Energy and Buildings 122 (2016) 6379

Contents lists available at ScienceDirect

Energy and Buildings

journal homepage: www.elsevier.com/locate/enbuild

Experimental investigation of the thermal performances of an

extensive green roof in the Mediterranean area
Piero Bevilacqua , Domenico Mazzeo, Roberto Bruno, Natale Arcuri
Department of Mechanical, Energy and Management Engineering, University of Calabria, Ponte P. Bucci 46/C, ZIP 87036 Arcavacata di Rende, Cosenza, Italy

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

Article history: Nowadays green roofs are a well-consolidated technology used in several countries since these systems
Received 11 January 2016 can be employed both to generate considerable energy savings and to improve the thermal performances
Received in revised form 22 March 2016 of buildings. This paper deals with an experimental analysis of an extensive green roof, installed on a
Accepted 23 March 2016
building of the University of Calabria (Italy, Lat. 39.3 N). The thermal performances of different layering
Available online 31 March 2016
solutions are analysed under typical Mediterranean climate conditions with reference to a traditional
roof system. The analysis showed that the green roof is able to reduce the temperature at the interface
Keywords:
with the structural roof, on average, by 12 C with respect to a black bituminous roof in summer and to
Green roof
Experimental analysis
maintain, on average, a value that is 4 C higher in winter. The measured temperatures allowed to calculate
Dynamic properties the transferred heat through the building roof, showing negative heat uxes for the whole period and a
Heat uxes reduction of the thermal energy entering the indoor environment of 100% in summer, proving the passive
Energy savings cooling effect of the green roof, and a reduction between 30% and 37% of the thermal energy exiting the
Mediterranean climate indoor environment in winter. The experimental data have been employed also to conduct a dynamic
characterization of the different vegetated solutions, showing that the different green roofs solutions are
able to generate values of decrement factor varying between 0.135 and 0.193 and a time lag between
3.1 h and 4.8 h, which can substantially improve the dynamic properties of traditional roof structures,
especially in the case of roofs with limited dynamic performances.
2016 Elsevier B.V. All rights reserved.

1. Introduction the improvement of stormwater management. Nevertheless, these

In recent decades, the increasing awareness of the indefensibil-
ity of the actual model of economic growth has led to new concepts
of sustainable development. In the building sector, which is respon-
sible for a relevant demand of primary energy, the trend in the quest
of adequate technical solutions is continuously increasing. Many
systems have been investigated to evaluate performances in terms
of environmental, economic and energy benets. Passive systems
such as green roofs, also named eco-roofs, living roofs or roof gar-
dens [1], belong to the studied systems. Nowadays green roofs are a
well-consolidated technology used in several countries around the
world as an innovative roong solution. Vegetated roofs have been
used since ancient times with the most famous example being the
Babylonian gardens in Mesopotamia [2], but more recent examples
can be found especially in Northern European countries. Indeed
this technology has historically proven validity in the improve-
ment of the building envelope insulation in cold climates and in
Corresponding author.
E-mail address: piero.bevilacqua@unical.it (P. Bevilacqua).
systems allow the achievement of several other benets, such as
the reduction of rainwater runoff [35], the mitigation of urban
heat island effect (UHI) [6,7], the improvement of membrane dura-
bility and extension of roof life [8,9] and the reduction of heating
and cooling demands for the buildings air-conditioning [1015].
Finally, green roofs sequester carbon dioxide and reduce local-
ized air pollution [1619], increase the water runoff quality [20],
create habitat for wildlife and recreational opportunities [2124]
and decrease urban noise pollution [25,26]. Therefore, green roofs
actually represent an innovative construction system that can
increase the sustainability of buildings and cities.
Vegetated roofs are categorised as intensive, simple intensive
(or semi-intensive) and extensive greenings according to the level
of maintenance required, but often only the extensive and inten-
sive categories are considered. Intensive roofs are deep, heavy,
require a high level of maintenance and, usually, they are designed
for complete accessibility. This type of roof is usually applied to
new buildings in which the extra weight is kept in considera-
tion during the design of the structural components. The extensive
ones, instead, are characterized by lightweight growing media
of reduced depth; they require low maintenance and are less

http://dx.doi.org/10.1016/j.enbuild.2016.03.062
0378-7788/ 2016 Elsevier B.V. All rights reserved.
64 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
likely to be designed for frequent human access. Typical extensive
green roofs are equipped with self-sustaining and native species
of plants, therefore they require low irrigation since the vege-
tation is well adapted to local climatic conditions [2729]. The
plants represent a necessary element to realize the evapotranspi-
ration process in the substrate layer, crucial for the achievement of
cooling effects [30]. Moreover, the reduced system weight makes
them suitable for installation on existing buildings. A typical exten-
sive green roof layering, together with the materials generally
employed, can be found in [3134]. Different constructive solutions
are nowadays available, that are often not standardized and may
be composed of different types of layers, materials, thicknesses and
vegetation. Consequently a more comprehensive datasets to report
quantitative information about the performance for the different
technologies adopted is required [29].
From an energetic point of view, as the surface of the roof
increases, in relation to external walls surface, the heat ux through
the building cover assumes a more relevant contribution in the
energy requirements of the building upper oor air-conditioning
[35]. In this regard, the physical phenomena involved in the energy
balance of the several layers of an extensive green, such as veg-
etation, substrate and drainage roof, lead to the reduction of the
thermal loads of the indoor environment, both in summer and
winter. Regarding the vegetation layer, the heat uxes with the out-
door environment involve absorption of solar radiation, long-wave
radiative exchange with the sky, convective exchange with external
air, precipitation and evapotranspiration processes. Similar physi-
cal phenomena are involved in the thermal balance of the soil, with
the addition of the conductive heat ux and the sensible heat stored
in the structure. The theoretical investigation of these systems
is very difcult due also to the strong variation of the thermo-
physical properties of the soil and drainage layer with the water
content (WC) [36]. In summer, the shading effect of vegetation,
the increased thermal mass of the system and evapotranspiration
phenomena, permit the reduction of heat uxes entering the build-
ing through the roof. In winter, the green roof offers additional
insulation to reduce heat losses from indoor spaces to the outdoor
environment.
In the recent literature many researchers focused their stud-
ies on the mathematical modelling of the heat and mass transfer
in a green roof in order to be able to assess the resulting ben-
ets [3642]. Such topic, given the complexity of the physical
phenomena involved, is still under development. Consequently
many authors addressed the problem by using mathematical model
solved with a numerical approach; however the results obtained
provide only a qualitative accuracy. Niachou et al. [43] estimated
the impact of a green roof on the total energy consumption of
buildings with various types of roofs, with and without insula-
tions, through a simplied mathematical approach, simulating the
presence of the vegetated roof by properly modifying the thermal
conductance of the building cover. They observed that the great-
est yearly energy saving, in the case of non-insulated buildings,
was of 37%. Spala et al. [44], using the dynamic simulation pro-
gram TRNSYS, calculated the energy consumption of a two-storey
building in Athens. They observed a remarkable reduction in the
cooling load with a maximum of 39% for the whole building, and of
58% for the last oor. Moreover, they observed a maximum decre-
ment of heating loads of 8% for the whole building, and 17% for the
last oor. However, not many details were given about the math-
ematical approach used. Another group of researchers, by using a
more detailed green roof model based on the numerical procedure
implemented in EnergyPlus, calculated the energy performances
of an ofce building with various typologies of green roof and a
high reective coating, in different European climate conditions
[45]. They concluded that in warm climates green roofs are suit-
able for reducing the energy demand for the space cooling, without
penalizing the scarce heating demand, and in cold climates they
are useful for reducing the thermal losses in winter. However the
thermal transmittance value in the study was not varied, since they
considered as a case study a well-insulated building.
Other researchers have evaluated the performance of green
roofs through experimental analysis. The results obtained by Liu
and Minor [9] in the city of Toronto showed that green roofs are
effective in reducing heat uxes through the roong system, thus
lowering the energy demand for space conditioning in the build-
ing. The additional green roof systems, compared to the reference
roof consisting of steel deck, gypsum board, vapor retarder, ther-
mal insulation, breboard and modied bituminous membrane,
reduced the heat uxes through the roof by 7090% in summer and
1030% in winter. The green roofs also reduced the roof membrane
maximum temperature in summer by more than 20 C and daily
temperature uctuations experienced by the roof membranes by
about 30 C. Their results refer to a humid continental climate local-
ity, and in the study the temperature inside the building was kept
constant so that any heat ow between the building and its environ-
ment created energy demand for space conditioning. Therefore, the
energy demand due to the roof was estimated by the sum of the heat
entering and leaving through the roof. In similar climatic condi-
tions another study, conducted by Teemusk and Mandar in Estonia,
compared experimentally the temperature regime of a lightweight
aggregates based roof garden with a modied bituminous mem-
brane roof in different season [8]. The results of their study revealed
that the roof garden can decrease the temperature uctuations sig-
nicantly in summer, protecting roof membrane from rapid cooling
and freezing in autumn and spring, and provide effective thermal
insulation in winter. In a Midwestern U.S. climate Getter et al. [46]
quantied the thermal properties of an inverted 325 m2 extensive
green roof installed over a traditional gravel insulated roof, taken
as reference. Their results demonstrated how an extensive green
roof inuences temperatures and heat uxes in the roof during dif-
ferent seasons of the year, with a reduction of heat uxes, through
the building envelope, by an average percentage of 13% in winter
and 167% in summer. In terms of maximum and minimum temper-
ature, the monthly average values over the course of the year were
consistently higher for the gravel-ballasted roof than the green roof,
with values up to 20 C warmer during summer. The study reported
in [47], regarding ve typical days, showed that a green roof test
bed in Singapore (China), can reduce the internal air temperature
by an average value of 0.5 C if compared with a bare roof.
Even though many examples of experimental analysis can be
found in the literature, most of the available researches on green
roofs rely on simulated data. Furthermore one of the main factors
affecting the performance of a green roof is the climatic conditions,
therefore generalizable results are difcult to assess. In this regard
the performance of green roofs in winter period is still a matter
of debate where some scientists claimed it as a medium to save
energy and some viewed it as a cause of more energy consumption
[1].
Other experimental studies have been conducted in Mediter-
ranean climate conditions. Coma et al. [48] evaluated the energy
consumption and thermal behaviour of three identical house-like
cubicles located in Puigverd de Lleida (Spain): two of them were
equipped with extensive green systems with different drainage
layer materials, pozzolana and rubber crumbs, installed on a tradi-
tional non insulated at roof, and one with a traditional insulated
roof. They measured the electrical energy consumption of a heat
pump system for more than one year. Their results indicated that
both extensive green roof cubicles generated less energy con-
sumption (16.7% and 2.2%, respectively) than the reference one
during warm periods, whereas higher energy consumption (6.1%
and 11.1%, respectively) during heating periods were detected.
DOrazio et al. [49] assessed the yearly thermal performance of
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 65

Table 1
Climatic data of Cosenza.

Units Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Monthly average C 8.1 8.8 11.3 14.4 18.1 23.1 26.0 25.8 22.7 17.8 13.4 9.4
daily temperature

Maximum monthly C 17.3 18.0 21.6 26.6 30.1 32.5 36.5 34.8 34.1 27.0 22.0 18.0
average daily temperature

Minimum monthly C 1.8 1.9 0.4 2.4 6.3 9.5 11.9 13.0 10.2 5.3 0.6 1.3
average daily temperature
Average monthly mm/month 139.4 119.3 95.2 72.5 52.1 26.3 14.9 22.7 56.3 99.4 140.2 151.4
daily rainfall
Number of Rainy Days days 10 9 8 10 6 4 3 3 8 8 10 13
Monthly average MJ/m2 7.7 11.8 17.3 21.8 25.7 29.6 28.9 26.0 20.0 12.9 9.4 7.7
daily solar radiation on the
horizontal surface

an extensive green roof and compared it with that of six other
type of roof, of similar size. The roof systems were installed on a
real-scale high-insulated experimental building under temperate
climate in the vicinity of Ancona (Italy) and the study was con-
ducted in absence of air conditioning. Their results demonstrated
that in winter the green roof is able to guarantee further insu-
lation even in saturation conditions while, in summer, the green
roof mitigates incoming heat uxes and ceiling temperatures, even
though its performance is partially hindered by high insulation. In
another work, a 2 C decrement in summer internal air temperature
by the installation of a green roof on buildings located in Athens was
determined [43]. For the same climatic condition and in absence of
air-conditioning, in [50] are presented the results of a long-term
experimental analysis to identify the thermal behaviour of a green
roof installed over a bare at roof with a low thermal insulation,
that was considered as reference. During the experiment, in both
buildings, the storey under the roof remained unoccupied during
summer while in the rest of the year, only the last storey under
the green roof was occupied by one person during ofce hours.
Their result showed that the green roof cannot provide consider-
able additional thermal insulation in order to minimise heat loss
within the roof during the cold season, while it can contribute sub-
stantially to building energy conservation during the warm period
of the year. Schweitzer and Erell [51] evaluated the energy per-
formances and assessed the cooling effect of extensive green roofs
through reduced-scale models of buildings equipped with planted
roofs in the campus of Tel Aviv University (Israel). They monitored
the daily evaporated water and calculated the associated latent
heat. Their results highlighted that the cooling efciency of all roof
variants studied was very low, since the reduction of the building
energy load attributed to the green roof system was less than 5%
of the calculated evapotranspiration heat. Olivieri et al. [52] con-
ducted an experimental measurements for the assessment of the
summer performance of extensive green roofs in a locality near
Ancona (Italy). The roof cover of the real-scale experimental build-
ing was insulated with a double layer of expanded polystyrene.
The whole roof was divided into six types of roof covering, made
of either brick or metal, with a slab in either pinewood or con-
crete and with or without under covering ventilation. Their results
highlighted that when the roof has dense vegetation the entering
thermal gain went down by about 60% compared with the roof with
no vegetation. Moreover the roof with vegetation acted as a passive
cooling system since the outgoing energy throughout the summer
was about 9% greater than the incoming energy during the same
period.
In this study, an experimental analysis of an extensive green roof
installed on a real-scale University building in Cosenza (Italy, Lat.
39.3 N) is presented. The paper concerns the evaluation of the tem-
perature levels and the thermal performances of different layering
solutions compared to a reference conguration representing the
original conguration of the roof surface before the installation of
the green roof. The external coating of the reference roof is made of a
black bituminous waterproof layer, which is a solution still adopted
for at roof buildings even though does not represent the majority
of the roofs in the area of the study. The analysis was conducted
both in summer and winter seasons in a typical Mediterranean
climate. The analysis showed the thermal behaviour of the green
roof in terms of average values and temperature uctuation in the
layers of the different stratigraphies, demonstrating the ability of
this system to provide benets both in summer and winter. More-
over the effect of the green roof on the conditions of the indoor
environment was assessed for three internal rooms of the last
oor of the building properly monitored. Furthermore, the mea-
surements of layers temperature have permitted the assessment
of the heat uxes through the structural roof, and to quantify the
achievable energy savings for the air-conditioning of the consid-
ered building per square meter of roof. Finally the experimental
data were used in an innovative way to perform a dynamic charac-
terization of different green roof solutions, by means the evaluation
of the decrement factor and the time lag of the thermal wave trans-
ferring through this roong solutions, providing data hardly found
in the literature.

Fig. 1. Plots layering and position of the installed sensors.

66 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379

2. Methodology

m2 K/W

m2 K/W
W/m K

W/m K

W/m K

W/m K
kg/m3

kg/m3

kg/m3
J/kg K

J/kg K

J/kg K
Unit
2.1. Climatic conditions

Value

0.034

0.098
The University of Calabria is located in Southern Italy, where the

1307
1210

1200
1400

1000
0.27

0.71
0.16

400
0.4
climatic conditions are typically Mediterranean, dened as subtype
Csa according to the Kppen climate classication. The summers are

Thermal Conductivity

Thermal Conductivity

Thermal Conductivity

Thermal Conductivity
hot and dry, while winters are rather wet. Cosenza is situated in a

Thermal Resistance

Thermal Resistance
valley surrounded by mountains, so it shows a particular micro-
climate. The temperature excursion, both at a daily and at a yearly

Specic heat

Specic heat

Specic heat
level, is very high. In particular, the minimum temperatures reach

Property

Density

Density

Density
low values during the night due to the heat exchange and the ther-
mal inversion, while in summer the temperatures are very high,
especially in non windy conditions. Precipitations are heavier dur-
ing the colder months. The monthly average daily values of the

Thickness [mm]
main climatic parameters are reported in Table 1.

2.2. Experimental set-up and monitoring system

Plot 3

1.8
55

75
80

30
3
The investigated building has a square plan with side of 21.30 m

m2 K/W

m2 K/W
and a height of 18 m; it consists of ve oors and one underground.

W/m K

W/m K

W/m K
kg/m3

kg/m3

kg/m3
J/kg K

J/kg K

J/kg K
The experimental extensive green roof with a surface of 200 m2 was

Unit

installed in October 2012 on the 380 m2 of the available plane sur-
face in the Department of Mechanical, Energetic and Management

Value

0.098
1925
1307
1210

1000
0.27

0.71

900

400
Engineering of the University of Calabria. The structural part of the

0.4

0.2

plane roof consists of reinforced concrete joists and polystyrene
lightening elements, with a total thickness of 31 cm. Four plots

Thermal Conductivity

Thermal Conductivity

Thermal Conductivity
Thermal Resistance

Thermal Resistance
were installed on the building roof for a total green roof area of
about 150 m2 . In particular three green roof plots, with a surface of

Specic heat

Specic heat

Specic heat
approximately 50 m2 each, were built with specic layering (Plots

Property
13 of Fig. 1) whereas one plot was left as the reference traditional

Density

Density

Density
roof (RR). Plot 1 and Plot 2 are made of the following layering, start-
ing from the bottom to the top: an additional recovery waterproof

layer; a lightweight concrete layer to provide a 1% slope for rainfall
Thickness [mm]

water management (from 10 cm to 5 cm); a root-barrier waterproof

layer (INDEX Italia Defend and ZINCO Harpoplan ZD UV); a water
storage and protection layer (ZINCO TSM32 and ZINCO BSM64); a
Plot 2

drainage and water storage layer (CLIMAGRN Climadrain 55 and

75
80

40

7
4
CLIMAGRN FKD 40); a lter sheet (ZINCO TG and CLIMAGRN Fil-
ter sheet 105); growing media (HARPOSEIC Terra Mediterranea)
m2 K/W

m2 K/W
W/m K

W/m K

W/m K
Thermal properties and geometrical characteristics of the materials of the different plots.

kg/m3

kg/m3

kg/m3
J/kg K

J/kg K

J/kg K
and Mediterranean vegetation. The main difference between Plot 1
Unit

and Plot 2 lies in the hydraulic properties of the material installed

since they are made of different commercial materials; however

Value

0.098
1925
1307
1210

1000

this aspect is not investigated in the study. Plot 3 was made with the
0.27

0.71

900

400
0.4

0.2

same layers but additional insulation layer of 3 cm (ISOLKAPPA Isol-

plan 120) was placed under the water storage layer and vegetation
Thermal Conductivity

Thermal Conductivity

Thermal Conductivity

was not planted. The thermal properties and geometrical character-

Thermal Resistance

Thermal Resistance

istics of the different layers of the plots, provided by manufacturers

of the different layers, are reported in Table 2. The growing medium
Specic heat

Specic heat

Specic heat

is mainly constituted by of lapillus of different grain size, a reduced

Property

Density

Density

Density

content of pumice and a percentage of organic matter of less than

6%. The planted species in Plot 1 and Plot 2 are sedum type, in partic-

ular Dianthus grantianopolitanus, Carpobrotus edulis and Cerastium

tomentosum.
Thickness [mm]

The last plot, with a surface of approximately 50 m2 , is the

Reference Roof and represents the original conguration of the
roof surface, before the installation of the green roof, with an
Plot 1

55

75
80

external coating made of a bituminous waterproof. The surface is

3
4

dark-coloured and the estimated value of the reection coefcient

Drainage and storage

(albedo) according to [6] is 0.05. Such a roof cover solution, even if

does not represent the totality of roofs in the area, is of great inter-
est from a thermal point of view because it induces high solar gain
Lightweight
Waterproof

in winter but also high loadings due to solar radiation in summer.

Protection
Insulation
Substrate

Meteorological data are monitored by a weather station, shown

Table 2

Layer

crete
con-

in Fig. 2, placed at the south-eastern corner of the roof, that mea-

sures: ambient air temperature, relative humidity, precipitations,
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 67

Fig. 2. Experimental site and meteorological station.

atmospheric pressure, wind direction and speed. The tempera-
ture sensor with solar shield has an operative range from 40 C
to +80 C with an accuracy of 0.4 C, the relative humidity sen-
sor works in the range of 0 100% with an accuracy of 3%. The
wind speed sensor has an operative range from 1.5 m/s to 79 m/s
and an accuracy of 5%. The rain gauge sensor has a resolution of
0.2 mm. For the measurement of total horizontal solar radiation a
secondary standard EPLAB pyranometer is used whereas the ther-
mal infrared sky radiation is measured with an EPLAB precision
infrared radiometer.
In each plot two acquisition point (A and B), properly spaced
from their edges to avoid boundary effects, were installed for the
monitoring of the main layers variables. In each point, along the
vertical prole, a probe simultaneously measure the temperature
and volumetric water content (WC) in the substrate. Two resis-
tance temperature detectors (RTD) were installed at the interface
between the lightweight concrete and the root-barrier waterproof
layer, and between the traditional roof surface and the lightweight
concrete. A further RTD was placed on the top of the insulation layer
in Plot 3 (Fig. 1). In the reference Roof, accordingly, only surface
temperature is measures by RTDs properly shielded and positioned
in two acquisition point.
The soil probe has accuracy of 2% for the measurement of soil
moisture in an operative range from 0% to saturation and an accu-
racy of 1% for the measurement of temperature with an output
from 50 C to +50 C. The RTD sensors are four wires PT100 class
1/3 with accuracy of 0.1 C at 0 C and 0.27 at 100 C.
The green roof is equipped with a drip irrigation system man-
aged by modules that operate the activation and shut-down
according to the values provided by adequate water presence sen-
sors installed in the drainage layers. The irrigation is provided
during hotter months in summer to prevent water stress of plants.
In Plot 1 irrigation operates from 22:00 to midnight while in Plot
2 it operates from midnight to 2:00. The articial daily watering

Fig. 3. Roof plan with Plots and Monitored internal rooms of the building.
68 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
amount is 3.56 kg/m2 day in both plots. Plot 3 does not require irri-
gation since the vegetation is absent. Irrigation water is taken from
a dedicated storage tank that collect rainfall for grey water reuse
for the watering of soil and plants. When the tank does not contain
a sufcient level of water, irrigation is provided by water from the
supply network.
The effect of the green roof on the internal conditions have been
assessed by considering only the top oor of the building, since
it mostly benets from the presence of the green roof. For this
purpose, sensors for the monitoring of internal air temperature, rel-
ative humidity and room ceiling surface temperature were installed
in the internal environments in three rooms located directly below
the three green roof plots (Room 1 is positioned beneath Plot 1
and so on). The same measurements were made in the Reference
Room, located in correspondence of the reference roof. Tempera-
tures were measured with RTD sensors with an operative range
from 40 C to +75 C and an accuracy of 0.5 C. The relative
humidity sensor work in the range 0100% with an accuracy of 2%.
These sensors are equipped with an internal data logger that stores
the measurements with a sample rate of 5 min.
From Fig. 3 it is possible to observe that the four rooms have
different expositions since they are located on different sides
of the building. Such an aspect would make any comparison
between indoor conditions of internal rooms difcult since oth-
ers parameters, such as solar radiation incident to the wall and
penetrating glazed surfaces, contribute to the thermal balance of
the indoor air temperature nodes. However in this particular case
the south-facing building wall is shaded by the adjacent building
and consequently is not exposed to direct solar irradiation during
the day. Furthermore Room 1, Room 3 and Reference Room do not
present east and west exposed windows whereas Room 2 has an
east-facing window that is located in a lodge and partially shielded
by an external window grate; therefore the solar radiation pene-
trating trough this glazed surface is limited. Considering also that
the four rooms have comparable plan surfaces it can be assumed
that, with a certain margin of error, a comparison between indoor
conditions can be conducted to evaluate the benets obtainable
by means of the different green roof congurations compared to
the reference case. In this regard, considering that in every room
the external wall surface is small compared to the roof surface,
heat uxes through the roof cover assume a cardinal role in the
determination of indoor room air temperature.
The last oor of the investigated buildings, including the four
investigated rooms, is dedicated to work activities and occupied
by research and teaching personnel. The building is occupied from
9:00 AM to 6:00 PM with one hour time break for lunch, so
accordingly, occupants schedule is mainly dictated by the work
regulations. The electric appliances and lighting use follows conse-
quently the work time schedule. In the considered academic year
Room 1, Room 3 and Reference Room were occupied by one person
each whereas Room 2 was not occupied. The rooms are equipped
with one desktop laptop with a nominal power of about 150 W and
are lighted by surface mounted luminaires equipped with uores-
cent lamp with a total power approximately of 230 W per room.
However, it must be highlight that, the investigated periods
were chosen during holidays, or during suspended work activities.
For this reason, as a consequence, the measured indoor variables
are representative of free oating conditions with the absence of
internal heat gains. In this regard, the occupant behaviour does not
affect the results of the analysed winter and summer periods.
A dedicated data acquisition system DAQ acquires all the moni-
tored variables. Every sensor is connected to the acquisition board
by dedicated modules. A proper acquisition software generates
a data structure (values array), where one sample per minute is
stored for each channel and saved into the PC le system, in a matrix
structure.
3. Thermal behaviour analysis
The investigations were carried out with reference to a repre-
sentative period in the hottest and in coldest weeks of the year, by
analysing the response of the different layers of the plots. Present-
ing one-week results rather than daily ones allows for misleading
conclusions related to short-term events such as brief rainfall
episodes and temporary overcast sky conditions to be eliminated.
Furthermore, the effect of green roofs on indoor environment con-
ditions was assessed to demonstrate its inuence in improving the
thermal conditions of indoor spaces.
3.1. Summer behaviour
Fig. 4 reports the climatic conditions for the selected summer
period in 2013 (from 22/08/2013 to 28/08/2013), identied by clear
sky condition, high external air temperature and an absence of
precipitations. Total horizontal solar radiation reached a peak of
862 W/m2 , the external air temperature varied between 19.9 C and
32.8 C, relative humidity between 35% and 80% and atmospheric
pressure between 974.8 hPa and 980.6 hPa.
Temperatures in the substrate and volumetric WC in the acquisi-
tion points (A and B) of Plot 1 are shown in Fig. 5a. The temperature
in the growing media follows the pattern of external air temper-
ature and ranges between 19.3 C and 35.3 C with a mean value
of 26.1 C in point A and 25.1 C in point B. The peculiarity of the
irrigation system allows for an evaluation of the spatial inuence of
WC on soil temperature proles. It is clear from Fig. 5a that higher
WC is related to lower temperatures. Acquisition point B is more
affected by the dropped water so that when irrigation starts, the
water content abruptly rises to 13.7% and then declines during the
day due to the evapotranspiration effects. As a consequence of WC
growth, an immediate drop in substrate temperature (TsB ), which
falls to the minimum daily value, is observed. In point A, where WC
showed more limited daily variations, of between 10% and 12%,
the correspondent substrate temperature showed slightly higher
values, especially during the irrigation time schedule. However, a
correlation between the WC and temperature in the growing media
appears; a greater WC can lead to more appreciable temperature
differences. Accordingly, Jim et al. [53] indicated the substrate-
moisture effect in delaying temperature rise in soil in their results.
The pattern on the top of the lightweight concrete layer evi-
dently indicates the ability of upper layers (substrate and drainage)
to attenuate temperature oscillations (Fig. 5b). The maximum daily
temperature excursion, dened as the range between maximum
and minimum temperatures, is considerably lower and the tem-
perature varies between 22.2 C and 26.1 C in point A with a steady
component equal to 24.1 C and between 21.3 C and 25.7 C in
point B with a steady component equal to 23.2 C. A further notable
contribution in attenuation of the maximum daily temperature is
given by the lightweight concrete, and the temperature measured
on the bottom of this layer can be considered fairly stable around
the steady value of 24.9 C in point A and 24.0 C in point B. The
effect of substrate moisture in suppressing the vertical tempera-
ture gradients is also detected in the lower layers of the layering.
The time lag of the temperature peak produced by the lightweight
concrete is 5 h, whereas the total time lag due to the presence of
the additional green stratigraphy is approximately 10 h.
Similar results were found in Plot 2 where a different irriga-
tion schedule was set. In this case, the greater WC values produce
a lower temperature in the growing media but the drastic drop
detected in plot 1 is not observable. However, in acquisition point
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 69

Fig. 4. Climatic condition of the selected summer period. Solar radiation, sky temperature and external air temperature (a). Rainfall, windspeed and relative humidity (b).

B, where a lower WC is registered, again temperature tends to be
higher either in minimum and maximum daily values. The range
is between 20.9 C and 36.0 C around a mean value of 27.6 C in
point A and between 19.3 C and 34.3 C around a mean value of
25.7 C in point B. This difference is reected in lower layers; again,
the temperature measured under the lightweight concrete is lower
in point B yet the deviation is not so accentuated.
In Plot 3 the combined effect of absence of the vegetation
and presence of an additional insulation layer determines rela-
tively higher temperatures value in the growing media that exceed
the operative temperature range of the sensor. Furthermore, the
absence of irrigation induces low levels of WC that varies around
the value of 3.5% and insignicant differences were measured
between the different acquisition points, so that only the proles of
point A are displayed (Fig. 6a). Such a low water level could not be
sufcient to produce relevant cooling effects linked to evaporation,
a further element that contributes to increment the growing media
temperature that reached 50 C with a maximum daily temperature
excursion of 25 C and a steady value of 36.4 C. The temperature
above the insulation layer ranges between 24.4 C and 37.3 C: the
effect of this additional layer in dampening the temperature uc-
tuation is detected clearly in the prole of the temperature above
the lightweight concrete layer, which varies between 27.4 C and
30.4 C. Again, the temperature under this layer can be considered

Fig. 5. Temperature proles in the layers. Substrate temperature and water content in acquisition points of Plot1 (a). Temperature at the bottom and top of the lightweight
layer in acquisition points of Plot1 (b).

Fig. 6. Temperature proles in the layers. Substrate temperature and water content in Plot3 (a). Temperatures underlying the substrate in Plot 3 (b).
70 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379

reasonably constant, oscillating around 28.8 C (Fig. 6b). In the ref-
erence roof, owing to the high values of external air temperature,
of incident solar radiation and of the solar absorption coefcient
of the black-coloured external surface, high values of the external
surface temperature peak were detected and the maximum daily
temperature excursion reached extremely high values (Fig. 7a). The
temperature ranges from 17.7 C to 64.8 C with a mean value of
36.5 C. Due to the incidence of solar radiation, the membrane tem-
perature rises, reaching the peak at the same time of the maximum
value of solar radiation. Then, the temperature decline with a lower
slope since the external air temperature predominates the pattern
of the supercial temperature when solar radiation assumes lower
values. At night, the longwave radiative heat exchange with the sky
permits membrane cooling and allows attainment of the minimum

Fig. 7. Surface temperature in Reference Roof, air temperature and solar radiation in the selected summer period (a). Comparison between average external surface
temperature in Reference Roof and temperature below the lightweight concrete on the other plots (b).

Fig. 8. Evapotranspiration heat and solar radiation incident on Plot 1 in the considered summer period.

Fig. 9. Trend of internal air temperature of the four rooms located underneath the roof plots for the considered summer period (a). Inverse cumulative distribution of internal
air temperature in August 2013 (b).
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
daily value that is found before sunrise. Similar ndings regarding
the temperature values of a reference roof are reported in [54,55].
In Fig. 7b a comparison between the average external surface
temperature of the reference roof and the temperature below the
lightweight concrete of the other plots is made. It is evident that
the green roof is able to produce lower temperature levels at the
interface with the structural roof, due to the shading effect of
vegetation, evapotranspiration phenomena and thermal resistance
growth. Plot 1 and Plot 2 present very similar temperature values
at the bottom of the green stratigraphy that is approximately 12 C
lower than the average supercial temperature of the reference
roof. Due to the presence of insulation, as expected, Plot 3 guaran-
tees a temperature level that is, on average 4 C higher than Plot 1
and Plot 2.
The plots irrigation enhances the cooling effect by means of
an evapotranspiration phenomenon. This aspect is an important
contribution to the energy balance of the green roof. Therefore a
qualitative estimation of the energy associated with evapotranspi-
ration has been conducted. From Fig. 5a it is possible to observe
that in point B the irrigation of the vegetation causes an imme-
diate growth of the substrate water content. This value however,
decreases during the following hours until the irrigation system
starts to operate again. The minimum daily value remains fairly
constant in all the day of the considered period. From these obser-
vations it can be concluded that the quantity of water provided by
irrigation is eventually converted into latent heat through evapo-
transpiration. In this case considering the difference between the
maximum and minimum WC daily value, the amount of evapo-
transpirated water and the associated energy can be determined.
According to [37] the water ux in a green roof can be expressed
as:
ET
=

where is the latent heat of vaporization (2422 kJ kg1 at 35 C) and
ET is the thermal ux due to the evapotranspiration process. Rear-
ranging the equation, the heat associated with evapotranspiration
LET in kWh, can be expressed as:
LET = Vw  
where Vw in m3 , is the daily amount of water loss in the substrate
and is the water density. The last term was then compared to the
solar radiation incident on the plot surface, since this is the main
loading of the thermal balance in summer condition. The measure-
ments of the pyranometer provide hourly values of global radiation
on the horizontal surface; the integral of such values over the day-
light period provides the daily energy collected by the green roof
plot surface. Results of the calculation are provided as an average
of the evapotranspiration heat for both acquisition point A and B.
Similar results were obtained in Plot 2, whereas in Plot 3, as already
observed in the layers temperature gures, the evapotranspiration
effects are of negligible account. For these reasons only the values
obtained for Plot 1 in are indicated Fig. 8.
Evapotranspiration accounts for from 26% to 35% of the inci-
dent solar radiation on the plot surface. The results, in terms of
latent heat, are in agreement with [51] who performed experiments
in a reduced-scale models of buildings equipped with planted
roofs, monitoring the total daily amount of water evaporated and
calculating the latent heat from the latent heat of vaporization; fur-
thermore, they are also in agreement with [38] which, through a
quasi-steady state heat and mass transfer green roof model val-
idated with robust experimental data obtained from laboratory
experiments, calculated the evapotranspiration term of the energy
balance for different simulation cases. Conversely in [39] using a
mathematical model it was found that for a typical summer in
71
China, 58% of the heat from a green roof was lost by evapotran-
spiration.
To assess the effectiveness of green roofs in producing energy
saving in the building, the effect on the indoor environment was
considered. Fig. 9a reports the trend of the internal air temperature
together with the external air temperature trend for the same sum-
mer period considered in the thermal analysis of the layering for
the four rooms located just beneath the correspondent plots. It is
important to underline that for all the analysed period, the building
was not occupied, therefore, the measured variables are represen-
tative of free oating conditions. The gure clearly shows that the
reference room presents the highest values of internal air temper-
ature followed by Room 2, Room 3 and nally Room 1, where the
temperatures are lower than 28 C for the whole period. Room 2
is evidently affected by other factors since it presents a more pro-
nounced daily temperature excursion, due to the major presence of
glazed surfaces and to the greater vertical dispersing walls surface
whereas in the other rooms the pattern of air temperature is much
less variable with limited daily excursion. On average, Room 1 pre-
sented an internal air temperature 2.3 C lower than the Reference
Room, whereas Room 2 and Room 3 were respectively 0.8 C and
1.2 C lower.
A further conrmation of the benecial effect of the green roof
derives from the analysis of the temperature cumulative distri-
bution for the entire month of August 2013, as shown in Fig. 9b.
The graph reports on the y axis the percentage of time, to which
corresponds a value of internal air temperature greater than the
correspondent value in the x axis. The curve relative to the Ref-
erence Room reports temperature levels signicantly greater than
the other rooms, denoting an undisputed benecial effect of the
green roof on the indoor environment. Temperatures higher than
28 C are observed with the time percentage of 95% for Reference
(1) Room, 89% for Room 2, 75% for Room 3 and 60% for Room 1.
3.2. Winter behaviour
Fig. 10 reports the climatic conditions for the selected winter
week in 2014 (from 24/01/2014 to 30/01/2014) characterized by
low values of solar radiation with a peak of 569 W/m2 , external air
(2) temperature variable between 1.4 C and 15.3 C, relative humidity
between 40.3% and 90.8%, atmospheric pressure between 820 hPa
and 992 hPa and the presence of two rainfall events.
In the substrate, greater values of WC are related to lower tem-
perature values. In Plot 2 the temperature in the substrate, in
agreement with the pattern of external air, varied between 4.0 C
and 15.3 C. However, the difference in WC between the two dif-
ferent acquisition points does not produce a relevant difference in
the values of substrate temperature for the same points. In winter,
the main force that drives the evapotranspiration phenomena is
solar radiation, which is more limited in magnitude; furthermore,
the high external relative humidity values restrain the evaporation
potential, therefore the overall effect does not allow for appreci-
ation of substantial differences in the temperature trend. Similar
results are found in Plot 1. In the layers below the growing media,
the temperature showed a variable trend but, nevertheless, the
temperature at the bottom of the lightweight concrete were rela-
tively high if compared to the external air, as can be seen in Fig. 11b.
Similar results were found for the substrate in Plot 3, so that only
the pattern of the variables below this layer is shown. The presence
of the insulation layer is undoubtedly benecial since in this plot the
temperature at the bottom of the lightweight concrete maintained
the highest values of the plots, for the considered period presenting
values that were always higher than the external air temperature
(Fig. 12). In the reference roof, again the surface temperature was
strongly inuenced by the external air temperature and solar radi-
ation and shows a variable trend, reaching a maximum value of
72 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379

Fig. 10. Climatic condition of the selected winter period. Solar radiation, sky temperature and external air temperature (a). Rainfall, wind speed and relative humidity (b).

34.6 C. The minimum value, on a daily level, is considerably lower
and fell below 0 C, due to longwave heat exchange with the sky.
A comparison between the temperatures registered at the bottom
of the lightweight concrete and the average surface temperature
measured on the reference roof, is shown in Fig. 13, observing how
the green roof can maintain a higher temperature than the average
surface temperature of the reference roof. In particular, in Plot 3
the temperature at the base of the green layering was on average
4 C higher than the surface of the reference roof, whereas in Plot 1
was 2.5 C higher. Plot 2 showed the worst performance of all the
three considered layering as for part of the week the temperature
was almost coincident with the average surface temperature of the
reference roof.
Regarding the internal air conditions, Room 3 showed the high-
est time percentage of temperature for each temperature shown in
the x axis (Fig. 14); this behaviour is due to the presence of the insu-

Fig. 11. Temperature and water content in Plot 2 substrate (a) and temperatures of the underlying layer in Plot 2 (b) for the selected winter period.

Fig. 12. Temperature underlying substrate in Plot 3 in the selected winter period.
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 73

lation layer. Room 1 and Room 2 showed performances close to that
of the Reference Room. The worst results were achieved in Room 2,
which did not demonstrate a different behaviour compared to the
Reference Room, in agreement with the result of the winter ther-
mal analysis of the layering. However, if temperature values higher
than the winter comfort condition level (20 C) are considered, Plot
1 and Plot 2 were able to produce air temperature values on the
correspondent room that are higher, or at least similar, to that of
the Reference Room. Even though winter performances are limited
if compared to the results of the summer period, the green roof can
contribute to thermal insulation of the building envelope, reducing
heat losses toward the outdoor environment.
4. Heat uxes evaluation
In this section an assessment of the heat uxes entering and
exiting the indoor environment through the structural roof of the
building, for the same summer and winter periods analysed in the
previous sections, has been conducted. The heat uxes delivered
through the roof cover allow a quantication of the energy benets
related to the green roof installation. The temperature measured
at the indoor ceiling surface of each room, the temperature at
the bottom of the lightweight concrete layer of each plot and the
surface temperature in the reference plot were used as boundary
conditions in a nite difference scheme. Grid independence was
priory tested, running simulations with different numbers of nodes.
In particular, in a preliminary test phase, simulations varying the
number of nodes from 10 to 20 with a step of 2 were performed.
From the results it was possible to conclude that fteen is a suf-
cient number of nodes to carry out subsequent simulations.
With this procedure the capacitance effect of the structural roof
is properly taken into account. The solution of the nite differ-
ence scheme provides the temperatures values in each considered
node, as shown in Fig. 15. The conductive heat ux was evaluated
between the ceiling surface node and the previous one. Once the
hourly heat uxes are known, the integral in the considered period
provides the energy delivered to the indoor rooms, allowing for
the determination of the possible energy saving per square meter
of roof due to the presence of the green cover.
4.1. Summer evaluation
Fig. 16a reports the heat uxes for the considered summer
period. In Plot 1 and Plot 2 the resulting heat uxes are always
negative indicating that the rooms are cooled by the ceiling that
disperses heat in the upper layers. The minimum value reached
is 5.3 W/m2 for Plot 1 and 6.7 W/m2 for Plot 2. In Plot 3, as
conrmed by the thermal analysis, the additional insulation layer
worsens the performance of the green layering with resultant pos-
itive conductive heat ux in most part of the period, indicating that
energy is entering the indoor environment. In the reference roof

Fig. 13. External surface temperature in Reference Roof, air temperature and solar radiation in the selected winter period (a). Comparison between average external surface
temperature in Reference Roof and temperature below the lightweight concrete on the other plots (b).

Fig. 14. Inverse cumulative distribution of internal air temperature for January 2014.
74 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379

the pattern of the heat ux is highly variable, in agreement with
the surface temperature of the plot, with peaks of about 50 W/m2 ,
considerably greater than other plots. During night time, when the
surface temperature falls to the minimum daily value, an inver-
sion of the heat uxes occurs, but the entity is quite limited with a
minimum value that does not exceed 14 W/m2 .
The integration of the quantied heat uxes in the considered
period provides the specic energy, expressed in Wh per square
meter of roof surface, that enters or leaves the correspondent
indoor rooms. As summarized in Table 3, in the reference roof
the total entering energy is 3025 Wh/m2 , one order of magnitude
higher than Plot 3, where it results as being only 547 Wh/m2 with
a reduction of about 82%. In particular, Plot 2 is able to produce
lower conductive heat ux than Plot 1; however they both nullify
the ingoing heat ux showing only outgoing uxes.
4.2. Winter evaluation
The same considerations were conducted for the considered
winter period. Fig. 17 shows the results of the nite elements sim-
ulations. For the three plots and the reference roof, the calculated
heat uxes are always negative in the winter period and with an
extreme variable pattern. This is a consequence of the variable trend
of the temperatures in the green roofs layers, as discussed in the
previous paragraphs. Fig. 16b demonstrate how, in cold weather
conditions, the indoor rooms disperse heat through the roof, nev-

Fig. 15. Temperature distribution of the nodes in the four plots for the summer period. T5, T7, T9, T11 and T13 are the temperatures of the correspondent nodes; Tbot is the
temperature at the bottom of the lightweight concrete layer; Tceil is the internal ceiling temperature and Tsup is the surface temperature of Reference Roof.

Fig. 16. Conductive heat ux calculated between the last two nodes of the structural in summer (a) and winter (b).
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 75

Table. 3
Specic thermal energy entering and exiting the indoor rooms through the structural roof in summer and in winter.

Summer Winter
2 2
Eout [Wh m ] Ein [Wh m ] in [%] Eout [Wh m2 ] Ein [Wh m2 ] out [%]

Plot 1 1211 0 100 3447 0 30

Plot 2 1828 0 100 3406 0 31
Plot 3 16 547 82 3124 0 37
Ref. Roof 235 3025 4954 0

ertheless in the reference roof the heat ux is notably higher, in
absolute value, than the green plots, by indicating the advantageous
effect of the presence of a green cover on the roof. In particular in
Plot 1 the minimum value reached by the heat ux is 42 W/m2 ,
in Plot 2 it is 37 W/m2 , in Plot 3 it is 32 W/m2 and in Reference
Roof it is 49 W/m2 . With reference to the energy loss through the
roof, Table 3 shows how Plot 1 and Plot 2 can reduce the loss by
30% and 31% respectively, whereas Plot 3, due to the presence of an
insulation layer, can provide a higher reduction of approximately
37%.
5. Dynamic characterization of the extensive green roof
In order to assess dynamic thermal properties of building enve-
lope elements, as traditional roof, it is usual to dene proper
parameters. These are, for the steady periodic regime, the heat stor-
age capacity, expressed in terms of decrement factor and time lag.
In non-sinusoidal conditions, the decrement factor is dened as
the ratio between the maximum daily excursions of the internal
and external temperature uctuations [56]. The time lag is calcu-
lated with reference to maximum temperature peaks in summer as
dened in [56].
Ti,max Ti,min
f = (3)
Te,max Te,min
 = tTi,max tTe,max (4)
where the subscripts i and e refer to internal and external conditions
and min and max refer to the daily maximum and minimum values.
In this section, a theoretical and experimental evaluation in
order to quantify the values assumed by these parameters for
the vegetation-substrate-drainage system was conducted for the
investigated green roof plots.
In theoretical analysis, the dynamic parameters were evaluated
considering the sol-air temperature as the external boundary con-
dition, which is usually employed to schematize the joint action
of the external loads [57,58] whereas, the temperature at the bot-
tom of the drainage layer was considered as the internal boundary
condition in the three plots. The sol-air temperature is obtainable

Fig. 17. Temperature distribution of the nodes in the four plots for the winter period. T5, T7, T9, T11 and T13 are the temperatures of the correspondent nodes; Tbot is the
temperature at the bottom of the lightweight concrete layer; Tceil is the internal ceiling temperature and Tsup is the surface temperature of Reference Roof.
76 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
from the external surface energy balance and, for a traditional roof,
is given by [57,58]:
hc hr 1
Tas = T + T +
hc + hr amb hc + hr sky hc + hr
The physical phenomena involved in a green roof are more com-
plex than a traditional roof and the evapotranspiration contribution
plays an important role in the energy balance. The evapotran-
spiration phenomenon leads to benets due to cooling of the
external ambient air with a consequent reduction of the equiva-
lent external temperature. In such conditions, the decrement factor
decreases and the time lag increases. The theoretical analysis was
conducted, similar to a traditional roof, without taking into account
the contribution related to evapotranspiration, obtaining there-
fore precautionary values of the dynamic parameters. However,
the absorption coefcient and the convective and radiative heat
transfer coefcients of the external surface of the green roof were
appropriately modied for the considered surfaces.
In Eq. (5), the sky temperature is obtained assuming it as a black
body. The foliage and soil absorption coefcients for the three plots
are calculated according to Del Barrio relations [40], properly mod-
ied to consider the fractional vegetation coverage of the plots:
f = (1 s s )(1 + s s )f
   the two acquisition points ranges between 0.100 and 0.178 in Plot
s = (1 s ) s f + 1 g 1 f
where the fractional vegetation coverage (f ) and the Leaf Area
Index (LAI) have been estimated through direct measurements on
the experimental site, obtaining a value of f = 0.59 and LAI = 1.5
for plot 1 and f = 0.56 and LAI = 1.5 for plot 2. The longwave trans-
mittance s represents the short wave transmittance related to the
vegetation as a function of the LAI index and was assumed equal
to 0.41 according to the values provided in [40]. The albedo of the
vegetation s was assumed equal to 0.26, that is a value proposed
in literature for short glass plants [38]. The albedo of the soil g
was measured in laboratory, obtaining a value of 0.17. The mean
absorption coefcient for each plot is calculated as:
 
= f f + 1 f s
The foliage and soil convective heat transfer coefcients are pro-
vided by the relation [10]:
 
hc,f = 1.1LAI a cp,a Cf Waf
hc,s = a cp,a Cs Waf
where a (1.23 kg/m3 ) is air density, cp,a (1005.6 J/kg K) is air specic
heat capacity, Cf e Cg are the bulk heat transfer coefcients respec-
tively in the vegetation and in the soil and Waf the wind speed in the
canopy. This formulation is more appropriate for a green roof since
takes into account the wind speed, the foliage and soil roughness.
Considering the contemporaneous presence of vegetation and soil
on the green roof plots, the mean convective heat transfer coef-
cient is calculated as a weighted average, in function of the foliage
coverage, between soil and vegetation heat transfer coefcients, as
shown by the following relation:
 
hc = f hc,f + 1 f hc,s
The mean radiative heat transfer coefcient for each plot is cal-
culated as [59]:
 3
 Time lag and decrement factors for the green roof Plots according to the theoretical
hr = fg 4Ths
where  is the Stefan-Boltzmann constant, Ths is the average
temperature between the plot surface temperature and the sky
temperature and fg , according to [10], is dened as:
f g
fg =
1
where 1 is dened as [10]:
1 = g + f g f (14)
(5)
The obtained values of mean absorption coefcient, mean con-
vective heat transfer coefcient and mean radiative heat transfer
coefcient are: = 0.35, hc = 11.01 and hr = 5.09 for Plot 1; = 0.36,
hc = 11.35 and hr = 5.12 for plot 2; = s = 0.83, hc = hc,s = 9.42 and
hr = 5.69 for Plot 3.
Fig. 18 shows the trend of the calculated sol-air temperature
for the three plots. As expected, the sol-air temperature is affected
by solar radiation and sky temperature; from the graphs it can be
clearly observed how for Plot 1 and Plot 2 the curves are very similar
with daily peaks of about 44 C, whereas in Plot 3 it reaches much
higher values than the other plots with daily peaks of about 72 C.
According to the procedure described and considering the same
acquisition points used in the thermal behaviour analysis, for each
plot the decrement factor and the time lag were evaluated in the
investigated summer period on a daily basis. In particular, in the
theoretical analysis the sol-air temperature describes the external
environment and only the effect of the additional green layers is
considered. Fig. 19a and Fig. 19b shows the daily values of the decre-
ment factor and of the time lag for the acquisition points of the three
(6) plots. For the considered summer period, the decrement factor of
(7)
1 and between 0.131 and 0.185 in Plot 2, whereas in Plot 3, since
there is not the effect of the vegetation, this ranges between 0.154
and 0.226. Regarding the time lag, Plot 3 always has lower values
than the other plots ranging from 2 h to 4 h, whereas Plot 1 and Plot
2 show almost identical daily values for the analysed seven days in
the range 4 h to 6 h.
The average values obtained in the investigated period, accord-
ing to the theoretical analysis, are listed in Table 4.
The relatively high time lag values observed conrm the positive
effect of the green stratigraphy in summer conditions. Plot 1 and
Plot 2 present very close time lag values, which are slightly lower
than ve hours since they consist of similar layer materials. In Plot 3,
the time lag value is even more reduced highlighting the important
(8) role played by vegetation plays in the dynamic characterization of
a green roof. For the same reason, the decrement factor in Plot 3 is
higher than the other two plots. However, the ability of the system
to suppress the temperature oscillations is conrmed by the results
(9) of this analysis.
(10) The results of the experimental evaluations were obtained
considering the temperature in the substrate as upper boundary
condition and that below the drainage layer, as the lower boundary
condition. Since the effect of the vegetation layer and of a part of the
substrate is excluded, these evaluations are more conservative. In
Fig. 19c and Fig. 19d the daily values of decrement factor and of time
lag are shown for the acquisition points of the three plots. For the
considered summer period, the decrement factor varies between
0.234 and 0.325 in Plot 1, where the differences between point A
and B are more evident, but are, however, contained. In Plot 2, in
all the investigated period, the values of the decrement factor for
the acquisition points are fairly similar, varying between 0.295 and
0.336. In plot 3, substantially higher values from 0.403 to 0.524
(11)
Table 4
(12) and experimental analysis.
(13)
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379 77

were observed, owing to the absence of vegetation. The time lag
was also variable during the days and ranges between 2 h and 6 h in
Plot 1, between 3 h and 5 h in Plot 2 and between 1 h and 3 h in Plot
3. The average values obtained in the investigated period, according
to the experimental analysis, are reported in Table 4. As expected,
time lags are smaller and decrement factors greater since in this
case the analysis excludes the effect of the vegetation layer and of
a part of the substrate, with a consequent reduction of the temper-
ature uctuation that represents the external boundary condition
of the system. The results of Table 4 conrm the importance of
the vegetation in summer, especially considering the decrement
factors, which in the last case, are almost double compared to the
results of the theoretical investigation.
A nal assessment of the dynamic properties results from a com-
parison with time lag and decrement factor values in presence and
absence of the green roof. For the reference roof, decrement factor
and time lag were evaluated considering the experimental exter-
nal and internal surface temperature. The mean values obtained in
the investigated summer period are 0.157 for the decrement factor
and 6.571 h for the time lag. Multiplying such decrement factor to

Fig. 18. Sol-air temperature for the three Plots in the considered period.

Fig. 19. Theoretical daily values of decrement factor (a) and time lag (b) in the summer period for the three plots. Experimental daily values of decrement factor (c) and time
lag (d) in the summer period for the three plots.
78 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
the experimental ones reported in Table 4 for the additional green
layers, we can obtain the total decrement factor of the whole roofs.
The total time lag is obtained instead by adding the experimental
time lags in Table 4 to that of the reference roof. The total experi-
mental decrement factor is equal to 0.045, 0.050 and 0.071 for Plot
1, Plot 2 and Plot 3 respectively while the total experimental time
lag is equal to 10.79 h, 10.57 h and 8.29 h, respectively.
Such a procedure allows to predict, known the dynamic proper-
ties of any traditional roof, as they are modied after the installation
of the green roof analysed in the present study. For instance, con-
sidering the dynamic properties of two type of roofs (R1 and R2)
studied by Kaska et al. [60], with a decrement factor of 0.035 and
0.48 and a time lag of 5.12 h and 2 h, the additional green layering
allows to reach values of 0.005 (R1) and 0.075 (R2) for the decre-
ment factor and of 11.69 h (R1) and 8.57 h (R2) for the time lag.
These results highlight the substantial improvement of dynamic
properties, especially in the case of traditional roofs with limited
dynamic performance (such as roof R2), that can be achieved with
the installation of extensive green roofs.
6. Conclusions
An experimental analysis of an extensive green roof consisting of
three different 50 m2 surface plots installed on a University build-
ing in South Italy, was presented. The study was conducted both
in summer and winter seasons in a typical Mediterranean climate,
analysing the thermal performances of different layering solutions
compared to a reference traditional roof of same area. The reference
roof is a black bituminous cover that represent the original cong-
uration of the roof surface. Results obtained allowed a comparison
with a low albedo roof that, even if does not represent the totality
of roofs in the area, is of great interest from a thermal point of view.
The results showed that in summer the green roof was able to main-
tain a noticeable lower temperature at the base of the additional
green stratigraphy. The plots without thermal insulation were able
to produce a temperature approximately 12 C lower than the aver-
age surface temperature of the reference roof at the interface with
the structural roof. The presence of an insulation layer determined
a worsening of the performances, with a temperature at the same
position approximately 4 C lower than the reference roof. In the
selected winter period, the green layering with thermal insulation
demonstrated that it could guarantee a temperature 4 C higher
than the average of the surface of the reference roof at the base of
the additional layers.
The presence of the green roof determined also a reduction of the
internal environment temperature on average of 2.3 C in the inves-
tigated summer period, demonstrating the quality of this solution
to decrease the cooling demand for the building air-conditioning.
In winter, if temperature values higher than the winter comfort
condition level (20 C) are considered, the green roof can produce
indoor room air temperature that is higher or at least similar to that
of the reference room.
The monitored quantities permitted to calculate the heat uxes
through the structural roof of the building in the analysed peri-
ods. The green roof plots generated negative heat uxes during
the investigated summer period, producing a passive cooling of the
building room through the structural roof, and provided a reduction
of the ingoing thermal energy of 100% in summer when compared
to the ingoing energy into the reference room. In winter, an atten-
uation between 30% and 37% of the outgoing heat ux has been
obtained; therefore in both analysed seasons it is possible to obtain
considerable energy savings for the air-conditioning of the involved
building. The experimental data were nally employed to conduct
a dynamic characterization of the different green roofs solutions.
The results of the dynamic characterization indicated that, under
the considered climatic conditions, the different green roofs solu-
tions, with a minimum average decrement factor value of 0.135
and a maximum average time lag of 4.8 h, can generate a substan-
tial improvement of dynamic properties of the envelope, especially
in the case of roofs with limited dynamic performance.
Acknowledgements
This research was supported by the Italian National Operational
Programme (PON) Research and Competitiveness for conver-
gency regions 2007/2013 Axis I Support to structural change
operational objective 4.1.1.1. Scientic-technological areas gener-
ating processes of transformation of the production system and the
creating new sectors Action II: Interventions supporting indus-
trial research through the project PON01 02543. Integrated and
sustainable management service for the water-energy cycle in urban
drainage systems.
References
[1] O. Saadatian, K. Sopian, E. Salleh, C.H. Lim, S. Riffat, E. Saadatian, A. Toudeshki,
M.Y. Sulaiman, A review of energy aspects of green roofs, Renew. Sustain.
Energy Rev. 23 (2013) 155168.
[2] T. Ayata, P.C. Tabares-Velasco, J. Srebric, An investigation of sensible heat
uxes at a green roof in a laboratory setup, Build. Environ. 46 (2011)
18511861.
[3] J. Mentens, D. Raes, M. Hermy, Green roofs as a tool for solving the rainwater
runoff problem in the urbanized 21st century? Landscape Urban Plann. 77
(2006) 217226.
[4] J.Y. Lee, H.J. Moon, T.I. Kim, H.W. Kim, M.Y. Han, Quantitative analysis on the
urban ood mitigation effect by the extensive green roof system, Environ.
Pollut. 181 (2013) 257261.
[5] M. Carbone, G. Garofalo, G. Nigro, P. Piro, Green roofs in the mediterranean
area: interaction between native plant species and sub-surface runoff, Appl.
Mech. Mater. 737 (2015) 749753.
[6] T. Susca, S.R. Gafn, G.R. Dellosso, Positive effects of vegetation: urban heat
island and green roofs, Environ. Pollut. 159 (2011) 21192126.
[7] D.E. Bowler, L. Buyung-Ali, T.M. Knight, A.S. Pullin, Urban greening to cool
towns and cities: a systematic review of the empirical evidence, Landscape
Urban Plann. 97 (2010) 147155.
[8] A. Teemusk, . Mander, Greenroof potential to reduce temperature
uctuations of a roof membrane: a case study from Estonia, Build. Environ. 44
(2009) 643650, http://dx.doi.org/10.1016/j.buildenv.2008.05.011.
[9] K. Liu, J. Minor, Performance Evaluation of an Extensive Green Roof, Greening
Rooftops for Sustainable Communities, Washington, D.C, 2005, pp. 111.
[10] D, Sailor A green roof model for building energy simulation programs, Energy
Build. 40 (2008) 14661478, http://dx.doi.org/10.1016/j.enbuild.2008.02.001.
[11] T.G. Theodosiou, Summer period analysis of the performance of a planted roof
as a passive cooling technique, Energy Build. 35 (2003) 909917.
[12] M. Zinzi, S. Agnoli, Cool and green roofs. An energy and comfort comparison
between passive cooling and mitigation urban heat island techniques for
residential buildings in the Mediterranean region, Energy Build. 55 (2012)
6676.
[13] I. Jaffal, S.-E. Ouldboukhitine, R. Belarbi, A comprehensive study of the impact
of green roofs on building energy performance, Renew. Energ. 43 (2012)
157164.
[14] S. Parizotto, R. Lamberts, Investigation of green roof thermal performance in
temperate climate: a case study of an experimental building in Florianpolis
city, Southern Brazil, Energy Build. 43 (2011) 17121722.
[15] J. Coma, G. Prez, C. Sol, A. Castell, L.F. Cabeza, Thermal assessment of
extensive green roofs as passive tool for energy savings in buildings, Renew.
Energ. 85 (2016) 11061115.
[16] O. Saadatian, K. Sopian, E. Salleh, C.H. Lim, S. Riffat, E. Saadatian, A. Toudeshki,
M.Y. Sulaiman, A review of energy aspects of green roofs, Renew. Sustain.
Energy Rev. 23 (2013) 155168.
[17] B.A. Currie, B. Bass, Estimates of air pollution mitigation with green plants and
green roofs using the UFORE model, Urban Ecosyst. 11 (2008) 409422.
[18] D.B. Rowe, Green roofs as a means of pollution abatement, Environ. Pollut.
159 (2011) 21002110.
[19] J.-J. Baik, K.-H. Kwak, S.-B. Park, Y.-H. Ryu, Effects of building roof greening on
air quality in street canyons, Atmos. Environ. 61 (2012) 4855.
[20] K. Vijayaraghavan, U.M. Joshi, R. Balasubramanian, A eld study to evaluate
runoff quality from green roofs, Water Res. 46 (2012) 13371345.
[21] B. Dvorak, A. Volder, Green roof vegetation for North American ecoregions: a
literature review, Landscape Urban Plann. 36 (2010) 351360.
[22] S. Schrader, M. Boning, Soil formation on green roofs and its contribution to
urban biodiversity with emphasis on Collembolans, Pedobiologia 50 (2006)
347356.
 P. Bevilacqua et al. / Energy and Buildings 122 (2016) 6379
[23] N. Williams, J. Rayner, K. Raynor, Green roofs for a wide brown land:
opportunities and barriers for rooftop greening in Australia, Urban For. Urban
Green. 9 (2010) 245251.
[24] M. Ignatieva, G.H. Stewart, C. Meurk, Planning and design of ecological
networks in urban areas, Landsc. Ecol. Eng. 7 (2010) 1725, http://dx.doi.org/
10.1007/s11355-010-0143-y.
[25] T. Van Renterghem, D. Botteldooren, In-situ measurements of sound
propagating over extensive green roofs, Build. Environ. 46 (2011) 729738,
http://dx.doi.org/10.1016/j.buildenv.2010.10.006.
[26] T. Van Renterghem, M. Hornikx, J. Forssen, D. Botteldooren, The potential of
building envelope greening to achieve quietness, Build. Environ. 61 (2013)
3444, http://dx.doi.org/10.1016/j.buildenv.2012.12.001.
[27] U. Berardi, A. GhaffarianHoseini, A. GhaffarianHoseini, State-of-the-art
analysis of the environmental benets of green roofs, Appl. Energ. 115 (2014)
411428, http://dx.doi.org/10.1016/j.apenergy.2013.10.047.
[28] T. Theodosiou, Green roofs in buildings: thermal and environmental
behaviour, Adv. Build. Energy Res. 31 (2009) 271288.
[29] R. Fioretti, A. Palla, L.G. Lanza, P. Principi, Green roof energy and water related
performance in the Mediterranean climate, Build. Environ. 45 (2010)
18901904, http://dx.doi.org/10.1016/j.buildenv.2010.03.001.
[30] P. Bevilacqua, J. Coma, G. Prez, C. Chocarro, A. Jurez, C. Sol, M. De Simone,
L.F. Cabeza, Plant cover and oristic composition effect on thermal behaviour
of extensive green roofs, Build. Environ. 92 (2015) 305316, http://dx.doi.org/
10.1016/j.buildenv.2015.04.026.
[31] K. Liu, Engineering performance of rooftop gardens through eld, in: RCI 18th
International Convention and Trade Show, Tampa, Florida, 2003, pp. 115.
[32] C.J. Molineux, C.H. Fentiman, A.C. Gange, Characterising alternative recycled
waste materials for use as green roof growing media in the U.K, Ecol. Eng. 35
(2009) 15071513, http://dx.doi.org/10.1016/j.ecoleng.2009.06.010.
[33] L. Rincn, J. Coma, G. Prez, A. Castell, D. Boer, L.F. Cabeza, Cabeza
Environmental performance of recycled rubber as drainage layer in extensive
green roofs. A comparative Life Cycle Assessment, Build. Environ. 74 (2014)
2230, http://dx.doi.org/10.1016/j.buildenv.2014.01.001.
[34] S.B. Mickovski, K. Buss, B.M. McKenzie, B. Skmener, Laboratory study on the
potential use of recycled inert construction waste material in the substrate
mix for extensive green roofs, Ecol. Eng. 61 (2013) 706714, http://dx.doi.org/
10.1016/j.ecoleng.2013.02.015.
[35] R. Martens, B. Bass, S. Saiz Alcazar, Roofenvelope ratio impact on green roof
energy performance, Urban Ecosyst. 11 (2008) 399408, http://dx.doi.org/10.
1007/s11252-008-0053-z.
[36] D.J. Sailor, M. Hagos, An updated and expanded set of thermal property data
for green roof growing media, Energy Build. 43 (2011) 22982303, http://dx.
doi.org/10.1016/j.enbuild.2011.05.014.
[37] R.M. Lazzarin, F. Castellotti, F. Busato, Experimental measurements and
numerical modelling of a green roof, Energy Build. 37 (2005) 12601267,
http://dx.doi.org/10.1016/j.enbuild.2005.02.001.
[38] P.C. Tabares-Velasco, J. Srebric, A heat transfer model for assessment of plant
based roong systems in summer conditions, Build. Environ. 49 (2012)
310323, http://dx.doi.org/10.1016/j.buildenv.2011.07.019.
[39] C. Feng, Q. Meng, Y. Zhang, Theoretical and experimental analysis of the
energy balance of extensive green roofs, Energy Build. 42 (2010) 959965,
http://dx.doi.org/10.1016/j.enbuild.2009.12.014.
[40] E. Palomo, D. Barrio, Barrio Analysis of the green roofs cooling potential in
buildings, Energy Build. 27 (1998) 179193, http://dx.doi.org/10.1016/S0378-
7788(97)00029-7.
[41] S.E. Ouldboukhitine, R. Belarbi, I. Jaffal, A. Trabelsi, Assessment of green roof
thermal behavior: a coupled heat and mass transfer model, Build. Environ. 46
(2011) 26242631, http://dx.doi.org/10.1016/j.buildenv.2011.06.021.
[42] R. Djedjig, S.E. Ouldboukhitine, R. Belarbi, E. Bozonnet, Development and
validation of a coupled heat and mass transfer model for green roofs, Int.
Commun. Heat Mass Transfer 39 (2012) 752761, http://dx.doi.org/10.1016/j.
icheatmasstransfer.2012.03.024.
[43] A. Niachou, K. Papakonstantinou, M. Santamouris, A. Tsangrassoulis, G.
Mihalakakou, Analysis of the green roof thermal properties and investigation
of its energy performance, Energy Build. 33 (2001) 719729, http://dx.doi.
org/10.1016/S0378-7788(01)00062-7.
79
[44] A. Spala, H.S. Bagiorgas, M.N. Assimakopoulos, J. Kalavrouziotis, On the green
roof system. Selection, state of the art and energy potential investigation of a
system installed in an ofce building in Athens, Greece, Renew. Energ. 33
(2008) 173177, http://dx.doi.org/10.1016/j.renene.2007.03.022.
[45] F. Ascione, N. Bianco, F. de Rossi, G. Turni, G.P. Vanoli, Green roofs in
European climates. Are effective solutions for the energy savings in
air-conditioning? Appl. Energ. 104 (2013) 845859, http://dx.doi.org/10.
1016/j.apenergy.2012.11.068.
[46] K.L. Getter, D.B. Rowe, J.A. Andresen, I.S. Wichman, Seasonal heat ux
properties of an extensive green roof in a Midwestern U. S. climate, Energy
Build. 43 (2011) 35483557, http://dx.doi.org/10.1016/j.enbuild.2011.09.018.
[47] X. Qin, X. Wu, Y. Chiew, Y. Li, A green roof test bed for stormwater
management and reduction of urban heat island effect in Singapore, Br. J.
Environ. Clim. Change 2 (2012) 410420.
[48] J. Coma, G. Prez, C. Sol, A. Castell, L.F. Cabeza, Thermal assessment of
extensive green roofs as passive tool for energy savings in buildings, Renew.
Energ. 85 (2016) 11061115, http://dx.doi.org/10.1016/j.renene.2015.07.074.
[49] M. DOrazio, C. Di Perna, E. Di Giuseppe, Di Giuseppe Green roof yearly
performance: a case study in a highly insulated building under temperate
climate, Energy Build. 55 (2012) 439451, http://dx.doi.org/10.1016/j.
enbuild.2012.09.009.
[50] T. Theodosiou, D. Aravantinos, K. Tsikaloudaki, Thermal behaviour of a green
vs. a conventional roof under Mediterranean climate conditions, Int. J.
Sustain. Energy 33 (2014) 227241, http://dx.doi.org/10.1080/14786451.
2013.772616.
[51] O. Schweitzer, E. Erell, Evaluation of the energy performance and irrigation
requirements of extensive green roofs in a water-scarce Mediterranean
climate, Energy Build. 68 (2014) 2532, http://dx.doi.org/10.1016/j.enbuild.
2013.09.012.
[52] F. Olivieri, C. Di Perna, M. DOrazio, L. Olivieri, J. Neila, Experimental
measurements and numerical model for the summer performance assessment
of extensive green roofs in a Mediterranean coastal climate, Energy Build. 63
(2013) 114, http://dx.doi.org/10.1016/j.enbuild.2013.03.054.
[53] C.Y. Jim, L.L.H. Peng, Substrate moisture effect on water balance and thermal
regime of a tropical extensive green roof, Ecol. Eng. 47 (2012) 923, http://dx.
doi.org/10.1016/j.ecoleng.2012.06.020.
[54] D. Morau, T. Libelle, F. Garde, Performance evaluation of green roof for
thermal protection of buildings in Reunion island, Energy Procedia 14 (2012)
10081016, http://dx.doi.org/10.1016/j.egypro.2011.12.1047.
[55] M.T. Simmons, B. Gardiner, S. Windhager, J. Tinsley, Green roofs are not
created equal: the hydrologic and thermal performance of six different
extensive green roofs and reective and non-reective roofs in a sub-tropical
climate, Urban Ecosyst. 11 (2008) 339348, http://dx.doi.org/10.1007/
s11252-008-0069-4.
[56] K.J. Kontoleon, D.K. Bikas, The effect of south walls outdoor absorption
coefcient on time lag, decrement me viewed it as a cause of more energy
consumpt http://dx.doi.org/10.1016/j.enbuild.2006.11.006.
[57] G. Oliveti, N. Arcuri, D. Mazzeo, M. De Simone, A new parameter for the
dynamic analysis of building walls using the harmonic method, Int. J. Therm.
Sci. 88 (2015) 96109, http://dx.doi.org/10.1016/j.ijthermalsci.2014.09.006.
[58] D. Mazzeo, G. Oliveti, N. Arcuri, Inuence of internal and external boundary
conditions on the decrement factor and time lag heat ux of building walls in
steady periodic regime, Appl. Energ. 164 (2016) 509531, http://dx.doi.org/
10.1016/j.apenergy.2015.11.076.
[59] G. Oliveti, N. Arcuri, S. Ruffolo, Experimental investigation on thermal
radiation exchange of horizontal outdoor surfaces, Build. Environ. 8 (2003)
8389, http://dx.doi.org/10.1016/S0360-1323(02)00010-0.
[60] O. Kaska, R. Yumrutas, O. Arpa, Theoretical and experimental investigation of
total equivalent temperature difference (TETD) values for building walls and
at roofs in Turkey, Appl. Energ. 86 (2009) 737747, http://dx.doi.org/10.
1016/j.apenergy.2008.09.010.