Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements

(1)

Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements

WARD SNOECK

Master’s Degree Project Stockholm, Sweden 2013

XR-EE-ICS 2013:011

(2)

Master Thesis

Quality of Service Monitoring in the Low Voltage Grid by using Automated

Service Level Agreements

Author:

Ward SNOECK

Supervisor:

Shahid HUSSAIN Examinator:

Lars NORDSTR ¨ OM

A thesis submitted in fulfilment of the requirements for the degree of MSc. Energy Innovation and Smart Cities

in

Industrial Information Systems and Control Department of Electrical Engineering

June 2013

(3)

Declaration of Authorship

I, Ward SNOECK, declare that this thesis titled, Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements and the work presented in it are my own. I confirm that:

This work was done wholly or mainly while in candidature for a Master degree at this University.

Where any part of this thesis has previously been submitted for a degree or any other qualification at this University or any other institution, this has been clearly stated.

Where I have consulted the published work of others, this is always clearly at- tributed.

Where I have quoted from the work of others, the source is always given. With the exception of such quotations, this thesis is entirely my own work.

I have acknowledged all main sources of help.

Where the thesis is based on work done by myself jointly with others, I have made clear exactly what was done by others and what I have contributed myself.

Signed:

Date:

i

(4)

Nicola Tesla

(5)

Abstract

In the coming years, the LV grid is expected to be updated to become a ’smart grid’.

More monitoring and control mechanisms will be implemented allowing a more flexible market and more efficient problem solving. This thesis investigates the possibility to use Service Level Agreement contracts as a tool to actively and automatically monitor quality of service in the low voltage grid.

To this end, a thorough study of the already existing automated SLA monitoring sys- tems, designed for telecommunication and e-business, is made, finding inspiration on how it should be implemented specifically for the LV grid. It is concluded that in theory, it should be possible to create an automated SLA monitoring platform for the LV grid.

Subsequently a conceptual framework is proposed for the automated SLA monitoring platform.

In a next step of development, the proposed framework is implemented in a Java-code agent based programming environment (JACK). The program is surrounded by XML and txt inputs and outputs, increasing its flexibility to communicate with other systems.

Finally, the programmed automated SLA monitoring platform is put to the test and several proposals are highlighted on how the program might be used.

A second topic in this master thesis is the analysis of the Grid4EU demo 2 project in the Uppsala region (about 80 km north of Stockholm). This project envisions a smart grid, constructed with elements provided by several partners in the Grid4EU consortium. To present the smart grid architecture in a clear, universally understandable and standardized way, the SGAM model is used to represent the Demo 2 smart grid architecture.

iii

(6)

I would like to express my sincere appreciation to everybody who supported me in any way to complete this master thesis.

On a first note I would like to thank Prof. Dr. Lars Nordstr¨ om for the useful advice, remarks and engagement through the completion of this thesis.

Also many thanks to my supervisor, Shahid Hussain for the daily commitment, advice and many discussions that really brought this thesis from a theoretical and conceptual idea to a practical realisation and demonstration. Further also many thanks to the ICS department at KTH.

Thirdly, Ulf Ysberg and Anders Kim Johansson from Vattenfall provided me with the tools to complete large parts of my thesis. This has strongly improved the quality and practical use of the thesis. Many thanks for this amazing opportunity to work together with Vattenfall.

On a fourth note, my gratitude goes to the teaching staff at both institutes where I pursued my dual degree, KTH Stockholm and KU Leuven. In particular Prof.Dr.Ir.

Herbert De Gersem, the KULAK teachers, Prof. Dr. Ir. Johan Driesen and the KUL teachers, Hossein Sharokni and Prof. Dr. Nils Brandt and the KTH teachers.

My classmates and fellow students at both KU Leuven and KTH Stockholm deserve a special mention for all the good times we shared during the past few years. Also much thanks to Eva Daels, who kept on believing and supporting me.

Last but not least, I owe a lot to my parents for making my whole engineering education possible.

iv

(7)

List of Figures

1 Research Methodology . . . . 3

1.1 Grid4EU . . . . 5

2.1 Cost of Power Quality . . . 10

2.2 Power Quality from customer point of view . . . 11

2.3 Power Quality from utility point of view . . . 12

2.4 Impulsive Transient . . . 13

2.5 Oscillatory Transient . . . 14

2.6 SLA@SOI basic components . . . 23

2.7 SLA@SOI top level view . . . 25

2.8 Multi Level SLA framework . . . 27

2.9 SLA Ontology framework . . . 28

2.10 SLAng Reference Architecture . . . 30

2.11 RBSLM Architecture . . . 31

2.12 WSLA Architecture . . . 34

2.13 Smart Grid Plane . . . 35

2.14 SGAM Layers . . . 35

2.15 Use Case Diagram . . . 39

2.16 Business Layer . . . 40

2.17 Component Layer . . . 42

2.18 Information Layer . . . 43

2.19 Communication Layer . . . 44

3.1 Conceptual Architecture of the SLA monitoring platform . . . 47

3.2 Sequence Diagram of the SLA platform . . . 52

3.3 Event based implementation of the SLA monitor in JACK . . . 56

4.1 Design view Load Template . . . 63

4.2 Design view Import Data . . . 65

4.3 Design view Data Acquisition . . . 67

4.4 Design view Comparator . . . 69

4.5 Design view SLA Action . . . 70

4.6 Design view PQ Index Calculations . . . 72

5.1 Real time monitoring: Output of SLA1 . . . 75

5.2 Real time monitoring: Output of SLA2 . . . 75

5.3 Real time monitoring: Output of SLA3 . . . 76

5.4 Real time monitoring: Output of SLA4 . . . 76

ix

(12)

5.5 Real time monitoring: Output of SLA5 . . . 77

5.6 Testing average values of SLA 1 . . . 79

5.7 Testing SLA 1, Power Quality . . . 79

5.8 Testing average values of SLA 2 . . . 80

5.9 Testing SLA 2, Power Quality . . . 80

5.10 Testing average values of SLA 3 . . . 81

5.11 Testing SLA 3, Power Quality . . . 81

5.12 Testing average values of SLA 4 . . . 82

5.13 Testing SLA 4, Power Quality . . . 82

5.14 Testing average values of SLA 5 . . . 83

5.15 Testing SLA 5, Power Quality . . . 83

5.16 Power Factor Compensation . . . 85

(13)

List of Tables

5.1 Determining the most appropriate SLA . . . 78

xi

(14)

AC Alternating Current

AMM Advanced Metering Management AMR Automatic Meter Reading

ASAI Average Service Availability Index ASUI Average Service Unavailability Index

CAIDI Customer Average Interruption Duration Index CIM Common Information Model

CIS Customer Information System DC Direct Current

DER Distributed Energy Resources DG Distributed Generation DMS Data Management System GIS Geographic Information System KPI Key Performance Parameter

LV Low Voltage

MDM Metering Data Management

MDMS Metering Data Management System NIS Network Information System

QoS Quality of Service

RES Renewable Eenergy Source RTU Receiver Transmitter Unit

SAIDI System Average Interruption Duration Index SAIFI System Average Interruption Frequency Index SGAM Smart Grid Architecture Model

SLA Service Level Agreement

xii

(15)

Abbreviations xiii

TSO Transmission System Operator UI User Interface

VPN Virtual Private Network

XML eXtensible Markup Language

(16)

Research Background

The Industrial Information and Control Systems (ICS) group at the KTH electrical engineering department does research and teach within the field of Computer Applica- tions in Power System Management. The purpose of this research and education is to develop methods and tools for designing and developing smart applications for power systems. One identified area is monitoring the smart meter and provide information to all stakeholders in a reasonable/pre-negotiated time frame as stated in a service level agreement. The idea behind SLA based monitoring is a purposeful monitoring based on stated stakeholders concerns (QoS values).

This theoretical research and practical development of this thesis assignment is per- formed at ICS. The assignment is divided in two main parts: the analysis of the Grid4EU Demo2 project of Vattenfall Distribution in Uppsala, and the development of an auto- mated SLA monitoring platform, as a smart grid application for the LV grid.

Research Question

The main research questions in this thesis are the following:

Is it possible to use Service Level Agreements as a tool to actively and automatically monitor quality of service in the low voltage grid?

How should a service level agreement platform for the LV grid work and how can we apply such a platform?

Is it possible to use the SGAM Model to represent the GRID4EU Demo 2 smart grid project in a clear and standardised way?

1

(17)

Chapter 2 Research Methodology 2

Quantitative vs. Qualitative research methods

To answer the research questions, it is determined whether a quantitative or a qualitative approach would be most favourable.

Qualitative research is used when one does not really know what to expect, to define a problem or develop an approach to the problem. This kind of research is clearly valuable to this research assignment. The research question is very open. We will be able to answer the question in a qualitative way: yes or no. To this end, the literature review will be based on qualitative ideas and inspiration sources.

Quantitative research is applicable when one really wants an answer to the research question, based on hard numbers. The quantitative part of this research project is performed after the qualitative research question has been answered. The quantitative research will provide us with an answer on ”how” the monitoring would be possible instead of just a ”yes” or ”no”.

Aims

The thesis’ aims are:

• Provide a clear understanding of what Power Quality and Quality of Service in the LV grid means.

• Provide a clear understanding of what SLAs are

• Show that inspiration for the energy sector SLAs can be found with SLAs in other domains

• Prove that SLAs can be used as a tool to monitor quality of service in the LV grid

• Provide information on how to achieve the previous

• Build a working test implementation of the system

• Prove that the implementation works

• Provide a thorough understanding of the GRID4EU demo 2 smart grid project

Applied Methodology

(18)

On a first note, a background study is performed, throwing a light on power quality.

A literature review is performed on SLAs, existing SLA monitoring platforms for web services and telecommunication. Also the Grid4EU Demo2 project is highlighted with a focus on the Smart Grid Architecture Model (SGAM).

Secondly, the SLA monitoring platform for the electricity grid is designed. If the de- signing step delivers a positive output, the qualitative answer to the research question is obtained.

To provide a quantitative answer to the research question, the third step is the develop- ment of the SLA monitoring platform in the multi agent Java programming environment JACK. A schematic representation of the thesis’ initial set-up is found figure 1.

Figure 1: Conceptual scheme of research methodology

To conclude, the platform is put to the test. Several examples are analysed. It is demon-

strated how the platform can be applied for post analysis and for real time analysis.

(19)

Chapter 1

Introduction

The traditional electricity grid, as it was constructed throughout the 20th century, re- quires a profound update. As the energy markets opened up for liberalised incentives the amount of distributed generation (DG) has strongly expanded. This nowadays leads more and more to an increased stress on the grid, reaching unacceptable levels. Simply expanding the grid as it is constructed now, would be highly inefficient. Due to the nature of the changes, the grid needs to be partially reinvented. Grid intelligence and communication is required for grid operation with the implementation of renewable and distributed sources. The market liberalisation also enabled an expansion of the number of actors in the grid, as every individual is now allowed to enter the energy market and be an active player. The Smart Grid is needed in order to meet the future requirements of the transforming energy sector.

Smart grids are already widely examined and accepted to be the next step in power grid development. The interpretation of the therm smart grid is still very wide. The first prototypes of smart grids are already emerging from research into practice. The research results of this thesis will be made available to the smart grid pilot project of Grid4EU in Uppsala, Sweden (Grid4EU, 2013).

4

(20)

Figure 1.1: The Grid4EU project in Uppsala

1.1 Why do we need smart grids?

1.1.1 Distributed Generation

With the liberalisation of the electricity sector, the market has opened up for private initiative. Everybody can decide to be an energy producer, consumer or both. En- ergy production is therefore moving from very dense centralised generation to more widespread, less dense generation. For the first time, very small or micro power plants enter the market. The current grid is not designed to cope with this kind of generation topology. (R.Belmans et al., 2010)

With all these new initiatives, a demand for coordination and monitoring in the grid is more imminent than ever. A more complex generation system needs a more intelligent

’smart’ communication system. The smart grid provides this.

1.1.2 Renewable Generation Technologies

Many of the DG units being implemented in the modern energy system are renewable

energy technologies. Although their implementation is still very limited, renewable en-

ergy sources (RES) such as wind and photovoltaics are already common sights in the

European energy system. Being environmentally friendly these RES have strong advan-

tages. They help in the reduction of the CO

2

footprint and give us the advantage to

increase self subsistence in energy. They play an important role in the possibility of a

future fossil fuel free energy system.

(21)

Introduction 6

However there are some drawbacks on the implementation of RES in the grid. For instance photovoltaic cells produce DC. The grid frequency is standardised on 50 Hz AC (or 60 Hz for instance in the USA). Also wind turbines do not always run at the same rotational speed, so the frequency generated in the nacelle will differ from time to time. To inject the power in the electrical grid, power electronics need to bring the power to the right frequency and voltage. This can cause harmonic distortion in the grid. On top of that, we do not choose when the wind blows and the sun shines. On a sunny and windy day, it could occur that too much energy is injected in the grid, or too little. If consumption cannot follow generation at that point, grid balancing problems might occur. Consequentially problems in voltage and frequency might manifest with the possibility of equipment failure or total blackout. The smart grid can detect these problems and by analysing the acquired data, effective actions and measures can be taken.

1.1.3 Prosumers and Bidirectional Power Flow

Generated electricity used to flow from a huge central generation point to thousands small end points. That time is coming to an end. With the upcoming of DG, a new type of actor emerged in the electricity market: the prosumer. The prosumer will both consume and produce energy. The power flow will be bidirectional. The traditional grid is not adapted to this kind of energy flows. Extra metering and monitoring needs to be implemented in order to map the power flows in this new kind of grid. The smart grid will have possibilities to perform such measuring and monitoring. The improved possibilities for interaction between customer and service provider could lead to opportunities to create new business cases and offer different kinds of contracts between customer and service providers.

1.1.4 Security of Supply

Undelivered energy is many times more expensive than the actual cost of the energy, in terms of economic damage. It is of the utmost importance that energy supply is secured.

Black outs are unacceptable for industry and for instance public services like hospitals.

Even if back up power is installed, this is a financial setback, as back up power usually

(22)

is expensive. The smart grid needs to provide sufficient security of supply, even while large numbers of DG and power electronic devices are included in the energy system.

1.1.5 Power Quality Problems

Voltage sags, Voltage rises, underfrequency, overfrequency, harmonic distortion, outages, interruptions and black outs are power quality problems closely associated with the quality of service (QoS). The better the quality of service in the energy system is, the more reliable and the more attractive this grid becomes for investors, companies and factories. The smart grid will improve the QoS and will enable new business models to use Power Quality and QoS as a tool, as will be performed in this thesis. For energy utilities it is a necessary evolution to improve the power quality in order not to lose customers and facilitate industrial growth, as it is not unlikely for a voltage drop to shut down a production line that needs several hours to restart, costing the customer big amounts of money. Also the loads are becoming increasingly sensitive to power quality problems, as computers are now omnipresent. Those IT devices require a high level of power quality and service trustworthiness.

1.2 Business opportunities in the smart grid - Using auto- mated SLA monitoring

Smart grids will include a large amounts of data exchange. This data, measurements from various places in the grid, will eventually enable the utilities to monitor everything that happens on a real time basis (or to start with, on a daily, hourly or quarterly basis).

This opens a lot of new opportunities in the electricity market, as actors are now much more flexible and dynamic, through this real time data, and can now take actions more accurately, effectively and swiftly, improving the energy services.

The proposed tool in this thesis to enforce these more dynamic market models are Ser-

vice Level Agreements (SLAs). With all the data available, SLAs are very can be very

precisely managed and monitored, leaving a very responsible and trustworthy relation-

ship between utility and customer, with lots of new possibilities for their interaction.

(23)

Introduction 8

SLAs have been explored in several other sectors, such as web services and telecommu-

nications (Cattaneo, 2009). For the energy system, the concept is rather unexplored, or

at least not implemented in a very extensive way like in the other sectors where SLAs

exist. These SLAs in the energy system cannot be monitored very thoroughly. This

thesis proposes a monitoring platform for SLAs between actors in the energy system.

(24)

Literature Review

2.1 Power Quality

To understand the services in the energy sector, comprehensive knowledge of power quality in the electricity grid is a prerequisite. Therefore, the basics of power quality are explained. Generally, one is able to say that a certain level of power quality is a service delivered by the energy utility, the service provider, to the customer. It indicates how trustworthy energy supply is. It also reflects the actual quality of the energy provided.

The total economic loss of bad power quality in the EU is estimated at about 150 billion Euro (1250 billion SEK) annually, mainly in the industrial sector (Targosz, 2007). Figure 2.1 also shows the biggest economic damage is inflicted by short interruptions, dips, transients and surges.

To explain the aspects of power quality, the book Electrical Power Systems Quality of Roger C. Dugan is used as the main reference (Dugan et al., 2003).

2.1.1 Definition

Power Quality may be defined in many different ways and from many different perspec- tives. For every actor in the electricity grid, another definition might apply. For instance a Transmission System Operator (TSO) will link power quality directly to reliability, grid stability and balance, being able to show how reliably the system is transmitting the

9

(25)

Chapter 3. Literature Review 10

Figure 2.1: The annual cost of power quality in the EU ( Targosz, 2007).

energy ordered by the customer. A power grid equipment manufacturer will have an- other view on power quality. The manufacturer will asses the quality of power in the grid with a view on how well the power system enables the equipment to work properly.

The book of Dugan argues that electricity is still a consumer driven issue. This point of view seems very interesting for the purpose of this thesis work. As later on in this thesis we will look into Service Level Agreements (SLAs), involving power quality indices, it is clear that the best approach for power quality is the one in the contracts and agreements between utility and customer. The definition for a power quality problem from the book is a good start for the purposes of this thesis:

A power quality problem is any power problem manifested in voltage, current or frequency deviations that results in failure or incorrect operation of customer

equipment (Dugan et al., 2003)

This thesis revolves around the term ’power quality’ as it is agreed upon in an agreement between service provider and customer. We therefore expand the definition as given in the book, now also focussing on breaching the power quality agreements in the contract and incorrect operation of utility equipment:

A power quality problem, or QoS problem in the electricity grid, is any power problem

manifested in voltage, current or frequency deviations that results in a failure to meet

(26)

the agreements involving power quality as determined in the negotiated Service Level Agreement as enforced between service provider and customer in the electricity market,

which in turn, might lead to failure or incorrect operation of customer or utility equipment

2.1.2 Perceptual causes

It is crucial to understand that it is mainly the energy utility’s responsibility to secure an acceptable (and agreed upon) level of power quality. However it is often the case that power quality problems are caused by nature, the customers themselves or a neighbour- ing customer. Where the power quality problems come from, differs from the customers point of view (figure 2.2) to the utilities point of view (figure 2.3).

Figure 2.2: Power Quality from customer point of view ( Dugan et al., 2003)

From figures 2.2 and 2.3 we can clearly see that natural events by far cause most power

quality problems. Examples are trees falling on overhead lines or lighting strikes creating

transient voltage surges. Another important trend is that customer and utility tend to

blame each other for power quality problems. The customer perceives power quality

problems much more inflicted by utilities and the other way around. Because power

quality is not sufficiently monitored, the real cause is often hard to find. There is also an 8

percent chance of the neighbour inflicting the damage. These diagrams clearly show that

there is a need for more information and monitoring on where power quality problems

(27)

Chapter 3. Literature Review 12

Figure 2.3: Power Quality from utility point of view ( Dugan et al., 2003)

come from, in order to take the right action. Given the economic losses associated with power quality problems (150 billion Euro per year in the EU (Targosz, 2007)) there is a need for more information on who causes which damage. An automated SLA management platform (as will be developed in this thesis) delivers the possibilities to solve this problem and take action to solve the financial aspect of this information gap.

2.1.3 Types of Power Quality problems

Power quality is very closely related to voltage quality. Power quality problems are therefore mostly voltage problems. The electricity system is operated in a way that the voltages are controlled along the branches of the grid, resulting the HV,MV and LV grid.

How much current is flowing through differs from line to line and depends on the loads that are attached to that specific line. The more power the loads demand, the more current will flow through the line. Also the voltage waveform needs to be sinusoidal at 50Hz (or 60Hz in some countries like the US). In the end, it is always the magnitude and frequency of the voltage supply that needs to be preserved within the agreed limits.

Power quality problems can be divided in 2 main groups: transient phenomena and non

transient phenomena. Transients automatically disappear after a short time (typically a

few cycles), however during their occurrence they can inflict considerable damage. Non

transient phenomena don’t disappear quickly. Some even don’t automatically disappear

(28)

at all. Specific action should be taken in order to solve these kind of power quality problems.

2.1.3.1 Transients

Impulsive transients are changes in voltage and/or current that are unipolar (either negative or positive). Usually they exist of a very steep swell or sag in voltage/current and decay at a much smaller slope. Impulsive transients are therefore characterised by their magnitude and decay time. In the power system, they might be causing subsequent oscillatory transients. A common example of impulsive transients are lightning strikes (figure 2.4).

Figure 2.4: Impulsive transient caused by a lightning strike ( Dugan et al., 2003)

Oscillatory transients are changes in voltage and/or current that are bipolar (both

negative or positive, oscillating at a certain frequency). They are categorised in high

frequency (>500 KHz), mainly caused by an impulsive transient, medium frequency (5-

500 KHz), caused by for example capacitor or cable switching, and low frequency (<5

KHz), mainly caused by capacitor bank energisation. In the distribution grid oscillatory

transients at less than 300 Hz are also found because of transformer energisation and

ferroresonance. An example of an oscillatory transient can be found in figure 2.5.

(29)

Chapter 3. Literature Review 14

Figure 2.5: An oscillatory transient ( Dugan et al., 2003)

2.1.3.2 Non-transient phenomena

Long duration voltage variations last longer than 1 minute. Examples are overvolt- ages, undervoltages and sustained interruptions which means a complete loss of voltage.

Short duration voltage variations last shorter and are categorised in instantaneous (0,5 - 30 cycles), momentary (30 cycles - 3s) and temporary (3s - 1 min). Examples are interruptions, voltage sags and voltage swells.

Voltage imbalance occurs when the 3 phases of the voltage do not have the same vector length anymore or their mutual phase difference does not equal 2π/3.

Waveform distortion is the spectral content of a deviation from the ideal sine wave.

The types of waveform distortion are harmonics, interharmonics, DC offsets, notches and noise.

Voltage fluctuations often occur in a random way. It causes flicker which can be experienced by the human eye in for example lighting.

Power frequency variations are disturbances where the fundamental frequency in the grid deviates from 50 (or 60) Hz.

2.1.4 Power Quality Indices

To quantify power quality, power quality indices have been developed. These indices

identify different aspects of reliability in the electricity grid. The most common and

(30)

general ones are System Average Interruption Duration Index (SAIDI) and Sys- tem Average Interruption Frequency Index (SAIFI). SAIDI gives an idea on how long an average customer can expect to be interrupted and SAIFI how many times N

i

is the customer with index i. λ

_i

is the number of interruptions sustained by customer i.

U

i

is the total outage duration of customer i.

SAIF I = Σλ

i

N

i

ΣN

_i

(2.1)

SAIDI = ΣU

_i

N

_i

ΣN

i

(2.2)

There are also more specific indices available, most of them to be calculated from SAIDI and SAIFI. Some examples are included here. The Customer Average Interrup- tion Duration Index (CAIDI) how long the average interruption will last. Average Service Availability Index (ASAI) and Average Service Unavailability Index (ASUI) indicate how long service will be (un)available on a yearly basis, in hours.

CAIDI = ΣU

_i

N

_i

Σλ

i

N

i

(2.3)

ASAI = Σ8760N

_i

− ΣU

_i

N

_i

Σ8760N

_i

(2.4)

ASU I = ΣU

i

N

i

Σ8760N

_i

(2.5)

2.1.5 How to monitor Power or Voltage Quality?

It used to be expensive to perform extensive voltage monitoring but the amount of

monitors in operation has significantly increased over the past few years. This mainly

due to the decreased cost of monitoring equipment (M. Bollen, 2013). Also the cost of

communication services, data storage, data processing and visualisation has dropped,

whilst their performance has significantly increased. M.Bollen et al recommend that the

results attained from the voltage quality monitoring programs are used for tackling the

(31)

Chapter 3. Literature Review 16

new challenges in the energy system, for instance the implementation of a higher level of Distributed Generation (DG).

How much monitors do we need and where should we put them? M. Bollen suggests the following allocation of monitoring resources:

• In the (E)HV networks it is sufficient to monitor at the connections between all EHV and HV , the EHV/MV and the HV/MV substations. Also monitoring is advised at customers and producers.

• In the MV networks it is recommended to monitor at the secondary sides of the HV/MV transformers and at the connection points of MV customers.

• In the LV networks it is recommended to measure at a selection of random cus- tomer connection points in order to obtain a statistically relevant sample. Smart meters will play a significant role in the future too. Note that for the purposes of Service Level Agreements, close monitoring at customers side is a requirement and a statistically significant portion of monitoring might not be trustworthy enough!

For grid measurements, it is considered good practice to measure the phase-to-phase voltage except in the LV grid: there the phase-to-neutral should be monitored in order to evaluate the voltage quality. Publishing of the results should be performed in a unified manner. All data should also be made available for all actors in the grid, and the use of internet is encouraged for the publication of this data.

2.1.6 Who should finance power quality monitoring?

Its clear that practically every stakeholder in the electricity market economically benefits

from a good power quality. For example the DSO delivers better a better service, or a

factory suffers less damage by outages of a production line. Consequently its is fair for

all stakeholders to contribute to the investment in better power quality. M. Bollen et

al (M. Bollen, 2013) suggest the financing of power quality monitoring should include

two main steps: the cost assessment and the financing plan. In the cost assessment

it is important the entire cost is considered and documented in an inventory. Both

investment and maintenance cost should be included in the cost assessment. To finance

(32)

the power quality monitoring program, it is possible to find funding through grid tariffs.

This is common practice now. However not every customer expects the same power quality and therefore not every customer contribute the same amount to power quality monitoring programs. Also, for a customer situated in a rural area it is more expensive do grid investments, yet this rural customer might have to pay the same grid tariffs.

With the use of SLAs as a tool to perform power quality monitoring, this problem of

’unfairness’ can be solved.

2.1.7 European Quality of Supply Standard: EN50160

The European Norm 50160 is an implemented standard on power quality, enforced in the European Union. This standard is one of the pioneering incentives in power quality regulation. The standard clearly has a lot of benefits, however nowadays the original implementation of the standard is rather outdated and shows several flaws. It has recently been improved.

2.1.7.1 Original shortcomings, upon enforcement of the standard

According to the paper of B.Kingham (Kingham, 2012) there are several flaws to be considered when analysing the EN50160 standard:

• Because the European Norm is commonly agreed upon by the members states of the European Union, it reflects the least ambitious plan for Power Quality limits and regulation. Local regulation plans in the most industrially developed countries within the European Union could be way stronger.

• For dips, swells and interruptions, there are only indicative values in the norm and no compliance limits.

• The norm only applies to the LV and MV networks, up to 35 kV.

• There are no measurement methods defined. Every operator can therefore freely

choose the way of measuring, which enables manipulation to a certain extent and

makes comparison between different operators and countries impossible.

(33)

Chapter 3. Literature Review 18

• Evaluation of the PQ is set on less than 100 percent of time, leaving freedom for power quality problems about 8 hours per week, regardless of severity of a possible event, or problem

2.1.7.2 Suggested improvements made by the European regulators

The B.Kingham paper (Kingham, 2012) states that the European regulators proposed some improvements in the ERGEG paper ”Towards Voltage Quality Regulation in Eu- rope”. Some of these recommendations have already been implemented recently, mainly in the field of voltage variations. The suggestions include the following:

• Definitions for power quality problems should be more comprehensive and unified in order to get a clear global understanding of the problem and comparable mea- surements can be consulted all around Europe. Measurement methods therefore need to be standardized as well.

• Indicative values should be avoided in the text of EN50160, but should be included in annexes.

• There should be binding limits for voltage variations that apply 100 percent of the time.

• The high voltage grid should be included in the norm and the concept of normal operating conditions should be clarified.

• Product standards should be reconsidered and clarified towards the customer when they buy electrical equipment to be attached to the grid.

2.1.7.3 Additional remark on the standardisation of Power Quality and the EN50160

A comprehensive European norm on power quality is of vital importance for the com-

petitiveness of our regions for the viability of our industry. A reliable energy system

is considered a prerequisite for industry when deciding to settle in region. If there is

no adequate quality of supply, industry will decide to move to a place where it a good

energy system is guaranteed.

(34)

In the smart grid, there is a possibility to monitor this norm. With a framework around service level agreements, each individual requirement for QoS agreed between the cus- tomer and the utility can be monitored, even if these requirements are stricter than the European norm.

2.2 Service Level Agreements

The aim for this thesis is to explore the possibilities to implement SLAs in smart grid operation and design a platform to monitor the QoS and manage SLAs. To this end, it is firstly defined what an actual SLA is, and where this concept is already applied in the format with automatic monitoring. It is explained why SLAs are beneficial for use in the future energy market equipped with a smart grid. To conclude, related work on the design, both conceptual and implemented, SLA monitoring platforms is discussed.

2.2.1 Defining Service Level Agreements

In many commercial service activities the services do not always meet the expectations.

Either high expectations are not met due to a low service quality, or low expectations are overwhelmed by (too) high service quality (Cattaneo, 2009). Both phenomena are less efficient than the service delivery that realistically meets the requirements of the customer. SLAs are good tools in order to establish such an optimal service delivery to expectation relationship. It enables service providers to deliver what they promise, and if not, action can be taken, for instance in the form of financial compensation, penalties or rewards.

”An SLA is a formal agreement between the service providers and the customers in the

context of service provisioning” (Hussain et al., 2012) The SLA is the part of a service

contract where the actual service, or level of service, is determined and agreed upon. So

both the customer and the service provider exactly know what to expect and how much

effort and financial means to spend on the service. The higher the quality, reliability and

availability of the service, the more expensive the contract will be and the more effort

the service provider will invest in it. Obviously, these kinds of contracts and agreements

are typically negotiable.

(35)

Chapter 3. Literature Review 20

Other sources define SLAs more elaborately as follows:

An SLA is an agreement between a service provider and a customer. The SLA describes the service, documents the service level targets and specifies the responsibilities of the service provider and the customer. A single SLA may cover multiple services or multiple customers. (SLA@SOI Consortium, 2011)

An SLA is an arrangement between a customer and a provider, describing technical and non-technical characteristics of a service, including QoS requirements and related set of metrics with which provision of these requirements is measured. (Lamanna et al., 2003) G. Dobson et al states carefulness should be adopted when using the terms SLA and QoS. Both are not the same thing although confusion sometimes may occur. (Dobson and Sanchez-Macian, 2006). The SLA is contains statements and rule sets concerning QoS.

D. Lamanna et al also point out the existence of horizontal SLAs, governing the in- teraction between coordinated peers, for example two competing companies using the same infrastructure. Vertical SLAs govern the interaction between subordinated pairs, one company delivering a service to another (Lamanna et al., 2003). In the smart grid mainly vertical SLAs are appropriate.

An SLA typically includes a sign off page where the contract is officially enforced. Also a service description and the scope of the agreement are declared. Other aspects like service hours, reliability, customer support, service performance and security are often included in the SLA (Cattaneo, 2009). Of course there are many more aspects, depending on the sector where the SLA is active.

SLAs often emerge from negotiations. The SLA will determine how stakeholders act towards each other and interact with each other. An SLA, enforced between the stake- holders will be the platform for this interoperability. An SLA sets up the coordination among stakeholders in a business case and relevant services with set points and agree- ments to be monitored (M.Comuzzi et al., 2009).

The SLA will often involve and be based on indices that represent the quality of service delivered. This entails an impartial tool to monitor how the service contract is respected.

It also gives the opportunity to indicate whether sanctions are justified if one of the

(36)

involved actors fails to comply with the SLA. This way SLAs can contribute to the trustworthiness of systems and agreements.

In short: SLAs are agreements between service providers and customers that enable the service providers to set SMART -targets, where SMART stands for Specific, Mea- surable, Achievable, Realistic and Timely. With SLAs, better service can be provided.

(Cattaneo, 2009)

2.2.2 Use of SLAs today

SLAs are common and already widely used in the several sectors, more specifically in telecommunications and e-Business. In the these sectors, QoS needs to be monitored and violations of the agreement need to be found, instantiating actions to solve the problems. They will serve as examples for our own electricity grid related purposes.

Since the deregulation of the telecommunications market, SLAs have become more and more of interest. Nowadays, SLA management and monitoring has been developed and is up and running. Examples of tools already deployed by IT service providers to perform SLA monitoring are CISCO Total Service Management, ViewGate Networks’ Inteligo and Oblicore Guarantee (Kosinski et al., 2008). It is crucial that SLA management tools operate with real time measurements and do not give a mere historical overview.

(However, the SLA monitoring platform designed in this thesis will also be able to run on historical data.)

The results obtained by the telecommunications industry are valuable for application in

smart grids. The deregulation of the energy sector shows large parallels to the dereg-

ulation of the telecommunications sector. That the telecommunications market moved

towards a service-based market, indicates that the energy market might implement such

structures as well. There still are some differences to consider and solutions need to be

found for this. In the smart grid, different groups of actors (more than 2 actors) will have

to work together to enforce the SLAs. Also SLAs in the smart grid will have different

sets of real time constraints. Measurements might not come in real ’real time’ format

but rather once every hour or 15 minutes. These sets of constraints are a challenge to

be tackled (Hussain et al., 2012), (Kosinski et al., 2008).

(37)

Chapter 3. Literature Review 22 2.2.3 Related Work: existing SLA monitoring architectures

To find inspiration for an automated SLA monitoring platform for power systems, several (generic) approaches in other sectors are studied. First SLA@SOI is discussed as this is a completely worked out generic platform for SLA management and will prove to be valuable. The other frameworks looked into are multi level SLA Management, SLA ontologies, SLAng, rule based SLAs and Web SLAs. These frameworks are sometimes already very elaborately developed, which means that in the next sections only focus on the basic principles.

2.2.3.1 SLA@SOI

The SLA@SOI framework is a generic and operational SLA management framework (SLA@SOI Consortium, 2011). The goal of the project is to provide a business ready service oriented infrastructure empowering the service economy in a flexible and de- pendable way. This economy should be based on trustworthy service delivery and clear negotiated agreements, transparent SLA management and more automation and au- tomated monitoring of service provision. The framework is mainly focussed on the initiation and enforcement of the SLA. Monitoring and evaluation of the service is in- cluded, but not in a very prominent manner (which, in power systems, it should be).

The framework treats the sensors and effectors in the system as agents. This indicates that the SLA@SOI framework is valuable to the purposes of this thesis.

SLA@SOI proposes a somewhat hierarchical management structure. The main archi- tecture consists of 5 components: the Business Manager, the SLA Manager, the Service Manager, the Manageability Agent and the Service Evaluation. A schematic represen- tation can be found in figure 2.6.

The business manager is the main, and most powerful component in the framework.

This component steers the whole working of the SLA management tool to attain the business goals and keep on track with customer relations. The actual functions of the business manager are:

• Customer and service providers interaction management.

• Searching and publishing products when communicating with customers.

(38)

Figure 2.6: SLA@SOI basic components ( SLA@SOI Consortium, 2011)

• Negotiation and establishment of agreements or contracts with customers.

• Notification of bills and penalties to customers and service providers.

Directly subjected to the business manager, the SLA manager is assigned to manage a set of SLA templates and also the SLAs that are running. Different aspects of the services in which the SLA is active may require different SLA managers. Thus, if an agreement between a customer and the business manager is enforced, it may consist of several different aspects which require a separate administrative management systems.

SLA@SOI identifies 3 SLA managers: business, software and infrastructure. This clearly divides the management in submanagers, each managing a different aspect related to the SLA between customer and provider.

The actual responsibilities of the SLA manager consist of the following:

• searching and publishing of SLA templates

• negotiation of specific SLA content and rule sets with customers and 3rd parties including conversion between different SLA formats

• SLA planning and optimisation

(39)

Chapter 3. Literature Review 24

• SLA provisioning and adjustment

The Service Manager manages the equipment/elements needed to instantiate a ser- vice. Different technical domains each have a different service manager, managing all the equipment needed for a service to be delivered. The tasks of the service managers include:

• publishing of service implementations

• maintenance of a service landscape

• reservation and booking of service instances

• mediation of management or adjustment operations to service instances and man- ageability agents

• triggering of actual service provisioning

Next to the service manager in the SLA@SOI hierarchy there is the service evaluation.

This component relies on information about the service quality characteristics, coming from several sources (run-time, design-time or historical). In the SLA@SOI framework there is only one service evaluation per SLA manager.

The manageability agent is the connection of the SLA management platform to the actual infrastructure and resources. It is the lowest ladder in the SLA management hierarchy. Here, information on actual sensors and effectors used for providing services is located. Each sensor could be a separate agent, but several can be collected in one agent. The specific tasks of the manageability agent are:

• sensing and monitoring the status of service instances and resources

• searching for and executing manageability actions

Figure 2.7 shows the interaction of a specific SLA@SOI monitoring platform. In the

reference document a full explanation with sequence diagrams and function clarification

can be found (SLA@SOI Consortium, 2011).

(40)

Figure 2.7: SLA@SOI top level view ( SLA@SOI Consortium, 2011)

2.2.3.2 Multi Level SLA Management

M. Comuzzi et al present a paper on multilevel SLA management approach (Comuzzi et al., 2010). Stated that current SLAs typically are typically oriented at customer side, for the service provider it is in many cases impossible to technically monitor service performance. Also in the power system, customer side monitoring equipment is a pre- requisite. In order to align the SLAs with measurements from the monitoring equipment, SLAs need to be translated to metrics and parameters.

The work in Comuzzi’s paper has been used for the completion of the SLA@SOI man-

agement framework. The proposed approach serves exactly the same aim as SLA@SOI.

(41)

Chapter 3. Literature Review 26

The architecture of the proposed SLA management framework can be found in figure 2.8 and consists of 6 modules:

• Negotiation is performed by the negotiation module. The SLA will be con- structed out of a repository, which predicts a model based on performance re- quirements and an SLA template registry.

• The provisioning module stores all the established SLAs. Also the software and the service schedules are kept here.

• The monitoring module decomposes the SLAs into separate, measurable rules or rule sets. Measured data will enter here and will be compared to the SLA rules, determining whether the SLA is breached and how severely the breaching was.

• The adjustment module is notified of SLA violations and will instantiate the correct action to counter the violation. The decisions on how to counter the violations are based on adjustment patterns and a manageability model.

• The e-contracting module manages all business related aspects, like interaction with customers and making business offers.

• The infrastructure module will manage all technical equipment and physical resources. To connect to these physical resources, a cloud-like data transmission is used.

2.2.3.3 SLA Ontologies

The need for a unified conceptual scheme for SLA architectures is expressed by Dob- son and Sanchez-Macian (Dobson and Sanchez-Macian, 2006). This paper identifies 3 viewpoints on QoS: technical, experienced by user and the quality of business. There is a need to define different kinds of rules to map and translate these 3 different quality levels to get a clear overview on the performance of the SLA. According to these rules, monitoring and QoS adaption is possible.

A different, already advanced architecture is proposed in the SLA Ontology approach

(Fakhfakh et al., 2008). This paper states that there are many different ways (languages)

to implement SLA specifications. However, it remains difficult to perform contract

(42)

Figure 2.8: Multi Level SLA framework ( Comuzzi et al., 2010)

monitoring and detect violations. There is also a problem of differences in knowledge, expertise and the used SLA language between customer and provider. More elaborative and more structured descriptions of SLA are needed to provide an architecture (more than just a language) to implement monitoring possibilities.

The paper of Fakhfakh et al contributes to the lack of structured architecture with the establishment of an ontology based SLA model. The different parts of the SLA Ontology framework can be found in figure 2.9. Due to the complexity of the model, only the basic parts are explained here.

The model starts from the basic principle of SLAs and distinguishes 4 concepts defining the SLA: The involved parties (service provider, customer and 3rd parties), the definition of the service, agreed and required quality of service obligations and the application domain, describing the contract’s context.

In order to acquire measurable input for the QoS, the necessary variables are imple- mented in the SLAParameter concept. These SLAParameters are based on metric data aggregated from measurements and are used for defining the QoS in an algorithmic way.

The SLAParameters are also defined by a unit (m, s, W...).

(43)

Chapter 3. Literature Review 28

Figure 2.9: SLA Ontology framework ( Fakhfakh et al., 2008)

To determine the total view on QoS, the obligation concept accesses a function that

defines the Service Level Objectives out of the predicate function. The predicate func-

tion contains the raw rule sets for QoS, understandable for computers and electrical

equipment. Each predicate is expressed by a Semantic Web Rule Language (SWRL)

rule.

(44)

2.2.4 Related Work: existing SLA descriptive languages

In order to automate SLA monitoring, the content of the SLA negotiated agreement needs to be described in a way that computers and machines can understand, interpret and use it. Most literature touches the Extensible Markup Language (XML) standard to describe SLAs. XML is a standard of the World Wide Web Consortium for descriptive languages by which it is possible to implement structured information in plain text. The extension is both readable for machines as humans. The languages studied in sections 2.2.4.1, 2.2.4.2, 2.2.4.3 are XML based languages or assume the SLAs are written in an XML format. It is considered wise to use XML as input for the monitoring platform, and therefore this will certainly be implemented in this thesis’ eventual outcome. Using XML will increase the platform’s flexibility and applicability to different systems.

2.2.4.1 SLAng

SLAng is an XML based language by which SLAs can be expressed. The SLAng syntax is based on the XML scheme. Figure 2.10 shows the very basic and generic reference architecture which forms the basis for the language development. Applications are clients that use the components and web services to deliver end user services. The web services can invoke components, which represent the underlying resources (measurement data etc...). Containers provide the components with management of the underlying resources services communication, persistence, transactions and security. The containers need support for QoS negotiation, establishment and monitoring. The underlying resources also include network and storage providers.

The SLA definition language (SLAng) (Lamanna et al., 2003) was developed with regards to the following requirements for an SLA language:

• Parametrisation: each SLA is built around a set of values (parameters) that describe the service. An SLA is also specifically and qualitatively characterized by the set of parameters describing the service.

• Compositionality: The SLA needs to accommodate cooperation between differ-

ent domain entities. Several different inputs from a domain need to be able to

work together to deliver services.

(45)

Chapter 3. Literature Review 30

Figure 2.10: SLAng Reference Architecture ( Lamanna et al., 2003)

• Valitation: Before an SLA is operational it should be validated by all parties involved. It should also be approved in syntax and consistency. In other words:

the traditionally signing of the contract is unchanged.

• Monitoring: All involved parties should be able to check whether the SLA is respected or breached. Therefore a logging system should be included

• Enforcement: The SLA should be enforced in an automated system with the implementation of an SLA platform using network routers, database management systems, middleware and web servers.

2.2.4.2 Rule Based SLA

Rule Based SLA (RBSLA) is an XML based language (RuleML) language to describe

SLAs. The syntax is machine readable and can be fed into a rule engine with the purpose

of monitoring the contract performance. This happens automatically from the instance

the contract is running. Also execution of contractual rules is performed automatically

(Paschke, 2005). The motivation for developing this language, similarly to SLAng, is

the need for a formal, machine readable and interchangeable way to exchange contract

information between business partners and organisations. The language should allow a

(46)

high degree of freedom and flexibility and not constrain the possibilities of interaction between organisations. This is the main reason why no imperative programming lan- guages (Java, C++, Visual Basic) are not favourable, but declarative rule languages, like RuleML is.

The basic building blocks of RuleML are extended with a specific RBSLA language set of building blocks in order to describe SLAs. A more extensive explanation can be found in the paper of Paschke (Paschke, 2005).

Papers (Paschke et al., 2005) and (Paschke, 2006) show an architecture of a rule based service level management tool, based on RBSLA (fig 2.11).

Figure 2.11: RBSLM Architecture ( Paschke et al., 2005)

The Mandarax rule engine executes the formalized contract rules, which are represented

on the basis of the contract log framework. RBSLA description of the contract is pro-

vided for machine readability, re-usability, rule interchange, serialisation, tool based

editing and verification. RBSLA is transformed into executable contract log rules with

a mapping. The GUI enables writing, editing and maintaining the SLAs, stored in the

contract base. The repository contains SLA templates for the construction or adapta-

tion of SLAs. A big advantage of this architecture is that it specifically included existing

(47)

Chapter 3. Literature Review 32

external tools and databases can be implemented. The service dash board shows the monitoring results and violations of the service contracts.

To conclude, this architecture proposal describes a declarative rule based approach in order to achieve SLA automated management instead of an approach based on classical procedural programming.

2.2.4.3 WSLA

The WSLA framework is constructed to specify and monitor SLAs specifically for web services. Just like SLAng and RBSLA, the WSLA language is based on the XML scheme.

A detailed description of the language can be found in (A. Keller, 2003).

(A. Keller, 2003). Figure 2.12 presents the WSLA architecture. The architecture is build around a generic life cycle for SLAs between customers and service providers. The SLA exists out of 5 stages, represented in the architecture:

• SLA negotiation and establishment is performed by the SLA establishment service tool. The tool allows the signatory parties establish prices, retrieving met- rics offered by the service provider, aggregate and combine them into SLA param- eters, define secondary parties and their tasks and construct the SLA document.

• SLA deployment includes the distribution of the parts of the SLA to the neces- sary components. It is possible that not the entire SLA is accessible for all involved parties, but only parts of it are. During the deployment the right parts will be sent to the right parties. It is important that all parties are able to interact and communicate. That’s why a standard language is proposed (in this case according to the Service Deployment Information format).

• Service Level Measurement and Reporting

– The Measurement service will monitor the system and how the SLA param-

eters are behaving during run time. It will retrieve data directly from the

managed resources. It is also possible for multiple measurement services to

monitor the same SLA parameters or metrics that are part of an SLA param-

eter.

(48)

– the Condition Evaluation Services compare the measured SLA parameters with the agreed SLA thresholds and specifications. It also notifies the man- agement system of its findings.

• Corrective Management Actions After violation of the SLA, corrective actions need to be instantiated. This functionality is provided by two services:

– If the Management Service receives a notification of a violation, it will search for the appropriate action to solve the problem, according to what is specified in the SLA. Before performing correcting actions, it consults the business entity to see whether the proposed action is allowable. When the business entities’ consent is received, action is performed in the management system.

– The conceptual Business Entity contains information about the business goals, business knowledge, policies of the included parties in an SLA. This information can remain invisible for the opposing parties (it can for instance contain information on which customer should be prioritised in service deliv- ery).

• SLA termination occurs when one or more parties decide to step out after

breaching of one or more SLA clauses, the termination date is specified, or a

renegotiation is in order.

Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements

Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements

WARD SNOECK

Master’s Degree Project Stockholm, Sweden 2013

XR-EE-ICS 2013:011

Master Thesis

Quality of Service Monitoring in the Low Voltage Grid by using Automated

Service Level Agreements

Author:

Ward SNOECK

Supervisor:

Shahid HUSSAIN Examinator:

Lars NORDSTR ¨ OM

A thesis submitted in fulfilment of the requirements for the degree of MSc. Energy Innovation and Smart Cities

in

Industrial Information Systems and Control Department of Electrical Engineering

June 2013

Declaration of Authorship

I, Ward SNOECK, declare that this thesis titled, Quality of Service Monitoring in the Low Voltage Grid by using Automated Service Level Agreements and the work presented in it are my own. I confirm that:

This work was done wholly or mainly while in candidature for a Master degree at this University.

Where any part of this thesis has previously been submitted for a degree or any other qualification at this University or any other institution, this has been clearly stated.

Where I have consulted the published work of others, this is always clearly at- tributed.

Where I have quoted from the work of others, the source is always given. With the exception of such quotations, this thesis is entirely my own work.

I have acknowledged all main sources of help.

Where the thesis is based on work done by myself jointly with others, I have made clear exactly what was done by others and what I have contributed myself.

Signed:

Date:

i

Nicola Tesla

Abstract

In the coming years, the LV grid is expected to be updated to become a ’smart grid’.

Subsequently a conceptual framework is proposed for the automated SLA monitoring platform.

In a next step of development, the proposed framework is implemented in a Java-code agent based programming environment (JACK). The program is surrounded by XML and txt inputs and outputs, increasing its flexibility to communicate with other systems.

Finally, the programmed automated SLA monitoring platform is put to the test and several proposals are highlighted on how the program might be used.

iii

I would like to express my sincere appreciation to everybody who supported me in any way to complete this master thesis.

On a first note I would like to thank Prof. Dr. Lars Nordstr¨ om for the useful advice, remarks and engagement through the completion of this thesis.

Also many thanks to my supervisor, Shahid Hussain for the daily commitment, advice and many discussions that really brought this thesis from a theoretical and conceptual idea to a practical realisation and demonstration. Further also many thanks to the ICS department at KTH.

Thirdly, Ulf Ysberg and Anders Kim Johansson from Vattenfall provided me with the tools to complete large parts of my thesis. This has strongly improved the quality and practical use of the thesis. Many thanks for this amazing opportunity to work together with Vattenfall.

On a fourth note, my gratitude goes to the teaching staff at both institutes where I pursued my dual degree, KTH Stockholm and KU Leuven. In particular Prof.Dr.Ir.

Herbert De Gersem, the KULAK teachers, Prof. Dr. Ir. Johan Driesen and the KUL teachers, Hossein Sharokni and Prof. Dr. Nils Brandt and the KTH teachers.

My classmates and fellow students at both KU Leuven and KTH Stockholm deserve a special mention for all the good times we shared during the past few years. Also much thanks to Eva Daels, who kept on believing and supporting me.

Last but not least, I owe a lot to my parents for making my whole engineering education possible.

iv

Contents

Declaration of Authorship i

Abstract iii

Acknowledgements iv

List of Figures ix

List of Tables xi

Abbreviations xii

1 Introduction 4

1.1 Why do we need smart grids? . . . . 5

1.1.1 Distributed Generation . . . . 5

1.1.2 Renewable Generation Technologies . . . . 5

1.1.3 Prosumers and Bidirectional Power Flow . . . . 6

1.1.4 Security of Supply . . . . 6

1.1.5 Power Quality Problems . . . . 7

1.2 Business opportunities in the smart grid - Using automated SLA monitoring 7 2 Literature Review 9 2.1 Power Quality . . . . 9

2.1.1 Definition . . . . 9

2.1.2 Perceptual causes . . . 11

2.1.3 Types of Power Quality problems . . . 12

2.1.3.1 Transients . . . 13

2.1.3.2 Non-transient phenomena . . . 14

2.1.4 Power Quality Indices . . . 14

2.1.5 How to monitor Power or Voltage Quality? . . . 15

2.1.6 Who should finance power quality monitoring? . . . 16

2.1.7 European Quality of Supply Standard: EN50160 . . . 17

2.1.7.1 Original shortcomings, upon enforcement of the standard 17 2.1.7.2 Suggested improvements made by the European regulators 18 2.1.7.3 Additional remark on the standardisation of Power Qual- ity and the EN50160 . . . 18

2.2 Service Level Agreements . . . 19

2.2.1 Defining Service Level Agreements . . . 19

v

2.2.2 Use of SLAs today . . . 21

2.2.3 Related Work: existing SLA monitoring architectures . . . 22

2.2.3.1 SLA@SOI . . . 22

2.2.3.2 Multi Level SLA Management . . . 25

2.2.3.3 SLA Ontologies . . . 26

2.2.4 Related Work: existing SLA descriptive languages . . . 29

2.2.4.1 SLAng . . . 29

2.2.4.2 Rule Based SLA . . . 30