Electrical Power and Energy Systems 55 (2014) 704–713

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier .com/locate / i jepes

Load-following performance analysis of a microturbine for islandedand grid connected operation

0142-0615/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.ijepes.2013.10.018

⇑ Corresponding author. Tel.: +91 0326 2235644; fax: +91 0326 2296563.E-mail addresses: gauri1983@gmail.com (G. Shankar), vivek_agamani@yahoo.

com (V. Mukherjee).

G. Shankar, V. Mukherjee ⇑Department of Electrical Engineering, Indian School of Mines, Dhanbad, Jharkhand, India

a r t i c l e i n f o

Article history:Received 19 February 2013Received in revised form 27 September2013Accepted 12 October 2013

Keywords:MicroturbineLoad-following performanceDistributed energy resourcesSynchronous generator

a b s t r a c t

Modeling, simulation and performance analysis of a microturbine (MT) generator (MTG) system is carriedout in this paper. The MTG system is consisting of a MT coupled with a synchronous generator. The pro-posed model incorporates power, speed and voltage controller for maintaining constant speed and volt-age under variable loading condition. Modeling and simulation tasks are performed in MATLAB-SIMULINK platform for different loading conditions under isolated and grid connected modes. Perfor-mance study of the MTG system is carried out with and without both speed and voltage controller. Itis observed from the simulation work that the MTG along with speed and voltage controller performsquite well under load disturbances, thereby, renders its suitability as a viable option for playing a key roleas distributed generation for both isolated and grid connected mode of operation.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Distributed generation (DG) system is going to play a key role inbridging the gap between the rate at which electrical energy de-mand is increasing and the generation capacity being added. A re-cent trend of decentralization in electric power utility is creatingmore opportunities for high penetration of DGs, serving as compli-mentary options to the centralized energy system. DG may beoperated in dual mode with grid or without grid. Nevertheless,stand-alone DG systems are preferred more in hilly areas and re-mote villages where accessibility to the main grid is really a bigchallenge [1–3].

Apart from these, DGs are technically stable, economically fea-sible and environment friendly. These are small and efficient mod-ular generation systems [4–6]. Recently, there is a growing concernamong the researchers across the globe in developing microtur-bines (MTs) for DG applications owing to their quick start capabil-ity and easy controllability which may be useful for efficient peakshaving. Also, MTs render reliable and efficient operation alongwith lower maintenance cost and low greenhouse gas emission[7,8].

Microturbine generation (MTG) is a multi-fuelled generatingsystem, incorporating simple cycle gas turbine technology withpower generating capacity ranging between 25 and 500 kW. Itsuits best to meet peak load requirements of the consumer because

of its quick start capability. Mainly, two types of MT are reported inthe literature. One of them is very high speed, single-shaft MTGwhere generator and turbine are mounted on the same shaft whilethe other one is the split-shaft MTG system where a generator isconnected via a gearbox to a power turbine [9–11]. Addition ofDG affects the overall dynamics of the power distribution network,thereby, accurate modeling of MTG and its control have becomeinevitable to predict its grid and off-grid interaction in advance.Due to these reasons, researchers around the globe have been con-centrating hard to explore accurate dynamic model of the MTGsystem.

Detailed theory of the gas turbine is well presented by Cohenet al. in [12]. In [9,13,14], modeling of single-shaft heavy dutygas turbine and its performance dynamics with acceleration, tem-perature and speed control are discussed. A review of different gasturbine model, developed till now, is presented and compared in[15,16]. So far as microgas turbine is concerned, its governing prin-ciple resembles heavy duty gas turbine theory. Modeling, simula-tion and control of load-following performance for grid/off-gridoperations of MTG are well pursued in [17–21]. These works dealwith single-shaft microturbine coupled with high speed perma-nent magnet synchronous generator (PMSG). High frequency elec-trical power generated by PMSG, eventually, cannot be useddirectly by the consumer. As a result, interfacing of power elec-tronic devices between the MTG and the end user is inevitable.The usage of power electronic components results in conversionlosses and makes the overall system operation and control morecomplex. To overcome these complexities while modeling ofsingle-shaft MTG with power electronic components, split-shaft

G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713 705

modeling of MT with either induction generator (IG) or synchro-nous generator (SG) is reported in [9,22–25]. The gas turbine gov-ernor model (GAST) developed by General Electric (GE) is the mostcommonly used model to study the dynamic performance of gasturbine. In [25,26], dynamics behavior of parallel operation of hy-brid fuel cell and MTG system forming a microgrid is explored.Authors of [8,23,24,27] have used the GAST model for differentload-following performance studies of split-shaft MTG system un-der with or without grid connected mode of operation. Sis-worahardjo et al. in [24] have made a comparison betweencontroller based on artificial neural network and conventional PIcontroller for standalone MT power plants. Oguz et al. have madesimilar type of approach in [27].

In the present work, GAST model of a split-shaft MTG is consid-ered and simulated in MATLAB-SIMULINK� platform [28]. Slowdynamics of electro-mechanical system of a MTG is explored con-sidering different load scenarios for islanding as well as grid con-nected mode of operation. Along with active power controllerand speed controller, an additional voltage controller is also incor-porated in the present work for load-following performance studyof the proposed MTG system.

The rest of the paper is organized as follows. In Section 2, modelof split-shaft MT along with its control mechanism is presented.The parameters used in the studied MT model are illustrated inSection 3. Simulation results are presented and discussed in Sec-tions 4. Finally, conclusions of the present work are drawn inSection 5.

2. Split-shaft MT model

Most of the MTs use a single-shaft design configuration in whichshaft rotates both the inlet air compressor and the generator. Single-shaft models are designed to operate at higher speed, typically, in therange of 50,000–120,000 rpm. These systems generate high-frequency alternating current (AC) which is rectified to direct currentand, finally, inverted to 50–60 Hz AC using power electronic inter-face system [17–21]. Single-shaft MTs are incorporated with mainlythree primary controllers viz. speed controller, temperature control-ler and acceleration controller. Each of these controllers modulatesfuel control block and overall turbine dynamics, as and when re-quired. Speed controller (being the primary controller) acts owingto the mismatch between reference speed and actual rotor speed.Temperature controller monitors the upper limit of the outputpower generated. Acceleration controller acts mainly to regulatethe rate of acceleration of rotor during start-up [23].

On the other hand, twin-shaft MTs use two turbines. One isused to drive the air compressor while the other is used to drivethe generator via a gear box. Exhaust coming out from the com-pressor turbine powers the generator turbine as shown in Fig. 1.The gear box is used to reduce the speed to 3600 rpm. With thepressure ratio splited between the two turbines, the lower outputshaft speed of the second stage turbine is more conducive to di-rectly accommodate a conventional generator such as an IG or aSG without any requirement of power electronic interface [29]. Arecuperator and a waste heat recovery or a heat exchanger is incor-porated to enhance overall efficiency and power generation output.The recuperator captures thermal energy of the exhaust from thepower turbine for preheating compressed air whereas heat ex-changer captures the exhaust energy to meet the heating loadrequirement of the consumers, if any. [8,22–27].

In this paper, a twin-shaft model of MT is used. As main focus ofthe present work is to explore the slow dynamics of the electro-mechanical systems under normal operating condition, recupera-tor and the heat exchanger (both being only system efficiencyraising components) are omitted from the proposed model. Forsimplicity, actual temperature control which uses thermocouple

for sensing exhaust temperature and acceleration control blocksare neglected in our studied model. It may be noted here thatthe temperature and acceleration control blocks do not participateunder normal operating condition. In the present work, a speedcontroller is used in place of governor of MT system [23].

2.1. Split-shaft MT and its control

In the present work, the most commonly used GAST model ofthe MT (developed by GE) is simulated to study the load-followingbehavior. SIMULINK-based GAST model of MT with speed control-ler and active power controller is shown in Fig. 2. The parameterdetails of the MT model may be found in [23] and are listed in Ta-ble 1. As shown in Fig. 2, speed control is realized by incorporatinga conventional proportional-integral (PI) controller which is usedto control the error between reference speed (xref) and the actualspeed xact of the rotor. Similarly, from Fig. 2, it may be observedthat the mechanical power output from the turbine (which in turngoverns active power output from generator) is controlled by ananother PI controller. The input to this PI controller is the differ-ence between reference active power (Pref) and the actual activepower generated (Pact). The outputs of the speed controller (x1), ac-tive power controller (x2) and temperature control block (x3) aregiven as inputs to a low value gate (LVG) which in turn governsthe convenient fuel flow rate. Output of the LVG is given to fuelopening valve (FOV) block, represented by a first order transferfunction (having time constant T1) with maximum and minimumvalve opening limits denoted by FOVmax and FOVmin, respectively.Depending upon LVG output, FOV actuates the fuel system block(having fuel system control time constant of T2) to produce re-quired fuel flow rate. Exhaust temperature block is representedby a first order time constant block with load limit time constantshown as T3. The values of load limit (Lmax) and FOVmax are takenas 1.2 (assuming that MT has 120% peak power capacity).

2.2. Simplified SG model and its control

A simplified SG model, as given in [23], is used for the simula-tion work of the present paper. In the present work, a predefinedSG model (as available in SimPowerSystems toolbox of MATLAB™[28]) is used which models both the electrical and mechanicalcharacteristics of the SG. The parameters of the studied SG modelare mentioned in Table 2. The real power output of the generatoris controlled by a PI controller. In the present work, terminal volt-age of the MTG system is maintained at desired level by using avoltage regulator whose input is the output of an another PI con-troller. This voltage control mechanism, in islanded mode, is be-yond the scope of the work reported by Saha et al. in [23]. Butthis voltage control mechanism, in islanded mode, is consideredin the present work. Input to the PI controller used in voltage con-trol mechanism is the mismatch between reference voltage (Vref)and the actual terminal voltage per phase (Vphase).

These controllers must work in unison to achieve two impor-tant targets viz. (a) meeting the power requirement at the con-sumer end and (b) keeping both the terminal voltage and thespeed deviation (i.e. frequency) within their prescribed limits. Ahigh quality power is expected from the MTG system when thesetwo targets are fulfilled. The block diagram of the MTG systeminterconnected with the utility grid is shown in Fig. 3. A 150 kVA,440 V, 60 Hz MTG system is interconnected with the utility gridvia a 200 kVA, 11 kV/440 V, 60 Hz, Y–D transformer. The utilitygrid is modeled as a simple RL equivalent source with short circuitlevel 500 kVA with a load of 5 kW [23]. The off-grid and gridconnected operation of the system may be realized by opening orclosing of the circuit breaker located at the point of commoncoupling (PCC).

Fig. 1. Block diagram of twin-shaft MT.

Fig. 2. Block diagram of GAST MT model [8].

Table 1Parameters of the MT model.

Symbol Meaning Value

Dtur Turbine damping 0.03FOVmax Maximum valve position 1.2FOVmin Minimum valve position �0.1Ki_pc Power controller integral gain 50.0Ki_sc Speed controller integral gain 40.0Ki_vc Voltage controller integral gain 0.5Kp_pc Power controller proportional gain 10.0Kp_sc Speed controller proportional gain 680.0Kp_vc Voltage controller proportional gain 0.005KT Temperature control loop gain 1.0Lmax Load limit 1.2Prated,MT Rated power, kVA 150Pref Reference power, p.u. 1.0T1 Fuel system lag time constant, s 10.0T2 Fuel system lag time constant, s 0.1T3 Load limit time constant, s 3.0Vref Reference voltage, p.u. 1.0xref Speed reference, p.u. 1.0

Table 2SG parameters.

Symbol Meaning Value

f Rated frequency, Hz 60 HzH Inertia constant, s 0.822KD Damping factor, p.u. 60.0P Number of poles 2Prated Rated power, kVA 150R Internal resistance, p.u. 0.02Vref Rated line to line voltage, V 440X Internal reactance, p.u. 0.3

706 G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713

3. Model parameters

The parameters used for the simulation of the studied MT, SGand grid are taken from the work of Saha et al. [23] and are illus-trated in Tables 1–3, respectively.

Fig. 3. Simulation block diagram of MTG with grid.

Table 3Distribution grid parameters.

Parameter Value

3-Phase source base voltage, kV 113-Phase source short circuit kVA, kVA 500Rated frequency, Hz 603-Phase source X/R ratio 6Distribution transformer power rating, kVA 200Distribution transformer primary voltage, kV 11Distribution transformer secondary voltage, kV 440

Fig. 4. MT mechanical power output and generator electrical power output(islanding mode-Scenario 1).

Fig. 5. SG rotor speed (islanding mode -Scenario 1).

G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713 707

4. Results and discussions

In the present work, load-following performances of the MTGsystem under isolated and grid connected modes are consideredand simulated. The analysis is carried out in per unit system con-sidering 150 kVA as the base kVA, 3600 rpm as the base speedand 11 kV as the base voltage.

4.1. Islanding operation of MTG

Scenarios considered vis-à-vis observations of load-followingperformances of the proposed MTG system in islanding mode ofoperation are presented below.

4.1.1. Islanding mode-Scenario 1In this scenario, performance of an isolated MTG system is ana-

lyzed without speed and voltage controller. In this scenario, simu-lation is performed for t = 90 s and set point for Pref is changedmanually with load. Initially, MTG system is assumed to be oper-ated with 30 kW (0.2 p.u.) of load. At t = 40 s, an additional loadof 90 kW (0.6 p.u.) is connected and corresponding change in Pref

is made to change turbine mechanical power output to meet theload requirement. Mechanical power output of turbine (PMECH)and electrical power output of generator (PSG) are plotted inFig. 4. The rotor speed (xr) and Vphase of the SG due to step changein load pattern are plotted in Figs. 5 and 6, respectively. It may beobserved from Fig. 4 that though the load is increased from 0.2 p.u.to 0.8 p.u but due to decrease in Vphase (as voltage controller is ab-sent) from 1 p.u. to 0.96 p.u. as shown in Fig. 6, the actual electricalpower delivered to the load is less than 0.8 p.u. As a result, thoughPMECH changes as per the predefined value of Pref but the PSG could

Fig. 6. SG terminal voltage per phase (islanding mode-Scenario 1).

Fig. 7. MT mechanical power output and generator electrical power output(islanding mode-Scenario 2).

Fig. 8. SG rotor speed (islanding mode-Scenario 2).

Fig. 9. SG terminal voltage per phase (islanding mode-Scenario 2).

708 G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713

not change up to 0.8 p.u. Hence, it is observed from Fig. 4 thatPMECH becomes greater than PSG which causes the rotor to acceler-ate and due to absence of speed controller, xr shoots up above1.0 p.u as shown in Fig. 5. However, if Pref would not have changedwith load, then PMECH would have been less than PSG causing rotorto decelerate and decrease in xr might have been observed in theabsence of both speed and voltage controller.

4.1.2. Islanding mode-Scenario 2In this scenario, both speed and voltage controllers are incorpo-

rated along with the active power controller of the MTG model andits overall performance is analyzed. The simulation is carried outfor t = 90 s. Simulation started with an initial load of 30 kW(0.2 p.u.), at t = 20 s an additional load of 90 kW (0.6 p.u.) isswitched on and further at t = 60 s, there is step increase in loadby 30 kW (0.2 p.u.). Profiles of PMECH and PSG for this scenario areplotted in Fig. 7. Profiles of xr and Vphase of the SG due to stepchange in load pattern (as considered in this scenario) are plottedin Figs. 8 and 9, respectively. It may be observed from Fig. 7 thatowing to step change in load demand at t = 20 s and t = 60 s, PMECH

follows PSG after overcoming the oscillatory response at the time ofchange in load. It may be noticed from Fig. 8 that the conventionalPI speed controller used in the model is successful in bringing backthe speed to 1.0 p.u. after the occurrence of the load disturbancesat t = 20 s and t = 60 s. It may be observed from Figs. 7 and 8 thatthe maximum settling time both in case of PMECH and xr is approx-imately 20 s for a large change in load of 0.6 p.u. occurring at

t = 20 s which is higher as compared to disturbances caused dueto small change in load of 0.2 p.u. at t = 60 s. A maximum peakovershoot of 0.831 p.u. and settling time of 1 s is observed in theprofile of PSG during larger load change at t = 20 s. A maximum0.1% dip in rotor speed from its nominal value is observed att = 20 s. Fig. 9 shows that Vphase momentarily dips due to step loadchange at t = 20 s and t = 60 s, but the installed voltage controllerchanges the generator internal voltage (Eg) accordingly to maintainVphase constant at 1.0 p.u. It is observed that due to large change inload at t = 20 s, a voltage dips to a minimum value of 0.782 p.u.from its nominal with settling time less than 0.5 s as comparedto load change at t = 60 s. However, due to fast acting dynamicsof voltage regulator circuit as compared to MT, Vphase could reachto 1.0 p.u. with a settling time of less than 1 s.

4.1.3. Islanding mode-Scenario 3In earlier two scenarios, MTG performances with resistive load

are considered. But in the present scenario, a R–L load is applied toan initial resistive load and the MTG performance is analyzedincorporating all controllers as used in scenario 2. Simulation forthis scenario is considered for 90 s. Simulation started with an ini-tial resistive load of 30 kW (0.2 p.u.). At t = 20 s, an another R–Lload of (90 kW + j 30 kVAR) is switched on and further at t = 60 s,an another R–L load of (15 kW + j 15 kVAR) is connected. Profilesof PMECH and PSG for this scenario are shown in Fig. 10. Profiles ofxr and Vphase of the SG due to step change in load pattern are plot-ted in Fig. 11(a) and (b), respectively. Profiles of PSG and reactivepower output of generator (QSG) are shown in Fig. 12. FromFig. 10, it may be inferred that the load-following performance ofthe active power controller installed in the model are being

Fig. 10. MT mechanical power output and generator electrical power output(islanding mode-Scenario 3).

Fig. 11. SG (a) rotor speed and (b) terminal voltage per phase (islanding mode-Scenario 3).

Fig. 12. Real and reactive power output of SG (islanding mode-Scenario 3).

G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713 709

monitored in this scenario also, as stated in the earlier two scenar-ios. As compared to previous case (Islanding mode-Scenario 2), att = 20 s, due to reduced power factor, oscillation with a little higheramplitude in the profile of PSG is observed, whereas the settlingtime remains more or less same, being less than 1 s in both cases.As compared to previous case (Islanding mode-Scenario 2), the ac-tion of speed and voltage controller may be noticed from Fig. 11in maintaining the speed and the voltage, respectively, at desiredlevel. Because of reactive power loading of 0.2 p.u. at t = 20 s, a lit-tle more drop in speed and the voltage is observed as compared toprevious case. However, settling time is observed to be more orless same from the profile of xr and Vphase, in both the cases. FromFig. 12 it may be observed that the profiles of both PSG and QSG fol-low the load requirement.

4.2. Grid connected operation of MTG

Scenarios considered vis-à-vis observations of load-followingperformances of the proposed MTG model in grid connected modeof operation are presented below. Both speed and voltage control-lers are taken into account in all the scenarios considered of thismode of operation.

4.2.1. Grid connected mode-Scenario 1In this scenario only resistive load is considered and simulation

is carried out for 90 s. Simulation started with an initial load of30 kW (0.2 p.u.) on MTG and 60 kW (0.4 p.u.) on grid. At t = 10 s,MTG is synchronized to the grid using technique as given in [30].At t = 15 s, an another load of 0.2 p.u. is applied to MTG and MTGis made to disconnect from grid at t = 20 s. Again at t = 40 s, MTGis re-synchronized and further an additional load of 0.2 p.u is ap-plied both on MTG and the grid at t = 50 s and t = 60 s, respectively.Finally, at t = 70 s once again MTG is disconnected from the grid.

Load-following performances of PMECH and PSG of the proposedMTG system are presented in Fig. 13. Profiles of PGRID and PSG arecompared in Fig. 14(a) and (b), respectively The variations of xr

and Vphase for this scenario may be observed from Fig. 15(a) and(b), respectively. These load-following performances justify the ac-tions of their respective controlling mechanisms. It may be ob-served from Fig. 13, prior to synchronization at t = 10 s andt = 40 s, both MTG and the grid were operating in steady state inislanding mode. As result, due to proper synchronization no shar-ing of load from MTG to grid and vice versa is observed. However,as the control topology remains the same in MTG during islandingand grid connected mode of operation, hence from Fig. 14, it may

Fig. 13. MT mechanical power output and generator electrical power output (gridconnected mode-Scenario 1).

Fig. 14. Real power output of (a) grid and (b) SG (grid connected mode-Scenario 1).

Fig. 15. (a) SG rotor speed and (b) SG terminal voltage per phase (grid connectedmode-Scenario 1).

Fig. 16. MT mechanical power output and generator electrical power output (gridconnected mode-Scenario 2).

710 G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713

be noticed that any change in load either at MTG side (0.2 p.u. ofload change each at t = 15 s and t = 50 s) or at grid side (0.2 p.u.of change in load at t = 60 s) during grid connected mode, is fullyshared by the grid only. This happens because droop governorbased load sharing has not been considered in this model. Loadshared by the grid from MTG during grid connected mode of

operation is fully transferred back to MTG when disconnected fromthe grid at t = 20 s and t = 70 s and PMECH follows PSG, accordingly asshown in Fig. 13. Fig. 15 shows that a change of 0.033% in xr is ob-served with a settling time of 10 s and 1.73% change in Vphase withsettling time of 2 s, when MTG is disconnected at t = 20 s from gridwhich is found higher as compared to when MTG once again dis-connected at t = 70 s.

4.2.2. Grid connected mode-Scenario 2In this scenario, simulation is performed for t = 90 s with R–L

load and the performance of MTG in grid connected mode of oper-ation is analyzed. Simulation started with initial R–L load of(30 kW + j 15 kVAR) on MTG and (45 kW + j 30 kVAR) on grid.MTG is made to synchronized with grid at t = 10 s and a load of(30 kW + j 15 kVAR) is further connected to MTG at t = 15 s. MTGis made to disconnect from grid at t = 20 s. Again at t = 40 s, MTGis re-synchronized and further an additional load of (15 kW + j15 kVAR) is applied on MTG at t = 50 s and a load of (30 kW + j15 kVAR) is connected to the grid at t = 60 s. Finally, at t = 70 s onceagain MTG is disconnected from the grid. Profiles of PMECH and PSG

following the load disturbances for this scenario are illustrated inFig. 16. Profiles of PGRID and PSG are presented in Fig. 17(a) and(b), respectively and variations of reactive power of grid (QGRID)and QSG are illustrated in Fig. 18(a) and (b), respectively. Profilesof xr and Vphase for this scenario may be observed from Fig. 19(a)and (b), respectively. It may be noticed from Fig. 16 that theload-following performance of the MTG is maintained in this sce-nario as well. When MTG is disconnected from grid at t = 20 sand t = 70 s, the loads shared by the grid from MTG are again trans-ferred back to MTG and consequently, PMECH changes accordinglyto meet the load changes occurred at MTG side. A maximum peakovershoot of 1% with settling time of 9 s is observed in the profileof PMECH following a large change in load at t = 20 s as compared toload change at t = 70 s. In this scenario too, at the time of synchro-nization at t = 10 s and t = 40 s, no load sharing between MTG andgrid is observed as seen from Figs. 16 and 17. However, it is alsoobserved that during grid connected mode of operation, loadchanges at MTG side at t = 15 s (30 kW + j 15 kVAR) and t = 50 s(15 kW + j 15 kVAR) is fully supported by the grid. And whenMTG is isolated from grid at t = 20 s and t = 70 s, automatically,load shared by the grid is transferred back to MTG. Oscillatory nat-ure in the profiles of PMECH and PSG of Figs. 16 and PGRID and PSG of 17are observed during load switching and synchronization withacceptable overshoots and settling times because of reactive nat-ure of the load. It may be noted from Fig. 19 that in this scenariotoo, both the speed as well as the terminal voltage per phaseprofiles are being settled down to 1.0 p.u. following the load

Fig. 17. Real power output of (a) grid and (b) SG (grid connected mode-Scenario 2).

Fig. 18. Reactive power output of (a) grid and (b) SG (grid connected mode-Scenario 2).

Fig. 19. SG (a) rotor speed and (b) terminal voltage per phase (grid connectedmode-Scenario 1).

G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713 711

disturbances as well as connection and disconnection events asconsidered in this scenario. Overshoots, undershoots and settlingtime in the profiles of xr and Vphase at the time of load disturbances,connection and disconnection events at different instant are wellunder acceptable limits.

4.3. MTG performance

For performance analysis of the MTG model, the model in thepresence of properly tuned active power controller, speed control-ler and the voltage controller is simulated for t = 350 s with resis-tive load and the results obtained is compared with other results.In this study, simulation started with no-load on MTG and a load

Fig. 20. MT mechanical power output (Ref [23] and simulated).

712 G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713

of 30 kW (0.2 p.u.) and 120 kW (0.8 p.u.) are connected to MTG att = 50 s and t = 200 s, respectively. The profiles of PMECH and xr arecompared with that of results reported in [23] and are shown inFigs. 20 and 21, respectively. It may be observed from Fig. 20 thatwhen a load of 0.2 p.u. is connected to MTG at t = 50 s, no peakovershoot and settling time of 15 s is observed as compared to a

Fig. 21. SG rotor speed (Ref [23] and simulated).

Fig. 22. Mechanical power output and generator electrical power output of (a)split-shaft MTG and (b) single-shaft MTG.

peak overshoot of 3.6% with settling time of 50 s as reported in[23]. At t = 200 s, a large load of 0.8 p.u. is connected to MTG anda peak overshoot of 2.5% with setting time of 45 s is observed ascompared to peak overshoot of 29.5% and settling time of 100 sas reported in [23]. From Fig. 20 it may be observed that decreasein rotor speed found to be same (in simulated result and in [23])following load switching at t = 50 s and t = 200 s. A higher peakovershoot is observed in the profile of xr of [23]. However, settlingtimes in the profile of simulated xr are found to be 5 s and 30 s cor-responding to load switching at t = 50 s and t = 200 s, respectively.The simulated result is much better as compared to settling time of80 s and 100 s in profile of xr as reported in [23], corresponding toload disturbances occurred at t = 50 s and t = 200 s, respectively.Higher settling time is observed in the profiles of PMECH and xr att = 200 s due to large switching of load of 0.8 p.u. as compared to0.2 p.u. of load disturbance at t = 50 s.

4.4. Comparison between split-shaft and single-shaft MT

A comparison of the load-following performances of split-shaftand single-shaft MT of similar rating under isolated mode of oper-ation is carried out. In this study, simulation started with an initialload of 15 kW (0.1 p.u.). An another additional load of 15 kW(0.1 p.u.) is connected at t = 5 s and t = 15 s. Profiles of PMECH andPSG for both the split-shaft and single-shaft MT are presented asFig. 22(a) and (b), respectively. It may be noticed from Fig. 22 thatthe load-following response of both the MTs are similar in natureand, thus, either of these two models may be used for DGs applica-tion. However, power electronic interface devices used in single-shaft model of MT for which many control strategies have beensurfaced in the recent literatures are not associated with split-shaftMT.

5. Conclusion

In this paper, the dynamic model of MTG system with power,speed and voltage controller is simulated for load-following perfor-mance study. The performances of the MTG under different loadingpattern for both isolated as well as grid connected mode are stud-ied and analyzed. Due to slow dynamics of MTG, little delay in in-crease and decrease in mechanical power in following the loaddynamics are observed. Even though, MTG could able to traceout the changing load profile keeping speed of rotor and terminalvoltage per phase at desired level under their respective controlleractions. The results obtained in islanding mode are compared toearlier reported results assuming similar MTG system with sameload patterns. Comparison shows better results due to the presenceof properly tuned power, speed and voltage controllers. Further, acomparison of load-following response between split-shaft andsingle-shaft MT is also carried out. The study of the present workshows that MTG system has a bright future in micro-grid applica-tions for meeting isolated or grid connected consumer loaddemand.

References

[1] Dursun Erkan, Kilic Osman. Comparative evaluation of different powermanagement strategies of a stand-alone PV/Wind/PEMFC hybrid powersystem. Int J Electr Power Energy Syst 2012;34:81–9.

[2] Ou Ting-Chia. A novel unsymmetrical faults analysis for microgrid distributionsystems. Int J Electr Power Energy Syst 2012;43(1):1017–24.

[3] Mohammadi S, Soleymani S, Mozafari B. Scenario-based stochastic operationmanagement of microgrid including wind, photovoltaic, micro-turbine, fuelcell and energy storage devices. Int J Electr Power Energy Syst2014;54(1):523–35.

[4] Geem ZW. Size optimization for a hybrid photovoltaic-wind energy system. IntJ Electr Power Energy Syst 2012;42(1):448–51.

G. Shankar, V. Mukherjee / Electrical Power and Energy Systems 55 (2014) 704–713 713

[5] Kaundinya DP, Balachandran P, Ravindranath NH. Grid-connected versusstand-alone energy systems for decentralized power-a review of literature.Renew Sustain Energy Rev 2009;13(8):2041–50.

[6] Davis MW. Distributed resource electric power systems offer significantadvantages over central station generation and T&D power systems–part I. In:Proc IEEE power engineering society summer meeting, vol. 1. 2002. p. 54–61.

[7] Scott WG. Micro-turbine generators for distribution systems. IEEE Ind ApplMag 1998;4(3):57–62.

[8] Zhu Y, Tomsovic K. Development of models for analyzing the load-followingperformance of microturbines and fuel cells. Electr Power Syst Res2002;62(1):1–11.

[9] Rowen WI. Simplified mathematical representations of heavy-duty gasturbines. Trans ASME J Eng Power 1983;105(4):865–9.

[10] Pai Fu-Sheng. An improved utility interface for microturbine generationsystem with stand-alone operation capabilities. IEEE Trans Ind Electron2006;53(5):1529–37.

[11] Watts JH. Microturbines: a new class of gas turbine engines. Gas Turbine NewsBrief 1999;39(1):5–11. <http://www.asme.org/igti/ggtn/archives.html>.

[12] Cohen H, Rogers GFC, Saravanamuttoo HIH. Gas Turbine Theory. Longman;1987.

[13] Hannett LN, Khan AH. Combustion turbine dynamic model validation fromtests. IEEE Trans Power Syst 1993;8(1):152–8.

[14] Massuco S, Pitto A, Silvestro F. A gas turbine model for studies on distributedgeneration penetration into distribution networks. IEEE Trans Power Syst2011;26(3):992–9.

[15] Yee SK, Milanovic JV, Hughes FM. Overview and comparative analysis of gasturbine models for system stability studies. IEEE Trans Power Syst2008;23(1):108–18.

[16] Shalan HEMA, Hassan MAM, Bahgat ABG. Comparative study on modeling ofgas turbines in combined cycle power plants. In: Proc 14th Internationalmiddle east power system conference; 2010, p. 970–76.

[17] Gaonkar DN, Patel RN, Pillai GN. Dynamic performance of microturbinegeneration system connected to a grid. Electr Power Compon Syst2008;36(10):1031–47.

[18] Guda SR, Wang C, Nehrir MH. Modeling of microturbine power generationsystems. Electr Power Compon Syst 2006;34(9):1027–41.

[19] Hajagos LM, Berube GR. Utility experience with gas turbine testing andmodeling. In: Proc IEEE power engineering society winter meeting, vol. 2.2001. p. 671–7.

[20] Grillo S, Massuco S, Morini A, Pitto A, Silvestro F. Microturbine controlmodeling to investigate the effects of distributed generation in electric energynetworks. IEEE Syst J 2010;4(3):303–12.

[21] Fethi O, Dessaint LA, Al-Haddad K. Modeling and simulation of the electric partof a grid connected microturbine. In: Proc IEEE power engineering societygeneral meeting, vol. 2. 2004. p. 2212–9.

[22] Al-Hinai A, Schoder K, Feliachi A. Control of grid-connected split-shaftmicroturbine distributed generator. In: Proc of 35th Southeasternsymposium on system theory; 2003; p. 84–8.

[23] Saha AK, Chowdhury S, Chowdhury SP, Crossley PA. Modeling andperformance analysis of a microturbine as a distributed energy resource.IEEE Trans Energy Convers 2009;24(2):529–38.

[24] Sisworahardjo NS, El-Sharkh MY, Alam MS. Neural network controller formicroturbine power plants. Electr Power Syst Res 2008;78(8):1378–84.

[25] Kamel RM, Kermanshahi B. Design and implementation of models foranalyzing the dynamic performance of distributed generators in the microgrid part I: micro turbine and solid oxide fuel cell. Trans D: Computer Sci EngElectr Eng 2010;17(1):47–58.

[26] Uzunoglu M, Onar O, El-Sharkh MY, Sisworahardjo NS, Rahman A, Alam MS.Parallel operation characteristics of PEM fuel cell and microturbine powerplants. J Power Sources 2007;168(2):469–76.

[27] Oguz Y, Ustun SV, Yabanova I, Yumurtaci M, Guney I. Adaptive neuro-fuzzyinference system to improve the power quality of a split shaft microturbinepower generation system. J Power Sources 2012;197(1):196–209.

[28] SimPowerSystems User’s Guide, ver. 4.6. The Mathworks Inc.; 2008.[29] Borbely AM, Kreider JF. Distributed generation: the paradigm of the new

millennium. USA: CRC Press; 2001.[30] Cho Changee. New Ideas for soft synchronizer applied to CHP cogeneration.

IEEE Trans Power Del 2011;26(1):11–21.