Computer Simulation Modeling and Analysis of the Dynamic … · 2012. 3. 20. · 1 Summary This report outlines the modelling, simulation analyses and investigation of dynamic behaviour

'-

DEVON . OTTAWA . VARENNES

r COMPUTER SIMULATION MODELINGAND ANALYSIS OF THE DYNAMICBEHAVIOUR'OF A RECIPROCATING

ENGINE BASED DISTRIBUTEDGENERATION UNIT DURING

ISLANDING TRANSITION

CLEAN ENERGY TECHNOLOGIESTECHNIQUES D'ÉNERGIE ÉCOLOGIQUE

-~~--------~- _._-- ~ ~~~--- ~- -----C TEe CENTRE DE LA TECHNOLOGIE DE L'ÉNERGIE DE CANMET

1+1 Natural ResourcesCanada

Ressources naturellesCanada Canadã

Report – CETC 2007-187 (TR) August 2007

COMPUTER SIMULATION MODELING AND ANALYSIS OF THE DYNAMIC BEHAVIOUR

OF A RECIPROCATING ENGINE BASED DISTRIBUTED GENERATION UNIT DURING ISLANDING TRANSITION

Prepared by:

Farid Katiraei

CANMET Energy Technology Centre – Varennes Natural Resources Canada

August 2007

Report – CETC 2007-187 (TR) August 2007

DISCLAMER

This report is distributed for informational purposes and does not necessarily reflect the views of the Government of Canada nor constitute and endorsement of any commercial product or person. Neither Canada nor its ministers, officers, employees or agents makes any warranty in respect to this report or assumes any liability arising out of this report.

CITATION

Katiraei, F. “Computer Simulation Modelling and Analysis of the Dynamic behaviour of a Reciprocating Engine based Distributed Generation unit during Islanding Transition”, CANMET Energy Technology Centre (CETC)-Varennes, QC, Canada. Report # 2007-187 (TR), Natural Resources Canada. August 2007, 54 pp.

ACKNOWLEDGEMENT

Financial support for this report was provided in part by Natural Resources Canada through the Program on Energy Research and Development (PERD).

Report – CETC 2007-187 (TR) i April 2007

TABLE OF CONTENT

1 Summary ................................................................................................................................................ 1 1 Sommaire................................................................................................................................................ 2

2 Introduction ............................................................................................................................................ 3

3 Anti-islanding protection of synchronous generator based DG units..................................................... 4

4 Study objective ....................................................................................................................................... 6

5 Study system........................................................................................................................................... 6

6 Distributed Generation model ................................................................................................................ 7 6.1 Synchronous generator model ...................................................................................................... 8 6.2 Gas engine model ......................................................................................................................... 9 6.3 Speed/Load control....................................................................................................................... 9

7 Islanding study ..................................................................................................................................... 12 7.1 Dynamic performance of the engine-generator .......................................................................... 12

7.1.1 Case DP-1: Isolated operation and load-drop test ......................................................... 13 7.1.2 Case DP-2: change in the load reference set-point ........................................................ 14

7.2 Islanding transitions.................................................................................................................... 19 7.2.1 Case IS-1: +50% and +25% power mismatch levels..................................................... 20 7.2.2 Case IS-2: 0% power mismatch for RL and PQ load models........................................ 20 7.2.3 Case IS-3: 0% power mismatch for resistive load......................................................... 21 7.2.4 Case IS-4: - 15% power mismatch (over generation).................................................... 22

8 Discussions........................................................................................................................................... 28

9 Conclusions and Future investigations................................................................................................. 30

10 Reference.............................................................................................................................................. 31 Appendix A Appendix B Appendix C Appendix D

Report – CETC 2007-187 (TR) ii April 2007

Report – CETC 2007-187 (TR) iii April 2007

1 Summary

This report outlines the modelling, simulation analyses and investigation of dynamic behaviour of a 180-kW biogas synchronous generator distributed generation (DG) unit, which is connected to a medium voltage rural feeder, following disconnection from the main grid and islanding. This work was mainly performed to assist development and field verification of an economically viable anti-islanding protection scheme as an alternative to a transfer trip to facilitate grid interconnection and project commissioning of the DG unit.

The main objective of the study is to identify power quality related parameters that can be measured locally at the generator site and be used to detect a loss of the main grid condition (islanding) with high degree of reliability. The modelling and simulation studies are particularly focused on analyses of frequency/voltage transients, and active/reactive power variations in the first two-second interval following the disconnection from the grid.

An innovative anti-islanding protection scheme based on a combination of voltage measurement and detection of the reactive power direction is proposed. The application of the method is recommended for anti-islanding protection of small synchronous generator based DG unit (bellow 500-kW) connected to rural feeders with highly inductive load, where few or no capacitor banks are installed.

The reported simulation studies showed that, subsequent to islanding, the voltage deviations and the changes in the reactive power flow direction of the generator are significant enough to trigger tripping of a protective relay with a selected set of voltage and reactive power thresholds. The main reason for large fluctuations observed in the reactive power and voltages of the DG is that output power of the DG unit is controlled for constant power operation at a leading power factor. Hence, loss of the main grid leads to shortage of reactive power on the feeder and voltage collapse that enforces changes in the reactive power direction of the DG unit.

Report – CETC 2007-187 (TR) 1 April 2007

1 Sommaire Le présent rapport présente les résultats de modélisation, de simulation et d’analyse concernant le comportement dynamique d'une unité de génération décentralisée à alternateur synchrone à biogaz de 180 kW raccordée à une artère rurale de moyenne tension après la déconnexion du réseau principal et l'îlotage. Les travaux ont été effectués principalement dans le but de faciliter la mise au point et la vérification sur place d'un plan de protection contre l’îlotage viable sur le plan économique comme solution de rechange à un télédéclenchement pour faciliter l'interconnexion des unités de génération décentralisée au réseau et leur mise en service.

Le principal objectif de l'étude est l'identification des paramètres liés à la qualité d’onde qui peuvent être mesurés à l'emplacement du générateur et servir à la détection de la perte de la connexion au réseau principal (îlotage) avec un degré élevé de fiabilité. Les études de modélisation et de simulation sont axées en particulier sur des analyses des comportements transitoires de la fréquence et de la tension et les variations de la puissance active et réactive dans les deux secondes suivant la déconnexion du réseau.

Une technique novatrice de protection contre l'îlotage reposant sur une combinaison de mesures de la tension et de détection de la direction de la puissance réactive est proposée. L'application de la méthode est recommandée pour la protection contre l'îlotage de petites unités de génération décentralisée à alternateur synchrone (moins de 500 kW) raccordées à des artères rurales supportant une charge hautement inductive et pour lesquelles il y a peu ou pas de banc de condensateurs installé.

Les études de simulation ont montré qu'après l'îlotage, les variations de la tension et les changements dans la direction du flux de la puissance réactive du générateur sont assez importants pour occasionner le déclenchement d'un relais de protection avec une sélection donnée de seuils de tension et de puissance réactive. La principale raison qui explique les grandes variations observées dans la puissance réactive et les tensions de l’unité de génération décentralisée est que la puissance de sortie de l’unité est commandée en vue d'un fonctionnement à puissance constante avec un facteur de puissance capacitif. Ainsi, la perte du réseau principal occasionne un manque de puissance réactive sur l’artère et une chute de tension entraînant par le fait même des modifications dans la direction de la puissance réactive de l’unité de génération décentralisée.


2 Introduction

For distributed synchronous generation, one of the key utility concerns is the detection of loss of the grid leading to unintentional islanding of the generator on the feeder [1]. In such situations, it is important that the distributed generation (DG) is rapidly disconnected from the utility feeder for reasons of:

• De-energizing the feeder for the safety of the utility linemen,

• Preventing poor power quality electricity supply to other feeder customers,

• Preventing out-of-phase reclosing leading to equipment damage from abnormal voltages and/or electromechanical shock.

Grid interconnection standards and guidelines such as IEEE Std. 1547, [1], or the draft CSA C22.3 No. 9, [2], require equipping a rotating machine-based DG unit with a Transfer-Trip (TT) scheme for anti-islanding protection if the minimum load on connecting feeder is less than twice the generator capacity (2 to 1 rule of thumb). Although an islanding detection time of less than 2 seconds is specified by IEEE Std.1547, some utilities require faster detection and DG disconnection times of less than one second. This generally depends on the protection coordination scheme and reclosing time of the automatic reclosure of the feeder. However, a fast TT scheme is very expensive and normally the economics of a small DG project (below 500-kW) does not allow such expenses. The turnkey cost of a utility-approved transfer trip with a safe disconnection time of less than 0.5 seconds is estimated in the range of $120k - $250k CAD. As discussed in the aforementioned standards, alternative anti-islanding protection methods can potentially be investigated; however, utility acceptance of any proposed method requires a field test or a third party certification, [3].

Many anti-islanding detection schemes have been proposed and investigated in the literature, [4] and [5], either categorized as passive or active protection techniques. The active methods, which are based on applying a small perturbation to the distribution grid through the interconnecting source and analyzing the response, are mainly applicable for inverter-based DG units. A combination of the customary generator over/under voltages and frequency protections and/or a variation of those methods are normally used for rotating-machine based DG units.

The aim of extensive research applied in this area is to suggest precise anti-islanding schemes which can distinguish normal disturbances in the system, due to faults or load switching (nuisance tripping), from a loss of the main grid condition. A reliable scheme should also have no or negligible Non Detection Zones (NDZ), [6]. That means the scheme should detect unintentional islanding of a DG under any load and generation condition.


One approach normally applied to rotating machine based DG unit protected by passive anti-islanding scheme is to identify measurable parameters of the system with the most significant variations upon an islanding condition. Then, a detection scheme based on a logical combination of tripping signals generated by violating threshold levels for those parameters is defined to overcome the NDZ concern. However, the main challenge is to choose appropriate threshold values.

Selection of a set of reliable parameters and choice of detection set-points are highly dependent on several aspects of the systems including the DG technology and control strategies. Hence, computer-based modeling and dynamic simulation analyses are suggested to investigate alternative solutions to the transfer-trip scheme by examination of the voltage and frequency dynamics at the point of DG connection subsequent to an islanding transient.

This report outlines details of a computer simulation model developed and several case studies are performed to analyze system transients at different operating points to examine a range of variations of several power quality parameters. An innovative anti-islanding protection method based on directional reactive power detection is also proposed and validated for the case of a farm-based bio-gas synchronous generator interconnection.

3 Anti-islanding protection of synchronous generator based DG units

Conventionally, under/over voltage, under/over frequency and directional real power relays are used for unintentional islanding (or loss of the mains) protection of distributed generation when the generation on the feeder is significantly below the feeder load, [7] and [8]. Typical transient phenomena expected subsequent to DG islanding are:

• There may be a sudden increase in DG output in order to meet the feeder loads;

• There may be a rapid decrease in frequency;

• Typically, automatic voltage regulation (AVR) control of a synchronous generator is in power factor or reactive power regulation mode so that they rely on grid support for voltage regulation. Loss of the main grid can therefore lead to rapid voltage variations.

Typical tripping points for such protections are given in Table 1, [8]. In most cases the under/over voltage and frequency protections of a small rotating machine based DG unit (below 500-kW) can properly respond to abnormal voltage and frequency excursions caused by loss of the mains condition. However, the ranges of voltage and frequency variations are highly dependent on the load type, DG control strategies, and the power mismatch level. Hence, there is always a concern with falling into a non-detection zone of these protection relays that must be prohibited by applying more reliable protection schemes or secondary protections.


Table 1 - Typical set points for a dedicated synchronous generator protective relay [8]

Protection Pick-up Time-Delay (cycles) Phase over-voltage (59) 110% (ΔV = +10%) 60 Phase over-voltage (59) 120% (ΔV = +20%) 6 Peak over-voltage (59I) 120% (ΔV = +20%) 10 Phase under-voltage (27) 90% (ΔV = -10%) 60 Phase under-voltage (27) 50% (ΔV = -50%) 6 Under-frequency (81) 59.5 Hz (Δf = -0.5) 6 Over-frequency (81) 60.5 Hz (Δf = +0.5) 6 Negative Sequence Over-voltage (47) 25% 60

Alternative protection solutions that utilize frequency and phase angle characteristics of the generator voltage waveform have been also implemented and mainly applied for applications with multiple DG units connecting to a common feeder. The most recognized protective relays of this category are the rate-of-change-of-frequency (ROCOF) and the vector surge (or vector shift) relays, [9].

The ROCOF relay measures and compares the rate of change of frequency (df/dt) of the generator voltage for two or more consecutive cycles. The vector surge relay alternatively traces the phase angle of the terminal voltage and responds to variation from one cycle to another cycle. Although a small power mismatch may result in only slow variations of the power frequency, a sudden jump in the voltage phase angle and unusual rate of change of frequency can be expected when the main source of synchronization is lost. However, large NDZ and nuisance tripping have been reported for frequency-based relays caused by large step changes in the load of feeders with low X/R ratio, and/or in case of DG units with low inertia, [9].

Rural feeders are mainly constructed of overhead lines and generally have resistive and inductive loads, with little capacitive load. If a synchronous generator based DG is set-up so that it is always a sink for reactive power when the grid is present, the loss of the grid will force the synchronous generator to begin supplying reactive power, which can then be used to detect an islanding condition. The commercially available generator AVR controls can be adjusted for a negative reactive power set-point by which the terminal voltage of the generator is adjusted to achieve a pre-fault reactive power absorption condition. Utilizing an off-the-shelf protective relay sensitive to changes in the direction and magnitude of the reactive power flow, e.g. a loss of the excitation relay (40Q) [10], which is set to trip for a particular reactive power export level, can therefore be applied as an islanding detection scheme.

The applicability of voltage/frequency based relays and a directional reactive power scheme for anti-islanding protection of a synchronous generator are examined in the following sections.


4 Study objective

The principal objective of this study is to investigate the dynamic behaviour of a reciprocating engine based distributed generation (DG) unit during and subsequent to disconnection of the main grid. The study is particularly focused on analysis of frequency and voltage transients and real/reactive power variations in the two-second interval following the disconnection from the grid. Then the engine-generator response characteristics for various levels of the power mismatch between load and generation are examined and ranges of voltage and frequency excursions, rate of change of the power frequency and reactive power variations are determined.

It is assumed that the DG unit operates in a constant real power generation mode, whereby the power generation of the unit can only be adjusted externally by changing a load reference set-point provided by an operator. The terminal voltage of the DG unit is also controlled to operate in a grid-following mode while adjusting the reactive power and power factor of the unit as required. A leading power factor of 0.97 is applied to the generator, ensuring reactive power import for the DG plant despite the variations in the on-site load. This operating strategy is a necessary requirement in order to use a directional reactive power scheme as an alternative islanding detection method.

5 Study system

The DG unit under investigation consists of a 313-kW, 480-V synchronous generator (SG) with a 200-kW biogas fuelled, internal combustion (reciprocating) engine (ICE) prime mover. The DG unit is connected to a three-phase radial feeder at 8.3-kV level. The radial feeder is connected to a distribution substation at about 3-km distance from the point of common coupling (PCC) of the DG. The radial feeder mainly supplies several farm facilities (single-phase and three-phase loads) and some residential customers. The minimum load of the feeder is about 18-A at 8.3-kV which is the equivalent of 258-kVA or 245-kW at power factor of 0.95 lagging. The maximum generation capacity of the DG unit with biogas fuel is 180-kW.

Figure 1 shows the one-line diagram of the local distribution system including the distribution substation, the distribution feeder, the DG unit connection, and the equivalent load of the feeder (fixed and variable loads). The circuit breaker CB1 represents the automatic reclosure of the feeder. CB1 is used to disconnect the feeder to simulate an unintentional islanding situation and/or a loss of the grid condition. The grid and the distribution substation are modeled as an equivalent representation of a voltage source in series with an RL impedance. The distribution feeder is also represented by an equivalent π (pi) model with line parameters given in Appendix A. The line parameters are based on utilizing a 336 kcmil conductor size, which is typical for rural feeders. The distribution load of the feeder is lumped together and modeled as either constant impedance or constant power load with lagging power factor of 0.95 or higher. The system model is developed in the PSCAD/EMTDC software environment1.

1 www.pscad.com


http://www.pscad.com/

Figure 1 – One line diagram of the study system

6 Distributed Generation model

The DG model for this study is comprised of a round-rotor synchronous generator model as the grid interface medium and a reciprocating gas engine model. Figure 2 shows the layout of the DG with a dual-fuel gas engine prime mover. The prime mover is an internal combustion Otto gas engine with six cylinders that operates based on a four-stroke cycle to drive a synchronous generator at 1800 revolution per minute (rpm), [11].

GenPrime Mover

Figure 2 - A single-shaft gas engine

Both the gas engine and the synchronous generator are equipped with several controllers and protection devices to perform the various tasks of engine start-up, generator voltage/speed adjustment, automatic synchronization with the grid, and real/reactive power generation controls for sound and proper operation during grid connected or stand-alone operation. An overall block


diagram of the control units and interconnection signals are shown in Figure 3. The detailed description of each control unit is provided in the following sections.

+ fd

Gen. Con. Unit

3PH

Generator

GasEngine AC

FieldE

ControlSynchronization

Protection

Governor Control Voltage

Regulator

Fuel

- CB

CircuitBreaker

sVsI

gVgI

refV

refLEngine Generator Control

Figure 3 - Engine and Generator control systems

6.1 Synchronous generator model

The synchronous generator model along with the excitation system for voltage regulation and/or reactive power compensation is described in this section. A detailed electromagnetic transient representation of the synchronous generator is used to model the electromechanical dynamics of the generator and other rotating masses, [12]. In this model, the DG interface medium is represented by a seventh-order set of ordinary differential equations (ODEs) in a stationary direct-quadrature (dq0) reference frame. The synchronous generator parameters based on the manufacturer specifications are given in Appendix B, [13].

The synchronous generator employs a separate excitation system to power the field winding through an Automatic Voltage Regulator (AVR), [14]. The separate excitation power is provided by a permanent magnet generator and a semi-conductor rectifier as part of the rotor. An IEEE type ACA2 ac excitation system is the closest model that represents a self-excited voltage regulator and was adopted for this study, [15]. The excitation model and parameters are given in Appendix A.

The control input for the AVR is a voltage reference set-point provided by the DG control system, Figure 3. The set-point is a constant value if the generator operates in voltage regulation mode. However, the voltage reference set-point can be varied, for instance by using a power factor controller (PFC) as shown in Figure 4, to control the reactive power output of the generator and achieve specific power factors. The PFC model of Figure 4 is based on a proportional-integral (PI)


controller that uses the terminal voltage and output current of the generator to calculate the real/reactive power supply of the generator in order to adjust AVR voltage reference (Vfref), [16]. Either a power factor set point or a VAr set-point can be used for the PFC.

Q Vf E

To generatorfield

Excitationsystem

Q

Power Factor Control (PFC)

Figure 4 – Reactive power control based excitation system

6.2 Gas engine model

A block diagram representation of the prime mover model (gas-engine) and control parts are shown in Figure 5. The model is composed of a speed/load control, a fuel actuator, and a representation of the engine, [17]. Inputs to the model are a load reference set-point (Lref) and engine-generator speed deviation (Δω). The load reference set-point is applied through the engine-generator main controller and synchronization unit as shown in Figure 4, [16]. A generic prime mover model, along with the corresponding parameters used in this study, is given in Appendix C. The type and characteristics of the speed/load control has a major impact on the dynamic performance of the prime mover. Typical governor controllers are explained in the following subsection, which covers common control units used by different manufacturers.

Speed/Load Control

Fuel Actuator GasEngine

FdL ref Tm

Prime MoverGovernor

Figure 5 - Gas engine control blocks for dynamic model

6.3 Speed/Load control

Generally, governor systems are categorised according to the type of controls and feedback signals used for the speed compensation, [18]. Three types of governor control systems are introduced in the following section that are widely used by the combustion engine manufacturers for speed and load


control of reciprocating engines. These are: (i) an “isochronous” or a constant speed governor, (ii) a “speed-droop” governor, and (iii) a speed-droop with power compensation, [18].

The simplest governor model is entitled “isochronous control” that varies the mechanical torque of the gas engine based only upon deviations in the generator speed (Δω). The isochronous control provides a fast response governor control and is only used for a single generator that forms an isolated grid. A variation of the basic isochronous control that employs a load adjustment input (Lref) and a speed-droop coefficient (R) is shown in Figure 6, [12]. The control block introduced by the transfer function G(s) in Figure 6 characterizes an integral controller with a time constant equivalent to the overall delays of a speed transducer and a control loop. The governor output is the fuel adjustment signal (Fd) that controls the gate position of the fuel valve.

FG(s)

=1L

Figure 6 - Speed/load control block for a simple integral controller

The governor control block of Figure 6 is very sensitive to a large load change and/or frequency excursions that may occur during transition from a grid-connected mode to an islanded mode. Hence, a second variation to this model is introduced in Figure 7, which is called governor “Type A”. This governor employs a lead-lag speed compensator Gt(s) to damp out speed variations by applying first a transient droop; then, stabilizing the generator speed based on the steady-state droop coefficient. A droop coefficient of R=4% is typical for small generators. The lead and lag time constants selected for this study are Tp = 1.059 s, and Ts = 3.05 s, respectively.

FGt(s)

=1L

Figure 7 – Speed/load control block type A

Governor Type A can be used for applications that require generator operation in both grid-connected and stand-alone operation. Governor Type A basically operates in a constant power mode


during the grid connected mode (Δω = 0). The real power set-point is specified by the load reference (Lref). However, for an isolated operation, a power compensation control strategy or a load following capability is applied while the power frequency is kept constant.

The commercial governor units from different manufacturers may vary based on the type of speed/load control methods. Figure 8 shows a governor control block that incorporates a power compensation loop for load sharing and uses a proportional-integral plus derivative (PID) controller for the speed regulation. This governor is referred to as governor Type B for this study. It should be noted that the power compensation loop is normally de-activated during the grid-parallel operation, if power sharing among multiple generators is not required.

F=1L

P

Figure 8 - Speed/load control block Type B

In Figure 8, Kp, Ki, KD are proportional, integral and derivative gains of the speed controller, respectively. Also KDroop is the power-freq. droop coefficient and TTD is the time constant of the power measurement sensor. Reference [19] describes an analog speed control unit type “ESD5330” manufactured by the Governor America Corp. (GAC) which employs a similar control strategy.

Dynamic performance analysis and “load-drop” tests are normally used to determine appropriate governor response time and adjust the gain factors of the speed/load controller. The governor parameters for this study are calculated according to the observations from a series of “load-drop” tests performed on the engine-generator during the preliminary engine commissioning. The gain factors and the droop coefficient selected for the speed/load controller of Figure 8 are listed in Table 2.

Table 2 - Speed/load control parameters

Speed controller Power compensation control KD = 0.05 KDroop = 0.04 KP = 9 TTD = 0.005 KI = 10


7 Islanding study

The study system of Figure 1 is used to investigate the dynamic behaviour of the DG unit and voltage/frequency transients after CB1 disconnection, while the DG unit operates in a constant power generation mode. The DG unit is equipped with an excitation and governor control system as discussed in Section 6.2. A comparison between the engine-generator operation based on governor Type A and Type B is performed to investigate the effect of speed control methods on the voltage and frequency response characteristics. Different operating points of the system are also investigated by varying the feeder load and the DG output power to analyze the impact of changes in the power mismatch level (imbalance) between load and local generation on the range of voltage and frequency variations.

The following simulation studies first examine the dynamic performance of the gas-engine generator in the grid-connected mode for step changes in load demand and the power generation set-point of the generator. Then, unintentional islanding of the feeder for various operating points and control modes of the DG unit are simulated and system transients are analyzed. Although the protection relay equipment for the line and the DG unit are not modelled, the simulation results will be used to analyze the effectiveness of typical under/over voltage and frequency protections for DG interconnection operation with the medium voltage (MV) distribution feeder of the study system and to investigate additions of new protective schemes for an anti-islanding protection. The protection study is based on the calculation of the range of voltage (ΔV) and frequency (Δf) deviations and the rate of change of frequency (df/dt) along with the time required for the system voltage and/or power frequency to reach certain thresholds (Δt) from the instant of grid disconnection.

7.1 Dynamic performance of the engine-generator

This section examines the dynamic performance of the engine-generator for step changes in the load and generation in the isolated and the grid-connected mode. The generator operates at constant power in parallel with the grid. A minimum load of 245-kW with lagging power factor of 0.95 is assumed for the feeder. The following study cases are investigated:

Case DP-1: Isolated operation and load-drop test for a sudden reduction in the generator loading,

Case DP-2: A step change in the load set-point of the generator from 20% to 100% (36-kW to 180-kW) during the grid-connected mode.

The second case is simulated for DG utilizing either type A or type B governor, with the load represented by a constant impedance model (RL load) or a constant power model (PQ load).


7.1.1 Case DP-1: Isolated operation and load-drop test

In stand-alone mode, the synchronous generator operates as a grid forming source employing a voltage regulation control and load following strategy to supply the load at the nominal voltage (1.0-p.u.) and fixed frequency (60-Hz). A load drop test (also known as load rejection test) is normally used to calibrate the frequency/voltage regulation controllers of an engine-generator during the commissioning phase. A similar test can be performed to analyze the stand-alone dynamic performance of the DG unit and to determine inertia constant and generator control parameters.

Figure 9 shows the power frequency variations of the engine-generator under study from field test results for two cases of load-drop tests. In each case, the generator was first loaded using a resistive load bank, then the load was suddenly dropped by 60-kW. The frequency curves for two initial operating points of 60-kW and 90-kW are overlaid on a same graph for comparison.

Load drop tests

59.80

60.00

60.20

60.40

60.60

60.80

61.00

61.20

61.40

61.60

61.80

15 25 35Time (sec.)

Freq

. (Hz

)

Test 1: 60-kW to zero load

Test 2: 90-kW to 30-kW load

Figure 9 - (Case DP-1) Power frequency variations for load drop-tests based on field measurements

In Figure 9, the response time of the generator and the speed control loop for frequency adjustment is about 12 seconds. A 60-kW load drop has also caused a frequency deviation in the range of 1.6 to 1.7 Hz. The frequency of slow oscillations (electromechanical mode of the generator) based on the first swing is about 1.7 rad/sec. The engine-generator inertia is estimated to be about H = 0.9 s.

The information extracted from test results of Figure 9 is then used to tune the gain factors of the speed/load controllers of the generator to achieve similar dynamic performance. Figure 10 shows the simulation results of the load-drop test from the DG model developed for the studies presented in this report. The generator loading is reduced from 90-kW to 30-kW at t = 20 s. The power frequency variations in Figure 10-a illustrate a response time of about 12 seconds for frequency regulation with a maximum deviation of about 1.7 Hz.


7.1.2 Case DP-2: change in the load reference set-point

Figure 11 shows DG response to an 80% change in the load reference set-point of the engine when governor type B, Figure 7, is used for speed/load control. The total feeder load is 245-kW/81-kVAr represented by a constant impedance model (RL). The load reference set-point of the engine is changed to full generation (180-kW) at t = 20 s, while activating the PFC+AVR controller to regulate reactive power of the DG unit. Real and reactive power variations of the DG unit (Pgen and Qgen) are shown in Figure 11-d illustrating a response time of about 12 s. to settle the output power of DG to the new operating point. The step change in the output power of DG causes deviations in the rotational speed of the generator (ωr), Figure 11-b, and slightly increases the bus voltage (Vline), Figure 11-c. The increase in the bus voltage is mainly because of the reduction in the power exported from the grid (Pgrid), Figure 11-e.

Impact of changing the speed/load control method on the dynamic performance of the engine-generator is shown in Figure 12. This figure compares performance of the governor Type B, Figure 7, with those of Type A, Figure 6, for the similar transients investigated in Figure 11. The simulation results of Figure 12 show that governor Type A provides a slightly faster speed and torque control response than the governor Type B, Figure 12-d. Governor Type A also exhibited larger speed deviations for the generator, Figure 12-b. However, the overall behaviour of the engine-generator for changes in the governor type is relatively similar. Governor type B has been used in most of the following reported studies.

In the next study, a comparison between a constant power load model (PQ) and a resistive-inductive load model (RL) is performed to investigate the impact of the load model on dynamics of the gas-engine generator. The constant power load model characterizes the real/reactive power component of the load as a function of variations in the voltage and frequency. The voltage/frequency relationship of the load and corresponding indices, which are adopted for this study, are described in Appendix A, [18]. The study is based on an 80% step change in the load reference set-point of the engine and the governor Type B is employed.

Figure 13 shows the comparison between the simulation results of this study with those of the constant impedance load model illustrated in Figure 11, whereby no significant change is observable. Hence, change in the load model has no (or negligible) impact on the dynamics of the engine-generator during the grid-connected operation, since voltage and frequency variations are negligible. The constant impedance load model is commonly used in the power system stability analyses, since it represents the most conservative load response for islanding transients considering the load dependency on voltage/frequency changes, [20].


DP-1: 60-kW load drop

Sec. 20.0 25.0 30.0 35.0 40.0

58.50 59.00 59.50 60.00 60.50 61.00 61.50 62.00 62.50

a) fr

eq. (

Hz)

freq.

58.50 59.00 59.50 60.00 60.50 61.00 61.50 62.00 62.50

b) G

en. S

peed

(Hz)

wr

0.900 0.920 0.940 0.960 0.980 1.000 1.020 1.040

c) V

(p.u

.)

Vline

0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100

d) P

, Q (M

W/M

VAr)

Pgen Qgen

0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100

e) P

(MW

)

Pload

Figure 10 - (Case DP-1) Simulation of a 60-kW load drop test based on computer model

Time (s)


DP-2: 80% step-change, RL load

Sec. 15.0 20.0 25.0 30.0 35.0 40.0

59.70

59.80

59.90

60.00

60.10

60.20

a) fr

eq. (

Hz)

freq.

59.980 60.000 60.020 60.040 60.060 60.080 60.100

b) G

en. S

peed

(Hz)

wr

0.9950 0.9960 0.9970 0.9980 0.9990 1.0000 1.0010 1.0020 1.0030 1.0040 1.0050

c) V

(p.u

.)

Vline

-0.100 -0.050 0.000 0.050 0.100 0.150 0.200 0.250

d) P

, Q (M

W/M

VAr)

Pgen Qgen Pload

0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 0.260

e) P

(MW

)

Pload Pgrid

Time (s) Figure 11 - (Case DP-2) Governor type B, Constant impedance load: 80% increases

in DG output power


DP-2: 80% load change, RL load

Sec. 15.0 20.0 25.0 30.0 35.0 40.0

59.960 59.980 60.000 60.020 60.040 60.060 60.080 60.100

a) G

en. S

peed

(Hz)

wr2 wr

0.9950 0.9960 0.9970 0.9980 0.9990 1.0000 1.0010 1.0020 1.0030 1.0040 1.0050

b) V

(p.u

.)

Vline2 Vline

0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200

c) P

(MW

)

Pgen2 Pgen

-0.0600 -0.0550 -0.0500 -0.0450 -0.0400 -0.0350 -0.0300 -0.0250 -0.0200 -0.0150 -0.0100

d) Q

(MVA

r)

Qgen2 Qgen

0.2200 0.2250 0.2300 0.2350 0.2400 0.2450 0.2500 0.2550 0.2600

e) P

(MW

)

Pload2


DP-2: 80% load change, RL load

Sec. 15.0 20.0 25.0 30.0 35.0 40.0

59.960 59.980 60.000 60.020 60.040 60.060 60.080 60.100

wr2 wra)

Gen

. Spe

ed (H

z) Governor type A Governor type B

0.9950 0.9960 0.9970 0.9980 0.9990 1.0000 1.0010 1.0020 1.0030 1.0040 1.0050

Vline2 Vline

b) V

(p.u

.)

0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200

Pgen2 Pgen

c) P

(MW

)

Governor type A

-0.0600 -0.0550 -0.0500 -0.0450 -0.0400 -0.0350 -0.0300 -0.0250 -0.0200 -0.0150 -0.0100

Qgen2 Qgen

d) Q

(MVA

r)

0.2200 0.2250 0.2300 0.2350 0.2400 0.2450 0.2500 0.2550 0.2600

Pload2 Pload

e) P

(MW

)

Figure 12 - (Case DP-2) Comparison between Governor Type A (Blue curves) and Type B (Green curve): 80% step change in DG output power

Time (s)


DP-2: 80% load change, RL / PQ load models

Sec. 15.0 20.0 25.0 30.0 35.0 40.0

59.980

60.000

60.020

60.040

60.060

60.080

a) G

en. S

peed

(Hz)

wr wr2

0.9950 0.9960 0.9970 0.9980 0.9990 1.0000 1.0010 1.0020 1.0030 1.0040 1.0050

b) V

(p.u

.)

Vline Vline2

0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200

c) P

(MW

)

Pgen Pgen2

-0.0550 -0.0500 -0.0450 -0.0400 -0.0350 -0.0300 -0.0250 -0.0200 -0.0150

d) Q

(MVA

r)

Qgen Qgen2

0.2200 0.2250 0.2300 0.2350 0.2400 0.2450 0.2500 0.2550 0.2600

e) P

(MW

)


DP-2: 80% load change, RL / PQ load models

Sec. 15.0 20.0 25.0 30.0 35.0 40.0

59.980

60.000

60.020

60.040

60.060

60.080

a) G

en. S

peed

(Hz)

wr wr2

0.9950 0.9960 0.9970 0.9980 0.9990 1.0000 1.0010 1.0020 1.0030 1.0040 1.0050

b) V

(p.u

.)

Vline Vline2

0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200

c) P

(MW

)

Pgen Pgen2

-0.0550 -0.0500 -0.0450 -0.0400 -0.0350 -0.0300 -0.0250 -0.0200 -0.0150

d) Q

(MVA

r)

Qgen Qgen2

0.2200 0.2250 0.2300 0.2350 0.2400 0.2450 0.2500 0.2550 0.2600

e) P

(MW

)

Pload Pload2

Time (s) Figure 13 - (Case DP-2) Comparison between the constant impedance load model (Blue curves) and the

constant power load model (Green curves): 80% step change in DG output power


7.2 Islanding transitions

The distribution feeder in Figure 1 is equipped with a automatic reclosure (represented by CB1) that employs a reclosing time of 1.5 seconds for the first reclosure attempt following a fault downstream of the feeder and the CB1 tripping. The utility company requires disconnection of the DG unit before restoration of the grid supply to avoid an out-of-step reclosing and damage to the line equipment. Normally, a Transfer-Trip (TT) scheme based on fast communication of the breaker position and deactivation of the inter-tie breaker of the DG unit is requested to ensure rapid disconnection prior to reclosing of the CB1. However, TT scheme is very expensive and obviously the economics associated with a small DG project does not allow such expenses. Investigation of alternative solutions is suggested in the DG standards and interconnection guidelines such as IEEE Std. 1547, [1], and the draft CSA C22.3, [2], to facilitate implementation of reliable and cost-effective options.

Assessment of an effective islanding detection protection requires dynamic analysis of the local distribution system behaviour subsequent to disconnection from the grid to assess the reliability of the solution. The main objective is to study Non Detection Zones (NDZ) of an anti-islanding protection scheme and apply corrective measures to prevent unintentional islanding of the local distribution system. Various levels of power mismatch between the load and the local generation at the time of disconnection are examined to determine possible NDZ.

The following case studies are performed to investigate voltage and frequency transients of the system and dynamic response of the DG unit for unintentional islanding of the local distribution system in Figure 1:

Case IS-1: +50% and +25% power mismatch levels between the load and generation (under generation),

Case IS-2: 0% power mismatch for RL and PQ load models (close match between load and generation),

Case IS-3: 0% power mismatch for resistive load,

Case IS-4: -15% power mismatch (over generation).

For all the studies, it is assumed that the DG unit operates in a constant power mode with a leading power factor of 0.97 to import reactive power from the grid, Section 4. However, variations in the load model and/or the operating point of the DG unit will be applied to investigate uncertainty in the load and the system parameters. The focus of the study is on the two-second interval following the disconnection, since the first reclosing attempt is assumed to occur in 1.5 s.


7.2.1 Case IS-1: +50% and +25% power mismatch levels

For this study, two power mismatch levels of +50% and +25% between the load and the DG output power prior to disconnection of the CB1 are analyzed. Considering the feeder load at the minimum level, 245-kW with a lagging power factor of 0.95, the DG output power is adjusted to 123-kW for the 50% mismatch and to 180-kW for +25% mismatch. The reactive power of the DG unit is controlled through the excitation system to import -31.00-KVAr from the grid.

Figure 14 shows the frequency, voltage and power variations of the load and DG subsequent to disconnection of CB1 at t = 20s. The blue curves illustrate variations of each quantity for +50% mismatch and the green curves correspond to +25% mismatch. It should be noted that after disconnection from the grid the electrical output power of the DG unit is determined by the load current in the isolated system, Figure 14-c, and Figure 14-d. The difference between the electrical load and the mechanical power on engine-generator shaft accelerates and/or decelerates the synchronous generator which causes frequency deviations. The voltage variations are dominantly affected by the changes in the reactive power balance and mainly follow the variations in the internal voltage of the synchronous generator which is in turn controlled by the excitation system. On the other hand, changes in the bus voltage vary the real power of the load, which affects the power balance and subsequently causes frequency deviations.

As expected, the voltage and frequency excursions for the case of +50% mismatch are significantly higher that those of +25% mismatch, Figure 14-a, and Figure 14-b. In both cases, the bus voltage at the PCC drops below 0.8 p.u., and the reactive power of the DG unit shows a sudden change in the direction, from -31.00-kVAr to +80-kVAr, Figure 14-d. However, the power frequency, depicted by the rotational speed of the generator, appears to behave differently for the case of +50% mismatch compared with that of +25%, Figure 14-a.

The frequency variations are highly dependent on the dynamics of the load. Figure 14-e shows that the load is affected by variations in the line voltage when a constant impedance load model is used. Therefore, for the case of +50% mismatch, the significant difference between the load and generation causes a steady reduction in the power frequency to 57.3 Hz within the one-second interval following the grid disconnection, Figure 14-a. However, when the power mismatch level is reduced to +25%, it changes the slope at which the frequency oscillations are initiated (positive slope).

It should be noted that, in both cases, disconnection of the grid causes a sudden drop in the line voltage, which affects the reactive power output of the DG and reverses the flow direction toward the grid, Figure 14-d.

7.2.2 Case IS-2: 0% power mismatch for RL and PQ load models

Figure 15 shows the system response to an accidental switching event that leads to disconnection of the main grid and islanding of the distribution feeder, which is modeled by opening CB1 at t=20 s.


Prior to disconnection of the grid, the feeder load and generation is closely matched to represent an unintentional islanding case with 0% power mismatch. The DG unit supplies 180-kW of real power at 0.97 leading power factor. The load is modeled by a parallel RL impedance representing 180-kW/60-kVAr at the rated voltage.

The simulation results of Figure 15-b, and Figure 15-c show that disconnection from the grid causes significant deviations in the line voltage and power frequency, although the real power of the load and DG is closely matched. This is mainly caused by the sudden voltage drop in the system due to the shortage of the reactive power required by the load, the distribution lines and the DG with a leading power factor. Figure 15-d also shows a sudden change in the reactive power output of the generator initiated by the voltage drop. Either voltage variations or the change in the reactive power direction of DG unit can be used to detect loss of the grid for this case.

The range of frequency excursions shown in Figure 15-b is highly dependent on the load type. In Figure 16 a comparison is made between the simulations results for two load models, namely RL and PQ load models, for an islanding scenario similar to that of Figure 15 with 0% power mismatch. The real power of the load in response to islanding transients is varied differently for the case of a RL model in comparison to a PQ model, Figure 16-e. Hence, the frequency deviations for a PQ load representation (Δf = 0.82 Hz) are lower than a RL load model (Δf = 1.25 Hz), Figure 16-a. However, the voltage and frequency variations are still large enough to be detected by under/over voltage and frequency protection schemes. Figure 16-d also shows similar changes in the reactive power of the generator subsequent to disconnection which can be detected by a directional reactive power relay.

7.2.3 Case IS-3: 0% power mismatch for resistive load

Figure 17 shows the simulation results of an islanding case based on a close match between the feeder load and power generation of the local source at the time of the grid disconnection. Simulation conditions are the same as those of Section 7.2.2; only the feeder load is assumed to be purely resistive (pF = 1.0). System responses for two types of governor models, Type A (blue curves), and Type B (green curves) are demonstrated in Figure 17. Change in the governor model only affects the frequency deviations. Comparing the simulation results of Figure 17 with those of Figure 15, significant reductions can be observed in the range of line voltage variations and the frequency excursions. Although the frequency deviations in this case may not exceed the under/over frequency protection thresholds of the frequency protective devices (typically ± 0.5 Hz), the voltage variations are large enough to be detected by under/over voltage protection of the generator, Figure 17-b. Voltage variations have also caused a sudden change in the reactive power direction of the generator, Figure 17-d, detectible by a directional reactive power relay.

This case may be considered as the worse case islanding scenario based on the transient behaviour observed from the simulation results and limited range of variations in the parameters which can be


used for islanding detection. However, by applying appropriate settings, a protection method based on changes in the reactive power direction can be suggested as an effective detection measure.

7.2.4 Case IS-4: - 15% power mismatch (over generation)

This case analyzes an islanding condition for which the DG output power is 15% more than the feeder load at the time of disconnection. The over generation condition will not be observed in the study system of Figure 1 as the rated capacity of the DG unit is significantly lower than the minimum load of the feeder. However, this case aims to investigate an operating condition of the study system expected in the near future, in which a second DG unit of similar technology and maximum capacity of 100-kW will be installed on the same feeder. The second DG will be located about 3-km downstream of the first DG. The future installation of the additional DG unit would represent a case of 15% over generation if both DG units operate at their rated capacities (280-kW generation) and the feeder load is about 245-kW (minimum loading condition of the feeder).

Figure 18 shows the simulation results for a -15% power mismatch condition. Prior to the disconnection of the grid, the DG unit supplies 180-kW and absorbs -31-kVAr. The feeder load is reduced to 157-kW/51-kVAr to model the suggested power mismatch level. As observed in Figure 18-b, and Figure 18-c, over generation of the DG unit with respect to the feeder load leads to large voltage and frequency excursions beyond the normal tripping thresholds of the under/over voltage and under/over frequency protection devices. Direction of the reactive power flow of the DG unit in Figure 18-d is also affected by the islanding condition. Changes in the reactive power direction can be detected by a directional power relay to trip the tie-line breaker upon reverse reactive power flow.



IS-1: 50% and 25% mismatch

Sec. 19.40 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60

57.0

58.0

59.0

60.0

61.0

a) G

en. S

peed

(Hz)

wr wr2

0.650 0.700 0.750 0.800 0.850 0.900 0.950 1.000

b) V

(p.u

.)

Vline Vline2

0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 0.260

c) P

(MW

)

Pgen Pgen2

-0.040

-0.020

0.000

0.020

0.040

0.060

0.080

d) Q

(MVA

r)

Qgen Qgen2

0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 0.260

e) P

(MW

)

Pload Pload2

+ 25% mismatch

+ 50% mismatch

Time (s)

Figure 14 - (Case IS-1) Islanding at t=20 s. for +50% (blue curves) and +25% (green curves): RL load model, governor Type B

IS-2: 0% mismatch, RL load

Sec. 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60 21.80 22.00

59.80 60.00 60.20 60.40 60.60 60.80 61.00 61.20

a) fr

eq. (

Hz)

freq.

59.80 60.00 60.20 60.40 60.60 60.80 61.00 61.20

b) G

en. S

peed

(Hz)

wr

0.650 0.700 0.750 0.800 0.850 0.900 0.950 1.000

c) V

(p.u

.)

Vline

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

d) P

, Q (M

W/M

VAr)

Pgen Qgen Pload

-0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175 0.200

e) P

(MW

)

Pload


IS-2: 0% mismatch, RL load

Sec. 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60 21.80 22.00

59.80 60.00 60.20 60.40 60.60 60.80 61.00 61.20

a) fr

eq. (

Hz)

freq.

59.80 60.00 60.20 60.40 60.60 60.80 61.00 61.20

b) G

en. S

peed

(Hz)

wr

0.650 0.700 0.750 0.800 0.850 0.900 0.950 1.000

c) V

(p.u

.)

Vline

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

d) P

, Q (M

W/M

VAr)

Pgen Qgen Pload

-0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175 0.200

e) P

(MW

)

Pload Pgrid

Figure 15 - (Case IS-2), Islanding of the distribution feeder at 0% power mismatch, RL load, Governor Type B



IS-2: 0% mismatch RL / PQ load

Sec. 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60 21.80 22.00

59.80 60.00 60.20 60.40 60.60 60.80 61.00 61.20

a) G

en. S

peed

(Hz)

wr2 wr

0.650 0.700 0.750 0.800 0.850 0.900 0.950 1.000

b) V

(p.u

.)

Vline2 Vline

0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240

c) P

(MW

)

Pgen2 Pgen

-0.040 -0.030 -0.020 -0.010 0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070

d) Q

(MVA

r)

Qgen2 Qgen

0.080 0.090 0.100 0.110 0.120 0.130 0.140 0.150 0.160 0.170 0.180

e) P

(MW

)

Pload2 Pload

RL load model

PQ load model

PQ load model

RL load model

Time (s)

Figure 16 - (Case IS-2) Islanding at 0% power mismatch, comparison between RL load model (green curves) and PQ load model (blue curve)


IS-2: 0% mismatch, Resistive Load (pf=1)

Sec. 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60 21.80 22.00

59.950 60.000 60.050 60.100 60.150 60.200 60.250 60.300 60.350 60.400

a) G

en. S

peed

(Hz)

wr2 wr

0.840 0.860 0.880 0.900 0.920 0.940 0.960 0.980 1.000 1.020

b) V

(p.u

.)

Vline2 Vline

0.120 0.130 0.140 0.150 0.160 0.170 0.180 0.190

c) P

(MW

)

Pgen2 Pgen

-0.0350 -0.0300 -0.0250 -0.0200 -0.0150 -0.0100 -0.0050 0.0000 0.0050 0.0100

d) Q

(MVA

r)

Qgen2 Qgen

0.130

0.140

0.150

0.160

0.170

0.180

e) P

(MW

)

Pload Pload2

Governor type B

Governor type A

Time (s)

Figure 17 - (Case IS-3) Islanding of the distribution feeder for 0% power mismatch and resistive load: Governor Type A (Blue curves), Governor Type B (Green curves)


IS-4: -15% mismatch, RL load

Sec. 19.60 19.80 20.00 20.20 20.40 20.60 20.80 21.00 21.20 21.40 21.60 21.80 22.00

59.75 60.00 60.25 60.50 60.75 61.00 61.25 61.50 61.75 62.00

a) fr

eq. (

Hz)

freq.

59.75 60.00 60.25 60.50 60.75 61.00 61.25 61.50 61.75 62.00

b) G

en. S

peed

(Hz)

wr

0.650 0.700 0.750 0.800 0.850 0.900 0.950 1.000

c) V

(p.u

.)

Vline

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

d) P

, Q (M

W/M

VAr)

Pgen Qgen Pload

-0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175

e) P

(MW

)

Pload Pgrid

Time (s)

Figure 18 - (Case IS-4) Islanding of the feeder based on -15% power mismatch by reducing the minimum load of the feeder

8 Discussions

The study results of a series of dynamic simulation analyses were presented that investigate the transient behaviour and voltage/frequency dynamics of a rural distribution feeder with DG interconnection subsequent to disconnection from the main grid. The principal objective of this study was to analyze the range of variations and typical patterns of changes for specific power quality parameters that can potentially be used to design passive anti-islanding protection schemes. The main parameters that were observed are:

frequency deviations (Δf),

voltage variations (ΔV),

changes in the reactive power direction (ΔQ).

Table 3 summarises approximate range of variations of the aforementioned power quality parameters over a 1 second interval following the grid disconnection (Δt). These values are obtained from the study results reported in the previous islanding simulation cases.

Table 3 - Summary of the observed variations in the selected power quality parameters

Power mismatch

Generation kW/kVAr

Load kW/kVAr Max Δf

0 < Δt < 1 s. Max ΔV 0 < Δt < 1 s.

Max ΔQ 0 < Δt < 1 s.

+50% 123 / -31 245 / 80 - 2.5 Hz - 23 % -31 to +80 kVAr +25% 180 / -31 245 / 80 ± 0.9 Hz - 33% -31 to +70 kVAr 0% (RL load) 180 / -31 180 / 60 + 1.25 Hz - 33% -31 to +55 kVAr 0% (R load) 180 / -31 180 / 0.0 ± 0.38 Hz - 15 % -31 to +7 kVar -15% 180 / -31 157 / 51 + 1.8 Hz -33 % -31 to +50 kVAr

As shown in Table 3, significant variations in the bus voltage above the typical ± 10% deviations (measured at the PCC) and sudden changes in the reactive power direction of the generator are expected under any load/generation mismatch condition. However, in some cases, for instance in case of supplying resistive load under a balance load and generation (0% mismatch), the frequency excursions may not be large enough to be detected based on typical ± 5% deviation settings.

The specific application reported in this study is based on a constant real/reactive power control mode. The DG controls tends to adjust the DG output power at pre-specified values even after disconnection from the grid. This control strategy will cause maximum deviations in the voltage and frequency of the generator if the output power of the generator (both real and reactive power) is not closely matched with the load prior to the disconnection from the grid.

As observed from the reported simulation cases, the governor model and type of speed/load controls have significant impact on the range of frequency excursions upon disconnection from the main grid. Typical speed/load control blocks were introduced and sensitivity of simulation results for two


speed/load control models with different control strategies was analyzed. The reported results in Table 3 are for a speed/load control model, whose control parameters were validated using the field tests results on the engine-generator.

The significant voltage drop and steady decreases in the PCC voltage after the grid disconnection, as shown in the reported studies, are the main result of the voltage control strategy applied for the DG unit. To achieve an importing reactive power condition, a power factor controller was used to generate corresponding voltage reference point for the Automatic Voltage Regulator (AVR) of the generator under any operating point. Hence, the reference point for the AVR is set by a proportional-integral controller (Figure 4). This mode of operation is only applicable during the grid-connected operation of the generator. The voltage control mode is automatically changed to a voltage regulation mode by the DG supervisory control system when the generator disconnection from the grid is recorded, for instance by monitoring the open/close position of the inter-tie circuit breaker.

Therefore, the assumption in the reported studies is that changes in the voltage control mode do not occur in the 2-second interval following the grid disconnection. That means the power factor controller still tends to adjust the reference point for the AVR to import reactive power. Hence, the AVR continues to reduce the generator voltage below the voltage at the PCC to change an exporting reactive power condition to an importing case. If this control strategy is continued, it will cause saturation of the PFC and the limitation of the AVR voltage reference point. Then, the output voltage of the generator will settle to a low voltage level for which a balance between the real/reactive power of the generator and the islanded load can be obtained. For clarification purposes, simulation results for a similar operating strategy and over an extended period of simulation is given in Appendix D.


9 Conclusions and Future investigations

The dynamic behaviour of a biogas engine generator for several islanding cases was studied and the voltage, frequency and reactive power deviation ranges were determined. The main objective of the study was to identify power quality related parameters that can be measured locally at the generator site and used to detect a loss of the main grid condition (islanding) with high degree of reliability.

An alternative anti-islanding protection scheme based on a combination of voltage measurement and detection of the reactive power direction is proposed. The reported simulation studies showed that, subsequent to islanding, the voltage deviations and the changes in the reactive power flow direction of the generator are significant enough to trigger tripping of a protective relay with a selected set of voltage and reactive power thresholds. An islanding detection method based on the frequency measurement scheme can also be applied as the secondary protection. However, special attention needs to be given to the selection of appropriate thresholds and tripping logic as a function of the various parameters. For instance, it is suggested to combine the directional reactive power tripping with a minimum real power export to avoid nuisance tripping on load changes or generator start-up during the grid-connected mode.

In continuation of the studies presented in this report, the following topics are suggested for future investigations:

Modeling and islanding protection studies of distribution systems with two or more DG units connected to the same feeder or installed on adjacent feeders which are supplied from a common distribution substation;

Investigate application limits of an anti-islanding protection scheme based on the directional reactive power; the proposed studies include analysis of the frequency and voltage transients for long feeders, underground feeders, and feeders equipped with fixed and/or switched capacitor banks at various points along the feeder;

Implementation of the typical protection relay models used for both feeder and generation protection and impact analysis of transient phenomena and system dynamics on protection equipment behaviour.


10 Reference

[1]. IEEE Std. 1547-2003, “IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems”, IEEE Standards Coordinating Committee 21, July 2003.

[2]. CSA Draft standard on Interconnection of Distributed Resources with Electricity Supply Systems, Canadian Standards Association Std. CSA C22.3 No. 9, 2007.

[3]. IEEE Std. 1547.1-2005, “IEEE Standard Conformance Test Procedures for Equipment Interconnecting Distributed Resources with Electric Power Systems”, IEEE Standards Coordinating Committee 21, July 2005.

[4]. W. Xu, K. Mauch, S. Martel, “An Assessment of Distributed Generation Islanding Detection Methods and Issues for Canada”, report # CETC-Varennes 2004-074 (TR), CANMET Energy Technology Centre – Varennes, Natural Resources Canada, July 2004, 53 pp.

[5]. W. Bower, M. Ropp, “Evaluation of islanding detection methods for photovoltaic utility-interactive power systems”, Report, International Energy Agency, IEA-PVPS T5-09, March 2002, [Online]. Available: http://www.iea-pvps.org/products/rep5_09.htm

[6]. Z. Ye, A. Kolwalkar, Y. Zhang, P. Du, and R. Walling, ““Evaluation of anti-islanding schemes based on non detection zone concept”, IEEE Transactions on Power Electronics, vol. 19, no. 5, pp. 1171–1176, 2004.

[7]. J. L. Blackburn and T. J. Domin, Protective Relaying: Principles and Applications. CRC Press, 2006.

[8]. G. Rockefeller, “Generator protection application guide”, Basler Electric, Rev. C.0, June 2001, [available on www.basler.com for download].

[9]. W. Freitas, et al., “Comparative analysis between ROCOF and Vector Surge relays for distributed generation applications”, IEEE Transactions on Power Delivery, vol. 20, no. 2, pp. 1315-1324, April 2005.

[10]. Basler Electric, “BE1-40Q; Loss of excitation relay”, General specification bulletin UBW-3, 12-04, [available on www.basler.com for download].

[11]. MAN, “Biogas Engine E 2876 LE 302”, Technical Data Sheet, 2006.

[12]. P. C. Krause, Analysis of Electric Machinery and Drive Systems, IEEE Press, 2002.

[13]. STAMFORT Power Generation , “HCI 434C/444C”, Technical Data Sheet, 2002, [online]: http://www.cumminsgeneratortechnologies.com/en/products/stamford/hcandp/hci4hcm4hcki4/


http://www.iea-pvps.org/products/rep5_09.htm

http://www.basler.com/

http://www.basler.com/

http://www.cumminsgeneratortechnologies.com/en/products/stamford/hcandp/hci4hcm4hcki4/

http://www.cumminsgeneratortechnologies.com/en/products/stamford/hcandp/hci4hcm4hcki4/

[14]. STAMFORT Power Generation, “MX341 Automatic Voltage Regulator (AVR)”, Technical Data Sheet, [online]: http://www.cumminsgeneratortechnologies.com/en/products/stamford/avr/forshort-circuitmaintenance/

[15]. IEEE Std. 421.5-1992, IEEE Recommended practice for excitation system models for power system stability studies, 1992.

[16]. Wexler Computer System Development Ltd., “GENCON II Pro Feature”, Technical Data Sheet, March 2006, [online]: http://www.wexler.co.il/features.pdf.

[17]. “Catalogue of CHP technologies”, U.S. Environmental Protection Agency Combined Heat and Power Partnership, EPA technical report, 2002.

[18]. P. Kundur, Power System Stability and Control, New York: McGraw-Hill, 1994.

[19]. Governors America Corp., “ESD5330 Series – Speed Control Unit”, Product Technical Information PTI 1041, 1999, [online]: http://www.governors-america.com/

[20]. IEEE Task Force, “Load representation for dynamic performance analysis [of power systems],” IEEE Trans. Power Syst., vol. 8, no. 2, pp. 472–482, May. 1993.

[21]. Beckwith Electric Co. “Intertie/generator protection relay M-3410A”, Product specifications, [online]: http://www.beckwithelectric.com/relays/m3410/m3410a.htm .


http://www.cumminsgeneratortechnologies.com/en/products/stamford/avr/forshort-circuitmaintenance/

http://www.cumminsgeneratortechnologies.com/en/products/stamford/avr/forshort-circuitmaintenance/

http://www.wexler.co.il/features.pdf

http://www.governors-america.com/

http://www.beckwithelectric.com/relays/m3410/m3410a.htm


Appendix A: Model parameters



Appendix A: Model parameters 1. Distribution feeder: Feeder parameters for a ACSR 336 kcmil conductor size (neutral 2/0) at 50oC are given in the following table.

R1 = 0.1883 Ω/km R0 = 0.5297 Ω/km X1 = 0.4054 Ω/km X0 = 1.2999 Ω/km B1 = 4.0301x10(-9) mho/km B0 = 1.576x10(-9) mho/km

2. IEEE Type ACA2 excitation model [15]:


3. PQ load model representation [18]: A combination of residential and commercial load in summer condition is considered. The parameters of the model are given as follows.

Kpv = 1.3 , Kqv = 3.2 Kpf = 1.1 , Kqf = -1.9


Appendix B: Synchronous generator parameters



Appendix B: Synchronous generator parameters [13]



Appendix C: Gas engine detail model



Appendix C: Gas engine detail model 1. Speed/load control:

The three types of speed/load control models used are described in Figure 6, Figure 7, and Figure 8.

2. Prime mover model: The prime mover model essentially consists of several control blocks representing the dynamic characteristic of various parts of a mechanical engine. The main blocks, as shown in the following control diagram, include (1) the fuel valve (gate), (2) the fuel injection system, and (3) cylinders behaviour.

FLimiter engine

eValve fuel system

T

The control parameters are: Tv = 0.12 s, Tf = 0.05 s, b= 25, Max = 1.2 p.u, Min = 0.0



Appendix D: Simulation study of voltage control strategy based on the reactive power import



Appendix D: Voltage control strategy based on a constant reactive power import Figure 19 shows the simulation results for the case of generator islanding. In order to investigate the voltage dynamic behaviour shown in the reported islanding cases of Section 6, an extended period of 20 seconds is simulated. It is also assumed that the voltage control strategy of the generator subsequent to disconnection from the grid remains the same as during the grid-connected operation. That means the voltage reference point of the generator AVR is adjusted by a PFC for a constant reactive power importing operation. However, to achieve a stabilized operating point, a governor model based on a 4% speed-droop control is utilized.

Prior to unintentional islanding, a +10% mismatch between the generation and load is applied that requires about 40-kW of real power import from the grid, Figure 19-d. The DG unit supplies 130-kW and imports -30-kVAr of reactive power. The islanding occurs at t = 20 s for which the rotational speed of the generator undergoes damped oscillations to settle the power frequency at a higher value based on the speed-droop characteristic, Figure 19-b. However, the output voltage of the generator shows a sudden drop and then a steady decrease until reaches a constant value at which the power balance between the load and generation is obtained, Figure 19-c.

The simulation results of Figure 19 shows that, during islanding transition and autonomous operation, the dynamic behaviour and the steady-state operating point of the system are highly dependent on the types of voltage and speed controls applied to the DG unit. If the voltage control strategy does not allow voltage regulation of the generator terminal voltage through AVR, power balance requirements are generally achieved at an operating point well beyond the standard voltage ranges.


IS: 10% mismatch, RL load

Sec. 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0

58.75 59.00 59.25 59.50 59.75 60.00 60.25 60.50 60.75 61.00 61.25

a) fr

eq. (

Hz)

freq.

58.75 59.00 59.25 59.50 59.75 60.00 60.25 60.50 60.75 61.00 61.25

b) G

en. S

peed

(Hz)

wr

0.800 0.825 0.850 0.875 0.900 0.925 0.950 0.975 1.000 1.025

c) V

(p.u

.)

Vline

-0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175

d) P

, Q (M

W/M

VAr)

Pgen Qgen Pload

0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180

e) P

(MW

)

Pload Pgrid

Time (s) Figure 19 - Islanding of the DG unit for an importing reactive power control without changes in the voltage control mode


Documents

Computer Simulation Modeling and Analysis of the Dynamic … · 2012. 3. 20. · 1 Summary This report outlines the modelling, simulation analyses and investigation of dynamic behaviour