You are on page 1of 8

Energy and Buildings 37 (2005) 12601267 www.elsevier.

com/locate/enbuild

Experimental measurements and numerical modelling of a green roof


Renato M. Lazzarin, Francesco Castellotti *, Filippo Busato
Department of Management and Engineering, University of Padova, Stradella S. Nicola 3, 36100 Vicenza, Italy Received 26 November 2004; received in revised form 5 February 2005; accepted 12 February 2005

Abstract Green roof utilisation has been known since ancient times both in hot and cold climates. Nowadays, it has been reconsidered at issue of energy saving and pollution reduction. In this paper, some measurement sessions on a green roof installed by the Vicenza Hospital are described. A data logging system with temperature, humidity, rainfall, radiation, etc. sensors surveyed both the parameters related to the green roof and to the rooms underneath. The aim is to evaluate the passive cooling, stressing the evapotranspiration role in summer time. Furthermore, the enhanced insulating properties have been tested during winter time. A predictive numerical model has been developed in a building simulation software (TRNSYS) to calculate thermal and energy performances of a building with a green roof, varying the meteorological dataset for a specic geographic zone. # 2005 Elsevier B.V. All rights reserved.
Keywords: Green roof; Energy saving; Evapotranspiration

1. Introduction In the last decades, the space left to the greenery in the urban landscape has decreased, allowing the uncontrolled growing of roads and buildings. The most worrying effects are the worsening of the air quality and the increasing of the average urban temperature. The last is due to the particular urban morphology that increases the thermal storage and lowers the thermal exchanges with the surrounding areas: more thermal energy is entrapped in the city and the urban air gets hotter than in the surrounding country. The phenomenon is called Urban Heat Island effect (UHI): it can increase air temperatures to even more than 10 8C [13]. It is possible to reduce this effect by increasing the vegetation cover of urban areas and buildings roofs. Moreover, the vegetation used to cover roofs can limit the heat ux through the roof itself reducing the thermal load of the rooms underneath [46]. Few experimental works and accurate analytical models are available in literature to

explain the evapotranspiration phenomenon as evidenced by many authors [7,8]. In the present paper, after a brief description of the characteristics of a green roof, the specic aim of the research is described. The potentiality of a green roof in the reduction of the cooling and heating load of a building is evaluated by means of experimental data. The evaporative cooling effect is investigated with a nite difference method: its contribution is computed as residual term in the energetic balance of the system. The whole system is also simulated developing a computing model that takes into account all the variables (both regarding the weather and the building).

2. Technical characteristics of a green roof The various layers of a green roof perform the different functions of a natural soil: giving the nutritional elements, storing water, letting transpiration and drainage in the meantime (Fig. 1). Generally speaking, on the pre-existing roof covered with a waterproong sheet, an anti-root barrier prevents the damage of the structure. A retention sheet collects the

* Corresponding author. Tel.: +39 0444 998778; fax: +39 0444 998888. E-mail address: caste@gest.unipd.it (F. Castellotti). 0378-7788/$ see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2005.02.001

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267

1261

Nomenclature A C e e* ET G Kc ks LAI r R Rn RH s t u adduction ux (W/m2) thermal accumulation (W/m2) air partial vapour pressure (Pa) air saturation vapour pressure (Pa) evapotranspiration ux (W/m2) conduction ux (W/m2) crop coefcient short-wave extinction coefcient leaf area index water specic heat of vaporisation (J/(kg K)) incident global solar radiation (W/m2) solar radiation entering the system (W/m2) relative humidity layer thickness (m) temperature (8C) wind speed measured at 2 m above the soil (m/ s)

Fig. 1. The different layers of a standard green roof.

Greek symbols a adduction coefcient (W/(m2 K)) g psychrometric constant (Pa/K) D slope of saturation vapour pressure versus temperature function (Pa/K) Dt time interval (s) c soil water content (kg/m2) l thermal conductivity (W/(m K)) r specic gravity (kg/m3) u water ux (kg/(m2 s)) Subscripts 0 reference value I, II, III soil layer a air c concrete roof d drainage ev evaporation i inside o outside r rainfall s soil sat saturation w waterproong sheet

nest soil particles, letting the water pass. The soil used in a green roof has a low volumetric mass (800900 kg/m3) and it is enriched with minerals and humus. The thickness can vary from 10 to 50 cm: it is the main characteristic that allows to distinguish the so-called extensive green roof from the intensive one. The extensive solution is suitable for lightweight and low height buildings: the utilised plants are species of sedum, shrubs and bushes that need low maintenance and can be self-generative. The intensive landscape, suitable for underground garages and heavy buildings, is a common roof garden, with bushes, ornamental plants and also trees, and needs regular garden maintenance.

3. The model of the physical system The experimental sessions described in this paper, regard the summers of 2002 and 2003 and the winter of 2004. The measurements were carried out on a green roof installed by the S. Bortolo Hospital in Vicenza, in the north-east of Italy. The green roof has an extension of about 1000 m2 and it consists of a 20 cm soil layer over an 11 cm drainage layer made of expanded polyethylene. The greenery is a kind of sedum, grown from premixed seed in the soil. The green roof system has been investigated in dynamic state in a mono-dimensional analysis at the nite differences [9]. The physical system is divided into different segments and nodes. The soil is described with three nodes (subscribes I, II and III), while one node describes the drainage layer (d), the waterproong sheet (w) and the structural concrete roof (c). The others elements are neglected for the limited mass. The upper border is the ambient air and the lower one is the underneath room. The whole system is described in Fig. 2, where the oriented thermal uxes are indicated with continuous lines. ETi is the thermal ux due to the evapotranspiration process of the generic node i: for the nodes II, III and d, it enters the nodes I, II and III, respectively, with opposite sign: that means that the evaporation heat of the node i is the condensation heat of the upper one. Gi,j is the conduction ux between two adjacent nodes and Ci is the thermal accumulation of each node. Ao and Ai indicate the external and internal adduction uxes, sum of convection and radiative uxes.

particles eventually fallen down from the upper layers. The drainage layer has the function of recreating the natural condition for the vegetation growing: it accumulates the water by means of little tanks in the upper side, but it permits the drainage of the excess water through some holes. The drainage layer can be made of different materials (recycled polyethylene or caoutchouc, for example) depending on the requested function: it can offer also an added thermal insulation. Over the drainage element, a lter sheet stops the

1262

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267

where t is the temperature of the node and s is its thickness. Yet, the water soil content alters the specic heat c as well as the specic gravity r; hence, the thermal accumulation of node I is calculated by
0 tI tI Dt where the temperature superscribe refers to the previous time interval that lasts Dt. Before describing the water balance of the system, it is necessary introducing the specic water content of the soil c and its specic water content in saturated condition csat. The ratio of them denes the relative soil humidity:

CI cI rI sI

RHs

c csat

(5)

At the same time as the energetic uxes, the water uxes take place: in Fig. 2, they are indicated with dotted lines. A rst kind of them depends on the evaporative ux; at node I is
Fig. 2. The nite differences model of physical system.

uev;I

ETI r

(6)

For example, one can write the balance for node I in terms of specic uxes, that is, watts per square meter: Rn Ao ETII GI;II ETI CI (1) The solar radiation ux coming into the system is Rn, that is, the net contribution after solar reection and absorption of 8 c ur uev;I uev;II Dt csat ; 1 < I c ur uev;I uev;II Dt; Dt : 0;I

where r is specic heat of vaporisation. The second kind of water ux depends on the saturation or dryness condition of the soil: if a node reaches the saturation, the excess water drains down to the lower one, while if it becomes completely dry, it recalls the needed water from the same one. At node I, the next balance permits to calculate the drainage ux towards the node II:

ud;I

if cI ur uev;I uev;II Dt > csat if cI ur uev;I uev;II Dt < 0 if 0 cI ur uev;I uev;II Dt csat

(7)

the greenery: it is dened by an exponential law [8]: Rn Reks LAI (2)

where ur is the rainfall ux.

The fraction of the incident radiation R entering the system is described by the leaf area index (LAI), that is the ratio between the green area and the underneath soil area, and by the short-wave extinction coefcient ks. To make the model easier, the outside convective and radiative thermal uxes are condensed into an adduction ux: Ao ao ta;o ts;o (3)

4. The experimental sessions All the meteorological variables were measured to evaluate the green roof behaviour through the above equations: solar global irradiation, air temperature and rainfall. Although the air relative humidity is not present in the previous relations, it was measured to be used in the predictive model that will be described in Section 5. Considering the need of knowing the temperature time history of the soil at different depth of the nodes, but also at the surface to calculate the adduction ux, the soil temperature was logged at the surface, at an intermediate depth and at the interface between soil and drainage layer. Yet, going downstream, the temperature between drainage layer and waterproong sheet was measured. Regarding the underneath room, other two variables were logged: the

where ao is the outside surface adduction coefcient, ta,o the outside air temperature and ts,o is the outside surface temperature. Taking into account that the thermal conductivity (l) in the soil depends on the water content of each layer, the conductive ux between node I and II is tI tII GI;II sI (4) sII 2lI 2lII

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267 Table 1 Set of sensors used in the experimental sessions Sensor Pyranometer Ambient air thermo-hygrometer Rain collector Soil thermometers Soil hygrometers Room thermometers Type Silicon photodiode RTD/capacitive sensor Tipping bucket Thermocouples CuC Resistive sensor Thermocouples CuC Geometric variables Single soil layer thickness Drainage layer thickness Waterproong sheet thickness Structural roof thickness Solar variables Short-wave extinction coefcient Leaf area index (session 2002) Leaf area index (session 2003) Leaf area index (session 2004) Physical variables Density of dry soil Density of saturated soil Density of drainage layer Density of waterproong sheet Density of structural roof Water content of saturated drainage layer Thermodynamic variables Specic heat of dry soil Specic heat of drainage layer Specic heat of waterproong sheet Specic heat of structural roof Conductivity of drainage layer Conductivity of waterproong sheet Conductivity of structural roof Outside adduction coefcient Inside adduction coefcient Table 2 Set of variables of the model Value 6.7 10.8 1 30 0.8 1.1 1.2 1.5 766 1270 25 1200 1000 4.5 1000 1200 1470 900 0.15 0.17 0.38 11 7 Unit cm cm cm cm

1263

temperature of the internal roof surface and inside air temperature, obviously to estimate the adduction ux. Finally, the soil humidity was measured: that piece of information has been used to tune the model, especially in terms of LAI and adduction coefcients. In fact, closing the energetic balance for each node, it is possible to calculate the evapotranspiration and so the water ux: minimising the difference between experimental and calculated results of soil water content, the optimum values of LAI and adduction coefcients were found. In Table 1, the different instruments are listed. All the signals were logged every second and the average value on 15 min was recorded. In the subsequent computations, the time resolution was set at 1 h. The two summer sessions covered August and September of 2002 and June and July of 2003. The winter session lasted from February to March 2004. For each node, the energetic balance was calculated starting from the set of values reported in Table 2. 4.1. Summer sessions Here, the discussion is limited to the session of 2003. In Fig. 3, the hourly mean irradiance is shown together the cumulative rainfall: it rained about 85 mm in the space of

kg/m3 kg/m3 kg/m3 kg/m3 kg/m3 kg/m2 J/(kg K) J/(kg K) J/(kg K) J/(kg K) W/(m K) W/(m K) W/(m K) W/(m2 K) W/(m2 K)

2 months. As said, the role of the evapotranspiration process was one of the investigated points. To this end, it is useful to compare the behaviour of the green roof in dry conditions and in wet conditions. The most dry and the most wet periods of the session are chosen to highlight the different energetic uxes. The length of the two periods is the same: 10 days. In the rst one (413 June), it rained 0.2 mm and the

Fig. 3. Measured hourly mean irradiance and cumulative rainfall for the summer session in 2003.

1264

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267

Fig. 4. Energy exchanges for 100 unit of incident solar irradiation between the nodes of the green roof model. Normal: green roof in dry conditions; italic: green roof in wet conditions.

soil relative humidity was about 10%, while in the second one (29 June8 July), it rained 28 mm and the upper soil layer reached the saturated condition. The incident solar irradiation in the two periods was almost the same (respectively, 186 and 184 MJ/m2) and so it is possible to normalise to 100 incident energy units, the energy exchanges between the nodes. The results are reported in the graph of Fig. 4: the greenery reected 23 units that were measured rotating the pyranometer of 1808 on its horizontal axis; it absorbed 39 units that are calculated as residual term after the transmitted fraction of Eq. (2).

The outside air conditions are reported in the graph of Fig. 5: in dry conditions the temperature at the surface reached even 55 8C and so the outgoing adduction ux (24 units) was higher than the corresponding one in wet conditions (13 units) when the surface temperature exceeded 40 8C only once. On the other hand, the wet soil gave rise to an evapotranspiration of 25 units whereas in dry conditions that contribution was limited to 12 units. This changed role between adduction and evapotranspiration produced a different thermal conduction towards the lower layer: in wet conditions the units were reduced by 80% with respect the dry conditions (Fig. 4). This trend continues and increases going downstream. The most important result regards the inside adduction: in dry conditions 1.8 units entered the underneath room, while in wet conditions 0.4 units left the conditioned room: not only the thermal gain due to the roof was eliminated, but also the wet green roof gave a contribution to a passive cooling. To evaluate the energy savings due to the installation of a green roof, it is possible to compare the previous results with those produced by a common roof with an additional insulating layer and a conventional covering (Table 3) with an albedo of 0.1 and an adduction coefcient of 14 W/ (m2 K). The behaviour of the traditional roof was estimated by means of a numerical simulation [10] that ran with the same boundary conditions of the experimental session that means with the measured solar irradiance and outside and inside air temperatures. The results are well-summarised in the graph of Fig. 6 that compares the attenuation of 100 units of incident solar irradiation coming down through the roof. The rst difference is pointed out by the solar reectivity: a common covering could have an albedo of 0.1, while the measured albedo for the green roof was 0.23. The second natural attenuation produced by the green roof is the solar

Fig. 5. Measured outside air and surface temperature for the summer session in 2003.

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267 Table 3 Stratigraphy of the common roof used in the comparison Element Thickness Conductivity Density Specic (cm) (W/(m K)) (kg/m3) heat (J/(kg K)) 0.38 0.06 0.17 0.7 1000 35 1200 1600 900 900 1470 880

1265

4.2. Winter session The length of the winter session is 60 days in which it rained about 250 mm, so the average soil relative humidity was very high. Despite the quite lower solar irradiance with respect to the summer sessions, the evapotranspiration ux was appreciable. In fact, the latent ux is also driven by the air vapour pressure decit (see next paragraph) that it is very high in winter. The direct consequence is a lowering in the surface temperature that rarely exceeded air temperature during the session (Fig. 7). The entity and the weight of each ux can be highlighted if a comparison with a traditional roof is done (Fig. 8). The relative signicant entity of the evapotranspiration has not to be misunderstood: the graph is normalized to 100 solar radiation units. Note that in winter the daily average solar irradiation was about 5 MJ/m2, while in summer the corresponding was about 18 MJ/m2. Anyway, the most signicant result regards the inside adduction: in winter mode the traditional roof has a better energetic behaviour, getting a slightly lower outgoing ux.

Concrete roof 30 Insulating layer 4 Water-proong sheet 1 Covering 3

Fig. 6. Comparison of the energetic exchanges of the dry or wet green roof with a traditional roof, starting from 100 incident solar irradiation units summer session.

5. The predictive numerical model All the previous discussions regarded the thermal and energetic behaviour of a green roof, getting the results from experimental data. Changing point of view, the evaluation of the same energy exchanges can be made beforehand with a predictive model. Starting from the meteorological variables that can be known statistically, and from the geometric and thermodynamic characteristics of a green roof, the model could estimate the evapotranspiration process and nally the inside adduction. Such information is useful to evaluate the thermal loads of the building and it has a fundamental importance to design a HVAC system.

absorption of the greenery. Without these processes the surface of the traditional roof can reach a temperature of 70 8C. That explains the very high adduction energy exchange for the traditional scenario in which, after all, 4.4 energy units enter the underneath room. Comparing that gure with the outgoing adduction energy for the wet green roof, it is immediate to realise that the entering ux is cancelled and even a slight outgoing ux is produced.

Fig. 7. Measured outside air and surface temperature for the winter session in 2004.

1266

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267

The model has been tuned calculating Kc, as the ratio of the measured value of ET and the calculated value of ET0, varying the relative humidity of the soil for the two summer sessions because of the well-water condition of 2002 and the stressed-water condition of 2003 (Fig. 9).

Fig. 8. Comparison of the energetic exchanges of the green roof with a traditional roof, starting from 100 incident solar irradiation unitswinter session.

There are different methods to evaluate the evapotranspiration process beforehand [11]. Here, an empirical method is used: it estimates the actual evapotranspiration ET starting from a reference value ET0: ET Kc ET0 (8) where Kc is an experimentally derived crop coefcient and ET0 is the maximum evapotranspiration relative to an actively growing and well-watered grass. That maximum can be estimated by a semi-empirical approach: DRn Ao gEA (9) Dg where D is the slope of the saturation vapour pressure versus temperature function, g the psychrometric constant, Rn + Ao is the available energy. EA takes into account the air vapour pressure decit: ET0 EA f ue e
*

(10)

e is the air saturation vapour pressure and e is the air partial vapour pressure. f(u) depends on wind speed u, according to the following relation: f u 0:261 0:54u (11)

Fig. 9. Crop coefcient varying soil relative humidity of well-watered condition (continuous line) and stressed-water condition (dotted line).

Fig. 10. Correlation between measured and calculated values of evapotranspiration for the three measurement sessions.

R.M. Lazzarin et al. / Energy and Buildings 37 (2005) 12601267

1267

To predict the energy exchanges of building with a green roof in dynamic state, the above-mentioned building simulation program has been used [10]. A dedicated module has been developed to simulate the green roof starting from its optical (LAI and ks), geometric and thermodynamic characteristics. ET is calculated by means of the previous relations and its contribution is introduced into the main building model as a ctive liminar coefcient that is added to convective and radiative ones: aET ET jta;o ts;o j (12)

where ta,o is the outside air temperature and ts,o is the soil surface temperature. The results of the simulation for the three measurement sessions are reported in the graphs of Fig. 10, where the measured and the calculated ET are compared. The correlation is good for the 2002 session when the greenery was in well-watered condition, while the correlation is bad in 2003 session when the soil was almost dry for all the summer. The correlation is good enough for 2004 winter session.

effect of evapotranspiration. During the winter the evapotranspiration process is driven above all by the air vapour pressure decit; its not negligible weight produced an outgoing thermal ux from the roof that is 40% higher than the corresponding one of a high solar absorbing and insulated roong. A predictive numerical model has been developed. It calculates the thermal and energy performances of a building with a green roof whose characteristics can be customised, varying the meteorological dataset for a specic geographic zone. The obtained correlation between measured and modelled evapotranspiration is satisfactory when the green roof is in well-watered conditions.

References
[1] H. Akbari, M. Pomerantz, H. Taha, Cool surfaces and shade trees to reduce energy use and improve air quality in urban area, Solar Energy 70 (2001) 295310. [2] M. Santamouris, N. Papanikolaou, I. Livada, I. Koronakis, C. Georgakis, A. Argiriou, D.N. Assimakopoulos, On the impact of urban climate on the energy consumption of building, Solar Energy 70 (2001) 201216. [3] R.M. Lazzarin, Il problema delle isole di calore nelle aree metropolitane (Heat island effect in the urban areas) in Il condizionamento dellaria (Air conditioning), Dario Flaccovio Editore, Palermo, 2003, ISBN 88-7758-499-8. [4] T. Takakura, S. Kitade, E. Goto, Cooling effect of greenery cover over a building, Energy and Buildings 31 (2000) 16. [5] S. Onmura, M. Matsumoto, S. Hokoi, Study on evaporative cooling effect of roof lawn garden, Energy and Buildings 33 (2001) 653666. [6] R.M. Lazzarin, A. DAscanio, A. Gasparella, Analysis of a green roof application to an industrial building, International Journal of Ambient Energy 24 (2003) 3543. [7] A. Meier, Strategic landscaping and air-conditioning savings: a literature review, Energy and Buildings 1516 (19901991) 479486. [8] E. Palomo Del Barrio, Analysis of the green roofs cooling potential in buildings, Energy and Buildings 27 (1998) 179193. [9] F. Castellotti, Studio sperimentale degli effetti energetici di un green roof sugli edici (Experimental analysis of the energy effects of a green roof on buildings), Ph.D. Thesis, Department of Management and Engineering, University of Padova, 2003. [10] TRNSYS: A Transient System Simulation Program, Solar Energy, Madison, 1997. [11] G. Rana, N. Katerji, Measurements and estimation of actual evapotranspiration in the eld under Mediterranean climate: a review, European Journal of Agronomy 13 (2000) 125153.

6. Conclusions The present paper describes the thermal and energetic behaviour of a green roof. The analysis begins from logged data in different measurement sessions, both in summer and in winter conditions. The physical system is dened by means of a nite differences model that calculates the evapotranspiration thermal ux as the residual term in the energy balance. The water uxes are calculated too. The role of the latent ux of the evapotranspiration is very important: during summer, with the soil in almost dry conditions the green roof allows an attenuation of the thermal gain entering the underneath room of about 60% with respect to a traditional roong with an insulating layer. That is due to the higher solar reection and absorption of the greenery, while the evapotranspiration is very limited. When the soil is in a wet condition, not only the entering ux is cancelled, but a slight outgoing ux is produced so that the green roof works as a passive cooler, thanks to the cooling

You might also like