impact of power demand and battery capacity in microgrid system is measured. Keyword: microgrid, renewable energy, emission reduction, optimization.
Optimal Energy Management of Microgrid Systems in Taiwan Yen-Haw Chen a, Yen-Hong Chen a, Ming-Che Hu b,* Abstract-
This
research
examines
operation of microgrid systems. disaggregated
from
main
optimal
Microgrids are
transmission
grid.
CPBM generation capacity of biomass power [kW] CPGT generation capacity of gas turbine power
Microgrids are able to integrate distributed
[kW]
renewable energy, take advantage of waste heat,
CPFC generation capacity of fuel cell power
provide higher power reliability, reduce electricity
[kW]
transmission loss, and decrease greenhouse gas
DMtELE
emissions. The study considers solar power, wind
[kW]
power, biomass power, gas turbine, fuel cell
DMtHEAT
generators, power storage, heat storage device,
ERBM CO2 emission rate from biomass electricity
electricity demand, and heat demand in microgrid
generation [kg/kWh]
system.
ERGT CO2 emission rate from gas turbine
In the paper, an optimization model of
electricity demand per hour at time t heat demand per hour at time t [J]
the microgrid system is proposed and the optimal
electricity generation [kg/kWh]
operating strategy of the system is presented.
ERFC CO2 emission rate from fuel cell electricity
A
case study of optimization model is analyzed in a
generation [kg/kWh]
microgrid system by Institute of Nuclear Energy
FRBM biomass needed for biomass electricity
Research (INER) in Taiwan.
[kg/kWh]
Additionally, the
impact of power demand and battery capacity in
FRGT gas needed for gas turbine electricity
microgrid system is measured.
[m3/kWh] FRFC hydrogen needed for fuel cell electricity
Keyword: microgrid, renewable energy, emission
[m3/kWh]
reduction, optimization
VCPV variable cost of solar power generation [$/kWh]
I.
Nomenclature
Indices t
variable
cost
of
wind
power
generation [$/kWh] time period, t =1, 2,…,T
Parameter BKt
VCWD VCBM
variable cost of biomass power
generation [$/kWh] hours per block [hour]
VCGT variable cost of gas turbine power
CPPV generation capacity of photo voltage
generation [$/kWh]
power [kW]
VCFC variable cost of fuel cell power generation
CPWD generation capacity of wind power [kW]
[$/kWh]
978-1-61284-220-2/11/$26.00 ©2011 IEEE
HRBM
combine heat production of biomass
t [kg]
power generator [J/kW]
etFC
HRGT combine heat production of gas turbine
[kg]
power generator [J/kW]
z
CO2 generated by fuel cell at time t total cost [$]
Variable nGT
number of gas turbines in microgrids
[dimensionless] n
FC
subsystems
are
stand-alone
separated
from
power
the
main
transmission grid. The power subsystems are
[dimensionless] electricity
Introduction Microgrids
number of fuel cells in microgrids
xtPV
II.
generated
by
photo
designed for small communities and they
voltage per hour at time t [kW]
produce electric power themselves to meet the
xtWD
demand.
electricity generated by wind power
Microgrids are proposed to provide
per hour at time t [kW]
energy locally and integrate distributed power
xtBM
generators and local renewable energy including
electricity generated by biomass per
hour at time t [kW]
wind, hydro, solar power, fuel cell and
xtGT
bioenergy.
electricity generated by gas turbine
per hour at time t [kW] xtFC
electricity generated by fuel cell per
Three microgrid systems are built and tested in Japan by the New Energy Industrial
hour at time t [kW]
Technology Development Organization (NEDO)
xtBT
[1]-[3].
electricity stored in battery at time t
microgrid
systems
contain
combined heat and power, gas engines, batteries,
[kW] xtBTI
electricity input of battery at time t
thermal storage tanks.
UK examines a
microgrid model including wing power, solar
[kW] xtBTO
The
electricity output of battery at time t [kW]
xtHS
heat stored in heat storage at time t
generating units, electricity storage, and thermal storage facilities to meet hospitals, hotels, and
[J] xtHSI
power, combined heat and power, boiler
heat input of heat storage at time t
leisure centers’ power demand [4]. Without
[J]
long
distance
transmission,
xtHSO heat output per hour from heat storage at
microgrids of community level provide more
time t [J]
stable power supply and higher reliability than
ytBM
biomass used by biomass at time t
the national power grid.
In main electricity
grid, high voltage power transmission is used to
[kg] ytGT
gas used by gas turbine at time t [m3]
reduce power transmission loss.
ytFC
hydrogen used by fuel cell at time t
electricity
[m3] etBM
loss
still
incurs
approximately 4% of total production cost [5]. CO2 generated by biomass at time t
Microgrid system generates power locally to meet regional demand which is able to reduce
[kg] etGT
transmission
However,
CO2 generated by gas turbine at time
transmission loss and provide high transmission
efficiency.
Microgrids are able to improve
capacity is renewable energy.
The buyback
energy utilization efficiency by taking advantage
price for renewable electricity is around 10 NTD
of
per kWh in Taiwan.
waste
heat
generation.
produced
from
electricity
Additionally, microgrids make
greenhouse
gas
emissions
reduction
by
providing higher power utilization efficiency and
The scheduled wind
power capacity is 2,000 MW in 2020 and the possible renewable energy capacity is 4,000 MW in 2020 [8]. The contribution of this paper is to
avoiding unnecessary transmission loss. This paper formulates an optimization
establish a framework to analyze optimal
This
operation of microgrid system in Taiwan. The
microgrid mathematical programming model is
framework contains an optimization model
formulated as a linear programming model. In
which maximizes total profit of microgrid
the
mathematical
system. A case study in Taiwan is conducted.
programming model is established on General
Renewable power production and demand of the
Algebraic Modeling System (GAMS) platform
microgrid system is measured and examined
and solved by CPLEX solver.
based on this framework.
model to analyze microgrid system.
research,
the
microgrid
GAMS is a
This paper is
modeling system designed for formulating,
organized as follows.
solving, and analyzing optimization problems.
literature review.
GAMS is a algebraic language which is useful to
formulated in the Section 3.
complex and large size problems [6]. CPLEX
case study is conducted and the results of the
is linear programming problem solver with one
microgrid model are showed in the Section 4.
of the most efficient algorithm.
The discussion and conclusion are in the
The model
simulates the supply of renewable energy
Section 2 makes
The optimization model is Accordingly, a
Section5.
including solar and wind power generators and the power demand of a single house.
A
maximal profit of microgrid is sought by GAMS
III.
Literature Review This section reviews previous studies
including establishment of microgrid system and
platform and CPLEX solver. A case study of Institute of Nuclear Energy
mathematical model of microgrid system.
The
Research’s (INER’s) microgrid system in Taiwan
following studies introduce field tests of
is tested and analyzed.
Microgrid system is
microgrid system in Japan and UK. Asano and
expected to provide higher renewable energy and
Bando [1]-[3] present an economic analysis for
reduce
microgrids field test in Japan. The New Energy
more
greenhouse
gas
emission.
Currently, the power market in Taiwan is
Industrial
dominated by a single state-owned utility
Organization (NEDO) of Japan field tests three
company
eight
regional power grids with generation resources
independent power providers (IPPs). The total
in 2005. In the demonstration of microgrids,
installed capacity is 38,346 MW that serves an
combined heat and power, gas engines, batteries,
annual demand of approximately 230,000 MW
thermal storage tanks are included.
[7]. Approximately, 5% of the total installed
paper, system configuration and operation
–
Taipower—along
with
Technology
Development
In the
strategies are determined.
The results show
economic incentive and supply reliability of
generation technologies and the barriers that microgrids have to overcome. Compared with previous studies, the
microgrid system. Hawkes and Leach [4] investigates system
contribution of our study is to formulate an
design and unit commitment in microgrids.
optimization model of microgrid system.
Also,
This linear programming model minimizes total
this paper presents a framework for finding
cost including fuel, maintenance cost, and
optimal operating strategy in microgrid system
annualized capital cost.
In this UK microgrid
(see Section 3). Additionally, the framework is
model, wing power, solar photovoltaic power,
tested and demonstrated by an INER microgrid
combined heat and power, boiler generating
in Taiwan (in Section 4).
units, electricity storage, and thermal storage facilities are considered to supply demand in a
IV.
Model
hospital, a hotel, and a leisure center. In the
The optimization model of the microgrid
result, this study shows optimal capacity
system seeks the minimal total cost. Total cost
investment and operating schedule of each
of the microgrid system is represented by z and
generating units in microgrids.
introduced in the following. The variable costs
The following studies test microgrid system
of solar, wind, biomass, gas turbine, and fuel cell
and analyze the advantages of the system.
power generation are VCPV, VCWD, VCBM, VCGT,
Lasseter and Piagi [9] addressed distributed
and VCFC in $/MWh. Notice that resource cost
power generating facilities make economic and
are included in variable cost, including biomass
environmental efficiency improvement recently;
feedstock, natural gas, and hydrogen cost.
distributed generation becomes good solutions to
Given time period can be different at each time, the
provide electricity.
Microgrids are subsystems
model assumes BKt is number of hours at time
combining distributed generators and associate
period t. The power productions are xtPV, xtWD,
demands.
xtBM, xtGT, and xtFC in MW at time t, respectively.
By islanding local generation and
demand, microgrid subsystems offer better
The
reliability in electricity supply.
operation
Additionally,
optimization in
the
model
obtains
microgrid
optimal
system
by
waste heat is able to be exploited efficiently by
maximizing total profit.
locating heat generating sources near local
profit contains generation cost terms of solar
demand.
power (VCPV×BKt×xtPV), wind power (VCWD×
In Eq. (1), the total
Abu-Sharkh et al. [10] discussed the
BKt ×xtWD), biomass power (VCBM ×BKt ×xtBM),
Microgrid systems which are more reliable and
gas turbine power (VCGT ×BKt ×xtGT), and fuel
efficient than traditional centralized generating
cell power (VCFC×BKt×xtFC). The total cost of
facilities become an alternative energy supply
the microgrid system is computed in Eq. (1).
system in recent years.
Microgrids produce
electric power on site, bringing high energy use
Min
z = Σt=1T[(VCPV×BKt×xtPV)+(VCWD×
efficiency and great integration of renewable
BKt×xtWD)+(VCBM×BKt×xtBM) +(VCGT×BKt×xtGT)
energy sources. This research also investigated
+(VCFC×BKt×xtFC)
(1)
The constraints of the model are discussed
To investigate greenhouse gas emission of
in the following. The generation capacities of
microgrid system, CO2 emission rate from
solar power, wind power, biomass power, gas
biomass electricity generation is represented by
PV
,
ERBM in kg/kWh, CO2 emission rate from gas
CPBM, CPGT, and CPFC in MW, respectively.
turbine electricity generation is represented by
turbine, and fuel cell generators are CP , CP
are number of gas turbines and
ERGT in kg/kWh, and CO2 emission rate from
number of fuel cells the decision maker decides
fuel cell electricity generation is represented by
to invest in microgrid system, respectively.
ERFC in kg/kWh.
The power generation capacity constraints are
CO2 emission of biomass, gas turbine, fuel cell
imposed for solar, wind, biomass, gas turbine,
generators.
and fuel cell power in Eqs. (2)-(6), respectively.
emission of biomass, gas turbine, fuel cell
xtBTIis
is power output
generators. Binary constraints and nonnegative
of battery. On the other hand, xtHSIis the heat
constraints for decision variables are listed in Eq.
n
GT
and n
FC
WD
the power input and xtHSO
input and
xtBTO
is heat output of heat storage
etBM, etGT, etFC are denoted by
Eqs. (14)-(16) compute CO2
(17) and Eq. (18).
device. In Eq. (7) and Eq. (8), current energy storage is equal to previous storage plus input minus output in storage devices. electricity storage at time t+1,
xt+1BT,
Then
is updated
Subject to xtPV ≤ CPPV WD
xtWD
≤ CP
by Eq. (7) and heat storage is described in Eq.
xtBM ≤ CPBM
(8).
xtGT xtFC
∀t
(2)
∀t
(3)
∀t
(4)
GT
GT
∀t
(5)
FC
FC
≤ n ×CP
∀t
(6)
constraints are established in Eq. (9) and Eq.
xt+1BT
xtBT+xtBTI-xtBTO
∀t
(7)
(10). On the demand side, the model assumes
xt+1HS = xtHS+xtHSI-xtHSO
∀t
(8)
DMtELE
xtPV+xtWD+xtBM+xtGT+xtFC-xtBTI+xtBTO
Electricity
and
heat
energy
balance
is electricity demand per hour at time t
and the unit is kW. DMtHEAT is heat demand BM
≤ n ×CP =
DMtELE BM
∀t
HR
and HRGT are combined heat production ratio of
= DMtHEAT
J/kW.
Energy demand is met by energy
generation and storage sources in Eq. (9) and Eq. (10). Fuel for biomass, gas turbine, and fuel cell generators are calculated in Eqs. (11)-(13), respectively.
Biomass feedstock demand for BM
biomass electricity is denoted by FR
in
∀t
(10)
(ytBM)
∀t
(11)
=
(ytGT)
∀t
(12)
=
(ytFC)
BM
×BKt×(xtBM)
FR
GT
FR
×BKt×(xtGT)
FC
×BKt×(xtFC)
FR
BM GT
ER
×BKt×(xtGT)
FC
ER
GT
=
∀t
(13)
(etBM)
∀t
(14)
=
(etGT)
∀t
(15)
=
(etFC)
∀t
(16)
×BKt×(xtBM)
ER
×BKt×(xtFC)
n ,n
(9)
× (xtBM)+HRGT × (xtGT)-(xtHSI)+(xtHSO)
per hour at time t and the unit is Joule. HR
biomass and gas turbine generators; the unit is
=
FC
=
= {0, 1}
(17)
kg/kWh, gas demand for gas turbine electricity
xtPV, xtWD, xtBM, xtGT, xtFC ≥ 0 ∀t
is denoted by FRGT in m3/kWh, and hydrogen
xtBT, xtBTI, xtBTO, xtHS, xtHSI, xtHSO ≥ 0 ∀t
demand for fuel cell electricity is denoted by FC
FR
3
in m /kWh.
ytBM,
ytGT, ytFC, etBM, etGT, etFC ≥ 0
∀t
each scenario, utilization of battery is examined
(18)
including average battery storage (over 24 The optimization model of microgrid
hours), total input of battery, total output of
system is an integer programming model. The
battery, and the battery efficiency (=total output/
model is formulated on GAMS system and
battery capacity).
solved by CPLEX solver.
between the microgrid system and the main grid
The optimization
Moreover, power trading
model is then used to perform a case study in
system is investigated.
The results of power
Section 4.
sell, purchase, total demand, and total profit of trading are illustrated in Tables 1-3(appendix 1).
V.
Table 1 show the demand increase
Case Study and Results This section presents a case study and
decreases profit from 1,742.0 NTD (New Taiwan
discusses the results of the microgrid system.
Dollar) to -1,376.6 NTD in no battery scenario
First, a case study of microgrid system in Taiwan
(Scenario 1). 10 kW battery scenario (Scenario
is introduced.
2) demonstrates that the increase of demand
Then the microgrid system is A
lowers 1746.8 NTD of basic case to 202.1 NTD
microgrid system of renewable energy is built by
of double demand and -1,372.2 NTD of triple
Institute of Nuclear Energy Research (INER) in
demand in Table 2.
Taiwan and the case study is analyzed in this
scenario (Scenario 3) shows the increase of
research.
demand decreases the profit from 1,749.0 NTD
investigated and the results are discussed.
The supply side of the microgrid
Also, 20 kW battery
system includes a solar power system of 100 kW
to -1,367.8 NTD in Table 3.
The additional
High Concentration PhotoVoltaic (HCPV), a 25
cost comes from the increase of demand and the
kW wind power generator, a 150 kW wind
purchase cost from main grid. Table 2 shows that 10 kW battery scenario
power generator, 2 kW fuel cell generators, and The
has highest battery efficiency of 299.7% in basic
demand side of the system includes a house and
demand case and lowest rate of 200.0% in triple
two office building with Direct Current (DC)
demand case.
electricity demand.1
scenario, the battery efficiency increases from
a 10 kWh electricity storage battery.
However, in 20 kW battery
This section begins by simulating the
179.2% to 200.0% when demand increases.
impact of increasing demand in no battery
Table 2 shows that low demand provides higher
scenario (Scenario 1), 10 kW battery scenario
battery efficiency (299.7% for Case 1) than
(Scenario 2), and 20 kW battery scenario
higher demand cases (203.2% for Cases 2 and
(Scenario 3) in the microgrid system.
200.0% for 3) here.
Assume
Case 1 with low demand
electric power demand increases from basic
in hour 13-24 doesn’t need to fill up whole
demand (Case 1) to double demand (Case 2) and
battery at hour 12. Hence, basic demand case
triple demand (Case 3) in each scenario.
(Case 1) of 20 kW battery scenario has only
In
179.2% battery efficiency (see Table 3). 1
Since photovoltaic solar power generates DC power, this research considers DC demand rather than Alternating Current (AC) to reduce conversion loss between DC and AC.
VI.
Conclusions
obtained with low battery efficiency for large
This paper formulates an optimization for The model is able to find
battery capacity case. Hence, it concludes that
optimal power operation in microgrid with
batteries of appropriate size are better than
minimal cost.
batteries with undersize or oversize.
microgrid system.
This research examines the
Microgrid is a power supply subsystem.
impact of power demand and power storage
Microgrid system is capable of integrating
device in microgrid system.
renewable energy and local supply demand.
The results demonstrate low demand In term of the usage of
Microgrid provides higher energy utilization
battery, low demand provides more extra power
efficiency by lower transmission loss and
to charge battery which yields high battery
exploiting waste heat. Additionally, microgrid
efficiency.
reduces greenhouse gas emission because of
shrinks the profit.
In contrast, the case with lower
demand might need less power than high
high power generation efficiency.
demand cases and then the battery efficiency
analyzes the optimal operation of microgrid.
could be lower in this situation. In conclusion,
Further studies include energy policy simulation
high profit is achieved for high demand scenario
in
and low profit is gained for low demand scenario.
environmental impact of microgrid system,
The
be
renewable power generation under uncertainty in
determined by both battery efficiency and power
microgrid, and multiobjective analysis for
supply situation.
microgrid system.
appropriate
battery
size
should
microgrid
system,
The paper
economic
and
By fixing demand, small battery capacity results low profit but high battery efficiency. Alternatively, high profit of energy sale is VII. Appendix Table 1 Results of no battery scenario (Scenario 1) in the microgrid
Battery Case
Power trading
Average
Total
Total
Efficiency
Total
Total
Total
Total
storage
input
output
(=output/capacity)
sell
purchase
demand
profit
(kWh)
(kWh)
(kWh)
(%)
(kWh)
(kWh)
(kWh)
(NTD)
493.6
1742.0
987.1
197.7
1480.7
-1376.6
Basic 0.0 0.0 0.0 643.6 58.6 demand (Case 1) Double 0.0 demand 0.0 0.0 439.2 347.8 (Case 2) Triple 0.0 0.0 0.0 371.3 773.4 demand (Case 3) The generation of solar power and wind power are 210 and 869 kWh, respectively.
Table 2 Results of 10kW battery (Scenario 2) scenario in the microgrid
Battery Case
Power trading
Average
Total
Total
Efficiency
Total
Total
Total
Total
storage
input
output
(=output/capacity)
sell
purchase
demand
profit
(kWh)
(kWh)
(kWh)
(%)
(kWh)
(kWh)
(kWh)
(NTD)
30.0
30.0
299.7%
621.4
36.4
493.6
1746.8
20.3
20.3
203.2%
419.2
327.8
987.1
202.1
20.0
20.0
200.0%
351.3
753.4
1480.7
-1372.2
Ba sic demand 2.0 (Case 1) Double demand 5.0 (Case 2) Triple demand 3.8 (Case 3) The gene ration of solar
power and wind power are 210 and 869 kWh, respectively.
Table 3 Results of 20kW battery (Scenario 3) scenario in the microgrid
Battery Case
Power trading
Average
Total
Total
Efficiency
Total
Total
Total
Total
storage
input
output
(=output/capacity)
sell
purchase
demand
profit
(kWh)
(kWh)
(kWh)
(%)
(kWh)
(kWh)
(kWh)
(NTD)
35.8
35.8
179.2%
611.4
26.4
493.6
1749.0
40.3
40.3
201.6%
399.2
307.8
987.1
206.5
40.0
40.0
200.0%
331.3
733.4
1480.7
-1367.8
Ba sic demand 4.8 (Case 1) Double demand 9.2 (Case 2) Triple demand 2.0 (Case 3) The gene ration of solar
power and wind power are 210 and 869 kWh, respectively.
[3] S. Bando, H. Watanabe, H. Asano, S, Tsujita, "Impact of
VIII. References
various characteristics of electricity and heat demand on
[1] H. Asano, S. Bando,"Economic analysis of microgrids,"
the optical configuration of a microgrid," Electrical
in proc. 2007 The Fourth Power Conversion Conference conf., pp. 654-658. [2] S. Bando, H. Asano, T. Tokumoto,T. Tsukada, T. Ogata, "Sensitivity analysis of the capacity of battery and photovoltaic generation and contracted demand of
Engineering in Japan, vol. 169(2), pp. 6-13, 2009 [4] A. D. Hawkes, M. A. Leach,
"Modelling
high level
system design and unit commitment for a microgrid," Applied Energy , vol.86:pp1253-1265, 2009 [5] M. Armando. Leite da Silva, João Guilherme de Carvalho
"Transmission
purchased power in a microgrid". IEE of Japan
Costa.
Transactions on Power and Energy, vol. 127(7):pp.
energy market,
783-790, 2007.
vol. 18(4), pp.1389-1394, 2003
loss allocation: part I—single
" IEEE Transactions on Power Systems,
[6]R.E. Rosenthal, GAMS: a user’s guide. [Online]. Available: http://www.gams.com/dd/docs/bigdocs/GAMSUsersGuid e.pdf, 2008. [7] K.M Wang.,
"
supply industry,
The deregulation of Taiwan electricity
"
Energy Policy, vol.34, pp2509-2520,
2006 [8] Bureau of Energy in Taiwan. Renewable Energy Policy. 2010. [Online] Available: http://www.moeaboe.gov.tw/Policy/Renewable/meeting/ SEmeetingMain.aspx?pageid=commend. [9] R. H. Lasseter, P. Piagi. solution,
"
"Microgrid:
a conceptual
presented at PESC’04, Aachen, Germany,
2004. [10] S. Abu-Sharkh, RJ. Arnold, J. Kohler, R. Li, T. Markvart, JN. Ross, K. Steemers, P. Wilson, R. Yao, "Can microgrids make a major contribution to UK energy supply?" Renewable and Sustainable Energy Reviews Vol.10, pp.78–127, 2006