,
STOCKHOLM SWEDEN 2018
Reactive power management
capabilities of Swedish
sub-transmission and medium voltage
level grid.
Öland's case.
MICHAL TOMASZEWSKI
KTH ROYAL INSTITUTE OF TECHNOLOGY
Abstract
Abstract
H¨ogre genomslagskraft av f¨ornyelsebara energik¨allor i eln¨ateten ¨ar b˚ade en utmaning och m¨ojlighet f¨or att optimalt kunna utnyttja potentialen av vindkraft och PV k¨allor, med avseende p˚a att stabilisera driften av framtida elkraftsystem. Tv˚av¨agsfl¨oden mellan distribution- och transmissionsoperat¨orer orsakar betydande problem att h˚alla sp¨anningen i n¨atet inom till˚atna gr¨ansv¨arden. Denna uppsats inneh˚aller en beskrivning av ¨Olands mellan- och l˚agsp¨anningsn¨at, p˚a 0.4 kV till 130 kV i syftet att utf¨ora en kvasistatisk analys av aktiva och reaktiva effektfl¨oden i systemet. M˚alet med analysen ¨ar att optimera det reaktiva effektutbytet i kopplingspunkten med fastlandets n¨at. I det analy-serade systemet, finns det en enorm potential p˚a 190% genomslagskraft av vindkraft. Kapaciteten p˚a vindkraftsparker kopplade till medtagna sam-lingsskenor i systemet uppg˚ar till 136,1 MW, som tillgodoser upp till 90.5 MW last. Med industrim¨assigt begr¨ansad reaktiv effektkapabilitet, uppg˚ar vindkraftsparkernas bidrag till n¨astan 66 MVAr, vilken m¨ojligg¨or kompensa-tion f¨or underskott och ¨overskott av reaktiv effekt i n¨atet. Det presenterade systemet ¨ar kopplat till fastlandet genom en kopplingspunkt, d¨ar fastlandet ¨
ar simulerat som en Thevenin ekvivalent. Huvudsakliga m˚alet med denna uppsats ¨ar att testa och analysera g˚angbara l¨osningar f¨or att minimera det reaktiva effektutbytet vid kopplingspunkten i St¨avl¨o, som kopplar ihop ¨Oland med resterande n¨at i Sverige, samtidigt som alla n¨odv¨andiga villkor enligt nu-varande n¨atkoder i Sverige bibeh˚alls, liksom termiska gr¨anser f¨or ledningarna och sp¨anningsgr¨anser f¨or systemet. Ytterligare beskrivs den b¨asta tillg¨angliga tekniken som finns idag f¨or reaktiv effektkompensation, och de mest lovande teknikerna f¨or att effektivt och verkningsfullt kontrollera reaktiva effektfl¨oden. Droop-kontroll-metodologier, med fokus p˚a globala och lokala till¨ampningar, och smarta n¨at-m¨ojligheter testas och modelleras av f¨orfattarna och presen-terar djupg˚aende i detta arbete. Dessutom j¨amf¨ors ekonomiska kostnader f¨or olika kontrollmetoder. Analyser av aktiva effektf¨orluster i systemet samt kost-nader f¨or implementation av alternativa l¨osningar presenteras, d¨ar de flesta g˚angbara l¨osningar behandlas, och ger en ¨oversk˚adlig bild av framtida per-spektiv och utmaningar i elkraftsystemet. Det visas att vindturbiners kontroll av reaktiv effekt, kan f¨orb¨attra driften av eln¨aten, genom att minimera det reaktiva effektfl¨odesutbytet i gr¨ansen mellan distribution- och transmission-soperat¨orers n¨at. Ytterligare pekar resultat p˚a att extra underst¨od av reaktiv effekt fr˚an vindturbiner kan leda till f¨orminskning av aktiva f¨orluster i sys-temet. Det presenterade systemet modelleras i mjukvaruprogrammet PSS/E dedikerat f¨or elkraftsingenj¨orer med hj¨alp av Python. Analys av data gjordes antingen i Python- eller R-relaterade milj¨oer. Detta arbete har gjorts tillsam-mans med KTH och E.ON Energidistribution AB.
Acknowledgements
I would first like to thank my thesis advisor Stefan Stankovic of the Electric Power and Energy Systems at KTH Royal Institute of Technology in Stockholm. I would like to express my sincere gratitude for the continuous support of the research, for his patience and motivation. He consistently allowed this paper to be my own work, but steered me in the right the direction whenever he thought I needed it.
I would also like to thank Ingmar Leisse, the expert who was involved in this research project. His expertise allowed me to take a glimpse and understand var-ious challenges of electric power systems. I would like to sincerely thank for the continuous support during my stay at the E.On company.
I would also like to acknowledge professor Lennart S¨oder of the Electric Power and Energy Systems at KTH Royal Institute of Technology in Stockholm as the second reader of this thesis, and I am gratefully indebted for his very valuable comments on this thesis.
Table of contents 6 List of Figures 8 List of Tables 15 1 Introduction 1 1.1 Introduction . . . 1 1.1.1 Background . . . 1 1.1.2 Motivation . . . 2 1.2 Research objectives . . . 4 1.3 Thesis organization . . . 4
2 Theoretical background and reactive power control techniques 5 2.1 Theoretical background . . . 5
2.1.1 Analytical expression of power flows . . . 5
2.2 Reactive power control techniques . . . 9
2.2.1 Voltage control . . . 9
2.2.2 Impedance control - Compensators . . . 10
2.2.3 Synchronous condensers . . . 11
2.2.4 Conclusions . . . 12
3 TSO-DSO Communication and literature review of reactive power control methods 13 3.1 TSO-DSO communication . . . 13
3.2 State of the art . . . 15
3.2.1 Introduction . . . 15
3.2.2 Local objective control methods . . . 16
3.2.3 Global objective control methods . . . 19
3.2.4 Controllability of wind turbine generators . . . 21
3.2.5 Conclusions . . . 22
4 System description 23
4.1 Oland . . . .¨ 23
4.1.1 General characteristics . . . 23
4.1.2 Buses . . . 26
4.1.3 Compensators . . . 27
4.1.4 Lines and cables . . . 28
4.1.5 Transformers . . . 28
4.2 Model and data validation . . . 29
4.2.1 Location of measurement points . . . 29
4.2.2 Initial model validation . . . 32
4.2.3 Final model validation . . . 35
4.2.4 Model validation results . . . 38
4.2.5 Olands electricity system characteristics . . . .¨ 41
5 Simulation Results 51 5.1 Introduction . . . 51
5.2 Reactive power support capabilities of wind farms . . . 53
5.3 Control Schemes . . . 54
5.3.1 Local objective control schemes . . . 55
5.3.2 Global objective control schemes . . . 62
5.3.3 Costs & Savings . . . 83
6 Conclusions and future work 85 6.1 Conclusions . . . 85
6.2 Future work . . . 86
Bibliography 89 A Additional figures 93 A.1 Oland . . . .¨ 94
A.2 PSS/E model . . . 95
A.3 Smart grid algorithm . . . 96
A.4 Costs and Savings . . . 97
2.1 Representation of a line . . . 6 2.2 Apparent power transfer through a line under constant receiving voltage
and sending voltage. . . 8 4.1 Average monthly Swedish consumption profile in year 2016. . . 24 4.2 Active power consumed and generated by loads and generators located
on the island. . . 25 4.3 Reactive power consumed and generated by loads and generators located
on the island. . . 25 4.4 Location of measurement point in real grid, case 1. . . 30 4.5 Location of measurement point in real grid, case 2. . . 30 4.6 Active Power difference between simulation and measurement without
correction. . . 31 4.7 Reactive Power difference between simulation and measurement without
correction. . . 31 4.8 Active Power difference between simulation and measurement after
cor-rection of line and transformers losses. . . 34 4.9 Reactive Power difference between simulation and measurement after
correction of line and transformers losses. . . 34 4.10 Relation in difference between measurement and simulation and active
power input to the wind farm at bus 37790. . . 36 4.11 Active Power difference between simulation and measurement after
lin-ear regression data correction. . . 37 4.12 Reactive Power difference between simulation and measurement after
linear regression data correction. . . 38 4.13 Logical process and steps taken to obtain accurate input data to the
PSS/E model. . . 38 4.14 WAE of active power. Comparison between different data correction steps. 40 4.15 WAE of reactive power. Comparison between different data correction
steps. . . 41 4.16 Active and reactive power consumption by loads in ¨Oland in year 2017. 42 4.17 Active and reactive power generation by generators in ¨Oland in year 2017. 43
4.18 Maximum (left) and minimum (right) voltage profile in year 2017 present in southern (top figure) and northern (bottom figure) part of ¨Oland. . . 44 4.19 Maximum (left) and minimum (right) voltage profile in year 2018 present
in southern (top part of the figure) and northern (bottom part of the figure) part of ¨Oland. . . 45 4.20 Active and Reactive power exchanged at the point of common
connec-tion at Stavl¨o station . . . 47 4.21 Measured active against reactive power exchange at the point of
con-nection at Stavl¨o station in year 2017. . . 48 4.22 Simulated active against reactive power exchange at the point of
con-nection at Stavl¨o station in year 2017. . . 49 5.1 Simulated active against reactive power exchange at the point of
con-nection at Stavl¨o station in year 2017. . . 52 5.2 Simulated active against reactive power exchange at the point of
con-nection at Stavl¨o station in year 2018. . . 53 5.3 Exemplary droop curve used in one of the St. Istad wind farms. . . 56 5.4 Reactive power exchanged at swing bus in year 2017 after local objective
control scheme have been implemented. Scenario without STATCOM functionality. . . 60 5.5 Reactive power exchanged at swing bus in year 2017 after local
objec-tive control scheme have been implemented. Scenario with STATCOM functionality. . . 60 5.6 Reactive power exchanged at swing bus in year 2018 after local objective
control scheme have been implemented. Scenario without STATCOM functionality. . . 61 5.7 Reactive power exchanged at swing bus in year 2018 after local
objec-tive control scheme have been implemented. Scenario with STATCOM functionality . . . 61 5.8 Exemplary of general droop curve created based on the reactive power
exchanged at swing bus and active power capacity used in wind farms. . 62 5.9 Influence of multiplication coefficients of droop curve on the absolute
reactive power exchanged at swing bus. Scenario with STATCOM func-tionality. . . 64 5.10 Influence of multiplication coefficients of droop curve on the absolute
reactive power exchanged at swing bus. Scenario without STATCOM functionality. . . 64 5.11 Reactive power exchanged at swing bus in year 2017 after global
objec-tive control scheme have been implemented. Scenario without STAT-COM functionality. . . 68 5.12 Reactive power exchanged at swing bus in year 2017 after global
5.13 Reactive power exchanged at swing bus in year 2018 after global objec-tive control scheme have been implemented. Scenario without STAT-COM functionality. . . 69 5.14 Reactive power exchanged at swing bus in year 2018 after global
objec-tive control scheme have been implemented. Scenario with STATCOM functionality. . . 69 5.15 Reactive power exchanged at swing bus in year 2017 after global smart
grid control scheme have been implemented. Scenario without STAT-COM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 73 5.16 Reactive power exchanged at swing bus in year 2017 after global smart
grid control scheme have been implemented. Scenario with STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 73 5.17 Reactive power exchanged at swing bus in year 2018 after global smart
grid control scheme have been implemented. Scenario without STAT-COM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 74 5.18 Reactive power exchanged at swing bus in year 2018 after global smart
grid control scheme have been implemented. Scenario with STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 74 5.19 Reactive power exchanged at swing bus in year 2017 after global smart
grid control scheme have been implemented. Scenario without STAT-COM functionality and objective of ±5 MVAr of reactive power ex-changed at swing bus. . . 77 5.20 Reactive power exchanged at swing bus in year 2017 after global smart
grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ±5 MVAr of reactive power exchanged at swing bus. . . 77 5.21 Reactive power exchanged at swing bus in year 2018 after global smart
grid control scheme have been implemented. Scenario without STAT-COM functionality and objective of ±5 MVAr of reactive power ex-changed at swing bus. . . 78 5.22 Reactive power exchanged at swing bus in year 2017 after global smart
grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ±5 MVAr of reactive power exchanged at swing bus. . . 78 5.23 Reactive power exchanged at swing bus in year 2017 after global smart
5.24 Reactive power exchanged at swing bus in year 2017 after global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ±10% of active power exchanged at swing
bus. . . 82
5.25 Reactive power exchanged at swing bus in year 2018 after global smart grid control scheme have been implemented. Scenario without STAT-COM functionality and objective of ±10% of active power exchanged at swing bus. . . 82
5.26 Reactive power exchanged at swing bus in year 2018 after global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ±10% of active power exchanged at swing bus. . . 83
A.1 Oland’s map of wind parks of capacity exceeding 5 MW . . . .¨ 94
A.2 Oland’s PSS/E model representation.¨ . . . 95
A.3 Smart grid control algorithm . . . 96
A.4 Potential savings of global and local objective droop control schemes in year 2017. Assumed reactive power cost 26.3 SEK/MVArh and 42.2 SEK/MVARh. . . 97
A.5 Voltage profile in year 2017 present in the southern part of ¨Oland after local control scheme have been implemented. Scenario with STATCOM functionality. . . 98
A.6 Voltage profile in year 2017 present in the southern part of ¨Oland after local control scheme have been implemented. Scenario without STAT-COM functionality. . . 98
A.7 Voltage profile in year 2017 present in the northern part of ¨Oland after local control scheme have been implemented. Scenario with STATCOM functionality. . . 99
A.8 Voltage profile in year 2017 present in the northern part of ¨Oland after local control scheme have been implemented. Scenario without STAT-COM functionality. . . 99
A.9 Voltage profile in year 2018 present in the southern part of ¨Oland after local control scheme have been implemented. Scenario with STATCOM functionality. . . 100
A.10 Voltage profile in year 2018 present in the southern part of ¨Oland after local control scheme have been implemented. Scenario without STAT-COM functionality. . . 100
A.11 Voltage profile in year 2018 present in the northern part of ¨Oland after local control scheme have been implemented. Scenario with STATCOM functionality. . . 101
A.13 Voltage profile in year 2017 present in the southern part of ¨Oland after global droop control scheme have been implemented. Scenario with STATCOM functionality. . . 102 A.14 Voltage profile in year 2017 present in the southern part of ¨Oland after
global droop control scheme have been implemented. Scenario without STATCOM functionality. . . 102 A.15 Voltage profile in year 2017 present in the northern part of ¨Oland after
global droop control scheme have been implemented. Scenario with STATCOM functionality. . . 103 A.16 Voltage profile in year 2017 present in the northern part of ¨Oland after
global droop control scheme have been implemented. Scenario without STATCOM functionality. . . 103 A.17 Voltage profile in year 2018 present in the southern part of ¨Oland after
global droop control scheme have been implemented. Scenario with STATCOM functionality. . . 104 A.18 Voltage profile in year 2018 present in the southern part of ¨Oland after
global droop control scheme have been implemented. Scenario without STATCOM functionality. . . 104 A.19 Voltage profile in year 2018 present in the northern part of ¨Oland after
global droop control scheme have been implemented. Scenario with STATCOM functionality. . . 105 A.20 Voltage profile in year 2018 present in the northern part of ¨Oland after
global droop control scheme have been implemented. Scenario without STATCOM functionality. . . 105 A.21 Voltage profile in year 2017 present in the southern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 106 A.22 Voltage profile in year 2017 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 106 A.23 Voltage profile in year 2017 present in the northern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 107 A.24 Voltage profile in year 2017 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 107 A.25 Voltage profile in year 2018 present in the southern part of ¨Oland after
A.26 Voltage profile in year 2018 present in the southern part of ¨Oland af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 108 A.27 Voltage profile in year 2018 present in the northern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 109 A.28 Voltage profile in year 2018 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of 0 MVAr of reactive power exchanged at swing bus. . . 109 A.29 Voltage profile in year 2017 present in the southern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 110 A.30 Voltage profile in year 2017 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 110 A.31 Voltage profile in year 2017 present in the northern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 111 A.32 Voltage profile in year 2017 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 111 A.33 Voltage profile in year 2018 present in the southern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 112 A.34 Voltage profile in year 2018 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 112 A.35 Voltage profile in year 2018 present in the northern part of ¨Oland after
global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 5 MVAr of reactive power exchanged at swing bus. . . 113 A.36 Voltage profile in year 2018 present in the northern part of ¨Oland
A.37 Voltage profile in year 2017 present in the southern part of ¨Oland af-ter global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 114 A.38 Voltage profile in year 2017 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 114 A.39 Voltage profile in year 2017 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 115 A.40 Voltage profile in year 2017 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 115 A.41 Voltage profile in year 2018 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 116 A.42 Voltage profile in year 2018 present in the southern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario without STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 116 A.43 Voltage profile in year 2018 present in the northern part of ¨Oland
af-ter global smart grid control scheme have been implemented. Scenario with STATCOM functionality and objective of ± 10 % active power exchanged at swing bus. . . 117 A.44 Voltage profile in year 2018 present in the northern part of ¨Oland
3.1 Comparison of local and global control schemes. . . 15
4.1 Bus summary of ¨Oland’s model of electric grid . . . 26
4.2 Bus summary of ¨Oland’s electric grid system model . . . 27
4.3 Compensator summary of ¨Oland’s electric grid system model . . . 27
4.4 Lines summary of ¨Oland’s electric grid system model . . . 28
4.5 Two-winding transformers summary of ¨Oland’s electric grid system model 28 4.6 Three-winding transformers summary of ¨Oland’s electric grid system model . . . 29
4.7 Active and reactive power flow through a subsystem (MV line and two step-up transformers) presented in figure 4.5. . . 33
4.8 Active and Reactive power flow through a simple subsystem (MV line and two step-up transformers) presented in figure 4.11 in step two of data correction. . . 35
4.9 Comparison of statistical parameters before and after implementation of model optimization algorithms. . . 39
4.10 Net and absolute active and reactive power exchange at swing bus and generation by wind farms in years 2017 and 2018. . . 48
4.11 Number of hours lying outside of assumed limits of ±5 MVAR, ±10 MVAR and 10% of absolute value of active power in years 2017 and 2018. 50 4.12 Absolute value of reactive power exceeding assumed limits of ±5 MVAR, ±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018. . . 50
5.1 Reactive power capabilities of WTG. Industry standard reference value - top, scaled reactive power support - bottom . . . 54
5.2 Reactive power capabilities of St. Istad wind farm generators. . . 57
5.3 Net and absolute active and reactive power exchange at swing bus and generation by wind farms in years 2017 and 2018 after local objective control scheme have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 58
5.4 Number of hours lying outside of assumed limits of ±5 MVAR, ±10 MVAR and 10% of absolute value of active power in years 2017 and 2018 after local objective control scheme have been implemented. B -Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 58 5.5 Absolute value of reactive power exceeding assumed limits of ±5 MVAR,
±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018 after local objective control scheme have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. 59 5.6 Comparison of different droop curves analyzed and their fit to the data. 63 5.7 Net and absolute active and reactive power exchange at swing bus and
generation by wind farms in years 2017 and 2018 after global objective control scheme have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 66 5.8 Number of hours lying outside of assumed limits of ±5 MVAR, ±10
MVAR and 10% of absolute value of active power in years 2017 and 2018 after global objective control scheme have been implemented. B -Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 66 5.9 Absolute value of reactive power exceeding assumed limits of ±5 MVAR,
±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018 after global objective control scheme have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. 66 5.10 Net and absolute active and reactive power exchange at swing bus and
generation by wind farms in years 2017 and 2018 after smart grid control scheme with objective of 0 MVAr have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 71 5.11 Number of hours lying outside of assumed limits of ±5 MVAR, ±10
MVAR and 10% of absolute value of active power in years 2017 and 2018 after smart grid control scheme with objective of 0 MVAr have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 71 5.12 Absolute value of reactive power exceeding assumed limits of ±5 MVAR,
±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018 after smart grid control scheme with objective of 0 MVAr have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 72 5.13 Net and absolute active and reactive power exchange at swing bus and
generation by wind farms in years 2017 and 2018 after smart grid control scheme with objective of ±5 MVAr have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 75 5.14 Number of hours lying outside of assumed limits of ±5 MVAR and ±10
5.15 Absolute value of reactive power exceeding assumed limits of ±5 MVAR, ±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018 after smart grid control scheme with objective of ±5 MVAr have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 76 5.16 Net and absolute active and reactive power exchange at swing bus and
generation by wind farms in years 2017 and 2018 after smart grid con-trol scheme, with objective of 10% of active power exchanged, have been implemented. B - Base scenario, S - STATOM scenario, NS - No STAT-COM scenario. . . 80 5.17 Number of hours lying outside of assumed limits of ±5 MVAR, ±10
MVAR and 10% of absolute value of active power in years 2017 and 2018 after smart grid control scheme, with objective of 10% of active power exchanged, have been implemented. B - Base scenario, S - STATOM scenario, NS - No STATCOM scenario. . . 80 5.18 Absolute value of reactive power exceeding assumed limits of ±5 MVAR,
±10MVAR, and 10 % of absolute value of active power in years 2017 and 2018 after smart grid control scheme, with objective of 10% of active power exchanged, have been implemented. B Base scenario, S -STATOM scenario, NS - No STATCOM scenario. . . 80 5.19 Cost of electrical equipment needed to build bus station connected to
DSO Distribution System Operator
TSO Transmission System Operator
DG Distributed Generator
DGU Distributed Generation Unit
AC Alternating Current
DC Direct Current
EHV Extra High Voltage
HV High Voltage
MV Medium Voltage
LV Low Voltage
WTG Wind Turbine Generator
PV Photovoltaic
MW Megawatt
GW Gigawatt
MVAr Mega Voltage Ampere Reactive
GWh Gigawatt hour
TWh Terawatt hour
MWh Megawatt hour
kV Kilovolt
ENTSO-E European Network of Transmission System Operators for Electricity
FACTS Flexible Alternating Current Transmission Systems
POC Point of Connection
PCC Point of Common Coupling
US Sending Voltage
UR Receiving Voltage
IS Sending Current
IR Receiving Current
PS Sending Active Power
PR Receiving Active Power
QS Sending Reactive Power
QR Receiving Reactive Power
Un Reference Voltage
X Reactance
R Resistance
Z Impedance
PST Phase Shifting Transformer
SVC Static Var Compensator
TCR Thyristor Controlled Reactor
TSR Thyristor Switched Reactor
TSC Thyristor Switched Capacitor
STATCOM Static Compensator
VSC Voltage Source Converter
SSSC Static Synchronous Series Compensator
UPFC Unified Power Flow Controller
p.u. per unit
WAE Weighted Average Error
NC DCC Network Codes Demand Connection Code
VPRF Variation of Reactive Power
RPC Reactive Power Characteristics
OPRD Optimal Reactive Power Dispatched
BAU Business as Usual
DFIG Doubly Fed Induction Generator
Introduction
1.1
Introduction
Throughout the time, electric power systems evolved and various energy carriers were used to satisfy the demand. Since last several years, with development of intermittent renewable energy sources like wind generators and photovoltaic pan-els, energy systems came into era of sustainable and local generation, but most importantly, generation independent of scarcity of natural resources like coal, gas or uranium.
1.1.1
Background
In 2017, total generation from all available energy sources was around 24817400000 MWh. Giving on average 20985 megawatt hours consumed more every hour, due to immense electrification lasting more than 135 years. Mostly in alternating form, electricity is being supplied to the consumers through transmission and distribu-tion lines. Some of them are several hundreds meters long with nominal voltage as low as 400 Volts. Others are extremely long, reaching up to 1050 kilometers while voltage, in alternating form, goes up to 1000 kV, [Liu, 2014]. Even higher voltages are obtained with direct current transmission systems. In most extreme cases, electricity is transmitted to the length of 2385 km by a single DC link, what is almost 6% of Earth’s perimeter. Accessible capacity of such link is equal to 7.1 GW, what is being equivalent to constant supply of nearly 34 million house-holds in Brazil, every hour, [Mendon¸ca et al., 2017]. Energy systems all around the world are being a subject of many concerns of reliability and sustainability in the times of extensive investments in renewable energy sources, especially in wind and photovoltaic farms. Development of renewable energy sources imposed mean-ingful changes regarding topology of the electric grids and direction of electricity flow through low and medium voltage distribution grids as well as high voltage electric transmission grids. One of the most problematic issues is reversed power flow from low to higher voltage grid, leading to unwanted voltage increase under
low load conditions. Resulting in raise in the investment of compensating devices and transformers.Unfortunately, renewable energy generation sources are usually located in the lower, distribution voltage grids, where lines and equipment is not designed to withstand such high currents and voltages. Furthermore, alternating current flowing through the conductor, transforms the conductor to an inductor, making the current lag the voltage. In result, lines are absorbing reactive power, and dropping the voltage at the end of the line. On the other hand, in cables, that have bigger capacitance values than overhead lines, due to Ferranti Effect, voltage lag is generated with respect to the current and lines supply the reactive power. In result there is increase of the voltage at the end of the line. In both cases compen-sating devices as shunt capacitors or shunt reactors are needed to supply or absorb extra reactive power and to keep the voltages and power flow within permissible limits, making grids safe and reliable [Chavan et al., 2016].
1.1.2
Motivation
”The price of failure is too high” [Energy, 2010]. Directive of European Parlia-ment was a breakthrough in terms of evolution of energy systems we knew before. Progress of energy policies was never so rapid, setting new thresholds and changing the energy industry forever, giving boost to more local and sustainable sources of energy [Union, 2009]. As the legislation states, by 2020, Europe should be facing a goal to achieve 20 % of the overall share of energy coming from renewable en-ergy sources. From 2007, when first communication from European Commission was published, to 2018, two years before the deadline to fulfill the requirements of European Parliament, wind power grew by more than 300% and surpassed, oil, nuclear, hydro and coal settling as the second largest generation source by capacity in Europe, reaching almost 170 GW. Only in 2017, wind assets in Europe increased by 15.7 GW, with most of it in Germany and United Kingdom, [WindEurope, 2018]. Further targets are coming into force from the policy makers, setting new objectives for energy sector until 2030 and beyond. What will lead to subsequent increase in renewable energy sources share in final energy generation, [Commision, 2014]. On the the other hand, consumption and in result generation of electricity in Europe has been steady for the last couple of years, pointing that old generating sources are being rather replaced by new capacities in wind and solar and are being phased out from the system. This results in several, significant challenges that grid operators and distribution companies are facing, [Coster et al., 2011].
the protection and relay system to work in top-down coordination. Additional gen-erating units connected to lower voltage grid might cause unsynchronized reclosing or are in need for adjusted location of disconnectors and relays. Lastly, additional generators located at the lower voltage grids might influence power quality under sudden drop of wind speed or cloudy periods over PV panels. This resulted in development of new grid codes and technical requirements for connection of dis-tributed generators. ENTSO-E and other EU organizations force new legislation stating connection demands for generators, [Com, 2016]. Moreover, with new net-work codes, cooperation between DSO and TSO is at the stake of imminent need of improvement in European countries. ENTSO-E, organization connecting 43 elec-tricity transmission system operators from 36 countries across Europe, points out to improve interaction between transmission and distribution system operators for the benefit of the consumer. Series of recommendations aim to serve as basis for discussion between all electricty market entities and participants to render success-ful story written by both. Introduction of vast amount of local and dispatchable generation sources like wind farms, PV panels or geothermal power plants, force the distribution system operators to extensively control active and reactive power flows being transmitted to higher voltages and transmission grid.
In the future power grid, excess of energy generated in lower voltage parts of the system, will be pushed up, to transmission lines vastly rising the complexity of the system in terms of efficient and effective control. For example, back in the day, voltages were kept under the allowable limits by central synchronous generators connected to high voltage substations, and transformers that varied the respective tap ratios. Now, TSO instead of balancing the voltage and frequency dips by large generation units in the transmission grid, installs compensating equipment like shunt reactors and capacitors or FACTS. The same occurs for DSO. Since 2011, E.On invested plenty of money into compensating devices. During the last 7 years, company installed more than 22 shunt capacitors at different voltage levels, starting as low as 6 kilovolts.
1.2
Research objectives
Objective of the thesis is to:1. Analyze viable solutions extending capabilities of distributed generation units, particularly, wind turbine generators, to help in addressing challenges of ex-cess reactive power exchange between sub-transmission and transmission level grids in Sweden,
2. Compare results of control strategies, both technically and financially, that could be implemented in existing topology of electric grid,
3. Contrast effects of potential grid codes limiting the exchange of reactive power between sub-transmission and transmission level grids.
Control methods, extending capabilities of DGUs, are evaluated basing on po-tential grid codes stating limits of reactive power exchange at the point of con-nection (POC) in Sweden. Comparison of results puts main focus on contrasting available technical solutions and their advantages. As, for instance, Q(P ) droop curves implemented locally in clustered wind generators with local objectives as well as with global objectives in the excluded system. Furthermore, smart grid ap-proach is tested, evaluating potential of the wind farms to support voltage control within ¨Oland boundaries. In the last objective of the paper, limits stated by po-tential grid codes are divided into three different limits of reactive power exchange at the POC.
1.3
Thesis organization
Theoretical background and
reactive power control techniques
2.1
Theoretical background
2.1.1
Analytical expression of power flows
Power flow through a line can be described with use of basic Kirchhoff’s Laws and a Two-port scheme. Assuming that lines are, under normal conditions, transmitting electricity in a perfect 3-phase balance, one could represent 3-phase line by a single phase line equivalent (same procedure can be applied for cables). A simple system shown in figure 2.1, with a line between two nodes, where receiving node is written with subscriptRand sending node written with subscriptS and where Z is a line’s
impedance and YSR is a line’s shunt admittance, this representation is called π
model of a line. Kirchhoff’s current and voltage law state that, respective sum of currents flowing into a node and from a node is equal to 0, whereas sum of electrical potential differences over a closed loop is equal to 0. Relating voltage and current flowing through a line, which can be described by a simple matrix of size 2x2, containing four constants A, B, C, D, multiplied by receiving end voltage and current, giving sending end voltage and current.
US IS = A B C D ∗ UR IR
Sending and receiving end voltages and currents can be related as has been done in equation 2.1 and equation 2.2.
UR+ Z ∗ (IR+ UR∗ YSR) = US (2.1)
IR+ US∗ YSR+ UR∗ YSR= IS (2.2)
Sending end voltage is equal to the sum of receiving end voltage and voltage drop (increase) over a line due to series impedance and shunt admittance. Sending end
Figure 2.1: Representation of a line
current is equal to sum of receiving end current, current due to shunt admittance and current due to series impedance of a line. Expressing the left side of the equation (2.1) and equation (2.2) in form of a multiplication of receiving end voltage and receiving end current, and rewriting, we find sending end voltage and current.
UR∗ (1 + Z ∗ YSR) + Z ∗ IR= US (2.3)
UR∗ YSR∗ (2 + Z ∗ YSR) + (Z ∗ YSR+ 1) ∗ IR= IS (2.4)
One can notice that equations look similar to the ones seen in the matrix form of the twoport scheme. Comparing left side of the equations to the right side of the twoport matrix, we can find that constants A,B,C and D are equal to:
A = 1 + Z ∗ YSR (2.5)
B = Z (2.6)
C = YSR∗ (2 + Z ∗ YSR) (2.7)
D = Z ∗ YSR+ 1 (2.8)
Finding the equality A = D, which states that the analyzed line is symmetric. Moreover, depending on the voltage of the line, as well as its length, equations can be simplified. If a line has low values of parallel and shunt admittances, which is valid for most of overhad medium and high voltage lines. We can assume, that YSR is equal to 0, imposing D = 1 and C = 0. Putting values of A,B,C and D into
equations (2.3) and (2.4), we come to:
UR+ B ∗ IR= US (2.9)
Taken into account assumptions, which closely represent the reality, simplified our analyzed model to following statements:
Sending end voltage is equal to receiving end voltage plus voltage drop caused by the line impedance.
Sending end current is equal to receiving end current.
It could be also presumed to take the receiving end voltage as a reference value UR= |UR|∠0◦. Equation (2.3) can be rewritten, with respect to the receiving end
current IR, inserting A = |A|∠α, B = |B|∠β and US = |US|∠δ, gives a formulation
for a receiving end current: IR= |US| |B| ∗ e j(δ−β)−|A| ∗ |UR| |B| ∗ e j(α−β) (2.11)
Putting (2.11) into formula for apparent power at the receiving end, SR= UR∗ I ∗
R,
where I∗Ris the complex conjugate of the receiving end current, we have: SR= |US| ∗ |UR| |B| ∗ e j(β−δ) −|A| ∗ |UR| 2 |B| ∗ e j(β−α) (2.12)
Knowing that SR= PR+ jQR, equation (2.12) can be split into active and reactive
power at the receiving end, giving: PR= |US| ∗ |UR| |B| ∗ cos(β − δ) − |A| ∗ |UR|2 |B| ∗ cos(β − α) (2.13) And QR= |US| ∗ |UR| |B| ∗ sin(β − δ) − |A| ∗ |UR|2 |B| ∗ sin(β − α) (2.14)
For a line with a negligible parallel or shunt impedances (as stated before in the assumptions), the parameters are, thus, (A = 1 and B = Z, giving equation (2.15) and equation (2.21)). PR= |US| ∗ |UR| |Z| ∗ cos(β − δ) − |UR|2 |Z| ∗ cos(β) (2.15) QR= |US| ∗ |UR| |Z| ∗ sin(β − δ) − |UR|2 |Z| ∗ sin(β) (2.16)
This general equations describing active and reactive power flow, conclude that: 1. Maximum power that can be transmitted through the line is limited if
receiv-ing end voltage URand sending end voltage US are kept constant. If voltages
at the sending and receiving end are equal, there is no active power flow, 2. Active power flowing through a line is strongly linked to the angle δ, reaching
3. Reactive power flow is strongly linked to the sending end voltage |US|.
Figure (2.2) shows relation between equations (2.13) and (2.14). Maximum reactive power transfer occurs at specific point when β − δ = 45◦, which is equivalent to tanβδ = 1. Further simplifications can be introduced, for purely inductive line
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 X Y β − α β − δ o n P Q S φ
Figure 2.2: Apparent power transfer through a line under constant receiving voltage and sending voltage.
(where Z ≈ X), giving |B|∠β ≈ X∠90◦, equations (2.13) and (2.14), result in: PR= |US| ∗ |UR| |X| ∗ sin(δ) (2.17) QR= |US| ∗ |UR| X ∗ cos(δ) − |UR|2 X (2.18)
Similar equations are derived in relation to the sending end node (taking refer-ence for the current from sender to receiver IR= IS):
PS = |US| ∗ |UR| |X| ∗ sin(δ) (2.19) QS = |US| ∗ |UR| X ∗ cos(δ) − |US|2 X (2.20)
From the equations (2.17)(2.18)(2.19)(2.20) we see that controllability of active and reactive power can be influenced by several variables, namely, sending voltage US,
receiving end voltage UR, impedance X and power angle δ.
2.2
Reactive power control techniques
As stated in section 2.1.1, there are four ways to control reactive power flow and voltage withing electric grid. This section highlights most used methods used in real power systems.
2.2.1
Voltage control
2.2.2
Impedance control - Compensators
Formula 2.21 states that reactive power flow could be varied by impedance fluctu-ation. QR= |US| ∗ |UR| |Z| ∗ sin(β − δ) − |UR|2 |Z| ∗ sin(β) (2.21)
Impedance of the line can be varied by connecting shunt or series compensators directly with the line or by putting compensator at the substation bridging two ends of a line. There are several types of compensators that could aid to stabilize and optimize performance of the grid. Reactors and capacitors are used to, respec-tively, compensate for the effects of the line capacitance and line reactance to limit the voltage rise or voltage drop. Shunt connected compensators are, as the name indicates, placed between lines and the neutral point, what partially or completely reduces shunt susceptance, minimizing the reactive power flow. Shunt reactors, for example, limit energization overvoltages and fundamental frequency overvoltages. This is extremely common problem under no-load periods, when voltage at the receiving end becomes too high. On the other hand, shunt capacitors increase the voltage and are used to ensure that voltages at all points remain within the accept-able limits in a high load circumstances. Furthermore, when lines are inductive, shunt capacitors optimize power factor, by supplying reactive power at points where it is consumed, in order to minimize the reactive power being transmitted from i.e. generators. In other words, reactive power, should be theoretically generated as close to the reactive load as possible, so the lines are naturally loaded (reactive power is consumed or generated by the line itself and covers reactive power needs of a line). Contrarily, series reactors and capacitors compensate series impedance of the line. High impedance of the line causes high voltage drop along the line lowering the maximum capacity of the line. Series capacitors compensate inductive reactance, whereas series reactors are often used to reduce fault currents and match impedance of parallel feeders.
2.2.2.1 SVC
transistor-switched reactors to make the reactive power controllability more finely tuned by making many discrete steps.
2.2.2.2 STATCOMs
Static synchronous compensators (STATCOMs) objective is to provide precise, rapid and flexible adjustment of reactive power. In contrary to previously enlisted compensators, STATCOMs are self-commutated devices. This compensation tech-nique base on modification of magnitude and polarity of the imaginary part of the current flowing through the device. STATCOMs are often used as dynamic power factor compensators i.e. in industrial areas, with high peaks of reactive power, for example when starting up and magnetizing the motors. Beside that, it could be also used for voltage compensation at the receiving end of a line, where load is connected.
2.2.2.3 SSSC
Static synchronous series compensators depend upon voltage source converter (VSC) and are connected in series with the transmission line through a transformer. Oper-ation principle of the SSSC is to generate voltage in quadrature with the line current and display capacitive or inductive equivalent impedance by rising or declining the power flow. One of the big drawbacks of SSSC is that it can only transmit reactive power in one direction.
2.2.2.4 UPFC
Last methods used in practice are UPFC (Unified Power Flow Controller), which could be understood as an SSSC and STATCOM connected by a common dc link. Reactive power in the shunt or series converter can be controlled independently upon each other, giving great flexibility. Unified power flow controller can generate any voltage phasor in series with the transmission line, assuming the voltage phasor is within permissible limits. UPFC, apart from controlling the reactive power and therefore voltage, can be used to magnify or decrease the active power flow through a line.
2.2.3
Synchronous condensers
thus, synchronous condensers are, very often, hydrogen cooled machines. This type of compensation has been widely used before development of semi-conducting materials and introduction of power electronics in power systems. Nevertheless, in some cases, synchronous condensers are still used to adjust voltage under high wind fluctuations that could lead to rapid voltage drop. What is more, constantly rotating synchronous condensers boost system with additional inertia. Which can be used to, for example, recover voltage stability in the system.
2.2.4
Conclusions
This chapter provided a theoretical background on power flow formulations and highlighted potential possibilities to compensate excess of shortage of reactive power in real power grids. Power system engineering is a very complex and challenging branch of electrical engineering, where plenty of compromises have to be made. Assuming that the receiving end voltage is kept constant, by injecting the reactive power, sending end voltage is boosted. This is a huge concern when deliberating about future power grid because bidirectional flows are tough to control, due to following reasons:
1. Excessive injections of reactive power from distribution grid to transmission grid vastly increases the voltage in higher voltage grids that can cause damages to the equipment, thus, it has to be compensated and controlled by expensive means,
2. Some electrical machinery and equipment can compensate reactive power be-ing either inductive (capacitors) or capacitive (inductors). This imposes that advanced control techniques and vast communication has to be implemented in power systems,
3. Compensation of reactive power at POC, sometimes will come at the cost of active power, increasing the losses in the system, leading to higher operational cost of the grid. This is, as shown later in the thesis, possible when extra reactive power have to be absorbed by wind turbine generators causing voltage drop in the lines.
TSO-DSO Communication and
literature review of reactive power
control methods
3.1
TSO-DSO communication
Data acquisition devices are mostly located at large loads, generators, points of interconnection between low, medium and high voltage parts of the grid. Readings are seldom used for real-time computing to improve the way grid operates. The era of smart grids could be a close or far future. Without a doubt smart grids will be es-sential to the electric power grid, automatically and intuitively controlling voltages, power flows, load shedding and frequency based on i.e. current electricity prices on the market, risk for the distribution companies and profit for the producers. It becomes even more promising under assumption that electric vehicles could be a grid balancing mean, remotely supplying or absorbing electricity while we sleep or work. However, before communication network will be set down, grid should be equipped with advanced data acquisition apparatus. To dynamically control out-put of the wind farm in order to balance reactive power being exchanged at desired point or within specific area, dynamic line rating systems should be of particular interest for grid operators, [Fernandez et al., 2016].
In most cases, current transformer measuring power flow from a wind turbine is located at the very output of the step-up transformer or at the end of the line connecting remotely located wind generators with the rest of the system (depend-ing on the agreement between grid operator and producer). Next measur(depend-ing point, could be located several substations later, leading to lack of proper accessibility to the data needed for sufficient control of output of wind turbine. Furthermore, when it comes to clusters of wind generators, there are cases, when current transformer is measuring the output of several WTGs. This causes several problems for controlla-bility of the wind generators, for example, measurements from the aggregated wind
farm model under very windy conditions, when there is nearly complete utilization of wind generators capacity and with condition when one of the WTG is not op-erating due to maintenance. Simply, readings will be misleading even for the most excellent control algorithms. Especially when it comes to integration of wind parks, currently there is very limited or none one-way communication with WTG, which is based on static or quasi-static models [Glinkowski et al., 2011]. More issues appear around privacy, cyber security, computing power of modern processing units and how data should be shared between different entities in the energy sector.
The last problem, communication between different entities, particularly in be-tween DSO and TSO is being highlighted in this section. Recent guidance on reactive power management at the T-D (Transmission-Distribution) interface re-garding national implementation for network codes on grid connection has been published in 2014. In the document ENTSO-E requests by the means of NC DCC (Network Codes Demand Connection Code), states that TSO should not require less than maximum range of 48 percent of the larger of the maximum import ca-pability. However, the way the information is processed between transmission and distribution system operators is antique. In Europe, only in Germany, automatic active control of exchange of reactive power between transmission and distribution systems is being developed. Other countries are to follow, not having such a huge penetration of distributed generators along the grid as Germany does.
However, TSO-DSO cooperation is set up by widely accepted rules that differ from country to country. For instance in Spain, during peak periods, the power exchanged between transmission and distribution systems has to have power factor of cos(φ) > 0.95 and during off-peak periods DSO is obligated to not exchange reactive power with TSO under penalty jeopardy. In most of the countries, there is an obligation for the DSO to keep the ratio of real power to apparent power as close to 1 as operator is able to. For example, the lowest possible power factor is kept in Poland and is equal to 0.928, [ENTSO-E, 2016], giving some flexibility when it comes to reactive power compensation and improvement of grid operation at the stake of better communication between DSO and TSO. Vast investments in shunt and series compensators by most of the European distribution system operators, indicate that the regulations could not exactly be fulfilled and financial analysis confirmed feasibility of the projects (it is more economically feasible for distribution system operator to install shunt compensator than pay to TSO for the excess of reactive power being exchanged).
example, after cabling of the MV grid in France, local voltage increase and difficul-ties with reactive power flow management occured, as a consequence of saturation of the existing HV/MV transformers on load tap changers. When reactive power starts to flow from the MV to the HV network, the level of reactive power leading to saturation depends upon the characteristics of HV/MV transformer. Neverthe-less, it has to be analyzed case by case, depending on the topology of the grid. However, the path had been paved by other countries of how important the proper approach to the challenges of cabling is. The higher the HV voltage, the lower is the limit of the reactive power flow. According to European network code on demand connection, distribution networks are required to have the capacity to re-strain reactive power flows towards transmission systems especially at low active power consumption. Currently in Sweden, exchange of active and reactive power is individual matter in the contract settled for particular point of connections.
3.2
State of the art
3.2.1
Introduction
By the European standard EN50160 the end consumers in power grids is ensured that the voltage at points of coupling is kept within a bandwidth of plus minus 10% of the nominal voltage Unin 99%/95% of the time. During periods of low loads and
high generation and high loads and low generation, voltage limits could be violated. Reactive power is needed to transport active power through a line, and surpluses of it cause many problems with voltage stability in both distribution and transmission systems. This chapter, apart from ”typical” control methods i.e. on load tap changers or installation of shunt or series compensators, discuss some pioneering approaches in reactive power and voltage controllability in medium voltage grids. Method of the control of reactive power is of huge importance, however also the scale of the control system to be implemented is decisive. Control systems can be divided into two categories. In first category, there are techniques based on local compensation with local objectives and local communication. Second category encloses techniques based on global compensation with global objectives and global communication. Pros and cons of both approaches are presented in table 3.1.
Control Method Advantages Disadvantages
Local Technically simpler to implement Conflicts between controllers
Little need of communication equipment and topology knowledge Local objectives
Global Control of set of inputs and desired outputs Technically troublesome
Global and local objectives ensured without conflicts High investment costs
Table 3.1: Comparison of local and global control schemes.
time frame. This cost and pay-back time will vary between different cases, location, voltage levels and complexity, therefore, proper balance between two solutions is needed. Comparison of the two approaches are presented further in the thesis in section 5.3. Communication between distributed generators has been widely discussed by many authors on various levels, starting from type of communication (wireless or wired), voltage levels, going to frequency of the metering and delay of the control method. In this section, most focus will be put upon algorithms and methodologies to effectively control the wind turbine generators as well as PV generators.
3.2.2
Local objective control methods
power support by 55% of the initial value under only power factor control and by 95.8% under power factor and voltage control. Smart grid solution with controllable power factors, resulted in higher flexibility and better results, leading to reduction of 97% of reactive power exchange comparing to BAU scenario. Authors indicate that EPO can be adopted for seasonal power factor setting requirement, as it is done in Spain and is presented by [Espejo Mar´ın and Garc´ıa Mar´ın, 2010].
In distribution grids, which for most of the time have an inductive character because of the loads, [Concordia and Ihara, 1982], a reduction of the total reactive power consumption of distributed generation units, indeed, improves the reactive power balance of the control area. Authors in Franz et al. [2015] test minimization algorithms on disperesed generation units. Analyzed droop control methods are improved cos φ(P ) profile, which are implemented as a control scheme to reduce voltage rises. Algorithm, however, is independent of the current network charac-teristics and the current characcharac-teristics at the point of common coupling. Results of the two optimized cos φ(P ) curves for reactive power support are highlighted and analyzed. In standard cos φ(P ) control scheme, reactive power support starts at a generation of 50% of the rated power P, whereas, the highest inductive re-active power consumption occurs at maximum generation (cos φ = 0.95). In the first control method, point k(PP
r), at which unit starts to supply reactive power is
being varied and is set in a range from 0.5 to 1. Point k is determined so that the DGU causes the same maximum voltage deviation 4V at the PCC independently of the feed in power when reactive power support is demanded. Second method varies cos φ by modifying the maximum feed in a range of 0.95 to 1. However, control method B is only applicable if there is a voltage margin left regarding the limits of the technical guidelines. New optimized cos φ profile had been found for aggregated distributed generation units. Results indicated that first method saved 3% of the active power losses throughout the system, while the second one lead to savings of 3.3%. Cuts in a range of 8.1 and 9.4% of reactive power, were respec-tively achieved, for control method 1 and control method 2. Concluding, reduction of reactive power demand by distributed generators has direct financial benefits for the DSO. Especially if there is a network charge in force for the overlaid TSO that depends on the power factor, [TenneT, 2018].
control is a well known concept in conventional power systems used primarily for the load sharing among multiple generation units. Frequency of each generator is allowed to droop in accordance with delivered active power in order to share the system load. In the proposed method voltage sensitivity matrix is employed to coordinate the slope factor and the voltage threshold of each PV system along a radial feeder by considering the maximum critical voltage deviation at the last bus on the feeder. Voltage sensitivity was used as a measure to quantify the sensitivity of voltage magnitudes and angles with respect to injected active and reactive pow-ers at the bus. Each characteristic is specified by two main parametpow-ers the voltage threshold and the slope factor, which are determined based on the voltage sensi-tivity analysis and the multi objective approach in order to balance the individual reactive power distribution against the total reactive power consumption and line losses. Comparable approach, with distinction that WTGs are analyzed instead of PV generators, is performed in this paper.
In Tayab et al. [2017] authors analyze potential of implementation of droop con-trol techniques in microgrid applications. Assuming that inverter output impedance, in the conventional droop control, is purely inductive (neglecting resistance) due to its high inductive line impedance and large induction filter installed. In an inductive system the active and reactive power drawn from each inverter can be expressed as: P = E ∗ V X ∗ sin(α) and Q = E ∗ V X ∗ cos(α) − V2 X (3.1)
Since the conventional method can not provide a balanced reactive power sharing among parallel connected inverters under line impedance mismatch, imbalance in reactive power sharing is a serious problem. Similar problems occur for parallel WTGs when it comes to finding proper distribution of reactive power among every single generator. To fight this problem virtual impedance loop based droop control had been developed. In adaptive droop control the maximum reactive power taken from each unit is stored and compared with the reference value. If the maximum reactive power is less than the reference value then the voltage amplitude follows the traditional Q/E equation, however, when it is bigger, the voltage amplitude follows equation (3.2).
E = E∗− nadd(Q − Qref) (3.2)
When Q > Qref the voltage amplitude changes the slope of E(Q). On the other
hand, when Q once again becomes lower than Qref the voltage amplitude change
the line function again. Robust and adaptive methods were derived, but they will not be further deliberated in this thesis.
method to control wind farms located in 110 kV grids, ensuring desired reactive power exchange value at the transmission system interface. Obtained results proved existence of the problem of effective and efficient control of full reactive power ca-pabilities of wind farms which could benefit grid operators. Authors demonstrated that, not only by hardware like shunt and series reactors or capacitors, effective com-pensation is possible, but also novel techniques might be used to fulfill grid code requirements and conditions. Furthermore, authors in [Stock et al., 2016] used opti-mal power flow (OPF) model with predictive control for power management of 110 kV grids with high power penetration. Results are extremely appealing reaching reduction of active power losses by 3.3% and reduction in number of transformer tap operations by 29.7%. Further work is, obviously, desirable.
3.2.3
Global objective control methods
By the development of power electronics and changes in technical requirements for the connection of generators to the grid, most of nowadays WTGs or PV farms have the ability to control reactive power output or voltage by several means. Firstly, reactive power can be in a function of controllable power factor. For example in [Maih¨ofner et al., 2017], authors state that distributed energy resources can be an economically successful replacement of shunt reactors and on load tap changers in medium-voltage grid, when comprehensive approach is performed. Idea of shifting of usage of ancillary services from high voltage placed conventional power plants to medium and low voltage located distributed generators is one of the potential solutions during phasing-out of German nuclear power plants and part of the con-ventional power plants.
is increased losses (outcome of higher power flow through the line). Furthermore, simulation showed that DERs need to be controlled for over 6000 hours during the year and at least one capacitor bank is needed for the same amount of time of the year. Concluding, the reactive power exchange between HV and UHV grids can be curtailed under vertical reactive power flexibility control during hours of high active power supply by distributed energy resources. Whereas, STATCOM functionality provides high vertical reactive power flexibility independent of the existing load conditions.
In [Auer et al., 2016], estimation of the future technical potential of distribution’s grid reactive power demand, supporting the voltage management in the transmis-sion grid, is conducted. Paper presents set of scenarios, yet, due to the fact that the forthcoming premises, at least in Europe is to rely on renewable energy sources, results of the 100% renewable electricity scenario are presented in this thesis. Au-thors underline the fact that wind and solar power plants are built according to their optimal yield instead of vicinity to the customers, making the distance on which energy is transmitted to increase. To optimize the reactive power provision at the connection between HV (High Voltage) and UHV (Ultra High Voltage) a standard interior point method for non linear optimization is considered. In each iteration analyzed constraints were: apparent power of the generators Q2< C2− P2, where:
• Q reactive power • C installed capacity • P active power
Allowed voltage fluctuations at grid nodes were set to 4U = ±10% and all tap changer positions were assumed to be in discrete steps varying between 0.95, 1, 1.05, power factor was fixed and set to 0.95 inductive for loads. Each evaluated distribution grid in Germany had a potential peak load of 132 MW. Analysis showed that for 98% of the nodes the local reactive power demand can be compensated with reactive power generation from the distribution grids in the near neighbourhood of radius equal or smaller than 30 km. For the remaining 2%, up to 200 MVAr is needed for local compensation, which was being mostly reactive rather than capacitive deficit, in result leading to more investments in extra reactive power supply (shunt and series capacitors) rather than reactive power absorbers (shunt and series reactors). Active power losses are considerably higher and are mostly being accounted for capacitive generation than for inductive one, growing rapidly at maximum capacitive Q supply. This indicates that, under circumstances of maximum deficits of capacitive reactive power, it is more efficient to compensate it locally, rather than utilize DGs to do so.
wind (31 GW) is analyzed and had been optimized under realistic electricity market behaviour (merit-order based). Simulations have shown that limitations of power plants are critical with respect to voltage stability, particularly during outage of one of the nuclear power plants. Case with constant power factor of WTGs built before 2011 set to 1 and constant power factor of WTGs built after 2011 was set to 0.95 inductive, show that the stability limit is higher, than in a scenario with controllable cos φ(P ) to be 0.95 for all assets, significantly improving the stability of the German’s grid. Reactive power controls of distributed renewable energy sources have powerful impact on voltage stability of the system, enormous penetration of the wind and lack of progress in the variability of power factor of the wind generators can cause instabilities, voltage sags and in consequence blackouts.
Authors in [Hinz and Moest, 2018], evaluate reactive power support on over 5000 nodes of transmission grid in Germany of 110 kV. Linearized load flow model is utilized to estimate the reactive power potential from every transmission grid substation. What is more, reactive power potential is used to estimate the system cost impact of the distribution grid flexibilities on the model. Finally, the reactive power potential for individual nodes is calculated. Unfortunately, authors took as-sumption of static production and static grid model of German TSO. Furthermore, case with low energy feed in by wind turbine generators and perfect cooperation between different entities participating in control systems, is analyzed as without any reactive power capabilities. Results are promising, savings up to 8.9 millions of euro annually could be achieved, compensating up to 10 GVAr of average reactive power potential.
3.2.4
Controllability of wind turbine generators
Common grid codes state that, wind generators should be able to perform fault ride-through as well as be able to set extended system voltage and frequency vari-ation limits, active power regulvari-ation, frequency control, power factor and voltage regulation capabilities, [Tsili and Papathanassiou, 2009]. In Germany WTG should be able to increase or decrease active power output with a ramp rate of 10 % of grid connection capacity per minute. Whereas in Ireland, ramp rate is set up to be between 1 and 30 MW per minute. For analyzed in this thesis case, the most important capabilities of wind parks are voltage and power factor controllability. For reactive power, in Denmark wind turbine is required to be able to control re-active power from 0 to 100 % of re-active power supplied between -10% and 10% of wind turbine active power capacity. Mostly, reactive power is being used to local voltage control activities. For instance, British grid code specify that there should be an automatic system regulating voltage in the vicinity of the wind farm (at the connection point) for high capacity generating units.
decades, since, this machines can be operated at both sub-synchronous and super-synchronous speeds. At the point, when the rotor excitation voltage applied to the rotor and the direct axis are aligned to stator flux, the resultant q axis rotor current, which is in quadrature with the stator flux, affects the real power output, where the d axis rotor current Idr, being in phase with the stator flux, affects the
reactive power output. In the paper, five different operating modes are considered (maximum stator reactive power absorption mode, rotor unity power factor mode, minimum DFIG loss mode (converter loss and copper loss are minimized), stator unity power factor mode (DFIG reactive power is supplied solely by the rotor and there is no reactive power exchange between the stator and the grid), maximum reactive power generation mode.) It was observed that, when it comes to reactive power Qs(reactive power supplied) is 0 for stator unity power factor mode, negative
for maximum Qsabsorption mode, rotor unity PF mode, minimum DFIG loss mode
and is positive for maximum Qs generation mode. It was concluded that both Qs
and Qr increases following gradual grow in direct rotor current. However it may
be desirable to operate DFIG under the maximum Qs, generation mode 5, under
low voltage fault condition in order to deliver as much reactive power as possible to the grid.
Authors in [Rahimi, 2017] analyze control of rotor and grid side converters in doubly fed induction generator in order to contribute maximally to voltage reg-ulation in terms of reactive power. In DFIG-based wind turbines, the rotor side converter (RSC) controls active and reactive power injected by the generator into the grid, and the grid side converter (GSC) keeps the voltage of the DC-link con-stant regardless of the direction of the rotor power flow, [Rahimi and Parniani, 2010]. Furthermore, authors presented possibility for the GSC that can also be utilized for voltage control and reactive power support. Obtained results indicate that reduction of voltage fluctuations are possible with presented in the paper coor-dinated control method. When it comes to wind turbines with full scale converters, there is complete controllability of reactive power being absorbed or supplied by the wind farm. Full scale converter wind turbines can actively participate in voltage and frequency regulation [Blaabjerg et al., 2012].
