thesis_LauraRamirezElizondo_library.

Optimal Usage of Multiple Energy
Carriers in Residential Systems
Unit Scheduling and Power Control
Laura M. Ramı́rez Elizondo
.
Optimal Usage of Multiple Energy
Carriers in Residential Systems
Unit Scheduling and Power Control
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus Prof.ir. K.C.A.M. Luyben,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen op dinsdag 26 maart, 2013 om 15:00 uur
door
Laura Marı́a RAMÍREZ ELIZONDO,
elektrotechnisch ingenieur,
geboren te San José, Costa Rica.
Dit proefschrift is goedgekeurd door de promotor:
Prof.ir. L. van der Sluis
Samenstelling promotiecommissie:
Rector Magnificus
Prof.ir. L. van der Sluis
Prof.dr. F. Munteanu
Prof.dr. C.M. Franck
Prof.dr.ir. B.J. Boersma
Prof.dr. C. Witteveen
Prof.ir. P.G. Luscuere
Dr.dr.h.c.ir. G.C. Paap
Voorzitter
Technische Universiteit Delft, promotor
Gheorghe Asachi Technical University of Iaşi
ETH Zürich
Technische Universiteit Delft
Technische Universiteit Delft
Technische Universiteit Delft
Technische Universiteit Delft, heeft als
begeleider in belangrijke mate aan de
totstandkoming van het proefschrift
bijgedragen.
Dit onderzoek werd uitgevoerd in het kader van het onderzoekprogramma “Innovatiegerichte Onderzoekprogamma’s - Elektromagnetische Vermogenstechniek” (IOP-EMVT),
dat financieel wordt ondersteund door SenterNovem, een agentschap van het Nederlandse
Ministerie van Economische Zaken.
Published and distributed by: Laura M. Ramı́rez Elizondo
E-mail: [email protected]
ISBN 978-94-6203-295-8
Cover design: Carolina Ramı́rez Elizondo
All illustrations included in this thesis: Oscar Calderón
c 2013 by Laura M. Ramı́rez Elizondo
Copyright All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without written
permission of the author.
Printed by Wöhrmann Print Service B.V., Zutphen, the Netherlands.
To José, Mami, Papi, Tita, Carolina and Eduardo, who I deeply love.
Summary
The world’s increasing energy demand and growing environmental concerns have motivated
scientists to develop new technologies and methods to make better use of the remaining
resources of our planet. The main objective of this dissertation is to develop a scheduling
and control tool at the district level for small-scale systems with multiple energy carriers
and to apply exergy-related concepts for the optimization of these systems. The tool is
based on the energy hub approach and provides insights and techniques that can be used
to evaluate new district energy scenarios. The topics that are presented include the multicarrier unit commitment framework, the multi-carrier exergy hub approach, a hierarchical
multi-carrier control architecture, a comparison of multi-carrier power applications and
the implementation of a multi-carrier energy management system in a real infrastructure.
The dissertation consists of seven chapters. Chapter 1 includes the project framework,
motivation, problem definition and research questions. Chapter 2 describes the model behind the scheduling tool that was developed for this PhD project, which is used to optimize
systems containing multiple energy carriers. Later in Chapter 3 the optimization tool is
adapted to include exergetic efficiency as assessment parameter of such systems. Chapter 4
presents the control architecture that was designed to cope with the dynamic behaviour of
the systems under study. In Chapter 5 the optimization tool is used to analyze the impact of
different emerging trends in district-level power systems, such as the active participation of
micro-CHP technologies, the incorporation of renewable sources and the application of the
virtual power plant concept. Chapter 6 provides insights regarding the implementation of
the optimization and control tool in real systems. Finally, Chapter 7 presents the conclusions
and recommendations of this dissertation.
The main contributions of this dissertation are listed below:
• A general multi-carrier unit commitment framework for energy systems that contain
multiple energy carriers was developed. The framework can be used with any kind of
energy carrier and for different possible couplings and power scales (Chapter 2).
• A technique to include storage was developed and implemented as part of the optimization tool. The results show that this technique can be valuable for peak-shaving
purposes at the generation side (Chapter 2).
• The exergy hub approach was introduced. The exergy hub provides a visual indication of the exergetic efficiency of the units. In Section 3.4 both the energy hub and
the exergy hub are depicted next to each other in order to reveal that a unit that is
considered to be very efficient from an energy point of view can be considered to be
very inefficient from an exergy perspective.
i
ii
Summary
• A comparison between the results for the optimal dispatch obtained from an exergetic
efficiency optimization and from an energetic efficiency optimization is performed for
the first time for multiple energy carriers (Chapter 3).
• The results of the scheduling optimization tool are compared to identify which configuration gives the best energy and exergy performances for specific loads. A sensitivity
analysis is performed in which the ratio between heat and electricity consumption is
varied to observe the influence of the type of load in the scheduling (Chapter 3).
• A two-level control strategy was designed for the application in systems with multiple energy carriers. Most of the control strategies found in the literature only focus
on electricity flows, thus the strategy proposed is valuable for multi-carrier systems
(Chapter 4).
• The optimization tool was extended in order to study the benefits of having an aggregator in charge of the optimization (Chapter 5).
• A comparison among three micro-CHP technologies was presented to show that different benefits can be obtained by using combined heat and power technologies with
different electricity-to-heat efficiency ratios (Chapter 5).
• An example was presented in which the influence of incorporating electric vehicles
at a neighbourhood was analyzed (Chapter 5).
• Several practical considerations were presented regarding the implementation of the
tool in real systems. A partial implementation in the renewable energy laboratory
DENlab was performed, which provides an added value to the theoretical results that
were accomplished in this dissertation (Chapter 6).
Laura M. Ramı́rez Elizondo
Samenvatting
De toenemende wereldwijde vraag naar energie en de groeiende zorg voor het milieu hebben wetenschappers gemotiveerd om nieuwe technologieën en methoden te ontwikkelen om
beter gebruik te maken van de resterende grondstoffen van onze planeet. Dit proefschrift
vloeit voort uit deze bezorgdheid. De belangrijkste doelstelling van dit proefschrift is het
ontwikkelen van een planningsprogramma gebruikmakend van de energy-hub-aanpak voor
kleinschalige, (woon)wijkniveau-systemen bestaand uit meerdere energiedragers en het evalueren van het gebruik van een exergie-analyse als assessmentinstrument voor dergelijke
systemen. Het proefschrift biedt inzichten en reikt technieken aan die kunnen worden toegepast bij het optimaliseren van systemen met meerdere energiedragers. De gepresenteerde
onderwerpen zijn: een raamwerk voor multi-carrier unit commitment, de multi-carrier
exergy-hub-aanpak, een hirarchische multi-carrier besturingsarchitectuur, een vergelijking
van multi-carrier toepassingen en de implementatie van een multi-carrier energy management system in bestaande infrastructuur.
Dit proefschrift bestaat uit zeven hoofdstukken. Hoofdstuk 1 omvat het projectkader,
de motivatie, de probleemstelling en onderzoeksvragen. Hoofdstuk 2 beschrijft het model
achter het planningsprogramma dat is ontwikkeld gedurende dit promotieonderzoek. Dit
programma wordt gebruikt voor de optimalisatie van systemen bestaand uit meerdere energiedragers. Een uitbreiding die de exergetische efficiëntie als regelparameter opneemt in de
optimalisatieprogramma wordt in hoofdstuk 3 uitgewerkt. Hoofdstuk 4 presenteert de regelarchitectuur die ontworpen is om om te gaan met het dynamische gedrag van de bestudeerde
systemen. In hoofdstuk 5 wordt het ontwikkelde optimalisatieprogramma gebruikt om de invloeden van verschillende opkomende trends in wijkniveau-energiesystemen te analyseren.
Voorbeelden hiervan zijn de actieve participatie van micro-WKK-technologieën, de integratie van hernieuwbare energiebronnen en de toepassing van het virtual power plant-concept.
In hoofdstuk 6 wordt vervolgens ingegaan op de verworven inzichten met betrekking tot
de implementatie van het optimalisatieprogramma en het regelprogramma gebaseerd op de
regelarchitectuur. Tot slot worden er in hoofdstuk 7 de conclusies en aanbevelingen van dit
proefschrift gepresenteerd.
De belangrijkste bijdragen van dit proefschrift zijn:
• De ontwikkeling van een algemeen multi-carrier unit commitment kader voor energiesystemen bestaand uit meerdere energiedragers. Dit kader kan worden toegepast bij
ieder type energiedrager en voor verschillende koppelingsmogelijkheden en vermogens schalen (hoofdstuk 2).
iii
iv
Samenvatting
• Een techniek om opslag mee te nemen is ontwikkeld en geı̈mplementeerd als onderdeel van het optimalisatieprogramma. De resultaten tonen aan dat deze techniek
waardevol is voor het afvlakken van pieken aan de generatie kant (hoofdstuk 2).
• De introductie van de exergy-hub-aanpak. De exergy-hub geeft een visuele indicatie
van het exergetisch rendement van de eenheden. In hoofdstuk 3 worden zowel de
energy-hub als de exergy-hub naast elkaar afgebeeld om aan te geven dat een eenheid die wordt beschouwd als zeer efficiënt vanuit energetisch oogpunt kan worden
beschouwd als zeer inefficiënt uit exergieperspectief (hoofdstuk 3).
• Een vergelijking van de resultaten voor de optimale inzet verkregen uit een exergetisch - rendement - optimalisatie en van een energetisch - rendement - optimalisatie.
In de literatuur is een dergelijke vergelijking niet eerder uitgevoerd binnen de context
van energievoorzieningssystemen met meerdere energiedragers (hoofdstuk 3).
• De resultaten van het optimalisatieprogramma worden vergeleken om te bepalen welke
configuratie de beste energie- en exergieprestaties geeft bij specifieke belastingen.
Een gevoeligheidsanalyse wordt toegepast waarin de verhouding tussen warmte en
elektriciteit wordt gevarieerd om te observeren hoe het type belasting de planning
beı̈nvloedt (hoofdstuk 3).
• Een gecascadeerde regelstrategie is ontworpen voor toepassing in systemen met meerdere energiedragers. De meeste regelstrategieën in de literatuur zijn alleen gericht
op elektriciteitsstromen, dus de voorgestelde strategie is waardevol voor multi- dragersystemen (hoofdstuk 4).
• Verscheidene opkomende trends zijn gesimuleerd om inzicht te krijgen in hun gevolgen voor energievoorzieningssystemen op woonwijkniveau. Het optimalisatieprogramma werd uitgebreid om de voordelen van een aggregator die verantwoordelijk
is voor de optimalisatie te bestuderen (hoofdstuk 5).
• Een vergelijking van drie micro-WKK-technologieën is gemaakt om aan te tonen dat
verschillende voordelen kunnen worden verkregen door gebruik te maken van warmtekrachtkoppelingstechnologieën met verschillende elektriciteit-warmteverhoudingen
(hoofdstuk 5).
• Een voorbeeld waarin de gevolgen van elektrische voertuigen op het systeem op wijkniveau werd geanalyseerd (hoofdstuk 5).
• Verscheidene praktische overwegingen met betrekking tot de implementatie van het
programma in echte systemen. De gedeeltelijke implementatie in het duurzame energie lab DENlab voorziet in toegevoegde (praktische) waarden aan de theoretische
resultaten die werden bereikt in dit proefschrift (hoofdstuk 6).
Laura M. Ramı́rez Elizondo
Contents
Summary
i
Samenvatting
iii
List of Tables
ix
List of Figures
xi
Nomenclature
xv
Acknowledgments
xxiii
1 The Project: Introduction
1.1 Project Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Motivations to Integrate Decentralized and Renewable Generation . . . . .
1.2.1 Increasing Global Energy Demand . . . . . . . . . . . . . . . . . .
1.2.2 High Dependence on Fossil Fuels . . . . . . . . . . . . . . . . . .
1.2.3 Uneven Distribution of Fossil Fuel Reserves . . . . . . . . . . . . .
1.2.4 Growing Environmental Concerns . . . . . . . . . . . . . . . . . .
1.3 Changes that Trigger the Evolution of Power Systems . . . . . . . . . . . .
1.3.1 Optimization of Traditional Power Systems . . . . . . . . . . . . .
1.3.2 Control of Traditional Power Systems . . . . . . . . . . . . . . . .
1.3.3 Emergent Participants: Cogeneration, District Heating and Renewable Energy Technologies . . . . . . . . . . . . . . . . . . . . . .
1.3.4 Role of Power Electronics in Future Power Systems . . . . . . . .
1.3.5 Smart Grids: Intelligence in Future Power Systems . . . . . . . . .
1.4 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4.1 The Energy Hub Approach . . . . . . . . . . . . . . . . . . . . . .
1.4.2 Exergy Analysis as Assessment Tool for Energy Systems . . . . . .
1.4.3 Intelligent Energy Management Systems at District Level . . . . . .
1.5 Research Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5.1 Main Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5.2 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . .
1.6 Overview of this Dissertation . . . . . . . . . . . . . . . . . . . . . . . . .
1.6.1 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.6.2 Main Contributions . . . . . . . . . . . . . . . . . . . . . . . . . .
v
1
1
3
4
4
5
5
6
6
7
9
12
12
14
14
15
16
17
17
17
18
18
18
vi
Contents
2 The Framework: Multi-Carrier Unit Commitment
2.1 Basic Concepts and Definitions . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Energy Hub Element . . . . . . . . . . . . . . . . . . . . . . .
2.1.2 Multi-Carrier Optimal Dispatch . . . . . . . . . . . . . . . . .
2.1.3 Multi-Carrier Unit Commitment . . . . . . . . . . . . . . . . .
2.2 Modeling of Multi-Carrier Energy Systems . . . . . . . . . . . . . . .
2.2.1 Constraints and Assumptions . . . . . . . . . . . . . . . . . . .
2.2.2 Energy Hub Conversion Model . . . . . . . . . . . . . . . . .
2.2.3 Multi-Carrier Optimal Dispatch Model . . . . . . . . . . . . .
2.2.4 Multi-Carrier Optimal Dispatch Model Including Storage . . . .
2.3 Multi-Carrier Unit Commitment Framework . . . . . . . . . . . . . . .
2.3.1 Multi-Carrier Unit Commitment Model . . . . . . . . . . . . .
2.3.2 Multi-Carrier Unit Commitment Technique to Include Storage .
2.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.1 Simulation 1: Multi-Carrier Unit Commitment without Storage
2.4.2 Simulation 2: Multi-Carrier Unit Commitment with Storage . .
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
21
21
21
22
22
22
22
24
26
28
28
31
32
32
39
44
3 The Approach: Energy or Exergy Optimization?
3.1 Basic Concepts and Definitions . . . . . . . . . . . . . . . . . .
3.1.1 Energetic Efficiency . . . . . . . . . . . . . . . . . . .
3.1.2 Exergetic Efficiency . . . . . . . . . . . . . . . . . . .
3.1.3 Energy Content of a Fuel . . . . . . . . . . . . . . . . .
3.1.4 Specific Internal Energy . . . . . . . . . . . . . . . . .
3.1.5 Entropy Change . . . . . . . . . . . . . . . . . . . . .
3.1.6 Specific Enthalpy . . . . . . . . . . . . . . . . . . . . .
3.1.7 Carnot Cycle . . . . . . . . . . . . . . . . . . . . . . .
3.1.8 First Law and Second Law of Thermodynamics . . . . .
3.1.9 Exergy Definition . . . . . . . . . . . . . . . . . . . . .
3.1.10 Energy and Exergy General Equations . . . . . . . . . .
3.1.11 Generalities of the Exergy Method . . . . . . . . . . . .
3.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Exergy Optimization Approach . . . . . . . . . . . . . . . . . .
3.3.1 Exergy Hub Conversion Model . . . . . . . . . . . . . .
3.3.2 Problem Statement . . . . . . . . . . . . . . . . . . . .
3.4 Exergy Calculation Models and Data . . . . . . . . . . . . . . .
3.4.1 System Representation: Energy Hub versus Exergy Hub
3.4.2 Energetic and Exergetic Efficiencies of Generation Units
3.4.3 Natural Gas Supply and Biomass Supply (Ξg , Ξb ) . . . .
3.4.4 Electricity Supply (Ξe ) and Electricity Demand (Γe ) . .
3.4.5 Heat Load (Γq ) . . . . . . . . . . . . . . . . . . . . . .
3.4.6 Compression Heat Pump (εBeq ) . . . . . . . . . . . . . .
3.4.7 Gas-fired CHP Unit . . . . . . . . . . . . . . . . . . . .
3.4.8 Gas-Fired Furnace (εD
gq ) . . . . . . . . . . . . . . . . .
3.4.9 Biomass-fired CHP Unit . . . . . . . . . . . . . . . . .
3.4.10 Cost Data . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
47
47
47
47
47
48
48
48
48
49
49
50
51
51
54
54
54
56
56
57
58
58
58
60
60
60
60
61
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
3.5
3.6
vii
3.4.11 Load Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.1 Simulation 1: Optimal Dispatch - Difference among Optimizing
Energetic Efficiency, Exergetic Efficiency and Costs . . . . . . . .
3.5.2 Simulation 2: Optimal Scheduling - Difference among Optimizing
Energetic Efficiency and Exergetic Efficiency . . . . . . . . . . . .
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4 The Control: Multi-Carrier Hierarchical Control Architecture
4.1 Basic Concepts and Definitions . . . . . . . . . . . . . . . . . . . .
4.1.1 Participation Factors . . . . . . . . . . . . . . . . . . . . .
4.1.2 Law of conservation of energy in an open system . . . . . .
4.1.3 Heat Conduction . . . . . . . . . . . . . . . . . . . . . . .
4.1.4 Heat Convection . . . . . . . . . . . . . . . . . . . . . . .
4.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Integrated Control Architecture . . . . . . . . . . . . . . . . . . . .
4.3.1 Unit Control . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2 Main Control . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.3 Unit Control Applied . . . . . . . . . . . . . . . . . . . . .
4.3.4 Main Control Applied . . . . . . . . . . . . . . . . . . . .
4.4 Dynamic Component Models and Controls . . . . . . . . . . . . .
4.4.1 Wind Turbine . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.2 Lead-Acid Battery . . . . . . . . . . . . . . . . . . . . . .
4.4.3 Gas-Fired Turbine . . . . . . . . . . . . . . . . . . . . . .
4.4.4 Solid Oxide Fuel Cell . . . . . . . . . . . . . . . . . . . .
4.4.5 Furnace . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.6 District Heating Load . . . . . . . . . . . . . . . . . . . . .
4.4.7 Heat Storage Tank . . . . . . . . . . . . . . . . . . . . . .
4.4.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5.1 Simulation 1: Electricity Control Subsystem Demonstration
4.5.2 Simulation 2: Heat Control Subsystem Demonstration . . .
4.5.3 Simulation 3: Complete System Demonstration . . . . . . .
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5 The Scenarios: Multi-Carrier Power Applications
5.1 Basic Concepts and Definitions . . . . . . . . . . . . . . . .
5.1.1 Virtual Power Plant . . . . . . . . . . . . . . . . . .
5.1.2 Aggregator . . . . . . . . . . . . . . . . . . . . . .
5.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . .
5.3 Optimization Scenarios . . . . . . . . . . . . . . . . . . . .
5.3.1 Scenario 1: Base Case . . . . . . . . . . . . . . . .
5.3.2 Scenario 2: Individual Optimization . . . . . . . . .
5.3.3 Scenario 3: Individual Optimization with Storage . .
5.3.4 Scenario 4: Collaborative Optimization . . . . . . .
5.3.5 Scenario 5: Collaborative Optimization with Storage
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
61
62
62
66
67
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
69
. 69
. 69
. 70
. 70
. 70
. 71
. 72
. 72
. 72
. 77
. 77
. 79
. 79
. 81
. 82
. 85
. 87
. 88
. 90
. 91
. 92
. 92
. 95
. 97
. 101
.
.
.
.
.
.
.
.
.
.
103
103
103
104
104
105
106
106
107
108
110
.
.
.
.
.
.
.
.
.
.
viii
Contents
.
.
.
.
.
.
.
.
.
.
.
.
.
110
111
112
112
112
114
116
117
120
124
127
129
131
.
.
.
.
.
.
.
.
.
133
133
136
137
138
139
139
139
143
147
7 The Outcome: Conclusions, Recommendations and Coming Work
7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2 Recommendations for Further Research . . . . . . . . . . . . . . . . . . .
7.3 Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
149
149
152
153
A Assumptions and Considerations
155
B Complementary Simulation Results
157
Bibliography
161
List of Publications
171
Curriculum Vitae
173
5.4
5.5
5.6
5.3.6 Inclusion of Renewables . . . . . . . . . . . . . . . . . . . . . .
5.3.7 Inclusion of Electric Vehicles . . . . . . . . . . . . . . . . . . .
Input Data for the Simulations . . . . . . . . . . . . . . . . . . . . . . .
5.4.1 Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4.2 Component Parameters . . . . . . . . . . . . . . . . . . . . . . .
5.4.3 Load Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5.1 Simulation 1: Comparison of Micro-CHP Technologies and Prices
5.5.2 Simulation 2: Introducing an Aggregator . . . . . . . . . . . . .
5.5.3 Simulation 3: Introducing Storage . . . . . . . . . . . . . . . . .
5.5.4 Simulation 4: Introducing Renewables . . . . . . . . . . . . . . .
5.5.5 Simulation 5: Introducing Electric Vehicles . . . . . . . . . . . .
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6 The Implementation: Multi-Carrier EMS
6.1 Literature Review . . . . . . . . . . . .
6.2 Implementation of a Multi-Carrier EMS
6.2.1 Forecast Model . . . . . . . . .
6.2.2 Optimization Module . . . . . .
6.2.3 Real-Time Control Module . . .
6.3 Implementation at DENlab . . . . . . .
6.3.1 Description . . . . . . . . . . .
6.3.2 Example at DENlab . . . . . .
6.4 Conclusions . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
List of Tables
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
Prices for small commercial costumers at Nordhausen, Germany . .
Prices for small commercial costumers used in this case study . . .
Parameters of the components used in this case study . . . . . . . .
Electric and heat loads for the period under study in kW . . . . . . .
Simulation 1: Multi-carrier unit commitment - Case 1 (best strategy)
Simulation 1: Multi-carrier optimal dispatch - Case 1 (best strategy)
Simulation 1: Multi-carrier unit commitment - Case 2 (best strategy)
Simulation 1: Multi-carrier unit commitment - Case 3 (best strategy)
Simulation 1: Multi-carrier unit commitment - Case 4 (best strategy)
Parameters of the storage element . . . . . . . . . . . . . . . . . .
Simulation 2: Multi-carrier unit commitment - Step 1 (best strategy)
Simulation 2: Multi-carrier unit commitment - Step 3 (best strategy)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
32
33
33
34
36
36
37
38
38
40
41
43
3.1
3.2
3.3
3.4
3.5
3.6
3.7
Component parameters . . . . . . . . . . . . . . . .
Fuel costs . . . . . . . . . . . . . . . . . . . . . . .
Heat loads for the scheduling example . . . . . . . .
Heat loads for the scheduling example (continued) .
Simulation 1: Optimal dispatch . . . . . . . . . . . .
Simulation 2: Scheduling for different loads - Case 1
Simulation 2: Scheduling for different loads - Case 2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
59
61
62
62
64
65
65
4.1
4.2
Main control summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Components and control subsystems . . . . . . . . . . . . . . . . . . . . .
78
91
5.1
5.2
5.3
5.4
5.5
5.6
Gas and electricity prices . . . . . . . . . . . . . . . . . . . . . .
Parameters of storage devices . . . . . . . . . . . . . . . . . . . .
Parameters of generation devices . . . . . . . . . . . . . . . . . .
Gas and electricity prices . . . . . . . . . . . . . . . . . . . . . .
Simulation 3: Multi-carrier unit commitment results for Scenario 4
Simulation 3: Multi-carrier unit commitment results for Scenario 5
6.1
Components and control subsystems . . . . . . . . . . . . . . . . . . . . . 143
ix
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
112
112
113
117
125
125
List of Figures
1.1
1.2
Four parts of the “Intelligent Power Systems” research project . . . . . . .
Energy hub representation . . . . . . . . . . . . . . . . . . . . . . . . . .
2
14
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
Representation of an energy hub with multiple converters . . . . . .
Energy hub representation for the example . . . . . . . . . . . . . .
Storage elements in an energy hub . . . . . . . . . . . . . . . . . .
Flow chart of the proposed technique . . . . . . . . . . . . . . . . .
Energy hub representation for Simulation 1 . . . . . . . . . . . . .
Electric and heat loads for the period under study in kW . . . . . . .
Simulation 1: Multi-carrier optimal dispatch - Case 1 (best strategy)
Simulation 1: Multi-carrier optimal dispatch - Case 4 (best strategy)
Energy hub representation for Simulation 2 . . . . . . . . . . . . .
Simulation 2: Multi-carrier optimal dispatch - Step 1 (best strategy)
Simulation 2: Multi-carrier optimal dispatch - Step 2 . . . . . . . .
Simulation 2: Energy content of the storage element - Step 2 . . . .
Simulation 2: Multi-carrier optimal dispatch - Step 4 . . . . . . . .
Simulation 2: Energy content of the storage element - Step 4 . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
24
25
27
31
32
35
36
39
39
41
42
42
43
44
3.1
3.2
Exergy hub representation . . . . . . . . . . . . . . . . . . . . . . . . . .
Energy hub representation . . . . . . . . . . . . . . . . . . . . . . . . . .
56
56
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14
Two-level hierarchical control architecture . . . . . . . . . .
Flowchart to compute participation factors . . . . . . . . . .
Flowchart to define the participation of the storage elements
Flowchart to calculate the desired power output . . . . . . .
Energy hub used in the example . . . . . . . . . . . . . . .
Wind turbine . . . . . . . . . . . . . . . . . . . . . . . . .
Schematic of the micro-CHP gas-fired generator . . . . . . .
Schematic of the solid oxide fuel cell . . . . . . . . . . . . .
Energy hub used in the example . . . . . . . . . . . . . . .
Simulation 1: Participation factors of controllable units . . .
Simulation 1: Frequency of the system and rotor speeds . . .
Simulation 1: Electricity demand and electricity supply . . .
Simulation 1: Electricity storage behavior . . . . . . . . . .
Simulation 2: Indoor temperature and DHS line temperatures
73
75
76
77
78
79
82
86
91
92
93
94
94
95
xi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xii
List of Figures
4.15
4.16
4.17
4.18
4.19
4.20
4.21
4.22
Simulation 2: Heat demand and heat supply . . . . . . . . . . . . . . . . . 96
Simulation 2: Heat storage behavior . . . . . . . . . . . . . . . . . . . . . 96
Simulation 3: Forecasted and actual output of the wind turbine . . . . . . . 97
Simulation 3: Forecasted and actual electricity load . . . . . . . . . . . . . 98
Simulation 3: Heat load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Simulation 3: Combined control error and frequency of the system . . . . . 99
Simulation 3: Temperature of the district heating hot line and the space heating 99
Simulation 3: Total electricity demand and electricity supply . . . . . . . . 100
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
5.14
5.15
5.16
5.17
5.18
5.19
5.20
5.21
5.22
5.23
5.24
5.25
5.26
5.27
Scheme for Scenario 1 - Base case . . . . . . . . . . . . . . . . . . . . . . 106
Scheme for Scenario 2 - Individual optimization . . . . . . . . . . . . . . . 106
Scheme for Scenario 3 - Individual optimization with storage . . . . . . . . 107
Scheme for Scenario 4 - Collaborative optimization . . . . . . . . . . . . . 108
Scheme for Scenario 5 - Collaborative optimization with storage . . . . . . 110
Inclusion of renewables in Scenario 5 . . . . . . . . . . . . . . . . . . . . 110
Inclusion of electric vehicles in Scenario 5 . . . . . . . . . . . . . . . . . . 111
Average electricity and heat demand patterns for a week in winter . . . . . 114
Average electricity and heat demand patterns for a week in summer . . . . 114
Single and average electricity and heat demand patterns for a week in winter 115
Single and average electricity and heat demand patterns for a week in summer115
Simulation 1: Total costs and primary energy consumption for a week in
winter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Simulation 1: Total costs and primary energy consumption for a week in
summer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Simulation 1: Gas and grid energy consumption for a week in winter . . . . 119
Simulation 1: Gas and grid energy consumption for a week in summer . . . 119
Legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Simulation 2: Total costs and primary energy consumption . . . . . . . . . 121
Simulation 2: Gas and grid energy consumption for a week in winter . . . . 122
Simulation 2: Gas and grid energy consumption for a week in summer . . . 122
Simulation 3: Total costs and primary energy consumption . . . . . . . . . 126
Simulation 3: Gas and grid energy consumption . . . . . . . . . . . . . . 126
Example of a solar power pattern . . . . . . . . . . . . . . . . . . . . . . 127
Simulation 4: Gas and grid energy consumption . . . . . . . . . . . . . . 128
Simulation 4: Total costs and primary energy consumption . . . . . . . . . 128
Electric vehicle aggregated load pattern for one day . . . . . . . . . . . . . 129
Simulation 5: Gas and grid energy consumption . . . . . . . . . . . . . . 130
Simulation 5: Total costs and primary energy consumption . . . . . . . . . 130
6.1
6.2
6.3
6.4
6.5
6.6
6.7
Diagram of the multi-carrier energy management system .
Results of the wind forecast model . . . . . . . . . . . . .
Solar panels on the roof . . . . . . . . . . . . . . . . . . .
PLC and back-to-back converter in DENlab . . . . . . . .
Motor-generator sets used to emulate different components
DENlab configuration diagram . . . . . . . . . . . . . . .
Energy hub representation . . . . . . . . . . . . . . . . .
5.13
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
136
138
140
140
141
141
143
List of Figures
xiii
6.8
6.9
6.10
6.11
Response of the solid oxide fuel cell . . . . . .
Electricity demand and electricity supply . . . .
Response of the system’s frequency and voltage
Response of the furnace . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
144
144
145
145
B.1
B.2
B.3
B.4
B.5
Simulation 3:
Simulation 3:
Simulation 3:
Simulation 3:
Simulation 3:
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
157
158
158
159
159
Response of the wind turbine
Response of CHP A . . . . .
Response of CHP B . . . . .
Response of CHP C . . . . .
Response of the battery bank
.
.
.
.
.
.
.
.
.
.
Nomenclature
Acronyms
AC
alternating current
ACE
area control error
AFR
air-to-fuel ratio
AIMMS
Advanced Interactive Multidimensional Modeling System
BTU
British Thermal Units
CHP
combined heat and power
CO2
carbon dioxide
DC
direct current
DEMS
district energy management systems
DENlab
renewable energy laboratory at Delft University of Technology
DHC
district heating and cooling
ECN
Energy Research Centre of the Netherlands
EMS
energy management system
EMVT
electro-magnetic power technology
ETH
Eidgenössische Technische Hochschule Zürich
HEMS
home energy management systems
HHV
higher heating value
HMI
human-machine interface
ICT
information and communication technology
IEA
International Energy Agency
IEEE
Institute of Electrical and Electronics Engineers
xv
xvi
Nomenclature
IOP
innovation-oriented research programs
IPCC
Intergovernmental Panel on Climate Change
LHV
lower heating value
MSc
master of science
p.u.
per unit
PhD
doctor of philosophy
PID
proportional integral derivative
PLC
programmable logic controller
PQ
active power - reactive power
PV
photovoltaic
SCADA
supervisory control and data adquisition
SEPATH
Simulatie van Energievraag Patronen van Huishoudens
sms
short message service
US
United States of America
VoFEN
Vision of Future Energy Networks
VPP
virtual power plant
WADE
World Alliance for Decentralized Energy
Functions
k
Fβ jp (Pαi )
function decribing the conversion from carrier αi to carrier β j
Fobj
objective function
Sets and Subsets
Cαi = {A, B, . . .} set of converter elements inside the energy/exergy hub
Ein = {e, g, q, . . . } set of carriers at the input side of the energy/exergy hub
Eout = {e, g, q, . . . } set of carriers at the output side of the energy/exergy hub
Subscripts
0
reference state/ initial state
e
electricity
g
natural gas
Nomenclature
q
heat
act
actual
afr
air fuel ratio
air
air
amb
ambient
col
cold
des
desired
dif
difference
ele
electrical
exc
exchange
exh
exhaust
exp
export
flu
flue gases
fue
fuel
fur
furnace
gen
generation
grd
public electricity grid
hex
heat exchanger
hhd
household
hot
hot
hub
individual hub
imp
import
in
in/inside
max
maximum
mec
mechanical
min
minimum
out
out/outside/output
pip
pipeline
xvii
xviii
Nomenclature
rad
radiator
rec
recuperator
ref
reference
ren
renewable sources
rof
roof
rot
rotor/rotational
rtd
rated
sch
scheduled
sur
surroundings
swh
service water heating
tnk
storage tank
tot
total
tur
turbine
veh
electrical vehicles
wal
wall
wat
water
wdw
window
wnd
wind
Variables and Constants Introduced in Chapter 1
λj
power-frequency characteristic of control area j
[MW/Hz]
f
frequency
Pi
active power of generator i
[MW]
Pj
active power export of control area j
[MW]
[Hz]
Variables and Constants Introduced in Chapter 2
Pαi
upper power limit at the energy hub’s input
[kW]
Pαi
upper power limit at the input of converter k p
[kW]
Pαi
lower power limit at the energy hub’s input
[kW]
lower power limit at the input of converter k p
[kW]
kp
k
Pαpi
Nomenclature
xix
Fcost
least total cost to arrive at a state
[e/(time period)]
Fprod
production costs
[e/(time period)]
Ftran
transition cost
[e/(time period)]
Idp
combination of commited units
[-]
Jdp
combination of commited units
[-]
Ndp
number of strategies that are saved at each period
[-]
t
time
[s]
T dp
time interval for the forward dynamic programming algorithm
[-]
Xdp
number of states that are evaluated within each period
[-]
Ėαi
energy change per time period
K1,αi
cost coefficient associated with input carrier αi
[e/(time period)]
K2,αi
cost coefficient associated with input carrier αi
[e/(time period · kW)]
K3,αi
cost coefficient associated with input carrier αi
[e/(time period · kW2 )]
k
p
K4,α
i
[kW]
cost coefficient of coverter k p associated with input carrier αi [e/(time period)]
e
Lβ j
active power carried by β j out of the hub’s internal cluster
[kW]
eαi
P
active power carried by αi to the hub’s internal cluster
[kW]
Dαi
power that flows to a storage element located at the hub’s input side
[kW]
e+αi
charge/stand-by storage efficiency
[-]
e−αi
discharge storage efficiency
[-]
eαi
charge/discharge storage efficiency
[-]
eq
Mβ j
equivalent storage element of the equivalent storage vector
[kW]
Mβ j
power that flows to a storage element located at the hub’s output side
[kW]
k
wαpi
status variable of converter k p
[-]
wαi
status variable of all converter associated with carrier αi
[-]
Eαstbi
stand-by energy losses per time period
[kW·(time period)]
αi
energy/exergy carrier at the hub’s input
[-]
βj
energy/exergy carrier at the hub’s output
[-]
energy conversion efficiency from carrier αi to carrier β j of converter k p
[-]
k
ηαpi β j
xx
Nomenclature
cαi β j
coupling factor
[-]
kp
converter inside the energy/exergy hub
[-]
Lβ j
active power carried by β j out of the energy hub
[kW]
active power carried by β j out of converter k p
[kW]
Pαpi
active power carried by αi to converter k p
[kW]
Pαi
active power carried by αi to the energy hub’s input
[kW]
k
Lβpj
k
k
vαpi
dispatch factor of carrier αi at converter k p
[-]
Variables and Constants Introduced in Chapter 3
upper limit at the exergy hub’s input
[kW]
Ξ αi
upper limit at the input of converter k p
[kW]
Ξ αi
lower limit at the exergy hub’s input
[kW]
Ξ αpi
lower limit at the input of converter k p
[kW]
µche
chemical potential
nche
quantity of moles per unit mass
3
velocity
[m/s]
w
work per kg
[J/kg]
φm
mass flow
[kg/s]
Ξ αi
kp
k
k
[J/mole]
[mole/kg]
εαpi β j
exergy conversion efficiency from carrier αi to carrier β j of converter k p
[-]
ε
exergetic efficiency
[-]
η
energetic efficiency
[-]
ζαi
exergy factor of carrier αi
[-]
ΥLHV,αi
lower heating value of carrier αi
Γβ j
exergy flow carried by β j to the hub’s output
[kW]
Γout
sum of exergy flow used by the loads at the hub’s output
[kW]
Lout
sum of power used by the loads at the hub’s output
[kW]
Ξαi
exergy flow carried by αi to the hub’s input
[kW]
Ξinp
sum of exergy flow brought to the system by the input fuels
[kW]
Pinp
sum of power brought to the system by the input fuels
[kW]
[kJ/kg]
Nomenclature
xxi
a
specific exergy
[J/kg]
bαi β j
coupling factor
[-]
g
gravity
[m/s2 ]
h
specific enthalpy
[J/kg]
m
mass
[kg]
p
pressure
[Pa]
Q
heat
S
entropy
s
specific entropy
T
temperature
u
specific internal energy
v
specific volume
z
height
[J]
[J/(K)]
[J/(K·kg)]
[K]
[J/kg]
[m3 /kg]
[m]
Variables and Constants Introduced in Chapter 4
ǫfur
furnace parameter that is based on experimental data
κ
thermal conductivity
ρ
density
τ
time constant
[s]
ξfur
factor representing the losses of the furnace
[-]
pf,βp j
participation factor
[-]
λ
tip speed ratio
[-]
ω
rotational speed
θ
pitch angle
[-]
ιfur
factor related to the excess air in the furnace
[-]
Φq
heat flow
τ
torque
A
area
cp
performance coefficient
k
[-]
[W/(m·K)]
[kg/m3]
[rad/s]
[W]
[N·m]
[m2 ]
[-]
xxii
Nomenclature
c
specific heat capacity
d
thickness
E
energy
J
moment of inertia
raft
stoichiometric air fuel ratio
U
heat transfer coefficient
V
volume
C
constant related to the performance coefficient of the wind turbine
[J/(K· kg)]
[m]
[J]
[kg·m2]
[-]
[W/(m2 · K)]
[m3 ]
[-]
Variables and Constants Introduced in Chapter 5
hd
households
[-]
Variables and Constants Introduced in Chapter 6
σ2t
conditional variance of ̺t
̺t
innovations or residuals of the time series
ςt
standardized residuals
[-]
[m/s]
[-]
Vectors and Matrices Introduced in Chapter 2
Meq
equivalent storage vector
Ė
vector containing the energy change during one time period
S
storage coupling matrix
D
vector containing the power that flows to storage elements located at the hub’s
input side
M
vector containing the power that flows to storage elements located at the hub’s
output side
e
L
output power vector of the hub’s internal cluster
e
P
input power vector of the hub’s internal cluster
C
energy hub’s coupling matrix
L
energy hub’s output power vector
P
energy hub’s input power vector
Vectors and Matrices Introduced in Chapter 3
Γ
exergy hub’s output vector
Ξ
exergy hub’s input vector
B
exergy hub’s coupling matrix
Acknowledgements
First of all, I would like to thank my daily supervisor Dr.dr.h.c.ir. Bob Paap for inviting
me to work on his project and for all the enriching conversations we have had over the
past years. Bob retired two years ago and for that reason he was not allowed to be my
co-promotor, however his involvement in this project continued during all this time and for
this I am deeply grateful. I would also like to thank his wife, Rietje Paap for making me
feel welcome at their home each time I visited Bob to discuss the results of my simulations,
particularly during the final year of my PhD. It was a great pleasure to have Bob as my
supervisor and I am sure that we will continue to have only pleasant meetings in the future.
I, of course, still look forward to seeing his sailing boat one day!
Secondly, I would like to thank my promotor Prof.ir Lou van der Sluis for his good
advice and for introducing me to his network of colleagues. As a result of his contacts, I am
currently conducting a research project in which the results of my PhD project are realized
in practice. In addition, I am currently drawing up some very interesting research proposals
with colleagues from other disciplines. I sincerely appreciate his continued support.
This PhD project was financed by SenterNovem. I would like to thank the members of
the IOP-EMVT program for all the interesting discussions that we had during our periodical
meetings. I would like to thank Prof.ir. M. Antal and Ir. G.W. Boltje for their involvement
at these gatherings. I would also like to thank Ir. Hans Buitenhuis, director of DWA, for his
input during the project.
During my PhD I had the opportunity to complete an internship at ABB Corporate
Research in Västerås, Sweden. I would like to thank Dr. Muhamad Reza and Dr. Georgios
Demetriades for giving me the opportunity to spend three months conducting research in
their group. This was a very valuable experience from a research point of view, but it also
gave me the chance to meet interesting people that I continue to keep in touch with today.
Due to the interdisciplinary nature of my topic I have had the opportunity to collaborate
with colleagues from other fields. I have been able to co-author publications with a few
colleagues and with others, draw up proposals that we look forward to working on in the
near future. I want to thank Ir. Alicja Lojowska, Ir. Sabine Jansen, Dr. Rudy Negenborn,
Dr.ir. Richard Toonssen, Ballard Asare-Bediako, MSc., Dr. Matthijs Spaan and Dr. Stefan
Witwicki for the joint-collaborations and Dr. Mathijs de Weerdt and Dr. Matthijs Spaan for
the collaborations to come. I would also like to thank Dr. Martin Geidl for the valuable contact that we had at the beginning of my PhD and Dr.ir. Madeleine Gibescu for introducing
me to the software AIMMS by means of an illustrative example. I also want to thank Dr.ir.
Michiel Houwing, Dr. Kas Hemmes, Dr. Poppong Sakulpipatsin, Ir. Theo Woudstra and
Dr.ir. Nico Woudstra for the valuable conversations we had about this research topic.
xxiii
xxiv
Acknowledgements
I would like to thank Croon Elektrotechniek, especially Ir. IJsbrand Ponsioen for giving
me the opportunity to conduct research on one of their current projects. It has been very
enriching to work with the entire Zero-Watt team.
It has been a great pleasure for me to work at the Power Systems Group at the Delft
University of Technology. I want to thank my colleagues for all the memorable moments we
shared during lunch time and during the coffee breaks. Special thanks to Alicja Lojowska
for the joint-collaborations and for being such a great officemate and Ana Ciupuliga for
the interesting conversations on nutrition, yoga and music :). During my PhD I supervised
several MSc students. Some of them worked on topics related to this thesis and others did
not, but they all contributed to my development as research supervisor. Thanks Eva Usó,
Victor Vélez, Remco Ammerlaan, Bernardo Valladares, Pradeep Jayakumar, Hugo Cruz
and Naoual Oukira. I want to especially thank Remco and Victor for their participation in
the papers we published and Bernardo for his work at DENlab.
I would like to thank my colleagues at the Amsterdam University of Applied Sciences
for making me feel welcome from the beginning. I am very grateful that I have had the
opportunity to be able to combine teaching and research and I would like to especially thank
the education manager Jorien Schreuder, Cleantech’s project manager Katrien de Witte, and
the two Lectoren of Cleantech, Dr. ir. Robert van den Hoed and Ir. Inge Oskam for allowing
me to change a few working hours in order to finalize the details of my PhD. I would also
like to thank my colleagues at the School of Electrical Engineering of the Universidad de
Costa Rica, especially Prof.dr. Jorge Romero, its Director, for supporting me with my
studies abroad and for the collaborations to come.
There are two very important colleagues that I would like to thank: Dr.ir. Barry Rawn
and Dr.ir. Dhiradj Djairam. We started a “Journal Club”, consisting of periodical meetings
to generate new research ideas and support the publication of journal entries and papers.
Their support and interest encouraged me to choose them as my paranimfs. Thank you
Barry and Dhiradj for accepting! I am grateful for all your support.
In order to keep things in perspective during a PhD, it is important to continue to do
the things one loves. A very important activity that I was able to continue to do the entire
duration of my PhD was to teach piano. I would like to deeply thank Marleen and Pieter
van Rijckevorsel, Anne and Tim van Wingerde, Marı́a Sofı́a Clercx Lao, their parents and
family for not only giving me the opportunity to teach them to play the piano but also
the opportunity to enjoy making wonderful music with them. Being able to witness their
significant progress has given me energy and inspiration during all this time. I am very
thankful for all the special moments that we have shared during our periodic concerts. You
really fill my heart with joy.
I would not have been able to teach piano if it was not for my piano teacher Prof. Flora
Isabel Elizondo, who accepted me as her undergraduate piano student whilst I was completing my bachelor’s degree in engineering. Completing a bachelor’s degree in music, was
a very important achievement in my life. Attending her nice and motivating lessons was
always a favourite activity for me. I want to thank her for her support and encouragement in
my decision to come to the Netherlands and for having travelled from Costa Rica to be able
to attend my PhD defense. I would like to give very special thanks to Prof. Emilio Alpı́zar
(electrical engineering professor at my home university), who motivated me to start a PhD
abroad and who also supported me on my decision to follow both engineering and music.
Acknowledgements
xxv
Surrounding yourself with nice people fills you with energy and brings so much pleasure. For this reason I would like to thank all the friends that I have met here, for their
continued support and appreciation. I want to deeply thank Milena Markova, Johan Wolmarans, Taryn Bonnette, Marco Álvarez, Mónica Morales, Marcelo Ackermann, Neslihan
Parmaksizoǧlu, Adolfo Chaves, Marcelo Gutiérrez, Jodi Kooijman, Coralie Selin, Antoine
Ripol, Ana Lao, Luud Clercx, Barry Lennon, Natalie Kretzschmar, Julian Schmied, Mustafa
Ibrahim, Alon Yehezkely, Shelly Nevo, Miriam ter Brake and Marieke Nekeman for all the
dinners/concerts/coffee breaks/movies/activities and all the time that we have spent together.
You all have a very important place in my heart. I would also like to thank all my good
friends who live in Costa Rica, with whom I meet every year when I go home to visit. It is
wonderful to be able to maintain our friendship even though there is such a great distance
between us. Even though I do not have a list with their names on here, they know who they
are :) I look forward to seeing you all in December to celebrate the completion of my PhD!
Finally, I want to thank my friends Taryn Bonnette, Barry Lennon, Barry Rawn and Stefan
Witwicki for helping me with last-minute revisions as well as Jodi Kooijman and Arjen van
der Meer for assisting me with the Dutch translations. Furthermore, I want to deeply thank
Oscar Calderón for the amazing illustrations he contributed to my thesis.
A loving family is one of the most valuable treasures a person can have. I cannot describe how much I love and cherish my family. My Mom and Dad have been my greatest
inspiration. I have always looked up to their perseverance and multiple talents. They gave
my brother, my sister and me the best opportunities possible to develop our abilities and they
always encouraged us to follow our passions. I love them from the bottom of my heart! My
brother Eduardo and sister Carolina are wonderful! Their creativity is something I really
admire, my brother creates meaningful and fun videogames and my sister wonderful paintings, besides working as industrial engineer. I am proud of them for their unique talents and
for the goodness in their hearts. Pi and Ami, you are awesome! I want to thank my Mom for
the beautiful dress she made for my defense and my sister for the great painting she made
for the cover of this thesis and that she designed based on the energy hub representation.
Thanks to the rest of my family, including my dearly beloved grandparents Lucila (Tita),
Dulia† and Jorge† and my family in law for their constant support, their constant thoughts
and for the special moments we always share every time we go to Costa Rica.
And finally José. My true soulmate. Thank you for the unconditional love and support
that you have given me during all these years. You supported me from the very beginning
with the idea of coming to the Netherlands and starting an MSc, the idea to continue and
complete a PhD, followed by the idea of pursuing my love for yoga in completing a teacher
training :) It is awesome to have found someone with whom I can feel free to do the things
I love and be supported whilst doing them. It is wonderful to share time with a person with
whom life is so full of love and enjoyment and to live with someone who every single day
puts a smile on my face ♥.
Laura M. Ramı́rez Elizondo
Delft, March 2013.
Chapter 1
The Project
Introduction
The world’s increasing energy demand and growing environmental concerns have motivated scientists to develop new technologies and methods to make better use of the remaining resources of our planet. This dissertation emerges in response to these concerns as
it investigates different scenarios in which the implementation of a scheduling and control
tool for the optimization of small-scale systems with multiple energy carriers at district level
proves to be beneficial. Furthermore, it analyzes the application of exergy-related concepts
for the optimization of such systems.
This first chapter describes the research framework and serves as theoretical background
for the other chapters. The chapter is organized in the following way. Section 1.1 introduces
the research program under which the project was performed. In Section 1.2 some of the
main aspects that have motivated the study and implementation of decentralized generation
are discussed. Section 1.3 describes the evolution of power systems due to the integration
of cogeneration and renewable energy sources. Section 1.4 presents the problem definition
of this thesis. In Section 1.5, the objectives and research questions are formulated. Finally
Section 1.6 contains the outline of this dissertation and a summary of the main contributions.
Parts of this chapter have been published in [1–3].
1.1 Project Framework
This PhD project was performed within the “Intelligent Power Systems” research framework of the IOP-EMVT program “Innovation-Oriented Research Programs - Electro - Magnetic Power Technology”, financed by SenterNovem, a former agency of the Dutch Ministry
of Economic Affairs. The project “Intelligent Power Systems” was initiated by the Electrical Power Systems Group and the Electrical Power Processing Group of Delft University of
Technology and by the Electrical Energy Systems Group and the Control Systems Group
of Eindhoven University of Technology. The project focuses on the effects caused by the
introduction of distributed (renewable) generators on existing power systems. A large implementation of decentralized technologies leads to horizontally-operated power systems
consisting of a large number of small to medium-sized distributed generators [4].
1
2
1 The Project: Introduction
The project “Intelligent Power Systems” consists of 4 parts, in which more than 10 PhD
students were involved. Figure 1.1 shows a schematic representation of the research project.
Each part is briefly described below:
• Inherently Stable Transmission Systems In this part, the influence of uncontrolled
decentralized generation on the stability and dynamic behavior of the transmission
network is investigated. The topics of research include the control of decentralized
and centralized power plants and the application of power electronics and monitoring
systems to allow their control.
• Manageable Distribution Networks This part focuses on how to use power electronic interfaces to support the electrical grid and on how to make the distribution
network more active. Moreover, the stability of the distribution network and the effect of the stochastic behavior of certain decentralized generators on the voltage level
is studied.
• Self-Controlling Autonomous Networks This part of the project is focused on the
development of control techniques to operate autonomous networks in an optimal
and secure way. The local networks studied are operated autonomously, but they can
remain connected to the electrical grid for security reasons.
• Optimal Power Quality The goal of this part is to provide elements for discussion
between the polluter and the electrical grid operator, who has to apply measures in
order to comply with the standards and electrical grid codes.
Figure 1.1: Four parts of the “Intelligent Power Systems” research project
This dissertation falls under the topic “Self-Controlling Autonomous Networks”. The
title of the PhD project under which this dissertation was performed is “Intelligent Energy
Supply at Household and District Level”. The project is divided into two parts: the part that
focuses on the household level is being performed at the Electrical Energy Systems Group
of Eindhoven University of Technology, while the part that focuses on the district level
was performed at the Electrical Power Systems Group of Delft University of Technology.
This dissertation corresponds to the second part of the project, entitled “Intelligent Energy
Supply at District Level”.
1.2 Motivations to Integrate Decentralized and Renewable Generation
3
1.2 Motivations to Integrate Decentralized and Renewable
Generation
Due to the economic and industrial development that took place during the twentieth century, the need for electricity has become crucial in our society. Nowadays, the availability
of electricity has a strong influence in the way we live, not only in aspects related to the way
we accomplish our tasks, but also in aspects related to the way we interact and communicate
with others. It is for this reason that scientists are strongly encouraged to look for ways to
provide a long-term energy supply.
The interest in decentralized energy systems with high penetration of combined heat
and power technologies (CHP) and renewable sources has increased during the last years.
Decentralized generation systems are systems in which the electricity production occurs at
or near the point of use, irrespective of the size or technology [5]. They can be connected
to the electrical grid (on-grid) or they can operate autonomously with no connection to the
grid (off-grid).
The energy sector is facing unprecedented challenges in relation to the world’s energy
supply. The International Energy Agency (IEA) estimates that between 2009 and 2035 the
total primary energy demand will increase by 48,5% if current policy scenarios are kept
unchanged, and by 22,8% if a post-2012 climate policy framework is applied to stabilize
the concentration of global greenhouse gases at 450 ppm CO2 equivalent [6]. The IEA
indicates that developing countries may play an important role in the increase of energy
demand.
The predicted depletion of fossil fuels and the need to reduce carbon dioxide emissions
have stimulated the quest for more sustainable options. In 2007, the European Council
adopted ambitious energy and climate change objectives for 2020. Furthermore, the European Council has stated a long-term commitment to reduce the carbon dioxide emissions in
a margin of 80% to 95% by 2050. The objectives are presented below as published in [7]:
• To reduce greenhouse gas emissions by 20%, rising to 30% if the conditions are right.
• To increase the share of renewable energy to 20%.
• To make a 20% improvement in energetic efficiency.
Four main reasons have served as motivation to look for more efficient and sustainable
energy technologies:
• There is a continuous increase in energy demand due to the growing population and
to the adoption of more electricity-consuming devices.
• There is a high dependency on fossil fuels and a predicted depletion of their reserves.
• There is an uneven distribution of fossil fuel reserves over the world.
• The environmental concerns are increasing due to the global warming phenomenon.
These four aspects motivate new technologies and policies to emerge. In each of the
following subsections, a general description of each of these aspects is given and a brief
description is provided about how decentralized renewable generation can help to improve
each of the aspects.
4
1 The Project: Introduction
1.2.1 Increasing Global Energy Demand
Three of the main challenges that emerge from the forecasted increase in energy demand
are: to ensure sufficient energy supplies, to reduce the dependency on fossil fuels, and to
tackle the environmental impact of our consumption habits [8]. Renewable sources can be
used to ensure sufficient energy supplies and to reduce the dependency on fossil fuels since
they are abundant and they are spread all over the planet. On the other hand, decentralized
energy systems can contribute to supply energy to new consumers in developing countries,
where in many cases the existing infrastructure is not suitable to supply remote areas with
electricity. Because of the fact that in decentralized systems the production is carried out at
the point of use, no costs are associated with electricity transmission at high voltage levels;
this makes these technologies appropriate and attractive.
In order to tackle the environmental impact of our consumption habits, we are forced to
stop placing economical revenues above environmental conservation and to make considerable changes in our life-style. This does not only apply at societal level, but at individual
level as well. We can progressively substitute products that we regularly buy with ecofriendly products. Moreover, not turning off the lights when we are not using them and not
bringing a reusable bag to the supermarket to put in our groceries should be unacceptable in
this day and age. We are responsible of preventing the following saying from holding true:
Only when the last tree has died, the last river has been poisoned and the last
fish has been caught will we realize that we cannot eat money.
Cree Indian saying
1.2.2 High Dependence on Fossil Fuels
Fossil fuels have traditionally been the most important energy sources to supply both the
primary energy and the electricity demand. According to the World Energy Statistics of
the International Energy Agency about 86,6% of the primary energy demand was supplied
by fossil fuels in 1973; this percentage was reduced to 80,9% in 2009 [6]. A reason for
that decrease was the higher participation of nuclear and hydro energy, which increased
from 2,7% in 1973 to 8,1% in 2009. Alternative technologies such as geothermal, solar,
wind and heat increased from a 0,1% to a 0,8% in the same period of time. According to
the same statistics report, in the case of the fuel shares of electric generation, fossil fuels
occupied 75,1% of the total shares in 1973. Later in 2009, this percentage was reduced to
67,1%. In this case, the participation of alternative sources increased from 0,6% in 1973
to 3,3% in 2009. Even though some reduction in the total shares of fossil fuels has been
achieved, the energy scenario is still strongly dominated by these sources.
There is an ongoing controversy about the forecasted depletion of fossil fuels, however
actions should be taken in a short term to reduce the existing high dependency. Renewable
energy sources have the potential to relieve fossil fuels from being the major energy sources,
in this way their remaining lifetime can be extended. Moreover, decentralized systems with
combined heat and electricity technologies can attain higher efficiencies than conventional
systems, which also reduces fossil fuel consumption. In the near future, the adoption of
alternative sources and more efficient technologies will be crucial for our further development. The remaining fossil fuels could better be reserved for the production of other
important products, such as plastics.
1.2 Motivations to Integrate Decentralized and Renewable Generation
5
1.2.3 Uneven Distribution of Fossil Fuel Reserves
The fossil fuel reserves are unevenly distributed throughout the planet; this produces tension
among countries. The proven oil and gas reserves are concentrated in a reduced number of
countries; for example, over half of the global proven gas reserves are concentrated in three
countries: the Russian Federation (27%), Iran (15%) and Qatar (14%) [9]. An increasing
number of nations are relying on gas imports to meet their energy demands [10]. According
to the International Energy Agency, the concentration of fossil fuel resources is the most
enduring energy security risk [9, 11]. The dependence on global oil consumption is expected
to increase, moreover prices are expected to fluctuate due to the short-term demand and
supply shifts [12]. This means that tensions among countries are likely to increase in the
coming years even more.
By introducing renewable energy, a wider range of sources can be used to cover the
national energy demand. Renewable sources are more evenly distributed than fossil fuels.
Therefore, in order to take advantage of this favorable characteristic, a country must first
identify the sources that are available and consequently encourage policies for their integration. By doing this, the country will not only be able to supply its own energy demand, but
also to reduce import costs, create local jobs and open the door to new investments.
1.2.4 Growing Environmental Concerns
The Intergovernmental Panel on Climate Change (IPCC) concluded that the global mean
temperature has increased by 0,6◦C during the 20th century [13, 14]. They argue that this
increase is likely to have resulted from a rise in the amount of greenhouse gases in the
atmosphere. Moreover, they indicate that there are strong evidences to infer that this warming effect is a consequence of human activities. A excerpt from [10], in which the effects
associated with climate change are listed, is cited below:
Climate change has been linked to increased temperatures causing droughts,
famines, insect infestations, the witness of new diseases, surges in existing maladies, and fires; floods and widespread human displacements; violent storms
resulting great human suffering; unseasonable blizzards and cold temperatures
and almost every other type of weather imaginable. The inefficient generation
of electricity in centralized plants is therefore a major cause of climate change
and the resulting conflict and insecurity that is resultant from it.
According to the World Alliance for Decentralized Energy (WADE), approximately one
third of the global CO2 emissions in 2005 resulted from heating, cooling and power supply systems in residential and commercial sectors [14]. WADE estimated that in the United
States of America, 20% of the total growth of CO2 emissions can be displaced by integrating
decentralized combined heat and power technologies in buildings due to the increased thermal efficiency of these systems. Similarly, renewable energy technologies utilize sources
that are continuously replenished, abundant and do not produce extra CO2 emissions, therefore they are suitable to complement both conventional and new technologies.
It can be concluded from this section that the inclusion of renewable and (decentralized) combined heat and power technologies in energy supply systems has the potential to
promote a more efficient and sustainable use of resources.
6
1 The Project: Introduction
1.3 Changes that Trigger the Evolution of Power Systems
The power system’s evolution that is taking place at this moment is driven by different factors. First, the expected depletion of fossil fuels and environmental concerns have encouraged society to look for a more sustainable development path. Second, new technical developments have allowed scientists to aim for more intelligent and efficient ways to control
and operate power systems. Third, international agreements set around the Kyoto Protocol
in 1997 have stimulated governments to pursue a new direction towards sustainability.
One of the main differences encountered by integrating decentralized and renewable
technologies in relation to traditional systems is that the power flow direction in the network is not predictable anymore: one-way traffic becomes two-way traffic [15]. This has
a significant impact in protection systems but also in the way power systems are optimized
and controlled. Traditionally, networks had a vertical structure with a single power flow
direction. Power systems consisted of large centralized power plants connected to transmission networks that fed distribution networks, from which the load was supplied. Due
to the introduction of decentralized generation, power systems are becoming horizontallyoperated systems in which power is not only produced at the traditional generation level, but
also at the distribution level. Two other significant differences that have taken place during
this evolution are the fact that besides large power plants (of several MWs), small-scale generation units (of a few kWs) have been introduced and the fact that some of the sources vary
stochastically, like in the case of solar radiation and wind. Due to these differences, the use
of power electronic devices and the development of more flexible bi-directional techniques
play an important role for a successful integration of new technologies.
This section introduces important aspects related to the recent evolution of power systems. First, a brief summary of some of the main optimization and control techniques used
in traditional power systems is presented. Later, a description of three of the main participants in this evolution is provided. Finally, the role of power electronics and the role of
information and communication technologies (ICT) is discussed.
1.3.1 Optimization of Traditional Power Systems
The optimization of traditional power systems is based on scheduling and dispatching the
power units involved according to the electricity demand. The terms energy management
system (EMS), economic dispatch and unit commitment are described below.
Energy Management Systems
In traditional power systems the transmission and distribution of electrical energy is monitored, coordinated and controlled in a control center where an energy management system
serves as interface between the operator and the power system [15]. In an energy management system, a Supervisory Control and Data Acquisition system (SCADA) is in charge
of collecting real-time measured data. The data are converted to digital data through a
computerized process. By means of a human-machine interface (HMI) the human operator
has access to the processed data. The processed data can be sent as input to other parallel programs for further computation, for example to an optimal power flow program, a
contingency analysis program or a unit commitment program.
1.3 Changes that Trigger the Evolution of Power Systems
7
Remote terminal units can be connected at several relevant locations. Furthermore a
programmable logic controller (PLC) can be used to configure and direct the signals and
digital data. A communication infrastructure allows the data exchange in the system. The
SCADA can be coupled to a control system containing the control algorithms in order to
execute further actions. A state estimator works in combination with the SCADA system
to overcome inconsistencies, for example measurements that may be corrupted, redundant
measurements, wrong measurements, etc [15]. Therefore, the state estimator keeps the
integrity of the real-time database.
Economic Dispatch
The solution of an economic dispatch problem provides the power output that each available
generation unit is required to deliver in order to supply a specified load condition in the best
economical way [16]. The resulting dispatch is obtained by a program that minimizes the
overall cost of fuel that is needed to serve the specified load. In traditional power systems
the load considered is the electric load. In this dissertation, the load to be considered can
also be of another energy form, for example a heat load.
Unit Commitment
The total power demand varies during the day, for this reason the electricity utility has
to decide in advance which generators to start up or shut down, and in which sequence
this should be done; this procedure is called unit commitment [16]. The unit commitment
program schedules the units according to a predicted or forecasted load over a future period
of time [16]. Some of the main factors that are taken into account for the optimal scheduling
of units are the production costs, the start-up and shut-down costs, the operating fuel costs,
the fuel types and the length of the forecasting period.
1.3.2 Control of Traditional Power Systems
In traditional electrical power systems, two main types of control can be identified: active
power control and reactive power control. The term active power control is related to performing frequency control, whereas the term reactive power control is related to performing
voltage control [15, 17]. Frequency and voltage measurements are used to determine the
quality of power supply, therefore active and reactive power control are vital to achieving a
satisfactory performance [15, 17]. Frequency should remain nearly constant to ensure an almost constant speed at the induction and synchronous motors. A constant speed at the drives
is important, since the performance of the generation units depends on the performance of
the auxiliary drives associated with the fuel, feed-water and combustion [17].
In traditional power systems, the active power and frequency control is executed at control centers, where data about the system’s frequency and about the power flows at interconnecting lines are constantly measured and collected by a SCADA system. Each area is
equipped with an automatic generation control, which executes different actions according
to the measured control error. In conventional systems, active power control is classified
into three different control mechanisms known as primary control, secondary control and
tertiary control; all generators that operate above a certain power rate are required to participate in at least the primary control.
8
1 The Project: Introduction
Primary Control
Primary control refers to the control actions that take place after a change in the system’s
frequency has occurred. The frequency of an electrical network depends on the active power
balance [17]. A power unbalance occurs when there is a mismatch between the active power
that is generated and the active power that is consumed, thus a change in power demand at
one point in the network is reflected by a change in frequency in the whole system [17]. In
order to compensate for the change in frequency, a speed governor can be set with a droop
that follows a certain frequency-power characteristic [15]. In this way the mechanical power
supplied to the generator is either decreased or increased until the power balance is restored.
The per unit droop [p.u.] is given by (1.1):
Droop =
△ f/ frtd
△Pi /Pi,rtd
(1.1)
where △ f [Hz] is the frequency change in the system, frtd [Hz] is the system’s nominal rated
frequency, △Pi [MW] is the change in active power of generator i and Pi,rtd [MW] is the
rated power of generator i.
Secondary Control
After the primary control has been applied, the power balance is restored, but as a consequence, the system operates at a lower or higher frequency than the rated one. Secondary
control is used to modify the setting of the speed governor in a way that the frequency
is brought back to its rated value; this is done by temporarily increasing the prime mover
power that raises the kinetic energy of the generation unit [15]. In interconnected systems,
there are several areas involved. The Area Control Error (ACE) [MW] indicates the surplus
or lacking amount of power that has to be generated or injected to a particular area:
ACE j = P j,act − P j,sch + λ j ( fact − fsch )
(1.2)
where P j,act [MW] is the actual power export of control area j, P j,sch [MW] is the scheduled
power export of control area j, λ j [MW/Hz] is the network power-frequency characteristic
of control area j, fact [Hz] is the actual frequency and fsch [Hz] is the scheduled frequency.
Tertiary Control
Tertiary control is not necessarily applied consecutively after the secondary control actions
have been accomplished. Tertiary control is related to the economic dispatch of components: an optimal economic dispatch is calculated for each operating condition. The term
economic dispatch was described in Section 1.3.1.
The principles of these three types of control are still used nowadays, however as the participation of (micro-)cogeneration units and the participation of renewable energy sources
increase, the control mechanisms will require more flexibility and complexity. In Chapter 4 the hierarchical principle of this traditional control is used, however additional control
subsystems are defined in order to allow the control of multiple energy carriers.
1.3 Changes that Trigger the Evolution of Power Systems
9
1.3.3 Emergent Participants: Cogeneration, District Heating and Renewable Energy Technologies
There are three emergent participants that have played an important role in the evolution
of power systems, mainly because of their differences in relation to traditional generation
plants. These are: cogeneration technologies, district heating systems and renewable energy
technologies. In Section 1.2, potential benefits of integrating cogeneration and renewable
technologies were discussed, in the following subsections attention will also be given to the
challenges that these participants have created in traditional power system infrastructures.
Cogeneration Technologies
Combined heat and power or cogeneration units are defined as generation units that simultaneously generate electricity and useful heat from the same fuel input; the fuel can be coal,
biomass, natural gas, nuclear material, sun radiation or heat stored in the earth [18]. Some
of the benefits that have been associated with cogeneration technologies are the following
[18–20]:
• Cogeneration dramatically increases the energetic efficiency of the system.
• Due to the higher efficiency, reduction of carbon dioxide emissions and other pollutants can be achieved.
• It allows increased energy security through reduced dependence on imported fuel.
• Cogeneration promotes cost savings for the energy consumers.
• Decentralized CHP units reduce the need for transmission and distribution networks.
• Local energy resources can be encouraged, particularly through the use of biomass,
waste and geothermal resources in district heating and cooling (DHC) systems.
Due to the resulting combined efficiency (electrical efficiency and thermal efficiency),
CHP units allow 75% to 80% of the fuel input to be converted into useful energy, and up to
90% in highly efficient plants, therefore by using fuels in a more efficient way, both energy
costs and CO2 emissions can be reduced [18]. The main difference of a cogeneration plant
with respect to a traditional plant is that its useful output is not only electricity, but also heat.
Another difference is that domestic CHP (micro-CHP) units may be able to inject power into
the electrical grid, for example in the case when more electricity than necessary is produced.
These two characteristics increase flexibility in the energy system. Nevertheless, in order to
take advantage of this flexibility, new scheduling strategies are necessary, since the control
and optimization of the system should not overlook the heat production. This dissertation
emerged as a response to filling that gap.
District Heating Technologies
A district heating system consists of buildings, pipes, a pump station and a heat production
station, where heat can be produced from a geothermal field, combustion and combined
heat and power units, among other technologies. Nowadays, the heat demand of some domestic appliances, such as washing machines and dryers is mostly supplied with electricity.
10
1 The Project: Introduction
However, this practice is unfavorable since first class energy is used for a purpose where a
second class energy should be used instead, in spite of this, state-of-the-art policies do not
encourage district heating supply [21].
Large penetration of district heating systems has taken place particularly in Scandinavian countries, where they occupy over 50% of the heat market; however, district heating
comprises a small fraction of the total heat market of the European Union [22]. There is
still potential for large penetration in other countries, but national and international policies
should be adapted in order to promote it. Some of the main characteristics of district heating
are summarized below [23]:
• Existing district heating and cogeneration facilities reduce the global carbon dioxide
emissions from fuel combustion by 3-4% annually (in relation to traditional systems).
As a reference, the Kyoto Protocol sets a target of 5% average reduction per year in
industrialized countries.
• District heating systems are very suitable to be fed by cogeneration plants; this raises
the overall efficiency of power and heat production as mentioned above.
• District heating systems can be fed with energy coming from several sources, including industrial waste heat, heat from incinerators, geothermal energy and biomass,
among others.
Due to the fact that district heating systems can be fed from several different sources,
including CHP units, a multi-carrier scheduling strategy can be beneficial. In Chapter 4 a
district heating load is used in the illustrative example.
Renewable Energy Technologies
Renewable energy is energy derived from resources that are not substantially depleted by
continuous use. Ideally, these resources do not entail significant pollutant emissions or other
environmental problems, and do not involve the perpetuation of substantial health hazards or
social injustices [24]. In [15], the structural changes that will occur in existing distribution
and transmission networks due to a large-scale implementation of renewable energy sources
are attributed to four main differences with respect to traditional systems:
• Most renewable energy generators are connected to the distribution network, in contrast to traditional large-scale generators, which are connected to the transmission
network.
• Most renewable energy generators are connected to the electrical grid by means of
power electronic interfaces, in contrast to large generation plants which are coupled
to the electrical grid directly.
• The output of most renewable energy generators depends on natural and uncontrollable sources, in contrast to traditional plants, which are driven by controllable
sources like fossil fuels and hydro power, among others.
• The outputs of several renewable energy generators have an intermittent character,
which can lead to power fluctuations in the electrical grid. This does not apply to
traditional generators.
1.3 Changes that Trigger the Evolution of Power Systems
11
As it was mentioned earlier, according to the International Energy Agency, the participation of alternative sources such as geothermal, solar, wind and heat, in the total primary
energy supply will increase from 0,8% in 2009 to 11,8% if current policy scenarios are
kept unchanged and to 18,6% if a climate policy framework is applied to stabilize the concentration of global greenhouse gases at 450 ppm CO2 equivalent [6]. This means that
transmission and distribution operators should start adapting their monitoring, control and
optimization infrastructures in a short term in order to smoothly cope with the changes.
The Clean Energy Progress Report of 2011 shows the results of a sound analysis performed by the Clean Energy Ministerial Secretariat, run by the US Department of Energy
in relation to the progress that has been done towards clean energy implementation [25].
Several key findings were published in this report, from which the following were selected
due to the relation that can be established with this dissertation:
• Thanks to favorable policy support, solar PV and wind power are achieving strong
growth. However, in order to achieve sustainable energy goals a doubling of all
renewable energy use is required by 2020.
The fact that a doubling of all renewable energy use is required means that the development and implementation of suitable control systems is crucial. Chapter 4 includes
an illustrative example of a control strategy applied to a system with high penetration
of wind. Such studies are important to gain insights and be able to cope with the
different challenges that stochastically-varying sources bring.
• Progress has been made to transform the market for some key energy-efficient products, including compact fluorescent light bulbs. However, in the buildings and industry sectors, significant under-investment remains. Much more policy effort is needed
to capture the near- term profitable and low cost energy savings opportunities.
The scheduling tool described in Chapter 2 is a first step towards the development of
tools that can be applied to the building, residential and industry sectors in order to
show potential cost/energy saving opportunities and in this way attract investors.
• Electric vehicles are poised to take off. Major economies have announced targets that
together would reach about 7 million vehicle sales per year by 2020. However, this
will only account for about 2% of light-duty vehicle stocks worldwide. Fuel economy
of conventional light-duty vehicles will need to improve faster to achieve a global
target of 50% improvement by 2030 compared to 2005 levels.
Even though the topic of electric vehicles is only analyzed by means of an illustrative example in Chapter 5, it shows some insights about the participation of electric
vehicles at district level.
• Increased attention and resources are required to expand smart grid pilot projects on
a regional levels.
The framework presented in this thesis will be applied to a pilot project during the
years 2012 and 2013. The author of this dissertation has been working on this project
since January 2012. The results of the pilot project will provide valuable information
about further possibilities for implementation at local and regional levels.
12
1 The Project: Introduction
1.3.4 Role of Power Electronics in Future Power Systems
In conventional power systems, fully controllable generators are in charge of performing
voltage and frequency control. However, due to the increment in the participation of noncontrollable units, power electronic devices are taking part in performing these actions.
Power electronics allow flexible control of electrical power, they allow DC-to-AC (inverter)
conversion, as well as DC-to-DC, AC-to-DC and AC-to-AC conversions.
Decentralized generation units, such as fuel cells and especially the ones powered by
renewable energy sources with stochastically varying nature, like photovoltaic systems and
wind turbines, are often connected to the distribution network by power electronic interfaces
[15]. In distribution systems, like the ones addressed in this dissertation, the main applications are focused on the control of voltage and power flow, but also on the improvement of
power quality [26]. Two modes of operation of power electronic interfaces are: PQ control
and voltage source control. These mechanisms are briefly described below. For modeling
purposes, the general procedure adopted in the literature is to model converters according
to their control functions, this means that fast switching transients, harmonics and inverter
losses are neglected [27].
Voltage Source Control
When a power electronic interface operates in voltage source control mode, the converter
provides power in a way that the voltage level and frequency are kept at a reference level:
the converter operates as the master converter of the system. Small-scale power systems
that are not connected to the main electrical grid rely on a master converter for their voltage
and frequency levels.
PQ Control
When a power electronic interface operates in PQ control mode (active power - reactive
power control mode), the converter provides active and reactive power according to the
respective power set-point. In this case, the converter operates as slave.
1.3.5 Smart Grids: Intelligence in Future Power Systems
In the inaugural edition of the IEEE Smart Grid Newsletter, the term smart grid was defined
in the following way [28]:
. . . a smart grid involves the increased use of digital information and controls
technology to improve the reliability, security, and efficiency of the electrical
grid . . . (and) dynamic optimization of grid operations and resources, with full
cyber-security.
The infrastructure for information and communication technology (ICT) accounts for
3% of the world’s electricity usage, but its role in the future energy scenario seems to be
far greater than that [29]. Recent and future developments in power systems, such as smart
grids and intelligent buildings require ICT infrastructures; this makes energy supply infrastructures dependant on ICT.
1.3 Changes that Trigger the Evolution of Power Systems
13
The smart grid concept can be applied at any voltage level and the functionalities associated with the term are quite broad. IC technologies will become crucial actors in the
smart grid scenario [30]. However many gaps have to be filled before being able to implement smart grids in the current power system infrastructures. A clear description about the
current status in relation to ICT and the smart grid concept was found in [31]:
. . . currently there are no well-established modeling, analysis, or decision-making
paradigms in support of deploying information communications technology
(ICT) needed to facilitate new functionalities essential for sustainable energy
services. What is available are fragmented coarse models of socio-ecological
systems (SESs), climate change energy models, man-made electric power grids,
as well as fragmented approaches to ICT developed for other applications and
believed to be directly applicable to smart grid design and operations.
Several authors have already identified difficult challenges that generate from the introduction of smart grids. The ones that were considered most important within the context of
this dissertation were selected and summarized below:
• Data might reveal information about the presence of people at their home and about
the appliances they use. This might affect their privacy. Therefore, customers might
be unwilling to provide their information [32].
• Large amounts of data will be generated with the introduction of smart grid technologies. Techniques for managing, analyzing and acting on this data will need to be
developed [30]. Moreover, maintenance, management and storage of data may be a
tedious job [32].
• Customer gateways are prone to physical as well as cyber security risks. For this
reason, energy meters need proper shelter to be physically secure [32].
• Old power plants will not easily be switched in response to highly variable new plants
[31].
• Implementation of smart meter systems involves an investment of several billion dollars for deployment and maintenance [32].
• At the distribution level, it is likely that traditional customers will sell power back
to the electrical grid. Depending on the penetration of this kind of loads, the entire
distribution system protection and control infrastructure will need to be redesigned to
manage different flow patterns. Given that many components have aged, it is a good
time to rethink the design of future distribution systems [31].
From the list above, it can be concluded that still many steps need to be taken in order to
incorporate the smart grid concept into the current power system’s infrastructure. Moreover,
as earlier discussed in this chapter, cogeneration and district heating technologies are likely
to play an important role in the future due to the higher efficiencies that can be attained
with them. These technologies may be coupled to the smart grid optimization and control
platform. In most of the papers dealing with smart grids, only electrical flows are taken into
account, however in order to make a better use of the synergies provided by the couplings
between different energy carriers, smart grid strategies should also take other energy carriers
into consideration.
14
1 The Project: Introduction
1.4 Problem Definition
Three main concepts serve as platform for the problem definition of this dissertation. These
concepts are: the energy hub approach, exergy analysis as assessment tool for energy systems and intelligent energy management systems at district level. A brief discussion of each
aspect is given below.
1.4.1 The Energy Hub Approach
The energy hub approach was developed as part of the project “Vision of Future Energy
Networks - VoFEN” at ETH Zürich. The objective of the VoFEN project is to find optimal
structures for energy systems in the future. In the VoFEN project, the interaction and conversion possibilities between different energy carriers are considered to increase the flexibility
of energy supply systems. An energy hub is flexible in supply due to the fact that there
are different energy carriers available at its inputs and also by the fact that internal conversion and storage are possible [33]. This flexibility allows the use of optimization tools
to determine the best way to supply a load, after taking the constraints of the system into
account.
An energy hub is a unit where multiple energy carriers are converted, conditioned and
stored [34–37]. It can serve as an interface for different energy infrastructures and/or loads
[33]. The energy hub shown in Figure 1.2 contains a hybrid input port with electricity and
natural gas as energy carriers, and a hybrid output port with electricity and heat as products.
The couplings that exist among the inputs and outputs are contained inside the energy hub.
Pe
Pg
Le
v gA Pg
A
v gB Pg
B
v gC Pg
C
Lq
Figure 1.2: Energy hub representation
Depending on its functionality, there are three types of elements that an energy hub can
contain: direct connections, converters and storage elements [33]. Direct connections are
elements that deliver an input carrier to the output port without converting it into another
energy form or changing its quality in a significant way. Converter elements transform energy carriers into different energy forms or qualities. Storage elements are used to represent
both direct and indirect storage of energy carriers [33]. The description and mathematical
representation of the energy hub is presented in Chapter 2.
1.4 Problem Definition
15
The following problems have already been solved using the energy hub approach: multicarrier optimal dispatch, multi-carrier optimal power flow, optimal hub coupling and optimal hub layout. Those problems were covered in the dissertation “Integrated Modeling
and Optimization of Multi-Carrier Energy Systems” [33]. The problems of scheduling (unit
commitment) and real-time control of such systems also require a flexible framework since
the flow interactions in systems with multiple energy carriers create challenges in terms
of planning, scheduling and control. Since these problems have not been solved in the
literature for systems with multiple energy carriers, the development of a suitable unit commitment framework for these systems was defined as one of the problems to be solved in
this dissertation. The energy hub approach is therefore used as platform.
1.4.2 Exergy Analysis as Assessment Tool for Energy Systems
Due to the increasing interest in promoting energy conservation, exergy analysis has the
potential to become an important tool for the study and design of energy plants and systems.
Exergy analysis can be used to evaluate the potential to produce work throughout the system.
This can provide a proper measurement of the losses in the system, which is necessary to
achieve an effective energy conservation during the system’s design and operation [38].
According to [38], three aspects that have prevented engineers from performing exergy
analyses are:
• The analysis presented in books is normally based on the first law of thermodynamics.
• Examples of the second law of thermodynamics are usually limited to simple processes and cycles, where the benefits of using the second law are not apparent.
• The design and operation conditions for energy plants have usually been based on
initial costs and not on taking the most advantage of the source, where the second law
of thermodynamics plays a role.
During the last decade, the interest in using exergy for the analysis of systems has increased. At Delft University of Technology, several efforts have been made to support activities in the field of exergy. A trigger for the research presented in this dissertation was an
example published in the area of built environment [39, 40]. This example is part of the dissertation “Exergy Efficient Building Design” defended at Delft University of Technology,
which provides insights about the possibilities of using exergy analysis as assessment tool
in the built environment [40]. The author analyzes the effectiveness of using an electricitydriven heat recovery unit in a dwelling ventilation system from an exergy perspective.
The author presented a steady-state energy and exergy analysis for a dwelling ventilation
system with and without the use of a heat recovery unit and compared the results. From the
results, he concluded that it could make sense to use the heat recovery unit only when the
environmental temperature is low enough to compensate for the electricity input, which has
a high exergy value [39]. Thus, it can be inferred from the example that from an exergetic
point of view, the heat recovery unit has to be scheduled only when the heat demand is high
due to low outside temperatures, but from an energetic point of view this is not the case.
As previously claimed by other authors [41, 42], the author stated that an exergy analysis provides a common basis to evaluate systems that contain heat and electricity flows
16
1 The Project: Introduction
despite their different abilities to produce work in relation to a given environment. The
study opened a path for further research. After revising the aforementioned work, the challenge in this dissertation became to provide a tool that would not only be able to schedule
the units beforehand (contrary to the case in which the best schedule is selected after having
compared the results of selected scenarios), but that could also provide the optimal dispatch
of the units involved according to the exergetic/energetic efficiency of the system. Another
aspect that was considered important for the optimization tool was to allow the inclusion of
several units and multiple energy carriers. The result of the work is condensed in Chapter 3.
As mentioned in Section 1.2, one of the main goals to be reached by including renewable
and combined heat and power technologies is to achieve a more efficient and sustainable
use of resources. Exergy analysis can be used to calculate the available work throughout a
system and to identify losses; this provides valuable information for the planning of the next
generation energy supply systems. Due to its potential as assessment tool and because there
is very limited literature about the use of exergy analysis in systems with multiple energy
carriers, this was defined as one of the problems to be addressed in this dissertation.
1.4.3 Intelligent Energy Management Systems at District Level
Several research institutes, professional societies and governmental agencies have decided
to incorporate and promote the development of concepts related to smart grids in their portfolio. The topic of intelligent energy management systems falls under the broad coverage of
smart grids. As discussed earlier in this chapter, one of the topics that still needs to be explored is the feasibility of developing and implementing an intelligent energy management
system for small-scale systems containing multiple energy carriers at district level. Furthermore, special focus must be given to scheduling of the generation units involved. The
scheduling tool to be described in Chapter 2 allows the analysis of different load scenarios
and system configurations that have arisen in response to several emerging trends in power
systems. The trends to be included in the analysis are:
• The growing participation of micro-CHP units
• The application of the virtual power plant (VPP) concept, in which several micro
generation units are controlled by an aggregator to achieve an optimized operation
• The inclusion of storage
• The inclusion of renewable sources
• The inclusion of electric vehicles.
The application of the scheduling tool for the analysis of these aspects was selected as
one of the targets of this dissertation. Additionally, a general control scheme to be applied in
systems with multiple energy carriers is presented in Chapter 4. The results of the comparisons among scenarios are presented in Chapter 5. The control scheme is part of the energy
management system described in Chapter 6. The examples presented in this dissertation
refer to residential loads.
1.5 Research Objective
17
1.5 Research Objective
1.5.1 Main Objective
The main research objective is based on the three concepts discussed in the problem definition. The resulting formulation is:
To develop a scheduling tool for small-scale systems that contain multiple energy carriers at district level (residential) by using the energy hub approach
and to evaluate the use of an exergy analysis as assessment tool for such systems.
1.5.2 Research Questions
The following research questions are based on the main research objective. The first research question is the main research question of this work and therefore it is answered
throughout all the chapters of this dissertation.
1. What kind of optimization tool can be used for the scheduling and control of systems
containing multiple energy carriers in residential areas?
2. What framework can be adopted in order to schedule and optimize the units involved
in an energy supply system with multiple energy carriers at district level?
3. What is the potential relevance of using the exergy concept to analyze energy supply
systems with multiple energy carriers?
4. What kind of real-time control strategy can be applied in a system containing multiple
energy carriers at district level?
5. How can the scheduling tool described in this dissertation be used for the analysis of
emerging trends in power systems and what benefits can be obtained from it?
6. What design can be proposed for a multi-carrier energy management system and can
it be implemented at DENlab?
Each chapter in this dissertation is dedicated to one of the listed research questions.
In order to answer Research Question 6, a partial physical implementation in the renewable energy laboratory DENlab was performed. More information about this laboratory is
presented in Chapter 6.
18
1 The Project: Introduction
1.6 Overview of this Dissertation
1.6.1 Outline
This dissertation consists of seven chapters. This section concludes Chapter 1, in which the
project framework, motivation, problem definition and research questions were introduced.
Chapter 2 describes the model behind the scheduling tool that was developed for this PhD
project, which is used to optimize systems containing multiple energy carriers. Later in
Chapter 3 the optimization tool is adapted to include exergetic efficiency as assessment
parameter of such systems. Chapter 4 presents the control architecture that was designed to
cope with the dynamic behaviour of the systems under study. The application of the control
architecture is shown by means of an example. In Chapter 5 the optimization tool is used to
analyze the impact of different emerging trends in district-level power systems, such as the
active participation micro-CHP technologies, the incorporation of renewable sources and
the application of the virtual power plant concept. Chapter 6 provides insights regarding
the implementation of the optimization and control tool in real systems. Finally, Chapter 7
presents the conclusions and recommendations of this dissertation.
1.6.2 Main Contributions
The main contributions of this dissertation are listed below:
• A general multi-carrier unit commitment framework for energy systems that contain
multiple energy carriers was developed. The framework can be used with any kind of
energy carrier and for different possible couplings and power scales (Chapter 2).
• A technique to include storage was developed and implemented as part of the optimization tool. The results show that this technique can be valuable for peak-shaving
purposes at the generation side (Chapter 2).
• The exergy hub approach was introduced. The exergy hub provides a visual indication of the exergetic efficiency of the units. In Section 3.4 both the energy hub and
the exergy hub are depicted next to each other in order to reveal that a unit that is considered to be very efficient from an energy point of view can be considered to be very
inefficient from an exergy perspective. Some authors consider that energy efficiencies
can be misleading [43], thus the exergy hub representation could serve as a way to
avoid this (Chapter 3).
• A comparison between the results for the optimal dispatch obtained from an exergetic
efficiency optimization and from an energetic efficiency optimization is performed.
In the literature, this kind of comparison has not been performed in the context of
energy supply systems containing multiple energy carriers (Chapter 3).
• The results of the scheduling optimization tool are compared to identify which configuration gives the best energy and exergy performances for specific loads. A sensitivity
analysis is performed in which the ratio between heat and electricity consumption is
varied to observe the influence of the type of load in the scheduling. In the literature,
the attempts that have been made to analyze systems containing several generation
1.6 Overview of this Dissertation
19
units from an exergy perspective are very limited and they do not focus on finding
an optimal scheduling by means of a mathematical optimization algorithm as done in
this dissertation (Chapter 3).
• A two-level control strategy was designed for the application in systems with multiple energy carriers. Most of the control strategies found in the literature only focus
on electricity flows, thus the strategy proposed is valuable for multi-carrier systems
(Chapter 4).
• Several emergent trends were simulated to provide insights regarding their impact in
district-level residential energy supply systems. The optimization tool was extended
in order to study the benefits of having an aggregator in charge of the optimization
(Chapter 5).
• A comparison among three micro-CHP technologies was presented to show that different benefits can be obtained by using combined heat and power technologies with
different electricity-to-heat efficiency ratios (Chapter 5).
• An example was presented in which the influence of incorporating electric vehicles
at a neighbourhood was analyzed (Chapter 5).
• A design of a multi-carrier energy management system was presented. A partial
implementation in the renewable energy laboratory DENlab was performed, which
provides an added value to the theoretical results that were accomplished in this dissertation (Chapter 6).
Chapter 2
The Framework
Multi-Carrier Unit Commitment
This chapter answers Research Question 2. It presents the optimization framework, from
now denoted multi-carrier unit commitment, that was developed to schedule and optimize
controllable units in energy systems with multiple energy carriers. The algorithms were
programmed in the optimization software for research applications AIMMS [44].
Section 2.1 contains basic definitions related to the framework. Section 2.2 presents a
summary of work that has been done in relation to modeling systems with multiple energy
carriers. In Section 2.3, the multi-carrier unit commitment framework is modeled and described. Section 2.4 contains simulation results to illustrate the optimization tool. Finally
in Section 2.5 the conclusions of this chapter are presented. Parts of this chapter have been
published in [1, 45].
2.1 Basic Concepts and Definitions
2.1.1 Energy Hub Element
There are three types of energy hub elements: direct connections, converters and storage
elements [33]. Direct connections deliver input carrier α to the output port without converting it into another energy form. Conversely, converter elements transform energy carrier
α into a different energy carrier β. Finally, storage elements consist of an interface and an
internal (ideal) storage. Through the interface, power in the form of energy carrier α may
be conditioned and/or converted into energy carrier γ, which is then stored internally. It is
assumed that when a storage element exchanges energy carrier α, the element is considered
a storage element of energy carrier α, even if energy carrier γ , α is stored internally [33].
2.1.2 Multi-Carrier Optimal Dispatch
Multi-carrier optimal dispatch is a method to determine an optimal operation policy for a
number of converter units processing multiple energy carriers [33]. This term is closely
linked to the energy hub concept, which was introduced in Section 1.4.1.
21
22
2 The Framework: Multi-Carrier Unit Commitment
2.1.3 Multi-Carrier Unit Commitment
Multi-carrier unit commitment is introduced in this dissertation as the computational procedure that makes scheduling decisions in advance about the units that must be committed to
supply a forecasted load in energy systems containing multiple energy carriers. The procedure determines the sequence in which the start-up and shut-down of the units (energy hub
elements) should be executed as well as the optimal dispatch of each committed unit at each
time period.
2.2 Modeling of Multi-Carrier Energy Systems
This section introduces the modeling concepts that serve as the base for the development of
the multi-carrier unit commitment framework. In the next few pages, the work published
in [33, 36, 37] is condensed, thus this section presents a summary of works performed by
other authors in relation to modeling systems with multiple energy carriers. Even though
the model presented is based on the literature, a modified nomenclature is introduced to
describe the model. This modified notation is more consistent and general; this allows the
model to be easily extended, as done in Section 2.3, in Chapter 3 and in Chapter 5.
The main constraints, assumptions and equations that are necessary to solve the multicarrier optimal dispatch problem are included in this section. Moreover, a simple example
is added to illustrate the concepts.
2.2.1 Constraints and Assumptions
The following assumptions are considered:
• If not mentioned explicitly, it is assumed that there is a unidirectional power flow
from the input to the output ports.
• Losses only occur at the converter elements inside the energy hub. Connecting networks are assumed to be lossless if not mentioned explicitly.
• An energy converter device is represented as a black box characterized by its energetic
efficiency or a function representing the input-output dependency.
• Even though the notation is based on the notation presented in [33], several changes
were introduced to allow a more precise definition of the variables.
2.2.2 Energy Hub Conversion Model
The energy conversion model proposed in [33] is used in this chapter without significant
alterations apart from the notation. For a system with one input energy carrier α and one
output energy carrier β, the input and output power flows are coupled by the coupling factor
cαβ [-], as shown in (2.1) [33, 36, 37]:
Lβ = cαβ Pα
(2.1)
2.2 Modeling of Multi-Carrier Energy Systems
23
where Pα [kW] is the steady-state input power flow and Lβ [kW] is the steady-state output
power flow of the energy hub. Due to the conservation of power, the output power of the
converter must be equal or smaller than the input power, thus the coupling factor is limited
by the following constraint:
Lβ ≤ Pα ⇒ 0 ≤ cαβ ≤ 1.
(2.2)
In the case of multiple energy carriers, the input energy carriers are denoted by α1 ,
α2 , . . ., αnin , while the output energy carriers are denoted by β1 , β2 , . . . , βnout . In general the
following notation will be used for the input and output energy carriers respectively: αi and
β j , where i = 1, 2, . . . , nin , j = 1, 2, . . . , nout , nin ∈ N and nout ∈ N . The variables αi and
β j represent specific energy carriers from the sets Ein = {e, g, q, . . . } and Eout = {e, g, q, . . . }
that may include electricity ‘e’, gas ‘g’, heat ‘q’, among others energy carriers. The matrix
that represents a system with multiple energy carriers is shown in (2.3):

Lβ1   cα1 β1
 
Lβ2   cα1 β2

..  =  ..
.   .
 
Lβn
cα1 βnout
out
| {z } |







L
cα2 β1
cα2 β2
..
.
cα2 βnout
{z
C


. . . cαnin β1   Pα1 

 
. . . cαnin β2   Pα2 


..   .. 
..
.
.   . 


. . . cαnin βnout Pαnin .
} | {z }
(2.3)
P
When several converters k p are considered, the set of converters Cαi = {A, B, . . .} is
introduced, where p = 1, 2, . . . , ncon and ncon ∈ N. Each converter k p has an energy carrier
αi as input. Therefore, when energy carrier αi serves as input for several hub elements or
k
converters, dispatch factors vαpi [-] are introduced. If the total input power Pαi [kW] splits
k
into NCαi converters, dispatch factor vαpi specifies the percentage of input power Pαi that
k
flows into converter k p [33]. The input power of the converter is denoted by Pαpi [kW] and
given by:
k
k
Pαpi = vαpi Pαi .
(2.4)
A general expression of coupling factor cαi β j [-] is given in (2.5). The expression takes
into account the participation of each converter k p , as shown:
X k k
cαi β j =
vαpi ηαpi β j
(2.5)
k p ∈Cαi
k
where vαpi is the dispatch factor that represents the percentage of input power Pαi that is
k
provided to converter k p and ηαpi β j [-] is the efficiency of conversion from energy carrier αi
to β j of converter k p . The efficiency of conversion can also be given as a function of the
k
input power flow: Fβ jp (Pαi ) [-]. Likewise, the coupling factor cαi β j results from adding the
individual contributions of all converters that produce energy carrier β j at their output:
X k k
cαi β j =
vαpi Fβ jp (Pαi ).
(2.6)
k p ∈Cαi
The constraints related to the dispatch factor are given by:
X k
k
0 ≤ vαpi ≤ 1 and
vαpi = 1.
k p ∈Cαi
(2.7)
24
2 The Framework: Multi-Carrier Unit Commitment
Figure 2.1 shows an energy hub in which a single input power Pαi is provided to several
converters k p .
k
vα1 Pα
i
i
Pαi
k
vα2 Pα
i
i
kn
vα con Pα
i
i
k k
vα2 η 2 Pα
i αi β j i
k
η 2
αi β j
k2
kn
k k
vα1 η 1 Pα
i αi β j i
k
η 1
αi β j
k1
η
con
cα β Pα
i j i
Lβ j
kn
kn
vα con η con Pα
αi β j i
i
kncon
αi β j
Figure 2.1: Representation of an energy hub with multiple converters
2.2.3 Multi-Carrier Optimal Dispatch Model
This subsection summarizes the problem statement of the multi-carrier optimal dispatch according to [33, 36]. The problem statement includes the objective function to be minimized
as well as the equality and inequality constraints.
For an energy hub, the power balance equality constraint states that the output power
vector L is equal to the coupling matrix C multiplied by the respective input power vector
P. This equality constraint is expressed in (2.8):
L − CP = 0.
(2.8)
The equality constraint for the dispatch factors is shown in (2.9) (derived from (2.7)):
X k
1−
vαpi = 0.
(2.9)
k p ∈Cαi
The inequality constraints for the problem statement are given by the lower and upper
limits of the hub power inputs (Pαi , Pαi ) [kW], the lower and upper limits of the individual
k
kp
converters’ power inputs (Pαpi , Pαi ) [kW] and the limits of the dispatch factors:
Pαi ≤ Pαi ≤ Pαi
k
k
Pαpi ≤ vαpi Pαi ≤
0≤
k
vαpi
≤ 1.
(2.10)
kp
Pαi
(2.11)
(2.12)
2.2 Modeling of Multi-Carrier Energy Systems
25
The objective function Fobj [e/(time period)] for the multi-carrier optimal dispatch is a
function of the energy hub’s variables. For example, when the operational costs are minimized, the objective function can be modeled as a quadratic function of the respective input
powers, where K1,αi [e/(time period)], K2,αi [e/(time period · kW)] and K3,αi [e/(time period · kW2 )] are cost coefficients:
Fobj =
X
(K1,αi + K2,αi Pαi + K3,αi P2αi ).
(2.13)
αi ∈Ein
Example: Multi-Carrier Optimal Dispatch Coupling Matrix Formulation
The example shows how to obtain the factors of the coupling matrix. The energy hub in
the example contains 3 converters and 1 direct connection as shown in Figure 2.2. The hub
inputs are natural gas and electricity coming from the electrical grid. The energy hub output
consists of an electric load and a heat load. Therefore, the input carrier set is defined as
Ein = {e,g}, where ‘e’ stands for electricity coming from the grid and ‘g’ stands for natural
gas. On the other hand, the output carrier set is defined as Eout = {e,q} where ‘e’ stands
for electricity and ‘q’ stands for heat. The set of converters is defined as Cαi = {A,B,C}.
Converter A and converter B represent two gas-fired CHP units and converter C represents
a gas-fired furnace. The direct connection represents the public electrical grid.
From Figure 2.2, it can be observed that there is no gas output Lg or any heat input Pq
at the energy hub under consideration, thus they are not included in the equality constraint
matrix shown in (2.14):
#" #
" # "
c
cge Pe
Le
= 0.
(2.14)
− ee
ceq cgq Pg
Lq
There is no conversion from electricity to heat, thus the corresponding coupling factor
ceq is equal to zero. There is a direct connection from the electricity input Pe to the electricity
output Le , for this reason the coupling factor cee is equal to 1:
ceq = 0, cee = 1.
Pe
Pg
(2.15)
Le
v gA Pg
A
v gB Pg
B
v gC Pg
C
Lq
Figure 2.2: Energy hub representation for the example
26
2 The Framework: Multi-Carrier Unit Commitment
The conversion from gas to electricity is performed by converter A and converter B.
B
Their individual contributions depend on the dispatch factors vA
g and vg and the efficienA
B
cies of the converters, given by ηge and ηge respectively. The coupling factor cge (gas to
electricity) results from the addition of both contributions (see (2.5)):
A
B B
cge = vA
g ηge + vg ηge .
(2.16)
The conversion from gas to heat is performed by converters A, B and C. Similarly, the
coupling factor cgq (gas to heat) results from the addition of their individual contributions:
A
B B
C C
cgq = vA
g ηgq + vg ηgq + vg ηgq .
(2.17)
2.2.4 Multi-Carrier Optimal Dispatch Model Including Storage
In accordance to the definition of storage element given in Section 2.1.1, a storage element
that exchanges power in the form of energy carrier αi is considered to be a storage element of
αi , even if another carrier is stored internally. The energy content inside the storage element
increases the amount of Ėαi △t during time period △t, where Ėαi [kW] is the power flowing
into the storage element. In order to determine the power Ėαi that is added or subtracted
from the storage element, the power Dαi [kW] flowing to the storage element, is multiplied
by the efficiency factor eαi [-]:
Ėαi = eαi Dαi .
(2.18)
The factor eαi is considered as the charge/discharge efficiency and it depends on the
direction of the power flow, i.e., if the storage element is being charged (e+αi ) or discharged
(e−αi ) [33]:


e+αi ,
if Dαi ≥ 0 (charging or stand-by)

eαi = 
(2.19)

1e−α , otherwise (discharging).
i
Storage elements can be placed between converters, at the hub’s input side or at the hub’s
output side. The following equations can be obtained from Figure 2.3. From the figure it
can be observed that power Dαi flows to a storage element placed at the hub’s input side and
power Mβ j [kW] flows to a storage element placed at the hub’s output side. Therefore, the
eαi [kW] and output power e
input power P
Lβ j [kW] of the converter cluster (represented by
the shaded area) are given by:
eαi = Pαi − Dαi
P
(2.20)
e
Lβ j = Lβ j + Mβ j .
(2.21)
[L + M] = C [P − D]
| {z }
| {z }
(2.22)
L = CP − Meq
(2.23)
In vector form, the balance equation around the converter cluster is:
which can be rewritten as [33]:
e
L
e
P
2.2 Modeling of Multi-Carrier Energy Systems
27
Figure 2.3: Storage elements in an energy hub
where
Meq = CD + M.
(2.24)
In this way all storage devices are mapped to the output side of the energy hub. The
storage coupling matrix S contains all the couplings between the equivalent storage vector
Meq and the vector Ė, which contains the changes in energy content of the storage elements.
The matrix that results is:
 eq  


sα2 β1 . . . sαnin β1   Ėα1 
 Mβ1   sα1 β1

 
 eq  
sα2 β2 . . . sαnin β2   Ėα2 
 Mβ2   sα1 β2


(2.25)
 .  =  .
..   .. 
..
..
 ..   ..
.
.   . 
.
 eq  


Mβn
sα1 βnout sα2 βnout . . . sαnin βnout Ėαnin
out
|
{z
}
|
{z
}
| {z }
S
Meq
Ė.
Thus, after applying (2.25), the power balance equation can be reformulated as follows:
L = CP − SĖ
(2.26)
where vector L contains the hub’s power outputs, C contains the coupling factors, P represents the vector of the energy hub’s power inputs, S is the storage coupling matrix and
vector Ė contains the changes in energy content of the storage elements.
In the case of multi-period problems, time variable t is introduced. The power balance
equality becomes:
(2.27)
Lt = Ct Pt − St Ėt
where Ėt contains the changes in energy content of the storage elements within period t. For
a storage element that exchanges energy carrier αi , the respective equation is [33]:
Ėαt i = Eαt i − Eαt−1
+ Eαstbi
i
(2.28)
where Eαt i is the stored energy at the end of period t, and Eαstbi represents the stand-by energy
losses per period.
In this case, the equality constraint for the dispatch factors is:
X k ,t
1−
vαpi = 0.
(2.29)
k p ∈Cαi
28
2 The Framework: Multi-Carrier Unit Commitment
The inequality constraints for the multi-period multi-carrier optimal power flow that
result from adding the storage constraints are:
P ≤ Pt ≤ P
(2.30)
kp
k ,t
k
Pαpi ≤ vαpi Ptαi ≤ Pαi
0≤
k ,t
vαpi
(2.31)
≤1
Dαi ≤
Dtαi
Mβ j ≤
Mβt j
(2.32)
≤ Dαi
(2.33)
≤ Mβ j
(2.34)
Ė ≤ Ėt ≤ Ė
0
(2.35)
nt
E −E =0
(2.36)
where t = 1, 2, . . . , nt , and nt ∈ N. Energy carriers αi are stored at the input side of the
hub, and energy carriers β j are stored at the output side. The last constraint states that the
energy content at the initial time (t = 0) must be equal to the energy content at the last time
period (t = nt ). This constraint is added in order to bring the storage content at the end of
the simulation back to the initial storage content.
2.3 Multi-Carrier Unit Commitment Framework
This section provides an extension to the work presented in the PhD dissertation “Integrated
Modeling and Optimization of Multi-Carrier Energy Systems” [33], which was briefly introduced in Section 1.4.1. In this work, the author presented a general steady-state modeling
and optimization framework using the energy hub approach for the following problems:
multi-carrier optimal dispatch, multi-carrier optimal power flow, optimal hub coupling and
optimal hub layout. The main contribution of this chapter is to extend the model in order to
solve the multi-carrier unit commitment problem.
2.3.1 Multi-Carrier Unit Commitment Model
The problem statement described in Section 2.2.3 is used to solve a multi-carrier optimal
dispatch problem for a defined operating point. An extension is necessary in order to solve
the multi-carrier unit commitment problem, where the scheduling of units is involved. One
of the main differences in relation to the problem statement presented in Section 2.2.3 is
that the multi-carrier unit commitment does not deal with the power inputs of the energy
hub, but in fact with the power inputs of the energy hub elements. Therefore, a new binary
variable is introduced and the problem statement of the multi-carrier optimal dispatch is
modified, as described later in this section.
The new variable specifies the status (on/off) of the energy hub elements; the correk
sponding symbol is wαpi . The new variable is included in the coupling factors as follows:
cαi β j =
X
k p ∈Cαi
k
k
k
wαpi vαpi ηαpi β j or cαi β j =
X
k p ∈Cαi
k
k
k
wαpi vαpi Fβ jp (Pαi ).
(2.37)
2.3 Multi-Carrier Unit Commitment Framework
29
k
The status variable wαpi is equal to 1 when the hub element is turned on and is equal to
0 when the hub element is turned off. The equality constraint that deals with the dispatch
factors depends now on the converters that are running, thus it becomes:
X k k
1−
wαpi vαpi = 0.
(2.38)
k p ∈Cαi
Regarding the inequality constraints, the minimum power input limit of converter k p is
k
now multiplied by the status variable wαpi . In this way, the optimization program allows the
input power to become zero when the respective energy hub element is turned off, however
k
when it is turned on, the lower limit remains Pαpi :
k
k
k
kp
wαpi Pαpi ≤ vαpi Pαi ≤ Pαi .
(2.39)
The inequality constraint of the hub power input also depends on the status variables.
The minimum power input limit of energy carrier αi is now multiplied by the binary variable
wαi , which is defined in terms of the status variables as follows:
P
wαi Pαi ≤ Pαi ≤ Pαi

P
kp


0, if k p ∈Cαi wαi = 0
where wαi = 

1, otherwise.
(2.40)
When all NCαi converters associated with a specific energy carrier αi are turned off,
k
wαpi = 0, the optimization program allows the input power of this particular energy
carrier to be zero. Conversely, when at least one converter associated with this energy carrier
P
k
is running, k p ∈Cαi wαpi , 0, the lower limit remains Pαi .
k p ∈Cαi
The objective function must be modified accordingly. For example, in the case of conkp
sidering economic cost as the objective function, coefficient K4,α
can be included to repi
resent costs associated with the individual converters, as for example no-load costs. The
resulting equation is:
Fobj =
X
αi ∈Ein
(K1,αi + K2,αi Pαi + K3,αi P2αi ) +
X
k
k
p
(K4,α
wαpi ).
i
(2.41)
αi ∈Ein ,k p ∈Cαi
The recursive algorithm known as forward dynamic programming is used to compute
the minimum cost in the time interval T dp with combination Idp , as given in (2.42) [46]:
Fcost (T dp , Idp ) = min[Fprod (T dp , Idp ) + Ftran ((T dp − 1), Jdp : T dp , Idp ) + Fcost ((T dp − 1), Jdp )]
(2.42)
where
(T dp , Idp )
combination Idp at time period T dp
((T dp − 1), Jdp )
combination Jdp at time period (T dp − 1)
Fprod (T dp , Idp )
production cost for state (T dp , Idp )
Fcost (T dp , Idp )
least total cost to arrive at state (T dp , Idp )
Ftran ((T dp − 1), Jdp : T dp , Idp ) transition cost from state ((T dp − 1), Jdp )
to state (T dp , Idp ).
30
2 The Framework: Multi-Carrier Unit Commitment
In the case of 4 units, arbitrary examples of unit combinations can be: Idp = [1001],
Idp = [0011], Idp = [1111], etc. The unit is operating (committed) when the digit assigned
is 1.
A strategy is the transition or path from one state at a given time period to a state at
the next time period [46]. Two more variables that are used for the multi-carrier unit commitment are Xdp and Ndp , where Xdp is the number of states that are evaluated within each
period and Ndp is the number of strategies that are saved at each period. These variables
have influence on the computational effort [46]. If fixed costs are considered, they must be
included in the equation, but outside the brackets.
Example: Multi-Carrier Unit Commitment Matrix Formulation
The example is based on the energy hub shown in Figure 2.2. It shows the coupling factors
and the constraints of the multi-carrier optimal dispatch problem after being adapted to suit
the multi-carrier unit commitment problem. In the example, there is no conversion from
electricity to heat, thus the corresponding coupling factor ceq is equal to zero. The power
balance equality constraint for the example is:
#" #
" # "
c
cge Pe
Le
= 0.
(2.43)
− ee
0 cgq Pg
Lq
The coupling factor cee , cge and cgq are now:
cee = we
cge =
cgq =
A A
B B B
wA
g vg ηge + wg vg ηge
A A
B B B
wA
g vg ηgq + wg vg ηgq
(2.44)
(2.45)
+ wCg vCg ηCgq .
(2.46)
The dispatch factor equality constraint for the example is:
A
B B
C C
1 − (wA
g vg + wg vg + wg vg ) = 0.
(2.47)
The constraints in the problem statement are:
"
# " # " #
we Pe
P
P
≤ e ≤ e
(2.48)
wg Pg
Pg
Pg
   A  
0 vg  1
   B   
(2.49)
0 ≤ vg  ≤ 1
0
1
vCg
 A A   A   A 
wg Pg  vg Pg  Pg 
 wB PB  ≤ vB P  ≤  B 
(2.50)
 g g   g g   Pg 


C
C C
C


v
P
wg Pg
g g
Pg
where the value of the status variable we (related to the electric input Pe ) is always 1, since
it represents the connection to the public electrical grid at all times. The power flow Pe at
the input is bidirectional, which means that electricity can be sold back to the grid when the
CHPs produce more electricity than required by the load Le . Other restrictions related to
the multi-carrier unit commitment problem that can be considered are the minimum up and
minimum down times for the converters. The start-up costs of the converters are included
in the optimization under transition costs in (2.42) .
2.3 Multi-Carrier Unit Commitment Framework
31
2.3.2 Multi-Carrier Unit Commitment Technique to Include Storage
This section presents a technique to include storage as part of a general unit commitment
framework for energy systems with multiple energy carriers. The multi-carrier unit commitment problem is solved using forward dynamic programming.
The technique is depicted in Figure 2.4. Four consecutive steps are involved. At first, a
unit schedule is obtained after running the multi-carrier unit commitment problem without
considering storage, just as described in Section 2.3.1. This first schedule is considered as
the starting point for Step 2.
At Step 2, the storage units are included as part of the system, as described in Section 2.2.4. At this stage, the program assigns a new dispatch to the scheduled generation
units and the storage units. Thus, if there are units to be discarded after the inclusion of
storage, it is expected that these units will dispatch at (close to) minimum capacity.
At Step 3 a fictitious bidirectional load is included instead of the storage units. This
bidirectional load is loaded in accordance to the power values of the storage units that were
obtained in Step 2 for each time period. The program runs the multi-carrier unit commitment
again and a new schedule is obtained. It is expected that the units that were brought to their
minimum capacity at Step 2 will now be turned off, if no minimum up or minimum down
constraints are violated.
Step 4 runs a new multi-carrier optimal dispatch, including storage according to Section 2.2.4. This is considered the final solution.
start
STEP 1: Solve the multi-carrier unit commitment problem using dynamic programming without including storage according to Section 2.3.1.
STEP 2: Use the scheduling solution of Step 1 and solve the multicarrier optimal dispatch including storage according to Section 2.2.4.
STEP 3: Incorporate the storage power flow results obtained at
Step 2 as an additional bidirectional load and solve the multicarrier unit commitment problem according to Section 2.3.1.
STEP 4: Use the scheduling solution of Step 3 and solve the multicarrier optimal dispatch including storage according to Section 2.2.4.
end
Figure 2.4: Flow chart of the proposed technique
32
2 The Framework: Multi-Carrier Unit Commitment
2.4 Simulation Results
This section presents the results of two representative simulations. In the first simulation, no
storage is included. In the second simulation, heat storage is considered. All the algorithms
used in the simulations follow the descriptions that were presented in this chapter.
2.4.1 Simulation 1: Multi-Carrier Unit Commitment without Storage
In order to keep continuity in this chapter, the energy hub that was used in examples 1 and 2
is also used for the simulations contained in this section. The diagram of the energy hub can
be observed in Figure 2.5. The system is connected to the electrical grid through input Pe .
The data for electricity and gas costs were obtained from a German website that specifies
real prices for small commercial customers [47]. The data can be found in Table 2.1. The
multi-carrier unit commitment in this section is specified for time periods of 15 min, for this
reason the data found in the website, specified in hours, were adapted to fit the selected time
periods, as shown in Table 2.2.
Pe
Pg
Le
v gA Pg
A
v gB Pg
B
v gC Pg
C
Lq
Figure 2.5: Energy hub representation for Simulation 1
Table 2.1: Prices for small commercial costumers at Nordhausen, Germany
Energy
Carrier
Electricity
Gas
Heat
Costs
Use per year
(kWh)
Fixed
(e)
Consumption
(e/kWh)
30 001 - 100 000
2 375 - 12 692
not specified
132,65
73,50
30,68
0,1897
0,0684
0,0558
2.4 Simulation Results
33
Table 2.2: Prices for small commercial costumers used in this case study
Energy
Carrier
Electricity
Gas
Heat
Costs
Use per year
(kWh)
30 001 - 100 000
2 375 - 12 692
not specified
Consumption
e
kW·(15 min)
0,0474
0,0171
0,0140
Table 2.3: Parameters of the components used in this case study
Component
Parameter
Unit
Maximum electrical output
Minimum electrical output
Maximum change in power
Start-up duration
Minimal down time
Minimal up time
Maximum heat power
Start-up cost
No-load cost
kW
kW
%Pαi ,max
min
min
min
kW
e/start
e/15 min
CHP A
A
CHP B
B
Furnace
C
35,00
3,00
50,00
6,00
30,00
60,00
45,00
2,50
0,43
23,00
2,00
50,00
6,00
30,00
60,00
30,00
1,50
0,30
50,00
6,00
80,00
1,00
-
In this simulation, the price of electricity that is sold back to the public grid is considered
to be 80% of the cost of electricity that is bought from the public grid, since utilities usually
buy energy for a lower price than they sell it [34]. Furthermore, the specifications for the
CHPs and the furnace were based on data found in [48], which were adapted for this case
study. The values that are used in this work are shown in Table 2.3.
The dependency between the inputs and outputs of the converters is represented by the
conversion efficiency. The values of the efficiencies were obtained from data of similarly
sized components found in [33, 34]. The following values were used:
A
ηA
gq = 0, 45 ηge = 0, 35
(2.51)
ηBgq = 0, 40 ηBge = 0, 30
(2.52)
ηCgq =
(2.53)
0, 40.
34
2 The Framework: Multi-Carrier Unit Commitment
The minimum input power is calculated from the minimum electrical power output given
in Table 2.3 (the values are given in kW):
PA
g=
PBg =
LA
e
ηA
ge
LBe
ηBge
=
3
= 8, 6
0, 35
(2.54)
=
2
= 6, 7.
0, 30
(2.55)
The maximum gas input of each converter is calculated below:
A
Pg =
B
Pg =
C
Pg =
LA
q
ηA
gq
LBq
ηBgq
LCq
ηCgq
=
45
= 100
0, 45
(2.56)
=
30
= 75
0, 40
(2.57)
=
80
= 200.
0, 40
(2.58)
The maximum gas input to the hub Pg is the sum of the gas input to the converters:
A
B
C
Pg =Pg + Pg + Pg = 375.
(2.59)
The software program SEPATH [49] is used to obtain the load patterns of 10 households
(aggregated). The software randomizes a large selection of data that was obtained through
a large survey [49]. After entering some information about the households to be studied,
a week load pattern for electricity and heat is generated. In this simulation, the results are
depicted for a period of 3 hours in time intervals of 15 minutes. The time range considered
starts at 15:45 hours and ends at 18:45 hours. A random Saturday winter day pattern is
used. In the Netherlands people habitually have dinner around 18:00 and during winter,
sunset occurs somewhere between 16:30 and 18:30. Due to these reasons, an increase in
the load pattern takes place during the selected period. The electric and heat load for the 12
intervals is shown in Table 2.4 and in Figure 2.6.
During the simulation, the 4 best strategies are saved by the program at each step. This
corresponds to variable Ndp (see Section 2.3.1). The best final result is the optimal strategy
for the multi-carrier unit commitment, in this case, the lowest total cost for the 12 periods
that were considered.
Table 2.4: Electric and heat loads for the period under study in kW
Time Period
Load
Le
Lq
1
2
3
4
5
6
7
8
9
10
11
12
10,1
10,6
9,9
22,9
7,4
26,4
8,5
37,6
11,9
52,9
14,8
54,1
14,6
72,3
19,2
87,9
19,8
88,2
16,3
48,1
11,0
32,4
5,9
33,3
2.4 Simulation Results
35
Electric and Heat Load of 10 Households
100
Electric load
Heat load
80
Power [kW]
60
40
20
0
15:45
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.6: Electric and heat loads for the period under study in kW
Four cases are presented in this section:
• Case 1: this is the base case; no-load costs, start-up costs and minimum up and down
time constraints are omitted.
• Case 2: this case includes no-load costs.
• Case 3: this case includes no-load costs and start-up costs.
• Case 4: this case includes no-load costs, start-up costs and minimum up and down
time constraints.
Case 1
In this case only the operational costs are included. From the results shown in Table 2.5
it can be observed that the multi-carrier unit commitment is ruled by the efficiency of conversion of the components involved. The reason for this is that the costs increase as the
efficiency decreases, since more fuel is necessary to provide the same output. CHP A (converter A) has the highest efficiency, followed by CHP B (converter B) and the furnace (converter C). Thus, the furnace operates only when the heat load is above 75 kW, which is the
maximum heat power that can be delivered by the CHPs. This occurs in time intervals 8 and
9, which correspond to the time periods that go from 17:30 to 18:00 hours of the selected
day. The distribution of load between the converters is the result of the multi-carrier optimal
dispatch for each of the commitments shown in Table 2.5. The power values in Table 2.6
B
correspond to the electricity input Pe and the gas power inputs to each converter PA
g , Pg and
C
Pg .
36
2 The Framework: Multi-Carrier Unit Commitment
Table 2.5: Simulation 1: Multi-carrier unit commitment - Case 1 (best strategy)
Time Period
Converter
Component
1
2
3
4
5
6
7
8
9
10
11
12
A
B
C
Grid
CHP A
CHP B
Furnace
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
1
0
0
1
1
0
0
Table 2.6: Simulation 1: Multi-carrier optimal dispatch - Case 1 (best strategy)
Time Period
2
3
4
5
6
7
8
9
10
11
12
1,9
23,6
0
0
-7,9
50,9
0
0
-13,1
58,7
0
0
-20,7
83,6
0
0
-29,0
100,0
19,8
0
-27,0
100,0
22,8
0
-40,9
100,0
65,3
0
-38,3
100,0
75,0
32,3
-37,7
100,0
75,0
33,0
-21,0
100,0
7,8
0
-14,2
72,0
0
0
-20,0
74,0
0
0
Optimal Dispatch for Case 1
125
100
75
50
Power [kW]
Pe
PA
g
PBg
PCg
1
25
0
−25
−50
−75
−100
15:45
Electrical grid
Converter A − CHP
Converter B − CHP
Converter C − Furnace
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.7: Simulation 1: Multi-carrier optimal dispatch - Case 1 (best strategy)
2.4 Simulation Results
37
An interesting result can be observed at time period 2, where it is cheaper to sell electricity back to the grid than to use the furnace to produce the difference in the heat demand
(Table 2.4 shows that from the first to the second period the electric load varies only 0,2 kW,
but the heat demand is more than doubled). This occurs due to the fact that the total efficiency of CHP A is two times higher than the efficiency of the furnace. Moreover, the cost
of gas is lower than the cost of electricity and the CHPs are gas-fueled. The total cost for
the 12 time intervals is e30.
Case 2
In this case the no-load costs are included. The no-load costs for CHP B are lower than for
converter A, as can be observed from Table 2.3. During the first time period this difference
in no-load costs appears to be predominant in comparison to the difference in efficiencies.
This can be observed in Table 2.7, where CHP B is used instead of CHP A, which has higher
efficiency, but higher no-load costs. Similarly, in time periods 5, 6 and 10 it is cheaper to
use the furnace, with no-load costs equal to zero, than to use CHP B, which has a higher
efficiency. The total cost for the 12 time intervals in this case is e36,58.
Table 2.7: Simulation 1: Multi-carrier unit commitment - Case 2 (best strategy)
Time Period
Component
1
2
3
4
5
6
7
8
9
10
11
12
Grid
CHP A
CHP B
Furnace
1
0
1
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
1
1
1
0
1
1
1
1
0
1
1
1
1
1
1
1
1
1
1
0
1
1
1
0
0
1
1
0
0
Case 3
The start-up costs play an important role in this case since they are quite high: e2,5 for
CHP A, e1,5 for CHP B and e1 for the furnace, as shown in Table 2.3. It is assumed that
only CHP A is committed at the beginning of the simulation. The best strategy found by the
program is to use the furnace continuously from time interval 5 to time interval 10, instead
of turning it off during period 7, as proposed in the previous case.
Table 2.8 shows the multi-carrier unit commitment for this case. The total cost for the
12 intervals is e39,130. It is important to recall that the program chooses the best feasible
path according to the options available for the 12 time periods. Thus, other solutions are
possible, which may give a different commitment but with the downside of a higher total
cost.
38
2 The Framework: Multi-Carrier Unit Commitment
Table 2.8: Simulation 1: Multi-carrier unit commitment - Case 3 (best strategy)
Time Period
Component
1
2
3
4
5
6
7
8
9
10
11
12
Grid
CHP A
CHP B
Furnace
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
0
0
1
1
0
0
Case 4
This case considers the minimum down and up times. The minimum up time for CHP B
is 60 min, i.e. 4 time intervals. The solution in Case 3 (which was the cheapest when
considering the no-load costs and start-up costs) is not feasible in this case because of the
minimum up time limitation. The multi-carrier unit commitment program selects the path
that produces the lowest overall cost, which in this case corresponds to the strategy shown
in Table 2.9. The total cost for the 12 intervals is e39,212.
Table 2.9: Simulation 1: Multi-carrier unit commitment - Case 4 (best strategy)
Time Period
Component
1
2
3
4
5
6
7
8
9
10
11
12
Grid
CHP A
CHP B
Furnace
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
0
1
1
0
0
In Figure 2.8 the optimal dispatch for this case is graphically depicted. It can be observed that CHP B is kept off during the simulation. The furnace is used to supply the heat
load that was supplied by CHP B in Case 1. Moreover, the electricity that is sold to the grid
is reduced in comparison to Figure 2.7 because CHP B does not participate in this case; as a
consequence, the electricity produced is less. Due to the relatively high price that is paid for
the electricity that is sold back to the grid (80% of the regular price for electricity), electricity is consumed from the grid only during the first time period, in the rest of the simulation
the optimal results are obtained when electricity is injected back to the grid.
2.4 Simulation Results
39
Optimal Dispatch for Case 4
125
100
75
Power [kW]
50
25
0
−25
−50
Electrical grid
Converter A − CHP
Converter B − CHP
Converter C − Furnace
−75
−100
15:45
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.8: Simulation 1: Multi-carrier optimal dispatch - Case 4 (best strategy)
2.4.2 Simulation 2: Multi-Carrier Unit Commitment with Storage
In this simulation, heat storage is added at the output side of the energy hub. Figure 2.9
shows the resulting energy hub. The specifications for the storage unit are based on the data
found in [33]. Some intermediate results are shown in this section to illustrate the four-step
technique that was described in Section 2.3.2. The program runs and executes the four steps
consecutively.
Le
Pe
Pg
v gA Pg
A
v gB Pg
B
v gC Pg
C
Lq
STORAGE
Figure 2.9: Energy hub representation for Simulation 2
40
2 The Framework: Multi-Carrier Unit Commitment
The data for the prices, the hub elements and the load are the same that were used in
Simulation 1, shown in Table 2.1, Table 2.3 and Table 2.4. The start-up prices are included
but the minimum up and down times are set to zero in this simulation in order to show a
clear variation in the commitment results due to the inclusion of storage and not to other
constraints. Another change with respect to the previous simulation is that the minimum
input power for the furnace is considered 1 kW (PCg = 1), instead of zero as in the previous
simulation. The data of the storage unit can be found below:
Table 2.10: Parameters of the storage element
Parameter
Unit
Storage
Minimum energy content
Maximum energy content
Maximum power (charge)
Maximum power (discharge)
Charging efficiency eαi
Stand-by losses
Initial energy content
Final energy content
kW·(15 min)
kW·(15 min)
kW
kW
%
kW/step
kW·(15 min)
kW·(15 min)
68,00
100,00
15,00
-15,00
0,85
0,50
80,00
80,00
In order to understand the results obtained from this simulation, let’s first recall the steps
included in the technique:
• Step 1: Solve the multi-carrier unit commitment problem using dynamic programming without including storage according to Section 2.3.1.
• Step 2: Use the scheduling solution from Step 1 and solve the multi-carrier optimal
dispatch including storage according to Section 2.2.4.
• Step 3: Incorporate the storage power flow results obtained from Step 2 as an additional bidirectional load and solve the multi-carrier unit commitment problem according to Section 2.3.1.
• Step 4: Use the scheduling solution from Step 3 and solve the multi-carrier optimal
dispatch including storage according to Section 2.2.4.
Step 1
After Step 1 has been completed, the unit schedule presented in Table 2.11 is obtained.
At this step storage is omitted, thus the results are equal to the results obtained at Case 1
in Section 2.4.1. Likewise, the multi-carrier economic dispatch is the same as shown in
Table 2.6; they are shown in Figure 2.10. The furnace is turned on at time periods 8 and 9,
where the load is the highest.
2.4 Simulation Results
41
Table 2.11: Simulation 2: Multi-carrier unit commitment - Step 1 (best strategy)
Time Period
Converter
Component
1
2
3
4
5
6
7
8
9
10
11
12
A
B
C
Grid
CHP A
CHP B
Furnace
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
1
0
0
1
1
0
0
Optimal Dispatch for Step 1
125
100
75
Power [kW]
50
25
0
−25
−50
−75
−100
15:45
Electrical grid
Converter A − CHP
Converter B − CHP
Converter C − Furnace
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.10: Simulation 2: Multi-carrier optimal dispatch - Step 1 (best strategy)
Step 2
At Step 2, the multi-carrier unit commitment shown in Table 2.11 is fixed as a pre-defined
parameter. In this case the storage unit is included as part of the system and the multicarrier optimal dispatch is solved. The results of the optimal dispatch can be observed in
Figure 2.11. It can be noted that the furnace works at its minimum capacity (PCg = 1) when
the load is the highest, between 17:30 and 18:00. It is expected that it will be turned off as
a result of Step 3, if no constraints are violated.
The storage bulk is charged during the first periods and discharged during the load peak.
The energy content of the storage unit can be observed in Figure 2.12. It can be noted that
the initial and final energy content is the same, as defined in one of the storage constraints.
42
2 The Framework: Multi-Carrier Unit Commitment
Optimal Dispatch for Step 2
125
100
75
Power [kW]
50
25
0
−25
−50
Electrical grid
Converter A − CHP
Converter B − CHP
Converter C − Furnace
−75
−100
15:45
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.11: Simulation 2: Multi-carrier optimal dispatch - Step 2
Energy Content of the Heat Storage Unit − Step 2
120
100
Energy [kW(15 min)]
80
60
40
20
Storage reference
Energy content op heat storage unit
0
15:45
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.12: Simulation 2: Energy content of the storage element - Step 2
2.4 Simulation Results
43
Step 3
In this step the dispatch values obtained from Step 2 for the storage unit are loaded as a
fictitious bidirectional load. The program runs the multi-carrier unit commitment again and
a new schedule is obtained. It is expected that the units that were brought to the minimum
capacity at Step 2, will now be turned off. The results of the multi-carrier unit commitment
in Table 2.12 show that the furnace is kept off during the whole simulation now that storage
is considered, just as expected. This proves that the use of storage can influence the results
of the multi-carrier unit commitment.
Table 2.12: Simulation 2: Multi-carrier unit commitment - Step 3 (best strategy)
Time Period
Converter
Component
1
2
3
4
5
6
7
8
9
10
11
12
A
B
C
Grid
CHP A
CHP B
Furnace
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
0
1
1
0
0
1
1
0
0
Step 4
Optimal Dispatch for Step 4
125
100
75
Power [kW]
50
25
0
−25
−50
−75
−100
15:45
Electrical grid
Converter A − CHP
Converter B − CHP
Converter C − Furnace
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.13: Simulation 2: Multi-carrier optimal dispatch - Step 4
44
2 The Framework: Multi-Carrier Unit Commitment
Step 4 provides the final solution, shown in Figure 2.13. It can be noticed that CHP A has a
higher production during the first periods than in the case where no storage was considered
(see Figure 2.10). The extra heat is used to charge the storage tank at the beginning of the
simulation. Consequently the storage unit is used to supply the load during the peak that
occurs between 17:30 and 18:00. Thus, this technique enables generation peak-saving by
the appropriate storage optimization.
With regard to the results of Step 1 (no storage) and Step 4 (with storage), a cost reduction was achieved from e32,5 to e28,5; this represents a 9,2%. In future research, the
installation and operation costs of the storage unit can be considered to evaluate if the reduction in operation costs due to the storage optimization can surpass the investment costs.
The energy content of the storage unit can be observed in Figure 2.14.
Energy Content of the Heat Storage Unit − Step 4
120
100
Energy [kW(15 min)]
80
60
40
20
Storage reference
Energy content op heat storage unit
0
15:45
16:15
16:45
17:15
Time
17:45
18:15
18:45
Figure 2.14: Simulation 2: Energy content of the storage element - Step 4
2.5 Conclusions
In this chapter, a general unit commitment framework for energy systems that contain multiple energy carriers was presented. The method was demonstrated with illustrative examples.
Electricity and gas were used as input energy carriers due to the fact that the infrastructures
for these energy carriers already exist and the prices are available. However, the framework
proposed is suitable for any kind of energy carrier and for different possible couplings and
power scales.
The use of the energy hub concept proved to be a suitable way to handle the conversion
couplings that may exist between carriers. This way of analysis provides flexibility and was
successfully applied for the development of the multi-carrier unit commitment framework.
2.5 Conclusions
45
The inclusion of storage proved to be valuable for generation peak-shaving purposes.
The technique proposed allows an optimized use of storage in systems with multiple energy
carriers. In the example, it was shown that storage has a significant influence on the results
of the multi-carrier unit commitment and the optimal dispatch.
Chapter 3
The Approach
Energy or Exergy Optimization?
This chapter deals with Research Question 3 of this dissertation. Section 3.1 includes the
main theoretical concepts that are applied. In Section 3.2 a literature review of exergyrelated studies is presented. Section 3.3 introduces the optimization framework. Later in
Section 3.4 the system under consideration and the component models are described. In
Section 3.5 the results of the dispatch optimization and the scheduling optimization are
presented. The conclusions of this chapter are presented in Section 3.6. Parts of this chapter
have been published in [50].
3.1 Basic Concepts and Definitions
In order to perform an exergy-related study, it is necessary to get familiar with several
theoretical concepts. Due to the multi-disciplinary nature of this dissertation, these concepts
are included in the following subsections in order to assist the reader with understanding the
general approach presented in this chapter.
3.1.1 Energetic Efficiency
The energetic (or thermal) efficiency is the ratio between the energy of the product and the
energy of the source. It is denoted by η [-].
3.1.2 Exergetic Efficiency
The exergetic efficiency is the ratio between the exergy of the product and the exergy of the
source. It is denoted by ε [-].
3.1.3 Energy Content of a Fuel
The energy content of a fuel is equal to the amount of heat that is produced during combustion and is calculated as the enthalpy difference between the substance and the combustion
47
48
3 The Approach: Energy or Exergy Optimization?
products, both at 25◦ C and 1 bar. There are two ways to compute the energy content of a
fuel; it can be done in terms of the higher heating value (HHV) or in terms of the lower heating value (LHV). Since water is a product of combustion, HHV is the heat content of water
when the water resulting from the combustion process is in liquid form and the LHV is the
heat content of water when the water resulting from the combustion process is in gaseous
form. The HHV gives an indication of the best possibilities of a fuel.
3.1.4 Specific Internal Energy
Specific internal energy u [J/kg] is the total amount of energy stored in a substance per kg.
The change in internal energy is measured by adding or removing energy under constant
volume [51].
3.1.5 Entropy Change
Entropy change ∆S [J/K] is associated with the extraction of an amount of heat ∆Q [J] from
a reservoir with a given temperature T [K], as shown in (3.1):
∆S =
∆Q
.
T
(3.1)
The increase of entropy principle states that the entropy creation due to irreversibility
is zero for ideal processes and positive for real ones, thus the magnitude of the entropy
creation due to irreversibility is a measure of the irreversibility of a process [38, 43].
3.1.6 Specific Enthalpy
Specific enthalpy h [J/kg] can be defined as the total thermomechanical energy content of a
mass for a given temperature per kg [38]. The change in enthalpy is measured by adding
or removing energy under constant pressure. For solids and liquids, the difference between
specific internal energy u [J/kg] and specific enthalpy h [J/kg] is very small, however for
gases this difference cannot be neglected [51]. The equation that describes enthalpy is given
as follows, where p [Pa] is pressure and v [m3 /kg] is the specific volume:
h = u + pv.
(3.2)
3.1.7 Carnot Cycle
The Carnot cycle is the most efficient cycle between any two temperatures [38, 51]. Thus
the efficiency of the Carnot cycle (Carnot factor) [-], is the maximum efficiency that can
be attained when work is produced out of heat that is extracted from one thermal energy
reservoir and transferred to another thermal energy reservoir [51]. It is given by:
!
T ref
(3.3)
Carnot factor = 1 −
T
where T [K] is the absolute temperature of the thermal energy reservoir and T ref [K] is the
reference temperature, for example the temperature of the environment.
3.1 Basic Concepts and Definitions
49
3.1.8 First Law and Second Law of Thermodynamics
The first law of thermodynamics, also known as the law of conservation of energy states
that energy can neither be created nor destroyed, but can be converted from one form to
another. The energy content is the amount of energy required to bring a substance from
the reference state to the actual state; it depends on conditions as temperature and pressure,
and it is determined in relation to a reference state [51]. An energy balance analysis around
a process or system accounts for all its energy. Thus, a general energy balance equation
considers the energy that enters and leaves the process under study [38].
The second law of thermodynamics indicates that heat cannot be fully converted into
work. Thus, the most energy-efficient closed cycle to perform the heat-to-work or work-toheat conversion is the Carnot Cycle under ideal performance conditions [38]. The KelvinPlanck formulation of the second law states the following [51]:
It is impossible for any system to operate in a thermodynamic cycle and deliver
a net amount of energy by work to its surroundings while receiving energy by
heat transfer from a single thermal energy reservoir.
Thus, the second law of thermodynamics indicates that heat cannot be fully converted
into work. Entropy production must be minimized in order to achieve an efficient energy
use [38].
3.1.9 Exergy Definition
Different definitions of exergy can be found in the literature, some of them are:
• Exergy is defined as the work potential that is available in gas, fluid or mass as a result
of its non-equilibrium condition relative to some reference condition [38].
• The exergy content of an energy carrier is the maximum amount of work that can be
extracted from it, in general [51]:
– Electricity is work, therefore, the exergy content is equal to the energy content.
– For fuels, the exergy content is more or less equal to the energy content.
– The exergy content of heat flows is smaller than the energy content.
• The property exergy defines the maximum amount of work that may theoretically be
performed by bringing a resource into equilibrium with its surroundings through a
reversible process [52].
• Exergy of a thermodynamic system is the maximum theoretical useful work (shaft
work or electrical work) obtainable as the system is brought into complete thermodynamic equilibrium with the thermodynamic environment while the system interacts
with this environment only. The total exergy of a system consists of: physical exergy
(due to the deviation of the temperature and pressure of the system from those of the
environment), chemical exergy (due to the deviation of the chemical composition of
the system from that of the environment), kinetic exergy (due to the system velocity
measured relative to the environment) and potential exergy (due to the system height
measured relative to the environment) [53].
50
3 The Approach: Energy or Exergy Optimization?
3.1.10 Energy and Exergy General Equations
The general energy equation is obtained from the energy balance statement of the first law
of thermodynamics, thus for a process or a system under study, the energy entering the
system is equal to the energy leaving the system [38]. The equation is divided by mass
m [kg] at both sides, thus it contains specific energy values:
+
gz1
|{z}
Potential energy in
|
=
gz2
|{z}
+
Potential energy out
|
321
2
|{z}
+
Kinetic energy in
322
2
|{z}
Internal energy in
{z
Flow work in
+
u2
|{z}
Internal energy out
{z
Work in
}
Energy in
+
Kinetic energy out
p1 v1 + win
|{z}
|{z}
+
u1
|{z}
p 2 v2
|{z}
Flow work out
Energy out
+ wout
|{z}
(3.4)
Work out
}
where g [m/s2 ] stands for the gravitational acceleration, z [m] stands for height/altitude,
3 [m/s] stands for velocity, u [J/kg] stands for specific internal energy, p [Pa] stands for
pressure, v [m3 /kg] stands for specific volume, win [J/kg] stands for the work per kg that
goes into the system, wout [J/kg] stands for the work per kg that goes out of the system.
Exergy is an explicit property at steady-state conditions and its value can be calculated
at any point of the system relative to a reference condition [38]. The general specific exergy
equation is:
a=
u−u
|{z}0
Internal energy
− T 0 (s − s0 ) + pv − p0 v0 +
| {z } | {z }
Entropy
Work
X
32
+ g(z − z0 ) + (µche − µ0 )nche
| {z } |
2
{z
}
|{z}
Momentum
Gravity
(3.5)
Chemical
where u [J/kg] stands for specific internal energy, T [K] stands for temperature, s [J/(K·kg)]
stands for specific entropy, p [Pa] stands for pressure, v [m3 /kg] stands for specific volume,
3 [m/s] stands for velocity, g [m/s2 ] stands for the gravitational acceleration, z [m] stands
for height/altitude, µche [J/mole] stands for chemical potential and nche [mole/kg] stands for
quantity of moles per unit mass. Subscript ‘0’ denotes the reference state. The general
specific exergy equation is often used under conditions where the gravitational, chemical
and momentum terms can be neglected. In this dissertation, the work contribution in (3.5)
is already included in the enthalpy definition formulated in (3.2). After using (3.2), in the
case of steady flow systems, the resulting specific exergy a [J/kg] is [38]:
a = (h − h0 ) − T 0 (s − s0 )
| {z } | {z }
Enthalpy
(3.6)
Entropy
where subscript ‘0’ denotes the reference state. To compute the available potential work by
bringing the system from operational state 1 to operational state 2, the following equation
is used:
a = (h2 − h1 ) − T 0 (s2 − s1 ).
(3.7)
3.2 Literature Review
51
Equation (3.7) is used for several calculations in Section 3.4. In an exergy analysis the principles of conservation of mass and conservation of energy in combination with the second
law of thermodynamics are applied [40]. The conventional definition of energy provides
the amount of Joules or British Thermal Units (BTUs) that are involved in the system or
process, but does not provide information about the potential to make work of the Joules or
BTUs involved. Here is where exergy provides valuable information by indicating the work
potential availability in those Joules or BTUs.
3.1.11 Generalities of the Exergy Method
The exergy method answers the question of where, how, why and how much of the available
work is lost in the system [38]. This method is based on evaluating the work that is available
at different points of the system; in this way, the quantity and location of the lost/useful work
can be detected [38]. This method can be used for designing new systems and for evaluating
existing ones.
If no useful work is done in the process, the change represents a loss in available work
[38]. In [38], the author discusses the importance of choosing a proper reference condition
for the system. He argues that using a common reference base like the surrounding environment as the ultimate heat sink is suitable for making valid comparisons between different
types of plants. Nevertheless, he points out that there are specific conditions where it is more
practical to choose other references according to the characteristics of the system. In this
dissertation, different plants are compared; furthermore the surrounding environment can
act as the infinite sink. For these reasons the reference condition chosen is the surrounding
environment.
A general exergy equation can be obtained by adding up all exergies related to an specific
point, as shown in (3.5). Exergy losses are generated by the production of entropy due to
non-ideal performance of processes. A high exergy loss in a particular system may indicate
that there is a poor match between the quality of energy supplied and the quality of energy
required in that specific system, which results in large irreversible losses [38].
3.2 Literature Review
During the last decade, several articles, books and academic documents have promoted exergy analysis as a better tool to assess energy systems than a traditional energy analysis.
For example in [43], benefits of using exergy-based rather than energy-based measures for
efficiency and losses are presented. The author argues that efficiencies based on energy
can often be nonintuitive or misleading, partly because energy efficiencies do not provide
a measure of how close a process approaches ideal conditions. The author also states that
energy losses can be large in quantity, nevertheless, they may not be that significant thermodynamically due to the low potential to make work of the energy that is lost.
The author M. Rosen has published several papers in the area of exergy [42, 43, 54–58].
One of the main messages given in his work is that exergy analysis provides insights into
efficiency improvement and environmental-impact reduction in systems and processes. He
argues that in complex systems with multiple products (e.g., cogeneration and trigeneration
plants), exergy methods can help evaluate thermodynamic values of different product energy
52
3 The Approach: Energy or Exergy Optimization?
forms, even though they normally exhibit very different characteristics. This statement is
closely related to the kind of systems regarded in this dissertation, where multiple energy
carriers are involved. Additionally, he indicates that in order to motivate engineers to use
exergy, joint efforts must be made to point out the benefits of using exergy methods clearly
and unambiguously. This chapter is an effort to show the added value of using exergy to
analyze energy supply systems with multiple energy carriers.
The exergy analysis has been introduced to various fields of study, among them, renewable energy, exergoeconomics, industrial ecology, building design and the transportation
sector [59–67]. In the area of building design, the topic of exergy has gained attention
particularly during the last decade; related works can be found in [40, 68–71].
The following shortcomings were found in the literature that was analyzed:
• Most literature is focused on simple and single processes and not on complex systems
with multiple energy carriers.
• In several papers, the term optimization is commonly referred to as the improvement
of a system with respect to an initial selection of parametric set-points and not as the
local or global minimum in a solution.
• Some papers do not provide insight about the benefits of using an exergy analysis
with respect to an energy analysis.
In order to overcome these limitations, in this chapter an optimization tool capable of
finding the optimal scheduling of units in a system with multiple energy carriers is introduced. In the following paragraphs some examples are discussed in which an exergy optimization of systems was aimed but in which one or more of the listed limitations were
found.
In the area of renewable energy, most of the exergy-related papers focus on a single
type of energy source. For example, in [59] an energy and exergy analysis of an integrated
solar combined cycle system is performed using the design plant data. Moreover, numerous
exergy studies have been performed about geothermal energy systems [60–62]. Geothermal energy is to some extent considered a renewable energy source since it usually has a
projected life of 30 to 50 years [60].
Another example in the topic of renewable energy can be found in [63], where energy
and exergy analyses are performed to four different wind power systems, including horizontal and vertical axis wind turbines. Their work is based on [64], where an efficiency formula
based on exergy values for wind energy systems is developed and described. The technique
utilizes the wind chill temperature associated with the wind velocity to predict the entropy
generation of the process. Their approach is valuable for the design of wind turbines since it
quantifies the exergy loss that occurs in the process of generating electricity from the kinetic
energy of the wind. Similarly in [65], the exergetic efficiency of a wind turbine is calculated.
The authors define the exergetic efficiency as a measure of how well the stream of exergy of
the fluid is converted into useful turbine work output or inverter work output. They consider
that the availability of the blowing air, in other words the exergy of the blowing air, is simply
the kinetic energy it possesses.
In [66] an optimization strategy is proposed where the cost/efficiency ratio is varied and
the optimal allocation of renewable energy is determined. One of the limitations in the approach is that it assumes fixed percentages of utilization of the sources for the optimization.
3.2 Literature Review
53
In [72], three different renewable energy systems are studied: solar energy, wind power and
geothermal energy. The paper provides the equations to obtain the exergy efficiencies of
each system, gives the corresponding results and makes a comparison between renewable
and non-renewable sources.
Exergoeconomics is a branch of engineering that combines thermodynamic evaluations
based on an exergy analysis with economic principles in order to provide useful information for the design and operation of a cost-effective system [53]. In the literature, a large
percentage of exergoeconomic studies have been performed in geothermal systems, as in
[73–75]. In [76], prices for energy and exergy of various energy sources along with their
CO2 equivalents are calculated. Detailed tables are provided for the different fuels considered. Even though this kind of studies provides an interesting insight by defining prices and
costs based on exergy values of the fuels, an exergoeconomical-oriented approach will not
be followed in this dissertation.
The field of industrial ecology emerged from efforts to reduce depletion and to move
towards a more sustainable utilization of resources [52, 77]. In [52] an analogy between
ecosystem evolution processes and industrial processes is made. By means of this analogy,
consumption is interpreted as a process of exergy removal. The authors argue that this approach allows an improved understanding and analysis of the interrelated roles that cycling,
cascading, efficiency gains and renewed exergy use may play. This is an example where
finding analogies facilitates the use of the exergy analysis in systems where this concept
had not been used before.
In the area of building design, the topic of exergy has gained attention particularly during the last decade, related works can be found in [40, 68–71]. Fossil fuels burn at very high
temperatures, therefore their work potential is largely wasted when fossil fuels are utilized
for hot water heating, space heating or even industrial steam production in building infrastructures, where low-temperature heat is desired [40]. This is also the case in residential
district infrastructures, like those considered in this dissertation. The building sector has a
very low exergetic efficiency of energy utilization, as a result atmosphere is polluted unnecessarily [40]; likewise, residential district infrastructures, having similar load patterns and
heating infrastructures, have a considerable low exergetic efficiency, and thus, have a high
potential for improvement.
Only one paper was found in the literature in which the concepts of exergy and the
energy hub were combined [67]. In [67] the objective function to be minimized is the inverse
equation of the total exergetic efficiency of the system, thus the optimization maximizes the
system’s exergetic efficiency. Moreover, the interactions between the energy carriers are
represented by the energy hub coupling matrix, which contains the energy efficiencies of
the units. In the paper, a comparison is made between the optimal dispatch of units obtained
by maximizing the exergetic efficiency and the optimal dispatch obtained by minimizing the
costs. Even though the results give some insight by stating that the most efficient solution in
terms of exergy may not always be the same as the one obtained from the most economical
solution, the paper does not give any indication about the results that would be obtained from
maximizing the energetic efficiency in relation to maximizing the exergetic efficiency. This
comparison would provide insight about differences between the analysis and optimization
of systems using energy and exergy, which is a topic of dispute.
54
3 The Approach: Energy or Exergy Optimization?
In complex systems involving more than one generation unit and multiple energy carriers, the differences that may exist between making an energy or an exergy analysis have
not clearly been presented in the literature yet. This chapter provides concrete comparisons
between performing an energy and an exergy optimization in a multi-carrier energy supply
system.
3.3 Exergy Optimization Approach
3.3.1 Exergy Hub Conversion Model
In the energy hub approach, a coupling matrix is used to represent the interactions among
the energy carriers that are contained in the energy hub. In this chapter, the systems to be
studied are represented in terms of exergy values, for this reason the term exergy hub was
selected to refer to the approach presented. The exergy hub model uses a coupling matrix
that contains the steady-state exergy efficiencies of each of the hub elements. Just like in the
energy hub approach, the input and output exergy flows are coupled by a coupling factor.
For a system with one input energy carrier α and one output energy carrier β, the coupling
factor bαβ [-] is used to define the relation between the steady-state input exergy flow Ξα
[kW] and the steady-state output exergy flow Γβ [kW] of the exergy hub:
Γβ = bαβ Ξα
(3.8)
where input energy carrier α is converted into the output energy carrier β. For several energy
carriers the variables αi and β j are used as defined in Chapter 2. The matrix that represents
a system with multiple energy carriers is shown in (3.9):

 


 Γβ1   bα1 β1
bα2 β1 . . . bαnin β1   Ξα1 




 Γ   b
bα2 β2 . . . bαnin β2   Ξα2 
 β2   α1 β2
(3.9)
..
..   .. 
..
 ..  =  ..
.
 .   .
.
.   . 






Γβ n
bα1 βnout bα2 βnout . . . bαnin βnout Ξαnin .
{z
} | {z }
| {zout} |
Γ
B
Ξ
When several converters k p are considered, the coupling factor bαi β j [-] that takes into account the participation of each converter k p can be expressed as:
X k k
bαi β j =
vαpi εαpi β j
(3.10)
k p ∈Cαi
k
where vαpi [-] is the dispatch factor that represents the percentage of input exergy flow Ξαi
k
[kW] that is provided to converter k p and εαpi β j [-] is the exergetic efficiency of conversion
from energy carrier αi to β j of converter k p .
3.3.2 Problem Statement
The output exergy flow vector Γ is equal to the coupling matrix B multiplied by the respective input exergy flow vector Ξ. This represents the exergy balance equality constraint:
Γ − BΞ = 0.
(3.11)
3.3 Exergy Optimization Approach
55
k
The sum of dispatch factors vαpi related to each energy carrier αi is equal to 1. Thus, the
dispatch factor equality constraint is:
X k
1−
vαpi = 0.
(3.12)
k∈Cαi
The inequality constraints are given by the lower and upper limits ofthe hub’s exergy
kp
k
flow inputs Ξ αi , Ξ αi [kW], of the converter units’ power inputs Ξ αpi , Ξ αi [kW] and of the
dispatch factors:
Ξ αi ≤ Ξαi ≤ Ξ αi
k
Ξ αpi
≤
0≤
k
vαpi
k
vαpi Ξαi
≤
kp
Ξ αi
≤ 1.
(3.13)
(3.14)
(3.15)
Three different objective functions are used for the optimizations presented in this chapter. The first objective function is based on the exergetic efficiency of the system, the second
one is based on the energetic efficiency and the third one on the price of the fuels involved.
In the first case, the objective function that is minimized is inversely proportional to the
exergetic efficiency of the system under study εtot [-], in this way the exergetic efficiency of
the system is maximized. The corresponding equation is:
Fobj1 =
Ξinp
1
=
εtot Γout
(3.16)
where εtot is the total exergetic efficiency of the system, Ξinp [kW] is equal to the sum of the
exergy flow brought to the system by the input fuels and Γout [kW] is equal to the sum of
the exergy flow used by the loads at the output side of the hub.
Analogously, in the second case, the objective function that is minimized is inversely
proportional to the energetic efficiency of the system under study ηtot [-]. The corresponding
equation is:
Pinp
1
=
(3.17)
Fobj2 =
ηtot
Lout
where ηtot [-] is the total energetic efficiency of the system, Pinp [kW] is equal to the sum of
the power brought to the system by the input fuels and Lout [kW] is equal to the sum of the
power used by the loads at the output side of the hub.
Finally, the objective function of the cost minimization [e/(time period)] is a quadratic
equation that depends on the energy prices and the input powers of the hub:
X
(K1,αi + K2,αi Pαi + K3,αi P2αi )
(3.18)
Fobj3 =
αi ∈Ein
where K1,αi , K2,αi and K3,αi are cost coefficients.
56
3 The Approach: Energy or Exergy Optimization?
3.4 Exergy Calculation Models and Data
3.4.1 System Representation: Energy Hub versus Exergy Hub
The hub that will be used for the illustrative examples consists of two combined heat and
power units, a furnace, a heat pump and a connection to the public electrical grid. The hub’s
inputs are natural gas, biomass and electricity coming from the grid. The hub’s load consists
of an electric load and a heat load. The set of input carriers is defined as Ein = {e,g,b}, where
‘e’ stands for electricity coming from the grid, ‘g’ stands for natural gas and ‘b’ stands for
biomass. The set of output carriers is defined as Eout = {e,q} where ‘e’ stands for electricity
and ‘q’ stands for heat. The selected system serves as an example to demonstrate the tool.
Figure 3.1: Exergy hub representation
Figure 3.2: Energy hub representation
3.4 Exergy Calculation Models and Data
57
Direct connection A represents the direct connection with the electrical grid, converter B
represents the electricity-driven heat pump, converter C represents the gas turbine (gas-fired
CHP), converter D represents the gas-fired furnace and converter E represents the fuel cell
system (biomass-fired CHP). Thus, the set of converters is defined as Cαi = {A,B,C,D,E}.
The exergy hub gives a better representation of what some authors call the real efficiency
of the units. In the exergy hub, each hub element has an exergetic efficiency associated with
it. By comparing Figure 3.1 and Figure 3.2 it is possible to identify which components
appear to be very efficient from an energetic point of view, but are not so efficient from an
exergetic point of view. A good example is the gas-fired furnace (converter D), which has
an energetic efficiency of 96%, whereas an exergetic efficiency of only 11%.
It is interesting to observe that by comparing both the exergy hub and the energy hub
representations it is possible to identify that a component may be more efficient from an
energetic point of view than another, but less efficient from an exergetic point of view. For
example the biomass-fired CHP (converter E) in Figure 3.2 has a total energetic efficiency
of 92%, higher than the energetic efficiency of the gas-fired CHP unit (converter C), which
has an energetic efficiency of 88%. However, from an exergetic point of view the biomassfired CHP has an exergetic efficiency of 29%, lower than that of the gas-fired CHP unit,
which has an exergetic efficiency of 43%. This difference plays an important role in the
optimization results, as will be shown in Section 3.5.
As mentioned in Section 3.1.9, exergy is the maximum theoretical useful work that can
be obtained from a source, thus from a sustainable point of view, it is better to look for ways
to use the largest amount of exergy that a source can offer than to look only at its energetic
performance and be satisfied with a good indicator value.
3.4.2 Energetic and Exergetic Efficiencies of Generation Units
In order to make an exergy analysis, the exergy content of the exergy flows involved is
determined. Depending on the energy carrier, different operations are required to calculate
the exergy content of the energy flow. The component efficiencies are usually modeled as
constants, whereas in reality those will vary over the input range of the component. Ideally,
the coupling matrix B would include an extensive formula for each component in order to
describe the dynamic behavior and its effect on the component efficiency. This formula
would be a combination of all the physical processes in the hub elements. Nevertheless,
in this chapter constant values are used in order to reduce the complexity of the examples.
In the following subsections, the hub elements will be briefly described. The respective
data can be found in Table 3.1. The total energetic efficiency of component k p , denoted by
k
k
ηαpi ,tot [-], is given by (3.19), the total exergetic efficiency εαpi ,tot [-] is given by (3.20):
k
k
k
k
k
k
ηαpi ,tot = ηαpi e + ηαpi q
εαpi ,tot = εαpi e + εαpi q .
k
(3.19)
(3.20)
k
where ηαpi e [-] is equal to the electrical conversion energetic efficiency of converter k p , ηαpi q
k
[-] is equal to the thermal conversion energetic efficiency of converter k p , εαpi e [-] is equal
kp
to the electrical conversion exergetic efficiency of converter k p and εαi q [-] is equal to the
thermal conversion exergetic efficiency of converter k p .
58
3 The Approach: Energy or Exergy Optimization?
3.4.3 Natural Gas Supply and Biomass Supply (Ξg , Ξb )
The fuel inputs, natural gas and biomass, are chemical energy flows and thus chemical
exergy flows. Chemical energy is often calculated using the lower heating value [78], repre
sented in this thesis by ΥLHV,αi J/kg of carrier αi . In order to calculate the specific exergy
aαi [J/kg] of the fuel input αi , the exergy factor ζαi [-] is introduced. This factor varies per
fuel type and is close to unity. The relationship is described in (3.21) and (3.22), where
φm,αi kg/s is the mass flow and Ξαi [kW] is the exergy flow. The exergy factor of natural
gas is ζg = 1, 04 [79], the exergy factor of biomass is ζb = 1, 1069 (deduced from values in
[80]):
aαi = ΥLHV,αi ζαi
1
.
Ξαi = aαi φm,αi
103
(3.21)
(3.22)
3.4.4 Electricity Supply (Ξe ) and Electricity Demand (Γe )
An electrical energy flow has an equal exergy content to its energy content (it can be fully
converted into work). This is both applicable for the grid input power as for the electrical
output power of the CHPs:
Ξe = Pe
(3.23)
Γe = Le .
(3.24)
In order to calculate the system’s efficiency, the electricity that comes from the grid is
divided by the exergetic/energetic efficiency of the centralized generation plant with which
it is produced. The centralized plants consist of steam generation units fueled by natural
gas. These generation units have efficiencies of only 33% to 35% [81]. In this work it is
assumed that the energetic efficiency of the centralized generation plant is ηgen,e = 33,28%
and the exergetic efficiency is εgen,e = 32%. These values were selected arbitrarily.
3.4.5 Heat Load (Γq )
The district heating demand represents the heat load in the system under study. The thermal
energy flow represents the mass flow φm,wat [kg/s] of water. The heat load decreases the
temperature (and thereby the enthalpy) of the working fluid. In the model representing the
physical processes, the temperature levels are controlled on reference temperature levels:
T col = 53 ◦ C and T hot = 80 ◦ C. Subscript ‘hot’ represents the hot water flow and subscript
‘col’ represents the cold water flow of the district heating at 1 atm. The pressure is assumed
ref
constant. This gives reference specific enthalpy values href
col , hhot [J/kg] and specific entropies
ref
ref
scol , shot [J/(K·kg)]. The exergy content of the thermal energy flow is described by (3.26),
which is derived from (3.25), where T 0 [K] is the temperature of the reference environment
and aq [J/kg] is the thermal specific exergy. The thermal output exergy Γq [kW] is the exergy
change of the working fluid:
aq = h − h0 − T 0 s − s0
(3.25)
1
ref
ref
ref
.
(3.26)
Γq = φm,wat href
hot − hcol − T 0 shot − scol
103
Component
Parameter
Input carrier
Maximum input carrier αi
Maximum input carrier αi
Minimum input carrier αi
Minimum input carrier αi
Electrical energetic efficiency ηkαi e
Electrical exergetic efficiency εkαi e
Thermal energetic efficiency ηkαi q
Thermal exergetic efficiency εkαi q
Total energetic efficiency ηkαi ,tot
Total exergetic efficiency εkαi ,tot
Maximum electrical output
Maximum thermal output
Maximum thermal output
Flow
Unit
Power
Exergy
Power
Exergy
kW
kW
kW
kW
%
%
%
%
%
%
kW
kW
kW
Power, exergy flow
Power
Exergy flow
Electricity Grid
A
Heat Pump
B
Gas-Fired CHP
C
Furnace
D
Biom.-Fired CHP
E
Electricity
3000,00
3000,00
-3000,00
-3000,00
100,00
100,00
100,00
100,00
3000,00
-
Electricity
1000,00
1000,00
75,00
75,00
283,00
34,43
283,00
34,43
2830,00
344,26
Natural gas
1996,00
2075,84
145,00
150,80
38,23
36,76
49,76
5,82
87,99
42,58
763,03
993,16
120,82
Natural gas
2255,00
2345,20
0,00
0,00
95,54
11,17
95,54
11,17
2154,35
262,07
Biomass
2171,28
2403,39
151,99
168,24
23,04
20,81
69,10
7,59
92,14
28,41
500,21
1500,37
182,52
3.4 Exergy Calculation Models and Data
Table 3.1: Component parameters
59
60
3 The Approach: Energy or Exergy Optimization?
The simulation makes use of time series data of the heat and electric loads. In Section 3.4.4 it was shown that for the electrical flows no further calculations are needed. The
load data of the thermal energy requires conversion however.
Combining (3.26) and (3.27) leads to (3.28), which can be used to convert thermal
power to a thermal exergy flow:
1
ref
Lq = φm,wat href
hot − hcol
103

ref
ref 
s − scol 

.
Γq = Lq 1 − T 0 hot
ref 
href
hot − hcol
(3.27)
(3.28)
3.4.6 Compression Heat Pump (εBeq )
In order to increase the flexibility of the hub, a compression heat pump is used. The heat
pump is electrically driven and uses a ground reservoir as low temperature heat reservoir.
The efficiency values are based on [80].
3.4.7 Gas-Fired Combined Heat and Power Unit (εCge , εCgq )
The values for the gas-fired CHP unit are based on the GE Jenbacher JMS 312 GS-NL. The
model used in this simulation has a 60% higher rated power than the one found in the data
sheet. A fixed efficiency is used from the data sheet [82]. The thermal exergetic efficiency
is based on the assumption that heat addition/transfer occurs by the working fluid operating
at temperatures between 53 ◦ C and 80 ◦ C.
3.4.8 Gas-Fired Furnace (εDgq )
The gas-fired hot water furnace data are based on the Byworth FM2500 [83]. The thermal
exergetic efficiency is calculated in a similar way as the gas-fired CHP.
3.4.9 Biomass-Fired Combined Heat and Power Unit (εEbe , εEbq )
The biomass-fired CHP is a system that converts biomass to electricity and heat. This system
consists of 3 subsystems: syngas production, fuel cell system and combustion. The first
subsystem converts biomass to synthetic gas, using a fast internal circulating fluidized bed
gasifier [80]. This process consists of 2 fluidized beds. One is a gasifier to convert the
biomass to syngas, the other is a combustor which provides the heat for the gasifier, by
combusting the unconverted biomass.
After the gasification the syngas is cleaned and compressed, the syngas is distributed to
the fuel cell system, which consists of a solid oxide fuel cell [80]. The unconverted syngas
that leaves the fuel cell is combusted, providing heat of which a part is used to preheat the
syngas and the air entering the fuel cell.
3.4 Exergy Calculation Models and Data
61
3.4.10 Cost Data
The data for the cost of electricity and natural gas were obtained from a German website
that specifies real prices for small commercial customers [47]. The data for the cost of
biomass were found at the website of the Biomass Energy Centre, owned and managed by
the UK Forestry Commission [84]. The cost of biomass considered in this work is obtained
by multiplying the cost of natural gas found in [47] by the ratio between the cost of biomass
and the cost of natural gas given in [84].
The price of biomass considered in this study corresponds to the price of wood pellets.
The scheduling computations in this chapter are specified for time periods of 15 min, the
data were adapted accordingly, see Table 3.2. The data are used for the cost optimization.
Table 3.2: Fuel costs
Energy
Carrier
Electricity
Natural Gas
Biomass (wood pellets)
Costs
Use per year
(kWh)
Fixed
(e)
30 001 - 100 000
2 375 - 12 692
not specified
0,0377
0,0208
not specified
Consumption
e
kW·(15min)
0,0474
0,0171
0,0143
3.4.11 Load Data
The load consists of the electricity and heat demand of a district containing 250 houses,
for which the components were designed. The software program SEPATH [49] is used to
obtain the load patterns. After entering information about the type of households, a load
pattern of the electricity and heat demand is generated for a week.
The load data used for the simulations in this chapter were selected from the output data
of the program. For the optimal dispatch example, the following data are used:
Le = 370, 00 kW
(3.29)
Lq = 1352, 50 kW
(3.30)
Γq = 164, 53 kW.
(3.31)
For the scheduling optimization example, the electric load chosen is Le = 255 kW, the
exergetic and energetic heat loads correspond to the outside temperatures; they are shown
in Table 3.3 and Table 3.4.
62
3 The Approach: Energy or Exergy Optimization?
Table 3.3: Heat loads for the scheduling example
Environment Temperature (◦ C)
Lq
Γq
-4
-3
-2
-1
0
1
2
3
4
5
1157,41
140,80
1111,11
135,16
1064,81
129,53
1018,52
123,90
972,22
118,27
925,93
112,64
879,63
107,00
833,33
101,37
787,04
95,74
740,74
90,11
Table 3.4: Heat loads for the scheduling example (continued)
Environment Temperature (◦ C)
Lq
Γq
6
7
8
9
10
11
12
13
14
15
16
694,44
84,48
648,15
78,85
601,85
73,21
555,56
67,58
509,26
61,95
462,96
56,32
416,67
50,69
370,37
45,05
324,07
39,42
277,78
33,79
231,48
28,16
3.5 Simulation Results
In this section the results of the dispatch optimization and the scheduling optimization of
the hub elements are presented, depicted in Figure 3.1 . The result of the exergy hub optimization provides the optimal input exergy flows. These results were converted into energy
flows to allow the comparison among the evaluated cases.
3.5.1 Simulation 1: Optimal Dispatch - Difference among Optimizing
Energetic Efficiency, Exergetic Efficiency and Costs
The optimal dispatch problem is solved applying the exergy hub approach presented in
Section 3.3 and using different objective functions for the optimization.
Case 1: Exergetic Efficiency Maximization
In this case, the objective function that is minimized is inversely proportional to the exergetic efficiency of the system under study, in this way the exergetic efficiency of the system
is maximized. For the exergy hub under study, electricity, natural gas and biomass are the
inputs of the system. The electrical input is divided by the exergetic efficiency of the centralized generation plant εgen,e with which it is produced. The incoming exergy and the exergy
used can be calculated with the following equations:
!
X Ξ
e
(3.32)
+ Ξg + Ξb
Ξinp =
εgen,e
X
Γe + Γq
Γout =
(3.33)
3.5 Simulation Results
63
where subscript ‘e’ refers to electricity, subscript ‘g’ refers to natural gas, subscript ‘b’ refers
to biomass and subscript ‘q’ refers to heat, Ξ is the input exergy flow and Γ is the output
exergy flow of the exergy hub. The resulting objective function is:
P Ξe
+
Ξ
+
Ξ
Ξinp
g
b
εgen,e
=
Fobj1 =
.
(3.34)
P
Γout
Γe + Γq
Case 2: Energetic Efficiency Maximization
In this case, the objective function that is minimized is inversely proportional to the energetic
efficiency of the system under study, in this way the energetic efficiency of the system is
maximized. Here again, the electrical input is divided by the energetic efficiency of the
centralized generation plant ηgen,e :
Pinp =
X
Lout
Pe
+ Pg + Pb
ηgen,e
X
Le + Lq
=
!
(3.35)
(3.36)
where P is the input power flow and L is the output power flow. The resulting objective
function is:
P Pe
+
P
+
P
Pinp
g
ηgen,e
b
Fobj2 =
=
.
(3.37)
P
Lout
Le + Lq
Case 3: Cost Minimization
In this case a cost minimization is performed. Equation (3.38) shows the objective function
with coefficients K1,αi , K2,αi and K3,αi . The coefficient K1,αi does not affect the optimization
results and for this reason it is omitted.
The coefficient K2,αi for each energy carrier is obtained from the data presented in the
last column of Table 3.2, for electricity K2,e = 0, 0474, for natural gas K2,g = 0, 0171
and for biomass K2,b = 0, 0143. The coefficient K3,αi is considered to be 0,001 for all
energy carriers. This value was found acceptable for the electricity coefficient. Due to the
fact that no information about this coefficient was readily available for the other energy
carriers, the same value was used for all of them. The results that were obtained from the
optimization provide enough insight even when using the same coefficient value K3,αi for
all the input carriers, thus this assumption was considered satisfactory for this study. The
objective function is shown below:
Fobj3 =
X
αi ∈Ein
(K1,αi + K2,αi Pαi + K3,αi P2αi ).
(3.38)
64
3 The Approach: Energy or Exergy Optimization?
Table 3.5: Simulation 1: Optimal dispatch
Power input to each hub element
Optimization
Grid
(kW)
H. Pump
(kW)
Gas-fired
CHP (kW)
Furnace
(kW)
Biom.-fired
CHP (kW)
εtot
%
ηtot
%
Cost
e
Input carrier
Case 1 - εtot
Case 2 - ηtot
Case 3 - Cost
e
0,00
0,00
177,28
e
75,00
75,00
351,77
g
876,29
145,00
340,57
g
627,20
130,37
0,00
b
151,99
1365,48
271,41
27,19
26,30
23,16
91,58
92,30
78,24
150,39
218,20
69,18
Results and Discussion
The results that are shown in Table 3.5 are obtained from a program that was developed
by the author of this thesis using the programming software AIMMS. In this program the
equations of the problem statement (including equalities and inequalities) presented in Section 3.3 were implemented. The program was used to carry out three different simulations.
Each simulation contains a different objective function, namely (3.34), (3.37) and (3.38).
The program converges to a solution when the evaluation of the objective function reaches
a minimum. Additionally, the program provides the value of the energetic efficiency, the
exergetic efficiency and the cost for each simulation so that the results of the three cases can
be analyzed and compared.
In this simulation only one load condition is simulated, thus only the optimal dispatch
for that specific load is obtained. As mentioned in Section 3.4.11, the electric load is
Le = Γe = 370, 00 kW and the heat loads are Lq = 1352, 50 kW and Γq = 164, 53 kW.
From Table 3.5 it can be observed that in both Case 1 and in Case 2, the heat pump is used
at its minimum capacity, with a power input of 75 kW. This is because its main input is electricity, which has been produced with a relatively inefficient plant, both from an energetic
and an exergetic point of view. It is interesting to observe that in Case 1, where exergetic
efficiency is maximized, the biomass-fired CHP is used at its minimum capacity, and the
gas-fired CHP is the component with the highest dispatch. Conversely, in Case 2, where
energetic efficiency is maximized, the biomass-fired CHP has the highest dispatch whereas
the gas-fired CHP is working at its minimum capacity. This can be explained by the fact
that from an exergetic point of view the gas-fired CHP is the most efficient component and
from an energetic point of view the most efficient component is the biomass-fired CHP. In
relation to the operating costs it can be observed that for this case, the exergetic efficiency
optimization resulted in lower costs than the energetic efficiency optimization.
It can be observed that according to the results of the optimization, the furnace is not
used at all in Case 3, and both the electricity input from the grid and the electricity-driven
heat pump play an important role in the dispatch, which was not the case in the previous
cases. This can be explained by the fact that the optimization looks only at the cost of the
input fuels or electricity and not if the electricity was produced with high efficient plants or
not. This example shows that the results from a energy optimization can be very different
from those of an exergy optimization. As was expected, due to the different nature of the
objective function, a cost optimization can give considerably different results with respect
to the other two cases.
3.5 Simulation Results
Table 3.6: Simulation 2: Scheduling for different loads - Case 1
Environment Temperature (◦ C)
Component
Grid
Heat Pump
Gas-Fired CHP
Furnace
Biomass-Fired CHP
-4
1
1
-3
1
1
-2
1
1
-1
1
1
0
1
1
1
1
1
2
1
1
3
1
1
4
1
1
5
1
1
6
1
1
7
1
1
8
1
1
9
10
1
1
1
1
11
1
1
12
13
14
15
16
1
1
1
1
1
1
1
1
1
1
12
13
14
15
16
1
1
1
1
Table 3.7: Simulation 2: Scheduling for different loads - Case 2
Environment Temperature (◦ C)
Component
Grid
Heat Pump
Gas-Fired CHP
Furnace
Biomass-Fired CHP
-4
1
1
-3
1
1
-2
1
1
-1
1
1
0
1
1
1
1
1
2
1
1
3
1
1
4
1
1
5
1
1
6
1
1
7
1
1
1
8
1
1
1
9
10
11
1
1
1
1
1
1
1
1
1
1
1
1
65
66
3 The Approach: Energy or Exergy Optimization?
3.5.2 Simulation 2: Optimal Scheduling - Difference among Optimizing Energetic Efficiency and Exergetic Efficiency
The main objective of this section is to show that in complex systems with several components and multiple energy carriers, the optimal scheduling of units (multi-carrier unit
commitment) coming from an exergetic efficiency optimization may vary from an energetic
efficiency optimization. In order to solve the scheduling optimization, several algorithms
were added to the problem statement described in Section 3.3.2. The multi-carrier unit
commitment algorithms were programmed according to the method described in Chapter 2.
The optimization is solved with a mixed-integer solver due to the ‘on’ and ‘off’ states of
the units. The objective functions that were used in Section 3.5.1 are used to obtain the
multi-carrier unit commitment.
Results and Discussion
The heat load is varied in order to show the effect on the scheduling of the components
with respect to the exergy and energy optimization. The electric load is kept constant at
Γe = Le = 255 kW and the outside temperature is changed in order to increase or decrease
the heat demand of the group of houses considered. The heat load values can be found in
Table 3.3 and Table 3.4. It can be observed in Table 3.6 and in Table 3.7 that for temperatures equal and below 6 ◦ C in both cases the hub was optimized in a way that the most
efficient component delivers the total demand in combination with the furnace (exergetically
or energetically efficient respectively).
In Case 1, where exergetic efficiency is optimized, it is the gas-fired CHP the one that
is scheduled. Conversely in Case 2, where energetic efficiency is optimized, the biomassfired CHP, which has the higher energetic efficiency, is the one that is scheduled. It is
interesting to observe that at higher temperatures, where the heat demand is lower, both
optimization solutions give the same scheduling: the gas-fired CHP and the furnace supply
the load. This can be explained because of the load ratio: the electricity demand is equal or
higher than the heat demand and thus, the higher electricity conversion efficiency of the gasfired CHP in relation to the biomass-fired CHP is more important than the overall efficiency
of the biomass-fired CHP in comparison to the gas-fired CHP.
This small comparison is just an example of the possibilities of scheduling and analysis
that the scheduling tool presented in this chapter provides. Further sensitivity analyses can
be performed to evaluate the influence of other conditions or parameters, like for example
the efficiency of conversion of the external generation plants.
There are still many disagreements and uncertainties in the scientific community about
the influence or advantages that an exergy analysis can provide in relation to an energy
analysis and if practical differences can be obtained. This is a valuable example because
it shows that the results actually vary: the most energetically efficient solution does not
always fit the results of an exergy optimization. This means that depending on the goal of
the system designers, different configurations can be believed to be the optimal one. If we
were to obtain the maximum potential from the available sources, then we would need to
follow the results of an exergy optimization, however this does not necessarily mean that
costs will be reduced.
3.6 Conclusions
67
3.6 Conclusions
We are all responsible of finding ways to make a better use of available resources. In this
chapter the exergy hub approach was introduced. In the exergy hub approach, exergy efficiencies (instead of energy efficiencies) are taken into account for the optimization of
systems that contain multiple energy carriers.
A comparison of results obtained by using different objective functions is shown in
Section 3.5.1 and Section 3.5.2. From the results it was observed that the dispatch and
scheduling of components differ considerably when performing an exergy optimization in
relation to an energy optimization. It was observed that the costs were lower for the case in
which the exergetic efficiency was maximized, with respect to the case in which the energetic efficiency was maximized. However, this applies only to the case that was analyzed,
thus no direct relation between an energy or exergy optimization and a reduction of costs
can be established as a conclusion.
Due to its flexibility, the scheduling tool that was presented in this chapter can be used
to evaluate different configurations and control strategies of systems with several generation
units and multiple energy carriers. The tool can be easily adapted to evaluate configurations
designed for the built environment, where research in the topic of exergy has increased
during the last years. The strength of the tool is that it can be used in complex systems,
like energy hubs with many links between components, or in annual simulations where the
calculations become extremely tedious when done by hand.
Exergy analysis is a good tool to determine where the losses of the system are located.
Moreover, since exergy is the maximum theoretical work that can be obtained from an energy flow, by choosing the most exergetically efficient configuration we are making the best
use of the work potential of the energy sources available. Unfortunately, the existing generation systems were not designed to make the best use of the work potential of the sources
and in many cases, sources with high work potential are used to supply low temperature
heat; those processes are characterized by having a high entropy production. Thus, in order
make a better use of the work potential of the sources, it would be necessary to re-evaluate
the existing equipment/machinery and re-design generation units in general according to an
exergy-oriented criteria. This is an important challenge for the future.
This dissertation concentrates on the optimization of systems that contain traditional
and state-of-the-art equipment, however this equipment was not designed taking exergy into
account. Furthermore, the focus of the rest of the dissertation is given to the optimization
of total costs and to keep track of the primary energy source utilization, thus even though
the tool provided valuable information about how to dispatch and schedule units to attain
a higher exergetic efficiency in the system, the energy hub was selected as the appropriate
approach to be used in the rest of the chapters.
Chapter 4
The Control
Multi-Carrier Hierarchical Control Architecture
This chapter answers Research Question 4 of this dissertation. In Chapter 2 a general unit
commitment framework for systems with multiple energy carriers was presented. Nevertheless, the topic of real-time control also deserves to be studied, since mismatches will exist
between the forecasted values and the actual ones.
This chapter presents an integrated control architecture for multiple energy carrier systems. The application of the control architecture is illustrated with an example. Section 4.1
contains basic concepts related to the control architecture. Section 4.2 includes a literature
review of control strategies for decentralized energy supply systems. In Section 4.3 the
integrated control architecture is described. Section 4.4 briefly describes the models that
were used and their link to the control architecture. Section 4.5 presents results obtained by
applying the control architecture to an example. Finally in Section 4.6 the conclusions of
this chapter are included. Parts of this chapter have been published in [85].
4.1 Basic Concepts and Definitions
In this chapter several models of heat components are used, for this reason some basic heat
transfer concepts and equations are introduced. Moreover, the definition of participation
factors is included. Energy can be transferred in three ways: conduction, convection and
radiation. In this work only conduction and convection are considered.
4.1.1 Participation Factors
Due to the performance of control actions, when a mismatch exists between the predicted
demand and the actual demand, the units involved participate at compensating this mismatch. A participation factor defines the percentage in which each of the units has to
participate in order to reduce the error due to the mismatch.
69
70
4 The Control: Multi-Carrier Hierarchical Control Architecture
4.1.2 Law of conservation of energy in an open system
The total energy change per unit of time [W], or power, is equal to the result of adding the
power that flows into the system in the form of γ, given by φm,γin uγin [W], the power that
goes out of the system, given by φm,γout uγout [W]), the power added Φin [W] and the heat
exchanged with the surroundings per unit of time Φout [W]:
dE
= φm,γin uγin − φm,γout uγout + Φin − Φout
(4.1)
dt
where φm,γin [kg/s] is the mass flowing into the system , φm,γout [kg/s] is the mass flowing out
of the system, u [J/kg] is the specific internal energy and t [s] is time. The total energy E [J]
and the specific internal energy u [J/kg] can be expressed by:
3
E = ρVcT
(4.2)
u = cT
(4.3)
3
where ρ [kg/m ] is density, V [m ] is volume, c [J/(K·kg)] is specific heat capacity and
T [K] is temperature. Equation (4.4) results from substituting the energy and specific
internal energy expressions in (4.1). It is assumed that no mechanical power is being added
(Φin = 0) and that the input and output mass flows are the same (φm,γin = φm,γout = φm,γ ) :
dT
mc
(4.4)
= φm,γ cγ T in − T out − Φout
dt
where m [kg] stands for mass (m = ρV). If the system is stationary, the heat exchanged with
the surroundings Φout , denoted from now by Φq , is equal to the heat transferred by the mass
flow φm,γ :
(4.5)
Φq = φm,γ cγ T in − T out .
4.1.3 Heat Conduction
Heat conduction occurs when heat is transferred by interactions between atoms/molecules,
but there is no transport of atoms or molecules themselves [51]. The equation that describes
the heat conduction process is:
κ
(4.6)
Φq = A∆T
d
where Φq [W] is heat transferred per unit of time , κ [W/(m·K)] is the thermal conductivity,
d [m] is the thickness of the wall, ∆T [K] is the temperature difference that drives the heat
transfer phenomena and A [m2 ] is the area of the wall.
4.1.4 Heat Convection
Heat convection occurs when heat is transferred through the transport of material. The
equation that represents this process is given by:
Φq = U A∆T
(4.7)
where Φq [W] is the heat transferred per unit of time, U [W/(m2 ·K)] is the heat transfer
coefficient, A [m2 ] is the area and ∆T [K] is the temperature difference that drives the heat
transfer phenomena.
4.2 Literature Review
71
4.2 Literature Review
This section summarizes several papers and reports where control architectures for distributed power systems are proposed. Most of the architectures have two-level hierarchical
configurations and only take electrical parameters and electrical interactions into account.
For example, in [86] a droop control method is applied on a system that contains renewable
energy generators and storage. In [86], the first level is called PQ Droop Control Method
and the second level is called Management of the Distributed Power Station. The first level
manages the power output of each unit separately, while the second level monitors the whole
system and controls the set-points of the units depending on the state of charge of the battery. The control is designed for two operating conditions: normal interconnected mode
and emergency mode. In both cases the control unit optimizes the power output of the generators by communicating new droop settings based on the information collected from the
inverters, decentralized generation units and battery banks. However, the selection of the
suitable droops is a trade-off between the stability margin, dynamic performance and shifts
in the droop operating point [86].
Another example can be found in [87], where control and power management strategies
based on locally measured signals without communication were proposed under various
micro-grid operating conditions. In the paper, the active power of each decentralized generation unit is controlled based on a frequency droop characteristic and a frequency restoration
strategy [87].
Apart from the control techniques based on droop mechanisms, other control schemes
were also found in the literature. For example in [88] a supervisory hybrid control scheme
for a micro-grid system using hybrid control techniques was presented. The hierarchical
control consists of a supervisory controller at the top level that interacts with the unit level
regulators. The supervisory control regulates the operation of the micro-grid by means of
transition management schemes. The predetermined routes for transition to other operating
states (i.e on-grid, off-grid, etc) are given by a finite hybrid automata representation of the
micro-grid system. In their work the micro-grid was partitioned in modules to reduce the
complexity of the problem. The control technique is applied to a wind energy conversion
and storage system. By means of an example the authors show how a voltage fall higher
than 0,7 per unit causes the supervisory controller to initiate a transition from the on-grid
operating state to the off-grid operating state.
Additional actions, like the use of storage and load shedding have also been included
in the control mechanisms for future power systems and micro-grids. For example, [27]
describes and evaluates the feasibility of two control strategies needed for islanded operation
of micro-grids. Particular attention is given to storage devices and load shedding strategies.
The concept of using a control scheme based on droop concepts to control inverters in an
off-grid AC system was studied by using two different control strategies: Single Master
Operation and Multi Master Operation, where several inverters are operated with voltage
source inverter control. A micro-grid central controller, which is at the top control level, is
installed at the medium/low voltage substation. A second hierarchical control level, located
at the loads, groups of loads and micro sources, exchanges information with the micro-grid
central controller, which provides the respective set-points. In the examples a fuel cell was
included, which is a combined heat and power device; however no attention was given to
the heat control or heat flows in general in their approach.
72
4 The Control: Multi-Carrier Hierarchical Control Architecture
Other schemes have been found in which the main objective is to support the grid. For
example, in [89] a control strategy is proposed in which inverters are used for grid-forming,
grid-supporting and grid-parallel operation. The strategy is applicable to interconnected
grids. The control architecture consists of three control levels: unit control and local control
to regulate and maintain voltage and frequency, and a main supervisory control to optimize
and control power dispatching and load sharing. These controls are based on conventional
primary, secondary and tertiary control respectively, described in Section 1.3.2.
From the literature reviewed, it was observed that a two-level architecture can be adjusted to fulfill the needs of future power systems, where decentralized generation will play
a more important role. Nevertheless the topic of integrated control for systems with multiple
energy carriers like electricity and heat has not been broadly studied yet. For this reason this
topic will be addressed in this chapter.
4.3 Integrated Control Architecture
The integrated control architecture that is adopted in this dissertation is hierarchical. It
contains two levels, as this was found to be well-accepted in recent literature for distributed
energy systems. Nevertheless, it is not limited to electrical parameters but also includes
parameters related to the flows of other energy carriers involved. The names given to the
two levels are: unit control and main control. The nomenclature introduced in Chapter 2 is
used in this chapter. In Figure 4.1 the two-level architecture is depicted.
4.3.1 Unit Control
The unit control is a local control, which can be embedded at each controllable unit. It
works independently at each of the units, but it can receive control signals from the main
control. The parameters required to perform the unit control are monitored locally at the
unit. In this context, a unit can be an individual component like a battery, but it can also be
a group of components if these components work as a whole and have a common controller.
An example can be a fuel cell system equipped with a battery. This fuel cell system can
have its own local control, which for example can prevent the fuel cell from suffering sharp
current changes by redirecting them to the battery. Since the fuel cell and the battery work
as a whole in this example, they are considered to be a single unit.
4.3.2 Main Control
The main control regulates the whole system. It is in charge of monitoring the system,
allocating the generation output and bringing the system’s relevant parameters, or control
parameters back to their nominal value. The main control commands the execution of the
multi-carrier optimal dispatch and multi-carrier unit commitment. Furthermore, the main
control allocates load changes to the controllable units. In the case of applying demand-side
management, the main control also regulates the controllable loads. In the case of off-grid
electrical systems more considerations must be taken into account than in the case of gridconnected systems, since there should be a master converter in charge of keeping the voltage
and frequency of the system within acceptable limits.
4.3 Integrated Control Architecture
Figure 4.1: Two-level hierarchical control architecture
73
74
4 The Control: Multi-Carrier Hierarchical Control Architecture
The power to be generated by each component is calculated through a process that starts
by determining the mismatch between the system’s control parameter and its nominal value.
The sign of the error signal indicates the proper corrective action to be executed, i.e. to
increase or decrease the power supply. The error signal can feed a proportional or integral
controller in order to determine the power that must be supplied or curtailed to bring the
control parameter back to its nominal value.
The main control is divided into control subsystems. Each control subsystem is associated with an energy carrier β j from the output side of the energy hub. In the case of a
hub with two energy carriers at the output side, like electricity and heat, the main control
has two control subsystems, one for electricity and another one for heat. Each generation
unit is assigned to a particular control subsystem. The particular energy carrier is selected
according to the requirements of the system. For example in an off-grid system, a good
operation of the electricity control subsystem is crucial for the stability of the entire system;
particularly because the frequency and the voltage level are defined by a master converter,
and not by the electrical grid. In this case all cogeneration units producing electricity at
one of the outputs are assigned to the electricity control subsystem, and therefore, regulated
with respect to their electrical output. The main control allocates the power difference between the forecasted and actual values to the units of all control subsystems by means of
participation factors. Each controllable unit has its own participation factor, given by:
k
k
pf,βp j
Lβpj ,sch /Lβ j ,rtd
k
+ pf,βp j ,min
= P kp
Lβ j ,sch /Lβ j ,rtd
(4.8)
k p ∈Cα
k
k
where pf,βp j [-] is the participation factor of unit k p at control subsystem β j , variable Lβpj ,sch is
k
the scheduled power output in the form of energy carrier β j of controllable unit k p , pf,βp j ,min
is the minimum participation of unit k p (it can be set to zero) and Lβ j ,rtd is the rated output
power. The rated output power can be calculated by adding up the rated power of all the
P k
converter units associated with energy carrier β j , this is equal to: Lβ j ,rtd = Lβpj ,rtd . Power
is given in kW.
k
The desired power output calculated from the control actions Lβpj ,des of unit k p is:
k
k
k
Lβpj ,des = Lβpj ,sch + pf,βp j ∆Lβ j ,dif
(4.9)
where ∆Lβ,dif represents the difference between the total scheduled power output and the
actual power demand. In the case of storage devices, the following aspects are considered:
• The availability to charge or discharge is evaluated before the main control sends
the correction signals. The storage device is available to charge or discharge when
the state of charge is lower or higher than the upper or lower state of charge limit
respectively. If the requirement is not met, then the storage is set to stand-by. The
output power of the storage devices is calculated using magnitudes, not vectors.
• If the scheduled storage action is to charge, but the system requires an increase in
supply, the control system passes a zero to (4.9). Likewise, if the scheduled action is
to discharge but the system requires a decrease in supply, a zero is passed to (4.9).
4.3 Integrated Control Architecture
Figure 4.2: Flowchart to compute participation factors
75
76
4 The Control: Multi-Carrier Hierarchical Control Architecture
Flowcharts will be used to describe and explain the actions taken by the control subsystems that are part of the main controller. A general flowchart is presented in Figure 4.2, but
it can be adapted for any control subsystem considered. Figure 4.2 depicts the first section
of the control sequence, which depends on the status of the main control parameter of the
control subsystem. The first decision determines if an increase or decrease in the power generation is required. After the total increase or decrease of power generation is determined,
the participation of each unit is defined by means of its participation factor, introduced in
(4.8), and the storage considerations previously mentioned.
The flowchart that shows how the storage dispatch is executed is shown in Figure 4.3.
In this case the actual corrective action required by the system and the scheduled storage
charging status are compared. In the case that the charging status does not match, e.g. the
system requires generation to be decreased, but discharging is scheduled, the control passes
a zero value, otherwise the scheduled storage dispatch is used to compute the participation
factors. Moreover the control can identify if the storage device can be charged or discharged
according to its state of charge and energy content.
Figure 4.3: Flowchart to define the participation of the storage elements
4.3 Integrated Control Architecture
77
After the participation factors are calculated, the main control sends the desired setpoints to each unit per control subsystem according to (4.9), see Figure 4.4.
Figure 4.4: Flowchart to calculate the desired power output
4.3.3 Unit Control Applied
The system shown in Figure 4.5 was selected for the application of the integrated control. It
consists of 2 gas turbines (CHP), a solid oxide fuel cell (CHP), a furnace, a wind turbine, a
battery bank, a heat storage tank and a district heating infrastructure, which is used to model
the heat demand. The system operates off-grid.
At first, it is important to identify which units are controllable and what type of unit
control each of them possesses, since they may vary considerably with respect to each other.
A description of each unit control is given in Section 4.4. The description includes the type
of control, the aim of the control, the control parameter that serves as trigger and the control
actions that are undertaken. The models that are used to represent each of the hub elements
in the simulation can also be found in Section 4.4.
4.3.4 Main Control Applied
The main control monitors the system and takes actions in order to keep the control parameters at the nominal value. As mentioned in Section 4.3.2, there is a control subsystem for
every energy carrier β j at the output side of the energy hub, in this example they correspond
to electricity and heat.
The control parameters are the frequency, which must remain at 50 Hz for the electrical system, and the temperature, which must remain at 100◦ C at the district heating hot
water line. Since the system is not connected to the electrical grid, all components that produce electricity are assigned to the electricity control subsystem, even though they might
also produce heat. The rest of the components are assigned to the heat control subsystem.
Specific details regarding the control parameters and error signals considered at the main
control for both control subsystems are presented in Table 4.1.
78
4 The Control: Multi-Carrier Hierarchical Control Architecture
Figure 4.5: Energy hub used in the example
Table 4.1: Main control summary
Control Subsystems
Aspect
Heat Subsystem
Electricity Subsystem
Control Type
Proportional controller
Droop-based control
Main control parameters
Hot water pipeline temp.
Nominal value: 100 ◦ C
System’s frequency
Nominal value: 50 Hz
Other variables
Indoor temperature
System’s voltage
Error signal
Simple error
Combined error: frequency
and load error
Error signal sign
Positive: Increase supply
Negative: Decrease supply
Positive: Increase supply
Negative: Decrease supply
Storage dispatch
Positive: Charge
Negative: Discharge
Positive: Charge
Negative: Discharge
Minimum participation factor
Storage: 0,1
Furnace: 0,1
Storage: 0,1
Gas turbine CHP : 0
SOFC CHP : 0
4.4 Dynamic Component Models and Controls
79
4.4 Dynamic Component Models and Controls
Several of the component models presented in this section were implemented with the help
of the MSc project [90], which was performed under the framework of this PhD project
and under the guidance of the author of this dissertation. The models are introduced in this
section because they are part of the system where the integrated control is tested. Each
subsection is dedicated to a particular generation unit, where first some generalities of the
unit are given to familiarize the reader, then the model of the unit is presented and finally
the unit control and the main control link are described according to the control structure
introduced earlier in this chapter. The models to be described in this section are the wind
turbine, the battery, the gas-fired turbine (CHP), the gas-fired solid oxide fuel cell (CHP),
the gas-fired furnace and the heat components of the district heating system that are used to
represent the heat load.
4.4.1 Wind Turbine
Overview
An horizontal axis wind turbine with a nominal power of 120 kW was chosen for the simulations. Horizontal axis wind turbines are by far the most common form of wind turbines manufactured today [24]. It consists of a rotor (containing the blades and the hub),
a low-speed rotating shaft, a gearbox, a high speed rotating shaft, a generator and a power
converter. A diagram of the wind turbine is shown below, based on [91]:
1. Foundation
2. Connection to electrical grid
3. Access ladder
4. Tower
5. Wind orientation control
6. Nacelle
7. Generator
8. Anemometer
9. Brake
10. Gearbox
11. Rotor blade
12. Blade pitch control
Figure 4.6: Wind turbine
13. Rotor hub
The kinetic energy of the wind is captured by the blades and is converted to rotational
mechanical energy. This is transmitted to a gearbox that is connected to a high-speed shaft
that turns the rotor of the electricity generator. Many of these wind turbines have an embedded control that is linked to the power converter in order to let the wind turbine work in
optimal power generation mode.
80
4 The Control: Multi-Carrier Hierarchical Control Architecture
The kinetic energy contained in the wind is given by the relation:
1
m 32
(4.10)
2 wnd wnd
where mwnd [kg] is the wind mass and 3wnd [m/s] is the wind speed.
The mass of air that flows per second φm,air [kg/s] can be obtained by multiplying the air
density ρair [kg/m3 ], the wind speed 3wnd [m/s] and the area swept by the rotor Arot [m2 ]:
Kinetic Energy =
φm,air = ρair Arot 3wnd .
(4.11)
After substituting (4.11) in (4.10), the power contained in the wind is obtained. The
power in the wind is equal to the kinetic energy per second [J/s] and is given in Watts [W]:
1
ρair Arot 33wnd .
(4.12)
2
Not all this power can be extracted by a wind turbine due to the losses associated with
the energy conversion process. The model that represents the power extracted from the rotor
blades is described below.
Pwnd =
Model
Low-speed shaft For simplicity and due to the fact that some of the values that were needed
are unknown and are not provided by the manufacturer, the stiffness of the shaft is
assumed to be infinite.
Rotor The equation used to compute the power extracted from the rotor blades is:
1
cp ρair Arot 33wnd
(4.13)
2
where cp [-] is the performance coefficient of the rotor efficiency and represents the
fraction of wind power that is captured by the rotor blades. In practice, this coefficient
has a value between 0,4 and 0,5 in the case of high-speed two-blade turbines, and
between 0,2 and 0,4 in the case of low-speed turbines. Its maximum theoretical value
is 0,59 [92]. The performance coefficient can be expressed as a function of the tip
speed ratio λ and the pitch angle θ; C1 , C2 , C3 , C4 , C5 and C6 are constants. The
performance coefficient can be expressed by:
!
C
C2
− 5
cp = C1
− C3 θ − C4 exp λi +C6 λ
(4.14)
λi
Pwnd =
where
1
0, 035
1
=
−
.
λi λ + 0, 08θ θ3 + 1
(4.15)
Induction Generator In this study, a simple generator model is used. This model is given
in per unit. The dynamics of the generator are modeled by the acceleration equation:
dω
1 τmec − τele
(4.16)
=
dt
2J
where ω is the rotational speed, J is the moment of inertia, τmec represents the mechanical torque and τele represents the electrical torque.
4.4 Dynamic Component Models and Controls
81
Unit Control
Even though the wind turbine is considered to be an uncontrollable unit because its source
of energy is the wind, which changes stochastically, there are two parameters that fall into
the unit control: the rotor speed and the pitch angle. The description of the unit control is
given below.
Rotor Speed Control
Function: keeps the optimal power supply of the wind turbine.
Trigger: responds to a rotor speed error signal.
Action: adjusts the rotor speed by modifying the electromagnetic torque.
Pith Angle Control
Function: performs power limitation by keeping the desired rotor speed.
Trigger: responds to a rotor speed error signal.
Action: adjusts the wind turbine blades.
Main Control Link
There is no control signal sent from the main control, it is assumed that the power contribution of the wind turbine is supplied to the system as it comes.
4.4.2 Lead-Acid Battery
Overview
The electricity storage selected consists of deep-cycle lead-acid batteries. The battery bank
has a nominal battery stack capacity of 1250 Ah. Lead-acid batteries are the most commonly used batteries in electrical power system applications, including applications where
renewable energy technologies are present. Moreover, this kind of batteries has been widely
used in the automotive industry. Lead-acid batteries have a wide range of sizes as well as
an acceptable cost in the market. In [93], several advantages and disadvantages of batteries
were gathered. Here only a selection is listed from [93–95]. These are some advantages:
• There are many sizes and designs available.
• The technology is well-understood, mature and reliable.
• It is possible to easily determine their state of charge.
• Low self discharge and low maintenance.
• The components are easily recycled.
82
4 The Control: Multi-Carrier Hierarchical Control Architecture
On the other hand, some disadvantages of lead-acid batteries are:
• Low cycle life.
• Only a limited full discharge cycles are allowed.
• They have low energy density.
• Materials can be toxic when they evaporate, moreover lead and the electrolyte used
are environmentally unfriendly.
• No fast charging is allowed.
Model
The battery model available at the library of MATLAB Simulink SimPowerSystems was
used in this work.
Unit Control
No unit control is applied. The control signals come directly from the main control.
Main Control Link
The main control sends a control signal to the power converter that indicates if the battery
must charge or discharge and how much according to the storage considerations.
4.4.3 Gas-Fired Turbine
Overview
Two gas turbines with a nominal electric power output of 100 kW each were selected for
the example. The type of turbine considered in this work has a split-shaft design, which according to [96], can be modeled as a single-shaft turbine. Rowen [97] provided a simplified
single-shaft gas turbine model that has been widely used in the literature. A block diagram
of this model can be observed in Figure 4.7. In the following subsections each block in the
diagram will be described. The PID control and the excitation control are described under
unit control. This model is given in per unit.
Figure 4.7: Schematic of the micro-CHP gas-fired generator
4.4 Dynamic Component Models and Controls
83
Model
Fuel System The input to the fuel system is the fuel command. The fuel command results
from multiplying the governor output and the per unit mechanical rotor speed ω. In
this study, pressure control may be neglected because of the small volume between
these valves [97]. The fuel system consists of two valves in series: one corresponding
to the valve positioning system and the other one to the actuator fuel system. Each
subsystem has a constant associated with it. The constant τV is associated with the
valve positioning system, and the constant τA is associated with the volumetric time
constant related to the downstream piping and fuel gas distribution manifold [97].
The fuel flow is the output of this block.
Compressor-Combustor-Turbine Assembly The fuel flow is the input for the compressorcombustor-turbine assembly. The model incorporates three transport delays, one associated with the compressor discharge volume τCD , another associated with the combustion reaction time τCR and another associated with the gas from the combustion
system through the turbine τTD . The turbine torque τtur and the turbine exhaust temperature T exh are given by [97]:
τtur = 1, 3 φm,fue − 0, 23 + 0, 5 (1 − ω)
(4.17)
T exh = T rtd − 390(1 − φm,fue ) + 306(1 − ω)
(4.18)
where T rtd is the rated exhaust temperature of the turbine and φm,fue is the fuel flow.
The subtraction of 0,23 from the fuel flow represents the percentage of fuel needed to
keep the gas turbine under operation at a nominal speed, no-load condition.
Synchronous Generator The torque τtur that results from the compressor-combustor-turbine
assembly model is the input for the synchronous generator model. The built-in model
of a synchronous generator that is available at the library of MATLAB Simulink SimPowerSystems is used for the simulation.
Exhaust Heat Recovery System The exhaust temperature that results from the compressorcombustor-turbine assembly model T exh is the input for the exhaust heat recovery system, which consists of a recuperator and a shell and tube heat exchanger in countercurrent configuration. The exhaust gases that come from the recuperator are transported to the heat exchanger.
The recuperator exhaust temperature T rec can be calculated using the following equation:
T rec = 0, 405(T exh − T amb ) + T amb
(4.19)
where T amb is the ambient temperature and the coefficient 0,405 results from a calculation that involves the inlet and outlet exhaust heat flow at rated values, the turbine
exhaust temperature, the recuperator exhaust temperature and the exhaust gases mass
flow taken from the micro-CHPs datasheet. The heat transport takes place at the contact area between the hot-side and cold-side exchanger chambers.
The equations that were introduced for the convection and conduction processes in
Section 4.1 are used in this section to model the heat recovery heat exchanger. The exhaust gases coming from the recuperator enter the hot chamber of the heat exchanger
84
4 The Control: Multi-Carrier Hierarchical Control Architecture
and warm up the water coming from the return pipe of the district heating system.
The hot side of the heat exchanger can be described by the following equation:
mexh cexh
dT out,hot
dt
= φm,exh cexh T in,hot − T out,hot − Φq,hex
(4.20)
where the incoming temperature at the hot side of the heat exchanger T in,hot is equal
to the recuperator exhaust temperature T rec obtained with (4.19).
The cold side is represented by:
mwat cwat
dT out,col
dt
= φm,wat cwat T in,col − T out,col + Φq,hex .
(4.21)
In both cases Φq,hex is given by:
Φq,hex = Uhex Ahex ∆T LMTD,hex
(4.22)
where subscript ‘hex’ indicates CHP heat exchanger, subscript ‘exh’ represents the
exhaust gases coming from the recuperator of the CHP unit, subscript ‘wat’ represents
the water flowing through the return pipe of the district heating system, subscript
‘hot’ represents the hot side of the heat exchanger and subscript ‘col’ represents the
cold side of the heat exchanger. The equations are based on (4.4) and (4.7). The
logarithmic mean temperature difference ∆T LMTD,hex is given by:
∆T LMTD,hex
T in,hot − T out,col − T out,hot − T in,col
=
.
T
−T
ln T in,hot −Tout,col
out,hot
(4.23)
in,col
Unit Control
This unit is equipped with a PID control and an excitation control. The excitation control is
a built-in MATLAB Simulink model.
PID Control
Function: keeps the desired frequency and desired power dispatch.
Trigger: responds to rotor speed error and load error signals.
Action: adjusts the fuel injection.
A Proportional-Derivative-Integral (PID) control that acts on the speed and load errors
is selected for the gas turbine. The gains KP , KI and KD are tuned in order to obtain
the desired response of the turbine-generator set. In this dissertation, the PID gains
that are used are the ones obtained in [90] by using the method suggested in [98] with
a step signal as system load input. The value of the governor output is the steady state
per unit value of the turbine’s mechanical power.
The speed error and the load error are obtained by subtracting the actual values to the
respective assigned reference values. The generators work in droop mode: they share
the load changes according to their droop. The droop gain that was selected for the
4.4 Dynamic Component Models and Controls
85
generators is 4%, i.e. a hundred percent change on the output power will result in a
four percent change in the frequency. The deviation in the frequency due to droop
control is corrected by changing the operating point of the governor. This is done by
means of adjusting the load reference set-point. In this way the desired frequency can
be recovered. The minimum limit represents the gas turbine’s capability of absorbing
power from the connected system while maintaining fuel burning in the combustor
[97], whereas the maximum limit accounts for the engine temperature control.
Excitation Control
Function: keeps the desired excitation voltage level.
Trigger: responds to a voltage error signal.
Action: adjusts excitation.
The generator is connected to the built-in excitation model to provide field voltage
control. The model is available at the library of MATLAB Simulink SimPowerSystems. The model regulates the terminal voltage of the generator at a reference value.
Main Control Link
The set-point of the gas input comes from the main control. The participation of the combined heat and power unit is calculated by the main control as described in Section 4.3.2.
4.4.4 Solid Oxide Fuel Cell
Overview
The interest in fuel cells has increased during the last decade for several reasons, among
them: their high efficiency in comparison to heat engines, the very low (or zero) harmful
emissions, the low maintenance costs, the lack of vibrations and the very low noise emissions [99]. In this work a solid oxide fuel cell (SOFC) of a rated electric power output of
75 kW is considered. Solid oxide fuel cells are sometimes called the third generation of
modern fuel cells [99]. The temperature at their stack ranges between 650◦C and 1000◦C.
Due to their high temperatures, they are intended for stationary applications. Their power
output ranges from a few kWs to several MWs. The high quality heat can be used for
cogeneration purposes.
One of the main advantages of SOFCs is that because of the high temperatures involved,
they do not need expensive catalysts and do not get poisoned by carbon monoxide, allowing fuel flexibility, however sulfur components must be removed before they enter the fuel
cell. Furthermore, the high temperatures are linked to their main disadvantages: they have
slow start-up times (when starting from cold) and the high temperatures produce corrosion
of components. For this reason, research is nowadays focused on lowering the operating
temperature down to 600◦C in order to reduce the strict material requirements and reduce
the start-up times [100].
86
4 The Control: Multi-Carrier Hierarchical Control Architecture
Fuel cells have a basic structure consisting of an electrolyte and two electrodes. The
main functions of the electrolyte are to act as an ion conductor so that ions can migrate from
one electrode to another, to act as an electron insulator, so that electrons are forced to flow
through the external circuit and to act as a barrier to separate the reactants [99]. On the other
hand the main functions of the electrodes are: to provide a place for the electrochemical
reaction to occur, to collect the electrons and to provide a flow path for the electron transfer.
The fuel cell voltage is equal to the potential difference between the cathode and the anode
and the fuel cell’s current is defined as the current that flows from the cathode to the anode
through the external circuit (opposite to the electron flow), to which the load is connected.
Figure 4.8 shows a graphic representation of the SOFC.
Electric Current
e
e
Fuel In
e
H2
O
Air In
e
O
Cathode
Electrolyte
H2 O
Excess
Fuel and
Water
Anode
O2
Unused
Gases
Out
Figure 4.8: Schematic of the solid oxide fuel cell
Model
Fuel Cell Model The fuel cell model available at the library of MATLAB Simulink SimPowerSystems was used in this work.
Heat Recovery System The heat exchanger is modeled using the same equations that were
used to model the heat exchanger of the gas turbine. The hot side of the heat exchanger can be described by the following equation:
mair cair
dT out,hot
dt
= φm,air cair T in,hot − T out,hot − Φq,hex
(4.24)
the air is heated thanks to the fuel cell’s stack.
The cold side is represented by:
mwat cwat
dT out,col
dt
= φm,wat cwat T in,col − T out,col + Φq,hex .
(4.25)
In both cases Φq,hex is given by:
Φq,hex = Uhex Ahex ∆T LMTD,hex
(4.26)
4.4 Dynamic Component Models and Controls
87
where subscript ‘hex’ indicates SOFC heat exchanger, subscript ‘air’ represents the
hot air coming from the stack of the SOFC, where the high temperatures are produced, subscript ‘wat’ represents the water flowing through the return pipe of the
district heating system, subscript ‘hot’ represents the hot side of the heat exchanger
and subscript ‘col’ represents the cold side of the heat exchanger. The equations are
based on (4.4) and (4.7). The logarithmic mean temperature difference ∆T LMTD,hex
is given by:
T in,hot − T out,col − T out,hot − T in,col
∆T LMTD,hex =
.
(4.27)
T
−T
ln T in,hot −Tout,col
out,hot
in,col
Unit Control
A simple delay transfer function is used as unit control for the fuel cell.
Main Control Link
Due to the fact that fuel cells are DC devices, the SOFC is connected to the system by means
of an inverter. The inverter receives the power/current settings from the main control.
4.4.5 Furnace
Overview
The furnace model is based on [101] and was implemented and described in [90]. A furnace
of a nominal power of 550 kW was selected. The furnace system consists of the furnace and
a heat exchanger that transfers heat coming from the flue gases of the furnace to the district
heating water.
Model
Furnace The heat that comes from the furnace Φq,fur can be obtained by using (4.5). The
factor ξfur [-] is added to this equation to represent the losses in the furnace:
Φq,fur = φm,flu cflu T in,fur − T out,fur ξfur
(4.28)
where φm,flu is the mass flow of the flue gases, cflu represents the specific heat capacity of the flue gases, T in,fur is the predicted combustion temperature during the fuel
combustion adiabatic process, also denoted T aft and T out,fur represents the temperature
of the gases at the exit of the furnace. The equations that are used to calculate the
combustion temperature and the temperature of the gases at the exit are given by:
ΥLHV,fue + rafr ιexs cair T air − 80
(4.29)
T in,fur = T aft =
1 + rafr ιexs cflu
T out,fur = T aft ǫfur
(4.30)
88
4 The Control: Multi-Carrier Hierarchical Control Architecture
where ΥLHV,fue is the lower heating value of the fuel, rafr is the stoichiometric air fuel
ratio, ιfur [-] is a factor related to the excess air and T air is the temperature of the outside
air and ǫfur [-] is a dimensionless furnace parameter that is based on experimental data.
The mass flow of the flue gases φm,flu results from the addition of the fuel mass flow
φm,fue and the air mass flow φm,air :
φm,flu = φm,fue + φm,air
(4.31)
where the mass flow of the air depends on the stoichiometric air fuel ratio rafr and a
factor related to the excess air ιexs :
φm,air = φm,fue rafr ιexs .
(4.32)
Unit Control
No unit control is included. The set-points come directly from the main control.
Main Control Link
The set-point for the gas flow input comes directly from the heat control subsystem of the
main control.
4.4.6 District Heating Load
Overview
A brief introduction to district heating systems was already given in Chapter 1. The district
heating system considered in this work consists of two types of loads: the service water
heat load and the space heating heat load. Each load is supplied with a separate heat exchanger. The district heating considered contains a return pipeline. The temperature is
kept at 100◦C at the hot water pipeline by the control. A brief description of the model is
provided below.
Model
In the model considered, pressure differences are neglected. The equations presented in
Section 4.1 are used to represent the different components.
Service Water Heating Load The relation between the service water heat demand Φq,swh
and the mass flow φm,wat of the water flowing through the district heating pipelines
is obtained by using (4.5). In this case the subscript ‘swh’ is used to represent the
service water heating and ‘wat’ is used to represent the flowing water.
Φq,swh = φm,wat cwat T in,swh − T out,swh
(4.33)
where T in,swh , given in Kelvin, and is equivalent to 15 ◦ C and T out,swh is equivalent to
80◦ C.
4.4 Dynamic Component Models and Controls
89
Space Heating Load The space heating demand represents the heat load required to keep
the indoor temperature at 20◦ C. The heat Φq,rad is transferred to the rooms by means
of radiators. Heat leaves the rooms through walls Φq,wal , windows Φq,wdw and the roof
Φq,rof . The resulting energy balance equation is:
mair cair
where
dT in,hhd
dt
= Φq,rad − Φq,wal − Φq,wdw − Φq,rof
Φq,wal = Uwal Awal T in,hhd − T out,hhd
Φq,wdw = Uwdw Awdw T in,hhd − T out,hhd
Φq,rof = Urof Arof T in,hhd − T out,hhd
(4.34)
(4.35)
(4.36)
(4.37)
where Φq is theheat transferred
per unit of time, U is the heat transfer coefficient, A
is the area and T in − T out is the difference that drives the heat transfer phenomena.
Subscript ‘hhd’ means household.
Radiators The radiator is represented by the following equation:
mwat cwat
dT out,rad
dt
where:
= φm,wat cwat T in,rad − T out,rad − Φq,rad
Φq,rad = Urad Arad ∆T LMTD,rad
(4.38)
(4.39)
where subscript ‘rad’ indicates radiator and subscript ‘wat’ represents the water flowing through the radiator. The equations are based on (4.4) and (4.7). The logarithmic
mean temperature difference ∆T LMTD,rad is given by:
T in,rad − T out,rad
∆T LMTD,rad =
(4.40)
T
−T
ln T in,rad −Tin,hhd
out,rad
in,hhd
where T in,rad is the temperature at the entrance of the radiator, T out,rad is the temperature at the exit of the radiator and T hhd is the household’s indoor temperature.
Heat Exchangers The heat exchangers are modeled in a similar way to those of the CHP
units and the SOFC.
Heat Losses The heat losses in the pipes Φq,sur are modeled as heat conduction to the surrounding soil from the pipe insulator. The equation that represents the transfer is:
mwat cwat
where:
dT pip
dt
= φm,wat cwat T sur − T pip − Φq,sur
Φq,sur = Upip Apip ∆T LMTD
(4.41)
(4.42)
where subscript ‘pip’ indicates return pipeline, subscript ‘sur’ represents the surroundings and subscript ‘wat’ represents the water flowing through the pipeline. The
equations are based on (4.4) and (4.7).
90
4 The Control: Multi-Carrier Hierarchical Control Architecture
Unit Control
Radiator Control This control keeps the desired indoor temperature by controlling the amount
of water mass that flows into the radiator. It consists of a proportional control.
4.4.7 Heat Storage Tank
Overview
The heat storage considered in this work consists of a water tank with a maximum designed
capacity of 190 kWh. The tank has a cylindrical shape and is buried in the soil. A heat
exchanger is connected between the tank and the hot pipeline of the district heating infrastructure.
Model
In this model the temperature of the storage medium is considered homogeneous.
Tank The storage tank is represented by the following equation:
mwat cwat
dT in,tnk
dt
= Φin − Φout
(4.43)
where Φout is equal to the losses to the surroundings:
Φout = Φq,los = Utnk Atnk (T in,tnk − T out,tnk )
(4.44)
and Φin is equal to the heat coming from the system Φq,sys times the storage conversion efficiency ηtnk [-]:
Φin = Φq,tnk = Φq,sys ηtnk
(4.45)
where the storage efficiency ηtnk is given by:



ηtnk , charging
ηtnk = 

 1 , discharging.
ηtnk
Subscript ‘tnk’ indicates storage tank and subscript ‘wat’ represents the water contained in the tank. The heat transfer coefficient Utnk depends on heat transfer parameters of the insulator and the soil.
Unit Control
No unit control is applied.
Main Control Link
The main control sends a control signal that indicates if the heat storage must charge or
discharge and how much according to the storage considerations.
4.4 Dynamic Component Models and Controls
91
4.4.8 Summary
In Figure 4.9 the energy hub for the example is presented and Table 4.2 shows the hub
elements and their relation to the multi-carrier hierarchical control architecture.
Figure 4.9: Energy hub used in the example
Table 4.2: Components and control subsystems
Hub Element
Hub’s
Converter
Control
Subsystem
Set-point from
Main Control
Unit Control
CHP (gas turbine)
A
electricity
gas flow
CHP (SOFC)
B
electricity
current
PID control
excitation control
delay transfer function
CHP (gas turbine)
C
electricity
gas flow
PID control
excitation control
Boiler
D
heat
gas flow
none
Wind turbine
E
none
none
Battery bank
F
electricity
current
rotor speed control
pitch angle control
none
Heat tank
G
heat
gas flow
none
92
4 The Control: Multi-Carrier Hierarchical Control Architecture
4.5 Simulation Results
The optimal scheduling and power dispatch are obtained from the multi-carrier unit commitment program, however, in the first two simulations the purpose is to show the operation
of the hierarchical control architecture, thus the selected dispatch does not correspond to the
optimal one. Simulation 1 shows the results of the electricity control subsystem, Simulation
2 shows the results of the heat control subsystem and Simulation 3 presents the results of
both subsystems for a simulation period of 24 hours in which the optimal scheduling and
optimal power dispatch are determined by the multi-carrier unit commitment program.
4.5.1 Simulation 1: Electricity Control Subsystem Demonstration
This section shows the operation of the electricity control subsystem. The units are working
at no-load and the assigned dispatch is set to zero for all units (this is equivalent to assuming
that the forecasted electric load is equal to zero). This is done in order to show how the
two-level hierarchical control accounts for unpredicted load variations in the system. The
electric load pattern and the wind power supply were arbitrarily chosen.
Participation Factor − CHP A
Unit
0.5
0
0
1
2
3
4
5
6
7
8
9
10
7
8
9
10
7
8
9
10
8
9
10
Participation Factor − CHP B
Unit
0.5
0
0
1
2
3
4
5
6
Participation Factor − CHP C
Unit
0.5
0
0
1
2
3
4
5
6
Participation Factor − Electricity Storage
Unit
0.5
0
0
1
2
3
4
5
Time [min]
6
7
Figure 4.10: Simulation 1: Participation factors of controllable units
As a consequence of having no initial assigned dispatch, the participation factor of each
unit is defined according to the unit’s rated capacity. At the beginning of the simulation all
4.5 Simulation Results
93
units are available. At the 5th minute of the simulation CHP A is forced to stop delivering
power, thus the main control reassigns the required power demand to the remaining available
units. The change in the participation factor of each controllable unit can be observed in
Figure 4.10. The rated capacity of CHP A, CHP B, CHP C and the storage are 100 kW,
75 kW, 100 kW and 100 kW respectively. Thus the initial participation factors are 0,26 for
CHP A, CHP C and the storage and 0,21 for CHP B. After 5 minutes, the new participation
factors are 0,36 for CHP C and the storage and 0,29 for CHP B (the SOFC).
System Frequency
e
W [p.u]
1.02
1.01
1
0.99
0.98
0
1
2
3
4
5
6
7
8
9
10
7
8
9
10
7
8
9
10
Rotor Speed CHP A
m
W [p.u.]
1.05
1
0.95
0
1
2
3
4
5
6
Rotor Speed CHP C
m
W [p.u.]
1.05
1
0.95
0
1
2
3
4
5
Time [min]
6
Figure 4.11: Simulation 1: Frequency of the system and rotor speeds
Figure 4.11 shows the control parameter of the electricity control subsystem: the system’s frequency. Additionally, the rotor speeds of the gas turbines are depicted (CHP A and
CHP C). It can be observed that thanks to the control strategy, the frequency was kept stable.
In Figure 4.12 three graphs are presented: the total electricity demand, the wind turbine’s
supply and the power supply delivered by the gas turbines (CHP A and CHP C) and the fuel
cell (CHP B). It can be observed that when the output of the wind turbine increases, the
power supplied by the CHPs is reduced accordingly in relation to the participation factors.
Figure 4.13 shows the state of charge of the electricity storage. The total electricity
demand and storage supply are also depicted in the upper graph in order to show the correspondence between the state of charge and the power requirements. As it was described in
Section 4.3, in order to charge or discharge the storage units, the control system first checks
its availability. In the simulation the state of charge remained close to 80%.
94
4 The Control: Multi-Carrier Hierarchical Control Architecture
5
Power [W]
Electricity Demand
x 10
4
2
0
0
1
2
3
4
5
6
Wind Turbine Supply
7
8
9
10
1
2
3
4
5
6
Electricity Total Supply
7
8
9
10
7
8
9
10
5
x 10
Power [W]
2
1
0
1.5
0
5
x 10
Power [W]
1
0.5
0
Electrical Power CHP A
Electrical Power CHP B
Electrical Power CHP C
Electrical Power Storage
−0.5
−1
0
1
2
3
4
5
Time [min]
6
Figure 4.12: Simulation 1: Electricity demand and electricity supply
5
x 10
5
Total Electricity Demand
Electricity Supply E.Storage
4
3
60
2
40
1
20
0
0
0
1
2
3
4
5
Time[min]
6
Storage Charging Commands
7
8
−1
10
9
Storage Discharging Commands
E. St. Ch. Available
E. St. Ch. Requested
E. St. Disch. Available
E. St. Disch. Requested
1
Unit
1
Unit
State of Charge [%]
State of Charge
80
0
0
1
2
3
4 5 6
Time [min]
7
8
9
10
0
0
1
2
3
4 5 6
Time [min]
7
Figure 4.13: Simulation 1: Electricity storage behavior
8
9
10
Power [W]
100
4.5 Simulation Results
95
4.5.2 Simulation 2: Heat Control Subsystem Demonstration
This section shows the heat control subsystem under operation. The space heating and
service water loads were arbitrarily chosen and the output of the CHPs is given as an input.
Figure 4.14 contains the results of the control parameters: the hot line temperature of the
district heating system and the indoor temperature. The return line temperature and the
outdoor temperature are also included. It can be observed that the room temperature was
kept at 20◦ C during the whole simulation period and the hot line close to 100◦C.
Indoor and Outdoor Temperatures
40
Indoor temperature
Outdoor temperature
Temperature [°C]
30
20
10
0
−10
0
15
30
45
60
75
90
105
120
District Heating Line Temperatures
120
Hot line temperature
Return line temperature
Temperature [°C]
110
100
90
80
70
0
15
30
45
60
Time[min]
75
90
105
120
Figure 4.14: Simulation 2: Indoor temperature and DHS line temperatures
Figure 4.15 contains 2 different graphs. The first graph shows the total space heating
demand (in correspondence with the outside temperature variation) as well as the space
heating demand and the service water demand. In the second graph, the power supplied by
the furnace, the CHPs and the heat storage is depicted. It can be observed that the furnace
and the storage tank supply heat so that the power balance was accomplished during the
whole simulation.
Figure 4.16 shows the total heat demand and storage heat. The temperature of the storage tank, the storage change and discharge commands and the storage availability are also
presented to show the correspondence with the heat supplied by the tank.
96
4 The Control: Multi-Carrier Hierarchical Control Architecture
Heat Demand
2000
Total heat demand
Heat demand: space heating
Heat demand: service water
Power [kW]
1500
1000
500
0
0
15
30
45
60
75
90
105
120
Heat Supply
2000
Total heat supply
Heat supply: furnace
Heat supply: CHPs
Heat supply: heat storage
Power [kW]
1500
1000
500
0
−500
0
15
30
45
60
Time [min]
75
90
105
120
Figure 4.15: Simulation 2: Heat demand and heat supply
Storage Temperature and Storage Supply
150
1500
1000
50
500
0
0
0
15
30
45
60
Time [min]
75
Storage Charging Commands
90
−500
120
105
Storage Discharging Commands
Available: charge
Charge requested
Available: discharge
Discharge requested
1
[−]
1
[−]
Temperature [°C]
100
2000
0
0
15
30
45 60 75
Time [min]
90
105 120
0
0
15
30
45 60 75
Time [min]
Figure 4.16: Simulation 2: Heat storage behavior
90
105 120
Power [kW]
Total heat supply
Heat supply: storage
Temperature heat storage
4.5 Simulation Results
97
4.5.3 Simulation 3: Complete System Demonstration
In this section a demonstration of the complete system’s operation is presented. The simulation was done for 24 hours. The wind speed forecast was obtained from a model that
was developed for this project. This forecast model will be described in more detail in
Section 6.2.1. Figure 4.17 shows both the forecasted and the actual electric output of the
wind turbine and Figure 4.18 shows the forecasted and actual electric load values used in
the simulation. It is assumed that the actual heat load values are equal to the predicted ones.
Figure 4.19 shows the total heat load, the space heating load and the service water load. As
it can be noted in Figure 4.17 and Figure 4.18, the actual values of the wind turbine’s power
output and electric load differ from the forecasted ones, this gives room for the control
system to operate and keep the control parameters within a suitable range of operation.
The forecasted values discussed above are the input to the multi-carrier unit commitment
program. Previous to the multi-carrier hierarchical control simulation, the multi-carrier unit
commitment program is used to find the optimal solution. In this way the scheduled values
can be obtained. In Figure 4.1, the set-points coming from the optimization shown at the top
of the two-level hierarchical control scheme are equal to the results of the multi-carrier unit
commitment program.
Electric Output of the Wind Turbine
200
Power [kW]
150
100
50
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Predicted Electric Output of the Wind Turbine
200
Power [kW]
150
100
50
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 4.17: Simulation 3: Forecasted and actual output of the wind turbine
98
4 The Control: Multi-Carrier Hierarchical Control Architecture
Electricity Demand
240
Power [kW]
180
120
60
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Predicted Electricity Demand
240
Power [kW]
180
120
60
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 4.18: Simulation 3: Forecasted and actual electricity load
Heat Demand
1000
Total heat demand
Heat demand: space heating
Power [kW]
750
500
250
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Heat Demand
200
Heat demand: service water
Power [kW]
150
100
50
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 4.19: Simulation 3: Heat load
4.5 Simulation Results
99
Combined Control Error of the System
20
[−]
10
0
−10
−20
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
System Frequency
Frequency [p.u]
1.01
1.005
1
0.995
0.99
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 4.20: Simulation 3: Combined control error and frequency of the system
District Heating Hot Line Temperature
Temperature [°C]
102
101
100
99
98
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Space Heating Temperatures
25
Temperature [°C]
20
15
10
5
0
Indoor temperature
Outdoor temperature
−5
−10
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 4.21: Simulation 3: Temperature of the district heating hot line and the space heating
100
4 The Control: Multi-Carrier Hierarchical Control Architecture
Power [kW]
Electricity Demand
240
Electric load
120
Power [kW]
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Electricity Supply
180
Electric output: wind turbine
90
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
180
Electric output: CHP A
Electric output: CHP B
Electric output: CHP C
Electric output: battery bank
Power [kW]
90
0
−90
−180
0
1
2
3
4
5
6
7
8
Figure 4.22: Simulation 3: Total electricity demand and electricity supply
In Figure 4.20 and in Figure 4.21 it can be observed that even though there were differences with respect to the forecasted values, the system’s frequency, the hot water pipeline
temperature and the room temperature were kept within the acceptable boundaries.
Figure 4.22 shows three graphs, the first one shows a curve with the electricity demand,
the second one shows the wind turbine’s power output and the third one present the power
supplied by the CHP units and the storage elements during the 24-hour simulation. CHP B
has a minimum operation setting of 25% of its maximum capacity, for this reason it operates
during the whole simulation. It can be observed that the storage unit plays an important role
in keeping the power balance of the system.
The reader can compare the scheduled values and the actual values in the figures shown
in Appendix B. They show the response of the individual units. It can be noted that the
overall shape of the actual dispatch is similar to the scheduled one but that the instantaneous
values differ from each other due to the influence of the performed control actions. This
response coincides with the expectations.
4.6 Conclusions
101
4.6 Conclusions
This chapter presented a two-level hierarchical integrated control architecture for systems
that contain multiple energy carriers. In the example electricity and gas were considered
as input and electricity and heat as the output. The general architecture was adapted to fit
the selected example. In this way the applicability of the control scheme was tested. It
showed to be robust and stable for normal operating conditions. Other control actions like
load shedding can be applied using the same architecture.
By using an integrated control strategy, the flexibility of having more energy carriers
was explored. In future work it would be interesting to include more energy carriers and
test the applicability of the control scheme that was designed. Furthermore, the interaction
with smart protection algorithms should also be studied.
Chapter 5
The Scenarios
Multi-Carrier Power Applications
This chapter answers Research Question 5 of this dissertation. Section 5.1 presents theoretical concepts that have not been introduced in previous chapters. Section 5.2 presents a brief
literature review on works where the concept of aggregating household generation units
has been applied. In Section 5.3 the different scenarios to be simulated are described. Section 5.4 includes the input data that are used in the simulations: the prices, the parameters of
the storage devices and the parameters of the generation components. Section 5.5 presents
the results of the different simulations and includes a discussion based on the comparison
of scenarios. Finally in Section 5.6 the conclusions of this chapter are drawn.
5.1 Basic Concepts and Definitions
5.1.1 Virtual Power Plant
In the literature, different definitions of a virtual power plant can be found. A definition that
fits the scope of this dissertation is the following [102]:
A VPP combines different types of renewable and non-renewable generators
and storage devices to be able to pose on the electricity market as a single
power plant with defined hourly (15 min) output.
The control of a virtual power plant can be done directly or indirectly. When done directly, the aggregator has control on the signals that are sent to the generation units, whereas
when it is done indirectly the aggregator only sends price signals and the household owing
the generation unit reacts on them [103]. From this classification it can be inferred that
direct control gives the best results, since the aggregator not only sends the prices for the
households to react according to their will, but sends the set-points to the micro-generation
units itself. Therefore, the aggregator sends and receives information from the users.
103
104
5 The Scenarios: Multi-Carrier Power Applications
5.1.2 Aggregator
The term aggregator can be used to refer to different entities, for example retail companies,
distribution system operators or integrated utilities, among others. In this dissertation the
aggregator will represent an integrated power management node having ICT communication
with all houses that have agreed to participate in the group optimization. Therefore, the
definition of aggregator used in this dissertation is the following [103]:
An aggregator is an actor that trades with aggregate power flows to/from households and/or operates a virtual power plant by controlling micro-CHPs and
other controllable units available.
Several benefits of using an aggregator were found in the literature. Some of the most
important ones are the following:
• The controllers at the household level remain simple (only sensing and actuation devices are required) since the optimization control is lifted from the household level to
the aggregator level.
• Investments in communication and control equipment become lower when it is done
for clusters of households.
• There is higher predictability of aggregated energy demand than that of individual
households.
5.2 Literature Review
This literature review comprises articles in which the virtual power plant concept is studied.
The articles dealing with the virtual power plant concept can be divided into two categories,
those that deal with medium-sized generation plants that are clustered to virtually function as a single power plant and those papers that deal with the aggregation of small-sized
generation units placed in households and building infrastructures. Since this chapter pays
attention to the latter, papers focusing on micro-generation units of a few kWs are considered in this review.
Several papers focus on the development of virtual power plant optimization frameworks. Some of the methods proposed include service oriented architecture [104] and
multi-agent-based control. For example, in [105] a multi-agent-based control was applied to
a test facility consisting of renewable sources and controllable units. The work focused on
the implementation of such multi-agent-based control in a real environment. Furthermore,
a large percentage of the available papers focus on market-based virtual power plants. In
this context, price-signal based control and bidding scenarios are studied [106–109].
The following paragraphs refer to papers that deal with the control of micro-CHPs. They
were selected for this review due to their relevance to this research. In [110] a simulation
tool based on the software packages Microsoft Excel and Visual Basic is presented. The
simulation tool can simulate one day or one year. The input of the tool are electrical and
thermal consumption data, given in intervals of 15 minutes. The output data are the thermal
and electrical power generated, the content of the heat accumulator, the number of start-ups
5.3 Optimization Scenarios
105
of the micro-CHP and the electricity supplied from the grid [110]. Even though the paper
states the potential benefits of using the VPP concept with several houses, the results shown
refer to the output of a single household, thus the interactions between the cluster of houses
is not presented. One of the objectives of this chapter is to show the influence that the
aggregator has on the interaction among different households.
In [111] each device is represented by an agent that seeks for an economical optimum. In
this case no central optimization entity is required. The communication with the auctioneer
is very limited since it only exchanges bids between the agents and the agent platform. The
agent decides when to start producing according to the bids. The network structure used is
the PowerMatcher. The PowerMatcher performs the price-forming process; it coordinates
demand and supply of a cluster of devices located directly below it [111]. In this chapter
the optimization is shifted to the aggregator due to the reasons mentioned in Section 5.1. In
[111] no storage, renewable sources or heat flows were included.
In [112] a methodology of a traceable and modular VPP is presented. The system allows
energy trade, network services and balancing. In the paper, some opportunities of using a
VPP are listed, for example, to facilitate decentralized energy owners to trade their production, to provide VPP stakeholders with the opportunity of getting benefits from participating
in the primary, secondary and tertiary control and to support network operators in congestion and load management. The paper states that further tests are required to execute and
evaluate the added value of implementing such system. In this chapter several comparisons
are presented in which the added value of having a VPP is shown.
Lastly, [103] presents a VPP control strategy to resolve wind imbalance. Results are
shown for a winter week in January. The author focuses of the economic incentives in VPP
to invest in micro-CHP units. The simulations were performed using stirling engines placed
inside the households. The author concludes that the imbalance volume due to wind and
the associated costs can be reduced to 33% and 20% respectively. The research gives good
insights about the potential benefits of applying the VPP concept. The author states that
further comparisons between other micro-CHP technologies are important. As a response
to this argument, this chapter presents a comparison among three different household-level
micro-CHP technologies.
5.3 Optimization Scenarios
This section introduces the optimization scenarios that were considered:
Scenario 1 Base Case
Scenario 2 Individual Optimization
Scenario 3 Individual Optimization with Storage
Scenario 4 Collaborative Optimization (Aggregator)
Scenario 5 Collaborative Optimization with Storage (Aggregator)
The inclusion of renewables and electric vehicles is evaluated for different of the aforementioned scenarios in separate simulations. Each scenario is described in the following
subsections.
106
5 The Scenarios: Multi-Carrier Power Applications
5.3.1 Scenario 1: Base Case
Distribution
transformer
heat
heat
electricity
gas
Figure 5.1: Scheme for Scenario 1 - Base case
This scenario is considered the base scenario for comparison. The following assumptions
are considered:
• Each household has a furnace to supply its heat load.
• Each household is connected to the electrical grid, from which its electric load is
supplied.
• There are no renewable energy technologies or storage elements installed.
5.3.2 Scenario 2: Individual Optimization
Distribution
transformer
heat
heat
H. Opt
electricity
gas
Figure 5.2: Scheme for Scenario 2 - Individual optimization
H. Opt
5.3 Optimization Scenarios
107
In this optimization micro-CHP units are placed at each household.
• Each household has a micro-CHP unit to supply (part of) its heat and electricity.
• Each household has a furnace as auxiliary device to supply (part of) its heat load.
• Each household is connected to the electrical grid, from which (part of) its electric
load is supplied. Excess electricity produced by the micro-CHP units can be fed back
to the grid.
• There are no renewable energy technologies or storage elements installed.
• Each household is optimized individually.
• Each household is represented by an individual energy hub.
The optimization tool developed using the modeling assumptions and algorithms presented in Section 2.3.1 is used to find the solution of this scenario.
5.3.3 Scenario 3: Individual Optimization with Storage
Distribution
transformer
heat
heat
STORAGE
H. Opt
STORAGE
H. Opt
electricity
gas
Figure 5.3: Scheme for Scenario 3 - Individual optimization with storage
The only difference with respect to the previous scenario is the addition of storage devices.
Each household is provided with a heat storage water tank and/or a battery. In this case
the four-step technique introduced in Section 2.3.2 is used for the optimization of each
individual household.
108
5 The Scenarios: Multi-Carrier Power Applications
5.3.4 Scenario 4: Collaborative Optimization
Aggregator
Distribution
transformer
heat
heat
electricity
gas
Figure 5.4: Scheme for Scenario 4 - Collaborative optimization
In this optimization, micro-CHP units are placed at each household and an external neighborhood aggregator is in charge of the optimization of the micro-CHP units.
• Each household has a micro-CHP unit to supply (part of) its heat and electricity.
• Each household has a furnace as auxiliary device to supply (part of) its heat load.
• Each household is connected to the electrical grid, from which (part of) its electric
load is supplied. Excess electricity produced by the micro-CHP units can be fed back
to the grid.
• There are no renewable energy technologies or storage elements installed.
• Each generation unit is operated according to the optimization performed by the aggregator. The aggregator induces exchange among the households; this minimizes the
power interaction with the electrical grid and enhances the efficient power utilization
within the neighborhood.
• The aggregator determines how much of the power is imported or exported from other
households. The respective household will pay or receive money for it according to
the price defined for exchanged electricity within the neighborhood.
• Each household is represented by an individual energy hub.
5.3 Optimization Scenarios
109
Model Extension
In order to show the influence of an aggregator, additional constraints were implemented in
the optimization model that was described in Chapter 2. For each household (represented by
an energy hub), the electricity generated by the generation units installed at the household
Lhe,d hub is given by the multiplication of the electricity coupling vector and the vector of
input carriers delivered to the hub. When several households hd are considered, where
d = 1, 2, . . . , nh and nh ∈ N, the following matrix form is used. Note that in this case
the result does not take the contribution of the electricity coming/going to the grid or the
electricity exchanged with the neighbors into account (all power values are in kW):
 hd 
 Pαi 
i
 . 
hd
. . . cαnin e  .. 
 h 
Pαdnin hub.
h
Lhe,d hub = chαdi e
(5.1)
The household power unbalance Lhe,d unb is defined as the electricity generated by the
generation units of the household Lhe,d hub minus the electricity demand of the household
itself Lhe d . At the same time, the power unbalance Lhe,d unb is equal to the power exchanged
with other households Phe,d exc plus the power coming or going to the grid Phe,d grd (in case that
no further exchange is allowed). The constraints associated with this condition are shown
below:
Lhe,d unb = Lhe,d hub − Lhe d .
(5.2)
Lhe,d unb
(5.3)
=
Phe,d exc
+
Phe,d grd .
The total electricity exchange Phe,d exc per household is equal to adding the power exported
to other households Phe,d exp minus the power imported from other households Phe,d imp :
Phe,d exc = Phe,d exp − Phe,d imp .
(5.4)
Since it is not expected that a household will import and export electricity at the same
time, the following constraint is defined:
Phe,d exp Phe,d imp = 0.
(5.5)
The power considered as exchanged power only refers to the power exchanged among
the households in the neighborhood, thus the sum of the power exchanged by all households
in the neighborhood is equal to zero:
X
Phe,d exc = 0.
(5.6)
hd ∈Hhd
The power that can be exported to other households is equal or lower than a positive
power unbalance and the power that can be imported from other households is equal or
lower than the magnitude of a negative power unbalance. This can be accomplished in the
optimization program by means of the following constraints:
Phe,d exc Phe,d unb >= 0
Phe,d exc Phe,d exc
<=
Phe,d unb Phe,d unb .
(5.7)
(5.8)
110
5 The Scenarios: Multi-Carrier Power Applications
5.3.5 Scenario 5: Collaborative Optimization with Storage
Aggregator
Distribution
transformer
heat
heat
STORAGE
STORAGE
electricity
gas
Figure 5.5: Scheme for Scenario 5 - Collaborative optimization with storage
The only difference with respect to the previous scenario is the addition of storage devices.
Each household is provided with a heat storage water tank and/or a battery. The four-step
technique introduced in Section 2.3.2 is used. The new constraints that were introduced
in Scenario 4 are also included in Step 1, Step 2, Step 3 and Step 4 of the optimization
technique.
5.3.6 Inclusion of Renewables
Aggregator
Distribution
transformer
heat
heat
STORAGE
electricity
gas
Figure 5.6: Inclusion of renewables in Scenario 5
STORAGE
5.3 Optimization Scenarios
111
For modeling purposes, the power coming from uncontrollable renewables (Phe,d ren ) is
considered as an extra input at the energy hubs representing each household. In Figure 5.6,
the representation of Scenario 5 with renewables is depicted. The equation for the power
unbalance can be written as:
Lhe,d unb = Phe,d exc + Phe,d grd + Phe,d ren .
(5.9)
5.3.7 Inclusion of Electric Vehicles
Aggregator
Distribution
transformer
heat
heat
STORAGE
STORAGE
electricity
gas
Figure 5.7: Inclusion of electric vehicles in Scenario 5
In Figure 5.7, the representation of Scenario 5 with electric vehicles and renewables is
d
depicted. For modeling purposes, the power demand from the electric vehicles (Lhe,veh
) is
added to the electricity demand of each household. The resulting equation for the power
unbalance is:
d
Lhe,d unb = Lhe,d hub − Lhe d + Lhe,veh
.
(5.10)
112
5 The Scenarios: Multi-Carrier Power Applications
5.4 Input Data for the Simulations
5.4.1 Prices
Price policies for the electricity that is fed back to the grid are still under development, thus
four different price options are analyzed to determine their influence in the optimization,
see Table 5.1. In option 1 no reimbursement is regarded, in option 2 the price currently
paid in the Netherlands is considered [15]. In the third option, 55% of the price charged for
electricity from the grid is considered and in the fourth case the same price for electricity
from the grid is taken. To motivate the exchange with other neighbors, the price for the
electricity that is imported/exported is 77,5%/72,5% of the price of electricity from the grid.
Table 5.1: Gas and electricity prices
Energy Carrier
Consumption Costs
Gas
Electricity
Electricity back to grid
Electricity exchanged (imported)
Electricity exchanged (exported)
Source
Option 1
(e/kWh)
Option 2
(e/kWh)
Option 3
(e/kWh)
Option 4
(e/kWh)
0,0684
0,1897
0,0000
-
0,0684
0,1897
0,0500
0,1470
0,1375
0,0684
0,1897
0,1897
0,1470
0,1375
0,0684
0,1897
0,1043
0,1470
0,1375
[47]
[47]
[15]
-
5.4.2 Component Parameters
The parameters of the storage devices can be found in Table 5.2. They include the energy
content of the storage device, the maximum power that can be withdrawn or delivered to it,
the charging efficiency and the stand-by losses. The parameters of the generation devices
used in the simulation are shown in Table 5.3. They include the maximum and minimum
input carrier boundaries, the electrical and thermal efficiencies, the total energetic efficiency
and the source from which the data were taken.
Table 5.2: Parameters of storage devices
Parameter
Minimum energy content
Maximum energy content
Maximum power (charge)
Maximum power (discharge)
Charging efficiency eαi
Stand-by losses
Unit
kWh
kWh
kW
kW
%
kW
Storage
Source
Electricity
Heat
1,00
5,00
0,80
-0,50
90,00
4e-3
9,60
14,00
2,00
-2,00
95,00
1e-4
[113],[103]
[113],[103]
[113],assumption
[113],assumption
[113],[103]
[113],assumption
5.4 Input Data for the Simulations
Table 5.3: Parameters of generation devices
Parameter
Input carrier
Minimum input carrier
Maximum input carrier
Electrical energetic efficiency
Thermal energetic efficiency
Total energetic efficiency
Electrical to thermal efficiency ratio
Source
Symbol
αi
Pα i
Pα i
k
ηαpi e
kp
ηα i q
k
ηtotp
kp
k
ηαi e /ηαpi q
Unit
kW
kW
%
%
%
%
Component
Grid
CHP 1
Stirling Engine
CHP 2
Gas-Fired CHP
CHP 3
Solid Oxide FC
Furnace
Electricity
0,00 or -100,00
100,00
100,00
100,00
[103]
Gas
0,00
6,67
15,00
85,00
100,00
0,18
[103],[113]
Gas
0,00
10,00
30,00
70,00
100,00
0,43
[114]
Gas
0,00
3,33
60,00
25,00
85,00
2,40
[103]
Gas
0,00
25,00
90,00
90,00
[103]
113
114
5 The Scenarios: Multi-Carrier Power Applications
5.4.3 Load Patterns
The data of the heat and electric load patterns were obtained from data sets generated in
[103]. Individual data sets for 200 houses were provided. Figure 5.8 and Figure 5.11 show
average load patterns for one week in winter and one week in summer.
6
Heat
Electricity
Average Power Consumption [kW]
5
4
3
2
1
0
0
12
24
36
48
60
72
84
96
Time [hours]
108
120
132
144
156
168
Figure 5.8: Average electricity and heat demand patterns for a week in winter
0.8
Heat
Electricity
Average Power Consumption [kW]
0.6
0.4
0.2
0
0
12
24
36
48
60
72
84
96
Time [hours]
108
120
132
144
156
168
Figure 5.9: Average electricity and heat demand patterns for a week in summer
5.4 Input Data for the Simulations
115
For illustration purposes, Figure 5.10 and Figure 5.9 show the load patterns of one randomly selected household and the respective average load patterns of four consecutive days
in winter and four consecutive days in summer.
8
Heat single household
Heat average
Electricity single household
Electricity average
Average Power Consumption [kW]
6
4
2
0
0
12
24
36
48
Time [hours]
60
72
84
96
Figure 5.10: Single and average electricity and heat demand patterns for a week in winter
2
Heat single household
Heat average
Electricity single household
Electricity average
Average Power Consumption [kW]
1.5
1
0.5
0
0
12
24
36
48
Time [hours]
60
72
84
96
Figure 5.11: Single and average electricity and heat demand patterns for a week in summer
116
5 The Scenarios: Multi-Carrier Power Applications
5.5 Simulation Results
In this chapter five simulations are shown. The micro-CHP technologies considered are: a
stirling engine, a gas-fired micro-CHP and a solid oxide fuel cell, with an electricity-to-heat
efficiency ratio of 0,18, 0,43 and 2,40 respectively; they are referred to as CHP 1, CHP 2,
and CHP 3 in the discussions and in the figures. The simulations are described below:
Simulation 1: Comparison of Micro-CHP Technologies and Prices In this simulation,
three micro-CHP technologies with different electricity-to-heat efficiency ratios are
compared. In order to do so, the electricity and heat demand curves of a single household were selected. Four different price options are considered. The optimization is
performed according to Scenario 2 (individual optimization), consequently there is
no aggregator involved. The results of the Scenario 1 (base case) are also presented
as reference for comparison.
Simulation 2: Introducing an Aggregator The results obtained when an aggregator is introduced are evaluated using the load data of five households. Three micro-CHP technologies with different electricity-to-heat efficiency ratios are considered for comparison. The optimizations are performed according to Scenario 1 (base case), Scenario
2 (individual optimization) and Scenario 4 (collaborative optimization).
Simulation 3: Introducing Storage In this simulation, the influence of introducing storage in a cluster of three households is evaluated. The optimizations are performed
according to Scenario 1 (base case), Scenario 2 (individual optimization), Scenario 4
(collaborative optimization) and Scenario 5 (collaborative optimization with storage).
Simulation 4: Introducing Renewables In this simulation the influence of introducing solar panels in a cluster of five households is evaluated. The results of Scenario 5 with
several storage combinations was evaluated.
Simulation 5: Introducing Electric Vehicles By means of this simulation it is possible to
evaluate the influence of introducing electric vehicles to a district. Two different cases
are considered. In the first one, electric vehicles are charged at will, in the second one
incentives are given so that the electric vehicles are charged at non-peak periods. In
the optimization, a cluster of 200 households is considered. The results correspond
to Scenario 2 (individual optimization) with electric vehicles and Scenario 4 with
electric vehicles (collaborative optimization).
5.5 Simulation Results
117
5.5.1 Simulation 1: Comparison of Micro-CHP Technologies and Prices
Four price options were considered for comparison in this simulation; they are shown in
Table 5.1. This is done in order to show how different price policies affect the optimization
results. Three micro-CHP technologies are compared: a stirling engine, a gas-fired microCHP and a solid oxide fuel cell, with an electricity-to-heat efficiency ratio of 0,18, 0,43 and
2,40 respectively. The parameters can be found in Table 5.3. A single household with average electricity and heat patterns is used for the analysis. The primary energy consumption
is calculated in the following way:
Psource =
X
Pe
ηgen,e
+ Pg
!
(5.11)
where Pe is the electricity power consumed from the electrical grid and Pg is the gas power
input. The primary energy from the electricity that comes from the grid is calculated by
dividing the electricity power with the energetic efficiency of a conventional large steam
generation unit, which is considered to be 35% [81].
Table 5.4 shows the price policies considered for the simulation and the colors that are
used to differentiate the cases in Figure 5.12, Figure 5.13, Figure 5.14 and Figure 5.15.
Table 5.4: Gas and electricity prices
Energy Carrier
Gas
Electricity
Electricity back to grid
Consumption Costs
Base Case
(e/kWh)
Option 1
(e/kWh)
Option 2
(e/kWh)
Option 3
(e/kWh)
Option 4
(e/kWh)
0,0684
0,1897
-
0,0684
0,1897
0,0000
0,0684
0,1897
0,0500
0,0684
0,1897
0,1897
0,0684
0,1897
0,1043
The optimization in this simulation is based on Scenario 2, in which the household
under study is optimized individually. The results of Scenario 1, which corresponds to
the base case, are also shown as reference. The total accumulated costs and the primary
energy consumption required to supply the load profiles of a chosen week in winter and a
chosen week in summer are presented in Figure 5.12 and Figure 5.13 respectively. The gas
consumption, reverse energy and grid energy consumption can be observed in Figure 5.14
and Figure 5.15 for winter and summer respectively.
By introducing a micro-CHP & furnace configuration instead of the traditional grid
& furnace configuration (base case), the total operational costs are reduced in all cases
considered, regardless of the price option selected or the micro-CHP technology used, see
Figure 5.12 and Figure 5.13. It can also be observed that the lowest costs are obtained when
CHP 3 is used; this corresponds to the solid oxide fuel cell unit in which the electricityto-heat efficiency ratio is the highest. This is evident particularly in the summer, where the
heat demand is much lower than in winter.
118
5 The Scenarios: Multi-Carrier Power Applications
Total Cost
100
Cost [EUR]
80
60
40
20
0
CHP1
CHP2
CHP Technology
CHP3
Consumption of Primary Sources
1200
Energy [kWh]
960
720
480
240
0
CHP1
CHP2
CHP Technology
CHP3
Figure 5.12: Simulation 1: Total costs and primary energy consumption for a week in winter
Total Cost
20
Cost [EUR]
16
12
8
4
0
−4
CHP1
CHP2
CHP Technology
CHP3
Consumption of Primary Sources
300
Energy [kWh]
240
180
120
60
0
CHP1
CHP2
CHP Technology
CHP3
Figure 5.13: Simulation 1: Total costs and primary energy consumption for a week in summer
5.5 Simulation Results
119
Gas Consumption
Reverse Energy
960
240
Energy [kWh]
300
Energy [kWh]
1200
720
480
240
0
180
120
60
CHP1
0
CHP2
CHP3
CHP Technology
CHP1
CHP2
CHP3
CHP Technology
Grid Energy Consumption
80
Energy [kWh]
64
48
32
16
0
CHP1
CHP2
CHP Technology
CHP3
Figure 5.14: Simulation 1: Gas and grid energy consumption for a week in winter
Reverse Energy
100
240
80
Energy [kWh]
Energy [kWh]
Gas Consumption
300
180
120
60
0
60
40
20
CHP1
CHP2
CHP3
CHP Technology
0
CHP1
CHP2
CHP3
CHP Technology
Grid Energy Consumption
60
Energy [kWh]
48
36
24
12
0
CHP1
CHP2
CHP Technology
CHP3
Figure 5.15: Simulation 1: Gas and grid energy consumption for a week in summer
120
5 The Scenarios: Multi-Carrier Power Applications
Furthermore, it can be observed in Figure 5.14 and Figure 5.15 that when the reverse
power is not economically rewarded (price option 1) and when the currently available reverse power tariff of 0,05 e/kWh is considered (price option 2), there is no incentive for
the micro-CHPs to produce more electricity than the one required by the household owing
the micro-CHP. When the same price of electricity is paid back (price option 3) and when
the price for the reverse power is 0,1043 e/kWh (price option 4), the micro-CHP produces
the heat required to match the load without the support of the auxiliary unit and sells all
the extra electricity to the grid. The solid oxide fuel cell unit is the technology that takes
the most advantage of these two last price options. It can be noted that in the summer the
total costs are even negative in the case where the same price is paid back to the micro-CHP
owner (price option 3), see Figure 5.13.
With respect to the total primary energy consumed, it can be observed in Figure 5.12 and
Figure 5.13 that in the two first cases (price option 1 and price option 2) the total amount is
reduced up to a 35% in the case of the CHP 3 (solid oxide fuel cell), and between 10% and
20% in the case of the other two technologies. The reduction is achieved both in summer and
winter. This is particularly interesting from a sustainable point of view. The total primary
energy is however increased in the case of the solid oxide fuel cell for the other two cases
(price option 3 and price option 4). This occurs because the optimization is based on costs,
thus the solid oxide fuel cell produces as much electricity as possible to earn the most; this
produces the increase. Nevertheless, it is important to recall that the electricity is produced
with a higher electrical efficiency, which in the end is better from a sustainable perspective
when the complete district is considered as the boundary.
Conclusion
The gas energy consumed is increased in all cases considered, this is expected because the
three micro-CHP technologies are fed with gas, see Figure 5.14 and Figure 5.15. As it was
mentioned before, the reverse energy increased in the cases where the reverse power was
economically rewarded with 0,1897 and 0,1043 e/kWh (price option 3 and price option 4),
particularly in the case of the solid oxide fuel cell. Finally, the electricity consumption from
the grid is significantly reduced in all cases considered. The minimum exchange with the
grid is attained when using a solid oxide fuel cell unit because of its electricity-to-heat ratio.
5.5.2 Simulation 2: Introducing an Aggregator
The influence of introducing an aggregator can be observed in this simulation. By considering only five households it is easier to identify the impact of the control strategy used in each
optimization scenario. Five different sets of electric and heat load patterns were arbitrarily
chosen for the cluster of households for a period of one week, for both summer and winter.
It is assumed that all five households have the same micro-CHP technology installed. Three
different cases are compared; in each case a different micro-CHP technology is considered.
The micro-CHP technologies considered are: a stirling engine, a gas-fired micro-CHP and
a solid oxide fuel cell, with an electricity-to-heat efficiency ratio of 0,18, 0,43 and 2,40 respectively, they are denoted CHP 1, CHP 2, and CHP 3. Price option 2 was selected for the
simulation, this means that 0,5 e/kWh is paid for the reverse power.
5.5 Simulation Results
121
The optimization is based on Scenario 1, Scenario 2 and Scenario 4, which were described in Section 5.3. The legend associated with the graphs is depicted below.
Scenario 1 Base case - No aggregator
Scenario 2 Individual optimization - No aggregator
Scenario 4 Collaborative optimization - Aggregator
Total Cost − Summer
300
300
240
240
Cost [EUR]
Cost [EUR]
Total Cost − Winter
180
120
60
0
180
120
60
CHP 1
0
CHP 2
CHP 3
CHP Technology
3600
3600
2700
1800
900
CHP 1
CHP 2
CHP 3
CHP Technology
Consumption of Primary Sources − Summer
4500
Energy [kWh]
Energy [kWh]
Consumption of Primary Sources − Winter
4500
0
Figure 5.16: Legend
2700
1800
900
CHP 1
CHP 2
CHP 3
CHP Technology
0
CHP 1
CHP 2
CHP 3
CHP Technology
Figure 5.17: Simulation 2: Total costs and primary energy consumption
Figure 5.17 shows the total costs per household and the consumption of energy from
primary sources by each of them. It can be observed that for the case in which CHP 3
(solid oxide fuel cell) is installed, the benefits obtained from having an aggregator are the
highest, both in summer and winter. Figure 5.18 and Figure 5.19 show the gas consumed
by each household, the energy consumed from the grid by each household, the electrical
energy injected back to the grid by each household, and the energy exchanged (imported
or exported) by each household. It can be noted that the gas consumption is increased in
all cases. This is because the micro-CHPs are fed with gas, thus this result was expected.
The grid consumption is reduced in all the cases because all these micro-CHPs also produce
electricity. However it is important to notice that in the case of CHP 3 for Scenario 4 (after
including the aggregator) the electricity consumption from the grid for the week selected in
summer becomes almost one tenth of the power consumption of Scenario 1 (the base case)
and becomes almost zero for the week selected in winter.
122
5 The Scenarios: Multi-Carrier Power Applications
Grid Consumption − Winter
400
3200
320
Energy [kWh]
Energy [kWh]
Gas Consumption − Winter
4000
2400
1600
800
0
240
160
80
CHP 1
0
CHP 2
CHP 3
CHP Technology
200
240
160
180
120
60
0
CHP 2
CHP 3
CHP Technology
Energy exchanged (Positive) − Winter
300
Energy [kWh]
Energy [kWh]
Reverse Energy − Winter
CHP 1
120
80
40
CHP 1
0
CHP 2
CHP 3
CHP Technology
CHP 1
CHP 2
CHP 3
CHP Technology
Figure 5.18: Simulation 2: Gas and grid energy consumption for a week in winter
Grid Consumption − Summer
300
800
240
Energy [kWh]
Energy [kWh]
Gas Consumption − Summer
1000
600
400
200
0
180
120
60
CHP 1
0
CHP 2
CHP 3
CHP Technology
100
8
80
6
4
2
0
CHP 2
CHP 3
CHP Technology
Energy exchanged (Positive) − Summer
10
Energy [kWh]
Energy [kWh]
Reverse Energy − Summer
CHP 1
60
40
20
CHP 1
CHP 2
CHP 3
CHP Technology
0
CHP 1
CHP 2
CHP 3
CHP Technology
Figure 5.19: Simulation 2: Gas and grid energy consumption for a week in summer
5.5 Simulation Results
123
For all cases, i.e. all micro-CHP technologies, the total energy exchange is higher during
winter, where both the heat and electric loads are higher than during summer. However, it
should be noted that the effect of the aggregator is not always significant. For example, the
exchange that takes place in summer for CHP 1 is equal to zero, thus in this case there is
no influence of the aggregator in the results of the optimization. Nevertheless, there are
cases in which the effect of the aggregator is relevant, for example the highest exchange
is performed at the case in which CHP 3 is installed for the selected week in winter, see
Figure 5.18.
As mentioned before, the energy consumed from the grid is considerably reduced particularly in the cases of CHP 2 and CHP 3, both in summer and winter, especially when
the aggregator is in charge of the cluster optimization. Additionally, it is important to observe that there is no energy injected back to the grid when using CHP 3. These two results
show that there is a reduction in the overall energy exchange with the grid when using this
particular micro-CHP technology.
Conclusion
After comparing the results for the analyzed cases including different micro-CHP technologies, it can be inferred that for this load by installing a micro-CHP with a high electricityto-heat efficiency ratio and introducing an aggregator, the energy exchanged with the grid
can be reduced significantly for the electric and heat loads considered. This is a very important result that can be taken into account by distribution network operators in charge of the
planning of distribution infrastructures.
By doing similar analyses it would be possible to evaluate the response to load patterns
of households of different districts and locations. In this way, it would be possible to revise
the need for a modification in the distribution infrastructure. For example, after introducing
such a micro-CHP technology and an aggregator, it might not be necessary to expand the
distribution network anymore; the infrastructure might be kept the same for a longer period
of time, even when an increase in the electricity demand is expected. Such an analysis
can provide valuable information to be used to perform a more appropriate planning, to
establish new policies, to define where to direct investments and to define incentives for
consumers that can lead them to pick a micro-CHP technology with a suitable electricityto-heat efficiency ratio for their particular location.
An aggregator can be used to tackle problems related to reverse energy flows generated
from installing generation units at distribution level. By giving incentives to the customers
to install suitable micro-CHP generation units to be controlled by a local aggregator, it can
be possible to reduce investments intended to increase the capacity and/or to modify the
current distribution network infrastructures.
124
5 The Scenarios: Multi-Carrier Power Applications
5.5.3 Simulation 3: Introducing Storage
This section shows the influence of introducing storage in the optimization. A cluster of
three households is used for the simulation. The optimizations are performed according
to Scenario 1 (base case), Scenario 2 (individual optimization), Scenario 4 (collaborative
optimization) and Scenario 5 (collaborative optimization with storage) for a period of 24
hours. The micro-CHP technology used for the simulation is the stirling engine.
It is interesting to observe that in this example because of the use of storage, the furnace
is no longer used. This can be observed by comparing the multi-carrier unit commitment of
Scenario 4 and Scenario 5 in Table 5.5 and Table 5.6.
In Figure 5.20 it can be observed that total cost paid by each household for its electricity
and heat consumption in Scenario 2 is lower than in Scenario 1. Furthermore, a reduction is
evident in the total energy consumed from primary sources. A reduction of more than 25%
is achieved by introducing the micro-CHP units. This means that a significant reduction
in cost and energy consumption is obtained after performing an individual optimization
per household according to the framework proposed in Chapter 2. This had already been
discussed in Simulation 1.
In Scenario 4 an aggregator is introduced. The aggregator controls the output of the
connected units and obtains data from the electricity exchange measurements of all households. Due to the presence of an aggregator, the costs associated with each household and
consumption of primary energy sources decrease even further. This is because the aggregator takes advantage of forecasts in order to induce an optimal energy exchange among
the households. Further savings are achieved by introducing storage, since it allows more
flexibility for the optimization.
Figure 5.21 shows the exchanged energy, both imported and exported, for each household. Household 3 imports energy from household 1 and household 2. As a consequence,
the energy consumption from the electrical grid and the reverse energy to the grid are significantly reduced when compared to Scenario 2.
Conclusion
These results provide valuable information. Storage is advisable when a significant reduction in the costs and consumption of primary energy is attained. As it was discussed in
Chapter 2, the application of storage can affect the multi-carrier unit commitment of the
micro-generation units placed at the individual households. It is recommended to use this
kind of optimization tools during the planning phase of a district. In this way it can be evaluated if the benefits obtained from using storage overcome the investment and maintenance
costs.
Multi-Carrier Unit Commitment Solution of Scenario 4 with 3 households
Time Period
Component
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
2
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
0
1
0
1
0
1
0
1
0
1
0
3
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
0
1
0
1
0
1
0
1
1
1
0
1
0
1
0
1
1
1
0
1
0
1
0
1
0
1
0
1
0
5.5 Simulation Results
Table 5.5: Simulation 3: Multi-carrier unit commitment results for Scenario 4
Table 5.6: Simulation 3: Multi-carrier unit commitment results for Scenario 5
Multi-Carrier Unit Commitment Solution of Scenario 5 with 3 households
Time Period
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
2
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
3
micro-CHP
Furnace
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
125
Component
126
5 The Scenarios: Multi-Carrier Power Applications
Total Cost
50
Cost [EUR]
40
30
20
10
0
Hh 1
Hh 2
Hh 3
Household
Total
Consumption of Primary Sources
600
Scenario 1
Scenario 2
Scenario 4
Scenario 5
Energy [kWh]
480
360
240
120
0
Hh 1
Hh 2
Hh 3
Household
Total
Figure 5.20: Simulation 3: Total costs and primary energy consumption
Grid Energy Consumption
480
48
Energy [kWh]
60
360
240
120
0
36
24
12
Hh 1
Hh 2
Hh 3
Household
0
Total
Hh 1
Reverse Energy
2
6
1
4
2
0
Hh 2
Hh 3
Household
Total
Exchanged Energy
8
Energy [kWh]
Energy [kWh]
Energy [kWh]
Gas Consumption
600
0
−1
Hh 1
Hh 2
Hh 3
Household
Total
−2
Hh 1
Hh 2
Hh 3
Household
Figure 5.21: Simulation 3: Gas and grid energy consumption
Total
5.5 Simulation Results
127
5.5.4 Simulation 4: Introducing Renewables
In this simulation the following cases are evaluated:
• Scenario 4
• Scenario 4 with renewables
• Scenario 5 with renewables and heat and electricity storage
• Scenario 5 with renewables and only heat storage
• Scenario 5 with renewables and only electricity storage
A group of five households is used for a simulation period of 24 hours. It is assumed that
each household has 7 solar panels installed on its roof. The patterns for the power coming
from the solar panels were based on data available at DENlab. The real data of the power
output and solar radiation is logged at the laboratory. Figure 5.22 shows a solar pattern.
Example of a Solar Power Pattern
0.5
0.4
Power [kW]
0.3
0.2
0.1
0
0
4
8
12
Time [h]
16
20
24
Figure 5.22: Example of a solar power pattern
From the simulation results shown in Figure 5.24 it can be observed that the difference
in total costs and primary energy consumption is small for the cases analyzed. A remarkable
result is that the power exchange within the neighborhood is reduced when the heat storage
is available. The addition of a battery does not produce significant benefits in terms of power
exchange.
Conclusion
It can be concluded that due to the low installation capacity and the load profiles considered,
the results were not considerably changed by introducing renewables and storage. In such a
condition it would be advised to avoid installing batteries. Batteries are disposed after a few
years of utilization, which in the end does not represent a sustainable option. This result
shows that storage is not necessarily beneficial in a system with renewables.
128
5 The Scenarios: Multi-Carrier Power Applications
Grid Consumption
10
320
8
Energy [kWh]
Energy [kWh]
Gas Consumption
400
240
160
6
4
80
0
2
Winter
0
Summer
Season
Reverse Energy
10
Sce. 4
Sce. 4 with Ren.
Sce. 5 with Ren. & Stor.
Sce. 5 with Ren. & Heat Stor.
Sce. 5 with Ren. & Elec Stor.
120
8
Energy [kWh]
Energy [kWh]
180
60
0
Summer
Season
Energy exchanged (Positive)
300
240
Winter
6
4
2
Winter
0
Summer
Season
Winter
Summer
Season
Figure 5.23: Simulation 4: Gas and grid energy consumption
Total Cost
30
Cost [EUR]
24
18
12
6
0
Winter
Summer
Season
Consumption of Primary Sources
400
Energy [kWh]
320
240
160
80
0
Winter
Summer
Season
Figure 5.24: Simulation 4: Total costs and primary energy consumption
5.5 Simulation Results
129
5.5.5 Simulation 5: Introducing Electric Vehicles
In this section the results after applying two different price policies are compared: in the first
one it is assumed that the electric vehicles are charged at will, in the second differentiated
price tariffs are provided so that the electric vehicles are charged at non-peak periods. A
neighborhood of 200 households is considered for the simulation. Furthermore, an electric
vehicle penetration of 20% was considered. Two optimization scenarios were selected for
this simulation. In the first one, an individual optimization (Scenario 2) is performed and in
the second one, the results of the collaborative optimization (Scenario 4) are shown.
The dataset for the electric vehicle load was obtained from a model developed in the
IOP project “Role of Energy Storage in Future Power Systems”, where electric vehicle load
patterns are obtained based on a Monte Carlo simulation approach. The model variables
are characterized by a stochastic behaviour and are correlated; a multivariate distribution
function was built by means of copula function and the respective marginal empirical distributions [115].
The original dataset inserted into the model is statistical information obtained from the
transportation data of 2008, provided by the Dutch Ministry of Transportation. The dataset
includes information about commuting activities like time of departure, time of arrival, address of departure, address of arrival, transport means, trip distance, etc [115]. Only hometo-home trips occurring within one or two consecutive days were considered. The simulated
single and double home-to-home trips were combined with a typical electric vehicle charging profile; this allowed the computation of the load patterns. Different electric vehicle penetration levels were modeled. Moreover, by applying price incentives, load shifts towards
off-peak hours were also modeled. More information on the mathematical derivation of the
model can be found in [115]. Figure 5.25 shows the load curves of the electric vehicles for
a random day for illustrative purposes. Both price policies are depicted.
Example of Electric Vehicle Patterns
150
Uncontrolled charging
Differentiated price
120
Power [kW]
90
60
30
0
0
4
8
12
Time [Hours]
16
20
24
Figure 5.25: Electric vehicle aggregated load pattern for one day
130
5 The Scenarios: Multi-Carrier Power Applications
4
2
Gas Consumption
x 10
Grid Consumption
400
320
Energy [kWh]
Energy [kWh]
1.6
1.2
0.8
0.4
240
160
80
0
0
Uncontrolled Differentiated
Price Policy
Reverse Energy
Energy exchanged (Positive)
300
2000
Scenario 2
Scenario 4
1600
Energy [kWh]
240
Energy [kWh]
Uncontrolled Differentiated
Price Policy
180
120
60
1200
800
400
0
Uncontrolled Differentiated
Price Policy
0
Uncontrolled Differentiated
Price Policy
Figure 5.26: Simulation 5: Gas and grid energy consumption
Total Cost
1200
Cost [EUR]
960
720
480
240
0
Uncontrolled
Differentiated
Price Policy
4
2
x 10
Consumption of Primary Sources
Energy [kWh]
1.6
1.2
0.8
0.4
0
Uncontrolled
Differentiated
Price Policy
Figure 5.27: Simulation 5: Total costs and primary energy consumption
5.6 Conclusions
131
Individual electric vehicle patterns were added as loads to the individual energy hubs
representing the households in the optimization model. In Figure 5.27 it can be observed
that there is a very small overall difference in the total costs and primary energy consumption
between the uncontrolled case and the case in which the differentiated tariff is applied.
Nevertheless, a more evident difference can be observed between the results of Scenario
2 (individual optimization) and Scenario 4 (collaborative optimization). Not only the total
costs and the consumption of primary sources is reduced at Scenario 4, but there is no need
for grid consumption at Scenario 4, since energy is exchanged within the neighborhood, as
shown in Figure 5.26. Therefore, the influence of a collaborative optimization is stronger
than the influence of the tariff differentiation.
Conclusion
This section provides valuable information to infer that policies should start focusing more
on enabling collaborative optimization channels instead of putting so much effort in coming
up with new differentiated tariffs that may not produce such a significant difference in terms
of total costs and overall grid consumption and that might bring new local problems due
to the additional peak that can occurs when electric vehicles start charging as soon as the
differentiated tariff begins.
5.6 Conclusions
Different comparisons were performed in this chapter, from which valuable insights were
obtained. For example in the first simulation it was possible to observe that not every microCHP technology is capable of providing substantial benefits to a district in terms of cost
savings and reduction of energy consumption. For the load patterns that were analyzed, the
solid oxide fuel cell provided the best results. In Simulation 2 it was possible to observe
the influence of having a collaborative optimization. The use of an aggregator proved to be
an effective way to reduce the energy exchange with the grid. As a consequence, by using
an aggregator the need for an expansion in the infrastructure can be reduced. Simulation 3
provided results in which the influence of storage was observed. Just like it was previously
shown in Chapter 2, by using storage the multi-carrier unit commitment can be altered. In
this case, an example of a cluster of 3 houses was selected for illustrative purposes.
When a low capacity of solar panels is installed, like in Simulation 4, the influence on
the collaborative optimization is limited. It was observed that storage does not necessarily
benefit a system with renewables. Batteries are disposable and their installation should only
be performed in cases where a significant benefit can be attained. Heat storage is in that
sense a better option, especially when a large amount of heat is produced by the micro-CHP
units.
Finally in Simulation 5 the inclusion of electric vehicles was studied. It was concluded
that policy makers should focus their effort on enabling the use of local aggregators. This
not only reduces the costs and the consumption of primary sources but allows a better way to
control possible power peaks. Using differentiated tariffs enhances the risk that all electric
vehicles are plugged at the same time, which generates problems related to power capacity
of the power distribution network. In conclusion, the scenarios under which a collaborative
optimization was performed provided the best results in the analysis.
Chapter 6
The Implementation
Multi-Carrier EMS
This chapter answers Research Question 6 of this dissertation. It is focused on practical aspects related to the implementation of small-scale energy management systems designed for
systems with multiple energy carriers. In Section 6.1 a literature review is presented about
EMS products available in the market and EMS technologies under development. In Section 6.2 the modules involved in the EMS are described. Section 6.3 presents a description
of the laboratory in which the algorithms were tested. Section 6.4 contains the conclusions
of this chapter. Parts of this chapter have been already published in [116].
6.1 Literature Review
This section presents a brief literature review on devices and EMS technologies designed
for small-scale energy systems that have been developed during recent years. The literature
review considers EMS technologies that are available in the market, as well as EMS technologies under development and EMS prototypes. The energy management systems that
were taken into account in this section are designed for small-scale applications, some of
them are designed for home applications (home energy management systems - HEMS) and
others for district applications (district energy management systems - DEMS).
The Power Router is focused on the optimization of power flows of individual households. It is in charge of monitoring the power coming from renewable sources like PV
panels and in charge of commanding when to charge or discharge the batteries and when to
sell electricity to the grid. Moreover, the device can be programmed with feed-in tariffs to
schedule and optimize the use of self-generated energy.
The Power Router consists of an inverter that has two DC inputs (150-600V, 15A each
string) and independent maximum power point trackers. This allows the maximization of
yield. Moreover, there are three power output capacities available: 3kW, 3,7kW and 5kW.
The control module decides if it is better to use the power coming from the renewable energy
sources to charge the battery, to use it at the household or to feed it into the grid [117]. An
important feature is that the Power Router can be used in island mode to supply backup
133
134
6 The Implementation: Multi-Carrier EMS
power in the case of a grid power interruption. However this option only works if there is
sufficient input from the renewable sources. The owner has access to the energy balance,
revenue and solar yield via internet. Furthermore, the owner can connect from a computer
or mobile phone and retrieve the required information. The Power Router is a product of
the Dutch company Nedap. At this moment, the Power Router considers only electricity
flows at household-level applications, thus it does not support multiple energy carriers or
the coordination with other households.
Plugwise is a platform that can be used to achieve a desired energy consumption pattern
by switching home appliances automatically and by monitoring the energy consumption
[118]. According to the developers of Plugwise, up to 30% of the electricity consumption
can be saved by switching off devices. Some of the main features of the Plugwise platform are: monitoring consumption, remote control and visualization of consumption via
charts. The overall optimization is not done automatically by the program, instead, the user
makes decisions based on the monitoring results, for example to switch on a washing machine at night to take advantage of the off-peak tariff or to switch off devices when not at
home, etc. The switching is performed automatically. The platform is focused on electricity
flows, however, the developers are looking forward to including gas consumption data and
to operate a thermostat via the Plugwise platform.
The Energy Guardian is a smart-metering platform. The system is capable of collecting real-time electricity data and of helping determine how to optimize consumption by
automatically switching equipment, such as computers or large energy consumption units
like air conditioners and chillers. It can also help regulate the local voltage of the whole
installation.
The energy data can be viewed online and can be used to control the switching of certain
equipment. Energy consultants can access the data of the site remotely; in this way they can
summarize and interpret the findings, but also identify the steps that have to be taken in order
to improve the energy usage. The corresponding software provides the following features
[119]:
• Real-time display of energy usage (data can be viewed per hour, day, week, month or
year).
• It can be used to break down energy data (for example, to monitor a single equipment).
• Remote control of specific devices via the internet.
• Energy alerts by sms or email (to warn if energy usage of a particular device changes
unexpectedly).
• Control of electrical appliances individually or in groups.
• It can be used to compare energy use of different groups or individual devices.
• Regression analysis: access real-time or historical data in order to spot trends and
opportunities for energy saving.
6.1 Literature Review
135
The Energy Guardian platform focuses on electricity flows. The overall optimization is
not done automatically by the program; instead, the energy consultants identify which steps
can be taken in order to achieve an improvement. The owner has access to the information,
therefore he/she can also participate in the decision-making process.
The PowerMatcher is a software platform that was developed by the Energy Research
Centre of the Netherlands (ECN). It provides an optimization and coordination protocol of
a large number of small units including distributed generation units, electricity storage and
demand response loads.
The PowerMatcher takes the electricity price into account to determine when to charge
and discharge storage units. It is designed for different power scales, from household-level
to areas containing large number of units and several MWs. The system is based on industry
standards in both the ICT and energy sectors, thus it can be deployed in existing systems.
Moreover, the PowerMatcher is designed to support the virtual power plant concept in which
the clustering of units is aimed.
The software is based on agent control, in which a logical tree is used for the optimization. In the tree structure each leave corresponds to an agent that is associated with a unique
objective. Moreover, the root of the tree is formed by the auctioneer agent. This is a unique
agent in charge of handling the price forming. The different types of agents are described
below [120]:
• Local device agent: This agent represents a particular device. This agent coordinates
its actions with all other agents in the cluster. The agent communicates its bid to
the auctioneer and receives price updates. In this way the amount of power to be
produced or consumed is determined.
• Auctioneer agent: This agent performs the process of price-forming. It receives the
market bids of the connected agents and searches for the equilibrium price.
• Concentrator agent: This agent represents a sub-cluster.
• Objective agent: This agent is in charge of defining the objective of a cluster.
The last version of the software is now being tested at a real demonstration project. The
main focus is given to electricity flows and electrical interactions, not to multiple energy
carriers.
Other technologies can be found in the market, however they can only be used to monitor
the flows and to display the results. Three of them are listed below:
• Smart Homes and Cities: Siemens’s IEEE 802.15.4 standard for lighting and climate
control management.
• Wiser Home Energy Management System: Schneider in-home display connected to
smart meter.
• Panasonic Home Energy Management System: It monitors and displays the energy
flows in the household.
Most of the products that are found in the market are focused on the household-level and
on monitoring electricity power flows. Considerable efforts still have to be made in order to
come with an EMS capable of coordinating and optimizing several households and multiple
energy carriers.
136
6 The Implementation: Multi-Carrier EMS
6.2 Implementation of a Multi-Carrier EMS
The energy management system presented in this section incorporates the techniques and
methodologies presented in the previous chapters. The multi-carrier EMS consists of three
main types of modules: the forecast module, the optimization module and the real-time
control module. The multi-carrier EMS is designed to include high penetration of stochastically changing generation as well as multiple energy carriers, such as heat, gas and electricity coming from combined heat and power units. The approach differs from traditional
energy management systems, where only electricity flows are taken into account, and from
EMS techniques in which the decision is based on data observation from the user and is not
done automatically. Figure 6.1 shows a schematic representation of the multi-carrier energy
management system. Each module is described in the following subsections. The overall
management is accomplished by interactions among the modules.
When only one energy hub is considered, the forecast, optimization and control modules
are placed at a local level (for example at the power substation or household that is represented by an energy hub). In the case of multiple hubs (for example, multiple households as
in Chapter 5), the forecast and optimization modules are placed at the aggregator level and
the control modules are located at the local level (for example at the household level).
Figure 6.1: Diagram of the multi-carrier energy management system
6.2 Implementation of a Multi-Carrier EMS
137
6.2.1 Forecast Model
The forecast module is in charge of generating forecasts for the time series of the load and
the renewable sources, i.e. wind speed and solar radiation. In this project, the persistence
forecast method was used to determine the load forecast. However a better forecasting
method is going to be developed in future work. For the wind speed, a forecasting model
was developed using measurements taken by an anemometer that is located at the roof of
the Electrical Engineering building of TU Delft. The forecasting model is described below.
The wind power generation for a few hours ahead is characterized by high uncertainty
due to the stochastic nature of wind. Due to the influence that the wind power output has
on the optimization results, the accuracy of the wind speed forecast has a significant role in
the response of the system. For this reason, it is crucial to have reliable information about
the future wind speed values. Forecasting in micro-grids is mainly performed in short terms
with high temporal resolution, generally for the next 1-4 hours [121]. In this work, forecasts
are performed every 15 minutes for a forecast horizon of 4 hours. Moreover, wind speed
measurements are regularly made available (at least every 15 minutes). Whenever a wind
speed measurement is recorded, the forecasting model is used to predict the wind speed for
the next 16 quarters; in this way, the forecasts are regularly updated.
The model presented in this section was developed as part of a joint-collaboration paper
between Alicja Lojowska and the author of this dissertation [116]. In order to build the forecasting model, the guidelines for modeling wind speed time series presented in [122] were
followed. For this purpose, wind speed time series measurements recorded in October 2006,
in DENlab (Delft, the Netherlands) were used. The time series comprises minute-based
measurements, thus 15-minute averages were derived so that the new time series complies
with the unit scheduling frequency of 15 minutes. First, the time series was transformed
to stationary by removing features like diurnal seasonality and non-gaussian distribution.
Then, by means of statistical tools, a suitable model in the class of ARMA-GARCH models
was specified and tested. The model that was found using good statistical practice is the
ARMA(1,2)-GARCH(1,1)-T model and it is presented below:
3w,t = 0, 993w,t−1 − 0, 36̺t−1 − 0, 09̺t−2 + ̺t
(6.1)
̺t = ςt σt
(6.2)
σ2t
= 0, 01 +
0, 66σ2t−1
+
0, 23̺2t−1
(6.3)
where 3w,t [m/s] denotes the wind speed at time t and ̺t denotes the innovations or residuals
of the time series. Moreover, σ2t is the conditional variance of ̺t and ςt stands for standardized residuals which are independent, identically Student-T distributed with 5 degrees
of freedom. The model was validated with respect to the main features of wind speed: distribution, autocorrelation and persistence and this resulted in a confirmed adequacy of the
model. The forecasting model that was built using data from October 2006 can be applied to
obtain wind speed predictions for any other October. This is possible because wind speed is
characterized by annual seasonality and wind speed behavior does not change significantly
from year to year[122]. Figure 6.2 presents measurements recorded in DENlab in October
2007 and the 1-step predictions made using the wind speed time series model. The forecasted values are satisfactory for this study. The forecasts further ahead in the future are
associated with higher uncertainty and therefore may deviate more from observations.
138
6 The Implementation: Multi-Carrier EMS
Wind Speed Forecast Results
5
4.5
4
Wind speed [m/s]
3.5
3
2.5
2
1.5
1
0.5
Actual wind speed
Forecasted wind speed
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure 6.2: Results of the wind forecast model
6.2.2 Optimization Module
The optimization model is based on Chapter 2 and Chapter 5. By using the multi-carrier
unit commitment framework and its extension (presented in Chapter 5), the optimization
module defines a group of set-points for optimal operation. These set-point are calculated
considering the operating rules and constraints of the components as well as the economic,
technical and/or environmental objectives that are included in the objective function. The
input parameters for the optimization module are the following:
• Load forecast
• Forecast of renewable sources (wind speed, sun radiation, etc)
• Initial status of storage devices
• Fuel prices
• Electricity exchange prices among neighbours in the case that several households are
involved.
• Physical constraints of the components
• Other constraints and operating rules of the system.
6.3 Implementation at DENlab
139
6.2.3 Real-Time Control Module
The scheduling and dispatch obtained from the optimization module is the input for the
real-time control module. This module is based on the hierarchical architecture presented
in Chapter 4. The control module makes decisions according to the status of the system’s
control variables and components. It performs the proposed optimal dispatch and calculates
the error in order to define the proper actions to be executed. The control module is placed
at the energy hub’s local level. This energy hub can be a micro-grid configuration, like the
example shown in Chapter 4 or a household containing controllable generation units, like
the ones considered in Chapter 5.
The three modules described above make up the multi-carrier energy management system proposed in this dissertation. They are coupled to each other and allow the proper
operation of systems with multiple energy carriers. The multi-carrier EMS has the potential
to be applied at household level and at district level.
6.3 Implementation at DENlab
6.3.1 Description
DENlab is a renewable energy laboratory located at the Power Systems Group of Delft
University of Technology [123]. This laboratory can be used for demonstration projects of
micro-grid set-ups containing renewable sources. A Programmable Logic Controller is in
charge of the automation of the electromechanical processes in the laboratory. DENlab provides flexibility to emulate different load patterns and generation units by means of 9 power
converters, 2 motor-generator-sets of 5,5 kW and 30 kW and profibus communication. The
laboratory has 180 solar panels that can be connected to the test facility or redirected to the
grid. Figure 6.3 shows the solar panels. The power capacity of the laboratory is 50 kW.
The operational characteristics of the components to be studied at the laboratory can
be programmed at the main computer using specialized software. The program can then
be downloaded to the PLC. Figure 6.4 shows a picture of the PLC that is used at DENlab.
The PLC sends the set-points to nine power converters that are physically placed at the
laboratory, in this way the power supplied or consumed by the converters can be controlled,
as well as the output of the motor-generator-sets. Therefore, a variety of components can
be emulated and the real power flow at DENlab can be monitored. Figure 6.5 shows the
motor-generator-sets placed at DENlab.
At the laboratory, the components of the system can be emulated in different ways.
In the following paragraphs, three types of components are described to show how this is
done: a rotating AC device, an electric load and a DC-operating device. Figure 6.6 shows a
diagram of DENlab; this can help the reader to understand the description that is given.
140
6 The Implementation: Multi-Carrier EMS
Figure 6.3: Solar panels on the roof
Figure 6.4: PLC and back-to-back converter in DENlab
6.3 Implementation at DENlab
141
Figure 6.5: Motor-generator sets used to emulate different components
Public Grid
Autonomous Grid
Converter-set A
AC
DC
AC
DC
50 kVA
1:1
25 kVA
Converter-set B
AC
DC
AC
DC
25 kVA
Converter C
AC
AC
M
5.5 kW
G
Soft
Starter
G
Soft
Starter
Converter D
AC
AC
M
30 kW
10 kVAR
5 kVAR
Grid Connection
3 kVAR
Reactive power
compensation
Converter-set E
AC
DC
AC 50 kVA
DC
50 kVA
460/ 400V
Converter-set F
AC
DC
DC
AC 50 kVA
50 kVA
1:1
PVinstallation
Figure 6.6: DENlab configuration diagram
142
6 The Implementation: Multi-Carrier EMS
Firstly, the emulation of a rotating AC machine is done with the help of a power converter and a motor-generator-set. In this case, a wind turbine is chosen for the example.
Wind speed data is obtained from an anemometer placed at the roof of the building; this
is the input of the wind turbine model. The wind turbine model calculates the power that
would be supplied by a wind turbine for the respective wind speeds. The converter’s current
set-point is calculated from the power output of the model. The converter-sets are connected
to the motor-generator-sets, see converter-sets C and D in Figure 6.6. As a consequence, a
change in the set-point of the power converter makes the motor-generator-set to turn slower
or faster. The currents that flow to the autonomous grid in DENlab represent the currents
that would flow in an analogous physical system.
Secondly, the way to emulate an electric load is described. Load profiles are obtained
with the software SEPATH [49]. An electric load dataset is used to calculate the set-point of
the back-to-back converter that is used to emulate a load of 10 households; this corresponds
to converter-set E in Figure 6.6. Due to the fact that in this case a power load is represented,
the power flows in the opposite direction to that of the generation units. Therefore, in
Figure 6.6 power will flow from the autonomous grid to the public grid and not the other
way around, like in the previous case.
Thirdly, a description is given to indicate how to emulate DC-operating devices. In this
case back-to-back converters are used. The DC set-points of the current are provided to
a AC-to-DC converter and this is transformed back to AC by a DC-to-AC converter. It is
possible to define a two-way flow, like in the case of charging or discharging a battery, or
a one-way flow like in the case of a fuel cell. Converter-sets A, B and F are used for this
purpose.
The activities performed at the laboratory as part of this PhD project include:
• The system configuration was programmed in STEP7, which is the software that is
used to control the PLC. This was done in separately organized modules. The input
and output ports were mapped to specific memory words and each power electronic
converter was assigned a fixed number and a set of memory words to avoid wrong
interactions within the program. Due to the modularity introduced, the emulation
characteristics of one converter can be easily changed. The new structure allows
flexibility in the laboratory for the implementation of different/new components.
• An energy management system based on [123] was implemented to test if the system
performed correctly under the new software configuration. This system only considers electricity flows in the system.
• A simple multi-carrier energy management system was implemented in order to test
the algorithms designed within this PhD project. The interactions among the forecast
module, the optimization module and the real-time control module were analyzed.
The laboratory was originally designed to evaluate only electricity flows, thus it was
not possible to implement a real multi-carrier system. Mathematical models were used to
represent the heat elements instead.
6.3 Implementation at DENlab
143
6.3.2 Example at DENlab
This section shows the results of the most recent tests that were performed at the laboratory.
The autonomous energy hub in this example consists of a fuel cell, a furnace and a battery
system. The implementation was done during the MSc project [124], guided by the author
of this dissertation. The objective of this project was to implement a fuel cell model at
converter-set F (which was the last one to be acquired at the laboratory), and integrate it to
the multi-carrier EMS.
A partial multi-carrier EMS was implemented at the laboratory, however only simple
tests will be shown in this section. Figure 6.7 shows the energy hub representation. The fuel
cell was modeled using converter-set F and the battery system was modeled using convertersets A and B, see Figure 6.6. The electric load represents the consumption of 10 households.
The electric load pattern was obtained from the software SEPATH. Every minute a new setpoint for the electric load is sent to converter-set E. For this example an artificial heat load
was selected in order to show significant load variation within the 20-minutes simulation
that is presented.
Le
Pg
Lq
Figure 6.7: Energy hub representation
Table 6.1: Components and control subsystems
Hub Element
Battery Bank 1
Battery Bank 2
Fuel Cell
Boiler
Converter
in DENlab
Control
Subsystem
Set-point from
Main Control
Unit Control
A
B
electricity
electricity
current
voltage
F
none
electricity
heat
current
gas flow
none
frequency regulator
voltage regulator
delay transfer function
none
The control module follows the multi-carrier hierarchical control architecture presented
in Chapter 4. Due to the fact that there are two energy carrier forms at the output side of the
energy hub, two control subsystems were defined: the electricity control subsystem and the
heat control subsystem. Table 6.1 shows the components’ subsystem assignment.
144
6 The Implementation: Multi-Carrier EMS
Characteristic Curve of the Solid Oxide Fuel Cell
60
50
Voltage [V]
40
30
20
10
Measured data DENlab
Simulation MATLAB
Hydrogen Consumption [p.u.]
0
0
16
32
48
64
80
96
Current [A]
Hydrogen Consumption
112
128
144
160
1
0.5
0
0
1
2
3
4
5
6
7
8
9 10 11
Time [min]
12
13
14
15
16
17
18
19
20
Figure 6.8: Response of the solid oxide fuel cell
Electricity Demand and Electricity Supply
50
40
30
20
Power [kW]
10
0
−10
−20
−30
−40
−50
Battery bank
Electric load
Fuel cell
0
1
2
3
4
5
6
7
8
9 10 11
Time [min]
12
13
14
15
16
Figure 6.9: Electricity demand and electricity supply
17
18
19
20
6.3 Implementation at DENlab
145
Frequency Measured at DENlab‘s Autonomous Bus
Frequency [Hz]
51
50.5
50
49.5
49
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
18
19
20
Line Voltage Measured at DENlab‘s 3−Phase Autonomous Bus
500
Voltage [V]
450
400
350
300
250
200
0
1
2
3
4
5
6
7
8
9 10 11
Time [min]
12
13
14
15
16
17
Figure 6.10: Response of the system’s frequency and voltage
Heat Load
45
Heat load difference
Furnace
40
35
Power [kW]
30
25
20
15
10
5
0
0
1
2
3
4
5
6
7
8
9 10 11
Time [min]
12
13
14
Figure 6.11: Response of the furnace
15
16
17
18
19
20
146
6 The Implementation: Multi-Carrier EMS
The following performance criteria for the frequency is considered: the frequency should
be kept between 49 and 51 Hz during at least 95% of the time and it should not be less than
42,5 Hz or higher than 57,5 Hz [125]. The voltage and frequency are measured at DENlab’s
autonomous bus. Converter-set B (battery bank 2) acts as the master of the system, thus it
operates in a voltage source control mode, whereas converter-sets B and F operate in PQ
control mode. These terms were introduced in Section 1.3.4.
The fuel cell model represents a solid oxide fuel cell of 46,5 kW. The model is based
on [126]. The fuel cell is assigned to the electricity control subsystem. The main control
defines the power that has to be supplied by the fuel cell and sends a current set-point to
converter-set F. A delay transfer function is used to make the fuel cell’s operation smoother.
This transfer function acts as the fuel cell’s unit control.
The fuel cell was modeled both in DENlab and MATLAB. A test was made to compare the results of both models. Figure 6.8 shows the current-voltage characteristic curve.
As it can be observed, the results are satisfactory. The second figure shows the hydrogen
consumption for a period of 20 minutes.
The battery system consists of two battery banks of 25 kW each. The total system’s
storage capacity is 100 kWh. Each battery bank contributes with half of the power required
from the electricity storage. However, due to its unit control for being the master, one of the
battery banks (assigned to converter-set B) provides extra compensation if required in order
to keep the frequency and voltage stable. Both battery banks are considered as one single
hub element and they are assigned to the electricity control subsystem.
In Figure 6.9 the outputs of the fuel cell and the battery system are shown. The battery
system and the fuel cell complement each other to supply the load. By keeping the power
balance, the system’s voltage and frequency are kept within the operating limits. This can
be observed in Figure 6.10. In this example the battery set-point was manually changed to
observe the reaction of the fuel cell to the change.
In order to test the heat control subsystem it is assumed that the difference between the
heat load and the heat supplied by the fuel cell has a sinusoidal shape. Therefore, the furnace (assigned to the heat control subsystem) must supply this difference. In Figure 6.11 the
response of the furnace can be observed. The model of the furnace is based on the equations
presented un Chapter 4. The main control sends the set-point of the natural gas flow to the
furnace model. No unit control is applied in this case.
The following improvements have to be performed in a future study:
• A more detailed model of the heat exchangers is required in order to make a better
analysis. At this point only the more representative capabilities of the multi-carrier
EMS were tested.
• Renewable energy sources should be included as well as suitable forecasts.
• A more complex hub should be tested in which the motor-generator sets are also
included.
• A longer simulation should be performed. For example, a whole week including the
solar power input.
6.4 Conclusions
147
6.4 Conclusions
This chapter gives an overview of the design of a multi-carrier energy management system.
In Section 6.2.1 the forecasting model used in this dissertation was described for the wind
speed time series. Using a similar methodology, the load forecast can be improved as well.
This can substitute the persistence load forecast models that are currently being used. In the
proposed multi-carrier EMS, the optimization module is based on Chapter 2 and Chapter 5,
while the real-time control module is based on Chapter 4.
As a result of the work performed in DENlab, it was possible to program and implement
a simple multi-carrier EMS. Moreover, a set of guidelines were defined to introduce new
students to the laboratory and to help them start working in new projects without interfering with earlier projects, a separate manual was developed for this. Nevertheless, further
work is required in order to attain a proper evaluation and to be able to test more complex
configurations that also include renewable sources.
Chapter 7
The Outcome
Conclusions, Recommendations and Further Work
This dissertation provides insights and techniques that can be applied for the optimization
of systems with multiple energy carriers. The topics that were presented include the multicarrier unit commitment framework, the multi-carrier exergy hub approach, a hierarchical
multi-carrier control architecture, a comparison of multi-carrier power applications and the
implementation of a multi-carrier energy management system in a real infrastructure. Section 7.1 contains the main conclusions of this dissertation, Section 7.2 presents the recommendations for further research and Section 7.3 briefly describes a research project that has
been started as a follow-up of this dissertation.
7.1 Conclusions
Multi-Carrier Unit Commitment Framework
Several recent studies analyze the active participation of mini and micro combined heat and
power units, but they mostly focus on the electricity outputs of these units and overlook
the influence of the heat outputs that are also present. Making this assumption is simplistic
and may result in erroneous expectations. For example, a configuration might appear to
be adequate to supply a certain electric load, but might produce a heat overload if no heat
is required in the system. This kind of mistakes can be avoided by using a general multicarrier unit commitment framework like the one presented in Chapter 2. The framework
provides enough flexibility to simulate the various scenarios that were presented in this
dissertation, however it can be easily adapted for other power scales and energy carriers.
The inclusion of storage for the balancing of power has traditionally been considered
as an independent procedure to be performed only after the unit commitment solution is
known, in other words, the unit commitment solution is firstly obtained and the storage
is applied afterwards. In this dissertation the inclusion of storage was taken into account
during the calculation of the multi-carrier unit commitment solution as part of a four-step
technique. By using this technique the multi-carrier unit commitment solution can be influenced by taking the storage availability into account. The technique demonstrated to
149
150
7 The Outcome: Conclusions, Recommendations and Coming Work
be valuable for peak-shaving purposes at the generation side as shown in the example of
Chapter 2.
Multi-Carrier Exergy Hub Approach
It is valid to say that, since exergy is the maximum theoretical work potential that can be
obtained from an energy flow, by choosing the most exergetically efficient configuration
we are making the best use of the work potential of the energy source. However, the existing generation systems were not designed to make the best use of the work potential
of the sources. Thus, in order to do so, it would be necessary to re-evaluate the existing
equipment/machinery and re-design generation units in general. The objectives of this dissertation did not focus on designing or improving individual components, but on optimizing
the interaction among several of them. Therefore, the exergy hub approach was introduced
in Chapter 3 as an optimization tool for systems with multiple energy carriers. In the exergy
hub approach, exergy efficiencies (instead of energy efficiencies) were taken into account
for the optimization of systems that contain multiple energy carriers.
Different objective functions were evaluated to show that there is an intrinsic difference
between defining the most efficient system from an exergetic point of view and the most
efficient system from an energetic point of view. Moreover, an economic optimization does
not necessarily correspond to the results of the other two optimization objectives; thus a
compromise should be made in order to attain the optimal system.
The tool presented provides flexibility to easily compare different configurations. The
maximum exergetic efficiency can be obtained as a result of the optimization. One of the
strengths of this tool is that it can perform complex and long-term calculations that would be
extremely tedious when done by hand. In the example presented in Chapter 3 it was possible
to observe that the scheduling of the generation units was affected by the output temperature.
Moreover, the optimal configuration that resulted from maximizing the exergetic efficiency
of the system differed from the optimal configuration that resulted from maximizing the
energetic efficiency.
Hierarchical Multi-Carrier Control Architecture
Traditional control techniques are no longer suitable to account for the interactions introduced by combined generation units and renewable sources. In Chapter 4 a general hierarchical control architecture was presented for systems with multiple energy carriers. In the
chapter, the dynamic behavior of the generation units was considered, thus dynamic models
were used for the simulations. The results show that the multi-carrier hierarchical control
architecture is capable of dealing with perturbations and load changes in the system. The
main control parameters were kept within the defined boundary conditions throughout the
simulations.
7.1 Conclusions
151
Comparison of Multi-Carrier Power Applications
The main conclusions of Chapter 5 are enumerated below:
• Not every CHP technology is capable of providing substantial benefits to a district in
terms of energy supply. For the load patterns that were analyzed, the solid oxide fuel
cell provided the best results due to the electricity-to-heat ratio of the load patterns
used in the analysis.
• The use of an aggregator proved to be an effective way to reduce the power exchange
with the grid. By using an aggregator the need for further investments to expand the
electricity supply infrastructure can be reduced.
• By using storage the unit commitment can be influenced. In this case an example of
a cluster of 3 houses was selected for illustrative purposes. If the benefit of having
electricity storage is similar to the benefit of having heat storage, it is advisable to
opt for heat storage since it is a more sustainable technology, especially when a large
amount of heat is produced by the micro-CHP units.
• When a low capacity of renewable sources is installed, like in Simulation 4, their
influence in the overall results at the collaborative scenario is limited. If the use of
batteries does not significantly change the results, it is recommended to avoid their
installation. Batteries are disposable and their installation should only be performed
in cases where a significant benefit can be attained.
• Regarding electric vehicles, policy makers should focus their effort on enabling the
use of local aggregators. This not only reduces the costs and the consumption of
sources but allows a better way to control possible power peaks. The application of
differentiated tariffs creates the risk of having all electric vehicles being plugged at the
same time, which demands a high power capacity of the energy supply infrastructure.
Multi-Carrier Energy Management System
Chapter 6 provided a global overview of the design of a multi-carrier energy management
system and presented a brief description of its parts. Additionally, a partial implementation
in the renewable energy laboratory DENlab was performed and guidelines were defined
for the future usage of the laboratory, however more tests have to be done to evaluate the
robustness of the multi-carrier energy management system that was implemented. This kind
of physical installations are very valuable to understand how such a system works and what
limitations are likely to be found during the implementation.
152
7 The Outcome: Conclusions, Recommendations and Coming Work
7.2 Recommendations for Further Research
Multi-Carrier Unit Commitment Framework
The research presented in this dissertation was focused on the optimization of existing systems with the help of the multi-carrier unit commitment framework. Investment costs were
not part of the study because they are important only when choosing a technology at the
beginning and not during the optimization of an existing system. The combination of renewable sources, electric vehicles and the way to exchange power between houses is independent of the investment costs. Nevertheless, during the planning stage of an energy
system it would be interesting to make a comparison among technologies in which the savings attained by using an optimization tool and/or the intervention of an aggregator can be
subtracted from the investment costs. In this way a better decision can be made with respect to the technology to be selected for a certain household or district. Therefore, such an
optimization tool can be used to support the planning stage of an energy system.
Multi-Carrier Exergy Hub Approach
The topic of exergy was studied in this thesis, however only one chapter was dedicated
to it. This means that still more research can be done particularly to study systems in
which the energy sources have a low exergy content like for example heat obtained from
a solar thermal system. In this thesis only sources with high exergy content like natural
gas, biomass and electricity were evaluated. During the last phase of this PhD project
various discussions were carried out with Sabine Jansen from the Architecture Faculty; she
researches how to apply exergy studies to improve building design. From the discussions,
the following topics were defined as possible research topics to be carried out with the
optimization tool in the future:
• Include solar energy: Determine if it is exergetically better to use solar panels or
solar collectors under different building scenarios. For example, define the optimal
percentage of solar panels and solar collectors that can be placed on a roof given the
constraint of the area available for their installation.
• Include storage: Investigate whether the optimization can provide insight on how to
further reduce the required input of exergy through the application of storage.
• Include low-exergy sources: Evaluate the optimization results using a combination of
sources with low exergy content like geothermal energy and solar collector systems.
Hierarchical Multi-Carrier Control Architecture
The hierarchical multi-carrier control architecture that was presented in Chapter 4 was tested
in an off-grid energy system configuration. In order to test the architecture even further the
following possibilities are proposed:
• Test the control architecture in a system that is connected to the grid to allow back
and forward electricity flows.
7.3 Further Work
153
• Test the control architecture in a system containing more energy carriers, both at the
input side and at the output side of the energy hub.
• Evaluate the control interactions in a system with more that two control subsystems,
for example with an electrical control subsystem, a heat control subsystem and a cold
control subsystem.
Comparison of Multi-Carrier Power Applications
More simulations can be done in order to make a deeper analysis of the possibilities provided by the optimization tool. For example,
• Make a sensitivity analysis related to the penetration of renewables in which the number of panels per household is different.
• Incorporate the buffer possibilities of including electric vehicles.
• Make longer simulations, for example, for a year, in order to observe the effect of
considering longer periods of time, especially in terms of storage utilization.
Multi-Carrier Energy Management System
Further tests need to be performed at DENlab. It is recommended that all the students to
be involved in the laboratory follow the guidelines that were defined as part of this PhD
project so that their work will not interfere with earlier implemented projects. Regarding
the forecasting models, the load forecast can be improved using a similar methodology to
the one presented for the wind speed forecast model. In this way a better forecast can be
obtained.
7.3 Further Work
Research Project
A research project was started at the beginning of 2012 in which the algorithms and techniques developed during this PhD project are applied at a real office building located in The
Hague. The project is being performed by two MSc students of the Power Systems Group of
Delft University of Technology with the supervision of the author of this thesis. The results
of the project will be published at the end of 2013. The reader is encouraged to contact the
author of this thesis for more information of the project.
DENlab
Two students recently started their MSc projects at DENlab. They will continue performing
tests related to the implementation of the multi-carrier energy management system described
in Chapter 6 in DENlab with the direct supervision of the author of this thesis.
Appendix A
Assumptions and Considerations
• The optimization is solved with a mixed-integer nonlinear solver due to the ‘on’ and
‘off’ states of the units.
• The BARON solver (Branch-And-Reduce Optimization Navigator) was used to solve
the mixed-integer nonlinear problem of the multi-carrier unit commitment. BARON
is a global optimization (GO) solver: it is a computational system for solving nonconvex optimization problems to global optimality [44]. The solver can solve purely
continuous, purely integer, and mixed-integer nonlinear problems, it can also be used
to find the k-best solutions.
• The results of the optimization were corroborated using MATLAB ‘fmincon’ function. Different unit combinations were tested and the results coincided with those obtained by AIMMS. The function ‘fmincon’ attempts to find a constrained minimum
of a scalar function of several variables starting at an initial estimate. This kind of
optimization is commonly known as constrained nonlinear optimization or nonlinear
programming.
• The results of the optimal dispatch in Chapter 3 were corroborated with the hand
calculations made by exergy researcher Sabine Jansen and the results coincided.
• The algorithms used for the simulations proved to be satisfactory for 250 households
or less, which is a quantity of households that can be connected to a distribution
transformer.
• A proof of scalability is outside the scope of this research.
• It is recommended to do further research in order to evaluate the performance of the
algorithms, especially if larger systems are considered. The objective of this thesis
was not to find the algorithm with the best performance, but to evaluate the benefits
that could be obtained from a multi-carrier unit commitment optimization and the operation of an aggregator.
155
156
A Assumptions and Considerations
• In this work the 4 best solutions were saved at each of the periods of the multi-carrier
unit commitment problem. Tests were made in which 6 an 8 solutions were saved,
however the results did not improve by increasing the number of saved solutions in
the cases that were analyzed.
• A difference is considered significant if more than 10% is achieved in relation to the
reference case.
• The following performance criteria are considered in the simulations: the frequency
should be kept between 49 and 51 Hz during at least 95% of the time and it should
not be less than 42,5 Hz or higher than 57,5 Hz [125]. The temperature of the hot line
should not be lower than 95 ◦ C and the room temperature should remain between 19
and 21 ◦ C.
• Some of the data that were used for the simulations are classified as confidential. For
this reason not all data that are required for the simulations are disclosed, however the
reader can contact the author if there is further interest to obtain information about
these data.
Appendix B
Complementary Simulation
Results
This appendix includes simulation results obtained for the example presented in Section 4.5.3.
Wind Speed
Speed [m/s]
15
10
5
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Pitch Angle and Power Coefficient
Degrees
4
3
2
Power coefficient
Pitch angle
1.5
2
1
1
0.5
0
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
0
Electric Output of the Wind Tubine
Power [kW]
120
60
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure B.1: Simulation 3: Response of the wind turbine
157
Scheduled Gas Input of CHP A
Power [kW]
400
200
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Participation Factor of CHP A
[−]
1
0.5
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Actual Gas Consumption of CHP A
Power [kW]
400
200
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure B.2: Simulation 3: Response of CHP A
Scheduled Gas Input of CHP B
Power [kW]
200
100
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Participation Factor of CHP B
[−]
1
0.5
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Actual Gas Consumption of CHP B
Power [kW]
200
100
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure B.3: Simulation 3: Response of CHP B
Scheduled Gas Input of CHP C
Power [kW]
400
200
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Participation Factor of CHP C
[−]
1
0.5
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Actual Gas Consumption of CHP C
Power [kW]
400
200
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure B.4: Simulation 3: Response of CHP C
Scheduled Electric Output of the Battery Bank
Power [kW]
180
90
0
−90
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Participation Factor and Voltage at the Battery Bank
180
1
Voltage
[−]
Participation factor
160
0.5
140
Voltage [V]
−180
120
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Actual Electric Output of the Battery Bank and State of Charge
Power [kW]
180
Electric output: battery bank
90
90
State of charge
80
0
70
−90
−180
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [h]
Figure B.5: Simulation 3: Response of the battery bank
60
SOC[%]
0
Bibliography
[1] L. M. Ramirez-Elizondo and G. C. Paap, “Unit commitment in multiple energy carrier systems,” in Proceedings of the North American Power Symposium (NAPS),
2009.
[2] L. M. Ramirez-Elizondo, A. Lojowska, and G. C.Paap, “Design of an energy management system to be implemented at denlab,” in Proceedings of the 6th International
Conference Electrical & Power Engineering, Iasi, Romania, 2010.
[3] L. M. Ramirez-Elizondo, “The increasing importance of small and medium scale
renewable energy systems,” Maxwel: Magazine of the Electrotechnische Vereeniging
TU Delft, vol. 12.4, pp. 14–17, 2009.
[4] M. Reza, P. Schavemaker, W. Kling, and L. van der Sluis, “A research program on
intelligent power systems: self controlling and self-adapting power systems equipped
to deal with the structural changes in the generation and the way of consumption,” in
CIRED, 2003.
[5] The World Alliance for Decentralized Energy, http://www.localpower.org/.
[6] International Energy Agency, Key World Energy Statistics 2011, 2011.
[7] European Comission, Energy 2020: A strategy for competitive, sustainable, and secure energy, 2011.
[8] European Comission, Energy Issues: Fieldwork: March - April 2006, 2006.
[9] International Energy Agency, Contribution of Renewables to Energy Security, 2007.
[10] World Alliance for Decentralized Energy, Security via Decentralized Energy, 2005.
[11] International Energy Agency, Energy Security and Climate Policy Assessing Interactions, 2007.
[12] International Energy Agency, World Energy Outlook 2006, 2006.
[13] Intergovernmental Panel on Climate Change, Mitigation, 2001.
[14] World Alliance for Decentralized Energy, Building Integrated Cooling, Heat &
Power For Cost-Effective Carbon Mitigation, 2005.
161
162
Bibliography
[15] P. Schavemaker and L. van der Sluis, Electrical Power System Essentials. Wiley &
Sons, 2008.
[16] J. Grainger and W. Stevenson, Power System Analysis. McGraw-Hill, 1994.
[17] P. Kundur, Power System Stability and Control. McGraw-Hill Professional, 1994.
[18] International Energy Agency, Cogeneration and Renewables, 2011.
[19] International Energy Agency, Combined Heat and Power: Evaluating the Benefits of
Greater Global Investment, 2008.
[20] International Energy Agency, Co-generation and District Energy: Sustainable Energy Technologies for Today and Tomorrow, 2009.
[21] L. Haiyan and P. Valdimarsson, “District heating modelling and simulation,” in
Thirty-Fourth Workshop on Geothermal Reservoir Engineering, (Stanford, USA),
pp. –, Stanford University, 2009.
[22] International Energy Agency, District Heating and Cooling: Environmental Technology for the 21st Century, 2002.
[23] International Energy Agency, Coming from the Cold: Improving District Heating
Policy in Transition Economies, 2004.
[24] G. Boyle, Renewable Energy: Power for a Sustainable Future. Oxford University
Press, 2004.
[25] International Energy Agency, Clean Energy Progress Report, 2011.
[26] R. de Graaff, Flexible distribution systems through the application of multi back-toback converters: Concept, implementation and experimental verification. PhD thesis,
Technische Universiteit Eindhoven, 2010.
[27] J. Lopes, C. Moreira, and A. Madureira, “Defining control strategies for microgrids
islanded operation,” IEEE Transactions on Power Systems, vol. 21, no. 2, pp. 916 –
924, 2006.
[28] IEEE, Smart Grid Newsletter, January 2011.
[29] S. Ruth, “Green it more than a three percent solution?,” IEEE Internet Computing,
vol. 13, pp. 74 –78, july-aug. 2009.
[30] G. Arnold, “Challenges and opportunities in smart grid: A position article,” Proceedings of the IEEE, vol. 99, pp. 922 –927, june 2011.
[31] M. D. Ilic, “Dynamic monitoring and decision systems for enabling sustainable energy services,” Proceedings of the IEEE, vol. 99, pp. 58 –79, jan. 2011.
[32] S. Depuru, L. Wang, V. Devabhaktuni, and N. Gudi, “Smart meters for power grid:
Challenges, issues, advantages and status,” in Power Systems Conference and Exposition (PSCE), pp. 1 –7, March 2011.
Bibliography
163
[33] M. Geidl, Integrated Modeling and Optimization of Multi-Carrier Energy Systems.
PhD thesis, Eidgenössische Technische Hochschule ETH Zürich, Nr. 17141, 2007.
[34] M. Geidl and G. Andersson, “Optimal power flow of multiple energy carriers,” IEEE
Transactions on Power Systems, vol. 22, pp. 145 –155, February 2007.
[35] M. Geidl and G. Andersson, “Optimal coupling of energy infrastructures,” in Power
Tech, 2007 IEEE Lausanne, pp. 1398–1403–, 2007.
[36] M. Geidl and G. Andersson, “A modeling and optimization approach for multiple
energy carrier power flow,” in Power Tech, Proceedings of the 2005 IEEE, Russia,
pp. 1–7, 2005.
[37] M. Geidl, G. Koeppel, P. Favre-Perrod, B. Klockl, G. Andersson, and K. Frohlich,
“Energy hubs for the future,” Power and Energy Magazine, IEEE, vol. 5, no. 1,
pp. 24–30, 2007.
[38] J. Athern, The Exergy Method of Energy Systems Analysis. New York: John Wiley &
Sons, 1980.
[39] P. Sakulpipatsin, E. Boelman, and J. Cauberg, “Exergy analysis as an assessment tool
of heat recovery of dwelling ventilation systems,” International Journal of Ventilation, vol. 6, pp. 77–86, 2007.
[40] P. Sakulpipatsin, Exergy efficient building design. PhD thesis, Technische Universiteit
Delft, 2008.
[41] E. Boelman, “Exergy needs for winter ventilation in buildings,” Sustainable Building
Conference, 2002.
[42] M. A. Rosen and I. Dincer, “Exergy as the confluence of energy, environment and
sustainable development,” International Journal of Exergy, vol. 1, no. 1, pp. 3 – 13,
2001.
[43] M. A. Rosen, “Clarifying thermodynamic efficiencies and losses via exergy,” Exergy,
An International Journal, vol. 2, no. 1, pp. 3–5, 2002.
[44] J. Bisshop, AIMMS, Optimization Modeling. Paragon Decision Technology: Haarlem, the Netherlands, 1999.
[45] L. M. Ramirez-Elizondo, V. Velez, and G. C. Paap, “A technique for unit commitment
in multiple energy carrier systems with storage,” in Proceedings of the 9th International Conference on Environment and Electrical Engineering (EEEIC), 2010.
[46] A. J. Wood and B. F. Wollenberg, Power generation, operation, and control. New
York: J. Wiley & Sons, 1996.
[47] Nordhausen am Harz Tarifinformation,
accessed May
http://www.nordhausen.de//wirtschaft bildung/st strom gas.php4.
1,
2009,
164
Bibliography
[48] M. Hable, Beitrag zur Energieeninsatzoptimierung mit evolutionaeren Algorithmen
in lokalen Energiesystemen mit kombinierter Nutzung von Waerme- und Elektroenergie. PhD thesis, Technische Universitaet Dresden, 2004.
[49] J. Uitzinger, SEPATH Simulatie van EnergievraagPATronen van Huishoudens.
IVAM, Roetersstraat 33 - 1018 WB Amsterdam - Postbus 18180 - 1001 ZB Amsterdam, July 2003. project: P1999-019.
[50] L. M. Ramirez-Elizondo, G. C. Paap, R. Ammerlaan, R. Negenborn, and R. Toonssen, “On the energy, exergy and cost optimization of multi-energy-carrier power
systems,” International Journal of Exergy, vol. (accepted for publication), 2012.
[51] K. Blok, Introduction to Energy Analysis. Amsterdam: Techne Press, 2007.
[52] L. Connelly and C. P. Koshland, “Exergy and industrial ecology–part 1: An exergybased definition of consumption and a thermodynamic interpretation of ecosystem
evolution,” International Journal of Exergy, vol. 1, no. 3, pp. 146 – 165, 2001.
[53] G. Tsatsaronis, “Definitions and nomenclature in exergy analysis and exergoeconomics,” Energy, vol. 32, no. 4, pp. 249–253, 2007.
[54] M. A. Rosen, “The exergy of stratified thermal energy storages,” Solar Energy,
vol. 71, no. 3, pp. 173 – 185, 2001.
[55] M. A. Rosen, “Does industry embrace exergy?,” International Journal of Exergy,
vol. 2, no. 4, pp. 221 – 223, 2002.
[56] M. A. Rosen, “Exergy conservation: An alternative to conserving the already conserved quantity energy,” International Journal of Exergy, vol. 2, no. 2, pp. 59 – 61,
2002.
[57] M. A. Rosen and D. S. Scott, “Entropy production and exergy destruction: Part i–
hierarchy of earth’s major constituencies,” International Journal of Hydrogen Energy, vol. 28, no. 12, pp. 1307 – 1313, 2003.
[58] M. A. Rosen and D. S. Scott, “Entropy production and exergy destruction: Part ii–
illustrative technologies,” International Journal of Hydrogen Energy, vol. 28, no. 12,
pp. 1315 – 1323, 2003.
[59] A. Baghernejad and M. Yaghoubi, “Exergy analysis of an integrated solar combined
cycle system,” Renewable Energy, vol. 35, no. 10, pp. 2157–2164, 2010.
[60] L. Ozgener, A. Hepbasli, and I. Dincer, “Energy and exergy analysis of geothermal
district heating systems: An application,” Building and Environment, vol. 40, no. 10,
pp. 1309–1322, 2005.
[61] L. Ozgener, A. Hepbasli, and I. Dincer, “Exergy analysis of two geothermal district heating systems for building applications,” Energy Conversion and Management,
vol. 48, no. 4, pp. 1185–1192, 2007.
Bibliography
165
[62] N. Yildirim, M. Toksoy, and G. Gokcen, “Piping network design of geothermal district heating systems: Case study for a university campus,” Energy, vol. 35, no. 8,
pp. 3256–3262, 2010.
[63] K. Pope, I. Dincer, and G. Naterer, “Energy and exergy efficiency comparison of horizontal and vertical axis wind turbines,” Renewable Energy, vol. 35, no. 9, pp. 2102–
2113, 2010.
[64] A. D. Aazahin, I. Dincer, and M. A. Rosen, “Thermodynamic analysis of wind energy,” International Journal of Energy Research, vol. 30, no. 8, pp. 553–566, 2006.
[65] O. Ozgener and L. Ozgener, “Exergy and reliability analysis of wind turbine systems:
A case study,” Renewable and Sustainable Energy Reviews, vol. 11, no. 8, pp. 1811–
1826, 2007.
[66] L. Suganthi and A. A. Samuel, “Exergy based supply side energy management for
sustainable energy development,” Renewable Energy, vol. 19, no. 1-2, pp. 285–290,
2000.
[67] T. Krause, F. Kienzle, S. Art, and G. Andersson, “Maximizing exergy efficiency in
multi-carrier energy systems,” in Power and Energy Society General Meeting, 2010
IEEE, pp. 1 –8, 2010.
[68] C. T. Yucer and A. Hepbasli, “Thermodynamic analysis of a building using exergy
analysis method,” Energy and Buildings, vol. 43, no. 2-3, pp. 536–542, 2011.
[69] A. Schlueter and F. Thesseling, “Building information model based energy/exergy
performance assessment in early design stages,” Automation in Construction, vol. 18,
no. 2, pp. 153–163, 2009.
[70] L. Tronchin and K. Fabbri, “Analysis of buildings’ energy consumption by means of
exergy method,” International Journal of Exergy, vol. 5, no. 5-6, pp. 605–625, 2008.
[71] D. Schmidt, “Low exergy systems for high-performance buildings and communities,”
Energy and Buildings, vol. 41, no. 3, pp. 331 – 336, 2009.
[72] C. Koroneos, T. Spachos, and N. Moussiopoulos, “Exergy analysis of renewable energy sources,” Renewable Energy, vol. 28, no. 2, pp. 295 – 310, 2003.
[73] O. Arslan and R. Kose, “Exergoeconomic optimization of integrated geothermal system in simav, kutahya,” Energy Conversion and Management, vol. 51, no. 4, pp. 663
– 676, 2010.
[74] L. Ozgener, A. Hepbasli, I. Dincer, and M. A. Rosen, “Exergoeconomic analysis of
geothermal district heating systems: A case study,” Applied Thermal Engineering,
vol. 27, no. 8-9, pp. 1303 – 1310, 2007.
[75] A. Hepbasli, “A review on energetic, exergetic and exergoeconomic aspects of
geothermal district heating systems,” Energy Conversion and Management, vol. 51,
no. 10, pp. 2041 – 2061, 2010.
166
Bibliography
[76] H. Caliskan and A. Hepbasli, “Energy and exergy prices of various energy sources
along with their co2 equivalents,” Energy Policy, vol. 38, no. 7, pp. 3468 – 3481,
2010.
[77] L. Connelly and C. P. Koshland, “Exergy and industrial ecology. part 2: A nondimensional analysis of means to reduce resource depletion,” International Journal
of Exergy, vol. 1, no. 4, pp. 234 – 255, 2001.
[78] D. Castorph, M. Kollera, and P. Waas, “Brennwert und heizwert (the high and low
heating values),” tech. rep., Fachhochschule Munchen, 1999.
[79] H. Baehr, “Die exergie der brennstoffe,” Brennstoff-Wärme-Kraft, vol. 31, no. 7,
pp. 292 – 297, 1979.
[80] R. Toonssen, Sustainable Power from Biomass: Comparison of technologies for centralized or de-centralized fuel cell systems. PhD thesis, Technische Universiteit Delft,
2010.
[81] Natural
Gas.org.
Accessed
August
http://www.naturalgas.org/overview/uses eletrical.asp.
23,
2011,
[82] GE Jenbacher GmbH & Co OHG, Jenbacher gas engines Technische Beschrijving
JMS 312 GS-N.L, 2010.
[83] Parkwood Boiler Works, Keighley, West Yorkshire BD21 4NW, UK, Byworth Boilers, Hot water boilers brochure.
[84] Biomass
Energy
Centre,
accessed
http://www.biomassenergycentre.org.uk/.
August
18,
2011,
[85] V. Velez, L. M. Ramirez-Elizondo, and G. C. Paap, “Distributed energy management
systems control strategy for multiple energy carriers,” in IEEE 16th International
Conference on Intelligent Systems Aplications to Power Systems, 2011.
[86] Y. Zilong, W. Chunsheng, L. Hua, and X. Honghua, “Design of energy management system in distributed power station,” in International Conference on Sustainable Power Generation and Supply, 2009.
[87] F. Katiraei and M. R. Iravani, “Power management strategies for a microgrid with
multiple distributed generation units,” vol. 21, no. 4, pp. 1821–1831, 2006.
[88] M. S. Khan and M. R. Iravani, “Supervisory hybrid control of a micro grid system,”
IEEE Electrical Power Conference, pp. 20–24, 2007.
[89] E. Ortjohann, W. Sinsukthavorn, A. Mohd, M. Lingemann, N. Hamsic, and D. Morton, “A hierarchy control strategy of distributed generation systems,” in International
Conference on Clean Electrical Power, pp. 310–315, 2009.
[90] V. Velez, “Design of a control strategy for optimal operation of an autonomous distributed generation system for electricity and heat,” Master’s thesis, Delft University
of Technology, 2010.
Bibliography
167
[91] A. Nordmann, “Wind turbine illustration.” Wikipedia Commons.
[92] C. Wang, Modeling and control of hybrid wind/photovoltaic/fuel cell distributed generation systems. PhD thesis, Montana State University, 2006.
[93] E. Uso, “Energy storage in small-scale power systems,” tech. rep., Delft University
of Technology, 2007.
[94] D. Linden and T. Reddy, Handbook of batteries. McGraw-Hill, 2002.
[95] I. Buchmann, Battery University. Richmond: Cadex Electronics Inc., 2007.
[96] Y. Zhu and K. Tomsovic, “Development of models for analyzing the load-following
performance of microturbines and fuel cells,” Electric Power Systems Research,
vol. 62, no. 1, pp. 1 – 11, 2002.
[97] W. I. Rowen, “Simplified mathematical representations of heavy-duty gas turbines,”
Journal of Engineering for Power, vol. 105, no. 4, pp. 865–869, 1983.
[98] G. Ellis, Control System Design Guide: Using Your Computer to Understand and
Diagnose Feedback Controllers. Elsevier Science Inc., 2003.
[99] L. Xianguo, Principles of Fuel Cells. Taylor & Francis: New York, 2006.
[100] E. Carlson, Z. R., S. Sriramulu, K. Roth, and J. Brodrick, “Using phosphoric-acid
fuel cells for distributed generation,” ASHRAE, vol. 49 (1), 2007.
[101] A. Block, Heat Transfer in Steam Boiler Furnaces. Leningrad, USSR: Hemisphere
Publishing Corporation, 1988.
[102] G. Koeppel, “Distributed generation literature review and outline of the swiss situation,” tech. rep., November 2003.
[103] M. Houwing, Smart Heat and Power, Utilizing the Flexibility of Micro Cogeneration.
PhD thesis, Technische Universiteit Delft, March 2010.
[104] P. Andersen, B. Poulsen, M. Decker, C. Traeholt, and J. Ostergaard, “Evaluation of a
generic virtual power plant framework using service oriented architecture,” in IEEE
2nd International Power and Energy Conference, pp. 1212 –1217, 2008.
[105] A. Dimeas and N. Hatziargyriou, “Agent based control of virtual power plants,” in
International Conference on Intelligent Systems Applications to Power Systems, pp. 1
–6, 2007.
[106] S. You, C. Traeholt, and B. Poulsen, “A market-based virtual power plant,” in International Conference on Clean Electrical Power, pp. 460 –465, 2009.
[107] E. Mashhour and S. Moghaddas-Tafreshi, “Bidding strategy of virtual power plant for
participating in energy and spinning reserve markets, part i: Problem formulation,”
IEEE Transactions on Power Systems, vol. 26, no. 2, pp. 949 –956, 2011.
168
Bibliography
[108] E. Mashhour and S. Moghaddas-Tafreshi, “Bidding strategy of virtual power plant
for participating in energy and spinning reserve markets, part ii: Numerical analysis,”
IEEE Transactions on Power Systems, vol. 26, no. 2, pp. 957 –964, 2011.
[109] E. Mashhour and S. Moghaddas-Tafreshi, “A review on operation of micro grids
and virtual power plants in the power markets,” in 2nd International Conference on
Adaptive Science Technology, pp. 273 –277, 2009.
[110] C. Schulz, G. Roder, and M. Kurrat, “Virtual power plants with combined heat and
power micro-units,” in International Conference on Future Power Systems, pp. 5 pp.
–5, 2005.
[111] B. Roossien, M. Hommelberg, C. Warmer, K. Kok, and J.-W. Turkstra, “Virtual
power plant field experiment using 10 micro-chp units at consumer premises,” in
CIRED Seminar on Smart Grids for Distribution, pp. 1 –4, June 2008.
[112] E. Bakari and W. Kling, “Development and operation of virtual power plant system,”
Proceedings of the 2011 IEEE PES Innovative Smart Grid Technologies Conference
Europe (ISGT Europe), 2011.
[113] F. Adamek, Demand Response and Energy Storage for a Cost Optimal Residential
Energy Supply with Renewable Generation. PhD thesis, ETH Zurich, 2011.
[114] Bluegen, http://www.bluegen.info/.
[115] A. Lojowska, D. Kurowicka, G. Papaefthymiou, and L. van der Sluis, “Stochastic
modeling of power demand due to evs using copula,” 2012.
[116] L. M. Ramirez-Elizondo, A. Lojowska, V. Velez, and G. C. Paap, “Design of a smallscale energy management system to be implemented in denlab,” Bul. Inst. Polit. Iasi,
t. LVII (LXI), f. 4, 2011.
[117] Nedap Energy Systems, http://powerrouter.com/, Remote Monitoring and Management, 2011.
[118] Plugwise, http://www.plugwise.com/nl/idplugtype-f/.
[119] Energy Guardian, http://www.energy-guardian.com/.
[120] PowerMatcher, http://www.powermatcher.net/.
[121] N. D. Hatziargyriou, A. Dimeas, A. G. Tsikalakis, J. A. P. Lopes, G. Karniotakis, and
J. Oyarzabal, “Management of microgrids in market environment,” in International
Conference on Future Power Systems, 2005.
[122] A. Lojowska, D. Kurowicka, G. Papaefthymiou, and L. van der Sluis, “Advantages
of arma-garch wind speed time series modeling,” in IEEE 11th International Conference on Probabilistic Methods Applied to Power Systems (PMAPS), 2010.
[123] A. M. van Voorden, G. C. Paap, and L. van der Sluis, “Set-up of a renewable energy
laboratory,” in IEEE Young Researchers Symposium, 2002.
Bibliography
169
[124] B. Valladares, “Development of an autonomous distributed energy management system for a diverse energy sources,” Delft University of Technology, The Netherlands,
2012.
[125] M. Bollen, J. Zhong, O. Samuelsson, J. Bjornstedt, and L. Striab, “Performance indicators for microgrids during grid-connected and island operation,” in PowerTech,
2009.
[126] C. Wang and M. H. Nehrir, “A physically based dynamic model for solid oxide fuel
cells,” IEEE Transaction on energy conversion, 2007.
List of Publications
L. M. Ramirez-Elizondo et al. On the Energy, Exergy and Cost Optimization of MultiEnergy-Carrier Power Systems, International Journal of Exergy. Journal paper accepted
for publication.
F. S. Melo, A. Sardinha, S. Witwicki, L. M. Ramirez-Elizondo and M. Spaan, ‘Decentralized Multiagent Planning for Balance Control in Smart Grids’, Procs. of the 1st Int’l Workshop on Information Technology for Energy Applications (IT4ENERGY’2012), Lisbon,
Portugal, Sep. 6-7th, 2012, Vol. 923 of CEUR Workshop Proceedings, ISSN 1613-0073,
online CEUR-WS.org/Vol-923.
B. Asare-Bediako, L. M. Ramirez-Elizondo, P. F. Ribeiro, W. L. Kling, G. C. Paap, ‘Analysis
and Development of Electricity and Heat Load Profiles for Intelligent Energy Management
Systems’, UPEC 2011.
L. M. Ramirez-Elizondo, A. Lojowska, V. Velez, and G. C. Paap, ‘Design of a small-scale
energy management system to be implemented in denlab’, Bulletin of the Inst. Polit. Iasi, t.
LVII (LXI), f. 4, 2011.
V. Velez, L. M. Ramirez-Elizondo, and G. C. Paap, ‘Distributed energy management systems control strategy for multiple energy carriers’, in IEEE 16th International Conference
on Intelligent Systems Aplications to Power Systems, 2011.
L. M. Ramirez-Elizondo, V. Velez, and G. C. Paap, ‘A technique for unit commitment in
multiple energy carrier systems with storage’, in Proceedings of the 9th International Conference on Environment and Electrical Engineering (EEEIC) 2010, pp. 106 109, 16-19
2010.
L. M. Ramirez-Elizondo, A. Lojowska, and G.C. Paap, ‘Design of an energy management
system to be implemented at DENlab’, in Proceedings of the 6th International Conference
Electrical & Power Engineering, Iasi, Romania, 2010.
L. M. Ramirez-Elizondo, ‘The increasing importance of small and medium scale renewable
energy systems’, Maxwel: Magazine of the Electrotechnische Vereeniging TU Delft, vol.
12.4, pp. 1417, 2009.
171
172
List of Publications
L. M. Ramirez-Elizondo and G. C. Paap, ‘Unit commitment in multiple energy carrier systems,’ in Proceedings of the North American Power Symposium (NAPS), 2009, pp. 16, 4-6
2009.
L. M. Ramirez-Elizondo, G. C. Paap, Nico Woudstra, ‘The Application of a Fuel Cell Electrolyzer Arrangement as a Power Balancing Set-Up in Autonomous Renewable Energy Systems’, Proceedings of the 40th IEEE North American Power Symposium, Calgary, Canada,
September 28-30, 2008. (Best paper award)
L. M. Ramirez Elizondo, G. C. Paap, ‘De Elektriciteitsvoorziening van een Autonoom Systeem d.m.v. een Brandstofcel’, Poster, Symposium Vermogensconversie IOP-EMVT, April
9, 2008, Arnhem, The Netherlands.
L. M. Ramirez Elizondo, G. C. Paap, ‘Dynamic Modeling and Control of a PEMFC- Supercapacitor Autonomous Power System’, Proceedings of the 4th IEEE Benelux Young Researchers Symposium in Electrical Power Engineering, February 7-8, 2008 Eindhoven, The
Netherlands.
L. M. Ramirez Elizondo, G. C. Paap, ‘Control Strategy for a PEMFC-Supercapacitor Autonomous Power System’, Proceedings of the Second European Fuel Cell Technology and
Applications Conference EFC2007, December 11-14, 2007, Rome, Italy.
Curriculum Vitae
Laura M. Ramı́rez Elizondo was born in San José, Costa Rica. In 2003, she received her
bachelor’s degree in Electrical Engineering at the Universidad de Costa Rica. Additionally,
on August 2005 she obtained a bachelor’s degree in Music with a major in piano at the
same institution. She graduated with honors from her M.Sc. studies in Electrical Power
Systems at Delft University of Technology in 2007. Laura worked on her PhD project
from September 2007 to December 2011. Since January 2012 she has been working as
part-time researcher at the Electrical Sustainable Energy Department of Delft University of
Technology and as part-time researcher and lecturer at the Amsterdam University of Applied
Sciences. Her topics of interest include sustainable development, system integration, music,
arts, ashtanga yoga and nutrition.
Recognitions
2008 Best student paper award at the 2008 IEEE North American Power Symposium, Calgary, Canada
2007 Cum Laude graduate (graduated with honours), MSc degree in Electrical Power Systems with a minor in Sustainable Development, Delft University of Technology, the
Netherlands
2005 Awarded with a Nuffic (Netherlands Fellowship Program) scholarship to perform
MSc studies in the Netherlands
1997 Bronze medal at the National Mathematics Olympics for high school students, Costa
Rica
1996 Latin American youth representative at the Young General Assembly, organized for
the 50th Anniversary of the United Nations, United Nations Office at Geneva, Switzerland
1995 Costa Rican youth representative at the Young General Assembly, organized for the
50th Anniversary of the United Nations, San Francisco, California, the United States
of America
1993 Gold medal at the National Mathematics Olympics for primary school students, Costa
Rica
173

Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project