Analysis of The Green Roofs Cooling Potential in Buildings
Analysis of The Green Roofs Cooling Potential in Buildings
Analysis of The Green Roofs Cooling Potential in Buildings
and Buildings
27 ( 1998)
179-193
lnformatiyue
et Syst&nrs
Ener&iques,
Eco!e
Nation& 77455
et Chausstes, Cede,
Pascal,
CitP Descartes-Champs-sur.Marne,
Received
29 April
1997; received
in revised form
13 May
1997
Abstract
This paper presents a mathematical model yielding a sensible, albeit simplified representation of the dynamic thermal behaviour of actual green roofs. Several parametric sensitivity analyses have been carried out to assess the cooling potential of green roofs in summer. The main conclusion of these analyses is that green roofs do not act as cooling devices but as insulation ones, reducing the heat flux through the roof. A relatively small set of parameters have been identified as relevant for green roof design: the leaf area index (LAI) and the foliage geometrical characteristics, the soil apparent density, its thickness, and its moisture content. 0 1998 Elsevier Science S.A.
Keywords: Green roofs; Cooling potential; Modelling
1. Introduction
Natural cooling techniqueshave been used over the centuries, but their application has been very scarce in recent decades. The introduction of mechanicalair conditioning systems into the building, with their great ene:-gyexpenditure, hasbecomethe standardalternative usedto natural cooling. A wide variety of natural cooling techniquesrelieson alternative roof designs. However, it can be observed that the most common roof constructionsin south E.uropeare inadequate during the warmestdays of the year. Sincethe number of studiesregardingthis matter are scarce.the europeancommunity hasrecently engageda research project (ROOFSOL: Roof Solutions for natural cooling, [ I] ) focused on the theoretical and experimental analysisof different roof solutions for cooling in the mediterranean region, mainly basedon evaporative and radiative cooling principle <. Green roofs constitute one of the solutionsto be analysed in ROOFSOL. Vegetation would protect ths roof from direct solar radiation and could cool it through enhancedevaporation. This roof solution is furthermore compatible with the functional, aesthetic and ecological criteI,ia applied to the designof the surroundings. The objective of this paper is two fold: ta proposea mathematical model yielding a sensible,albeit simplified representation of the dynamic thermal behaviour of real green
* E-mail: [email protected] 0 1998 Elsevier Science
roofs, and to analyse their potential as cooling devices in summertime. The paper is organized asfollows: first, the componentsof a green roof are identified (Section 2). In Sections3-5, the models describing, as rigorously as possible, the thermal behaviour of each component are presented.The green roof model is then obtained (Section 6) by coupling the models of its components. Finally, several parametric sensitivity analysesare performed in Section 7 in order to identify key parameters an efficient green roof design. for 2. Green roof description Three componentscan be distinguishedin a green roof: the canopy (leaf cover), the soil and the structural support (see Fig. 1).
OUlD0OR CONDfTIONS
l*DOOR
CONDITIONS
0378.7788/98/$19,CKl
PIISO378-7788(97)00029-7
180
E. Palomo
Del
Barrio/Energy
and
Buildings
27 (199c()
179-193
Indoor conditions are imposed by the building. They are usually defined by the thermal state of the indoor air and the temperature of the surfaces seeing the roof bottom. The quantities defining the outdoor conditions are the solar radiation flux, the thermal radiation flux coming from the sky, the temperature and moisture content of the air, the wind speed and the wind direction. The roof is supposed to be large enough to assume horizontal homogeneity. Heat and mass fluxes are assumed to be mainly vertical, so that one-dimensional models can be used to describe the thermal behaviour of the roof components.
3. Roof support model The roof support is supposed to be an homogeneous layer of a solid material with constant thermophysical properties. The energy balance equation governing the evolution of its thermal state is, in one dimension:
PCP
aTs(z>t) -= at
A a2T,(z,t) s at2
where T,(z,t) representsthe temperatire field, p is the density, cp the specific heat, and h, the thermal conductivity of the material. The following boundary conditions, decoupling the supportfrom the rest of the roof (see Section 6 for coupling), are assumed:
The forma1theory of thesephenomenawas first statedby Philip andde Vries [ 21, de Vries [ 31, andLuikov [ 41. Since then, a large amount of theoretical and experimental work hasbeen performed. The porous medium is assumed equivalent to a fictitious continuum, and the macroscopicbalance equationsare obtainedby averaging microscopicbalanceand transfer equations (at the pore scale) over a representative volume [ 51. To warrant the validity of suchan approach,the size of the representative volume should be considered as controlled by the size of the moisture content heterogenities and should be viewed asrelatively quite large to the typical size of pore or grain. Under the following hypothesis: thermal and moisture content gradients in horizontal directions are assumedto be zero (one dimensional model) ; the soil matrix is assumed equivalent to an homogenous and isotropic continuum. Its properties do not explicitly dependon the direction; the liquid and vapour phases always in equilibrium; are pores are assumedsmall. The total pressure is then regardedasa constantand the net air transfer is assumed to be zero. The macroscopicbalanceequationscan be written as:
[h(w,T) +no,,(co,T),~
1. >
>
aw(Z,t) -=at
where z = 0 meansthe top of the support layer and z = L the bottom in contact with the indoor air. L is the thicknessof the support layer, T,, the indoor air temperature, ~spport,,~rp the imposedtemperature at the top surface,and T~upport,bottom the temperatureof the support at its bottom surface.
dKtz,t)
a:
qJO(z,t)
4. Soil model A soil is a porous medium in which three phases can be distinguished: the solid matrix (minerals and organic material) , the liquid phase(water) and the gaseous one (air and water vapour) Qualitatively, we can say that in unsaturated soils the heat will be transported in these three phases. The respective dominant mechanisms be: conduction in solwill ids and liquid phases,convection in liquid and gaseous phases, and latent heat transfer by vapour diffusion in pores. Furthermore, the heat transfer will dependon the water content and on the temperature. This leacs to a mutual dependence and continuous redistribution af heat and moisture which is quantitatively described by a set of coupled nonlinear equations.
local temperaturein the porousmedium domain, (C) local volumetric moisturecontent in the porous mediumdomain, ( - ) weighted heat capacity, (J kg- K- ) effective thermal conductivity, (W m- K-) latent heat of vaporization, (J kg- ) non-isothermalvapour diffusivity coefficient, (kg m* s- K- ) isothermalvapour diffusivity coefficient, (kgms-) isothermalmass(vapour + liquid) diffusivity coefficient, (m* s- ) non-isothermalmass(vapour + liquid) diffusivity coefficient, ( m2s- K- ) hydraulic conductivity, (m s- ) water sink representingthe root extraction term, (s-l)
and Buildings
27 (1998)
179-193
181
As a first approach to the problem of assessing the potential cooling role of green roofs, we will assume constant (vs. time and space) moisture profiles within the soil. Accordingly, Eq. (3) becomes:
(7) where D is the diffusion coefficient of water vapour in air (m2 s- ), A is the latent heat of vaporization (J kg-), R, the gas constant of water vapour (J kg- K- ) , T the absolute temperature ( K), P the total pressure (Pa) and p\ the partial vapour pressure (Pa). The latter one is given by the following thermodynamic relation: P,, =psw s* R,~ (8)
=i
[h(o,T)
+AD,,&.T)+j-
dT(z,t) i >
(4)
All the coefficients in these equations are dependent variables of the moisture content and the temperature. They must be specifically determined for each soil. However, somefunctional relations between these coefficients and the soil temperature and moisture can be proposed considering that first, we are not interested in a particular soil, but in soils in general, and second, that curves and table representations of coefficients dependencies is not the better way of parametrizing a model for sensitivity analyses. The following correlation model for estimating soil thermal conductivity was proposed in Ref. [ 61:
with ps the saturated water pressure (Pa), g the acceleration due to gravity (m s-~), and $ the total water potential in equilibrium with vapour (m). The vapour diffusion in a porous medium is obviously slower than in free air. Many authors agree in expressing the coefficient D in Eq. (7) as a linear function of the medium porosity ( % of total volume occupied by the soil particles) : D=ccDo E (9)
where p is the soil apparent density in (kg m3), w the volumetric moisture content, A the thermal conductivity in (J m- S - K-), and pep the thermal capacity in l:J me3 K-). A large sample of soils, with apparent density lialues from 1100 to 1500 (kg m-) and volumetric moisture contents from 4 to 25%, has been used in Ref. [6] to obtain the previous correlation. h values calculated with this formula involved a relative error lower than 7% in most cases. To be coherent with the correlation mod21 (Eq. (5) ), the following formula, proposed by the same author, will be used for estimating the soil thermal capacity. As in the previous case, variables must be given in the InternatisDnal Unit System. pc,(w) =4180(0.2+o)p (6)
where D, is the vapour diffusion coefficient in free air (=0.611X10-4m2s- , Int. Crit. Tables), E the soil porosity, and cx a constant value: 0.58-0.67 (e.g., [ 71). The density and porosity of disperse bodies, including soil, is an index of the mutual arrangement of the particles, their mutual proximity, and the degree of contact. For ideal systems of balls, independently of the systems of packing (cubic, hexagonal, tetrahedral. n-stage hexagonal,. . .), the relationship between the density and porosity is expressed by the following formula (e.g., Ref. [ 61): E= 1-p 100
1 PS
(10)
It can be demonstrated (see Appendix .4) that under the following hypothesis: 1. the air and water vapour follow the ideal gas law; 2. the net transfer of air in the soil is negligible; 3. pores are small enough for the total pressure to be regarded as constant; 4. the soil moisture content is always above the wilting point and the total water potential in equilibrium with vapour is then greater than - lOJ.2 cm. The non-isothermal soil vapour diffus vity DvT (Kg m s- K-l), can be written as
The wilting point is an extremely important parameter of soil. The store of water up to such a moisture content is absolutely .mavailable to plants.
where p represents the apparent density (mass of the particles referred to the total volume of the disperse body, including pores), and pI the so-called specific gravity of particles. This latter is defined as the product of the true density (mass of the dry particles referred to the volume occupied by the solid phase) by the acceleration of gravity. An usual value of ps for soils is 2700 kg rne3. According to Vershinin et al. [6], it may be generally stated that the entire variety of pores and densities occurring in natural soils can be explained by a certain ideal packing of soil particles. If the apparent density of soil and the specific gravity of its constituent particles are known, the total porosity of the system can be then estimated by Eq. ( 10). As we said before, the store of water up to the wilting point is absolutely unavailable to plants. On the other hand. moisture contents above the soil fieldcapacity (soil capillarywater capacity) may causes root damages (asphyxia). We will then assume that soil moisture content is always between these two extremes. To estimate the total water potential in equilibrium with vapour. 4, the following linear functional relation will then be used:
182
E. .Dnlomo
Del
Barrio/Energy
and Buildings
2 7 (I 99X)
179-I
93
wwP is the volumetric moisture content at the wilting point, where $= I,&,= - 10e4 I5 cm, and o>, is the volumetric moisture content at field capacity, with $I= I+!+~ - lo-* = cm. The model describing the soil thermal behaviour is then defined by Eqs. (4)-( 11), with boundary conditions of Dirichlet type:
the with TsOI~.tOP Tsoi~.bottom temperatures imposed at the and top and bottom of the soil layer. As in the previous case (Section 3)) assuming Dirichlet boundary conditions the soil behaviour is decoupled from the other roof components. Coupling the component models is treated in Section 6. Model input parameters are the thickness of the soil layer (L), the soil apparent density (p), the soil moisture content (w), the soil wilting point ( oWP) and the soil field capacity (w+,).
5. Canopy model The canopy is composed by the leaves and the air within the leaf cover. The main processes ccntributing to the definition of the canopy thermal state are 1: Fig. 2) : see I. solar radiation absorbed by the lealfes; 2. longwave radiative exchanges (TIR) between the leaves and the sky, the leaves and the ground surface, and among the leaves themselves; 3. convective heat transfer between the leaves and the canopy air, and between the ground surface and the canopy air; 4. evapotranspiration in leafs. This process includes three phenomena: water evaporation inside the leaves (stomata1 cavity), vapour diffusion to the leaves surface, and convective vapour transport from the leaves surface to the air: of wate: vapour in the soil sur5. evaporation/condensation face and vapour convective transfer between the soil surface and the air;
HEAT d VAKWR
TRANSFFR
transfer
processes in a canopy.
6. convective heat and vapour transfer between the air within the canopy and the free air. The complexity of a canopy as a system of sources and sinks of heat and mass is such that an exact description of its physical behaviour is almost impossible. While attempting to figure out a simpler representation (model) of a canopy, one is facing two types of problems. The first one is the inherent spatial complexity and dishomogeneity of the foliage. This implies that, for an accurate description. the number of simultaneous equations to be solved could be five times higher than the number of leaves in the canopy. The second one is the turbulent nature of the air stream within (and above) a canopy. Its consequence is that the direction and magnitude of the fluxes of energy and mass vary at any moment and cannot be exactly predicted. In spite of this. in much of the literature concerned with the coupling of plants with their environment, heat and mass transfer to and from a canopy are described as vertical fluxes along a concentration gradient. across some typical resistances. However. the assumption that transfer takes place along a vertical direction only, implies an averaging out of the variations along an horizontal plane. On the other hand. the known empirical relations between fluxes and gradients warrant the validity of this approach only as far as time averages of fluxes and gradients are concerned. The conditions for this approach to yield a sensible, albeit simplified representation of a real canopy, have been largely investigated. One-dimensional models of a canopy may be expected to be representative only on a horizontal scale large enough to be regarded as homogeneous, i.e., much larger than the typical dimension of any foliage element as leaves, branches, interrow spacing or row width. There is another isotropy problem, however. No natural canopy may be regarded as homogeneous in the vertical direction, vertical gradients of the functions of state cannot be neglected. It is normally attempted to reproduce the profiles of temperature and humidity within a canopy by representing the latter as a stack of semitransparent, semipermeable homogeneous layers. The number of layers have been described as anything between one (e.g., [ 81) and a continuum which is discretized by a numerical procedure [ 91. Obviously, only multilayer models are able to reproduce temperature profiles and vapour content within the canopy. It is clear however, that the profiles themselves are not useful information for the application we are interested in (cooling potential of green roofs), and one could be satisfied with mean values. In this paper therefore, a canopy will be regarded as one homogeneous layer, characterized by one value of leaves temperatures and one value of temperature and vapour content in the air within (see Fig. 3). Such a layer is bounded by the ground surface at the bottom. and an ideal surface at the top, which is in turn, homogeneous. Under the previous assumptions, the macroscopic or t<mean>>energy and vapour balance equations in a canopy are:
E. Palomo
Del Barrio
/Energy
and Buildings
27 (1998)
I79-193
183
5.1. Net thermal radiation The longwave transmittance, 7,(LAI), of a canopy having leaf area index LA1 is defined as the quotient between the longwave radiative flux entering either the upperor the lower surface of a section of the canopy and the tlux leaving the other end of that section. It is also <<called transmittanceof a canopy of black leaves, since, in the longwave range, transmittance and reflectance of the leaf tissueare negligible. The canopy will transmit only the radiation which is not even once intercepted by a leaf. It can be shown [ 101 that for diffuse longwave radiation, T,(LAI) is affected only by the geometricalproperties of the canopy. Indeed. under some assumptions,the theoretical functions for extinction of radiation in a turbid mediummay be applied to a canopy (e.g., [ 111). to yield: 7,(LAI) = exp( - k, LAI) (14)
Fig. 3. Sketch ofthe single-layerrepresentation ofa cano3y. radiation flux; TIR = longwave radiation exchanges.
SOLAR
= Solar
leaves temperature (average in 1he control volume), (K) air temperature (average in the control volume), (K) air specific humidity (average in the control volume), (kg kg-) leaves specific thermal capacity, (J m p-3 K- ) average leaves thickness, (m) air specific thermal capacity, (J m-j K- ) air density, (kg mw3) canopy layer thickness. (m) solar radiation absorbed by the leaves. (W
m-)
with k,, the extinction coefficient for longwave radiation, that can be analytically calculated for someidealized leaf angle distributions. Some values of it, deducedfrom the literature (see e.g., [ 121)) are suppliedin Table 1. A canopy absorbs fraction equalto ( 1- T,) the longwave a radiation it receives. The net thermal radiation flux in a canopy is then written as:
%d,TIR = ( 1 71) [&k, + 07.: -
2fl7-3
(15)
where TIky is the sky temperatureand 7, the temperatureof the ground surface. 5.2. Net solar radiation The shortwave radiation transmitted by a canopy is the sum of the unintercepted radiation and the radiation that is either transmitted or reflected downwards (or both) by any leaf within the canopy. The shortwave transmittanceT,(LAI) for diffuse shortwave radiation of a canopy with a LAI, can be represented[ 131, with reasonable accuracy, by an exponential law: TV exp( - k,LAI) = (16)
net thermal radiation flux on le:%ves, me2) (W sensibleheat flux between the foliage and the canopy air, ( W m-) energy flux due to leavestranspiration, (W me*) sensibleheat flux between the canopy air and the foliage ~~~~~~~~~~~(P~~,,~.~~), me) = (W sensible heat flux between the canopy air and the ground surface, ( W m -) sensibleheat flux between the canopy air and the outdoor air, (W m-> vapour flux between the canopy air and the foliage, (kg m-) vapour flux betweenthe canopy air and the ground surface, (kg mP2) vapour flux betweenthe canopy air and the outdoor air, (kg m-*) The equationsbelow are written per unit of ground area. The total leaf area (one face only) contailed in a volume of unit baseis namedLeaf Area Index (LAI) .
The coefficient of extinction k, has to be a function of the optical propertiesof the leaves,and in Ref. I 141is calculated through:
Table 1 Values of the extinction coefficient for longwave radiation. the literature, for idealized leaf angle distributions Leafs distribution Horizontal Conical (u = 45) Vertical (a = 90) Spherical The angle between symbol LY. k, 1: I.05 0.829 0.436 0.684; 0.8 I the leaves and a horizontal plane is represented by the as deduced from
184
/Energy
and Buildings
27 (1998)
179-l
93
k.=[(l-r)*s
Pt]k,
(17)
with TV pt the transmittance and reflectance of the leaf tissue, et respectively. For a <<mean)>green leaf. the equation before results in k, = 0.74 k, [ 151.For species with mainly horizontal leaveswe have k, g 1.10,and for species with mainly vertical leaveswe get k, z 0.29. The reflectance of a canopy is always smallerthan the one of the leavescomposingit. The mutual shadingof leavesand the multiple scattering within the canopy result in a sort of <(cavity>> effect, which causesan additional absorption of radiation. It will be assumed[ 121 tha. the canopy behaves as a densecanopy (stand of unitary interception for diffuse radiation, or completely covering the soil) in respect to the fraction of the incident radiation it intercepts. Hence: p,(LAI) = (1 - r,(LAI))pz (18)
where 1 is the leaves characteristic length and u the wind speed.a, b, m and n are empirical coefficients (a = 1174, b=207,m=OS,n=0.25fortomatocrops). Using Eq. (21)) Stahghellini [ 161 showsthat for growing wind speeds dimension,the consequence anerror in the or of estimation of the external resistanceshrinks. For instance, variations of wind speedabove say 0.2 m s- , have small effect on the external resistance,whereasit may matter a lot for air flow in the range of 0.05 or 0.1 m s - .
5.4. Transpirationjux
where px is the reflectance of a densecanopy. A canopy absorbsa fraction equal to 1 - rs-- ps of the shortwave radiation it receives. The net solar radiation flux absorbedby a canopy is then written as: (Prad.sol=[1-T\-(1-~s)prl(1+7,P~)~s (19)
If vapour pressure chosenasan appropriateforcing funcis tion, the energy flux consumed let water evaporatein leaves to may be representedby a law analogousto the one usedin Section 5.3:
(~trans.pa 2 LA1-y(ryJri)(eppe) =-
(22)
where cps represents solar radiation at the top of the canthe opy, and ~,p~q~ solar radiation reflected by the ground. the
5.3. Convective heat transfer between leaves and air
The transfer of sensibleheat between the foliage and the air within the canopy will be represented the Newton law by of convection: 2
conv,p-a= - (Pconv,a-p - ZLAI?( = cp
Tp- T,)
where r, indicates a meancanopy resistance sensibleheat to transfer, or <<canopy external resistanceawhich is, in fact, defined by Eq. (20). The experimental data available in the literature are not conclusive about a unique equation to predict external resistance of leaves exposedto low windspeeds,in the presence of external sources of turbulence, and being only a few degreeswarmer or cooler than the surrounding air. We will adopt here the correlation model proposed by Ref. [ 161, which is basedon an exhaustive experimental work:
al re=(llTp-Tal +bu2)(Sm -1 >
where y is the thermodynamic psychometric constant (Pa K- ), ep and e, (Pa) are the vapour pressureat the leaf surface and at the canopy air, respectively, peis the canopy external resistance(s m- ), defined by Eq. (20), and pi is the so-calledinternal resistance vapour transfer of the canto opy, or bulk stomata1 resistance,which is. in fact, defined by Eq. (22). The internal resistance a canopy is known to be affected of by a number of physiological and environmental parameters. Only the latter group will be considered here. Among the environmental parameters, shortwave irradiance ( qso,) appearsto be the most important one. Leaf to air vapour pressuredifference ( ep- e,), leaf surface temperature( T,) and COz concentration of the air are known to play also a significant role. A very interesting discussionabout the physical meaning of the internal resistanceaswell asabout its functional relations with each driving variable, can be found in Refs. [ 16181. Stahghellini [ 161 showsthat the apparentbehaviour of the internal resistance a canopy is almostexclusively deterof mined by the internal resistanceof the leaves composingit. Accordingly, the observed behaviour of the internal canopy resistancecan be representedby a phenomenologicalmodel quite similar to the one suggested the internal resistance for of a leaf: (23) where rmin (s m- ) is the minimum possible resistance, whosemagnitudehasa purely physiological origin. The symbols f, representdimensionless functions larger than unity, quantifying the relative increase of the internal resistance, whenever oneof the parameters limiting the rate of transfer is of water vapour. The following function was proposed to representsucheffects:
(21)
* The heat flux density, cp(W mm*), througn a finite surface in contact with a fluid is assumed to be proportional to the difference of the internal energy concentration, pcPr (J m-l), between de surface and the fluid: constant, K( m s ), is called cp= KPC~ (Twrtae - TAuld). The proportionality <<bulk heat transfer coefficient~r, which is related to the heat transfer resistance, r (s m- ), by: K = I ir. The commonly ttsed convective coefficients, h (W mm2 Km ). whichare alsocalled <<bulkheat transmissioncoefficientsn, are then defined by: h = tcpc, = pc,,ir.
E. Palomo Table 2 Values of the canopy model parameters Model General Optical External Internal resistance resistance p,=o.12 a=1174 b=207 r,,, = 82 T,,, = 24.5 c, =4.3 cz = 0.54 h,=2 parameters
Del Barrio/Energy
and Buildings
27 (1998)
179-I 93
185
adopted Freezed
analysis Free parameters LA1 = 2-5 L=O.2-1 m k,=O.4-1.0 k, = 0.74 k, I= 0.05-0.30 m .f= 0.5-2.0
Air exchanges
R= 125-1000
ach
where y is the thermodynamic psychometric constant (Pa K- ) and A the latent heat of vaporization (J kg- ). f,(T,)=l-tC,(T,-T,)2 Ti(C02) = 1 +c,(c0,-200)* (24) 5.6. Heat and vapour transport between the air within the canopy and outdoor air Heat and vapour fluxes caused by mass transfer between the canopy air and the air outsidethe canopy boundary layer, will be describedby Pco,,,v,a--% = -h,,(T,-T,)[W me2]; h,,=R Lpc, [W m-K-1
and
c
r, =f
Fi(e,--e,)
= 1 +Cs(e,-e,)
For a tomato crop, the following values are proposed in [16]:p,,,,,=82(sm-),C,=4.3,C,=0.54,T,=24.5(C), C3 = 2.3e - 2, CI = 6.1 e - 7 and Cs = 4.3. In Section 7, for sensitivity analysis purposes, the internal resistance of a canopy will be represented by:
ri,tomak,
(29)
(2-5)
the coefficientfallowing us to represent a canopy evaporating more cf< 1) or less (j> 1) than a tomato c!-op. 5.5. Heat and vapourjw the air between the ground surface and &,,=-&h,,[K rnp2 s-r Pa-] (30)
The transfer of sensible heat between the ground surface and the air within the canopy will be represented by the Newton law of convection: Pconv,g-a - %I\ ,a-g= -h,(T,-Ta) = (26)
where T, is the temperature of the air canopy, Tg the temperature of the soil surface, and II, (W m- K-) the bulk convective coefficient of heat transport. In a similar way, the transfer of water vapour between the ground surface and the air within the canopy will be represented by: (Pvapg-a -4Tapg= = -I;,(e,-4 (27)
respectively, with T, the outdoor air temperatureand e, the outdoor air vapour pressure.L is the canopy height and R represents air exchangerate ( s- ) This approachis quite the similar to the one usedfor describingheat flux by ventilation in buildings. The canopy modelis then defined by Eqs. ( 13)-( 30). The model inputs are solar radiation flux ( cp,), sky temperature (T,,,), ground surface temperature (T,), dry bulb outdoor temperature( TX), andoutdoor air vapour pressure(e,) . The total numberof parametersin the model is 21 (seeTable 2).
6. Green roof model The green roof model is composedby the support model describedin Section 3, the soilmodel presented Section 4, in and the canopy model in Section 5, together with the coupling models which will be here described. Such models representthe real boundary conditions at the canopy-soil and canopy-support interfaces, satisfying the physical constraint of continuity for the statesvariables and the flux densities.
where e, is the vapour pressure of the canopy air (Pa), eF the vapour pressure at the soil surface (Pa), and &( kg s- m- Pa-) the bulk convective coefficient of v apour transport. Since heat as well as vapour transport is convective and not diffusive (Lewis number close to 1) . we can assume (28)
186
E. Palomo
Del
Barrio
/ Enerp
and
Buildings
27 (1998)
179-193
6. I. Canopy-soil
with the partial water pressurep, calculated by Eq. (8) at the temperature and moisture content of tl-.e soil top surface. Continuity of the heat flux density at the canopy-soil interface. implies
(h+AD,*pp :=o = (1 -Pg):;R
This equation represents the energy balance at the soil top surface, when no energy storage is assumed. Continuity of the vapour flux density at the canopy-soil interface, implies
- &-
(33)
two interfaces (canopy-soil and soil-support models) where the relations between two connectedmodulesare described. Each module hasfrontiers to communicatewith the outside. A frontier is a set of some input-output variables (e.g., temperatureand heatflux at the top surface of the supportmodel). A detailed description of the adoptedmodeling approachcan be found in Refs. [ 20,211. The resolution method is basedon the assumptionthat a global solution of the problem may be achieved by a set of local resolutions and relations traducing the constraints expressedon the coupling interfaces [ 211. Each interface sends constraintsto the modulesto be coupled, which intend to reach a state satisfying theseconstraints. In particular, a Newton iteration algorithm hasbeenusedto find the interface state variables which satisfy the constraintsof flux densities continuity of the connected modules.Differential equations inside the modules (domain state variables updating) are integratedusing a method belonging to the generalfamily of BDF (Backward Differentiation Formulas) methods. specially adaptedfor solving stiff problemsof ordinary differential equations [ 221. The finite volumes method has been usedto translate continuous models (soil and support models) into spacediscretemodels.
4.2. Soil-support
7. Sensitivity analysis interface we can write 7.1. Climatic duta A sequenceof 10 days of hourly meteorological data in Athens, from 1/OS/ 1982 to 10/08/1982, hasbeen used for the sensitivity analysis. Global solar radiation data are represented in Fig. 4, outdoor temperature measurements in Fig. 5, relative humidity in Fig. 6. and wind speedin Fig. 7. The sky temperature has been estimated using the model proposedby Berdhal and Fromberg [ 231. 7.2. Sensitivity analysison the canopy The values of the canopy model parametersthat we are using for the sensitivity analysesare listed in Table 2. They
- , h + 1D,,)-
dT(
;.f)
=-A-.~,
iK(z,t)
ii; --I,
Continuity
of heat flux
a.-
(34) The roof support is supposed be INaterproof and imperto meableto water vapour. The greenroof modelhasbeentramlated underan <<objectoriented>> form, and implemented in MATLAB [ 191. It is composedby three modules(canopy. soil and support models) limited by Dirichlet boundary conditions on their frontiers, where temperature (or moisture content) is fixed, and
on an horizontal
wrface.
Athens
(Greece).
August,
1982.
Tern
36 ,per i4
20 0 30
100
I inl
c (h)
I50
200
2 50
Fig. 5. Outdoor
dry hulh
tempcr;riurc.
Athens
(Greece).
August,
1982.
9 Wind Spee 8
150
200
250
August,
19X7.
have been gathered in five categories (first column in the table) depending on the parameter purpose. Parameters of a same group have been split in freezed and free parameters. The free ones are those that group a class of parameters (c.p., parameterjof evapotranspiration), or those having a greater influence on a particular process (e.g.. extinction coefficients
have a greater effect on canopy solar transmittance than reflectances), or those beset with a high degree of uncertainty (e.g.. parameter R). For a 1 m height canopy, an K value of 1000 ach implies a canopy air behaviour close to the one of the outdoor air. that is a very well ventilated canopy. On the contrary, H = 125
188
E. Palmno
Del
Barrio
/Energy
and Buildings
27 (1998)
179-I
93
20
40
60
80
100
120
140
(I
20
40
60 Time(h)
80
100
120
140
with
Fig. I I. Effect of the leaf area index on the solar transmittance, X-,=0.67,f=1.L=1m,R=lOOOach.and/=0.10m.
with k, = 0.48,
:(,
2
20 40 60 Time
(h)
X0
lh-==----d 0
20
40
60 Time
80
140
(h)
ach means that the air speed inside the canopy is lower than 0.03 m s-, which is the typical range of air velocity in greenhouses without forced ventilation. Taken values of the R parameter between 125 and 1000 ach allows us to explore, in terms of sensitivity analysis, extrerne situations.
During simulations, the temperature of the soil surface is assumed to be constant, 7g= 20C. as well as its water vapour pressure (relative humidity of 90% in pores). The results achieved regarding the effect of different parameters on the canopy solar radiation transmission, the foliage temperature and the hydrothermal state of the air within the canopy, are as follows. Effect of the coefficients of extinction, k, and k,: These parametersmainly affect the transmissionof solar radiation by the canopy. The amountof solar radiation incoming to the soil is drastically reduced when k, moves from 0.30 to 0.74, and the foliage temperature slightly increases(see Figs. 8 and 9). The effect of theseparameters the canopy air state on is negligible. Effect of LAI: The behaviour of this parameter is quite similar to the one of extinction coefficients. Increasing LA1 from 2 to 5 implies a strong reduction of the canopy solar transmittance as well as a rise of the foliage temperature during the day (seeFigs. 10and 11) The effect of this parameter on the canopy air state is alsonegligible. Effect of the parameterfregulating the canopy transpiration rate: This parameterhas no effect on the solarradiation transmissionprocess.It essentially modifies the foliage temperature during the day. When f grows from 0.5 to 2, the foliage temperature increasesof about 4C at noon (see Fig. 12). The influence of this parameteron the air state is negligible, maybebecause very high value for the parameter a R hasbeenused. Effect of the air exchangesrate. R: The foliage temperature is very sensitive to this parameter.Differences of about 34C can be observed, day and night, when increasingR from 125achto 1000sch.The foliage temperatureincreasesduring the day hours and diminish during the night (see Fig. 13). After sunrise,the temperature of the air within the canopy diminish when R decreases(see Fig. 14). and the vapour
We are only looking balance in the soil. at the quantities than could modify the energy
E. Palomo
34 Foliage Temperature 32
Del
Barrio/Energy
und Buildings
27 (19%)
179-193
189
(C)
16
0
I
20 40 60 Tune (h) SO 100 I20 140 0 20 40 60 Time (h) 80 100 120 140
Fig. 12. Effect of the parameterfon the foliage temperature, k,=0.67,LAI=5,1,=1 m,R=1000ach,andI=O.IOm.
with k,=O.48,
Fig. 15. Effect of the parameter R on the water vapour pressure in the air within the canopy, with k,=0.48. k, = 0.67, LA1 = 5, L= 1 m, f= 1. and l=O.lO m.
32
26
Effect of the canopy height. L: This parameterhasan influenceon the air temperatureand humidity contentquite similar to the one of R. The air vapour concentrationincreases during the day when L decreases, and the air temperaturediminish during the night. The effect on foliage temperature is negligible. Effect of the leaves dimension, 1: No perceptible changes in the behaviour of foliage and air temperatures,or of the air vapour concentration, have been observed when varying I from 0.05 to 0.20 m. I
20 40 60 Time (h) 80 ICI0 120 140
16 0
7.3. Sensitivity
R on the foliage
temperature,
with k, = 0.48,
34 Air Tel 32
I I-
A 10 cm layer of concrete is chosenfor the roof support. The correspondingthermal parameters are: Density 2500 (kg mP3) Specific heat 820 (J kg-) Thermal conductivity 1.75 (W rn.- K-) Layer thickness 0.10 (m) Convective coefficient h 3 (W m- KP) The free parametersof the canopy model (see Fig. 2) are freezed at the following values: k, = 0.48, k, = 0.67, L = I m, R = 1000 or 125 sch. LA1 = 5, andf= 0.5, 1 or 2. The following are the adopted values for the soil model parameters. 1100or 1500 (kg rne3> Apparent density (p) 0.1,0.2 or 0.3 (m) Thickness (L) Volumetric moisture 20or40 (lo) content (~1) 20or40 (70) Field capacity ( wfc) lOor (9%) Wilting point (w,,)
30
28
26
24
20
60 Tune (h)
80
1Ull
120
140
Fig.
14. Effect
k, = 0.48,
of the parameter R on the canopy air temperature. k, = 0.67, LA1 = 5. L = I m.f= I, and I = 0.10 m.
with
concentration increases significantly during the day hours (see Fig. 15) . As the previous one, this parameter doesnot modify the solartransmittanceof the canopy.
The air temperature in the building is assumed constant and equal to 25C. The effect of the thickness of the soil on the heat flux through the roof is shown in Fig. 16. The heat flux entering
190
E. Palomo
Del
Barrio/Energy
and Buildings
27 (I 998)
179-193
] _ .I -2
\:
so
I50
200
250
Fig. 16. Effect of the thickness of the soil layer (L) on the heat flux through the roof. p= I, w=40%, wuP =20%. 0,,=40%. LAI=5.f= I.
50
IS0
200
250
Fig. 17. Effect of the soil apparent density flux through the roof. w=40%, w,,=20%,
into the building decreases when such a parameter increases. An augmentation of the inertia level is added to the rise of the insulation level. The heat flux through a roof having a light soil ( p= 1100 kg m-) is compared with the one of a roof with a heavy soil (p= 1500 kg m-) in Fig. 17. It can be seen that, in opposite direction, the role of the apparent density is quite similar to the one of thickness. When density decreases, thermal conductivity diminish and the heat flux through the roof is reduced. The effect of the soil moisture content is shown in Fig. 18, for a light and a heavy soil. It must be noticed that the soil thermal conductivity and the capacity increase with the soil moisture content so that the thermal diffusivity diminish. This is the main reason why when increasing the moisture content in the soil the heat flux through the roof is reduced. The augmentation of the evaporation at the soil surface is also a contributing factor. When freezing the soil moisture content, there is no influence of the soil wilting point and field capacity on the heat flux through the roof (see Fig. 19). On the other hand, the effect of the crop transpiration ability (parametera is negligible (see Fig. 20) as well as the one of the parameter R, which representsthe air exchange rate between the canopy andthe outdoor air. We can seein Fig. 21 that while increasing R from 125 ach to 1000 ach, the heat flux through the roof decreases only by 0.5 W m . In Fig. 22>the behaviour of the heat flux through the roof for two extreme casesis shown. The first one represents conditions againstcooling: a high soil density (p = 1500kg rn-~3), a low soil moisture content ( w = 20%), an intermediate value of LA1 (LA1 = 3). and a factor f= 2 increasing the canopy internal resistance vapour transfer. On the conto trary. the secondone is selectedto promote cooling: a light soil (r= 1100 kg m-) with 40% of moisturecontent, and a
,v
-I ' 0
50
IS1
200
I 250
Fig. 18. Effect of the soil volumetric moisture content (Win the figure) on the heat flux through the roof. L = 0.2 m, tiwr = 20%. wtC = 40S, LA1 = 5, f= 1.
Fig. 19. Effect of the soil wilting point and field capacity (W,.+ and W,, in the figure) on the heat flux through the roof. w = 403, L = 0.20 m, LA1 = 5,
f= I.
E. Palomo
4 Iled Flux 35
Del Burrio/
Energy
and Build&s
27 (1998)
179~-193
191
-rho-l -rho=1
Fig. 20. Effect of the canopy transpiration (paramete1.f) on the heat flux through the roof. w = 40%, w,, = 2076, utc = 40%. LA1 = 5, L = 0.20 m.
Fig. 22. Heat flux through two roofs defining almost extreme cooling. wwP =20%, cl& =40%. L=O.?O m.
conditions
for
Flux (Wm?)
-05 1 0
I
20 40 60 80 llme (h) 103 120 140
Fig. 21. Effect of the parameter R on the heat flux throL.gh the roof. r= 1100 kgm-. w=40%, wNp= 20%, qL = 40%. LA1 = 5, L == 0.20 m.
very well developed canopy (LA1 = 5). We can see that, in both cases, the green roof does not cool the building. It only behaves as a very good insulation device.
8. Conclusions A mathematical model yielding a sensible, albeit simplified representation of the dynamic thermal behaviour of real green roofs has been proposed and used for parametric sensitivity analyses. It has been shown that a well designed and managed green roof could behaves as a high quality insulation device in summer, reducing the heat flux through tne roof. The main design parameters have also been identified. The LAI, that represents the total leaves surface contained in a volume of unit base, and the leaf argle distributions, which determine the short and longwave transmittance of the
canopy. Both parameters determine the role of the canopy as a shadowing device. The thickness of the soil layer, its apparent density, and its moisture content determine the soil thermal diffusivity. It increases with the apparent density, and decreases with the soil moisture content. The canopy evapotranspiration and the air exchanges rate between the canopy and the free air, which have an important role when looking at the hydrothermal state of the canopy, have a secondary effect on the heat flux through the green roof. Designing green roof for reducing heat flux through the roof in summer means: 1. selecting plants with a large foliage development and/or with a mainly horizontal leaf distribution, in order to warrant a low solar radiation transmission. Remember that the essential role of the canopy in the green roof is shadowing; 2. selecting light soils, that reduce the thermal conductivity as well as weight. A large field capacity, which is both beneficial for insulation purposes ( soil thermal diffusivity reduces when soil moisture content increases) and the for plants development, is also desirable. Peats could be a good choice, However, optimizing the green roof performance just for summer time seems to be inappropriate because winter time may be cold and unpleasant in the mediterranean region. Improving green roof for summer may causes problems in winter time. No analyses of green roofs behaviour in winter have been carried out in this paper, mainly due to the inadequacy of the soil modeling approach for this purpose. Concerning modeling, it seems that the adopted degree of simplification for the canopy is good enough for our purposes. The reason is that shortwave radiation transmission is the only process in the canopy which has an important effect on the green roof behaviour. On the contrary. the soil model has to be improved, at least by including moisture balance equa-
192
E. Palomo
Del
Burno
/ Energy
and
Buildings
27 (1998)
179-193
tions. As we mentioned before, the soil moisture content is one of the significant roof parameters that will play an important role in winter.
Substituting Eq. (A.4) into Eq. (A.3), assumingwater potential asa function of water content and temperature,and using Eq. (AS), Eq. (A.3) becomes
J,,= ---
(A.61 When I/J> - lo6 cm, it can be shown that AIT> -g( ti/T-t%jt/aT) ([Nak75]), and we obtain
J,,= ---
where J, is the water vapour flux in the z direction, D the diffusion coefficient of water vapour in air, p\. the density of the water vapour, and u the velocity of convection. For the air movement. a similar equation can be written
1
L D,,$ dZ
(A.7)
which is the equation proposedby Philips and de Vries [ 21 for vapour transport in soils. Comparing Eq. (A.7) with the generalequation
J, = - D,,,$
with J, the air flux and pa the air density. [Jnder the following hypothesis: the air and water vapour obey the ideal gaslaw:
Pv Pv = CT; Pa Pa = R,Ti P=Pv+Pa
we obtain the following functional forms for the non-isothermal and isothermalvapour diffusivities:
D VT = _ --. DA
pv
D vw
RZTP-p,
a* aw
with py = partial vapour pressure; = partial air pressure; pa P, = gas constant of water vapour: R, = gas constant of air; T= absolutetemperature;P = total pressure; the net transfer of air is zero:
J, 3 =()zzQ~=~~.
References
ROOFSOL: Roof Solutions for natural cooling. Contract No. JOR3CT960074, Commission of the European Communities, DG XII Science, Research and Development. [ 2 ] J. Philip, D.A. de Vries, Trans. Am. Geophys. Union 38 ( 1957). [ 3 ] D.A. de Vries. Trans. Am. Geophys. Union 39 ( 1958). 141 A.V. Luikov. Heat and mass transfer in capillary porous bodies, Pergamon. Oxford, 1966. [5 ] S. Whitaker. Adv. Heat Transfer 13 ( 1977) 666. 161 P.V. Vershinin et al., Fundamentals of agrophysics, in: A.F. Ioffe, I.B. Revut (Eds.), Israel Program for Scientific Translations, 1966. [7] M.W. Tschapek, El agua en el suelo, C.S.I.C. ( 1966). Manuales de ciencia actual, (Vol. 2) [ 8 ] J.L. Monteith, Gas exchange in plant communities, in: L.T. Evans (Ed.), Environmental control of plant growth, Academic Press, New York. 1963. [9] A. Perrier. Etude et essai de modelisation des &changes de masse et denergie au niveau des couverts vegetaux: profils micro-climatiques, vapotranspiration et photosynthse nette, Ph.D. University of Paris VI, 1976. [lo] J. Ross, The radiation regime and architecture of plants stands, Dr. Junk Publishers. The Hague, 198 1. ] 111 J.M. Norman, Radiative transfer in vegetation, in: D.A. de Vries, N.H. Afgan (Eds.), Heat and Mass Transfer in the Biosphere. ScriptaBokks, Washington, 1975. [ 121 C. Stanghellini, Radiation absorbed by a tomato crop in a greenhouse, Institute of Agricultural Engineering (IMAG), Wageningen. Research Report 83-5, 1983. [ 131 H. Kasanga, M. Monsi. Jap. J. Bot. 14 ( 1954). [ 141 I. Goudriaan, Crop micrometeorology: a simulation study, Simulation Monographs, Pudoc. Wageningen, 1977. [ 151 J. Ross. Radiative transfer in plants communities. in: J.L. Monteith (Ed,). Vegetation and Atmosphere, Academic Press, London, 1975. [I]
pa az
poresare smallenoughfor the total pressure beregarded to asa constant (aPl(z=O). Eq. (A. 1) becomes
J,,= ----
D p R,TP-p,
dPv c3z
IA 1,
which is the equation that Philip and de Vries [ 21 assumes to be valid for the diffusion of water vapour in soils. To show the effects of soil water ccntent and temperature on the vapour transfer, we introduce the following thermodynamic relations Kelvin equation dlnp, A =dT R,T
Clapeyron equation
(A.51
where ps is the saturatedwater vapour pressure,g the acceleration due to gravity, A the latent heat of vaporization, and $ the total water potential in equilibrium with vapour and negative in soils.
/ Energy
and Buildings
27 (1998)
179-193
193
[ 161 C. Stahghellini. Transpiration of greenhouse crops. Ph.D. Dissertation, Agricultural University, Wageningen, 1987. [ 171 I.R. Cowan, J.H. Troughton, Planta 97 ( 1971). [ 181 G.D. Farquhar, T.D. Sharkey, Rev. Plant Physiol. 32 ( 1981). [ 191 MATLAB, High-performance numeric computation and visualization software, The MathWorks. [ 201 R. Ebert, B. Peuportier, G. Lefebvre, Simulation of the thermal building behaviour based on an object-oriented ADA implementation, Conf.
Proc. Building Simulation 1991, IBPSA, Sophia-Antipolis, France, 1991. [21] R. Ebert, Developpement dun environnement de simulation de systemes complexes. Application au bltiments, Ph.D. Ecole Nationale des Ponts et Chaussees, Paris, 1993. [ 221 L.F. Shampine, M.W. Reichelt, The matlab ODE suite, Matlab Internet services: http: / /www.mathworks.com. [ 231 P. Berdhal. R. Fromberg, Solar Energy 29 (4) ( 1982).