support legacy protocols
as well and therefore support a real‐world environment where non‐61850 devices need to be
supported as well.
Our study demonstrated that it is possible to provide data consumers (applications or analytic
solutions) with measurement value and event information from Vattenfall primary stations in a
standardized manner. By implementing canonical, application agnostic services, a loosely coupled
SOA is implemented that enables the reuse of services and information flows. This in turn,
enables multiple applications to consume the information.
Our study showed that a service oriented architecture together with the IEC 61850 and IEC CIM
standards allows for reuse of information integration flows. This greatly improves robustness in
the solution, while providing flexibility and isolates the system components against changes in
other components, as opposed to the traditional hardwired point‐to‐point ("Spaghetti")
integrations. It also provides fewer flows and components to monitor and administer, thereby
improving security. This is a prerequisite if Vattenfall wants to move towards the capability of a
more agile development approach of "Smart" applications at the enterprise level.
BIBLIOGRAPHY
[1] International Electrotechnical Commission (IEC), “Smart Grid Standards Map,”
http://smartgridstandardsmap.com/.
[2] G. Robinson, J. Horstman, M. J. Mousavi, and G. Rajappan “Data Analytics for the Smart Grid”
(Book Chapter in Smart Grids: Advanced Technologies and Solutions, Second Edition,
December 2017).
[3] H. Vardhan, R. Ramlachan, W. Szela, and E. Gdowik “Deploying digital substations:
Experience with a digital substation pilot in North America” (2018 71st Annual Conference for
Protective Relay Engineers (CPRE), March 2018).
[4] S. Fuloria and R. Anderson “Towards a security architecture for substations” (2011 2nd IEEE
PES International Conference and Exhibition on Innovative Smart Grid Technologies,
December 2011).
[5] IEC TC57, IEC 61968‐4:2019, “Application integration at electric utilities ‐ System interfaces for
distribution management ‐ Part 4: Interfaces for records and asset management”.
10