Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

A methodology to Optimize Benefits of Microgrids

Mike Quashie, Student Member IEEE


Geza Joos, Fellow IEEE
Dept. Of Electrical and Electronic Engineering
Mcgill University,
Montreal, Canada
mike.quashie@mail.mcgill.ca,
geza.joos@mcgill.ca

Abstract—This paper proposes a generalized methodology to reduction in the cost of energy, investment deferral, reduction
determine the optimal configuration of microgrids that in power losses and reduction in greenhouse gas emission.
maximizes its benefits. The technique proposed herein, These benefits may affect stakeholders within the power
incorporates a probabilistic modeling of variable loads and system such as the utility, microgrid customers, independent
distributed generators into a deterministic power flow problem power producers, policy makers and society. Most of these
to solve an optimization problem. The losses from the power benefits are interrelated and an improvement in one may affect
flow problem is monetized and supplied to the objective function another. However, there are benefits which may affect
as one of its input multiple cost functions. The optimization particular stakeholders such as infrastructure investment
problem then concatenates the monetized losses, monetized
deferral, may benefit the utility or the distributed network
emission and the cost components of generation into a single
operator [5]. The cost associated with implementing the
objective and then minimizes it. The optimization’s aim is to
minimize the energy cost to the utility while maximizing loss
infrastructures of microgrid is of great concern to stakeholders
reduction and emission reduction under all practical constraints. hence the need to understand how to maximize the value
The proposed algorithm is applied to a real feeder in North derived from its benefits. This would inform a better
America and the results showed a significant decrease in the cost investment decision in microgrid and enhance the drive
of energy. This demonstrates the effectiveness of the towards its advancement. Various studies appear in literatures
methodology proposed in microgrid planning. towards quantifying and optimizing some of these benefits.
Morris et al. developed a framework for quantifying these
Index Terms— Benefits, Distributed Generation, Microgrid, benefits considering stakeholders participation. In [1] an
Monetized Losses, Optimization. operational and planning methodology is also proposed to find
the minimum cost of energy using HOMER software [1]. The
I. INTRODUCTION use of HOMER, however, limits the flexibility of the
Over the last few years, the traditional power system has methodology proposed. Dicarto et al. also proposed an
undergone tremendous changes. Government policies, optimization procedure for economic evaluation of microgrids
technological advancement, economic and environmental [6]. However this paper proposes a more flexible optimization
incentives, seem to be changing the features of electricity methodology which aims to maximize microgrids’ benefits. It
generation, transmission and distribution [1] while distributed also includes more of microgrids’ benefits in its optimization
generation gradually increases its presence in the power problem such as distribution losses which the US energy
systems. Recent power outages during hurricane Katrina and information agency and energy statistics, Canada, estimates to
Sandy have exposed the vulnerability of a centralized power be 65.661 billion kilowatt hours for the year 2011 in Canada
system and highlighted the benefits of microgrids. Researchers and about the same figure for previous years [7].
view microgrids as perhaps the epitome of the move toward Herein, the paper outlines a methodology to determine the
distributed power, where distributed energy resources (DER) optimal configuration of microgrids to maximize its benefits.
would coordinate to serve the needs of the local distribution The optimization problem concatenates the benefits as
network and provide services to the main grid [2], [3]. The multiple cost functions into a single objective function and
term microgrid refers to a distribution system with distributed then minimizes it. The benefits considered include minimizing
energy resources that may operate connected to the main distribution losses, reduction in generation cost and
power grid or in isolation (islanded mode) [4]. Microgrid is greenhouse gas emission. Section II of the paper outlines the
seen as the technology that can improve local system modeling of the stochastic characteristics of the renewable
reliability and aid in the integration of renewable energy resources considered, Section III describes the formulation of
resources. Alongside reliability, microgrids may provide the optimization problem, Section IV details the

This work was sponsored in part by National Science and Engineering


Research Council (NSERC) of Canada through a grant to the NSERC Smart
Microgrid Network (NSMG-NET)

978-1-4799-1303-9/13/$31.00 ©2013 IEEE


implementation of the methodology on a real feeder and the HOURLY WIND POWER OUTPUT
results and analysis are presented in Section V. 0.1

II. MODELING OF RENEWABLE ENERGY RESOURCES


0.09

HOURLY OUTPUT IN PER UNIT (P. U)


A. Wind Energy
Uncertainties associated with renewable resources such as 0.08
wind and solar energy requires special consideration in power
system planning and analysis. Meteorological conditions such 0.07
as wind speed, solar irradiance, ambient temperature and the
like, highly influence the output of wind and solar energy.
0.06
Careful analyses have to be done to provide proper models
that reflect these characteristics. Herein, the weibull
distribution function is used to model the wind speed. The 0.05 Winter

weibull distribution has been advocated as the best model for Spring
Summer
wind speed [8]. Historical data of wind speed over a six year 0.04
Fall
0 5 10 15 20 25
period is obtained from Canadian weather energy and Hours
engineering datasets (CWEEDS). The data is utilized to
estimate an hourly frequency distribution of the wind speed Figure 1. Hourly Wind Power Output
for each season of the year, with each season represented by a
24hr daily distribution. The mean and standard deviation of B. Solar Energy Model
the data are computed as shown in (4) and (5) respectively. Unlike wind speed, the variability of solar irradiance is not
The weibull probability distribution function (pdfs) needed to totally random. Its value is certain to be zero at nights hence
model the hourly wind speed is then computed as a function of the use of weibull pdfs may not be appropriate in its modeling
the mean speed (xm) and standard deviation (σ) as shown in [9]. Herein, the Bimodal Beta pdfs model is used in describing
(1). The shape parameter (k) and scale parameter (c) are also the solar irradiance based on historical data as outlined in the
obtained as in (2) and (3) respectively. The output power (PW) sub section above. The mean and standard deviation of the
is computed as a function of the modeled hourly speed as in data is used to compute the parameters of the Beta distribution
(7). The rated power (Pr), cut in speed (xc), rated speed (xr) as shown in (8) and (9).
and cut out speed (xo), is provided on the manufacturer’s
specification. Fig. 1 shows the variable output of the wind ( )
= −1 (8)
generator.
( )
= (1 − ) −1 (9)
( )= ( ) (1)
Subsequently the probability distribution function is
=( ) .
(2) calculated as in (10) with ( ) as a gamma function. The
output power (Ps) is then calculated as a function of the solar
= (3) irradiance (s) and ambient temperature (T) as shown in (11).
( ) The maximum power output (Pmax), reference temperature
(Tref) and reference irradiance (Sref) is predetermined or
= ∑ (4) provided on the manufacturer’s specification. Fig. 2 also
. illustrates the output of the solar panel.
σ= ∑ ( − ) (5)
( )( )
( )= (10)
( ) ( )

Where n is the data size. When the value of k equals 2, = ∗ ∗ 1+ ( − ) (11)


f(x) becomes (6) as shown below and termed Raleigh
distribution.

( )= ( ) (6)

0, 0≤ ≤
P ∗ , < ≤
= (7)
P, < ≤
0, <
HOURLY SOLAR OUTPUT A. Monetized Losses(ML)
0.06 Winter
Spring The power flow problem uses the forward and backward
Summer
Fall
iteration method to compute losses in the network. It takes the
0.05
variable output of the renewable resources and the variable
HOURLY OUTPUT IN PER UNIT (P.U)

load as input and computes the losses for various


0.04 configuration and sizing of the renewable resources for each
time period for each node (j) as in (15). The losses are then
0.03 monetized using wholesale price of energy (CP) as shown in
(16). It is worth noting that reduction in losses benefit the
distribution network operator as a stakeholder.
0.02
(, , , ) (, , , )
(t, n, µ, σ) = ∑ × (15)
(, , , )
0.01

ML (t, n, µ, σ) = (t, n, µ, σ) × CP (16)


0
0 5 10 15 20 25 B. Generation Cost Component(GCC)
HOURS
The generation cost component as explained earlier is the
Figure 2. Hourly Solar output sum of the generation cost of electricity from the grid and that
of distributed generators (DGs) as in (17). Herein, the paper
The output of the renewable resources shows the need for assumes a flat rate for the cost of electricity from the grid,
storage during the some hours of the day. The paper assumed since detailed computation of the operational cost of the
that the distributed resources may have enough storage and the central grid is outside the scope of the paper. However, for the
work herein does not elaborate on storage sizing, location and DGs, the annualized cost of investing in the DGs is outlined in
scheduling. The load is assumed to follow the IEEE RTS (19) with the assumption that the operational cost associated
hourly distribution load curve. with the DGs is negligible. Annualized cost herein, refers to
the cost that needs to be recovered for a payback period of y
III. OPTIMIZATION ALGORITHM FORMULATION years, at an interest rate of r [10], [11]. Also, the capital costs
The technique proposed in this paper incorporates a (CC) considered in the paper includes the cost of renewable
probabilistic modeling of variable loads (PLoad) and distributed resources, its installation and other system component costs in
generators into a deterministic power flow problem to solve an the microgrid. The salvage value (SV) is discounted to the
optimization problem. The losses (PLOSS) from the power flow present value and subtracted from the capital cost. The
problem is monetized and supplied to the objective function as difference is then annualized using the capital recovery factor
one of its input multiple cost functions. The objective function (CRF) as outlined in (18) and (19) [11]. With the annualized
is formulated as a sum of the monetized losses (ML), cost cost of energy computed, the average cost of energy (AVC)
component of generation (GCC) and cost of emission of from the DGs for wind (CPW) and solar (CPS) could be
greenhouse gases (GHC) as in (12). The objective function is obtained as in (20) and used in (18).
then minimized subject to energy balance, penetration limit
(Ppen), of the distributed resources, technical and reliability ( , )=
constrains using the heuristic search method and cost benefit ∗ + , (t, n, µ, σ) ∗ +
analysis as in (13) and (14). The cost component of generation , (t, n, µ, σ) ∗ (17)
herein refers to the cost of the grid’s power (PG) and that of ( )
distributed generators (DG). In essence the objective function = (18)
( )
represents the total cost of energy to the utility operator in this = ×( − ) (19)
instance. ($/ )
= (20)
( / )

min = (ML(t, n, µ, σ) + GCC(t, n) + GHC(t, n)) (12) C. Cost of Emissions(EC)


Mathematical models presented in this sub-section are
Subject to: used to monetize benefits associated with pollutant emissions,
which are the primary environmental benefit of microgrids
, − , (t, n, µ, σ) + , (t, n, µ, σ) + [12]. Benefits associated with environment are not specific to
, (t, n, µ, σ) = , (t, n) (13) utilities nor customers but the society as well. Herein, the
paper considers reduction in carbon dioxide (CO2) associated
,(t, n, µ, σ) + , (t, n, µ, σ) ≤ , × with use renewable resources compared to the use of fossil
, (t, n) (14) fuel. Carbon tax imposed on a ton CO2 emitted is used in
monetizing the emission associated with the central power
Where t is the time (hour of the day), n, p the number of generation to the grid as outlined in (21).
configuration, j is the bus (node) number while μ and σ
represents the estimated mean and standard deviation of the $
= × × (21)
variable characteristics of the renewable resources.
IV. IMPLEMENTION OF ALGORITHM V. RESULTS AND ANALYSIS
The proposed algorithm is tested on International Council on The proposed algorithm is used in evaluating the cost of
Large Energy system’s (Cigre’s) medium voltage North energy for different positioning and sizing of DGs in the
American distribution network benchmark. This benchmark above network. The algorithm uses the heuristic method to
is capable of operating as a microgrid with distributed determine the optimal configuration that results in the
generators (DGs) and storage connected [12]. The minimum cost of energy as outlined in section III. Different
distribution network with no DGs and storage connected is scenarios were considered in determining the optimal
considered as the base case. The candidate nodes for DG configuration: (I) No DGs installed, (II) PV only installed,
installation are nodes 4, 7 and 10. The historical (III) Wind only installed, and (IV) Both PV and Wind
meteorological data of Varennes, Quebec is taken as input for installed. Though all scenarios showed some decrease in the
average peak cost of energy as compared to the base case, the
the renewable resources model. The peak load of the network
fourth option provided the most decrease. The optimal
is given as 12.775MW [12]. The renewable resources configuration was also found to be 1.2 MW of wind turbine on
considered here are solar and wind with a maximum rating of node 7 and 30 KW solar panel on node 4. A decrease of 16.1
30KW and 1.5MW respectively. The cost of electricity percent is observed in the average cost of energy for the
supplied by the grid is considered to be $0.04/KWh, Quebec optimal case in comparison to the base case as illustrated in
wholesale electricity cost [13]. That supplied by the Solar fig. 4. The monetized benefits accumulating from the loss
Panel is obtained as $0.20/KWh while the wind turbine minimization and greenhouse gas reduction, offset the cost
supplies its energy at $0.13/KWh. A carbon tax of $10 per associated with energy generation from the DGs and other
ton of CO2 is considered in estimating the monetized microgrid infrastructures, resulting in the lower energy cost
environmental benefit [14]. It is also assumed that the source for the optimal case. The variable nature of the cost function is
of the grid’s supply is diesel hence generates 778g of CO2 due to the variability in demand and generational output
per KWh [15]. The network, with DGS connected operates as alongside the losses (PLOSS) component of the cost function.
a microgrid connected to the main grid at the point of The efficiency of the network is also improved by 10 percent
common coupling as shown in fig. 3. The DGs shown on the through loss minimization for the optimal case in comparison
figure represent candidate nodes for possible placement of to the base case. The other scenarios also resulted in some loss
DG. The load is assumed to follow the IEEE RTS hourly minimization as illustrated in table 1. However, the
distribution load curve. The network load is considered to improvement in efficiency was very minimal for case II with
increase by 10 percent annually. solar only. The output of solar panels during some hours of the
night is certain to be zero for some seasons so the network
then operates as in the base case resulting in the minimum
efficiency improvement. Scenario III on the other hand
showed significant improvement closer to that of the optimal
case.

COST OF ENERGY TO THE DISTRIBUTION SYSTEM OPERATOR


0.06

0.058

0.056
COST OF ENERGY(S/KWh)

0.054

0.052

0.05

0.048

0.046

0.044

0.042 Base Case


Optimal Case
0.04
0 10 20 30 40 50 60 70 80 90 100

WINTER SPRING SUMMER FALL


HOURS

Figure 4. Comparison of Hourly Total Cost of Energy to the Distributed


Network Operator for a Year (each 24hr represents a Season)

Figure 3. Cigre's North American medium Voltage Distribution Network


Benchmark with DG connected to operate as Microgrid [12].
TABLE I. LOSSES IN THE POWER NETWORK FOR VARIOUS PowerTech, 2009 IEEE Bucharest, vol., no., pp.1-6, June 28
SCENARIOS 2009-July 2 2009
[7] Energy statistics handbookfirst quarter 2012 [online]. Available:
Average Losses(MWper hr) http://www.statcan.gc.ca/pub/57-601-x/57-601-x2012001-eng.pdf
[8] Momoh, James A.; D'Arnaud, Keisha; , "Optimizing grid
Seasons Winter Spring Summer Fall Annual(MWh) connected renewable energy resources with variability," Power
I. Base Case(No DG) and Energy Society General Meeting, 2012 IEEE , vol., no., pp.1-
6, 22-26 July 2012
0.6757 0.6743 0.6770 0.6743 243.1123 [9] Yasser Moustafa Atwa, “Distribution system planning and
II. Solar Only Reliability assessment under high DG penetration," Ph.D.
dissertation, Dept. of Electrical and Computer Engineering,
0.6620 0.6605 0.6631 0.6605 238.1372 Waterloo, Canada.
[10] S. M. Moghaddas-Tafreshi, H. A. Zamani, and S. M. Hakimi,
III. Wind Only
“Optimal sizing of distributed resources in micro grid with loss of
0.6037 0.5992 0.6160 0.6082 218.4336 power supply probability technology by using breeding particle
swarm optimization,” Journal Renewable Sustainable Energy, vol
IV. Wind and Solar
3, 043105, pp1-17, July, 2011.
0.6035 0.5988 0.6155 0.6078 218.3088 [11] Masters, Gilbert M.,"Renewable and efficient electric power
systems,” New Jersey: Wiley, 2004, p. 372-376.
[12] M. Wakefield, “Methodological approach for estimating the
benefits and costs of smart grid demonstration projects,” EPRI,
VI. CONCLUSION Palo Alto, CA: 2010. 1020342.
The proposed methodology has been developed and [13] CIGRE TF C6.04.02, “Benchmark systems for network
applied to cigre’s medium voltage North American integration of renewable and distributed energy resources,”
distribution network benchmark, capable of operating as a technical brochure, version 21, pp25-39, August 2010.
microgrid. Application of the methodology to the microgrid [14] Hydro Quebec, “Distribution tariff effective April,2012”,
planning process resulted in significant decrease in the cost of available at http://www.hydroquebec.com/publications/en/rates/
energy to the distribution network operator. A significant pdf/distribution_tariff.pdf
improvement in the network’s efficiency (loss minimization) [15] BC Ministry of Finance, “British Columbia carbon tax.” available
as well as reduction in CO2 emissions is also observed in on the Internet at http://www.fin.gov.bc.ca/tbs/tp /climate/A4.htm.
applying the methodology to the microgrid planning process. [16] Benjamin K. Sovacool, Valuing the greenhouse gas emissions
The economic, technical and environmental benefits of the from nuclear power: A critical survey, Energy Policy, Volume 36,
Issue 8, August 2008, Pages 2950-2963, April, 2008.
microgid which are maximized through the use of the
proposed methodology help offset the cost associated with
microgrid’s implementation, building a better business case
for microgrid advancement. The methodology could be
extended to include storage sizing and scheduling in further
studies. This methodology is useful for long term planning
purposes of microgrids.

VII. REFERENCES
[1] Wencong Su; Zhiyong Yuan; Mo-Yuen Chow; , "Microgrid
planning and operation: Solar energy and wind energy," Power
and Energy Society General Meeting, 2010 IEEE , vol., no., pp.1-
7, 25-29 July, 2010.
[2] Venkataramanan, G.; Marnay, C.; , "A larger role for microgrids,"
Power and Energy Magazine, IEEE , vol.6, no.3, pp.78-82, May-
June 2008
[3] Marnay, C.; Venkataramanan, G.; Stadler, M.; Siddiqui, A.S.;
Firestone, R.; Chandran, B.; , "Optimal Technology Selection and
Operation of Commercial-Building Microgrids," Power Systems,
IEEE Transactions on , vol.23, no.3, pp.975-982, Aug. 2008
[4] Katiraei, F.; Iravani, R.; Hatziargyriou, N.; Dimeas, A.; ,
"Microgrids management," Power and Energy Magazine, IEEE ,
vol.6, no.3, pp.54-65, May-June 2008
[5] Morris, Greg; Abbey, Chad; Joos, Geza; Marnay, Chris; “A
Framework for the Evaluation of the Cost and Benefits of
Microgrids,” CIGRÉ International Symposium, Bologna, Italy,
September 2011
[6] Dicorato, M.; Forte, G.; Trovato, M.; , "A procedure for
evaluating microgrids technical and economic feasibility issues,"

You might also like