Modeling, Simulation, and Performance Analysis of Power Management Strategies for an Islanded Microg

International Journal of Energy Science (IJES) Volume 3 Issue 6, December 2013 www.ijesci.org

doi: 10.14355/ijes.2013.0306.02

383

Modeling, Simulation, and Performance

Analysis of Power Management Strategies for

an Islanded Microgrid Faruk A. Bhuiyan*1, Amirnaser Yazdani2, Serguei L. Primak3

*1,3Electrical and Computer Engineering Dept., Western University, London, Ontario, Canada 2Electrical and Computer Engineering Dept., Ryerson University, Toronto, Ontario, Canada

*[email protected]; [email protected]; [email protected]

Abstract

Considering both real and reactive power, this paper

presents the modeling, simulation, and performances

analysis of various power management strategies (PMSs) for

the sizing of an enhanced renewable energy penetrated

islanded microgrid (IMG), which consists of an integrated

photovoltaic‐wind‐diesel‐battery system. The models of the

PMSs are illustrated by flowcharts which can be utilized to

determine optimal sizes of the IMG components. When the

IMG contains both a battery energy storage system (BESS)

and a diesel generator system (DGS), there are several ways

to meet the load demand. The various options of fulfilling

the load demand make the PMSs complex. The complexities

in PMSs have significant impact on the incurred costs

through the fuel usages rate of DGS and the deterioration

level of BESS. The PMSs are presented by considering

unknown component sizes of the IMG. The paper considers

the effect of reactive power on BESS charging. The

effectiveness of the modified PMSs is demonstrated through

simulation studies in the MATLAB/Simulink environment.

Then the performances of the PMSs are compared and the

cost‐varying areas are identified meticulously by utilizing

the simulation results.

Keywords

Battery; Dispatch; Microgrid; Photovoltaic; Power Management;

Wind

Introduction

Electrification of remote communities remains a challenge

in Canada and elsewhere due to the economical and

technical barriers. Most of the Canadian remote

communities have been supplied by diesel generator

systems (DGSs), while the rest have used low to

medium penetration wind/pv diesel systems (Arriaga

et al. 2012; Weis et al. 2008). Thus far, many studies

(Bernal−Agustin et al. 2009) of islanded microgrid

(IMG) have been performed and many systems are

installed with various configuration of components.

Among the installed systems, many projects have

failed due to improper design. The systems design can

be improved if optimization methods are applied

(Zhou et al. 2010). The sizing optimization of an IMG

can be performed by either analytical techniques or

chronological time‐series/probabilistic simulations.

Analytical techniques are difficult to implement due to

the large number of variables, non‐linearity in the

models, and complexity of the configurations. The

timeseries/probabilistic simulation techniques require

well‐defined power management strategies (PMSs)

and are computationally intensive. Power management

in an IMG can be performed by either controlling the

energy resources or/and load management

(Lujano−Rojas et al. 2012); the accommodation of load

management strategies during the design phase may

not be a good idea.

Barley et al. (1996) proposes a number of control

strategies for the operation and simulation of a

wind‐photovoltaic‐diesel‐battery system. Utilizing

load setpoint to start and stop the diesel engine, and

state of energy (SOE) setpoint to charge the battery

energy storage system (BESS), an optimization method

for the control strategies is proposed in Ashari et al.

(1999). Bagen et al. (2005) presents a simulation

technique for the operating strategies of stand‐alone

power systems by giving the priorities to the operation

of nonconventional generating units, conventional

generating units, and BESS in sequence. Dufo−López et

al. (2005) combines two of the operating strategies of

Barley et al. (1996) in a photovoltaic (PV) system with

energy storage and identified a critical load. Katiraei et

al. (2007) proposes an energy‐flow model for an

autonomous wind‐diesel system for the investigation

of the the daily and monthly performances. Various

PMSs for standalone hybrid power system integrated

with hydrogen energy storage are proposed in Ipsakis

www.ijesci.org International Journal of Energy Science (IJES) Volume 3 Issue 6, December 2013

384

et al. (2009). Zhou et al. (2011) proposes the

‘source‐following’ and ‘grid‐following’ dynamic

control approaches in order to enable exchange power

amongst the sources, and to manage energy of the

grid‐connected system. Considering real power only,

Vrettos et al. (2011) proposes three operating modes for

a small isolated high‐penetration renewable energy

system integrated with a BESS. Ohsawa et al. (1993)

has applied artificial neural network to the control

strategies of PV‐diesel power system, although few

researchers (Lin et al. 2011; Yasin et al. 2011) have

proposed intelligence‐based approaches for the

dynamic control of autonomous hybrid power

systems.

Various software tools (HOMER ; HYBRID2; iHOGA;

INSEL ; RETScreen ; TRANSYS ) are used for

simulation, optimization, and performance evaluation

of IMGs. Amongst them, HOMER (Hybrid

Optimization Model for Electric Renewable) has been

widely used for optimization studies, HYBRID2 is

popular for simulations, and HOGA (Hybrid

Optimization by Genetic Algorithms) is primarily used

for multi‐objective optimization.

However, the aforementioned software tools and

papers have not modeled PMSs for the objective of

determining the component sizes considering both real

and reactive powers of the load. This paper models the

PMSs in detail by utilizing flowcharts, where both real

and reactive powers of the load are considered.

Study System

Fig. 1 shows a schematic diagram of an IMG whose

main components are DGS, a wind power system

(WPS), a photovoltaic power system (PVS), an energy

storage system, a dump load, a primary load, and

power‐electronic converters. In this configuration, the

energy storage system is assumed to be a battery

energy storage system (BESS) connected to the power

system through a bi‐directional power‐electronic

converter. In turn, the PV power system comprises

multiple PV modules interface with the point of

common coupling (PCC) through a DC/AC converter.

Also, the BESS is composed of a bank of series‐/

parallelconnected batteries and a power‐electronic

converter. The WPS delivers power at unity power

factor to the PCC, represented as Pw. The delivered real

and reactive powers of the PVS are denoted

respectively by Ppv and Qpv. The real and reactive

powers delivered by the DGS are denoted by Pdi and

Qdi, respectively. The power delivered by the batteries

is referred to as the discharged power and denoted by

Pb. The real and reactive powers of the

power‐electronic converter of the BESS are denoted as

Pcon and Qcon, respectively; the typical high efficiency of

the power‐electronic converter implies that Pb and Pcon

are almost equal and are therefore used

interchangeably in this paper. The real and reactive

power components of the primary load are

represented by Pl and Ql, respectively. The dump load

is assumed to be a resistive load and represented as Pdl.

At any instant, stable operation of the IMG

corresponds to following power‐balance equations (1):

Pw(t) + Ppv(t) + Pdi(t) + Pcon(t) − Pl(t) − Pdl(t) = 0

Qdi(t) + Qpv(t) + Qcon(t) − Ql(t) = 0

(1)

FIG. 1 SCHEMATIC DIAGRAM OF AN ISLANDED MICROGRID

Reactive Power and Charging/Discharging of BESS

Figs. 2 (a) and (b) show that the primary load is

decoupled into real power, Pl, and reactive power, Ql.

Depending on the control strategy, the load reactive

power, Ql, can be supplied by the converters of the

PVS, BESS or DGS. To deliver the rated power of the

PVS into the PCC, the PVS converter rating, pvconS ,

must be the same as the rated power of the PVS, pvratP .

The power delivered by the PVS usually falls below pv

ratP in a year. Therefore, the PVS converter can be

utilized to supply reactive power of the load. If the

instantaneous real power of the PVS is Ppv(t), then the

PVS can deliver a reactive power of

2 2( ) ( )pv pv pvconQ S P where 2 2( ) ( )pv pv

conS P

(2)


385

(a)

(b)

FIG. 2 REAL AND REACTIVE POWER PHASORS AT (a) EXCESS RENEWABLE

GENERATION, AND (b) SHORTAGE OF RENEWABLE GENERATION

Figs. 2(a) and (b) indicate that a part of the load

reactive power is supplied from PVS, as Qpv, while the

remaining load reactive power can be supplied from

the BESS converter and/or DGS. Fig. 2 further

illustrates that the real power supplying capacity of the

BESS converter decreases when the load reactive

power needs to be compensated. Thus, the BESS

converter real power, conavaP , is formulated as,

22( ) (max 0,( )pvcon con lava rat

lrem

P S Q Q

Q

(3)

where conratS is the rating of the BESS converter, and

lremQ is the remaining load reactive power demand; conratS needs to be higher than l

remQ when DGS does not

run. However, if DGS runs and delivers real power,

the load reactive power can be shared by the

converters and DGS,

(max 0,( ) )

(max 0,( ) )

conpvcon lrat

con dirat rat

lrem

dipvdi lrat

con dirat rat

lrem

SQ Q Q

S SQ

SQ Q Q

S SQ

(4)

where Qcon and Qdi are the reactive‐power output of the

BESS converter and DGS, respectively. In (4) diratS is the

rated power of DGS, and, lremQ should not be higher

than the aggregated value of conratS and di

ratS .

Consequently, the effective real power of the BESS

converter ( )conavaP and the DGS max( )diP decrease as

2 2

2 2max

( ) ( )

( ) ( )

con con conava rat

di con dirat

P S Q

P S Q

(5)

The rate of charge of BESS, ( )bavaP , depends on state of

energy (SOE), Eb(t), and the rated charging power,

maxbP , is

maxmax

( )min ,max(0,

Δ

b bb b

avaE E t

P Pt

(6)

If bavaP in (6) is larger than con

avaP , then the charging of

the BESS needs to be lowered to conavaP ; otherwise, the

BESS can be charged by bavaP . Fig. 2 (a) shows surplus

power, Psur, at higher renewable power generation. The

actual charging of the BESS, equation (7), is the lowest

of the three values bavaP , con

avaP and Psur.

min( , , )b con bsur ava avaP P P P (7)

If the surplus power Psur is higher than the absolute

value of Pb, then the remaining power must be

dumped. Fig. 2 (b) illustrates the real and reactive

powers of DGS and BESS converter. When DGS and

BESS converter run, the load reactive power is shared

by DGS and BESS as described by (4). During the BESS

discharge, the available discharging power from BESS

can be formulated as

minmax

( )min ,max(0, )

Δ

b bb b

avaE t E

P Pt

(8)

Since the BESS discharges through its converter, the

maximum discharge power of the BESS is determined

by the real power that flows through the converter. If

the net load, Pde f, needs to be delivered by the BESS,

then the BESS discharge power is

min( , , )b con bava ava defP P P P (9)

Depending on the PMS, Pde f may need to be supplied

by both the BESS and the DGS. Otherwise, power

shortage will occur. In Fig. 2 (b), the negative Psur

means positive Pde f .

The SOE in the BESS can be expressed as

( Δ ) ( )(1 ) ( )Δb b b bE t t E t P t t (10)

where is the self discharge coefficient, b is the

efficiency during charging and discharging, and Δt is


386

the time step.

Power Management Strategies and Their Flowcharts

Two distinct types of control in an IMG are (i) dynamic

control, which deals with the frequency and

magnitude of the voltage, and (ii) PMS, which controls

the energy resources of an IMG. This paper deals with

four different PMSs, which are described in the

following subsections.

Power Management Strategy‐A (PMS‐A)

Fig. 3 illustrates the algorithm of the PMS‐A, utilizing

the renewable resources first for the real load power

demand. The load reactive‐power demand is

compensated from converters (PVS and BESS) and

DGS. The real power limit of the BESS converter ( conavaP )

and that of the DGS ( maxdiP ) are then calculated

(explained earlier). If the renewable resources cannot

meet the net real load i.e., Pde f > 0, then the BESS is

employed to compensate the demand (at ( 2b

defP P ));

otherwise, the DGS shall start. The “power shortage”

in Fig. 3 indicates that the resource sizes do not meet

the load demand. The BESS enters into charging mode

if the aggregated delivered power from the renewable

resources exceeds the primary load demand (Psur > 0).

The actual charging power (Pb) depends on the

capacity and power rating of the batteries, and the real

power handling capability of the BESS converter. If the

surplus power cannot be absorbed by the BESS (i.e.

1b

surP P ), the rest is burnt in the dump load. The

PMS‐A does not allow DGS to charge the BESS. The

exception is when the DGS operates at a low load with

a non‐zero minimum setpoint ( mindiP ) setting. When the

net load demand falls below the minimum setpoint of

the DGS, the DGS then runs at its minimum operating

point. The extra generation from the DGS is then used

to charge the BESS. Otherwise it can be dumped. The

PMS‐A is a modified form of load following strategy.

FIG. 3 FLOWCHART OF PMS‐A


387

Power Management Strategy‐B (PMS‐B)

The PMS‐B allows the DGS for charging the BESS.

Fig.4 illustrates a flowchart for the PMS‐B. First the

aggregate renewable power is compared with load real

power. The real powers from the converter of BESS

and the DGS are then calculated upon supplying load

reactive power (described earlier). If the surplus

generation, i.e. Psur > 0 is higher than Pb, then the BESS

enter the charging mode at its maximum rate and the

rest is burnt to the dump load; otherwise at Psur < Pb,

the DGS runs for a while to charge the BESS at a

maximum rate. At min( )b bE t E of the BESS and/or at

‘on’ status (k = 1) of DGS, if aggregate surplus

generation from the renewable resources is lower than

Pb, then the DGS starts/continues operation for

charging the BESS. The DGS is commended to stop

when SOE in the BESS reaches at bsocE . If Pde f > 0, and

( )b bsocE t E in BESS, and the DGS is off (k = 0), the net

real load (Pde f) is supplied by the BESS; if fails, the DGS

starts to compensate the remaining net load. When

DGS is already in ‘on’ status (k = 1) and SOE of the

BESS stays above minimum ( min( )b bE t E ) and below

bsocE , the DGS continues its operation for charging

BESS at its maximum rate. The energy at BESS may fall

at/below minimum level ( min( )b bE t E ) with DGS ‘off’

status. In such a situation, the DGS goes into operation

mode for charging BESS. As soon as the SOE of the

BESS reaches at bsocE , the DGS stops for charging the

BESS. The bsocE is considered a lavel which is lower than

maxbE . The Fig. 4 represents all possible scenarios under

unknown sizes. The “power shortage” of Fig. 4 is true

if the reserve and renewable resources cannot meet the

primary load demand. The PMS‐B is a modified form

of SOC setpoint dispatch strategy.

FIG. 4 FLOWCHART OF PMS‐B AND PMS‐C


388

FIG. 5 FLOWCHART OF PMS‐D

Power Management Strategy‐C (PMS‐C)

When the DGS is utilized for charging the BESS at

maxbE instead of b

socE , then the PMS‐B becomes PMS‐C. It

is a modified form of the cycle charge strategy. The

strategy may need the DGS to continue at its operation

mode for a longer period than that of PMS‐B.

According to PMS‐B and PMS‐C, the DGS runs either

at full power or at a rate not exceeding the aggregate of

BESS charging power and net load demand. The

flowchart for the PMS‐C is given in Fig. 4 where both

real and reactive power balances are depicted. The

PMS‐C also indicates that the dumping of power,

generated from the DGS, is required. Fig. 4 illustrates

that “power shortage” may occur under various

conditions, especially with improper sizes of

components in IMG. Fig. 4 illustrates the DGS ‘on’/‘off’

status by a flag variable ‘k’.

Power Management Strategy‐D (PMS‐D)

Utilizing the per unit energy cost curve of DGS (Barley

et al. 1996), the calculation steps of critical load Pd and

cycle charge load Pc are shown in the end of this

subsection. The Pd is the intersecting point of direct

diesel cost and BESS deterioration cost. It is more

economical to use the batteries when net load stays

below Pd; otherwise the operation of DGS is more

economical. Fig. 5 illustrates the PMS‐D which utilizes

the critical load, Pd, upon fulfilling load reactive power

demand. The PMS‐D is the modified form of the frugal

discharge strategy. Fig. 5 illustrates that when the net

load Pdef is above Pd, the DGS provides power to the

load at its fullest capacity; otherwise, the BESS delivers

the remaining load power. Alternatively, if the net

load, Pdef, falls below Pd, the BESS discharges power at

its extent. Power failure may occurs if both the DGS

and the BESS cannot meet the net load demand, which

may occur due to inappropriate component sizes.

Subsection‐1 shows the procedure for deriving Pd and

Pc.

1) Determination of Critical and Cycle Charge Load: The


389

per unit costs of energy for the DGS includes capital,

operation, maintenance, and replenishment. Capital

costs depend on the rating of the DGS. The running

costs depend on operation and maintenance costs. The

operation cost is mainly the fuel costs, which including

fuel, transportation, and inventory holding cost. The

fuel consumption of the DGS is not linear. The fuel

consumption can be assumed as a quadratic function

with a cost at no‐load operation. Thus, the fuel

consumption (L/h) of DGS for Pdi, can be expressed as

(Zhu 2009).

21 2 3( )di di di di

c ratF g P g P g P (11)

Where 1g , 2g , and 3g are fuel consumption

coefficients in L/kW2h, L/kWh, and L/kWh, respectively.

Here diratS is the same as di

ratP taken at unity power

factor. The fuel consumption cost in $/kWh for DGS can

thus be expressed as

1 2 3( )di

di di ratf fdi

PC g P g g C

P (12)

where Cf is in $/L and includes fuel, transportation and

inventory holding cost.

Maintenance cost of the DGS varies at the level of

produced power and number of start‐stop. The per

kWh hourly maintenance cost, dimhC , is given as

didi mmh di

rat

CC

P (13)

Where dimC is the maintenance cost for di

ratP capacity of

DGS.

The hourly running cost of the DGS for per kWh diesel

generated energy is the sum of the fuel consumption

cost (operation cost) and the hourly maintenance cost

expressed as

1 2 3( )di di diom f mh

diratdi

PC g P g g C C

P (14)

where diomC is the hourly operation and

maintenance/running cost. The capital cost of the BESS

depends on size. The BESS wear cost is treated as the

cost of delivered energy from BESS and the

maintenance cost. The BESS wear cost is expressed as

bb bcw mhb

eqc rat

CC C

DOD E (15)

where bwC , b

cC , DODeqc, bratE , and b

mhC are wear cost,

capital cost, equivalent depth of discharge, kWh rating

of BESS, and hourly maintenance cost of BESS

respectively. Number of cycle to failure (Ncf) for the

BESS can be expressed as (Drouilhet et al. 1997).

1

0

( 1)u DODce

cf uN e

NDOD

(16)

where Nce, u1, and u0 are parameters for the equation.

The DOD is the depth of discharge for the BESS. Based

on battery manufacturer life cycle data, (16) can be best

fitted for a specific BESS. As an example,

manufacturer’s life cycle data for NiCd cell is best

fitted, which provides the parameter values u0=1.67,

u1=‐0.52, and Nce = 2055. Once the best fitted values are

fixed, then the equivalent depth of discharge can be

calculated as

1

1( ) ( )

M

eqc n ncfn

DOD N DODM

(17)

where M is the total number of observation point to

figure out the values of Ncf for different DOD utilizing

(16). This DODeqc is required for (15). For charging the

BESS, DGS can be used to run in excess to the net‐load

and thus requires extra generating costs. If the round

trip efficiency of the BESS and the charger is R , then

the cost of cycle charge can be written as

dii fdi

cR

F cC

(18)

The sum of battery wear costs and cycle charge costs

can be written as

dii fe di b b

c c w wR

F cC C C C

(19)

where diiF is the extra fuel required to charge the BESS.

The DGS optimum stopping set‐point occurs when the

BESS wear costs and DGS running costs becomes equal.

Thus, the equation (14) and (15) can be equalized to

calculate the roots of Pdi, which is expressed as

2 22 1 32

1 1

( ) 4

2 2

di b dib diw ratf mh fw f mh

df f

g c C C g g c PC g c CP

g c g c

(20)

From (14) and (19) the cycle charge load is expressed as Pc

and can be calculated by (21)

2 22 1 32

1 1

( ) 4

2 2

di e die dic ratf mh fc f mh

cf f

g c C C g g c PC g c CP

g c g c

(21)

Simulation Results

To demonstrate the performances of the PMSs, several

case studies are conducted in the MATLAB/Simulink

environment by utilizing the values of Table I. The


390

values of Table II are used on (20) to determine the

critical load, Pd, and thus to simulate PMS‐D. The bwC of

(20) is determined using (15) and (17). The hourly

average time series wind speed data, for a Canadian

site, is obtained from Weather Canada; and hourly

average time series solar irradiation data, based on

latitude and longitude, is produced by using HOMER

package. The wind power and PV power are produced

by employing (Bhuiyan et al. 2010) and (Lee et al. 2008),

respectively. To generate Fig. 6, probabilistic treatment

is included on the time series data at 15 minutes

interval. For other figures, year around hourly average

time‐series data is used. In Fig. 1, the Pb is positive for

delivered power of the BESS; however, in all figures

except Fig. 6, the BESS power is represented by −Pb.

The real and reactive load power components are

calculated utilizing IEEE RTS load model (IEEE RTS

1979) and a time series of load power factor (Saha et al.

2010).

TABLE I SIMULATION PARAMETERS

Simulation Parameters Values

WPS rating 600 kW

PV power rating 150 kW

Diesel generator rating 320 kVA

Inverter rating 640 kVA

BESS capacity rating 7.2 MWh

BESS power rating 450 kW

Base power 300 kW

Base power factor 0.9

Base BESS discharge time 1 hr

Base BESS capacity 300 kWh

Efficiency of PV system ( pv ) 15%

Cut‐in wind speed 3.5 m/s

Rated wind speed 12 m/s

Cut‐out wind speed 23 m/s

Period under observation 8760 hrs

BESS minimum level Emin 40%

BESS SOC level Esoc 60%

Initial BESS SOC 80% of rated

Self discharge of battery 0.2% per hour

Efficiency of inverter (discharging) 95%

Efficiency of rectifier (charging) 95%

TABLE II CRITICAL LOAD, Pd, DETERMINATION PARAMETERS

FOR PMS‐D

Parameter Value Comments

1g 0.00012 L/kW2h equation (11)

2g ‐0.011 L/kWh equation (11)

3g 0.16 L/kWh equation (11)

diratP 300 kW equation (11)

Fuel price, fc 1 $/L equation (12)

dimhC 0.11 $/h for 1 kWh equation (13)

/b bc ratc E 400 $/kWh equation (15)

bmhC 0.05 $/h for 1 kWh equation (15)

eqcDOD 1100 cycles equation (17)

Impact of Converter Reactive Power on BESS

Charging

This case demonstrates the impact of converter

reactive power on the BESS charging for the PMS‐A of

the IMG which is considered composed of without

PVS. The rating of the BESS converter is taken 0.5 p.u..

As Fig.6 (a) shows, Pw remains above 1 p.u., which is

more than the primary load demand. Therefore, DGS

does not require to run, and thus Pdi and Qdi in Fig. 6 (a)

are zero from 1520th hour to 1525th hour. Fig. 6 (b)

indicates that the SOE of BESS has not reached to the

maximum level until 1525th hour. Thus, the excess

wind power is stored in the BESS. At 1520.25th hour,

the reactive power demand is 0.5 p.u., which is

supplied by the BESS converter, as shown in Fig. 6 (a).

Although there is excess wind power, and BESS has

the capacity remaining for charging, the actual BESS

charging power (Pb) is zero due to the lower size

converter.

FIG. 6 IMPACT OF CONVERTER REACTIVE POWER ON BESS CHARGING

Simulation Results for PMS‐A

Fig. 7 illustrates the effectiveness of the PMS‐A by the

real power, reactive power and SOE of the BESS. From

2171th hour to 2192th hour, the aggregate renewable


391

power is sufficient for primary load as shown in Fig.

7(a); thus, the BESS is in charging mode. As Fig. 7(a)

shows, the aggregate renewable power is low for

primary load from 2238th hour to 2285th hour. Thus,

the shortage is compensated by the BESS; if fails, the

DGS runs e.g. 2286th hour to 2378th hour. The

operation from 2114th hour to 2170th hour in Fig. 7(b)

indicates that the SOE of BESS stays at minimum. Fig.

7(a) further demonstrates that short‐term start/stop of

DGS (e.g. 2325th hour) may occur depending on wind

and PV resource output. The Pdl of Fig. 7(a) indicates

that the excess renewable power is dumped when

BESS and converter cannot accommodate. As Fig. 7(c)

shows, the load reactive power is mostly delivered by

the PVS converter and then the rest is supplied by the

BESS converter and/or DGS.

FIG. 7 POWER AND ENERGY OF IMG COMPONENTS FOR PMS‐A

Simulation Results for PMS‐B

Fig. 8 demonstrates the capability of PMS‐B by the SOE

of the BESS, the real and reactive powers of the IMG

components, presented from 3300th hour to 3475th

hour. Fig. 8 (a) shows that the delivered wind power is

low for many hours, between 3300th hour and 3475th

hour. The BESS does not have enough SOE for

discharging as shown in Fig. 8 (b). Therefore, the DGS

operates for load real power and for charging the BESS

(e.g. 3305th hour to 3316th hour). Once the SOE of

BESS reaches at bsocE , then the DGS stops. The BESS can

deliver the part of net load until the SOE of BESS stays

above minbE , and thus Fig. 8(b) illustrates that the DGS

starts again when the SOE of BESS touches at minbE .

Thus, the DGS mostly runs with heavy load,

subsequently its short‐period start/stop decreases. Fig.

8(c) demonstrates further that the reactive power is

supplied mostly by the PVS converter; after that the

BESS delivers the remaining reactive power and at the

end the DGS helps.

FIG. 8 POWER AND ENERGY OF IMG COMPONENTS FOR PMS‐B

Simulation Results for PMS‐C

Fig. 9 illustrates the productiveness of the PMS‐C by

the real power, reactive power and SOE of the BESS.

From 3395th hour to 3410th hour in Fig. 9 (b) indicates

that the DGS operates for charging the BESS up to the

maximum level. As Fig. 9(a) shows, the DGS operates

almost at full load for charging the BESS and for

fulfilling the net load demand. By observing Figs. 8

and 9, it can be stated that the operating periods of

DGS in this strategy are longer than those of PMS‐B.

As the SOE of BESS goes at maximum level by the

DGS operation, Fig. 9(a) illustrates that the dumping

power has increased (e.g. from 3350 th hour to 3367th

hour). Thus, the renewable energy penetration for the

primary load is expected to decrease.

FIG. 9 POWER AND ENERGY OF IMG COMPONENTS FOR PMS‐C

Simulation Results for PMS‐D

Fig. 10 demonstrates the performance of PMS‐D by the

simulation results, presented from 2105th hour to

2110th hour. As Fig. 10(a) shows, the DGS operates


392

when net load exceeds Pd; otherwise the BESS delivers

the net load, e.g. 2111th hour to 2117th hour. Thus, the

operation of the DGS and the BESS is complementary

for the net load demand unless any constraints work.

Therefore, the DGS operation contains frequent

start/stop, and it can be inferred that this causes higher

maintenance costs for DGS. Fig. 10(b) demonstrates

that the SOE decrease rate in BESS is lower compared

to that in other strategies. The slow decreasing rate of

SOE in the BESS may enhance the life cycles of the

batteries. Fig. 10 indicates that the charging rate of the

BESS is much higher than that of the discharging rate.

FIG. 10 POWER AND ENERGY OF IMG COMPONENTS FOR PMS‐D

SOEs of BESS for the PMSs

Figs. 11 (a), (b) and (c) respectively illustrate the SOEs

of the BESS for the PMS‐A, PMS‐C, and PMS‐D. As Fig.

11(a) shows, the SOE of BESS for PMS‐A remains low

from 2900th hour to 5500th hour due to seasonal

variations of wind speed. The magnified plot in Fig.

11(a) demonstrates that the SOE of BESS has decreased

significantly below minimum level due to

self‐discharge of the BESS. The situation becomes

worse when the BESS reaches at minimum level along

with low renewable power generation (e.g. 1376th hour

to 1475th hour). Thus, the PMS‐B and PMS‐C utilize

the DGS for charging BESS; Fig. 11(b) shows that the

SOE of BESS does not reach below minimum level. Fig.

11(b) further shows the SOE of BESS reaches

frequently at maximum level. The SOE of the BESS for

PMS‐B is excluded as it is very similar to PMS‐C. It can

be inferred that the SOE of BESS in PMS‐B would not

frequently reach the maximum level like PMS‐C. The

SOE of BESS for PMS‐D, (shown in Fig. 11(c)),

illustrates that the rate of discharge from BESS is lower,

as the BESS only discharges for low net load. Figs. 11

(a) and (c) show that the SOEs and self discharges of

BESS are similar both in PMS‐A and in PMS‐D.

FIG. 11 SOE OF BESS FOR (a) PMS-A, (b) PMS-C, AND (c) PMS-D

Reactive Power Management in IMG without PVS

The reactive powers of the IMG, which is constituted

without PVS for the PMS‐A, PMS‐B, PMS‐C and

PMS‐D are shown in Figs. 12 (a), (b), (c) and (d),

respectively. All of the sub‐figures of Fig. 12 illustrate

that the Qpv is zero. Thus, the BESS converter mostly

delivers the reactive power (Qcon), when the DGS does

not require to share. Fig. 12 (a) demonstrates that the

reactive power demand at around 2121th hour is 0.45

p.u., which is shared by both the BESS converter and

the DGS; where the Fig. 12 (b) (c) and (d) indicate that

the reactive powers are delivered only from the BESSs.

FIG. 12 REACTIVE POWER MANAGEMENT IN IMG OF NO PVS

Comparison of PMSs

All of the aforementioned studies are represented for

few hours of total simulation which is for the clarity of

the figures. Table III compares the performances of the

strategies based on a year around (8760 hrs) simulation.

As Table III shows, the SOE of BESS stays at minimum

level for 3898 hours, 156 hours, 72 hours, and 1143

hours, respectively, for PMS‐A, PMS‐B, PMS‐C, and


393

PMS‐D. The number of hours, the SOE of the BESS at

minimum level, is significantly lower for PMS‐B and

PMS‐C compare to others. The number of hours, the

SOE stays above the 97% of its maximum level, is the

highest for the PMS‐C compare to the others’.

Consequently, the renewable energy penetration is the

lowest and thus the dumping power is the highest for

the PMS‐C. Although the DGS is used for charging the

BESS in PMS‐B and PMS‐C, the number of DGS

operating hours are lower in the strategies due to near

full load operation. Alternately, the DGS operates

longer hours with variable load in PMS‐A and PMS‐D.

Moreover, the self discharges in PMS‐A and PMS‐D

are higher (presented earlier). The low operating hours

of the DGS indicate low maintenance and operating

costs needed for the DGS.

TABLE III PERFORMANCES COMPARISON OF PMSs

Items PMS‐A PMS‐B PMS‐C PMS‐D

SOE at min. 3898 hrs 156 hrs 72 hrs 1143 hrs

SOE above 97%

of max. 1559 hrs 1824 hrs 2506 hrs 1864 hrs

DGS operation 3671 hrs 1763 hrs 1678 hrs 3614 hrs

Renew. Energy

penetration 69.15% 66.83% 62.62% 66.55%

Dumping

energy (kWh) 1183 1375.2 1620.3 1414.5

BESS self disch. high low low high

DGS operation variable rated rated variable

BESS expected

life‐cycle low moderate moderate low

Expected

maint. cost high low low high

Conclusions

The paper has modeled few PMSs and it has compared

the performances of them. The paper has also

presented the insight complexities in the model of the

PMSs. Moreover, the impact of converter reactive

power on BESS charging is incorporated in the study.

The performance analysis, based on case studies and

long term simulation, indicates that the PMS‐B is

comparatively better PMS than that of the others. The

study further illustrates the situation and impact of

BESS self‐discharge, SOE condition of BESS at

maximum and minimum level, hours of DGS

operation and frequent start/stop of DGS. Thus, the

analysis has figured out few sensitive parameters,

which need to be taken into account during system

design and feasibility study. In this study the necessity

of sizing optimization for the IMG is focused, so that,

the optimal sizes and the optimal power management

strategy can be determined effectively by utilizing the

flowcharts of the PMSs.

ACKNOWLEDGEMENT

The first author would like to thank Ministry of

Training, Colleges, and Universities, Ontario, and

Natural Sciences and Engineering Research Council

(NSERC), Canada, for awarding Ontario Graduate

Scholarship (OGS) and Canada Graduate Scholarship

(Doctoral), respectively.

REFERENCES

Arriaga M., Canizares C. A., and Kazerani M. “Re‐newable

energy alternatives for remote communities in northern

ontario, Canada.”IEEE Trans. on Sust. Energy 4(3); (2013):

661−70.

Ashari M., Nayar C. V. “An optimum dispatch strategy

using set points for a photovoltaic (PV)‐diesel‐battey

hybrid power system.” Solar Energy 66(1); (1999): 1−9.

Bagen, Billinton R. “Evaluation of different operating

strategies in small stand‐alone power systems.” IEEE

Trans. Energy Convers 20(3); (2005): 654−60.

Barley C. D., Winn C. B., “Optimal dispatch strategy in

remote hybrid power systems.” Solar Energy 58(4‐6);

(1996): 165−79.

Bernal‐Agustin J. L, Dufo‐Lot’pez R. “Simulation and

optimization of stand‐alone hybrid renewable energy

systems.” Renew and Sust Energy Rev. 13(8); (2009) :

2111−18.

Bhuiyan F. A., Yazdani A. “Reliability assessment of a

wind‐power system with integrated energy storage.” IET

Renew Power Generation 4(3); (2010); 211−20.

Drouilhet S., Johnson B. L., Drouilhet S. and Johnson L.: “A

battery life prediction method for hybrid power

applications.” National Renewable Energy Labora‐tory,

US Dept. Energy, Chicago, IL 1997; Contract

DE‐AC36‐83CH10093.

Dufo−López R., Bernal‐Agustín J. L. “Design and control

strategies of PV‐Diesel systems using genetic

algorithms.” Solar Energy 79(1); (2005): 33−46.

Ipsakis D., Voutetakis S., Seferlis P., Stergiopoulos F., and

Elmasides C. “Power management strategies for a

stand‐alone power system using renewable energy

sources and hydrogen storage.” International Journal of

Hydrogen Energy 34(16); (2009): 7081−95.


394

Katiraei F., Abbey C. “Diesel plant sizing and perfor‐mance

analysis of a remote wind‐diesel microgrid.” IEEE PES

Gener Meeting 2007; p. 1−8.

Lee D. J., Wang L. “Small−signal stability analysis of an

autonomous hybrid renewable energy power

generation/energy storage system part I: time‐domain

simulations.” IEEE Trans. Energy Convers 23(1); (2008):

311−20.

Lin W. M., Hong C. M., and Chen C. H. “Neural‐network

‐based MPPT control of a stand‐alone hybrid power

generation system.” IEEE Trans. Power Elec‐tronics

26(12); (2011): 3571−81.

Lujano‐Rojas J. M., Monteiro C., Dufo‐López R., and

Bernal‐Agustin J. L. “Optimum load management

strategy for wind/diesel/battery hybrid power systems.”

Renewable Energy44 (2012): 288−95.

Ohsawa Y., Emura S., and Arai K. “Optimal operation of

photovoltaic/diesel power generation system by neural

network.” Proc. of the second international forum on

application of neural networks to power system (1993); p.

99−103.

Saha T. K., Kastha D. “Design optimization and dynamic

performance analysis of a stand‐alone hybrid wind‐diesel

electrical power generation system.” IEEE Trans. Energy

Convers 25(4); (2010): 1209−17.

Subcommittee P. M. “IEEE reliability test system.” IEEE

Trans. on Power Appar. and Syst. PAS‐98(6); (1979):

2047−54.

Vrettos E. I., Papathanassiou S. A. “Operating policy and

optimal sizing of a high penetration RES‐BESS system for

small isolated grids.” IEEE Trans. Energy Convers 26(3);

(2011): 744−56.

Weis T. M., Ilinca A. “The utility of energy storage to

improve the economics of wind−diesel power plants in

Canada.” Renewable Energy 33(7); (2008): 1544−57.

Yasin A., Napoli G., Ferraro M, Testa A., and Antonucci V.

“Fuzzy logic based management of a stand‐alone hybrid

generator.” Clean Electric Power Confer, Italy (2011); p.

690−6.

Zhou W., Lou C., Lu L., and Yang H. “Current status of

research on optimum sizing of stand‐alone hybrid˝

solarUwind power generation systems.” Applied Energy

87(2); (2010): 380−9.

Zhou T., François B. “Energy management and power

control of a hybrid active wind generator for distributed

power generation and grid integration.” IEEE Trans.

Indus. Electronics 58(1); (2011): 95−104.

Zhu, J.: Optimization of power system operation, Wiley‐IEEE

Press, 2009.

HOMER:ʺhttp://www.nrel/gov/HOMER.“

HYBRID2:ʺhttp://www.ceere.org/rerl/projects/software/hybri

d2/.“

iHOGA:ʺhttp://www.unizar.es/rdufo/hoga‐eng. htm.“

INSEL:ʺhttp://www.insel.eu.“

RETScreen:ʺhttp://www.retscreen.net/.“

TRANSYS:ʺhttp://www.trnsys.com/.“

Available at:“http://climate.weatheroffice.gc.ca/ʺ. ac‐cessed

Nov. 2012.

Modeling, Simulation, and Performance Analysis of Power Management Strategies for an Islanded Microg

Documents

Transcript of Modeling, Simulation, and Performance Analysis of Power Management Strategies for an Islanded Microg