Prospective Products and Beneﬁts of the Green AGH Campus Project

Providing Scaled-down Future Smart Grid Experience

Igor Wojnicki, Sebastian Ernst and Leszek Kotulski

Department of Applied Computer Science, AGH University of Science and Technology, Krakow, Poland

Keywords:

Smart Grid, Optimization, Energy Distribution, Grid Design, Control, Optimization, Artiﬁcial Intelligence,

ICT.

Abstract:

Smart grid stakeholders have partially opposing goals. New tools and utilities are needed to reach them. The

paper discusses prospective products and beneﬁts of the Green AGH Campus Project. The project provides

energy management infrastructure being a test bed for future smart grid solutions. It is based on three main

components: an Advanced Distribution Management System, a Power Data Warehouse and a Simulation Sys-

tem. Coupled together, they are capable of providing advices regarding either optimization of grid operations,

or its design and topology. The optimizations are multicriteria and multivariant balancing stakeholders’ needs.

1 INTRODUCTION

“A smart grid is an electricity network that can cost

efﬁciently integrate the behaviour and actions of all

users connected to it – generators, consumers and

those that do both – in order to ensure an econom-

ically efﬁcient, sustainable power system with low

losses and high levels of quality and security of sup-

ply and safety” (European Commission, 2011). Par-

ties involved in smart grid management are identiﬁed

as either operators or consumers. The operators are

responsible for managing the grid. They are subdi-

vided further mainly into: Transmission System Op-

erators (TSO), Distribution System Operators (DSO),

Energy Generators (producers).

TSOs manage the very high voltage grid such as

400 or 225 kV. They connect all regional electricity

grids with each other. TSOs are responsible for cor-

recting the imbalance in the network regarding energy

demand and supply. They are decoupled from retail

and generation. DSOs are responsible for energy dis-

tribution at the regional level. They handle high volt-

age (below 60 kV), medium voltage (1–30 kV) and

low voltage. Their main responsibility is to deliver

energy to the consumers. Producers provide electric-

ity, which is distributed through TSO and DSO grids.

Consumers use electricity delivered by DSOs and

are subdivided into residential and commercial ones.

Taking into account a decentralized energy generation

a new category of consumers has emerged, so-called

prosumers: producing consumers. An introduction

of distributed generation, including renewable energy

sources, is a major game changer for the electricity

delivery. It is no longer a uni-directional ﬂow from

the producers through TSO and DSO grids to the con-

sumers but a multidirectional ﬂow which can be con-

trolled only by proper ICT (Information and Commu-

nications Technology) systems. There are very high

requirements for such systems. They need to be: geo-

graphically distributed, heterogeneous, real-time and

must allow for partial disconnections. What is more

they generate massive amounts of data. The electric

grid, in this context, is perceived as the most wide

spread adoption of the Internet of Things concept.

The stakeholders of the energy business have mul-

tiple, even opposing, goals. Among others these are:

non technical loss identiﬁcations (theft), energy price

and usage optimization, energy quality assurance,

emergency response time minimization etc. There

are multiple efforts carried out to provide proper ICT

tools. They regard grid planning, simulation and real-

time management.

There are several big IT players involved, includ-

ing: IBM, Oracle, SAP, Google, Cisco

. The main

focus of rapidly progressing smart grid solutions is to

provide more insight into processes taking place on

the grid and beyond to provide integration with Smart

City concepts. There are also emerging businesses

offering added values on top of data analytics such as

source: European Utility Week 2014, Amsterdam:

http://www.european-utility-week.com

395

Wojnicki I., Ernst S. and Kotulski L..

Prospective Products and Beneﬁts of the Green AGH Campus Project - Providing Scaled-down Future Smart Grid Experience.

DOI: 10.5220/0005494403950400

In Proceedings of the 4th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS-2015), pages 395-400

ISBN: 978-989-758-105-2

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

Opower or Nest Labs.

This paper focuses on the DSO perspective and

surroundings. It involves the end customers, includ-

ing prosumers, and sellers. The aim is to present an

overview of products and beneﬁts of the Green AGH

Campus project (Szmuc et al., 2012). The project

takes into consideration a campus area. It is a model

of a scaled down urban area representing a distribu-

tion network. Thus it serves as a testbed to research

new energy management strategies, develop new tools

for future smart grid, being a living lab. The project

is led by AGH University of Science and Technology,

hence its name.

2 STAKEHOLDERS

There are three major stakeholders taken into consid-

eration. These are: a DSO, an end customer, and sales

(the producers or prosumers), see Fig. 1 for details.

They are targeted at simple yet divergent goals. A

DSO’s goal is to maintain grid operations. A cus-

tomer wants to lower his electricity bills. While sales

want to maximize their income.

There are two main observable drivers than: cus-

tomer satisfaction and compliance with regulations.

The customer satisfaction drives all stakeholders how-

ever after analyzing it, it turns out, that it regards dif-

ferent, even opposing goals such as: income maxi-

mization and energy bill reduction. The main assess-

ments of this driver are: non interruptible delivery,

better power quality and wise energy usage. The non

interruptible delivery and better power quality con-

tribute to the DSO’s main goal which is maintaining

grid operations. The wise energy usage contributes

towards the income maximization for the sales as well

as energy bill reduction for the customer.

The compliance with regulations drives the DSO.

It is a complex driver which main components can be

characterized as: regarding renewables, CO2 emis-

sion and reliability

. The compliance is usually

forced upon DSOs by regulators and law. Assess-

ing this compliance results in: non interruptible deliv-

ery and better power quality, which regard maintain-

ing grid operations and income maximization goals

which are main objectives for the DSO and sales re-

spectively.

So, as it can be seen from the above, these three

parties and their goals are highly entangled. Fulﬁll-

ing them is a great challenge. Providing proper tools

which can support reaching the goals is the ultimate

EU climate and energy package http://ec.europa.eu/

clima/policies/package/index en.htm

goal for the proposed research. The tools regard ICT

solutions, including algorithms, software, and anal-

ysis, as well as methodologies and good practices.

They focus on two major areas being either topology

optimization or control optimization (see Fig. 2), the

former regarding grid design, the latter regarding grid

operations. Thus these are the requirements that re-

alize the goals (energy bill reduction, income maxi-

mization, maintaining grid operations). Particularly

optimizing grid operations is a complex requirement

which consists, among others, of providing dynamic

tariffs, restore service after a failure as soon as possi-

ble while minimizing number of disconnected loads,

and stabilize grid to provide uninterrupted operations.

3 PRODUCTS AND BENEFITS

There are three main product categories of the pro-

posed Green AGH Campus Project. These are: scien-

tiﬁc, commercial, didactic.

The scientiﬁc category ranges from researching

topics related to distributed computing, agent-based

systems, optimization, to artiﬁcial intelligence (AI).

Distributed computing in this contexts regards analy-

sis of large volumes of data, often referred to as Big

Data. Signiﬁcance of this topic is reinforced by sole

nature of power grid which is distributed by nature.

Amount of data generated by grid sensors, including

smart meters, and their high scale is uncomparable

and without precedence. Microgrid and islanded op-

erations force smart grid solutions to be more ﬂexible

and capable of distributed processing. Contemporary

grid managements are usually centralized, thus fur-

ther applied and interdisciplinary research regarding

agent-based applications and systems is required. As

it has been already researched (Wojnicki et al., 2013;

Kotulski, 2008; Kotulski and Sedziwy, 2012), apply-

ing agent-based approach both to process large vol-

umes of data and provide optimized control could of-

fer proper solutions.

The above systems need to be supported with ar-

tiﬁcial intelligence, mostly to optimize their modus

operandi. It is expected to research and deploy ma-

chine learning with advanced pattern recognition and

planning and preference modeling (Klimek et al.,

2013). It could serve identiﬁcation of processes tak-

ing place on the grid and supporting operators with

admissible solutions.

Considering the above there is a need for knowl-

edge management at the grid level as well as to pro-

vide proper inputs and outputs for AI methods. Simi-

larly, there is a need for formal speciﬁcations for such

distributed control systems to verify their behavior

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

396

Figure 1: Stakeholders and goals.

Figure 2: Goal realisation.

and assure their safety.

Another aspect of AI which ﬁnds applications

here is inference. Real-time inference is needed to

provide context aware control and early malfunction

detection. It also includes automatic state interpreta-

tions providing advices and serving as analytics plat-

form, thus supporting operators and engineers at the

control center, at DSO, or applications to other par-

ties. One of important themes here is so-called non-

technical loss identiﬁcation, which is regards energy

theft.

Multiple optimization methods and concepts ﬁnd

applications here as well. Especially multidimen-

sional, multicriteria and multivariant algorithms.

They can be used both for power grid design or con-

trol.

Last but not least there is a need for historic data

analysis. It allows to make either control, design or

business decisions based on insufﬁcient data, thus ap-

proximating or ﬁlling gaps with the interpreted his-

torical ones. It is not only “foreseeing” the future but

also fulﬁlling gaps in data streams representing states

of the entire grid system overload, lost communica-

tion etc.

The commercial type of outcomes regards actual

products and services that could be offered directly to

the involved parties. Besides new grid control algo-

rithms which needs to be tightly integrated with DMS

systems, or methods for grid design, being products

for DSOs, the most prominent one would be establish-

ing and providing dynamic tariffs. It would be a result

of multicriteria analysis using researched earlier do-

mains, since there are different goals of the involved

stakeholders. Thus dynamic tariffs can improve main-

taining grid operations, maximizing sales proﬁts and

minimizing energy bills of the end customers.

The didactic category of the outcomes is not sub-

ject of this paper however it should be noted here

since the project is developed by AGH University.

As a result all other outcomes: research results, al-

ProspectiveProductsandBenefitsoftheGreenAGHCampusProject-ProvidingScaled-downFutureSmartGrid

Experience

397

gorithms, methodologies, tolls can ﬁnd a way to be

integrated with curricula.

4 ARCHITECTURE

The proposed concept is based on three main com-

ponents. These are: an aDMS (Advanced Distribu-

tion Management System), a Power Data Warehouse,

and the Simulation System (see Fig. 3: Distribution

Management System, Data Warehouse, Simulation

respectively). It is assumed that the aDMS works in

a redundant mode, based on two instances: the ac-

tual aDMS, managing the distribution of energy and a

shadow instance, also called the shadow environment.

The shadow instance is not connected to physical de-

vices. Its role is to verify the energy management pro-

cess under varying (real or simulated) circumstances

and parameters. These include raw telemetric data

from ﬁeld devices, raw simulated data, simulated and

real decisions made by the operators and engineers

as well as structural modiﬁcations to the managed

network. This approach allows, on one hand, efﬁ-

cient management of power, and on the other – opti-

mization of the network operation based on any crite-

ria, including simulation-based implementation of the

assumed management process and network structure

changes.

Using an ETL (Extraction-Transformation-

Loading) process, the data warehouse collects

information regarding the network operation from

the aDMS. This allows for its further analysis using

BI (Business Intelligence) methods and extraction

of results and conclusions which can serve as

input to the simulation environment. Simulations

allow for veriﬁcation of the aDMS operation and

implementation of arbitrarily complex optimization

processes, based on multiple criteria. Separation

of the simulation system from the aDMS makes

it independent from technological and conceptual

limitations of current aDMS systems and makes its

performance less critical. Performed analyses do

not affect the management of the actual network,

and therefore cannot destabilize it. It must be noted

that while the simulation system uses information

about the network structure from the ‘production’

aDMS instance, the generated operation parameters

are submitted to the shadow instance.

aDMS systems extend classical SCADA solutions

by assuring appropriate scalability and integration

with the power grid operator’s business processes.

They usually have a modular structure, which typi-

cally includes the following components: SCADA,

NMS (Network Management Systems), OMS (Out-

age Management Systems), FDIR (Fault Detection,

Isolation and Recovery) and VVC (Volt-Var Control).

The SCADA module is responsible for communica-

tion with local SCADA systems and directly with

ﬁeld devices; it acquires telemetric data and assures

efﬁcient remote control. The NMS module provides

network management features, offering an operator’s

interface (including e.g. schematic and geographic

visualization, an event log, notiﬁcations, event reac-

tions, etc.) and cooperating with the SCADA module

by allowing for appropriate control of the entire man-

aged network. The OMS module allows for mainte-

nance planning and scheduling and manages fault no-

tiﬁcations and handling. It often integrates data from

smart meters, allowing for automatic detection of dis-

connected grid segments. The above modules consti-

tute a DMS-class system. By further developing its

functionality, an aDMS (Advanced DMS) system can

be created. This is usually achieved by integrating

additional modules, responsible for network opera-

tion optimization, such as FDIR and VVC. The FDIR

module allows for automatic isolation of faults and

provides automatic switching (or recommendations

for the operator) to minimize losses, e.g. the number

of affected customers. VVC stabilizes network volt-

ages and power distribution. Such modules often use

additional helper modules which perform power ﬂow

calculations on a given network structure or simulate

its behavior. aDMS systems are also often equipped

with modules which allow ﬁeld crews to directly in-

teract with the system, which gives them information

about the current network status and provides means

of reporting diagnostic and repair actions.

Simulation of power grids is perform both within

aDMS-class systems and by means of external tools.

Simulations are usually performed on three logical

levels: device, network and comprehensive simula-

tion. Device-level simulation involves replacement of

real ﬁeld device input and output data with simulated

information. This allows for simulation of the DMS

system behavior as well as its reaction to generated

events. Solutions such as SimSCADA can be used for

this purpose. Network-level simulation goes a step

further, offering input and output data in conjunction

with the grid topology. This allows for generation

of events involving groups of devices as they inter-

act within the network. Often, such simulations also

allow for generation of events which, under normal

circumstances, are the decisions of the NMS or the

system operator. SimNet is a ready-to-use solution

of this class. Finally, comprehensive simulations in-

volve all possible signiﬁcant grid components as well

as factors which affect them: end devices, network

parameters and topology, source and load character-

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

398

Figure 3: Architecture and requirements realisation.

istics and weather conditions. One of the solutions

which can be used in such scenario is GridLAB-D,

described further in this section.

Data warehouses allow for central storage and

analysis of business data. They usually store data

from long time spans which can be used to obtain new

information, signiﬁcant for the business operation.

Therefore, a data warehouse provides a uniﬁed data

store oriented around business entities, which is: inte-

grated (as it contains data from different subsystems),

time-variant (as they contain data from various peri-

ods) and non-volatile (as the data is loaded incremen-

tally from transactional systems). Data warehouse so-

lutions are provided by, among others, Oracle, IBM,

SAS and Microsoft. These solutions often use pro-

prietary or open software components for data ac-

quisition, discovery and analysis, including relational

databases (Oracle, DB2, SQL Server, PostgreSQL,

Informix), non-relational data stores (HBase, Mon-

goDB, Cassandra) and AI-based solutions (classiﬁca-

tion, machine learning, statistical analysis, decision-

support systems, etc.).

GridLAB-D, developed by the Paciﬁc Northwest

National Laboratory and supported by the US Depart-

ment of Energy, is an advanced, open-source power

system simulation tool based on the BSD license.

This system is one of the leading grid simulation

tools, which provides accurate results thanks to pre-

cise models. Thanks to GridLAB openness, it is pos-

sible to modify any of its components, which is often

necessary to adjust it to local conditions and require-

ments.

GridLAB-D is a modular system: besides the

“central” power ﬂow simulation module, it has ded-

icated modules for precise simulation of loads (in-

cluding residential and commercial customers), dis-

tributed generation, weather, energy market compo-

nents, reliability testing (including metrics such as

SAIFI or SAIDI), power storage or electric vehicle

charging.

However, while the system has been widely used

in the USA (as the project was started by the US De-

partment of Energy) and Australia (e.g. within the

Smart Grid, Smart City project

), its use in Europe

was very limited. Therefore, it is crucial to verify

how each component complies to the characteristics

of European (and Polish) power grids and the local

regulations. Evaluation of each module will provide

the information whether it needs to be modiﬁed and to

what extent – from adjusting parameters to algorithm

and model modiﬁcations. This process will be per-

formed by the University’s software and power grid

staff, as well as external experts.

The foreseen changes include customization of the

modules for proper operation with voltages and phase

conﬁgurations used in Europe and better coherence

with real devices and installations. For instance, the

residential module includes predeﬁned load schemes

for houses and apartments and algorithms to multiply

the instances with set parameter variations, which al-

lows for easy and scalable simulation of loads gener-

http://www.smartgridsmartcity.com.au

ProspectiveProductsandBenefitsoftheGreenAGHCampusProject-ProvidingScaled-downFutureSmartGrid

Experience

399

ated e.g. by newly-built housing estates. However, its

characteristics are typical for American households,

with deep penetration of air conditioning systems, dif-

ferent thermal insulation parameters and heating solu-

tions as well as different characteristics of the appli-

ances themselves. It is a similar case with the com-

meracial module, which simulates loads originating

from commercial and ofﬁce buildings.

Practically all modules are predicted to require

modiﬁcations. GridLAB-D’s supported weather data

formats are vastly popular in the USA, but less so in

Europe; data formats and conversion workﬂows need

to be established. Models for DG devices (PV, wind

turbines, heat pumps) need to be customized to bet-

ter match European devices. Quality metrics used by

Polish and European utilities need to be implemented

and conﬁgured; also, for economics-related simula-

tions, the energy trading schemes may need to be ad-

justed to local characteristics.

The above components cooperate (see Fig. 3) to

form an Advice which in turn can be used by the

aDMS. Providing the Advice fulﬁlls requirements:

Optimize grid operations and Optimize design and

topology, which in turn realize the goals (see Fig. 2).

5 CONCLUSIONS

The Green AGH Campus project is based on state-

of-the-art solutions and methods to ensure it is on

the cutting edge of distribution grid management re-

search. The concept, based on three components

(aDMS, Power Data Warehouse, Simulation), is an

innovative approach which allows for better exploita-

tion of data and detection of knowledge (rules, pat-

terns) regarding grid operations. Because AGH Uni-

versity is not a TSO or DSO, but a research-oriented

scientiﬁc institution, the potential for improvement

and experimental veriﬁcation of results is much big-

ger than in case of industrial, production systems.

One possible criticism may address the fact the

system will be initially deployed on the university

campus power network, which is of a signiﬁcantly

smaller scale than the intended application. However,

all components are assessed with regard to their scal-

ability, and tight integration of simulation solutions

allows to perform experiments on models of larger,

real-life distribution grids.

Finally, one of the important aspects of the Green

AGH Campus projects is the possibility to develop a

range of services, from operation parameter optimiza-

tion to strategic investment planning, which can be

offered to operators on the Polish and European mar-

kets.

ACKNOWLEDGEMENTS

This research is ﬁnanced by AGH University of Sci-

ence and Technology contract number 11.11.120.859.

REFERENCES

European Commission (2011). Smart grid task force man-

date.

Klimek, R., Wojnicki, I., and Ernst, S. (2013). State-

space reduction through preference modeling. In

Rutkowski, L., Korytkowski, M., Scherer, R.,

Tadeusiewicz, R., Zadeh, L. A., and Zurada, J. M.,

editors, Artiﬁcial Intelligence and Soft Computing,

volume 7895 of Lecture Notes in Computer Science,

pages 363–374. Springer Berlin Heidelberg.

Kotulski, L. (2008). Distributed graphs transformed by

multiagent system. In Rutkowski, L., Tadeusiewicz,

R., Zadeh, L., and Zurada, J., editors, Artiﬁcial Intel-

ligence and Soft Computing ICAISC 2008, volume

5097 of Lecture Notes in Computer Science, pages

1234–1242. Springer Berlin Heidelberg.

Kotulski, L. and Sedziwy, A. (2012). On the effective distri-

bution and maintenance of knowledge represented by

complementary graphs. T. Computational Collective

Intelligence, 6:105–120.

Szmuc, T., Kotulski, L., Wojszczyk, B., and Sedziwy, A.

(2012). Green AGH Campus. In Donnellan, B.,

Lopes, J. A. P., Martins, J., and Filipe, J., editors,

SMARTGREENS, pages 159–162. SciTePress.

Wojnicki, I., Kotulski, L., and Ernst, S. (2013). On scalable,

event-oriented control for lighting systems. In Bar-

bucha, D., Le, M. T., Howlett, R. J., and Jain, L. C.,

editors, KES-AMSTA, volume 252 of Frontiers in Arti-

ﬁcial Intelligence and Applications, pages 40–49. IOS

Press.

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

400