Design Optimization of A Residential PV-Battery Microgrid With A Detailed Battery Lifetime Estimation Model

This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2020.2965894, IEEE
Transactions on Industry Applications
1
Design Optimization of a Residential PV-Battery

Microgrid with a Detailed Battery Lifetime
Estimation Model
Mansour Alramlawi, Student Member, IEEE , Pu Li
Department of Process Optimization
Institute of Automation and Systems Engineering
Ilmenau University of technology
Ilmenau, Germany
mansour.alramlawi@tu-ilmenau.de, pu.li@tu-ilmenau.de
Abstract—The interest in installing PV-based microgrids has sources. As an example, the work in [2] proposed an opti-
increased significantly in the last years due to the urgent need for mal sizing approach for a grid-connected PV-battery system
reducing greenhouse gas emissions and improving the reliability considering different electricity pricing tariffs to decrease the
as well as the quality of power supply, in particular in developing
countries. The design of a microgrid is challenging since multiple annual operation cost of the system. In [3] an optimal sizing
components and complicated operating conditions have to be method for a grid-connected PV-battery system is proposed
taken into account. In this paper, a comprehensive method for to minimize the consumer electricity bill. A techno-economic
optimal design of a class of residential PV-battery microgrids is optimal approach was proposed in [4] to minimize the total
proposed to determine the optimal number of lead-acid batteries net present value of a PV-battery system for a smart household
and PV-panels, the optimal battery bank depth of discharge
(DOD) value, and the optimal tilt angle of the PV-panels. The considering the annual cost of electricity consumption.
aim of the optimization is to minimize the levelized cost of energy From a different perspective, several approaches were pro-
(LCOE) where the limitation of the annual total loss of the power posed to optimize the residential off-grid microgrid sizes to
supply (T LP S) and the operational constraints are considered. decrease the levelized cost of energy and increase the system
In addition, a detailed model for battery lifetime estimation is reliability. For instance, in [5] an optimal design approach
introduced based on the physico-chemical mechanism of the
lead-acid battery. Moreover, the effect of design parameters, was presented to determine the optimal size of a PV-wind-
e.g., the size and DOD of the battery bank on the battery battery system to minimize the annual capital cost while
capacity degradation, the PV-Panel number and battery lifetime considering a specified T LP S value. In addition, the work
on LCOE, is investigated. Computation results demonstrate that in [6] proposed an optimal sizing method for an islanded
the proposed method is capable of optimizing the size of the PV-wind-battery system, where the uncertainty of the PV and
microgrid components and the other design parameters while
satisfying all technical and operational constraints. wind power output as well as the load demand was considered.
In [7] a household PV-battery system was optimized, taking
Index Terms—Microgrid, grid blackout, lead acid battery
model, battery aging, optimal design.
the effect of the load increment on the system reliability into
account. In [8] an optimal sizing approach for a standalone
I. I NTRODUCTION residential microgrid was introduced to minimize the operation
Photovoltaic (PV) energy represents one of the most impor- cost, greenhouse gases emissions and critical dump energy. An
tant clean and cheap energy sources. However, the intermittent optimization algorithm based on the iterative simulation was
behaviour of the generated power from PV-systems imposes proposed in [9] to optimize the PV-array and battery sizes
challenges in its design and operation. Due to the high in a standalone PV system, where a detailed dynamic model
technology development in recent years, it has become feasible of the lead-acid battery was utilized to simulate the battery
to integrate PV-systems with conventional electrical power performance. However, most of the previous studies [2]- [9]
sources to build a local microgrid that is able to supply the did not consider the battery lifetime degradation effect.
required energy in a reliable and cost-effective manner [1]. For Due to high capital cost and short lifetime of a battery
this purpose, it is important to make an optimal design for the storage system (BSS), compared to other components in a
microgrid which is a challenging task since a microgrid usually microgrid, more and more attention has been drawn to opti-
consists of multiple components and its operation conditions mizing its size and operation by utilizing different approaches
are complicated and constrained. for battery lifetime estimation. The work in [10] proposed a
multi-objective optimization approach to minimize T LP S and
Several studies have appeared in recent years to handle the
the cost of the generated energy, where the life loss cost of
problem of residential microgrid optimal sizing with different
the lead-acid battery was considered based on the weighted-
aims and design variables. For grid-connected microgrids,
Ah counting method. In [11] an optimal sizing method for
different approaches were proposed to decrease the consumed
a residential microgrid was proposed, where a linear battery
power cost by increasing the dependency on renewable energy
capacity degradation model was used to estimate the lead-acid
The authors gratefully acknowledge the generous financial support of the battery lifetime degradation by considering only the discharged
German Academic Exchange Service (DAAD). power effect. A two-stage approach was proposed in [12] to
0093-9994 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2020.2965894, IEEE
2
determine the battery size, the DOD value, and the battery system operation cost. Recently, an economic model predictive
lifetime by counting the battery operation cycles. Recently, a controller based operation strategy was proposed in [29] to
bi-level optimization model was introduced in [13] to optimize optimize the performance of a microgrid considering the grid
the capacity of the components in an isolated microgrid. blackout problem. An optimal power dispatch framework was
In [13] the battery lifetime was estimated by considering used in [30] to minimize the operation cost of a grid-connected
the total cumulative transfer energy based on the number microgrid while ensuring the system reliability during the
of charge/discharge cycles. The work in [14] proposed a blackout periods.
simplified method to optimize the size of a grid-connected PV- Based on the preliminary results in [31], this paper presents
battery system to minimize the total cost and meanwhile the an optimal design approach for residential PV-Battery mi-
battery lifetime was considered by using a generalized lifetime crogrids to compute the optimal number of PV-panels and
estimation model for both lithium-ion and lead-acid battery. In batteries, the optimal value of the battery bank depth of
addition, the optimal sizing of different battery technologies discharge, and the optimal tilting angle of the PV-panel. In
was investigated in [15] to reduce the total expansion cost of comparison to the previous studies on optimal design of
a grid-tied microgrid, considering the influence of the battery residential microgrids, the contributions of our work in [31]
depth of discharge and the number of charge/discharge cycles. and its extension in this paper can be summarized as follows
A linear optimization approach was used in [16] to optimize • A new optimal design method for residential microgrids
the battery size and select the battery technology (lead-acid is proposed to minimize LCOE while satisfying the
or lithium-ion) in a PV-battery residential microgrid while required annual T LP S percentage, where battery lifetime
minimizing the battery bank degradation cost. In [17] an and long term daily grid blackout are considered.
optimal design method for an off-grid PV-battery system was • A detailed model for residential microgrids considering
introduced to determine the PV-system and the battery bank practical operational constraints is applied to describe
capacity. Moreover, an electrochemical model was used in [17] the microgrid operation. Meanwhile, a comprehensive
to estimate the lithium-ion battery lifetime. A simplified op- lead-acid battery model that includes the battery current,
timization approach was proposed in [18] to optimize lithium voltage and state of charge performances is applied to
ion battery size in a household PV-battery system, where the describe the battery operation.
discharged power effect on the battery lifetime was considered. • An improved method to calculate the LOCE utilizing
From the above discussions, it can be noticed that lithium- an accurate estimation of the number of lead-acid battery
ion and lead-acid batteries are dominant battery technologies replacements during the microgrid lifetime by consider-
used in microgrid applications. However, due to its low cost, ing the impact of SOC, discharging current, number of
high safety and high recyclability rate, lead-acid battery is cycles, acid stratification and sulfate-crystal structure on
the most commonly used electrochemical energy storage in the battery lifetime.
various fields [19], [20]. Furthermore, the lead-acid battery is • Computation studies are made to reveal the effect of
an effective solution for a wide range of applications such as battery depth of discharge and the number of batteries
back-up power supplies and hybrid PV systems especially in on the battery lifetime.
developing countries [21], which is in line with our aim at The paper is constructed as follows. Section II describes
developing an optimization method to find the optimal size the PV array, the battery model and the grid-tie line model.
of PV-systems working in the area suffering from a daily and Section III presents the lead-acid battery aging model in detail.
long periods of grid blackouts. The microgrid operation algorithm is presented in Section IV.
Due to insufficient energy generation, blackout remains The formulation of the optimization problem and the solution
a recurrence in many developing countries. Scheduled or method are described in Section V. The computation results of
unscheduled blackouts on distribution systems can cause a a case study are given in Section VI. The paper is concluded
major economic loss to the customers [22], [23]. To avoid grid in Section VII.
blackouts or reduce their negative impacts, diesel generators
[24] or batteries [25] are usually employed. Recently, consid- II. M ICROGRID MODEL
erable attention has been paid to microgrids as a cost-effective
A residential PV-battery microgrid with a DC-coupled ar-
solution to the blackouts problem due to its ability to operate
chitecture is studied in this paper, see in Fig. 1. The PV-
both in islanded and in grid-connected modes [26]. Therefore,
array is connected to the system via a DC-charger. The lead-
various methods have been proposed to optimize the design
acid battery bank is charged by an AC-charger from the grid
and operation of microgrids while taking the blackout problem
and a DC-charger from the PV-array, and it is discharged
into account. In [24] an optimal design approach was proposed
via a DC/AC inverter connected to a residential load. In
to optimize the battery bank and the PV-array size in a residen-
addition, controllable switches are used to hold the operational
tial PV-battery system considering scheduled grid blackouts.
constraints in the microgrid. These components are modelled
Another design approach was presented in [27] aiming at
below in detail.
minimizing the capital cost, operation cost, and the cost due
to unserved load during power outages. The proposed work in
[28] aimed to determine the optimal size of the battery bank A. PV-array
in a PV-battery system to ensure a continuous power supply The output power from each PV-cell in the PV-array is
during the grid blackout periods while minimizing the total affected by the received solar irradiance and the ambient
3
Fig. 1: The proposed PV-battery microgrid schematic diagram [32].
temperature [29]. In this paper, an electrical equivalent circuit where Pdisp.pv (t) is the dispatched power from the PV-array.
from a single diode PV-cell with a series resistance [29], [33]
is used to calculate the available power to be dispatched from B. Battery model
a PV-cell Pav.c (t), as follows
For an accurate description of the battery performance, the
Pav.c (t) = F F (t)Voc (t)Isc (t), (1) battery state of charge (SOC) and its voltage Vb (t) need to be
modelled in detail [35]. In this paper, the battery voltage during
where F F (t) is the PV-cell fill factor [29]. The open circuit
charging and discharging is calculated based on a modified
voltage Voc (t) and short circuit current Isc (t) of the PV-cell
Shepherd equation [36] as follows
are given by
Voc (t) = Voc.stc + Kv (Tc (t) − 25) (2) Vb (t) = Voc (t) + Ib (t)Rcell (t) (8)
where
GT (t)
Isc (t) = (Isc.stc + Ki (Tc (t) − 25))

(3)  V0,c − gc (1 − SOC(t)), if Ib (t) > 0
1000
Voc (t) = , (9)
where Voc.stc and Isc.stc are the open circuit voltage and short  V0,d − gd (1 − SOC(t), if Ib (t) < 0
circuit current of the PV-cell under standard test conditions,
respectively, Kv and Ki are the related coefficients, respec- and
tively.
 ρ ρc Mc SOC(t)
 ( CNc + CN Cc −SOC(t) ), if Ib (t) > 0
The PV-cell temperature Tc (t) is affected by the total solar Rcell (t) =
 ( Cρd + ρd Md (1−SOC(t))
irradiance GT (t) and the measured ambient temperature Ta (t), N CN Cd −(1−SOC(t)) ), if Ib (t) < 0.
as follows (10)
N OCT − 20 where Voc (t) is the battery-cell open circuit voltage, Ib (t) is
Tc (t) = Ta (t) + GT (t). (4)
800 the battery cell current which is positive during charging and
It is to note that the total solar irradiance is the summation negative during discharging, V0,c/d is the nominal open circuit
of the direct beam GB , diffused GD and reflected GR solar cell voltage, Rcell (t) is the battery cell internal resistance, and
irradiance [34], arriving onto the PV-panel CN is the nominal battery capacity. Mc/d , ρc/d and gc/d are
parameters with values taken from [35].
GT (t) = GB (t, β) + GD (t, β) + GR (t, β) (5) The relation between charging/discharging power Pch/dis (t)
where β is the PV-panel tilting angle. The total generated of the battery cell and Ib (t) is given by
power from the PV-array Pav.pv depends on the number of PV- p
−Voc (t) + Voc 2 (t) + 4P (t)R
X cell (t)
panels connected in series Ns.p and in parallel Np.p , therefore, Ib (t) = , (11)
2Rcell (t)
Pav.pv (t) = Pav.c (t)Nc.p Ns.p Np.p (6) where PX (t) is positive during charging and negative during
discharging in which the superscript X can be ch/dis.
where Nc.p is the number of PV-cells in the PV-panel. To For an accurate calculation of the battery SOC, the gassing
ensure that the dispatched power from the PV-array not exceed current Igas (t) [37] of the lead acid battery is considered, i.e.
the generated power from the PV-array, following constraints
should be satisfied Z t
1
Pdisp.pv (t) ≤ Pav.pv (t) (7) SOC(t) = SOC(t − 1) + (Ib (t) − Igas (t))dt, (12)
CN 0
4
where CN is the battery cell nominal capacity in Ah. The electrode and reduces the surface area of the electrochemical
gassing current is calculated using Eq. 13 based on the Tafel reaction [40].
approximation approach The remaining (rest) capacity of the battery Crem.b (t) is
modeled as follows
CN
Igas (t) = Igas,0 exp(cu (Vb (t) − Vgas,0 )
100Ah Crem.b (t) = Cdis.0 − Ccorr (t) − Cdeg (t), (18)
+ cT (Tb (t) − Tgas,0 ), (13)
where Cdis.0 is the normalized discharge capacity of the bat-
where Igas,0 , Vgas,0 and Tgas,0 are the normalized gassing tery. Ccorr (t) and Cdeg (t) are the battery capacity decrement
current, voltage, and temperature, respectively, for a 100 Ah due to corrosion and degradation, respectively. It is to note that
nominal battery capacity. cu and cT are the voltage and the battery should be replaced when its remaining capacity is
temperature coefficient, respectively. lower than 80% of the nominal capacity [37] otherwise, the
It is to note that SOC(t) should be kept between the upper rate of battery aging will accelerate and the battery will be
and lower bounds, i.e., subject to a sudden malfunction [41]. The effect of battery
SOCmin ≤ SOC(t) ≤ SOCmax (14) corrosion and degradation is explained below in detail.
where SOCmin = 1 − DOD and SOCmax is the normalized

battery capacity. Moreover, to prevent the simultaneous charg- A. Corrosion
ing and discharging of the battery, the following constraint is Corrosion layers increase the total internal resistance of the
used battery cell. The thickness of the corrosion layer ∆W (t) is
s1 (t) + s2 (t) + s5 (t) ≤ 1 (15) correlated to the active material loss in the battery cell [37].
where s1,2,3 are control signals of the controllable switches, The battery cell capacity loss due to the corrosion layer is
see Fig. 1. expressed as
∆W (t)
Ccorr (t) = Ccorr,max (19)
C. Grid-tie line ∆W max
A microgrid can work in a grid-connected or an islanded where ∆W max is the corrosion layer thickness when the
mode. To describe this aspect, we use a grid-tie line status battery cell reaches its maximum lifetime and ∆W (t) is given
parameter αg and formulate it as follows by

0, if blackout occured
∆W (t − ∆t)
1/0.6
αg (t) = (16)
1, otherwise. ∆W (t) = ks + ∆t, Vcorr < 1.74
ks
It is to note that the dispatched power from the main grid (20)
is limited by the grid-tie line capacity in the grid-connected ∆W (t) = ∆W (t − ∆t) + ks ∆t, Vcorr ≥ 1.74
mode. Therefore, the following constraint should be held (21)
during the operation
where ks is the corrosion speed parameter and Vcorr is the
Pdisp.g (t) ≤ αg (t)Pg.max , (17) corrosion voltage [37].
where Pg.max is the maximum capability of the grid-tie line.
B. Degradation
III. L EAD - ACID BATTERY AGING MODEL
Degradation describes the battery capacity loss due to bat-
Estimating the battery lifetime is an essential issue in energy
tery cycling. The nominal number of cycles is provided in the
system planning and operation. Therefore, many approaches
battery datasheet which is determined under the standard oper-
have been worked out in the literature to estimate the battery
ating condition. However, the battery capacity deterioration is
lifetime [38]. The work in [37] proposed a detailed model for
affected by the real operating SOC and the discharge current
lead-acid battery lifetime estimation based on the operating
values. Therefore, the weighted number of cycles ZW (t) is
conditions of the battery [39]. This model used in this study
used to calculate the battery loss due to degradation as follows
to estimate battery life by considering the main effects of
[37]
decreasing the battery capacity. The corrosion of the positive
electrode and the degradation of the active mass in the ZW (t)
Cdeg (t) = Cdeg,max × exp(−cZ (1 − )) (22)
battery are the two important ageing mechanisms which were 1.6Z0
investigated in [37]. Corrosion has a direct effect on the battery
where Cdeg,max is the capacity at the end of battery lifetime,
cell resistance. Adding a new corrosion coating with weaker
cZ is a parameter (equal to 5.0 [37]) and Z0 is the nominal
conductivity will increase the total internal resistance of the
number of battery cycles given in the datasheet. ZW (t) is the
battery cell and decrease the conductivity between the grid
weighted number of cycles calculated by
and the active material. Meanwhile, battery degradation causes
capacity loss due to changes in the mechanical structure of the Z t
1
electrode. This will lead to a decrease in the porosity of the ZW (t) = |Id (τ )|fSOC (τ )facid (τ )dτ (23)
CN 0
5
where Id (τ ) is the battery discharging current, fSOC describes Algorithm 1: Microgrid operation algorithm.
the effect of SOC and discharging current on the battery Input: GT (t), Ta (t), αg (t).
degradation process expressed as Calculate the available power from the PV-array using
fSOC (t) = 1 + (cSOC,0 the explained model in section II.A;
+ cSOCmin (1 − (SOCmin (t))|t0 )fI (t)(t − t0 )
t Phase I, covering the load
if Pav.pv (t) ≥ Preq.l (t) then
(24)
S3 (t) = 1;
where cSOC0 and cSOCmin are parameters taken from [35]. Pdisp.pv (t) = Preq.l (t);
t
(SOCmin (t))|t0 is the lowest state of charge since the last full else if Pdis.max (t) + Pav.pv (t) ≥ Preq.l (t) then
charge at time t0 . fI (t) is the current factor representing the S3 (t) = 1 & S5 (t) = 1;
effect of the discharged battery current which will be increased Pdisp.pv (t) = Pav.pv (t);
by decreasing the discharging current value. The influence of Pdis (t) = Preq.l (t) − Pdisp.pv (t);
the acid stratification on the weighted number of cycles (see * Update the battery SOC(t), see section II.B;
Eq. 23) is modeled by the factor facid (t) which is calculated * Update the battery aging variables, see section III;
by s else if αg (t) == 1 then
Iref S4 (t) = 1;
facid (t) = 1 + fstrat (25) Pdisp.g (t) = Preq.l (t);
|I(t)|
else
where T LP S(t) = T LP S(t − 1) + 1 ;
Z t end
fstrat (t + 1) = fstrat,0 + (f+ (t) − f− (t)) (26) Phase II, charging the battery bank
0
if SOC(t) ≤ SOCmax & S5 (t) == 0 then
where f+ (t) and f− (t) represent the increment and the decre- if Pav.pv (t) − Pdisp.pv (t) ≥ 0 then
ment in the acid stratification, respectively. S2 (t) = 1;
Pch = min (Pav.pv (t) − Pdisp.pv(t) , Pch.max (t));
IV. M ICROGRID OPERATION STRATEGY else if αg (t) == 1 then
The task of the microgrid is to cover the load demand from S1 (t) = 1;
the available sources according to a predefined operation strat- Pch (t) = Pch.max ;
egy. Therefore, the microgrid works to satisfy the following else
power balance equation at each time step. Pch (t) = 0;
end
Pl.g (t) + Pl.pv (t) + Pdis (t) = Preq.l (t) (27) * Update the battery SOC, see section II.B;
where Pl.g (t) and Pl.pv (t) are the dispatched power from the * Update the battery aging variables, see section III;
grid and the PV-array to cover the load, respectively, Pdis (t) end
is the discharged power from the battery bank, and Preq.l (t)
is the required power by the load.
In this study, the rule-based operation strategy [42] is used A. objective function
to operate the microgrid considering the operation constraints
described in [32]. The operation strategy used is summarized The levelized cost of energy (LCOE) [6] concerns the total
in algorithm 1. In phase I, the aim of the microgrid is to meet capital investment, the maintenance and operation costs of the
the load demand. If the available power from the PV-array is microgrid. Therefore, the objective function is defined as
greater than the required load, it will be used to cover the
load; otherwise, the battery bank and the PV-array will work min LCOE (28)
Uc ,Ui
together to cover the load. If the battery bank is not able to
cover the deficit energy, the power from the main grid will be where Ui = [Npv.m , Nbatt ] and Uc = [DOD] are the
used if it is available; otherwise, the load will not be covered vectors of the integer and the continuous decision variables,
and the T LP S counter will increase. In phase II, if the battery respectively. Since the optimal tilting angle of PV-panels is
needs to be charged, the extra power for the PV-array will be independent of the other design parameters, it is possible to
utilized to charge the battery bank ; otherwise, the battery will calculate its optimal value independently by changing the tilt
be charged from the main grid if it is connected. angle β from 0° to 90°. In this way, the computation time of
solving the whole optimization problem can be reduced.
V. O PTIMIZATION MODEL The LCOE is calculated by dividing the total annual dis-
Our aim of optimal design is to find the optimal number patched power cost (T AP C) by the total dispatched power
of batteries and PV-panels, the DOD value, and the PV- from all sources in the microgrid Pdisp.t , namely
panel tilting angle. Meanwhile, the microgrid should be able
to provide a cost-effective and reliable power supply to the T AP C
LCOE = PTmax . (29)
load.
t=1 Pdisp.t (t)
6
T AP C consists of the annualized capital cost (ACC) as well

as the annualized maintenance and operation cost (AM OC),
i.e.,
T AP C = ACC + AM OC. (30)
The capital recovery factor (CRF ) is used to convert the total

capital cost of the microgrid (T CC) to ACC, as follows
ACC = T CC × CRF (31)
where
i(1 + i)lts
CRF = , (32)
(1 + i)lts − 1
and
T CC = CCpv + CCb.b + Crep.b (33)
where i is the interest rate and lts is the total lifetime of the
system. CCpv and CCb.b are the PV-array and the battery
bank capital cost, respectively. Crep.b is the battery bank
replacement cost expressed as
Nrep.b
X 1
Crep.b = × CCb.b (34)
n=1
(1 + i)n×ltb
where
lts
Nrep.b = (35)
ltb
Fig. 2: Flowchart of the simulation-based optimal design
where ltb is the expected battery lifetime (equal to the time that method.
battery lifetime limit is reached), see section III. The capital
cost of each component is calculated by B. Constraints
In addition to the operational constraints of the PV-array,

CCpv = Npv.m Cpv.m (1 + BOSpv ), (36)
the grid-tie line, and the battery bank discussed in section II,
and an additional constraint is needed to ensure the reliability of
the microgrid.
CCb.b = Nbatt Cbatt (1 + BOSbatt ), (37) In the case when a blackout occurs and the available power
from the battery and the PV-array is not enough to cover the
where Npv.p is the total number of PV-panels in the PV-array load, the load will lose the power supply. Considering this
and Nbatt is the battery number in the battery bank. Cpv.p issue, the annual total loss of power supply (T LP S) [5] can
and Cbatt are the cost of a PV-panel and the cost of a single be calculated as follows
battery, respectively. BOSpv and BOSbatt are the balance of
system cost for PV-panel and battery bank, respectively. The PTmax
Lp (t)
t=1
annual maintenance and operation cost is defined as follows TLPS = × 100 (40)
Tmax
AM OC = Cm.pv + Cm.b + Cdisp.g , (38)
where Tmax equals to 8760 hours and Lp (t) is a binary
where Cm.pv and Cm.b are the yearly maintenance cost of variable, i.e., Lp (t) = 1 when the available power from the
the PV-array and the battery bank, respectively. Cdisp.g is the microgrid is lower than the required load, else Lp (t) = 0.
annual cost of the dispatched power from the main grid which To consider T LP S, the following constraint is added to the
is calculated by optimization problem
TX
max T LP S ≤ T LP Smax (41)
Cdisp.g = Cg Pdisp.g (t)∆t (39)
t=1
where T LP Smax is the maximum allowable value of T LP S
where Cg is the cost of the energy dispatch from the grid per specified by the user.
kW/h and Pdisp.g (t) is the total dispatched power from the In addition, it can take place that the main grid and the
grid at time t. AC/DC inverter are not able to cover the load simultaneously
7
1.5 1.5
1 1
Preq.l (kW)
Preq.l(kW)
0.5 0.5
Winter Spring Summer Fall Winter Spring Summer Fall

0 0
6 12 18 24 6 12 18 24
Time (h) Time (h)
Fig. 3: The considered load profiles for a residential load in Fig. 4: The considered load profiles for a residential load in
weekends [31]. workdays [31].
TABLE I: The Lead-acid battery parameters.
due to the issue of the AC power synchronization. This can
Parameter CN Z0 Cdis,0 V0,c V0,d
be considered by adding the following constraints (Ah) (-) (-) (V) (V)
s3 (t) + s4 (t) ≤ 1 (42) Value 100 750 1.185 2.21 2.13
s5 (t) + s4 (t) ≤ 1 (43)

TABLE II: The PV-panel parameters [46].
The constraint in Eq. (42) prevents the main grid and the
PV-array from covering the load simultaneously and Eq. (43) Parameter Pmax.stc Voc.stc Isc.stc N OCT Kv Ki
(W) (V) (A) ◦C (%/◦ C) (%/◦ C)
avoids covering the load by the main grid and the battery
simultaneously. Value 250 37.6 8.92 46 -0.32 0.05
C. Solution Method TABLE III: Cost model parameters.

In this study, a simulation-based method is used to de- Parameter unit capital cost maintenance cost BOS
termine the optimal design parameters of the microgrid. In PV-array 550 $/kWp 0.5%CCpv 0.5
this method, the simulation is used to evaluate and verify the Battery 150 $/kWh 1%CCb.b 0.1
effect of the design parameters given by the optimizer [43]. As
shown in Fig. 2, each solution from the optimizer is evaluated
based on the microgrid model. At first, the annualized mainte- respectively. Meanwhile, the minimum daily grid blackout is
nance and operation costs are calculated by hourly simulating considered to be 8 hours [32]. It is also noted that in this paper
the system for one-year. Then, the simulation continues until a 1-hour time step is used for the simulation and optimization.
the battery’s remaining capacity is 80 % of the nominal value. Fig. 5 illustrates the operation of the microgrid on one summer
Finally, LCOE is calculated and evaluated by the optimizer. day. It can be seen from Fig. 5 (a) that the PV-array has the
In this study, the genetic algorithm is used as the optimizer first priority to cover the load even if the power from the grid
to find the optimal design parameters because of its ability to is available, see Fig. 5 (b). However, in specific hours when
solve complex optimization problems irrespective of the model all the power sources in the microgrid are not able to cover
of the system [44]. the load, a loss of power supply occurs as seen in Fig. 5 (a) at
hours 5 and 21. In addition, it is clearly seen from Figs. 5 (a)-
VI. RESULTS AND DISCUSSIONS (d) that all operation constraints of the microgrid components
The microgrid studied in [32] is considered as a case study are satisfied.
in this paper. To consider the daily and seasonal variation in The microgrid optimal sizing approach proposed in Section
the residential load profile, four seasonal days in workdays V is utilized to determine the PV-panel number, the battery
and weekends are used, as shown in Fig. 3 and Fig. 4, number and the DOD value of the battery bank in the
respectively. The main parameters of the battery cell are given microgrid shown in Fig. 1. Meanwhile, the optimum tilt angle
in Table I, other parameters can be found in [45]. The PV-panel is taken at which the radiation on the titled PV-panel becomes
parameters are listed in Table II. The capital cost, maintenance the maximum. The isotropic model [47] is used to calculate the
cost, and the balance of system cost of each component are total solar radiation falling on the tilted PV-panel by varying
shown in Table III. Note that the maximum allowed power to the tilt angle β from 0° to 90°, see Fig. 6.
be extracted from the grid-tie is 5 kW and the cost of the The presence of integer decision variables with nonlinear
consumed power from the grid-tie is Cdisp.g = 0.15$/kW h, constraints in the optimization problem leads to a complex
8
2 2400
Annual radiation kWh/m2 /yr

Preq.l (t) Pl.pv (t) Pl.g (t) Pdis (t)
(a)
1.6 2200
Power (Kw)
2000
1.2
Optimal angle
1800
0.8
1600
0.4
1400
0 1200
1 5 9 13 17 21 24
Time (h)
0 10 20 30 40 50 60 70 80 90
PV-panel tilt angle (dgree)
6
P disp.g (t) P av.g (t) (b) Fig. 6: Relation between the PV-panel tilt angle and the falling
5 solar radiation
4 TABLE IV: Optimal design results.
Power (Kw)
3
Parameter Nbatt Npv.m DOD T LP S LCOE
(-) (-) (-) (%) ($/kWh)
2 Value 7 15 0.76 0.99 0.18
0 mixed-integer nonlinear programming (MINLP) problem. In

5 10 15 20
this paper, genetic algorithm is used to solve the optimization
Time (h)
problem under the MATLAB 8.2 environment. It to note that
3.5 the genetic algorithm stops when the average relative change in
Pav.pv (t) Pdisp.pv (t)
3
(c) the generations of the best objective function value is less than
or equal to the specified threshold which is selected here to
2.5
be 10−4 . The optimal design results for the desired T LP S of
Power (Kw)
2 1% are shown in Table IV. It is noteworthy that the optimized

battery capacity has a total nominal capacity of 700Ah and
1.5
the PV-array capacity is 3.75 kWp .
1 The effect of the design parameters is illustrated in Fig.7.
0.5
From Fig. 7 (a) it can be noticed that the number of batteries
has the highest impact on the T LP S. Moreover, it can be
0 seen from Fig. 7(b) that increasing the battery lifetime has a
5 10 15 20
Time (h) considerable effect on the LCOE that cannot be ignored. The
reason behind is that the increase of the battery lifetime will
4 1.2
Pch (t) Pdis (t) SOC (t) (d) decrease the number of battery replacement which decreases
3.5 the replacement cost in LCOE, see Section V-A. The effect
1
3 of the number of PV-panels on the LCOE is illustrated in Fig.
Normalized SOC
0.8 7)(c). It can be seen that increasing the PV-panel number will
Power (Kw)
2.5
decrease LCOE. However, when the number of PV-panels
2 0.6 increases so that the energy generated by the panels becomes
1.5 greater than the needed power to cover the load and to charge
0.4
1
the battery bank, LCOE is increased.
0.2 The battery capacity degradation is calculated using the
0.5
weighted number of cycles which considers the influence of
0 0 SOC, the discharge current rate, and the acid stratification,
2 4 6 8 10 12 14 16 18 20 22 24
Time (h) in addition to the nominal number of cycles stated in the
datasheet. To study the impact of discharging current, the
Fig. 5: The microgrid operation in an arbitrary day. (a) battery lifetime in the microgrid is calculated with different
Required load power and available power to cover load. (b) numbers of batteries. From Fig. 8, it can be seen that
Available power and the dispatched power from the grid. (c) increasing the number of batteries will lead to an increase
Available power and the dispatched power from the PV-array. of the battery lifetime. However, at a specific number of
(d) SOC, charging and discharging power of battery bank batteries, this increment has a negative effect. The reason
behind is that increasing the number of batteries at the same
9
0.45 0.24
24
0.4 21 (a) 0.22
(b )
0.35 18
0.2
LCOE ($)
TLPS (%)
LCOE ($)
0.3 15
0.18
12
0.25
9 0.16
0.2
6
0.14
0.15 3
0.1 0 0.12
1 4 7 10 13 16 19 22 25 2 3 4 5 6 7 8 9 10
Nummber of batteries Battery Liftime (years)
0.3 20 0.2
2.1
18 (c) 0.198 (d)
0.24 16
1.8
14 0.196
TLPS (%)
TLPS (%)
LCOE ($)
LCOE ($)
12
0.18 0.194 1.5
10
8 0.192
0.12 1.2
6
0.19
4
0.9
0.06 2 0.188
0
2 6 10 14 18 22 26 30 34 38 42 46 50 0.186 0.6
10 20 30 40 50 60 70 80 90
Nummber of PV-pannels
PV-panel tilting angle
Fig. 7: (a) Impact of Number of batteries on the LCOE and the T LP S. (b) Impact of Number of PV-panels on the LCOE
and the T LP S. (c) Impact of battery lifetime on the LCOE (d) Impact of PV-panel tilting angel on the LCOE and the
T LP S.
operating conditions will decrease the discharged current computation studies are made to reveal the effect of battery
from each battery cell, which will increase the influence of depth of discharge and the number of batteries on the battery
the current factor in the degradation of the battery capacity lifetime. The extension of this study by using lithium-ion
[37]. batteries for comparison would be a meaningful future work.
It was reported in [37] that the SOC has a high impact on
the battery lifetime. Operating the battery at a low DOD
value leads to an increase of the active mass degradation in
the battery. Here, the effect of battery DOD is illustrated in R EFERENCES
Fig. 9. It can be seen that increasing the DOD value indeed [1] S. S. Thale, R. G. Wandhare, and V. Agarwal, “A novel reconfigurable
decreases the battery lifetime. This effect arises due to the microgrid architecture with renewable energy sources and storage,” IEEE
increased mechanical stress on the active masses and from Transactions on Industry Applications, vol. 51, no. 2, pp. 1805–1816,
March 2015.
the increasing size of the sulfate crystals [37]. [2] L. Zhou, Y. Zhang, X. Lin, C. Li, Z. Cai, and P. Yang, “Optimal
sizing of pv and bess for a smart household considering different price
mechanisms,” IEEE Access, vol. 6, pp. 41 050–41 059, 2018.
[3] R. Khalilpour and A. Vassallo, “Planning and operation scheduling of
VII. C ONCLUSION pv-battery systems: A novel methodology,” Renewable and Sustainable
Energy Reviews, vol. 53, pp. 194–208, 2016.
A new optimal design approach for PV-battery microgrids [4] O. Erdinc, N. G. Paterakis, I. N. Pappi, A. G. Bakirtzis, and J. P.
is proposed in this paper. The proposed approach is able to Catalão, “A new perspective for sizing of distributed generation and
energy storage for smart households under demand response,” Applied
calculate the optimal number of PV-panels and batteries, the Energy, vol. 143, pp. 26–37, 2015.
optimal value of the battery bank depth of discharge, and the [5] H. Yang, Z. Wei, and L. Chengzhi, “Optimal design and techno-
optimal tilting angle of the PV-panel. Meanwhile, the levelized economic analysis of a hybrid solar–wind power generation system,”
Applied Energy, vol. 86, no. 2, pp. 163–169, 2009.
cost of energy is minimized and the required annual total
[6] A. Maleki, M. G. Khajeh, and M. Ameri, “Optimal sizing of a grid
loss of power supply is satisfied. A detailed model for a resi- independent hybrid renewable energy system incorporating resource
dential microgrid considering practical operational constraints uncertainty, and load uncertainty,” International Journal of Electrical
is applied to describe the microgrid operation. Moreover, an Power & Energy Systems, vol. 83, pp. 514–524, 2016.
[7] F. Benavente, A. Lundblad, P. E. Campana, Y. Zhang, S. Cabrera, and
improved method to calculate the LOCE utilizing an accurate G. Lindbergh, “Photovoltaic/battery system sizing for rural electrifica-
estimation of the number of lead-acid battery replacements tion in bolivia: Considering the suppressed demand effect,” Applied
during the microgrid lifetime by considering the impact of energy, vol. 235, pp. 519–528, 2019.
[8] U. Akram, M. Khalid, and S. Shafiq, “An improved optimal sizing
SOC, discharging current, number of cycles, acid stratification methodology for future autonomous residential smart power systems,”
and sulfate-crystal structure on the battery lifetime. Moreover, IEEE Access, vol. 6, pp. 5986–6000, 2018.
10
Normalized batttery capacity

1.2
Battery life limit

0.8
8 batteries 10 batteries 12 batteries 14 batteries 16 batteries 18 batteries

0.6
0 5000 10000 15000 20000 25000 30000 35000 40000 45000
Time (h)
Fig. 8: The impact of battery bank size on battery capacity.

1.2
Normalized batttery capacity
Battery life limit

0.8
DOD = 1 DOD = 0.9 DOD = 0.8 DOD = 0.7 DOD = 0.6 DOD = 0.5 DOD = 0.4
0.6
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 55000
Time (h)
Fig. 9: The impact of depth of discharge on battery capacity.
[9] I. A. Ibrahim, T. Khatib, and A. Mohamed, “Optimal sizing of a and amelioration using battery energy storage systems,” IEEE Transac-
standalone photovoltaic system for remote housing electrification using tions on Industry Applications, vol. 52, no. 1, pp. 5–10, Jan 2016.
numerical algorithm and improved system models,” Energy, vol. 126, [23] A. Esmaeilian and M. Kezunovic, “Prevention of power grid blackouts
pp. 392–403, 2017. using intentional islanding scheme,” IEEE Transactions on Industry
[10] M. I. Hlal, V. K. Ramachandaramurthy, A. Sarhan, A. Pouryekta, and Applications, vol. 53, no. 1, pp. 622–629, Jan 2017.
U. Subramaniam, “Optimum battery depth of discharge for off-grid solar [24] J. Khoury, R. Mbayed, G. Salloum, and E. Monmasson, “Optimal sizing
PV/battery system,” Journal of Energy Storage, vol. 26, p. 100999, 2019. of a residential pv-battery backup for an intermittent primary energy
[11] R. Atia and N. Yamada, “Sizing and analysis of renewable energy and source under realistic constraints,” Energy and Buildings, vol. 105, pp.
battery systems in residential microgrids,” IEEE Transactions on Smart 206–216, 2015.
Grid, vol. 7, no. 3, pp. 1204–1213, 2016. [25] J. Mitra, “Reliability-based sizing of backup storage,” IEEE Transactions
[12] T. M. Masaud and E. El-Saadany, “Correlating optimal size, cycle life on Power Systems, vol. 25, no. 2, pp. 1198–1199, May 2010.
estimation, and technology selection of batteries: A two-stage approach [26] H. F. Habib, M. El Hariri, A. Elsayed, and O. A. Mohammed, “Uti-
for microgrid applications,” IEEE Transactions on Sustainable Energy, lization of supercapacitors in protection schemes for resiliency against
2019. communication outages: A case study on size and cost optimization,”
[13] G. Ma, Z. Cai, P. Xie, P. Liu, S. Xiang, Y. Sun, C. Guo, and G. Dai, “A IEEE Transactions on Industry Applications, vol. 54, no. 4, pp. 3153–
bi-level capacity optimization of an isolated microgrid with load demand 3164, 2018.
management considering load and renewable generation uncertainties,” [27] B. Yan, P. B. Luh, G. Warner, and P. Zhang, “Operation and design
IEEE Access, vol. 7, pp. 83 074–83 087, 2019. optimization of microgrids with renewables,” IEEE Transactions on
[14] S. Barcellona, L. Piegari, V. Musolino, and C. Ballif, “Economic Automation Science and Engineering, vol. 14, no. 2, pp. 573–585, 2017.
viability for residential battery storage systems in grid-connected PV [28] J. Zhou, S. Tsianikas, D. P. Birnie III, and D. W. Coit, “Economic and re-
plants,” IET Renewable Power Generation, vol. 12, no. 2, pp. 135–142, silience benefit analysis of incorporating battery storage to photovoltaic
2017. array generation,” Renewable Energy, vol. 135, pp. 652–662, 2019.
[15] I. Alsaidan, A. Khodaei, and W. Gao, “A comprehensive battery energy [29] M. Alramlawi, E. Mohagheghi, and P. Li, “Predictive active-reactive op-
storage optimal sizing model for microgrid applications,” IEEE Trans- timal power dispatch in pv-battery-diesel microgrid considering reactive
actions on Power Systems, vol. 33, no. 4, pp. 3968–3980, 2018. power and battery lifetime costs,” Solar Energy, vol. 193, pp. 529 – 544,
[16] H. Hesse, R. Martins, P. Musilek, M. Naumann, C. Truong, and 2019.
A. Jossen, “Economic optimization of component sizing for residential [30] S. J. Hossain, T. G. Paul, R. Bisht, A. Suresh, and S. Kamalasadan,
battery storage systems,” Energies, vol. 10, no. 7, p. 835, 2017. “An integrated battery optimal power dispatch architecture for end-user-
[17] M. Astaneh, R. Roshandel, R. Dufo-López, and J. L. Bernal-Agustı́n, driven microgrid in islanded and grid-connected mode of operation,”
“A novel framework for optimization of size and control strategy of IEEE Transactions on Industry Applications, vol. 54, no. 4, pp. 3806–
lithium-ion battery based off-grid renewable energy systems,” Energy 3819, 2018.
conversion and management, vol. 175, pp. 99–111, 2018. [31] M. Alramlawi, Y. Souidi, and P. Li, “Optimal design of pv-battery
[18] S. de la Torre, J. M. González-González, J. A. Aguado, and S. Martı́n, microgrid incorporating lead-acid battery aging model,” in 2019 IEEE
“Optimal battery sizing considering degradation for renewable energy International Conference on Environment and Electrical Engineering
integration,” IET Renewable Power Generation, vol. 13, no. 4, pp. 572– and 2019 IEEE Industrial and Commercial Power Systems Europe
577, 2018. (EEEIC/I&CPS Europe). IEEE, 2019, pp. 1–6.
[19] K. Mongird, V. Fotedar, V. Viswanathan, V. Koritarov, P. Balducci, [32] M. Alramlawi, A. Gabash, E. Mohagheghi, and P. Li, “Optimal operation
B. Hadjerioua, and J. Alam, “Energy Storage Technology and Cost of hybrid pv-battery system considering grid scheduled blackouts and
Characterization Report,” 2019. battery lifetime,” Solar Energy, vol. 161, pp. 125–137, 2018.
[20] S. Few, O. Schmidt, and A. Gambhir, “Energy access through electric- [33] E. Lorenzo, Solar electricity: engineering of photovoltaic systems.
ity storage: Insights from technology providers and market enablers,” Earthscan/James & James, 1994.
Energy for sustainable development, vol. 48, pp. 1–10, 2019. [34] H. Häberlin, Photovoltaics: system design and practice. John Wiley &
[21] J. Yang, C. Hu, H. Wang, K. Yang, J. B. Liu, and H. Yan, “Review on Sons, 2012.
the research of failure modes and mechanism for lead–acid batteries,” [35] J. Sachs and O. Sawodny, “A two-stage model predictive control strategy
International Journal of Energy Research, vol. 41, no. 3, pp. 336–352, for economic diesel-pv-battery island microgrid operation in rural areas,”
2017. IEEE Transactions on Sustainable Energy, vol. 7, no. 3, pp. 903–913,
[22] T. Zhang, S. Cialdea, J. A. Orr, and A. E. Emanuel, “Outage avoidance 2016.
11
[36] C. M. Shepherd, “Design of primary and secondary cells ii. an equation

describing battery discharge,” Journal of the Electrochemical Society,
vol. 112, no. 7, pp. 657–664, 1965.
[37] J. Schiffer, D. U. Sauer, H. Bindner, T. Cronin, P. Lundsager, and
R. Kaiser, “Model prediction for ranking lead-acid batteries according to
expected lifetime in renewable energy systems and autonomous power-
supply systems,” Journal of Power sources, vol. 168, no. 1, pp. 66–78,
2007.
[38] R. Dufo-López, J. M. Lujano-Rojas, and J. L. Bernal-Agustı́n, “Com-
parison of different lead–acid battery lifetime prediction models for use
in simulation of stand-alone photovoltaic systems,” Applied Energy, vol.
115, pp. 242–253, 2014.
[39] M. Mohsin, A. Picot, and P. Maussion, “Lead-acid battery modelling
in perspective of ageing: a review,” in 2019 IEEE 12th International
Symposium on Diagnostics for Electrical Machines, Power Electronics
and Drives (SDEMPED). IEEE, 2019, pp. 425–431.
[40] H. Bindner, T. Cronin, P. Lundsager, J. F. Manwell, U. Abdulwahid, and
I. Baring-Gould, “Lifetime modelling of lead acid batteries,” 2005.
[41] Y. Mekonnen, H. Aburbu, and A. Sarwat, “Life cycle prediction of
Sealed Lead Acid batteries based on a Weibull model,” Journal of Energy
Storage, vol. 18, pp. 467–475, 2018.
[42] G. K. Venayagamoorthy, R. K. Sharma, P. K. Gautam, and A. Ahmadi,
“Dynamic energy management system for a smart microgrid,” IEEE
transactions on neural networks and learning systems, vol. 27, no. 8,
pp. 1643–1656, 2016.
[43] W. Mefteh, “Simulation-based design: Overview about related works,”
Mathematics and Computers in Simulation, vol. 152, pp. 81–97, 2018.
[44] G. Stojanovski and M. Stankovski, “Model predictive controller employ-
ing genetic algorithm optimization of thermal processes with non-convex
constraints,” Genetic Algorithms in Applications, pp. 19–34, 2012.
[45] J. Sachs, Model-based optomization of hybrid energy systems. Stuttgart
University, 2016.
[46] YGE 60 cell series 2, Yingli Solar, December 2019. [Online]. Available:
www.yinglisolar.com
[47] A. K. Yadav and S. Chandel, “Tilt angle optimization to maximize
incident solar radiation: A review,” Renewable and Sustainable Energy
Reviews, vol. 23, pp. 503–513, 2013.

Design Optimization of A Residential PV-Battery Microgrid With A Detailed Battery Lifetime Estimation Model

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Design Optimization of A Residential PV-Battery Microgrid With A Detailed Battery Lifetime Estimation Model

Uploaded by

Copyright:

Available Formats

This article has been accepted for publication in a future issue of this journal, but has not been

Design Optimization of a Residential PV-Battery

Fig. 1: The proposed PV-battery microgrid schematic diagram [32].

where SOCmin = 1 − DOD and SOCmax is the normalized

T AP C consists of the annualized capital cost (ACC) as well

The capital recovery factor (CRF ) is used to convert the total

ACC = T CC × CRF (31)

In addition to the operational constraints of the PV-array,

Winter Spring Summer Fall Winter Spring Summer Fall

s3 (t) + s4 (t) ≤ 1 (42) Value 100 750 1.185 2.21 2.13

s5 (t) + s4 (t) ≤ 1 (43)

C. Solution Method TABLE III: Cost model parameters.

Annual radiation kWh/m2 /yr

0 mixed-integer nonlinear programming (MINLP) problem. In

2 1% are shown in Table IV. It is noteworthy that the optimized

Normalized batttery capacity

Battery life limit

8 batteries 10 batteries 12 batteries 14 batteries 16 batteries 18 batteries

Fig. 8: The impact of battery bank size on battery capacity.

Battery life limit

Fig. 9: The impact of depth of discharge on battery capacity.

[36] C. M. Shepherd, “Design of primary and secondary cells ii. an equation

You might also like