Conference PaperPDF Available

Flexible Fog Computing Architecture for Smart Microgrids


Abstract and Figures

With an increase in the adoption of renewable energy sources and storage, microgrids and smart local energy systems have made considerable headway in recent years. The unreliable nature of renewable resources and the high costs of storage infrastructure, however, pose challenges to designers. Research in distributed control, multi-agent systems and advanced analytics can aid in improving the efficiency and reliability of microgrids. However, the integration of these services and technologies into a single system is difficult due to their isolated nature and varying preference in protocols and platforms. This paper proposes a flexible fog computing-based, distributed deployment and virtualisation architecture that solves some of the integration challenges while offering increased flexibility and scalability. This architecture is implemented and deployed on an existing UKRI-funded microgrid demonstrator and evaluated on its ability to integrate the control, energy pricing, and analytics elements as well as on the extended features it offers to the microgrid.
Content may be subject to copyright.
Flexible Fog Computing Architecture for Smart Microgrids
Nandor Verba
*, Elena Gaura
Centre for Data Science
Coventry University, Coventry, United Kingdom
Pablo R. Baldivieso-Monasterios, George Konstantopoulos
Department of Automatic Control & Systems Engineering
University of Sheffield, Sheffield, United Kingdom
Euan A. Morris, Stephen D.J. McArthur
Department of Electronic and Electrical Engineering
University of Strathclyde, Glasgow, United Kingdom
With an increase in the adoption of renewable energy sources and storage, microgrids and smart
local energy systems have made considerable headway in recent years. The unreliable nature
of renewable resources and the high costs of storage infrastructure, however, pose challenges
to designers. Research in distributed control, multi-agent systems and advanced analytics can
aid in improving the efficiency and reliability of microgrids. However, the integration of these
services and technologies into a single system is difficult due to their isolated nature and varying
preference in protocols and platforms. This paper proposes a flexible fog computing-based,
distributed deployment and virtualisation architecture that solves some of the integration
challenges while offering increased flexibility and scalability. This architecture is implemented
and deployed on an existing UKRI-funded microgrid demonstrator and evaluated on its ability
to integrate the control, energy pricing, and analytics elements as well as on the extended
features it offers to the microgrid.
Smart local energy systems, fog computing, integration, flexibility, scalability, distributed
control, multi-agent systems, data to knowledge chains
With the increased attention and funding for renewable energy in recent years, various energy
storage and generation technologies have reached higher technical and digital maturity and cost-
effectiveness [1]. The concepts of ‘Industry 4.0’ that define the technological trends and
requirements of the fourth industrial revolution, can be used as a roadmap for the energy
industry. This roadmap leads to the formulation of the ‘Energy 4.0’ initiative [2] [3]. A key
enabling concept is that of a smart grid - which intends to bring flexibility, integrability and
scalability to the energy system. Such “smart” properties can be realized through the application
of innovations in fundamental disciplines such as: distributed control; multi-agent systems
* Corresponding author
(MAS) and data science. The first focuses on guaranteeing reliability and robustness while
MAS provides flexibility by endowing each component with intelligence in its decision making.
The data science/analytics components provide Forecasting and Systems state information that
is crucial in decision making. The existing smart grid infrastructure comprises several
interconnected distributed energy resources (DER), and different types of loads, industrial or
residential. Distinctive features of energy networks include the fluctuating nature of each DER,
generation capabilities and the heterogeneity of its components [4]. For local smart grids to
benefit from the advances in distributed control, multi-agent systems and big-data analysis, and
thus increase their efficiency and their benefit to users, they need to integrate the relevant
progress from these fields into a cohesive framework that enables seamless communication
between its elements and users while also allowing new features and services to be integrated.
Centralised deployment of analytics, control and negotiation services is a common practice;
this solution, however, presents drawbacks in terms of scalability, computational burden, or
requires substantial changes within an existing system. From a data standpoint, the paradigm
of ‘Fog Computing’, as suggested by Cisco in [5], allows for non-centralised solutions which
mitigate some of these drawbacks. From a control perspective, distributed solutions are
favoured over centralised ones; the nature of the smart grid itself lends to such distributed
solutions [6]. In the case of smart grids, power inverters are the interface between a DER and
the electricity network. These power inverters can be used to incorporate different
functionalities to the smart grid allowing different DERs to be dynamically connected as parts
of the Industry 4.0 revolution [7]. Smart grids pose significant challenges for control in terms
of both the potentially high number of DERs involved and the potentially spread-out
geographical locations of each DER. The issues from high dimensionality are related to the
computation capacity, while a sparse distribution of DER locations may put a strain on the
communication channels [8].
Intelligent agents can also provide non-centralised solutions through multi-agent systems. MAS
solutions are also distributed in nature and are a collection of intelligent agents which
communicate and interact in the same environment. The effectiveness of MAS approaches has
been demonstrated with many functions pertinent to smart microgrids. Some capabilities which
have been demonstrated can be seen in [9] where brokerage of energy markets is demonstrated
to control the dispatch of energy resources, and in [10] where MAS facilitation of peer to peer
energy trading is investigated. Flexible dispatch of energy resources using MAS is also
demonstrated in [11], where distributed control of an integrated energy system including
renewable generation, electric vehicle demand, and energy storage is investigated. In addition,
estimation of network parameters and constraints is explored in [12] where the generation
parameters of solar PV are estimated using MAS, with a predictive model created which can
be used to inform control decisions. Other areas of interest where MAS solutions have been
demonstrated within a microgrid environment include active network management
applications, including for system restoration [13], and demand response management[14].
The intersection of new distributed control techniques and distributed computation has been
proven to be fertile ground for new research. Recent applications [15] have implemented
predictive controllers using distributed computation techniques, while the authors in [16] have
explored notions of clustering in micro-grids. These may open the door to further developments
in energy management using MAS and predictive control, such as in [17, 18], where the
approach focuses on battery optimisation and leverages implicitly on the existence of a
communication network. The nature of the smart grid lends itself to a distributed setting both
in the computation as well as in the control sense, for which a robust communication platform
capable of handling large amounts of data is needed [19].
Transforming the data generated by local smart grids into knowledge and information that can
be used by digital and human actors in energy systems has been an ongoing challenge. Beyond
the issue of defining what variables to measure and how often to transmit them, with the
increased digitalisation of these systems, an emerging problem is how to store and analyse this
data while providing real-time responses and not diminishing the environmental benefits of
these green energy systems by storing large quantities of data [20]. Adapting traditional extract,
transform, load (ETL) [21] approaches that are standard in data analysis, to the needs of real-
time energy systems requires a flexible processing architecture.
The various research fields that interact to add value to smart microgrids have different
requirements, protocols and associated platforms which create a highly heterogeneous
environment that slows down development and reduces the benefit of such energy systems.
This work examines how a messaging and data broker-based architecture can answer the energy
systems specific requirements of Distributed Control, Multi-Agent Systems and Big-Data
Analysis while providing the required flexibility, integrability and scalability.
This paper first provides an overview of the benefits, requirements and protocols that drive the
research in MAS, Distributed Control and data analytics scoped to smart microgrids. This set
of requirements is expanded with features that stem from the need to make local energy systems
more scalable, flexible and interoperable. Architecture and communication models from the
Internet of Things (IoT), Cloud and Fog Computing are combined to formulate the proposed
architecture. This architecture is implemented and deployed on a smart microgrid installation
following a use-case specification that is designed to test the capabilities and features of the
architecture. The architecture is evaluated based on the criteria presented in Table 1 and extra
functionality is highlighted in the review section.
Table 1. Requirements for each domain
Detailed Requirement
Evaluation Criteria
Real-time/ Live
communication through
industry-standard protocols.
Platforms need to be able
to broker live messages to
and from these systems
Some Controllers need to
run on specialist hardware.
Any required information
needs to be transmitted to
and from these locations.
Retention of functionality in
case of network failures and
exceptional events.
Self-contained units need
to be able to house all their
required services.
Negotiation through tailored
protocols, standards and
The agent ontologies need
to be translated for legacy
systems and elements
from other domains.
Agents are deployed in a
tailored environment with
proprietary messaging.
The system needs to be
able to virtualise this
environment and recreate
it on site.
Virtualise networking,
storage, processing and
Ability to deploy services
and devices.
Polyglot nature
Be able to “speak” multiple
protocols and languages to
enable interfacing with these
Provide support for
synchronous and
asynchronous messaging.
Big Data
Provide the required
management, storage and
processing capability.
Provides interfaces to
platforms that are the state
of the art for this field.
Data Stream
Support for high throughput
message passing, processing
and decision making.
Able to translate messages
to stream and attach
workers to these streams.
ADEPT use-case
The use-case was selected due to its high technology readiness level (TRL) implementation of
a smart microgrid and due to it containing a variety of storage and generation components. It is
a good candidate for the architecture as this system is designed to be vertically and horizontally
scaled by extra units and by changing the layout of these units.
Figure 1. ADEPT Smart Energy System
The main aim of the Advanced multi-Energy management and oPTimisation time-shifting
platform (ADEPT) [22] is to create an intelligent microgrid system based on several distributed
energy resources (DERs) that combine a variety of smart features. The ADEPT concept is based
on the combination of two novel technologies: i) a novel smart inverter design that seamlessly
integrates different DERs and supports the grid in providing virtual inertia with current limiting
capabilities under faulty or abnormal grid conditions and ii) an intelligent monitoring and
control platform to manage the power flow between the DERs based on a model predictive
control (MPC) framework. The image in Fig. 1. shows the deployed on-site system, with the
industrial consumer in the background.
The extension aims to add a layer of virtualisation on top of the ADEPT system that would
allow more complex distributed control, modelling, and digital twin elements to be deployed
as well as facilitate offline processing and live monitoring. It requires the existing MPC
Controllers to be extended so they can receive price set points from agents ‘wrappers’ and
tailored forecasting information from the analytics platforms. The agents' wrappers require the
current set-points set by the controllers as well as market data and forecasting data to be able to
agree on set prices for peer energy sharing and purchasing. The analytics platform monitors all
these systems to be able to quantify how well they function and to be able to forecast
consumptions and estimate system states.
Flexible and Scalable Deployment Architecture
The proposed solution is based on a messaging-based flexible fog architecture that allows
dynamic message passing between peer systems and the migration of services from various
nodes while retaining their functionality. This framework also allows various protocols and
technologies to be used internally that are application tailored. Through the virtualisation of
Networking, Storage and Processing elements this Fog Computing architecture allows
resources to be deployed at various points of the system while maintaining their integrity and
functionality. This is done by containerising the applications and through the advanced routing
present in the messaging system. The concepts that spring from the Fog of Things paradigm
[23] suggest virtualising smart devices connected to the system. It’s this concept that allows us
to plug-in virtual devices to real systems and segregate parts of the network.
The federated messaging framework can be seen in Fig 2, where it shows how a single system
can be connected to a set of its visible peers in a local cluster while retaining connectivity
through cluster leads with neighbouring clusters. This configuration allows for fractal-like
scaling of the architecture where the clustering concept can be extended to local regions of
clusters and even larger entities. This allows the systems to retain some of the advantages of a
hierarchical system while benefiting from the advantages of peer-to-peer communication.
Figure 2. Federated Messaging Framework
The AMQP/MQTT messaging broker is at the heart of the system, as it allows various
components to be interconnected without the need to reconfigure them. The system can be
broken down into 3 distinct areas. The physical area contains all of the devices connected to the
system directly or through the brokers. Here, the brokers may translate CAN, Modbus or other
protocol messages to AMQP and back for control or for sensing needs. Depending on the type
of protocol used, the brokers may be configured to read certain address values at regular
intervals and transmit them to the processing elements. The brokers also make the connection
between Legacy systems and the proposed digital architecture. They can act in two different
ways. In the first case, they are deployed alongside components where they broker messages
from that components platform to the wider system. The second method has these brokers
function as wrappers where they control all the communication from and to a service acting as
a virtualisation layer on top of them. This is best used for physical devices and their digital
Figure 3. Extension Data Flow Diagram
The data flow diagram from Fig. 3. represents what parameters are passed between each group
of components and what are the abstraction elements or brokers on top of them. The red
containers represent physical entities or services, while the blue containers are either wrappers
around physical entities or brokers between services and the messaging system. The dotted
tables then represent what data is passed and of what type: 24h is aggregated data for the day;
24h-F is forecasted data for the next 24 hours. The other elements are considered real-time or
the latest available information. The aggregated and forecasted elements come from the local
or centralised data-store and analytics components while the real-time data is captured from the
messaging system.
To evaluate the proposed architecture, a deployment use-case is proposed that verifies the
requirements from Table 1. This use-case is then implemented and evaluated on how it satisfies
the needs of the various systems and also on what extra functionality these add to the digital
Evaluation Use-Case
The high-level view evaluation use-case can be seen in Fig. 4. and is defined in such a way that
it makes use of all the added functionality and showcases how various components can be
deployed both centrally and locally while retaining their functionality.
Figure 4. Extension Deployment Use-Case
This use-case is designed to connect the existing testing and development sites from Sheffield,
Glasgow and Coventry. Each site has its own set of functionalities with some of them being
ported to the other two sides and all the communication routed through the federated messaging
system. The Sheffield system is the main connection point to the physical ADEPT unit. It also
houses the OpalRT real-time controllers and a Docker container with forecasting and agent
components. Glasgow has a SPADE [24] based evaluation platform for agent-based energy
negotiation that is extended with a set of virtual devices and forecasting components. Coventry
houses most of the offline analytics and monitoring components together with a domestic load
monitoring system and a set of virtual devices.
The purpose of the multi-agent layer in the architecture is to increase both the functionality and
flexibility of the distributed MPC controller. An overview of this layer is shown in Fig. 5, where
MPCs are ‘wrapped’ within agents. These agents control the flow of information between the
MPCs, their connected devices, and a market agent. The market agent sets the current cost of
purchasing electricity from the wider grid. The cost of purchasing electricity from the grid is
used to update the cost function used by the MPC to make control decisions. Under normal
circumstances, the MPCs inform each other of control actions they intend to make, but with the
addition of the agent wrapper, this mere informing of control actions can be replaced with the
negotiation of control actions between MPCs.
Figure 5: Multi-Agent System Layer of Cloud Computing Architecture.
Deployment Evaluation
The communication requirements of MAS, the MPC Controllers and of big data analysis are
satisfied through brokers to and from the AMQP messaging system. Tailored brokers were
implemented to: pass messages from the agents' framework, through their XMPP messaging
system, to AMQP; pass messages from the real-time and MATLAB based controllers using
UDP ports; to store logging and monitoring data in the data stores and to retrieve the requested
information for the agents and MPC controllers.
The Locality requirements of the distributed controllers and MAS are partially answered by the
brokers that allow messages to be passed to the right sources and partially through the
virtualisation of device messages that allow them to be transmitted to the right control
environment. The use of Docker-based environments to house these components also adds a
layer of flexibility that enables locality-based deployments.
The robustness criteria are satisfied through the implementation of the resource virtualisation
element by allowing services, data and connections to be replicated and migrated to the locality
of the microgrid units thus being more resilient to connection issues with centralised services.
The polyglot interfaces requirements for connectivity to other systems can be partially solved
through brokers and common API standards. The messaging architecture also allows messages
to be transmitted to and from multiple peers making extensions and integration more seamless.
This messaging architecture also feeds into the data stream analysis component that can display
and process data and send on-demand responses as it is coming in through the system.
Added Functionality
Migration allows services and components to be moved from central repositories to the edge of
the network where the microgrid is deployed. This functionality allows extra microgrid side
processing capabilities to be better utilised reducing the demand on the central service and
allowing users to optimise their deployments to improve reliability and latencies.
Self-containment is the property of the microgrid where it can house all of the digital services
it needs to function. This can significantly increase its resilience to network failures and through
design allows more peer to peer connectivity and interoperability by needing to serve
connection interfaces.
Software in the Loop (SIL)/ Hardware in the Loop (HIL) testing allows components of the
microgrid or several microgrids to be digitally isolated from the rest of the systems and tested
through various use-cases. This is made possible due to the messaging architecture that can be
used to re-configure communication channels.
Security Isolation of internal protocols and services behind messaging services adds a layer of
protection through obscurity to these smart grids. Using these types of Secure Sockets Layer
(SSL) enabled protocols for external communication also increases the security robustness of
the microgrid. Furthermore, as only cluster heads are required to have open AMQP ports, these
further increase the security of peer systems.
This paper proposes a flexible fog computing architecture designed for smart microgrids that
allows them to integrate advanced control, price negotiation and analytics tools while increasing
their capability to scale and connect to peer systems and services. The requirements of the
architecture were formulated based on current trends in the three domains and evaluated based
on a scaling and hybrid deployment use-case. The implementation and deployment showed that
the barriers of the three areas can be overcome through brokers and a reconfigurable messaging
architecture and that it can also extend the capabilities of the existing system.
Future work can be divided into three lines of work. The first is to evaluate the architecture to
quantify how it influences the latencies and reliability of the system with the intent to formulate
a system topology model. This model can then be used for topology optimisation and
evaluation. The second line of work is to develop tailored MPC controllers that can make use
of the system's new functionalities. The third line of work will consist of designing negotiation
agents for price-setting that can improve the cost-effectiveness of these systems.
This work was supported by the UK Research and Innovation (UKRI) through the Engineering
and Physical Sciences Research Council (EPSRC) as part of the Energy Revolution Research
Consortium (ERRC) with the reference EP/S031863/1. The ADEPT demonstrator was
developed by Infinite Renewables in collaboration with GS Yuasa Batteries, Sheffield
University and Swanbarton with the support of an Innovate UK grant with the reference
1. McCrone, A., Moslener, U., d’Estais, F., Usher, E. and Grüning, C., Global trends in
renewable energy investment, Bloomberg, 2017.
2. Lasi, H., Fettke, P., Kemper, H. G., Feld, T., & Hoffmann, M. Industry 4.0. Business &
information systems engineering 6, No. 4, pp 239-242, Aug. 2014.
3. Lang, M., From industry 4.0 to energy 4.0. future business, models and legal relations.
Digitalisierung in der Energiewirtschaft XX, Jahrestagung Institut für Berg-und Energierecht,
4. Ferraro, P., Crisostomi, E., Raugi, M., & Milano, F. (2017). Analysis of the Impact of
Microgrid Penetration on Power System Dynamics. IEEE Trans. Power Syst., No. 32, 2016.
5. Bonomi F, Milito R, Zhu J, Addepalli S. Fog computing and its role in the internet of things.
Proceedings of the first edition of the MCC workshop on Mobile cloud computing, pp 13-16,
17 Aug 2012.
6. Guerrero, J. M., Chandorkar, M., Lee, T., & Loh, P. C. (2013). Advanced Control
Architectures for Intelligent Microgrids; Part I: Decentralized and Hierarchical Control. Ind.
Electron. IEEE Trans., No. 60, pp 1254-1262. 2012.
7. Xue, Y., & Guerrero, J. M. (2015). Smart inverters for utility and industry applications. PCIM
Eur. 2015; Int. Exhib. Conf. Power Electron. Intell. Motion, Renew. Energy Energy Manag.
Proc., pp 1-9, 2015.
8. Guerrero, J. M., Loh, P. C., Lee, T. L., & Chandorkar, M., Advanced control architectures
for intelligent microgridsPart II: Power quality, energy storage, and AC/DC microgrids. IEEE
Trans. Ind. Electron., No. 60, pp 1263-1270, 2012.
9. A. Y. Rodríguez González et al., A competitive and profitable multi-agent autonomous
broker for energy markets, Sustain. Cities Soc., No. 49, August 2018.
10. Satheesh Kumar GS, Nagarajan C. Design and implementation of intelligent energy control
for tariff management using multi‐agent system in smart/micro grid. Concurrency and
Computation: Practice and Experience. Vol. 31, No. 12, 25 June 2019.
11. Zeng C, Jiang Y, Liu Y, Tan Z, He Z, Wu S. Optimal Dispatch of Integrated Energy System
Considering Energy Hub Technology and Multi-Agent Interest Balance. Energies. Vol. 12, No.
16, pp 3112, Jan 2019.
12. C.-M. Huang, Y.-C. Huang, S.-P. Yang, K.-Y. Huang, and S.-J. Chen, Parameter Estimation
and Power Prediction for PV Power Generation Using a Multi-agent Algorithm, 2019 IEEE Int.
Conf. Ind. Technol., pp 679684, 2019.
13. Dong Z, Lin L, Guan L, Chen H, Liang Q. A service restoration strategy for active
distribution network based on multiagent system. 2018 IEEE Innovative Smart Grid
Technologies-Asia (ISGT Asia), pp 384-389, 22 May 2018.
14. S. Davarzani, R. Granell, G. A. Taylor, and I. Pisica, Implementation of a novel multi-agent
system for demand response management in low-voltage distribution networks, Appl. Energy,
Vol. 253, pp 113516, June 2018.
15. Skarin, P., Eker, J., Kihl, M., & Arzen, K.-E., Cloud-Assisted Model Predictive Control.
2019 IEEE International Conference on Edge Computing (EDGE), pp 110112, 2019.
16. Bullich-Massagué, E., Díaz-González, F., Aragüés-Peñalba, M., Girbau-Llistuella, F.,
Olivella-Rosell, P., & Sumper, A., Microgrid clustering architectures. Applied Energy, No. 212,
pp 340361, 2018.
17. Parisio, A., Rikos, E., & Glielmo, L., A model predictive control approach to microgrid
operation optimization. IEEE Transactions on Control Systems Technology, Vol. 22. No. 5, pp
18131827, 2014.
18. Kou, P., Liang, D., & Gao, L., Distributed EMPC of multiple microgrids for coordinated
stochastic energy management. Applied Energy, No. 185, pp 939952, 2017.
19. Serban, I., Cespedes, S., Marinescu, C., Azurdia-Meza, C. A., Gomez, J. S., & Hueichapan,
D. S., Communication requirements in microgrids: A practical survey. IEEE Access, No. 8, pp
4769447712, 2020.
20. T. Arthi and H. Shahul Hamead, Energy aware cloud service provisioning approach for
green computing environment, 2013 International Conference on Energy Efficient
Technologies for Sustainability, Nagercoil, pp 139-144, 2013.
21. Vassiliadis P., A survey of extracttransformload technology, International Journal of
Data Warehousing and Mining (IJDWM)., Vol. 5. No. 3, pp 1-27, 1 June 2009.
22. ADvanced multi-Energy management and oPTimisation time shifting platform (ADEPT),
UK Research and Innovation, [Accessed: 24-Feb-
23. Verba N., Chao K. M., James A., Goldsmith D., Fei X., Stan S. D., Platform as a service
gateway for the Fog of Things. Advanced Engineering Informatics, Vol. 33, pp 243-57, 1 Aug.
24. Gregori ME, Cámara JP, Bada GA. A jabber-based multi-agent system platform,
Proceedings of the fifth international joint conference on Autonomous agents and multiagent
systems, pp 1282-1284, 8 May 2006.
... Our study considers scenarios that cover a wide range of DER configurations and renewable inputs. For the wind turbine data, the scenarios were generated based on the data collected in [32] from real turbines in South Wales,UK. The data for PV panels is generated following a Clear-sky models and a burr distribution noise for Sheffield,UK. ...
... Our study considers scenarios that cover a wide range of DER configurations and renewable inputs. For the wind turbine data, the scenarios were generated based on the data collected in [36] from real turbines in South Wales,UK. The data for PV panels is generated following a Clear-sky models and a burr distribution noise for Sheffield,UK. ...
Full-text available
Continuous integration of renewable energy sources into power networks is causing a paradigm shift in energy generation and distribution with regards to trading and control; the intermittent nature of renewable sources affects pricing of energy sold or purchased; the networks are subject to operational constraints, voltage limits at each node, rated capacities for the power electronic devices, current bounds for distribution lines. These economic and technical constraints coupled with intermittent renewable injection may pose a threat to system stability and performance. We propose a novel holistic approach to energy trading composed of a distributed predictive control framework to handle physical interactions, i.e., voltage constraints and power dispatch, together with a negotiation framework to determine pricing policies for energy transactions. We study the effect of forecasting generation and consumption on the overall network's performance and market behaviours. We provide a rigorous convergence analysis for both the negotiation framework and the distributed control. Lastly, we assess the impact of forecasting in the proposed system with the aid of testing scenarios.
Full-text available
Progress in Microgrid (MG) research has evolved the MG concept from classical, purely MG power networks to more advanced power and communications networks. The communications infrastructure helps control and manage the unreliable power outputs that most standard power generation elements of the MG (e.g., wind turbines and photo-voltaic panels) deliver. Although communication technologies do offer certain advantages for sensing and control, they generate other complications due to packet loss and packet latency, among other transmission impairments. In this work, we discuss the impact of communications on MG performance, establishing the requirements of data exchanges and system response in the three levels of a hierarchical control approach: primary, secondary, and tertiary. With a focus on the secondary level — responsible for ensuring the restoration of electrical parameters — we identify standards, networking protocols, and communication technologies relevant for the interoperability of MGs and clusters of MGs, including both modes of operation: isolated and grid-connected. We review theoretical approaches and practical implementations that consider the effects of the communications network on the general performance of the MG. Moreover, we undertake an experimental analysis of the influence of wired and wireless communication networks on MG performance, revealing the importance of designing future smart control solutions more robust to communication degradation, especially if wireless technologies are integrated to provide scalable deployments. Aspects such as resilience, security, and interoperability are also shown to require continuing efforts in research and practical applications.
Full-text available
Free and competitive energy markets are a recent and increasing phenomenon in several countries. Understanding these new energy markets and estimating their possible evolutions are current challenges of the research community. To avoid real market risks, the research community have developed autonomous traders and tested them in the Power Trading Agent Competition (PowerTAC), a complex energy market simulator. In this paper, we present COLDPower'16, a competitive autonomous trader composed of expert agents in specific kinds of markets and customers that combines local strategies into a global strategy to maximize profit. The local strategy of each tariff expert agent uses reinforcement learning algorithms, while the local strategy of the wholesale expert agent estimates future energy prices and the amount of energy that can be negotiated to buy energy when prices are low and sell energy when prices are high. COLDPower'16 was tested in Power TAC 2016. It achieved 2th place in the final round of this international competition with 7 autonomous agent brokers.
Full-text available
Incorporation of the renewable resources (RR) with regular grid has many challenges like location difference between the renewable energy source and energy distribution point, and characteristic constraints like variability in weather may reduce wind energy, and reduced rain may result in lower hydroelectric energy and so on. In this work, an efficient method is proposed to integrate the RR energy in to the regular power grid. By integrating the RR with conventional energy grid, the user consumption charges can be considerably reduced and, the exploitation of fossil fuels can be gradually reduced. The work also proposes an efficient mechanism to employ real‐time tariff management by demand control center that collects the usage information in real‐time scale. Experimental simulations are performed using MATLAB/SIMULINK environment. For real‐time simulations, fuzzy controller‐ and TCP/IP‐based communication technologies are used. The results show that the proposed work exhibit a better performance against SCADA system, and it shows effective implementation of tariff and energy utility management.
Full-text available
The paper proposes a stochastic model to analyse the dynamic coupling of the transmission system, the electricity market and microgrids. The focus is on the impact of microgrids on the transient response of the system and, in particular, on frequency variations. Extensive Monte Carlo simulations are performed on the IEEE 39-bus system, and show that the dynamic response of the transmission system is affected in a non trivial way by both the number and the size of the microgrids.
Full-text available
Internet of Things (IoT), one of the key research topics in recent years, together with concepts from Fog Computing, brings rapid advancements in Smart City, Monitoring Systems, industrial control, transportation and other fields. These applications require a reconfigurable sensor architecture that can span multiple scenarios, devices and use cases that allow storage, networking and computational resources to be efficiently used on the edge of the network. There are a number of platforms and gateway architectures that have been proposed to manage these components and enable application deployment. These approaches lack horizontal integration between multiple providers as well as higher order functionalities like load balancing and clustering. This is partly due to the strongly coupled nature of the deployed applications, a lack of abstraction of device communication layers as well as a lock-in for communication protocols. This limitation is a major obstacle for the development of a protocol agnostic application environment that allows for single application to be migrated and to work with multiple peripheral devices with varying protocols from different local gateways. This research looks at existing platforms and their shortcomings as well as proposes a messaging based modular gateway platform that enables clustering of gateways and the abstraction of peripheral communication protocol details. These novelties allow applications to send and receive messages regardless of their deployment location and destination device protocol, creating a more uniform development environment. Furthermore, it results in a more streamlined application development and testing while providing more efficient use of the gateway’s resources. Our evaluation of a prototype for the system shows the need for the migration of resources and the QoS advantages of such a system. The examined use case scenarios show that clustering proves to be an advantage in certain use cases as well as presenting the deployment of a larger testing and control environment through the platform.
In this era of advanced distribution automation technologies, demand response is becoming an important tool for electricity network management. The available flexible loads can efficiently help in alleviating the network constraints and achieving demand-supply balance. Therefore, this forms the rationale behind this paper, which aims to implement a multi-agent system framework in order to achieve flexible price-based demand response. A genetic algorithm-based multi-objective optimization technique is applied to determine the optimal locations and the amount of required demand reduction in order to keep the network within statutory limits. The methodology is based on probabilistic estimation of the granularity of total available flexible demand from shiftable home appliances in each low-voltage feeder. Moreover, an optimal decision making for the start time of appliances upon receiving a real-time price signal is proposed. This is accomplished by considering the willingness to participate as well as price demand elasticity of the different clusters of customers. To fully demonstrate the feasibility and effectiveness of the proposed framework, a modified IEEE 69 bus distribution network comprising 1824 low voltage residential customers has been implemented and analyzed.
Microgrids are considered one of the most promising solutions to integrate renewable distributed generation into the electric power system. During the last decade, the microgrid concept has been studied and developed and nowadays it is becoming a reality. Hence, in the coming years a transformation of the current electric power system to a multi-microgrid power system can be expected. In this direction, the study of multi-microgrids is currently being explored. Accordingly, this paper examines the possible multi-microgrid architectures to form a grid of microgrids. For this purpose, the microgrid as a single entity and its possible interactions with external grids is first defined. Then, the possible multi-microgrid architectures are defined in terms of layout, line technology and interface technology. Finally, a comparison between the different architectures is performed in terms of cost, scalability, protection, reliability, stability, communications and business models. This analysis is expected to be of great utility for grid planners and policy makers, who can select the most adequate architecture in function of their necessities.
The concept of multi-microgrids has the potential to improve the reliability and economic performance of a distribution system. To realize this potential, a coordination among multiple microgrids is needed. In this context, this paper presents a new distributed economic model predictive control scheme for the coordinated stochastic energy management of multi-microgrids. By optimally coordinating the operation of individual microgrids, this scheme maintains the system-wide supply and demand balance in an economical manner. Based on the probabilistic forecasts of renewable power generation and microgrid load, this scheme effectively handles the uncertainties in both supply and demand. Using the Chebyshev inequality and the Delta method, the corresponding stochastic optimization problems have been converted to quadratic and linear programs. The proposed scheme is evaluated on a large-scale case that includes ten interconnected microgrids. The results indicated that the proposed scheme successfully reduces the system wide operating cost, achieves the supply-demand balance in each microgrid, and brings the energy exchange between DNO and main grid to a predefined trajectory.