This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON SMART GRID

Multi-Objective Optimization and Design of

Photovoltaic-Wind Hybrid System for Community
Smart DC Microgrid
Mohammad B. Shadmand, Student Member, IEEE, and Robert S. Balog, Senior Member, IEEE

AbstractRenewable energy sources continues to gain popularity. However, two major limitations exist that prevent widespread adoption: availability of the electricity generated and the
cost of the equipment. Distributed generation, (DG) grid-tied photovoltaic-wind hybrid systems with centralized battery back-up,
can help mitigate the variability of the renewable energy resource.
The downside, however, is the cost of the equipment needed to
create such a system. Thus, optimization of generation and storage
in light of capital cost and variability mitigation is imperative to
the financial feasibility of DC microgrid systems. PV and wind
generation are both time dependent and variable but are highly
correlated, which make them ideal for a dual-sourced hybrid
system. This paper presents an optimization technique base on
a Multi-Objective Genetic Algorithm (MOGA) which uses high
temporal resolution insolation data taken at 10 seconds data rate
instead of more commonly used hourly data rate. The proposed
methodology employs a techno-economic approach to determine
the system design optimized by considering multiple criteria
including size, cost, and availability. The result is the baseline
system cost necessary to meet the load requirements and which
can also be used to monetize ancillary services that the smart
DC microgrid can provide to the utility at the point of common
coupling (PCC) such as voltage regulation. The hybrid smart DC
microgrid community system optimized using high-temporal resolution data is compared to a system optimized using lower-rate
temporal data to examine the effect of the temporal sampling of
the renewable energy resource.

PV surface area.
,

Wind footprint area.

Battery capacity.
Ratio of power imported from grid to load.
Yearly O&M per unit.
Inflation rate.
Power purchased from utility at time .
Operational duration under consideration.
Index of availability.

DNM

Demand not met (kWh/year).

Yearly demand.
Demand at time .
Minimum allowable storage level at time .
State of charge of battery bank at time .
Price of grid power.
Interest rate.

Index TermsGenetic algorithm, microgrid, optimization, photovoltaic, PV-storage system, smart grid, wind turbine.

Escalation rate.
Photovoltaic system efficiency.
Wind turbine system efficiency.

I. NOMENCLATURE
Initial cost.

II. INTRODUCTION

Operation & Maintenance cost.

Life cycle of the system.
Cost of power imported from grid.
Manuscript received August 12, 2013; revised November 23, 2013; accepted
March 23, 2014. Funding for the 27.6 kW photovoltaic system studied in this
paper at Texas A&M University, College Station, Texas was provided by a grant
from the US Department of Energy (DOE) and the Texas State Energy Conservation Office (SECO). Paper no. TSG-00652-2013.
The authors are with the Renewable Energy and Advanced Power Electronics Research Laboratory in the Department of Electrical and Computer
Engineering, Texas A&M University, College Station, TX 77843 USA (e-mail:
mohamadshadmand@gmail.com; robert.balog@ieee.org).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSG.2014.2315043

C distribution systems are ideal for integrating distribute renewable energy sources and energy storage
into point-of-use energy systems [1][3]. Renewable energy
adoption has increased with 60% annual growth in the installed
capacity of photovoltaic (PV) systems from 2004 to 2009, and
80% in 2011 [4]. However, two major fundamental limitation
exists that prevent truly widespread adoption: availability
of electricity generated and cost of equipment. At the same
time, DC systems have been gaining popularity because of
the high efficiency, high reliability and easy interconnection
of renewable sources compared to AC systems [1], [5]. A DC
microgrid system with distributed PV and wind generation and
employing centralized battery storage, illustrated in Fig. 1, is an

1949-3053 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2

IEEE TRANSACTIONS ON SMART GRID

Fig. 2. Example of high temporal resolution PV data on a cloudy day.

Fig. 1. One line system diagram.

attractive technology solution for communities to go-green

while simultaneously ensures reliable electricity.
To mitigate fluctuations in the generation portion of the hybrid system, a battery bank is used for energy storage. This
will absorb the surplus power and supply deficit power under
different operating conditions. The proposed methodology has
been previously used for assessing standalone distributed photovoltaic systems and has been experimentally verified through
long-term field testing [6].
DC distributed generation (DG), grid-tied PV-wind systems
with centralized battery back-up, illustrated in Fig. 1, have been
proposed for community-scale microgrids such as the Pecan
Street project in Austin Texas [7] and other communities [8]. An
example of the power electronic interface between the utilitypower grid and the community distribution bus is the solid state
transformer [9], [10] being developed by the FREEDM Center
at NC State, Fig. 1. The control and management of the islanded
PV-wind and storage systems have been proposed in [11][14].
Due to the stochastic behavior of wind and solar energy, one
of the most vital factors in optimal sizing of the hybrid systems is the temporal resolution of collected data to provide sufficient resolution [15], particularly on cloudy days, Fig. 2. Optimization techniques such as Multi-Objective Genetic Algorithm
(MOGA) [16][18] are able to take advantage of high temporal
resolution data for renewable resources instead of relying only
on mathematical probabilistic models. The main contribution of
this paper is to employ an energy availability technical assessment in conjunction with economic assessment to optimize the
sizing of the PV panel, wind turbine, battery storage, and the
power imported from grid in order to determine the system that
would guarantee a reliable energy supply with the lowest investment.
The objective of this paper is to simultaneously maximize the
power availability and minimize the cost which will optimize

system size with highest availability. The problem is firstly visualized based on MOGA technique and Pareto Front for purpose
of engineering trade off when analyzing high temporal resolution data. Then a utility function will be determined in order to
do a decision making for the multi-objective problem. Uncertainty analysis is added because of the stochastic behavior of
the insolation and wind speed data.
The average month technique [19], [20] is not used in this
paper because the system designed in this way may not be able
to satisfy the load during some periods of real-work operation.
On the other hand, the calculation of each subsystem (wind,
photovoltaic, storage, and ratio of power imported from grid)
separately for the worst month in each resource makes the total
system oversized. The optimized hybrid system, based on accurate and enhanced 10 seconds insolation data rate of photovoltaic system, is compared to conventional PV-Wind optimized systems based on hourly insolation data. Using MOGA as
the optimization technique when analyzing high temporal resolution insolation data shows that the system availability is maximized for lowest possible cost comparing to conventional hybrid system sizing [21][23].
Due to the different operating life of various components, reliability analysis is critical [24], [25]. In this paper, though no
reliability analysis details will be discussed, the results of the
extra cost overhead to the owner due to maintenance and repairs will be included. In addition, the mathematical modeling
of the system considers the economic aspects such as inflation,
interest, and escalation rates which makes the model more realistic.
III. CASE STUDY FOR APARTMENT COMPLEXES
The case study is a grid-tied community living environment
in College Station, Texas. The case study uses high temporal
resolution data collected from a 27.6 kW PV system installed on
Texas A&M University campus [26]. The system, configured as
five independent residential-scale arrays, has the PV generation
(ac output) data sampled every 10 seconds, Fig. 2. Details of the
hybrid system and selected site are:
The apartment complex consists of 70 units with 28 two
story buildings, the load schedule is shown in Table I
The monthly average load requirement for the apartment
complex is approximately equal to 82,920 kWh
There is 5.27 kWh/m2 of available incident solar energy
The site has class-1 wind [27], and Wind Finder [28] is
used to gather wind resource data
The hourly demand of the case study is illustrated in Fig. 3.
The load profile of each apartment is estimated over an entire
year as it is illustrated in Fig. 4. A center of mass approach is

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
SHADMAND AND BALOG: MULTI-OBJECTIVE OPTIMIZATION AND DESIGN OF PHOTOVOLTAIC-WIND HYBRID SYSTEM

IV. PROBLEM STATEMENT

TABLE I
SUMMARY OF APARTMENT COMPLEX ENERGY USAGE

In this section, the objectives of the system will be mathematically formulated for optimization. These objectives are to
simultaneously maximize the power availability and minimize
the cost which minimizes the system size with highest possible
availability. The output power of PV and wind generators has
the highest priority to feed the DC bus, and if the power generated is inadequate, the battery bank can be discharged to a certain amount to feed the bus. If there is still insufficient power,
a certain amount of power can be purchased from grid to feed
the load. Thus, the power imported from grid has the lowest priority.
A. Cost
Major component of the system cost consists of the price of
PV panels, wind turbines, and battery bank. The total system
cost ($/year) includes initial cost and operational & maintenance
cost (O&M), this can be formulated as
(1)

Fig. 3. Daily load profile.

and
indicate the initial cost and Operation &
where
Maintenance (O&M) cost of each individual components respectively.
and
are the life cycle of the system and
cost of power imported from grid respectively. So the first objective can be formulated as
(2)
where
,
,
and are the design parameters
in this project.
For the photovoltaic sub-system, the initial and O&M cost
can be formulated as
(3)
(4)
For the wind turbine sub-system, the initial and O&M cost
can be similarly formulated as
(5)

Fig. 4. Estimated monthly residential load profiles.

(6)
utilized to estimate the load profiles for apartment units which
lie between energy profiles. For example the unit with 1150 kWh
which is between 1000 kWh and 1500 kWh, we get
, where
. Therefore, 1150 kWh
has 70% resemblance to 1000 kWh load pattern and 30% to
1500 load pattern. By using this methodology and base on the
information provided in Table I, the monthly average load requirement for the apartment complex is determined to be approximately 82,920 kWh.

For the battery bank, the initial cost and O&M cost can be
formulated as
(7)

(8)

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4

IEEE TRANSACTIONS ON SMART GRID

For the battery bank, since the operational life cycle is less than
PVs and wind turbines, it is expected that they must be replaced
several times during the system life span. This replacement cost
is taken into consideration as O&M costs in (8).
The cost of importing power from the grid can be formulated
as

TABLE II
NUMERICAL VALUE OF HYBRID SYSTEM PARAMETERS

(9)

TABLE III
CONSTRAINTS NUMERICAL DATA

B. Availability
Availability, the fraction of the time when energy is available,
is a key figure of merit for the proposed system. It is important to make a clear distinction between availability and reliability. Reliability is the ability of the system to operate without
failure; availability is the ability of the system to supply power
to the load. As an example, a highly reliable photovoltaic energy system, where the components are not prone to failure, can
have low availability if there is insufficient energy storage to
support the loads power requirements during the night or during
an overcast day.
A specified level of availability can be achieved with many
configuration of a system. The availability can be formulated
for duration under consideration as
(10)
The DNM can be formulated as

V. OPTIMIZATION
A. Design Constraints
A physical constraint which must be added to the optimization algorithm is the available area for PV panels and wind generators installation:
(16)
(17)

(11)
where
is a step function which is zero if the supply power
is greater or equal demand and one if the demand is not met.
The imported power from grid is:

Obviously the lower bounds can be zero, but in order to make

the system more reliable for the purpose of uncertain analysis,
the lower bound for wind turbines is decided to be about 100 .
For the selected apartment complex the upper bound was determined to be approximately 12% of the available area, which was
approximately 4,221
.
The imported power from the grid should be within a certain
range:

(12)

(18)

(13)

The fraction of the power to be imported from grid can vary

from zero to one, this can be seen from (12).
Finally, the total generated power should not exceed the demand in order to avoid oversizing the system and adding excessive cost. This constraint is formulated as:

where

(14)
The second objective can be formulated as:

(19)
(15)
The hybrid system parameters values are given in Table II.

The numerical values of design constraints are given in Table III

for the presented case study.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
SHADMAND AND BALOG: MULTI-OBJECTIVE OPTIMIZATION AND DESIGN OF PHOTOVOLTAIC-WIND HYBRID SYSTEM

B. NSGA-II Optimization for the Hybrid System

Multi Objective Genetic Algorithm (MOGA) which is commonly called Non-dominated Sorting Genetic Algorithm-II
(NSGA-II) [18], [29] is used as an optimization algorithm or
search method to find a set of equally good solutions for the
objectives mentioned in Section III as a form of a Pareto frontier. However other optimization techniques can be used, but
this method has been one of the most popular heuristic search
methods for multi-objective optimization [18]. The Genetic
Algorithm is a well-known non-gradient-based search method
which mimics the natural evolution process.
The key distinction between single-objective and multi-objective optimization is that in the case of multi-objective optimization, there may be multiple feasible solutions that satisfy
the optimization criterion. Further, it may not be possible to
identify one solution as being better than another if neither is
dominated by the other in some sense. In other words, in multiobjective optimization, there could exist a set of equally-good
solutions rather that a single solution as we expect in single-objective optimization problems. The results of multi-objective
optimization can be also be described as a set of non-dominated
solutions, the so called Pareto frontier.
MOGA uses Genetic Algorithm (GA) as its core with two
important new concepts in order to achieve good multi-objective optimization. These two concepts are non-dominated
sorting and crowding distance as described in [16], [18]. By
using these two concepts and GA principle, MOGA algorithm
can be formed which is called Non-dominated Sorting Genetic Algorithm-II (NSGA-II) [18]. This algorithm is briefly
explained in this section.
In iteration , parent population
and offspring population
, each with solutions, are combined to form a bigger population with
solutions. Non-dominated sorting is then performed to find solutions with similar ranks.
A sorting process is then performed by selecting solutions
with the lowest rank and then solutions with the next lowest rank
and so on. This process continues until the number of solutions
in the parent population exceeds . Then, for the latest sorted
subpopulation included in the parent population, only the solutions with a larger crowding distance are selected until parent
population has exactly solutions. Crossover and mutation operators are then performed to find the next offspring population.
Fig. 5 illustrates the flowchart of NSGA-II algorithm used in
this paper for optimizing the hybrid system.
The comprehensive design procedure of the hybrid system
for DC smart microgrid is illustrated in Fig. 6. Without loss of
generality, the models of the various components, including the
wind generator, solar cells, and power electronics interfaces can
be made arbitrarily complex to improve the fidelity of the model.
In a PV-wind hybrid energy harvesting system, there are many
factors which contribute to the overall conversion efficiency.
One of the most important conditions is the geographical location where the system is deployed which determines the latitude
and the meteorological conditions. Therefore, the first step in the
design procedure is to specify the geographical location. Based
on the location, conditions such as available wind and solar

Fig. 5. MOGA (NSGA-II) optimization algorithm for hybrid system.

resources and weather data are used to determine the ratings

for the PV array and wind generator. Usable roof area, desired
availability of power (loss of load probability), desired lifetime,
and limits on maximum imported grid power are considered as
design constraints for the objective functions. Finally, by using
the optimization tool illustrated in Fig. 5, a set of equally good
solutions, the Pareto Frontier, is found that maximize the availability and minimize the cost of the hybrid system.
Fig. 7 illustrates a set of optimal design solution for the hybrid
system. The PV panel cost (
), wind turbine cost (
),
battery bank cost (
), and price of grid power (
) are
assumed 450
, 100
, 100 $/kWh, and 0.10 $/kWh
respectively. Fig. 7 illustrates the trade-off between cost and
availability. This is due primarily from the need to supply the

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
6

IEEE TRANSACTIONS ON SMART GRID

VI. DECISION MAKING

The set of optimal solution given by the Pareto Frontier is
illustrated in Fig. 7 and can be used to visualize the optimal solutions and perform engineering tradeoff studies. In this paper
the utility theory [30], [31] is used to decide the optimal solution
based on the preferences for the smart DC microgrid. Therefore,
a function need to be defined to clearly rank-order the alternatives for decision making, this function is commonly called the
utility or value function [31]. Usually the term value function
denotes the decision under certainty and the term utility function denotes a decision under uncertainty.
Two decision making scenarios are investigated in this paper:
with and without uncertainty in the solar radiation, wind speed,
and demand data. Firstly a utility function will be formulated
for decision on system design without uncertainty, then the expected value of utility function will be used for decision by
taking into consideration the uncertainties.
A. Decision Without Uncertainty

Fig. 6. Comprehensive design and optimization procedure of the hybrid system

for smart DC microgrid.

Objectives and attributes are used as tools for modeling the

preferences or the utility function, and then an irrevocable allocation of resources will be done. An attribute, or figure of merit
(FOM), is the measure of progress toward an objective. The attributes in this paper are the availability and cost. So a function
need to be defined that relates every point in an -dimensional
attribute space to a scalar value or utility as follow

(20)
which allow the designer to rank order the alternatives. The next
step is to convert all attributes to same scale, frequently called
pricing out. The summary of the decision modeling without
uncertainty is illustrated in Fig. 8. By using the procedure illustrated in Fig. 8, the general form of value function for the hybrid
system optimization is given by

(21)

Fig. 7. Pareto frontier reveals the set of optimal solutions found from the evaluation of (1)(15) for the proposed hybrid system.

peak load during as determined by the system energy availability. Thus, the Pareto Frontier is a tool that enables engineering tradeoff analysis to choose the unique design from the
set of feasible designs based on particular preferences.
Due to stochastic behavior of the solar and wind, in this paper
the decision making is done for two scenarios: without uncertainty and with uncertainty. A more detail discussion on uncertainty analysis is presented in the next section.

and
are the weighing factors that can be defined
where
by the designer. If
, it means the designer are indifferent between the availability and cost of the hybrid system for
the DC distribution systems. Consequently, the proposed value
function provides the designer the ability to perform the engineering trade study. Finally by minimizing
using
GA optimization toolbox in MATLAB, the irrevocable decision
on the design variables can be obtained.
The optimized design variables of the hybrid system for two
scenarios are given in Table IV. In the first design scenario, is
assumed to be equal to
which means an indifference to cost
and availability. For the second design scenario, more weight
is given to the availability than cost (
,
) in
the
function. As expected, inspection of the Pareto

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
SHADMAND AND BALOG: MULTI-OBJECTIVE OPTIMIZATION AND DESIGN OF PHOTOVOLTAIC-WIND HYBRID SYSTEM

Fig. 8. Decision model without uncertainty.

OPTIMIZED HYBRID

TABLE IV
SYSTEM DESIGN VARIABLES
WITHOUT UNCERTAINTY

FOR

TWO SCENARIOS

Fig. 9. Procedure of optimizing the utility function with uncertainty on sola

insolation, wind speed, and demand data.

TABLE V
OPTIMIZED HYBRID SYSTEM DESIGN VARIABLES FOR TWO SCENARIOS WITH
UNCERTAINTY

Front of the optimal solutions of the system (Fig. 7), reveals that
the cost of the system increased significantly.
B. Decision With Uncertainty
A normal probability distribution is assumed over the attribute vectors. The solar insolation, wind speed, and demand
have the following normal distribution:

on availability such as two days without solar insolation is possible, but will result in a larger and more costly system. Thus,
the main contribution of this paper is to introduce a step by step
optimization design procedure for desired availability and cost.

(22)

VII. DISCUSSION

(23)

This section compares the proposed modeling approach of

hybrid system for DC microgrid based on high-temporal resolution data to conventional sizing approach [21], [32] based
on hourly National Solar Radiation Database (NSRDB) [33].
The histogram of power flow at the point of common coupling
(Fig. 1), based on a second data set, is used for comparison and
verifying the optimization technique. Fig. 10 illustrates the histogram of power flow of the conventional sizing using hourly
data, and Fig. 11 illustrates histogram of power flow of the proposed multi criteria design approach based on high-temporal
resolution data. The histogram of power flows are plotted for
the optimization without uncertainty and design scenario 1 of
Section VI.
As shown in Fig. 11, a significant unmet demand cant be
seen. As mentioned in previous section, an optimization design
based on collected data does not guarantee worst case system
availability. The unmet demand can be further reduced by adjusting the weight factor of availability and cost in the utility
function. When comparing the proposed modeling approach to
conventional sizing approach, a significant further increase in
the availability of the community DC microgrid as a load on the
utility grid can be seen while minimizing the system cost. Operationally this means the system becomes easier to dispatch and

Now their corresponding parameter in system modeling should

be modified as:
(24)
(25)
(26)
In order to optimize the utility function (21) under uncertainty, the expected value of the utility function must be determined. In this paper Monte Carlo is used for uncertainty propagation purpose. The proposed methodology is illustrated in
Fig. 9. The simulation is performed for 100 iterations; the results are given in Table V for the two design scenarios discussed
in the previous section. As shown, the cost of the system is increased significantly when considering uncertainties. Interestingly, the analysis revels similar availability comparing to results given in Table IV, but the cost of the system what is increased substantially.
It is important to note that an optimization design based on
collected data does not guarantee worst case system availability.
A worst case optimized system design specifying absolute limits

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
8

IEEE TRANSACTIONS ON SMART GRID

find the final optimal solution. The optimization is done for two
scenarios: with certainty and uncertainty on available resource.
The proposed methodology guarantees a reliable energy supply
with lowest investment.
Costs for PV-Wind based renewable energy have declined
steadily over the last 30 years as the technology has improved,
and the trend is expected to continue. Overall, the cost differences between clean and traditional energy are less extreme
than critics often imply and the differential continues to decline
steadily and the proposed sizing methodology is a push in this
direction by integrating the resources and load profiles with economics to provide an optimal solution which has lowest investment cost.
REFERENCES
Fig. 10. Histogram of power flow at the point of common coupling for 7 days
using the conventional sizing approach based on hourly data.

Fig. 11. Histogram of power flow at the PCC for 7 days using the proposed
modeling approach based on high-temporal resolution data.

control. Overlaying the capability of a smart grid communication system, the excess energy generated, shown as exporting
can be selectively injected into the utility grid by adjusting the
power point tracking of each power sources.
VIII. CONCLUSION
An important contribution of this paper is a sizing model that
considers desired availability and cost simultaneously. The proposed methodology avoids oversizing the system for high percentage of availability by using accurate and enhanced hightemporal resolution data. The main objective of this paper is to
provide a general model that quantifies the availability and cost
of hybrid renewable energy systems for smart DC microgrid.
In this paper high temporal resolution data for PV system is
used to optimize the hybrid system based on Multi-Objective
Genetic Algorithm (MOGA). MOGA is used to plot the Pareto
Front in order to visualize the problem for engineering tradeoff.
A utility function based on availability and cost formulated to

[1] W. Baochao, M. Sechilariu, and F. Locment, Intelligent DC microgrid

with smart grid communications: Control strategy consideration and
design, IEEE Trans. Smart Grid, vol. 3, pp. 21482156, 2012.
[2] Emerge Alliance, [Online]. Available: http://www.emergealliance.org/
[3] A. Mohamed and O. Mohammed, Connectivity of DC microgrids involving sustainable energy sources, in Proc. IEEE Industry Applicat.
Society Annu. Meeting (IAS), 2011, pp. 18.
[4] Z. Peng, W. Yang, X. Weidong, and L. Wenyuan, Reliability evaluation of grid-connected photovoltaic power systems, IEEE Trans. Sustainable Energy, vol. 3, pp. 379389, 2012.
[5] R. Majumder, A hybrid microgrid with DC connection at back to back
converters, IEEE Trans. Smart Grid, vol. 5, no. 1, pp. 251259, Jan.
2014.
[6] J. W. Kimball, B. T. Kuhn, and R. S. Balog, A system design approach
for unattended solar energy harvesting supply, IEEE Trans. Power
Electron., vol. 24, pp. 952962, Apr. 2009.
[7] A. Toliyat, A. Kwasinski, and F. M. Uriarte, Effects of high penetration levels of residential photovoltaic generation: Observations from
field data, in Proc. IEEE Int. Conf. Renewable Energy Research and
Applicat. (ICRERA), 2012, pp. 16.
[8] B. Fahimi, A. Kwasinski, A. Davoudi, R. S. Balog, and M. Kiani,
Charge it!, IEEE Power Energy Mag., vol. 9, pp. 5464, 2011.
[9] S. Xu, A. Q. Huang, S. Lukic, and M. E. Baran, On integration of
solid-state transformer with zonal DC microgrid, IEEE Trans. Smart
Grid, vol. 3, pp. 975985, 2012.
[10] S. Bhattacharya et al., Design and development of Generation-I silicon based Solid State Transformer, in Proc. IEEE Appl. Power Electronics Conf. and Exposition (APEC), 2010, pp. 16661673.
[11] A. Bidram and A. Davoudi, Hierarchical structure of microgrids control system, IEEE Trans. Smart Grid, vol. 3, pp. 19631976, 2012.
[12] J. A. Peas Lopes, C. L. Moreira, and A. G. Madureira, Defining control strategies for MicroGrids islanded operation, IEEE Trans. Power
Syst., vol. 21, pp. 916924, 2006.
[13] G. T. Heydt, The next generation of power distribution systems,
IEEE Trans. Smart Grid, vol. 1, pp. 225235, 2010.
[14] G. T. Heydt et al., Pricing and control in the next generation power
distribution system, IEEE Trans. Smart Grid, vol. 3, pp. 907914,
2012.
[15] M. B. Shadmand and R. S. Balog, Optimization of photovoltaic-wind
hybrid system for apartment complexes and other community living
environments by minimizing excess capacity, in Proc. IEEE Photovoltaic Specialists Conf. (PVSC), 2012, pp. 531536.
[16] C. M. Fonseca and P. J. Fleming, Genetic algorithms for multiobjective optimization: Formulationdiscussion and generalization, in Proc.
ICGA, 1993, pp. 416423.
[17] G. Ying, S. Lei, and Y. Pingjing, Study on multi-objective genetic
algorithm, in IEEE World Congr. Intelligent Control Automat., 2000,
vol. 1, pp. 646650.
[18] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, A fast and elitist
multiobjective genetic algorithm: NSGA-II, IEEE Trans. Evol.
Comput., vol. 6, pp. 182197, 2002.
[19] M. M. H. Bhuiyan and M. A. Asgar, Sizing of a stand-alone photovoltaic power system at Dhaka, Renewable Energy, vol. 28, pp.
929938, 2003.
[20] L. Fernandes and J. Figueiredo, An economic analysis of a power production hybrid system, in Proc. Int. Conf. Eur. Elect. Market, 2008,
pp. 16.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
SHADMAND AND BALOG: MULTI-OBJECTIVE OPTIMIZATION AND DESIGN OF PHOTOVOLTAIC-WIND HYBRID SYSTEM

[21] D. Abbes, A. Martinez, and G. Champenois, Eco-design optimisation of an autonomous hybrid wind??photovoltaic system with battery
storage, IET Renewable Power Generation, vol. 6, pp. 358371, 2012.
[22] W. D. Kellogg, M. H. Nehrir, G. Venkataramanan, and V. Gerez, Generation unit sizing and cost analysis for stand-alone wind, photovoltaic,
and hybrid wind/PV systems, IEEE Trans. Energy Conversion, vol.
13, pp. 7075, 1998.
[23] X. Sa, V. nchez, J. M. Ramirez, and G. Arriaga, Optimal sizing of a
hybrid renewable system, in Proc. IEEE Int. Conf. Indust. Technol.
(ICIT), 2010, pp. 949954.
[24] A. M. Leite da Silva et al., Distributed energy resources impact on
distribution system reliability under load transfer restrictions, IEEE
Trans. Smart Grid, vol. 3, pp. 20482055, 2012.
[25] S. V. Dhople and A. D. Dominguez-Garcia, Estimation of photovoltaic system reliability and performance metrics, IEEE Trans.
Power Syst., vol. 27, pp. 554563, 2012.
[26] The Home of Current Solar Projects at Texas A&M University SolarAgs [Online]. Available: http://solarags.tamu.edu/
[27] Zoning Issues and Wind Energy in Virginia US Department of Energy [Online]. Available: http://www.windpoweringamerica.gov/pdfs/
small_wind/va_zoning.pdf
[28] Wind Reports/Forecast W. Finder [Online]. Available: http://www.
windestimator.com
[29] D. E. Goldberg and J. H. Holland, Genetic algorithms and machine
learning, Mach. Learning, vol. 3, pp. 9599, 1988.
[30] P. C. Fishburn, Utility theory for decision making, DTIC Document,
1970.
[31] R. L. Keeney, Decisions with Multiple Objectives: Preferences and
Value Trade-Offs. Cambridge, U.K.: Cambridge Univ. Press, 1993.
[32] H. Beltran, E. Perez, N. Aparicio, and P. Rodriguez, Daily solar
energy estimation for minimizing energy storage requirements in PV
power plants, IEEE Trans. Sustain Ener., vol. 4, pp. 474481, 2013.
[33] NREL, National Solar Radiation Data Base (NSRDB) 19611990,
National Renewable Energy Laboratory (NREL).
Mohammad B. Shadmand (S09) received the B.S.
degree in electrical engineering from Qatar University, Doha, Qatar, in 2010. He received the M.S. degree in electrical engineering from Texas A&M University, College Station, TX, USA, in 2012, where he
is currently working toward the Ph.D. degree in electrical engineering.
He is a researcher with Renewable Energy &
Advanced Power Electronics Research Laboratory
since 2010. His research interests include FEA
of high-frequency magnetic components, model

predictive control, matrix converter, performance analysis of PV systems, and

switching power supplies.
Mr. Shadmand was awarded second place in the IEEE Industrial Application
Society Graduate Thesis Contest for his M.S. thesis in 2013.

Robert S. Balog (S92M96SM07) received the

B.S. degree in electrical engineering from Rutgers,
The State University of New Jersey, New Brunswick,
NJ, USA, and the M.S. and Ph.D. degrees in electrical
engineering from the University of Illinois at UrbanaChampaign, Urbana, IL, USA.
From 1996 to 1999, he was an Engineer with
Lutron Electronics, Coopersburg, PA, USA. From
2005 to 2006, he was a Researcher with the U.S.
Army Corp of Engineers, Engineering Research
and Development Center, Construction Engineering
Research Laboratory, Champaign, IL, USA. From 2006 to 2009, he was
a Senior Engineer at SolarBridge Technologies, Champaign, IL, USA. He
then joined Texas A&M University, College Station, TX, USA, where he
is currently an Assistant Professor with the Department of Electrical and
Computer Engineering. He is the holder of 17 issued and pending U.S. patents.
His current research interests include power converters and balance-of-systems
technologies for solar photovoltaic energy, particularly microinverters for ac
photovoltaic modules, and highly reliable electrical power and energy systems
including DC microgrids
Dr. Balog is a Registered Professional Engineer in the State of Illinois. He
received the IEEE Joseph J. Suozzi INTELEC Fellowship in Power Electronics
in 2001. He is a member of Eta Kappa Nu, Sigma Xi, National Society of Professional Engineers, American Solar Energy Society, and Solar Electric Power
Association. He was the recipient of the 2011 Rutgers College of Engineering
Distinguished Engineer Award.