Frequency Response From Wind Turbines

1st year Transfer Report Frequency Response from Wind Turbines Institute of Energy - Cardiff University PhD Title : System Support from Demand and Generation Flexnet Workstream – Smart, Flexible Controls Supervisors : Dr. Janaka Ekanayake, Prof. Nick Jenkins Student : Ian Moore Date : 19th November 2009

i

Synopsis Background As of October 2009 the UK reached 4 GW of installed wind capacity. By 2012 it is expected that there will be “12GW of wind schemes either operational, being built or already with planning permission” [1]. Any future large installed capacity of wind power will need to provide essential system ancillary services of which frequency response is one such service. Unlike conventional synchronous based generating plant, modern wind turbines due to the nature of their power electronic based interface do not, by default, provide inertial support during the initial stage of a system frequency event. Research Question This PhD project focuses on the following two areas : “What is the best method of implementing primary frequency response from the planned future capacity of wind power ?” “What are the effects on the wind turbine when providing this frequency response ?” Planned Work To address this question the planned work encompasses : • • •

Model based simulation of wind turbines at the power system level Construction of a wind turbine test rig for machine level investigation Experimental testing of system response and machine interaction

This report provides an overview of the initial work conducted in the first year which includes a literature review, basic simulation and test rig construction. Project Methodology Desired system response In order to ascertain what response is required from wind turbines it is necessary to review the response which is provided by existing synchronous based generation plant. It seems reasonable to assume that wind turbines will be required to emulate a response similar to the existing response characteristic of synchronous plant. The current method of how frequency response performance is maintained was reviewed. A review of the UK grid connection code was undertaken. Existing intermittent generation such as wind farms, classified as ‘Power Park Modules’ by National Grid Company, are not required to provide frequency response although the capability must be present. It is noted that sustained secondary frequency response would only be possible by operating a power park at a sustained lower output during normal operation with an associated lower energy revenue generation.

ii

Implementation of response The report includes a review of the basic functioning of power electronic converter schemes and common control methods for FPC and DFIG machines. These were found to be based around vector and/or load angle control of generator-side and grid-side Voltage Source Converters. A simple scheme for primary frequency response by restoration of inertial response was simulated. This was achieved by adding the negative rate of change of system frequency processed via a 1st order delay, to the turbine torque setpoint. This method showed a successful contribution to primary response by extraction of turbine kinetic energy. Paper review discovered another author presenting an alternative method of implementing a modified inertial response which maximised extracted kinetic energy. The simulation undertaken used a high wind penetration scenario of 20GW of wind, operating alongside 40GW of synchronous plant. All loads and generating plant used single lumped equivalent models. Exploration of how a dispersed resource of wind farm capacity can best provide primary frequency response will be investigated to include partitioning the wind turbine capacity into a multi-machine model. Effects on wind turbine machine In addition to providing a response suitable for the power system at the wind farm point of connection, the effects possible on the turbine, of implementing the frequency response must be investigated. Any demanded increase in torque and hence rotor or stator currents, outside of a turbines normal designed operating mode could result in unwanted mechanical stresses and oscillations in the drive train or tower. The following areas will be investigated : • • •

Torque effects/oscillation Aerodynamic effects – may explore the possibility of using ‘GH BLADED’ Converter ratings

Test Rig Development This is now nearing final completion with voltage and current sensors at the assembly stage. Basic inverter operation producing a sinusoidal wave by PWM switching at a low current and voltage was achieved. Further refinements or problems which need resolving are : Over-current shutdown - Suspected noise causing over-current shutdown is expected to be rectified in a second revision of the gate driver board which is under construction for the generator side VSC. Pendulum machine disturbance - Initial problems with unexplained shutdown of the DC Pendulum machine controller were seen to disappear when the mains supply was fitted with a smoothing reactor. However a small disturbance did once again occur (Nov 09) even with the reactor fitted although this was not as severe are previously (machine continued to run). AVR slew rate - The AVR component of the controller appears to have a slew rate limited input which may cause problems in providing accurate voltage regulation of a synchronous generator when connected.

iii

Acknowledgements I would like to thank my supervisors Professor Nick Jenkins and Dr. Janaka Ekanayake for their assistance, direction and encouragement. Also I would like to acknowledge Dr.Nolan Caliao and Ajith Tennakoon for their work in initial testing and commissioning of the wind turbine test rig. Additionally thanks are due to Denley Slade and the staff in the Electronics Workshop for their skills and efforts constructing the test rig and assistance in design. Finally thanks of course to the sponsors FLEXNET

iv

Contents 1 Introduction..............................................................................................................................2 1.1 Project Overview...............................................................................................................2 1.2 Motivation.........................................................................................................................2 1.2.1 Frequency Response Services....................................................................................3 1.2.2 Frequency Response from Wind Turbines.................................................................3 1.3 Objectives..........................................................................................................................4 2 Wind Turbine Generator Systems............................................................................................5 2.1 Energy Extraction.............................................................................................................5 2.2 Modes of Operation..........................................................................................................7 2.3 WT design features..........................................................................................................8 2.3.1 Fixed speed versus variable speed.............................................................................8 2.3.2 Power Control............................................................................................................8 2.3.3 Electro-mechanical coupling......................................................................................9 2.3.4 Mechanical Coupling.................................................................................................9 2.3.5 Other design features.................................................................................................9 2.4 Generator Types................................................................................................................9 2.4.1 Induction Generator...................................................................................................9 2.4.2 Doubly Fed Induction Generator.............................................................................10 2.4.3 Full Power Converter...............................................................................................11 2.5 Power Electronic Converter Fundamentals.....................................................................12 2.5.1 Converter Classification...........................................................................................12 2.5.2 Bridge Configurations..............................................................................................12 2.5.3 Square-wave Switching Scheme..............................................................................13 2.5.4 Pulse Width Modulation Switching Schemes..........................................................16 2.5.5 Back to Back Frequency Converter.........................................................................16 2.6 Electrical Machine Control............................................................................................18 2.6.1 Load Angle Control Theory.....................................................................................18 2.6.2 Vector Control..........................................................................................................19 2.7 Wind Turbine Generator Control Schemes.....................................................................22 2.7.1 Full Power Converters.............................................................................................22 2.7.1.1 Generator Side Control...................................................................................22 2.7.1.2 Grid Side Control............................................................................................25 2.7.2 Doubly Fed Induction Machines..............................................................................27 2.7.2.1 DFIG Control Scheme ...................................................................................27 2.7.2.2 Other DFIG Control Schemes.........................................................................28 3 Connection Requirements & Response Capability................................................................29 3.1 UK Requirements for grid connection............................................................................29 3.1.1 General.....................................................................................................................29 3.1.2 Steady State Reactive Power and Voltage Control..................................................30 3.1.3 Fault Ride Through Capability................................................................................31 3.1.4 Power System Stabiliser and Black Start Capability...............................................32 3.1.5 Frequency Response................................................................................................32 3.1.6 Reserve.....................................................................................................................34 3.2 Desired Response............................................................................................................35 3.2.1 NGC Benchmarking of Plant FR capability............................................................35 3.2.2 Performance of Synchronous Plant..........................................................................35 3.2.3 System Requirements for FR...................................................................................36 3.3 Primary Response Capability from WTs........................................................................37

v

3.3.1 Primary Frequency Response Schemes...................................................................37 3.3.2 Machine Effects & Converter Current Limits..........................................................41 3.4 Secondary Response Capability from WTs....................................................................43 3.4.1 Wind Farm Control Scheme....................................................................................43 3.4.2 Frequency Controller...............................................................................................44 4 Modelling...............................................................................................................................45 4.1 Introduction.....................................................................................................................45 4.1.1 Equations of Motion.................................................................................................45 4.2 System Model.................................................................................................................47 4.2.1 Synchronous Plant Response...................................................................................48 4.2.2 Reduced Order Machine Model...............................................................................48 4.3 Control Scheme...............................................................................................................49 4.4 Simulink Model...............................................................................................................49 4.5 Setup................................................................................................................................50 4.6 Results.............................................................................................................................50 4.6.1 Open Loop Wind Turbine Response........................................................................50 4.6.2 Closed Loop Wind Turbine Response.....................................................................52 4.6.3 Wind Turbine and Synchronous Response..............................................................52 4.7 Simulation setup and results summary table...................................................................53 4.8 Discussion of results.......................................................................................................55 5 Experimental Wind Turbine Test Rig....................................................................................56 5.1 Overview.........................................................................................................................56 5.2 Design.............................................................................................................................58 5.2.1 DC Motor/ Pendulum Motor – Block A..................................................................58 5.2.2 Generator Machine – Block B.................................................................................60 5.2.3 Back to Back PWM converters – Block C & D.......................................................62 5.2.4 Generator and Grid Side Controller – Block E & F.................................................66 5.2.5 Power System – Block G.........................................................................................68 5.2.6 General Assembly and Connection..........................................................................68 5.2.7 Setup.........................................................................................................................69 5.3 Results.............................................................................................................................69 5.3.1 Measurement and Open Loop Control Test.............................................................69 5.3.2 Bridge Inverter Test.................................................................................................72 6 Further Work..........................................................................................................................75 6.1 Risks................................................................................................................................76 6.2 Gantt Chart......................................................................................................................76 7 Appendices.............................................................................................................................78 7.1 Simulations......................................................................................................................78 7.1.1 Simulation Baseline Record.....................................................................................78 7.1.2 Setup.........................................................................................................................79 7.1.3 M-Files.....................................................................................................................80 7.1.4 Model Parameters....................................................................................................87 7.2 Laplace Transformation..................................................................................................89 7.3 Experimental Test Rig....................................................................................................91 7.3.1 Procedure for Use.....................................................................................................91 7.3.2 Equipment Specifications........................................................................................92 ..........................................................................................................................................97 7.3.3 Hardware Design......................................................................................................98 8 References............................................................................................................................104

vi

List of Abbreviations ADC – Analogue to Digital Converter ASIC – Application Specific Integrated Circuit AVR – Automatic Voltage Regulator DAC – Digital to Analogue Converter DFIG – Doubly Fed Induction Generator DSP – Digital Signal Processor FPC – Full Power Converter FR – Frequency Response K.E – Kinetic Energy IG – Induction Generator IM – Induction Machine MOSFET – Metal-oxide-semiconductor Field Effect Transistor MPPT – Maximum Power Point Tracking PM - Permanent magnet ROCOF – Rate of Change of Frequency RSC – Rotor Side Converter SG – Synchronous Generator SVM – Space Vector Modulation Sync – Synchronous uP - Microprocessor VSC – Voltage Source Converter WT – Wind Turbine

1

1 Introduction 1.1 Project Overview This report concentrates on the theory, simulation and planned experimental implementation of Frequency Response (FR) from Power Converter based WTs which are currently the favoured choice of technology for wind power plant. This PhD forms part of the work of the FLEXNET project which is investigating the design, operation and optimisation of the future electricity system for the UK. For experimental work a 1kW laboratory based wind turbine test rig is being constructed at Cardiff University for practical implementation of wind turbine power electronic controls and evaluation of other novel machine/plant control schemes.

1.2 Motivation Increasing worldwide demand for Energy, worries about depletion of existing fossil fuel reserves, nuclear proliferation, economic security and, by no means least Climate Change, are some of the main reasons for the planned expansion of the wind power sector in the UK and worldwide. For the UK which has a privatised electricity system, whether market based methods are used or central planning, as is the case in some nationalised industries, uncertainty still exists in regard to the future make-up of the electrical energy supply. Some features of a future UK energy sector are likely to be : • • • •

Variability of output from the expansion of new Renewable sources such as wind Inflexibility of any future baseload Nuclear plant Declining indigenous Gas reserves Impetus to reduce CO2 emissions particularly from coal

As with some other forms of Renewables such as Solar, Wave and Tidal these forms of Energy ‘harvesting’ are stochastic in nature and cannot be dispatched with the same ease as conventional generating plant such as Coal, Gas and Hydro. Also of note is that much of these Renewables are of a distributed nature and thus are of a smaller unit size and are geographically dispersed. Thus in general there is an impetus in the industry for requirements for more flexibility from plant in terms of participating in provision of energy output, reactive power and additional ancillary services due to this new power system topology. For the UK in particular, the planned large scale expansion of wind power is necessitating research into how this relatively new type of plant can be successfully integrated into the electrical power system. Displacement of conventional plant under a large penetration of wind will require some ancillary services to be provided from WTs, of which FR is one such service.

2

1.2.1 Frequency Response Services

Frequency (Hz)

Figure 1.1 demonstrates what happens when a Frequency ‘event’ occurs on the electricity network. These rare events when there is a sudden unexpected large deficit between load and generation which can result in an unwanted drop in system frequency (potentially causing system collapse). Immediately after the step imbalance, stored inertial energy in spinning machinery begins to be consumed by the generators helping to maintain their electrical output. After a short delay in response, conventional synchronous plant contracted to have spare reserve power begin to open steam governors and thus increase the generator power output. This then restores the frequency on a temporary basis until ‘secondary’ slower response plant can come ‘on-line’.

50.2

continuous service event 10 s 30 s O

60 s time

49.8 49.5 49.2

occasional services X primary secondary response response

to 30 min

Figure 1.1 – Modes of operation of frequency control on the UK grid [2] For maximisation of energy capture and thus increased energy revenues, WT plant in the UK would participate in only the ‘primary’ response shown in the figure above. This short term response would only affect the energy output of the machine for up to 30 seconds.

1.2.2 Frequency Response from Wind Turbines Fre que ncy (H z)

Key technical aspects to investigate in the research area can be identified as below : •

Converter based turbines have no natural inertia response, thus access to this stored K.E will need to be enabled

•

The quantity of WTs on the system will vary at different times, thus management continuous service and dispatch of this response will be likely different to that from conventional event 50.2 Synchronous plant 10 s 30 s 60 s

•

O time plant will again The positive power primary response currently given by Sync 49.8 need to be replicated in some form by WTs 49.5

3

occasional services

49.2 X

primary response

to 30 min secondary response

•

Capability of WTs to provide this service, dynamic effects on their machines and loads on the turbine from providing this response.

1.3 Objectives This project aims to contribute to the subject of investigation of provision of FR from power converter based wind turbines by : a)

Modelling the GB system FR and obtaining/developing/refining a suitable control method capable of an appropriate response on the Power System.

b)

Undertaking experimental testing of such a control scheme in order to verify its functioning and discover some of the practical implementation issues/characteristics.

c)

Explore further Turbine system issues and machine effects such as aerodynamic effects, possibly using proprietary WT software simulation tools.

These objectives are further elaborated in the chapter ‘Further Work’.

4

2 Wind Turbine Generator Systems The majority of WT capacity in existence today consists of machines in the range of 500kW to 2MW, horizontal axis machines having two or three blades. These types of WTs are the most common and are still the favoured choice for development with the trend being towards even larger machines [3].

2.1 Energy Extraction The theoretical maximum quantity of energy that can be extracted from a horizontal flow of wind can be calculated as below : The available energy in a wind stream is given by

P = 1 2 ρ AU 3 air where ρ is the air density, U is the wind speed, and A is the area swept by the wind turbine blades. However, the energy which can be extracted by the wind turbine is less than the available energy in the wind and is given by

P = C P m p air Where Cp is called the power coefficient and depends on the tip speed ratio ρ which is the ratio between the velocity of the rotor tip and wind speed defined by

λ =

ω rR U

where ωr is the aerodynamic rotor speed and R is the radius of the rotor. Cp has a theoretical maximum of 0.59 (known as the Betz limit) but will be typically up to 0.4 for a commercial 3-bladed turbine in operation. For a given wind turbine design there is an optimum value of tip speed ratio λopt, which gives the maximum power extraction. Figure 2.1 shows a typical performance curve for a modern high-speed wind turbine, the maximum efficiency of the turbine of approximately 0.45 occurring at a tip speed ratio of 7.

5

Figure 2.1 – Variation of coefficient of power with tip speed ratio [4] A principle cause of the variation of the efficiency of power extraction of the blade assembly is due to the variation of the angle of attack α of a fixed blade with the incident wind. These components are shown in Figure 2.2 where U is the wind perpendicular to the turbine axis, W is the apparent wind relative to the rotating blade and β is the blade pitch angle. At a low tip speed ratios the blade is in a stall condition, at higher tip speed ratio the blade has a low angle of attack and drag effects predominate, both of these effects thus causing less than optimum power extraction [4] .

U

Figure 2.2 – Velocity components in the plane of a blade cross-section [4] If the operational speed of a turbine rotor is allowed to change in order to keep the tip speed ratio more or less constant an increased quantity of energy can be captured compared to a fixed speed machine. shows the set of torque speed curves which define the performance characteristics for a wind turbine machine for different wind speeds. As expected higher wind speeds result in a higher combination of torque and rotor speed hence giving a higher power output. This product of torque and rotor speed gives rise to the maximum power curve shown.

6

Generator torque[Nm]

14000 12000 10000 8000

Speed limit Aerodynamic torque B A Curve for maximum power

12 m/s

C

11 m/s

6000

10 m/s 9 m/s

4000

8 m/s 7 m/s

2000

6 m/s

Rotor speed [rpm]

0 0

500

1000

1500

2000

2500

3000

3500

Figure 2.3 –Torque speed characteristics of a variable speed wind turbine [5]

2.2 Modes of Operation Practical design constraints mean that the WT must be constrained from operating above a maximum torque and maximum rotor speed. Both of these limits, and also the electrical machine rating, coincide approximately with the rated wind speed of the WT which normally is around 12m/s. Additionally there is a minimum wind speed below which operation of the WT provides no benefit. Referring to Figure 2.4, these operating modes for a variable speed type WT can be summarized as follows : Below cut-in : Machine is not rotating and produces no power output. Max Power Tracking : B to C - Electrical torque is controlled such that the machine speed tracks the maximum power curve. Constant Speed : A to B and C to D – Due to power converter constraints at low speed and aerodynamic noise constraints at high speed, torque is allowed to vary but speed is held almost constant [6] Pitch Control or Stall : D to E - Speed and torque are both limited to their maximum values by adjustment of the blade pitch angle. This stalling or feathering of the blade reduces the torque produced. If no pitch mechanism is provided, stalling of the blades at high wind speeds can be achieved through suitable blade design. Shutdown : Above a certain speed the WT is brought to a halt to avoid damage.

7

Generator Torque

D

E

Rated Torque Shutdown Speed

C

Cut-in Speed

Speed Limit

B

A Generator Speed

Figure 2.4 – Operating modes for a WT For a fixed speed WT the modes would be the same with the exception of the maximum power tracking mode which would instead be replaced with a constant speed operational mode.

2.3 WT design features 2.3.1 Fixed speed versus variable speed Fixed speed – This type of turbine is designed to operate within a relatively narrow shaft speed range which is determined by the variation in slip of the induction generator connected to it. This might be up to 1-2% at rated output [7]. With changing wind speed and an essentially constant shaft speed, the tip speed ratio will change and thus the turbine will not operate at its optimum power production point except for a single wind speed. Variable speed – By allowing a more constant tip-speed ratio the turbine can be operated to extract more power from the wind. There is a trade-off between complexity and this extra performance. This design requires a either a full power converter design to allow the shaft speed to vary independently of grid frequency, or alternatively a partial power converter design (DFIG) which allows a greater range of slip than a conventional IG machine. Typical slip ranges achievable with this configuration are -40 to +30% [8] A slight variation of the variable speed design is a wound rotor induction machine with a variable resistance rotor. A commercial implementation of this is known as Opti-slip manufactured by Vestas. This design can allow variations in speed of typically 0-10% above synchronous speed [8]. Notably this design avoids the necessity for slip rings and their associated maintenance by incorporating all of the switching components on the rotor.

2.3.2 Power Control Fixed blades with appropriately designed aerodynamic profiles or ‘pitching’ blades are used to reduce the input mechanical shaft torque under excessively high wind speed conditions.

8

2.3.3 Electro-mechanical coupling Direct grid connection of a synchronous machine driven by a wind turbine rotor is not possible as the relatively rigid electrical coupling would give rise to high mechanical stresses and also unwanted variations in electrical output during aerodynamic disturbances. These might include tower shadow effect. Hence IG based machines which have inherent damping due to their slip operation, or alternatively full converter interfaced machine connection topologies are used which are fully decoupled via a dc-link.

2.3.4 Mechanical Coupling Geared drives – In order to use industry standard 4-pole electrical machinery or similar a stepup gearbox is necessary. A typical 500kW 40 meter diameter turbine has a rotor speed of 33 r.p.m [4] Direct drive – By using a large diameter machine with multiple poles a ‘gearless’ drive arrangement is possible. This has reliability advantages as gearboxes have proved to be a common failure in early turbines. These down-time and maintenance issues are especially relevant for offshore turbines.

2.3.5 Other design features Output voltage ratings are typically 690V with a step-up transformer located close-by. Newer designs are beginning to use higher generator output voltages.

2.4 Generator Types The below topologies represent conventional designs typically implemented in horizontal 2 or 3 bladed versions common today from 250kW to 5MW.

2.4.1 Induction Generator An induction Generator (IG) based WT is shown in Figure 2.5. This turbine operates in a similar fashion to large scale induction motors which are commonly found in industry. Similarly these WTs are equipped with power factor correction through capacitor banks to compensate for reactive power consumption and a soft-starter to reduce in-rush currents on start-up. Benefits – Low cost, low maintenance squirrel cage rotor construction Disadvantages – Reactive power consumption during faults, possibility of overvoltage in islanding condition, less than optimum power extraction capability (varying tip-speed ratio), essentially uncontrolled i.e no control of real or reactive power.

9

Softstarter

Capacitor bank

Squirrel-cage induction generator

Figure 2.5 - Induction Generator based turbine [9]

2.4.2 Doubly Fed Induction Generator A Doubly Fed Induction Generator (DFIG) based WT is shown in Figure 2.6. By using a wound rotor and a bi-directional part-scale converter this machine is able to operate over a range of shaft speeds. In addition to real power control, terminal voltage can be controlled through reactive power import and export capability. Control of this reactive power via the rotor is effectively amplified and it is possible to use only a partial scale converter located here to achieve the same reactive power control as a full scale one located on the stator [7].

Wound rotor induction generator

Power Converter

Crowbar

Figure 2.6 – Doubly Fed Induction Generator [9]

10

2.4.3 Full Power Converter A full power converter (FPC) based WT is shown in. By inserting a full bridge converter and inverter between the electrical generator and the grid, complete rotational decoupling of the turbine and grid system is accomplished. There is freedom to use an IG or SG machine. Also if the generator requires no magnetizing current as in the case of a permanent magnet based SG, the converter can be a simple diode bridge rectifier.

Induction/Synchronous generator

Power converter

Figure 2.7 – Full Power Converter based WT [9]

Benefits – Complete control of output to grid side and hence ease of compatibility with grid connection requirements. Dc-link provides damping for torque oscillations caused by varying windspeed. Disadvantages – Additional cost of power electronics

11

2.5 Power Electronic Converter Fundamentals Both the DFIG and FPC type of WT employ a power electronic assembly to convert from ac to dc and back to ac. For the DFIG this synthesised ac waveform is used principally to enable control of the rotor magnetisation current and hence overall control of the machine and power export through the directly connected stator. For the FPC all of the power generated must pass through the power electronic assembly. To obtain the complete ac to ac conversion two individual converters are connected ‘back to back’ via a dc-link.

2.5.1 Converter Classification The general name ‘Converter’ is given to a circuit which can operate as both an inverter and a rectifier although the term inverter is commonly used in its place [10]. Two basic types of converter exist namely the current source converter and the voltage source converter. Current Source Converter This requires a stiff dc current source ideally with infinite Thevenin impedance at the input. It can be constructed from a variable voltage source with feedback current control and a series inductor. Asymmetric voltage blocking devices such as MOSFETs are not suitable for use in a CSC, instead devices such as thyristors must be used [10]. Voltage Source Converter or Voltage Fed Converter (VSC) This converter configuration uses a stiff dc voltage source at one side and converts it to ac at the other, the Thevenin impedance of the source ideally being zero, a capacitor can be added to assist in this [10]. This type of converter is commonly used for WT power electronics applications.

2.5.2 Bridge Configurations Depending on harmonic noise requirements, component cost considerations etc it is possible to choose between a number of bridge configurations for the synthesis of 3-phase ac. Two-level inverter This is the simplest topology and is shown in Figure 2.8. It consists of three pairs of half bridges a,b and c with a total of 6 semiconductor devices which are indicated in the figure as simple switches. Multi-stepped inverters These use a greater number of switching devices to obtain a more accurate sinusoidal output, an advantage being the reduced size of ac and dc filtering.

12

DC - Link Upper

Va

a

Vdc b Lower

c

Figure 2.8 – 6-Pulse three phase inverter bridge

2.5.3 Square-wave Switching Scheme In its simplest method of operation, fabrication of a three phase output to the 3-phase balanced load shown in Figure 2.8 is done as follows. This method is called square-wave switching : General Rules •

Switches are either fully ON or OFF (this reduces switching losses)

•

No two switches in the same half bridge can conduct at the same time. This would be a short circuit of the DC supply with destructive currents and is known as ‘shoot-through’. For prevention of this scenario it is common practice for switching logic to incorporate a ‘deadband’ period which enforces a delay period between alternation of switching of the upper and lower bridge devices.

•

When switching OFF a device, a path must be made for the conduction of the inductive current in order to prevent overvoltage across the device and hence destruction of the semiconductor switch. ‘Freewheeling’ diodes provide the path for this current decay, ready for the reversal of direction of the current into the opposite bridge leg.

Simple switching scheme •

Identical drive signals are sent to all of the bridges but are phase shifted by 2π/3 (this creates the 3 separate phase voltages).

13

•

Since we are generating a.c signals current is expected to be always non zero so the legs will switch alternately between upper ON and lower OFF and lower ON and upper OFF.

Hence this scheme is implemented using the following sequence indicated in Table 2.1. This pattern alternately applies either +Vdc or –Vdc to a single phase in series with the other two phases in parallel. Thus allowing conduction of current through the upper device or the lower device and producing an alternating current. Voltages developed across the phases are : •

legs in parallel will have 1/3 of Vdc

•

single conducting leg 2/3

•

upper conducting phases have positive voltage

•

lower conducting phases negative voltages. Table 2.1- Switching states and applied voltages for square-wave scheme Half Bridge States

Va

Vb

Vc

Ua Ub Uc

(of Vdc)

(of Vdc)

(of Vdc)

Comment

La Lb Lc 0

0

0

0

0

0

1

1

1

0

0

0

OR

Three phases at same potential. No freewheel path. Invalid states.

1

1

1

0

0

0

0

0

0

0

0

0

All ON

0

0

0

All OFF 1

0

1

0

1

0

1

0

0

0

1

1

1

1

0

0

0

1

0

1

0

1

0

1

0

1

1

1

0

0

0

0

1

1

1

0

Destructive short circuit current. Invalid state. Bridge idle

1/3

- 2/3

1/3

Valid State

2/3

- 1/3

- 1/3

Valid State

1/3

1/3

- 2/3

Valid State

-1/3

2/3

- 1/3

Valid Stat

-2/3

1/3

1/3

Valid State

-1/3

- 1/3

2/3

Valid State

14

The resulting applied voltage waveforms from this scheme are shown below in Figure 2.9. Note that the upper gate drive switching signals are the same wave form shape as the voltages developed Va0, Vb0 and Vc0. The lower gate drive signals are the inverse of these. Va0 is the voltage of the mid point bridge of leg a with respect to a fictitious midpoint on the DC supply voltage. Va is the phase voltage of phase ‘a’. PI +0.5 Vdc

Phase a

Va0 -0.5 Vdc

2 PI /3 +0.5 Vdc

Phase b

Vb0 -0.5 Vdc

+0.5 Vdc

Phase c

Vc0 -0.5 Vdc

Vdc Vab

Vdc Vbc

Vdc Vca

2/3 Vdc Va 1/3 Vdc

Figure 2.9 – Switching waveforms and applied voltage for square-wave scheme

15

Observations that can be made on this switching sequence are : •

Either the upper or lower switch must be ON and the other OFF as conduction in one direction or the other is always required to fabricate the alternating current.

•

Since we have 3 phases the resultant overlap means either 2 upper devices and 1 lower or vice-versa 1 top device and 2 lower devices must always be active. This gives the resultant applied voltage to the phases as shown in the table.

2.5.4 Pulse Width Modulation Switching Schemes An improvement on the square-wave switching method is to use Pulse Width Modulation. This technique modulates the length of the ON pulse and thus provides voltage control of the synthesized output. In its most basic format a switching duty cycle is output depending on the desired ON-OFF time ratio. The frequency of repetition of this duty cycle is determined by the ‘carrier wave’ frequency. More advanced PWM based schemes can produce a greater accuracy in sinusoidal output with reduced harmonic content. Two popular methods are listed below : SPWM – Sinusoidal Pulse Width Modulation is a common technique which varies the output pulse width in proportion to the magnitude of an internally generated sine-wave reference signal. This scheme will synthesize a good approximation of a sine-wave when an inductive load is connected to an inverter bridge. The switching signals and output waveform of such a SPWM scheme are demonstrated in Figure 5.69. SVM – Space Vector Modulation utilises the concept of a rotating space vector. Although computationally intensive one of its benefits is that it can ‘optimise the harmonic content’ of an isolated 3-ph neutral load. This is of relevance to machine loads as these are often do not have the neutral connected [10].

2.5.5 Back to Back Frequency Converter Connection together of the dc sides of two 3-phase VSCs completes a circuit known as the back-to-back frequency converter shown in Figure 2.10. This circuit enables frequency to vary independently on both sides of the inverters and also power can flow in either direction. In addition to its application in variable speed WT systems it is widely applied in industrial drive applications and is known in its most flexible configuration as a 4-quadrant variable speed ac drive.

16

Figure 2.10 – Back to back frequency converter An observation of this circuit is that when equipped with MOSFET or IGBT devices, due to the ‘body diode’ on each transistor the generator side (left hand side of Figure 2.10) switches are able to freewheel and thus act as a diode bridge when power is required to flow from the 3-phase to the dc-link. The grid side switches of course require appropriately modulated gate drive pulses in order to transfer current to the 3-phase grid side connection (right hand side of Figure 2.10).

17

2.6 Electrical Machine Control Two control methods which commonly appear in inverter bridge control schemes for use with electrical machines are load angle control and vector control. For vector control use of the d-q reference frame is necessary. These two control methods and the d-q reference frame are discussed below.

2.6.1 Load Angle Control Theory This method seeks to control the flow of power through an inductive element by manipulating the angle between the voltage at the sending and receiving end sources. The principle is the same as controlling the flow of power through a predominantly inductive power system element such as a transmission line. In the case of a VSC, a coupling reactor provides the inductive element across which the phase angle is measured. The real power flow is determined by the angular difference between the voltages. The reactive power flow is determined by the difference in magnitude of the voltages. For the circuit (a) and phasor (b) shown in Figure 2.11

VS

jX

VR

A) i

Sending Source

Receiving Source A) VS jXI

B)

δ VR

φ I

Figure 2.11 – Load angle power transfer (a) circuit and (b) phasor diagram

The relationship between power flows, angle and voltages can be derived as below :

Complex power SS

 V − VR = VSIS* = VS  S jX 

  

*

18

(2.1)

 VS* − VR*  − jX = VS 

(2.2)

VS = V S ejδ and VR* = VR

Since then

 VS2 VS VR*  − j  =j X X

VS2  V V cos δ + jVS V R sin δ  − j S R  SS = PS + jQS = j X X  

(2.3)

Therefore PS =

VS V R sin δ X

(2.4)

QS =

VS2 VS VR − cos δ X X

(2.5)

where δ is the load angle, φ is the power factor angle, VS is the sending end voltage, VR is the receiving end voltage and X is the inductive reactance between them. The steady state active and reactive power flow equations (2.4) and (2.5) form the basis for this control method whereby the voltage magnitude seen at the receiving end VR and the load angle δ are controlled to provide the required real and reactive power flows. For a VSC placed at the receiving end, the PWM reference sine wave and duty cycle can be adjusted to give the required load angle and magnitude independently of the generator at the sending end.

2.6.2 Vector Control In order to implement a vector based control scheme it is necessary to convert the 3-phase rotating quantities of a machine into a stationary reference frame in two dimensions. A brief explanation of this is given below for an IG although the basic principles are equally applied to a SG machine also. The per-phase equivalent machine circuit shown in Figure 2.12 for an IG is only valid for steady-state conditions. For high performance control of a machine a controller based on equations derived from this model is not sufficient.

19

Figure 2.12 – Equivalent circuit model of induction generator [11] An IM can be considered to be a ‘transformer with a moving secondary, where the coupling coefficients between the stator and rotor phases change continuously with the change of rotor position [10]’. It is possible to accurately model the machine to include the varying inductances but the differential equations involved would make it too complex to implement as a control system. Hence some form of simplification is required. A 3-phase machine can be represented as an equivalent 2-phase machine with ds, qs being direct and quadrature stator components and dr, qr being direct and quadrature rotor components. These however still have time varying components. Parks Transformation - R.H.Park in the 1920’s solved this problem by replacing the stator voltages, currents and flux linkages instead with variables rotating at a synchronous speed in a fictitious rotor winding. This is known as ‘Parks Transformation’. This transformation eliminates time varying inductances. Later H.C Stanley and then Krause and Thomas ‘showed that time varying inductances could be eliminated by referring the stator and rotor variables to a common reference frame which may rotate at any speed (arbitrary reference frame)’ [10]. d-q reference frame – This final transformation leads to all the machine variables being described in the d-q reference frame as shown in Figure 2.13 . The derivation of these equations is too complicated for this report however the reader is referred to [12] and [13].

Figure 2.13 – dq representation of an induction machine [9]

20

Vector Control (Field Oriented Control) - The basic idea behind vector control (which is also known as field oriented control) is to force an IM to behave as a separately excited dc brushed machine. By doing this electromagnetic torque can be made nearly instantaneously equal to a demanded torque. Permutations of this control include stator flux-oriented, rotor flux-oriented and air gap flux-oriented methods, of which all of these can be indirect or direct methods. Direct and indirect methods refer to the necessity or not, for use of sensors to measure air gap flux linkages [12]. Rotor Flux Oriented Control – This is one example of the possible vector control methods and its application to an IM. In this control scheme the rotor flux linkage vector is kept perpendicular to the rotor current vector. The vector diagram for this scheme is shown in .

Figure 2.14 – dq representation of stator and rotor currents with rotor flux orientation [7]

This alignment is accomplished by setting the q-axis rotor flux ψqr to zero and secondly by setting the d-axis rotor current idr to zero. In terms of practical implementation, a result of this control method is that : • •

‘torque control’ of the machine is determined by the q-axis stator current iqs ‘flux control’ (for control of voltage) is determined by ids.

Some coupling does exist between these d and q control terms which can be compensated for in a practical control scheme.

21

2.7 Wind Turbine Generator Control Schemes Examples of vector control schemes and load angle control schemes for WTs are shown in this section. In all of these schemes the dc-link voltage is assumed to be regulated at a constant voltage by managing the flow of real power from the generator to the grid by an appropriate means. If this real power is not transferred there is a danger of the WT overspeeding or the dc-link voltage rising to an excessively high value. This is a particular problem during grid fault conditions when low grid-side terminal voltage can prevent export of real power. Control of back-to-back converters can be split into grid-side and generatorside control functions.

2.7.1 Full Power Converters

2.7.1.1 Generator Side Control For an IG or a Synchronous machine the generator side converter control can be realised either as load angle control or vector control. Control schemes presented here export power to the dc-link to ensure the turbine tracks its maximum power curve. Generators which have no reactive power transfer capability such as PM synchronous machine can use a simple uncontrolled diode bridge rectifier for the generator side converter. Load Angle Control – Figure 2.15 shows the implementation of this control strategy [14]. The desired power Pref is determined by the turbine maximum power point look-up table. The internal generator voltage eG is calculated from the rotor base speed ωb and the rated voltage EG. Manipulation of these values gives inputs to the PWM control of angle control αG and terminal voltage control vG.

ωr

P

pref

αG = ω

pref xG eG vG

vG

r

eG =

ωr EG ωb

αG

vG =

vds =

2vG sin α G

vqs =

2vG cos α G

PWM control

eG2 − qref xG eG

Figure 2.15 – Generator side load angle control strategy for FPC

Vector Control for Synchronous Generator – A generator side vector control strategy is shown for a synchronous based WT in Figure 2.16. This uses a d-axis defined along the flux linkage vector ψm. In practical terms this means that if the flux linkage vector ψm is known, the torque of the machine can be controlled by the q component of the stator current iqs. Correspondingly control of the d component of the stator current ids exerts control of the reactive power production of the generator. The two PI controllers scale the current demands into suitable d and q components of voltage, for the VSC PWM control signals.

22

pGE vqs

SYN Generator

VSC vds

vds

Flux control in d-axis

+

KP +

+

+

−

Swing equation

Tin − Te = J

ids eG iqs ψ m

eG −

ids − ref

P 2 ωr

iqs

ω X d ids ωb

ids eG

iqs

+ +

K KP + I s

d ωm dt

ωm

ids

ω X q iqs ωb

vqs

KI s

Te

Tin

− +

iqs − ref

1 Te − ref . k ψm

Torque control in q-axis

ψm

ωr Te − ref

Figure 2.16 – Generator side vector control strategy for a synchronous machine [14] Vector Control for PM Synchronous Generator - For a permanent magnet machine the direct stator current reference ids in Figure 2.16 can be set to zero as no reactive power is transferred. Vector Control for Induction Generator – The generator side vector control strategy for an induction generator based WT can be implemented by selecting the d-axis to align with the rotor flux. Then by regulating the d axis stator current ids air gap flux is controlled and by regulating the q axis stator current iqs torque is controlled. Such a control scheme is shown in Figure 2.17. Since this is an asynchronous machine the VSC carrier frequency must be set according to the desired slip s and actual rotor speed ωr as calculated.

23

Figure 2.17 – Generator side vector control for induction machine [9]

24

2.7.1.2 Grid Side Control The grid side control scheme is required to transfer the power incoming from the generator side VSC, out to the grid side. This ensures a constant dc-link voltage. Secondly the grid side VSC must provide reactive power transfer as appropriate to its agreed grid connection. An FPC connected to the grid is shown in Figure 2.18. Real power to transfer is Pgr and reactive power to transfer is qGR. An inductive reactor Xgr is connected between the converter and the WT terminal outputs, the terminal output voltage being v1

Figure 2.18 –FPC connected to grid [9] Load Angle Control – A control scheme to maintain correct real power export (and hence regulate dc-link voltage) and also terminal voltage control is shown in Figure 2.19.

vdc + v DCref−

KP +

−

KI s

pGRref−

f ( pGR − ref , vs , θ

)

θ

g ( qGR − ref , vs , θ

)

vGR

vs

vsref −

+

−

K KP + I s

qGRref−

Figure 2.19 – FPC grid side load angle control [14]

25

Dc-link error and terminal voltage error are passed through PI controllers to produce demanded values for real and reactive power transfer. These commanded values pass through appropriate function to give the load angle setpoint θ and the converter terminal voltage vGR. As the grid connection is an infinite bus vs cannot be changed. Vector Control – This method seeks to control the grid side PWM converter by manipulating the VSC output VGR and the terminal voltage VS in the dq reference frame. The scheme is shown in Figure 2.20. Control of real power and reactive power are implemented by q axis and d axis currents respectively.

vdc −

KP +

vDC − ref

vs − ref +

vqs

+

−

KI s

K KP + I s

iq − ref

id − ref

R-F Trns

R-F Trns

+

−

iq

Xs

id

Xs

+

vs

−

K KP + I s

+

+

K KP + I s

vqGR

−

+

+

vdGR

+

vds

Figure 2.20 – FPC grid side vector control [15] Vector Control with AVR controlled dc-link - A slight variation on the basic vector control method presented is when a wound rotor synchronous machine uses its AVR to maintain the dc-link voltage. In this case the dc-link error summing point with PI control in the upper loop of Figure 2.20 is replaced by a maximum power point reference and power summing point instead [16].

26

2.7.2 Doubly Fed Induction Machines A property of DFIG machines is that transfer of reactive power to the network is possible via the rotor side converter or the stator side converter. The advantage of reactive power transfer via the rotor is that it is magnified by a factor of 1/s. The converter is also capable of transferring real power to the rotor from the network when in sub-synchronous operation and from the rotor to the network in super-synchronous operation [7]. The schematic for a typical DFIG and its overall control scheme is shown in Figure 2.21.

Figure 2.21 - Schematic of DFIG WT [11] The crowbar protection shown operates by shorting the rotor through a resistance to protect the rotor from over-current in a fault condition.

2.7.2.1 DFIG Control Scheme An example of a commonly implemented scheme is detailed as below. Generator Side Control – The generator side controller provides control of stator power ps though torque control and additionally controls stator reactive power qs. Similar to vector control implemented for FPC machines, by splitting the controlled current into the two orthogonal components in the dq reference frame, individual control of torque and terminal voltage is facilitated. In the case of the DFIG, the rotor currents idr and iqr are controlled as opposed to the stator currents ids and iqs of a synchronous or singly fed induction machine. This type of controller is also known as PVdq or current-mode control [7]. Implementation of the torque control loop is shown in Figure 2.22. Torque demand Tc from the maximum power point curve (not shown) is summed with an optional term which provides synthesis of the inertial action found in synchronous machines. The resulting torque demand is converted to a rotor quadrature reference value whereby it is then summed with the actual current value. This is then output through a PI control, summed with a decoupling term and finally passed to rotor side VSC (not shown) as a q-axis voltage setpoint. Note that in this figure, ς is the Laplace s constant.

27

Figure 2.22 – DFIG rotor side torque control loop [11] Implementation of the voltage control loop is shown in Figure 2.23. The upper part of the input to the first summing block represents the generator magnetising component of the idr current. The lower part represents the component controlling reactive power flow with the network [Cartwright]. After passing through a delay term, then summation with measured idr a PI control converts the demanded current into a d-axis voltage term. After adding a decoupling term the demanded Vdr is passed onto the VSC PWM input (not shown).

Figure 2.23 – DFIG rotor side voltage control loop [11]

Grid Side Control - The grid-side controller maintains the dc-link voltage by import or export of real power.

2.7.2.2 Other DFIG Control Schemes Rotor Flux Magnitude and Angle Control (FMAC) – This scheme adjusts the magnitude and angle of the rotor flux vector. It has an advantage of low interaction between the voltage and power loops and has good system damping and voltage recovery after faults [7]

28

3 Connection Requirements & Response Capability 3.1 UK Requirements for grid connection In order for plant operators to connect their generators to the electricity network whether at distribution or transmission level the plant must comply with what are known as ‘Grid Codes’. These specifications which are issued by the respective power system operator are necessary so that plant behaves appropriately under all known operating conditions and also abnormal conditions. For wind farms it is noted that the requirements apply at what is known as the ‘Point of Connection’ where the aggregated power enters the network rather than the individual WT terminals themselves. The following sub-sections concerning the UK gridcodes are taken directly from [17]

3.1.1 General The connection of new generation in Great Britain is governed by the Grid Codes of National Grid plc [17]. A collection of non-synchronous generating units that are powered by an intermittent source, joined together by a system with a single electrical point of connection (may include a DC Converter) to the GB transmission system is categorised as a “Power Park Module”. The Grid Code only applies to a Power Park Module such as a wind farm, not individually to power park units (i.e. individual wind turbines). Almost all the performance requirements that are mandatory for the power park module are applicable to modules installed in England and Wales with a completion date on or after 1 January 2006. However, the performance requirements applicable to power park modules in Scotland vary and the Codes are under continual review. Therefore it is recommended to refer to the most up-to-date Grid Codes. For a Generating Unit or Power Park Module using an intermittent power source, the requirement is that the active power output shall be independent of system frequency for system frequency changes within the range 50.5 to 49.5 Hz and should not drop with system frequency by greater than the amount specified in Figure 3.24 for system frequency changes within the range 49.5 to 47 Hz.

100% of Active power output

95% of Active power output

47.0

49.5

50.5

Frequency

Figure 3.24 - Requirement placed on the output power of a generating plant in terms of frequency [17]

29

At the point of connection the active power output under the steady state conditions of any Generating Unit, DC Converter or Power Park Module directly connected to the GB Transmission System should not be affected by voltage changes in the normal operating range, that is ±5% continuously or ±10% for 15 minutes for 400 kV, ±10% continuously for 275 or 132 kV and ±6% continuously for less than 132 kV.

3.1.2 Steady State Reactive Power and Voltage Control Conventional synchronous plant is required to control the voltage and also absorb or generate reactive power, in accordance with the needs of the power system. Normally the transmission system operator determines the operating settings of these generators. All Power Park Modules (excluding those connected to the total system by a current source dc converter and those connected at 33kV or below) must be capable of supplying rated MW output at any point between the limits 0.95 power factor lagging and 0.95 power factor leading. With all plant in service, the reactive power limits defined at lagging and leading power factor as a function of the active power output are defined in Figure 3.25. These reactive power limits will be reduced pro rata to the amount of plant in service. The Power Park Modules are also required to provide continuously acting automatic voltage control system to provide control of the voltage and operation of the plant without instability over the entire operating range of the plant. The automatic control system shall be designed to ensure smooth transition between the shaded area bounded by CD and the non-shaded area bounded by AB in Figure 3.25.

Point A is equivalent (in MVAr) to: 0.95 leading Power Factor at Rated MW output Point B is equivalent (in MVAr) to: 0.95 lagging Power Factor at Rated MW output Point C is equivalent (in MVAr) to: -5% of Rated MW output Point D is equivalent (in MVAr) to: +5% of Rated MW output Point E is equivalent (in MVAr) to: -12% of Rated MW output Figure 3.25 - NGC plc reactive power requirement [17]

30

3.1.3 Fault Ride Through Capability Each power park module and any constituent power park unit shall remain transiently stable and connected to the system without tripping for a close-up solid three-phase short circuit fault or any unbalanced short circuit fault on the GB transmission system operating at voltages of 200 kV or above for a total fault clearance time of up to 140 ms. In this case, (a) during the period of the fault each power park module shall generate maximum reactive current without exceeding the transient rating limit of the generating unit or power park module and/or any constituent power park unit. (b) each power park module shall be designed such that upon both clearance of the fault on the GB transmission system and within 0.5 seconds of the restoration of the voltage at the grid entry point active power output shall be restored to at least 90% of the level available immediately before the fault. shows the typical fault recovery for cases with two circuit breakers and three circuit breakers (see top right hand corner of each diagram for the configuration).

Figure 3.26 - Typical fault recovery for two-ended and three-ended circuits [17] For voltage dips of greater than 140 ms, each power park module and any constituent power park unit shall remain transiently stable and connected to the system without tripping for balanced voltage dips and associated durations any where on or above the solid line of Figure 3.27. In this case,

31

(a) provide active power output, during voltage dips at least in proportion to the retained balanced voltage at the grid entry point except in the case of a asynchronous generating unit or power park module where there has been a reduction in the intermittent power source in the time range in Figure 3.27 that restricts the active power output below this level and shall generate maximum reactive current without exceeding the transient rating limits of t the generating unit or power park module and/or any constituent power park unit. (b) restore active power output within 1 second of restoration of the voltage to at least 90% of the level available immediately before the fault.

Voltage as a % of nominal

Voltage duration

Figure 3.27 - Minimum voltage dips above which generators should be stable and connected [17]

3.1.4 Power System Stabiliser and Black Start Capability The requirements for excitation control facilities, including PSS can be agreed when signing the bilateral agreement. The GB Grid Code states that black start capability is agreed at a number of strategically located power stations.

3.1.5 Frequency Response According to the Grid Codes each power park module must be capable of operating in a manner to provide frequency response at least to the solid boundaries of Figure 3.28. Each power park module must be capable of providing some response, in keeping with its specific operational characteristics, when operating between 95% to 100% of registered capacity as illustrated by the dotted lines in Figure 3.28. If the frequency response capability falls within 32

the solid boundaries, the power park module is providing response below the minimum requirement which is not acceptable. The capability profile specifies the minimum required level of primary, secondary and high frequency response. The definitions of these responses are based on the curves shown in Figure 3.29. The phrase “Minimum Generation (MG)” applies to the entire power park module operating with all generating units synchronised to the system. The Designed Minimum Operating Level (DMOL) is the output at which a power park module has no high frequency response capability. It must be less than or equal to 55% of the registered capacity.

Figure 3.28 – Minimum frequency response profile for a ±0.5 Hz frequency change [17]

33

Figure 3.29 - Interpretation of Primary, Secondary and High Frequency response [17]

3.1.6 Reserve The power park modules are not obliged to provide reserve. However, the power park modules can participate for providing fast reserve or short term reserve under an ancillary services agreement or under a bilateral agreement. Definition of the two terms fast reserve and short term reserve are as follows : (a) Fast Reserve - provides the rapid and reliable delivery of active power through an increased output from generation or a reduction in consumption from demand sources, following receipt of an electronic dispatch instruction from National Grid. (b) Short Term Operating Reserve - Short Term Operating Reserve (STOR) is a service for the provision of additional active power from generation and/or demand reduction. The STOR service could be a committed service or a flexible service.

34

3.2 Desired Response In order for continued operation of a future power system with a varying mix of wind power and conventional synchronous plant, an assumption is made that any future wind capacity will need to provide a primary response of a format similar to that required for existing synchronous based generating plant. An important characteristic of primary response is that it should be “released increasingly with time, through automatic governor action, in the period 10-30 seconds after the incident and sustained for a further 20 seconds.” [2].

3.2.1 NGC Benchmarking of Plant FR capability For a generating plant participating in FR services a typical response characteristic to a change in frequency is shown in Figure 3.30

Figure 3.30 – Typical plant frequency response characteristics [2]

The upper graph shows a negative system frequency deviation of 0.8Hz after 10 seconds and 0.5Hz after 60 seconds. The sample at 10 seconds being representative of a point in the primary response period (0 to 30 seconds) and the 60 second point being representative of a point in the secondary response period (30 seconds to 30 minutes). The process of benchmarking a generating plant is outlined in [2]. The middle graph shows a suitable frequency test waveform which might be input to a generator system. The lower graph shows the positive power response which would be expected from a generating plant.

3.2.2 Performance of Synchronous Plant By taking the response of plant to differing frequency deviations a response profile can be constructed as shown in Figure 3.31. The Minimum Stable Generation (MSG) and Generator Registered Capacity (GRC) represent the limits of operation of the generator. As can be seen part-loaded plant has a proportionally greater capability for an increase in plant output when system frequency drops. For an increase in system frequency, highly loaded plant has a greater capability for reduction in plant output. Larger deviations in frequency show an increased quantity of response as expected. 35

Figure 3.31- Typical Genset Frequency response profile [2] These response profiles are of a similar format to minimum frequency response profiles as specified in grid code requirements e.g Figure 3.28. It is noted in [2] that due to the nature of ‘air breathing engines’ such as gas turbines, a drop in frequency (i.e synchronous shaft speed) may effect such a plant’s capability to output a positive power response.

3.2.3 System Requirements for FR The quantity of frequency responsive generation on the system will need to vary depending on the system demand. As demand drops a greater quantity of FR plant will need to be online. For the primary response phase for the UK system this is shown in Figure 3.32. If the ‘largest credible generation loss risk’ occurs, which is 1320MW, a similar quantity of response to cover this loss is still needed to control the frequency deviation. Additionally since there are less loads on the system any characteristic contribution from load reduction due to frequency drop will be less. Other factors which contribute to this include low frequency load tripping and speed of response of contributing generators. A similar need for increasing response for lower system demand also occurs for the secondary response service [2].

Figure 3.32 – System frequency response requirements [2] 36

3.3 Primary Response Capability from WTs 3.3.1 Primary Frequency Response Schemes Individual WT modelling Figure 3.33 shows an overview of an example WT model used in frequency response modelling. Included is a wind model, a two-mass model of the rotating components, and a high level turbine controller principally for limitation of power capture at above rated speeds consisting of a blade pitch angle control block.

Figure 3.33 – Wind turbine model for frequency modelling [18] Auxiliary Signal for Inertial Response One simple method used by [19 ,20] for FR is shown in . By introducing a proportional and 1st order delay term acting on the system frequency signal, a modified Torque/Power signal is produced. This signal is combined with the existing torque setpoint of the machine which finally results in a modified current setpoint.

Figure 3.34 – Block diagram of auxiliary torque/power signal for DFIG inertial response [20]

37

The differential gain term provides a torque in proportion to the rate of change of frequency similar to the torque occurring in a synchronous machine due to inertial energy release. The delay term provides ‘shaping’ of the response. The performance of a 1.5 MW DFIG turbine with this FR loop, connected onto an 8MVA stand-alone diesel system is shown in the graphs in Figure 3.35 for a disturbance of -0.15p.u [20]. The solid line indicates the response with the WT FR loop included, the dashed line is with no response from the WT. With the FR loop included the lower left graph shows a constant slip angle during the frequency disturbance thus indicating that the turbine rotor speed is now ‘coupled’ back to the system frequency. In the right-hand middle graph, the improvement in performance of the system is evident from the reduction in ROCOF and also the reduction in frequency excursion magnitude.

Figure 3.35- DFIG (at 100% power) response to a frequency disturbance with (solid) and without (dashed) supplementary inertia effect [20]

38

Identical Inertial Restoration A further refinement of the ‘delayed df/dt’ method is possible, in order to obtain an identical SG type response, by dynamic modification of the gain value used for K. If slip is forced to stay constant, because this is a variable speed turbine the K.E extracted for a particular change in frequency will vary depending on the operating speed of the turbine. By suitably scaling gain value K to depend on speed, the response loop can be shown to give a constant FR contribution regardless of variations in operating speed [20]. Maximum Energy Extraction Algorithm A new ‘stepwise’ algorithm is presented in [20] which combines a stepwise increase in electromagnetic torque followed by an adjustable ramp-down period. The profile for this torque response is shown in Figure 3.36(b). Its operation can be described as follows : •

•

Onset of frequency event – After detection of a drop in system frequency at time t0, an increase in torque is commanded ∆T. This results in an increased power output for the machine hence a reduction in system ROCOF. The increase in electrical torque above the aerodynamic torque results in turbine deceleration at a rate proportional to the difference in torque and the machine inertia value. Gentle rampdown and avoidance of stall condition – Stall is the operation of a blade when a low lift to drag ratio occurs [4]. To avoid operation of the turbine in such a low efficiency condition during the primary response phase a lower limit of operational efficiency is set. In this algorithm it has been chosen to be at point ωcrit shown in Figure 3.36(a). Time t1 and t2 are chosen such that time t2, the end of the wind turbine contribution to the frequency response phase, will coincide with the chosen ωcrit. A method is shown in the paper to perform this calculation.

The size of ∆T, on set of the ramping down, end of the response time t2 and quantity of K.E extracted could be adjusted to suit the response required. In the paper ∆T was set at 20% and the ramp down length (t2 – t1) to 10 seconds.

Figure 3.36- (a) Variation of aerodynamic torque with generator speed for wind speed = 11.5 m/s (rated power). (b) Stepwise torque method for FR. [20] Results of this algorithm are shown in Figure 3.37. A sustained positive power response of approximately 13 seconds is produced. At 15.6 seconds the electro-mechanical torque and the aerodynamic torque become equal and the deceleration of the machine ends. 39

Figure 3.37- Performance of stepwise method for maximum K.E extraction [20] Comparison of Schemes The performance of three methods of inertial response from DFIG WTs is undertaken in [20] and is presented in Figure 3.38. •

• • •

Power Control - dotted line. ‘Power control’ which limits the power increase to 10% and the net exchange of active power to zero. This is thought to be a scheme which would be attractive from a power system operator’s perspective, although it was noted that this scheme might cause turbine stall. Inertial response – Dash dot. This is the simple ‘delayed df/dt’ scheme using a proportional gain and delay on a frequency input to produce a modified torque demand. ‘Step ramp’ algorithm – solid line. No response from WT – dashed line.

As can be seen in the right hand graphs the step-ramp algorithm as configured shows the greatest speed decrease out of all of the methods and provides a significant reduction in frequency excursion.

40

Figure 3.38- Comparison of primary FR methods. Dashed, no support. Solid, Step-wise 20% torque increase. Dash dot, Inertial response (K=39, T = 0.1). Dotted, Power control (10% step and net exchange is zero. [20]

3.3.2 Machine Effects & Converter Current Limits Regulation Parameter Choice The effects of using different machine parameters for inertial regulation purposes are highlighted in [20]. Differences between using Pstator, Tel, Ptotal and the effect on rotor power and possible stalling of the turbine are discussed. Additional effects of introducing MPT to the overall control and its anti-stalling effect is shown. Converter Over-currents Reference [20] draws attention to the over-current limitation of the DFIG’s RSC and hence the FR capability of the machine at different operating points. This is demonstrated in Figure 3.39. The upper left graph shows the limited capability to provide a positive power response when the turbine is operating at 80% of rated power. A 0.25 MVA RSC is used on the 1.5MVA turbine in the simulation.

41

Figure 3.39 - Influence of DFIG initial loading to the delivery of inertia effect. Dash-dot, rated power. Dashed, 80% loading. Solid, 40% loading. [20]

42

3.4 Secondary Response Capability from WTs Although the primary objective of this work is provision of primary response, it is of interest to consider wider wind farm control and provision of reserve as when implemented they may have closely related control and communication aspects.

3.4.1 Wind Farm Control Scheme A wind farm controller which includes FR functionality is presented in [18]. This is shown in Figure 3.40. Individual WTs communicate their available power with the central controller and receive dispatched power and voltage setpoints according to the Voltage Monitor and Frequency Controller.

Figure 3.40 – Wind Farm Control Strategy [18]

‘Five principle tools’ to deal with FR/reserve are identified in [18] as : • • • • •

Absolute Limit – Capping of maximum power output. Ramp Limit – Restriction in rate of rise or fall of power. Balance Control – Similar to conventional generator control in providing a known ramp up or ramp down capability. Delta Control – Tracking of maximum power output capability but at a constant percentage below e.g actual production could be set at a constant 5% below available production to provide spinning reserve. Maximum Export Limit (MEL) – This is an upper limit on power export defined for UK generators. This limit can be changed by the transmission operator to ensure appropriate levels of FR are in place.

43

3.4.2 Frequency Controller A WT controller is presented in [18] which has capability of implementing Delta, MEL or Balance control respectively. Depending on grid frequency and power available from the turbine an appropriate reference power for the turbine converter is produced. The results for a multi-megawatt FPC equipped induction machine WT for a negative and a positive frequency disturbance using MEL control are shown in Figure 3.41. Wind speed input, turbine speed, pitch angle, available power output and actual power output are shown for the three control methods possible. Wind speed of 12 m/s is used along with a turbulence intensity of 18%.

Available Power

Output Power

Figure 3.41 – MEL Control [18]

[B-H] concludes that rapid power response for participation in FR is possible from a large FPC induction machine based WT and aggregation of outputs from a dispersed resource of WTs would be expected to smooth the outputs seen in the results. Note that the model used does not feedback into a system frequency model.

44

4 Modelling 4.1 Introduction Method - The modelling here is undertaken by transferring all of the mathematical relationships which describe the system into a set of s-domain transfer functions. This common procedure enables the solution of differential equations by standard algebraic methods [21]. Matlab-Simulink is used to solve these equations and also provides a useful graphical interface to describe the model and then view the results in the time-domain. Assumptions – For simplification and speed of simulation all of the loads are lumped together as one single mass. This amalgamated representation is also the case with the response from the sync plant and the wind turbine plant. Note that future simulation work is planned with a ‘partitioned’ response. Simulation Objectives – The primary objectives of the simulations presented here in this chapter are to : • • • •

demonstrate relationship between frequency and power on a simplified UK system show how this frequency is regulated by sync plant and additionally WT plant verify that the model produces the same results as given in [19] present a baseline model to be used for investigation of more advanced WT response

Note - As in common with most power systems modelling all units are expressed in p.u quantities unless otherwise stated

4.1.1 Equations of Motion The imbalance between electrical torque Te and mechanical torque Tm and the relationship between these and rotational inertia are the fundamentals of load-frequency studies. These are derived below for a single machine [13] During an imbalance between power and load the net accelerating torque Ta is : Ta = Tm - Te

(4.1)

where Tm , Te are positive for a generator (N.m)

The combined inertia is accelerated accordingly and gives rise to the ‘Swing equation’: J

dω m = Ta dt

(4.2)

J is moment of inertia (kg.m2), ωm angular velocity (rad/s)

For power system studies inertia is normally given in terms of the per unit inertia constant H which is defined as : H=

K.E VAbase

=

0.5 Jω 02m VAbase

(4.3)

where ω0m is rated angular velocity (rad/s)

45

Substituting for J in the swing equation : 2H

d  ωm  dt  ω 0 m

 Tm − Te  =  VAbase / ω 0 m

(4.4)

Because Tbase = VAbase / ωm the p.u equation of motion can be expressed as : 2H

dϖ r = Tm − Te dt

where we define ω r as

(4.5)

ωm ω 0m

Dispensing with the ‘m’ subscript we can define ωr and ω0 as the angular velocity and rated angular velocity respectively in rad/s. Including a component of damping torque proportional to speed deviation we get : 2H

dω r = T m − T e - KD∆ ω dt

(4.6)

r

Rearranging to obtain the acceleration : 1 dω r = ( T m − T e - KD∆ ω 2H dt As ∆ ω

r

= ω r- ω

0

r

)

(4.7)

then

dω r d (∆ ω r ) dt = dt By substituting for ω

r

(4.8)

we can now express our swing equation in final form as :

1 d (∆ ω r ) = ( T m − T e - KD∆ ω 2H dt

r

)

(4.9)

For load-frequency studies the preferable quantities to analyse are Power and frequency as opposed to Torque and frequency [13]. For a small deviation (delta) from initial values (subscript 0) and all values in p.u : P = ωr T P = P0 +∆P T = T0 + ∆T ωr = ω0 + ∆ωr

46

P0 + ∆P = (ω0 + ∆ωr) (T0 + ∆T)

(4.10)

Assuming that the product of ∆ωr and ∆T is comparatively small then : ∆P = ω0 ∆T + T0 ∆ωr

(4.11)

Also since ω0 = 1 and if ∆ωr is small then ∆P = ∆T Hence for small speed deviations, generator power Pgen and load power Pload we can express the swing equation as : 1 d (∆ ω r ) = ( P gen − P load - KD∆ ω 2H dt Note that because of p.u quantities ∆ ω system.

r

r

)

(4.12)

is of course identical to ∆ f on a synchronous

4.2 System Model Figure 4.42 shows equation (4.12) implemented as ∆P model in the s-domain along with the plant response which forms a closed loop feedback system. Transfer of the equation on to the s-domain is shown in the appendices. From an initial steady state operating point, disturbances are applied to the system and the effects observed with differing plant frequency response controls. Sensitivity of loads with respect to a change in frequency is given by the system damping D. Stored inertia and therefore initial rate of change of frequency for a power disturbance is dictated by Heq, the equivalent combined inertia of the system (not including any additional synthesised wind turbine inertia). Synchronous governor response is determined by the combined transfer functions of the Droop, Governor and Turbine transfer functions which are shown in the next section. Wind turbine response is given by inserting the model of Figure 4.44 in to the system model. Both the wind turbine response and synchronous response represent aggregated models of the individual plant on the system.

Disturbance

∆f setpoint =0

Sync Gov. Response +

∆P

+ +

1 D + 2Heqs

Wind Turb. Response

Figure 4.42 - GB delta power inertial model 47

∆f

4.2.1 Synchronous Plant Response This response consists of the expected ∆P increase from all of the synchronous plant on the system. The quantity of this plant and its scheduled level of response (i.e it’s droop) is determined through the setting the gain in the ‘Governor Droop’ block. The below figure is taken from the Simulink model and is of the form of a generating unit with a reheat steam turbine [13]. The simulation model uses a time constant of 12 seconds for the Turbine Re-heat [19].

-1 1 1 In 1

w r_ d

-K G o v e rn o r D ro o p

1

1

2s+1

0 .2 s + 1

0 .3 s + 1

12 s+1

G o v e rn o r

T u rb in e

T u rb in e R e -h e a t

1 O ut1

P d_sync

Figure 4.43 – Steam Turbine Transfer function

4.2.2 Reduced Order Machine Model

Tdem

X1

Kp + Ki/s

X2 [1 + sT1]

Tshaft X3

∆T -

Frequency Response Control Block

+ +

+

G(s)

+

∆Tf_ctrl

∆f

-

Manageable simulation of the power system for frequency studies requires use of a reduced order aggregated wind turbine model. For the wind turbine plant the linearised ∆T model used is shown below in Figure 4.44[19]. This model is accurate for small changes in torque about an initial operating point. Parameters used for this model are calculated using the data for a 2MW IG machine given in appendices Table 7.3. Note this machine model can also be used to represent a DFIG with appropriate parameter selection.

η

x

Tsetp

ωr

1 .∫ J

Max power lookup curve

0.6 p.u -

Initial Torque

+

0.6 p.u

Initial Aero Torque

Figure 4.44 - Linearised ∆T wind turbine model

48

ωr

∆P (2MW)

4.3 Control Scheme By increasing the Torque set point of the wind turbine in response to a frequency deviation a short term increase in electrical power output of the machine can be achieved. Supplementary control to enable this inertial/fast primary response action is shown in Figure 4.45. The combination of the transfer functions below provides an increase in commanded power output in response to the change of frequency. The synthesis of the inertia is provided by the gain k1 acting on the rate of change of frequency deviation signal. Appropriate adjustment of this gain can give replacement of the inertia. Depending on the gain chosen the turbine can be made to done one of the following [19] : •

make the rotor speed track the system frequency

•

provide a set quantity of inertia independent of rotor operating speed (this is the ideal ‘like for like’ replacement of Synchronous plant inertia

•

customised synthesis of inertial response as chosen by the system operator

Additional shaping of the response is provided by the ‘washout’ filter in combination with gain k2 and also the first order delay block using Tf in combination with gain kf. These additional first order delay terms extend the overall response towards the end of the primary response phase.

k1 ∆f

Tf_ctrl

k2

Figure 4.45 - Wind turbine frequency response control block

4.4 Simulink Model The model used to produce the results in this report is shown in Figure 4.46 and is based on the Matlab mdl file used in [19]. The complete model showing subsystem components is shown in the appendices. The only difference between the models is that the ‘No of m/c’ gain block uses a different value and to compensate for this, a gain block needed to be added before the ‘freq sig lag’ switch. More detailed comments about the differences between these models are available in the appendices.

kf Tf s + 1

49

G B S y s w ith F a s t P R fr o m W in d Ia n M o o r e - 1 4 A p r il 2 0 0 9

w_d

G B _ fa s t _ P R _ 1 . m d l

R e v is io n 1

G B _ fa s t _ P R .d o c

Ia n M o o re

DO C Text

1 6 th O c to b e r 2 0 0 9 In 1 O u t 1

P d_ syn c P d_syn c

s y n c _ g o v _ re s p o n s e

1 1 6 .2 1 s + 1

w _d w_r

2 H te rm

50

f

In 1 O u t 1

f1 .m a t

W T _ re s p P d _ d is tu r b

Figure 4.46 - Simulink model (top level only shown) for investigation of primary response from WTs

4.5 Setup The 2020 high wind penetration scenario given in appendices Table 7.4 was selected for simulation. This scenario assuming that all plant on the system is at full output uses the following parameters : Total Capacity Synchronous Capacity Wind Turbine Capacity Total equivalent inertia (H)

63.54 GW 40.78 GW 19.4 GW 3.11

The wind capacity was operating at an initial output of 0.6pu before the load disturbance.

4.6 Results Simulations were conducted for various control parameters, load disturbances and participating frequency response combinations.

4.6.1 Open Loop Wind Turbine Response For a system disturbance of +0.05pu which equates to an increase in load (or loss of generation) of 317.5 MW, with no response from wind turbine capacity and only (the inherent) inertial contribution from spinning synchronous plant, the system frequency is shown in Figure 4.47. The system comes to equilibrium due to the decrease in power taken from frequency dependant loads. The initial rate of change of frequency is determined by the inertia of the spinning synchronous plant.

50

For the same disturbance the open loop response of the wind turbine capacity is shown in Figure 4.48. A peak positive power increase of approximately 10% is shown for the first 15 seconds followed by a net reduction in power output for the period 15 to 30 seconds after the disturbance; in the second phase, the turbine operating at a reduced power output, due to being driven off its optimum operating point, on the maximum power point curve. Variation of the supplementary controller parameter k2 shows the effect on gain of the response during its positive and negative output. 50

F re q u e n c y [H z ]

4 9 .9 5

4 9 .9

4 9 .8 5

4 9 .8

4 9 .7 5

70

75

80

85

90 T im e [ s ]

95

100

105

110

Figure 4.47 – System frequency for a generation disconnection of 0.05 p.u (no response from generation)

0 .7 4 K 2 = 0 .5 , K 1 = 3 , T w = 1 K 2 = 1 .0 , K 1 = 3 , T w = 1

0 .7 2

W in d T u rb in e T o rq u e [ p . u ]

0 .7 0 .6 8 0 .6 6 0 .6 4 0 .6 2 0 .6 0 .5 8 0 .5 6 0 .5 4

70

75

80

85

90 T im e [ s ]

95

100

105

110

Figure 4.48 – Variation in Wind Turbine Power in response to frequency deviation shown in figure for varying k2

51

4.6.2 Closed Loop Wind Turbine Response For a larger disturbance of +0.0189 pu (1320 MW), Figure 4.49 shows the no-response and the wind turbine only responses effect on system frequency. This shows a reduction of frequency excursion measured at the beginning of the secondary response phase from approximately 49.05 Hz to 49.35 Hz. Additionally the initial rate of frequency change is reduced by a factor of five from – 0.1 Hz/sec to -0.02 Hz/sec.

5 0 .2 W it h o u t W T f re q u e n c y re s p o n s e W it h W T f re q u e n c y re s p o n s e 50

F re q u e n c y [H z ]

4 9 .8

4 9 .6

4 9 .4

4 9 .2

49

70

75

80

85

90 T im e [ s ]

95

100

105

110

Figure 4.49 – System frequency after 1320 MW generation disconnection, no plant response and Wind Turbine only response

4.6.3 Wind Turbine and Synchronous Response For the combined response from synchronous plant and wind turbine plant, Figure 4.50 shows the results obtained for the same disturbance as in the previous case. The addition of the wind turbine response clearly shows a reduction in the initial frequency excursion from approximately -0.25 Hz to -0.125Hz. 50 W it h o u t W T f re q u e n c y re s p o n s e W it h W T f re q u e n c y re s p o n s e

F re q u e n c y [H z ]

4 9 .9 5

4 9 .9

4 9 .8 5

4 9 .8

4 9 .7 5

70

80

90

100

110 T im e [ s ]

120

130

140

150

Figure 4.50 – System frequency after 1320MW generation disconnection, synchronous governor response and wind turbine plus synchronous governor only response 52

4.7 Simulation setup and results summary table Figure of [19]

Scenario

Disturb pu

Freq dev Hz (pu)

Unless specified otherwise 3a 3b

-3

0.25Hz disturb No response FPC Open Loop

3c

DFIG

4a

0.25Hz disturb same 3a O.L 0.5Hz disturb

4b

0.9Hz disturb

K2

1

Tw

Freq Signal Gain

Freq Signal Lag Block

Sync Wind Droop Cap. setting (no of m/c)

Results (Te_WT), f

1

k1, X1, X3 are negative

0.25 (-0.005)

na

na

0

0

0.005pu

0.25 (-0.005)

50

OFF

0

0

OFF

0

0

Not simulated here

OFF

0

0

Did previously simulate

OFF

0

0

Various

0.01pu

-0.5 (-0.01)

Various

50

Time constant 5sec not 11? 0.6 to 0.72

Sync Only

Good

0.6 to 0.85

Less than Paper 0.6 to 1.2

0.6 to 1.15

0.59 to 1.18

100 0.0189pu

-0.9 (-0.019)

500

ON (1/20s+1)

-11

19.4/63.54

-0.65Hz Freq at 30sec (-0.013pu)

Good Paper is -0.55Hz at 30 sec (0.6to0.7pu Te at Wind Turbine)

0.0189pu

-0.9 (-0.019)

500

ON (1/20s+1)

-11

0

2.5sec to initial spike of 0.25Hz (-0.005pu) Settles -0.08Hz (-0.0016pu)

Similar

Wind + No Sync 8. 2020 High Wind

Comments On match to paper

0.005pu

O.L 6. 2020 High Wind

K1

Sync + Wind

19.4/63.54

Page 53 of 112

Similar

Table Key : O.L – Open Loop, Assumed : Output Power (pu) in paper = Te_WT (pu) in this model

Page 54 of 112

4.8 Discussion of results The simulations undertaken clearly demonstrate the capability of DFIG and FPC based wind turbines in provision of frequency control in the primary response phase. The desired response from conventional synchronous generation is for a gradual release in the period 0 to10 seconds, followed by a sustained response in the subsequent 10 to 30 second period [2]. It is logical to assume that future power system requirements will require this form of response from wind turbine plant. Figure 4.49 shows a good contribution to control of frequency when there is no synchronous response on the system. Figure 4.50 also shows good response when a combination of synchronous and wind turbine response operates. However for this latter case the overshoot in frequency correction (at 93 seconds) is unwanted and optimisation of controller parameters and design would be warranted. In addition to the key research objective of investigating an optimum response from the UK ‘fleet’ of WTs further evaluation of the following points is suggested : • •

Overload capability of existing marketplace WT Converter units Effect of response on WT rotor speed and onset of stall conditions

These would enable increased refinement of parameters for optimum response and further ascertain the practicality of implementation of the FR scheme.

Page 55 of 112

5 Experimental Wind Turbine Test Rig 5.1 Overview A wind turbine test rig consisting of a 1kW motor generator set with a 3-phase full inverter/converter bridge is being developed in the Institute of Energy here at Cardiff University. This is to be controlled via a dSPACE® ‘rapid control prototyping’ embedded system. Important capabilities of the Test Rig are : • • • •

accepts various driven machines i.e Synchronous, IG or DFIG generator auto-generates downloadable controller code from Matlab-Simulink models easily configurable inverter/converter bridges allows practical evaluation of different turbine control-schemes and machine topologies

Photographs of the test rig are shown below in Figure 5.51 to Figure 5.53.

Figure 5.51 - Early development of rig showing dc pendulum machine left and 3-ph sync generator to right

Page 56 of 112

Figure 5.52 – Complete wind turbine test rig

Figure 5.53 – From left to right : Transformer and line module, full bridge back to back converters and resistive load bank

Page 57 of 112

5.2 Design The functional blocks shown in Figure 5.54 are described in this section. Extra specifications and details of custom made circuitry relating to these hardware components can be found in the appendices.

Figure 5.54 – Wind Turbine Test Rig functional blocks

5.2.1 DC Motor/ Pendulum Motor – Block A Since the purpose of this Rig is primarily to investigate wind turbine electrical machines and their associated control aspects, instead of a real wind tunnel and blade assembly a DC motor is used to supply the mechanical power to the wind generator electrical machine. This is a 4quadrant machine capable of acting as a motor or a generator in both rotational directions. The controller for this machine, which is provided by the manufacturer, is capable of closed loop control of speed and torque both via an internal setpoint or through an external input signal. It also has a mode for direct control of armature current. Its functioning is indicated in Figure 5.55 below and also by graphics on the controller front panel. A transfer function for such an ‘armature’ controlled machine can be found in appendices Figure 7.72. Since this is a shunt DC machine with constant field excitation the current of the machine and hence also its torque is controlled directly by the turn-on duty cycle of the armature thyristors (contained within the Power Electronics block). The operation of a dc shunt machine means that the terminal voltage applied to the machine is countered by a back-emf generated by the machine itself which is proportional to the armature speed. The proportional elements of the control loop shown provide appropriate gain to increase or decrease machine applied voltage (and hence current and therefore torque) reducing the error in the loop and finally by the integrator action reducing the error effectively to zero.

Page 58 of 112

Constant Voltage Field I set

T setpoint

+ -

PI I set

n setpoint

+ -

+ -

PI

Thyristor ON - Time

Power Electronics

PI

Armature Current

T

DC Sep Exc Shunt Machine

n

Figure 5.55 – DC pendulum controller function

Proposed Control of Aerodynamic Torque For simple replication of the aerodynamic operation of the turbine an appropriate control loop is proposed which will determine the torque setpoint for the driving DC pendulum machine according to a basic wind speed versus torque output look up table. This look-up table corresponding to the basic turbine aerodynamic performance for different wind speeds. This will be implemented via the main embedded controller. The control action is indicated in Figure 5.56. Note that only a single Torque Speed curve characteristic is shown which corresponds to a single wind speed.

Torque/V Speed/V

Speed Torque Characteristic

ADC

Controller Load

Torque set point DAC

Pendulum motor

Generator

PI Excitation ADC

Figure 5.56 – Proposed implementation of aerodynamic torque characteristic

Page 59 of 112

[22]

5.2.2 Generator Machine – Block B Via a flexible shaft coupling various machines can be directly connected to the DC motor. For the synchronous wound rotor machine (4-pole non-salient) an open loop ‘excitation voltage controller’ provides a current in proportion to the mark-to-space ratio switching control input. Specifications for these machines can be found in the appendices. Interconnection of the blocks A and B along with relevant signal scaling and unit part number identifying the components are shown in Figure 5.57.

Page 60 of 112

Figure 5.57 – Pendulum motor (Block A) to Generator machine (Block C) connection

Page 61 of 112

5.2.3 Back to Back PWM converters – Block C & D Two identical ‘6-pulse’ bridges are arranged in a back to back configuration, via a dc-link, to form a complete a.c to a.c convertor. This circuit configuration is very flexible and enables independent control of the grid and generator. The bridge arrangement along with placement of necessary voltage and current sensors is shown in Figure 5.60 MOSFETS BUZ 384 N-channel devices are employed which are nominally rated to 10.5A. These were part of a pre-manufactured assembly which include extra components for suppression of noise. These extra circuit components are shown in the appendices but essentially consist of a series resistor and capacitor connected across the drain and source. Such an ‘R-C turn-off snubber’ prevents voltage spikes and oscillations across the MOSFET during device turn-off [23]. Gate Drives To convert the logic switching output signals from the DSpace controller to an appropriate level suitable for operating the MOSFETs an integrated 3-phase bridge driver chip, International Rectifier IR2133 was used. This IC provides essential features of : • • •

Level shifting via external bootstrap capacitors to provide high turn on voltage required for the floating upper bridge gate drive outputs Electrical isolation to protect the embedded controller from high voltages in power side component failure scenarios Additional protection logic including ‘deadband’

Operation of Bootstrap circuitry In order to drive an IGBT or MOSFET with the lowest ‘on-state’ voltage drop across the device (and hence lowest power loss) the gate voltage must be 10 to 15V above the source voltage [24]. This means the gate drive supply would need to be in excess of the DC rail voltage. The option used to provide this supply for their IR2133 integrated circuit is a ‘bootstrap’ arrangement which is indicated in Figure 5.58.

Page 62 of 112

Figure 5.58 – Bootstrap supply schematic [25]

When the lower leg power transistor is conducting the bootstrap capacitor charges up to Vcc via the bootstrap diode. This lower device then terminates conduction and the upper device begins conduction. At this stage Vs rises to the dc-link voltage Vdc whereby the bootstrap capacitor voltage VB is now lifted to Vdc + Vcc. This process repeats, as every time the transistor is turned on, the bootstrap capacitor discharges to provide the power supply for the amplifier driving the upper gate drive. Figure 5.59 shows the final drive stages for the gate outputs which consist of an upper and lower pair of transistors arranged to drive each gate in a push-pull configuration.

Figure 5.59 – Final output stage in a typical monolithic gate drive [24]

Page 63 of 112

These driver circuits are connected to the embedded controller via opto-isolating transistors for protection. Switching Scheme The dSPACE controller comes equipped with pre-configured PWM outputs as below : • •

3-phase PWM or SVM (1 set of 3-phases) 1-phase PWM (4 individual channels)

Thus for the test rig implementation one converter will need to be driven using the 3-phase PWM outputs, the other will need fabricating from 3 single PWM channel outputs. The fourth PWM signal is to be used for the DC link regulation duties. In non- SVM mode all of the PWM signals can be modulated by an appropriate sine wave as demonstrated later in this chapter. Note that the generator side MOSFETS can provide rectification without being switched by virtue of their reverse bias rectifying property (body diodes) and thus will allow power flow from the ac generator side to the dc-link and hence produce a voltage on the bus.

Page 64 of 112

Ph 1

Ph 2

Ph 3

Ph 1

Ph 2

Ph 3

Red Yell Blue

M

Ph 1

Ph 2

Ph 3

Ph 1

Ph 2

Ph 3

LEM LV 25-P AVR

IR 2133 Gate Drive

IR 2133 Gate Drive LEM LV 25-P

dSPACE DS1103 PPC

VLL_gen

VLL_grid

LEM LTS 25-NP

LEM LV 25-P

Current Sensor

LEM LTS 25-NP

Figure 5.60 – Wind Turbine Test Rig back to back converter bridge configuration

Page 65 of 112

5.2.4 Generator and Grid Side Controller – Block E & F These are the two main control duties to implement. These will regulate the appropriate voltages, currents and power flows on the generator, DC Link and grid side to implement the required wind turbine control scheme and correct synthesis of the grid side a.c power output. A typical DFIG control scheme is shown in Figure 5.61 below. The scheme uses vector control of the generator via the rotor current to give reactive power control using id and real power control using iq. The grid side converter maintains real power flow to or from the grid in order to regulate the dc link voltage. Appropriate load angle control and variations on basic vector control are planned to be implemented for both FPC and DFIG machines. In addition to speed, position information from a tachometer on the machine shaft is available to the embedded controller. This will be used in the d-q transformation within the generator and grid side PWM control schemes.

Figure 5.61 – Block diagram for the control of a DFIG [26]

Page 66 of 112

Embedded Control System Block E & F are both implemented on a dedicated control system. The embedded controller, software, instrumentation and development interface is capable of automatically generating downloadable code from Matlab-Simulink blocks. Additionally modification of parameters and viewing of variables and data is possible in real time including construction of custom display instrumentation. Notable features of the embedded controller platform are : • • •

fast main CPU (1 GHz) slave motor control DSP range of peripheral communication interfaces

Figure 5.62 shows a system diagram for dSPACE hardware platform.

Figure 5.62 – DS1103PPC embedded controller architecture [27]

Page 67 of 112

Sensors/Instrumentation Industry standard closed loop voltage and current transducers are used : • •

Voltage : LEM LV-25P, galvanically isolated, closed loop compensated transducer using hall effect Current : LEM LTSR 15-NP, galvanically isolated, closed loop hall effect, ASIC based

Circuit configuration for these sensors can be found in the appendices

5.2.5 Power System – Block G This consists of basic configurable power system load elements including a 3-phase transformer, line module with series resistive inductive and shunt capacitive elements and a resistive load bank. This will enable testing of the wind turbine machine for various real and reactive power loadings including step changes in output. In the later stage of the project it is planned to connect the test rig to another motor/generator load system in the laboratory to see how the wind turbine machine responds to step changes in system frequency. More information on this power trainer can be found from relevant equipment manual [28].

5.2.6 General Assembly and Connection Cabling Twisted pair cabling is used for the gate drive to MOSFET gate connection in keeping with recommended practices for minimisation of radiated noise and to reduce the possibility of spurious switching signals [29, 30] . Earthing To ensure noise free operation, isolation for protection of equipment, and personal safety from electric shock, please note the overall earthing arrangements highlighted below and in Table 5.2 • • •

dSPACE CP1103 connector panel has all BNC bodies tied to mains earth. Voltage and current sensors are provided with insulated BNC connectors Gate drive BNC’s are insulated

The driven generator, both converters and the dc-link are floating with respect to mains earth. Table 5.2 – Earthing scheme for wind turbine test rig Item AC driven machine Gen – Side Inverter DC-Link Grid Side Inverter AC – Grid Side Voltage Sensors

Earthing Isolated from mains

Voltage 0-400 Vac

Referenced to AC driven machine

0-600 Vdc

Comment Upper gate drive BNCs at Vdc-link dc voltage above 50V Upper gate drive BNCs at Vdc-link

Isolated from mains Floating BNCs

Current Sensors

Floating BNCs

0-500V in / +/- 5V out +/- 5V out

DSpace Break Out box

BNCs Earthed

+/- 30V dc

Page 68 of 112

Connection of these using metal BNC may compromise operator safety Connection of these using metal BNC may compromise operator safety Logic 0V tied to mains earth internally

Note connection of any neutral points on the driven machine, DC-Link or bridge generated a.c may impact/necessitate careful reconsideration of the above and review of any safety measures taken. Protection A slow blow fuse rated to 1.6A is fitted to the synchronous rotor excitation winding in order to protect the rotor against over-current. A slow blow fuse rated to 1.6A is fitted to the dc-link to prevent excess current in the dc-link

5.2.7 Setup The basic procedure to create, implement and run a control scheme for the test rig is detailed in the appendices along with comments on specific use of the relevant software packages.

5.3 Results The initial testing of the rig was undertaken in open-loop configuration using a low level of excitation for the a.c generator and hence a low value of d.c link (approx 30 V) and a correspondingly low voltage fabricated a.c wave output.

5.3.1 Measurement and Open Loop Control Test Open Loop Control - This test demonstrates the open loop control capability of the control system by outputting a control voltage determined by the centre left ‘slider control’ shown in Figure 5.64. This signal is output from the 6th DAC channel (shown in Figure 5.65) on the controller and is connected to the external speed input signal of the pendulum machine controller. The simulink model for this is shown in Figure 5.65 and the subsystem block in Figure 5.63. Instrumentation Capability –Basic measurements of currents, voltages, shaft speed and torque are input to the relevant ADC channels and then displayed on the PC instrument layout. Additionally appropriate blocks were used to calculate shaft power and also the real and reactive power components of the measured currents and voltage to give the outputs P_2 and Q_2 shown. Closed Loop Control – Relevant simulink blocks to implement closed control of voltage via the synchronous machine AVR control input are shown at the bottom of Figure 5.65. This functionality was not tested however as no over-voltage protection scheme was present on the AVR input.

1 C o n sta n t

1

rpm _set_raw

rp m _ se t_ g a i n

rpm _set

rp m _ li m ite r

-K -

vo l t p e r rp m e xt

0 .1 d s5

DAC DS1103DAC_C6

Figure 5.63 – Open loop speed setpoint control

Page 69 of 112

Figure 5.64 – Instrumentation layout for basic measurement experiment

Page 70 of 112

R T I D a ta

10 M UX ADC DS1103M UX_ADC_CO N1

-K -

-K -

rpm

a s1

rp m p e r vo lt

10

-K -

0 .1

v o l t p e r rp m

DAC

d s1

DS1103DAC_C1

DAC DS1103DAC_C3

a s2

+ /- 1 is + /-1 0 V e q u iv

-K -

Nm

N m p e r vo lt

0 .1

vo lt p e r N m

DAC

d s2

DS1103DAC_C2

DAC C o n ve n tio n : P o s itiv e T o rq u e fo r m o to r-g e n -l o a d

DS1103DAC_C4

Nm

-K -

power_shaf t

rad_s

T e rm i n a to r

P ro d u ct

ra d _ s p e r rp m P_2 Q _2

T e rm in a to r2

V

PF

PQ

sp e e d _ se t_ D A C 6

I

f(u )

Active & Reactive Power

sig n arm l s

S_2

sq rt_ p sq r_ q _ sq r

A m ps_rm s

RMS sig n arm l s

T e rm i n a to r3

D ivid e

3

p o w e r _ lo a d _ p e r _ p h a s e

P ro d u ct1

p o w e r _ lo a d

T e rm in a to r1

3 p h a se s

V d if f _ r m s

RMS1 ADC DS1103ADC_C19

ADC DS1103ADC_C20

10

-K -

a s4

-K -

V d if r

V d ifr p e r vo lt

10

10

a s3

vo lt p e r V d ifr

0 .1

Am ps

A m p s p e r vo lt

1 C o n sta n t1

1

vo lt p e r A m p

0 .1

DAC

d s4

DS1103DAC_C7

0 .1

DAC

d s3

DS1103DAC_C8

P ID v o lt _ s e t _ r a w

vo lt_ se t_ g a in

1 0 0 V e rro r g e t 1 0 0 % d u ty

v _error

P I D C o n t r o l l eR r a t e L i m i t e r M a x _ d u t y

duty _set

-K vo lt p e r d u ty

d u ty cycle : 0 -1 0 V c trl v o l ta g e g ive s 0 -1 0 0 %

0 .1 d s6

Figure 5.65 – Matlab-Simulink model for basic testing of instrumentation capabilities

Page 71 of 112

DAC DS1103DAC_C5

5.3.2 Bridge Inverter Test Square Wave switching - This test demonstrates the simple square wave switching scheme as detailed in chapter X. By setting the ‘sin_true’ constant to zero as shown in Figure 5.66 a 50:50 duty cycle square pulse is sent to each PWM output channel. Channel b and c are delayed by the appropriate 2Pi/3 phase angle. Note that by sending a pulse with a low value of 0 and a high value of 1 means that no modulation is present on the output signal from block ‘DS1103SL_DSP_PWM3’. For experimentation some variation of the frequency and pulse widths can be achieved by varying the input parameters shown on the instrument layout of Figure 5.67. Additionally there is a ‘Plotter’ component in the top right hand corner to verify that the modulating signals chan_a, chan_b and chan_c are correct.

R T I D a ta NO TES :

te st1 .m d l Ia n M o o re O ct 2 0 0 9

p e rio d = 2 * p i / f_ ra d s E xe rc ise s b a sic 3 -p h P W M syn th e sis

d o u b le ((i n t1 6 ((2 * p i / f_ ra d s)* 1 0 0 0 )))/1 0 0 0 co n ve rts to 3 d p p re cisio n

f_ ra d s = 5 0 * 2 * p i n e e d to cre a te in W o rksp a ce

S co p e

S in e W a ve 3

S cope 2

chan_a

1

chan_a

sin _ tru e

D u t y c y c le a

S w itch

D u t y c y c le b

chan_b p u ls e _ a

P u lse G e n e ra to r

T ra n sp o rt D e la y

D u t y c y c le c chan_c

T ra n sp o rt D e la y1

PW M Stop

DS1103SL_DSP_PW M 3

S cope 3 1 _O N_O FF

Figure 5.66 – Simulink model for bridge inverter testing

Page 72 of 112

Figure 5.67 – Instrumentation layout for bridge inverter testing

Page 73 of 112

Figure 5.68 shows the phase a gate drive signal and the voltage developed across phase a (resistive) load for the square wave scheme driving a load with resistive and inductive elements. For this RL load, as expected there is a delay in the change in current and hence delay in voltage rise as the respective magnetic fields change in the inductive components. Gate drive logic signal Voltage across load

Figure 5.68 – Fabrication of a.c waveform in RL load using square wave switching (CH1 50.0V/div, CH2 10.0V/div, time base 10.0ms/div) Sinusoidally Modulated PWM Sinusoidal modulated PWM signals are created by inputting a sinusoidal signal which varies between zero and one into the DS1103SL_DSP_PWM3 block. This block outputs a centrally aligned PWM signal with a duty cycle in proportion to the block input. For example a zero input gives a 0% duty cycle output, an input of 0.5 gives a 50% duty cycle output, an input of one gives a 100% duty cycle output. The instrument layout shown in Figure 5.67 includes the facility for frequency adjustment of the generated sinusoidal wave which is fed into the DS1103SL_DSP_PWM3 block. Figure 5.69 shows the results for this switching scheme with a series resistive inductive load. This waveform shows a considerable improvement over the previous switching schemes and bears close resemblance to the ideal sinusoidal shape required of a grid side inverter. Some noise is present which is to be expected on a basic Sinusoidal modulated PWM scheme. Gate drive logic signal

Voltage across load

Figure 5.69 – Fabrication of a.c waveform in RL load using Sinusoidally modulated PWM (CH1 50.0V/div, CH2 2.00V/div, time base 10.0ms/div) Page 74 of 112

6 Further Work The contribution made by the PhD is planned in three main areas of work and are planned as below 1 - Modelling of appropriate Response from Wind Turbines : This work involves using Matlab-Simulink to further investigate and optimise a suitable control algorithm for provision of FR from WTs. The key thrust of this work is to use multimachine representation of WTs and varying their response either individually or collectively to provide the optimum response. 2 - Experimental testing : Testing of the FR control scheme will be undertaken in order to verify its functioning and discover some of the practical implementation issues/characteristics. This will firstly involve stand-alone operation of the WT rig and its response to a change in load. A second more realistic stage of testing is planned where the WT will be connected onto a larger Power system whereby a change of frequency will be instigated on the larger system by again a step change in load or step adjustment of the main power system generation output. Observation of the effects of the implemented FR on the test rig DC driving machine may feed into the 3rd work topic detailed below. 3 - Exploration of further Turbine system issues : Whilst providing a positive power response to the grid, a major concern with any WT system, is the effect that this transient change in power will have on the mechanical and aerodynamic assembly of the turbine. Excessive stresses may lead to premature failure or maintenance and downtime issues for the WT. This work topic will explore these areas from the point of view of the effect on FR on the WT assembly. Note this work may involve the use of proprietary WT software simulation tools such as those from Garrad Hussan in Bristol.

Page 75 of 112

6.1 Risks Some potential problems identifiable in the project are – WT Rig • General noise from switching causing malfunctioning/spurious operation of the WT rig • Harmonic noise from WT rig interfering with operation of Power System Trainer • Learning curve associated with developing a complex embedded controller Simulation • Balance between accuracy and simulation run time of models • Access to accurate data for Turbine machine and Power system modelling Turbine System issues • Learning curve for aerodynamic theory • Obtaining use of 3rd party software

6.2 Gantt Chart The sequence of tasks and approximate time allocations are indicated in the Figure 6.70. Development of the WT test rig is described in more detail than the simulation work as this practical element of the project involves less uncertainty in terms of task sequence.

Page 76 of 112

Figure 6.70– Gantt Chart for PhD Page 77 of 112

7 Appendices 7.1 Simulations 7.1.1 Simulation Baseline Record Model Name : GB Sys with Fast PR from Wind Date : 13 April 2009 Author : Ian Moore

File : GB_fast_PR.mdl

References Paper : J.Ekanayake. N.Jenkins, “Frequency Response from Wind Turbines”, Wind Engineering Vol 32 No 6 2008 Change History Previous Version : Fig7.mdl, Janaka, March 2009 Janakas Mods (compared to Paper) : Added a first order lag block for smoothing, which in fact provided the extended power response for figs 6 & 8 Ians Mods : Addition of Freq Signal Gain (needed because of pu WT confusion at system summing point) / Addition of switch to remove lag block so to enable sharp response of fig 3 and 4 results Ians Minor mods : Signal names added for scope identity /Converted pu for output to freq for format like in paper General Settings Machine Type IG, parameters as appendix X kp,ki = 10,1 although paper states 0.5,0.5 Note : doesn’t seem to make a difference Summary Comments : Fig 3a – Can’t compare full model as don’t have it, and unsure about how to verify effect on shaft speed 0.06 Fig 3 and 4 are done without the Freq Signal delay block. Is this of importance? 4b shows variation in gain General comment on paper : Fig3,4 work doesn’t quite join up to Fig 6,8 Extra Questions : (Possibly Minor) Sync Droop appears set at 11%, 1/11 not 4 % ? Procedure : Run the setup file then the simulation file.

Page 78 of 112

7.1.2 Setup The Simulink Solver settings are as shown in the dialogue below

Page 79 of 112

7.1.3 M-Files Inertia_setup.m %Frequency response from wind turbines - JE NJ GS 2008 %Constants for simplified WT inertia model %12 March 2009 %Ian Moore %This file works out the parameters to manually insert in the main model %The values are taken from Janakas paper %Later models have their own parameters input so some of the control %parameters below are not used. e.g K1 controller gain etc % GB sys Heq = 4.55 D = 1 %2MW IG/DFIG Wind Turbine in p.u S=2e6 % VA Rs=0.00491 % Stator Resistance Rr=0.00552 % Rotor Resistance Xls=0.09273 % Stator Reactance Xlr=0.1 % Rotor Reactance Xm=3.96545 % Magnetising Reactance H=4.5 % Lumped Inertia Constant Lm=Xm Xr=Xlr Lr=Xr

% Because X=wL at 1pu w % Xlr % Rotor Leakage Inductance

%Control Model Parameters, Voltage path v_qr / v_qs kp=0.5 % All in pu ki=0.5 % Supplementary Control Gains in pu K1=3.0 K2=1 Tw=1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %Work out the Inertia from H given w_e=1500*2*pi*2/60 % Rotational speed electrical, 2 pairs of poles on this machine w_m=(1500*2*pi)/60 J=(2*H*S)/(w_m*w_m) % kg.m sq %Dont actually use this, see notes %Initial power output for integrator 2MW %LUT %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Lss=Xm+Xls Lrr=Lr+Lm L0=(Lrr-((Lm*Lm)/Lss)) %%%%%%%%%%%%%%%%% Choose IG or DFIG %%%%%%%%%%%%%%%% %X1=Lss/Lm %DFIG %X2=1/Rr %X3=Lm/Lss

Page 80 of 112

%T1=L0/(w_e*Rr) X1=Lrr/Lm %IG X2=1/Rs X3=Lm/Lrr T1=Lss/(w_e*Rs)

%Note its w_s in paper

%%%%%%%%%%%%% Convert X2/[1+sT1] to 1/(as+1) format s_term = T1/X2 non_s_term = 1/X2 %%%%%%%%%%%%%%% END %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Page 81 of 112

GB_Fast_PR_setup.m %Set some defaults so we can run a sim directly from the GUI %without mfile having to run an mfile first %Note these are global so we can use the workspace editor to change %paramaters quickly rather than changing an m-file global Ld_Db; global sync_droop; global GW_WT;

Ld_Db = 0.0189; sync_droop = -11; GW_WT = 19.4;

global global global global global global

K1; K2; Tw; Kf; Kf_Lag; f_Lag_ON;

K1 = -3; K2 = 1; Tw = 1; Kf = 500; Kf_Lag = 20; f_Lag_ON = 0;

global global global global global global

B_K1; B_K2; B_Tw; B_Kf; B_Kf_Lag; B_f_Lag_ON;

B_K1 = -3; B_K2 = 1; B_Tw = 1; B_Kf = 500; B_Kf_Lag = 20; B_f_Lag_ON = 0;

Page 82 of 112

GB_Fast_PR_sim.mdl % File name - GB_fast_PR_sim.m % Ian Moore - 20/4/09 % Requires - GB_fast_PR_setup.m Run once first % Does - Runs the mdl sim and plot results function GB_fast_PR_sim clear mdl_file = 'GB_fast_PR_3' %global sigsOut global tout; global f; global Te_WT; global param; global Ld_Db; global sync_droop; global GW_WT; global global global global global global

K1; K2; Tw; Kf; Kf_Lag; f_Lag_ON;

global global global global global global

B_K1; B_K2; B_Tw; B_Kf; B_Kf_Lag; B_f_Lag_ON;

% Set the plots to cycle through linestyle by default set(0,'DefaultAxesColorOrder',[0 0 0],'DefaultAxesLineStyleOrder','-|--|-.|:') %%%%%%%%%%%%%%%%%%%%%%%%%% 3a OL plot of f %%%%%%%%%%%%%%%%%%%%%%%%%%%% figure(1); clf; %Clear the figure % [ Ld_Db, sync_droop, GW_WT, k1, k2, Tw, Kf, Kf_Lag, f_Lag_ON] param = [ 0.005, 0, 0 , -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(f.time,f.signals.values,'b');xlabel('Time [s]');ylabel('Frequency [Hz]'); xlim([70 110]);ylim([49.75 50]); %title('Power System Frequency Deviation - Fig3a'); hold all; grid on; %Hold plot and cycle line colours %legend('Frequency (Hz)'); print -f1 -r600 -djpeg fig3a;hgsave('fig3a'); %%%%%%%%%%%%%%%%%%%%%%%%%% 4a Plot %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%% figure(2); clf;

%Clear the figure

%param = [ 0.005, 0, 0, -3, 0, 1, 500, 20, 1 ]; update(); sim(mdl_file); param = [ 0.005, 0, 0, -3, 0.5, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(Te_WT.time,Te_WT.signals.values,'b');xlabel('Time [s]');ylabel('Wind Turbine Torque [p.u]'); xlim([70 110]);ylim([0.54 0.74]);

Page 83 of 112

%title ('Open Loop Turbine Torque for 0.25hz Freq Deviation- Fig 4a'); hold all; grid on; %Hold plot and cycle line colours param = [ 0.005, 0, 0, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_Te_WT('r'); %param = [ 0.005, 0, 0, -3, 2, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_Te_WT('m'); %legend('K2 = 0, K1 = 3, Tw = 1','K2 = 0.5','K2 = 1.0','K2 = 2.0'); legend('K2 = 0.5, K1 = 3, Tw = 1','K2 = 1.0, K1 = 3, Tw = 1'); print -f2 -r600 -djpeg fig4a;hgsave('fig4a'); %%%%%%%%%%%%%%%%%%%%%%%%%% 4b Plot %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%% %Bigger Disturbance figure(3); clf; %Clear the figure param = [ 0.01, 0, 0, -3, 0, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(Te_WT.time,Te_WT.signals.values,'b');ylabel('Wind Turbine Torque (p.u)'); xlim([70 110]);ylim([0.4 1.2]); title ('Open Loop Turbine Torque for 0.5hz Freq Deviation - Fig 4b'); hold all; grid on; %Hold plot and cycle line colours param = [ 0.01, 0, 0, -3, 0.5, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_Te_WT('r'); param = [ 0.01, 0, 0, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_Te_WT('g'); param = [ 0.01, 0, 0, -3, 2, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_Te_WT('m'); legend('K2 = 0, K1 = 3, Tw = 1','K2 = 0.5','K2 = 1.0','K2 = 2.0'); print -f3 -r600 -djpeg fig4b;hgsave('fig4b'); %%%%%%%%%%%%%%%%%%%%%%%%%% Fig 6 Plot %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%% figure(4); clf; %Clear the figure param = [ 0.0189, 0, 0, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(f.time,f.signals.values,'b');xlabel('Time [s]');ylabel('Frequency [Hz]'); xlim([70 110]);ylim([49 50.2]); %title ('2020 High Wind System frequency Drop, No Synchronous Response, with and without Wind Response for 1320GW loss'); hold all; grid on; %Hold plot and cycle line colours param = [ 0.0189, 0, 19.4, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_f('r'); legend('Without WT frequency response','With WT frequency response'); print -f4 -r600 -djpeg fig6;hgsave('fig6'); %%%%%%%%%%%%%%%%%%%%%%%%%% Fig 8 Plot %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%% figure(5); clf; %Clear the figure param = [ 0.0189, -11, 0, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(f.time,f.signals.values,'b');xlabel('Time [s]');ylabel('Frequency [Hz]'); xlim([70 150]);ylim([49.75 50]); %title ('2020 High Wind System frequency Drop, Including Synchronous Response, with and without Wind Response for 1320GW loss'); hold all; grid on; %Hold plot and cycle line colours param = [ 0.0189, -11, 19.4 , -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file);plot_f('r'); legend('Without WT frequency response','With WT frequency response'); print -f5 -r600 -djpeg fig8;hgsave('fig8');

Page 84 of 112

%{ %%%%%%%%%%%%%%%%%%%%%%%%%% Fig FR Control %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%% Add workspace outs figure(6); clf; %Clear the figure param = [ 0.0189, -11, 19.4, -3, 1, 1, 500, 20, 1 ]; update(); sim(mdl_file); plot1 = plot(f.time,f.signals.values,'b');ylabel('Frequency (Hz)'); %set(plot1,'XLim',[70 130],'YLim',[48.8 50.2]) title ('2020 High Wind System frequency Drop, Including Synchronous Response, FR Control'); hold all; grid on; %Hold plot and cycle line colours legend('Without WT frequency response','With WT frequency response'); %} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% function plot_Te_WT(colour) plot(Te_WT.time,Te_WT.signals.values,colour) end function plot_f(colour) plot(f.time,f.signals.values,colour) end function update Ld_Db sync_droop GW_WT K1 K2 Tw Kf Kf_Lag f_Lag_ON end

= = = = = = = = =

param(1); param(2); param(3); param(4); param(5); param(6); param(7); param(8); param(9);

end %miscellaneous commands %get(plot1);xlim;

Page 85 of 112

G B S y s w ith F a s t P R fr o m W in d Ia n M o o r e - 1 4 A p r il 2 0 0 9

-1 1 -K -

G B _ fa s t _ P R _ 1 .m d l

1

2s+1

1

0 .2 s + 1

12 s+ 1

0 .3 s + 1

P d_ syn c

G B _ fa s t _ P R . d o c

1 C lo c k 0

s+ 1

P d _ W in d P d _ W in d

S w it c h

1

-K 1

1 9 .4 /6 3 .5 4

w _r

2 H te rm

N o o f m /c

20 s+1

w_d

6 .2 1 s + 1

-K 1

50

f1 .m a t

T_ d e m

k1

f re q s ig la g d u /d t

-3 T_ d e m

-K -

10 s+ 1

1

s

0 .0 1 2 9 s + 0 .0 0 4 9 1

-K -

Te_W T

k2 s s+1

1

PP dd_ _WW TT T e_ W T

1 /3 .5 0 .6

1 s

Pd_W T

L o o k -U p T a b le

0 .6

0 .9 5

Figure 7.71 – Complete frequency response model in Matlab-Simulink

Page 86 of 112

f

7.1.4 Model Parameters Synchronous plant Parameters taken from [3] Governor =

Turbine =

Droop =

 1   0.2 s + 1

 2s + 1   1   12s + 1  0.3s + 1

 1  11

2MW induction wind turbine model parameters Stator resistance (Rs) : 0.00491 pu Rotor resistance (Rr) : 0.00552 pu Stator reactance (Xls) : 0.09273 pu Rotor reactance (Xlr) : 0.1 pu Magnetising reactance (Xm) : 3.96545 Lumped inertia constant (H) : 4.5 sec Controller Parameters kp = 0.5, ki = 0.5 Closed Loop Simulation Parameters Tw = 1, k1 = 3, k2 = 1, kf = 500, Tf = 20 Calculation of Inertia Heq =

∑

i = coal , gas ,....

Hi *

Si S sys

Inertia constant H is the kinetic energy in watt-seconds divided by the VA base where ω0m is the rated angular velocity in rad/s. H =

1 Jω 02m 2 VAbase TABLE 7.3 - PARAMETERS FOR SIMPLIFIED WIND TURBINE MODEL Turbine Type DFIG

X1

Lss Lm

X2

1 Rr

Page 87 of 112

X3

Lm Lss

T1 or T2

L0 ω s Rr

FPC (IG based)

Lrr Lm

1 Rs

Lm Lrr

L0 ω s Rs

TABLE 7.4 - PLANT MARGIN AND OPERATING CAPACITY

Generator Type

Scenario - High Wind 2020 Installed Capacity

Plant Margin

Operating Capacity

(GW)

(GW)

(GW)

New Coal

3.7

1.3

2.41

FR

Coal

16.9

7.61

9.3

FR

Gas

27.3

12.29

15.02

FR

6

0

6

FR

Interconnector

3.3

0

3.3

FR

Other

6.8

2.04

4.76

FR

Nuclear

FR Sync Cap = 40.78 Onshore Wind

14.3

5.72

FR

Offshore Wind

34.2

13.68

FR

Other

5.6

3.36

No FR

118.1

63.54

Total Capacity

FR PEI based = 19.4

TABLE 3 H

EQ

Generator Type

ON SYSTEM BASE

63.5MVA

Scenario - High Wind 2020 Capacity(GW)

Hi

Heq

New Coal

2.41

4.5

0.17

Coal

9.3

4.50

0.66

Gas

15.02

6.00

1.42

Nuclear

6.00

3

0.28

Interconnector

3.30

0

0.0

Other

4.76

4.5

0.34

Onshore Wind

5.72

0

0

Offshore Wind

13.68

0

0

Other

3.36

4.5

0.24

Total

63.54

3.11

Page 88 of 112

7.2 Laplace Transformation For a rotating machine we know per unit torque is synonymous with per unit power if there is no change in speed. Although real plant steam turbines or diesel generating plant will have a particular Torque Speed characteristic we assume here that the plant will adjust its torque upwards slightly to maintain its setpoint Power output when a system frequency drop occurs. This enables us to present the modified swing equation X describing the relationship between power and frequency on the system. 1 d (∆ ω r ) = ( P gen − P load - KD∆ ω 2H dt

r

)

In terms of change in frequency and ∆P we can replace to give 1 d (∆ f ) = ( ∆ P - KD∆ f ) 2H dt This is converted into the S-Domain as below : Re-arranging to separate input and output

2H

d (∆ f ) + KD∆ f = ∆ P dt

For simplification of presentation (as is common with most authors) from now on we dispense with superbars as Power System load studies normally use per unit quantities. Taking the Laplace transform with all initial conditions as zero gives 2Hs ∆F(s) + KD ∆F(s) = ∆P(s) where for example the notation ∆F(s) indicates the term is the Laplace transform of the time domain function ∆f(t) and where s is a constant with the unit of 1/t Collecting the terms gives

Page 89 of 112

∆F(s) ( 2Hs + KD ) = ∆P(s) The transfer function of the system is G(s) = ∆P(s) / ∆F(s) = 1 / 2Hs + KD This equation can be seen implemented in Figure 4.42

Page 90 of 112

7.3 Experimental Test Rig 7.3.1 Procedure for Use

Setup The basic procedure to create, implement and run a control scheme for the test rig is as follows: • • • • •

Create Matlab-Simulink model to include appropriate dSPACE blocks Create object code using Real-Time Workshop from within Matlab-Simulink Download code (.sdf file) to the target system (DS 1103PPC) Create appropriate instrumentation layout from within dSPACE ‘control desk’ for viewing and/or modification of system data/parameters. Start execution of embedded controller code and start ‘animation’ of instrument panel

Please take of note the items in the following subheadings regarding the overall process of development.

Equipment Initialisation Make sure Ethernet link cable from controller to PC is connected before booting the PC.

Matlab Simulink Model Consider using ‘output’ connectors on sub system models and top-level blocks. Visibility of parameters to other blocks and also modification of these. Consider use of workspace parameters with respect to ability to modify parameters ‘in-line’. Organise a method for version control of Matlab models and corresponding instrument layout.

dSPACE Instrument Layout Consider re-usability of instrument layouts and appropriate naming.

Object Code No comments.

Control System Operation Observe appropriate operation for each item of equipment as below : Pendulum Controller This machine has no emergency stop. WARNING - Use of the rocker switch 13 [LD Manual] will result in the machine only temporarily returning to a zero setpoint if the external control input is used. Return switch 11 to the ‘internal’ setting position to avoid this immediately after performing an emergency shutdown of the pendulum machine with this switch. Page 91 of 112

Table 7.5 – Convention taken for rotation for test rig Reference : Looking at rear housing of the Pendulum machine case, output shaft facing away Quadrant

Pendulum

Rotation

1

motor

Clockwise

2

Generator

Counter clockwise

3

Motor

Clockwise

4

generator

Counter clockwise

7.3.2 Equipment Specifications Interconnection of the LD equipment uses the signal scaling indicated in the table below Table 7.4 – Signal scaling and naming for dc pendulum controller Signal Name M in

Description Torque transducer

Type Sensor Output

Design Scaling 1 V / 3 Nm

Measured

EXTERN

Speed/Torque setpoint

Control Input

+ 4.32 V for 1495 rpm

n out

Speed sensor output

Conditioned sensor output

10 V / 3500 rpm / 22 Nm / 20A +/- 14 V max/min

Mout

Torque sensor output

Conditioned sensor output

+/- 14 V max/min

Page 92 of 112

+11.7 V @ 1500 (table 4)

Direction Taking U1 as + ? Q1 op gives ? Speed + CW

Comments

+ CW

Gain Adjustable

Q1 op gives V Note U1,U2 is +

Gain Adjustable

Rate limited

Synchronous Machine 4- Pole non salient with damper winding, Class 1

Table 7.6- Nameplate data V

Y/Delta 400/230

A

1.52 / 2.66

kW

0.8 kVA / 0.8

Cos phi

0.8 – 1 – 0.8

U/min

1500 50Hz

Uerr = 220V Max 1.6A Is B/F IP20 Thermal CB included

Terminals : Stator - U1 U2, V1 V2, W1 W2 Excitation - F1, F2

Excitation Voltage Controller LD 745 021 Rectified AC by PWM of GTO/Thyristors

Table 7.7 - Nameplate data Output [+, -]

DC 0 - 200V

Input [0..10V, 0V]

Manual step buttons, TTL/24V edge triggered steps

Protection

S/C and O/L

Page 93 of 112

1A max

DC Pendulum machine Part no. PM 732 68 DC Shunt wound Table 7.8- Nameplate data 150 – 300V

Arm max 8.5A

1.0 – 2.0 kW 1500 – 3000 min-1 Err 200V 0.7A max I KL B/F IP20 Thermal CB included Torque Output [U1, U2]

1 Vdc / 3 Nm

Terminals Armature (rotor) A1, A2 Field (stator) E1, E2

Figure 7.72 – Shunt Field dc motor equation [21]

Page 94 of 112

Pendulum machine control unit LD 732 695 uP controlled Table 7.9– Control Modes modes TORQUE CONTROL

Closed Loop Torque

SPEED CONTROL

Closed Loop Speed

UNCONTROLLED

Uncontrolled (actually armature current ctrl)

Load char + Run-up char

Auto record of run-up and load

When sync m/c is on mains grid?

Figure 7.73 - Pendulum Controller Front Panel [31]

Page 95 of 112

dSPACE Controller Board Table 7.10– dSPACE Controller Board Specifications [27]

Page 96 of 112

Page 97 of 112

7.3.3 Hardware Design

Additional MOSFET Circuitry As supplied by LD-Didactic the MOSEFET modules come are configured as below (note these are the original BUZ 73 versions)

Figure 7.74 Internal wiring for LD Mosfet Units – Drawn by Paul Farrugia Note diagram is for the version supplied with 7A transistors and may have changed.

Gate Drive Boards A typical connection of the IR2133 device is shown below

Figure 7.75 – Typical connection for IR2133 [32]

Page 98 of 112

A functional block diagram for this device is shown below :

Figure 7.76 – Internal functioning of IR2133 [32] Overview of internal functioning : Schmitt Triggers : Removes noise from logic input to signal to give a clean rising edge or falling edge. Level translator and PW discriminator : Couples input logic to internal level with necessary noise immunity. e.g must tolerate Vss dropping below Vcom which can happen in practical gate drive layout implementations. To reduce noise a Pulse Width (PW) discriminator filters the gate switching signal. Pulse Generator : For the top gate drive the signal is changed into a pulse format in order to reduce power consumption used in the level translator stage. Delay : For the lower gate drive the signal passes via a delay to ensure a minimum deadtime (hence protecting against ‘shoot-through’) and then directly on to the ‘totem pole’ output stage. Pulse Discriminator : Changes the pulse format back into a square wave format. Vdd/Vbs Level Translator : Raises the internal logic level suitable to drive the upper gate drive transistors which are sitting at Vs (i.e same as the dc link voltage). The following circuit was designed in accordance with recommendations given in [24]

Page 99 of 112

C3

Place Bootstrap Capacitors at IC Pins

150pF 1

680R C4 150pF R8

VCC

8

2

7

4

6

3

5

Fclr D4 11DF4

SD

D5 11DF4

HCPL2531

680R

C5

U5

GND 1

R15

7

2

1K

6

4

8

5

150pF

VB3

VCC

VB2

C14 1uF

D7 Yellow LED

R19 VS3

VS2

VS1

680R C6 FAULT R10 680R C7

1

150pF R11 680R C8

U6

2

7

4

6

3

5

HIN1

R17

ITRIP Fclr I/P

9K

HIN2 SD I/P

HCPL2531

150pF

1

U7

R16 1K

8

R12

2

7

680R C9

4

6

3

5

HIN3 LIN1

VSS

VSS

COM

LO3 LO2 LO1 VS3 HO3 VB3

U8

R Shunt

R13

1

680R C10

2

7

4

6

3

5

150pF R14

8

<--- Signal

VSS Power -->

LIN2

Isolation Barrier

0.33

VSS

LIN3 ITRIP

FAULT LIN3 LIN2 LIN1 HIN3 HIN2 HIN1 VB1 HO1 VS1 VB2 HO2 VS2

J11

COM RV1 20R

R18

1 2

PSU RAC05 PSU 1 2

3 4

3 4

1K

C13 0.1uF 50V

C11 47uF 50V

RS: 193-301 VSS

Gate Driver Board Design Version 1 - Denley Slade 2009 Page 100 of 112

VSS

HO2 VS2

R26

33

R27

33

R28

33

R29

27

R30

27

R31

27

J5 1 2 J6 1 2

VCC HO3 VS3

LO1 COM

LO2 COM

LO3 COM

P1

SD will require a pullup HIN, LIN & FLTCLR all have 100K internal pullups

VSS

COM

VCC

POWER I/P 1 2

GND

28 27 26 25 24 23 22 21 20 19 18 17 16 15

Farnell: 1174239

HCPL2531

680R

ITRIP FAULT Fault CLR LIN3 CAO LIN2 CALIN1 CA+ HIN3 SD HIN2 VSS HIN1 COM VCC LO3 VB1 LO2 HO1 LO1 VS1 VS3 VB2 HO3 HO2 VB3 VS2

1 2

D9 RED

HO1 VS1

IR2133

HCPL2531

150pF

U9

J12 Csen

LED VSS Fclr I/P

1 2 3 4 5 6 7 8 9 10 11 12 13 14

R32 1K

Fclr

Standby

Csen

8

R25 1.5K

R23 1K

S3

SD I/P

VCC

Q4 BC117

D8 Green LED

VSS

SD Fclr

Q3 BC117

1K

Q2 BC108

1K

S1

150pF

R24 15K

C15 1uF

D3 1N4001

3

R21 10K

R22

S2 SD

VB1

C12 1uF

HCPL2531

R9

1 2 3 4 5 6 7 8 9 10

R20 1K

D6 11DF4

J7 1 2 J8 1 2 J9 1 2 J10 1 2

Use Twisted Pair Cable

U4

FAULT

R6

Current and Voltage Sensors Voltage sensor circuit using LEM LV25-P

Figure 7.77 - [33]

Current Sensing circuit using LEM - LTS 15 NP

Figure 7.78 - [34]

Page 101 of 112

Cabling Table 7.11 – Slave I/O connections [DS1103 Hardware Installation and Configuration Nov 2007, dSPACE ControlDesk Help File]

Page 102 of 112

Cable Colour coding for the 1st gate drive is as follows : 5 Yellow

ST2 PWM

7 Red

HIN 1

26 Orange

LIN 1

8 Green

HIN 2

27 White

LIN 2

9 Blue

HIN 3

28 Grey

LIN 3

19 Purple

VCC(+5V)

37 Black

GND

Page 103 of 112

8 References

Page 104 of 112

1

[] – BWEA, “Wind hits 4GW barrier - now powers 2.3 million homes in UK”, http://www.bwea.com/media/news/articles/pr20091020.html , accessed 19 November 2009 2

[] - Erinmez, I.A., Bickers, D.O., Wood, G.F., Hung, W.W., “NGC Experience with frequency control in England and Wales – Provision of frequency response by generators”, IEEE PES Winter meeting, 31 January – 4 February 1999, New York USA. J.F, McGowan J.G, Rogers A.L, “Wind Energy Explained – Theory, Design and Application”, John Wiley & Sons, 2002. 3

[ ] – Manwell

[] –Burton T, Sharpe D, Jenkins N, Bossanyi E, “Wind Energy Handbook”, John Wiley & Sons, 2001 4

[] - Ramtharan, G.; Ekanayake, J.B.; Jenkins, N., “Frequency support from doubly fed induction generator wind turbines”, IET Renewable Power Generation, Volume 1, Issue 1, March 2007, pp. 3-9. 5

[] – Fox B, et al, “Wind Power Integration – Connection and system operational aspects”, IET, 2007 6

[] – Anaya-Lara O, et al, “Wind Energy Generation – Modelling and Control”, John Wiley & Sons, 2009 7

8

[] – Ackermann T, et al, “Wind Power in Power Systems”, John Wiley & Sons, 2005

[] – Caliao D, “Modelling and Control of a Fully Rated Converter Wind Turbine” Thesis submitted to University of Manchester, 2008 9

10

[] – Bose B, “Modern Power Electronics and AC Drives”, Pearson Education, 2002

[] – Bossanyi E, “GH Bladed - Theory Manual”, Issue no.17, Garrad Hussan Partners Ltd, Bristol, 2007 11

12

[] – Krause P, et al, “Analysis of Electric Machinery and Drive Systems”, 2nd Edition, IEEE, 2002

13

[] – Kundur P, “Power System Stability and Control”, EPRI, 1994

[] - Ramtharan, G., Jenkins, N., Anaya-Lara, O., “Modelling and Control of Synchronous Generators for Wide-range Variable-speed Wind Turbines”, Wind Energy, Vol 10, 2007, pp 231– 246 14

15

[] - G Ramtharan, “Control of variable speed wind turbine generators”, PhD Thesis, 2008.

[] - Seul-Ki Kim, Eung-Sang Kim, Jae-Young Yoon and Ho-Yong Kim, “PSCAD/EMTDC Based Dynamic Modeling and Analysis of a Variable Speed Wind Turbine”, IEEE Power Engineering Society General Meeting, 2004. Volume , Issue , 6-10 June 2004 Page(s): 1735 - 1741 Vol.2 16

17

[] - National Grid Company plc, “The Grid Code,” Issue 3, Revision 25, 1 February 2008.

[] – Banham-Hall D, et al, “Grid Connection Oriented Modelling of Wind Turbines with Full Converters”, UPEC 2009 18

19

[] - J.B.Ekanayake N.Jenkins G.Strbac, “Frequency response from wind turbines”, Wind Engineering, Vol. 32, pp. 573-586. 2008 [] – Milanovic J, Kayikci M, “Dynamic Contribution of DFIG-Based Wind Plants to System Frequency Disturbance”, IEEE Transactions on Power Systems, Vol 24, no.2, May 2009 20

21

[] – Bolton W, “Control Engineering”, 2nd edition, Longman, 1998

[] – Caliao N, Tennakoon A, “Control Systems Development and Implementation with dSPACE and Wind turbine test rig, Internal report, Cardiff University 22

[] – Mohan N, et al, “Power Electronics: Converters, applications and design”, John Wiley and Sons, 1989 23

[] – International Rectifier, “Application Note AN-978 – HV Floating MOS-Gate Driver ICs”, IR, RevD 24

[] - Merello A, et al, “Design Tips – Using monolithic high voltage gate drivers DT04-4revA”, International Rectifier 25

[] – Heier S, “Grid Integration of Wind Energy Conversion Systems”, John Wiley and Sons, 2nd Edition, 2006 26

[] - dSPACE, “DS1103PPC Controller Board”, dspace_2008_ds1103_en_pi777.pdf, downloaded Dec 2008, http://www.dspaceinc.com 27

[] - TecQuipment, “PSS1 – Power System Simulator - Operator Manual”, TecQuipment, Nottingham UK 28

29

[] – Valentine R, “Motor Control Electronics Handbook”, McGraw-Hill, 1998

30

[] – Williams T, “The Circuit Designers Companion”, Newnes, 1991

[] – Leybold Didactic Gmbh, “Instruction Sheet – Control Unit for Pendulum Machine 732 695”, version 04/95 –Li 31

32

[] – International Rectifier, “IR2133 Datasheet PD60107 revX”, downloaded March 2009,

http://www.irf.com/ 33

[] – LEM, “LTS 15-NP Datasheet”, downloaded 20 Nov 2009, http://www.lem.com

34

[] - LEM, “LV-25P Datasheet”, downloaded 20 Nov 2009, http://www.lem.com