Offshore DC Grids Impact of Topology on Power Flow Control

download Offshore DC Grids Impact of Topology on Power Flow Control

of 6

Transcript of Offshore DC Grids Impact of Topology on Power Flow Control

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    1/6

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    2/6

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    3/6

    3

    from a Dispatch Center [5]; last, any power imbalance

    resulting from a contingency (loss of a converter, for

    example) or intermittent energy supply is automatically

    recovered, up to the others remaining capability.

    3.2 Dedicated P-V droop controls for specific behaviours

    The P-V droop is a generic control enabling to share the DC

    voltage regulation amongst a set of converters with norequirement for a master control or any communication

    between the converters. However, the implementation of

    dedicated P-V droop controls to share offshore wind power

    according to a pre-defined policy for all onshore converters is

    rarely discussed (if ever).

    In the remaining of this section, we discuss how dedicated P-V droop controls may be implemented to lead to different

    strategies with respect to onshore power injections, regardless

    of the offshore wind conditions. The principles are sketched

    based on the H grid topology, and will be illustrated inSection 4.2 on a more complex DCG. These behaviours also

    enable simpler operations for Dispatch Centers, since only the

    mid-point DC voltage should be set (instead of LRSP, slope

    and active power as described in [5]).

    Direct point-to-point injection.This behaviour ensures that

    the overall wind power generated by the ith

    wind farm (which,referring to Figure 1, is transmitted by WS-VSCito the DCG)

    is fully injected to the AC onshore mainland network by the

    GS-VSCiconverter (apart from the losses). Hence, the DCG

    behaves as if it was a set of radial wind farm connections

    (which would make sense for different producers to sell

    energy to their respective national markets, for instance), yetwith an additional interconnection feature, if required.

    To ensure this behaviour, the P-V droop control should be

    designed so that it controls the voltage for the H grid mid-

    points (Bi), rather than at the onshore converters DC

    connection point (Ci). This leads to a non-linear P-V droop

    characteristic (in red, Figure 3) since the slope depends on theDC voltage.

    Figure 3: P-V droop characteristics: dedicated point-to-pointinjection (plain), and linear approximation (dashed).

    Wind power mitigation.Another interesting behaviour is to

    use the DCG to mitigate all wind power injections, so that

    onshore converters get exactly the same amount of power.

    One of the main interest with this control is that mitigating

    different wind power sources leads to more constant

    injections on the AC network (all the more when the windfarms are numerous and geographically spread).

    As for the previous one, this behaviour is simply achieved by

    controlling the H grid mid-points voltages (Bi). Yet, theprinciple is to use the onshore converters DC voltage control

    to force the current to circulate through the DC tie (B1-B2)

    which constitutes the common mitigation branch, as will be

    illustrated in Section 4.2.

    Superimposing inter-area power exchange. The previous

    power injection policies are examples of dedicated P-V droop

    controls, for which the slope of the P-V characteristic is

    automatically adjusted depending on the DC system (cables

    resistance, converter losses) and the DC voltage; in addition,the voltage input parameter is set for the H grid mid-points.

    Hence, considering this is a no-load reference voltage, the

    implementation for any of those controls is straightforward,

    since a Dispatch Center should only provide the reference

    voltage as a set-point:

    Figure 4: Implementation of a dedicated P-V droop control

    for the H grid topology.

    Given the specific topology of the H grid, both the point-to-

    point injectionand wind power mitigationbehaviours may be

    complemented with a steady inter-area power exchange

    between GS-VSC1and GS-VSC2. This is easily controlled by

    setting the DC voltage orders for the onshore converters,

    according to Equation 1:

    (1)

    An illustration of how an inter-area exchange may be

    superimposed (and how this, in turn, requires supplementary

    constraints not to exceed the converters ratings) will beprovided in Section 4.2.

    Application to other DCG.It should be emphasized that the

    ability to implement similar behaviours using P-V droop

    controls relies on the DCG topology. For instance, assuming

    the same layout as in Figure 1, but where the B1-B2 is

    replaced by two new ones: A1-C2andA2-C1, the direct point-

    to-point injectionpolicy is not feasible without resorting to a

    master control to monitor in real-time the offshore wind

    generation, and to continuously adapt the DC voltagereference value accordingly.

    Pmax limit

    Pmin limit

    Vdc

    limit

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    4/6

    4

    4 H grid extension

    Assuming an existing H grid as depicted in Section 3, we

    discuss in the current one the possible extensions of such a

    DCG in order to connect a new offshore wind farm.

    4.1 Wind spillage illustration

    National or regional grid codes may differ with respect to thecapacity requirements to connect a new wind farm. For

    instance, RTE (the French TSO) has to ensure fully rated

    connections (i.e. the connexion has to be designed for the

    maximum export capacity of the wind farm). In other

    countries where wind spillage is allowed, guidelines provide

    indicators to estimate the optimal connection capacity (lowerthan the maximum wind farm generation capacity), depending

    for instance on the wind farm profile [7].

    To quantify possible wind spillage depending on cable

    ratings, we consider an already existing H grid described as

    follows (Figure 5).

    Figure 5: H grid extension with WF3(wind spillage allowed).

    The initial H grid is formed with two fully rated DC links (A1-C1, 1.2GW; and A2-C2, 1GW) connecting two wind farms

    which respective capacities are 1.2GW and 1GW; the

    capacity of the B1-B2 tie is not relevant for this illustration.

    The distance between the two existing wind farms (WF1and

    WF2) is 200km.

    Figure 6: Wind spillage estimate for various cable ratings.

    Figure 6 provides an estimate for the average wind powerspilled when connecting a new wind farm (WF3, rated for

    800MW) using two cables C1-C3 and C2-C3 as depicted in

    Figure 5, depending on the distance between C3and the other

    wind farms (for the sake of simplicity, we assume C3 is

    equidistant from them), and on the ratings of the two new

    cables (expressed as a ratio k of the total WF3 capacity). The

    wind spillage is computed using 10.000 correlated wind

    samples for each distance/rating cross-combination, thanks to

    the method exposed in Section 2. Since all power flows arenot controllable on this layout, it is not possible to push

    WF3 current toward the most appropriate node C1 or C3

    depending on remaining capacity, hence wind spillage isrelatively high, and cannot be limited by increasing the cables

    capacity beyond a certain threshold. Section 5 describes how

    power flow controlling devices would significantly improve

    wind spillage.

    This example is not intended to assess a break-even point

    beyond which wind spillage is economically sound: obviously

    enough, the results should be considered on a case-by-case

    basis since highly dependent on the examined parameters

    (new connection ratings, distance between wind farms), andthe assumptions (rating of the existing DCG, distance

    between WF1and WF2, capacities of the wind farms etc.), not

    to mention the cost for unsupplied energy. However, this

    gives a rough idea of the possible benefits of using as much as

    possible the existing assets when connecting a new wind

    farm, if wind spillage is an option.

    4.2 Prohibiting wind spillage: possible grid extensions

    This section illustrates possible H grid extensions to connect a

    new wind farm (WF3) with the ratings previously indicated in

    Section 4.1, assuming that wind spillage is not allowed or

    reveals economically unsatisfactory. As a consequence, WF3must be connected to the shore using a supplementary DC

    path, instead of re-using the existing assets (unless the H grid

    was initially oversized).

    Figure 7: H grid extensions without wind spillage (dotted):backbone (top) and meshed backbone (bottom) layouts.

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    5/6

    5

    Two possible layouts are sketched on Figure 7, where dotted

    lines represent new cables and converters to connect WF3 to

    the existing DCG. The main benefits of the meshed version of

    the backbone (compared to the tree-like one on top of Figure

    7) are: superior interconnection capacity between the three

    AC connection points of the DCG; ability to operate the DCGstill as a DCG (with interconnection capability) in N-1

    condition. Yet, both topologies may be used to provide

    similar power flow controls, as illustrated in the following.

    Considering the meshed backbone extension with the

    following parameters:

    - DC voltage: 320kV DC

    - Converters and wind farm ratings as indicated in

    Figure 7

    - Cable resistance (): A1-B1: 0.6; A2-B2: 0.5; A3-B3:

    0.6; B1-B2: 0.4; B1-B3: 0.7; B1-C1: 0.3; B2-B3: 0.6;

    B2-C2: 0.5; B3-C3: 1.2.

    An implementation of the controls described for the H grid (inSection 3.2) in the meshed backbone extension was realised

    in Scilab using the above parameters, with correlated wind

    speeds synthesized as shown in Section 2. Due to space

    limitations, only two examples are presented below.

    Direct point-to-point injection with constant inter-area

    power exchange. This simulation illustrates the behaviourpresented in Section 3.2, in which the P-V droop is tuned so

    that all onshore converters receive automatically the offshore

    wind power from the respective offshore converters they are

    directly connected to. In addition to that, we impose a

    constant 200MW power exchange from both GS-VSC1 and

    GS-VSC3to GS-VSC2.

    Figure 8: Point-to-point behaviour (with 200MW inter-areaexchange). Plain (offshore) and dotted (onshore) power

    curves are paired by colour.

    As illustrated in Figure 8, for each pair (GS-VSC i, WS-

    VSCi)i=1,2,3, the P-V droop control enables to reproduce

    exactly offshore wind power fluctuation on the onshore sideusing local measurements only (offshore and onshore power

    curves are paired by colour). Yet, the corresponding power

    curves are translated, as a consequence of the supplementary

    200MW power injection from GS-VSC1and GS-VSC3to GS-VSC2. Taking into account the ratings of GS-VSC2, the droop

    control suddenly raises the DC voltage (see Figure 9) so that

    the power injection never exceeds 1GW; thus, the excess of

    power is automatically shared between the two remaining

    converters (around t=300s). As soon as this constraint does no

    longer apply, the initial behaviour takes over.

    Figure 9: Point-to-point behaviour (with 200MW inter-area

    exchange). DC voltages at Ci(top),Bi(mid.),Ai(bottom).

    Mitigation (without inter-area power exchange). This

    simulation illustrates the P-V droop control designed to

    enable power mitigation, applied on the same network.

    Figure 10: Power mitigation behaviour. Offshore wind power

    (top), and corresponding onshore injections (bottom).

    As depicted in Figure 10, each onshore converter injects at

    any time one third of the total offshore power, up to itsmaximum capacity. Hence, GS-VSC3 increases the DC

    voltage around t=320s in order to cope with its power ratings

    (800MW); from this moment, the two others share exactly the

    remaining power, until GS-VSC2reaches in turn its maximum

    capacity (1GW) around t=330s. This situation lasts about

    100s, before the mitigation behaviour becomes effectiveagain. The same causes lead to same effects at t=710s.

  • 8/9/2019 Offshore DC Grids Impact of Topology on Power Flow Control

    6/6

    6

    5 Extending the power flow control with

    resistance modulation devices

    The H grid presented in Figure 1 and the two extensions inFigure 7 (backbone and meshed backbone layouts) are fully

    controllable using only onshore converters for DC voltagecontrol: in any DC cable, power flow control is achieved with

    appropriate voltage set-points from the onshore converters.

    On contrary, it is not possible to precisely control power in

    each DC cable for some topologies (such as the one depicted

    in Figure 5) due to the meshing and the large number of

    cables compared to the number of onshore converters.

    Partial power flow control may result in overloads. In such

    situations, [8] shows to what extend the use of supplementary

    DC power flow control (PFC) device may increase the regionof operation for a DCG, by extending power flow flexibility.

    The authors also propose a thyristor-based converter to

    achieve this PFC feature.

    In our study, we also considered similar ideas (by adapting to

    the DCG a series impedance modulation device [9],composed of electromechanical and power-electronic

    switches to connect/disconnect sub-conductors), yet from a

    slightly different standpoint: indeed, using PFC devices on

    partially controlled DCG should reduce (or even eliminate)

    wind spillage, hence making some topologies such as the onedepicted in Figure 5 more cost-effective. To assess this idea,

    new wind spillage estimates were computed on the same

    topology, yet assuming PFC devices on branches C1-C3 and

    C2-C3, and compared to the previous estimates from Figure 6.

    As expected (Figure 11), significant gains are achieved usingPFC devices, especially with higher cables ratings (whichenable greater flexibility for power dispatch on the grid).

    Figure 11: Wind spillage estimate for various cable ratings,

    with (plain) and without (dotted) PFC devices.

    6 Conclusions

    Based on ongoing work in the European project TWENTIES,

    this paper highlights power flow issues for an offshore DC

    grid connecting intermittent energy sources (wind turbines) tothe AC mainland grid. In order to assess the value for

    different grid extensions and ratings with respect to possible

    wind spillage, a process used to generate geographically

    correlated wind speeds is described first. As an illustration,

    expected wind spillage values are computed for a simple DC

    grid extension. Depending on the associated cost for

    unsupplied energy and local connection requirements,different DC grid topologies may therefore be preferred.

    The article also illustrates (through simulations run on a

    meshed 6-terminal grid) how dedicated Power-Voltage droopcontrols may take advantage of the grid topology to

    implement specific power flow policies (such as mitigating

    variable generation, for instance) relying only on local

    measurements. Yet, this does not apply to all DC grids: the

    topology has an impact on which behaviours may be

    implemented without resorting to a master control or

    communication.

    Last, all DC layouts do not allow for complete power flow

    control due to the meshing (whatever the kind of DC voltage

    control). An adaptation of the AC impedance modulationprinciple is suggested which, if used in partially controlled

    DC grids, would enable significant wind spillage savings.

    Acknowledgements

    This ongoing work is part of the European TWENTIES

    project, funded within the 7th

    Framework Program (FP7).

    References

    [1] D. Cirio, A. Pitto, A.M. Denis, O. Despouys, L. He,

    C.C. Liu, C. Moreira, B. Silva, Dynamics and controlof Multi-Terminal High Voltage Direct Current

    networks for integration of large offshore wind parksinto AC grids, CIGRE Session 44, Paris (August 2012).

    [2] E. Kahn, The reliability of distributed wind

    generators, Electric Power Systems Research, volume

    2, pp 1-14 (March 1979).

    [3] M. Doquet, Use of a stochastic process to sample wind

    power curves in planning studies, Power Tech,

    Lausanne, pp. 663-670 (July 2007).

    [4] http://www.atlanticwindconnection.com.

    [5] C. D. Barker, R. S. Whitehouse, Autonomous

    Converter Control in a Multi-Terminal HVDC System,

    ACDC, London (October 2010).[6] G. Pinares, M. Bollen, Understanding the Operation of

    HVDC Grids, CIGRE Symposium, Bologna (September

    2011).

    [7] P. Djapic, G. Strbac, Cost Benefit Methodology for

    Optimal Design of Offshore Transmission Systems,

    Centre for Sustainable Electricity and Distributed

    Generation, http://www.sedg.ac.uk/, (July 2008).

    [8] E. Veilleux, B.-T. Ooi, Multi-Terminal HVDC Grid

    with Power Flow Controllability, CIGRE Session 44,

    Paris (August 2012).

    [9] P. Couture, Switching modules for the extraction/

    injection of power (without ground or phase reference)

    from a bundled HV line,IEEE Transactions on PowerDelivery(July 2004).