Wind farm connection to an AC network



 

The performance requirements of wind power plants (WPPs) are defined by the respective regulatory bodies for power system operation and are usually described in the form of a grid code. These codes usually define the expected operational ranges of frequency and voltage, the requirements for reactive power/voltage control and active power/frequency control. This article discusses the issues encountered when a WPP is connected to a weak AC network.

The performance requirements must be complied with under all operating conditions which include: operation and isolation under islanding conditions, operation during peak/light load periods, and during generator/network outage conditions. Compliance must be demonstrated both pre and post connection by using adequate simulation studies.

Post commissioning monitoring should also confirm the ability of the plant to operate satisfactorily during and post disturbances (i.e. the ability to ride through disturbances) and to operate satisfactorily in harmony with the other connected generators and dynamic plant installed in the network.

Connection of a WPP in Tasmania

The Musselroe Wind Farm is located at Cape Portland in the far North East of Tasmania, Australia. It is a Class 1A wind site and is subject to predominate westerly winds of the “Roaring 40’s”. Construction of the 168 MW wind farm was completed in late 2013 by Hydro Tasmania.

The wind farm consists of 56, 3 MW wind turbine generators (WTG’s). These are double fed induction generators. The connection to the Tasmanian network required construction of a new 48 km, 110 kV single circuit transmission line from Derby to the wind farm substation with two 33/110 kV 90 MVA step-up transformers. The Connection Point (POCC) at the Derby substation has a very low fault level (360 MVA) resulting in an SCR range at the wind farm of 2.1. The power system and market is operated by the Australian Energy Market Operator (AEMO).

Fig 1: Simplified Wind Farm Model.

To meet the performance requirements of the National Electricity Market (NEM) in Australia, reactive plant and control systems were designed and built inclusive of the following components:

  • 2 x 14 MVA synchronous condensers for fault level (short circuit ratio) support
  • 4 x 4 Mvar STATCOMS for transient, dynamic and steady state reactive control reactive control
  • 4 x 10 Mvar switched shunt capacitors for steady state reactive power support and harmonic filtering.

A simplified model of the wind farm layout is shown in Fig. 1. This configuration was used in simulation software PSSE for dynamic modelling and analysis.

Issues and solutions

This section provides an overview of issues encountered during the design and construction phases of the project that impacted electrical performance of the wind farm.
Rate of change of frequency and anti-islanding protection

An anti-islanding scheme is a regulatory requirement imposed by the Tasmanian Transmission Network Service Provider (TNSP) and the market operator (AEMO) to avoid circumstances where electrical loads in the network become islanded with generating assets. Under such conditions either frequency and/or voltage could be sustained at levels exceeding values specified in the National Electricity Rules (NER) with conventional protections ether ineffective or too slow in acting.

Traditional anti-islanding schemes have used voltage, frequency and rate-of-change of frequency (RoCoF) as triggers to detect an islanded condition. However, these schemes may not always discriminate between low frequency conditions and sustained low frequency due to islanding particularly considering selected wind turbine technology (no frequency control), wide range of wind farm outputs and likelihood of matching demand with generation. To avoid this, an intelligent anti-islanding scheme was developed using a synchrophasor reference methodology. The details of this scheme have been documented in [1].

Another issue with the traditional anti-islanding schemes has been identified and it is caused by reduced inertia in the Tasmanian power system. In the event of credible contingency events (i.e. fault and trip of a generator), RoCoF is seen to be reaching levels that would see traditional anti-islanding schemes trigger on their RoCoF protection. This means there could be a trip of a generator somewhere in the system causing the frequency to deviate at such a rate that would trigger a cascade of tripping of generating assets with traditional anti-islanding schemes.

The issue is currently managed by activation of network constraints in dispatch engine that would force the system to avoid arriving at these scenarios. This has the impact of constraining off generation hence increasing the cost to the market. The devised synchrophasor solution at Musselroe does not suffer from this issue as it does not use RoCoF triggers.

Active power control during fault ride through

The WTGs at Musselroe have a control mechanism for handling fault and voltage depressions. This function when triggered reduces active current and increases reactive current during a voltage depression in order to support the voltage. The current changes are proportional to the voltage magnitude. Therefore the wind farm active power ramps down to a very low level during a voltage depression. The trigger for this operation is a voltage threshold (Fig. 2), that was set quite high at 0,85 pu due to the nature of the Musselroe wind farm’s performance requirements.

Fig. 2: Voltage detection leading to conflicting PPC and WTG controls.

The Power Plant Controller (PPC) is a PLC based central active and reactive feedback power control system for overall wind farm control. The PPC is designed to suspend any control during voltage disturbances. The PPC trigger threshold sensing voltage disturbances should align with the trigger setting on individual WTG’s.

During the occurrence of some shallow faults (high fault impedance) it was observed that the individual WTG actions were triggered on some WTGs while PPC actions would not suspend. This was due to filtering of the voltage signal on PPC. In wind sites with low short circuit ratios, changes to voltage magnitude in both upward and downward directions are very rapid and filtered voltage signal becomes very unreliable when applied to fault detection.

The consequence of this issue was seen when the PPC was trying to apply feedback control to the WTG’s during transients. Multiple active power feedback loops exist in the PPC control with associated delays in their signals. Therefore in transient events, there will always be an error seen by the PPC feedback. In the instances when the WTG’s low voltage triggers were activated but the PPC didn’t “freeze” over all wind farm controls, the WTG’s would try to recover to pre-disturbance active power level post fault and the PPC was issuing varied levels of set point targets due to the two aforementioned issues. This consequently excited active power oscillations post fault.

This issue was resolved by customising the PPC control to be open loop control such that the pre-disturbance set point would be maintained, even if the PPC did not pick up a fault. Fig. 2 shows the PPC missing fault detection.

Availability of accurate models and associated information

Many issues were experienced in the Musselroe project due to ability of the available models to represent the performance of the WPP under different (and uncommon) operating conditions and associated documentation. The NEM in Australia has extremely stringent modelling requirements, particularly for generator proponents. Due to the complex nature of wind turbines themselves, and the strenuous operating conditions they are subjected to, the models have to be reviewed and refined. It was believed that a lot of the refinements needed hadn’t been uncovered in the past but due to the low short-circuit-ratio at the WPP site additional stresses have exposed model limitations.

This was exacerbated by more diligent studies being undertaken on the wind farm performance for that very reason. As the site includes additional reactive plant, the complete wind farm had to be modelled including the dynamic performance of each device as well as the interaction between them which is a requirement for wind farms connected in the NEM. Another NEM requirement is that comprehensive mathematical models have to be produced for the complete wind farm in diagrammatic or block diagram format.

This was an extensive exercise as many manufacturers and equipment suppliers are reluctant to provide detailed information on their plant and associated control due to intellectual property concerns. This makes trouble shooting of issues difficult as well as holds up the NEM approval process if the diagrams are not seen as sufficient by the TNSP or AEMO.

Protection challenges

Many challenges were associated with coordinating the protection systems. One notable issue was coordinating the protection such that adequate performance including mitigation of cascading protection operation was maintained under internal fault scenarios. A complex range of scenarios were explored and modelled in PSCAD and PSSE and an appropriate inter-tripping matrix was devised to cover off all risks and meet performance requirements. A key issue to weigh up was trying to keep as much of the wind farm online to maximise production while still meeting performance requirements.

Impact of constraints on hydro generation

As Tasmanian generation system is predominantly hydro, consideration was given to understand the impact of the wind farm operation on Tasmania’s other hydro generation units. As the wind farm is large relative to the size of the power system, thermal, voltage, transient and oscillatory dispatch constraints within the region were modified to enable the system to operate in its new technical envelop. During high output of wind generation, both local constraints and inter-regional constraints are more onerous and more frequent. These conditions constrain generation from the traditional hydro generating assets impacting the system inertia, fault level, reactive power balance and voltage regulation.

Positive impact on network voltages

Due to the remote location of the connection point (PCC), before the wind farm was connected, the local voltage was difficult to regulate, although this was not considered as a significant issue. Availability of multiple steady state and dynamic reactive power sources supporting the wind farm improve voltages at the connection point (PCC).

A key point in establishing the steady state voltage control was introduction of a sliding voltage set point which changes with active power flow (Fig. 3).

Fig. 3: Voltage – Power control at the WPP.

Synchronous condenser Instability (pole slipping)

As the WTG operation ramps down active power, upon a detection of a fault on radial supply to the WPP, voltage angle separation during a fault would see the wind farm voltage angle deviating away from the network and be offset from its pre-disturbance position once the fault is cleared. As the active power of the wind farm is ramping up on returning from WTG voltage control operation, some of the initial active power rise was seen as a speed rise absorbed by the synchronous condenser which caused a further voltage angle deviation from the network.

Under some very weak grid scenarios where voltage angles are widespread across the network, the voltage angle was seen to be separated by over 360° from the synchronous condensers and the network in this scenario. This would indicate pole slipping of the condensers. The problem was rectified by slowing down the active power return rate from WTG operation such that the voltage angle of the farm could be maintained in synchronism with the system upon recovering from a fault.

Transfer of fault through HVDC

The HVDC interconnector connecting mainland Australia to Tasmania, Basslink, blocks active power flow during faults. A real system event has occurred where a fault on the mainland of Australia saw blocking of active power flow in Basslink and upon fault recovery, the ramp in active power pulled voltages down to a level which caused Musselroe WTGs to trigger fault & voltage depressions handling modes. This could be viewed as a transfer of AC disturbance through a DC link.

Mode cycling

Various scenarios modelled would see mode cycling of controllers. In some scenarios WTG’s were cycling through fault and voltage depressions handling modes and other scenarios show the STATCOMs cycling between transient and dynamic control mode.

The fault and voltage depressions handling mode cycling issue was rectified by modifying control parameters until the performance was acceptance to the TNSP and AEMO and a demonstration that it is extremely unlikely for the issue to occur. The cycling of STATCOM modes was rectified by modifying the control logic to allow for an alternative set of parameters to be applied for a specified time after switching between dynamic and transient control mode.

Acknowledgement

This article was first published in Connection of wind farms to weak AC networks, by Cigré Working Group B4.62 in December 2016 and is republished here with permission.

Reference

[1] Jimmy Chong: A Synchrophasor Approach For Islanding Protection, Transend Networks, Cigré Australia, Australian Panel B5, SEAPAC Conference, Brisbane, 12-13 March 2013.

Contact Prince Moyo, Cigré, Tel 011 800-4659, prince.moyo@eskom.co.za

The post Wind farm connection to an AC network appeared first on EE Publishers.

Source: EE plublishers

More news