Las filas

The controller software and hardware is assumed to be more reliable. The signals all come to this single point in the system. If this fails, it may cause the whole system to fail or become unstable. It is challenging to modularise the voltage controller and to add levels of parallelism. There can be only one voltage controller in the system with many current/power sources. Two strong voltage sources cannot be easily connected together without stability issues and “fighting”.

Assuming there is more than one outer loop controller on separate circuit boards and only the output of one of them is connected. An error detection algorithm is needed to detect the failure in order to switch. Failure to do so may result in an unstable system. Care must also be taken not to over complicate the system, which may make the system less reliable. It is also assumed that the error detection and control switching system will be more reliable. This part requires further work to understand the robustness of this kind of system.

5.4

Chapter Summary

• When comparing between the open loop and closed loop voltage control cases, the designed voltage controller was able to suppress the resonance in the LC filter that was triggered by current disturbance as found in the open loop case. In the case study, the LC filter resonant frequency was 2000 rad.s−1, the frequency bandwidth of the VSC1 inner loop control (ωni1) was 1428.6 rad.s−1

and the outer loop voltage controller was five times slower than the inner loop. Overall the speed of the outer loop voltage controller was seven times slower than the LC filter resonant frequency and yet it was able to suppress the filter resonance by controlling the filter capacitor to be a constant voltage source. Therefore, the voltage control bandwidth does not have to be faster than the resonant frequency of the LC filter. Constant voltage controlled helps to

5.4. Chapter Summary 155 minimise over-voltage in the offshore HVDC filter capacitor, therefore it may help to minimise filter capacitor tripping events as experienced by Borwin1. • One key advantage of using a closed-loop voltage control with a cascade con-

figuration is being able to limit the current reference in between the inner and outer loop of the controller. This is useful for protecting the converter from overloading. Simulation found that current saturation works for both balanced and unbalanced fault cases, and the control system is able to ride through the fault with low voltage and quickly recover back to the original state after the fault has cleared.

• The current carrying capability of the offshore HVDC converter station may be reduced due to a loss of a module and current saturation can be used to prevent overload in this scenario. If the current is saturated while power is still flowing into the offshore station from the WF, this will cause the AC-side to over-voltage at the filter capacitor. In order, to overcome this issue, the system must signal the WT turbines to reduce their power output.

• System dynamics depend on how the offshore WT power control interact with the network and VSC-HVDC station. For the open-loop method of setting the converter terminal voltage, it is not recommended to have WT control speed too close to the resonant frequency of the station’s LC filter. Further work is also needed to refine power control and compensation of the dynamics in the AC voltage. The power control method used is not perfect but it was enough to study the impact of a change in the power input on the HVDC substation converter (VSC1). If the resonance in the AC side voltage can be compensated, and true power control is implemented, the damping in the system can be improved compared to the case when current control was used in the WF.

• Some feed-forward signals for the controller can be estimated and therefore the sensors for this may be eliminated. However, the voltage controller is complicated and still require a number of sensors for feed-forward and two layers of feedback control, and is not very ideal for control system reliability.

A failure in one sensor signal can undermine the performance of the controller. Modularisation with redundancy for each sensor was suggested to improve sensor reliability. Small signal analyses were carried out to observe the system eigenvalue damping when errors occur in these signals.

Chapter 6

Practical Experimentation

An attempt was made to practically implement and validate the model of the con- trollers for the VSC-HVDC converter station and the aggregated WT converter as described in Chapter 4. This chapter will discuss about the limitations of the avail- able laboratory hardware used and what can be done to solve these issues. Time and funding were limited to continue the work, therefore considerations for further practical experimentation work on the AC voltage controller has also been made.

6.1

The Experimental Rig

The experimental test rig, as shown in Figure 6.1, is a large safety cabinet enclosure to house the converters and all electrically exposed devices, such as the current and voltage transducer boards, protection circuits, and filter circuits. It was designed to operate with a DC voltage source power supply rated up to 600 Vdc and 24 A. Interlocks are in place to isolate everything inside the enclosure to reduce risk of electrical hazard in case of unauthorised access. Cooling fans are also installed to keep everything inside the enclosure as cool as possible. The shelves inside test enclosure are reconfigurable and the total size is big enough to house two converters together.

The three-phase high power resistor load bank shown in Figure 6.2 has switches to discretely vary the resistance per phase. This will be used for initial testing of the converter and control system before connecting with other converters and devices,

Protectionycircuity andyinput/output portsyandyconnectors dSPACEyDS1103 Three-phasey DC-AC Converter LCyfilter Transducers DCyPowerySupply

Figure 6.1: The experimental test rig safety cabinet set-up.

such as a grid simulator.