MODELING SMART GRIDS AS COMPLEX SYSTEMS

THROUGH THE IMPLEMENTATION OF INTELLIGENT HUBS

Jos

e Gonz

alez de Durana, Oscar Barambones

University College of Engineering, University of the Basque Country, Nieves Cano 12, 01006 Vitoria-Gasteiz, Spain

Enrique Kremers, Pablo Viejo

European Institute for Energy Research, Karlsruhe Institute of Technology and EDF

Emmy Noether Str. 11, 76131 Karlsruhe, Germany

Keywords:

Electrical grid, Hybrid renewable energy systems, Energy saving, Microgrid, Smart meter, Intelligent hub,

Random graph, Complex system, Complex computer system, Scale free network, Agent based model.

Abstract:

The electrical system is undergoing a profound change of state, which will lead to what is being called the

smart grid. The necessity of a complex system approach to cope with ongoing changes is presented: combining

a systemic approach based on complexity science with the classical views of electrical grids is important for an

understanding the behavior of the future grid. Key issues like different layers and inter-layer devices, as well

as subsystems are discussed and proposed as a base to create an agent-based system model to run simulations.

1 THE ELECTRICAL GRID AS A

COMPLEX SYSTEM

The electrical grid as a whole can be considered as

a complex system (more properly a Complex Com-

puter System) whose aim is to assure a reliable power

supply to all its consumers. Only regarding the grid

from a multi-disciplinary point of view can help us

understand the behavior of these systems. Despite

conceptual advances in concrete ﬁelds like chaos the-

ory or emergence in non-linear or self-organized sys-

tems, which were studied in the last decades, a uniﬁed

theory of complexity does not yet exist.

Complex networks have been studied by several

scientists. Erd

os and R

enyi (1959) suggested the

modeling of networks as random graphs. In a ran-

dom graph (Bollob

as, 1998), the nodes are connected

by a placing a random number of links among them.

This leads to a Poisson distribution when considering

the numbers of connections of the nodes, thus there

are many nodes with a similar number of links.

Watts and Strogatz (1998) deﬁned β as the prob-

ability of rewiring an edge of a ring graph and called

these networks small-world. Analyzing networks

with values 0 < β < 1, they found that these systems

can be highly clustered, with a relatively homogenous

topology, and have small characteristic path lengths.

However, the study of networks in the real world

has shown that there are many examples where this is

not true but they exhibit a common property: the num-

ber of links k originating from a given node exhibits

a power law distribution P(k) ∝ k

−γ

, i.e. few nodes

having a large number of links. These networks are

called scale-free and they are located in between the

range of random and completely regular wired net-

works. Many systems in the real world such as neural

networks, social networks and also the power grid,

fulﬁll these properties.

Barab

asi and Albert (2002) mapped the topology

of a portion of the World Wide Web and found that

some nodes, which they called hubs, have many more

connections than others and that the network as a

whole exhibits a power-law distribution for the num-

ber of links connecting to a node. Using the Barab

asi-

Albert network model, Chassin and Posse (2005) an-

alyzed the topologies of the North American electric

grid to estimate their reliability and calculated the ex-

ponent of scale-free power law as being λ = 3.04 for

the U.S. eastern grid and λ = 3.09 for the western one.

Considering all of the advancements in complex-

ity science, in this paper we will show how an electric-

ity grid can be represented through a model as a com-

plex system that can be used for simulations. First, the

smart grid will be presented and some key issues dis-

146

González de Durana J., Barambones O., Kremers E. and Viejo P. (2010).

MODELING SMART GRIDS AS COMPLEX SYSTEMS THROUGH THE IMPLEMENTATION OF INTELLIGENT HUBS.

In Proceedings of the 7th International Conference on Informatics in Control, Automation and Robotics, pages 146-151

DOI: 10.5220/0003003701460151

 SciTePress

cussed. Then, the approach for modeling the grid is

explained and in the last section the simulation model

is presented.

2 THE SMART GRID

The term smart grid as introduced by Amin and Wol-

lenberg (2005), usually covers the entire spectrum

of the electrical system, reaching from transporta-

tion over distribution up to the delivery. In common

with earlier deﬁnitions, it contains two key elements:

digital data processing and communication networks.

Therefore, it can be said that what characterizes this

intelligent grid is the existence of a ﬂow of data and

information, between the supplier company and the

consumer, running in parallel with the energy ﬂow

(Singer, 2009).

With today’s smart grid goals in mind, energy sup-

ply companies are in a transition process between our

real electricity grid and the future smart grid, trying to

improve the conventional network infrastructure, es-

tablishing the digital level (essence of the intelligent

network) and also creating new business processes to

carry out the capitalization and commercialization of

the intelligent network.

The operation of the smart grid is far more com-

plicated than the conventional power grid and in order

to be operated, some special components like com-

puters, sensors, remote controlled switching devices,

as well as communication networks are necessary.

For example, the current power grid is still not ready

to admit microgrid connections. Connections made

at present are experimental and almost always done

manually, by taking care that a number of factors are

fulﬁlled e.g. before realizing a connection.

Trying to model a microgrid, the network to which

it will be connected should also be considered. Al-

though a large amount of work in this area has been

done, the main problem is that the current electricity

grids are not yet adequately prepared for the transi-

tion to the smart grid. Therefore, in this article, the

authors do not consider the inadequate existing grid,

but instead focus on a hypothetical future network:

the smart grid.

2.1 The two Layer Model

Concerning the upcoming challenges, especially fac-

ing the growing need of interaction of the different

units of the smart grid, a two layer model is proposed

(Kremers et al., 2010). These were identiﬁed as:

Physical Layer. The ﬁrst layer is the physical struc-

ture of the electrical grid itself, including all the

Figure 1: Different layers in an electrical microgrid.

power transmission lines. It includes the power

ﬂows as well as all the electrical devices related

to the correct operation of the grid.

Logical Layer. This second layer, which represents

the main part of the upcoming generation of elec-

trical grids, is not yet present, in contrast to the

physical layer. This layer includes all the infor-

mation exchange that has to be arranged to control

distributed generation (DG), dispatchable loads

and other smart equipment in future grids. It has

to be underlined that the communication paths do

not have to be the same as the links in the ﬁrst

layer, although they could be exploited for that

aim. An example of this is Power Line Commu-

nication (PLC).

The current electricity grid could be seen as part

of the ﬁrst layer, whereas the second layer is still

the focus of vast research and development. It rep-

resents all the information and communication tech-

nology linked in some way to the grid and its oper-

ation. It implements a system that allows real-time

communication between the elements of the grid. In

the E-Energie project (2010), an Internet of Energy is

suggested as an analogy to computer networks. This

medium could itself serve as a communication plat-

form. More examples of the implementation of the

logical layer are described in Kremers et al. (2010)

and could be PLC, existing communication networks,

wireless technologies, etc.

In the following sections, some key role playing

concepts of the smart grid will be exempliﬁed and

discussed. First of all, smart metering as a technol-

ogy under deployment is presented. Afterwards, the

MODELING SMART GRIDS AS COMPLEX SYSTEMS THROUGH THE IMPLEMENTATION OF INTELLIGENT

HUBS

147

concept of intelligent hub is introduced as a generic

modeling approach for intelligent devices in the fu-

ture grid. Finally, the microgrid as a sub-system of

the smart grid is discussed.

2.2 Smart Metering

Traditionally, an energy meter measures only the con-

sumption of the total energy during a speciﬁc time.

This is used for billing the customer the total amount

of energy he consumed. There is no way to obtain in-

formation on when the energy was consumed nor in

what way. Smart meters are intended to provide more

detailed information which will allow the supplier to

adjust the pricing for consumption based on different

parameters.

Electricity prices vary during a day or season, fol-

lowing the market offer-demand principle (especially

with the introduction of renewable, non-dispatchable

sources) or due to external factors such as tempera-

ture. Using a multiple tariff based system will allow

for the reﬂection of these pricing changes to the ﬁnal

customer and thus entice him to make a more eco-

nomical use of energy. These pricing signals shall

help to reduce peak loads and sell more energy in off-

peak periods, e.g. during the night.

The ESMA (European Smart Metering Alliance)

deﬁnes a smart meter as an advanced meter with

several functions, such as automatic data processing

and transfer, automatic performing of measurements,

which provides meaningful and up-to-date informa-

tions of consumption to the relevant actors and units

of the energy system. Additionally, smart meters can

provide support for measures to increase energy efﬁ-

cient consumption. Proof of the relevance of this de-

vice are the statements made by governments of dif-

ferent nations worldwide. For example, three of them

have been chosen:

• Malta, where a pilot project is currently under-

way, with more than 5,000 smart counters being

installed. The objective for 2012 is to have only

smart meters in use.

• The United Kingdom, where in December 2009

the U.K. Department of Energy and Climate

Change announced its intention to have smart me-

ters in all homes by 2020.

• The U.S. where, according to Edison’s Institute

for Electric Efﬁciency, many of the country’s

largest electricity distribution companies have

plans to install millions of meters in the coming

years, with deployments to be complete between

2012 and 2015.

Figure 2: Architecture of an Intelligent Hub.

However, in the authors’ opinion, energy savings

will only be achieved when the meters are reinforced

by new devices and directives such as information dis-

plays, time-varying pricing, energy audits and, in par-

ticular, some form of automatic load control. The fact

that companies such as Intel, Cisco or Google are de-

veloping hardware and software for this growing mar-

ket, seems to conﬁrm that idea.

2.3 Intelligent Hubs as Interaction

between the Layers

Having described the smart grid properties, the ques-

tion of how the assumed measures can be imple-

mented to make the grid smarter is apparent. The

approach taken in this study is to model some spe-

cially designed generic units called Intelligent Hubs

that:

• implement the communication functions of the

logical layer,

• monitor the physical grid,

• perform data processing and evaluation,

• can take actions on the physical grid,

• can act as a local decision unit, and

• handle any interactions between the logical and

physical layer.

ICINCO 2010 - 7th International Conference on Informatics in Control, Automation and Robotics

148

The introduction of the intelligent hub arises from

the idea that there are many different technologies

and implementation possibilities for new infrastruc-

ture equipment, but no standard deﬁnition of these

new intelligent units currently exists. For example,

at household level a local load shedding module con-

nected to a smart metering system could be imple-

mented, or at a substation level new technologies that

are be able to communicate with customers to send

e.g. grid state signals, etc. are possible.

They all have in common that they share at least

some of the characteristics of the intelligent hubs

named above. This allows us to model a generic in-

telligent hub, which accomplishes with the speciﬁca-

tions given and is able to simulate the behavior of

these future equipment elements, even though a con-

crete implementation is not realized. It is important to

underline that the intelligent hub is the link between

the two layers, logical and physical, thus gathering

information from both of them, being able to pro-

cess it and actuating on the physical layer to perform

changes. The acquisition of the data from the physical

grid is performed by sensors, for example measuring

units on the lines. The actuators are any kind of in-

teraction with the power grid, such as demand control

for example by direct (such as relays, operating on the

line), or indirect means (dynamic demand reduction

of the equipment).

The question of how far a smart meter can be seen

as an intelligent hub or vice versa has to be analyzed

further. There exist implementations of smart meters

that seem to accomplish with some of the intelligent

hub features (like load shedding functions), but in our

opinion this already goes beyond the concept of me-

tering. So, at least for modeling purposes, the smart

meter will be seen as a part of the intelligent hub or an

external unit linked to this, as the concept of the hub

involves a much broader list of features, which can be

summarized as the whole interaction between the two

layers – even at different levels of the grid.

2.4 Microgrids as Smart Grid

Subsystems

A microgrid is a set of small energy generators ar-

ranged in order to supply energy for a community of

users in close proximity. It is a combination of gen-

eration sources, loads and energy storage, interfaced

through fast-acting power electronics. Emerging from

the general trend of the introduction of Renewable

Energy Sources (RES), microgrids will mostly in-

clude this type of generation, so they form part of the

Hybrid Renewable Energy Systems (HRES). Micro-

grids represent a form of decentralization of electri-

cal networks. They comprise low- or medium-voltage

distribution systems with distributed energy sources,

storage devices and controllable loads.

During disturbances, the generation and corre-

sponding loads can autonomously disconnect from

the distribution system to isolate the load of the mi-

crogrid from the disturbance without damaging the

integrity of the transmission grid. This mode is called

islanding mode. From the point of view of the cus-

tomer, it can be seen as a low voltage distribution ser-

vice with additional features like an increase in lo-

cal reliability, the improvement of voltage and power

quality, the reduction of emissions, a decrease in the

cost of energy supply, etc.

PCC

Transmision Grid

High Voltge 380 kV

Medium Voltge 50 kV Distribution Grid

Figure 3: Integrated microgrid.

In Figure 3, a schematic drawing of an inte-

grated microgrid can be seen, showing the Point of

Common Coupling (PCC) and its electrical connec-

tions, but without representing the information chan-

nels. The authors have previously identiﬁed (Kremers

et al., 2010) that electrical grids, as well as micro-

grids mostly satisfy the principal characteristics that

distinguish them as true systems of systems, as de-

ﬁned by Maier (1998). The smart grid is constituted

as a large complex system with operational and man-

agerial independent elements (that are systems them-

selves), evolutionary development, emergent behav-

ior and geographical distribution.

3 A COMBINED APPROACH FOR

SMART GRID MODELING

For simulation purposes, one should not pay atten-

tion to accessory components but instead focus on the

essential parts. It may occur that modeling some im-

portant elements is unnecessary for the operation, as

MODELING SMART GRIDS AS COMPLEX SYSTEMS THROUGH THE IMPLEMENTATION OF INTELLIGENT

HUBS

149

for example some electronic components that while

being essential for the actual operation they are trivial

or nonexistent for simulation.

So in this article, we focus on a very important

element to consider both the actual operation of the

microgrid and the simulation: the smart meter. As-

suming that there is an intelligent hub at each network

bus, a smart meter at each load bus and there exists

communication among the nodes, the resulting com-

plex computer system can be used as a basis for smart

grid models.

Agent-based modeling tools are able to recreate

complex system behavior such as those described

here, unexpected emergent behavior in these systems,

internal and external events, communications within

the system, etc. In particular, local effects of the sin-

gle units comprising the system can be modeled and

their effects can be analyzed at the system level.

The combination of several approaches allows the

creation of models that might abstract some details

from the single unit models, but all in all create a

much more realistic representation at the system level.

The inclusion of some communication among the de-

vices is fundamental here. This combined approach

is the one followed in this work, as in the authors’

opinion it is very advantageous when modeling future

electrical systems.

Geographic network localization, distribution of

processing and databases, interaction with humans,

and unpredictability of system reactions to unex-

pected external events are also present in this kind of

network.

4 THE SIMULATION MODEL

A combined approach model has been developed us-

ing Anylogic (XJTek, 2010), in which the grid nodes

are represented by agents in the model, and each

agent is provided of respective subsystem models

(System Dynamics (SD), Discrete Events (DE), etc.).

This kind of modeling has been chosen to allow for

events such as the sudden elimination of one (or more)

nodes, or connections between nodes, to be possible

during the simulation, thus providing the possibility

of a dynamic simulation of the electrical grid, includ-

ing events such as failures, disasters and terrorist at-

tacks.

The two-layer structure deﬁned for the smart grid

will be a key element due to its representation in the

model. Apart from the agent-based approach, two

other modeling paradigms are used: the SD paradigm

for the physical layer and the DE paradigm for the

logical layer.

The model is completely open, so it can be used to

address a number of issues of design and computation

that arise in such networks. Some of them coud be:

• Real grid, smart grid and microgrid simulation

• Grid and microgrid architecture design

• Centralized and decentralized control design

• Load connection and disconnection

• Microgrid connection and islanding modes

• Branch or node (e.g. substation) deleting

• Energy savings strategies

The approach presented here is intended to result

in a suitable electrical microgrid model. It is a con-

tinuation of the authors’ previous studies in this area

(Gonz

alez de Durana et al., 2009; Gonz

alez de Du-

rana and Barambones, 2009). In these works, the

mesh method was used to obtain voltages at the mesh

nodes and currents through the branches, assuming

the voltages given by the generators are known. In a

further study, however, a new power ﬂow method has

been created (Kremers et al., 2010). This new method

was implemented using an combined approach in

which agents represent the buses of the grid.

5 CONCLUSIONS AND

OUTLOOK

A view of energy systems from the complex systems

approach has been given, underlining the importance

of viewing the energy system as such, especially for

its future development. Modeling the energy system

at system level is crucial to help us understand the

interactions of the single units, and be able to ob-

serve system phenomena such as emergence and the

behavior of the system as a whole. The requirement

of the introduction of new perceptions of the energy

system was shown with the proposition of the two

layer model to represent the smart grid. Further, a

device abstraction was done for modeling generic de-

vices called intelligent hubs that are able to interact

with this new environment.

An agent-based model for the simulation of mi-

crogrids is being implemented using AnyLogic. The

model offers a clear two-layer structure, which allows

for the representation of both physical and logical in-

teractions between the elements. The logical layer of-

fers a robust base to implement agent communication

in real time. The physical layer provides the technical

results of the power ﬂow calculation integrated into

the model. This allows real-time simulations of the

ICINCO 2010 - 7th International Conference on Informatics in Control, Automation and Robotics

150

grid to be computed, which can provide valuable in-

formation prior to the implementation of the real grid.

The ﬁrst models of the intelligent hub have been de-

veloped and are currently being tested.

The model is mainly intended to design and test

microgrids and can be used as a tool for the design,

development and demonstration of control strategies,

especially centralised supervisor control and decen-

tralised load-dispatch control, the design and demon-

stration of microgrid operation strategies, the design

and testing of microgrid communication buses and

optimal microgrid design.

ACKNOWLEDGEMENTS

This work was made possible through cooperation be-

tween the University College of Engineering in Vi-

toria and EIFER in Karlsruhe. The authors are very

grateful to the Basque Government for the support

of this work through the project S-PE09UN12 and

to the UPV/EHU for its support through the project

GUI07/08.

REFERENCES

Amin, S. and Wollenberg, B. F. (2005). Toward a smart

grid: power delivery for the 21st century. Power and

Energy Magazine, IEEE, 3(5):34–41.

Barab

asi, A. L. (2003). Linked: How everything is con-

nected to everything else and what it means. Penguin

Group New York.

Barab

asi, A. L. (2007). The architecture of complexity.

Control Systems Magazine, IEEE, 27(4):33–42. 0272-

1708.

Barab

asi, A. L. and Albert, R. (2002). Statistical mechanics

of complex networks. Reviews of Modern Physics, 74.

Bollob

as, B. (1998). Modern graph theory. Springer Verlag.

Chassin, D. P. and Posse, C. (2005). Evaluating north amer-

ican electric grid reliability using the barab

asi-albert

network model. Physica A: Statistical Mechanics and

its Applications, 355(2-4):667–677.

Erd

os, P. and R

enyi, A. (1959). On random graphs, i. Pub-

licationes Mathematicae (Debrecen), 6:290–297.

European Commission. (2006). European smartgrids tech-

nology platform: Visions and strategy for europe’s

electricity networks of the future.

E-Energie Projekt. (2010) Bundesministerium fuer

Wirtschaft und Technologie. http://www.e-energie.de

Gonz

alez de Durana, J., Barambones, O., Kremers, E., and

Viejo, P. (2009). Complete agent based simulation of

mini-grids. In The Ninth IASTED European Confer-

ence on Power and Engineering Systems, EuroPES

2009, volume 681, pages 046–188, Palma de Mal-

lorca, Spain. Acta Press.

Gonz

alez de Durana, J. M. and Barambones, O. (2009). Ob-

ject oriented simulation of hybrid renewable energy

systems focused on supervisor control. In IEEE Con-

ference on Emerging Technologies and Factory Au-

tomation, 2009, ETFA 2009, pages 1–8.

Hatziargyriou, N., Asano, H., Iravani, R., and Marnay, C.

(2007). Microgrids - an overview of ongoing research,

development, and demonstration projects. IEEE

power and energy magazine, (july/august 2007):17.

Jiyuan, F. and Borlase, S. (2009). The evolution of distribu-

tion. Power and Energy Magazine, IEEE, 7(2):63–68.

1540-7977.

Karpov, Y. G., Ivanovski, R. I., Voropai, N. I., and Popov,

D. B. (2005). Hierarchical modeling of electric power

system expansion by anylogic simulation software. In

Power Tech, 2005 IEEE Russia, pages 1–5.

Kremers, E., Lewald, N., Barambones, O., and Gonz

alez de

Durana, J. (2009). An agent-based multi-scale wind

generation model. In The Ninth IASTED European

Conference on Power and Engineering Systems, Eu-

roPES 2009, volume 681, pages 064–166, Palma de

Mallorca, Spain. Acta Press.

Kremers, E., Viejo, P., Gonz

alez de Durana, J. M., and

Barambones, O. (2010). A complex systems mod-

elling approach for decentralized simulation of elec-

trical microgrids. In 15th IEEE International Confer-

ence on Engineering of Complex Computer Systems,

page 8, Oxford.

Maier, M. W. (1998). Architecting principles for systems-

of-systems. Systems Engineering, 1(4):267–284.

Singer, J. (2009). Enabling tomorrow’s electricity system -

report of the Ontario smart grid forum.

Valov, B. and Heier, S. (2006). Software for analysis of inte-

gration possibility of renewable energy units into elec-

trical networks. In The 5th International Conference

Electric Power Quality and Supply Reliability, Viimsi,

Estonia. Proceedings Tallin University of Technology

pp 173-177.

Watts, D. J. and Strogatz, S. H. (1998). Collective dynamics

of ‘small-world’ networks. Nature, 393(6684):440–

442.

XJ Technologies website. (2010). Anylogic.

http://www.xjtek.com

MODELING SMART GRIDS AS COMPLEX SYSTEMS THROUGH THE IMPLEMENTATION OF INTELLIGENT

HUBS

151