Study of using solar energy system to supply a data center

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 633
STUDY OF USING SOLAR ENERGY SYSTEM TO SUPPLY A DATA CENTER
Mohammad Yousefi1, Pedram Ghalebani1
1Semnan University, Department of Electrical and Computer Engineering, Semnan, Iran.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Supplying the electricalenergyofadatacenteris
an important issue that plays a major role indesign, reliability
and operation of the data center. IT and datacenterindustries
will make very significant improvements over the next few
years which indicates an ever growing demand of energy for
data centers which can contribute to more consumption of
resources and more generation of carbon. More and more,
data centers are being asked not only to pay greater attention
to their energy consumption, but also to consider and
understand their relationship tothegreaterenvironmentsuch
as exploiting renewable energy. It implies that it is no longer
just about how energy is used, but also about how it is
produced. In order to use renewable energy asapowersupply,
its conformation to requirements of a data center should be
considered, which are desired reliability and power quality. In
this paper solar energy is investigated if it can meet the
requirements of data center power supply.
Key Words: Solar Cell, Data Center Power Supply,
Renewable Energy, Energy Demand, Solar Energy
1. INTRODUCTION
Providing the electrical energy needed for operation of the
data center is an ingenious and important approach.
Electrical system of the data center includes transformer,
generator, panel board, UPS and etc., and it can have
different layouts and configurations. It plays an important
role in determination of reliabilityofthedata centerandalso
affects the total cost of operation. If the harmonics ignored
properly, the grid connection can affect the power market
price [1]. They also can behave differently in islanding
detection in case of using synchronous generatorsisthegrid
connected system [2].
Nowadays the growing developments of data centers along
with other recent technologies such as plug-in electric
vehicles are imposing constantly increasing in demand of
energy to power plants [3]-[8], consideringthelimitationsof
fossil fuels such as the terminable sources used to generate
the electricity and the pollution caused by generation of
electricity from fossil fuels, the necessity of finding
alternative source of energy gains a great importanceand IT
industries are more convinced to exploit other sources of
energy, called renewable energy. In order to achieve this
goal different sources of energy, other than fossil fuel exist.
Solar energy, in the form of radiant light and heat from the
sun, has been harnessed and utilized by humans in different
ways using a range of ever-evolving technologies. Solar
radiation is considered as an available renewable energy on
earth. To harvest the solar energy, the most common way is
to use solar panels. The tilt angle of the solar panels plays an
essential rule in capturing maximum solar radiation.
Akhlaghi et al. [9] studied the sufficient number of optimal
tilt angle adjustment to maximize residential solar panels
yield. Every day earth receives a great and considerable
amount of energy, the amount of solar energy reaching the
surface of the planet is so vast that in one year it is about
twice as much as will ever be obtained from all of theEarth's
nonrenewable resources but only small portion of this
energy can be utilized. Developing accurate models can
effectively predict the availability of solar energy based on
weather forecasting [10]. In [11], a linear parametervarying
model is developed that iscapableofthetakinguncertainties
into account especially in climate related data. No need to
mention that, solar cells are subjected to severe weather
conditions and impact by foreign objects. Khalili et al. [12-
13] introduced a wave propagation basedhealthmonitoring
method using WSFE-based UEL which has a great potential
to be employed in order to detect and localize any possible
damage in solar cells and panels. Meshless methods such as
peridynamics [14-15] can also be used in health monitoring
of solar systems.
Solar power is obtained when the sunlight is converted into
electricity. It can be done, either directly usingphotovoltaics
(PV), or indirectly using concentrated solar power (CSP).
Solar energy is not available at night, and energy storage is
an important issue in order to provide the continuous
availability of energy. Because modern energy systems
usually assume continuous availability of energy, especially
for a data center, solar energy must be stored in different
forms. Gharghabi et al. [16-18] investigated the effect of the
various contingencies to happen on the structure of a data
center.
Since solar energy is not a permanent source of energy, and
providing a continuing and constant energy fora data center
is vital and indispensable, another reliable source of energy
is definitely needed. In this paper, the power grid is
considered to operate as a secondary source of energy, thus
it guarantees that energy is always delivered to the data
center, when the solar energy is not available foranyreason.
Although solar energy has been substantiallyoverlookedfor
years because of its unsecure and intermittent nature, the
current climate concerns and energy crisis urge
governments and academia to dedicate a large body of
research and investigation to enhancing solar energy

systems efficiency and reliability while these systems are
getting widespread applications in both industrial and
residential scales. One of the most important energy
consumers in industrial sector are refrigeration systems. S.
Aghniaey [19] in her paper analysed a solar-powered,
ammonia-water refrigeration cycle that is able to provide
temperatures as low as -40° C while using solar energy as
the power source. In another paper [20] she compared this
novel system with a conventional refrigerationsystemanda
12% increase in efficiency and 6% increase in COP was
reported for the novel solar-powered refrigeration system.
Solar systems are promising toolsincuttingcarbonfootprint
and reducing electricity demand and consumption in the
near future.
The objective of this paper is to study the combinationofthe
solar energy alongside the power grid to supply a data
center. In this paper, the considerations of the power supply
of a data center such as reliability and power continuity are
investigated. The radiation graph of sun in Tehran is taken
into account in simulationsandtheintermittent behaviourof
the solar energy is examined. The load model of the data
center is also implemented to present the variation of the
energy it needs. By modelling and implementing the power
supply system of the data center in MATLAB Simulink, the
variation of battery state of charge and outputpowerof each
power provider are shown and investigated. The simulation
shows the operations of the system for a week, when the
solar cells are starting to charge the batteries. In order to
increase the reliability of the system two batteries are used.
Finally it is investigated that solar energy alongside the
power grid can supply the data center.
2. BASIC WORKING PRINCIPLE
Figure1 showsaschematicofthesystemwithsolarpanel.
Fig. 1: Block diagram of the system
In this paper modeling of a grid connected solar cell
system is used to present power supply stability. The model
in Eq.1 ispresentedin[21]describeselectricalcharacteristics
of a solar cell.
(1)
In which Ir is irradiance; T is junction temperature; I_O is
saturation current; q is elementary charge of an electron;nis
an ideality factor; k is Boltzmann’sconstant;Np isthenumber
of cells in parallel and Ns is the number of cells in series.
By increase of irradiance the output power will increase.
But as it is presented higher temperature results lower
output power and vice versa. So as a result a place with good
irradiance and low temperature will boost the efficiency.
There are different methods usedforcontrollingthesolar
panels. Maximum Power Point Tracker (MPPT) is a method
that holds the output power of thePVpanels,ratherconstant.
Low cost MPPTs are designed to control the PV panels to
operate off-grid. For example MPPT model presented in [22]
does not need any external control signal to determine the
working point. In [23] a robust three phase model for MPPT
in PV panels connected to the grid is presented which by
using MPPT and a current controller regulates the current.
This model prevents the current overshoot. But in this paper
in order to control the PV panels the controller or MPPT is
not used, so that the effect of the variation of the sunlight in
day on the constant load can be illustrated and also power
continuity in a data center during the day can be
demonstrated.
Charge controller manages the charging and discharging
process of batteries. In fact it determines when a battery
must be charging or discharging over the load in order to
provide the power. For example in [24] a charge controller
for lead acid batteries by using PWM in PI algorithm for
standalone system is proposed. An improper algorithm of
charge controller can harm the lead acid batteries.
Scrutinizing the charge controller and analyzing it, is not the
objective of this paper.
In [25] by developing an intelligent charge controller
manufactured by Thompsonmicrocontroller,theefficiencyis
increased by using a Microcontroller. The charge controller
used in this paper is an intelligent charge controller which is
capable of programming and sending controlling signals to
start and stop, charging and discharging of batteries and
power grid. The logic statements used for charge controller
are as follows:
(2
)
A data center needs a huge amount of energy for IT
devices and cooling requirements. For example using
Combined Cooling, Heating and Power (CHP) causes a great
heat generation. A discussion has been made in [23] about
the application of the coolingsystem and IT devices together.
It must be taken into account that a data center consumes a
fairly constant energy and only during the night due to the
( ) 1 ( 1 99). 1
( ) 2 ( 2 99). 1. 2
( ) 1. 2.
( ) 1 ( 1 35). 1 ( 1 35).( 2 35).( 0)
( ) 2 ( 2 35). 2. 1
( ) 1. 2
a Chg Bat Dis
b Chg Bat Chg Dis
c ChGd Chg Chg DsGd
d Dis Bat Chg Bat Bat PVCnt
e Dis Bat Chg Dis
f DsGd Dis Dis
 
 

     
 

exp 1
100
PVr
PV P SC P o
S
qVI
I N I N I
nkTN
   
     
    

decrease of temperature, power consumption of the cooling
system reduces.
3. SIMULATION RESULTS
The data center used for simulationsneeds2MWelectric
power. The power demand of adatacenterisratherconstant,
but %20 decrease of power demand is taken into account at
nights within 12 AM and 5 AM due to reduction of cooling
system power demand andfeweroperationofcomputational
operation of data center. The electric power needed by data
center is provided by solar panels and power grid
simultaneously.
The system considered for simulation consists of two
chargeable batteries, 10000 and 8000 Ah, and also a huge
battery with %50 statement of charge which represents the
power grid and all of them can be charged by solar panels.
Voltage of batteries is 48V and voltage of power grid is 220V.
Single phase model of the system is implemented and the
simulations are based onthepowerproducedandconsumed.
Battery Charge Controller controls the process charging and
discharging of the batteries based on logical directives. In
order to convert DC to AC or vice versa, inverter or converter
with efficiency of 0.95 is used.
The irradiance graph of the sun in Tehran is shown in Fig.
2. Since Iran is located in a proper position that absorbs
ample amount ofsunlight, and the solar power in someareas
exceeds 1000 w/m^2, this country has the great potential of
using solar power.
Fig. 2: Irradiance of the sun in Tehran in summer and
winter
Fig. 2 shows the solar power of the sun in the Tehran in a
day, in summer and winter. As it can be seen these two
figures are different, that affects the output power of the PV
panels. The data center is located in Tehran, whose
geographical coordination is shown in Table 1:
Table 1: Geographical coordination of Tehran
N E
51°25'23.00" 35°41'46.00"
The amount of sunlight that PV panels absorb,
determines the output power of the PV panels. Based on
simulations results and also Eq. 1 output power of the PV
panels which consist of 10000 panels with 72 cells in series,
reaches 1.2 MW at maximum in midday. Anyway it is lower
than the power needed by data center most of the time. This
fact is illustrated in Fig. 3.
Fig. 3: Power generated by PV panels
During operation of PV panels, the batteries are charging
or discharging. The first battery works as the primary
battery, and the second battery is known as backup to
enhance the reliability of the system. This battery provides a
portion of the load, when the primary battery cannot supply
the load. State of charge (SOC) of two batteries is shown in
Fig. 4.
As it can be seen in Fig. 4 based on the algorithm of the
charge controller, the SOC of the batteries varybetween33%
and %99. And also while the primary is charging the backup
battery is discharging and supplying the power. In other
words it can be said that the backup battery works as the
complement of the primary battery.
Providing a continuous power source fordatacentersisa
delicate and important approach. Thus in any data centers
power sourcesotherthanpowergridsuchasuninterruptable
power sources (UPS) and generators, are exploited toensure
the constant source of energy for data center. It must be
noted that based on simulation results shown in Fig. 4,
Discharging process of the batteries which provides the
power needed by the data center, can also act as a power
backup supply when power grid fails and is not available.
This fact enhances the availability of the power. In [26] it has
been discussed that wind energy can be incorporated to
supply the energy to feed the data center. It was also
discussed [27] its exploitation in data modern electric ships.

(a)
(b)
Fig. 4: Variation of the SOC of (a) primary battery (b) backup
battery
In the configuration of the power system considered in
this paper the power grid is capable of providing the 2 MW
load of the data center and it undertakes the load whenever
the sun is not available or batteries are charging and cannot
providethe needed powertorunthedatacenter.Thusduring
the intervals which the batteries are charging,thepowergrid
supplies the total load. Figure 5 depicts the power provided
by the power grid.
Fig. 5: Power provided by grid
4. CONCLUSION
Based on simulationscarriedoutfollowingresultscanbe
concluded:
1) In order to enhance the reliability of the power
supply, two or more batteries are needed. But they can be in
smaller size. Thus in case of failure of one of the batteries
continuous operation of the solar energy supply is
guaranteed. It also gives more flexibilitytohavecontrol over
operation of the power supply.
2) It is shown that in the simulated configuration over a
week solar energy provides %18 of the total energy needed
by operation of the data center. This fact leads to reduction
of the cost of the operation. Consequently, it can justify the
high amount of capital the solar energy system needs to be
installed.
3) Using solar power in a data center in addition to
providing power and reducing the air pollution,canincrease
the sustainability of power system. If the power is supposed
to be generated locally for a limited period, for example
when power grid is not available, solar power increases the
reliability of power system alongside the UPS and
generators.
REFERENCES
[1] H. Norouzi, S. Abedi, R. Jamalzadeh, M. Ghiasi Rad,
S.H. Hosseinian, “Modeling and investigation of harmonic
losses in optimal power flow and power system locational
marginal pricing.” ENERGY Journal, Science Direct, P-68,
2014. 140e147.
[2] H. Jouybari Moghaddam, S.H. Hosseinian, B. Vahidi, M.
Ghiasi Rad, “Smart Control mode selection for proper
operation of synchronous distributed generators”, 2nd
Iranian Conference onSmartGrids,Iran(IslamicRepublicof)
, 2012.
[3] Babaei, Mehdi, Maziar Babaei, and Mohsen Niasati.
"Parametric analysis of overvoltages caused by back-
flashover in “Siah-bishe” 400kV GIS substation." Electric
Power and Energy Conversion Systems (EPECS), 2013 3rd
International Conference on. IEEE, 2013.
[4] S Jafarishiadeh, M Ardebili, A Nazari Marashi,
“Investigation of pole and slot numbers in axial-flux pm bldc
motors with single-layer windings for electricvehicles,”24th
Iranian Conference on Electrical Engineering (ICEE), pp.
1444-1448, 2016.
[5] Babaei, Maziar, Mehdi Babaei, and Ghasem Nourirad.
"Analysis of influential factors in determining Very Fast
Transient Overvoltages of GIS substations." Power
Engineering and Optimization Conference (PEOCO), 2014
IEEE 8th International. IEEE, 2014.
[6] S. Akhlaghi, A. Akhlaghi, and A. A. Ghadimi, “Performance
Analysis of the Slip ModeFrequencyShiftIslandingDetection
Method under different Inverter Interface Control Strategy,”
In Power and Energy Conference at Illinois (PECI), 2016.
Illinois, USA, IEEE, Feb, 2016.

[7] R. Jalilzadeh Hamidi and H. Livani, “Myopic real-time
decentralizedchargingmanagementofplug-inhybridelectric
vehicles,” Electric Power System Research, Vol. 143, pp. 522-
543, 2017.
[8] Seyyedmahdi Jafari Shiadeh, Mohammad Ardebili, and
Parvin Moamaei, “Three-dimensional finite-element-model
investigation of axial-flux PM BLDC machines with similar
pole and slot combination for electric vehicles”, In:
Proceedings of Power and Energy Conference, Illinois, pp.1–
4, 2015.
[9] S. Akhlaghi, M. Sarailoo, M. Rezaeiahari, H. A. Sangrody,
“Study of Sufficient Number ofOptimalTiltAngleAdjustment
to Maximize Residential Solar Panels Yield” IEEE Power and
Energy Conference at Illinois (PECI), 2017. Illinois, USA, Feb,
2017.
[10] Farshid Abbasi, Javad Mohammadpour, Roland Tóth,
and Nader Meskin. "A support vectormachine-basedmethod
for LPV-ARX identification with noisy scheduling
parameters," In Proceeding of European Control Conference
(ECC), pp. 370-375. 2014.
[11] Farshid Abbasi, Javad Mohammadpour, Roland Tóth,
and Nader Meskin. "A Bayesian approach for model
identification of LPV systems with uncertain scheduling
variables," In 54th Annual IEEE Conference on Decision and
Control (CDC), pp. 789-794, 2015.
[12] Khalili A., Jha R., SamaratungaD.“SpectrallyFormulated
User-Defined Element in Conventional Finite Element
Environment for Wave Motion Analysis in 2-D Composite
Structures” European Journal of Computational Mechanics,
November 11,2016,DOI:10.1080/17797179.2016.1253364
[13] Khalili A., Samaratunga D., Jha R., Lacy T. E.,
Gopalakrishnan S. “WSFE-based User-Defined Elements in
ABAQUS for Modeling 2D Laminated Composites with
Complex Features” 30th ASC TechnicalConference,Michigan
State University, EastLansing,MI,US,28-30September2015.
[14] Yaghoobi,Amin,andMiG.Chorzepa."Meshlessmodeling
framework for fiber reinforced concrete structures."
Computers & Structures 161 (2015): 43-54.
[15] Yaghoobi, Amin, and Mi G. Chorzepa. "Fracture analysis
of fiber reinforced concrete structures in the micropolar
peridynamic analysis framework." Engineering Fracture
Mechanics 169 (2017): 238-250.
[16] P. Gharghabi, J. Lee, M. S. Mazzola, and T. E. Lacy Jr.,
“Development of an Experimental Setup to Analyze
Carbon/Epoxy Composite Subjected to Current Impulses,”
Am. Soc. Compos. Thirty-First Tech. Conf., 2016.
[17] P. Gharghabi, P. Dordizadeh B., and K. Niayesh, “Impact
of Metal Thickness and Field-Shaper on the Time-Variant
Processes during Impulse Electromagnetic Forming in
Tubular Geometries,” Journal of the Korean Physical Society,
vol. 59, no. 61, 2011, p. 3560-3566.
[18] P. Dordizadeh-Basirabad, P. Gharghabi, and K. Niayesh,
“Dynamic Analysis of a Fast-acting Circuit Breaker
(Thompswon) DriveMechanism,”J.KoreanPhys.Soc.,vol.59,
no. 61, p. 3547, 2011.
[19] Aghniaey, Sama, and Seyed Mohammad Seyed
Mahmoudi. "Exergy analysis of a novel absorption
refrigeration cycle with expander and compressor." Indian
Journal of Scientific Research (2014).
[20] Aghniaey, Sama, S. Mohammad S. Mahmoudi, and Vahid
Khalilzad-Sharghi. "A comparison between the novel
absorption refrigeration cycle and the conventional
ammonia-waterabsorptionrefrigerationcycle."International
Conference on Heat Transfer, Fluid Mechanics and
Thermodynamics, 2014.
[21] Affordable Solar Group, LLC, “SMA Sunny Boy SB 2500
InverterwithESSSwitch,”May2008,http://www.affordable-
solar.com/sunny-boy-2500-inverter-export.htm
[22] Zhongyi He et al., "Low cost MPPT controller for off grid
solar Applications", IEEE ICMS, 2010.
[23] Il-Song Kim, "System dynamic model and charging
control of lead-acid battery for stand-alone solar PVsystem,"
Elsevier, Solar Energy 81, 2007, pp. 405–414.
[24] B.J. Huang, et al., "System dynamic model and charging
control of lead-acid battery for stand-alone solar PVsystem",
Elsevier, Solar Energy 84, 2010, pp. 822–830.
[25] H. Mashelini and X. F. CARELSE, "Microcontroller-Based
Charge Controller for Stand-Alone Photovoltaic Systems,"
Elsevier, Solar Energy Vol. 61, No. 4, 1997, pp. 225–230.
[26] M. Ghiasi Rad, M. Rahmani, P. Gharghabi, A. Zoghi and S.
H. Hosseinian, "Scheduling a Wind Hydro-Pumped-Storage
Unit Considering the Economical Optimization," American
Journal of Electrical and Electronic Engineering, vol. 5, no. 1,
pp. 16-22, 2017.
[27] Babaei, Maziar, et al. "Developmentofahybridmodelfor
shipboard power systems." Electric Ship Technologies
Symposium (ESTS), 2015 IEEE. IEEE, 2015.

Study of using solar energy system to supply a data center

More Related Content

What's hot (20)

Similar to Study of using solar energy system to supply a data center (20)

More from IRJET Journal (20)

Recently uploaded (20)

Study of using solar energy system to supply a data center