Applied Energy 128 (2014) 254–264

Contents lists available at ScienceDirect

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

Technical and economic impact of residential electricity storage at local

and grid level for Portugal
João M. Santos ⇑, Pedro S. Moura, Aníbal T. de Almeida
ISR – Dep. of Electrical and Computer Engineering, University of Coimbra, Polo II, 3030 Coimbra, Portugal

h i g h l i g h t s

 Residential energy storage simulation with real data of a complete year.

 Three different storage roles defined and simulated with management rules.
 Grid global impact assessed for residential storage massification.
 Storage can reduce cost to 35% of PV alone.
 The residential power flows with the grid can be reduced more than 20%.

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

Article history: The development of the Smart Grids will enable a more interactive and intelligent behaviour of the
Received 16 September 2013 houses that are connected to the grid. This modification of behaviour of the house must rely on a conju-
Received in revised form 12 December 2013 gation of technologies (on-site generation, demand-side management, demand response, etc.) but, in
Accepted 20 April 2014
general, the literature mentions storage as the ultimate response to conjugate generation, demand and
Available online 16 May 2014
grid interaction. In this context, the impact of in-house energy storage capacity deserves to be properly
analysed. This work focuses on the analysis of electricity storage inside the house and its influence on the
Keywords:
grid interaction, ensuring the demand satisfaction. First the typical demand and generation profiles are
Energy storage
Home energy management
analysed to characterise the need for storage. Different roles for the storage utilisation are analysed with
Distributed generation the objective of optimising self-consumption and mitigating the peak power flows from and to the grid.
Renewables integration For the given storage roles, the benefits are evaluated from a local and grid global perspective. The anal-
Zero energy houses ysis is done using data from real profiles of demand and generation, as well as by simulating the existence
of storage. The simulated storage capacity in this analysis has resulted in significant improvements in the
residential energy management. Furthermore, the effect of the simulated storage capacity is strongly
influenced by the sizing and operating strategy.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction interactive behaviour in the grid connection, in contrast to the typ-

The house of the future should progressively become a Nearly
Zero Energy House (ZEH), and, to achieve it, the architecture and
thermal optimisations are not enough, but also improvements in
the electrical systems are needed. The presence of self-generation
in the residential sector is growing and will certainly become an
issue in terms of grid management and reliability. Additionally,
the increasing amount of large scale generation from intermittent
sources available on the grid is already posing difficulties for the
grid management. In this context, the house should achieve a more
⇑ Corresponding author. Tel.: +351 936291236.
E-mail addresses: mirsa@isr.uc.pt (J.M. Santos), pmoura@isr.uc.pt (P.S. Moura),
adealmeida@isr.uc.pt (A.T. de Almeida).
ical passive situation. Since the local generation is highly variable
and intermittent, with a different profile from the demand, and
the local demand has limited flexibility, the distributed local
storage will certainly be an important resource for the house to
balance supply and demand. Without storage, the house uses only
the grid to compensate the generation–consumption mismatch,
causing an unnecessary additional power flow in the grid connec-
tion. At the same time, the cost of the storage technologies is
decreasing, and soon is expected to become economically suitable
for small applications. However, it is important to know how much
storage capacity is needed and how it should be managed to opti-
mise the expected benefits for the grid and for the customer.
The importance and the opportunity for storage in the electrical
system is addressed in several publications [1–6], some of them

http://dx.doi.org/10.1016/j.apenergy.2014.04.054
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
 J.M. Santos et al. / Applied Energy 128 (2014) 254–264
suggesting distributed applications and particularly at residential
level. A more specific evaluation of the use of energy storage in res-
idential buildings has already been done from several perspectives.
In [7] a perspective on the sizing of residential storage is given,
considering the benefit of keeping the grid connection, resulting
in some calculation rules for storage with two different objectives
of covering most of the electricity needs or only the peaks. In [8],
the evaluation of a residential reversible fuel cell Photovoltaic
(PV) storage system found numerous advantages of having the sys-
tem grid-connected instead of stand-alone, but in this case the
management option is to use the grid to supply the residential con-
sumption peaks. A different purpose for storage is exploited in [9]
where a storage system is optimised to take advantage from time-
of-use pricing to obtain cost savings, despite the system cost not
being considered. Another strategy is presented in [10,11] by the
combination of storage with demand side management for
self-consumption enhancement. With a more generic solution
approach it is proposed in [12] an algorithm that could take into
account multiple scenarios, domestic technologies and house con-
figurations. In [13], a technique for scheduling the management of
a residential installation with PV and energy storage is proposed
that assumes a predefined energy price evolution trough the day
and on the references prices defined by the customer. Another
potentially profitable PV storage system is proposed in [14], con-
sidering the provision of ancillary systems to the grid. In [15],
the energy storage is used in a residential distribution feeder to
shave the peak demand and the impact of PV installations at the
house level is also analysed. In a global perspective of the grid,
the storage importance is revealed in [16] where the fraction of
variable generation curtailed is clearly reduced with storage capac-
ity increase. The study in [17] also shows how the particular resi-
dential storage could attenuate the peak injection of electricity into
the grid that could lead to negative impacts in high residential PV
penetrations. An economic assessment of a residential storage sys-
tem with incentives is provided in [18] by quantifying the depen-
dence of the internal rate of return on the battery costs. In [19], the
economical optimal sizing of PV and battery system is determined
as a function of the remuneration policy. An approach to house
energy management including PV and storage together with the
charging load of PHEV is presented in [20] where the impact of dif-
ferent tariffs is also analysed. In [21] the authors have concluded
completely against the financial and environmental benefit of
using lead acid batteries to time-shift the demand in PV houses
with the current feed-in tariff.
The present work addresses the residential storage option with
the aim of demonstrating the importance of defining concrete
objectives for the storage management. The remainder of the paper
is structured as follows. In Section 2 an analysis of the residential
profiles is made. Section 3 is focused on the sizing aspects of the
storage capacity for general typical profiles of demand and gener-
ation. In Section 4 the global impact on the grid is evaluated
considering penetration of PV and storage systems. Section 5 pre-
sents the economic assessment of the storage use. Finally, Section 6
summarises the paper, emphasising its main conclusions.
2. Residential profiles analysis
The role of energy storage in residential buildings is mainly
influenced by the profiles of the local generation and demand, that
the storage system will help to match, and so the first step of such
assessment is to analyse them. The research work about residential
storage found in the literature is made for particular examples of
houses. In most of the cases the data used are from a unique
residential site and for a small time period, like one day
[14,22,13,9,20], one week [23], or for a large time interval like
255
one year or more [24]. In other cases the data from a set of houses
are used for an entire year [7,25,21]. In two of these, the houses in
the set are analysed each one independently and compared among
them. In two situations the data are from different metered houses
and in the other is from a model that simulates different types of
houses. In a similar manner, the generation data are from isolated
systems metered or simulated.
The results presented in the related work for the relationship
between demand and generation profiles are usually biased by
the ratio between total generation and total demand. In order to
perform an analysis independent from this ratio, the absolute
values of demand and generation were not considered, the values
relatively to the annual total being used. In this way, the assess-
ment was focused on the relative intensity variation through time
and the simulations can be performed considering any desired
ratio of total demand/generation. So, all the values of demand, gen-
eration, and storage capacity are given relatively to the annual
demand daily mean (AD). The data used in this assessment were
provided by the Portuguese mainland transmission system opera-
tor (REN). For the demand, the load profile provided to give the
residential consumption with 15 min resolution, when the cus-
tomer meter does not give readings with that resolution, was used
[26]. This profile represents a generic profile for the residential sec-
tor in the entire country. Since this profile is used relatively to the
average daily demand, the total annual demand does not affect the
analysis performed, and so the results are independent from vari-
ations in the house size, typology or income. Since the impact from
generation is fully dependent on the considered generation profile,
this work was done for the situation of residential generation
based on Photovoltaic (PV) technology. The reason for this is that
PV is the most common and suitable generation type for residential
application. The PV generation profile used is based on all the PV
systems that are connected to the grid in the whole system [27].
In order to analyse the cyclic profile of the demand, a histogram
of the demand for every 15 min interval of a day period was built.
The resulting graph for the entire year demand is shown in Fig. 1. It
is visible in the graph a significant coincidence around the same
demand value at the same time of the day. In fact there is, for most
of the day periods, a single frequency peak and the adjacent hills.
This reveals the daily cyclic profile of the demand since, when
considering a day period, the distribution of the demand is very
similar for the entire year. This is mainly found in the late night
and early morning period, while during the day the demand is
more dispersed and especially in the late afternoon and evening
the frequency peak is lower and the dispersion higher. It is also
remarkable that there is a peak in demand happening in a small
2.4
Demand (% annual daily average / 15min)
120
2.2
2 100
1.8
80
1.6
1.4 60
1.2
40
1
0.8
20
0.6
0 4 8 12 16 20 24
Hour of day
Fig. 1. Demand distribution for the entire year counting the number of days with
given demand in each 15 min period.
256 J.M. Santos et al. / Applied Energy 128 (2014) 254–264

portion of the day and in a small number of days while the rest of 3. Storage roles simulation
the demand values are most of the time below half of the maxi-
mum peak. For the PV generation, the daily cyclic profile and the In this section, a simulation approach is proposed to assess the
seasonal variation is shown in Fig. 2. impact of different storage operating strategies in the power flows
To analyse the coincidence between generation and demand, with the grid and in the storage capacity usage. Four different roles
the supply and demand cover factors, as formulated by [28], were for storage are defined and the respective algorithms for storage
used, that are given by operation are described. The simulation inputs are the residential
demand and generation profiles and the outputs are the storage
R t2 R t2
½t1;t2 min fPS ; PD g ½t1;t2 min fPS ; PD g utilisation and the power flow to and from the grid. The storage
t1 t1
CF S ¼ R t2 ; CF D ¼ R t2 ð1Þ is considered to have a round-trip efficiency of 90%. To keep the
t1 S
P t1 D
P
simulation technology agnostic, the storage capacity is considered
effective, that is, the simulation is done considering only the stor-
where PS ðWÞ is the local supply power and PD ðWÞ is the demand
age capacity available to use, and so the minimum State of Charge
power. These cover factors represent how much of the local gener-
(SOC) is, apparently, zero. The input data and the results are for a
ation is used by the demand and how much of the demand is sup-
storage utilisation during an entire year (2011).
plied by the local generation, respectively. In order to have profiles
The algorithm used for the simulation considers that the energy
representing a Net ZEH, the annual total demand is equal to the
ES that is transferred to (or from, if negative) the storage system on
annual total generation, what leads to a CF year and CF year
D of about
S each time slot i is given by
40.2% for the considered case study. However, due to the seasonal (
variation of the demand and generation profiles, the CF for each minðESC  EiSS ; EiSD Þ if EiSD P 0
month vary independently in a inverse way between supply and EiS ¼ ð2Þ
maxðEiSS g ; EiSD Þ if EiSD < 0
demand, as show in Fig. 3. It is important to note that these are
15 min mean values and so do not include instantaneous mis- with ESC being the storage system capacity, ESD the demand for
matches between demand and generation. With these cover factors energy storage, and g the storage efficiency. The stored energy ESS
it can be concluded that, despite the annual total generation and at the beginning of each time slot is given by
demand have the same values, they do not happen exactly at the (
same time in the whole year. More exactly, in an yearly average, Ei1 i1
SS þ ES if Ei1
S P0
EiSS ¼ ð3Þ
60% of the generation must be exported to the grid or wasted, while Ei1 1 i1
SS þ g  ES if Ei1
S <0
60% of the demand cannot be satisfied directly by the generation. In
the best cases, both of these values never go below 40%. This lack of Thus, the energy imported from the grid Egrid for each time slot
coincidence between demand and generation imposes the need to is
store energy in any manner. Eigrid ¼ EiD  EiG þ EiS ð4Þ
In order to analyse the peak occurrences, the demand values
were divided between ten equally spaced intervals and calculated where EG is the energy from PV generation and ED is the energy
the number of times that the demand value falls in each defined demand of the house. The demand value for energy storage ESD will
interval. This is shown in Fig. 4 with the demand intervals given depend on the role considered as described next.
relatively to the annual peak demand. The duration is also given
relative to the year duration and the frequency is cumulative. It 3.1. Roles definition
is clear that the higher values of demand are very rare with, for
instance, the demand being above 60% of the annual peak in only 3.1.1. Reach grid zero energy
11.5% of the year. Despite this duration represents more than a Due to the lack of matching between the PV and demand pro-
month it is also clear from Fig. 1 that it does not happen altogether, files, seen in Fig. 3, the houses with PV local generation will export
but in small periods each day. This means that the grid must be to the grid most of the local generation power in a given moment
prepared to supply a high value of demand from the house that and then, later, import from the grid the same amount of power to
happens only in an insignificant slot of time. satisfy the demand. This results in unnecessary power flows
between the house and the grid that imply higher losses and man-
agement needs from the grid point of view. The storage objective
90
considered for this role is to mitigate the influence of the demand
PV Generation seasonal mean (% annual peak)

80
Winter and generation profiles on the grid demand. In an ideal situation, a
Spring house that generates and consumes the same amount of energy
Summer
70 should have no exchanges with the grid. To achieve this objective
Autumn
the amount of PV generation energy that exceeds the demand is
60 stored whenever this happens (and if the storage system is not
full). This stored energy is afterwards used to satisfy the demand
50
when the PV generation is not sufficient. The storage demand for
40 each time slot i is given by

30 EiSD ¼ EiG  EiD ð5Þ

20
3.1.2. Reduction in peak demand from grid
10 Since the higher peaks of demand occur on a small portion of
time, as seen in Fig. 4, it could be beneficial for the grid and for
0
0 4 8 12 16 20 24
the customer to have a more stable demand by reducing its peaks.
Hour of day (legal)
Here the peak was defined as the maximum demand occurrence
on the entire year since this is the value that the grid must ensure
Fig. 2. Seasonal mean of PV generation for each 15 min slice of the day. and that the user must have available. The objective of this storage
 J.M. Santos et al. / Applied Energy 128 (2014) 254–264 257

70 60
CFmonth CFmonth
S
D
60 year

Demand Cover Factor (%)

CF

Supply Cover Factor (%)

50
S CFyear
D
50
40
40
30
30
20
20

10
10

0 0
1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Month Month

(a) (b)

Fig. 3. Supply (a) and demand (b) cover factors on month and year time frames.

factor = 1, and peak reduction = 0.5). The storage demand for

100
options (a), (b) and (c) is given, respectively, by Eqs. (6)–(8).
Demand Value (% Annual Peak Demand)

90
EiSD ¼ EiG  maxð0; EiD  EPin Þ ð6Þ
80

70 (
EiG  EiD if EiG P EiD
60 EiSD ¼ ð7Þ
EiG  EiD þ minðEPin ; EiD  EiG Þ if EiG < EiD
50

40 EiSD ¼ EiG  EiD þ EPin ð8Þ

30
where
20
EPin ¼ maxðEinD Þ  a 8i and EnD ¼ EG  ED ð9Þ
10

0 are the maximum peak from grid demand desired and the net
0 20 40 60 80 100 energy from grid, respectively, and a is the peak from grid reduction
Cumulative Duration (%) factor.

Fig. 4. Cumulative distribution of the frequency of the demand value relative to the 3.1.3. Reduction in peak energy injected into the grid
maximum annual peak.
It is also important to achieve stabilisation in terms of the
energy exported to the grid. The peak of PV generation could rep-
role is to reduce the peak consumption requested to the grid with- resent a problem for the grid in the management of electrical
out affecting the utilisation behaviour. This is done considering an systems with large amounts of installed PV capacity. Thus, a role
objective reduction relative to the maximum annual peak demand. to prevent this situation was defined, where the generated
This reduction is achieved by using the stored energy to supply all energy above a determined threshold should always be stored
the power demanded that is above a given threshold. Since this is instead of exported to the grid. The objective with the storage
an important condition for the grid management it is considered is to reduce the peak of energy injected into the grid to a fraction
that the desired threshold must never be exceeded. This simulation of the peak without storage. So, the storage demand is calculated
was performed assuming three possible management options to as
integrate the storage with the local generation and the grid demand: (
maxð0; EiG  EiD  EPout Þ if EiD < EiG
EiSD ¼ ð10Þ
(a) The PV energy is used to supply only the demand that EiG  EiD if EiD P EiG
exceeds the desired grid peak and the surplus PV energy is
where
stored. The stored energy is used to supply the demand that
is above the desired peak when the PV energy is insufficient. EPout ¼ maxðEinG Þ  b 8i and EnG ¼ ED  EG ð11Þ
(b) The PV energy is used to supply all the demand and the sur-
plus PV energy is stored. The stored energy is used to supply are the maximum peak to grid demand desired and the net energy
the demand that is above the desired peak when the PV to grid, respectively, and b is the peak to grid reduction factor.
energy is insufficient.
(c) The energy is stored from the grid when the demand is 3.1.4. Integration of wind power from the grid
below the desired grid peak. The PV energy is used to supply At the grid level there is also the problem of using the intermit-
only the demand that exceeds the grid peak desired and the tent generation when it is available. To help the grid in the man-
surplus PV energy is stored. The stored energy is used to agement of this problem, the house could import in advance the
supply the demand that is above the desired peak when energy that is going to be needed to supply the next future
the PV energy is insufficient. demand, if this energy is coming from these intermittent sources.
In this way, the house helps to integrate the renewable energy
To illustrate the behaviour of each management option an from the grid only by importing the same amount of energy from
example day is shown in Fig. 7 (storage size = 0.5, generation the grid as before but a little bit early. The integration of grid wind
258
generation capacity is done assuming that the energy imported
from the grid could be identified as coming or not from a wind gen-
eration source. Since the grid wind generation capacity is many
times larger than the house’s daily demand, the maximum daily
importable wind generation from the grid was considered to be
equal to the house’s daily demand. So this means that the house
will only import from the grid wind energy equivalent to what is
expected to be used in a daily period. For this storage role the
house is considered as not having local generation, so the only
source of energy is the grid. The algorithm works in a quite similar
fashion as for the grid zero energy role. Therefore, the demand is
preferentially satisfied with grid’s wind energy and when it
exceeds the demand it is stored. The stored energy is then used
when the wind energy is not sufficient to satisfy the demand and
if there is not enough stored energy the supply must rely on grid’s
non-wind energy. So, the storage demand is given by
EiSD ¼ EiGwind  EiD
where EGwind is part of the grid wind generation, with an annual
value equal to the house demand.
3.2. Simulation results
Considering a situation where the annual PV generated energy
is equal to the annual demand, Fig. 5 shows how the objective of
reaching grid zero energy is partially achieved, using the respective
role, for different storage capacities. The values are a daily average
of the year total energy flows. The storage utilisation has a strong
impact on the flows until a storage capacity around 0.6 AD, where
the impact starts becoming minor, with a reduction in grid depen-
dence to 40% of the initial state being achieved. This shows that a
storage dimensioned for the daily time frame is the most suitable
for these profiles. For larger (and unreasonable) storage capacities
the benefit is lower and comes from energy transactions between
months. For larger storage capacities the initial SOC has a signifi-
cant influence on the resulting flows. Anyway, the zero surplus
generation is never reached for storage capacity below 50 AD.
The grid zero energy state is achieved for a storage capacity of
100 AD. Since the annual generation is equal to the annual
demand, and the storage system has losses, this grid zero energy
situation results in a final SOC that is lower than initial SOC. The
difference is of 24.24%, meaning that the storage losses over a year
are equivalent to the demand of about 24 average days. Since the
storage system has losses, the hypothesis that they must be
Daily Energy Flow (× annual demand daily mean)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 0
0 0.2
Storage Size (× annual demand daily mean)
Fig. 5. Storage impact for reaching grid zero energy, considering both reasonable (a) and unreasonable (b) storage capacities (notice that the maximum storage size is 1 AD in
(a) and 100 AD in (b)).
J.M. Santos et al. / Applied Energy 128 (2014) 254–264
compensated by having the annual generation greater than the
demand, in order to achieve zero energy flows with the grid, was
assessed. However, in the analysed situation and assuming reason-
able values of storage capacity, this hypothesis is not confirmed
since the increment in generation capacity causes mainly an
increase in surplus generation and just a small decrease in grid
demand, as can be seen in Fig. 6.
Based on the simulation of reducing grid peak using storage
capacity for the whole year, the minimum storage capacity needed
to achieve each peak reduction value can be assessed. This result
can be seen in Fig. 8. For example, with a generation capacity of
1 and a storage size of 0.5, the grid peak demand can be reduced
to 50%. For the three options, the storage allows to achieve the
peak reduction objective keeping a SOC near to 100% in most of
the time. Despite this being needed to guarantee the maximum
grid peak desired, it is also responsible for the incapacity to store
most of the energy generated. The result is that most of the energy
ð12Þ that is not immediately consumed must be exported to the grid or
wasted. Incrementing the storage capacity while using the man-
agement strategy of peak reduction has an insignificant effect on
reducing the surplus generation. The option (b) reduces the storage
losses because it uses directly the PV generation to supply the
demand but, at the same time, the storage level is kept too low
for reducing the peak, what results in more storage capacity
needed to achieve the same peak reduction compared to option
(a). Option (c) is the one that keeps a higher SOC, since it charges
from grid, and so achieve the peak reductions with lower storage
capacities.
In the same way as for the peak from grid role, the needed stor-
age capacity to achieve each reduction in the peak export was
assessed (Fig. 9). The results are similar, but in this case are even
more linear. To achieve a peak to grid reduction in 50% it is needed
a storage capacity of 0.5 AD, for an annual generation equal to the
demand. The difference is that for this role the SOC is kept very low
most of the time, due to the intrinsic needed availability to receive
charge.
The results of the simulation for the grid wind integration role
are show in Fig. 10. With no storage capacity the non-wind energy
flux and the not used grid wind energy have already a quite low
value, about half of the values for PV, due to the more coincident
profile of wind generation and demand. The storage capacity
reduces both the non-wind energy flux and the unused wind
energy but the impact is felt slowly, and for a storage capacity
equal to the annual daily mean demand the values are reduced
to only 60% of the situation without storage.
Daily Energy Flow (× annual demand daily mean)
0.7
Surplus Generation
From Grid
0.6
Storage In
Storage Out
0.5
Surplus Generation
0.4
From Grid
Storage In
0.3 Storage Out
0.2
0.1
0.4 0.6 0.8 1 0 20 40 60 80 100
Storage Size (× annual demand daily mean)
(a) (b)
 J.M. Santos et al. / Applied Energy 128 (2014) 254–264 259

Daily Energy Flow (× annual demand daily mean)

1.4 could be the driving objective for the local storage capacity man-
Surplus Generation agement design.
1.2 From Grid The impact on the grid should be analysed from two important
Storage In
Storage Out
perspectives: in terms of energy balancing for the whole grid, and
1 in terms of transmission capacity of all the distribution branches of
the grid. This second aspect is out of the scope of the present work,
0.8 and so the grid is considered as a single node where all the gener-
ation and demand must be balanced.
0.6

4.1. Simulation framework

0.4

0.2
To accomplish the desired objective, a framework to simulate
the residential storage impact on the electric grid was established.
0 The simulation allows to analyse the impact with different config-
0 0.5 1 1.5 2 urations of storage management and penetrations on the system.
Annual generation daily mean (× annual demand daily mean) Since the analysis is performed on the base of the grid energy bal-
ancing, considering the grid as a single node, the simulation grid
Fig. 6. Generation impact for reaching grid zero energy (storage capacity = 1 AD).
scheme could be represented as in Fig. 11. This figure shows how
the grid management is done between all the generation and
demand. The generation is divided into generation from conven-
4. Grid impact of residential storage tional or stable sources (GS ), distributed intermittent generation
from the residential sector (GR ), and other intermittent generation
The evaluation of the impact on the grid from having distrib- (GI ). The demand is also divided between demand from the resi-
uted energy storage at the residential level must be considered dential sector (DR ) and demand from all other sectors (DO ). The res-
in the analysis of the residential storage role. As the residential sec- idential sector is considered to be divided between conventional
tor is responsible for a significant part of the grid total demand, any houses and houses with generation and, eventually, storage capac-
modification on the residential electricity management will have a ity, here called Zero Energy Houses (ZEH). The relative penetration
corresponding impact on the grid management. At the same time, (p) indicates the percentage of houses that are ZEH.
the residential sector is also responsible for a large amount of pho- The simulation uses as input the generation and load data with
tovoltaic generation that, most of the times, is fully exported to the a resolution of 15 min. The generation data are divided by source
grid, due to the existence of feed-in special tariffs. The impact on both as absolute values and as profiles. The profiles are used to
the grid could be a consequence of a storage capacity designed to simulate generation from a given source considering a desired
accomplish a local objective or, otherwise, the impact on the grid installed capacity. The load data used include the load for the
Energy (% annual demand daily mean)
Energy (% annual demand daily mean)

2.5 100 2.5 100 2.5 100

Energy (% annual demand daily mean)

Demand 90 Demand 90 90
Generation Generation
2 Grid Demand 80 2 Grid Demand 80 2 80
Storage SOC (%)

Storage State Storage State

Storage SOC (%)

Storage SOC (%)

70 70 70
1.5 60 1.5 60 1.5 60
50 50 50
Demand
1 40 1 40 1 40
Generation
30 30 Grid Demand 30
Storage State
0.5 20 0.5 20 0.5 20
10 10 10
0 0 0 0 0 0
00:00 06:00 12:00 18:00 00:00 00:00 06:00 12:00 18:00 00:00 00:00 06:00 12:00 18:00 00:00
Time Time Time
(a) (b) (c)
Fig. 7. Peak reduction example day profiles.

1.4 1.4 1.4

Annual generation daily mean
Annual generation daily mean Annual generation daily mean
(x annual demand daily mean)
( × annual demand daily mean)

( × annual demand daily mean)

( × annual demand daily mean)

1.2 (x annual demand daily mean) 1.2 1.2 (x annual demand daily mean)
2.0
Required Storage Size

Required Storage Size

Required Storage Size

2.0 2.0
1.0
1 1.0 1 1 1.0
0.5
0.5 0.5
0.0
0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

0 0 0
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Annual Grid Peak Demand (× Annual Peak Demand) Annual Grid Peak Demand (× Annual Peak Demand) Annual Grid Peak Demand (× Annual Peak Demand)

(a) (b) (c)

Fig. 8. Minimum required storage capacities vs. desired grid peak demand reduction.
260 J.M. Santos et al. / Applied Energy 128 (2014) 254–264

an entire year and the residential profiles to divide the load data
Required Storage Size (× annual demand daily mean)

0.7
between residential and other sectors.
Annual generation daily mean Given these input data, the simulation could be performed for
0.6 (× annual demand daily mean)
different storage management options and gives the resultant data
2.0 for the ZEH load and generation requested to the grid for the both
0.5
1.0
situations with and without storage.
0.5
0.4
4.2. Impact assessment
0.3
Using the simulation framework described, the impact from the
0.2 generation and storage capacity of the ZEH on the grid was
assessed. Since the objective is to analyse the impact on the grid
0.1 of the residential storage penetration, the scenario of having resi-
dential generation without storage was used as reference. This
0 ensures that the assessment is focused specifically on the impact
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1
of storage and only indirectly on the impact of residential genera-
Export to Grid Annual Peak (× Peak w/o storage)
tion penetration.
Fig. 9. Minimum required storage capacities vs. desired export peak to grid The data used for the analysis are from the Portuguese electrical
reduction. grid over the year 2011. The grid load data with a resolution of
15 min are provided by the Portuguese mainland transmission sys-
tem operator (REN) [27], being the residential load calculated con-
Daily Energy Flow (× annual demand daily mean)

sidering that it is responsible for 28.43% of the annual grid demand

0.35
Non−used Grid renewable [29]. The residential load with 15 min resolution is obtained by
Grid non−renewable applying the suitable demand profile, also provided by REN [26],
0.3
Storage In
to the annual residential demand. The PV generation profile is
Storage Out
0.25 extracted from the PV generation data provided by REN that
accounts for all the PV devices connected to the grid [27]. Since this
0.2 impact assessment is done for the whole grid, this PV generation
profile is taken as the mean profile representing the diversity of
0.15 geographical sites and system types available on the grid. This PV
generation profile is used to simulate the PV generation to the
0.1 desired installed capacity, that depends on the ZEH penetration
and on the capacity per house. For this a PV generation yield of
0.05
1500 kW h/kW p was used, as an average value for the entire coun-
try, calculated using the PV generation data and the PV installed
0
0 0.2 0.4 0.6 0.8 1 capacity referred in [30]. The number of residential buildings is
Storage Size (× annual demand daily mean) retrieved from the data of 2011 census [31]. The buildings are
divided between single family buildings and multiple family build-
Fig. 10. Storage impact for grid wind integration.
ings. For the first ones, a PV installed capacity of 2.45 kW p and a
storage capacity of 6 kW h (effective) was considered, and for the
last ones the double of these capacities. These are the capacities
that allow to have a PV annual generation equal to the residential
demand and a storage of 60% of the residential demand daily mean.
The storage round-trip efficiency is considered to be 90%.
Each storage role manages the storage for a defined objective
and so the impact should be analysed according to the different
objectives. For the grid zero energy role, it is expected that the
power flows of the residential sector with the grid are mitigated
since the locally generated energy is kept at the house instead of
being exported to the grid and this same energy could be used at
the house when needed, instead of having to import it from the
grid. Therefore, from the grid perspective, this could mean a double
reduction in power flows when comparing to a situation without
storage. This reduction is presented in Fig. 12, where the sum of
the imported and exported energy during an year relatively to
the total grid demand, for each level of ZEH penetration, is shown.
While without storage the power flows with the grid will increase
due to the need to export to the grid the generated energy that is
not immediately consumed, with the adequate storage role they
will diminish as the stored energy will reduce the import from
the grid. It is important to note that this figure is referenced to
the total grid demand but this value is also reduced with the ZEH
Fig. 11. Single node grid simulation scheme. penetration. The grid zero role option is the only one that causes
a reduction on the residential power flows with the grid and has
entire grid and the profiles for the residential sector. The simula- a significant impact on this indicator, being able to reduce the
tion uses the relative percentage of residential consumption for power flows with the grid in 60%. The other two roles will only
 J.M. Santos et al. / Applied Energy 128 (2014) 254–264 261

40 Storage Role
160

Daily Peak of Grid Demand Covered by

No Storage (Annual Max)
Annual Residential Grid Transactions

140 Grid Zero (Annual Max)

35
Import Peak Control (Annual Max)

Residential Generation (%)

Export Peak Control (Annual Max)
(% Total Grid Demand)

120
No Storage (Annual Mean)
30 Grid Zero (Annual Mean)
100 Import Peak Control (Annual Mean)
Export Peak Control (Annual Mean)
25 80

Storage Role 60
20
No Storage
Grid Zero 40
15 Import Peak Control
Export Peak Control 20

10 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
ZEH Penetration (%) ZEH Penetration (%)

Fig. 12. Impact on electricity power flows between the residential sector and the Fig. 14. Annual maximum and mean values of daily peak percentage of grid
grid caused by different roles of residential storage. demand covered by residential generation, for different storage roles and zero
energy houses penetration (the no storage lines are coincident with the import peak
control ones).

attenuate slightly or almost completely the power flows increase.

Considering that the power flow to and from the grid implies losses is a consequence of this role being the one that maximises the
in the transmission, the reduction in this power flow will reduce use of the storage capacity available.
these losses. Additionally, with a storage role that accommodate On the other hand, the export peak control role has the objec-
the PV generation without incrementing the power flow with the tive of limiting the peak of energy exported to the grid and the
grid, the residential PV penetration will not demand for a grid energy exported from the ZEH to the grid could cover an unman-
capacity expansion. ageable percentage of the grid demand (Fig. 14). In this figure it
The objective of reducing the peak demand from the grid is is presented the daily peak percentage of grid demand satisfied
achieved by the respective role considering that the demand exclusively by the residential generation exported to the grid
could never be higher than an objective value. The result obtained (net generation). A ZEH penetration of 50% would be responsible
for the maximum peak demand reduction of 55% leads to an for satisfying the annual maximum of more than half of the grid
impact on the grid demand shown in Fig. 13. In this figure is pre- peak, and with a penetration of 80% the generation will exceed
sented the annual maximum and mean values of grid demand for the demand peak. The adequate storage role could allow to incre-
every 15 min, relative to the total annual peak, and for each role, ment the penetration slightly more than 10%, for the same impact
assuming total ZEH penetration. The import peak control role is on grid peak demand cover. Even considering the annual mean
the only one that reduces the maximum annual peak. The typical value, it is possible to achieve more than 70% without storage.
day has two grid demand peaks, one around midday and the The grid zero role is, again in this case, the better solution when
other in the early night, being the residential sector mainly considering the annual mean. It reduces the value for the case of
responsible for the last one. Since this role is focused on the total penetration to 40% and allows an increment of 30% in the
reduction in the residential peak, it reduces the night grid peak ZEH penetration with the same grid impact. When analysing the
but keeps the midday one, while the other roles do the opposite. grid demand covered by residential generation in terms of day
However, the grid zero role performs a better reduction in the hour impact (Fig. 15) it is clear that the grid zero role could have
peak in terms of annual mean value than the import peak control a better performance in reducing the peak export to grid if the stor-
role. For the grid zero role, the annual mean demand value is ing process was more uniformly distributed during the generation
lower at any time of the day than for the other two roles. This hours.

100 75
(% annual demand peak, p = 100 %)

(% annual demand peak, p = 100 %)

95 Storage Role
Annual Max Hour Grid Demand

Annual Mean Hour Grid Demand

70
No Storage
90 Grid Zero
65 Import Peak Control
85 Export Peak Control

80 60

75 55

70 Storage Role

No Storage 50
65 Grid Zero
Import Peak Control 45
60 Export Peak Control

55 40
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Hour of Day Hour of Day
(a) (b)
Fig. 13. Impact of residential storage on total grid demand for each 15 min of the day, considering the annual maximum (a) and average (b) values.
262 J.M. Santos et al. / Applied Energy 128 (2014) 254–264

5. Economic assessment
To assess the value of energy storage, the economic aspects
must also be taken into account. Therefore, an analysis of the eco-
nomic benefits ensured by energy storage at the residential level is
performed. This analysis aims to assess how the pricing scheme
could reflect the storage importance from the grid perspective
and how the customer could benefit from that pricing. The objec-
tive is to compare economically the different storage roles between
them and with the situation of PV without storage. This compari-
son was done considering different pricing scenarios for each sys-
tem/role. For this effect the annual economic balance between the
system cost, the value of produced energy, and the value of energy
imported from the grid was calculated.
The annual system cost (C sys ) was calculated considering the
capital cost of the system and the respective lifetime:
C sys ¼ ðCapexPV þ Capexstor Þ  CRF þ C man
where CapexPV is the capital cost of the PV system, Capexstor is the
capital cost of the storage system, and C man is the annual cost of sys-
tem maintenance. The Capital Recovery Factor (CRF) is used to con-
vert the capital cost of the system into a stream of annual payments
over the system lifetime, and is given by
n
ið1 þ iÞ
CRF ¼ n1
ð1 þ iÞ
where i is the discount rate and n is the system lifetime in years.
Since the annual system cost is given per kW h of annual aver-
age daily demand (AD) the capital cost of PV assumes the installed
capacity needed to generate annually the same amount of energy
as the demand, and the capital cost of storage assumes a storage
device with a usable capacity as defined in Table 1 and assuming
a maximum depth of discharge (DOD) of 80%. The PV yield calcu-
lated from the REN data of 1500 kW h/kW p was used and the cost
predicted for 2020 of 1850 €/kW p for PV [32] and of 200 €/kW h
for lithium-ion storage [33] was assumed. The lifetime considered
is 20 years and the discount rate is 5%. The included annual costs of
maintenance are of 1.25% of the system cost.
Several scenarios that represent incentives for the customer to
adopt each role were defined as follows:
1. Self-consumption incentive scenario. The tariff for on-site con-
sumption of generated energy is 0.17 €/kW h, the feed-in tariff
is 0.17 €/kW h, and the buy tariff is 0.17 €/kW h.
2. Time-of-use scenario. The buy tariff is 0.10 €/kW h on off-peak
hours (10 p.m–8 a.m) and 0.30 €/kW h on peak hours (8 a.m–
10 p.m). The feed-in tariff is 0.17 €/kW h.
3. Peak penalisation scenario. The buy tariff is 0.17 €/kW h, with a
penalty of 0.51 €/kW h when the demand goes above 50% of the
ð13Þ annual peak, and the feed-in is 0.17 €/kW h.
4. Feed-in cut. The feed-in limit is set at 50% of the annual peak.
The tariff below this is of 0.17 €/kW h and the buy tariff is
0.17 €/kW h.
The economic benefit provided for each system is summarised
in Table 1, where the annual result, with the system cost included,
is presented as a percentage of the cost for the customer without
ð14Þ PV and storage. Thus, for each scenario, the best system/role is
the one with lower percentage result. The grid zero role is the
one that better take profit from the self-consumption incentive,
what also means that this is the best incentive if the grid objective
is to reduce the amount of energy transfers with the residential
sector. Even though the grid zero role being the one that most
reduces the grid demand on peak hours, it does not have the best
result with the time-of-use tariff. This happens because it transfers

Storage Role
70
Storage Role No Storage
Annual Mean Grid Demand Supplied by
Residential Generation (%, p = 100 %)

160 Grid Zero

Residential Generation (%, p = 100 %)

No Storage
Annual Max Grid Demand Supplied by

60
Grid Zero Import Peak Control
140 Import Peak Control Export Peak Control
Export Peak Control
50
120

100 40

80
30
60
20
40
10
20

0 0
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Hour of Day Hour of Day
(a) (b)
Fig. 15. Impact of residential storage on hourly distribution of grid demand covered by residential generation, considering the annual maximum (a) and average (b) values
(notice the different vertical axis limits and that the no storage lines are coincident with the import peak control ones).

Table 1
Economic results for different storage roles and pricing scenarios. The annual result for each scenario is given relative to the result for a customer without PV and storage.

Storage role No storage Grid zero Peak from grid reduction Peak to grid reduction
Usable storage size (AD) 0 0.6 0.4 0.5
Equivalent full cycles/year n.a. 243.9 66.83 104.9
Annual System Cost (€/kW hAD ) 58.50 76.12 70.25 73.19
Scenario 1 result (%) 26 9 40 30
Scenario 2 result (%) 53 67 60 63
Scenario 3 result (%) 73 64 47 73
Scenario 4 result (%) 73 92 87 85
 J.M. Santos et al. / Applied Energy 128 (2014) 254–264
energy from the day hours, which could be valued by the feed-in
tariff, to the off-peak time, when the benefit from not importing
from the grid is lower. Nevertheless, none storage role is better
than the PV alone for this scenario. The peak from grid reduction
role shows good response to the scenario where the higher peak
from grid is penalised, but the scenario where the feed-in peak
energy is not valuable does not give benefit for any storage role.
6. Conclusion
This work aimed to assess the relevance of having distributed
electricity storage capacity at the residential level, side by side
with the consumption and generation devices. A comprehensive
assessment of the storage roles was made considering three
aspects: the in-house energy management with the storage
included, its large-scale impact on the grid if the numbers of
houses adopting storage becomes significant, as well as the eco-
nomic scenarios for stimulating this adoption and the associated
benefits. The approach used in this work presents novelty consid-
ering the following aspects. Only the grid connected scenario is
considered. The demand and generation data used are neither lim-
ited to a particular small time frame nor to a particular residential
system, giving more generalised results. Different roles for storage
are considered, with the purpose of optimising the local benefit
and also the global grid impact. The scenarios considered for the
economic evaluation intend to reflect the technical benefits of
the grid and not the profit from particular niche opportunities that
will disappear with the massive penetration of these systems. The
simulation is done with storage operating rules, and so the
obtained results do not imply a priori knowledge of future data.
In contrast with the renewable generation and the demand, that
have limited flexibility, the storage role is fully dependent on the
way it is managed. Thus, some different roles for storage were
defined and for each one an adequate management strategy was
used. For this purpose the work has started by analysing the typical
generation and demand profiles of the residential sector. This has
shown that, on a daily time slicing, only a small part of the residen-
tial demand is coincident with the PV generation and that the
demand peak occurs out of the generation window. Also, the resi-
dential demand is half of the time lower than half of the maximum
peak. Considering the entire year, the difficulty in matching the PV
generation and demand is affected even more, as the generation is
lower in the winter when the demand is higher and the opposite
happens for the summer. This situation has revealed has a major
constrain for the sizing and the management of the storage system
since, for the same capacity, the mean state of charge is too low in
the winter and too high during the summer. What can be con-
cluded is that, despite a reasonable capacity of storage being very
useful to make daily matches of generation and demand, the match
gap between seasons is not achievable and the house will benefit
from being connected to the grid to import and export the extra
energy when needed.
Three storage roles were defined, each one with a different
objective: maximise the house independence from the grid; mini-
mise the power flow peaks from the grid; and minimise the power
flow peaks to the grid. A role with the objective of integrating the
grid wind generation was also defined, but the storage impact in
this role is much lower due to the higher correlation of the wind
generation profile with the demand. For the grid zero role an opti-
mal storage size zone was obtained, while for the other roles the
benefit is more linear with the storage increment. In terms of grid
impact, the effect of storage is also relevant and could be essential
for the grid management for higher penetrations of PV generation
in the residential sector. Despite the peak reduction roles being
effective in reducing the annual maximum power flow values, from
263
the grid perspective, the grid zero role provides better performance
since it reduces the impacts in the annual power flow average
values.
From the economic perspective, the defined pricing scenarios
take advantage of the storage presence, with the exception to the
peak to grid reduction role, which has no positive impact in the
defined pricing conditions. It was also found that the way of man-
aging the storage devices can reduce the storage system lifetime
because the storage has a very low average SOC due to the winter
season, when the generation is not sufficient to replace the used
stored energy. This can be improved by a managing scheme to mit-
igate the continuous deep discharge of the storage devices.
Acknowledgements
The authors would like to acknowledge the financial support
from the Portuguese Foundation for Science and Technology under
the reference FCT-DFRH-SFRH/BD/70250/2010.
This work has been framed under the Initiative Energy for
Sustainability of the University of Coimbra and supported by the
Energy and Mobility for Sustainable Regions Project CENTRO-07-
0224-FEDER-002004.
References
[1] Makansi J, Abboud J. Energy storage the missing link in the electricity value
chain; 2002.
[2] Eyer J, Corey G. Energy storage for the electricity grid: benefits and market
potential assessment guide. Tech. Rep. Sandia National Laboratories; 2010.
[3] Electricity Advisory Committee. Bottling electricity: storage as a strategic tool
for managing variability and capacity concerns in the modern grid. Tech. Rep.
U.S. Department of Energy; 2008.
[4] Beaudin M, Zareipour H, Schellenberglabe A, Rosehart W. Energy storage for
mitigating the variability of renewable electricity sources: an updated review.
Energy Sustain Dev 2010;14(4):302–14. http://dx.doi.org/10.1016/
j.esd.2010.09.007.
[5] Energy Commission of California. 2020 strategic analysis of energy storage in
California. Tech. Rep.; California Energy Commission; 2011.
[6] Energy storage: Asian Systems & Apps. Tech. Rep. August; Zpryme Research &
Consulting; 2012.
[7] Mulder G, Ridder FD, Six D. Electricity storage for grid-connected household
dwellings with PV panels. Solar Energy 2010;84(7):1284–93. http://dx.doi.org/
10.1016/j.solener.2010.04.005.
[8] Maclay J, Brouwer J, Scottsamuelsen G. Dynamic analyses of regenerative fuel
cell power for potential use in renewable residential applications. Int J
Hydrogen Energy 2006;31(8):994–1009. http://dx.doi.org/10.1016/
j.ijhydene.2005.10.008.
[9] de Ven PV, Hegde N, Massoulie L, Salonidis T. Optimal control of residential
energy storage under price fluctuations. In: IARIA energy conference 2011.
Venice, Italy; 2011. p. 159–62.
[10] Castillo-Cagigal M, Caamaño Martín E, Matallanas E, Masa-Bote D, Gutiérrez A,
Monasterio-Huelin F, et al. PV self-consumption optimization with storage and
Active DSM for the residential sector. Solar Energy 2011;85(9):2338–48.
http://dx.doi.org/10.1016/j.solener.2011.06.028.
[11] Matallanas E, Castillo-Cagigal M, Gutiérrez A, Monasterio-Huelin F, Caamaño
Martín E, Masa D, et al. Neural network controller for Active Demand-Side
Management with PV energy in the residential sector. Appl Energy
2012;91(1):90–7. http://dx.doi.org/10.1016/j.apenergy.2011.09.004. <http://
www.sciencedirect.com/science/article/pii/S0306261911005630>.
[12] Molderink A, Bakker V, Bosman MG, Hurink JL, Smit GJ. Domestic energy
management methodology for optimizing efficiency in Smart Grids. In: 2009
IEEE Bucharest PowerTech; 2009. p. 1–7. http://dx.doi.org/10.1109/
PTC.2009.5281849.
[13] Armas JM, Suryanarayanan S. A heuristic technique for scheduling a customer-
driven residential distributed energy resource installation. In: 2009 15th
International conference on intelligent system applications to power systems.
IEEE. ISBN 978-1-4244-5097-8; 2009. p. 1–7. http://dx.doi.org/10.1109/
ISAP.2009.5352954.
[14] Clastres C, Pham TTH, Wurtz F, Bacha S. Ancillary services and optimal
household energy management with photovoltaic production. Energy
2010;35(1):55–64. http://dx.doi.org/10.1016/j.energy.2009.08.025.
[15] Venu C, Riffonneau Y, Bacha S, Baghzouz Y. Battery storage system sizing in
distribution feeders with distributed photovoltaic systems. In: PowerTech,
2009 IEEE Bucharest. IEEE. ISBN 978-1-4244-2234-0; 2009. p. 1–5. http://
dx.doi.org/10.1109/PTC.2009.5282093.
[16] Denholm P, Hand M. Grid flexibility and storage required to achieve very high
penetration of variable renewable electricity. Energy Policy
2011;39(3):1817–30. http://dx.doi.org/10.1016/j.enpol.2011.01.019.
264 J.M. Santos et al. / Applied Energy 128 (2014) 254–264
[17] Hollinger R, Wille-Haussmann B, Erge T, Sönnichsen J, Stillahn T, Kreifels N.
Speicherstudie 2013. Tech. Rep. Fraunhofer Institute for Solar Energy Systems
ISE; 2013.
[18] Braun M, Büdenbender K, Magnor D, Jossen A. Photovoltaic self-consumption
in Germany – using lithium-ion storage to increase self-consumed
photovoltaic energy. In: 24th European photovoltaic solar energy
conference. Hamburg; 2009.
[19] Mulder G, Six D, Claessens B, Broes T, Omar N, Mierlo JV. The dimensioning of
PV-battery systems depending on the incentive and selling price conditions.
Appl Energy 2013;111:1126–35. http://dx.doi.org/10.1016/j.apenergy.
2013.03.059.
[20] Zhao J, Kucuksari S, Mazhari E, Son YJ. Integrated analysis of high-penetration
PV and PHEV with energy storage and demand response. Appl Energy
2013;112:35–51. http://dx.doi.org/10.1016/j.apenergy.2013.05.070. <http://
linkinghub.elsevier.com/retrieve/pii/S0306261913004832>.
[21] McKenna E, McManus M, Cooper S, Thomson M. Economic and environmental
impact of lead-acid batteries in grid-connected domestic PV systems. Appl
Energy 2013;104:239–49. http://dx.doi.org/10.1016/j.apenergy.2012.11.016.
<http://linkinghub.elsevier.com/retrieve/pii/S0306261912008094>.
[22] Testa A, Caro SD, Torre RL, Scimone T. Optimal design of energy storage
systems for stand-alone hybrid wind/PV generators. In: 2010 International
symposium on Power Electronics Electrical Drives Automation and Motion
(SPEEDAM); 2010. p. 1291–6.
[23] Maclay J, Brouwer J, Samuelsen G. Dynamic modeling of hybrid energy storage
systems coupled to photovoltaic generation in residential applications. J Power
Sources 2007;163(2):916–25. http://dx.doi.org/10.1016/j.jpowsour.2006.
09.086.
[24] Palomino G, Wiles J, Stevens J, Goodman F. Performance of a grid connected
residential photovoltaic system with energy storage. In: Conference record of
the twenty sixth IEEE photovoltaic specialists conference – 1997; 1997. p.
1377–80. http://dx.doi.org/10.1109/PVSC.1997.654347.
[25] Widén J, Wäckelgå rd E, Lund PD. Options for improving the load matching
capability of distributed photovoltaics: methodology and application to high-
latitude data. Solar Energy 2009;83(11):1953–66. http://dx.doi.org/10.1016/
j.solener.2009.07.007.
[26] REN. Electricity demand profiles. 2013. <http://www.mercado.ren.pt/PT/
Electr/InfoMercado/Consumo/Paginas/PerfisConsumo.aspx>.
[27] REN. Electricity generation share. 2013. <http://www.centrodeinformacao.
ren.pt/PT/InformacaoExploracao/Paginas/EstatisticaDiariaDiagrama.aspx>.
[28] Baetens R, Coninck RD, Helsen L, Saelens D. The impact of domestic load
profiles on the grid-interaction of building integrated photovoltaic (BIPV)
systems in extremely low-energy dwellings. In: Proceedings of the renewable
energy research conference. Trondheim, Norway; 2010. p. 3–14.
[29] Balanço Energético 2011. Tech. Rep.; DGEG; 2013.
[30] DGEG. Potência Instalada das Centrais de Produção de Energia Eléctrica a Partir
de Fontes Renováveis. 2013. <www.dgeg.pt>.
[31] de Estatística, I.N. Censos INE. 2013. <www.ine.pt>.
[32] Connecting the sun. Tech. Rep. September; European Photovoltaic Industry
Association; 2012.
[33] Hensley R, Newman J, Rogers M. Battery technology charges ahead. Tech. Rep.
McKinsey & Company; 2012.