Author: Ander GONZÁLEZ FERNÁNDEZ DE BOBADILLA A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in Engineering Science and Technology

(1)

Integration of photovoltaic sources and battery based storage systems – A DC analysis and distributed maximum power point tracking solution

Author: Ander GONZÁLEZ FERNÁNDEZ DE BOBADILLA A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in Engineering Science and Technology

“Docteur en Sciences de l’Ingénieur et Technologie”

Academic year 2018-2019

Supervisor: Professor Johan GYSELINCK Electrical energy group Bio, Electro And Mechanical Systems Department

Thesis jury:

Prof. Dr. Ir. Michel KINNAERT (Université libre de Bruxelles, Chair) Prof. Dr. Ir. Pierre HENNEAUX (Université libre de Bruxelles, Secretary) Prof. Dr. Ir. Omar HEGAZY (Vrije Universiteit Brussel)

Prof. Dr. Ir. Olivier DE BLECKER (Université de Mons) Prof. Dr. Ir. Lieven VANDEVELDE (Universiteit Gent)

Prof. Dr. Ir. Wilmar MARTÍNEZ (Katholieke Universiteit Leuven)

Prof. Dr. Ir. Giovanni PETRONE (Università degli Studi di Salerno)

(2)

(3)

“It is better, of course, to know useless things than to know nothing.”

Seneca the Younger

(4)

(5)

UNIVERSITÉ LIBRE DE BRUXELLES

Abstract

École polytechnique de Bruxelles

Bio, Electro And Mechanical Systems Department Doctor of Philosophy

Integration of photovoltaic sources and battery based storage systems – A DC analysis and distributed maximum power point tracking solution

by Ander G

ONZÁLEZ

F

ERNÁNDEZ DE

B

OBADILLA

In this thesis the integration of photovoltaic (PV) generation and energy storage into the electrical grid is discussed. Although the studied system is for grid tied ap- plications, here the integration of the PV generation and the energy storage system (ESS) on the direct current (DC)-side of the system is addressed. The work contained in this thesis focuses on the integration of the DC-working parts before interfacing them with the grid through the use of an inverter and seeks an increasing in the energy that the system can deliver.

First, a study of classical systems that present well-differentiated parts is pre- sented: PV generation, a lithium-ion battery based ESS, the utility grid and a residen- tial electricity consumer. PV installations of 3 and 10kWp are considered together with storage capacities ranging from 1 to 9kWh. This yields interesting insights on how the system works based on the timing of the generation and consumption of energy. The results are used to highlight the weaknesses of the selected converter arrangement for the interfacing of the PV source and the ESS. Results show that the system is rather stiff and lacks from conversion efficiency when it needs to work in a wide range of powers, mainly due to low consumer power demand during battery discharge. In this first part of the thesis, three solutions to workaround the efficiency problem are proposed: reducing the difference between the ESS and the DC-bus voltages, using isolated converters to interface the ESS, or adopting a new arrange- ment of the parts of the system. One of the first two proposed solutions should be adopted if the same system topology is to be kept. These two solutions address the efficiency problem when the ESS is involved in the energy conversion. The third solution is proposed as alternative to the classical systems that use a DC-bus to ex- change power with the different parts of the system.

The new proposed arrangement features a distributed maximum power point tracking (DMPPT) type system that includes storage at module level. DMPPT sys- tems are able to track the maximum power point tracking (MPPT) of each panel separately by connecting a small power electronic converter (PEC) to each PV panel.

They are specially useful when the PV installation receives uneven irradiance, i.e.

shadows are present in some of the panels, increasing the annual yield of PV energy

from 7 to 30% as reported in the literature. Unfortunately, this kind of systems can-

not always handle high irradiance mismatches, and fail to track the maximum power

point (MPP) throughout the whole installation in some cases. Including batteries at

module level instead of connecting them to the DC-bus, allows for increasing the

MPPT range of the system, virtually to any severity of irradiance mismatch (de-

pending on the state of charge (SoC) of the battery pack), as well as adding storage

capability to the system.

(6)

The novel proposed system is able to workaround the problems of using non- isolated converters, achieving PV energy conversion efficiencies from 86% (for at least 10% of the peak power) to 90% and storage charge/discharge efficiencies rang- ing from 86% to 95%. Besides, it brings the opportunity to exploit the synergies of having storage at module level in systems that combine renewable energies and storage. Moreover, DMPPT systems achieve superior PV generation under partially shaded conditions when compared to classical PV arrays increasing the PV genera- tion when compared to classical or centralized PV installations up to 45% in power as reported in the literature.

In the second part of the thesis, the proposed novel DMPPT topology is pre- sented. The whole system is fully designed from scratch, including PECs, sizing of the different parts of the modules, embedded control loops of the modules and su- pervisory control of the whole system. Finally, the results obtained from running the proposed system are shown and discussed, and suggestions given on how to operate and protect the system. Experimental results are obtained using a 1.5kWp PV power and 1.5kWh capacity test bench built for that purpose.

The proposed system is able to generate PV energy, store the energy coming from PV generation and inject the generated and stored energy into the grid. The pro- posed system extends the MPPT capability of storage-less series-connected DMPPT systems. This is achieved by using the batteries not only to store energy when re- quired, but also to compensate the power mismatch across DMPPT modules of the same string when the output voltage of the modules becomes a limit. It also presents a modular and upgradable approach to PV systems including storage. This modu- larity also brings fault tolerance, and an ability to continue working after failure of one or more of the DMPPT modules by partially or completely isolating the faulty module (depending on the nature of the fault). Moreover, the addition of the DC-DC converters allows for the use of different PV panels in the system, i.e. from different manufacturers or technologies.

In conclusion, the presented system is very flexible, can be designed for a wide

range of power levels and energy storage sizes, and presents improved reliability

when compared to other series-connected DMPPT systems.

(7)

Acknowledgements

Firstly, I would like to express my gratitude to my family and friends for their sup- port and motivation during not only my PhD, but in different times throughout my life.

I would like to thank my supervisor Prof. Gyselinck for giving me the chance of working with the BEAMS department during my PhD and for the supervision of my work. Besides my supervisor I would like to thank the other members of my PhD committee Prof. Kinnaert and Prof. Garone for the periodic supervision of my work and insightful comments and questions about my work.

My sincere thanks also goes to Prof. Iu and Prof. Fernando, who hosted me in the Power And Clean Energy (PACE) research group during my stay in the University of Western Australia, and who gave access to their research facilities. Thanks also to the fellow researchers in the PACE research group who welcomed me and to those who I had the chance to collaborate with. Besides, I would like to take the chance to thank the Fédération Wallonie-Bruxelles for partly supporting my stay in the PACE research group under Concours des bourses de voyage program.

I would also like to thank my colleagues at BEAMS as well for their help and the useful discussions that we had during these years.

Finally I would like to thank other staff members from the BEAMS department.

To Ariane for her helpful attitude in both personal and work matters, to the tech-

nicians of the department that help me with the work I needed to do, especially

to Pascal who has always been fast and efficient in doing so and to Axel Dero and

Michel Osée for helping me with different tasks I had to do during this time.

(8)

(9)

List of Figures

2.1 Different parts of the system under study and possible power-flow

paths. . . . 5

2.2 Possible system topologies for grid-tied PV and storage systems. . . . 6

3.1 Considered PV and storage system topology. . . . 9

3.2 I-V (top) and P-V (bottom) curves for Suntech Power STP245S-20-Wd module. . . 10

3.3 Clear-sky irradiance profiles of a whole day for winter and summer cases. . . 11

3.4 One full day household (load) consumption profiles for a summer day (top) and a winter day (bottom). . . 12

3.5 Diagram of the local-grid simulator. Inputs in green, outputs in red. . . 14

3.6 Block diagram of the ESS subsystem of the local-grid simulator. . . 15

3.7 Energy management algorithm flowchart. . . 16

4.1 Basic boost converter schematics. . . 20

4.2 Bidirectional boost converter schematics, including the ESRs of pas- sive components. . . 20

4.3 Equivalent circuit of switches/diodes during conduction. . . 20

4.4 Bidirectional boost converter equivalent circuits including ESR in pas- sive components. . . 22

4.5 Half-bridge schematics and averaged model equivalent circuit. . . 27

4.6 State dependent half-bridge equivalent circuits. . . 28

4.7 Half-bridge switching detail: positive I

_L

, rising edge in u. . . 31

4.8 Half-bridge switching detail: positive I

_L

, falling edge in u. . . 32

4.9 Half-bridge switching detail: negative I

_L

, rising edge in u. . . 33

4.10 Half-bridge switching detail: negative I

_L

, falling edge in u. . . 34

4.11 Three-level bidirectional boost converter schematics. . . . 36

4.12 Dual active bridge converter schematics. . . 37

4.13 Simplified equivalent circuit of dual active bridge converter. . . 38

4.14 DAB waveforms for r

t

V

2

> V

₁

. . . 39

4.15 Graphical description of the current change. . . 40

4.16 Simple battery model. . . 42

4.17 Battery internal voltage, V

_bat

, in function of the SoC for a 150V lithium- ion battery pack. Source: Simscape Power Systems, Matlab Simulink. . 43

4.18 Single diode PV model. . . 44

5.1 PV generation results for 3kWp system using 2- and 3-level boost con- verters. . . 46

5.2 PV generation results for 10kWp system using 2- and 3-level boost converters. . . 46

5.3 Maximum achieved conversion efficiency in the PV generation results

using 2- and 3-level boost converters. . . . 47

(14)

5.4 Efficiency of the 2-level and 3-level boost converters in the ESS. . . 48 5.5 Round-trip efficiency data of the ESS from Tables B.5 and B.6. . . 49 5.6 Round-trip efficiency data of the ESS from Tables B.6 to B.9 for 3kWp

and 10kWp PV systems with different storage sizes. . . 49 6.1 Storage including series-connected DMPPT system topology. . . 55 6.2 Comparison of the currents for interleaved and non-interleaved 3-

port boost converter. On the right the currents (or a part of) appearing in the input port(s), on the left the currents (or a part of) appearing in the output port(s). . . 60 6.3 Interleaved three-port boost converter topology with interleaving branch

in red. . . 60 6.4 Main operation modes of the interleaved three-port boost converter. . 62 6.4 Main operation modes of the interleaved three-port boost converter.

Continued. . . 63 6.5 Battery pack voltage vs. SoC for constant-current discharge at differ-

ent rates from 1C to 0.2C. . . 65 6.6 Battery pack voltage vs. SoC for constant-power discharge at different

power rates. . . 65 6.7 Battery pack built using Samsung SDI ICR18650-26F lithium-ion recharge-

able cells. Cells tied using nickel strips in 13 groups in series of 2 cells in parallel each. Without battery management system (BMS). . . 67 6.8 Finished battery pack built using Samsung SDI ICR18650-26F lithium-

ion rechargeable cells. Battery pack convered in blue wrap including a BMS voltage and current monitoring system. . . 67 6.9 Non-interleaved three-port boost converter schematics. . . 70 6.10 Block diagram of the converter using transfer functions derived from

presented small-signal model . . . 73 6.11 External circuit proposed in order to avoid simultaneous switching of

S

₁

and S

₂

. . . 74 6.12 Up- and down-counting modulation (trailing- and leading-edge mod-

ulation) to avoid simultaneous ON states of S

₁

and S

₂

(and S

⁰₁

and S

⁰₂

). 75 6.13 Proposed symmetrical modulation to avoid simultaneous on states of

S

₁

and S

2

(and S

₁⁰

and S

₂⁰

). . . 76 6.14 First power converter prototype. . . 78 6.15 Finished power converter prototype. The power converter interfaces

the PV panel, the battery and the output that is connected to the DC-bus. 78 6.16 Switching of switch S

₁

. V

_PV

= 32V, V

_b

= 48V, V

_o

= 60V, I

_PV

= 5A,

I

_b

= 0A. . . 80 6.17 S

3

/D

_PV

switching cell and equivalent circuit considered for snubber

design. . . 81 6.18 Switching of switch S

₃

without snubber circuit. V

_PV

= 32V, V

_b

= 48V,

V

_o

= _60V, I

_PV

= _5A, I

_b

= 1A. . . 82

6.19 Switching of switch S

₃

with different RC snubber cir-

cuits. V

_PV

= 32V, V

_b

= 48V, V

_o

= 60V, I

_PV

= 5A, I

_b

= 1A. . . 83

6.19 Switching of switch S

₃

with different RC snubber cir-

cuits. V

_PV

= 32V, V

_b

= 48V, V

_o

= 60V, I

_PV

= 5A, I

_b

= 1A. Con-

tinued. . . 84

6.20 AMC1100 based isolated voltage sensor. . . 84

6.21 HLSR 10-P based isolated current sensor. . . 85

(15)

6.22 Measurement conditioning circuit placed between the sensor and the

ADC input. Differential amplifier (blue) and anti-aliasing filter (red). . 85

6.23 Proposed control scheme with cascaded PV voltage (C

_Vpv

) and cur- rent (C

_Ipv1

) loops and independent battery current control (C

_Ib

). . . 86

6.24 Bode plot of the PV current uncompensated loop (T

u

). . . 88

6.25 Frequency response of the open-loop (solid blue) and closed-loop (dashed orange) compensated PV current-loop. . . 88

6.26 Frequency response of the open- (solid blue) and closed-loop (dashed orange) compensated battery discharge current-loop. . . 90

7.1 Selected series connection of the DMPPT system topology. . . 94

7.2 PV installation in the roof of the building, organized in two groups of three PV panels each. . . 95

7.3 Test bench including the PECs of the DMPPT system, batteries and an independent LabVIEW based data-acquisition system. . . 95

7.4 Module output voltage thresholds for protection and compensation triggering. . . 96

7.5 DMPPT system topology including different controls and communi- cation links. . . 98

7.6 Initialization of supervisory control, version 1. . . 100

7.7 Typical IV (red) and PV (green) curves of a PV panel for a given irra- diance and temperature. Left to MPP is region 1 where dP

_PV

/dV

_PV

is positive and right to MPP is region 2 where dP

_PV

/dV

_PV

is negative. . . 101

7.8 Equalize output voltages routine. . . 102

7.9 Supervisory control version 1. . . 104

7.10 Supervisory control version 2. . . 106

7.11 Supervisory control version 3. . . 108

7.12 Supervisory control version 4. . . 110

7.13 Battery-based power compensation on all modules routine, version a. Introduced in version 4 of supervisory control. . . 111

7.14 Supervisory control version 5. . . 112

7.15 Battery compensation on all modules routine, version b. . . 113

7.16 Battery compensation on all modules routine, version c. . . 114

7.17 Flowchart of the selected P&O MPPT algorithm. . . 115

7.18 Flowchart of the sequence to perform MPPT algorithm in N

_mod

num- ber of modules. . . 116

7.19 Flowchart describing the DC-bus voltage adaptation algorithm. . . 118

7.20 Flowchart of the DMPPT system stop algorithm. . . 120

7.21 Flowchart of the sequence to recover a module that tripped. . . 122

8.1 Measured inductor currents and simulated inductor L

₁

current during PV current reference step change from 2A to 5.5A. The PV current is equally shared by the two inductors. Output port connected to a 33Ω load. . . 126

8.2 Measurements of PV current, battery current, output current and out-

put voltage during a battery current step change from 0A to -1.2A

(charge) while PV current is 5.2A. Output port connected to a 33Ω load.126

8.3 Measurements of PV current, output current and PV voltage during

PV voltage reference step change from 34V (zero current) to 24V (pre-

viously identified MPP) using a Benq GreenTriplex PM245P00. Out-

put voltage is held constant to 60V. . . 127

(16)

8.4 Simulation of the trailing- and leading- edge modulation during bat- tery charge. The currents of the first branch are represented in blue, second branch in red and the waveforms concerning both branches in black. . . 128 8.5 Simulation of the symmetrical modulation during battery charge. The

currents of the first branch are represented in blue, second branch in red and the waveforms concerning both branches in black. . . 129 8.6 Measurements of the current waveforms of the converter during bat-

tery discharge for I

_PV

= 4A, I

_b

= 1A with port voltages held to V

_PV

= 32V V

_b

= 48V and V

_o

= 60V. Output port supplies a pro- grammable load. The currents of the first branch are represented in blue, second branch in red and the waveforms concerning both branches in black. . . 131 8.7 Measurements of the current waveforms of the converter during bat-

tery charge for I

_PV

= 5.5A, I

_b

= − 1A with port voltages V

_PV

= 32V V

_b

= 48V and V

_o

= 61V. Output port supplies a 33Ω resistive load.

The currents of the first branch are represented in blue and the cur- rents of the second branch in red. . . 132 8.8 Efficiency curves of the interleaved three-port boost converter. Con-

stant battery discharge power curves. . . 133 8.9 Efficiency curves of the interleaved three-port boost converter. Con-

stant battery charge power curves. . . 134 8.10 Initialization (version 1) of the DMPPT system. Supervisory control

version 1. No voltage equalization. . . 136 8.11 Initialization (version 1) of the DMPPT system. Supervisory control

version 1. With voltage equalization. . . 137 8.12 Initialization (version 1) of the DMPPT system. Supervisory control

version 4. Under partial shading condition. . . 138 8.13 Initialization (version 2) of the DMPPT system and MPPT. Supervi-

sory control version 5c. Under partial shading condition. . . 139 8.14 Zoomed version of Figure 8.13 showing the initialization of the sys-

tem using version 2 of the initialization routine. . . 140 8.15 DMPPT system stop routine (Figure 7.20). . . 142 8.16 Experimental results showing the DC-bus voltage adaptation algo-

rithm. Supervisory control version 1. . . 143 8.17 Experimental results showing the recovery of all the modules after

trip. Tripping instants marked with ellipses. Supervisory control ver- sion 1. . . 144 8.18 Experimental results showing DMPPT module 2 tripping at t = t

₁

and

going back to normal operation after recovering at t = t

₂

. Supervisory control version 5a. . . 145 8.19 DMPPT system under homogeneous irradiance performing MPPT al-

gorithm at 1Hz with clouds. Supervisory control version 1. Case 1. . . 147 8.20 Powers and efficiencies of DMPPT system under homogeneous irra-

diance case 1. . . 148 8.21 DMPPT system under homogeneous irradiance performing MPPT al-

gorithm at 1Hz with clouds. Supervisory control version 1. Case 2. . . 149 8.22 Powers and efficiencies of DMPPT system under homogeneous irra-

diance case 2. . . 150 8.23 Diagram of the construction of the selected panel consisting of six PC

cell strings and three bypass diodes. . . 151

(17)

8.24 Considered types of PV partial shading. . . 152 8.25 Three PV panels from the experimental setup. Left to right: type A

shadow, clear panel and type B shadow. . . 152 8.26 Normal operation of the DMPPT system with 2 type A and 2 type B

partial shading applied. Supervisory control version 5c. . . 153 8.27 Powers and efficiencies of the case in Figure 8.26. . . 154 8.28 Normal operation of the DMPPT system with 5 type B partial shading

applied. Supervisory control version 5b. . . 156 8.29 Powers and efficiencies of the case in Figure 8.28. . . 157 8.30 Normal operation of the DMPPT system with 4 type B partial shading

applied. Supervisory control version 5b. . . 158 8.31 Powers and efficiencies of the case in Figure 8.30. . . 159 8.32 Normal operation of the DMPPT system with 1 type B partial shading

applied. Supervisory control version 5a. . . 160 8.33 Powers and efficiencies of the case in Figure 8.32. . . 161 8.34 Normal operation of the DMPPT system with 1 type A and 2 type

B partial shading applied. Supervisory control version 5c. Cloudy weather. . . 162 8.35 Powers and efficiencies of the case in Figure 8.34. . . 163 8.36 Normal operation of the DMPPT system with 2 type A partial shading

applied. Supervisory control version 4. . . 164 8.37 Powers and efficiencies of the case in Figure 8.36. . . 165 8.38 Initialization and MPPT of the DMPPT system with 2 type A partial

shading applied. System losses stability after MPPT start. . . 167 8.39 Estimation of the efficiency range of the proposed DMPPT system un-

der test cases presented in Part I. . . 169 A.1 Dimensions of an E shaped core. . . 183 A.2 Built inductor (on 1cm grid paper). . . 185 B.1 PV generation after conversion and efficiency of conversion for the

two- and three-level boost converters on a summer day for a 3kWp PV installation. . . 190 B.2 PV generation after conversion and efficiency of conversion for the

two- and three-level boost converters on a winter day for a 3kWp PV installation. . . . 191 B.3 PV generation after conversion and efficiency of conversion for the

two- and three-level boost converters on a summer day for a 10kWp PV installation. . . 191 B.4 PV generation after conversion and efficiency of conversion for the

two- and three-level boost converters on a winter day for a 10kWp PV installation. . . . 192 B.5 Simulation results for 3kWp PV generation system on a clear-sky sum-

mer day and 2kWh ESS using a 2-level boost converter. . . 194 B.6 Simulation results for 3kWp PV generation system on a clear-sky win-

ter day and 2kWh ESS using a 2-level boost converter. . . 195 B.7 Simulation results for 10kWp PV generation system on a clear-sky

summer day and 6kWh ESS using a 2-level boost converter. . . 196 B.8 Simulation results for 10kWp PV generation system on a clear-sky

winter day and 6kWh ESS using a 2-level boost converter. . . 197

(18)

B.9 One day simulation results for 3kWp PV generation system in sum-

mer and 2kWh ESS using a 3-level boost converter. . . 198

B.10 One day simulation results for 3kWp PV generation system in winter and 2kWh ESS using a 3-level boost converter. . . 199

B.11 One day simulation results for 10kWp PV generation system in sum- mer and 6kWh ESS using a 3-level boost converter. . . 200

B.12 One day simulation results for 10kWp PV generation system in winter and 6kWh ESS using a 3-level boost converter. . . 201

C.1 Top layer of the PCB (signal). . . 204

C.2 Middle layer 1 of the PCB (ground plane). . . 204

C.3 Middle layer 2 of the PCB (supplies). . . 205

C.4 Bottom layer of the PCB (signal and power). . . 205

E.1 Control structure diagram. Coordinate change corresponds to (E.15), virtual input calculation to (E.16) and (E.17), and u feedback control laws to (E.20) and (E.21) respectively. . . 213

E.2 Converter start-up operation. Supply is first given by the battery and after output voltage reference is reached supply is smoothly shifted to PV port. . . 217

E.3 Details on current shifting from battery to PV port during start-up of converter. . . 217

E.4 Converter start-up operation. Output supplied with PV power, a ramp is applied in order to keep PV current under MPP current of the panel. 218 E.5 Operation of the converter supplied by PV power during a sudden load change. . . 219

E.6 Operation of the converter during d

₂

, d

⁰₂

and d

₃

duty-cycle changes. . . 220

(19)

List of Tables

3.1 Main characteristics of the Suntech Power STP245S-20-Wd module. . . 10

3.2 Selected PV system characteristics. . . 11

4.1 Bidirectional boost converter states. . . 21

4.2 Differential equations of the boost converter describing its different states. . . 23

4.3 Equations of the half-bridge during its different states. . . 27

4.4 Conduction losses during each state in the half-bridge model. . . 29

4.5 Conduction losses for each state expressed as function of i

_L

and D in the half-bridge model. . . 30

4.6 States for dual active bridge (DAB) converter depending on the volt- ages across the transformer windings. . . 38

6.1 Characteristics of the Benq GreenTriplex PM245P00 polycrystalline 260Wp PV panel at standard test conditions (STC). . . 64

6.2 Characteristics of the NMC battery pack in each module (from datasheet). 64 6.3 Empirical characterization of the battery pack (from constant power discharge curves). . . 66

6.4 Specifications of the PEC contained in the DMPPT modules. . . 68

6.5 Selected operating points for the design of the different control loops. . 87

6.6 Working modes of the modules and controller signal arrangement. . . 91

7.1 Overvoltage protections in the embedded control of the DMPPT mod- ules. . . 95

7.2 Summary of supervisory control versions including their advantages and drawbacks. . . 123

8.1 Summary of all the presented DMPPT system experimental results. Type of shading (if any) in each module and supervisory control ver- sion included. . . 135

A.1 Inductor design nomenclature. . . 183

A.2 Ratings and number of components for two- and three-level boost topologies. . . 186

A.3 Sizing of the components for 2- and 3-level boost converters in PV interfacing application for 3kWp and 10kWp systems. . . 186

B.1 Converter performance on a summer day. 3kWp PV system. . . 189

B.2 Converter performance on a summer day. 10kWp PV system. . . 190

B.3 Converter performance on a winter day. 3kWp PV system. . . 190

B.4 Converter performance on a winter day. 10kWp PV system. . . 192

B.5 Performance of non-isolated converters in ESS during summer for

3kWp generation system and k = 0.7 (k is the amount of load con-

sumption that is supplied by the system in per unit). . . 193

(20)

B.6 Performance of non-isolated converters in ESS during summer for

3kWp generation system and k = 1. . . 193

B.7 Performance of non-isolated converters in ESS during winter for 3kWp generation system. . . 193

B.8 Performance of non-isolated converters in ESS during summer for 10kWp generation system. . . 193

B.9 Performance of non-isolated converters in ESS during winter for 10kWp generation system. . . 193

D.1 PV to output efficiency measurements. . . 207

D.2 PV and battery to output efficiency measurements. P

_b

= 50W. . . 208

D.3 PV and battery to output efficiency measurements. P

_b

= 100W. . . 208

D.4 PV and battery to output efficiency measurements. P

_b

= 150W. . . 208

D.5 PV and battery to output efficiency measurements. P

_b

= 200W. . . 209

D.6 PV and battery to output efficiency measurements. P

_b

= 250W. . . 209

D.7 PV to battery and output efficiency measurements. P

_b

= − 50W. . . 209

D.8 PV to battery and output efficiency measurements. P

_b

= − 100W. . . . 209

D.9 PV to battery and output efficiency measurements. P

_b

= − 150W. . . . 210

D.10 PV to battery and output efficiency measurements. P

_b

= − 200W. . . . 210

D.11 PV to battery and output efficiency measurements. P

_b

= − 250W. . . . 210

E.1 Simulation parameters . . . 216

F.1 Cost of the main power circuit . . . 222

F.2 Cost of the power supplies . . . 223

F.3 Cost of other components . . . 223

(21)

Acronyms

AC alternating current. 5, 9, 19, 91

ADC analog to digital converter. 125, 127

BMS battery management system. xiv, 6, 50, 64, 66, 171, 178 CAM classical averaged model. 19, 24, 25, 37

CCM continuous conduction mode. 19

DAB dual active bridge. xix, 19, 36, 37, 51, 57, 58, 171

DC direct current. v, vi, 1, 2, 5, 6, 7, 9, 12, 13, 19, 37, 40, 41, 42, 45, 47, 50, 55, 56, 57, 59, 61, 66, 77, 91, 93, 97, 99, 101, 103, 105, 107, 109, 116, 117, 119, 121, 123, 125, 133, 135, 141, 155, 166, 171, 172, 173, 174, 175, 177, 178, 179, 181, 185, 187, 189, 221

DCM discontinuous conduction mode. 19

DMPPT distributed maximum power point tracking. v, vi, xix, 1, 2, 51, 55, 56, 57, 58, 61, 66, 70, 76, 90, 91, 92, 93, 94, 97, 99, 101, 103, 107, 109, 117, 119, 121, 124, 125, 130, 133, 135, 141, 146, 155, 166, 168, 169, 171, 172, 173, 174, 175, 178, 179, 203, 211, 221

DoD depth of discharge. 6, 13, 48, 64, 192 DSP digital signal processor. 85

EMC electromagnetic compatibility. 77

EMI electromagnetic interference. 77, 178, 203 EoL end-of-life. 6

ESR equivalent series resistance. 20, 23, 25, 42, 72

ESS energy storage system. v, 1, 2, 5, 6, 7, 9, 12, 13, 14, 15, 36, 37, 45, 47, 48, 50, 51, 55, 57, 168, 171, 172, 175, 177, 181, 185, 192, 194

GAM generalized averaged model. 19, 37 HF high frequency. 1, 36, 37, 41, 50, 58, 171 HV high-voltage. 7, 9, 36, 47, 50, 181, 182, 186 LPF low-pass filter. 84

LV low voltage. 36, 47, 50, 181, 182

(22)

MIMO multiple-input multiple-output. 211, 213

MPP maximum power point. v, 1, 7, 10, 9, 10, 51, 56, 64, 68, 127, 141, 166, 173, 174, 181, 184, 185, 207, 212, 215

MPPT maximum power point tracking. v, vi, 2, 13, 45, 55, 56, 57, 89, 93, 97, 99, 101, 103, 107, 109, 116, 121, 123, 133, 141, 155, 166, 171, 173, 174, 175, 178, 179, 189 OST one shoot trigger. 119, 141

P&O perturb and observe. 10, 45, 107, 109, 189 PCB printed circuit board. 76, 79, 203

PCC point of common coupling. 5, 9, 12, 48

PEC power electronic converter. v, vi, xix, 1, 2, 5, 13, 14, 19, 26, 35, 45, 50, 58, 61, 66, 89, 172, 174, 181, 185

PSM phase shift modulation. 36, 37

PV photovoltaic. v, vi, 1, 2, 5, 7, 9, 10, 9, 10, 12, 13, 14, 15, 19, 43, 44, 45, 46, 47, 48, 50, 51, 55, 56, 57, 58, 59, 61, 64, 66, 68, 71, 72, 84, 86, 87, 88, 89, 90, 91, 92, 93, 94, 97, 99, 101, 103, 107, 109, 117, 119, 121, 125, 127, 130, 133, 135, 141, 146, 155, 166, 168, 169, 171, 172, 173, 174, 175, 177, 178, 179, 181, 182, 184, 185, 189, 192, 207, 211, 212, 215, 216

PWM pulse width modulation. 50, 58, 74, 75, 74, 77, 86, 87, 89, 90, 94, 97, 119, 121, 125, 141, 172, 178, 215

ROAM reduced-order averaged model. 19 SISO single-input single-output. 213, 214, 215 SOA safe operating area. 93, 103, 116, 121, 123, 166

SoC state of charge. v, 5, 6, 13, 14, 42, 43, 48, 64, 155, 171, 178, 192 STC standard test conditions. xix, 10, 64, 66, 68, 181, 207

ZCS zero current switching. 58

ZVS zero voltage switching. 35, 57, 58

(23)

Nire familia eta lagunei

(24)

(25)

Chapter 1

Introduction

Recently there has been a growing interest in developing high power and high en- ergy density energy storage systems (ESSs). Being less dependent on fossil fuels is a goal that can be achieved developing other energy sources and storage technologies.

Among the profits of lowering the dependence on these kind of fuels are its increas- ing costs, geopolitical concerns, pollution and global warming. There are several ways to deal with this problem; increasing the penetration of the renewable sources and increasing the efficiency of the transportation systems are two important ways to overcome this problem. New systems in the network need for proper interfaces.

These are realized using power electronic converters (PECs) that may not work at high efficiencies depending on the conditions and their operating point.

In this thesis the integration of battery-based energy storage and photovoltaic (PV) sources is discussed, focusing on the direct current (DC) interfacing of both parts through the use of PECs. There are several ways to achieve this and none of them is a universal best solution. This means that depending on the application, location and size among other parameters, the solution to adopt will not be the same.

Here, small residential-size and small-office-sized PV and storage installations are studied. Usually these are 10kWp or less PV installations. Regarding storage, there are different sizes in the market. A frequently repeated size is 2kWh in residential applications, which typically comprise 3kWp to 5kWp PV installations for a single household.

The main objective of the work contained in this thesis is to increase the energy that can be obtained from a system that comprises PV generation and battery-based storage. In order to achieve this, three different paths are followed: (i) maximmiza- tion of the produced PV energy, i.e. obtention of maximum power at all times, (ii) increasing the conversion efficiency by reducing the conversion stages or operat- ing at more efficient points when possible, and (iii) advanced energy managements strategies that improve the operation of the whole system.

So as to know how the energy conversion is performed simulations of a classical PV and storage systems are run. These simulations serve to determine the problems that arise in such systems under certain conditions. Once the energy conversion is studied different solutions are proposed, these are: increasing the storage system voltage, using a isolated converter topology to interface the energy storage system, and adopting a distributed maximum power point tracking (DMPPT)

¹

architecture.

The first two solutions tackle the low energy conversion efficiency problem in the storage system interfacing without any further improvement. This is achieved by making the voltage difference between the high- and low-voltage sides of the con- verter lower in the first case, and by matching these voltages with a high frequency (HF) transformer in the second case. The DMPPT architecture is able to increase the

1A DMPPT system comprises modules that include PV panels and PECs to track the maximum power point (MPP) of each panel even under partial shading conditions.

(26)

energy conversion efficiency when the storage is used as well. Moreover, it has the ability to increase the PV generation under partial shading conditions. For having the same advantages as the two first proposed solutions, and the ability to further increase the energy output of the system by increasing the PV generation under par- tial shading conditions, the adopted and thus further studied solution, is adopting a DMPPT architecture to interface in the DC side the PV generation and the ESS.

In the first part of the thesis, classical systems using non-isolated converters are modeled and simulated. These use PECs to interface the PV, the ESS and the grid.

Also a household consumption is considered as load. Different PV installation and ESS sizes are simulated and the results are presented along with the conclusions drawn from the study. The results are obtained from a local-grid simulator devel- oped for this purpose. The main contributions in this part of the thesis are an insight into the working of the selected classic converter architecture and the local-grid sim- ulator, a methodology to simulate the energy exchange between the different defined parts of the system.

In the second part of the thesis a DMPPT system with storage at module level is presented. This system can cope with the difference in low- and high-voltage re- quirements so as to achieve an acceptable energy conversion efficiency. Moreover, it presents some other advantages such as, modularity, increased fault tolerance and increased PV generation under shaded conditions. Besides, the inclusion of stor- age at module level further increases the maximum power point tracking (MPPT) capability of the DMPPT system when compared to storage-less DMPPT systems.

Including storage at module level increases the flexibility of the system by adding a degree of freedom to the power-flow in the DMPPT module

²

. This system not only aims for the integration of the different parts, but for a synergy in the operation of the different parts together. The main contributions of this part of the thesis are:

the concept of including the storage system at module level instead of directly at the DC-bus, design of the DMPPT module based on a three-port converter for stan- dalone applications, sizing of the module (PV, storage and converter), embedded control for the DMPPT modules, a supervisory control

³

for the whole DMPPT sys- tem that extends the MPPT capabilities of the DMPPT system, and a protection layer that uses the DC-bus voltage control and the battery-based power compensation in order to protect DMPPT systems based on boost-type converters.

In the end of the thesis the conclusions from the presented work and a com- parison between the classical systems and the proposed novel DMPPT architecture with storage at module level are presented. There it is demonstrated that the system achieves superior performance when compared to the classical system by working at higher efficiency, regardless of the size of the PV and ESS installations. Besides, the system is able to increase the PV generation of the system under partial shading conditions by performing the MPPT at each panel, instead of a string with many PV panels.

2A DMPPT module comprises a PV panel, small battery pack and a PEC.

3The supervisory control sets the current and voltage references of individual modules in order to adapt the working point of the whole system.

(27)

Part I

Storage and renewable energy

generation system analysis

(28)

(29)

Chapter 2

Introduction to Part I

In this part of the thesis a study of the PV and storage systems is carried out. The system topologies are discussed and one of them is chosen for further analysis. In this study the different considered parts are: (i) PV generation source, (ii) lithium- ion battery based ESS, (iii) a load and (iv) the utility grid. These parts work with DC and alternating current (AC) voltages and need PECs to be interfaced with each other. The arrangement of these PECs and their topology is directly tied to the spec- ifications and requirements of the application. In Figure 2.1 the different parts of the system and the power transfer paths between them are illustrated. In this figure a point of common coupling (PCC) is included where every part of the system is connected directly or by means of a PEC in order to exchange energy.

In this study it is assumed that the load and the utility grid are connected di- rectly to the PCC. The power transferred from the PV and ESS to the PCC will be instantaneously consumed by the load or injected into the grid depending on the load consumption. The grid will absorb or inject the excess or lack of power that is supplied by the generation and storage system. In essence, the sum of powers in the PCC is always equal to zero. The interesting power transfer paths that are con- sidered are only the ones illustrated in Figure 2.1. It is not considered though that the ESS can be charged with power coming from the grid as the fluctuation of the energy price is not considered.

Some of the possible grid-tied PV and storage system topologies are shown in Figure 2.2. The (a) and (b) cases present systems that have completely decoupled the PV generation and the ESS. This is generally an expensive solution that presents however an easy control of the energy that is generated and stored. Particularly the case (a) is prone to be the most expensive. The removal of the DC-DC converter may

Battery

PV

Grid

5

2 1

Load

3

4 PCC

FIGURE2.1: Different parts of the system under study and possible power-flow paths.

(30)

Multi-port converters Shared DC-Link

Independent converters

Battery

DC/DC

DC/AC PV

PCC

DC/DC Battery

DC/AC PV

PCC

DC/DC Battery

DC/AC PV

PCC

DC/AC Battery

DC/AC PV

PCC

DC/AC

DC/DC DC/DC

Dedicated converters

Battery

DC/AC PV

PCC DC/DC

Battery

DC/AC PV

PCC DC/DC

(a) (b) (c) (d) (e) (f)

FIGURE2.2: Possible system topologies for grid-tied PV and storage systems.

increase the cost of the system due to required oversizing in PV, ESS and inverter semiconductors. This occurs due to the widely different voltages that present both parts, i.e. PV generation and ESS. Oversizing the semiconductors in the inverter will increase its price, but most of the increase in cost would come from oversizing the ESS. In the system presented in case (b) this problem is overcome including the DC-DC converters that handle these voltage variations. Case (c) presents a simi- lar approach where the additional inverter in (b) is removed. Both generation and storage parts are controlled by means of DC-DC converters and they share the same DC-bus. The DC-bus is connected to the PCC by means of an inverter that may be rated less than the sum of the maximum powers in the generation and storage parts of the system. In order to further reduce the cost of the system, one of the DC-DC converters is removed in cases (d) and (e). In these configurations, the power of the part connected directly to the DC-bus is controlled using the inverter and the DC- DC converter controls the power supplied or consumed by the other. The preferred solution is the case (e) where the batteries are interfaced to the DC-bus by means of a DC-DC converter. This allows for a good control of the battery current and keeps the batteries from undesired cycling coming from DC-bus voltage ripple which may deteriorate the batteries in the long term. Another reason to interface the ESS and the DC-bus using a converter is that connecting a battery pack to the grid through an inverter, i.e. without any intermediate voltage-conditioning stage, leads to battery oversizing in terms of voltage due to the voltage swing versus the change in state of charge (SoC) during charge-discharge cycles [1].

In case (d), when the ESS is at its lowest admissible SoC, the inverter still requires

the ESS to supply enough voltage to work properly. Thus, when the ESS is working

at its highest admissible SoC, the DC-bus voltage will be higher than necessary for

operation, leading to increased losses in the inverter. Besides, the minimum voltage

condition imposed by the inverter has to be met at the end-of-life (EoL) condition of

the ESS. This means that when the system is new, losses will be higher depending on

the battery management system (BMS) strategy. The voltage swing in a battery pack

depends on the cell chemistry used, the number of cells connected in series and the

allowed depth of discharge (DoD) or SoC trip. Usually the voltage required by the

inverter is high and the voltage swing of a battery pack will be important. As exam-

ple, for a typical lithium-ion cell with a nominal voltage of 3.7V, the higher and lower

voltage limits are 4.2V and 2.75V respectively. If a 100% DoD is allowed, the lower

(31)

cell voltage should meet the inverter DC voltage requirement. In Europe this is 230 √

2 = 325V, which divided by the lower cell voltage limit yields 325/2.75 = 118 minimum number of series-connected cells. For a 118 series-connected cell battery pack, the higher voltage limit is 4.2 · 118 = 496V, a 53% higher. A time-dependent aging reserve can be implemented in software, i.e. increasing the allowable DoD as the battery ages. This provides the same effective capacity with low dependence on battery lifetime [2], but the voltage shift problem is not addressed.

In case (e), where the PV generation system is directly connected to the DC-bus, two problems may appear. If the PV system is not big enough in terms of power, the voltage it generates in its output may not be high enough for the inverter to work properly under certain conditions. When the PV system is too large in terms of power, it will require the use of high-voltage (HV) semiconductors in all the parts connected to the DC-bus (i.e. inverter and ESS converter) in order to stand the PV voltage. PV systems in open-circuit typically output a voltage that is 20% to 30%

higher than their MPP voltage. This will worsen the on-state characteristics and increase the price of the used semiconductor devices. In addition the DC-bus voltage working range will be very broad.

Case (f) presents a multi-port converter to interface the PV generation and the ESS with the DC-bus. There is an increasing interest in using this kind of converters in such application. The use of this topology is not found in the industry and remains in research phase at this moment. The characteristics of the system will depend on the multi-port converter topology chosen.

After considering all the advantages and disadvantages of the different presented

topologies, the option illustrated in Figure 2.2c has been chosen for further study in

Part I of the thesis.

(32)

(33)

Chapter 3

Outline of the local-grid study

In this chapter, the outline of the local-grid study is described. The study of the system is broken down into different steps, then the different results are input into the local-grid simulator. The local-grid simulator is a tool developed in Simulink that describes the use of the ESS and the power-flow between the different parts of the system for the given consumption and generation inputs.

The system topology under study is shown in Figure 3.1.This comprises ded- icated DC-DC converters for the PV generation and the ESS interfacing. The HV ends of these converters are connected to the DC-bus of the inverter. The inverter is connected to the PCC on its AC side. The load and the utility grid are connected to the PCC as well.

PPV PINV PG

PL

PESS

PCC

FIGURE3.1: Considered PV and storage system topology.

The outline of this chapter is as follows: first the case studies are presented with specific subsections for the PV generation, load consumption and energy storage system; secondly the local-grid simulator is introduced, then the energy manage- ment algorithms are presented and the chapter concludes with a discussion section pointing out the specificity of the study.

3.1 Case studies

Different case studies are selected. These studies comprise 3 and 10 kWp PV installa-

tions, different storage capacities and two load consumption profiles. Two different

seasonal conditions are considered, summer and winter, and different non-isolated

topologies are used to interface all the parts of the system.

(34)

3.1.1 PV generation

PV generation systems are built connecting PV panels in series to form the so called strings or arrays. Strings will contain a different number of PV panels depending on the required voltage. When a string meets the voltage requirements, but it does not supply the required power, more strings are added in parallel to the first one. These strings are by default containing the same type and number of PV panels

¹

.

Here 2 case studies are analyzed, 3kWp and 10kWp PV systems. In these two cases the use of Suntech Power STP245S-20-Wd PV panels is considered. This is a polycrystalline (p-Si) PV panel that is commonly used as example in this kind of works. The output characteristics of the selected PV panel are shown and gathered in Figure 3.2 and Table 3.1 respectively. Characteristics of the two PV generation systems at standard test conditions (STC)

²

are gathered in Table 3.2.

0 5 10 15 20 25 30 35 40

Voltage [V]

0 5 10

Current [A]

1 kW/m²

0.5 kW/m²

0.1 kW/m²

0 5 10 15 20 25 30 35 40

Voltage [V]

0 100 200 300

Power [W]

1 kW/m²

0.5 kW/m²

0.1 kW/m²

FIGURE3.2: I-V (top) and P-V (bottom) curves for Suntech Power STP245S-20-Wd module.

TABLE3.1: Main characteristics of the Suntech Power STP245S-20-Wd module.

Characteristics Value Rated power 245W Open-circuit voltage (V

_oc

) 37.3V Short-circuit current (I

sc

) 8.52A MPP voltage @ STC (V

_mpp

) 30.5V MPP current @ STC (I

_mpp

) 8.04A

1This is a requirement in classical systems if the MPP is to be tracked accurately. If any of the PV panels in one string is different tracking of the MPP in all the panels of the string will not be possible.

2Cell temperature of 25^◦C and irradiance of 1000 W/m2 with an air mass 1.5 (AM1.5) spectrum.

(35)

TABLE3.2: Selected PV system characteristics.

Characteristics 3kWp system 10kWp system

Number of panels per string 13 15

Number of strings connected in parallel 1 3

Rated power 3185W 11kW

Open-circuit voltage (V

_oc

) 484.9V 559.5V Short-circuit current (I

_sc

) 8.52A 25.56A MPP voltage @ STC (V

mp

) 396.5V 457.5V MPP current @ STC (I

_mp

) 8.04A 24.12A

0 4 8 12 16 20 24

Time [h]

0 200 400 600 800 1000

Irradiance [W/m

2

]

Winter Summer

FIGURE3.3: Clear-sky irradiance profiles of a whole day for winter and summer cases.

(36)

Considering a clear-sky model, two different days are simulated for summer and winter irradiance conditions. Applied irradiance profiles are shown in Figure 3.3. In order to interface the PV generation with the rest of the system, the classic 2-level (section 4.1) and 3-level (section 4.3) boost converters are selected. The working point of the converters is assumed to be at the MPP of the panel throughout the whole simulated day using a perturb and observe (P&O) algorithm.

3.1.2 Load consumption

Load consumption profiles are an input to the local-grid simulator. These are used to calculate the power-flow at the PCC and determine the amount of energy that is transferred to the grid. Besides, this is part of the information input to the energy management algorithm. The algorithm uses this information in order to decide if the PV generation is stored in the ESS or whether the latter is used to supply the load.

The load is connected to the PCC and the only way to transfer the generated or stored energy is by means of the inverter that interfaces the DC-bus with the PCC.

In this study the inverter is considered as ideal, having a 100% efficiency. In this way the load consumption profile is added directly to the DC-bus in DC quantities and no reactive power supply from the system is considered. This study focuses thus on the optimization of the DC side of the system.

In order to keep seasonal consistence two different load profiles are used, one corresponding to a summer day and another corresponding to a winter day like in the PV irradiance. The load profiles are measurements obtained from a Belgian household at a high sampling frequency (1kHz), shown in Figure 3.4. Using a high sampling frequency allows for detection of short power consumption peaks, such as fridge compressor start.

0 4 8 12 16 20 24

Time [h]

0 2 4 6 8

Active power [kW]

Summer

0 4 8 12 16 20 24

Time [h]

0 2 4 6 8

Active power [kW]

Winter

FIGURE3.4: One full day household (load) consumption profiles for a summer day (top) and a winter day (bottom).

(37)

Load consumption profiles in Figure 3.4 are used in the simulations with 3kWp PV generation systems. Simulations using a 10kWp PV generation system use the same profile but the power is multiplied with a factor of 3. This would be similar to the consumption of three identical households. This approach does not take into account the different timing in the load usage that may smooth out the power con- sumption when different consumers are added together. Therefore, this is the worst case scenario in terms of instantaneous power demand.

3.1.3 Energy storage system

The ESS selected in this study is a lithium-ion battery pack

³

. Originally two different battery pack voltages were proposed, 50V and 150V. The use of the 50V battery pack was discarded in the converter design phase due to high voltage step-up required by the application.

Different capacities using the 150V battery pack are studied using the 2- and 3-level boost converters. The maximum and minimum SoC are 90% and 10% re- spectively, allowing for an 80% DoD. This prevents the battery from working in the ranges where its derivative of voltage with respect to the SoC is high. Besides, restricting the DoD increases the battery pack lifetime and avoids overcharge and overdischarge due to errors in the SoC estimation. This is not a problem in simula- tion, but it should be taken into account as SoC estimation is not 100% accurate in real systems.

The ESS SoC at the beginning of the simulation corresponds to the remaining charge from the previous day using the same consumption and generation profiles in the local-grid simulator.

3.2 Local-grid simulator

The local-grid simulator, shown in Figure 3.5, is a tool developed in Simulink that reproduces the interaction between the different parts of the system in Figure 3.1.

The simulator has two distinct parts, the control executing the energy management algorithm, and the ESS.

The energy management algorithm is implemented inside the control block. The control receives information about the variables of the system which are the PV gen- eration after conversion, load consumption, SoC of the ESS and the power that the ESS is supplying or consuming.

The PV generation is treated as input because no curtailing is considered. This means that the PV system will always generate as much energy as it can. The power signal provided to the local-grid simulator is the amount of power fed to the DC-bus, incorporating thus the effect of the selected PV PEC, including converter dynamics without switching behavior and losses, and MPPT performance. In a similar way, the load power demand is given as input, that includes all assumable load dynamics thanks to high sampling frequency of the recorded data. Changing user habits is discarded in this work.

The energy management algorithm sets the references of inverter power and ESS power based on the available data.

3Although not specified, from the obtained curves the use of an LCO or LTO chemistry can be assumed. These chemistries present a flat discharge voltage profile.

(38)

ENERGY STORAGE SYSTEM PV

LOAD

PCC

ENERGY MANAGEMENT

ALGORITHM

ESS POWER

SoC ESS POWER REFERENCE

GRID POWER

BATTERY POWER

RT η STATE

FIGURE3.5: Diagram of the local-grid simulator. Inputs in green, outputs in red.

More complex algorithms can be developed by providing more data to the con- trol block. Typical data fed to advanced systems are weather forecast and user re- lated information. This is out of the scope of this work.

The ESS block is built using the battery model presented in section 4.5, an ef- ficiency map of the selected PEC (Figure 5.4) and a coulomb counting system that estimates the SoC of the battery pack. This accounts for the losses of the converter working at the given conditions.

The efficiency map of the PEC is calculated in this work from standalone simu- lations under the range of working conditions. Each point requires a simulation of the converter. Points for different charge and discharge currents as well as different voltages are calculated. With the calculated points a 2-dimensional look-up table is built. Efficiency information in this table includes the switching and conduction losses of the PEC. The models used in order to simulate the different working points are discussed in Chapter 4 and the resulting efficiency maps are shown in Figure 5.4.

A block diagram of the ESS block is shown in Figure 3.6.

The battery current is obtained by dividing the power demand by the battery voltage. Then the efficiency of the converter is multiplied with the power demand for retrieving the required low voltage power. The low voltage power is then used to calculate the SoC of the battery which is fed-back in order to generate the inner battery voltage. The high voltage DC-bus voltage is kept constant at 680V, while different battery voltages have been considered in the range [150V,175V].

3.3 Energy management algorithms

The energy management algorithm decides in each moment the use of the available energy in the system. This algorithm tests the different case studies to achieve the highest self-consumption rate through improvement of the round-trip efficiency of the system. Advanced energy management methods are not pursued as the goal is to increase the production and the efficiency of the hardware, i.e. the best PEC topology and working points of the system are pursued.

Author: Ander GONZÁLEZ FERNÁNDEZ DE BOBADILLA A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in Engineering Science and Technology

Integration of photovoltaic sources and battery based storage systems – A DC analysis and distributed maximum power point tracking solution