Applied Energy 126 (2014) 142–150

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Economic impacts of small-scale own generating and storage units,

and electric vehicles under different demand response strategies
for smart households
Ozan Erdinc ⇑
Department of Electrical and Electronics Engineering, Faculty of Engineering and Architecture, Arel University, Tepekent Campus, Buyukcekmece, Istanbul 34537, Turkey

h i g h l i g h t s

 A MILP framework based HEM system modeling.

 Evaluation of different demand response strategies.
 Consideration of additional own generating unit, EV and ESS.
 Control of shiftable appliances further to reduce total daily cost.

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

Article history: With the increasing importance given to smart grid solutions in end-user premises, demand response
Received 5 March 2014 (DR) strategies applied to smart households are important topics from both real time application and aca-
Received in revised form 31 March 2014 demic theoretic analysis perspectives, recently. In this study, a mixed-integer linear programming (MILP)
Accepted 3 April 2014
framework based evaluation of such a smart household is provided. Electric vehicles (EVs) with bi-direc-
Available online 8 May 2014
tional power flow capability via charging and V2H operating modes, energy storage systems (ESSs) with
peak clipping and valley filling opportunity and a small scale distributed generation (DG) unit enabling
Keywords:
energy sell back to grid are all considered in the evaluated smart household structure. Different case
Demand response
Electric vehicle
studies including also different DR strategies based on dynamic pricing and peak power limiting are
Energy storage system conducted to evaluate the technical and economic impacts of ESS and DG units. Besides, shiftable loads
Home energy management such as washing machine and dishwasher are also considered in Home Energy Management (HEM)
Smart household system structure for the effective operation of the household. Moreover, a further sensitivity analysis
Vehicle-to-home is realized in order to discuss the impact of ESS and DG sizing on daily cost of smart household operation
considering further pros and cons.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction ing the last decade and notable investments have been declared for
this concept by leading country governments [1,2].
During the recent years, improving the efficiency and effective- During the operation of the existing electric power system
ness of electric energy usage has been a leading concern around structure, a considerable difference occurs between the electricity
the world due to several reasons including the finite sources of usage patterns daily and seasonally. High cost peaking power
conventional fossil fuels, the increasing impacts of global warming, plants are required to be operated during the sharp peaks of daily
the stochastic nature of main renewable energy systems, political and seasonally periods and even these peaks are likely to cause the
impacts of energy dependence/independence, etc. Thus, new con- need of constructing new power plants and upgrading the existing
cepts and ideas have been proposed to achieve this target. Among asset (lines, transformers, etc.) capacities for transmission and dis-
them, smart grid concept has drawn significant attention for the tribution purposes [3]. In order to provide the ability to also control
efficient and effective operation of the electric power system dur- the demand side of the energy balance between generation and
utilization to defer the necessity of new investments, smart grid
⇑ Tel.: +90 212 867 25 00; fax: +90 212 860 04 81. vision enables effectively accommodating all generation and stor-
E-mail addresses: ozanerdinc@arel.edu.tr, ozanerdinc@gmail.com age options with consumer participation in the demand side

http://dx.doi.org/10.1016/j.apenergy.2014.04.010
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
 O. Erdinc / Applied Energy 126 (2014) 142–150 143

Nomenclature

Indices Variables
t period of the day index in time units (h or min) PESS;ch ESS charging power (kW)
t
m shiftable appliance index
PESS;dis
t ESS discharging power (kW)

Parameters PESS;used
t power used to satisfy household load from the ESS
CEESS charging efficiency of the ESS (kW)
CEEV charging efficiency of the EV PEV;ch
t EV charging power (kW)
CRESS charging rate of the ESS (kW per time interval) PEV;dis
t EV discharging power (kW)
CREV charging rate of the EV (kW per time interval) PEV;used power used to satisfy household load from the EV
t
DEESS discharging efficiency of the ESS (kW)
DEEV discharging efficiency of the EV Pgrid power supplied by the grid (kW)
t
DRESS discharging rate of the ESS (kW per time interval)
PPV;sold
t power injected to grid from the PV (kW)
DEEV discharging rate of the EV (kW per time interval)
N1 maximum power that can be drawn from the grid (kW) PPV;used
t power used to satisfy household load from the PV
N2 maximum power that can be sold to the grid (kW) (kW)
Pother household power demand (kW) Pshift
t;m power required by shiftable appliance m (kW)
t
PPV;pro power produced by the PV (kW) Psold
t total power injected to the grid (kW)
t
SOEESS;ini initial state-of-energy of the ESS (kW h) SOEESS
t state-of-energy of the ESS (kW h)
SOEESS;max maximum allowed state-of-energy of the ESS (kW h) SOEEV
t state-of-energy of the EV (kW h)
SOEESS;min minimum allowed state-of-energy of the ESS (kW h) uESS binary variable. 1 if ESS is charging during period t, 0
t
SOEEV;ini initial state-of-energy of the EV (kW h) else
SOEEV;max maximum allowed state-of-energy of the EV (kW h) uEV binary variable. 1 if EV is charging during period t, 0
t
SOEEV;min minimum allowed state-of-energy of the EV (kW h) else
DT number of time intervals in one hour ugrid binary variable. 1 if grid is supplying power during
t
kbuy
t price of energy bought from the grid (cents/kW h) period t, 0 else
ksell
t price of energy sold to the grid (cents/kW h)

[4,5]. Related to the recent attention given to smart grid vision, The demand for non-shiftable appliances should be continuously
smart households that can monitor their use of electricity in supplied especially to sustain the comfort level of the end-users. It is
real-time and act in order to lower their electricity bills have also to be noted that energy efficiency concepts such as preventing the
gained specific importance by the research regarding possible stand-by mode operation of TV units, etc. are out of scope of this dis-
demand side actions. cussion on non-shiftable appliances. On the other hand, the energy
Demand side actions for smart households in a smart grid envi- demand of controllable appliances can be modified. Besides, the
ronment generally focus on demand response (DR) strategies energy demand of shiftable appliances can be totally shifted from
enabling interaction between utility and consumers. DR is a term peak to off-peak hours. The rated power is fixed for non-shiftable
defined as ‘‘changes in electric usage by end-use customers from their and shiftable appliances. However, the power consumption of con-
normal consumption patterns in response to changes in the price of trollable appliances changes between a maximum and minimum
electricity over time, or to incentive payments designed to induce band considering the operating condition of the relevant appliances
lower electricity use at times of high wholesale market prices or when due to existing conditions [3].
system reliability is jeopardized’’ by the US Department of Energy As a recently considered type of end-user appliance, electric
(DOE) and comprises incentive based programs and price based vehicles (EVs) has the potential of offering different pros and cons
programs (time-of-use, critical peak pricing, dynamic pricing, related to the operating mode [9]. For the charging operation of
etc.) [6,7]. EVs, considerable levels of power requirements exist as can obvi-
DR can be considered mature for industry, but is a relatively ously be derived from the example that the charging station power
new concept for residential users corresponding to 41% of electric- level of Chevy Volt – which is even a small-sized EV – is 3.3 kW
ity usage [5,8]. Home Energy Management (HEM) systems and [10]. As a different operating mode possible for EVs, the vehicle-
smart metering infrastructure play a vital role in effectively to-home (V2H) and even vehicle-to-grid (V2G) modes can also
applying DR strategies to residential areas. Generally, HEM sys- contribute to the efficient operation of HEM system significantly.
tems electronically receives the pricing data from the relevant load Using the existing energy in EV battery after returning home for
serving entity (LSE) using the smart metering infrastructure and the clipping of peak hour power demand via V2H mode can be a
aims to provide the most economic operation of home appliances potential application for DR strategies. In this regard, different
together with considering user preferences as seen in Fig. 1. Shift- types of energy storage systems (ESSs) can also provide peak clip-
ing demand from peak to off-peak hours is desired and this ping and valley filling by storing energy during off-peak periods
demand shifting depends on the appliance classification. Thus, and consuming this energy during peak times of general use [3,6].
home appliances can be divided into three categories from the There are many recent studies dealing with DR strategies for the
perspective of HEM [3]: optimum appliance operation of smart households. Li and Hong [3]
proposed an ‘‘user-expected price’’ based DR strategy for a smart
 Non-shiftable appliances (television – TV, refrigerator, etc.) household also including a battery based ESS for the aim of lower-
 Controllable appliances (Heating Ventilation and Air Condition- ing the total electricity cost by charging and discharging the ESS at
ing – HVAC, lightening, etc.) off-peak and peak price periods, respectively. However, the impact
 Shiftable appliances (dishwashers, washing machines, etc.) of including an additional EV load that can also be helpful for peak
144 O. Erdinc / Applied Energy 126 (2014) 142–150

Fig. 1. Block diagram of a fundamental DR strategy for smart households.

clipping in certain periods when EV is at home and the possibility strategies, a distributed small scale renewable energy generation
of an own production facility are not evaluated in Ref. [3]. Zhao system, the V2H capability of an EV together with an additional
et al. [6] considered the HEM strategy based control of a smart ESS. Two shiftable appliances (dishwasher and washing machine)
household including photovoltaic (PV) based own production facil- are also inserted in the HEM strategies. The impacts of EV addi-
ity and availability of EV and ESS. However, V2H and further pos- tional features, distributed generation facilities (a small-scale
sible V2G operating modes of EV are not taken into account in roof-top photovoltaic – PV-system in this study) and ESS utiliza-
Ref. [6]. Restegar et al. [11] developed a smart home load commit- tion are evaluated in terms of consumer electricity bill reduction
ment strategy considering all the possible operating modes of EV performance using a single HEM system formulated in a MILP
and ESS, however neglected the impact of an extra peak power lim- framework as the main contribution of the study. The proposed
iting strategy that is probable to be imposed by LSE as similarly not EMS enables load shifting and ESS and EV operation management
considered also in Refs. [3,6]. Pipattanasomporn et al. [12] and (by charging in off-peak hours and discharging in peak hours).
Kuzlu et al. [13] presented a HEM strategy considering peak power Different case studies are conducted considering also the impact
limiting DR strategy for a smart household, including both smart of different user preferences and ESS and PV sizes on system
appliances and EV charging. Shao et al. [14] also investigated EV operations.
for DR based load shaping of a distribution transformer serving a The paper is organized as follows: Section 2 gives the method-
neighborhood. Refs. [12–14] did not provide an optimum operating ology employed in the study. Afterwards, Section 3 includes the
strategy considering price variability with the aim of obtaining the case studies for evaluating daily DR based operation strategies
lowest daily cost apart from just limiting the peak power drawn for the smart household. Finally, concluding remarks are presented
from the grid by household in certain periods. Matallanas et al. in Section 4.
[15] applied an HEM system based on neural networks with exper-
imental results for a household including PV and ESS. However, the 2. System description and methodology
impacts of varying price as well as other types of DR strategies are
not evaluated in Ref. [15]. Angelis et al. [16] performed the evalu- The objective of the proposed HEM strategy is to minimize the
ation of a HEM strategy considering the electrical and thermal con- total daily cost of electricity consumption. The cost is the differ-
straints imposed by the overall power balance and consumer ence between the energy bought from the grid and the energy sold
preferences. Chen et al. [17] provided an appliance scheduling in back to the grid by the household-owned assets that are able to
a smart home considering dynamic prices and appliance usage pat- provide energy (PV in this study). The price variables are time
terns of consumer. Missaoui et al. [18] also provided a smart build- dependent, a fact that provides possibility to imply time varying
ing energy management strategy based on price variations and prices for both bought and sold energy.
external conditions as well as comfort requirements. The pricing !
data based energy management is also suggested by Hu and Li X Pgrid buy Psold sell
t
[19] together with a hardware demonstration. These papers Minimize TC ¼  k  t  kt ð1Þ
t
DT t DT
together with many other studies not referred here have provided
valuable contributions to the existing literature on the application where TC represents the total cost of daily operation. The
of smart grid concepts in household areas. However, many of the constraints presented below constitute the main body of the HEM
mentioned papers failed to address distributed renewable energy system operation. The model can be easily extended and adapted
contribution to reduce load demand on utility side, V2H option to other more specific implementations (e.g. by further modeling
of EV to lower the peak demand periods, and a possible ESS specific smart-appliances such as HVAC, water heaters, appliances
together with different DR strategies in a single study. with cycling operation and/or customer’s contract details). Any time
In this paper a mixed-integer linear programming (MILP) model granularity can be used simply by selecting the appropriate DT. For
of the HEM structure is provided to investigate a collaborative instance, the DT coefficient must be 4 for 15-min interval used in
evaluation of dynamic-pricing and power limiting based DR this study, as one hour comprises four 15-min intervals.
 O. Erdinc / Applied Energy 126 (2014) 142–150 145

2.1. Power balance SOEEV

t  SOEEV;max ; 8t ð15Þ

Pgrid þ PPV;used þ PEV;used þ PESS;used SOEEV

t  SOEEV;min ; 8t ð16Þ
t t t t
other EV;ch ESS;ch
X shift
¼ Pt þ Pt þ Pt þ Pt;m ; 8t ð2Þ
SOEEV EV;max
m t ¼ SOE ; if t ¼ t 1 ð17Þ

Eq. (2) states that the load consisting of the shiftable appliance SOEEV EV;min
t ¼ SOE ; if t ¼ t 2 ð18Þ
load, other part of residential load, the charging needs of the EV
and the ESS is either satisfied by the grid or by the combined pro- Eq. (10) enforces the fact that the actual power provided by the
curement of energy by the PV, the ESS and the EV. It is to be noted EV discharge can be used to cover a portion of the household
that the power supply of a shiftable appliance is considered unin- needs. Constraints (11) and (12) impose a limit on the charging
terruptible when the process starts for this appliance till the pro- and discharging power of the EV. The idle EV state can be described
cess ends. For example, if the washing machine is started, this by any of these constraints by the time the respective power
operation cannot be interrupted until the time of washing process variable is allowed to have zero value. Eqs. (13)–(17) describe
is finished and accordingly power supply should continously be the state-of-energy of the EV. Constraint (13) forces the state-
provided within this period. of-energy at every interval to have the value that it had at the pre-
vious interval plus the actual amount of energy that is transferred
2.2. ESS modeling to the battery if it is charging at that interval minus the energy that
is subtracted if the battery is discharging during that interval. At
the beginning of the time horizon the state-of-energy of the EV
PESS;used
t ¼ PESS;dis
t  DEESS ; 8t ð3Þ coincides with the initial state-of-energy of the EV, as described
by Eq. (14). Constraint (15) limits the state-of-energy of the EV bat-
PESS;ch
t  CRESS  uESS
t ; 8t ð4Þ tery to be less than its capacity. Similarly, constraint (16) prevents
  the deep discharge of the EV battery by imposing a least state-
PESS;dis
t  DRESS  1  uESS
t ; 8t ð5Þ of-energy limit. Finally, Eqs. (17) and (18) represent the option of
having the EV battery fully charged or discharged at the least
PESS;ch P ESS;dis state-of-energy at pre-selected time intervals.
SOEESS
t ¼ SOEESS
t1 þ CEESS 
t
 t ; 8t  1 ð6Þ
DT DT
2.4. PV modeling
SOEESS
t ¼ SOEESS;ini ; if t ¼ 1 ð7Þ
PPV;used
t þ PPV;sold
t ¼ PPV;pro
t ; 8t ð19Þ
SOEESS
t  SOEESS;max ; 8t ð8Þ
Similarly to Eqs. (3), (10), and (19) enforces the fact that the
SOEESS  SOEESS;min ; 8t ð9Þ actual power provided by the PV can be used to cover the house-
t
hold needs or injected back to the grid.
Eq. (3) enforces the fact that the actual power provided by the
ESS discharge can be used to cover a portion the household needs. 2.5. Total power injected to the grid
Constraints (4) and (5) impose a limit on the charging and dis-
charging power of the ESS. The idle ESS state can be described by
any of these constraints by the time the respective power variable Psold
t ¼ PPV;sold
t ; 8t ð20Þ
is allowed to have zero value. Eqs. (6)–(9) describe the state-of- The total amount of power injected to the grid consists of the
energy of the ESS. Constraint (6) forces the state-of-energy at every amount of power provided by the PV which is enforced by Eq.
interval to have the value that it had at the previous interval plus (20). The formulation in Eqs. (3), (10), and (20) can easily be
the actual amount of energy that is transferred to the battery if it is extended to cover possible power injection capabilities of EV and
charging at that interval minus the energy that is subtracted if the ESS to grid if EV and ESS are also considered to be capable of selling
battery is discharging during that interval. At the beginning of the back energy.
time horizon the state-of-energy of the ESS coincides with the ini-
tial state-of-energy of the ESS, as described by Eq. (7). Constraint 2.6. Power transaction restrictions
(8) limits the state-of-energy of the battery to be less than the
ESS capacity. Similarly, constraint (9) prevents the deep discharge
of the battery by imposing a least state-of-energy limit. Pgrid
t  N 1  ugrid
t ; 8t ð21Þ

2.3. EV modeling Psold

t  N2  ð1  ugrid
t Þ; 8t ð22Þ
Eqs. (21) and (22) implement the logic of power exchange. If
PEV;used
t ¼ PEV;dis
t  DEEV ; 8t ð10Þ power from the grid is needed to be drawn, then it is not possible
to inject power back to grid. The reverse case is also described by
PEV;ch
t  CREV  uEV
t ; 8t ð11Þ these constraints. N1 can be a time-dependent parameter or con-
stant positive integer value that imposes a limitation on the power
 
PEV;dis
t  DREV  1  uEV
t ; 8t ð12Þ that can be drawn from the grid. This limitation represents a
restriction posed by the aggregator or the responsible LSE for the
end-user electrification in order to face the situation where in its
P EV;ch PEV;dis
SOEEV EV
t ¼ SOEt1 þ CEEV 
t
 t ; 8t > 1 ð13Þ control area exist multiple households that own HEM system.
DT D
The implementation of a time-varying peak power drawn from
the grid limit as a different DR strategy can be easily adapted on
SOEEV
t ¼ SOE
EV;ini
; if t ¼ 1 ð14Þ
this formulation, by providing N1 as a time-dependent parameter
146 O. Erdinc / Applied Energy 126 (2014) 142–150

as considered in this study. Similarly, N2 imposes a limit on the

power that can be injected back to the grid and also can be
replaced by a time-dependent parameter.
The provided formulation can easily be adapted to any kind of
power flow and energy trading possibilities with minor additions
to relevant basis given in this study.

3. Test and results

To evaluate the total impact of different case studies in house-

hold operation on consumer electricity bills, the MILP model is
tested in GAMS v.24.1.3. [20] using the solver CPLEX v.12 [21]
and the relevant obtained results are discussed in this section. Fig. 2. The power consumption profile of the household.
The household load demand is provided considering power val-
ues of real household appliances given in Ref. [22] for a smart
home demonstration project. The relevant appliance data with
their considered usage periods within the evaluation period are
presented in Table 1. It is to be noted that washing machine and
dishwasher are considered to be operated by HEM system consid-
ering user preferences. The obtained relevant total household
demand variation apart from these two shiftable loads and EV is
given in Fig. 2. Besides, Fig. 3 shows a real-time measured power
production profile of a 520 W roof-top PV system on 26.02.2013.
This power production profile is adjusted to relevant different
power ratings for PV system that will be considered in the follow-
ing parts of the study.
A bi-directional EV operation including V2H (meaning that a
portion of the energy stored in EV battery is used to partly cover
the household load) option is considered. The specifications of a
Chevy Volt with a battery rating of 16 kW h is taken into account. Fig. 3. PV system power production curve.
The Chevy Volt is employed with a charging station limited to a
charging power of 3.3 kW [10]. The same power limit is also
assumed to be valid for the discharging operation in V2H mode. Integrating the two-way energy transactions between the end-
The charging and discharging efficiencies are considered 0.95. It user and the utility, the net-metering approach is employed. When
is also considered that the initial EV battery energy is 8 kW h the available energy from the household-owned resources is suffi-
(50% state-of-energy) when the owner arrives at home and the cient to cover the total of the needs, the excess of energy can be
lower limit of EV state-of-energy is restricted to 4.8 kW h (30% sold back to the grid and vice versa. For pricing the bought energy
state-of-energy). from the grid, a dynamic pricing based DR scheme is considered.
The following assumptions hold for the ESS; the charging and The time-varying price signal available for the consumer via the
discharging rate per 15 min is assumed to be 1/5 of its maximum smart meter is shown in Fig. 4 [23]. Besides, a flat rate of 4 cents/
energy capacity. Its initial state-of-energy is 1/2 of the relevant kW h is paid to the end-user for the energy sold-back to the grid.
maximum battery energy capacity and charging/discharging effi- All the values of common parameters mentioned till now for case
ciencies are 0.95. Lastly, the deep-discharging limit of battery studies that will be evaluated below are summarized in Table 2.
based ESS is 1/4 of its maximum energy capacity. In order to evaluate the economic and technical impacts of dis-
tributed generation and ESS availability, controllability of more
Table 1 numbers of shiftable loads and application of different DR strate-
Household appliance data. gies, five different case studies are conducted in this study which
can briefly be described as follows:
Appliance Power (kW) Total usage period (h-min)
Oven 2.4 30 min
Cooker hood 0.225 30 min
Microwave 1.2 0
Refrigerator 1.666 24 h
Iron 2.4 30 min
Toaster 0.8 15 min
Kettle 2 30 min
Hairdryer 1.8 30 min
Telephone 0.005 24 h
TV 0.083 6 h 45 min
Desktop Computer 0.15 2 h 15 min
Air conditioner 1.14 7 h 15 min
Hair straightener 0.055 0
Printer 0.011 0
Lighting 0.1 6 h 15 min
Other (fixed) 0.05 24 h
Washing machinea 1.4 1h
Dishwashera 1.32 30 min
a
The usage period throughout the day is decided by HEM system. Fig. 4. Time-varying dynamic price signal for DR.
 O. Erdinc / Applied Energy 126 (2014) 142–150 147

Table 2
Common values of parameters for case studies.

CEESS(charging efficiency of the ESS) 0.95

CEEV (charging efficiency of the EV) 0.95
CRESS (charging rate of the ESS (kW per time 1/5 of its maximum
interval)) energy capacity
CREV (charging rate of the EV (kW per time 3.3
interval))
DEESS (discharging efficiency of the ESS) 0.95
DEEV (discharging efficiency of the EV) 0.95
DRESS (discharging rate of the ESS (kW per time 1/5 of its maximum
interval)) energy capacity
DREV (discharging rate of the EV (kW per time 3.3
interval))
SOEESS,ini (initial state-of-energy of the ESS 1/2 of its maximum
(kW h)) energy capacity Fig. 5. The variation of power drawn from the grid for Case 1.
SOEESS,max (maximum allowed state-of-energy Varies for each case study
of the ESS (kW h))
SOEESS,min (minimum allowed state-of-energy of 1/4 of its maximum
the ESS (kW h)) energy capacity
SOEEV,ini (initial state-of-energy of the EV
(kW h))
SOEEV,max (maximum allowed state-of-energy of 16
the EV (kW h))
SOEEV,min (minimum allowed state-of-energy of 4.8
the EV (kW h))
DT (number of time intervals in one hour) 4
ksell (price of energy sold to the grid (cents/ 4
t
kW h))

 Case 1: This case is the base case where there is neither a

distributed generation nor ESS. Only price based DR activity is Fig. 6. The variation of EV power via V2H mode for Case 1.
considered. It is again noted that in this case and in all following
case studies, EV is considered to be available after 6 pm (time of
returning back home from work) in all below given conditions
with the possibility of V2H operation and this issue will not
be reminded again for other case studies.
 Case 2: In this case study, there is a PV system of 3 kW available
that can either sell energy back to grid or transfer this energy
for using within the household that is decided by HEM system.
There is not an ESS and only price based DR activity is
considered.
 Case 3: There is an additional battery based ESS of 3 kW h apart
from the available 3 kW PV system. Again, only price based DR
activity is considered.
 Case 4: In this case study, a peak power limiting DR strategy is
also considered in addition to price based DR strategy that is
much likely to be faced in real life as LSE can limit the power Fig. 7. The power consumption variation of EV and shiftable appliances for Case 1.
that be drawn from the grid in certain peak power periods to
avoid more sharp peaks similar to Refs. [12–14]. The ESS and
PV units in Case 3 are still available. This peak power limiting daily cost of 169.4 cents. If the shiftable appliances are assumed to
operation is conducted between 7 pm – 10:45 pm with a peak be uncontrolled by HEM system and are operated manually by the
power limit of 3 kW in this study. end-user as 1 h for washing machine and half hour for dishwasher
 Case 5: Under same assumptions, the peak power limitation by starting from 7 pm, this leads to a total cost of 177.5 cents (4.77% of
LSE is extended to cover all the night from 7 pm to 6:45 am in increase compared to base case) and shows the importance of
this case study due to the reasons that will be discussed below. incorporating shiftable appliances with HEM system.
As the next step, the obtained results concerning with Case 2
The results of Case 1 are presented in Figs. 5–7. The total house- are given in Figs. 8–10. As there is any ESS system and the EV is
hold demand observed at the grid connection point for Case 1 is available after the PV power drops to zero in the sample day, all
presented in Fig. 5. Besides, the power provided by EV via V2H the PV power is used to cover a portion of the household demand
operation mode is shown in Fig. 6. It is obvious that the HEM deci- in the relevant periods and all the excess energy is injected back to
des to use the available energy within the EV battery till the deep grid as seen in Fig. 8. The final power variation drawn from the grid
discharging limit at peak price periods to lower daily energy cost is also seen in Fig. 8. Besides, the EV based power provided for cov-
and re-charges the EV during lower price periods to the maximum ering a portion of the household demand similar to Case 1 is pre-
capacity with the EV charging power seen in Fig. 7. It can also sented in Fig. 9 together with the PV power used within the
clearly be seen from Fig. 7 that the HEM system shifts the use of household. As expected, EV system again provides power down
dishwasher and washing machine to the lowest price periods to its deep discharging limit in peak price periods and charged
around 4 am with the additional demand increase in these periods by the grid power in lower price periods as seen in Fig. 10. The
compared to Fig. 2. Such an operation in base case leads to a total HEM algorithm also shifts the use of shiftable appliances to the
148 O. Erdinc / Applied Energy 126 (2014) 142–150

Fig. 8. The variation of power drawn from the grid and injected back to grid for Fig. 11. The variation of power drawn from the grid and injected back to grid for
Case 2. Case 3.

Fig. 9. The variation of EV power via V2H mode and PV power used for Case 2. Fig. 12. The variation of EV power via V2H mode, ESS power and PV power used for
Case 3.

Fig. 10. The power consumption variation of EV and shiftable appliances for Case 2.
Fig. 13. The power consumption variation of EV, ESS and shiftable appliances for
Case 3.

same time intervals of Case 1 as also seen in Fig. 10. This case study
results with a daily cost of 116.8 cents with a significant reduction reduction of 33.71% compared to base case is obtained for Case 3.
of 31.05% compared to base case. This issue is obtained by the This slight reduction in the cost compared to Case 2 is obtained
availability of a power generating unit that can cover some part by the additional ESS availability to further help covering more
of household demand and sell energy beyond this. of the demand in peak price periods with charging–discharging
Case 3 with the additional availability of an ESS unit provides cycles related to varying prices.
the results shown in Figs. 11–13. The power in grid connection As another type of applicable DR strategy, peak power limiting
point in grid-to-home way (Grid) and home-to-grid way (PV_sold) in certain periods is also considered in Cases 4 and 5 as stated
are presented in Fig. 11. Here, there is a small increase in the before. The first peak power limitation in Case 4 for regular well-
demand early in the morning due to the decision of HEM to charge known peak power periods in the evening provides the results
the battery based ESS (BAT_T) to further use in peak price periods shown in Figs. 14–16. As seen, the power drawn from the grid is
(BAT-used) as seen in Figs. 12 and 13. The results relevant to EV bi- now more shifted to lower price periods as seen in Fig. 14. There
directional operation and control of shiftable appliances are similar are also some periods that there is some more power drawn from
to Case 1 and 2 as seen from Figs. 12 and 13 compared to Figs. 6, 7, the grid for charging ESS to be ready for covering a greater portion
9 and 10 except for some periods after midnight that the available of the household load demand where grid power availability is
energy in ESS is also used for helping to fully charge EV before the restricted by LSE. These extra charging–discharging cycles are
end of the horizon. A total daily cost of 112.3 cents with a obviously seen in Figs. 15 and 16. Such an extra power limitation
 O. Erdinc / Applied Energy 126 (2014) 142–150 149

without selling back to grid. This is due to the fact that ESS should
be ready for covering as much power as possible during evening
conditions where peak power limiting starts. In later hours of peak
power limiting, ESS and EV have charging and discharging cycles
for reaching all the targets of meeting the household demand
and having a fully charged EV as seen in Figs. 18 and 19. Also, it
is obvious that as the hours of new peaks are now restricted in this
test case, the use of shiftable appliances are moved to sooner hours
with higher prices as peak power limits cannot satisfy now both EV
charging and shiftable load utilization simultaneously. The men-
tioned case study results in a daily cost of 142.9 cents, which is
considerable higher compared to Case 4 due to the mentioned rea-
sons of necessity for moving the use of some appliances to peak
Fig. 14. The variation of power drawn from the grid and injected back to grid for price periods. For the five case studies presented in this study, all
Case 4.
the relevant results are summarized in Table 3. It is to be noted
that, the algorithm takes 0.13 s to solve the problem for Case-3

Fig. 15. The variation of EV power via V2H mode, ESS power and PV power used for
Case 4.
Fig. 17. The variation of power drawn from the grid and injected back to grid for
Case 5.

Fig. 16. The power consumption variation of EV, ESS and shiftable appliances for
Case 4.
Fig. 18. The variation of EV power via V2H mode, ESS power and PV power used for
Case 5.
based DR leads to a total cost of 113.1 cents, which is nearly the
same with Case 3. As the HEM system already uses the available
energy in EV battery during peak hours and shifts the most of
the power utilization to off-peak price hours in Case 3, a further
peak power limiting during peak hours did not make a significant
change in power utilization pattern and accordingly total daily
price.
All the case studies till now provide new significant peaks in
former off-peak periods. As the HEM system automatically shifts
the use of controllable and shiftable appliances to lower price peri-
ods, this is likely to happen in real life conditions. Thus, LSE may
have to take further limitations for these possible new peak
demand periods in future widespread deployment of such smart
grid based operations. To test further extension of DR based peak
power limiting operation, Case 5 is evaluated with the relevant
results given in Figs. 17–19. An interesting result seen in Figs. 17 Fig. 19. The power consumption variation of EV, ESS and shiftable appliances for
and 18 is that all PV power is utilized within the household Case 5.
150 O. Erdinc / Applied Energy 126 (2014) 142–150

Table 3 For this issue, different peak power limiting DR strategies are also
Comparison of different case studies. discussed in shorter and longer hours of power restriction and the
Case study Total daily price Cost significant impact of the chosen strategy directly on costs is clearly
(cents) reduction (%) presented. Moreover, a sensitivity analysis related to PV and ESS
Case 1 (with shiftable appliances) 169.4 Base case sizes is provided and it is further observed that including bigger
Case 1 (without shiftable appliances) 177.5 4.77 PV and ESS units expectedly decreases the total daily cost, which
Case 2 (with PV, without ESS) 116.8 31.05 is a trade-off for the end-user between additional new installation
Case 3 (with PV and ESS) 112.3 33.71
Case 4 (DR peak power limit for 4 h) 113.1 33.23
cost and daily operational cost reduction.
Case 5 (DR peak power limit for all night 142.9 15.64 In a further study, the impact of extra charge–discharge cycles
after 7 pm) related to two-way energy trading on battery life both for ESS
and EV is a point to take into account with pros/cons. Expanding
the HEM strategy to cover other controllable (HVAC, lighting,
etc.) appliances will also be considered in future studies.
Table 4
Sensitivity of daily cost on ESS and PV sizing.
References
ESS size (kW h) PV size (kW)
1 2 3 4 5 [1] Wissner M. The smart grid – a saucerful of secrets? Appl Energy
2011;88:2509–18.
1 150.8 133.6 115.2 95.8 76.0 [2] Siano P. Demand response and smart grids – a survey. Renew Sustain Energy
2 149.3 132.1 113.7 94.3 74.5 Rev 2014;30:461–78.
3 147.8 130.6 112.3 92.9 73.1 [3] Li XH, Hong SH. User-expected price-based demand response algorithm for a
4 146.3 129.2 110.9 91.4 71.6 home-to-grid system. Energy 2014;64:437–49.
5 144.9 127.8 109.4 89.9 70.2 [4] Borlease S. Smart grids: infrastructure, technology and solutions. CRC Press;
2013.
[5] Xue X, Wang S, Sun Y, Xiao F. An interactive power demand management
strategy for facilitating smart grid optimization. Appl Energy
as an example using a Dual Core Laptop with 2 GHz CPU and 8 GB 2014;116:297–310.
RAM, which can give an insight of the operating time required for [6] Zhao J, Kucuksari S, Mazhari E, Son YJ. Integrated analysis of high-penetration
the methodology. PV and PHEV with energy storage and demand response. Appl Energy
2013;112:35–51.
As a further analysis, sensitivity of the daily cost on sizing of ESS
[7] Venkatesan N, Solanki J, Solanki SK. Residential demand response model and
and PV units is evaluated and the relevant results are presented in impact on voltage profile and losses of an electric distribution grid. Appl
Table 4 for 25 different combinations. As expected, increase of both Energy 2012;96:84–91.
[8] Shen B, Ghatikar G, Lei Z, Li J, Wikler G, Martin P. The role of regulatory
PV and ESS sizes provides a decrement in the total daily cost of
reforms, market changes, and technology development to make demand
electricity usage by the increased flexibility for HEM system to response a viable resource in meeting energy challenges. Appl Energy [in
manage the power flow with more capability of peak clipping press].
and valley filling via ESS and power selling back possibility via [9] Weis A, Jaramillo P, Michalek J. Estimating the potential of controlled plug-in
hybrid electric vehicle charging to reduce operational and capacity expansion
PV unit. However, the installation of bigger systems should be costs for electric power systems with high wind penetration. Appl Energy
evaluated in more detail with relevant pros and cons considering 2014;115:190–204.
payback period of this installation with cost decrement, etc. The [10] GM Chevy Volt specifications. <http://gm-volt.com/full-specifications/.
[11] Restegar M, Firuzabad MF, Aminifar F. Load commitment in a smart home.
dependence of system to PV and ESS sizes can further be modeled Appl Energy 2012;96:45–54.
with an approximation method to examine the impacts of sizes out [12] Pipattanasomporn M, Kuzlu M, Rahman S. An algorithm for intelligent home
of the limits considered in this study to better evaluate the use of energy management and demand response analysis. IEEE Trans Smart Grid
2012;3:2166–73.
greater or smaller units from economical point of view. [13] Kuzlu M, Pipattanasomporn M, Rahman S. Hardware demonstration of a home
energy management system for demand response applications. IEEE Trans
4. Conclusions Smart Grid 2012;3:1704–11.
[14] Shao S, Pipattanasomporn M, Rahman S. Demand response as a load shaping
tool in an intelligent grid with electric vehicles. IEEE Trans Smart Grid
In this study a MILP model of a HEM system is proposed. The 2011;2:624–31.
operation of a smart household that owns a PV, an ESS that consists [15] Matallanas E, Cagigal MC, Gutierrez A, Huelin FM, Martin EC, Masa D, et al.
Neural network controller for active demand-side management with PV
of a battery bank and also an EV with V2H option is considered. energy in the residential sector. Appl Energy 2012;91:90–7.
Two-way energy exchange is allowed through net metering. The [16] Angelis F, Boaro M, Squartini S, Piazza F, Wei Q. Optimal home energy
energy drawn from the grid has a real-time cost, while the energy management under dynamic electrical and thermal constraints. IEEE Trans Ind
Inf 2013;9:1518–27.
sold back to the grid is considered to be paid a flat rate. This paper [17] Chen X, Wei T, Hu S. Uncertainty-aware household appliance scheduling
makes two basic assumptions. Firstly, the complete real-time pric- considering dynamic electricity pricing in smart home. IEEE Trans Smart Grid
ing signal is known perfectly before the beginning of the off-line 2013;4:932–41.
[18] Missaoui R, Joumaa H, Ploix S, Bacha S. Managing energy smart homes
optimization horizon. Also, the user preferences and consumption according to energy prices: analysis of a building energy management system.
behavior are assumed to be accurately known. Appliance data of a Energy Build 2014;71:155–67.
real smart household and real time measured production data of a [19] Hu Q, Li F. Hardware design of smart home energy management system with
dynamic price response. IEEE Trans Smart Grid 2013;4:1878–87.
roof-top PV system are employed.
[20] General Algebraic Modeling System. <http://www.gams.com>.
Several test cases were examined considering the impacts of PV [21] CPLEX 12 Solver Description. <http://www.gams.com/dd/docs/solvers/
and ESS availability and shiftable load controllability as well as cplex.pdfg.
applying different DR strategies. A cost reduction reaching more [22] Tascikaraoglu A, Uzunoglu M, Tanrioven M, Boynuegri AR, Elma O. Smart grid-
ready concept of a smart home prototype: a demonstration project in YTU. 4th
than 35% is obtained with additional PV and ESS units compared International conference on power engineering, energy and electrical drives
to the base case (Case 1). Also, the impact of HEM strategy on (POWERENG 2013), 13–17 May, 2013, Istanbul, Turkey. p. 1568–73.
the power pattern of a household is significant and accordingly [23] Tsui KM, Chan SC. Demand response optimization for smart home scheduling
under real-time pricing. IEEE Trans Smart Grid 2012;3:1812–21.
new peaks in former off-peak periods are quite likely to occur.