Coordinated DC voltage control

As a last step, coordinated voltage control in the layout in Figure 73 is studied. With the assumptions introduced above, the influence of the limited WPP capability on the DC voltage control scheme and in particular on the other elements sharing its burden in the network is discussed.

The relevant figures for the sample simulations shown in this section are reported in Table 14.

The offshore HVDC converter (HVDC OFF) is controlled in V/f fashion based on Figure 12.

Both HVDC2 and WPP are performing a droop control of V²DC according to Figure 11 (with no integral action) and Figure 74 respectively. Their gains Kp13 and KppcDC can have two values based on the cases below:

 Case 1: Kp = 4 pu and KppcDC = 1 pu. In this case, the main DC voltage control burden is shouldered by the converter HVDC2 and ultimately the AC system connected to it. Such case may be considered as the most intuitive approach, as a consequence of the WPP limitations outlined above and the desire to maximise wind power production and hence reduce down-regulation of the WPP during undisturbed operation. Essentially it is assumed that HVDC2 takes up approximately 80% of the steady-state control effort, leaving 20% to the WPP.

 Case 2: Kp = 1 pu and KppcDC = 4 pu. The scenario is the opposite of Case 1, with the WPP taking up most of the control burden in the DC network. In this case, the WPP provides 80% of the control effort, while HVDC2 only contributes with 20%. This situation could possibly become reality if the AC grid connected to HVDC2 was a particularly weak one, in which large power variations would endanger the frequency stability.

Initially, as seen in Table 14, HVDC1 and HVDC2 are evacuating the active power from the WPP according to the scheduled power flow. The base power for all converters and WPP is 1200 MVA. Converter HVDC1 is in this case P controlled in order to easily use it to provoke active power imbalances in the DC grid. The following events are simulated:

13 Kp is the power control proportional gain at the HVDC station – in the block P in Figure 11.

0.9 1 1.1 1.2 1.3 1.4 1.5

0.57 0.58 0.59 0.6

P OFF [pu]

0.9 1 1.1 1.2 1.3 1.4 1.5

1.001 1.002 1.003 1.004

Time [s]

V DC,OFF [pu]

Case 1 Case 2 Case 3

 At t = 1 s, a step in the power reference of converter HVDC1. The power reference is stepped up by 0.27 pu (ca. 325 MW).

 At t = 3 s, converter HVDC1 goes out of service, creating a power surplus in the DC grid of about 0.6 pu (720 MW).

Table 14 - Relevant data for coordinated DC voltage control simulations.

Initial power flow (losses neglected) Control HVDC2 P1,0 [MW] P2,0 [MW] POFF,0 [MW] Type Kp [pu]

395 280 675 V²DC droop 4-1

Control HVDC1 Control WPP

Type Kp [pu] Ti [s] KppcDC [pu] VDB [pu]

P control 1 0.1 1-4 0.0

The simulation results are reported in Figure 76. The left hand side shows the whole simulation, while the right hand side is a zoom-in of the most abrupt event at t = 3 s.

Figure 76 - Time domain simulation results for DC voltage control test. From top to bottom: HVDC2 active power and DC voltage, HVDC OFF active power and DC voltage.

In Figure 76, one can observe the following:

 Expectedly, Case 2 performs significantly worse than Case 1 in terms of limitation of transient DC voltage drifts. This is due to the limitations imposed by the WPP. In Case 2, DC voltage deviations can be twice as large as for Case 1.

 Larger voltage deviations also imply that converter HVDC2 is transiently called to partly make up for what the WPP cannot do. Until the WPP picks up the expected steady-state

0 1 2 3 4 5

Power balance control

share, HVDC2 is constrained to temporarily over- or under-shoot its power level to support DC voltage control though it was designed not to be the main responsible for it.

The acceptability of this ultimately depends on the characteristics of the AC grid HVDC2 connects to.

 The high gain KppcDC in the WPP for Case 2, combined with its poor dynamic performance, gives rise to the resonances observed above (Figure 75). Again, a detailed description of the phenomenon is out of scope here, but it highlights that a certain design and tuning effort is needed to optimise the performance when delayed power sources (WPP) need to contribute to DC voltage control.

(a) Discussion

From the brief analysis above, one can conclude that WPPs limitations challenge proper control of DC voltage in situations where WPPs need to provide a substantial contribution to the service.

Though this scenario seems unlikely, if forecasts concerning development of large, meshed offshore DC grids with massive amount of wind power become reality, there will have to be sufficient available regulation from the onshore converters to provide DC voltage support.

Moreover, WPPs limitations for down-regulation of their power production are only dynamic.

This means that other HVDC converters in the grid could provide supplementary DC voltage control support temporarily, then letting the WPPs reduce their power in steady-state. The performance of such an approach may even be enhanced by dynamically changing control gains at the HVDC stations [143]. From a practical perspective, the demand for power down-regulation in an offshore DC grid may be more common than for up-regulation, since such grid would presumably be a net producer of energy. However, this is just a speculation, and the actual scenario would also depend on who controls such a grid, as well as how.

Generally, it was shown that providing DC voltage control through the whole control chain of a WPP may be dynamically challenging even by brutally neglecting any communication delay and optimising the routing of the control signal in the WPPC. This means that other options (mentioned above) may have to be explored if WPPs are to participate quickly and reliably to the service. Such solutions would probably be based on a more distributed control philosophy, nearer to each converter and potentially eliminating elements that could worsen the performance (e.g.

PLLs) [80], [87]. Nonetheless, such options are not discussed further here, and would require an EMT modelling approach for AC quantities rather than the RMS used here.

Additionally, looking at the dynamic profile of the curves in Figure 76, it can be seen that the power ramp-rates can reach a significant modulus, and this has implications for the WTGs mechanical systems. WTGs provided with FRT capability are usually designed for even larger gradients. However, such design is done considering the FRT event as a very rare one, while fast DC voltage control may have to be provided continuously. From this standpoint, it is certainly best to exclude WPPs from control during small imbalances. A dead-band should thus always be used on WPPs required to do DC voltage control.

6.4 Summary

Participation of WPPs to active power balance control in AC-DC networks was the topic of this chapter. The analysis, conducted through dynamic simulations, aimed at (i) recommendation of control candidate for frequency control provision from WPPs connected to AC grids in a

point-to-point fashion with VSC-HVDC and (ii) analysis of WPPs capabilities to contribute to DC voltage control in DC grids.

Concerning frequency control support from a point-to-point VSC-HVDC connected WPP, according to the results illustrated along the chapter and further considerations, the preferred default solution for its implementation is a scheme based on long-distance communication of the onshore frequency directly to the WPPC. A communication-less scheme was compared to such default solution and results superior in systems where limitation of initial ROCOF is the most vital figure for the frequency control.

As for DC voltage control, it was demonstrated that state-of-art WPPs and their connection to HVDC grids may pose challenges to efficient implementation of the service. WPPs can contribute satisfyingly in steady-state, in the same way as they do for frequency control, but the dynamic requirements of DC voltage control are far beyond what WPPs can do nowadays. The consequence is that either (i) WPPs should take up a very small dynamic share of DC voltage control (first hundreds of milliseconds to a few seconds) and can possibly contribute more widely on a longer term (after a few seconds) or (ii) new more advanced solutions to improve the dynamic performance should be devised, probably making use of converter-nearer controls and a more comprehensive power control scheme spanning over both DC and AC system or (iii) other ways should be evaluated to enhance the available performance from WPPs (e.g. energy storage).