An Optimized Combination of A Large Grid Connected PV System Along With Battery Cells and A Diesel Generator
An Optimized Combination of A Large Grid Connected PV System Along With Battery Cells and A Diesel Generator
An Optimized Combination of A Large Grid Connected PV System Along With Battery Cells and A Diesel Generator
fferdowsi@fsu.edu
Nomenclature
B
()
Maximum energy produced by the diesel generator during the time study (KWh)
Operating and maintenance costs related to the generator power capacity ($/KW)
Operating and maintenance costs related to the battery energy capability ($/KWh)
Operating and maintenance costs related to the battery power capability ($/KW)
(i)
(i)
(i)
(i)
1. Introduction
Photovoltaic Systems (PV) are used to convert solar energy to electricity through solar
panels. Due to the environmental concerns, electricity productions based on green/clean fuel
sources (such as sun) have become increasingly important. In recent years, grid- connected
photovoltaic power plants of several megawatts have also seen commercial operations in many
developed countries. In addition, PV systems have a number of technical and economical
benefits, such as loss reduction and voltage profile improvement [1]. Unfortunately,
instantaneous changes in the irradiance reaching the PV arrays and unpredictable nature of the
solar power lead to a corresponding change in their output power. In most studies, the time
frame for fluctuations in irradiance is in the order of few minutes especially suitable for systems
with megawatts rating [2]. For large PV systems, the output power can change considerably in
several minutes time frame, but for a number of small systems distributed over a large area,
fluctuations are much less [3]. The frequency of these fluctuations are dictated by type and size
of moving clouds [4], location and penetration level of the PV systems. As far as the penetration
level is increased, the negative impacts of PV power fluctuations on some aspects of the
network operation such as stability of the system, high variations in node voltages and
scheduling of generation units would be increased [5-7].
The main target for utility is to have smooth output power of the PV system or to produce a
certain limit of power from the PV systems. By applying certain technologies and methods,
one could achieve to this task but, due to high costs of methods and the energy generated from
the PV system, it is worthwhile to have economical analysis and then choose and implement
suitable technology to smooth the power fluctuations.
In this matter, the use of energy storage systems (ESS) could be an effective method to
mitigate these negative impacts [8] and [9]. In selecting a proper storage system one has to
determine its applications. When a relatively high amount of power must be supplied in a very
short period of time which is called power application; capacitors, super conducting magnetic
storage systems and batteries with rapid reactions are vital [10]. But for energy application
such as storing energy during the peak generation period and injecting this energy during the
peak loading period, providing huge amount of energy during a certain period of time (hours
up to days), hydro pump stations, hydrogen based storage systems and batteries usually are
used. These mentioned technologies have different characteristics in performance. Hydro pump
stations are not suitable for using with large PV systems, because they need large flat areas.
Some of these technologies such as superconducting magnetic storage systems and ultra
3
capacitors have very low energy densities and thus cannot be used alone in power applications
[11, 12].
Categorizing batteries as the common member of both mentioned applications indicates that
they can meet the required power capacity and necessary amount of energy. Moreover, batteries
have quick response in both charge and discharge processes, so they are one of the most suitable
technologies to reduce the output power fluctuations of PV systems.
When energy storage is considered, the storage reserve capacity and battery system
capability has to be determined in terms of the days of battery back- up [13]. One approach for
determination of battery capacity for a given system is to estimate it by multiplying the number
of days that the system load could be supplied by the battery system with no power available
from PV system. This method does not reflect the continuous variation of system generation
and load on the storage capability. Also, it cannot be applied to small stand- alone power
systems, because the stored energy is associated with only the generation shortage due to the
solar resources unavailability. There are other methods that can be used to reduce power
fluctuations. Installing a dump load will dissipate the excess PV power. The dump load consists
of a resistance and a controller which controls the power flow through the load. In this method,
dumping part of the generated power considerably will lead to loss of revenues. A third method
is to curtail the power by operating below the maximum power point (MPP). This method only
requires modification of the control strategy of the power conditioning unit but still the
curtailed power shows considerable loss in revenues for all limits of power fluctuations.
This paper presents four different cases to implement the grid connected photovoltaic system
using: 1) suitable battery storage, 2) battery storage system with curtailment of power by
operating below the MPP, 3) battery storage and a back up diesel generator, and 4) combination
of battery storage and a back up diesel generator with the curtailment of power by operating
below the MPP. In all cases, the proper linear programming (LP) optimization problem with
the corresponding constraints is applied. The problem is modeled in the LINGO software. The
results obtained from solving the problem are compared in terms of costs and net benefit until
the most efficient and economical option is obtained.
Fig. 1 shows a typical power waveform for 24 hours period of time based on real temperature
and irradiance data obtained from a metrology station in southwestern United States with 10
minutes resolution [14]. As the figure shows, there are some times in which the generated
power changes considerably and the amount of these changes may be over 50%. Since these
amount of changes affects the power network adversely as mentioned, power fluctuations
reduction should be considered vital.
(1.5% of system capacity) per each 10 minutes time interval for all cases in this paper. Based
on this demand, one has to use a suitable battery with the PV in order to satisfy the required
operating conditions of the utility. In Ref. [15] mathematical analysis have been done in gridconnected photovoltaic/battery system shown in Fig. 2 to seek for the most optimum battery to
control fluctuations.
.
Four ordinary batteries, Lead- Acid, Ni- Cd, NaS and Li- Ion, were studied technically in
terms of their performance and then linear programming was applied as an optimization process
to obtain the most economical option. The specifications of the four types of batteries are given
in Table (1) [11], [16, 17]. Minimizing the cost function with the corresponding constraint used
in Ref. [15] showed that NaS battery has the lowest cost for 18 years (period of study)
assumption.
Therefore, NaS battery is used for all cases in this paper as the most optimum
battery .Table 2 represents the results of the battery optimization problem for the fluctuation
constraint given in [15] .
Na-S
Li-ion
Ni-Cd
Advantages
Disadvantages
Production cost
Harmful for
environment
300
1000
Production cost,
required special
charging circuit
1300
150
170
500
390
CP
CE
600
SP
SE
OP
OE
%85
%85
$
TB years
10
13
1.5
1.7
3.9
30
15
1.5
%10
LA
5.5
Imposition of
further cost %
(caused by
using batteries)
-29.4%
Na-S
6.2
-20.5%*
3604.084
9872.172
Ni-Cd
4.3
-44.8%
4192.814
8672.548
Li-ion
5.9
-24.3%
3600.104
9897.315
Net benefit
(107) $
Battery Power
Rating (KW)
Battery Energy
Rating (Kwh)
3963.031
9080.169
are allowed to inject variable generated power into the grid as long as the operating limits of
the network are not violated. To control the output of the PV system, a BS system is included.
To examine the economical aspect of installing BS system one could estimate the maximum
revenues that the system owner can obtain. An optimization algorithm which is brought up in
Ref. [11] is applied with slight changes to calculate this revenues as well as battery power and
energy ratings. The input to the optimization process are time series of PV power with no
battery included. The objective function (B) with used battery expenses and are given as
follows. It should be noticed that the selling price of energy is assumed to be 0.45 $/KWh and
the time study is 18 years:
7
= =1 (0.45
= =1
(1+)(1) +
= =1
(1+)(1) +
() = () + ()
| () ( 1)| 150
(1+) 1
(1+)
(1+) 1
(1+)
() = ( 1) 6 ( 1)
( = 1) = ( )
| ()|
()) (1+)
, , , 0
=1
(1+)
=1
(1+)
(1)
(2)
(3)
(4)
(5)
> 1
(6)
0 1
( %) ()
0 (, , )
, ,
(7)
(8)
(9)
(10)
(11)
The first term in (1) calculates the revenues gained from selling energy to the grid. The
second and third terms represent the net costs related to the battery power and energy ratings.
In (2) and (3), the first term estimates the worth of capital cost related to battery power and
energy ratings. The second term in both equations represents the worth of the annual operating
and maintenance costs and the third term considers the salvage value at the end of battery
lifetime. In (4) the injected power to the grid is calculated. Equation (5) ensures that power
fluctuations are in the desired range. In (7), the initial state charge of battery is determined.
Other constraints are given in (8) (11).
Fig. 3: Schematic of PV system with battery and diesel generator along with the curtailment
based on operating below the MPP
Operating below the MPP control method acts as the BS system with only having charging
process. The use of this method makes the system to operate in low level production.
Integration of an external power supply such as diesel generator would solve this low level
operation problem. Using the diesel generator along with the battery would reduce the size of
battery (cost of battery) and increase its lifetime, because it responds very fast and has less
expenses comparing to the battery.
The following relations with equations (12)-(16), represent the optimization problem for the
proposed model.
= =1 (0.45
1
1
6
()) (1+)
6 () [1 1 + 2 2 + 3 3 ]
(12)
= =1
(1+)(1) +
(1+) 1
(1+)
() = () + () () + ()
1
0 6 () _
=1
(1+)
, , , , , _ 0
(13)
(14)
(15)
(16)
(12) and (13) determines the system cost functions similar to equations (1) and (2) in the
previous case. Equation (14) calculates the injected power to the grid in each section of the
9
time. In (15) the operation of the diesel generator is limited by applying a specific constraint
on its required fuel during the period of study. Finally, the last equation (16) is the nonnegativity constraint.
4. Numerical Results
In this section, the linear programming based optimization problem is solved for the
following cases and the results are shown and compared with each other. The case study is a
10 MW PV power plant that forms an area of about 310 square meters of land. The weather
information such as temperature and irradiance is obtained with a resolution of ten minutes in
an area with longitude and latitude 33.96 N and 118.42 W with an elevation of 32 meters
above sea level in the United States [14]. The generated power from PV panels in each 10
minutes frame is estimated by means of the model described in [18]. Furthermore, the accuracy
of this mathematical model has been verified for some weather data via the system that has
been simulated in MATLAB/SIMULINK.
The study period is considered 18 years and it is assumed that changes in the climatic
conditions for every year are almost the same. Thus, temperature and radiation are sampled
during a year for each 10 minutes. Regardless of periods in which the radiation is less than 2
W/m2, the number of annual output power samples is about 26000. Since the AC power should
be injected to the grid, the inverter efficiency should also be taken into account. It is assumed
the inverters efficiency is constant and equal to 0.9. It should be also noticed that the
permissible power fluctuation range is considered 150 KW/10 min for all cases.
10
Ppv
PG
7:10 AM
8:20 AM
9:30
10:40
11:50
13:00
14:10
15:20
16:30
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
Power (KW)
8000
7000
6000
5000
4000
3000
2000
1000
0
-1000
8000
6000
4000
Pb
2000
Eb
0
-2000
-4000
7:10 AM
8:20 AM
9:30
10:40
11:50
13:00
14:10
15:20
16:30
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
considering the parameters related to the diesel generator). In this case, we need a NaS battery
with 2354.679 KW of power rating and 4507.632 KWh of energy rating. The maximum
curtailed power is also equal to 1497.947 KW.
Figs. 6 and 7 show the PV power before and after the reduction of fluctuations, battery
8000
7000
6000
5000
4000
3000
2000
1000
0
-1000
Ppv
PG
7:10 AM
8:20 AM
9:30
10:40
11:50
13:00
14:10
15:20
16:30
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
Power (KW)
4000
3000
2000
Pb
1000
Eb
0
-1000
-2000
7:10 AM
8:30
9:50
11:10
12:30
13:50
15:10
16:30
7:30
8:50
10:10
11:30
12:50
14:10
15:30
16:50
7:40
9:00 AM
10:20
11:40
13:00
14:20
15:40
17:00
5000
-3000
Pc
the battery life time and reduce the battery size and cost. Characteristics of the employed diesel
generator are given in Table (3) [19]. According to the lifetime of the generator, it should be
replaced 3 times after initial installation during the period of study. Moreover, in the
optimization problem, the fuel consumption of the diesel generator has been limited to 50000
liter per year (equation 15).
Table 3
($)
280
($)
28
($)
80
(hr)
20000
()
0.5
($)
($)
6000
T (years)
18
1.1
In this case, the battery power rating is reduced to 1864.083 KW and its energy rating is also
reduced to 3013.206 KWh. We will need a diesel generator with 1662.02 KW of power
capacity. Figs. 8 and 9 show the output power, battery power and energy ratings and also diesel
8000
7000
6000
5000
4000
3000
2000
1000
0
-1000
Ppv
PG
7:10 AM
8:20 AM
9:30
10:40
11:50
13:00
14:10
15:20
16:30
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
7:20
8:30
9:40
10:50
12:00 PM
13:10
14:20
15:30
16:40
Power (KW)
power profiles.
Fig. 8: Output power before and after applying the controlling method
13
3000
2000
1000
Pb
Eb
-1000
-2000
-3000
7:10 AM
8:30
9:50
11:10
12:30
13:50
15:10
16:30
7:30
8:50
10:10
11:30
12:50
14:10
15:30
16:50
7:40
9:00 AM
10:20
11:40
13:00
14:20
15:40
17:00
4000
Pd
Fig. 9: Battery power, Battery energy and diesel generator power profiles
8000
7000
6000
5000
4000
3000
2000
1000
0
-1000
Ppv
PG
7:10 AM
8:30
9:50
11:10
12:30
13:50
15:10
16:30
7:30
8:50
10:10
11:30
12:50
14:10
15:30
16:50
7:40
9:00 AM
10:20
11:40
13:00
14:20
15:40
17:00
Power (KW)
the output power, battery power and energy, curtailed power and diesel power profiles.
14
3000
2000
Pb
1000
Eb
Pc
-1000
-2000
7:10 AM
8:40
10:10
11:40
13:10
14:40
16:10
7:20
8:50
10:20
11:50
13:20
14:50
16:20
7:20
8:50
10:20
11:50
13:20
14:50
16:20
4000
Pd
Fig. 11: Battery power & energy, curtailed power and diesel power profiles
Numerical results corresponding to each method are summarized in Table (4) for making a
better evaluation of the four proposed methods.
Table 4: Numerical results of solving the LP optimization problem for the four proposed
methods
Using NaS
Battery
Battery +
Curtailment
Battery +
Diesel
6.39
6.81
6.88
Battery +
Diesel +
Curtailment
6.99
-14.88%
-12.82%
-11.75%
-10.38%*
1497.947
2702.266
2992.037
2354.679
1864.083
1335.165
6012.457
4507.632
3013.206
2859.388
1662.020
1532.335
4. Conclusion
High penetration of PV system into electric network could be detrimental to overall system
performance, because of having fluctuations in their output power. In this study, a hybrid
method
consisting suitable storage battery (NAS), backup diesel generator along with the
curtailment of the generated power by operating below MPP was proposed to smooth the output
power of the grid-interactive PV system. We focused on applying the most economical method
to this system in order to decline the output power fluctuation velocity of the PVs connected to
the grid. Four different cases were considered and in each case, accurate calculation and
15
analysis of different aspects were included to have minimum cost and maximum benefits in
terms of reducing battery size and expenses and the curtailed power of operation under MPP.
Comparison of the results for different cases showed that in the area of study applying the
combination of battery, power curtailment and diesel generator is the best method in terms of
both technically and economically. It is clear that such studies can be performed in other areas,
with regard to their whether circumstances.
References
[1] B.H. Chowdhury, A.W. Sawab, Evaluating the value of distributed photovoltaic
generations in radial distribution systems, Energy Conversion, IEEE Transactions on, 11
(1996) 595-600.
[2] G. Vijayakumar, M. Kummert, S.A. Klein, W.A. Beckman, Analysis of short-term solar
radiation data, Solar Energy, 79 (2005) 495-504.
[3] Solar Server, The worlds largest photovoltaic solar power plant is in Pocking, [Online].
Available: http://www.solarserver.com/ solarmagazin/anlage_0606_e.html, Oct. 2012
[4] E.C. Kern, E.M. Gulachenski, G.A. Kern, Cloud Effects on Distributed Photovoltaic
Generation: Slow Transients at the Gardner, Massachusetts Photovoltaic Experiment, Power
Engineering Review, IEEE, 9 (1989) 43-44.
[5] B.H. Chowdhury, Effect of central station photovoltaic plant on power system security,
in: Photovoltaic Specialists Conference, 1990., Conference Record of the Twenty First IEEE,
1990, pp. 831-835 vol.832.
[6] N. Okada, K. Takigawa, A voltage regulation method for dispersed grid-connected PV
systems under high-density connection, Solar Energy Materials and Solar Cells, 75 (2003)
637-646.
[7] T. Yun Tiam, D.S. Kirschen, Impact on the Power System of a Large Penetration of
Photovoltaic Generation, in: Power Engineering Society General Meeting, 2007. IEEE,
2007, pp. 1-8.
[8] D. Mears, H. Gotschall, H. Kamath, E.P.R. Institute, U.S.D.o. Energy, T. Insights, E.P.
Corporation, EPRI-DOE Handbook of Energy Storage for Transmission and Distribution
Applications, EPRI, 2003.
[9] J.J. Shea, Distributed power generation planning and evaluation [Book Review],
Electrical Insulation Magazine, IEEE, 17 (2001) 67-68.
[10] M.E. Glavin, W.G. Hurley, Ultracapacitor/ battery hybrid for solar energy storage, in:
Universities Power Engineering Conference, 2007. UPEC 2007. 42nd International, 2007, pp.
791-795.
[11] W.A. Omran, M. Kazerani, M.M.A. Salama, Investigation of Methods for Reduction of
Power Fluctuations Generated From Large Grid-Connected Photovoltaic Systems, Energy
Conversion, IEEE Transactions on, 26 (2011) 318-327.
[12] E. Spahic, G. Balzer, B. Hellmich, W. Munch, Wind Energy Storages - Possibilities, in:
Power Tech, 2007 IEEE Lausanne, 2007, pp. 615-620.
[13] R.A. Messenger, J. Ventre, Photovoltaic systems engineering, CRC Press/Taylor &
Francis, 2010.
[14] NREL Solar Radiation Research Laboratory (BMS). [Online]. Available:
http://www.nrel.gov/midc/srrl_bms/, Feb. 2013.
16
[15] F. Ferdowsi, A.S. Yazdankhah, B. Abbasi, Declining power fluctuation velocity in large
PV systems by optimal battery selection, in: Environment and Electrical Engineering
(EEEIC), 2012 11th International Conference on, 2012, pp. 983-988.
[16] Electricity Storage Association. Storage Technologies, [Online]. Available:
http://www.electricitystorage.og/site/home/, Oct. 2012.
[17] P. Poonpun, W.T. Jewell, Analysis of the Cost per Kilowatt Hour to Store Electricity,
Energy Conversion, IEEE Transactions on, 23 (2008) 529-534.
[18] M.E. Ropp, M. Begovic, A. Rohatgi, Determination of the curvature derating factor for
the Georgia Tech Aquatic Center photovoltaic array, in: Photovoltaic Specialists Conference,
1997., Conference Record of the Twenty-Sixth IEEE, 1997, pp. 1297-1300.
[19] HOMER Energy LLC., A users guide, The Micro-power Optimization Model,
Copyright 2012.
[20] Sayyid Mohssen Sajjadi, Ahmad Sadeghi Yazdankhah, Farzad Ferdowsi, A new
gumption approach for economic dispatch problem with losses effect based on valve-point
active power, Electric Power Systems Research, Volume 92, November 2012, Pages 81-86,
ISSN 0378-7796
[21] F. Ferdowsi, C. S. Edrington and T. El-mezyani, "Real-time stability assessment
utilizing non-linear time series analysis," North American Power Symposium (NAPS), 2015,
Charlotte, NC, 2015, pp. 1-6.
[22] F. Ferdowsi, A. Sadeghi Yazdankhah and H. , "A combinative method to control output
power fluctuations of large grid-connected photovoltaic systems," Environment and
Electrical Engineering (EEEIC), 2014 14th International Conference on, Krakow, 2014, pp.
260-264.
[23] F. Ferdowsi, C. S. Edrington and T. El-mezyani, "Small Signal Stability Assessment in
Power Electronic-Based Components," 2015 FREEDM Systems Center Annual Industry
Review and Conference, Raleigh, NC, Jan 2015.
[24] S. Paran, C. S. Edrington and B. Vural, "Investigation of HIL interfaces in nonlinear
load studies," North American Power Symposium (NAPS), 2012, Champaign, IL, 2012, pp. 16.
[25] O. Naeckel et al., "Power Hardware-in-the-Loop Testing of an Air Coil Superconducting
Fault Current Limiter Demonstrator," in IEEE Transactions on Applied Superconductivity,
vol. 25, no. 3, pp. 1-7, June 2015.
[26] A. Solouk, M. Shahbakhti, M. J. Mahjoob, " Energy Management and Control of a
Hybrid Electric Vehicle with an Integrated Low Temperature Combustion (LTC) Engine ",
ASME Dynamic Systems and Control Conference, 10 pages, Oct. 22-24, 2014, San Antonio,
Texas, USA.
[27] Saberi, H., Sabahi, M., Sharifian, M. B., & Feyzi, M. (2014). Improved sensorless direct
torque control method using adaptive flux observer. Power Electronics, IET, 7(7), 1675-1684.
[28] Saberi, H., & Sharifian, M. B. B. (2012, October). An improved direct torque control
using fuzzy logic controllers and adaptive observer. In Computer and Knowledge
Engineering (ICCKE), 2012 2nd International eConference on (pp. 83-88). IEEE.
[29] Saberi, H., Sharifian, M. B. B., & Amiri, M. (2012, May). Performance improvement of
direct torque control drives in low speed region. In Electrical Engineering (ICEE), 2012 20th
Iranian Conference on (pp. 505-510). IEEE.
[30] Rakhshan, M., Vafamand, N., Shasadeghi, M., Dabbaghjamanesh, M., & Moeini, A.
(2016). Design of networked polynomial control systems with random delays: sum of squares
approach. International Journal of Automation and Control, 10(1), 73-86.
17
18