Sumario Executivo.pdf

10
Multilevel modular converters for off-shore wind farm HVDC transmission system Valentin Plyusnin Under supervision of Prof. Dr. F. Silva Dep. of Electrical and Computer Engineering, IST, Lisbon, Portugal April 2014 Abstract - The aim of this work is to design a decentralized DC capacitor voltage balancing controller for high voltage multilevel modular converters in direct current cable transmission systems within and from off-shore wind farms to inland. The high voltage direct current (HVDC) system has been designed to transfer electric power from off-shore generators to inland substations. Carried out simulations of 81-level converter system included the stress testing of developed decentralized algorithms in order to confirm their performance. Additional closed loop power control system has been developed and implemented. It was demonstrated that the developed decentralized converter system is able to balance the capacitor divider voltages and can operate on a reduced subset of levels to provide redundancy. Functionality of the whole HVDC system has been tested by simulation of the dynamics of its parameters within the specified values. Introduction The high voltage direct current (HVDC) systems are becoming more widely used in networks for electric power transmission and distribution from off-shore wind farms to inland due to their advantages in comparison to other solutions. These systems are characterized by lower losses and the absence of reactive power problems. This means that there is no need to use an additional equipment for correction of power factor or compensation of reactive power. However, the necessity of additional equipment for conversion of electric power generated in AC mode makes expensive the use of HVDC system on short distances. When the distance increases larger than 80 km this solution becomes economically advantageous [1]. The concept of Modular Multilevel Converter (M2C) [5] is increasingly used in the HVDC transmission systems since this new design offers better characteristics and increased efficiency. Limited working space available on off-shore stations imposes restrictions on the size of converters and requires certain design features for further minimization. Unfortunately, most of the investigated multilevel converters topologies (diode-clamped type (NPC), capacitor-clamped type (flying capacitors)) does not meet these requirements [2, 5]. Unlike other systems, M2C satisfies to these requirements due to its modular structure. Besides this point, a lot of other important aspects have to be taken into account. Main requirements are: the fast and independent control of active and reactive power flow, the operation without bulky passive filters, the black start capability and the option of extending the DC- network to more than two stations. At severe fault conditions and disturbances, including short circuits, the DC-side must be managed fast and safely [2, 3, 4]. This paper includes 5 sections. A description of M2C is given in the first part. Second part provides a description of the proposed voltage control techniques and introduces a new decentralized cell selection algorithm to balance capacitor voltages. In the third section power injection control is developed. Simulation results are presented in a

Transcript of Sumario Executivo.pdf

  • Multilevel modular converters for off-shore wind farm HVDC

    transmission system

    Valentin Plyusnin

    Under supervision of Prof. Dr. F. Silva

    Dep. of Electrical and Computer Engineering, IST, Lisbon, Portugal

    April 2014

    Abstract - The aim of this work is to design a

    decentralized DC capacitor voltage balancing

    controller for high voltage multilevel modular

    converters in direct current cable transmission

    systems within and from off-shore wind farms to

    inland. The high voltage direct current (HVDC)

    system has been designed to transfer electric

    power from off-shore generators to inland

    substations. Carried out simulations of 81-level

    converter system included the stress testing of

    developed decentralized algorithms in order to

    confirm their performance. Additional closed loop

    power control system has been developed and

    implemented. It was demonstrated that the

    developed decentralized converter system is able

    to balance the capacitor divider voltages and can

    operate on a reduced subset of levels to provide

    redundancy. Functionality of the whole HVDC

    system has been tested by simulation of the

    dynamics of its parameters within the specified

    values.

    Introduction

    The high voltage direct current (HVDC) systems

    are becoming more widely used in networks for

    electric power transmission and distribution from

    off-shore wind farms to inland due to their

    advantages in comparison to other solutions.

    These systems are characterized by lower losses

    and the absence of reactive power problems. This

    means that there is no need to use an additional

    equipment for correction of power factor or

    compensation of reactive power. However, the

    necessity of additional equipment for conversion

    of electric power generated in AC mode makes

    expensive the use of HVDC system on short

    distances. When the distance increases larger than

    80 km this solution becomes economically

    advantageous [1].

    The concept of Modular Multilevel Converter

    (M2C) [5] is increasingly used in the HVDC

    transmission systems since this new design offers

    better characteristics and increased efficiency.

    Limited working space available on off-shore

    stations imposes restrictions on the size of

    converters and requires certain design features for

    further minimization. Unfortunately, most of the

    investigated multilevel converters topologies

    (diode-clamped type (NPC), capacitor-clamped

    type (flying capacitors)) does not meet these

    requirements [2, 5]. Unlike other systems, M2C

    satisfies to these requirements due to its modular

    structure. Besides this point, a lot of other

    important aspects have to be taken into account.

    Main requirements are: the fast and independent

    control of active and reactive power flow, the

    operation without bulky passive filters, the black

    start capability and the option of extending the DC-

    network to more than two stations. At severe fault

    conditions and disturbances, including short

    circuits, the DC-side must be managed fast and

    safely [2, 3, 4].

    This paper includes 5 sections. A description of

    M2C is given in the first part. Second part provides

    a description of the proposed voltage control

    techniques and introduces a new decentralized cell

    selection algorithm to balance capacitor voltages.

    In the third section power injection control is

    developed. Simulation results are presented in a

  • fourth section and finally, conclusions are drawn in

    the last part.

    1. Multilevel Modular Converter (M2C)

    1.1. Principle of M2C topology

    The modular multilevel converter (M2C) is a

    new type of voltage source converter (VSC) for

    medium or high-voltage DC power transmission

    [4]. Figure 1 presents the M2C circuit topology

    which is formed by 2-level cells (SM). These cells

    are externally driven in order to obtain the

    required multilevel output voltage waveforms.

    Each converter half-arm creates discrete voltage

    levels between zero and power supply value U. The

    sum of both half-arm voltages gives the output

    voltage in range of [U, -U].

    The N-level converter output voltage is given

    by (1), where is a discrete variable (2) and single

    cell capacitor voltage uci can be calculated from (3).

    uo = U (1)

    { 1 , N 3

    N 1 , , 0 , ,

    N 3

    N 1 , 1 } (2)

    uci = UdcN 1

    (3)

    Figure 1b schematically presents an equivalent

    circuit for converter arm modelling.

    1.2. Calculation of M2C single arm parameters

    Since a three-phase M2C converter can be

    considered a three-phase balanced system, the

    single phase equivalent model is here used to

    derive transmission power. Furthermore, the

    converter output voltages have nearly sinusoidal

    waveforms, so it can be approximated by an AC

    voltage source (V1), as it is shown on Figure 2, in

    which the equivalent circuit of M2C arm connected

    to the grid is illustrated.

    It is important to note that protection chokes

    arm inductors, which are inserted in each half-arm

    to limit the AC-current whenever the DC-Bus is

    short circuited (fault-condition), are not

    considered [4], since the circulating current is

    usually small enough.

    Figure 1a Three-phase schematic of M2C Figure 1b equivalent circuit of M2C single arm

    Figure 2 Equivalent circuit of M2C single arm connected to the grid

  • Current I is given by:

    I =V1 V2jXL

    (4)

    The AC power flow between the two sources is

    given by the following equations [1]:

    P12 = V1rms V2rms sin

    XL (5)

    Q12 = V1rms2 V1rms V2rms cos

    XL (6)

    The phase difference angle expression (7) can be

    obtained from (5) and (6).

    = acos (V1rmsV2rms

    P

    P2 + Q2) + atan2(Q, P) (7)

    In order to avoid problems in the series

    connection of semiconductors in each cell,

    according to the manufacturers available power

    devices, 3 kV voltage cells were assumed. DC input

    and AC output voltages (U and V1rms) for 81-level

    converter (single phase) can be calculated using

    (1)-(3).

    Figures 3a, 3b present temporal evolutions at

    steady state conditions of output current and

    voltage.

    Cell capacitors were designed accordingly to

    (8), as in [6].

    C = ict

    Vc (8)

    According to (8), the maximum displacement

    current that flows through the capacitor

    corresponds to the arm current, which can be

    calculated from [7].

    1.3. Lower losses and efficiency of M2C

    The efficiency is an important parameter of the

    M2C. All power losses have to be identified and

    taking into account in simulations. The efficiency of

    the M2C can be given by (9), where the input

    power (denominator) is represented by the sum of

    the output power (Po) and the power losses (Plosses)

    [9].

    =Po

    Po + Plosses (9)

    In this study the losses of the cell DC capacitors

    and arm inductors can be included, if significant

    [8], by considering the respective resistive terms

    associated to the semiconductor on-state

    resistances. Therefore, Plosses is a sum of the

    semiconductor on-state and switching losses. The

    on-state losses (PLon) are characterized by the

    semiconductor on-state resistance (Ron) and the

    root mean square value of the arm current Irms:

    PLon = Ron Irms2 (10),

    As it is shown in [8], the switching losses are

    dominant. To evaluate these losses the dissipated

    energy in each switching (11) is given as:

    Wpcn = Vcemax ipcn tr+tf

    2 (11),

    where the semiconductor fall and rise times (tf and

    tr) are provided by the manufacturer, Vcemax

    corresponds to the maximum cell voltage value

    and ipcn (red curve in Fig. 4) is the current that

    flows through the cell in the switch instant.

    Figure 3a M2C output current (81-level converter)

    Figure 3b M2C line-to-neutral output voltage (81-level

    converter)

  • Since half of the output current flows in each arm

    [7], the switching power losses of the

    semiconductors are given by:

    PT2=

    WpT2

    T2

    = Vcemax IcavT2

    tr + tf2

    (12)

    The developed approach to calculate M2C

    losses will be used in section 3 for the

    determination of a 25-level M2C efficiency.

    2. M2C Control Method Description

    2.1. Sigma-Delta Modulator

    To obtain the desired converter output voltage

    level, a Sigma-Delta modulator based on PWM

    modulation technique (Fig. 5) is used. In this PWM

    modulator the areas of the modulating reference

    voltage (vref) and of the scaled converter output

    voltage (vout) must be equal during one period,

    therefore the error between vref and vout signals is

    integrated (13). When the result of (13) exceeds

    specified limits, the voltage level is increased or

    decreased, as it is shown in Figure 6.

    1

    TC vref vout

    TC

    o

    dt = 0 (13)

    The reference voltage is generated in external

    control system designed in section 3.2.

    2.2. Cell capacitor voltage balancing technique

    A proper cell selection algorithm is necessary

    to ensure the cell capacitor voltage tracks a

    reference value, together with low switching

    frequency for each cell [3].

    Capacitors can be charged or discharged

    respectively upon the sign of the current that flows

    through. Since there are redundant levels (several

    cell combinations are possible for the same level

    realization), selection of different cells, can be

    made based on cell capacitor voltage comparison.

    Thus if the current sign is positive, cells should be

    ordered increasingly (low to higher capacitor

    voltages) and ordered in the opposite if the current

    sign is negative. It is important to notice that each

    M2C half-arm currents have different signs,

    therefore the sorting should be made separately

    for each half-arm. As it can be seen in Figure 7,

    once the algorithm is implemented, cell capacitor

    voltages are maintained within specified margins.

    With this voltage balancing centralized

    method, all cells voltage need to be compared in

    order to obtain an accurate result.

    Figure 5 Sigma-Delta control

    Figure 6 Modulator output level (1) and error (2) signals

    Figure 4 - M2C output current and single cell switching

    instants

  • This implies the use of a very sophisticated central

    control unit (CCU), capable of making a fast signal

    processing.

    2.3. Decentralized control of the M2C

    The new proposed decentralized cell capacitor

    voltage control methodology allows to each M2C

    cell to define its own control. As a result, each cell

    is capable to decide on its respective state based

    on the comparison of cell voltage with another cell.

    This means that the required input data could be

    significally decreased (level, cell voltage, another

    cell voltage) resulting in a simplification through

    cooperation of the controller unit.

    Behavior of the developed algorithm for

    decentralized controller can be divided for 2 parts.

    In the first part input variables are evaluated.

    Current sign is detected and cell voltage

    comparison is made. The result obtained in this

    part refers to the cell selection, made in pairs of

    cells. This decision will be updated in the second

    part of the algorithm, which refers to guarantee

    the desired converter voltage output level. Second

    part decision is based on the evaluation of internal

    variables - cell number and counter A. According to

    the level desired it is possible to know how many

    cells should be in DC voltage mode. Counter A

    variable, which is transmitted from cell to cell,

    permits to conclude how many cells behind the

    evaluated cell are already in this mode. Once

    known the cell number, the output level and the

    counter A value, the number of cells that should

    have already been turned on can be calculated.

    With this information, final decision about the cell

    state is taken. Figure 8 shows the flowchart of the

    decentralized algorithm.

    With this new strategy, modular controllers

    process a small amount of data each. It is

    important to notice that cell voltage comparison

    isnt made between random cells, but there is a

    specific configuration that guarantees the correct

    performance of these controllers. With this

    configuration, nearly equal switching frequencies

    of each cell were observed.

    From [10], it is known that each cell needs

    duplex optical-fiber (OF) cables to transmit the

    data to the CCU. With this new autonomous cell

    concept, only one OF cable is needed to transmit

    the information about the desired voltage level.

    Tests on single phase converters up to 81 level

    were performed. Figure 9 shows that this new

    control methodology permits to maintain the cell

    capacitor voltages satisfactory balanced even in

    unfavorable performance conditions, such as

    redundant operation with low switching

    frequency.

    Figure 7 Cell capacitor voltages of 9 level converter

  • 3. HVDC system

    3.1. Calculation of the HVDC system

    parameters

    HVDC system consists of a wind farm and grid

    circuits (RLe), two M2C converters operating in AC-

    DC and DC-AC modes, two transmission power

    cables and DC-Bus capacitors. All these

    components are schematically shown in Figure 10.

    Some extra power components (i.e. smoothing

    coils) are added. In this small example, it is

    assumed that 29 turbine wind farm (2 MW each)

    Figure 8 - Flowchart of the decentralized algorithm

    Figure 9 Low frequency redundant operation of 81 level

    M2C with 21 levels

  • generates 56 MW of electric power, which is

    transmitted by submarine power cables to the in-

    land substation. The calculated system parameters

    are presented in Table 2.

    Additional active and reactive power controls (PQ system) have been developed in order to minimize the deviations in power values.

    Table 2 HVDC system parameters

    WIND FARM

    Pfarm 56 MW

    V 35.5 kV

    IRMS 743.6 A

    Imax 1051.6 A

    L 7.4 mH

    R 0.2

    GRID

    Pgrid 54 MW

    V 35.5 kV

    IRMS 717 A

    Imax 1014 A

    L 7.2 mH

    R 0.21

    M2C 1

    N (number of levels) 25

    Vcell 3047.5 V

    Ccell 11.5 mF

    99.5%

    M2C 2

    N (number of levels) 25

    Vcell 3000 V

    Ccell 11.3 mF

    99.6%

    DC cable

    l (line length) 120 km

    RL 0.7372

    LL 0.084 H

    UDC 72 kV

    Ceq (DC-Bus eq. cap.) 0.84 F

    3.2. Power injection control

    To provide effective HVDC operation the PQ

    system must monitor and control the output

    currents (i1, i2 and i3) and DC-side voltage (Udc) (Fig.

    9). Following the approach described in [1, 6] the

    output currents in dq reference frame are obtained

    solving the system (14), which describes the

    processes of Figure 11.

    {

    diddt

    = R

    L id +

    1

    L (L iq + d Udc ed)

    diq

    dt=

    R

    L iq +

    1

    L ( L id + q Udc eq)

    (14)

    In (14), d and q are modulation indexes which,

    after the transformation to 3-phase reference

    frame, will be sent as input signals to the Sigma-

    Delta controller. Active power injected into the grid

    depends on the direct current component (id) (15).

    The quadrature current component (iq) is related to

    the reactive power which should be kept around

    zero (15).

    {p = ed id + eq iqq = ed iq + eq id

    {p = ed idq = ed iq

    (15)

    The equivalent M2C model in dq coordinates is

    shown in Figure 12. For the DC-side the following

    equation is obtained:

    C dUdcdt

    = d id + q iq UdcReq

    (16)

    Figure 10 The schematic HVDC system

    Figure 11 HVDC grid-side schematic

    Figure 12 Equivalent M2C model in dq reference frame

  • Once decoupled applying

    {Hd = (L iq + d Udc ed)

    Hq = ( L id + q Udc eq) (17),

    multiple input multiple output (MIMO) system in

    (14), similarly to [6], single input single output

    system (SISO) (18) is obtained.

    {

    diddt

    = R

    L id +

    1

    L Hd

    diq

    dt=

    R

    L iq +

    1

    L Hq

    (18)

    Applying Laplace transform to (14) and (16) results

    in transfer functions (19) and (20) which are

    represented by block diagrams in Figures 13 and

    14.

    id,q(s) = Hd,q

    sL + R (19)

    Udc(s) =Req

    s Req Ceq + 1 (d id + q iq) (20)

    C(s) block in Figures 13 and 14 represent the

    corresponding current and voltage PI controllers

    which are described in [6], so that the respective

    gains (Kpi,v, Kii,v) are given by (21) and (22) [6].

    {Kii = n

    2 L

    Kpi = 2 n L R (21)

    {

    Td =

    1

    2 n

    Kiv =n2 Td

    d Req Kpv = Kiv Req Ceq

    (22)

    The Idref signal in Figure 14 serve as input

    reference signal for the current control system (Fig.

    13). The output signals of the whole control system

    are 2 modulation indexes in dq reference frame

    (23).

    {

    d =

    Hd + L iq

    Udc

    q = Hq + eq + L id

    Udc

    (23)

    Using Park and Concordia transforms on (23) one

    can obtain Sigma-Delta input sinusoidal

    modulation indexes for each phase.

    4. Simulation Results

    This section presents HVDC system with PQ

    control and 25 level M2C converters modelled in

    Simulink environment. The results are presented in

    2 separate parts. The first part characterizes the

    inland grid-side, while another one refers to the

    offshore wind farm-side.

    4.1. Grid-side

    The simulation results for the output voltage

    and current are shown in Figures 15a and 15b. 25-

    level sinusoidal output waveform is obtained by

    each arm of the converter. Voltage and current

    total harmonic distortion (THD) were of 3.69% and

    1.43% respectively. These THD values are

    acceptable according to the high voltage

    standards, needing almost no filtering.

    Cell capacitor voltages (Figure 15c) are

    stabilized around the reference value after a short

    transient. This confirms the reliability of the

    developed decentralized controllers.

    Figure 13 Closed loop current control system

    Figure 14 Closed loop voltage control system on DC-side

    Figure 15a Grid-side 25 level M2C converter output

    voltages (1-1.08s)

    Figure 15b Grid-side 25 level M2C converter output currents

    (1-1.08s)

  • Figures 15d-15g demonstrate the operation of

    PQ control system. After a short transient the

    output currents become balanced, and the

    converter input voltage (UDC) is stabilized on the

    specified reference value. Active power injected

    into the grid is regulated to 54 MW. The reactive

    power is kept around zero.

    4.2. Wind farm-side

    Figures 16a-16f demonstrate the wind farm-

    side operation. THD values for the converter

    output voltages and currents were of 4.01% and

    1.43% respectively.

    Similarly to section 4.1, cell capacitor voltages

    (Fig. 16c) are balanced after a short transient.

    Figure 15c Grid-side 25 level M2C cell voltages (0-3s)

    Figure 15d Grid-side 25 level M2C converter output currents

    (0-1s)

    Figure 15g Reactive power injected into the grid (0-3s)

    Figure 15f Active power injected into the grid (0-3s)

    Figure 15e Underground DC power cable voltage (0-3s) Figure 16b - Wind farm-side 25 level M2C converter

    output currents (1-1.08s)

    Figure 16c Wind farm-side 25 level M2C cell voltages (0-3s)

    Figure 16a Wind farm-side 25 level M2C converter output

    voltages (1-1.08s)

  • Use of PQ control system provides fast

    stabilization of the wind farm-side M2C output

    currents at the wind farm active power kept equal

    to 56 MW and the reactive power around zero (Fig.

    16d-16f).

    5. Conclusions

    In this work an innovative algorithm for

    decentralized balancing and control of M2C cell

    capacitor voltages is presented. Tests

    demonstrated suitable performance of converters

    with up to 81 levels and revealed that this new

    technique can assure M2C redundant operation

    even with low switching frequency. Developed

    methodology for decentralized control makes it

    possible to provide the cell autonomy using

    decentralized controller units decreasing the

    required number of fibre-optical cables.

    HVDC system based on two 3-phase 25-level

    converters operating in both DC-AC (wind farm-

    cable) and AC-DC (cable-grid) modes has been

    designed. Additional PQ control also has been

    developed to minimize the deviations of system

    input/output powers (0.1%). Achieved

    efficiencies of the converters were of 99.5% and

    99.6%. The obtained current THD were 1.43% for

    both converters, while voltage THD achieved of

    4.01% for wind farm-side and 3.69% for grid-side.

    The usefulness of the designed system has been

    proved.

    References

    [1] J. P. Sucena Paiva, Redes de Energia Eltrica: uma anlise

    sistmica, 2 edio, IST Press, Instituto Superior Tcnico,

    Lisboa, Portugal, Dezembro 2007.

    [2] - J. Fernando Alves da Silva, J. J. E. Santana, S. Ferreira Pinto,

    CONVERSORES COMUTADOS PARA ENERGIAS RENOVVEIS,

    Instituto Superior Tcnico, Lisboa, Portugal, 2012.

    [3] - A. Hassanpoor, K. Ilves, S. Norrga, L. Angquist, H. Nee,

    Tolerance-band modulation methods for modular multilevel

    converters, Power Electronics and Applications (EPE) 15th

    European Conference, Lille, France, 2013, p. 1-10.

    [4] R. Marquardt, Modular Multilevel Converter: An

    universal concept for HVDC-Networks and extended DC-Bus-

    applications, Power Electronics Conference (IPEC), 2010

    International, pp 502 507, Sapporo, Japan, June 2010.

    [5] - A. Lesnicar, R. Marquardt, New Concept for High Voltage

    Modular Multilevel Converter, PESC Conference, Aachen,

    Germany, 2004, p. 5.

    [6] J. Pereira, Conversor de nove nveis para ligao de um

    gerador fotovoltaico rede eltrica, Dissertao de

    Mestrado, Instituto Superior Tcnico, Lisboa, Portugal,

    Outubro de 2011.

    [7] P. Asimakopoulos (CERN), Dr. K. Papastergiou

    (dir.)(CERN), Prof. M. Bongiorno (dir.) (Chalmers U. Tech.),

    Design and Control of Modular Multilevel Converter in an

    Active Front End Application, Chalmers U. Tech., Gothenburg,

    Sweden, 2013-09-01. - 100 p.

    [8] S. Allebrod, R. Hamerski, R. Marquardt, New

    Transformerless, Scalable Modular Multilevel Converters for

    HVDC-Transmission, Power Electronics Specialists

    Conference, 2008, pp 174 179, PESC 2008. IEEE.

    [9] - J. Fernando Alves da Silva, Electrnica Industrial:

    semicondutores e conversores de potncia, Fundao

    Calouste Gulbenkian, 2013.

    [10] - A. Lesnicar, R. Marquardt, "An innovative modular

    multilevel converter topology suitable for a wide power

    range", Proc. IEEE Bologna Power Tech Conf., vol. 3, 2003.

    Figure 16d Wind farm-side 25 level M2C converter output

    currents (0-1s)

    Figure 16e Active power injected into the DC power cable

    (0-3s)

    Figure 16f Reactive power injected into the DC power cable (0-3s)