Protective Selection without Relay Communication

To avoid the use of communication between distant relays, the selection is realised by the following assumptions and measurement allocations.

1) The CBs just connected to one radial cable or a wind farm will operate immediately only when their evaluated distance x is exactly or almost zero. The DC-chopper system across the wind farm DC-link needs to operate to dump all the redundant power. Meanwhile the wind farm needs to be stopped. If x is not zero, it will always wait for a delay time for the primary cable protection to operate and form a possible new power transmission route in the loop.

2) If the CB at one end of the cable detects exactly the cable length D as fault distance, this means the fault has occurred at the DC bus connected to other end of this cable. This CB will trip immediately because the CB near the DC bus cannot block fault current by itself due to the reverse diode of the CB in which reverse current flows. This is the main difference from AC protection, where the CB near an AC bus is the fastest primary protection. However, this depends on the accuracy of distance evaluation algorithm, especially for a ground fault with grounding resistance. Hence an accurate and efficient fault distance evaluation method is required.

3) If the evaluated distance value is negative, that means the fault did not occur on this cable, so it will wait for a delay time (as shown the CB relay point (m) in

Chapter 5 Protection Coordination of Meshed VSC-HVDC Transmission Systems for Large-Scale Wind Farms 139

Figure 5.10, for the fault (flt), v(m) < v(rm) , hence evaluated distance

(

)

)

( − <

=dv_m v_m v_rm

x ).

Three typical fault condition coordination decision examples are shown in detail in Table 5.4 to better describe the decision process. 1) For loop cable fault f3 in Figure 5.1, firstly the evaluated absolute distance range is used to classify all the relay points, where D is the cable length (e.g. 200 km in Figure 5.1). 2) Then for each category, the “+” distance relays are the primary protections because the fault is detected within its protection range. For others in this category, those only connected to a VSC source are in the second order, because if the primary protection CBs fail to trip, it can be seen as a bus fault, where the VSC definitely needs to be tripped. 3) The others in this category are ranked as the third order.

The ordering is carried on until all the categories are sorted to reach a tripping order result. These rules are the same as for a radial cable fault like f1. However, for DC bus fault, it is different, as stated the CB near the faulted DC bus cannot isolate fault current flows through reverse diodes. Therefore if a relay evaluated distance is exactly the cable length D, its CB has the priority to trip primarily, e.g. relay R[4], R[8], and R[12] in Table 5.4 under bus fault f2.

Communication may be needed for CB relays at the same DC bus but because they are physically close communication is practical. A cable ground fault with a large resistance is not as serious as a short-circuit fault and some time delay is permitted. But a ground fault on the DC bus can be precisely detected even without accurate grounding resistance evaluation. For example, it is easy for relay R[2] to identify a bus fault when the ground distance evaluation is exactly the same with R[5], R[6], and R[7]. The cable length inductance ratio with typical grounding resistance can be used as a reference for a fuzzy decision. For instance, evaluated distance from R[4] may not be exactly D, but 1.5D. However, it still needs to trip first as the primary protection. A more accurate and faster DC ground fault location and resistance assessment has been proposed in Chapter 4.

Chapter 5 Protection Coordination of Meshed VSC-HVDC Transmission Systems for Large-Scale Wind Farms 140 Ta ble 5 .4: Pro tectiv e Ord er Selectio n witho ut Relay Commun icatio n

Chapter 5 Protection Coordination of Meshed VSC-HVDC Transmission Systems for Large-Scale Wind Farms 141

5.5 DC Wind Farm Protection Simulation Results

A simulation system of the proposed Supergrid section is modelled in PSCAD/EMTDC. Network parameters of the system are shown in Figure 5.1. The PMSG and VSC parameters are provided in Tables 5.5 and 5.6. A detailed frequency dependent phase cable model is employed in the simulations. The same DC cable π-model parameters in Table 5.2 are used for critical tripping time tc calculations. The

proposed protection scheme is applied to this specific DC wind farm system to show the protection results. The faults simulated are short-circuit faults and ground faults at the three selected points in Figure 5.1 and Table 5.4. After the faults occur, the VSC IGBTs are blocked for self-protection. The per unit power calculation uses 600 MW as base value for each grid-side VSI connected to an AC grid. Finally in Section 5.5.4, the aforementioned cable modelling comparison is also performed on this system for a short-circuit fault.

Table 5.5: PMSG Parameters

Parameter Value Parameter Value

Rated power Pn 300 MW Pole pair no. Pp 100 Rated stator voltage Vsn 99 kV Phase resistance 0.068 p.u. Rated frequency fg 50 Hz Phase inductance 0.427 p.u.

Table 5.6: VSC Parameters

Value Parameter

Wind Farm VSC Rectifier AC Grid VSC Inverter

Rating Power 300 MW 600 MW

DC Voltage ±100 kV ±100 kV

DC-link capacitance 10 mF 20 mF

Choke inductance 18 mH 22 mH

Transformer voltages 99 kV / 96 kV 96 kV / 110 kV