IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO.

3, MAY/JUNE 2017 1769

Demand Side Management in Microgrids for Load

Control in Nearly Zero Energy Buildings
Luigi Martirano, Student Member, IEEE, Emanuele Habib, Giuseppe Parise, Life Fellow, IEEE,
Giacomo Greco, Student Member, IEEE, Matteo Manganelli, Student Member, IEEE, Ferdinando Massarella,
and Luigi Parise, Member, IEEE

Abstract—The paper presents a feasible model of architecture energy performance of buildings directive (EPBD) [3]. NZEB
for the technical building systems (TBS) particularly suitable for means a building that has a very high energy performance, as
nearly zero energy buildings (NZEBs). The suggested model ag- determined in accordance with Annex I of Directive [3]. The
gregates the users around an electric node in order to reach up the
threshold value of electric power, and to get a more virtuous and nearly zero or very low amount of energy required should be
flexible cumulative load profile. Present proposal is a full electric covered to a very significant extent by energy from renewable
common smart micro grid with a single point of connection, with sources, favoring energy from renewable sources produced on-
heating and domestic hot water generated by a centralized electric site or nearby. Let us note that the major appliances (both white
heat pump system. The renewable energy is provided by a photo-
and brown goods) and other electric loads not fixed (equipment
voltaic field connected to the common grid. A building automation
control system operates those electric TBS modulating the global supplied by plugs) are not taken into account in the NZEB qual-
load for a building demand response (DR). The effectiveness of the ification. The EPBD requires all new buildings to be NZEBs by
proposed model consists of exploiting thermal inertia as an energy the end of 2020 (public buildings must be NZEBs by 2018). A
storage, by forcing both local and central set points of heating and building may reach the NZEB qualification through a complex
air conditioning systems. The control is based on the integrated of efficient technical building systems (TBSs), building automa-
and common operation of all users and all systems of the building
as one unique “large user.” The integrated management of the grid tion control system (BACS), and renewable sources, besides a
is operated to control the whole electric demand exploiting the self- very high energy performance given by envelope insulation and
consumption, avoiding peaks, and maintaining a flat load profile. other passive measures [4]. The TBSs that must be provided
The suggested microgrid model allows concretely the possibility to vary with the type of building, but will generally include a
realize a building DR with benefits for the end-users in a consumer building electric service and power distribution system to serve
view point. A study of the effect of these control opportunities on
whole electric demand is done by simulation on a case study. the loads, a heating, ventilation, and air conditioning (HVAC)
system, a domestic hot water (DHW) system, electronic safety
Index Terms—Demand side management, electric power sys- and security systems, and a communication system (ICT). The
tems, microgrid, nearly zero energy buildings (NZEBs), thermal
energy storage.
extremely low amount of energy that NZEBs require (energy
spent ES2 ) comes mostly from renewable local sources (energy
I. INTRODUCTION generated ERES ) like: photovoltaic (PV), ground-source heat
pumps (GSHP) or thermal solar systems.
HE ENERGY policy approach is carrying out a long-
T time renewal process of the electric and in general energy
framework. The energy spent in commercial, residential, and in-
The complexity of TBSs present in a NZEB determines that
the TBSs have to be organized in a “microgrid approach;” it is
necessary to use the most innovative solutions; it is necessary
stitutional buildings is a great amount (in EU is estimated about
the presence of a building energy management system (BEMS)
40% of total energy consumption and about 90% in high-density
that is a SCADA system able to operate, manage, and control
urban areas) [1]. The general encouragement of the rational use
the technical systems; and the BEMS needs a BACS, a smart
of energy, also for residential users, introduced the new approach
metering system, and a building manager (BM).
of the nearly zero-energy buildings (NZEBs) by the European
The entire “NZEB cycle” has to be considered not just as a
Manuscript received July 28, 2016; revised November 10, 2016 and January
qualification process but as a permanent target, therefore, not
16, 2017; accepted February 8, 2017. Date of publication February 22, 2017; only designers and installers, but also operators and managers,
date of current version May 18, 2017. Paper 2016-ESC-0734.R2, presented are critical for the successful implementation of the objectives
at the 2016 IEEE Industry Applications Society Annual Meeting, Portland,
OR, USA, Oct. 2–6, and approved for publication in the IEEE TRANSACTIONS
of the energy performance in a complete life cycle point of view
ON INDUSTRY APPLICATIONS by the Energy Systems Committee of the IEEE (design-installation-operation) [5]–[7]. The BM is a key person
Industry Applications Society. in this process.
The authors are with the Faculty of Civil and Industrial Engineering, Sapienza
University of Rome, Rome 00185, Italy (e-mail: [email protected]; emanuele.
The paper presents a full electric model of microgrid for
[email protected]; [email protected]; [email protected]; matteo. residential/tertiary buildings characterized by an unique point
[email protected]; [email protected]; [email protected]). of connection (POC) with the electric grid.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
The approach proposed by the authors is the integration of
Digital Object Identifier 10.1109/TIA.2017.2672918 TBSs inside a residential building, referring to their energy

0093-9994 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
1770 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO. 3, MAY/JUNE 2017

interaction and management. In this way, the whole building

can be seen as a single subject that interacts with the dis-
tribution networks enhancing its energy efficiency and cost
performance.
Buildings actually have considerable amount of flexible loads.
For instance, the electricity load for cooling/heating could be
shifted with the use of building thermal mass and other thermal
storage facilities [8].
In this framework, the demand side management (DSM) can
reach a great impact in respect of the network, especially con-
sidering the presence of manageable loads like white appliances Fig. 1. On the left: yearly energy spent and yearly energy from renewables
[e.g., washing machines (WM)] and HVAC systems, by shifting in a NZEB. On the right: a sample daily power profile in the POC of a NZEB
loads in time. without controls (continuous line) and with a DSM strategy (dashed line).
The use of such DSM for power balance in smart grids is
referred as demand response (DR), which has been attracting
attentions in the recent years [10]. The paper is organized as follows. Section II presents a brief
Although there are some papers available on the integration overview about NZEB and introduces the suggested concept of
of controllable appliances and HVAC systems [10], [11] they KZEB. The suggested architecture of the microgrid is discussed
mainly deal on thermal storage or CHP generators management. in Section III. Section IV presents the detail about the electric
In NZEB, heat exchange to the ambient is limited, thus, it is and thermal models used to evaluate the impact of the controls in
possible to use building structures themselves as heat storage. the energy performance of the building. Section V presents the
This opportunity is analyzed in this paper. proposed BEMS for DSM and describes the operating controls.
Currently, the development and application of DR in build- Section VI presents the case study application and the results.
ings are still at the beginning stage. Particularly, a comprehen- Section VII concludes the paper.
sive review for providing sufficient knowledge and information
to all DR practitioners and from the view point of demand side
II. (NEARLY/NET) ZERO ENERGY BUILDINGS
is needed [8]. Most of the articles present in literature have
their special focuses with limited attention on building DR [12], NZEBs. As noted, the NZEB qualification approach consid-
[13]. The aim of the paper is to show that the advantages of the ers the global annual primary energy amount and not the in-
suggested method based on an user view point, by adopting a stantaneous balance of the power and so the impact to the
BEMS that manages both electric and thermal loads improving utility networks (electric and gas) [15]. So, it could be most
the energy performance of the building and operating a DR, appropriate to use the definition of “net zero” instead of
in comparison with other previous models. The suggested ap- “nearly zero” energy building. Fig. 1 shows that in a NZEB,
proach upgrades the NZEB qualification to “km zero energy the energy yearly balance ES 2 − ERES is equal to zero, but
building” (KZEB) one with a multiple goal: the instantaneous power exchanged with the grid cannot be
1) cost savings without reductions in comfort for the users; equal to zero. It can be bidirectional and can create chaotic
2) controlled load profile to reduce impact on the distribu- phenomena.
tor, improving the exploitation of local generation from
renewable energy sources (RES).
A. From NZEB to KZEB
The economic impact can reach in case of price-based DR
programs as dynamic electricity pricing scenarios [14]. It has been proved that in theory a DSM strategy operating
The most important innovations introduced by the suggested on the controllable loads and on the electric and thermal storage
microgrid model are as follows. can reach a “near zero power behavior” (like the dashed profile
1) The electric architecture with a unique POC and a com- in Fig. 1) so a building can become a KZEB [16]. The suggested
mon distribution system for the entire building. BEMS operates avoiding to change users’ behavior that must
2) The integration of the domotic systems of the single be free to use the facilities and the appliances in full freedom. It
apartments in a global building automation system with must be noted that while NZEB is a label qualification, KZEB
respect of the privacy of the users but with the sharing is a target that is obtained during the life, thanks to the dynamic
of some managements. control of the TBSs [17].
3) The management of the thermal load of the building
according to the electric profile in order to exploit the
natural inertia of the building as a storage system. B. Gas Free Buildings
4) The management of the smart appliance in a building The suggested architecture for the TBSs provides a gas free
point of view. supply, with all systems supplied by the electric network. In
The suggested micro grid model allows concretely the possi- the suggested microgrid, heat for HVAC and DHW systems is
bility to realize a building DR with benefits for the end-users in generated by electric heat pumps and cooking devices are full
a consumer view point. electric.
MARTIRANO et al.: DEMAND SIDE MANAGEMENT IN MICROGRIDS FOR LOAD CONTROL IN NEARLY ZERO ENERGY BUILDINGS 1771

whole building (great user) rather than of a single independent

user. The aggregation of the small users in a grid supplied by
a common network allows reaching better performance, both
economic and technical. The natural aggregation is at building
level with a HV/LV substation. The economic advantage will be
shared between the BM and the users. The BM buys energy and
distributes the cost among users with a regulation agreed (flat, at
consumption, with bonus, etc.). The optimal management of the
energy grids in a building requires the aggregation of the users
in a single POC with the grid so that the energy generated locally
can be fully locally consumed to reach up the threshold value of
demand and to get a more virtuous and flexible cumulative load
Fig. 2. NZEB, KZEB, and ZEB. profile.

C. Local generation B. HVAC and DHW Systems

Local electric generation is provided by a PV system. Heat In the suggested microgrid, heating and DHW are organized
generation is done by a geothermal source heat pump system. with a common central ground source heat pump (GSHP) (see
Geothermal source heat pumps use less energy than conven- Fig. 3). The heating distribution is common, starting from a
tional ones as the ground limits cold source temperature, thus, puffer for the heating and a boiler for the DHW. Each apartment
promoting a higher energy efficiency ratio. Simply implies electric Utility is equipped with an under floor heating system and a satellite
center (SC) with metering systems.
D. From NZEB to Zero Emission Building (ZEB)
C. Electric Power System
The use of a full electric system, without central gas heat and
without cogeneration makes the building a ZEB. Moreover, it is The suggested microgrid consists of an electric power sys-
possible to implement an advanced BEMS with a great impact tem characterized by a single point of connection (POC) with
on the electric exchanges with the network (see Fig. 2). the DSO in HV level, a substation HV/LV and a common LV
Experimental analysis of electric power balance in residential distribution, a main LV switch board (MSB) to supply both the
buildings with heating through GSHP with PV local generation common services (heating, elevators, etc.) and the single units,
has shown a very significant mismatch between energy gener- a PV system connected to the MSB, a main distribution starting
ation and demand leading to an exchange of 70% of locally from the MSB for all common TBSs (heating, lighting, eleva-
generated energy with the grid even though the building is not tors, etc.), a feeder distribution starting from the MSB up to
a NZEB [18]. This means that some control strategy should be each unit in a radial scheme, a feeder for each unit, and a power
implemented in order to reduce energy exchange with the grid. distribution for each unit by a local panel.
In this framework, in this paper building electric loads are BACS/HBES - The TBSs are flanked by a BACS [also called
divided in uncontrollable, plannable, and controllable loads as Home and Building Electronic Systems (HBES)] [20] that can
Home electronic System
discussed in Section IV, in order to check the effect of control on use an open standard as Konnex.
energy exchange with the grid. Actually, as shown in [18], power The complete architecture of the HBES consists of [21] an
demand of appliances, lighting, and other uses of electricity in independent system for each unit (HES) using a local BUS and
residential buildings is very relevant to energy balance so they completed with a local supervisory (LS), and a system for the
are included in the developed model. building (BES) with a backbone BUS connected to the main
supervisory system (MS) where is uploaded the BEMS.
III. ARCHITECTURE OF THE MICROGRID The HES connects the local technical systems of each unit
(lighting, alarms, blinds, heating, etc.) by smart actuators. It is
A. Microgrid With a Single POC
operated by the users to control all the home systems and the
The suggested architecture for the TBSs consists of a full appliances by a human system interface (HIS) that is a software
electric microgrid with a single POC with the network. Indeed, loaded on a tablet or smartphone connected in wifi to the LS.
the complexity of TBSs requires a special architecture of the The operation can be managed also remotely because the LS is
electrical power system that must overcome the traditional rigid connected to the router.
scheme toward an adaptive “microgrid.”. Furthermore, the pres- The BES connects the centralized technical services and it
ence of a BEMS/BACS to manage and control the TBSs allows is linked with all the HES by a coupler (CP in Fig. 3). It is
implementing advanced DSM logics without additional costs, used by the BEMS to manage the centralized services and the
considering that the hardware (BACS) and the software (BEMS) controllable loads by a MS that is the main brain.
are already available [19]. Embedded TBSs (lighting, HVAC, etc.) are controlled di-
In the suggested model, the property of the microgrid can be rectly by the HES. Usually, appliances cannot be monitored
of the building or of the DSO according to the existing regulatory and controlled directly by an HES as manufacturers mostly pre-
provisions, but it is managed by a BM with a perception of the fer to adopt proprietary protocols. Modern appliances cannot
1772 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO. 3, MAY/JUNE 2017

Fig. 3. Full electric smart microgrid architecture for a gas-free NZEB.

be switched ON/OFF by actuators of an HES, as simple power 4) advanced HBES that connects all the domotics unit and
switch ON/OFF does not work correctly with their electronic con- a common metering/monitoring system;
trollers. The preferable way to monitor and control this type of 5) an advanced BEMS to manage the thermal system con-
appliance is to integrate them with the HBES at a higher level sidering not only the overall energy but also the control
like TCP/IP network or cloud architectures. of the electric demand;
A strong thrust to the spread of BACS in housing will have 6) smart appliances connected to the BEMS available for
when smart appliance adopting open protocols will be available a common management.
in the market. The integration of smart appliance in distributed
BACS with open protocols is a key point to their diffusion and
full exploitation. IV. ELECTRIC AND THERMAL BEHAVIOR MODELS
Meantime, in the suggested microgrids, the WM and other Two distinct modeling approaches for private household con-
manageable loads are considered as smart appliances. They are sumption can be identified: top–down and bottom–up [22]. The
supposed to be able to communicate in wifi mode through the authors developed a bottom–up simulation tool of the energy
router of the unit to a cloud. behavior of the building to highlight the impact of the BEMS
In conclusion, the main innovative characteristics of the sug- strategy on the power and energy profiles [23], [24].
gested architecture are as follows: Bottom–up models create load profiles on the appliance or
1) full electric building (gas free), with zero emission and household level and then project these results to represent a
an extremely flexible load; building or a large area. The tool is based on a model that con-
2) unique POC with the electric utility for all the units at templates electric and thermal loads with a statistical approach
the MV level with an owned MV/LV substation; for the presence and the habits of the users. For each apartment
3) centralized heating generation for HVAC and DHW by the tool generates the electric and the thermal profiles. The sum
a GSHP and under floor heating systems; of the electric profiles of all the apartments is added to the
MARTIRANO et al.: DEMAND SIDE MANAGEMENT IN MICROGRIDS FOR LOAD CONTROL IN NEARLY ZERO ENERGY BUILDINGS 1773

TABLE I TABLE II
CHARACTERISTICS OF THE UNCONTROLLABLE LOAD CATEGORIES CHARACTERISTICS OF THE PLANNABLE LOAD CATEGORIES

Load Trend daily Trend time Average Daily

category activation number of use power(W) energy(kWh)

Lighting - - 50 0.7
Fridge and Standby - - 70 1.5
TV 3 2h 250 0.7
PC and multimedia 3 2h 250 0.7
Housework 1 15 m and 30 m 1300 0.2
Hygiene 1 15 m and 30 m 1500 0.6
Kitchen 2 15 m and 30 m 2300 1.5

electric profile of the entire thermal system evaluated as sum

of all the apartments and converted in electricity. This global
building profile is added to the generation profile, giving the
balance at the point of connection.

A. Electric loads for the Apartments 2) “Y” between zero and one, to be compared with the
The electric loads present in the apartments are subdivided value of the probability of activation at the same time X.
into the following. If Y is greater than the probability, the appliance will not be
1) Uncontrollable depend strongly on the habits of the activated and another trial at another time X will be done. For
occupants: No control strategy can be implemented. each type of load, the number of activation during each simula-
2) Plannable depend on the habits of the occupants: tion day and the time of use of each load during each activation
The ending time of the cycle is chosen by the occu- should be calculated. The probability functions of each load are
pants, but the actual starting time can be managed by based on load diagram of each appliance from [25]. For each
the BEMS. Once started, the cycle must end without type of load, the number of activation during each simulation
interruption within the time set by the user. The com- day and the time of use of each load during each activation
fort of the users is improved. In the suggested BEMS, should be calculated.In order to grant the accuracy of the model
this is the power for dishwashers (DW) and WM. and to increase the flexibility of the model, the daily number
3) Controllable may be operated and regulated by the of activations and the use time are simulated using probabil-
BEMS without a discomfort for the habits within cer- ity distributions based on user behavior and parameterized on
tain limits. In the suggested BEMS, this is the power statistical data and heuristic considerations.
for heat pumps that, moreover, exploits thermal inertia Modeling of power demand is divided into five steps.
of the building. 1) Residents going out and the time spent away from home
The simulation model starts from the behavior of individual are simulated with heuristic considerations through a
users, which are represented using a Markov chain Monte Carlo probability function. This provides a profile of avail-
modeling approach based on the daily load profile data to predict ability at home of each unit inhabitant.
the electrical load profile. The research presented in this paper 2) For each load category, the daily number of activations
uses a bottom–up approach, combining the probability of being is determined through a Gauss or a Poisson distribution
at home and the probability of appliance load and summing the centered at the statistical mean.
resulting loads to get the unit load. The global unit load is com- 3) Monte Carlo simulation is used to sample start time of
posed of seven uncontrollable load categories shown in Tables I each load. For each activation in each unit the Tstart is
and II categories of plannable ones (washing machine and DW) defined using a probability curve from statistical data
shown in Table II.In this model, the stay-at-home profile is ob- and availability at home of each resident, for each load
tained considering each occupant labeled as “working” or “not category. Lighting and refrigerator are continuous at
working” and it is possible to integrate the model with other mean measured power. Home availability of inhabitants
users’ characteristics, thanks to the modularity of the model, and sleeping time are considered for lighting too.
due to the MCMC construction. In order to run the model, it 4) Time of use is determined for each load category, after
is necessary to generate samples of each random variable for each load activation, through Gauss or uniform distri-
each Monte Carlo trial. In this analysis, two set of variables are bution based on statistical data provided in [25].
considered, to apply the Monte Carlo method to a single load 5) Building power demand curve is obtained by summing
using probability curve. The two random variables associated electrical load of each apartment.
with each Monte Carlo trial are as follows: Fig. 4 shows an example as through the Monte Carlo method
1) “X" between 1 and 96 (last sample time of a day), which a start time distribution and a load profile for each load may be
indicates the random time of start of the electric load; obtained.
1774 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO. 3, MAY/JUNE 2017

C is the heat capacity of the whole ambient evalu-

ated considering wall contribution up to the effec-
tive thicknesses of the layers interested by a 1 hour
period temperature solicitation;
Ai is the thermal power provided by internal loads (peo-
ple, cooking, etc.); input to the model;
AH ,U is the thermal power input from heating system;
H is the building-to-ambient heat transfer coefficient;
AH ,U is evaluated by (3), that models a PI temperature
controller with set-point temperature Tsp, with satu-
ration at AH ,U max .

 t
AH ,U (t) = Kp ∗ (Tsp (t) − Ti (t)) + Ki ∗ (Tsp (t)
0

− Ti (t)) dt. (3)

D. Thermal Loads for the Building

Once calculated the thermal requirements AH ,U of a single
unit, the model calculates the overall requirements of the build-
ing summing the thermal requirements of each single unit.

AH ,B = AH ,U .

Fig. 4. Results of a daily simulation with 92 DW: the histogram of the starting The heat pump operates controlling the puffer temperature.
times simulated (up) and the load profile simulated compared with the measured Its temperature is evaluated by energy balance equation with
one (down).
implicit time discretization

The electric behavior of each kind of load is simulated by

a vector P of 96 elements for a sample day, providing a mean Ta (t + Δτ ) =
value for each 15 minute, as proposed in [24]. Accuracy of the Ca ∗ Ta (t) + (AP ,B (t) − AH ,B (t)) Δτ + Ha ∗ Δτ ∗ Tac ◦
model, with special regard to stochastic contribution has been ( C)
Ca + Ha ∗ Δτ
tested. (4)

B. Electric loads for the Building Ta is puffer temperature;

The electric behavior of the whole building is given by a Ca is heat capacity of the puffer;
vector PE B obtained by summing load vectors of each unit. AP ,B is power input from heat pump;
The electric energy EE ,B ,d spent in a day in the building is AH ,B is power output to building heating system;
thus Ha is puffer-to-room heat transfer coefficient;
96
Tac is room temperature (wherein the puffer is installed).
 The heat pump operates with a step controller with hystere-
EE ,B ,d = PE B (q) /4 (kWh). (1)
sis loops, separating heat pump-puffer and puffer-units require-
p=1
ments by exploiting the puffer thermal inertia [26]. Then, electric
C. Thermal Loads for a Single Unit power absorbed by heat pump is calculated from the datasheet of
the unit at the operating temperature of the puffer, once evapora-
Heating requirements of each apartment, modeled as a single tor inlet water temperature is provided. Fig. 5 shows an example
zone, are evaluated by energy balance equation with implicit of the results of the thermal model.
time discretization Finally, power demand is averaged on 15 min, in order to get
the electric power due to heating (PHB ). The model furnishes a
Ti (t + Δτ )
vector of 96 elements PH B (q).
C ∗ Ti (t) + (Ai + AH ,U ) Δτ + H ∗ Δτ ∗ Te (t) ◦ DHW model is equivalent to heating puffer, providing PH W
= ( C)
C + H ∗ Δτ vector.
(2)
E. Balance on the Node of the Building
where
Δτ is the considered time step; PV generation is modeled from weather data. Thus, the power
Ti is internal ambient temperature; exchanged with the DSO is obtained by summing all loads and
Te is outside temperature, input to the model; PV power generation together.
MARTIRANO et al.: DEMAND SIDE MANAGEMENT IN MICROGRIDS FOR LOAD CONTROL IN NEARLY ZERO ENERGY BUILDINGS 1775

Fig. 6. Scheme of the thermal system management.

jectives of the suggested BEMS are reducing ES,T (like existing

BEMS), controlling the net global profile at the POC maximiz-
ing SCR, and minimizing Pp, and minimizing the cost of the
energy in case of dynamic costs.
The suggested BEMS operates avoiding to change users’ be-
havior that must be free to use the facilities and the appliances in
full freedom. It operates on scheduling of smart appliances and
Fig. 5. Thermal simulation of a given day. From top: internal temperature Ti
controlling heating system, exploiting its thermal inertia due to
of one apartment, heating thermal power of one unit AH , U , building aggregated building and system.
heating thermal power AH , B , puffer temperature Ta , and heat pump electric
power PH B .
A. Thermal System Management
The suggested BES is able to control the following in real
F. Key Performance Indicators (KPI) time (see Fig. 6):
The performance of different control strategies may be evalu- 1) set point temperature of boilers and puffer (TB );
ated through the following KPI, applied to the resulting profile. 2) comfort set point temperature of each units (TC );
1) ES,T : the yearly energy spent by the thermal system. 3) economy set point temperature of each units (TE );
2) ΔP a: variation of average power P a in comparison to 4) actual ambient set point temperature TS of each unit,
noncontrolled profile. equal to economy or comfort temperature.
3) ΔP p: variation of peak power P p in comparison to The HES is also able to control the following in real time
noncontrolled profile. remotely by HIS:
4) P a/P p: ratio between average and peak powers. 1) comfort and economy set point temperatures of each
5) ΔEG : variation of the energy exchange with the grid in units, forcing values set by BES in a range of ± 2°C;
comparison to noncontrolled profile. 2) actual ambient set point temperature TS of each unit,
6) EH: equivalent hours, equal to the ratio between daily equal to economy or comfort temperature.
absorbed energy from the grid and peak power. A traditional BEMS operates setting:
7) SCR: the self-consumption ratio, equal to the ratio be- TB N the normal operating boilers temperature;
tween self-consumed energy and locally generated en- TC N the normal operating set point in case of the pres-
ergy. ence of inhabitants in the unit (comfort mode);
TE N the normal operating set point in case of the ab-
V. BUILDING ENERGY MANAGEMENT SYSTEMS FOR DEMAND sence of inhabitants in the unit or during nighttime
SIDE MANAGEMENT (economy mode).
The general goal of a building management system (BMS)
B. Control T1
is to manage the HVAC system in order to guarantee an
acceptable comfort for occupants and a correct operation It is possible to reduce the electric peak load in case of high
of the system [27]. A BMS becomes BEMS if it is able to global load or to store energy in case of low (or negative) global
control the energy demand too [28]–[30]. The existing BEMSs load by forcing the set points TB of the central boilers. That is
control the energy performance of the building with the goal of
if p(t) > PM 1 → BEMS forces TB from TB N to TB L
reducing the amount of the yearly energy spent by the thermal
if p(t) < Pm 1 → BEMS forces TB from TB N to TB H
system ES,T maintaining a prospected level of comfort.
The new BEMS approach suggested in this paper reduces being
the yearly energy spent ES,T , as the existing ones, and more it p(t) load demand of the building at POC;
controls instant for instant the electric demand at the point of PM 1 threshold value of maximum power;
interface with the DSO, considering local generation. The ob- Pm 1 threshold value of minimum power;
1776 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO. 3, MAY/JUNE 2017

TB L lower value of the boiler temperature;

TB H higher value of the boiler temperature.

C. Control T2
It is possible to reduce the electric peak load in case of high
global load or to store energy in case of low (or negative) global
load by forcing set points TC of local thermostats
if p(t) > PM 2 → BEMS forces TC from TC N to TC L
if p(t) < Pm 2 → BEMS forces TC from TC N to TCH
being
PM 2 threshold value of maximum power;
Pm 2 threshold value of minimum power;
TC L lower value of comfort mode setpoint;
TCH higher value of comfort mode setpoint. Fig. 7. Scheme of the SA control.
TC L and TCH may vary from 0.5 to 1°C according to the
external temperature Te and season in order to avoid impact on kWh and the local generation. Each time slot in the day is char-
the user’s comfort. acterized by a color (for example: red = more expensive than
the base cost, yellow = base cost, green = cheaper than base
D. Control T3 one). The user considering his needs and the prospected cost,
decides the time (time x) at what the DW or the WM must be
It is possible to reduce global load by scheduling a switching
completed. The BEMS receives the request of a new cycle by
of ambient operation mode from economy (TE N ) to a pre com-
the LS connected to the HIS. So, the BEMS schedules (ex-ante)
fort mode (TP N ) in the first morning and of nonoccupied units
the starting time of the appliance.
before presumed evening reentry time.
The simple rules adopted by the BEMS are as follows.
Simulations have shown that this control is particularly effi-
1) The DW cycle must end before time x.
cient during mealtime in order to avoid contemporary between
2) The WM cycle must end at time x.
the high cooking loads and the other loads.
The rules adopted by the BEMS do not impact the user will,
Moreover, this control improves users comfort as it reduces
indeed improve the level of satisfaction of the service because
heating time from TE N to TC N at evening reentry.
he decides the exact time of ending of the cycle, according to
Control efficacy may be improved during building life time,
the effective cost and his needs.
using measured data available from BEMS itself. The BM may
The BEMS updates the ST introducing the new appliance. The
evaluate the effective load profiles and improve the controls on
BEMS sends periodically the ST to the cloud. The cloud acquires
the basis of actual data.
the updated ST and manages the starting time according to the
TP N is the pre comfort operating temperature set point lower
ST. The BEMS may operate also in real time in order to avoid
than TC N but higher than TE N . TP N may vary according to the
sudden building overloads and to consider sudden changing in
external temperature Te , in order to control comfort and energy
the PV generation.
management.
F. Impact of Controls on the Behavior of the Users
E. Control SA—Smart Appliance Management
The suggested BEMS considers the user behavior that is fun-
The suggested BEMS manages the global demand of the
damental in smart grid management. Traditionally, users are
building by scheduling the start times of smart appliance ac-
reluctant to change the habits. The suggested BEMS improves
cording to users’ will, following objective functions. As stated
the energy and power performances, without forcing the habits,
above, smart appliances considered in this paper are WM and
on the contrary improving the users comfort. Table III shows
DW. The user interacts with the smart appliances by means of
the impact of the suggested controls on the users’ habits and
the HIS as shown in Fig. 7.
comfort.
User behavior is an important part in smart grids and the
BEMS takes into consideration this aspect. In traditional appli-
G. Impact of Controls T1, T2, T3, and SA on the
ances, users decide starting time of the cycle, pushing the start
button. In the suggested model, with smart appliances, users still Energy Performance
decide operation cycle timing, but they define end-time rather The impact of these controls on the energy performance has
than start-time. By using a smart device, the user decides the been evaluated through simulations. Maximum percent reduc-
best time according to the effective cost of the cycle furnished tion of peak power ΔPp% and maximum percent deferrable
by the BEMS through a scheduling timetable (ST) that shows energy ES % (as if there were an energy storage) is shown in
the prospected costs according to the time. The prospected costs Table IV for each suggested control strategy. It should be noted
are evaluated by the BEMS considering the dynamic cost of the that the reduction of energy provided by grid ΔEG % is less
MARTIRANO et al.: DEMAND SIDE MANAGEMENT IN MICROGRIDS FOR LOAD CONTROL IN NEARLY ZERO ENERGY BUILDINGS 1777

TABLE III TABLE V

IMPACT OF THE CONTROLS ON THE USERS HABITS AND COMFORT BEMS SETPOINT PARAMETERS

Control Impact on users’ habits Impact on users’ comfort BEMS Control Strategy TE N -TC N TB N Smart appliance

T1- boiler NO NO °C °C
T2- local setting Yes, but the user can override Negligible, the difference is of N T3 (1 hour before) 16–20 45 By the users
the setting about ½ or 1 degree Ad1 T1, T2 16–19 40–43 By the users
T3- preheating NO Better, because the user find the Ad2 T1, T2, SA 16–19 40–43 By the BEMS with user approval
house warmer when he returns Ad3 T1, T2, SA 16–19 40–43 By the BEMS
home
SA- appliances NO, the user decides Better, because the user decides
the end of the cycle

TABLE IV
IMPACT OF THE CONTROLS ON THE ENERGY BEHAVIOR

Control Δ Pp % ES %

T1 0–30 0–15
T2 0–30 0–5
T3 0–30 0–10
SA 0–10 0–15
T1+T2+T3+SA 0–40 0–30

Fig. 8. Simulated building load, control mode N.

than ES % as not all the energy ES may be moved to the period
of local generation. For the energy evaluation, it is considered
the presence of a PV system with a nominal power of 0.75 times
the peak power of the building.

VI. CASE STUDY OF AN NZEB AND RESULTS

A residential/tertiary building is considered as case study
[31]. The building is located in Campobasso (Italy) and con-
sists of 92 apartments. The building is completely gas free. The
adopted architecture of TBSs is the one described previously.
Heat generation is provided by two geothermal heat pumps,
each with 80-kW rated thermal power. The maximum electric
power absorbed by the heat pumps and auxiliaries is equal to
55 kW. The geothermal field consists in 36 probes grounded at Fig. 9. Simulated building load, control mode Ad1.
100 m that provide 120-kW rated thermal power.
A PV generator is installed with 100-kW rated power.
The building is full wired with an HBES system complete Case Ad3: Heating (controls T1 and T2) and full smart ap-
of a smart metering for each unit: hot water, cold water, heat pliances management (appliances postponement
meters located in the SC, and electricity meter located in the totally delegated to the BEMS—100% of users
main switchboard MSB. Each unit is equipped with a smart are collaborative).
DW, a smart WM, an electric cooker, a refrigerator, and other The complete set point parameters are shown in the Table V.
typical domestic appliances. Control mode N results are shown in Fig. 8 for a sample day
Building load profile has been simulated for four different of March. Control modes Ad1, Ad2, and Ad3 results are shown
cases, in a given day of March. in Figs. 9, 10, and 11, respectively, for the same day.
Case N: Pre heating strategy (control T3) with TP N = Results are summarized in Table VI. The peak power of about
TB N , 1 hour before scheduled heating time in the 120 kW of control mode N is reduced to 93 kW in Ad3 control
morning. mode, with a 22% of decrease, whereas self-consumption of
Case Ad1: Heating (controls T1 and T2). energy generated by the PV reaches a value equal to 100%
Case Ad2: Heating (controls T1 and T2) and smart appli- (SCR = 1.00).
ances management considering the availability It is important to highlight that the Ad3 results are very similar
or not of users to postpone the loads (the simu- to the Ad2 ones, so the collaborative behavior of about 50% of
lation takes into consideration the collaborative the users for the smart appliance management is enough to
behavior of 50% of the users). obtain significant results.
1778 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 53, NO. 3, MAY/JUNE 2017

assistance of a common BACS and a BEMS to exploit its entire

potential. In NZEBs, the impact of these controls can reduce the
peak power up to 40%, to reach self-consumption ratio next to
100% in every condition and to increase the equivalent hours
of use up to 30%, making the best possible use of the TBS and
potentially reduce their nominal power. Simulations on a case
study building settled in central Italy with heating system and
smart appliances has shown that on a reference March day a
12% reduction of peak power and 23% reduction of average
power may be reached even if not all users fully collaborate.
Moreover, load control allows improving self-consumption of
PV generated power that requires fully collaborating users to
avoid grid feeding.
Fig. 10. Simulated building load, control mode Ad2.

REFERENCES
[1] Intergovernmental Panel on Climate Change (IPCC), An Assessment of
the Intergovernmental Panel on Climate Change, Synthesis Report, Va-
lencia, Spain, Nov. 12–17, 2007.
[2] McKinsey and Company, “Pathways to a low-carbon economy,” Tech.
Rep. Ver. 2 of the Global Greenhouse Gas Abatement Cost Curve., 2009.
[3] European Parliament and the Council of 19 May 2010 “The directive
2010/31/eu on the energy performance of buildings,” Official J. Eur. Union,
vol. 53, 2010.
[4] Smart net zero energy buildings, Net-ZEB strategic research center, 2013.
[Online]. Availanle: http://www.solarbuildings.ca.
[5] A. Marszal et al., “ECEEE, zero energy building: A review of definitions
and calculation methodologies,” Energy Build., vol. 43, pp. 971–979,
2011.
[6] Y. Lu, S. Wang, and K. Shan, “Design optimization and optimal control
of grid-connected and standalone nearly/net zero energy buildings,” Appl.
Fig. 11. Simulated building load, control mode Ad3. Energy, vol. 155,pp. 463–477, 2015.
[7] X. Cao, X. Dai, and J. Liu, “Building energy-consumption status world-
TABLE VI wide and the state-of-the-art technologies for zero-energy buildings during
RESULTS FOR A SAMPLE DAY IN MARCH the past decade,” Energy Build., vol. 128, pp. 198–213, 2016.
[8] K. Shan, S. Wang, C. Yan, and F. Xiao, “Building demand response and
control methods for smart grids: A review,” Sci. Technol. Built Environ.,
BEMS N Ad1 Ad2 Ad3
vol. 22, pp. 692–704, 2016.
[9] S. Wang, X. Xue, and C. C. Yan, “Building power demand response
Peak power Pp kW 120 105 95 93
methods toward smart grid,” HVAC&R Res., vol. 20, no. 6, pp. 665–87,
ΔP P p.u. - 0.87 0.79 0.78
2014.
Daily energy spent kWh 998 980 980 980
[10] C. Yan, X. Xue, S. Wang, and B. Cui, “A novel air-conditioning system
GSHP energy kWh 66 49 49 49
for proactive power demand response to smart grid,” Energy Convers.
Daily energy from the grid Ed kWh 641 623 583 577
Manage., vol. 102,pp. 239–246, 2015.
Δ Eg p.u. - 0.97 0.90 0.90
[11] T. Wakui, H. Kawayoshi, R. Yokoyama, and H. Aki, “Operation man-
EH h 8.27 9.3 10.2 10
agement of residential energy-supplying networks based on optimization
Self consumption kWh 357 356 390 392
approaches,” Appl. Energy, vol. 183.pp. 340–357, 2016.
Grid feeding kWh 35 36 1 0
[12] J. Aghaei and M. I. Alizadeh, “Demand response in smart electricity
SCR p.u. 0.90 0.91 0.99 1.00
grids equipped with renewable energy sources: A review,” Renew. Sustain.
Energy Rev., vol. 18, pp. 64–72, 2013.
[13] P. Siano, “Demand response and smart grids—A survey,” Renew. Sustain.
VII. CONCLUSION Energy Rev., vol. 30, pp. 461–78, 2014.
[14] M. H. Albadi and E. F. El-Saadany, “A summary of demand response in
The aggregation of the small users of a building or a clus- electricity markets,” Elect. Power Syst. Res., vol. 78, no. 11, pp. 1989–96,
ter of buildings in a grid supplied by an adequate and private 2008.
network connected with the utility (DSO) by a single POC al- [15] J. Salom, J. Widen, J. Candanedo, I. Sartori, K. Voss, and A. Marszal,
“Understanding net zero energy buildings: Evaluation of load matching
lows to control power use, both economically and technically. and grid interaction indicators, in Proc. 12th Conf. Int. Build. Perform.
The development of NZEBs pushes toward this direction. The Simul., Sydney, Australia, Nov. 14–16, 2011, pp. 2514–2521.
building or group of buildings represents the natural limit of the [16] G. Parise, L. Martirano, and L. Parise, “Ecodesign of ever net-load micro-
grids,” IEEE Trans. Ind. Appl., vol. 50, no. 1, pp. 10–16, 2014.
aggregation, like in heating systems. This allows to get a virtu- [17] P. Arboleya et al., “Efficient energy management in smart micro-grids:
ous and flexible cumulative load profile. Starting from an user ZERO grid impact buildings,” IEEE Trans. Smart Grid, vol. 6, no. 2
view point, the paper suggests a microgrid architecture, with a pp. 1055–1063, Mar. 2015.
[18] A. Franco and F. Fantozzi, “Experimental analysis of a self consump-
BEMS that manages both electric and thermal loads improving tion strategy for residential building: The integration of PV system and
the energy performance of the building and operating a DR. The geothermal heat pump,” Renew. Energy, vol. 86, pp. 1075–1085, 2016.
suggested approach upgrades the NZEB qualification to KZEB [19] I. A. Sajjad, M. Manganelli, L. Martirano, R. Napoli, G. Chicco, and
G. Parise, “Net metering benefits for residential buildings: A case study
one with cost savings and more comfort for the users and re- in Italy,”in Proc. 15th Int. Conf. Environ. Elect. Eng., Jun. 10–13, 2015,
duced the impact on the distributor. Themicro grid needs the Rome, Italy, pp. 1647–1652.
MARTIRANO et al.: DEMAND SIDE MANAGEMENT IN MICROGRIDS FOR LOAD CONTROL IN NEARLY ZERO ENERGY BUILDINGS 1779

[20] Energy Performance of Buildings –Impact of Building Automation, Con- Giuseppe Parise (M’82–SM’03–F’10–LF’15) re-
trol, and Building Management, Eur. Tech. Standard EN 15232, 2012. ceived the M.S. degree in electrical engineering from
[21] L. Martirano, R. Marrocco, F. Liberati, and A. Di Giorgio, “KNX protocol the Sapienza University of Rome, Rome Italy, in
compliant load shifting and storage control in residential buildings,” IEEE 1972.
Ind. Appl. Soc. Annu. Meeting, Oct. 18-22, 2015, Dallas, USA. pp. 1–6. Since 1973, he has been with the Department
[22] L. G. Swan and V. I. Ugursal, “Modeling of end-use energy consumption in of Electrical Engineering, Faculty of Engineering,
the residential sector: A review of modeling techniques, ”Renew. Sustain. Sapienza University of Rome, where he is currently
Energy Rev. vol. 13, no. 8, pp. 1819–1835, 2009. a Full Professor of electrical power systems.
[23] A. Capasso, W. Grattieri, R. Lamedica, and A. Prudenzi, “A bottom-up
approach to residential load modeling,” IEEE Trans. Power Syst., vol. 9,
no. 2, pp. 957–964, May. 1994.
[24] L. Martirano, S. Fornari, A. Di Giorgio, F. Liberati, “A case study of
a commercial/residential microgrid integrating cogeneration and electri-
cal local users,” in Proc. Conf. Environ. Elect. Eng., May 5-8, 2013,
pp. 363–368.
[25] S. Sibilio, A. D’Agostino, M. Fatigati, and M. Citterio, “Valutazione dei Giacomo Greco (S’14) received the B.D. degree in
consumi nell’edilizia esistente e benchmark mediante codici semplificati: automation engineering and the M.D. degree in elec-
Analisi di edifici residenziali,” Italian Ricerca Sistema Elettrico, ENEA, trical engineering from the Sapienza University of
Report RSE/2009/115, 2009. Rome, Italy, in 2012 and 2016, respectively. He is
[26] U. I. Dar, I. Sartori, L. Georges, and V. Novakovic, “Advanced control of currently working toward the Ph.D. degree at the
heat pumps for improved flexibility of Net-ZEB towards the grid,” Energy Sapienza University of Rome, Rome, Italy.
Build., vol. 69, pp. 74–84, Feb. 2014.
[27] E. R. Sanseverino, M. Ippolito, and G. Zizzo, “Impact of building automa-
tion control systems and technical building management systems on the
energy performance class of residential buildings: An Italian case study,”
Energy Build., vol. 69, pp. 33–40, 2014.
[28] G. Graditi et al., “Innovative control logics for a rational utilization of elec-
tric loads and air-conditioning systems in a residential building,” Energy
Build., vol. 102, pp. 1–17, 2015.
[29] A. Dounis, P. Tiropanis, A. Argiriou, and A. Diamatis, “Intelligent control Matteo Manganelli (S14) received the B.S. degree
system for reconciliation of the energy savings with comfort in buildings (Hons.) and the M.S. degree (Hons.) in energy en-
using softcomputing technique,” Energy Build., vol. 43, pp. 66–74, 2011. gineering from the Sapienza University of Rome,
[30] J. Figueiredo and J. Sá da Costa, “A SCADA system for energy manage- Rome, Italy, in 2010 and 2013, where he is currently
ment in intelligent buildings,” Energy Build., vol. 49,pp. 85–98, 2012. working toward the Ph.D. degree.
[31] L. Martirano et al., “Smart micro grids for nearly zero energy buildings,”
in Proc. IEEE Ind. Appl. Soc. Annu. Meeting, Oct. 1–6, 2016, Portland,
OR, USA, pp. 1–8.

Luigi Martirano (S’98–M’02–SM’11) received the Ferdinando Massarella received the M.S. degree in
M.S. and Ph.D. degrees in electrical engineering from electrical engineering from the Sapienza University
the Sapienza University of Rome, Italy, in 1998 and of Rome, Rome Italy, in 1998.
2003, respectively. He is a Professional Engineer in the field of the
In 2000, he joined the Department of Electrical electric power systems.
Engineering, Sapienza University of Rome, Rome,
Italy, where he is currently an Associate Professor of
electrical power systems.

Emanuele Habib received the M.S. degree in me- Luigi Parise (S’06–M’10) received the M.S. degree
chanical engineering and the Ph.D. degree in applied in electrical engineering from the Sapienza Univer-
physics from the Sapienza University of Rome, Italy, sity of Rome, Rome Italy, in 2009.
in 2002 and 2007, respectively. He is currently Research Fellow with the Sapienza
In 2011, he joined the Department of Astronautics, University of Rome, and a Tutor in electrical power
Electrical, and Energy Engineering, Sapienza Univer- systems in hospitals with the University Biomedical
sity of Rome, Rome, Italy, where he is currently an Campus of Rome, Rome.
Assistant Professor of applied physics.