Accepted Manuscript

Title: AN EXPERIMENTAL METHOD TO

QUANTITATIVELY ANALYSE THE EFFECT OF
THERMAL INSULATION THICKNESS ON THE
SUMMER PERFORMANCE OF A VERTICAL GREEN
WALL

Authors: F. Olivieri, R. Cocci Grifoni, D. Redondas, J.A.

Sánchez-Reséndiz, S. Tascini

PII: S0378-7788(16)31541-9
DOI: http://dx.doi.org/doi:10.1016/j.enbuild.2017.05.068
Reference: ENB 7649

To appear in: ENB

Received date: 14-11-2016

Revised date: 2-5-2017
Accepted date: 25-5-2017

Please cite this article as: F.Olivieri, R.Cocci Grifoni, D.Redondas, J.A.Sánchez-
Reséndiz, S.Tascini, AN EXPERIMENTAL METHOD TO QUANTITATIVELY
ANALYSE THE EFFECT OF THERMAL INSULATION THICKNESS ON THE
SUMMER PERFORMANCE OF A VERTICAL GREEN WALL, Energy and
Buildingshttp://dx.doi.org/10.1016/j.enbuild.2017.05.068

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
AN EXPERIMENTAL METHOD TO QUANTITATIVELY ANALYSE THE EFFECT OF THERMAL

INSULATION THICKNESS ON THE SUMMER PERFORMANCE OF A VERTICAL GREEN WALL

F. Olivieria, *, R. Cocci Grifoni b, D. Redondasc, J. A. Sánchez-Reséndizd, S. Tascinib

a
Universidad Politécnica de Madrid, E.T.S. Arquitectura, Department of Construction and Technology in Architecture,

Avda. Juan de Herrera 4, 28040 Madrid, Spain. Email: francesca.olivieri@upm.es

b
School of Architecture and Design, University of Camerino, Via Della Rimembranza, 63100 Ascoli Piceno, Italy. Email:

roberta.coccigrifoni@unicam.it; simone.tascini@unicam.it
c
Universidad Politécnica de Madrid, E.T.S. Edificación, Department of Applied Mathematics, Avda. Juan de Herrera 6,

28040 Madrid, Spain. Email: dolores.redondas@upm.es

d
Universidad Politécnica de Madrid, Innovation and Technology for Development Centre, Avda. Complutense, 28040,

Madrid, Spain. Email: sarjarq@gmail.com

*Corresponding Author: E-mail: francesca. olivieri@upm. es. Fax: +34913366560. Phone: +34913364239

Highlights

• A full-scale green wall was monitored during 2 summers in a Mediterranean climate

• Different insulation thicknesses were tested during monitoring

• A new methodology to simulate the thermal performance of green walls was developed

• The methodology was experimentally validated

• The results show that the cut-off insulation thickness is 9 cm

Abstract

Green façades and walls greatly contribute to reducing solar gains and dispersion through the building envelope. This

implies a lower energy load for both heating and cooling and the mitigation of thermal conditions in outdoor areas. Despite

this, more studies are needed regarding the influence of these systems on the thermal behaviour of insulated façades. In

this manuscript, we report the results of experimental research carried out on a vertical green wall in a continental

Mediterranean climate. The main goal of the research is to establish a thickness above which the behaviour of the green

façade becomes isothermal and its performance do not improve. To this end, we analyze and evaluate the effect of

insulation thickness on the energy performance of a green wall using a new methodology called green façade optimization
(GFO). Comparing the simulations to experimental data, collected in a full-scale experimental box during the summers

of 2011 and 2012, allowed the model to be validated. The results show that a green wall acts as a passive cooling system

when the façade is moderately insulated, up to an insulation thickness of 9 cm, above which more insulation becomes

redundant and inefficient.

Key words: Green façades, insulation thickness, parametric optimization, experimental measurements, thermal

performance.

1. Introduction

International and European regulations for energy efficiency are becoming increasingly rigid, and this often

translates into greater thermal insulation thicknesses for both roofs and walls.

On the other hand, research has demonstrated that green walls can be an effective strategy to cool interiors in the

summer and insulate in the winter [1, 2, 3, 4-13]. In fact, the presence of vegetation greatly contributes to the reduction

of solar gains and dispersion through the envelope [14, 15, 16, 17-20]. This translates into a lower energy load for both

heating and cooling and to mitigate thermal conditions in outdoor areas [21, 22, 23]. During the hottest hours of the day,

bare walls accumulate and release more heat to the environment than green walls. On the contrary, during the cool night-

time hours and in conjunction with the period of natural ventilation, green walls absorb more heat from the internal

environment, favouring cooling and contributing to lowering the operational temperature. This is particularly important

in the hottest climates [1, 14, 24, 25].

A lot of research has demonstrated that the use of green walls also reduces the urban heat island effect (UHI) [26-29],

improves outdoor thermal comfort [30], and has a positive effect on the filtration of air-borne pollutants [31-32]. In

addition, all of these aspects have been considered within the context of wider urban ecosystem services provided by

vegetation. In this respect, it is increasingly likely that the design of new buildings in the future will focus on the

simultaneous presence of classic thermal insulation solutions and green walls.

With regard to reducing a building’s energy demand, the majority of research has focused on hot climates or summer

conditions. Haggag and al. [14] experimented with a green wall installed in a school building located in the city of Al-

Ain, in a hot, arid climate (United Arab Emirates). The green wall was monitored and compared with a bare wall during

the summer, which is characterized by daytime temperatures range from 35 °C to 50 °C. The results of the study showed

that the vegetation layer partially blocked solar radiation, thereby regulating the temperature; the reduced external surface

temperature of the green wall yielded a 6°C decrease in temperature on the internal surface. As a consequence, the green

wall always maintained a lower indoor ambient temperature than the bare wall, a difference that ranged from 2°C during
the night time to almost 6°C during the peak day time exterior temperature. Despite this, indoor temperatures of about

45°C are still far from a comfortable temperature, which means that although the cooling load was reduced, there was still

a need for mechanical cooling.

In a study by Manso et al. [24], an innovative green wall system was presented. The aim of the study was to test the new

modular system in a Mediterranean climate through comparison with a reference wall. The experimental procedure was

performed in Covilhã, in the inland region of Beira, in Portugal. The research was based on comparative analysis of the

interior surface temperature and heat fluxes between the green wall and the bare wall carried out between September and

December. The results showed that with the green wall, the exterior surface temperature was reduced up to 15°C and

thermal gains due to conduction were reduced by 75%. In this case, further analysis may include the thermal behaviour

during the cool season in order to quantify the potential of green walls as passive cooling systems. In another study,

Hoelscher et al. [24] quantified the cooling effects of façade greening by analysing shading, transpiration, and insulation.

The research was conducted in Berlin and analyzed, among other aspects, the decrease in temperature on the interior and

exterior surfaces of the greened walls. The experiments were conducted during hot summer periods on three building

façades covered with three different climbing plants: Parthenocissus tricuspidata (site A), Hedera helix (Site B), and

Fallopia baldschuanica (Site C).

The highest temperature reduction on the exterior surface was reached for site A: the bare wall was heated up to 51. 5°C

on a hot summer day while the greened wall temperature only reached 38. 3°C on the wall surface behind the vegetation.

The interior surface temperature at Site A decreased by 0. 1–11. 3°C compared to the bare wall for a hot summer day.

Similar results were found for greening with H. helix (Site B) (up to 12. 3°C), while the cooling was weaker with F.

baldschuanica (Site C) (up to 6. 6°C).

The results presented in this research are very interesting and demonstrate the great efficiency of vegetal systems in hot

climates and during the summer. Despite this, nearly all rely on comparing the thermal behaviour of two walls—one with

and one without vegetation—under the same weather conditions. The results are valid for the specific situation but the

analysis cannot be extrapolated to different climate conditions or green wall systems.

This is important because in addition to summer cooling, green walls can also reduce heat losses during colder seasons.

With regard to this, Cameron et al. [5] investigated the effects of a greened wall during the cold season in Reading, United

Kingdom, which is characterized by a maritime-temperate climate. The aim was to investigate the insulation properties

of plants, and experiments were conducted during two winters using small-scale physical models to simulate heat loss

from a building. The physical model was made of a brick construction around a water tank maintained at 16 °C; energy

use was monitored. During the first winter, 12 brick cuboids were arranged in a matrix design; 8 additional bricks were

included in the second winter. Half of the cuboids were planted with Hedera helix, two plants per side, i. e. , eight plants
per cuboid. During the first winter, Hedera foliage covered approximately 80% of the planted cuboids to a depth of 30–

60 mm. By the following winter, the foliage had completely covered the roof and walls to a depth of 60–80 mm. There

were clear differences between unplanted cuboids and planted cuboids during periods of rainfall, i. e. , the walls behind

the foliage were often dry to the touch, while there was evident surface moisture on the bare walls. The advantage of the

vegetated cuboids was again clear during episodes of windy weather, where the greened cuboids were 43% more efficient

than the bare walls at reducing convective heat loss, particularly by reducing the wind chill. Furthermore, the research

demonstrated that foliage-covered brick cuboids maintained higher temperatures than the surrounding air, particularly in

the evening, with an associated potential to reduce peak energy demand. As is known, the trapping of warmed air is a

principal function of commercial insulation products and vegetated walls present similar characteristics. In general,

covering cuboids with Hedera helix reduced mean energy consumption by 21% compared to bare cuboids during the first

winter. During the second winter, a 37% mean savings was achieved.

In other research, Köhler [4] analysed the thermal insulation effect of a vegetated wall by using infrared measurements.

In this case, temperatures were measured on the exterior surface on the wall. The results showed that the temperature

buffer created by Hedera helix was about 3°C on early winter mornings.

The reduction of convective heat loss due to a vegetation layer was also studied by Perini et al. [3]. In particular, this

research focused on the influence of the vegetation on wind speed and its effects on the thermal resistance of the façade.

The experiment was conducted in three different Dutch cities by comparing three vertical plant-covered walls with the

bare walls next to them. The reduction in wind speed measured 0. 1 m in front of the walls varied from 0. 43 m/s to 0. 55

m/s, depending on the system. As a consequence of this reduction, the wind speed measured 0.1 m in front of the green

walls was in each case less than 0. 2 m /s, which implies that the external surface resistance (Re = 0. 04 m2K/W) could

be equal to the internal surface resistance (Ri = 0. 13 m2K/W), raising the total thermal resistance of the façade by 0. 09

m2K/W.

This research, which focused on winter conditions, showed that green walls can contribute to the thermal insulation of

the walls, since they act as an additional protective layer. On the other hand, it must be emphasized that the systems

analyzed were mainly characterized by climbing plants, and it should thus be determined whether the insulating effect is

maintained when using systems consisting of plants and a substrate.

Despite the drawbacks to these studies, as stated at the beginning of this section, it is likely that the in future, new buildings

and renovated green walls will coexist with conventional thermal insulation. Therefore, it is also necessary to consider

insulation in green walls for the optimum performance of the two systems throughout the year. Thermal insulation serves

to maintain well-being and protect internal environments from daily and seasonal variations . It represents one of

the most costly measures in renovation but it has a large influence on thermal energy savings .
It is known that thermal insulation conditions the effect of green walls [1, 16, 33, 34], but there are few studies that

quantify the thickness of insulation necessary to guarantee the required performance and to optimize its use . This

study therefore analyses the change in thermal behaviour of a green wall during the summer when different

thicknesses of thermal insulation are applied. The goal is to establish a thickness above which the behaviour of the

green façade become isothermal and its performance do not improve.

To do so, an experimental building façade with three different insulation thicknesses was used. The wall was

monitored in order to develop a numerical model that, once validated with experimental data, allowed the effect of

different insulation thicknesses to be studied. Using algorithms already developed to study green roofs (i. e. , Green

Roof Model/EcoRoof), the numerical model allowed the value of the variables that characterize the

thermohygrometric and evapotranspiration properties of the wall to be correctly identified . In particular, an

optimization analysis was used to determine some important physical variables in order to define the thermal

behaviour of the green wall. The optimization was handled entirely within the virtual environment provided by

Rhino, Grasshopper [35], and the evolutionary algorithm Galapagos. The simulation was performed using

EnergyPlus [36] along with the Ladybug [37] and Honeybee plug-ins for Grasshopper [38, 39]. This in turn allowed

the correct simulations to be obtained. The numerical model was verified with the set of experimental data and

prognostic tests.

2. Materials and methods

2.1 Local climate conditions in summer

The monitoring periods considered in this study are the summers of 2011 and 2012, and specifically the month of July in

both years. Monitoring was carried out in a full-scale experimental building (Fig. 1) in Colmenar Viejo (883 m above sea

level, 40°39’N, 3°45’W), a town located approximately 40 km north of Madrid in a mountainous area characterized by a

continental Mediterranean climate with hot, dry summers.

Local weather conditions during the monitoring periods were recorded by a weather station (Fig. 2) installed 100 m from

the experimental building. Global irradiance, air temperature, relative humidity, wind, and rainfall were each registered

every 5 minutes. The accuracy of the thermometers and hygrometers is ±0. 2K and ±2%, respectively [40, 41].

As shown in Table 1, important differences can be noted in the temperature and relative humidity registered in July in

2011 and 2012.

In July 2011, the horizontal global irradiance exceeded 800 W/m2 in 33% of daylight hours and 1000 W/m2 in 8% of

daylight hours. In July 2012, the horizontal global irradiance exceeded 800 W/m2 in 32% of daylight hours and 1000

W/m2, in 5% of daylight hours. This area therefore receives high solar radiation.
In July 2011, the outdoor temperature was over 26°C only in 34% of the hours and above 30°C in 13% of the hours, while

in July 2012 it was over 26°C in 68% of the hours and above 30°C in 31% of the hours. In both years, the temperature

fell below 15°C very rarely.

In July 2011, the relative humidity was below 20% only in 5% of the hours and over 50% in 22% of the hours. On the

other hand, in July 2012, the relative humidity was below 20% in 38% of the hours and over 50% in 15% of the hours.

The data show a climate that is fundamentally characterized by hot summers with low relative humidity. Within this

general description, it is clear that the summer of 2012 was hotter than the summer of 2011.

2.2 Experimental design

The full-scale experimental box formed part of an office building of the Intemper Company, located in Colmenar Viejo.

The three-storey office building has a rectangular floor plan; the first two storeys are the same size (13. 8 m x 40 m) and

are larger than the third (13. 8 m x 28. 9 m), which has a terrace facing south where the experimental façade prototype

was installed (Fig. 1). An unoccupied space to the south ensured a sunny experimental façade throughout the monitoring

periods. The prototype was composed of four spaces that were completely isolated from each other in order to create

virtually adiabatic spaces.

This work focuses on only one of the four spaces, which measured 1. 8 m x 1. 8 m x 2. 4 m (Fig. 3). The floor, ceiling,

and walls were composed of 0. 60 m of extruded polystyrene (λ= 0. 035 W/(mK)), yielding a total thermal resistance of

17. 8 m2K/W in the walls and 17. 2 m2K/W in the floor and ceiling. Only the wall facing south, where the green façade

was installed, had a different composition. This was composed of modular panels including the following parts: metallic

box (0. 6 m x 0. 6 m x 0. 08 m) containing a substrate wrapped in felt, drip irrigation system, coupling structure, and

vertical structure. The exterior finishing of the panel was composed of a sedum vegetation layer. During two of the three

monitoring phases, an extruded polystyrene layer (λ = 0. 035 W/(mK)) was embedded in the wall. The thickness of this

layer depended on the monitoring phase, which is detailed in Section 2. 4. The substrate's thermal conductivity was

determined through laboratory tests made following the UNE-EN 12667:2002 standard [42]. Both the dry substrate and

field capacity substrate (the amount of soil moisture held in the soil after excess water has drained away and the rate of

downward movement has decreased) were analysed and their respective thermal conductivities were determined. The dry

substrate's thermal conductivity was 0. 092 W/(mK), compared to 0. 191 W/(mK) for the damp substrate.

Table 2 summarizes the main characteristics of the layers forming the façades. As can be seen, the value of exterior

surface resistance (0. 13 m2K/W) exceeds the standard surface resistance value (0. 04 m2K/W) due to the added thermal

resistance provided by plants on exterior surfaces [3].

Water use is optimized by a cistern planter box at façade level, which allows the drainage solution to recirculate.
2.3 Data acquisition

The experimental box was monitored in order to obtain thermal data corresponding to each layer in the southern façade

and the interior of the space (Fig. 4 a and b). To this end, three surface PT100 thermoresistors (63 mm x 8 mm x 2 mm)

in three threads were installed in layers of the wall, from outdoors to indoors:

- External surface temperature between the sheet metal layer and the panel's felt layer (ExST).

- Surface temperature between the panel and the extruded polystyrene (PaST);

- Surface temperature on the interior surface of the extruded polystyrene, inside the module (InST).

Furthermore, two PT100 thermoresistors (length = 100 mm and diameter = 6 mm) in four threads were installed in the

central zone of the box, located near the floor and the ceiling:

- Internal temperature near the ceiling (CiT).

- Internal temperature near the floor (FiT).

All thermoresistors were paired in order to prevent the loss of data or reinstallation in case of errors or breakage. The

accuracy of the thermoresistors is ±0. 15 K [40, 41].

For each probe, temperatures were recorded every 5 minutes along with the date and time. The values can be represented

graphically by the software by selecting the probes desired for each moment as well as a time period. The SCADA-type

software was installed on a conventional PC. A M-340 programmable automation device from Schneider was used to

convert the analogue signals from the probes into temperatures.

2.4 Monitoring periods and selection of comparable days

As described in Section 2. 1, the monitoring period considered in this study was the month of July in 2011 and 2012.

During this time there were three monitoring phases corresponding to different configurations of the south façade (Fig

4b). Phase 1 was much longer than the other two phases (the wall was monitored with this configuration for some years,

see Table 3) but July 2011 was selected for this study.

During this phase, the south façade was consisted of modular panels composed of a metallic box (0. 6 m x 0. 6 m x 0.

08 m) containing the substrate wrapped in felt, drip irrigation system, coupling structure, and vertical structure. As

described in Section 2. 2, the exterior finishing of the panel was made up of a sedum vegetation layer. The U-value of the

façade was 1. 42 W/(m2K).

During phase 2, from 1 to 8 July 2012, a 3-cm layer of extruded polystyrene (λ = 0. 035 W/(mK)) was installed on the

internal side of the metallic box, achieving a U-value of 0. 64 W/(m2K). During phase 3, which took place from 12 to

25 July 2012, the thickness of insulation was increased to 7 cm, achieving a U-value of 0. 37 W/(m2K). Table 2
summarizes the main characteristics of the layers forming the south façade during the three monitoring phases.

In order to analyze the variation of the thermal behaviour of the façade based on the insulation thickness , three days

with similar weather conditions were chosen. This was done in order to eliminate meteorological variations so that

any thermal behaviour would only be due to the insulation thickness [43]. With this goal in mind, a mathematical

procedure was used to choose the three most similar days, each of which pertained to a monitoring phase. The data

considered were hourly values of temperature, humidity, radiation and wind speed (wind direction was disregarded),

all of which were available in the three phases of study.

“One day” meant any sequence of 24 consecutive hours; that is, a day could start at any time and include the

following 23 hours. The data were then cleaned since phase 1, which had more observations, also had a lot of

missing data. The database (Table 3) was split into 11 groups of consecutive hours, eliminating intermediate missing

data. Phase 2 provided 192 sets of hourly data, that is, i = 1,. . . , 169 possible days (i. e. , day 1: 1,. . . , 24; day 2:

2,. . . , 25 ;. . . ; day 169: 169,. . . , 192). Phase 3 provided 336 sets of hourly data, i. e. , 213 possible days.

Since phases 2 and 3 had fewer observations than phase 1, a Euclidean distance

d (f 2 i, f 3j), i = 1,. . . , 169, j = 1,. . . , 213

was calculated between each day of phase 2 and each day of phase 3.

The Euclidean distance means that a unit of temperature, for example, has the same importance (in distance) as a

unit of irradiance. However, looking at the descriptive measures of the data, it is clear that increasing the irradiance

by one unit of measurement is much easier than increasing the wind by one unit. For this reason, the relative

importance assigned to the variables was:

Irradiance >>> Relative humidity ≥ Temperature >>> Wind speed.

Table 4 indicates the average value and standard deviation of irradiance , relative humidity, temperature, and wind

speed recorded by the meteorological station during the monitoring periods.

Given all the Euclidean distances, the nearest days in phases 2 and 3 were selected by the automatic “K-means

clustering” partition process [44, 45]. The days selected were those whose distance was less than 13. 5. This value

was chosen because it marked a clear separation between a small group of close pairs and a large one of distant

pairs.

The Euclidean distance between selected days in phases 2 and 3 and all days in phase 1 was calculated:

d(f2i, f3j, f1k) = d(f2i, f3j) + d(f2i, f1k) + d(f3j, f1k).

These distances yielded the best days (i, j, k) of each phase.

The selected days are as follows:

- Phase 1: 05 July 2011 from 04:00 until 24 hours later.

 - Phase 2: 07 July 2012 from 04:00 until 24 hours later.

- Phase 3:12 July 2012 from 04:00 until 24 hours later.

As shown in Figure 5, the irradiance coincides perfectly on the three selected days (Fig. 5a). Differences can be seen,

however, for temperature (Fig. 5b) and relative humidity (Fig. 5c), although the graphs are similar. The relative humidity

during the three phases is similar up to 9:00 P. M. , after which point the curves of phases 1 and 2 continue with similar

trends while the curve for phase 3 remains below 30%. With regard to air temperature, the curves of phases 2 and 3

coincide from 8:00 A. M. to 3:00 P. M. , while the curve of phase 1 is consistently a couple of degrees above this. Starting

at 5:00 P. M. , the graphs of phases 1 and 3 coincide, while the graph of phase 2 lies about 2°C below with some variation

up to 4°C different.

3. Experimental study

Thanks to similar weather conditions for the three days selected for the study, it was possible to compare the thermal

behaviour of the southern wall during the three monitoring phases in order to identify the influen ce of the thermal

insulation thickness. For this goal, the following details were analyzed:

- the oscillation of the internal and external surface temperatures;

- the difference between the internal temperatures and the external surface temperature ;

- the difference between the internal surface temperature and the temperature on the panel surface .

3.1 Oscillation of the internal and external surface temperatures in the three case studies .

During phase 1, the maximum and minimum temperatures registered on the external surface of the module (External

Surface Temperature No Insulation = ExSTNI) were 22. 5°C and 14. 6°C respectively, yielding a thermal oscillation

of 7. 9°C. For the internal surface temperature (Internal Surface Temperature No Insulation InSTNI), the maximum

was 21. 7°C while the minimum was 17. 5°C, with an oscillation of 4. 2°C.

In phase 2, a layer of thermal insulation was attached to the internal side of the vegetation panel. In this case, the

maximum external surface temperature (External Surface Temperature 3 cm Insulation = ExST3I) was 23. 6°C and

the minimum was 14. 3°C, with an oscillation of 9. 3°C. This variation reduces to 4. 1°C for the internal surface

temperature (Internal Surface Temperature 3 cm Insulation = InST3I), with a maximum of 23. 3°C and a minimum

of 19. 2°C.

In phase 3, which was characterized by 7 cm of thermal insulation attached to the interior of the south ern wall, the

maximum external surface temperature (External Surface Temperature 7 cm Insulation = ExST7I) was 23. 3°C; the

minimum was 11. 7°C, with a thermal oscillation of 11. 6°C. However, the thermal oscillation of the internal surface

reduced to 3. 9°C, with a maximum of 23. 3°C and a minimum of 19. 4°C.
It is clear from the data that for the non-insulated wall, the thermal oscillation of the internal surface temperature is

only 37% lower than the external temperature. This increases to 52% for 3 cm of thermal insulation and reaches

77% for 7 cm of insulation. It is therefore evident how the insulation layer attenuates the temperature variation,

minimizing night-time cooling in particular. This data is illustrated in Fig. 6 a, b, and c and summarized in Table 5.

3.2 The difference between external surface temperature and internal temperature .

The effect of thermal insulation in the wall can also be seen by comparing the external surface temperature and the

internal temperatures. In fact, in phase 1, the temperature near the roof (CiTSI) reached a maximum of 23. 6°C, only

one degree higher than the maximum external surface temperature. The temperature near the floor obviously

remained lower, reaching 21. 9°C. However, the results are different in phases 2 and 3. Temperatures near the

ceiling reached 25. 6 and 25. 1°C for each phase respectively, about 2°C higher than the maximum external

temperature, while those near the floor were practically identical to the maximum surface temperature (Fig. 6 a, b

and c). This means that even though the effect of the vegetation is seen despite the presence of thermal insulation

[46], the thermal insulation ensures that the internal temperature remains higher . In spite of this, the effect is not

accentuated by the thickness of the insulation, since the behaviour of the module is very similar in this respect for

phases 2 and 3.

3.3 The difference between internal surface temperature and the temperature of the panel surface .

An important difference can be seen, however, between the internal surface temperature (InST) and the temperature

of the panel surface (PaST) due to the thickness of the insulation. In fact, while the 3 cm of insulation in phase 2

create a constant difference of about 2°C between the two sensors, the difference is much larger in phase 3 with

thicker insulation. It is never lower than 3. 3°C and it reaches a maximum of 8. 2°C (Fig. 6 b and c). In addition to

confirming that the potential effect of cooling due to the green surface is well attenuated by a 7 cm thick layer of

insulation, this study quantifies the effect, illustrating the reduction in attenuation to 33% during afternoon hours

and reaching a reduction of between 60 and 75% from the early morning to midday.

4. Numerical model

As stated in the introduction, the objective of the present work is to investigate the thermal behaviour of complex systems

such as green façades by means of building performance evaluations. Both modelling and simulation, as well as

experimentation, have all now become necessary steps in building design and operation (e. g. , understanding and

interaction).

Various models have been built to simulate the complex thermal phenomena of green walls, such as evapotranspiration,

the exchange of air between the vegetation and the external air, the hydrothermal state of the vegetation, and the flow of

heat through the green façade. The model used for this study was the building energy simulation program Energy Plus
[36], in conjunction with a parametric model designer called Grasshopper® [37, 38].

4.1 The parametric study

The EnergyPlus module is used to simulate thermodynamic aspects of green roofs, including: long- and short-wave

radiation exchanges under the green roof; the effects of the green roof on convective heat exchange;

evapotranspiration from the base and the plants; and conduction and accumulation of heat in the soil layer. To

simulate the green façade behaviour, the RoofVegetation module [47, 48] was considered. This simulation module

controls characteristic parameters of green wall, such as the leaf area index (LAI, i. e. , the ratio of the projected

green surface area to the total surface of the wall soil). There are many complex interactions that affect its energy

performance [49, 50]: the leaves reflect and absorb solar radiation and reduce the amount of radiation that reaches

the roof surface; there is long wave radiation exchange between the canopy elements, between the substrate surface

and the foliage and between the foliage and the environment; transpiration on the leaves; convective heat flux

between the leaves and the ambient air; conductance within the substrate which depends also on the amount of

moisture; and there is some solar reflection and absorption on the substrate surface .

Input to the simulation includes the characteristics of the material in the RoofVegetation module. Different aspects

of the green roof are specified, such as root depth, thermal properties, the density of leaf coverage, plant height, and

stomatal conductance, as well as soil humidity and irrigation. The complexity of the calculation system in

RoofVegetation module means that a large number of parameters are required to describe the details of the green

roof construction system. The RoofVegetation module only deals with the soil substrate, i. e. , the last component

of the roofing unit. It generally includes the drainage, insulation, and anti-root layers, which are modelled in the

EnergyPlus “Materials” sheets.

According to the EnergyPlus technical manual [36], the RoofVegetation module was developed specifically for flat

external surfaces, implying the presence of a significant soil layer and a specific heat -exchange mechanism.

Therefore, the use of this module is inappropriate for simulating the behaviour of highly inclined surfaces (e. g. ,

vertical walls). Vertical green walls can be simulated, however, with certain approximations, even though the results

have not been verified by the developers [46, 51]. Indeed, an adequate distinction should be made between irrigation

models in the two cases, since the vertical walls are not able to retain the same amount of water as roofs.

The latest and most advanced model [50] considers physiological processes within the plants, including

evapotranspiration and radiative and convective thermal exchange between the vegetation layer , the façade, the

surrounding environment, and the terrain. Input includes characteristic data regarding the vegetation layer (leaf

absorbance, leaf size, LAI, radiation attenuation coefficient, and stomatal conductance) and the meteorological data

at the location. This mathematical model was validated using a physical model at the Illinois Institute of Technology,
but neither this model nor other similar ones have been implemented in EnergyPlus. The software is therefore not

appropriate for simulating the behaviour of green walls in detail.

For this reason, a parametric method was developed to optimize the green façade (green façade optimization ,

GFO) in order to consider the characteristics of the growing media, irrigation and vegetation characteristics, and to

account for shading and insulation effects as well as evapotranspiration from the substrate and plants .

The method uses 3D parametric tools, Grasshopper® [38] in particular, which is a graphical algorithm editor

integrated with Rhinoceros, a 3D modelling program. Specifically, the research uses a variety of Grasshopper plug-

ins (Fig. 7), such as Ladybug, Honeybee and Galapagos, a mathematical computation solver. The method developed

is represented in a graphical flow diagram (Fig. 8) in which three open-source Grasshopper plugins were used:

1) Ladybug imports EnergyPlus weather data files (. EPW) into Grasshopper and creates 2D and 3D graphics

for weather data analysis, one for each phase (2011 and 2012).

2) Honeybee assigns the material characteristics to the geometry of the system as input to EnergyPlus and

connects Grasshopper to EnergyPlus.

3) Galapagos, an evolutionary computation solver, analyzes the result and minimizes the distance between the

measured and simulated temperatures to optimize the values. It is useful when the problem involves many

unknown parameters and it works by trial and error, adjusting the parameters in order to obtain increasingly

finer solutions approaching the required condition.

The 3D model of interest was a box with internal dimensions 1. 80 m (width) x 1. 80 m (depth) x 2. 40 m (high) as

depicted in Fig. 3 (Section 2. 2). Five of the six faces of the box were set to adiabatic to limit the study to the

behaviour of the last surface, which was exposed to the south. The material of this surface was the object of the

optimization process.

4.2 Identification of the thermal/physical green wall characteristics from Grasshopper + Galapagos and their

validation

Using experimental and simulated data, the Grasshopper script developed in this research uses the “evolutionary

solver” Galapagos to minimize the difference between them. Galapagos is able to define solutions for defined problem

with a defined fitness function. The criteria to be optimized should be defined numerically. These values are passed to

Galapagos in order to solve the problem. The solution is based on increasing/decreasing the numerical data in loops to

see whether the results are improving or worsening. As a result, the design outcomes will be optimized or a defined

problem will be solved by using a numerical based fitness function in its highest/lowest achievable values [52].
In Galapagos, various elements can be defined by the solver to set fitness function values. The solver helps the fitness

function by maximizing or minimizing the fitness value in order to define the best possible value. This approach offers a

unique system of analysis that can identify the right value to assign to the green wall (Fig. 9, the unknown parameters are

highlighted in orange).

The materials for each phase were set as follows (Fig. 10):

 First phase (1B): single layer (outside) with a RoofVegetation/EnergyPlus material that considered the soil and

green layer

 Second phase (2B): exterior layer as in phase 1B, as well as an internal layer with a 3-cm-thick insulating material

 Third phase (3B): exterior layer as in phase 1B, as well as an internal layer with a thicker (7 cm) insulating

material

For the first two phases, Galapagos was used to find optimized values to assign to the unknown variables in the

RoofVegetation module (leaf emissivity, visible absorbance, saturation, residual volumetric moisture content of the soil,

etc. ) and in the insulation material (thermal absorbance and density), which must be known [53]. These values were

determined during the first 1B (the bare wall) and second 2B (with the internal insulating layer) phases as represented in

Fig. 4b to obtain internal surface temperatures (exposed to the south) closer to the actual measured values (Fig. 11 and

Fig. 12).

The precise purpose of the optimization (phases 1B-2B) was to minimize the error as much as possible during daylight

hours. The optimization process requires a green wall model that is able to respond the simulation feedback. In other

words, optimization is an iterative process that starts with a defined geometry (structure), which is then simulated. Based

on the feedback obtained from the simulation engine, parameter values of the geometry (structure) are changed to obtain

better simulation results. The solution is based on increasing/decreasing the unknown numerical values in loops to see

whether the results move towards the measured conditions (i. e. , monitored temperatures). As a consequence, the

unknown parameters are optimized using a numerical fitness function at its highest/lowest achievable values [52].

A baseline scenario (default values) was used for comparison when studying the various parameters. The list of parameters

and their values is given in Table 8. For each parameter, the table shows the minimum and maximum values used during

the simulations. The results are summarized in Tables 6 and 7 and in Figs. 13, 14 (histograms).

The validation of the numerical results was made considering the third Phase 3B and using the results of previous

measurements (Section 3) to verify the correctness of the numerical simulation of thermal phenomena for the vertical

green wall. This technique is mainly due to the fact that the measurement shows the consistency of the model with the
reality. It is well known that achieving the greatest similarity between the measurement and simulation configurations is

essential [54]. The most important issue is to minimize any aspect that cannot be fully simulated. Attention was focused

on this aspect so that an evolutionary approach (i. e. , Galapagos) could be used to determine all these unknown

parameters, as stated previously.

The last phase (3B) was selected for validation, considering all of the previous material values (Table 1) with the

depth of the insulating layer changed from 3 to 7 cm. A period during daylight hours was considered because most of the

urban mitigation potential of green walls can be observed at that time [55]. Due to selected monitored data (see section

2. 4), a direct comparison was considered to validate simulated data. As stated in section 2. 4, July 2011 and 2012 were

considered and three days were selected as similar study scenarios. The third scenario (Phase 3, 12 July 2012) represents

boundary conditions for the validation, so that only one diurnal behaviour can be used. Numerous studies have shown

that a direct comparison point-by-point is feasible when small amounts of carefully selected data are compared (D. E.

Coleby 2002[60], (Archambeault & Connor, 2008 [57]). The results of the validation study are illustrated in Figure 13.

In the upper panel, the simulated values largely agree with the measured values and have the same behaviour. The

temperature comparison is shown in the bottom panel. A maximum difference of 1. 1°C can be seen between the simulated

and measured values in the middle part of the day.

This phase was useful for validating the parametric methodology and for demonstrating the predictive qualities of the

numerical model.

5. Results and Discussion

The main goal of this research was to understand the behaviour of insulation material in a vertical green wall. The

insulation effect is related to the insulation capacity of the different layers, which depend on their different compositions,

such as the substrate layer (thickness and materials), the air in the plant layer, other possible intermediate air layers, etc.

It is well known [58, 59] that there are no studies of green walls that extensively analyze the insulating effects linked to

the substrate layer and the thickness of the insulation. This is therefore a very important aspect of this research.

The thermal behaviour of green technological devices (green façades, living walls, green roofs) is still unknown in many

cases. It is therefore of interest to use a simulation tool to evaluate and establish the right thermal transmittance values

and other technical standards such as thermal comfort temperature for different climate conditions.

Various papers have demonstrated that the insulation layer attenuates the variation in temperature on the internal

surface of the façade and reduces the cooling effect due to vegetation [60, 33]. To quantify these effects for different

insulation thicknesses, the GFO methodology was used to define all optimized values for the insulation material
and the green layer (tables 6 and 7). As a consequence, the behaviour of the green wall was studied as a function of the

insulation thickness (varying from 3 to 13 cm). The results of this are shown in Figure 14.

Analyzing the temperature behaviour in the figure, it is clear that there is a cut-off thickness that makes the thermal

function of the green wall’s insulation almost useless. It is important to understand the amount of insulation that optimizes

the capacity of the green wall to correctly manage thermal fluxes and therefore to determine the limiting thickness of the

insulation beyond which there is no benefit directly connected to the thickness of the insulation.

Carefully observing the trends in the figure, it can be seen how slight variations in the simulated temperature can be seen

for thicknesses greater than 3 cm, while the curve is completely flat for values above 9 cm [61]. This result agrees to the

conclusion that in buildings with low insulation levels, the total load in the element with the addition of the green is lower

than that of a building with high insulation levels. This represents an important aspect for energy efficiency renovations

and urban comfort[62, 63,64] because when a green system is installed in an existing building with low insulation levels,

the energy performance of the building becomes similar to a new building with high insulation level, as stated by H. J.

Moon et al [65].

6. Conclusions

The main goal of this research was to understand the importance of insulation thickness in a green façade. A new

parametric optimization methodology called GFO (green façade optimization) was therefore developed and validated

using real data monitored in an experimental box located near Madrid (Spain). The GFO methodology was developed to

find all the unknown variables that well-known thermal simulation tools need to simulate the thermal behaviour of green

façades. Comparison of the simulations to experimental data allowed the model to be validated. The model was then used

to simulate the behaviour of the green wall, varying the insulation thickness from 3 cm to 13 cm.

The experimental data show that thermal insulation reduces temperature variations inside the building, minimizing night-

time cooling in particular, and that this effect increases with increasing insulation thickness. For the non-insulated wall,

the thermal oscillation of the internal surface temperature is only 37% lower than the external temperature . This

increases to 52% for 3 cm of thermal insulation and reaches 77% reduction from the oscillation recorded on the external

surface for 7 cm of insulation. Regarding the temperature on the panel surface, the results confirm that the potential

effect of cooling due to the green surface is well attenuated by a 7 cm thick layer of insulation: the reduction in

attenuation is around 33% during afternoon hours and reaches values between 60 and 75% from the early morning

to midday. As a result of this attenuation, and even though the effect of the vegetation is seen despite the presence

of thermal insulation, the thermal insulation ensures that the internal temperature remains higher. This effect could
be mitigated by the use of natural ventilation at night.

In addition, the GFO methodology showed an efficient insulation thickness up to 9 cm, above which more insulation

becomes redundant and inefficient because the green façade has the same thermal behaviour regardless of external

conditions. In fact, for an insulation depth of 3 cm, there is a lapse rate about 3°C during the day that decreases to about

2°C for 6 cm and about 1°C for 9 cm. Starting from a 10-cm insulation thickness, there is a temperature difference less

than 1ºC during the day, an effect that can be disregarded. This represents an interesting result because the correct

insulation depth of green wall can be considered economically feasible and should be implemented, as it will provide

higher rates of comfort accompanied by lower air conditioning energy costs. In fact, optimum envelope components and

optimum insulation thicknesses in green walls have the potential to reduce building energy consumption and therefore to

reduce CO2 emissions. In this context, the effect of the optimum insulation thickness on the environment due to optimum

energy consumption can be defined by architects and designers in the initial design stages for more sustainable planning.

Due to the research requirements, the simulations were conducted on buildings without heating and cooling systems and

without internal thermal loads. To quantify the effective margin of thermal gain offered by vertical and horizontal green

solutions in terms of energy, the environment, and cost, it would be useful to conduct more detailed simulations on the

buildings under real use conditions. This means quantifying the effective contribution offered by the green walls/roofs in

reducing the heat loads of the building, consequently modulating the size of the systems in order to maximize the benefit

in terms of cost and energy during the entire summer period, and also extending the analysis to the winter period. It would

also be useful to simulate and compare the temporal behaviour of green envelopes with different types of heating (radiant,

convective air and water) and conditioning (air, radiant) systems. This would allow for the identification of possible

successful one-to-one relationships from the point of view of reducing overall energy needs.

This can be considered just the initial result, as further investigations related to different climate areas are needed.

Acknowledgements

This research was financed by the Spanish Ministry of Economic Development through the research project MODIFICA.

References

1) M. Hunter, N. S. G. Williams, J.P. Rayner, L. Aye, D. Hes & S. J. Livesley Quantifying the thermal performance
of green façades: A critical review. Ecological Engineering (2014) 63, 102–113.
http://doi.org/10.1016/j.ecoleng.2013.12.021
2) S. F. Larsen, C. Filippín & G. Lesino, Thermal Simulation of a Double Skin Façade with Plants. Energy Procedia
(2014). 57, 1763–1772. http://doi.org/10.1016/j.egypro.2014.10.165
3) K. Perini, M. Ottelé, Fraaij a. L a., Haas EM, R. Raiteri, Vertical greening systems and the effect on air flow
4) M. Köhler, Green facades—a view back and some visions. Urban Ecosystems, (2008), 11(4), 423–436.
5) R. W. F. Cameron, J. Taylor, M. Emmett, A Hedera green facade - Energy performance and saving under
different maritime-temperate, winter weather conditions. Building and Environment, (2015), 92, 111–121
6) O. E. Bellini, L. Dagli, Verde verticale. Soluzioni tecniche nella realizzazione di living walls e green façades.
Maggioli Editore, (2009)
7) CY Cheng, KKS Cheung, LM Chu, Thermal performance of a vegetated cladding system on facade walls. Build
Environ (2010), 45:1779–87
8) G. Pérez, L. Rincón, A. Vila, JM. González, LF. Cabeza, Behaviour of green facades in Mediterranean
Continental climate. Energy Convers Manag, (2011), 52:1861–7
9) V. Tatano, Verde. Naturalizzare in verticale. Maggioli Editore, (2008)
10) C. van Uffelen, Facade Greenery Contemporary Landscaping. Braun Publishing AG, (2011)
11) Q. Chen, B. Li, Liu X, An experimental evaluation of the living wall system in hot and humid climate. Energy
Build, (2013), 61:298–307
12) K. Perini, P. Rosasco, Cost–benefit analysis for green façades and living wall systems. Build Environ (2013),
70:110–21
13) F. Olivieri, D. Redondas, L. Olivieri, J. Neila, Experimental characterization and implementation of an integrated
autoregressive model to predict the thermal performance of vegetal façades. Energy and Buildings, (2014), 72,
309–321
14) M. Haggag, A. Hassan & S. Elmasry, Experimental study on reduced heat gain through green façades in a high
heat load climate. Energy and Buildings (2014) 82, 668–674 http://doi.org/10.1016/j.enbuild.2014.07.087
15) M. T. Hoelscher, T. Nehls, B. Jänicke & G. Wessolek Quantifying cooling effects of facade greening: shading,
transpiration and insulation. Energy and Buildings, (2015), http://doi.org/10.1016/j.enbuild.2015.06.047
16) F. Olivieri, L. Olivieri, J. Neila, Experimental study of the thermal-energy performance of an insulated vegetal
façade under summer conditions in a continental mediterranean climate. Energy Build (2014) 61–77.
17) E. A. Eumorfopoulou, K. J. Kontoleon, Experimental approach to the contribution of plant-covered walls to the
thermal behaviour of building envelopes. Building and Environment, 44(5), 1024–1038
18) N. H. Wong, A. Y. Kwang Tan, Y. Chen, K. Sekar, P. Y. Tan, D. Chan, N. C. Wong, Thermal evaluation of
vertical greenery systems for building walls. Building and Environment, (2010), 45(3), 663–672
19) C. Y. Jim, H. He, Estimating heat flux transmission of vertical greenery ecosystem. Ecological Engineering,
(2011), 37(8), 1112–1122
20) H. F. Di, Wang, Cooling effect of Ivy on a wall. Exp Heat Transf, (1999), 12:235–45
21) G.Latini, R. Cocci Grifoni, S .Tascini, Thermal comfort and microclimates in open spaces (2010) Thermal
Performance of the Exterior Envelopes of Whole Building21s - 11th International Conference,
22) R. Cocci Grifoni, M. Pierantozzi, S. Tascini, G. Passerini, Assessing the representativeness of thermal comfort
in outdoor spaces WIT Transactions on Ecology and the Environment (2011) 155, 835-846.
23) Saiz, S., Olivieri, F., Neila, J. (2016). Green roofs: Experimental and analytical study of its potential for urban
microclimate regulation in Mediterranean–continental climates. Urban Climate, 17, 304-317
24) M. Manso, J. P. Castro-Gomes, Thermal analysis of a new modular system for green walls. Journal of Building
Engineering (2016) 7, 53–62. http://doi.org/10.1016/j.jobe.2016.03.006
25) S. Nadia, S. Noureddine, N. Hichem & D. Djamila, Experimental Study of Thermal Performance and the
Contribution of Plant-Covered Walls to the Thermal Behavior of Building. Energy Procedia (2013) 36, 995–
1001. http://doi.org/10.1016/j.egypro.2013.07.113
26) S. Sheweka, A. N. Magdy, The Living walls as an Approach for a Healthy Urban Environment. Energy
Procedia (2011), 6:592–9
27) N. H. Wong, Tan AYK, P. Y. Tan, N. C. Wong, Energy simulation of vertical greenery systems. Energy Build,
(2009), 41:1401–8
28) E. Ng, L. Chen, Y. Wang, C. Yuan, A study on the cooling effects of greening in a high-density city: An
experience from Hong Kong. Build Environ ,(2012), 47:256–71
29) E. Alexandri, P. Jones, I. Beausoleil-Morrison, Temperature decreases in an urban canyon due to green walls
and green roofs in diverse climates. Build Environ, (2008), 43:480–93
30) R. Cocci Grifoni, G. Latini, S. Tascini, The representative day technique in the analysis of thermal comfort in
outdoor urban spaces, (2011) PLEA 2011 - Architecture and Sustainable Development, Conference Proceedings
of the 27th International Conference on Passive and Low Energy Architecture, pp. 397-401
31) M. Ottelé, H. D. van Bohemen, A. L. Fraaij Quantifying the deposition of particulate matter on climber
vegetation on living walls, Ecol Eng ,(2010), 36:154
32) B. A. Currie, B. Bass, Estimates of air pollution mitigation with green plants and green roofs using the
UFORE model. Urban Ecosyst, (2008), 11:409–22
33) F. Bisegna, B. Mattoni, P. Gori, , F. Asdrubali, , C. Guattari, , L. Evangelisti, , S. Sambuco, F. Bianchi, Influence
of insulating materials on green building rating system results Energies, (2016) 9 (9), art. no. 712
34) K. J. Kontoleon, E. A. Eumorfopoulou, The effect of the orientation and proportion of a plant-covered wall layer
on the thermal performance of a building zone. Building and Environment, (2010), 45(5), 1287–1303
35) ZUBIN KHABAZI, Generative Algorithms (using Grasshopper), www. morphogenesism. com, (2010)
36) LBNL. 2010. EnergyPlus Engineering Reference. EnergyPlus Manual Documentation Version 5.0, April. The
Board of Trustees of the University of Illinois and the Regents of the University of California through the Ernest
Orlando Lawrence Berkely National Laboratory
37) S. R. Mostapha; P. Michelle, 2013. Ladybug: a parametric environmental plugin for grasshopper to help
designers create an environmentally-conscious design. In: Proceedings of the 13th International IBPSA
Conference Held in Lyon, France. http://www.ibpsa.org/proceedings/BS2013/p_2499.pdf
38) http://www.grasshopper.com/
39) S. R. Mostapha, P. Michelle, Ladybug: a parametric environmental plugin for grasshopper to help designers
create an environmentally-conscious design. In: Proceedings of the 13th International ,(2013) , IBPSA
Conference Held in Lyon, France. http://www. ibpsa. org/proceedings/BS2013/p_2499. pdf
40) http://www.global-download.scheider-electric.com , accessed 30th August2016
41) F. Olivieri, C. Di Perna, M. D’Orazio, 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 (2013) 63,1–14
42) http://www.aenor.es/aenor/normas accessed 29th August 2016
43) F. Gugliermetti, F. Bisegna, Meteorological days for HVAC system design in Mediterranean climate, Building
and Environment, (2003), 38 (8), pp. 1063-1074
44) J.A. Hartigan, Clustering algorithms. John Wiley & Sons, Inc. Hartigan, J. A., (1975)
45) M.A. Wong, Algorithm AS 136: A K-Means Clustering Algorithm, Journal of the Royal Statistical Society,
Series C (Applied Statistics) (1979) 28 (1): 100–108. JSTOR 2346830, 1979[8]
46) S. Flores Larsen, C. Filippín & G. Lesino, Modeling double skin green façades with traditional thermal
simulation software. Solar Energy (2015). http://doi.org/10.1016/j.solener.2015.08.033
47) UIUC, LBNL. 2007a. EnergyPlus Input Output Reference, U.S. Department of Energy
48) UIUC, LBNL. 2007b. EnergyPlus Engineering, Reference, U.S. Department of Energy
49) D. Holm, Thermal improvement by means of leaf cover on external walls - a simulation model. Energy and
Buildings (1989), 0
50) Susorova, M. Angulo, P. Bahrami, B. Stephens, A model of vegetated exterior facades for evaluation of wall
thermal Performance, Building and Environment 67, (2013)
51) F. Fantozzi, C. Bibbiani, C.Gargari, Parametri fisico-tecnici delle specie vegetali utilizzate per la realizzazione
di tetti e pareti verdi nelle regioni mediterranee, per la realizzazione di un data-base specifico da utilizzare in
programmi di simulazione energetica degli edifici (2013) Report RdS/PAR2013/137
52) Z.Khabazi, Generative Algorithms using Grasshopper (2012), http://www.morphogenesism.com/
53) P. Gori, F. Bisegna, Thermophysical parameter estimation of multi-layer walls with stochastic optimization
methods International Journal of Heat and Technology, (2010), 28 (1), pp. 109-116
54) R. Jauregui, F. Silva , Numerical Validation Methods, Numerical Analysis - Theory andApplication, (2011),
Prof. Jan Awrejcewicz (Ed.), ISBN: 978-953-307-389-7
55) Gagliano, M. Detommaso, F. Nocera, U. Berardi, The adoption of green roofs for the retrofitting of existing
buildings in the Mediterranean climate (2016) International Journal of Sustainable Building Technology and
Urban Development, 7 (2), pp. 116-129
56) D. E. Coleby, A. P. Duffy, Analysis of techniques to compare complex data sets. The current issue and full text
archive of this journal is available, (2002), 21, 540-553
57) B. Archambeault, Z. Yu, Application of the Feature Selective Validation Method to radio path loss
measurements. 2009 IEEE International Symposium on Electromagnetic Compatibility, (2009), 259-263.
58) G. Pérez, L. Rincón, A. Vila, J. M. González & L. F. Cabeza, Behaviour of green facades in Mediterranean
Continental climate. Energy Conversion and Management (2011) 52(4), 1861–1867
http://doi.org/10.1016/j.enconman.2010.11.008
59) G. Pérez, L. Coma, I. Martorell & L. F. Cabeza, Vertical Greenery Systems (VGS) for energy saving in buildings:
A review. Renewable and Sustainable Energy Reviews (2014) 39, 139–165
http://doi.org/10.1016/j.rser.2014.07.055
60) Gagliano, M. Detommaso, F. Nocera, G. Evola, A multi-criteria methodology for comparing the energy and
environmental behavior of cool, green and traditional roofs, (2015), Building and Environment, 90, pp. 71-81
61) Jaffal, S.-E. Ouldboukhitine, R. Belarbi, A comprehensive study of the impact of green roofs on building energy
performance (2012) Renewable Energy, 43, pp. 157-164
62) R.C. Grifoni, R. D'Onofrio, M. Sargolini, M. Pierantozzi, A parametric optimization approach to mitigating the
urban heat Island effect: A case study in Ancona, Italy (2016) Sustainability (Switzerland), 8 (9), art. no. 896
63) F. Nardecchia., F. Gugliermetti, F. Bisegna. How temperature affects the airflow around a single-block isolated
building, (2016) Energy and Buildings, 118, pp. 142-151.
64) F. Nardecchia., F. Gugliermetti, F. Bisegna., A novel approach to CFD analysis of the urban environment, (2015)
Journal of Physics: Conference Series, 655 (1), art. no. 012013.
65) H. J. Moon, K.A. An, S.W. Han, Energy Saving Effects of Green Roof in Existing Buildings with Low Insulation
Levels, (2014), ASim2014 Proceedings, International IBPSA Building Simulation Conference
Figure captions

Figure 1: General view of the experimental building.

Figure 2: Weather station installed next to the experimental building.
Figure 3: Schematic plan and section of the experimental box.
Figure 4: Section of the green wall.
Figure 5: Schematic section of the experimental box and position of the thermoresistors (a); configurations of the southern

façade during the three phases of monitoring (b).

Figure 6: Weather conditions during the three selected days: irradiation (a), temperature (b) and relative humidity (c).
Figure 7: Temperature values recorded during the three selected days: 5 July 2011, Phase 1, southern façade without

insulation, NI = No Insulation (a); 7 July 2012, Phase 2, southern façade with a 3-cm insulation layer , 3I = 3 cm of

insulation; (b) and 12 July 2012, Phase 3, southern façade with a 7-cm insulation layer, 7I = 7 cm of insulation.
Figure 8: Honeybee and Ladybug plug-in.
Figure 9: Graphical flow diagram
Figure 10: Parameters of the GreenRoof module.
Figure 11: Grasshopper optimized materials for each phase
Figure 12: First optimization, Phase 1B.
Figure 13: Second optimization, Phase 2B.
Figure 14: Third validation, Phase 3B.
Figure 15: Thermal behavior of green façade with different insulation thickness
Table 1: Local weather conditions during the monitoring periods. The number of hours is shown in percentage with relation to the total
number of hours of the monitoring periods.

Global Temperature Relative

irradiance humidity

[W/m2] [°C] [%]

>800 >1000 >26 >30 <15 <20 >50

July 2011 33% 8% 34% 13% 3% 5% 22%

July 2012 32% 5% 68% 31% 2% 38% 15%

Table 2: Main characteristics of the layers forming the façade during the three monitoring phases.

Green wall

Monitoring phase 1 Monitoring phase 2 Monitoring phase 3

Thickness λ R Thickness λ R Thickness λ R

[m] [W/(mK)] [m2K/W] [m] [W/(mK)] [m2K/W] [m] [W/(mK)] [m2K/W]

Rse 0.130 0.130 0.130

Galvanized 0.0010 50 0.006 0.0010 50 0.006 0.0010 50 0.006

steel

Polyester 0.0015 0.250 0.006 0.0015 0.250 0.006 0.0015 0.250 0.006

felt

Substrate 0.0800 0.190 0.420 0.0800 0.190 0.420 0.0800 0.190 0.420

(field

capacity)

Polyester 0.0015 0.250 0.006 0.0015 0.250 0.006 0.0015 0.250 0.006

felt

Galvanized 0.0010 50 0.006 0.0010 50 0.006 0.0010 50 0.006

steel

Extruded - - - 0.0300 0.035 0.857 0.0700 0.035 2.000

polystyrene

Rsi 0.130 0.130 0.130

Total 0.704 1.561 2.704

thermal

resistance
Table 3. Data base studied for the data analysis. The sample size is given in number of hours.

2009
e
d
c
b
a

0 200 400 600 800 1000 1200 1400 1600

a b c d e
abcde 287 1325 257 267 647

Sample size (hours)

2010
d
c
b
a
0 200 400 600 800 1000 1200 1400 1600

a b c d
abcd 1165 438 567 547

Sample size (hours)

2011
b
a
0 200 400 600 800 1000 1200 1400 1600

a b
ab 1438 525

Sample size (hours)

Table 4. Average value and standard deviation of irradiance, relative humidity, temperature, and wind recorded by the meteorological
station during the monitoring periods.
Irradiance Relative Humidity Temperature Wind
Average value 361.23 33.26 22.72 2.75
Standard deviation 386.61 12.36 4.61 2.06
Table 5. Maximum and minimum superficial temperatures, in the exterior and the internal layer of the vegetal façade during the three
selected days: NI (No Insulation layer, phase 1); 3I (3 cm insulation layer, phase 2); 7I (7 cm insulation layer, phase 2). The reduction
of the difference between the maximum and minimum surface temperature is expressed in each case, as a percentage of the difference
between the maximum and minimum internal surface temperature (ΔInST) and to the difference between the maximum and minimum
external surface temperature. (ΔExST).

Max(ExST) Min(ExST) ΔExST Max(InST) Min(InST) ΔInST Reduction

[max-min] [max-min]
[°C] [°C] [°C] [°C] [°C] [°C] [%]
NI 22.5 14.6 7.9 21.7 17.5 4.2 37
3I 23.6 14.3 9.3 23.3 19.2 4.1 52
7I 23.3 11.7 11.6 23.3 19.4 3.9 77
Table 6. Optimized values of the green layer determined with the parametric optimization routine.

Leaf emissivity 0.80

Conductivity of dry soil [W/(mK)] 0.2

Visible absorbance 0.70

Saturation volumetric moisture content of the soil layer 0.30

Residual volumetric moisture content of the soil layer 0.10

Initial volumetric moisture content of the soil layer 0.30

Table 7. Optimized values of the insulation material determined with the parametric optimization routine.

Density [Kg/m3] 0.80

Thermal absorbance 0.4

Table 8. List of vertical vegetation parameters studied
Parameter Name Baseline Min Max Comments
(in EnergyPlus) value
Vegetation Height of Plants {m} 0.2 0.01 1 0.1-0.5 are reasonable for living walls
Leaf Area Index 4.0 0.001 5
{dimensionless}
Leaf Reflectivity 0.5 0.1 0.5
{dimensionless}
Leaf Emissivity 0.8 0.8 1 Default=0.95
Minimum Stomatal 300 50 300
Resistance {s/m}
Growing Roughness VeryRough 6 values from VerySmooth to
Medium VeryRough
Thickness {m} 0.08 0.05 0.5 0.15 & 0.30 are common for green
roofs.. Living walls are slimmer
Conductivity of Dry 0.2 0.2 1 Typically 0.3-0.5 for green roof
Soil {W/m-K} substrate
Density of Dry Soil 1500 300 2000 Typically 400-1500
{kg/m3}
Specific Heat of Dry 1100 501 2000 Default=1000
Soil {J/kg-K}
Thermal Absorptance 0.90 0.81 1 Typically 0.90-0.98
Solar Absorptance 0.7 0.4 0.9 Typically 0.6-0.85
Visible Absorptance 0.7 0.51 1
Moisture in Saturation 0.3 0.11 0.1 Typically less than 0.5
Growing Volumetric Moisture
Media Content of Soil Layer
Residual Volumetric
Moisture Content of
Soil Layer
Initial Volumetric 0.1 0.01 1
Moisture Content of
Soil Layer
Irrigation Daily Rate 0.3 0.11 1
{cm/hr}