Electrical Power and Energy Systems 94 (2018) 67–76

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

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

Monitoring and controlling energy distribution: Implementation of a

distribution management system based on Common Information Model
L. Fiaschetti a,c,⇑, M. Antunez b,c, E. Trapani c, L. Valenzuela b,c, A. Rubiales a,c, M. Risso b,c, G. Boroni b,c
a
Comisión de Investigaciones Científicas de la Provincia de Buenos Aires (CIC-PBA), Buenos Aires, Argentina
b
Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina
c
Pladema Institute, Gral. Pinto 339, 7000 Tandil, Argentina

a r t i c l e i n f o a b s t r a c t

Article history: The management of renewable and distributed energy resources is changing the way in which electric
Received 13 January 2017 distribution systems are being operated. Many companies commonly create several custom system solu-
Received in revised form 24 May 2017 tions to manage power distribution. The functions provided by these systems are network visualization,
Accepted 23 June 2017
state estimation, system control and data acquisition, among others. Despite their relevance, many mid-
Available online 30 June 2017
size or small distribution companies have problems to install one of these solutions because they are
either very expensive, incompatible, or limited by difficulties in data exchange. As a possible solution
Keywords:
to this problem, this paper presents the development of a Distribution Management System (DMS) based
Open architecture design
Common Information Model
on open source technology. This system combines four main components: OpenDSS-framework,
Smart grid ActiveMQ-broker, applications for visualizing and editing electrical network, and Common Information
Electrical engineering Model (CIM). The tool is part of Tenergia R&D project 1 and has a collection of applications designed to
Distribution management system monitor and control the distribution networks located in Argentina.
Computer-aided engineering Ó 2017 Elsevier Ltd. All rights reserved.
Power system metering

1. Introduction commercial tools such as DIgSILENT PowerFactory [9], ETAP [10],

Schneider [11] are available.
There is a significant interest for the electrical industry to Several works [12,13] present the benefits of installing a DMS,
develop systems supported by computer-aided engineering [1]. both for the end users and the distribution company. The process
In order to attend decision making and maintenance activities, of instantiating a DMS requires specific software, communication
these systems should provide several functions: Network Connec- infrastructure and hardware for the automation of different net-
tivity Analysis (NCA), Distribution System State Estimation (DSSE), work components. In [14] it is shown that the software represents
Load Flow Applications (LFA), Volt/Var Control (VVC), Fault Man- 20% of the total value for automating the network.
agement & System Restoration (FMSR), Distribution Load Forecast- The implementation of such systems can be achieved develop-
ing (DLF), among others [2–4]. ing solutions that could use existing open source components, and
A Distribution Management System (DMS) is a set of software incorporating monitoring and communication network functions.
components which monitor and control the network with the In addition, the cost factor is interesting for the proposal. According
aim of optimizing the behavior of an electrical smart grid, and to reference, for an investment of 28 million dollars in the first
therefore enabling utility companies to provide power in a more 3 years, the capital was recovered after 5 years, and the profit
reliable, efficient and safer way [5–7]. Moreover, it improves out- reached 51% of the invested capital after 10 years [14]. Although
age reduction, outage time minimization, maintenance of accept- the values are interesting for any company, only a few distributors
able voltage levels and line loss reduction [8]. Many research and can afford such an investment. Considering that open source com-
ponents allow cost reduction, the architecture is adapted to the
needs of the region.
Moreover, this kind of systems have to be supplemented with
⇑ Corresponding author at: Pladema Institute, Gral. Pinto 339, 7000 Tandil, appropriate message passing and synchronization capabilities in
Argentina. a prescribed format. For that reason, standards are being used to
E-mail address: lfias@exa.unicen.edu.ar (L. Fiaschetti).
1
facilitate the data exchange between applications within a com-
This work is financed by R&D projects FITS 2013 – UREE10 – FONARSEC –
Ministry of Science, Technology and Productive Innovation of Argentina, and PIDDEF pany, as well as with other companies. These two standards are
36/14 – Ministry of Defense of Argentina. the IEC 61970–301 and the IEC 61968–11, which correspond to

http://dx.doi.org/10.1016/j.ijepes.2017.06.029
0142-0615/Ó 2017 Elsevier Ltd. All rights reserved.
68 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76

the Common Information Model (CIM) and Distribution extensions mentation of the DMS is presented, with characteristics such as
[15]. Although CIM is primarily designed for node-breaker repre- flexibility, extensibility and evolution, opposed to the monolithic
sentation, with certain extensions it can be leveraged to achieve architectures currently on the market. Instead of developing an
a standard interoperable bus-branch model representation and independent solution, technologies that are well tested are com-
exchange format. The power system network can be modeled bined and effectively used in the industry. Specifically, this DMS
using CIM with as many abstraction levels as required. In [16] is developed by connecting several well-known technologies with
three models (physical, topological, and dynamic) to represent a collection of applications designed to monitor and control distri-
power network information on the developed DMS are described. bution networks located in Argentina.
If more accessible analysis tools are needed, flexible and reliable The analysis tools need to have a clear representation to syn-
open source software packages should be used [17]. In this context, chronize real-world and simulation behavior. OpenDSS is a flexible
OpenDSS is a comprehensive open source tool for electric power and reliable open source tool to model and simulate the electrical
distribution systems developed by the Electric Power Research behavior of the distribution network.
Institute (EPRI) [18] to model and simulate the electrical behavior One important requirement of the current tools is the integra-
of the distribution network. Another characteristic of OpenDSS is tion of power flow modeling and simulation environments. A
that it includes a COM interface to communicate with other appli- mechanism to separate the application logic from the simulation
cations, providing flexibility and modifiability [19,20]. execution is another important requirement. In order to combine
Furthermore, being able to combine applications is important if different applications, it is essential to use a standard model with
a robust messaging scheme is required. Most messaging applica- several levels of abstraction so as to export and import a variety
tions are implemented as a message broker which connects pro- of tools. The chosen data model should enable the generation of
ducers with consumers, where producers send messages and specific input models with a coherent transformation.
consumers process them [21]. There are many message systems The main motivations for using CIM to model the network are
that can be used to communicate information within different the availability of tools and libraries to support this format and
parts of the systems, such as RabbitMQ, ActiveMQ, OpenMQ [22]. the high degree of compatibility with other industry standards.
The ActiveMQ broker is chosen because it is faster on message These facts facilitate the application integration of various dis-
reception. This is an essential requirement in a system where sev- tributed systems. Thus, CIM standards is used to facilitate the
eral devices are constantly sending information [22]. exchange of data between applications within a company and
The electrical companies have installed substantial monitoring the exchange of data between different companies.
equipment that regularly sends relevant information of the electri- In turn, these developments are encouraged by research-
cal network. In order to save this information, a variety of NoSQL and funding entities to research on strategic issues in science, technol-
SQL databases [23] are studied. In this sense, the NoSQL databases ogy and innovation policy. Thus, they support new ideas and
show more benefits when the analysis of large volumes of data strategies to develop products and services. In this regard, there
[24] is required, which is very important in the distribution sector. is an opportunity to implement a DMS that provides the specific
Thus, Apache Cassandra [25] is selected due to present a good per- functionality to the companies taking into account their local
formance in all the tests. Moreover, this is an open source database, needs. The developed tool is being applied in Tenergia R&D pro-
which is one of the requirements defined by this development. ject1. It is intended that the resulting tool will be generically
An EMS architecture is shown in [26]. Although it is useful in applied to regional studies in certain selected regions of Argentina
our proposal, it adopts only a static CIM to represent and update and, consequently, be available to a larger community.
information. This feature degrades real-time data management
performance. Besides, data acquisition and device control are not
3. Implementation
represented. Additionally, in [27] it is shown that software
OpenDSS provides many useful algorithms for the distribution sec-
As mentioned above, most DMS are based on monolithic archi-
tor. However, it cannot support real-time execution.
tectures. At present, the new challenges in the distribution sector
Finally, current systems are very expensive and lack features
need new integrated tools to help and improve the operation of
such as oriented services, modularity, common standards and flex-
the electrical system. Thus, it is necessary to incorporate quality
ibility. Another drawback worth mentioning is the dependence on
attributes such as flexibility, adaptability and interoperability
providers. Thus, the main purpose of this work is to design and
without losing performance and security. An important decision
implement a DMS architecture based on open source technology
is to use ActiveMQ as a message broker to communicate informa-
which provides the requirements mentioned above in contrast
tion between different parts of the system. The broker pattern
with existing monolithic structures. Moreover, the combination
allows to decouple the operation of the subsystems, and it man-
of OpenDSS, ActiveMQ, CIM and NoSQL databases on the proposed
ages communication between the modules. In addition, to make
architecture allows the incorporation of new algorithms and tools
a DMS safer, security tactics are implemented based on sessions
dynamically in contrast with the closed and private existing
and encryption. Besides, this architectural pattern allows the
systems.
implementation of several availability mechanisms which make
The following section shows the motivation for using the men-
the system more fault tolerant. Moreover, new technologies in data
tioned applications and the requirements for their interconnection.
bases are used to save all the obtained information. Mainly, the
The proposed architecture is demonstrated in Section 3 and
NoSQL databases are optimized to use data mining and machine
described in Section 4 considering the most important implemen-
learning algorithms which are very useful for the proposed system.
tation aspects. Section 5 depicts several tests which demonstrate
In this section, an analysis of open source components and
the feasibility of using the system over a real distribution network.
information standards integrated into this project is presented.
Finally, conclusions are presented in Section 7.

3.1. Applications for visualizing and editing electrical networks

2. Motivation and requirements
The proposed system has an operation diagram representation
The main purpose of this article is to contribute with solutions (schematic and geographic) which allows the user to create, mod-
applied to the distribution grid. Therefore the design and imple- ify and operate the power network. To implement the schematic
 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76 69

interface, several open source editing tools such as Jgraph [28], Cir- basically supports analysis commonly performed for utility distri-
cuitSimulator [29], JHotDraw [30], JDiagram [31] are studied. A bution systems including distributed generation analysis. An
variety of features related to customizable edition (component important property of OpenDSS is that it is very expandable and,
insertion, drag and drop, stencils, copy and paste, etc.) and license thus, can be easily modified [27]. OpenDSS is used in Distribution
are analyzed to select the most adaptive tool. Planning and Analysis, Analysis of Distributed Generation Inter-
As a result, Jgraph, which is developed in Java through the connections, Harmonics and Interharmonics Analysis and
Swing and AWT libraries, is the selected framework. In fact, there Neutral-to-earth Voltage Simulations, among others [38].
are articles where these libraries are proposed for similar develop- The connection between OpenDSS and the DMS can be seen in
ments [32]. It has a license BSD 3-Clause and provides functionality Fig. 2. OpenDSS counts with a COM interface to interchange data
for visualization and interaction through nodes, including features with other applications; this interface has a set of methods and
such as templates XML, which support the creation of new figures. properties which can be used through many programming lan-
This is useful for creating new electrical elements which corre- guages. There are several frameworks which can implement these
spond to different standards. Several visual features can be devel- connections. In this case, a variety of libraries for JAVA languages
oped with this framework, such as mechanisms of connection are tested, and JACOB Project is chosen due to the flexibility and
between components under the CIM model. reliability of its results. This library is an open-source project
Open source tools are assessed to implement the geographic which uses JNI (Java Native Interface) to make native calls into
representation or GIS (Geographic Information System). These the COM and Win32 libraries [39].
tools must allow electrical network visualization and edition, Communication between the proposed DMS and OpenDSS is
object tracking (electric crews) and alarm visualization, among implemented through the Builder and Adapter design patterns
others. The tools under evaluation are OpenLayers [33], GeoTools [40]. First, electrical network modeled into the CIM is accessed to
[34], OpenGeoSuite [35], Udig [33], gvSIG [33] and Open JUMP [33]. build the OpenDSS script. Then, the JACOB library is used to com-
The assessment considers the requirements mentioned above, pile the script and get the results of different algorithms provided
as well as quality attributes such as flexibility, adaptability, effi- by OpenDSS. Finally, the obtained results are adapted to an internal
ciency, usability, maintainability and portability. Therefore, format to be processed and displayed.
OpenLayers library is selected as the best option. This tool is devel- In Section 5, load flow analysis is performed for the IEEE nodes
oped in JavaScript and allows web or embedded Java development. test feeder [41] and compared with the available results to verify
Finally, to complete the GIS development, it is necessary to the correctness of the circuit modeled in the proposed DMS.
choose a free access map. Consequently, Open-StreetMap [36] is
selected. It provides free and editable maps of the entire world,
3.3. ActiveMQ-broker
maps which are built by volunteers with an open content license
(Fig. 1).
Communication between components which can be installed in
different devices, is provided by ActiveMQ software. ActiveMQ is
3.2. OpenDSS-framework an open source message broker with a full Java Message Service
(JMS) client. It supplies features such as communication with more
OpenDSS is an open source software package developed by EPRI than one client or server, security, computer clustering, virtual
[27,37], which can be used to simulate distribution networks. It memory, cache and the possibility of having access to any database
contains detailed models of several network components, for as a persistence provider. In addition, it employs message queues
example lines, loads, generators, monitors, etc. The program as a persistent communication mechanism among processing

Fig. 1. Schematic and geographic representation.

70 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76

Fig. 2. Communication between DMS and openDSS.

Fig. 3. Interaction of the applications with the activeMQ-Broker.

nodes. Message queuing facilitates data sharing and provides a every message the applications send to the topics. The PGP Data
publish/subscribe system which can be applied to the components Formatter has two keys, a public and a private one. The public
in order to communicate with other components [42]. Finally, Acti- key is necessary for the producer applications to encrypt the mes-
veMQ supplies a fail over protocol so as to maintain high availabil- sages. On the other hand, the private one allows the consumer
ity [43]. applications to decrypt the messages.
Functionality is based on topic management. Several topics are
created, and every module of the system is subscribed to them. The 3.4. CIM
topics are associated with different types of events generated in
the distribution networks. Each event contains relevant informa- CIM is developed by EPRI as a unique information model for
tion about some parts of the system. Hence, each part only needs electric power management systems to obtain interoperability
to be subscribed to the topics that contain the necessary informa- between internal modules in the system and export data to exter-
tion. Therefore, a modular system is implemented where each nal tools. This model is defined under the series of IEC 61970 and
module can work independently from the others. IEC 61968 norms corresponding to data transmission and distribu-
As shown in Fig. 3, the system is divided into different applica- tion, respectively. In addition, it is described by a UML (Unified
tions.Each application communicates with the others through Modeling Language) class diagram [47].
topics created in ActiveMQ-Broker. The applications not only con- As it was mentioned in [16] the power system network can be
sume the information of the topics, but also produce new informa- modeled using CIM at different levels of abstraction presenting
tion, which is sent to other topics. The applications that interact the respective level of detail. Thus, a modular division between sta-
with ActiveMQ-Broker are Alarm Processor, State Estimator, Data tic model and dynamic model is implemented to address the prob-
Acquisition System (DAS), Schematic and Geographic Representa- lem of real-time model exchange. The former enables the
tion and Network Connectivity Analysis (NCA). For example, the representation of the network topology and the connections
modules responsible for visualizing will be subscribed to the among them. The latter defines the state of the network measure-
new-measurements topic. When the Data Acquisition System pub- ments. Each application will use each model type when required.
lishes a new measurement on this topic, the ActiveMQ sends the Fig. 4 shows the main classes of these models. Consequently, the
new information, and the corresponding modules are informed in static model is the input of all algorithms presented into the
real time about the changes in the measurement values. DMS proposed and the dynamic model provides measurement val-
The authentication system supplied by ActiveMQ is used in the ues in real-time.
project [44]. Thus, a user name and password have to be employed The exchange of information in CIM is implemented through
for each application (producer and consumer) in order to log into RDF (Resource Description Framework) models [48] stored in a
the topic. Moreover, Apache Camel is used to encrypt all the mes- standard format, for example, RDF/XML (extensible Markup
sages [45]. The PGP Data Formatter [46] is implemented to encrypt Language).
 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76 71

Fig. 4. Components defined in the static and dynamic models.

Fig. 5. Updating process of the static model.

The type of the module determines the model that need to be larity, flexibility and independence between the supplier and the
updated. The static models are updated by the network editors platform.
with a low frequency. This must be reflected in all the modules The DMS architecture is divided into four main modules: DAS,
as a trigger engine (Fig. 5). applications, middleware services and persistence. DAS is the first
The dynamic model is updated by the broker and the DAS pub- communication point with several information sources such as
lishes measurement values with a high frequency. The modules measuring equipment and SCADA systems. All the received
subscribed to the corresponding topic updates the dynamic model information is published in the middleware service ActiveMQ,
automatically with the arrival of a measure (Fig. 6). which sends the captured information to the other parts of the
At present, CIM cannot model all the information that needs to system. This historical information is stored in a database to be
be used and exchanged by the electrical sector. Namely, in the pro- employed by several algorithms, as state estimation and demand
posed modular division, the network is formed by the interconnec- forecasting, among others. The process of storing the information
tion of physical elements, which can be stored in RDF format files. is executed by the persistence module, and the DAO pattern (Data
However, there are cases where these format files do not charac- Access Object) is used to implement it [49]. Thus, the design of the
terize the physical elements. For such cases, an extension of CIM database can be altered without modifying the code implementa-
is necessary in order to support some parameters, for example, tion. The applications module contains independent applications
resistance and reactance matrix of lines, reactance percent that provide the functionality of the DMS: Schematic and
between windings in transformers. Geographic representation, State estimator, NCA and Alarm
Processor.
4. Architecture The values on the screen are updated using a change notifica-
tion mechanism based on the MVC [40] pattern, which enables
An ADD methodology (Attribute Driven Design) is employed to the view and the controller components to be notified by the
build the DMS architecture 7. The main quality attributes required model. These components are registered in the model, providing
to build the architecture are usability, extensibility, flexibility, an immediate update. This feature allows an updated process that
interoperability and security. To meet these attributes, an Event do not violate the independence of the model.
Driven Architecture (EDA) which is based on a paradigm publish- The user applications could run on dedicated machines or exter-
subscribe was defined. EDA allows monitoring, detect and control nal servers. This facilitates modifiability and the use of components
different events in real time through several components. These that can be developed and updated independently. For example, to
components are uncoupled and can be modified isolatedly. This add the OMS (Outage Management System), only communication
property improves the flexibility and modifiability of the system with the middleware service has to be defined. This communica-
[26]. tion is performed by the topics and the OMS will be subscribed
EDA is complemented with the Service Oriented Architecture to an existing topic and will send the new information. Thus, these
(SOA), which allows every application to be freely connected to changes do not have an impact on the other applications and sys-
the platforms. In the same way, these features improve the modu- tem extensibility is maintained.
72 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76

Fig. 6. Updating process of the dynamic model.

Fig. 7. DMS architecture.

Fig. 8. Monitoring and controlling system components.

This architecture is instantiated and tested for many re-closers allows the system to retrieve and decode packages from
provided by a local distributor. In this case, DNP3 protocol and re-closers, and also to write registers into them. Moreover, it is
MODBUS are used to operate the equipments. In the first, differ- necessary to adapt the information provided from the devices
ent libraries to implement this protocol are analyzed. The to another format understandable by the DMS. Therefore, extra
selected library is OpenDNP3[50] because it presents more bene- functionality is added to the application layer to convert the val-
fits than the others. This implements all the DNP3 layers. It ues and commands to a JSON format as intermediary between the
 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76 73

Fig. 9. IEEE 4 nodes test feeder.

Fig. 10. Sequence diagram: power flow calculation.

devices and the DMS. Finish, libmodbus [51] is adopted to imple- selected by the user is translated to a set of tags which is sent
ment Modbus protocol. This library is a free software tool to through an ActiveMQ topic. Then, the DAS module subscripted to
send/receive data according to the Modbus protocol which is the same topic, receives that set of tags and converts it into a valid
written in C, and supports RTU (serial) and TCP (Ethernet) frame (depending on the protocol), resulting in a physical opera-
communication. tion into the device.
Under the proposed schema, every physical register of the Nowadays, the volt-var control algorithm is being developed.
equipment is identified by a unique code called Tag. The operation This algorithm is executed from Python with the OpenDSS library,
74 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76

Table 1 with the electrical parameter of each component). Next, the Script-
Refresh rate results. Builder module is responsible for building the OpenDSS script and
Visual interface making communication through the COM interface. Finally, the
# Measurements Max Min Average power flow is executed and the results are shown on the schematic
and geographic representation of the proposed DMS. As a result,
10 0.003 0.001 0.002
100 0.009 0.002 0.006
the values obtained for voltages, currents, and power flows are
1000 0.0018 0.015 0.0016 equal to the ones reported by OpenDSS. In addition, the performed
10000 0.0057 0.037 0.042 tests demonstrate that the times of script generation are insignifi-
cant compared to the communication times through the COM
interface.
and returns the optimal device configuration, which can be sent by
an ActiveMQ topic to change the turn ratio for LTC transformers, 5.2. Schematic representation update
control the reactive of the photo-voltaic system or change the state
of the capacitor bank 8. Firstly, the refresh rate is analyzed on the schematic network
representation module. This is essential for choosing the JGraph
library. In this test, different scenarios with 10, 100, 1000 and
5. Test cases
10000 tags are modeled. It must be mentioned that each scenario
is updated every 1 s (Table 1). After analyzing each case, it can
The new tools have to be robust and integrate seamlessly with
be concluded that the refresh rate increases linearly with the num-
other software systems. In this sense, several Test Feeders [41],
ber of tags on the screen. That is to say, if the number of tags
which are central to verify new distribution system software, have
increased by 10, the total time grows 2.6 ms (Fig. 11). This is the
been published. Test feeders will also be advancement drivers in
case of large scale distributors like Edenor [52] and Edesur [53].
distribution system analysis technology, being designed to stretch
incrementally the capabilities of the tools.
5.3. The performance of middleware services
Considering that this paper mostly refers to a new DMS, the
user’s interface will have to support new smart grid components
This section describes a performance test executed on Acti-
and behaviors. Middleware services and databases usually contain
veMQ broker. Several data sources are simulated while they are
software for transaction and persistence management, remote
concurrently publishing information of 10 tags. In all the cases
communication primitives and threading policies, which have a
500,000 tags are sent and the throughput is analyzed in each case.
great impact on the aspects mentioned above.
Fig. 12 shows that the throughput increases with the amount of
The following section shows three test cases linked to OpenDSS
data sources able to be processed, and an average of 30,000 tags
interface, ActiveMQ broker, database operations and visualization
per second are obtained. However, Table 2 shows that the percent
rate.
of errors grows with a higher amount of data sources. This happens
due to the overload of ActiveMQ buffers and the overhead of com-
5.1. OpenDSS interface munication between consumers and producers [54].

To test the OpenDSS interface, it is proposed to execute a power 5.4. The performance of databases
flow computation following the scheme explained in 3.2. For this
purpose, the IEEE 4 nodes test feeder is selected (Fig. 9). The Finally, a load test on the persistence module, which allows to
sequence diagram for this case can be visualized in Fig. 10. This evaluate the reading and writing latency, is performed. To measure
diagram shows the operation flow, starting from the creation of reading latency, the average time for executing a query is calcu-
the network on the schematic representation (network modeling lated. That query selects a total of 100, 10,000 and 40,000 measure-
ment values. The results can be seen in Table 3, where the response

·10-2
6
·104
Max 3

5 Min Max
Average 2.5

4
2
Tags per second
Time (ms)

3
1.5

2
1

1 0.5

0 0
0 0.1 1 0 100 1,000
Measurements ·104 Data sources
Fig. 11. Comparison of the refresh rate results. Fig. 12. ActiveMQ throughput.
 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76 75

Table 2
Results of the throughput test.

Middleware service (ActiveMQ)

# Data sources # Samples Average (ms) Min (ms) Max (ms) Error (%) Throughput (per sec)
1 500,000 2 1 245 0 2105
10 500,000 5 1 279 0 7367
100 500,000 20 1 470 0 17902
1000 500,000 50 1 1017 1.14 29757

1,600
Table 3
Results of the read latency test. Average
1,400
Database (Cassandra)
Min
1,200 Max
# Operations Average (ms) Min (ms) Max (ms) Error (%)
100 60.32 27.00 1196.00 0.00
1,000

Average (ms)
1000 51.75 34.00 1929.00 0.00
40000 528.98 593.00 4715.00 0.00
800

600
5,000
400
4,500 Average
Min 200
4,000
Max
3,500 0
0 1
Operations
Average (ms)

3,000
·104
2,500
Fig. 14. Comparison of the write latency results.
2,000

1,500 6. Discussion
1,000
The proposed DMS can edit and operate an electrical network
500 and perform several data analyses on a dedicated database. More-
0 over, the adopted architecture allows the incorporation of addi-
0 4 tional modules to bring solution to other required functionalities.
Operations Hence, this implementation is more suitable to be adapted to
·104 new requirements, in contrast with existing monolithic structures.
Fig. 13. Comparison of the read latency results.
Also, it can be integrated with other on-line systems to help deci-
sion makers during the system operation. In fact, the authors are
currently working on adding other features to the system such as
state estimation, volt/var control and fault management, all of
Table 4 them integrated in the implemented DMS. Furthermore, a new tool
Results of write latency test. for visualizing results in GIS environment is being developed.
It worth noting that security challenges were properly
Database (Cassandra)
addressed to avoid risk situations that could endanger the integrity
# Operations Average (ms) Min (ms) Max (ms) Error (%)
of the system’s data. As seen in Section 3, large amounts of infor-
100 44.30 25.00 843.00 0.00 mation are transmitted through the modules in order to be pro-
1000 28.74 20.00 775.00 0.00
cessed for each application. As a consequence, it was necessary
10000 24.30 17.00 2194.00 0.00
to implement tactics to avoid any external agent to publish infor-
mation into a broker and send it to the modules. We followed a
security approach based on session and encryption, which only
times oscillate between 60.32 and 528.98 ms. Fig. 13 shows that allows to authorized modules and external systems to publish
the response time increases when the number of consulted data information.
grows. Finally, it is worth mentioning that this tool will be used by
The average insertion time of a measurement value is analyzed three distribution companies in Argentina in the short term.
to evaluate the writing latency of the database. For this reason, sev-
eral numbers of writing operations (100, 1000, 10,000) are evalu-
ated. Table 4 shows the results of each test. Fig. 14 depicts that 7. Conclusion
the writing time decreases with the increase of writing operations.
This happens because Cassandra maintains in memory the result of Nowadays, there are several challenges related to distribution
the last operations. Therefore, it is not necessary to bring the infor- networks. The reliably operation of an electrical network is an
mation from the hard disk every time a query is executed. important concern for the power system operators. The applica-
76 L. Fiaschetti et al. / Electrical Power and Energy Systems 94 (2018) 67–76

tions with on-line monitoring must be able to recognize the dan- [17] Moffet M-A, Sirois F, Beauvais D. Review of open-source code power grid
simulation tools for long-term parametic simulation. CanmetENERGY, Tech.
gerous situations that can be leading to network collapse. Thus,
Rep. 137.
it is necessary to have a system that supports decision-making [18] E.P.R. Institute, EPRI, 2017. URL http://www.epri.com/.
which allows the analysis of the state of the power network. [19] Tsampasis E, Sarakis L, Leligou HC, Zahariadis T, Garofalakis J. Novel simulation
Moreover, the integration of renewable energy is taking an approaches for smart grids. J Sens Actuat Netw 2016;5(3):11.
[20] Kumar Y, Ringenberg J, Depuru SS, Devabhaktuni VK, Lee JW, Nikolaidis E,
important role in industry. Thus, it is necessary to have a system Andersen B, Afjeh A. Wind energy: trends and enabling technologies. Renew
that supports decision-making which allows the analysis the state Sustain Energy Rev 2016;53:209–24.
of the power network. This paper presents a DMS development [21] Kouzapas D, Yoshida N, Hu R, Honda K. On asynchronous eventful session
semantics. Math Struct Comput Sci 2016;26(02):303–64.
based on open source tools. Under this scheme, several subsystems [22] Ionescu VM. The analysis of the performance of RabbitMQ and ActiveMQ. In:
can be easily added. In this way, it is possible to adapt the system 2015 14th RoEduNet International Conference-Networking in Education and
to the new challenges that involve renewable energies. This new Research (RoEduNet NER). IEEE; 2015. p. 132–7.
[23] Rabl T, Gómez-Villamor S, Sadoghi M, Muntés-Mulero V, Jacobsen H-A,
paradigm allows the incorporation of tools that help operation. Mankovskii S. Solving big data challenges for enterprise application
It worth mentioning that important characteristics such as performance management. Proc VLDB Endowment 2012;5(12):1724–35.
interoperability, event orientation and flexibility were taken into [24] Moniruzzaman A, Hossain SA. Nosql database: new era of databases for big
data analytics-classification, characteristics and comparison, arXiv preprint
account during the development, and were implemented making arXiv:1307.0191.
use of state of the art architectural patterns such as SOA, EDA, pub- [25] Cassandra A. The apache software foundation. The Apache Cassandra project.
lish/subscribe and broker. In addition, the tool was integrated with [26] Pradeep Y, Seshuraju P, Khaparde S, Joshi RK. Flexible open architecture design
for power system control centers. Int J Electr Power Energy Syst 2011;33
NoSql storage in order to obtain data to analyze several aspects of
(4):976–82.
the power network. The design combines: OpenDSS, which pro- [27] Dugan RC. Reference guide: the open distribution system simulator
vides several electrical algorithms; Jgraph, as a graphical library; (OpenDSS). Electric Power Research Institute, Inc.
ActiveMQ, as a broker; Cassandra, as a storage mechanism; and [28] Alder G. Design and implementation of the JGraph swing component. Techical
Report 1 (6).
CIM as an intermediate model. Using this existing open source [29] Falstad P. Circuit simulator v1.2, 2013. URL http://www.falstad.com/circuit/.
software, it was possible to build a platform which provides func- [30] Garzón MA, Aljamaan H, Lethbridge TC. Umple: a framework for model driven
tions that allow enhanced real-time operation and simulation of a development of object-oriented systems. In: Software Analysis, Evolution and
Reengineering (SANER), 2015 IEEE 22nd International Conference on. IEEE;
changing electric distribution system. 2015. p. 494–8.
Finally, more Smart Grid/DMS applications can be implemented [31] MindFusion, JDiagram (2016). URL http://www.mindfusion.eu/onlinehelp/
over the framework to facilitate development tasks. jdiagram/.
[32] Zhou M, Zhou S. Internet, open-source and power system simulation. In:
Power Engineering Society General Meeting, 2007. IEEE. IEEE; 2007. p. 1–5.
Acknowledgments [33] Steiniger S, Hunter AJ. Free and open source GIS software for building a spatial
data infrastructure. Geospatial free and open source software in the 21st
century 2012:247–61.
The authors would like to thank the Ministry of Science, Tech- [34] p. OSGeo, The open source java gis toolkit (2016). URL http://www.geotools.
nology and Productive Innovation of Argentina, and the Ministry org/.
of Defense of Argentina for their financial support. [35] Boundless. OpenGeoSuite; 2016. URL http://suite.opengeo.org/.
[36] Barron C, Neis P, Zipf A. A comprehensive framework for intrinsic
OpenStreetMap quality analysis. Trans GIS 2014;18(6):877–95.
References [37] Dugan R, Arrit R, Henry R, McDermott T, Sunderm W. Opendss EPRI
distribution system simulator-harmonic load modeling documentation.
[1] Taylor T, Ohrn M. Network management for smart grids: innovative operations retrieved december, 2014; 2014.
centers to manage future distribution networks. Revista ABB 2009;3:45–9. [38] Dugan RC, McDermott TE. An open source platform for collaborating on smart
[2] Roytelman I, Landenberger V. Real-time distribution system analysis-integral grid research. In: 2011 IEEE Power Energy Soc General Meet. IEEE; 2011. p.
part of DMS. In: Power Systems Conference and Exposition, 2009. PSCE’09. 1–7.
IEEE/PES. IEEE; 2009. p. 1–6. [39] Adler D. The jacob project: A java-com bridge; 2005.
[3] Rezkalla M, Heussen K, Marinelli M, Hu J, Bindner HW. Identification of [40] Gamma E. Design patterns: elements of reusable object-oriented
requirements for distribution management systems in the smart grid context. software. Pearson Education India; 1995.
In: Power Engineering Conference (UPEC), 2015 50th International [41] Feeders DT. IEEE power and energy society (2014). URL https://ewh.ieee.
Universities. IEEE; 2015. p. 1–6. org/soc/pes/dsacom/testfeeders/.
[4] Bakken D. Smart grids: clouds, communications, open source, and [42] Snyder B, Bosanac D, Davies R. Introduction to apache ActiveMQ. Active MQ in
automation. CRC Press; 2014. Action 6–16.
[5] Meliopoulos AS, Polymeneas E, Tan Z, Huang R, Zhao D. Advanced distribution [43] A.S. Foundation, Apache ActiveMQ availability (2011). URL http://activemq.
management system. IEEE Trans Smart Grid 2013;4(4):2109–17. apache.org/failover-transport-reference.html.
[6] Vaccaro A, Loia V, Formato G, Wall P, Terzija V. A self-organizing architecture [44] A.S. Foundation, Apache ActiveMQ security (2011). URL http://activemq.
for decentralized smart microgrids synchronization, control, and monitoring. apache.org/security.html.
IEEE Trans Ind Inform 2015;11(1):289–98. [45] Rademakers T. Activiti in action: executable business processes in BPMN
[7] Terzija V, Valverde G, Cai D, Regulski P, Madani V, Fitch J, Skok S, Begovic MM, 2.0. Manning Publications Co.; 2012.
Phadke A. Wide-area monitoring, protection, and control of future electric [46] A.S. Foundation, Apache camel crypto; 2011. URL http://camel.apache.org/
power networks. Proc IEEE 2011;99(1):80–93. crypto.html.
[8] Gungor VC, Sahin D, Kocak T, Ergut S, Buccella C, Cecati C, et al. A survey on [47] The Common Information Model CIM: IEC 61968/61970 and 62325-A practical
smart grid potential applications and communication requirements. IEEE Trans introduction to the CIM, author=Uslar, Mathias and Specht, Michael and
Ind Inform 2013;9(1):28–42. Rohjans, Sebastian and Trefke, Jörn and González, José M, year=2012,
[9] D. PowerFactory, Version 14.0. DIgSILENT International, Germany. URL http:// publisher=Springer Science & Business Media.
www.digsilent.de/. [48] Cyganiak R, Wood D, Lanthaler M. Rdf 1.1 concepts and abstract syntax. W3C
[10] E.P.S.A.. O. Software, ETAP; 2017. URL https://etap.com/. Recommendation 2014;25:1–8.
[11] S.E.D. NS, Schneider; 2014. URL http://www.schneider-electric-dms.com/. [49] Song J, Sierra SC, Rodríguez JC, Perandones JM, Jiménez GDC, Buján JO, García
[12] Tanskanen A, Raussi T, Partanen J, Lohjala J. Cost and benefit analysis for a RM, Galdón AS, et al. Data consistency management in an open smart home
distribution management system in electricity distribution networks. Int J management platform. In: Modelling Symposium (EMS), 2014 European. IEEE;
Energy Sect Manage 2010;4(2):256–72. 2014. p. 366–71.
[13] Katic N. Benefits of smart grid solutions in open electricity market. Acta [50] Curtis K. A DNP3 protocol primer. DNP User Group; 2005.
Polytechnica Hungarica 2013;10(2):49–68. [51] Raimbault S. A modbus library for linux, mac os x, FreeBSD, QNX and win32;
[14] Katic N, Marijanovic V, Stefani I. Smart grid solutions in distribution networks 2016. URL http://libmodbus.org/.
cost/benefit analysis. In: CICED 2010 Proc. IEEE; 2010. p. 1–6. [52] E.D. y Comercializadora Norte Sociedad Annima, EDENOR; 2008. URL http://
[15] Simmins JJ. The impact of PAP 8 on the common information model (CIM). In: www.edenor.com.ar/cms/SP/EMP/ACE/COM.html.
Power Systems Conference and Exposition (PSCE), 2011 IEEE/PES. IEEE; 2011. [53] E.D.S.S. Annima, EDESUR; 2016. URL http://www.edesur.com.
p. 1–2. ar/masedesur/conoceedesur.aspx.
[16] Pradeep Y, Seshuraju P, Khaparde SA, Joshi RK. CIM-based connectivity model [54] A.S. Foundation, Apache ActiveMQ; 2011. URL http://activemq.apache.org/
for bus-branch topology extraction and exchange. IEEE Trans Smart Grid what-is-the-prefetch-limit-for.html.
2011;2(2):244–53.