Energy 135 (2017) 413e429

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Techno-economics and environmental analysis of energy storage for a

student residence under a South African time-of-use tariff rate
S.O. Masebinu a, *, E.T. Akinlabi a, E. Muzenda b, A.O. Aboyade c
a
Department of Mechanical Engineering Science, University of Johannesburg, South Africa
b
Department of Chemical Engineering Technology, University of Johannesburg, South Africa
c
Faculty of Engineering and the Built Environment, University of Johannesburg, South Africa

a r t i c l e i n f o a b s t r a c t

Article history: Time-of-Use has been introduced in South Africa as part of demand side management measures. Battery
Received 3 August 2016 energy storage (BES) can take advantage of energy price arbitrage under favourable pricing regimes.
Received in revised form However, the challenge is to what extent will the introduced policy favour the installation of BES at
9 May 2017
residential accommodations? The tools to assess suitability of installing BES exist but they come at a cost.
Accepted 20 June 2017
Available online 23 June 2017
In this study, we improved upon existing methodology and implemented it in Microsoft Excel to assess
techno-economic viability and environmental benefits of using BES. The approach showed that none of
the three BES technologies investigated was economically viable at the prevailing average rate of 0.1442
Keywords:
Demand side management
$/kWh for peak electricity. The Monte Carlo simulation implemented suggests that the minimum mean
Energy storage price of peak needed for the BES system to break even range between 0.2560 e 0.2919 $/kWh. At 50%
Environment discount in storage medium cost and 100% increase in the price of peak, the BES will only break even
Probabilistic simulation when the average price of peak is 0.2043 $/kWh at maximum cycling cost that range between 0.1077 e
Techno-economics 0.1560 $/kWh. The study concluded that reduction in the cost of storage medium has more impact on
Time-of-use economic viability than increasing only peak price of electricity.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Emerging economies will account for about 90% of the global net
energy demand growth by 2035 [1]. Electricity demand will be
higher among other forms of energy in the emerging economies [1].
This projected growth has increased the interest in improved en-
ergy efficiency to limit the risk and environmental impact of the
sources of energy which are mostly from fossil fuel [2,3]. South
Africa's economy, identified as an emerging economy, is energy
intensive and in recent years, economic growth coupled with
increased access to electricity by previously unserved communities,
has led to an increased demand for electricity, that at some point in
2008, exceeded generation capacity [2,4]. This situation forced
Eskom, the state-owned utility which accounts for more than 95%
of electricity generated and distributed nationwide [5], into
controlled load shedding with an estimated economic impact of
about US$268 million [5]. To avoid a repeat of that magnitude of
load shedding, Eskom embarked on two major mitigating
* Corresponding author.
E-mail address: masebinity@gmail.com (S.O. Masebinu).
strategies. The first of these strategies was to build new baseload
power plants, as well as re-commission and upgrade old power
stations where feasible, to support peaking power plant re-
quirements. The second strategy focused on implementing con-
sumer demand reduction through integrated demand management
(IDM) [5e7]. The target of the first strategy was to add 17,000 MW
to electricity generating capacity by the end of 2019 [6]. Efforts
under that strategy led to 3655 MW added to the grid in 2013 and
another 9564 MW is expected by the end of 2017 from two coal
power plants [7]. Also, 1600 MW of renewable sourced energy was
added to the grid in 2014 as part of the first strategy [8]. The second
strategy, IDM, aimed to increase energy efficiency and was made up
of a series of demand side management (DSM) programmes
namely; energy efficient DSM; demand response and energy con-
servation [5]. With the implementation of the IDM, 19 GWh
annualised energy savings was achieved in 2013/14 [7]. The suc-
cesses achieved on the two strategies, are under threat due to
Eskom's aging fleet. The utility reported that two-third of its power
stations are beyond the mid-point of their expected life span [7].
This is indicative that continued expansion and upgrading of the
country's power generation and grid infrastructure is required, to
ensure that generation keeps pace with demand in the medium to

http://dx.doi.org/10.1016/j.energy.2017.06.118
0360-5442/© 2017 Elsevier Ltd. All rights reserved.
414 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Nomenclature gk Yearly inflation rate for cost of component k [%]

i Annual interest rate [%]
NPCO&M Net present cost for operation and maintenance
Acronyms [$/kW-yr]
LD and HD Refers to low demand (summer) and high demand costO&M Unit operation and maintenance cost [$/kW]
(winter) gO&M Yearly operation and maintenance cost inflation rate
PrElpeak Price of electricity at peak TOU [$/kWh] [%]
PrElstand Price of electricity at standard TOU [$/kWh] gPrEl Yearly inflation rate for price of electricity [%]
PrEloffpeak Price of electricity at offpeak TOU [$/kWh] NPCwoutES Net present cost for system without battery energy
Epeak Summation of AC energy consumed during peak [kWh] storage [$]
Estand Summation of AC energy consumed during standard NPCwithES Net present cost for system with battery energy
period [kWh] storage [$]
Eoffpeak Summation of AC energy consumed during offpeak NPVsavings Net present saving [$]
period [kWh] LCOE Levelized cost of electricity [$/kWh]
Ibatt ðtÞ The current in/out of the battery [A] IRR Internal rate of return [%]
Imax Maximum current to charge the battery without costfixed year Annual fixed service and demand charges [$/yr]
damage [A] d Day of the year [1e365]
SOC State of charge of the battery t Hour of the year [0e8759]
Capbatt The battery capacity [kWh] h Hour of the day [0e23]
Pmax Maximum absorbable power from the AC grid [kW] Dt Time step of simulation [1 h used]
Pload ðtÞ Average demand over hour t [kW] NPCall comp The sum of NPC of all k components and the O&M
hAC AC-DC converter, the rectifier, efficiency [%] cost [$]
DC

hDC DC-AC converter, the inverter, efficiency [%} NPCE from batt The NPC of energy to charge the battery [$]
vh
AC
Hourly noise factor
hbattch Battery charging efficiency [%]
vd Daily noise factor
hbattdisch Battery discharging efficiency [%]
Elquality Life time benefit for electricity quality [$]
Pbi dir The rated power of the bi-directional converter [kW]
Gridcap ut Life time benefit for grid capacity utilization [$]
VDC The DC bus voltage [V]
CSavings Coal savings [kg]
Pfrombatt ðtÞThe load demand met by the battery during peak hours
Ccons peak Coal consumption during peak [kg]
[kW]
Ppeakdir ðtÞ The load demand not met by battery during peak hours Ccons offpeak Coal consumption during offpeak [kg]
[kW] CO2; Avoided CO2 avoided [kg]
d Self-discharge rate COAvoided CO avoided [kg]
lifek Life of component k [years] SO2; Avoided SO2 avoided [kg]
lifepro Life of project [years] NOx; Avoided NOx avoided [kg]
DoD Depth of discharge Sar Rate of sulphur removal [%]
Rk Number of replacement of component k within project ts Percentage of SO2 emission out of sulphur in coal [%]
life [#] hs Desulphurisation rate [%]
NPCk Net present cost of component k [$] NN Mass fraction of nitrogen in coal [%]
Ncycles @ %DoD Number cycles at depth of discharge [#] hn Efficiency of nitrogen to be converted into NOx [%]
Efrombatt Load met by the battery [kWh/day] g Percentage of NOx from the fuel out of the total
costk Cost of component k [$] amounts of NOX emissions [%]
Epeak dir Daily load during peak hours not met by the battery hN Nitrogen removal efficiency [%]
purchased from the grid [kWh/day]

long term. However, South Africa's international and domestic
commitments to climate change mitigation means that a significant
part of any new generation infrastructure, needs to be based on
sources “cleaner” than coal. However, cleaner, renewable sources
like wind and solar that forms the major part of the renewable
energy independent power producer procurement (REIPPP) are
intermittent and thus present challenges to grid stability and grid
management, especially at high penetration levels. It is widely
recognised that energy storage is one of the major ways to address
this challenge. But energy storage technologies are still relatively
expensive, requiring policy and regulatory interventions that
improve the commercial viability of their implementation. The
time-of-use (ToU) tariff system is one such regulatory tool. ToU
pricing divides a day into two or three segments with different
prices that remain fixed day-to-day over a season.
The ToU tariff system has been introduced by Eskom in some
part of the country. It is an element within the IDM, and has been
introduced to reduce peak demand aside a yearly increase in tariff.
A ToU allows for flexibility of demand deferment with potential to
reduce customer electricity bill. It can reduce whole market prices
of electricity, avoid or defer capacity investment, give utility an
opportunity to operate with more level demand curve, increase the
integration of distributed energy sources, and decrease environ-
mental pollution by avoiding the need for operating a power plant
at peak period [9,10]. Deployment of energy storage (ES) can be
advantageous due to the different pricing regimes for end cus-
tomers. ES system assists in storing energy during low electric price
regimes and discharge the stored energy when the applicable
charge is at peak ToU, usually with a high tariff. This DSM strategy
helps to reschedule energy consumption over a period of time and
 S.O. Masebinu et al. / Energy 135 (2017) 413e429
thus, the end user can reduce the overall cost of electricity by
shaving consumption, exploiting energy price differential or by
matching demand to supply [11e16]. The use of ES, from a
perspective of the consumer owing the ES, will be economical if the
total life cycle cost of the ES is less than the total savings in the
consumer's electricity bill. At grid level, ES can assist in load
balancing, power quality improvement and bridging of the power
gap [12,15,17e21]. At present, Eskom is exploring battery ES (BES)
at large scale under the program “Large Scale Energy Storage”
portfolio (LSES). At Eskom's LSES facility, lithium ion phosphate
battery and sodium nickel chloride battery technologies are being
tested and there are plans to install two more flow battery systems
[22]. The testing period is expected to last over three years with
over 1000 nameplate cycles anticipated [22]. While these tests are
being conducted for large scale systems, peak shaving at consumer
side which can immediately reduce peak demand also needs par-
allel investigation.
The technologies for ES are numerous and have been exten-
sively discussed in literature [11,23]. The technologies include
mechanical ES; pumped hydro (PHES), compressed air (CAES), and
flywheel; chemical; zinc bromide (ZnBr), vanadium redox flow
battery (VRFB), lead acid (PbA), sodium sulphur (NaS), nickel
cadmium(NieCd), iron chrome (FeCr), lithium ion (Li-ion) and
hydrogen; electrochemical; capacitors and supercapacitor; and
electromagnetic; superconducting magnetic [11,23]. PHES and
CAES are the most cost effective ES for large scale capacity while
batteries are dominating on small to medium scale applications
[23e25]. Of the battery technologies, PbA, NaS, and Li-ion are the
most suitable for residential, commercial and industrial applica-
tions, due to technology maturity, high efficiency and low main-
tenance cost [26,27]. In recent times, flow batteries have also been
deployed at medium scale level. A pilot project using vanadium
redox flow battery has been implemented in South Africa with
reported energy production cost that ranges between 0.3506 and
0.5409 $/kWh1 [28].
There has been a number of studies investigating the com-
mercial viability of BES under ToU regimes. Yan et al. [27] inves-
tigated BES for commercial buildings considering two ToU
regimes, peak and off peak, and concluded that PbA holds more
economic prospect than NaS and Li-ion batteries. Dufo-Lopez and
Bernal-Agustin [26] simulated PbA and Li-ion for grid connected
residential ES under a three regime ToU tariff structure. The study
concluded that at the present high cost of batteries, both PbA and
Li-ion are not economical. However, Li-ion holds more economic
prospect due to higher efficiency. The critical economic elements
for the profitability of the system are dependent on the unit cost of
storage medium and number of cycle to failure. Battke et al. [11]
conducted a probabilistic analysis on lifecycle cost of four batte-
ries across six applications. They concluded that the cost perfor-
mance of NaS, PbA and Li-ion are close over the applications
investigated, hence, no clear leading technology was identified.
Zakeri and Syri [23] expanded on the technologies considered by
Battke et al. [11] and showed the economic performance of
different ES scenarios. The study concluded that the cost of the
storage medium, replacement cost, interest rate, electricity prices
and discharge time has higher impact on the levelized cost of
energy (LCOE) than other parameters. They also reported that the
effect of uncertainties in techno-economic data can affect evalu-
ation results by 5e17%. Safaei and Keith [29] assessed ES from a
perspective of decarbonisation. The study proposed that only a
small proportion of ES is needed to decarbonise electricity as low
priced natural gas provides much more of grid management.
1
1 USD ¼ R14.49 18th April 2016.
415
However, they also stated that ES becomes highly important when
emission cut is constrained close to zero without any other
dispatchable-zero-carbon source allowed within the grid. In the
work of Kaldellis et al. [25], NaS and PbA, are two batteries with an
economic prospect for small to medium scale applications for a
small island. Whereas, PHES and CAES are more suitable for large
scale applications. Hessami and Bowly [30] investigated the eco-
nomic performance of different ES. They concluded that CAES is
more economically viable than batteries for large scale applica-
tion. From the reviewed literature, different methodologies and
input parameters determine which technology is more appro-
priate and economically viable. The literature suggests that while
PbA, NaS, and Li-ion are closely competitive, however, the degree
of uncertainty around their key economic parameters such as the
cost of storage, cycle life and efficiency, makes it difficult to decide
on the most appropriate technology with high degree of
confidence.
This study aims to contribute to the body of knowledge on ES by
using a student residence in an academic institution as a case study
in emerging economies like South Africa that is sometimes faced
with persistent load shedding problems, partly influenced by high
energy demand from student residences at peak periods. It is thus a
good example of a load center where BES could be employed for
peak shaving as part of IDM. The energy consumption profile of
student residences tends to follow similar consumption profile on a
daily basis, hence combining BES with the implementation of a ToU
regime might proffer some technical and economic advantages
both to the grid operator and to the consumer. The tools to assess
the technical and economic suitability of installing BES for resi-
dential accommodation exist but they come at a cost and require
expert training to use them. The development of a simple, adapt-
able and open access tool to help determine the commercial
feasibility of BES deployment would therefore be useful, not just for
energy managers of these institutions, but also for policy makers. In
this work, we have improved upon existing methodology and
develop a system that helps to address South Africa's specific ToU
regime. Also, the proposed set of equations has been implemented
in a widely assessable tool, Microsoft Excel Visual Basic Application,
with a spreadsheet front end for data input. The tool can be used for
a deterministic calculation without any additional add-in. To cater
for the uncertainty in input parameter, Palisade @Risk add-in was
used. The developed tool (downloadable from https://goo.gl/
iem29T) was then used to assess the technical and economic ben-
efits of implementing BES to take advantage of the ToU tariff system
in South Africa. This study can therefore be useful in shaping future
policy frameworks for encouraging ES as a grid management and
peak shaving mechanism. A further exploration of the emissions
avoided due to peak shaving is also included to emphasize the
environmental benefits of ES.
2. System description
The economic viability of peak demand smoothing using BES
compared to the cost of a system without BES using the City of
Johannesburg's City Power ToU (CoJ-ToU) tariff structure was
investigated in this study. In the CoJ-ToU tariff system, the price of
electricity varies for the different time of weekdays (Monday to
Friday), weekends, low demand season (September to May) and
high demand season (June to August) as shown in Fig. 1. The
applicable three ToU structures are:
1. Peak-period when demand is expected to be highest, requiring
high electricity production cost via peaking plants, thus the
highest price of electricity. The price of electricity during this
period is denoted as PrElpeak ;
416 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Fig. 1. CoJ applicable TOU regime for different conditions.

2. Standard-period when demand is expected to be normal or
moderate. The price of electricity during this period is denoted
as PrElstand ;
3. Off peak-period when electricity demand is expected to be low
and less expensive to produce. The price of electricity during
this period is denoted as PrEloffpeak .
For weekdays during low demand season in Fig. 1, off-peak TOU
(indicated by the green segment) starts from 2200hr and ends by
0600hr of the next day. It is followed by an hour standard TOU from
0600 to 0700hr (mid-peak indicated by the yellow segment) and
thereafter, a peak TOU from 0700 to 1000hr. Then a standard TOU
continues from 1000 to 1800hr, followed by a peak TOU from 1800
to 2000hr. The corresponding CoJ-ToU pricing regimes for boarding
houses, flats, and student commune for weekdays is presented in
Fig. 2. On Saturday only off peak and standard TOU are applicable
and for Sunday it is an all day off peak.
The methodology used to size the BES, to simulate the charge
and discharge sequence, and develop the economic model for using
BES and a system without BES is presented in the next section. The
metric to assess the economic benefit of using BES could be through
a number of approaches, including the net present value or cost
(NPV/C), the LCOE, internal rate of return (IRR) and the payback
period. All metrics except the payback were used to assess the
profitability of the system described in this study. Payback does not
account for benefit that occur after break even and therefore it does

Fig. 2. CoJ price of electricity under a TOU for weekdays.1.

 S.O. Masebinu et al. / Energy 135 (2017) 413e429 417

Fig. 3. Power demand for 17th April 2016.

not measure profitability adequately. It is only applicable if at the
end of a project life, the net cash inflow outweighs the net cash
outflow. Also, due to replacement of storage medium there is a
possibility of recording more than one payback period. Therefore, it
has not been considered in this study.
3. Methodology
3.1. Description of case study
A residence with 17 rooms occupied by students from the Uni-
versity of Johannesburg was used as the reference for this study.
Energy consumption data for the residence was logged using an
Engage Efergy energy data logger. The energy data logger consists
of three current transformer (CT) clamps, a transmitter, receiving
hub and a router. The power demand data for a typical weekday is
shown in Fig. 3.
average hourly load profile produced from the simulation is
shown in Fig. 4. The boxplot shows the aggregated minimum, first
quartile, mean, third quartile and maximum energy consumption
for 24-h. The outliers are represented by black dots and are
excluded from the calculation. Electrical loads tend to be noisy as
their energy consumption and time of use are not constant at
every time step. Therefore, to simulate consumption of energy
over a year from the two months' data collected, some degree of
variability termed as noise were introduced into the average daily
energy consumption data. The noise analysis was introduced to
account for hourly, daily and seasonal variation as presented in
Equation (1). The hour (t) for a particular day, d, in the year is
calculated in Equation (2).
Pload ðtÞ ¼ Pavg *ð1 þ vh vd Þ (1)
t ¼ ½ðd  1Þ*24 þ h (2)

3.2. Load data and energy consumption where t is each hour for the year from 0 to 8759, d is days of the
year from 1 to 365, and h is hours for a day from 0 to 23. Matlab
The load data retrieved from the data logger over the month of was used to simulate and generate the yearly data from the logged
April and May 2016 was the basis for the yearly simulation. The data as shown in Fig. 5 according to procedure presented by

Fig. 4. Summarised hourly energy consumption for April 2016.

418 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Fig. 5. Simulated load data.

Okundamiya et al. [31]. Column ðh  1Þ matrix is the hourly load,
Pload ðtÞ [kW], can be represented as a ðh  dÞ matrix by multi-
plying Pavg with a row ð1  dÞ matrix of ones. The hourly noise
function, vh , is a ðh  1Þ matrix whereas the daily noise function,
vd , is a ð1  dÞ matrix. The hourly and daily noise input values are
randomly generated from a normal distribution. The normal dis-
tribution has an average of zero and a standard deviation equal to
the hourly and daily noise factors respectively [32e34]. These
factors are considered here as 0.8 and 0.1 respectively, which were
calibrated from the real-time load data using regression tech-
niques [35]. The annual consumption under the three TOU re-
gimes was then calculated from the simulated Pload ðtÞ as shown in
Equation (3)
X
8759  demand charges.
ETOU ¼ Pload LD ðtÞ þ Pload HD ðtÞ (3)
t¼0
affect the stability, and reliability of electricity supply as well as
the utilization rate of generating plant capacity. All of these can
be minimised with the implementation of BES which will
shift peak load to off-peak period and reduce the peak to valley
gap.
The average daily Eoffpeak, Estand and Epeak as calculated
according to Equation (3) were 105.88, 102.26 and 55.23 kWh
respectively. The applicable ToU hours and tariffs were presented
in Figs. 1 and 2. The tariffs are for 2015/2016 CoJ-ToU rates for a
hostel, large commune, and flats. These tariffs are also applicable
over the weekend except that there are no peak charges.
Hence, the battery is constrained not to discharge over week-
ends. Other technical and economic inputs are presented in
Table 1. The fixed cost in Table 1 represents both service and

The average daily energy consumption of the residence was 3.3. BES sizing
256.59 kWh from the logged dataset. However, the simulated
data with noise inclusion and seasonal variation, the average The average energy required from the batteries was 22.09 kWh,
daily consumption is 263.37 kWh. The peak to valley ratio of the 40% of the average daily peak energy. The decision for 40% of peak
simulated data was 5.12 and the minimum daily load factor was was to cater for lightings and other low power appliances, while all
0.19. The small daily load factor and huge peak to valley ratio other heating appliances in the residence such as electric geyser,
cooker and flat iron are covered by the grid. The capacity of the
batteries, Capbatt , chosen were 32.54, 30.95 and 29.22 kWh for PbA,
Table 1
Techno-economic input parameters. NaS, and Li-ion respectively based on their mean round trip effi-
ciency presented in Fig. 8. With SOCmin of 20%, the available ca-
Parameters (unit) Values
pacity were 26.03, 24.75 and 23.38 kWh for PbA, NaS and Li-ion
Fixed cost ($/yr) 439 battery respectively. The Efrom batt was the product of the avail-
SOCmin (%) 20
able capacity with the battery discharge efficiency and the DC-to-
Life of project (yr) 25
Life of bi-directional converter (yr) 10 AC inverter efficiency. Equal battery charge and discharge effi-
Unit cost of bi-directional converter ($/kW) 345 ciency was considered. These were calculated by taking the square
Direct current voltage (V) 48 root of the round-trip efficiency [36]. With a VDC ¼ 48 V, the Ah
Time step (hr) 1
capacity required is 678, 645, and 609 Ah for PbA, NaS and Li-ion
Component inflation (%) 2
O&M escalation rate (%) 2
respectively. The choice of battery that can be implemented var-
Electricity escalation rate (%) 9 ies, for example, PbA battery in South Africa cost on average 162
Interest rate (%) 9 $/kWh [37]. The Trojan L16HC model, 420 Ah/6 V, can be used (2
AC-to-DC and DC-to-AC efficiency (%) 98 parallel strings of 8 in series). Similar considerations were also
given to NaS and Li-ion BES.
 S.O. Masebinu et al. / Energy 135 (2017) 413e429 419

Fig. 6. Grid connected ES system.

3.4. Sequence of operation for the BES
The methodology presented by Dufo-Lopez and Bernal-Agustin
[26] serves as the basis for analysis. The BES system schematic
presented in Fig. 6 consists of a battery bank and power conversion
system (PCS) that includes the following: battery charger (AC-to-
DC), inverter (DC-to-AC) and charge/discharge controller for the
voltage. Taking a time step, Dt , of 1 h, the SOC rises from
SOCðt  DtÞ to SOCmax . The SOC at any time t, SOCðtÞ, is calculated
using Equation (5) [36]. During the charging duration, the battery
does not supply any power to meet load. That is both Pfrom batt ðtÞ
and Ppeak dir ðtÞ are zero. All load requirements are met by the AC
grid.

"  #
ðSOCmax  SOCðt  1ÞÞ*Capbatt *1000 Pmax  ðPload ðtÞ*1000ÞÞ*hAC=DC P
VDC bi dir
Ibatt ðtÞ ¼ min Imax ; ; ; (4)
Dt VDC VDC

battery. The PCS could be a single bi-directional unit as assumed in

this paper while the control can be configured to suit the consumer
need. The BES will be used to provide part of the peak load. The
accrued benefit is derived from arbitrage; buying electricity during
off peak period at PrEloffpeak to charge the battery and discharging
during peak period at PrElpeak . The BES in this case is grid connected
and is charged by a conventional energy source. It is assumed that
the control strategy presented in Fig. 6 is ideal and component
efficiency is constant over each charge and discharge cycle. The
technical charging and discharging constraints imposed on the
system is described below (flowchart Fig. 7);
During off-peak period: switch 1 is closed while switch 2 is
open. If the battery is at a SOC lower than the SOCmax , current Ibatt ðtÞ
[A] flows into the battery through the rectifier to charge the battery.
From Equation (4) [36,38], the current Ibatt ðtÞ flowing into the
battery is the minimum current required to charge the battery to a
SOCmax over the offpeak hours, in this case, it is 6 h, Pmax is the
maximum absorbable power from the AC grid and VDC is the DC bus
Ibatt ðtÞ*Dt*hbatt
SOCðtÞ ¼ SOCðt  DtÞ þ ch
(5)
Capbatt *1000
VDC
During standard period: the AC grid satisfies all load
requirement of the residence. Switch 1 and 2 are open, hence no
current flows into the battery, that is Ibatt ðtÞ is zero, and the bat-
tery do not discharge to meet load, Pfrom batt ðtÞ is zero. The
available capacity of the battery is the difference between the
SOCmax and any self-discharge (d) during an idle period as calcu-
lated in Equation (6).
SOCðtÞ ¼ SOCðt  DtÞ*ð1  dÞ (6)
During peak period: switch 1 is open, switch 2 is closed and
current, Ibatt ðtÞ, flows through the closed switch Equation (7) [39].
The battery discharges to meet all load Pload ðtÞ or part of it
depending on the discharge constraint. The discharge constraint
420 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Fig. 7. Flow chart of the charging and discharging of the BES.

implemented in this study, was that the BES should deplete its load satisfied by the AC grid Ppeak dir ðtÞ is the difference between
capacity to SOCmin over the 5 h peak period. Therefore, the power the load Pload ðtÞ and the load satisfied by the battery, Pfrom batt ðtÞ
from the battery, Pfrombatt ðtÞ, can be calculated from Equation (8). calculated from Equation (10).
The SOC reduces from SOCðtÞ to SOCmin each day Equation (9). The
 S.O. Masebinu et al. / Energy 135 (2017) 413e429 421

Fig. 8. MCS input parameters.

"  #
SOCðt  1Þ  SOCmin Þ*ðCapbatt
VDC
*1000Þ
P ðtÞ*a*1000 Pbi dir
Ibatt ðtÞ ¼ min Imax ; D ; load ; (7)
Dt VDC *hDC=AC VDC *hDC=AC

investigation, the battery will discharge at high current and once

Ibatt ðtÞ*hDC=AC *VDC SOCmin is achieved, in about 2 h, the charging of the battery com-
Pfrombatt ðtÞ ¼ (8)
1000 mences. Therefore, the most critical consideration in estimating the
battery life for this case is the degradation caused by the energy
Ibatt ðtÞ*h Dt cycled [26]. Other factors which can also impact on the life of the
SOCðtÞ ¼ SOCðt  DtÞ  batt disch
(9) battery but has not been considered are handling and storage
Capbatt *1000
VDC condition, operating temperature, battery chemistry and corrosion.
The life of bi-directional converters is dependent on the manufac-
Ppeak dir ðtÞ ¼ Pload ðtÞ  Pfrom batt ðtÞ (10) turer's quality control, quality of installation, exposed environ-
mental conditions and usage. Generally, it is between 8 and 15
Having known the load satisfied by the BES and that not met by
years [26,40]. The average of 10 years will be used for the bi-
it, the annual energy from the BES and part of the peak purchased
directional converter while the battery life will be calculated from
from the grid can be calculated from Equation (11) and Equation
Equation (13) as adapted from Ref. [36].
(12) respectively.
Capbatt *DoD*Ncycles @ %DoD
X
8759  Lifebatt ¼   (13)
Efrombatt
Efrom batt ¼ Pfrom batt LD ðtÞ þ Pfrom batt HD ðtÞ (11) hbattdisch *hDC=AC
t¼0

X
8759 
Epeak direct ¼ Ppeak dir LD ðtÞ þ Ppeak dir HD ðtÞ (12) 3.5.2. Component replacement
t¼0
The number of replacement for each component k within the
project is the integer value obtained by dividing the project life,
Lifepro , by the component life, Lifek , as shown in Equation (14).

3.5. Cycle life and replacement of components  

Lifepro
Rk ¼ integer (14)
Lifek
3.5.1. Life of components
If the life of component k, Lifek , is less than the project life,
Lifepro , there will be a need to replace component k. We only
focused on the storage medium and the bi-directional converter for 3.6. Economic modelling
replacement though other components like fuse and switches
might also need replacement. The life of a battery can be estimated The net present cost of component k, NPCk , is the cost of acquiring
either by the number of cycles to failure or lifetime throughput the component, plus the cost of replacement due to failure or end of
energy before failure [40]. Due to the case study under component's life over the project lifetime minus the value of the
422 S.O. Masebinu et al. / Energy 135 (2017) 413e429

remaining component's life that has not been used at the end of the NPCwout BES is the sum of the fixed service charge with the sum-
project. NPCk is calculated by Equation (15). The first term is the cost mation of the product of the total energy consumed during each
of the component, the second term is the cost of replacement of the ToU by the applicable ToU tariff. The NPCwith BES calculates in
component over project life and the third term is the value of the addition to the cost of acquiring the BES and energy cost for
component not used at the end of the project life. charging the BES, the cost of the energy not met by the BES,
Epeak direct , for both low and high demand seasons. If the NPCwith BES
0 !
X
Rk
ð1 þ gk Þj*lifek is less than the NPCwout BES , implementing BES hold some prospect
NPCk ¼ costk þ @ costk * for further investigation. The NPCwout BES , is calculated by Equation
j¼1 ð1 þ iÞj*lifek (20) and NPCwith BES , by Equation (21).
!1
costk ð1 þ gk Þlifepro A "
 *½lifek *ð1 þ Rk Þ* (15) X
lifepro  
lifek ð1 þ iÞlifepro NPCwout BES ¼ PrElpeak LD *Epeak LD þ PrElpeak HD *Epeak HD
j¼1
Added to the NPC for all components is the net present operation  
þ PrElstand LD *Estand LD þ PrElstand HD *Estand HD
and maintenance cost for all components, NPCO&M , within the  
project life. þ PrEloffpeak LD *Eoffpeak LD þ PrEloffpeak HD *Eoffpeak HD
The NPCO&M is calculated by Equation (16) [41]. Hence the #
 ð1 þ g Þj
PrEl
overall NPC and O&M cost for all components over the project life is þ costfixed year *
given by Equation (17). ð1 þ iÞj
(20)
!
X
life pro
ð1 þ gO&M Þ j
NPCO&M ¼ costO&M * (16) NPCwith BES ¼ NPCall comp þ NPCE from batt
j¼1 ð1 þ iÞj "
X
lifepro 

þ PrElpeak LD *Epeakdirect
X LD
NPCall comp ¼ ðNPCk Þ þ NPCO&M (17) j¼1
 
k þ PrElpeak HD *Epeak direct HD þ PrElstand LD *Estand LD
The energy from the BES, which was earlier purchased from the  
þ PrElstand HD *Estand HD þ PrEloffpeak LD *Eoffpeak LD
grid is discharged during peak hours. Net present cost of energy

from the BES, NPCE from batt , is calculated by Equation (18) [36]. þ PrEloffpeak HD *Eoffpeak HD
Equation (18) sums up the total energy to charge the BES during #
 ð1 þ g Þj
low demand season and multiply it by PrEloffpeak LD. Similarly, the PrEl
þ costfixed year *
sum of the energy to charge the BES during high demand season is ð1 þ iÞj
multiplied by PrEloffpeak HD. The sum of both energy cost is calcu- (21)
lated over the life of the project including the charging and dis-
charging efficiencies and the AC-DC and DC-AC efficiencies. The If the difference between NPCwout BES and NPCwith BES is positive, i.e
result is then discounted to calculate the NPCE from batt . greater than zero, then deploying BES is said to be profitable. The
difference for the saving accrued for exploiting BES is presented in

2P8759  P
8759
 3
X batt LD ðtÞ*PrEloffpeak LD þ t¼0 Pfrm batt HD ðtÞ*PrEloffpeak HD
life Pfrm
pro
4
t¼0 ð1 þ gPrEl Þj 5
NPCE from batt ¼ * (18)
AC *hbatt
hDC ch *hbatt disch *hDC ð1 þ iÞj
j¼1 AC

Having known the NPCall comp , NPCE from batt and Efrombatt , the peak Equation (22)
price of electricity needed, PrElpeak needed , to evaluate the profit-
ability of deploying BES can be calculated as presented in Equation NPVsavings ¼ NPCwout BES  NPCwith BES (22)
(19). The numerator term gives the life cycle cost of acquiring the
The percentage of NPV saving for using BES with respect to case
BES while the denominator term gives the lifetime energy pro-
without BES is given by Equation (23).
duction of the BES. If the PrElpeak needed is less than the prevailing
weighted average PrElpeak , then making use of BES will be profit-
NPVsavings
able. The weighted average price of peak caters for different peak %NPVsavings ¼ *100% (23)
prices for the different seasons within the year. NPCwout BES
Another metric for economic evaluation is the LCOE for
NPCall comp þ NPCE from batt deploying BES. The LCOE aggregates the lifetime cost of the system
PrElpeak ¼  (19) and the lifetime energy production to determine the minimum
needed Plifepro ð1þgPrEl Þj
j¼1
Efrom batt * j price ($/kWh) at which energy generated by any system must be
ð1þiÞ
sold to break even over the life time of the project. It is desirable to
To further assess the profitability of utilising BES, comparing the have a selling price above the minimum LCOE. If the LCOE of a
NPC of a system without BES, NPCwout BES , to that with BES, system with BES, LCOEwith BES , is less than the LCOE of a system
NPCwith BES , is another metric for economic comparison. The without ES, LCOEwout BES , then the deployment of BES is said to be
 S.O. Masebinu et al. / Energy 135 (2017) 413e429
profitable. The LCOEwout BES is calculated by Equation (24) while
LCOEwith BES is calculated by Equation (25) [42]. The difference be-
tween the LCOEwith BES and PrElpeak needed lies in the additional cost
of the unmet load by the ES system purchased from the grid that
was included in the LCOEwith BES .
NPCwout BES
LCOEwout BES ¼ P  (24)
8759
t¼0 Pload ðtÞ *lifepro
NPCwith BES
LCOEwith BES ¼ P  (25)
8759
t¼0 Pload ðtÞ *lifepro
Another econometrics for investment decision making is the
internal rate of return (IRR) of an investment. It is the discounting
rate that equates the NPV of the negative and positive cash flows.
That is, the discounting rate at which the total NPVsavings is zero. This
is given by Equation (26) and iteratively calculated by changing i.
IRR ¼ i; when NPVsavings ðiÞ ¼ 0 (26)
3.7. Probabilistic approach
Known to BES are the lack of consistent data on technical and
economic parameters, uncertainties about feasible uses, and ul-
timately cost competitiveness. Although, an extensive review of
literature exists for technical characteristics and cost details
[23,43e49], the degree of variability of such reviews are high.
Despite the variability associated with the technical and economic
parameters for BES, there is still a need to appropriately match
pairs of technology and application towards a cost-effective de-
cision making. To cater for the uncertainties, a probabilistic
approach through Monte Carlo Simulation (MCS) was employed.
MCS is governed by the law of large numbers around the mean
distribution. The MCS model requires an adequate number of
simulation to achieve the distinctive calculation accuracy. 10,000
iterations as proposed by Baumann et al. [24] was therefore
implemented in this study. The reference technical values used for
simulation were derived from Baumann et al. [24] while the cost
of storage medium was retrieved from Sustainable [37]. Fig. 8
shows the minimum, first quartile, mean, third quartile and
maximum input parameters for the MCS of each BES technology
investigated. Fig. 8aed are the number of cycle life at 80% depth of
discharge, round trip efficiency, battery acquisition cost and
operation and maintenance cost of each BES technology respec-
tively. High variability was observed in the data as presented in
Fig. 8. The MCS was conducted using Program Evaluation and
Review Technique (PERT) distribution based on a beta distribution
as proposed by Battke et al. [11]. The PERT is insensitive to
extreme value unlike the triangular distribution and requires
three inputs; the minimum, mode (most likely parameter) and
maximum values for each parameter. The mean has been
considered as the most likely parameter in this study. The factors
considered for MCS, which is assumed to be mutually uncorre-
lated, are cycle life, round trip efficiency, storage cost, and oper-
ation and maintenance cost.
3.8. Other benefits of BES
In this study, only conventional electrical energy source was
considered to charge the BES. The use of BES supports the inte-
gration of renewable energy (RE) sources, which will be consid-
ered as part of a more sustainable project in the future.
423
Notwithstanding the exclusion of RE sources in this present study,
other techno-economic and environmental benefits of using ES
include improved electricity quality, maximizing grid capacity
and reduction in associated air pollution from conventional en-
ergy sources.
3.8.1. Improved electricity quality
BES system can flatten out sudden short-duration surges on
waveform caused by variation in voltage, frequency, harmonics,
and power factor to improve power quality and protect the cus-
tomer's sensitive processes and load [50e53]. Also, BES can provide
an uninterrupted power supply to support customer load during
power disruptions and assist in resynchronization of load with the
grid when power is restored [17,47,54,55]. According to Schoenung
and Eyer [56], the annual benefit for improving electricity quality is
about 750 $/kW. Equation (27) gives the project lifetime benefit for
improving electricity quality, Elquality .
Efrom batt
Elquality ¼ *lifepro *750 (27)
hdisch dur
3.8.2. Improved grid capacity utilization
BES system can reduce the urgency or defer the need for
investing in new transmission and distribution facilities and sub-
stations by storing energy until there is less stress on the grid
[11,51]. BES systems are used to avoid congestion-related charges
that occur during high demand period when cheap energy cannot
be delivered to some loads due to transmission facilities peaking
during such period. According to Schoenung and Eyer [56], the
annual benefit for grid capacity utilization, Gridcap ut , is about 650
$/kW. The project life time benefit is given in Equation (28).
Efrom batt
Gridcap ut ¼ *lifepro *650 (28)
hdisch dur
3.8.3. Reduced emission of grid power generating source
South Africa is the fifth largest coal producing country in the
world and third largest exporter of the commodity [57]. 77% of its
energy needs is being met by coal and about 90% of electricity
generation is from coal-fired plants [7,57]. By making use of BES,
emissions from coal-fired plant during peak period can be reduced
due to peak demand shifting. The emissions include CO2, SO2, NOx,
and CO.
The coal required to manufacture the BES is calculated based on
the battery medium weight and equivalent CO2 emitted presented
in Equation (29) according to [58]. The CO2 emission (kg CO2,eq./kg
of battery weight), SCO2 , is given as 0.9, 1.2 and 12.5 while energy
density, En denbatt , is given as 0.04, 0.2, and 0.13 kWh/kg for PbA,
NaS and Li-ion respectively [58]. On average, 2.6 ton of CO2 is
emitted per ton of coal combusted [27]. The coal savings, CSavings , for
using BES is given by Equation (30) [27]. It is estimated by finding
the difference between the mass of coal consumed to charge the
BES during off peak hours, Ccons offpeak , mass of coal avoided for
using BES during the peak period, Ccons peak and mass of coal
required to manufacture the BES. Other emissions are calculated
based on the CSavings . The avoided CO2 emission from the CSavings ,
CO2; Avoided , is given by Equation (31).
Capbatt
* SCO2
En denbatt
Crequired ¼ (29)
2:6
424 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Table 3
  Deterministic viability result.
CSavings ¼ Efrom batt *Ccons peak
Base Case Result Lead Acid NaS Li-ion
! !
X
8759 Cycles @ 80% DoD [#] 775 2659 3752
 Pbidir *tchar *Ccons offpeak *Lifepro  Crequired Round trip efficiency [%] 75 83 93
Storage Cost [$/kWh] 162 395 643
t¼0
O&M Cost [$/kW-yr] 9.39 17.94 10.21
(30) Results
Battery life 2.99 10.72 16.29
NPC Comp [$] 26,866 22,964 26,219
CO2;Avoided ¼ 2:6*CSavings (31)
NPC E-From Bat [$] 10,748 9271 7683
Price of Peak needed [$/kWh] 0.2620 0.2353 0.2595
Methodologies for estimating CO, SO2 and NOx emission from
Cost of battery [$/kWh Cycled] 0.2616 0.1859 0.2139
power generating plants are immature according to Yan et al. [27]. NPC w/o BES [$] 187,963 187,963 187,963
A recent approach presented by Yan et al. [27] for CO, estimated NPC with BES [$] 209,085 204,456 207,247
that on average, 0.23 kg of CO is emitted per ton coal, Equation (32). NPV Saving [$] 21,122 16,493 19,284
The avoided SO2 and NOx emission were estimated by Equation (33) LCOE w/o BES [$/kWh] 0.0782 0.0782 0.0782
LCOE with BES [$/kWh] 0.0870 0.0851 0.0862
and Equation (34) respectively [27]. In Equation (33), the ratio 64/ %NPV Saving 11.24% 8.77% 10.26%
32 represents SO2 molecular weight with respect to elemental Minimum daily load factor 0.2126 0.2206 0.2324
sulphur molecular weight while in Equation (34), the ratio 30.8/14 Peak to valley ratio 4.7045 4.5323 4.3020
represents NOx molecular weight to elemental nitrogen molecular Reduction in peak [%] 8.16% 11.52% 16.02%
weight. Other parameters have been defined in the nomenclature
list and the constants used is presented in Table 2 according to [27].
The results of the simulation are presented in Table 3. The re-
CSavings sults show that none of the BES yielded any positive return on
COAvoided ¼ 0:23* (32)
1000 investment over the 25 years project life at the current ToU tariff
rates. Of the three, NaS had the highest likelihood of profitability
64 with the least PrElpeak needed . However, the PrElpeak needed was 63%
SO2; Avoided ¼ *C *Sar *ts *ð1  hs Þ (33)
32 Savings higher than the average PrElpeak . As earlier presented, for a system
with BES to be profitable, either the NPCwith BES < NPCwout BES or
30:8 h LCOEwith BES < LCOEwout BES or the calculated
NOx;Avoided ¼ *CSavings *NN * n *ð1  hN Þ (34)
14 g PrElpeak needed < PrElpeak . None of the BES technologies had their
NPC or LCOE below the system without BES nor their
PrElpeak needed below the average selling price of peak of 0.1442
$/kWh. The NPVsavings of the three technologies were higher than
4. Results and discussion that of a system without BES by 11.24%, 8.77% and 10.26% for PbA,
NaS and Li-ion respectively. Considering PbA with a low storage
4.1. Simulation of BES medium cost, but also a low life cycle and low cycling efficiency,
these resulted in high cycling cost 0.2616 $/kWh/cycled, high
After supplying all needed parameters into the developed number of replacement with associated cost (NPCk ) as well as the
spreadsheet, we simulated the operation of the system. Fig. 9 shows high cost of the energy produced (NPCE from batt ). On the other
the mean demand (P_load [kW] No Storage) from the simulated hand, Li-ion had the highest storage medium cost but with a high
data extracted from the mean data logged, as well as charge, cycle life and efficiency which both saw to it low energy cycling
discharge and state of charge profile for PbA. The resulting peak to cost of 0.2139 $/kWh/cycled with high technical benefit of
valley ratio decreased from 5.12 to 4.70, 4.53, and 4.30 for PbA, NaS, reduction in peak yet its PrElpeak needed was higher than the
and Li-ion BES respectively as shown in Table 3. The minimum daily
average PrElpeak by 80% when compared to 63% of NaS and 82% of
load factor also increased from 0.19 to 0.21, 0.22, and 0.23. The
PbA.
increase in load factor, though not significant due to the size of BES
and percentage peak covered, notwithstanding, this indicates bet- Since the deterministic approach is not sufficient to truly assess
the impact of uncertainties in econometric analysis, as explained in
ter utilization of generating capacity. This is particularly important
where demand during off-peak without BES falls significantly section 3.7, a probabilistic approach was considered. Fig. 10 and
below base load. The valley created can be harnessed to charge the Fig. 11 show the mean PrElpeak needed and mean LCOE with 95% error
BES. Also, the reduction in peak to valley ratio due to BES ensures bar taking into consideration uncertainties in input parameters.
power quality stability and improvement such that the generating The 95% error bar indicates the range in which 95% of the results
plant ramping rate to cover short peaks is reduced. generated by the MCS lie. The average selling PrElpeak considering
summer and winter price differential and the LCOE of the system
without BES is represented by the dotted line. For a BES technology
Table 2 with a mean price below the dotted line, such BES is said to be
Environmental benefit Constants.
economically viable. The uncertainty analysis showed that the most
Parameters values financially viable technology is NaS with the least PrElpeak needed of
Sar 1% 0.2560 $/kWh (0.2160 $/kWh/cycled) while PbA is the least
ts 95% economically viable technology as shown in Fig. 10. The
hs 80%
PrElpeak needed for Li-ion and PbA were 0.2795 and 0.2954 $/kWh
NN 32%
hn 1.5% respectively. Factors that aided NaS economic performance were
g 99% the moderately low cost of storage medium compared to Li-ion, its
hN 0.8% high cycle life, and cycling efficiency. After consideration of
 S.O. Masebinu et al. / Energy 135 (2017) 413e429 425

Fig. 9. Charge discharge sequence after implementing BES.

uncertainties, the PrElpeak needed deviated from the deterministic
price by 26.29%, 8.87% and 7.56% with a wide range of lower and
upper bound for PbA, NaS and Li-ion batteries respectively as
shown in Fig. 10. The lower and upper bound for each BES tech-
nology shows that the impact of uncertainties in cost and technical
data exceed by far the differences in the mean PrElpeak needed across
technologies. Furthermore, the LCOE in Fig. 11 shows that despite
the variation in PrElpeak needed , a strong competition exists between
NaS and Li-ion BES for peak shaving application as the degree of
variation was minimal in both BES technologies.
4.2. Sensitivity analysis
In the past two years, an annual increase in the price of elec-
tricity has been observed. Both peak and off peak prices are affected
by the increase. Since none of the three BES technologies is
economically viable based on their current technical performance
characteristics and average PrElpeak , we investigated varying
PrElpeak as well a reduction in the cost of the storage medium on the
profitability of using BES. Sensitivity analysis was conducted within
the MCS framework to find the combination of the economic pa-
rameters within control of policy makers to make the system

Fig. 10. Price of peak needed compared to the average price of electricity.
426 S.O. Masebinu et al. / Energy 135 (2017) 413e429

Fig. 11. LCOE with BES compared to LCOE without BES.

profitable. The results reported are the mean values after 10,000
iterations of the MCS for each of the eleven scenarios considered.
The scenarios are presented below;
A - 25% increase in both summer and winter peak prices
B - 50% increase in both summer and winter peak prices
C - 100% increase in both summer and winter prices
D - 25% decrease in cost of storage medium only
E - 50% decrease in cost of storage medium only
F - 25% decrease in cost of storage medium þ50% increase
both peak prices
G - 25% decrease in cost of storage medium þ100% increase
both peak prices
H - 50% decrease in cost of storage medium þ50% increase
both peak prices
I - 50% decrease in cost of storage medium þ100% increase
summer peak price only
J - 50% decrease in cost of storage medium þ100% increase
summer peak price þ50% increase in winter peak price
K - 50% decrease in cost of storage medium þ100% increase
both summer and winter peak price
In all scenarios where the PrElpeak only was increased without a
decrease in cost of storage, the PrElpeak needed remained the same.
This is valid because only parameters of production cost such as
PrEloffpeak , costk etc. will alter the price of peak needed while the
selling price PrElpeak will only affect the return on investment. At
scenario A, B, and C, increasing the PrElpeak did not lower the price
of peak needed but increased the accrued NPVsavings by reducing the
margin between NPCwout BES and NPCwith BES as shown in Fig. 12. A
in similar trend was also observed for the LCOE in Fig. 13. NPVsavings for
using BES with 100% reduction in the price of peak compared to a
in system without BES increased by 63.68%, 81.26% and 69.92% for
PbA, NaS and Li-ion respectively as shown Fig. 14.
in Aside increasing PrElpeak , consideration was given to reduction
in the cost of storage, scenario D and E. This could be implemented
in through a subsidy programme. Conversely to scenario A, B, and C, a
reduction in cost of storage did not alter the average PrElpeak rather
in it lowered the PrElpeak needed for each BES technology. The response
of these economic indicators to change in parameters are useful as a
in guide to policy makers as two different approaches could be
employed to drive the profitability of the system. At 50% reduction

Fig. 12. Sensitivity analysis on price of peak needed.

 S.O. Masebinu et al. / Energy 135 (2017) 413e429 427

Fig. 13. Sensitivity analysis on LCOE.

Fig. 14. Effect of scenarios on NPV savings.

in cost of the storage medium, the cycled energy cost ($/kWh/cycle)
were 0.1960 for PbA, 0.1077 for NaS and 0.1201 Li-ion BES. NPVsavings
of the system increased by 54.81%, 57.29% and 59.68% compared to
50% increase in PrElpeak with the NPVsavings as 31.89%, 40.57% and
34.97% for PbA, NaS and Li-ion respectively. Cost reduction strongly
favours high technical performing technologies like Li-ion towards
profitability as seen by the 59.68% increase in savings.
Scenario F and G investigated 25% reduction on storage medium
cost with 50% and 100% increase in the PrElpeak . At 100% increase,
the price of peak was 0.2298 $/kWh while PrElpeak needed for PbA,
NaS and Li-ion were 0.2457, 0.2160 and 0.2305 $/kWh. Only NaS
was profitable at this condition with its PrElpeak needed < PrElpeak . At
50% storage medium cost reduction with 50% increase in the price
of peak, scenario H, the BES cycled energy cost were 0.1560 $/kWh/
cycle for the least efficient BES, PbA, 0.1077 $/kWh/cycle for NaS
and 0.1201 $/kWh/cycle for Li-ion. At these cycled costs, the
LCOEwout BES was 0.0904 $/kWh while for NaS with the highest
likelihood of profitability, the LCOE was 0.0905 $/kWh which was
not sufficient to break even as the LCOEwith BES > LCOEwout BES .
Since summer ToU tariff accounts for 273 days of the year, its
long duration impacts on the economics evaluation. We therefore
investigated the impact of increasing the selling price of peak over
summer period by 100% with 50% discount on storage cost, scenario
I. At this condition, only NaS yielded a positive return on invest-
ment with a NPVsavings of 0.2%. Increasing winter PrElpeak by 50%,
scenario J, the three BES technologies yielded a positive return on
investment. Prevailing PrElpeak was 0.2036 $/kWh while the
PrElpeak needed for PbA was 0.1960, NaS was 0.1753 and Li-ion was
0.1808 representing 0.47% (1088 $), 1.69% (3938 $), and 1.28% (2973
$) savings respectively. The IRR at this scenario was 10.06%, 14.1%
and 11.72% for PbA, NaS and Li-ion respectively. At 50% storage cost
reduction and 100% increase in both price of peaks, scenario k,
PrElpeak increased to 0.2298 $/kWh resulting in 1.93% (4745 $),
3.02% (7457 $) and 2.56% (6301 $) savings with IRR of 12.09%,
16.92% and 14% for PbA, NaS, and Li-ion respectively. From this
analysis, it can be deduced that though the system showed a pos-
itive return on investment, the savings were not quite huge.
Reduction in storage cost and increase in average PrElpeak are
needed for the profitability of implementing BES.
From the sensitivity analysis conducted above, the driving fac-
tors for the economic viability of implemented technologies were
the prevailing cost of storage, the cycle life, cycling efficiency and
cycled cost. This study shows that the present average PrElpeak
0.1442 $/kWh is below the PrElpeak needed to encourage the use of
428 S.O. Masebinu et al. / Energy 135 (2017) 413e429
grid-tied BES. For BES to be economically viable, the cycling cost of
storage medium should be less than 0.1560 $/kWh for PbA, 0.1077
$/kWh for NaS and 0.1201 $/kWh for Li-ion. At these energy cycling
costs, an average PrElpeak of 0.2042 $/kWh is required. A price above
the PrElpeak is needed if the deployment of grid-tied BES is to be
encouraged. Where no discount of the storage medium is provided,
that is at the current technology rate, the PrElpeak needed to break
even were 0.2919 $/kWh, 0.2560 $/kWh and 0.2795 $/kWh for PbA,
NaS and Li-ion respectively. In Denmark and Germany with rela-
tively developed ES markets, the average price of electricity is
0.3485 $/kWh in Denmark and 0.3382 $/kWh in Germany. The
average cost of electricity across Europe is 0.2422 $/kWh [59]. This
shows that the present PrElpeak is relatively cheap and thus
implementing BES is not economically viable. To support the up-
take of BES as a demand side management solution, thereby
reducing peaks, it might be feasible to support the investment of
consumers on the uptake of storage technologies. Such investment
coupled with high deployment rate will accelerate development
and technology improvement through research, create more
competition within the storage market and eventually drive the
cost of storage downward. This will make storage financially
attractive even without subsidy in the long run. While we strongly
encourage some form of investment on storage cost through sub-
sidy, the scheme should ensure that the peak load being avoided is
well managed in such a way the network benefit. Uncoordinated
charging of storage system can have a negative effect on grid
balancing itself and therefore negate the motive behind this
proposition.
4.3. Other benefits of BES
Peak shaving using BES provides other techno-economic and
environmental benefits. The environmental benefit majorly lies in
the avoided high coal consumption required during the peak period
and the environmental pollution associated. For using BES to shave
40% peak of the student residence under study, the life time benefit
to the grid for power quality management was 8577 $ and for
transmission and distribution deferral was 74,332 $. The lifetime
benefit for the three BES are the same as they are designed to
produce 22.09 kWh of energy irrespective of their construction
chemistry. However, the three technologies offer different envi-
ronmental benefits with Li-ion having the highest environmental
benefit, that is 24.07 ton CSavings , 62.58 ton CO2; Avoided , 5.54 kg
COAvoided , 19.25 kg SO2; Avoided and 3.18 kg NOx; Avoided though it had
the highest emission during manufacturing based on a 12.5 kg
CO2,eq./kg weight of battery. The high net environmental perfor-
mance of Li-ion BES is due to its high cycling efficiency, as it re-
quires less energy from the grid to charge the battery to the
required capacity. On the other hand, PbA battery had the least
environmental benefit as shown in Table 4. The environmental
impact of disposal of the BES was not considered in this study. This
Table 4
Technical and environmental Benefit of peak shaving.
PbA NaS
Elquality ($) 8577 8577
Gridcap_ut ($) 74,332 74,332
Coal Required (ton) 0.31 0.07
Coal Saving (ton) 7.11 16.37
Reduced CO2 emission (ton) 18.48 42.57
Reduced CO emission (kg) 1.63 3.77
Reduced SO2 emission (kg) 5.69 13.10
Reduced NOx emission (kg) 0.94 2.16
is because the disposal of BES can account as benefit derived from
recycling the BES. This cost/benefit for the end of life of the BES vary
depending on the location where the disposal is performed. We
therefore assume that the cost benefit of disposing the BES negates
the environmental cost to the consumer.
5. Conclusion
The technical and economic performance of three BES, PbA, NaS
and Li-ion batteries was investigated under a South African ToU
tariff rate. The study concluded that the prevailing average PrElpeak
is not favourable for the deployment of BES to take advantage of
energy arbitrage. The study proposed a minimum average PrElpeak
of 0.2043 $/kWh with cycling cost of 0.1560 $/kWh for PbA, 0.1077
$/kWh for NaS and 0.1201 $/kWh for Li-ion. The proposed costs are
the minimum if the BES system is to break even during it life time,
however, a PrElpeak above the minimum and energy cycling cost
below the proposed are desirable. Assessing the environmental
performance of the system, Li-ion battery yielded superior envi-
ronmental performance due to high round trip efficiency though
lagged on economic evaluation where NaS leads. Considering both
the technical and environmental performance of the three tech-
nologies, a close competition between NaS and Li-ion exist but Li-
ion leads on environmental consideration. PbA will compete
closely with the other two technologies if its cycle life and effi-
ciency are improved upon.
The constant price of peak needed for scenarios A, B and C
suggests that increasing the PrElpeak only was not sufficient to move
the system into profitability but lowering PrEloffpeak or securing a
cheaper source of charging the BES might be required to further
improve on the prospect of profitability. The option of lowering
PrEloffpeak , which might not suffice as a policy tool, calls for
consideration of alternative sources of energy which might be
cheaper to charge the BES and also meet load demand during peak.
For this, the authors recommend considerations of solar PV and
wind energy. The current rate of technology improvement of these
RE sources, reduction in the cost of acquiring them and increase in
PrElpeak in South Africa suggest that grid parity between conven-
tional and RE sources is imminent.
Acknowledgement
The authors wish to acknowledge the University of Johannes-
burg-University Research Committee for providing grants for this
research work [UJ-URC-201339837].
References
[1] IEA. Worl energy outlook. France: International Energy Agency; 2013.
[2] Becker B, Fischer D. Promoting renewable electricity generation in emerging
economies. Energy Policy 5//2013;56:446e55.
[3] Aboyade AO, Carrier M, Meyer EL, Knoetze H, Go € rgens JF. Slow and pressur-
ized co-pyrolysis of coal and agricultural residues. Energy Convers Manag 1//
2013;65:198e207.
[4] National Treasury. Electricity. Pretoria, South Africa: Department of National
Treasury; 2011.
[5] Pegels A. Renewable energy in South Africa: potentials, barriers and options
for support. Energy Policy May 2010;38:10.
Li-ion
[6] Eskom. Integrated report 2012. Johannesburg: Eskom; 2012.
8577 [7] Eskom. Integrated report 2014. Johannesburg: Eskom; 2014.
74,332 [8] DoE. State of renewable energy in South Africa. Pretoria: Department of En-
1.1 ergy; 2015.
24.07 [9] Yang P, Tang G, Nehorai A. A game-theoretic approach for optimal time-of-use
electricity pricing. IEEE Trans Power Syst 2013;28:9.
62.58
[10] Badtke-Berkow M, Centore M, Mohlin K, Spiller B. Making the most of time-
5.54
variant electricity pricing. Environmental Defense Fund, Environmental
19.25
Defence Fund; 2015.
3.18 [11] Battke B, Schmidt TS, Grosspietsch D, Hoffmann VH. A review and probabi-
listic model of lifecycle costs of stationary batteries in multiple applications.
 S.O. Masebinu et al. / Energy 135 (2017) 413e429
Renew Sustain Energy Rev 9//2013;25:240e50.
[12] Palizban O, Kauhaniemi K. Energy storage systems in modern gridsdmatrix of
technologies and applications. J Energy Storage 5//2016;6:248e59.
[13] Aneke M, Wang M. Energy storage technologies and real life applications e a
state of the art review. Appl Energy 10/1/2016;179:350e77.
[14] Zafirakis D, Elmasides C, Sauer DU, Leuthold M, Merei G, Kaldellis JK, et al. The
multiple role of energy storage in the industrial sector: evidence from a Greek
industrial facility. Energy Procedia 01/01 2014;46:178e85.
[15] Lin B, Wu W. Economic viability of battery energy storage and grid strategy: a
special case of China electricity market. Energy 4/1/2017;124:423e34.
[16] Tong S, Fung T, Klein MP, Weisbach DA, Park JW. Demonstration of reusing
electric vehicle battery for solar energy storage and demand side manage-
ment. J Energy Storage 6//2017;11:200e10.
[17] Aguado JA, de la Torre S, Trivin ~ o A. Battery energy storage systems in trans-
mission network expansion planning. Electr Power Syst Res 4//2017;145:
63e72.
[18] Kerdphol T, Qudaih Y, Mitani Y. Optimum battery energy storage system using
PSO considering dynamic demand response for microgrids. Int J Electr Power
& Energy Syst 12//2016;83:58e66.
[19] Kyriakopoulos GL, Arabatzis G. Electrical energy storage systems in electricity
generation: energy policies, innovative technologies, and regulatory regimes.
Renew Sustain Energy Rev 4//2016;56:1044e67.
[20] Barbour E, Wilson IAG, Radcliffe J, Ding Y, Li Y. A review of pumped hydro
energy storage development in significant international electricity markets.
Renew Sustain Energy Rev 8//2016;61:421e32.
[21] Mandelli S, Brivio C, Leonardi M, Colombo E, Molinas M, Park E, et al. The role
of electrical energy storage in sub-Saharan Africa. J Energy Storage 11//
2016;8:287e99.
[22] Langley P. Eskom LSES facility. In: Africa utility week: clean power Africa. Cape
Town, South Africa: Eskom; 2015. p. 20.
[23] Zakeri B, Syri S. Electrical energy storage systems: a comparative life cycle cost
analysis. Renew Sustain Energy Rev 7 Nov 2015;42:28.
[24] Baumann M, Zimmermann B, Dura H, Simon B, Weil M. A comparative
probabilistic economic analysis of selected stationary battery systems for grid
applications. IEEE Electr Power Energy Conf 2013:6.
[25] Kaldellis JK, Zafirakis D, Kavadias K. Techno-economic comparison of energy
storage systems for island autonomous electrical networks. Renew Sustain
Energy Rev 2//2009;13:378e92.
[26] Dufo-Lopez R, Bernal-Agustin JL. Techno-economic analysis fo grid-connected
battery storage. Energy Convers Manag 2015;2015:11.
[27] Yan X, Zhang X, Chen H, Xu Y, Tan C. Techno-economic and social analysis of
energy storage for commercial buildings. Energy Convers Manag 2//2014;78:
125e36.
[28] Gildemeister. CellCube-a storage system based on vanadium redox flow
technology. Germany: gildemeister energy solutions GmbH; 2015.
[29] Safaei H, Keith D. How much bulk energy storage is needed to decarbonize
electricity. Energy Environ Sci 2015;8:9.
[30] Hessami MA, Bowly DR. Economic feasibility and optimization of an energy
storage system for Portland wind farm (Victoria, Australia). Appl Energy
2011;2011:9.
[31] Okundamiya MS, Emagbetere JO, Ogujo EA. Techno-Economic analysis of a
grid-connected hybrid energy system for developing regions. Iran J Energy
Environ 31 Aug 2015;6:12.
[32] Homer Energy. Homer Pro. version 3.7 user manual. 2016. Homer Energy.
[33] Homer Energy. Load noise. 03/04/2016. 2015. Available:, http://usersupport.
homerenergy.com/customer/en/portal/articles/2188767-load-noise.
[34] Boait P, Advani V, Gammon R. Estimation of demand diversity and daily de-
mand profile for off-grid electrification in developing countries. Energy Sus-
tain Dev 12//2015;29:135e41.
[35] Okundamiya MS, Emagbetere JO, Ogujo EA. Evaluation of various global solar
radiation models for Nigeria. Int J Green Energy 2016;13:8.
429
[36] Dufo-Lopez R, Bernal-Agustín JL. Techno-economic analysis of grid-connected
battery storage. Energy Convers Manag 2//2015;91:394e404.
[37] Sustainable. Solar power, online eco store. 1/05/2016. 2016. Available:, http://
www.sustainable.co.za/.
[38] Dufo-Lopez R, Bernal-Agustin JL. Design and control strategies of PV-Diesel
systems using genetic algorithms. Sol Energy 2005;79:14.
[39] Dufo-Lopez R. Optimisation of size and control of grid-connected storage
under real time electricity pricing conditions. Appl Energy 2015;140:14.
[40] Okundamiya MS, Emagbetere JO, Ogujo EA. Techno-Economic analysis of a
grid-connected hybrid energy system for developing regions. Iran J Energy
Environ 31 Aug 2015;6:12.
[41] Dufo-Lopez R, Bernal-Agustin JL. A comparative assessment of net metering
and net billing policies. Study cases for Spain. Energy 2015;84:10.
[42] Darling SB, You F, Veselka T, Velosa A. Assumptions and the levelized cost of
energy for photovoltaics. Energy Environ Sci 2011;4:3133.
[43] Chen H, Cong TN, Yang W, Tan C, Li Y, Ding Y. Progress in electrical energy
storage system: a critical review. Progr. Nat Sci 3/10/2009;19:291e312.
[44] Díaz-Gonza lez F, Sumper A, Gomis-Bellmunt O, Villaf afila-Robles R. A review
of energy storage technologies for wind power applications. Renew Sustain
Energy Rev 5//2012;16:2154e71.
[45] Evans A, Strezov V, Evans TJ. Assessment of utility energy storage options for
increased renewable energy penetration. Renew Sustain Energy Rev 8//
2012;16:4141e7.
[46] Hadjipaschalis I, Poullikkas A, Efthimiou V. Overview of current and future
energy storage technologies for electric power applications. Renew Sustain
Energy Rev 8//2009;13:1513e22.
[47] Ibrahim H, Ilinca A. Techno-economic analysis of different energy storage
technologies. In: Zobaa AF, editor. Energy storage, technology and application.
InTech; 2013. p. 40.
[48] Ibrahim H, Ilinca A, Perron J. Energy storage systemsdcharacteristics and
comparisons. Renew Sustain Energy Rev 6//2008;12:1221e50.
[49] Karellas S, Tzouganatos N. Comparison of the performance of compressed-air
and hydrogen energy storage systems: karpathos island case study. Renew
Sustain Energy Rev 1//2014;29:865e82.
[50] Makansi J, Abboud J. Energy storage: the missing link in the electricity value
chain. Energy Storage Council; 2002.
[51] Parker CD. Leadeacid battery energy-storage systems for electricity supply
networks. J Power Sources 11/30/2001;100:18e28.
[52] Dubarry M, Devie A, Stein K, Tun M, Matsuura M, Rocheleau R. Battery Energy
Storage System battery durability and reliability under electric utility grid
operations: analysis of 3 years of real usage. J Power Sources 1/15/2017;338:
65e73.
[53] Mahela OP, Shaik AG. Power quality improvement in distribution network
using DSTATCOM with battery energy storage system. Int J Electr Power &
Energy Syst 12//2016;83:229e40.
[54] Koohi-Kamali S, Rahim NA. Coordinated control of smart microgrid during
and after islanding operation to prevent under frequency load shedding using
energy storage system. Energy Convers Manag 11/1/2016;127:623e46.
[55] Bakar NNA, Hassan MY, Sulaima MF, Mohd Nasir MNI, Khamis A. Microgrid
and load shedding scheme during islanded mode: a review. Renew Sustain
Energy Rev 5//2017;71:161e9.
[56] Schoenung SM, Eyer J. Benefit/Cost framework for evaluating modular energy
storage. California: Sandia National Laboratories; 2008.
[57] Nkomo JC. Energy and economic development: challenges for South Africa.
J Energy South Afr 2005;16:11.
[58] McManus MC. Environmental consequences of the use of batteries in low
carbon systems: the impact of battery production. Appl Energy 5//2012;93:
288e95.
[59] Eurostat. Electricity price statistic. 22/06/2016. 2016. Available:, http://ec.
europa.eu/eurostat/statistics-explained/index.php/Electricity_price_statistics.