Using high voltage DC transmission and multi-terminal DC networks for offshore wind farm Integration

Large offshore wind farms have been planned and some are currently under construction in Europe. 

Go to the profile of Stephen Finney
Oct 13, 2017
1
0
Upvote 1 Comment

Author(s): Lie Xu; Stephen Finney; Olivier Despouys

Abstract

Multi-terminal HVDC transmission technology using voltage source converters has been proposed for integrating large offshore wind farms into transmission grids. This article introduces some of the main concepts and challenges associated with the design and development of large offshore DC systems. Different converter topologies and offshore DC grid configurations are discussed. Droop based DC voltage control and power dispatch strategies are described and strategies for safe ride through onshore AC grid fault are explored. System behaviour and protection requirement during faults on the DC grid are highlighted.

Introduction

The UK Government has committed to ensuring 15% of the UK’s energy demand is provided by renewable sources by 2020 [1], and has established a series of carbon budgets that oblige a halving of greenhouse gas emissions by 2027 relative to 1990 levels [2]. These challenging targets require increased use of low-carbon energy sources. The low-carbon energy resources found around Britain’s coast and in its territorial waters – wind, wave and tidal – are vast and a significant investment in harnessing such resources is already underway and expected to continue. Equally, European Wind Energy Association estimates that 120 GW of offshore wind power will be installed in the next two decades in Europe, amounting to 10% of the installed generation capacity. However, the technologies involved need to be made more cost-effective and more robust for operation in challenging offshore environments. Furthermore, appropriate power networks need to be developed to bring the energy ashore and to the main demand centres.

In cost-benefit terms, a number of recent studies have indicated that, for a 1 GW offshore wind farm, high voltage DC (HVDC) becomes preferable to high voltage AC technology for the main connection to the onshore grid when cable lengths exceed 80 km. All of the UK’s Round 3 sites identified for the next phase of offshore generation development are at such distances from shore and are therefore highly likely to depend on HVDC technology. HVDC system using voltage source converter (VSC) is suitable for operation on weak AC systems such as those that may be expected to be developed as offshore hubs for collection of wind generation outputs. Although VSC technology currently has limited capacity and higher power loss than line commutated converter (LCC), it also promises significant advantages, being suitable for operation as part of a multi-terminal HVDC (MTDC) grid that, by avoiding conversion between AC and DC at particular bussing points, would reduce the total number of converters and can allow the efficient pooling and dispatch of energy over a vast geographical area [3].

The interconnection of various offshore wind farms into a single ‘Super Grid’ has also been proposed, with the purpose of supplying a large portion of the total power demands throughout Europe [4]. To this end, development of a ‘grid ring’ connecting various offshore wind farms, proposed for construction in the North Sea, has been put forward, with the aim of linking neighbouring countries through the construction of HVDC lines [5]. Fig 1 shows an example of offshore MTDC networks in the North Sea in which large numbers of offshore wind farms and power networks are connected through the DC grids.

Fig 1: Diagram of North Sea MTDC networks for connecting large offshore wind farms and power networks

To ensure security of supply and safe, stable system operation on a large scale, there are number of challenges that need to be addressed before such MTDC systems can be implemented. This article discusses some of the key issues of offshore DC network design, power flow control of large MTDC system, system operation and protection during AC and DC faults.

Principles of VSC based HVDC transmission

VSC technology provides the means by which a transmission level AC voltage can be synthesised from a HVDC supply. The fundamentals of VSC based HVDC transmission operation may be explained in simple terms by considering each terminal as shown in Fig 2 a. The equivalent circuit at fundamental frequency can be represented by considering a voltage source generated by the converter connected to the AC main via a three-phase reactor and power transformer as shown in Fig 2 b. Both the amplitude and the phase angle of the output fundamental voltage V c are controlled with respect to the voltage V s.

Fig 2: Basic VSC based HVDC Transmission

According to the fundamental frequency phasor diagram shown in Fig 2 c, the active and reactive power at the grid V s side can be expressed, respectively as

(1)

where δ is the phase angle between the V c and V s and X s is the reactance of the coupling inductor at fundamental frequency.

From (1), it can be seen that the active and reactive power are controlled independently through V csin δ and V ccos δ, respectively. As shown in Fig 2 c and (1), for active power flows

  • if V c phase-lags V s, active power P s (appears negative in (1)) flows from the AC to the DC side;
  • if V c phase-leads V s, the active power P s (appears positive in (1)) flows from the DC to the AC side.

For reactive power flows considering δ is usually very small (cosδ≃1)(cos⁡δ≃1)

  • if V c > V s, reactive power Q s in (1) appears positive referring to the converter providing capacitive reactive power to the AC network;
  • if V c < V s, reactive power Q s in (1) appears negative referring to the converter absorbing inductive reactive power from the AC network.

The typical P-Q diagram which is valid within the whole steady-state AC network voltage is shown in Fig 3. It illustrates the real power, P s and reactive power, Q s , capability of a VSC converter terminal, as a function of AC system voltage. Normally converters are sized based on the specific P-Q envelope, Q-V envelope and DC link voltage.

Fig 3: PQ diagram of VSC transmission

With the pulse width modulation (PWM) controlled VSC it is possible to create any phase angle and voltage amplitude (within limits set by the dc-link voltage magnitude) by changing the PWM modulation depth and the relative phase displacement, respectively. This allows independent control of the active and reactive power.

Converter topologies

VSC circuits are based around high capacity power semi-conductor devices (most commonly IGBT modules) which are controlled to switch between the ON (Conducting) state and OFF (Blocking) state. This switch mode operation minimises conversion losses by ensuring that the semi-conductor devices operate in regions of lowest power dissipation. Although semi-conductor losses are low in both the ON and OFF states, there is high momentary power dissipation during the transition between these states. This generates switching losses which in turn limit the frequency at which the devices may be operated.

Current technology can achieve IGBT voltage ratings up to the 6.5 kV range. This contrasts with HVDC transmission where DC line voltages in the order of ±400 kV are required. To achieve these voltage requirements converters must be made up of multiple series connected devices. To-date there are two principal types of converter circuit employed in HVDC.

Two-level converters

Fig 4 shows the circuit diagram for a two-level VSC of the type used in the early deployment of VSC based HVDC transmission systems. High voltage IGBT devices in the top and bottom of the converter leg allow the output to be switched between the positive (+½V dc) and negative (−½V dc) DC poles. The output voltage is thus made up of a series of rectangular pulses of controllable width as shown in Fig 4. By controlling the output pulse width, the converter output can generate a fundamental component with controllable frequency, phase and magnitude.

Fig 4: Two-level VSC circuit and waveforms

In addition to the controlled fundamental, the converter output voltage also contains significant switching frequency components. Power filters are therefore placed between the converter output and the AC network to prevent the injection of harmonic current into the AC network. The use of relatively high switching frequencies means that filter requirements are substantially reduced relative to thyristor based LCC HVDC systems.

To reach the operating voltages required for HVDC, high voltage switches within two-level converters must be implemented as a series combination of many low voltage devices, each of which must be operated such that voltage is shared evenly at all times. Dynamic voltage sharing during switching is particularly challenging by the speed of the IGBT devices where turn-off and turn-on times are in the order of a few microseconds. The need for a series of connected devices imposes a number of conflicting design compromises. Slowing the IGBT switching time can assist with dynamic voltage sharing but will tend to increase switching loss which in turn limits switching frequency. High switching frequency will allow PWM to achieve improved power quality and reduced filter size but will result in increased losses. Electromagnetic interference (EMI) is present in all power electronic systems and presents additional problems for two-level, high voltage VSC, where many devices must switch simultaneously. EMI may be controlled by slowing the switching times which will have a negative impact on switching loss and conversion efficiency.

Multi-level converters

Multi-level converters employ circuits which allow the DC link voltage to be subdivided into a series of switching, each of which is compatible with the rating of a single power semi-conductor device. This eliminates the need for a series connection of devices, device voltages are well controlled and switching speeds can be optimised for reduced losses. Multi-level converters are not restricted to two-level output but can use modulation schemes which can vary both the magnitude and period of the output voltage pulse. Use of multi-level modulation can deliver significant improvements in power quality, filter requirement and reduction in switching loss. The principal challenge for multi-level converters is the provision of intermediate voltage levels. A great many multi-level circuits have been proposed but to date only the modular multi-level converter (MMC) has proved practical for HVDC.

Fig 5 shows the circuit diagram and voltage waveform for the MMC converter. The MMC consists of a number of series connected cells which can be based on either half-bridge [6] or full-bridge circuits [7]. The capacitor of each cell supports a fraction of the total DC voltage and acts as a virtual voltage source that may be switched in and out of the series path according to the state of the cell switches. This allows the use of a stepped modulation scheme, as illustrated in Fig 5. This type of modulation requires cells to switch close to the fundamental frequency, resulting in very low switching loss. Practical MMC employs many cells and can achieve an output voltage that is a close approximation to a sine wave with low harmonic content. Additional power filters, other than the arm inductance, are not generally required.

Fig 5: MMC converter circuit and voltage waveforms

Unlike the DC link capacitor of a two-level converter the MMC cell capacitors contribute to the phase energy transfer between converter and AC grid and therefore experience energy oscillation at twice the AC network frequency. The cell capacitance must therefore be sized to ensure that the resulting voltage ripple is held to an acceptable level. The requirement to limit cell voltage ripple results in a total converter capacitance value in excess of that required by the two-level converter. However, the stepped modulation in Fig 5 shows that different cells are active at different points of the modulation cycle. The result of this is that the loading of cells is not equal. To avoid voltage unbalance between converter cells it is necessary to impose another layer of control which monitors cell voltage and manages the cell rotation such that their average energy transfer and voltage are equalised.

The main difference between cells using half-bridge and full-bridge circuits as shown in Fig 5 is in the behaviour during DC faults. As shown as the dashed red lines in Fig 5, during a DC fault and after the blocking of the IGBTs, AC fault current continues feeding the DC fault points through the diodes of the half-bridge cells. However, for full-bridge cells, the cell capacitor voltages act as an opposing voltage to block any AC fault current from flowing to the DC fault points, in the expense of extra semi-conductor devices and increase power losses during normal operation. Further details on DC fault will be discussed in DC Fault Response.

Offshore DC grid configuration

Existing HVDC connections for offshore wind farms are point-to-point ones, as most of the planned ones so far. However, new connection schemes are being considered using MTDC technology. MTDC grids connecting various wind farms to several onshore stations could provide additional benefits compared with direct point-to-point connections. The following is an excerpt of a wider comparison [8].

In addition to offering a transmission path for offshore power to markets, MTDC provides interconnection capacity if connected to different AC zones (or else superior transmission capacity for the AC zone it connects to).

Connecting geographically dispersed wind farms together through a MTDC will have a smoothing effect on the overall wind generation, which could not be achieved using several point-to-point connections.

Depending on the meshing, MTDC generally provides several possible paths for offshore wind power to shore, even in N − 1 situations.

In addition to ancillary services that can be provided to the onshore AC networks by adequate controls in point-to-point schemes (voltage support, inertia emulation, primary frequency support), MTDC system can offer additional AC congestion reduction by preventively shifting power injections at its ends, or power oscillation damping.

MTDC systems are likely to be developed and built up gradually, either starting from existing point-to-point DC connections linked together via a DC tie, or via different project phases (as planned for the Atlantic Wind Connection project in [9]). A consequence of this process is that the wide variety of layouts in which a MTDC network may result, because of various possible topological extensions and the different development policies. As an illustration for the latter, national codes may differ with respect to the capacity requirements to connect offshore wind farms (fully rated or not, for instance), thus resulting in possible wind spillage and finally, different MTDC requirements and layouts.

Several kinds of MTDC configurations, either tree-like or meshed ones, have to be considered even for a limited number of converters. This is depicted in Fig 6, which illustrates a simple scenario. A simple MTDC shown in Fig 6 b is resulted from linking two point-to-point connections shown in Fig 6 a with a DC tie. More complex layouts could be designed at this stage, and Fig 6 b is the simplest tree-like backbone grid, also referred to as the ‘H’ grid. Assuming the connection of a third wind farm to the ‘H’ grid, a number of different layouts may appear, as shown in Figs 6 cf. The two layouts shown in Figs 6 c and d are fully rated (that is, with a supplementary point-to-point connection which prevents wind spillage) backbone topologies, either tree-like or meshed. On contrary the ones shown in Figs 6 e and f take advantage of the existing assets to connect the new wind farm without developing new transmission capacity for the MTDC system, assuming wind spillage is allowed.

Fig 6: Simple MTDC layouts, starting from a simple ‘H’ grid, to possible extensions for connecting a new wind farm to the existing grid

a Two point-to-point connections

b A simple ‘H’ grid

c Fully-rated extension of the ‘H’ grid: tree-like backbone

d Fully-rated extension of the ‘H’ grid: meshed backbone

e ‘H’ grid extension to connect a new wind farm with possible wind spillage: tree-like

f ‘H’ grid extension to connect a new wind farm with possible wind spillage: meshed

Power flow control for offshore MTDC grids

As for point-to-point connections where one converter controls the DC voltage while active power is managed on the other end, there should be at least one converter station to set the DC voltage reference for a MTDC network. As described in [10], this task can only be allocated to onshore converters, as the offshore ones must control the island frequency and voltage of the offshore wind farms. For security reasons, it is commonly recognised that several converters should be responsible for controlling the DC voltage.

Control during normal operation

A simple and robust way to control the power share of the converters responsible for the DC voltage (and the amount of active power they transmit to the AC networks) is the use of voltage droop control [11], in which active power injections are adjusted by changing DC voltage set points according to linear characteristics between points B and C as shown in Fig 7.

Fig 7: DC voltage droop control including active power and voltage limits

The voltage droop characteristic between points B and C in Fig 7 can be interpreted as follows: intermittent offshore power injections (from offshore wind farms) lead to DC voltage variations. Considering one single onshore converter in charge of the DC voltage control, a voltage falling from V dc0to V dc, (that is: a decrease in offshore active power injected into the DC network) should be compensated by injecting less active power to the onshore AC network (and vice versa in case of an increase of DC voltage). Hence, multiple parallel onshore converters with the same droop characteristic can respond to the fall in DC voltage by decreasing their active power outputs simultaneously. Such decrease will counteract the reduction in DC voltage and the converters will settle the active power outputs and DC voltage at a steady-state point on the droop characteristic. The droop characteristic therefore allows multiple stations to share power using local measurements only, in an autonomous manner without communication.

In addition, the droop characteristics may be complemented to take into account the DC voltage limits and the power limits of the converter (see portions A-B and C-D in Fig. 7). Finally, constant power (or voltage) reference may be desired for some converters, while ensuring their support in case of large DC voltage variations; this feature is provided by combining droop control with some fixed point characteristics as shown in Fig 8.

Fig 8: DC voltage droop control combined with fixed voltage (left) or power (right)

In a MTDC system connecting offshore wind farms, onshore converters are usually the only ones to actively control power flows, via the DC voltage and power droops mentioned above. As they represent a limited number of control variables in the overall system, the ratio between onshore and offshore converters and the meshing of the grid have an influence on the capability to control power flows either for all the branches of the grid, or for some only. This capability, or the lack of it, is hereafter referred to as ‘full’ or ‘partial’ power flow control.

As demonstrated in [12], backbone topologies (including the ‘H’ grid) depicted in Fig. 6 are fully controllable with respect to power flow. Topologies where power flow can only be partially controlled require more stringent needs with respect to a MTDC master control, in order to prevent the grid from any overcurrent or overvoltage. Alternatively, new equipments [13] connected to some branches are proposed to provide additional flexibility on the overall power flow. Those devices can be controlled to modulate the branches resistance, thus shifting power towards the preferred path. As illustrated in [12], a coordinated control of onshore converters (using the droop control) with those devices can lead to significant savings in offshore wind spillage.

Control during onshore AC grid fault

Depends on whether the onshore AC networks connected to the MTDC system are separated or interconnected, an onshore AC fault affects the AC voltage at one or multiple onshore converter terminals. Consequently, the active power exchange between the relevant onshore converters and the grids may be significantly reduced because of the reduced AC voltages. In any case, to ensure safe operation of the MTDC system the generated and transmitted active power must be balanced during such a fault. If the total generated wind power exceeds the maximum power that can be transmitted by the onshore converters because of the reduced AC voltages, DC overvoltage and subsequent shutdown of the whole system could occur if due care were not taken. Under such conditions, it is important that either the wind power is reduced or active power is dumped in separate equipment such that the DC voltage of the MTDC system can be maintained. In addition, offshore AC systems need to be maintained with limited frequency and voltage variations so that when onshore fault is cleared, the system can recover quickly.

For the purpose of active power balancing during potential DC over voltage caused by onshore AC faults, a number of strategies may be employed [3]

  • Fast wind farm output power reduction. This can be achieved by using either fast telecommunication between the offshore converters and the wind turbines [14] or offshore wind farm network AC voltage and/or frequency modulation [15]. In the later case, the offshore AC system voltage and/or frequency is rapidly changed using the offshore HVDC converter when DC overvoltage is detected. Active power from the wind turbines which are specially designed to response to fast AC voltage and frequency variation is then automatically reduced. For such schemes, their reliability can be the main problem.
  • DC dumping resistors on the onshore converters. DC dumping resistors are place on the DC side of each (or selected) onshore converter and they are switched in during DC overvoltage. Thus the wind farms and offshore converters can operate as normal during the fault period with no disruption [16]. However, this method incurs extra hardware cost.

Fig 9 shows an example of using DC dumping resistors on two of the onshore converter DC terminals for dumping excessive power when the onshore converters cannot maintain DC voltage control because of faults on the onshore AC networks.

Fig 9: Example of using DC dumping resistors on onshore converter terminals for a 4-terminal MTDC system

DC fault response

DC faults on VSC based HVDC systems can generate large over-current and rapid collapse of DC voltages because of the discharge of the capacitors on the DC link. This also results in a large increase in AC side current fed to the site of the DC fault because of the presence of freewheeling diodes within each converter branch. Figs 10 a and b show the one-terminal equivalent circuits after the blocking of the converters (IGBTs are switched off) during a DC line-to-line fault for two-level based and MMC (half-bridge) based HVDC systems, respectively.

Fig 10: Equivalent circuit during a DC line-to-line fault after the blocking of the converters

a Two-level converter

b MMC converter using half-bridge cells

Figs 11 a and b show simulation results during a DC line-to-line fault for two-level based and MMC based point-to-point HVDC systems, respectively. Both schemes are rated at 1000 MW at ±400 kV. The DC cable length is 80 km and the fault occurs 40 km from both terminals. In the simulation, the faults occur at 0.08 s and prior to the faults both converters are operated at zero power. As can be seen from Fig 11 a, when the fault occurs, the DC capacitor on the converter terminal discharge quickly and consequently the DC voltage drops to zero within 3 ms in this case. This results in the increase of AC and DC fault current. The high capacitor discharging current i c is transferred to the converter diodes when the DC voltage becomes zero resulting in a large ‘jump’ of the DC fault current i dc at 0.083 s as seen in Fig 11 a. For MMC based system shown in Fig 11 b, the DC voltage collapses to zero in <1 ms because of the small DC capacitance (the MMC cell capacitors are bypassed after the blocking of the converters). Its maximum DC fault current is significantly lower compared with the two-level converter because of presence of the arm inductors and the absence of large DC capacitor discharging current. The presence of the arm inductors also results in considerable oscillations in the DC voltage as can be seen from Fig 11 b. For either configuration, the converter diodes need to be protected from such large fault current though the situation in two-level converter is much severe. In addition, the collapse of the DC voltage results in the complete loss of power transmission and adequate system recovery process needs to be designed to ensure a speedy ramp up of power transmission. The use of reverse blocking MMC based on full-bridge cells as shown in Fig 5 will eliminate any AC fault current though it does not prevent the collapse of the DC voltage.

Fig 11: Simulation results of a two-terminal HVDC system during a DC line-to-line fault on the middle of the DC cables

a Two-level converter based

b MMC based using half-bridge cells

The impact of faults on meshed MTDC networks will be severe, as the DC voltage across the entire system is likely to collapse within a few milliseconds because of the low DC network impedance. Consequently, a complete loss of power transmission capability, affecting both onshore AC networks and offshore wind farms, may be experienced during, and potentially after, faults. It might be expected that DC circuit breaker (DCCB) will be required to operate to ensure that single fault event neither causes disconnection of demand nor an excessive ‘loss of infeed’. To gain the benefits of DCCB in terms of allowing operation of un-faulted DC grid branches to ride-though faults on other branches, extremely fast fault detection and isolation (e.g. <5 ms) is required to minimise the currents interrupted by the DCCBs. However, the DC voltage across the entire system is likely to be severely affected and coordinated re-energisation of the entirely DC network would still be required prior to restoring power transmission.

Conclusion

This article discusses some of the key issues for implementing MHVDC system for connecting large offshore wind farms and power networks. The principles of power flow and control for VSC based HVDC system is introduced and different converter topologies are discussed. Offshore DC grid configurations are described, and power dispatch and DC voltage control using DC voltage droop character are analysed. System control and operation during faults on onshore AC network and DC network are explored for satisfactory fault ride through operation and protection requirement. Significant challenges need to be overcome before large scale MHVDC system can be developed.

References

  1. https://www.gov.uk/government/policies/increasing-the-use-of-low-carbon-technologies.
  2. https://www.gov.uk/government/policies/reducing-the-uk-s-greenhouse-gas-emissions-by-80-by-2050/supporting-pages/carbon-budgets.
  3. Xu L. Yao L.: ‘DC voltage control and power dispatch of a multi-terminal HVDC system for integrating large offshore wind farms’, IET Renew. Power Gen., 2011, 5, (3), pp. 223–233 (doi: 10.1049/iet-rpg.2010.0118).
  4. ‘European offshore Supergrid proposal’ Vision and Executive Summary, Airtricity 2006.
  5. [Online] available at: http://www.scotland.gov.uk/Topics/Business-Industry/Energy/Infrastructure/north-sea-grid.
  6. Lesnicar A. Marquardt R.: ‘An innovative modular multilevel converter topology suitable for a wide power range’. Proc. Power Technical Conf., Bologna, Italy, 2003.
  7. Mondal G. Critchley R. Hassan F. Crookes W.: ‘Design and simulation of a modular multi-level converter for MVDC application’. Proc. IEEE Int. Symp. (ISIE), Poland, June 2011, pp. 200–205.
  8. TWENTIES deliverable D5.4: ‘DC grids: motivation, feasibility and outstanding issues’, available at http://www.twenties-project.eu.
  9. http://www.atlanticwindconnection.com.
  10. TWENTIES deliverable D5.3b: ‘Advanced results of control and protection of DC networks: behaviour, optimisation and grid extension’, available at http://www.twenties-project.eu.
  11. Barker C. D. Whitehouse R. S.: ‘Autonomous converter control in a multi-terminal HVDC system’. Proc. of the IET ACDC Conf., London, UK, October 2010.
  12. Despouys O.: ‘Offshore DC grids: impact of topology on power flow control’. Proc. of the IET ACDC Conf., Birmingham, UK, December 2012.
  13. Veilleux E. Ooi B.-T.: ‘Multi-terminal HVDC grid with power flow controllability’. CIGRE Session 44, Paris, France, August 2012.
  14. Xu L. Andersen B. R.: ‘Grid connection of large offshore wind farms using HVDC’, Wind Energy, 2006, 9, (4), pp. 371–382 (doi: 10.1002/we.185) .
  15. Xu L. Yao L. Z. Sasse C.: ‘Grid integration of large DFIG based wind farms using VSC transmission’, IEEE Trans. Power Syst., 2007, 22, (3), pp. 976–984 (doi: 10.1109/TPWRS.2007.901306).
  16. Harnefors L. Jiang-Häfner Y. Hyttinen M. Jonsson T.: ‘Ride-through methods for wind farms connected to the grid via a VSC-HVDC transmission’. Proc. Nordic Wind Power Conf., Denmark, November 2007.
Go to the profile of Stephen Finney

Stephen Finney

Chair in power electronics, University of Edinburgh

No comments yet.