ÐØßÍÛ ÊÑÔÌßÙÛ ÝÑÒÌÎÑÔ ßÒÜ Ú×ÔÌÛÎ×ÒÙ ×Ò ß ÝÑÒÊÛÎÌÛÎóÚÛÜ Í×ÒÙÔÛóÐØßÍÛ ÝËÍÌÑÓÛÎóÛÒÜ ÍÇÍÌÛÓ ÑÚ ÌØÛ ÔÊÜÝ Ü×ÍÌÎ×ÞËÌ×ÑÒ ÒÛÌÉÑÎÕ
ß½¬¿ ˲·ª»®-·¬¿¬·- Ô¿°°»»²®¿²¬¿»²-·- ìðì
̸»-·- º±® ¬¸» ¼»¹®»» ±º ܱ½¬±® ±º ͽ·»²½» øÌ»½¸²±´±¹§÷ ¬± ¾» °®»-»²¬»¼ ©·¬¸
¼«» °»®³·--·±² º±® °«¾´·½ »¨¿³·²¿¬·±² ¿²¼ ½®·¬·½·-³ ·² ¬¸» ß«¼·¬±®·«³ ïíèí
¿¬ Ô¿°°»»²®¿²¬¿ ˲·ª»®-·¬§ ±º Ì»½¸²±´±¹§ô Ô¿°°»»²®¿²¬¿ô Ú·²´¿²¼ ±² ¬¸» ï笸
±º Ò±ª»³¾»®ô îðïðô ¿¬ ²±±²ò
Institute of Energy Technology
Lappeenranta University of Technology
Professor Olli Pyrhönen
Dept. of Electrical Engineering Institute of Energy Technology
Lappeenranta University of Technology
Reviewers Professor Leon M. Tolbert
Power Electronics and Engineering Laboratory
Dept. of Electrical Engineering and Computer Science University of Tennessee
Dr. Ambra Sannino
ABB Power Systems, FACTS Västerås, Sweden
Opponent Professor Victor Vtorov
The Faculty of Electrical Engineering and Automatics St. Petersburg Electrotechnical University (LETI)
ISBN 978-952-214-993-0 ISBN 978-952-214-994-7 (PDF)
ISSN 1456-4491
Lappeenrannan teknillinen yliopisto Digipaino 2010
Pasi Peltoniemi
Phase voltage control and filtering in a converter-fed single-phase customer-end sys- tem of the LVDC distribution network
Lappeenranta 2010 177p.
Acta Universitatis Lappeenrantaensis 404 Diss. Lappeenranta University of Technology
ISBN 978-952-214-993-0, ISBN 978-952-214-994-7 (PDF), ISSN 1456-4491
In recent years, the network vulnerability to natural hazards has been noticed. More- over, operating on the limits of the network transmission capabilities have resulted in major outages during the past decade. One of the reasons for operating on these limits is that the network has become outdated. Therefore, new technical solutions are studied that could provide more reliable and more energy efficient power distribution and also a better profitability for the network owner. It is the development and price of power electronics that have made the DC distribution an attractive alternative again. In this doctoral thesis, one type of a low-voltage DC distribution system is investigated. More specifically, it is studied which current technological solutions, used at the customer-end, could provide better power quality for the customer when compared with the current system.
To study the effect of a DC network on the customer-end power quality, a bipolar DC network model is derived. The model can also be used to identify the supply parameters when the V/kW ratio is approximately known. Although the model provides knowledge of the average behavior, it is shown that the instantaneous DC voltage ripple should be limited. The guidelines to choose an appropriate capacitance value for the capacitor lo- cated at the input DC terminals of the customer-end are given. Also the structure of the customer-end is considered. A comparison between the most common solutions is made based on their cost, energy efficiency, and reliability. In the comparison, special attention is paid to the passive filtering solutions since the filter is considered a crucial element when the lifetime expenses are determined. It is found out that the filter topology most commonly used today, namely the LC filter, does not provide economical advantage over the hybrid filter structure.
Finally, some of the typical control system solutions are introduced and their short- comings are presented. As a solution to the customer-end voltage regulation problem, an observer-based control scheme is proposed. It is shown how different control system structures affect the performance. The performance meeting the requirements is achieved by using only one output measurement, when operating in a rigid network. Similar per- formance can be achieved in a weak grid by DC voltage measurement. An additional improvement can be achieved when an adaptive gain scheduling-based control is intro- duced. As a conclusion, the final power quality is determined by a sum of various factors, and the thesis provides the guidelines for designing the system that improves the power quality experienced by the customer.
Keywords: Distribution of electrical energy, single phase system, converter control, passive filter UDC: 621.3.014.7 : 621.316.1 : 621.314.2
The research work of this thesis has been carried out at Lappeenranta University of Tech- nology during the years 2007-2010 in the Department of Electrical Engineering, where I have been working as a student of the Finnish Graduate School of Electrical Engineering (GSEE). The research work was conducted as a part of the "Power electronics in electric- ity distribution"-project, that was funded by the Finnish Funding Agency for Technology and Innovation (TEKES), and by several companies involved in the project.
I express my gratitude to my supervisor, Professor Juha Pyrhönen, for his guidance and encouragement and trust as he gave me almost free hands to accomplish the thesis.
I would also like to thank my other supervisor Professor Olli Pyrhönen for his support, although he got into the process only at the very end. I am grateful to Professor Jarmo Partanen for giving me the opportunity to work in such an interesting and multidiscipli- nary project, where different branches of electrical engineering collide. I also wish to thank Dr. Markku Niemelä for his valuable comments and support not only during the project, but the whole five years of graduate studies.
I would like to thank the preliminary examiners of this dissertation, Professor Leon Tolbert and Dr. Ambra Sannino for their valuable comments on the manuscript. I am very grateful for your contribution and help in improving the thesis.
I wish to thank my research colleagues in the project. It has been quite a journey for me to the mysteries of electricity distribution and you have broaden my horizon to another level. I’m also quite convinced that there is not another project going on anywhere in the world involving as many Pasis as in our project.
I owe a lot to my colleagues, who have shared an office with me for these past years.
Not only because I might have hold your research work down, but also for making the worst days quite comfortable. It also feels like we made some contributions to the field of researcher irony, that we so enthusiastically sometimes developed.
Many thanks are due to Dr. Hanna Niemelä for her contribution to improve the lan- guage of the manuscript. I would also like to thank the other personnel at the department.
The financial support by the Walter Ahlström Foundation is greatly appreciated.
Finally, I want to express my deepest gratitude to my wife, Laura, for tolerating my research enthusiasm that I so often have taken home with me and spent several hours exploring the questions of engineering. And also our kids Emilia, Noora, Riku, and Lau- riina; at the end of the day, you remind me of what is important in life.
Lappeenranta, October the 25th, 2010 Pasi Peltoniemi
I Pasi Peltoniemi, Pasi Nuutinen, Pasi Salonen, Markku Niemelä, Juha Pyrhönen; Out- put Filtering of the Customer-end Inverter in a Low-Voltage DC Distribution Net- work,In Proc. of EPE-PEMC 2008, 1–3 September, Poznan, Poland
II Pasi Peltoniemi, Pasi Nuutinen, Markku Niemelä, Juha Pyrhönen; Control of the Single-phase Customer-end Inverter in a Low-voltage DC Distribution Network,In Proc. of IECON 2008, 10–13 November, Orlando, USA
III Pasi Peltoniemi, Pasi Nuutinen, Markku Niemelä, Juha Pyrhönen; Voltage Oriented Control of LVDC distribution Inverter, In Proc. of APEC 2009, 15–19 February, Wash- ington DC, USA
IV Pasi Peltoniemi, Pasi Nuutinen, Markku Niemelä, Juha Pyrhönen; LQG-based Volt- age Control of the Single-Phase Inverter for Noisy Environment, In Proc. of EPE 2009, 8–10 September, Barcelona, Spain
Abstract 3
Acknowledgements 5
List of publications 7
Nomenclature 11
1 Introduction 19
1.1 Power distribution . . . 19
1.2 LVDC distribution system concept . . . 22
1.3 Voltage quality of distribution systems . . . 26
1.3.1 EN 50160 . . . 26
1.3.2 Other voltage quality standards . . . 27
1.3.3 Power acceptability curve . . . 28
1.3.4 Discussion . . . 30
1.4 Survey of voltage control methods . . . 31
1.5 Scope of the thesis . . . 35
1.6 Scientific contributions . . . 36
2 Customer-end system 39 2.1 Customer-end system . . . 39
2.1.1 DC/AC converter structures . . . 40
2.1.2 Filters . . . 45
2.1.3 Modulation methods . . . 51
2.2 Comparison of different customer-end structures . . . 55
2.3 Conclusions . . . 64
3 Modeling and system interactions 67 3.1 LVDC system model . . . 67
3.1.1 AC supply grid . . . 67
3.1.2 Cables. . . 71
3.1.3 Loads . . . 73
3.2 Interface between three subsystems. . . 80
3.3 LVDC network interactions . . . 81
3.3.3 Simulation study with the example LVDC network . . . 94
3.4 Summary and conclusions . . . 100
4 Voltage control of the customer-end converter 105 4.1 The control problem and objectives. . . 105
4.2 Multiloop control system . . . 107
4.2.1 Cascade PI control system . . . 107
4.3 Observer-based control . . . 114
4.3.1 LQG controller . . . 115
4.3.2 LQG-based controller structure for LVDC purposes . . . 117
4.3.3 Tuning for performance and stability. . . 120
4.3.4 Evaluation of the performance of different controller structures. . 122
4.3.5 Load adaptive controller . . . 125
4.3.6 Control implementation using a hybrid filter . . . 130
4.4 Summary and conclusions . . . 130
5 Experimental results 133 5.1 Laboratory setup . . . 133
5.2 Laboratory supply network . . . 136
5.3 Steady-state measurements . . . 137
5.4 Start-up performance . . . 145
5.5 Load transient measurements . . . 149
5.6 Summary and conclusions . . . 153
6 Summary and future work 155
References 159
Appendix A 167
Appendix B 173
Latin alphabet
A,B,C,D matrices of continuous-time state-space realization
x state vector
B flux density
C capacitance
c No-load gain
E energy
f frequency
G transfer function
h harmonic order
I RMS current
i current
J Bessel function
J cost function
k Adaptive slope
Kν exponent value for the voltage ratio L feedback gain matrix
L inductance
M Modulation index
m carrier index
N number of winding turns
n baseband index
P power, flickering index p instantaneous power R resistance, ratio
r resistance
s Laplace variable
T period, time
t time
U RMS voltage
u voltage
V volume
v speed of propagation Y,y admittance
Z,z impedance
M innovation gain
P error covariance matrix Q covariance and cost matrix R covariance and cost matrix
Greek alphabet
α,β fitting coefficient
ǫ permittivity
γ phase shift
λ wavelength
Φ,Γ matrices of discrete-time state-space realization µ voltage ratio, permeability
ω angular frequency
ω angular speed
φ phase shift, power factor
θ reference angle of the co-ordinate frame
Subscripts
δ stray inductance
k time index
s synchronous
2sw second harmonic group
ab phase-to-phase
ares antiresonance
avg average
blank blanking time
c capacitor
c carrier frequency
cable DC cable related variable
com commutation
cond conduction ctrl control system
D diode
dc direct current delay time delay error current error
F forward voltage
H hybrid filter
inv inverter
iron iron losses
L maximum demand current, inductor L1 converter side inductor
L2 load side inductor LC tuned filter LL line-to-line
load load experienced by the DC network
lt long-term
max maximum, saturation minpw minimum pulse
N neutral conductor
n negative pole
n nominal
o output
p positive pole
pri primary winding rated rated value
res resonance
rr reverse recovery
s sampling
s supply
sc short-circuit sec secondary winding
st short-term
sw switching
T transformer
tr transistor
Abbreviations
AC alternating current
BIBO Bounded input bounded output BSN B-spline network
CBEMA Computer Business Equipment Manufacturing Association CPC customer point-of-connection
DARE Discrete-time Algebraic Riccati Equation DC direct current
DPG Distributed Power Generation DPS Distributed Power System EMI Electromagnetic interference
HFL-ILC High-frequency-link integral-half-cycle HFT High-frequency transformer
HT Hilbert Transformation
IEC International Electrotechnical Commission IEEE Institute of Electrical and Electronics Engineers IGBT Insulated Gate Bipolar Transistor
ILC Iterative Learning Control
ITIC Information Technology Industry Council
KE Kalman Estimator
LAC Load Adaptive Controller LMI Linear Matrix Inequality
LPF Low-pass filter
LQG Linear Quadratic Gaussian LQR Linear Quadratic Regulator LTR Loop Transfer Recovery
LV Low voltage
LVDC Low-voltage direct current
MRAC Model Reference Adaptive Control
MV Medium voltage
NETL National Energy Technology Laboratory NPC Neutral-point clamped
OSAP One sample ahead preview PCC Point-of-common coupling PDM Pulse Density Modulation
PEBB Power Electronic Building Block PFC Power Factor Correction
PLL Phase-Locked Loop PT Park Transformation PWM Pulse Width Modulation QFT Quantitative Feedback Theory RHP Right Half-plane Pole
RMS Root Mean Square
SG Smart Grid
SMC Sliding-mode control
SOGI Second Order Generalized Integrator
TD Transport delay
THD Total harmonic distortion UPS Uninterruptible Power Supply
VSC Variable structure control ZCS Zero-current switching
ZEDS Zonal Electric Distribution System ZVS Zero-voltage switching
Introduction
1.1 Power distribution
A typical structure of the power distribution system in Finland reaches from 110 kV to the customer. The system is based on 50 Hz AC three-phase voltage. The voltage level starting from the 110 kV also includes 20 kV in the medium-voltage (MV) network and 400 V in the low-voltage (LV) network. According to Lakervi and Partanen (2008), a typical apparent power of the transformer is 10–40 MVA at 110 kV power stations that are supplying medium-voltage networks. The transformers supplying LV networks are usually around one hundred kVA. This network structure was built in Finland after the Second World War. It means that the age of the components of the distribution network is close to a point where they have to be replaced by new ones. Moreover, as the consump- tion of electric energy has grown ever since the construction of the network, the current distribution network will soon become inadequate to meet the demand. The situation is similar in various European countries and also in the US. Therefore, there is a vast num- ber of ongoing research projects about how to improve the current distribution network structure in order to meet the demands of the future for power quality, energy efficiency, and reliability. The reliability in the Finnish distribution network is mainly determined by the faults occuring in the MV network. Lakervi and Partanen(2008) claim that over 90 % of the outages experienced by the customers are due to faults in MV network. The reliability issues have also given rise to a power distribution concept called Smart Grids (SG) (EU, 2010; US, 2010), and the discussion on the reliability of the current network structure has got even more intense after the severe large-scale outages both in the US and Italy during 2003. The National Energy Technology Laboratory (NETL) in the US has
1 1 0 / 2 0 k V 2 0 / 0 . 4 k V
S ~ 1 0 - 4 0 M V A S ~ 1 0 0 k V A
Figure 1.1: Structure of the current AC distribution network.
listed the targeted properties of the SG (NETL,2010):
• self-healing from power disturbance events,
• enabling active participation by consumers in demand response,
• operating resiliently against physical and cyber attack,
• providing power quality for the 21st century needs,
• accommodating all generation and storage options,
• enabling new products, services, and markets, and
• optimizing assets and operating efficiently.
The SG concept is based on the future trend, which is to distribute the energy pro- duction to various levels of the transmission and distribution networks. This prevents the building of robust transmission networks that meet the requirements. In the literature, the distribution of energy production is often referred to as distributed power generation (DPG). An inherent feature of almost every DPG plant is that they apply power electron- ics to interface with the grid. The utilization of power electronics in large scale for power distribution purposes proposes various questions that have not been fully answered yet.
Specifically, these questions include the questions of the aforementioned power quality, reliability, and energy efficiency. In addition to these questions, there is also a question of the best possible solution to interface DPG plants in large scale to the network. One possible solution is to use a low-voltage DC (LVDC) distribution system. When the high level of intelligence and communication is included in the LVDC distribution system, it could also be considered to include the elements that make the SG. The concept of the LVDC distribution system is the topic of the next section.
By far, the LVDC is not the first distribution system that uses DC voltage to deliver power. Hammarstrom(2007) even questions if we are using the right type of voltage in our current distribution. In the paper, the conversion efficiency costs of adopting vari- ous premise AC and DC distribution system topologies are addressed. It is noticed that when a residence is supplied from a DC source, the total conversion efficiency within a residential DC distribution system could be similar to, or even better than, that for AC distribution. A residential small-scale DC system is proposed byEngelen et al. (2006).
Consequently, the residential DC distribution is not recommended, which is the opposite to that stated byHammarstrom(2007). Further research should instead focus on the ex- tension of DC power delivery to higher levels of the electricity grid. That is because only a very small increase in efficiency is attained by using a residential DC system. Further- more,Engelen et al.(2006) mention their concern about the losses that power electronics produce with partial loading and the safety issues that involve DC voltages and currents.
Brenna et al.(2004) have proposed another kind of DC distribution network that is sup- posed to integrate various generators, storage systems, and DC loads (Fig. 1.2). The DC network is interfaced with a local public grid by means of one or more inverters. A slightly similar DC distribution system is investigated byKarlsson (2003). The system is meant to interface the distributed power system devices. In the grid there is possibly
B a l a n c e d l o a d
B a l a n c e c o n v e r t e r
U n b a l a n c e d l o a d
G
I n t e r f a c e c o n v e r t e r
Figure 1.2: DC network structure proposed byBrenna et al.(2004).
multiple inverters that supply the DC network. The load sharing between the rectifiers is carried out by using a droop control method. It is found in the paper that the risk of entering converter overmodulation is a stronger limitation than stability when the DC bus cable has reasonable parameters.
In addition to residential networks and DPG interfaces, DC networks have also been used for other purposes. Opportunities and challenges of adopting a DC distribution scheme for industrial power systems is investigated by Baran and Mahajan (2003) (Fig.
1.3). The issues that are being focused on include the interactions between connected power converters and also the question of the network grounding. It is shown that con- verter interactions can be minimized with a proper filtering and with the control of the converters. However, only simulation results have been presented. A naval DC distrib- ution system is investigated byCiezki and Ashton(2000). It is claimed in the paper that significant gains can be realized in terms of survivability, weight, manning, and cost, when the current AC radial distribution system is replaced with a DC Zonal Electric Distribu- tion System (DC ZEDS). In the DC ZEDS system, the main bus DC voltage is stepped down within the zone and then converted to three-phase AC and lower DC voltage by ad- ditional power converters. Because of the large interconnection of power converters, the negative input impedance effects create the possibility of unwanted resonances. These stability effects have been discussed in the paper in addition to fault detection and load
D C b u s
D C D C D Cl o a d
D C D C D Cl o a d
G e n 1 G e n 2
Z o n e Z o n e
Rectifier 1 Rectifier 2
Figure 1.3: DC network application proposed byBaran and Mahajan(2003).
shedding problems. Thandi et al. (1999) introduce a Power Electronic Building Block (PEBB) based distribution power system (DPS), comprising of a front-end power factor correction (PFC) boost rectifier, a DC-DC converter, and a three-phase four-leg inverter.
A comprehensive modeling and analysis of the proposed system is performed. The ef- fect of impedance overlap on the system and individual subsystems are examined. It is found in the paper that the DC link voltage may become oscillatory. Three methods are mentioned to avoid stability problems, namely, (1) increasing the DC link capacitance, (2) increasing the filter damping, or (3) decreasing the load inverter bandwidth. Finally, Kakigano et al.(2007) have proposed a micro-grid type of DC network (Fig. 1.4). The DC network is meant to operate as a residential supply grid, where it is also possible that the customer’s power production devices, such as solar panels are connected. The structure of the DC link is bipolar and it uses 170 V, 0 V, and -170 V DC voltage levels.
Depending on the need, the DC voltage could be transformed into either a three-phase or single-phase AC voltage or to another DC voltage level.
Despite the vast amount of different DC distribution concepts, the LVDC concept differs from the ones already investigated. Furthermore, according to the results presented in the papers, a system such as LVDC is expected to have more potential than the system that is intended only for residential purposes.
1.2 LVDC distribution system concept
The main objective of the LVDC distribution system is to provide reliable electric energy transmission from the MV network to the LV customers. Furthermore, the required prop- erties of the system include the ability to provide good-quality voltage supply, where the
A C / D C
D C / D C
s e c o n d a r y b a t t e r y b i d i r e c t i o n a l
r e c t i f i e r
i n v e r t e r
G P C
P C
P C
P C
P C
6 . 6 k V / 2 3 0 V
s i g n a l l i n e p o w e r s u p p l y
S u p e r v i s o r P C
L o c a l P C
D C 1 0 0 V 1 f 1 0 0 V
3 f 2 0 0 V D C 1 7 0 V
D C 1 7 0 V
1 f 1 0 0 V
Figure 1.4: Residential DC network structure proposed byKakigano et al.(2007).
safety of the user is guaranteed. Although personnel safety and system protection are of primary importance, they are outside the scope of the thesis and therefore not investigated here. However, safety issues have been investigated with a similar LVDC distribution network. In (Salonen et al.,2009b), the proctection scheme for an LVDC distribution net- work is proposed. Furthermore, the paper presents various possible faults and protection requirements for the network. Different faults are also analyzed in (Salonen et al.,2008a).
In (Salonen et al.,2009a), the functionality of the proposed protection scheme is verified by measurements. The measurements are carried out with a LVDC distribution system prototype. The effect of the grounding on the network and the protection system structure is considered in (Salonen et al.,2008b)
The motivation to use the LVDC distribution network type of solution lies in the op- portunity to increase the reliability of electricity distribution and the end-customer voltage quality (Kaipia et al.,2009). Furthermore,Lassila et al.(2009) have calculated the life cy- cle costs, when different typically applied technologies are used to construct the example network. The life cycle costs are illustrated in Fig.1.5, where it is clearly visible that the price of power electronics govern the obtained economical benefit when compared with the technologies used today. In addition to these benefits, the network allows an increase in the level of intelligence in the distribution system and the use of as high transmission capacity as possible. A higher transmission capacity of the LV system enables the build- ing of longer LV networks. The ability to build longer LV networks also means that the length of the expensive MV networks can be decreased. By using underground cables when building LV networks, the reliability of the system can be increased by reducing the length of the MV network, and furthermore, the system becomes more tolerant of natural hazards when compared with overhead lines. Finally, the option to use a higher transmis- sion capacity by higher voltages means that in order to deliver the same amount of energy to the customer, a lower current is needed. The need for a lower current leads to lower transmission losses, when the same cable cross-sections are used. In the case of smaller cable cross-sections, the losses remain approximately the same but the investment costs
Figure 1.5: Comparison of costs, carried out with an example LVDC network struc- ture in (Lassila et al.,2009).
needed to build the system decrease. In spite of the technical benefits, it is the cost and the profitability that determine the possible success of the LVDC concept. Therefore, the solutions to be used in the system must be carefully chosen, because the direct and indi- rect costs are totally determined by them. The structure of the LVDC distribution system that is used throughout the thesis is introduced next.
When starting from the MV network interface, the system is connected using a trans- former the structure and secondary voltage level of which are determined by the applied DC voltage level. The DC voltage level is determined by the structure of the DC link, which also dictates the structure of the rectifier required. The LVDC distribution network can be realized by using either a unipolar or bipolar DC network structure (Fig. 1.6).
The maximum voltage that can be used in the DC network is determined by the EU low- voltage directive 2006/95/EC (LVD,2006) and it is 1500 VDC. In the unipolar structure, in order to maximize the transmission capacity, a full 1.5 kV should be used. However, such a high voltage level makes the use of the unipolar DC link structure undesirable, since the commercially available IGB transistors having high enough nominal voltage for 1.5 kV also have quite high nominal currents, which results in an increase in their cost. Furthermore, the use of the unipolar structure would also become challenging when considering the voltage ratings of customer-end devices. Therefore, the bipolar structure could be regarded as more appropriate for the DC network. In the bipolar DC network, the voltage levels used are 750, 0, and -750 VDC. In order to produce such voltage levels, either a twelve-pulse rectifier or a multilevel converter must be used to convert the three- phase AC voltage to three DC voltage levels. When a 12-pulse rectifier is used, it requires a transformer that has two secondary windings supplying the two rectifier bridges. The type of the power electronics switches in the rectifier is defined by the functional require- ments. The main functional requirement that determines the type is the direction of power flow. In case there is only a need to deliver the power from the MV to LV network, diodes or thyristors can be used. If there is also a need to transfer power from the LV to MV network, a typical choice is to use IGB transistors. To distinguish these two solutions, the former is referred to as rectifiers and the latter PWM rectifiers. Whereas rectifiers can be either uncontrolled or controlled, the PWM rectifiers are always controlled. The property that is achieved by making the rectifier controlled (e.g. a half-controlled thyristor-bridge) is the smooth start-up and the option to adjust the DC network voltage. With PWM rec- tifiers, including bidirectional power flow, many additional properties are achieved. They include active DC voltage control, unity power factor and reactive power control on the grid side, and nearly sinusoidal voltages and currents. The structure of the PWM recti- fier could be a 12-pulse rectifier using IGB transistors or a multilevel converter, such as a neutral-point clamped (NPC) converter. The use of a multilevel converter becomes an attractive solution when a PWM rectifier is needed, since it requires only a normal trans- former and low inductance on the grid side in order to achieve sinusoidal waveforms.
A minor drawback of multilevel converters is the additional complexity in their control methods.
Because the bipolar DC link structure was found to be a more appropriate solution, the supply voltage of inverters connected to the system is 750 VDC. The structure of the
N e t w o r kM V
T r a n s f o r m e r R e c t i f i e r
A C D C
1 5 0 0 V D C
0
A C D C
D C A C
C u s t o m e r i n v e r t e r
C u s t o m e r i n v e r t e r s
L P F
L P F
(a) Unipolar DC link structure
N e t w o r kM V
T r a n s f o r m e r R e c t i f i e r
A C D C A C
D C
YD
7 5 0 V D C
- 7 5 0 V D C
0
A C D C
A C D C
D C A C
D C A C
C u s t o m e r i n v e r t e r s
C u s t o m e r i n v e r t e r s
L P F L P F
L P F L P F
(b) Bipolar DC link structure
Figure 1.6: LVDC network with different DC link structures.
customer-end converter depends on whether a three- or single-phase supply is required.
Unlike in the rectifier case, controllable power electronics switches are always used at the customer-end. The reason for this is that the output voltage of the converter needs to be controlled. Since the average apparent power that is needed by a typical customer is about 10 kVA, IGB transistors can be considered suitable switches to be used. A typical switch- ing frequency that is applied when IGBTs are used is around 3–10 kHz. Fast switching of switches results in high-frequency harmonics. These harmonics are typically filtered out by using a low-pass filter (LPF) as illustrated in Fig. 1.6. As the filtering and control are needed to achieve nearly sinusoidal output voltage and current, they are also needed to comply with the standards. Compliance with the standards, when the phase voltage of 230 V (RMS), 50 Hz is supplied to the customer from the bipolar DC link, is also the objective of this thesis. Various voltage quality requirements given by standards are de- scribed in the next section. It is obvious that the requirements could be achieved in many ways; however, in an application that is used in the power distribution system, the cost, power quality, energy efficiency, and reliability are the main themes that dictate what is the best solution. The solutions of the customer-end converter and filtering are discussed in detail in the next chapter.
1.3 Voltage quality of distribution systems
1.3.1 EN 50160
The requirements for the quality of the phase voltage supplied by the distribution networks is determined byEN50160 (1994) standard in the European countries. In the standard, the requirements are given for the frequency of the phase voltage, voltage level, and limits for voltage variations. There are limits for fast transients, flickering, harmonic voltages, and transient overvoltages. The parts of the standard that are considered to be the most relevant from the perspective of the thesis are introduced in the following.
The frequency of the phase voltage is 50 Hz. In normal operation the average of the 10 s sample must be
• 50 Hz±1 % for 99.5 % of the year and
• 50 Hz±4 % for 100 % of the time.
The voltage level requirement is given as RMS voltage for the phase voltage and is Un =230 V. According to the standard, the variation of the voltage is allowed, but there are limitations on how much the voltage may vary. In normal operation, excluding faults and outages
• 95 % of the 10-minute averages of the RMS phase voltage must be inside∆Un =± 10 % and
• all of the 10-minute RMS phase voltage averages must remain between∆Un =+10 /-15 %.
For fast voltage transients that occur when loads are connected or other switchings are made in the network, the standard declares that
• the highest variation from the nominal value (230 V(RMS)) is 10 % for a few times per day.
• In normal operation the voltage should not vary more than 5 % from the nominal.
Furthermore, the standard defines the limit for flickering. It states that
• the long-term flickering index Plt should not exceed the value 1 for 95 % of the time.
Finally, the limits for the level of harmonic voltages are given in the standard. In normal operation, during every week
• 95 % of the 10-minute averages of every harmonic voltage RMS value must be less than given in Table1.1.
1.3.2 Other voltage quality standards
There are also other standards aiming at providing guidelines for the acceptable quality of voltage. These standards are not directly applied to building or designing of low- voltage AC distribution networks in Europe; however, they include valuable information that could be used as a basis for the system analysis. For example, they describe the requirements for the level of acceptable emissions and required immunity to disturbances for devices that are connected to the network.
Table 1.1: Acceptable harmonic voltage levels at the supply terminals in percents of the nominal phase voltageUnaccording toEN50160(1994)*.
Odd harmonics
Not multiple of 3 Multiple of 3 Even harmonics
Orderh Amplitude (%) Orderh Amplitude (%) Orderh Amplitude (%)
5 6 3 5 2 2
7 5 9 1.5 4 1
11 3.5 15 0.3 6–24 0.5
13 3 21 0.2
17 2
19 1.5
23 1.5
25 1.5
*The THD of the supply voltage including all harmonics up to the order of 40 should not exceed 8 %.
IEC 61000
IEC61000 (1992) is a series of standards that at a general level describe the electro- magnetic interference (EMI) testing procedures and requirements for electrical equip- ment connected to the network. The part of the series that could be considered relevant also from the distribution network point-of-view include at least IEC 61000-2-2 and IEC 61000-3-x. IEC 61000-2-2 defines the compatibility levels for low-frequency-conducted disturbances and signaling in a public low-voltage power supply systems (both single- and three-phase) with the nominal phase voltage up to 240 V (RMS). IEC 61000-3-x defines emission limits for electrical equipment connected to the grid. It sets limits for different types of disturbances. More specifically, 61000-3-2 includes limits for the har- monic current emissions whereas 61000-3-3 defines the limitations for voltage changes, voltage fluctuations, and flicker in public low-voltage supply systems. These standards are for all electrical equipment the nominal phase current of which is less than or equal to 16 A. For electrical equipment of a larger phase current than 16 A, the corresponding requirements are given in standards 61000-3-4 and 61000-3-5. The limitations for differ- ent disturbances deviate somewhat from those given in EN 50160. However, they have the same THD value (8 percent) when all the harmonics are included up to the order of 40. Furthermore, 61000-3-3 gives a value ofPst ≤1 for the short-term flicker-index and Plt ≤0.65 for the long-term index.
IEEE 519-1992
The standard defines the limits for harmonic current injection from end-users so that the harmonic voltage levels of the overall power system will be acceptable. The limits are given for the maximum individual harmonic components and for the total harmonic dis- tortion (THD). For systems the nominal voltage of which is below 69 kV, the THD recom- mendation is less than 5 %. The limit of 3 percent is given for each individual voltage har- monic. The standard also defines limit values for harmonic currents. An acceptable level is determined according to the voltage level and short-circuit ratio (i.e. Rsc = Isc/IL), whereIscis the short-circuit current andILis the maximum demand current in the point- of-common coupling (PCC). Short-circuit ratio describes the level of possible interaction between the load-injected harmonics and the supply system impedance at PCC. The limit values for individual current harmonics for a system the voltage of which is less than or equal to 69 kV according toIEEE519-1992(1993) are given in Table1.2.
1.3.3 Power acceptability curve
In addition to the EN and IEC standards, the voltage quality measure is given as the power acceptability curve. This curve encapsulates the essence of all requirements for the phase voltage of the distribution system. These include the requirements for transients, long- and short-term (e.g. voltage sags and overvoltages), as well as for steady state. The curve depicts the severity of the voltage variation∆U with respect to its duration. It is applicable both to the RMS voltage and the instantaneous amplitude of the phase volt- age. The curve was formerly known as the CBEMA curve according to the Computer
Table 1.2: Harmonic current distortion limits in percents of the maximum demand currentILaccording toIEEE519-1992(1993)*.
Un≤69 kV
Isc/IL h< 11 11≤h<17 17≤h<23 23≤h<35 35≤h
<20 4.0 2.0 1.5 0.6 0.3
20-50 7.0 3.5 2.5 1.0 0.5
50-100 10.0 4.5 4.0 1.5 0.7
100-1000 12.0 5.5 5.0 2.0 1.0
>1000 15.0 7.0 6.0 2.5 1.4
Business Equipment Manufacturing Association. However, the CBEMA curve has been replaced by the ITIC curve, according to the Information Technology Industry Council.
On behalf of the ITI council, the acceptable region has been extended when compared with the original CBEMA curve mainly for practical reasons. The ITIC curve, which is especially developed for 120 V computer equipment, is illustrated in Fig. 1.7. Although developed for 120 V computer equipment, it has also been applied to general power qual- ity evaluation (Dugan et al.,2003). Despite the broad application,Kyei et al.(2002) have
Figure 1.7: ITIC power acceptability curve for 120 V computer equipment (ITIC, 2009)
.
criticized the applicability of the curve; in particular, when different types of loads are used and also when three-phase system is used instead of single-phase system. However, it is shown that the power acceptability curve can be derived for different types of loads and systems using the CBEMA curve as the basis.
1.3.4 Discussion
In the EN 50160 and also in the IEC 61000-2-2 standard, the recommended voltage THD value is 8 percent when all harmonics are included up to the order of 40. The same value in IEEE 519-1992 equals 5 percent. With power electronic devices, the switching frequency can easily be chosen to a frequency that is noticeably higher than the 40th harmonic. In practice it means that the THD value calculated according to the standards above gives optimistic results when power electronic devices are concerned. Therefore, in this thesis, in addition to the method described in the standards, the THD value is calculated so that the first switching harmonic group is included. The value of 5 % is chosen as the limit for this THD value. Secondly, for example IEC 61000-3-6 defines certain planning levels for compatibility and harmonics. From that perspective, it is reasonable to choose somewhat stringent requirements compared with the recommended levels given in the standards. It is also expected in the future that the existing requirements will be replaced with stricter ones.
A peculiar difference between EN50160 and IEEE519-1992 is that the EN standard, which is based on the IEC standard, defines requirements for the phase voltage at the point where a customer connects to the network. On the contrary, in IEEE519-1992 the given requirements must be met in the point-of-common coupling (PCC). The PCC is the point where in the supply network branch the first customer is connected. By adopting this intepretation, the PCC in the LVDC network is located at the DC network. Furthermore, the standard does not define or limit the amount of distortion at the customer network as long as the harmonic level at the PCC meets the requirements. Obviously, the standard is not suitable to be used to design LVDC distribution networks from the customer point-of- view.
The requirements stated in EN 50160 can mostly be affected with the voltage control.
This includes the voltage level requirement and also the requirements for fast transients and flickering. Harmonic voltages that appear in power electronic equipment mainly con- sist of switching harmonics, and to cope with them, passive filtering is used. In the case where there is an outage in the DC network, the control system can maintain the voltage level only up to a certain level. In order to improve the power quality of the system com- pared with a current distribution system, which is one of the targets of the development, these DC network outages should not affect the end-customer voltages. In practice, it means that the DC network should include power production to provide energy while the MV network fault is cleared. To sum up, the focus of the thesis is to develop a control system that provides the good-quality voltage for the customers. The control system’s performance is evaluated in the normal operation of the DC network and it should meet the requirements for voltage level and fast transients.
1.4 Survey of voltage control methods
In this section, a survey of previously documented voltage control methods is made. The main focus is on the recent development, but also the most typical solutions are described.
The survey is limited to the single-phase inverters. The single-phase inverter solutions cover a broad area of applications. However, the uninterruptible power supply (UPS) can probably be regarded as the most relevant application from the perspective of the LVDC network customer-end inverter. The control of UPS devices has similar voltage quality requirements as an LVDC customer-end inverter, in other words, to provide a high- quality, constant-frequency, and constant-amplitude voltage to the customers. Although the section primarily aims at revealing the fundamental principles behind the different control methods, it also serves as a basis for development work.
A survey on control methods and modeling of single-phase UPS inverters is provided byDeng et al. (2005). In the survey, the control methods are divided into three subcat- egories, namely, the model-based instantaneous feedback control, feedforward learning control, and nonlinear control. These subcategories are further divided into categories that describe the control method more precisely. The survey by Deng et al., being in the heart of the single-phase inverter output voltage control, is partially repeated here for the sake of completeness and to serve the reader. Although being quite comprehensive, the survey lacks some methods that have also been used to control UPS. The scheme illustrat- ing the control solutions available is shown in Fig.1.8. The scheme is far from complete, but provides a glimpse of a broad field of available control methods, most of which have already been applied to the UPS output voltage control.
Probably the most common control scheme is the multi-loop scheme (Ryan et al., 1997; Loh et al., 2003; Rech et al., 2003; Deng et al., 2004, 2008b; Kukrer et al., 2003;
Jung et al.,1997;Naser and Quaicoe,1996;Buso et al.,2001). A typical multi-loop scheme is a cascaded scheme where different types of controllers can be applied. These controllers include at least different variations of proprotional (P), integration (I), and derivation (D) controllers and a deadbeat controller (Kukrer and Komurcugil, 1999; Buso et al., 2001;
Mattavelli,2005). In a typical UPS system presented in the literature, an LC filter is used (Grundling et al., 1997; Saritha and Jankiraman, 2006; Deng et al., 2004, 2007, 2008b;
V o l t a g e c o n t r o l m e t h o d s f o r s i n g l e - p h a s e i n v e r t e r s
M o d e l - b a s e d i n s t a n t a n e o u s
f e e d b a c k c o n t r o l F e e d f o r w a r d l e a r n i n g
c o n t r o l N o n l i n e a r c o n t r o l R o b u s t c o n t r o l
M u l t i l o o p - C o n t r o l S c h e m e s D e a d b e a t c o n t r o l l e r s
R e p e t i t i v e C o n t r o l S c h e m e s I t e r a t i v e L e a r n i n g C o n t r o l S c h e m e s
M o d e l R e f e r e n c e A d a p t i v e C o n t r o l S c h e m e s
N e u r a l N e t w o r k C o n t r o l S c h e m e s
H h C o n t r o l S c h e m e H 2/H h M i x e d S e n s i t i v i t y C o n t r o l S c h e m e s S l i d i n g - M o d e C o n t r o l l e r s
Q F T C o n t r o l S c h e m e s
Figure 1.8: Control methods used to control the output voltage of a single-phase in- verter .
Kukrer and Komurcugil, 1999; Kukrer et al., 2003). Similarly, the typical measurement arrangement of such a UPS system includes either the measured inductor or capacitor cur- rent and the measured output voltage. The outer loop of the control system is typically the voltage loop, and the inner loop is the current loop. The block diagram illustrating such a multiloop structure is shown in Fig. 1.9(a). For stability reasons, the inner loop band- width is typically 4–10 times as high as that of the outer loop. Tuning of the controllers can be carried out for example by pole placement or a robust control design method such as the quantitative feedback theory (QFT) developed byHorowitz (1963). The benefits of the cascade scheme include the ease of implementation along with the ease of imple- menting the current limits, and when properly tuned, a good transient performance can be achieved. The drawbacks are possible additional measurements needed for decoupling the variables, sensitivity to measurement noise, and possible parameter variations.
Another type of the instantaneous control scheme includes the observer-based solu- tions. In Fig. 1.9 (b), a basic scheme of an observer-based state-feedback control is shown. A state-feedback-based UPS control scheme is proposed by Komurcugil et al.
(2006). The control is tuned according to the Linear Quadratic Regulator (LQR) prin- ciple. However, the implementation of the control system was carried out in continuous time. In the LQR-based control, the state-feedback coefficients are determined according to the cost function, which is minimized when specific costs for states are determined.
The costs determine the feedback gain values and thereby also the behavior of the control system. The paper reports a good steady-state performance when a resistive load was used. Another observer-based control scheme operating in the dq frame is proposed by Saritha and Jankiraman(2006). The introduction of a dq frame improves the performance of the PI controllers, whereas there is a probable amplitude and phase error while using sinusoidal references. Also the decoupling of variables becomes easier in the dq frame.
Although the observer could also be used as an estimator in order to minimize the number of measurements, this option was not used in the paper.
Because the output voltage is cyclically varying, a learning control method such as repetitive control and iterative learning control (ILC) methods are found to be attractive
ur e f(t) V o l t a g e c o n t r o l
C u r r e n t c o n t r o l
C
i1Modulator uc
V o l t a g e
c o n t r o l l e r C u r r e n t
c o n t r o l l e r
LrL
ud c
(a) Multi-loop (cascade) control scheme.
ur e f(t) C
Modulator uc
V o l t a g e c o n t r o l l e r
LrL
ud c
x1 , e s t
O b s e r v e r x2 , e s t
S t a t e - f e e d b a c k c o e f f i c i e n t s
(b) Observer-based state-feedback con- trol.
Figure 1.9: Structure of the control system when applied to a single-phase inverter.
solutions. The structure of a learning control is illustrated in Fig. 1.10 (a). The same structure is valid also when a neural-network-based control is considered, by only chang- ing the type of the learning controller. Application of a repetitive control to a single-phase UPS application can be found at least inRech et al. (1997) and Rech et al.(2003). The fundamental principle behind the repetitive control is that the error between the reference and the output of the system is stored for one period and then fed back into the system during the next cycle. The use of the repetitive control guarantees excellent steady-state behavior because it eventually corrects the cyclic errors in the waveform. However, the drawback of such a controller is that it has poor dynamic properties. A typical solution to compensate the dynamics of the repetitive controller is to use a feedforward controller that has fast dynamical properties (Rech et al., 1997,2003;Deng et al.,2005). InRech et al.
(2003), different feedforward controllers are applied, namely a predictive PD controller, a predictive one sample ahead preview (OSAP) controller, and a model reference adaptive controller (MRAC). Another learning control method is the ILC method. The ILC method has been applied to a UPS application byDeng et al.(2007). The fundamental difference between the repetitive control and the ILC is that in the ILC, it is supposed that the system returns to the same initial condition after every cycle. The benefit obtained by using the ILC method is that it requires only one sensor. The main drawback with the ILC is that it is not possible to achieve a subcycle response. For this purpose, as with the repetitive control, different feedforward control methods could be applied.
Instantaneous learning or adaptation to the environment is achieved when adaptive control methods are used. The highest level of adaptation is achieved with control algo- rithms that are self-organizing by nature. One such adaptive control method is a neural network. A neural-network-based control algorithm for UPS is introduced byDeng et al.
(2008a) and by Sun et al. (2002). Although being a highly sophisticated control algo- rithm, it is rarely used in practice because of the complexity in design and implementa- tion. However, the method that avoids the use of nonlinear functions is introduced by Deng et al.(2008a). The proposed neural network method is based on a B-spline network
C
Modulator uc LrL
ud c
ur e f
L e a r n i n g C o n t r o l l e r F e e d f o r w a r d
C o n t r o l l e r
(a) Learning control scheme is appro- priate e.g. for repetitive, ILC, and neural network controls.
C
Modulator uc LrL
ud c
ur e f
C o n t r o l l e r A d a p t a t i o n a l g o r i t h m R e f e r e n c e m o d e l
eu
ur m
uc
uc
ua d a p t
(b) Model reference adaptive control (MRAC).
Figure 1.10: Structure of the control system when applied to a single-phase inverter.
(BSN) technique. With B-splines, any continuous function can be approximated to a de- sired degree of accuracy as long as the network used is large enough (Deng et al.,2008a).
It is also claimed that the design and implementation of the proposed control is easier as a result of the absence of nonlinear functions. In the paper, excellent performance is reported when only a single measurement is used. Yet another widely applied adaptive control algorithm is the model reference adaptive control method. The fundamental prin- ciple behind the MRAC method is that the control system includes the reference model.
By introducing an error signal that describes the difference between the reference model output and the output of the plant, the input signal of the plant is varied in order to achieve similar plant behavior as that of the reference model. This task is carried out by the adapta- tion algorithm. The most widely known algorithms include the gradient and least-squares algorithms and their modified versions. A typical structure of the MRAC is shown in Fig. 1.10 (b). An MRAC scheme that is used to control the UPS inverter is presented at least by Grundling et al. (1997) and by Carati et al. (2000). In addition to the MRA controller, the control system includes a repetitive controller used to improve the steady- state performance. With the MRA controller, the problems of unmodeled dynamics and parameter variation can be avoided. The drawback of the MRAC method is that it has to be made robust or the control system may become unstable even as a result of very minor disturbances. The problem is also known as drifting of parameters. However, to avoid stability problems, there are various robust adaptive control methods available (Tao, 2003;Ioannou and Sun,1995;Ioannou and Fidan,2007;Landau et al.,1998).
Over the past few years, the sliding-mode control (SMC) has proven to be very pop- ular mainly because of the fast transient response. In some cases, the dynamics of the controlled plant can be simplified by applying the SMC. Also the structure of the control system can be made very simple (Fig. 1.11(a)). Therefore, it could be expected that the SMC has also been applied to control a UPS inverter (Tai and Chen,2002;Kukrer et al., 2006). In the SMC, the switching commands of the inverter are determined by the switch- ing function. It is based on a variable structure control (VSC), which means that under
C
uc LrL
ud c
ur e f
C o n t r o l l e r
(a) Simplified scheme of the SMC.
C
Modulator uc LrL
ud c
ur e f H I n f .
C o n t r o l l e r ir e f
i1
(b) Configuration for theH∞control
Figure 1.11: Control systems’ principle structure when applied to single-phase in- verter.
different structures the control of the system remains constant. The drawback of the SMC is that it usually requires a quite high bandwidth to implement it. However, also discrete-time SMC has been proposed with fixed sampling frequency. Another well- known problem with SMC is the chattering problem and it must be taken care of. In the inverter applications, chattering will cause serious voltage harmonics, which are undesir- able.Tai and Chen(2002) propose a discrete-time SMC scheme that employs a dual-loop structure, where a current predictor (outer loop) utilizing the tracking error of the out- put voltage is devised to estimate the desired inductor current, while a current controller (the inner loop) is adopted to regulate the inductor current and to generate the desired command to the inverter.
Finally, the robustness of the inverter control, especially when power distribution is considered, is a very important property. Robustness guarantees that the system remains stable under variation of parameters, uncertainties and unmodeled dynamics. However, specific robust control methods such asH2andH∞are still typically considered to be too complicated to be used in practical applications. A design of anH∞ loop-shaping con- troller (Fig.1.11(b)) to a UPS application has been carried out byLee et al.(2001). Two control schemes are proposed that use only capacitor voltage measurement. In construct- ing the controller, the Linear Matrix Inequality (LMI) method is used. It is noted that the performance of the control can reach the performance of a properly tuned multiple loop feedback control.
Although various voltage control methods have already been presented in the literature and the field seems to be extensively studied, there are some differences between UPS and LVDC inverters, which motivates the present research. More specifically, the difference in the supply voltage between the UPS and the LVDC customer-end inverter is of interest.
In the UPS system, the DC voltage is typically supplied from a battery storage, whereas in the LVDC system the inverter is supplied from the DC network, which is connected to the grid. In this DC network, various other customers may be connected to the grid and may have an effect on the quality of supply voltage. Another fact that motivates the study of the control system design for an LVDC inverter is that the presented methods are typically used with devices the output power of which is around hundreds of watts. In practice, this means that the type of the power electronic switches is different, which allows the use of a higher switching frequency. A higher switching frequency again allows the use of a higher sampling frequency in the control system, and therefore the response of the system is faster than with a slower sampling frequency.
1.5 Scope of the thesis
The main purpose of the study is to find a suitable control method to control the output voltage of the customer-end system in order to meet the requirements set for the customer power quality. The control system will be primarily investigated and tested with a single- phase system. However, one of the targets is to make the control system universal in the sense that with minor changes it could be used with different types of converters including three-phase solutions. Additionally, to obtain the best possible solution for the
customer-end system, the subtasks include finding of the passive filtering solution and harmonic performance evaluation of different modulation methods. In the language of power distribution, the best possible passive filtering solution can be interpreted as being the one that produces the minimum costs while possessing all the desired properties and meeting all the requirements. The evaluation of the modulation methods is also based on the target to achieve a customer-end system that is cost effective, energy efficient, and reliable and produces a good power quality.
There are a few exceptions where the customer-end system can be regarded as not being responsible for the output power quality. These exceptions include the system op- eration where either the supply AC grid or the DC network experiences a fault (long- or short-term). In the case of an outage, the customer-end system can be regarded as respon- sible for supplying the standard quality voltage as long as there is enough energy in the DC network. After the voltage has gone below the minimum voltage limit, the supply is turned off.
In the normal no-load steady-state operation, the DC network voltage at the customer system terminals is 750 VDC. In the transient operation, the DC voltage varies according to stiffness of the supply grid. The DC voltage level is also dependent on the other cus- tomers connected to the network and on their instantaneous power demand. To analyze the effect of the DC supply network on the customer-end performance, a bipolar network is modelled. Based on the model, the load-dependent voltage drop in the network can be calculated. The purpose of the modeling is also to find out the worst-case performance from the supply point-of-view. Furthermore, the effect of an additional capacitance con- nected to the network is studied. The study provides a formulation by which one can determine the additional capacitance that is needed to obtain the desired DC voltage rip- ple at the input DC terminals of the customer-end system.
1.6 Scientific contributions
The main scientific contributions of the thesis can be listed as:
• Analysis of the customer-end passive filtering solutions. The study is based on finding out the minimum lifetime costs by producing a filtering solution that also meets the technical requirements set for a customer-end system(Publication I).
• Modeling and analysis of a bipolar low-voltage DC distribution system by applying an average model. The model can be used for more thorough stability studies and also to identify the supply parameters when the V/kW-ratio is known.
• Introduction of a gain scheduling control within the LQG control scheme. The motion of the poles as a function of load is utilized in order to reassign the integrator or PI controller gains to improve the overall performance.
• Modeling, analysis, and implementation of a PI-based cascade- and observer- based (LQG) output voltage control. The study discusses the advantages and disad- vantages of the choices that designer has to make when designing an output voltage
control for the customer-end system in the LVDC distribution network. One benefit of the proposed control schemes is that they can be easily adopted to the three-phase systems.(Publications II, III and, IV).
• The study of capacitor dimensioning carried out byKorshun(2009) is further ana- lyzed in the case when an additional capacitance is connected only to the customer- end. A special attention is focused on the current ripple attenuation property of DC networks .
In the thesis, also the cost-efficiency and reliability of basic customer-end system solu- tions for an LVDC distribution network are considered. In the study, the applicability of some of the most commonly used converter topologies to an LVDC system is considered.
The thesis is based on the results that are published in publications I, II, III, and IV.
The research work reported in the publications has been carried out mostly by the author.
In the measurements, the author received invaluable help from the co-author Nuutinen.
Preparation of the manuscripts was performed by the author and the co-authors.
Customer-end system
In this chapter, customer-end system solutions are considered. The primary target in the process of choosing the customer-end converter and filter and in developing of the control algorithm is to obtain economically the best possible solution. This goal must be achieved by not making any compromise in the performance of the system. That is, in addition to being economically feasible, the system must be energy efficient and reliable and meet the requirements set for power quality. Here, different structures of the customer-end system are introduced.
2.1 Customer-end system
The total customer-end system consists of the DC supply network, a DC/AC converter, a filter, a load, and a control system as shown in the block diagram of Fig. 2.1. The customer-end converter may also be a DC/DC converter. Especially, if the DC distribu- tion becomes widely applied, many of the electronic loads will change over to DC and then be supplied from a DC/DC converter. However, in this thesis it is considered that customers still have an AC network and all the loads use AC voltage. The customers’
AC network can be either a three- or single-phase network. Since various devices used
D C / A C
c o n v e r t e r F i l t e r C o n t r o l
s y s t e m D C
n e t w o r k u d c L o a d
id c ii n v
u i n v u oio
F e e d b a c k s i g n a l s s w
Figure 2.1: Customer-end system. In the figure, also the interacting voltages and cur- rents are presented. From the system, the measured signals are fed back to the control system, which produces the switching commands (sw) to the DC/AC converter.
