- Email: [email protected]
Assessing the natural ventilation potential of the Basel region
Energy and Buildings 39 (2007) 1159–1166 www.elsevier.com/locate/enbuild
Assessing the natural ventilation potential of the Basel region M. Germano * E´cole polytechnique fe´de´rale de Lausanne (EPFL), Laboratoire d’e´nergie solaire et de physique du baˆtiment (LESO), LE 0 01, Station 18, CH-1015 Lausanne, Switzerland Received 22 January 2006; received in revised form 15 July 2006; accepted 18 July 2006
Abstract A method is proposed to assess the natural ventilation potential (NVP) of the Basel Region, Switzerland, by taking into account the most comprehensive set of factors involved in natural ventilation. These factors are either driving forces, such as wind pressure and stack effect, or constraints, like noise pollution and atmospheric pollution. The process considers these factors in an ordinal qualitative scale and gives its result in this same scale. Climatic data are extracted from a mesoscale atmospheric simulation model referred to as Finite Volume Model (FVM), accounting for the effect of the urban fabric and taking as boundary conditions data coming from a model of the Swiss national weather service (MeteoSwiss). FVM provides an urban turbulence module which specifically simulates the effects of urban areas on the meteorology. Qualitative scales allow to bypass the problem of the inaccuracy of some parameters, which can be very high, especially in urban environment and in the pre-design phase of a construction project. Actually, the method is particularly suitable for designers intending to take early-stage decisions. # 2007 Elsevier B.V. All rights reserved. Keywords: Natural ventilation; Mutlicriteria analysis; Mesoscale atmospheric modelling
1. Introduction Factors involved in natural ventilation are incommensurable. This means that, if all simultaneously considered, they cannot relate or be reduced to another single criterion. Moreover, these factors cannot in most of the cases be evaluated easily with a good accuracy. Both these reasons led to the choice of multicriteria analysis as a tool for evaluating the potential for natural ventilation. This approach will help to cope with the incommensurability and the uncertainty of the problem, by using ordinal scales for each criterion as well as for the final evaluation result. Problems of natural and hybrid ventilation are usually solved by means of airflow simulation tools. These tools are either computational fluid dynamics, zonal, nodal, or empirical models. They solve the equations of fluid dynamics or simplified equations to compute for instance the airflows in the building or parts of it. However, they do not take into account all factors involved in natural ventilation, like external noise, external pollution or safety compromising. Furthermore, they tend to propagate and to increase the uncertainty of the
* Tel.: +41 21 693 45 45; fax: +41 21 693 27 22. E-mail address: [email protected]. 0378-7788/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2006.07.010
input parameters, which is for some parameters, like pressure coefficients, already high, especially in the early phases of a project. It has been proven [1] that the higher the complexity of the simulation tool the higher this uncertainty increases. 2. Criteria involved in natural ventilation 2.1. Driving forces The two driving forces of natural ventilation are wind pressure and stack effect. These forces induce a pressure difference on the building, which in turn generates airflows in it. The wind-induced pressure difference is given by D pw ¼ 12C p rv2
(1)
with Dpw is the difference between the static pressure on a given spot on the building envelope and the upstream reference static pressure (in an undisturbed zone), Cp the pressure coefficient for a given spot on the building and a given wind direction, r the air density, and v is the upstream reference wind speed. Pressure coefficients are determined either experimentally in a wind tunnel or numerically using computational fluid dynamics.
1160
M. Germano / Energy and Buildings 39 (2007) 1159–1166
The other driving force of natural ventilation is stack pressure, or pressure due to buoyancy. It is induced by density differences between the indoor and outdoor air. The Bernoulli equation, combined with the ideal gas equation of state, leads to the stack pressure difference between two openings separated by a vertical distance h: D ps ¼ ri gh
Ti Te Te
This assumption on wind will also hold in the present method for the three last criteria, which are stack effect, noise and pollution (obviously in a reverse way for the last two, since they are constraints and not driving forces). Finally, it is assumed that wind and buoyancy will never counteract each other. (It is always possible to configure opening in such a way.)
(2)
where ri is the internal air density, g the acceleration of gravity, Ti and Te are the internal and external air temperature. In the absence of wind, when Ti > Te, the air enters through the lower openings and goes out through the upper ones (upward flow). A downward flow takes place when Ti < Te. 2.2. Constraints The two constraints to natural ventilation retained here are noise pollution and atmospheric pollution. Others exist, such as compromising the safety of the building while leaving windows open, or the fact that the occupants do not interact in a proper way, but we just mention them. Noise is a constraint to natural ventilation when the building is occupied. Apart from ‘extra’ noise sources, such as airports or building sites, noise levels are closely linked with road traffic, which is in turn primarily linked with the urban feature of a given location. In the currently proposed method, the percentage of urbanization was then used to assess the qualitative noise level. In order to assess the outdoor air quality, the following pollutants are usually considered: nitrogen monoxide (NO), nitrogen dioxide (NO2), sulphur dioxide (SO2), carbon monoxide (CO), ozone (O3), and volatile organic components (VOC). These pollutants each have long-term, short-term limiting values or both. For instance, the Swiss environment protection law (LPE) imposes for nitrogen dioxide a daily limiting value of 80 mg/m3, whereas the annual limiting value is 30 mg/m3. 3. Method 3.1. Assumptions of the method The first assumption made in the current methodology is that the building will be built in a way that gets the most out of the potential for natural ventilation. The same goes in case of refurbishment. In addition, it is supposed that the building occupants are aware of natural ventilation and open the windows or ad hoc openings accordingly. The third assumption, which will be referred to as the monotonicity assumption, is that, given two city locations with different local wind speeds (resulting in particular from the influence of the building surroundings), wind-driven natural ventilation will be more effective at the location having the highest wind speed (provided that natural ventilation is assessed in the same type of building in both locations).
3.2. Justification of the resort to multicriteria analysis Multicriteria analysis is a technique devoted to lighten a decision problem and to help solve it. This problem is made up of several possibly conflicting objectives, translated into criteria. The expected result of a multicriteria evaluation is an action or a group of actions to be taken [2,3]. In our case, the actions are ‘to build at a given place’ and the decision to come to is ‘where to build’. This is why action and location have the same meaning in the following text. A general property of multicriteria aggregation methods is their ‘monotonicity’, that is to say that if any of the criteria improves (respectively worsens), then the global evaluation improves (respectively worsens). In our case, if for example, wind blows stronger in a second situation than in a first one (without changing the other criteria), then the global evaluation of the natural ventilation potential (NVP) will be better in the second situation. Similarly, if a location is noisier in a second situation than in a first one, then the global evaluation of the natural ventilation potential will be better in the first situation. This ‘monotonicity’ is fully coherent with the third aforementioned assumption. This is the first justification for using multicriteria analysis. As another reason, a qualitative multicriteria method is able to cope with the lack of information always encountered in the early stages of a building project. 3.3. Principles of the multicriteria analysis method Qualiflex The principles of Qualiflex [4–6] are explained hereafter in a simplified way thanks to an example made up of three actions a1, a2, a3 (three sites in our case) and three criteria c1, c2, c3. Let us assume that each action (each site in our case) is assigned a rank for each criterion and that each criterion is assigned a weight. Figures in italics in Table 1 represent the ranking matrix elements. In this example, for the first criterion (whose weight is 5), a1 is better placed than a2 and a3, which in turn are equally placed. The next step consists in considering all the possible action rankings Ri, for i = 1, . . ., (3!): R1 : a1 ; a2 ; a3 ;
R2 : a2 ; a1 ; a3 ;
R3 : a3 ; a1 ; a2 ;
R4 : a 3 ; a 2 ; a 1 ;
R5 : a2 ; a3 ; a1 ;
R6 : a1 ; a3 ; a2
For each ranking Ri and for each criterion, the concordance with the ranking matrix must be checked by comparing every couple of actions. Each time the relative position is the same
M. Germano / Energy and Buildings 39 (2007) 1159–1166 Table 1 Input table
Weight a1 a2 a3
c1
c2
c3
5 1 2 2
4 2 1 3
1 3 3 2
1161
2 5 1 4 + 2 1 = 12. And for the remaining rankings see Table 3. If these indices are considered, it comes out that the first ranking R1 accords best with the data of the problem. So the ranking R1 causing a1 to be better placed than a2, in turn better placed than a3 will be chosen.
The ranking matrix elements are represented by figures in italics.
3.4. Filling out the ranking matrix
(respectively different) in the ranking matrix of the problem data and in the ranking Ri, a so-called concordance index is incremented (respectively decremented) by one. In our example, let us consider ranking R4 and criterion c2. For each couple of actions:
The ranking matrix is filled out by computing each criterion ci (i = 1, . . ., 4) for each site aj (whether it is a ‘base site’ or a ‘site to be assessed’). In the current method the criteria are calculated for each hour and then summed up for every hour in the same fashion as is usually done for degree hours.
1. R4 causes a2 to be better placed than a1. So does the matrix for criterion c2. The concordance index is thus incremented. 2. R4 causes a3 to be better placed than a1. The matrix causes a1 to be better placed than a3 for criterion c2. The concordance index is thus decremented. 3. R4 causes a3 to be better placed than a2. The matrix causes a2 to be better placed than a3 for criterion c2. The concordance index is thus decremented.
3.4.1. Wind The first criterion c1 is related to the wind and was proposed to have the following definition:
Therefore, the value of the concordance index for R4 and c2 is 1 1 1 = 1. A minor modification has been brought to this procedure: the concordance index of a given ranking is decremented also if the ranking matrix gives two actions equally placed for a criterion and if the actions differ of more than one position in this given ranking. By repeating this operation for every couple (Ri, cj), the concordance indices take the values given in Table 2. A so-called global concordance index is then calculated for every ranking by adding the concordance indices of the ranking beforehand multiplied by the corresponding weight. For example, the global concordance index of R4 is: Table 2 Matrix of concordance indices
Weight R1 R2 R3 R4 R5 R6
c1
c2
c3
5 2 0 2 2 0 2
4 1 3 1 1 3 1
1 2 2 0 2 2 0
Table 3 Global concordance indices Global concordance index R1 R2 R3 R4 R5 R6
12 10 6 12 10 6
c1 ¼
N 1X vlocal N i¼1 i
(3)
where the summation is made over the entire set of N instants in the considered period and vlocal is the local wind speed at time i. i Note that as the current evaluation is ordinal and as the same building is supposed to be built at all sites (see the ‘third assumption’ made above), the coefficient (1/2)Cpr of Eq. (1) can be set aside, bypassing, in so doing, the problem of the pressure coefficients inaccuracy. The local wind speed vlocal was assessed by way of a i mesoscale atmospheric model referred to as FVM and developed by the Air and Soil Pollution Laboratory (LPAS) of the E´cole polytechnique fe´de´rale de Lausanne (EPFL) [7,8]. FVM took as boundary conditions data collected from the results of a complex atmospheric simulation tool referred to as LM. This latter tool used by the Swiss national weather service (MeteoSwiss) is a prediction model solving the primitive hydro-thermodynamical equations describing compressible non-hydrostatic flow in a moist atmosphere, by using the finite difference method (see, e.g. Ref. [9]). The mesoscale model FVM was applied to compute important urban effects – mainly wind speed reduction, heat island effect and pollutant transport – that are not accounted for in MeteoSwiss’ model. 3.4.2. Stack effect As was the case for wind speeds, outdoor temperatures are retrieved in FVM’s simulation results. The free-running temperature Tfr (i.e. the internal temperature of the building where no heating, cooling or ventilation is used) is calculated using a thermal simulation program, that takes into account the thermal mass of the building, the solar gains and the internal gains and obviously the external temperature. The internal building temperature is calculated starting from this freerunning temperature and by considering simultaneously comfort zones and respectively the heating, cooling or ventilation state. In the calculations undertaken here, the
1162
M. Germano / Energy and Buildings 39 (2007) 1159–1166
building of interest has been assumed to have the following characteristics. It was:
These internal and external temperatures at time i can be input in the following proposed definition of the stack effectrelated criterion: N 1X T ii T ei c2 ¼ (4) N i¼1 T ei As was the case when it came to the wind-related criterion, the coefficient righ of Eq. (2) can be set aside. 3.4.3. Noise levels As mentioned before, noise levels are retrieved from the percentage of urbanization of the site of interest. The fraction ai of rural to urban coverage has been determined as part of another project named KABA (Klimaanalyse der Region Basel, see http://www.unibas.ch/geo/mcr/Projects/KABA) whose main objective was ‘to design maps covering the trinational region of Basel (Switzerland, Germany and France)’. The noise-related criterion is proposed to have the following definition: N 1X ai N i¼1
(5)
where ai is fraction of rural to urban coverage (usually independent of time i). 3.4.4. Pollution levels Pollution concentrations result from their emissions and from their dispersion. Emissions being known, FVM can calculate their dispersion and thus their local concentrations. Nitrogen monoxide (NO) and nitrogen dioxide (NO2) have been taken as tracer gases to inform the overall pollution level. Considering these sole pollutants is quite reasonable because all the other pollutants tend to vary in the same way through time and because the current analysis is purely ordinal. The spatial distribution of these emissions has been assumed to follow the urban density distribution of the city. The pollution-related criterion is proposed to have the following definition: c4 ¼
N 1X C NOxi N i¼1
(6)
where CNOxi is the sum of the NO and of the NO2 concentrations. In addition to these criteria, the fraction of time when free cooling is possible has been computed by way of Eq. (7): qfc ¼
N 1X dfc N i¼1 i
dfc ¼
a brick fac¸ade building; south-oriented; full-time occupied.
c3 ¼
with
(7)
1 0
when T fr > T cu and T e T cu otherwise
(8)
in which Tfr is the free-running temperature, Tcu the upper limit of the comfort zone and Te is the external temperature. dfc is equal to one only if both the following conditions are fulfilled: the air is too warm indoors (in absence of heating, ventilation and cooling) (Tfr > Tcu); the outside air is cool enough to cool down the building to a comfort temperature (Te Tcu).
4. Results The method has been applied over the region of Basel and over a period when an intensive campaign of measurements has been carried out within the BUBBLE project. The BUBBLE project consisted in a large urban Planetary Boundary Layer (PBL) experiment carried out under the auspices of the European COST 715 action. Its aim was that of investigating the exchange processes occurring near the urban surface as well as the flows occurring in the upper part of the Urban Boundary Layer (UBL). The tools used were surface and remote sensing instrumentation on the one hand, and a mesoscale meteorological model on the other hand. FVM has proven to produce results that tally better with these on-site measurements than MeteoSwiss’ larger-scale model. The results hereafter show the natural ventilation potential of the Basel region during a 3-day period taking place during the mentioned measurement campaign, namely from the 25th until the 27th June 2002. The method can of course be applied over a longer period, as soon as FVM is validated over it, providing thus a more representative natural ventilation potential. The simulations undertaken with FVM discretize horizontally the Basel region in 40 40 cells, making up 1600 sites of 1 km2 each for the current method. In what follows, two (and later on eight) base sites have been arbitrarily selected among them. The 1598 remaining sites have been assessed and ranked by Qualiflex (see Section 3.3) in three classes according to whether they were ranked higher than the two base sites, lower than them or in between. This gives rise to shaded maps such as the one shown in Fig. 1 (bottom), for each of which Qualiflex has been run 1600 times (if the base sites themselves are counted). The topmost maps of figure represent the partial rankings of the 1600 sites in each of the four criteria. These rankings are the starting point of Qualiflex’s evaluation (see Section 3.3). With a view to refining the rankings, the number of base sites can be arbitrarily increased. Fig. 2 displays the same NVP assessment as earlier, this time with eight base sites, creating thus nine classes. The fraction of time when free cooling is possible, given by Eq. (7), is represented by Fig. 3.
M. Germano / Energy and Buildings 39 (2007) 1159–1166
1163
Fig. 1. Natural ventilation potential of the region of Basel from the 25th until the 27th June 2002. (Top) Partial rankings of the 1600 sites according to the four criteria: wind hours (foremost map), stack hours, noise hours and pollution hours (hindmost map). (Bottom) Global rankings from the point of view of the natural ventilation potential carried out by Qualiflex. In each map, the two base sites are represented by two squares and the built-up area by a black line. The three classes are higher NVP (light grey), intermediate (grey) and lower NVP (dark grey).
5. Results analysis The results of the natural ventilation potential assessment of the Basel region (Figs. 1 (bottom) and 2) show that the denser the city, the lower the potential for natural ventilation. This is due to the fact that wind hours, noise hours and pollution hours are obviously lower in built-up areas due to the wind speed reduction in the city and to the link of constraints with urbanization. This is a little less evident when it comes to stack hours, since the city-increased external temperature does not directly come into play. In fact, stack effect is reduced because the differences between the internal and the
external temperatures are less marked in the urban context. The fraction of time when free cooling is possible, represented by Fig. 3, is higher in the urban areas because more cooling is required and because, even so, the outside air is cool enough there. In the countryside on the other hand, the indoor air is simply already cool enough without ventilation and therefore passive cooling does not need to be applied. 6. Verification of the results An attempt is made in the current section to validate these last results, which one should be taken with an open mind
1164
M. Germano / Energy and Buildings 39 (2007) 1159–1166
ments. Measurements of ambient fine particulate matter (PM2.5), volatile organic components (VOC) compounds, carbon monoxide (CO) and so forth were undertaken inside and outside buildings. Amongst the four cities selected for this study was Basel. For this validation’s purpose, these measurements are an indirect way to assess natural ventilation by comparing internal and external concentrations of pollutants. Provided that the pollutant has no internal source and that it decays naturally, one can indeed reasonably suppose that, by high ventilation rates, its internal and external concentrations tend to level out. In a publication analyzing the results of the EXPOLIS study, Ha¨nninen et al. [11] use a so-called infiltration factor defined as F INF ¼
Fig. 2. Natural ventilation potential of the region of Basel during the IOP. Eight base sites (squares) are used here.
C ai Ca
(9)
where Cai is the concentration of ambient pollutant that has infiltrated indoors and Ca is the ambient (outdoor air) concentration. The indoor concentration Ci is the sum of Cai and of the indoor-generated concentration Cig:
insofar as the ideal conditions, i.e. the suppositions made in Section 3.1 are hardly ever fulfilled. Besides, an ideal verification should resort to ventilation rates measured in houses of a same given type, all over the studied domain and during the period of interest.
C i ¼ Cai þ Cig
6.1. The EXPOLIS study
where Ci is the total indoor concentration. This formula holds for sulphur (emitted as gaseous sulphur dioxide and oxidized to sulphate in the atmosphere) [11], which was chosen for this verification. Dockery and Spengler [12] elaborated on the mass-balance equation, assuming uniform mixing within the building and steady state conditions over the sampling period. This mass-balance equation gives the following expressions for Cai and Cig:
The EXPOLIS (Air Pollution Exposure Distributions within Adult Urban Populations in Europe) study [10] about population exposures to air pollution was chosen to support the current verification. This study focuses on working age urban populations in Europe, exposed to air pollutants in their homes, workplaces and other common urban microenviron-
(10)
In the absence of indoor-generated particles (Cig = 0), the infiltration factor can be simply written as F INF ¼
C ai ¼
Ci Ca
Pa Ca ; aþk
(11)
C ig ¼
Q Vða þ kÞ
(12)
where P is the penetration efficiency, a the air exchange rate, k the decay rate indoors, Q the source strength and V is the interior volume of the building, which are all constant under the assumed steady state conditions.The first equation of (12) and the definition of the infiltration factor give F INF ¼
Pa aþk
(13)
The air exchange rate thus reads a¼
Fig. 3. Fraction of time when free cooling is possible over the region of Basel during the IOP.
kF INF P F INF
(14)
Ha¨nninen et al. [11] assumed a value of approximately 1.0 for P and a value of 0.39 h1 for k. In order to be able to compare air exchange rates with the NVP results of this chapter, EXPOLIS sites were selected only if:
M. Germano / Energy and Buildings 39 (2007) 1159–1166 Table 4 Air exchange rates of the EXPOLIS sites Site Site Site Site Site Site
1 2 3 4 5
a (h1)
Longitude (m)
Latitude (m)
Measurements period
19.62 5.57 3.51 2.88 2.03
609,944 612,813 611,871 612,238 611,852
266,357 268,937 265,614 265,566 267,742
15–17 16–18 07–09 04–06 11–13
June 1997 July 1997 July 1997 June 1997 June 1997
the site was located in Basel; the measurement were made during the same period of the year as the period studied in the current chapter (that is 3 weeks before and 3 weeks after it); a window at least was open during the 48-h measurements. These conditions restricted the original set of 50 Basel sites to the 5 sites shown in Table 4, ordered decreasingly by air exchange rates. 6.2. Natural ventilation potential of the EXPOLIS sites There are two ways of comparing the ranking of these sites by air exchange rates and their ranking by the method described in Section 3. First, if the purpose is to validate the above NVP maps, a map as the one of Fig. 2 can be used. Fig. 4 shows a close-up of the NVP map together with the five EXPOLIS sites. As noise and pollution levels do not come into play in the EXPOLIS experiments, the weights of the noise and pollution criteria are set to zero in the multicriteria analysis. The same weights have been assigned to the wind and buoyancy criteria. The eight base sites used in Section 4 are used again here. The results are not as meaningful as could be since four sites fall in the same NVP zone. The reason for this is that the EXPOLIS study focus in locations downtown.
1165
Table 5 Rankings of the EXPOLIS sites by way of air exchange rates (first column) and by way of the method described in Section 3 (second column) Ranking by air exchange rates a (highest a in the upmost row)
Ranking by the method described in Section 3 (highest NVP in the upmost row)
Site Site Site Site Site
Site Site Site Site Site
1 2 3 4 5
1 3 4 2 5
A perhaps better and more straightforward way to make the comparison simply consists in ranking the EXPOLIS sites with the method described in Section 3. The result is shown in the second column of Table 5. 6.3. Verification results The number of EXPOLIS sites fulfilling the conditions to be accepted and their distribution throughout the FVM domain are clearly not sufficient to conduct a proper verification. Nonetheless, the map of Fig. 4 is in agreement with the ranking of the sites according to air exchange rates, since site 5, which has the lowest exchange rate, is located in a zone of lower NVP than the other sites. Moreover, Table 5 shows that, except site 2, the sites are ranked in the same order according to exchange rates and according to the method described in Section 3. 6.4. Causes of error In spite of the bad position of site 2, the results are satisfactory because the causes of error are multiple: The year of the EXPOLIS experiments (1997) is different from the year simulated by FVM (2002). Meteorological conditions may have changed from 1 year to the other. The buildings may not be identical to one another and may not be similar to the one assumed in Section 4. Some indoor-generated sulphur may have been emitted as the infiltration factor of site 5, which is greater that one suggests it. (Because F INF > 1, site 5 has in fact a negative air exchange rate. This value is obviously unphysical and should be disregarded. Despite this, one can still confidently state that site 5 has the lowest air exchange rate.) The penetration efficiency P and the decay rate k are just estimates [11] and may be different for each building. Natural ventilation may not have been exploited at best during the EXPOLIS experiments.
Fig. 4. Natural ventilation potential of the region of Basel during the IOP and location of the EXPOLIS sites. (Eight base sites are used here.)
Actually, the current verification is an indirect way of assessing the natural ventilation potential. In an ideal case, air changes measurements should have been undertaken in similar buildings during the BUBBLE experiment.
1166
M. Germano / Energy and Buildings 39 (2007) 1159–1166
7. Conclusion A semi-qualitative multicriteria analysis method has been implemented in order to assess the natural ventilation potential of the Basel (Switzerland) region. The method takes into account driving forces and constraints related to natural ventilation, in order to state whether or not a given site is appropriate for natural ventilation (and passive cooling). The results show that this natural ventilation potential, which is strictly related to airflow rate and to indoor air quality, is higher in rural areas than in urban ones. The results over the region of Basel have been validated with the help of the air pollution exposure study EXPOLIS. The validation is satisfactory, given the numerous causes of error arising. On the other hand, the fraction of time when passive cooling is possible, defined as the fraction of time when the internal building temperature can be brought down to thermal comfort by the sole means of natural ventilation, turns out to be higher in urban areas (up to 92%) and lower in rural ones (up to 25%). This arises from the fact that the building of interest (in the event an uninterruptedly occupied south-oriented brick fac¸ade building) needs less cooling in rural zones. On top of that, a further calculation of the degree hours of cooling saved by ventilation (not shown in this article) demonstrate that substantial energy savings can be made with the help of natural ventilation and that air-conditioning can be completely avoided in this region for the kind of building studied. Attention is drawn here to the fact that this method has also been implemented in a user-friendly software tool [13] (not described here) covering the whole Europe and has been the object of a validation process. These method and tool are meant for designers at the very early stages of a project of construction or of refurbishment. If, for instance, a well-ventilated building is found in a zone ranked by the method as ‘intermediate’, a designer can state with a good certainty that natural ventilation will be applicable in a building of the same type located in a zone ranked as of ‘higher potential’. It should be reminded at this point that the assessed natural ventilation potential is the potential of the site (and not, strictly speaking, that of the building). Once a site with a good potential is found, the designer’s part is to construct a building or to refurbish an existing one in a way that makes the most out of this potential. In other words, both an appropriate site and an appropriate building are necessary conditions for natural ventilation to be applied. Incidentally, this methodology is hard to validate: a proper validation would imply that buildings erected on the assessed
site do take advantage at the best of the potential of those sites, which is rarely the case in existing constructions. Acknowledgments I would like to thank most warmly Prof. Claude-Alain Roulet, my PhD supervisor, for his support, as well as Dr. Alain Clappier and Clive Muller at the Air and Soil Pollution Laboratory (LPAS) of the E´cole polytechnique fe´de´rale de Lausanne (EPFL) for allowing the use their atmospheric mesoscale model (FVM) and for their explanations. This work was partly supported by the Swiss State Secretariat for Education and Research (formerly known as the Swiss Federal Office for Education and Science). References [1] J.-M. Fu¨rbringer, Sensibilite´ de mode`les et de mesures en ae´raulique du baˆtiment a` l’aide de plans d’expe´riences, Ph.D. Thesis, Lausanne, Switzerland, E´cole polytechnique fe´de´rale de Lausanne, 1994. [2] A. Scha¨rlig, De´cider sur plusieurs crite`res. Collection Diriger l’entreprise, Presses Polytechniques et Universitaires Romandes, Lausanne, 1990. [3] A. Scha¨rlig, Pratiquer E´lectre et Prome´the´e. Collection Diriger l’entreprise, Presses Polytechniques et Universitaires Romandes, Lausanne, 1990. [4] J. Paelinck, Qualitative Multiple Criteria Analysis, Environmental Protection and Multiregional Development, Papers of the Regional Science Association, vol. 36, 1976, pp. 59–74. [5] J. Paelinck, Qualiflex, a flexible multiple-criteria method, Economic Letters 3 (1978) 193–197. [6] J. Paelinck, The Multicriteria Method QUALIFLEX: Past Experiences and Recent Developments, Working Paper 1979/15, Netherlands Economic Institute, Rotterdam, 1979. [7] A. Martilli, Development of an Urban Turbulence Parameterisation for Mesoscale Atmospheric Models, Ph.D. Thesis, E´cole polytechnique fe´de´rale de Lausanne, Lausanne, Suisse, The`se no. 2445, 2001. [8] A. Clappier, A. Martilli, P. Grossi, P. Thunis, F. Pasi, C. Krueger Bernd, B. Calpini, G. Graziani, H. van den Bergh, Effect of sea breeze on air pollution in the Greater Athens Area. Part I. Numerical simulations and field observations, Journal of Applied Meteorology 39 (2000) 546–562. [9] J. Steppeler, H.W. Bitzer, M. Minotte, L. Bonaventura, Nonhydrostatic atmospheric modeling using a z-coordinate representation, Monthly Weather Review 130 (8) (2002) 2143–2149. [10] M.J. Jantunen, O. Ha¨nninen, K. Katsouyanni, H. Knoppel, N. Ku¨nzli, E. Lebret, M. Maroni, K. Saarela, R. Sra´m, D. Zmirou, Air pollution exposure in European cities: the ‘‘EXPOLIS’’ study, Journal of Exposure Analysis and Environmental Epidemiology 8 (4) (1998) 495–518. [11] O.O. Ha¨nninen, E. Lebret, V. Ilacqua, K. Katsouyanni, N. Ku¨nzli, R.J. Sra´m, M. Jantunen, Infiltration of ambient PM2.5 and levels of indoor generated non-ETS PM2.5 in residences of four European cities, Atmospheric Environment 38 (2004) 6411–6423. [12] D.W. Dockery, J.D. Spengler, Indoor–outdoor relationships of respirable sulfates and particles, Atmospheric Environment 15 (1981) 335–343. [13] M. Germano, C. Ghiaus, C.-A. Roulet, Natural ventilation potential, in: G. Cristian, A. Francis (Eds.), Natural Ventilation in the Urban Environment. BEST (Buildings, Energy & Solar Technology) Series, Earthscan Publications Ltd., London, Sterling, VA, 2005 , , pp. 195–227 (Chapter 10).
Assessing the natural ventilation potential of the Basel region M. Germano * E´cole polytechnique fe´de´rale de Lausanne (EPFL), Laboratoire d’e´nergie solaire et de physique du baˆtiment (LESO), LE 0 01, Station 18, CH-1015 Lausanne, Switzerland Received 22 January 2006; received in revised form 15 July 2006; accepted 18 July 2006
Abstract A method is proposed to assess the natural ventilation potential (NVP) of the Basel Region, Switzerland, by taking into account the most comprehensive set of factors involved in natural ventilation. These factors are either driving forces, such as wind pressure and stack effect, or constraints, like noise pollution and atmospheric pollution. The process considers these factors in an ordinal qualitative scale and gives its result in this same scale. Climatic data are extracted from a mesoscale atmospheric simulation model referred to as Finite Volume Model (FVM), accounting for the effect of the urban fabric and taking as boundary conditions data coming from a model of the Swiss national weather service (MeteoSwiss). FVM provides an urban turbulence module which specifically simulates the effects of urban areas on the meteorology. Qualitative scales allow to bypass the problem of the inaccuracy of some parameters, which can be very high, especially in urban environment and in the pre-design phase of a construction project. Actually, the method is particularly suitable for designers intending to take early-stage decisions. # 2007 Elsevier B.V. All rights reserved. Keywords: Natural ventilation; Mutlicriteria analysis; Mesoscale atmospheric modelling
1. Introduction Factors involved in natural ventilation are incommensurable. This means that, if all simultaneously considered, they cannot relate or be reduced to another single criterion. Moreover, these factors cannot in most of the cases be evaluated easily with a good accuracy. Both these reasons led to the choice of multicriteria analysis as a tool for evaluating the potential for natural ventilation. This approach will help to cope with the incommensurability and the uncertainty of the problem, by using ordinal scales for each criterion as well as for the final evaluation result. Problems of natural and hybrid ventilation are usually solved by means of airflow simulation tools. These tools are either computational fluid dynamics, zonal, nodal, or empirical models. They solve the equations of fluid dynamics or simplified equations to compute for instance the airflows in the building or parts of it. However, they do not take into account all factors involved in natural ventilation, like external noise, external pollution or safety compromising. Furthermore, they tend to propagate and to increase the uncertainty of the
* Tel.: +41 21 693 45 45; fax: +41 21 693 27 22. E-mail address: [email protected]. 0378-7788/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2006.07.010
input parameters, which is for some parameters, like pressure coefficients, already high, especially in the early phases of a project. It has been proven [1] that the higher the complexity of the simulation tool the higher this uncertainty increases. 2. Criteria involved in natural ventilation 2.1. Driving forces The two driving forces of natural ventilation are wind pressure and stack effect. These forces induce a pressure difference on the building, which in turn generates airflows in it. The wind-induced pressure difference is given by D pw ¼ 12C p rv2
(1)
with Dpw is the difference between the static pressure on a given spot on the building envelope and the upstream reference static pressure (in an undisturbed zone), Cp the pressure coefficient for a given spot on the building and a given wind direction, r the air density, and v is the upstream reference wind speed. Pressure coefficients are determined either experimentally in a wind tunnel or numerically using computational fluid dynamics.
1160
M. Germano / Energy and Buildings 39 (2007) 1159–1166
The other driving force of natural ventilation is stack pressure, or pressure due to buoyancy. It is induced by density differences between the indoor and outdoor air. The Bernoulli equation, combined with the ideal gas equation of state, leads to the stack pressure difference between two openings separated by a vertical distance h: D ps ¼ ri gh
Ti Te Te
This assumption on wind will also hold in the present method for the three last criteria, which are stack effect, noise and pollution (obviously in a reverse way for the last two, since they are constraints and not driving forces). Finally, it is assumed that wind and buoyancy will never counteract each other. (It is always possible to configure opening in such a way.)
(2)
where ri is the internal air density, g the acceleration of gravity, Ti and Te are the internal and external air temperature. In the absence of wind, when Ti > Te, the air enters through the lower openings and goes out through the upper ones (upward flow). A downward flow takes place when Ti < Te. 2.2. Constraints The two constraints to natural ventilation retained here are noise pollution and atmospheric pollution. Others exist, such as compromising the safety of the building while leaving windows open, or the fact that the occupants do not interact in a proper way, but we just mention them. Noise is a constraint to natural ventilation when the building is occupied. Apart from ‘extra’ noise sources, such as airports or building sites, noise levels are closely linked with road traffic, which is in turn primarily linked with the urban feature of a given location. In the currently proposed method, the percentage of urbanization was then used to assess the qualitative noise level. In order to assess the outdoor air quality, the following pollutants are usually considered: nitrogen monoxide (NO), nitrogen dioxide (NO2), sulphur dioxide (SO2), carbon monoxide (CO), ozone (O3), and volatile organic components (VOC). These pollutants each have long-term, short-term limiting values or both. For instance, the Swiss environment protection law (LPE) imposes for nitrogen dioxide a daily limiting value of 80 mg/m3, whereas the annual limiting value is 30 mg/m3. 3. Method 3.1. Assumptions of the method The first assumption made in the current methodology is that the building will be built in a way that gets the most out of the potential for natural ventilation. The same goes in case of refurbishment. In addition, it is supposed that the building occupants are aware of natural ventilation and open the windows or ad hoc openings accordingly. The third assumption, which will be referred to as the monotonicity assumption, is that, given two city locations with different local wind speeds (resulting in particular from the influence of the building surroundings), wind-driven natural ventilation will be more effective at the location having the highest wind speed (provided that natural ventilation is assessed in the same type of building in both locations).
3.2. Justification of the resort to multicriteria analysis Multicriteria analysis is a technique devoted to lighten a decision problem and to help solve it. This problem is made up of several possibly conflicting objectives, translated into criteria. The expected result of a multicriteria evaluation is an action or a group of actions to be taken [2,3]. In our case, the actions are ‘to build at a given place’ and the decision to come to is ‘where to build’. This is why action and location have the same meaning in the following text. A general property of multicriteria aggregation methods is their ‘monotonicity’, that is to say that if any of the criteria improves (respectively worsens), then the global evaluation improves (respectively worsens). In our case, if for example, wind blows stronger in a second situation than in a first one (without changing the other criteria), then the global evaluation of the natural ventilation potential (NVP) will be better in the second situation. Similarly, if a location is noisier in a second situation than in a first one, then the global evaluation of the natural ventilation potential will be better in the first situation. This ‘monotonicity’ is fully coherent with the third aforementioned assumption. This is the first justification for using multicriteria analysis. As another reason, a qualitative multicriteria method is able to cope with the lack of information always encountered in the early stages of a building project. 3.3. Principles of the multicriteria analysis method Qualiflex The principles of Qualiflex [4–6] are explained hereafter in a simplified way thanks to an example made up of three actions a1, a2, a3 (three sites in our case) and three criteria c1, c2, c3. Let us assume that each action (each site in our case) is assigned a rank for each criterion and that each criterion is assigned a weight. Figures in italics in Table 1 represent the ranking matrix elements. In this example, for the first criterion (whose weight is 5), a1 is better placed than a2 and a3, which in turn are equally placed. The next step consists in considering all the possible action rankings Ri, for i = 1, . . ., (3!): R1 : a1 ; a2 ; a3 ;
R2 : a2 ; a1 ; a3 ;
R3 : a3 ; a1 ; a2 ;
R4 : a 3 ; a 2 ; a 1 ;
R5 : a2 ; a3 ; a1 ;
R6 : a1 ; a3 ; a2
For each ranking Ri and for each criterion, the concordance with the ranking matrix must be checked by comparing every couple of actions. Each time the relative position is the same
M. Germano / Energy and Buildings 39 (2007) 1159–1166 Table 1 Input table
Weight a1 a2 a3
c1
c2
c3
5 1 2 2
4 2 1 3
1 3 3 2
1161
2 5 1 4 + 2 1 = 12. And for the remaining rankings see Table 3. If these indices are considered, it comes out that the first ranking R1 accords best with the data of the problem. So the ranking R1 causing a1 to be better placed than a2, in turn better placed than a3 will be chosen.
The ranking matrix elements are represented by figures in italics.
3.4. Filling out the ranking matrix
(respectively different) in the ranking matrix of the problem data and in the ranking Ri, a so-called concordance index is incremented (respectively decremented) by one. In our example, let us consider ranking R4 and criterion c2. For each couple of actions:
The ranking matrix is filled out by computing each criterion ci (i = 1, . . ., 4) for each site aj (whether it is a ‘base site’ or a ‘site to be assessed’). In the current method the criteria are calculated for each hour and then summed up for every hour in the same fashion as is usually done for degree hours.
1. R4 causes a2 to be better placed than a1. So does the matrix for criterion c2. The concordance index is thus incremented. 2. R4 causes a3 to be better placed than a1. The matrix causes a1 to be better placed than a3 for criterion c2. The concordance index is thus decremented. 3. R4 causes a3 to be better placed than a2. The matrix causes a2 to be better placed than a3 for criterion c2. The concordance index is thus decremented.
3.4.1. Wind The first criterion c1 is related to the wind and was proposed to have the following definition:
Therefore, the value of the concordance index for R4 and c2 is 1 1 1 = 1. A minor modification has been brought to this procedure: the concordance index of a given ranking is decremented also if the ranking matrix gives two actions equally placed for a criterion and if the actions differ of more than one position in this given ranking. By repeating this operation for every couple (Ri, cj), the concordance indices take the values given in Table 2. A so-called global concordance index is then calculated for every ranking by adding the concordance indices of the ranking beforehand multiplied by the corresponding weight. For example, the global concordance index of R4 is: Table 2 Matrix of concordance indices
Weight R1 R2 R3 R4 R5 R6
c1
c2
c3
5 2 0 2 2 0 2
4 1 3 1 1 3 1
1 2 2 0 2 2 0
Table 3 Global concordance indices Global concordance index R1 R2 R3 R4 R5 R6
12 10 6 12 10 6
c1 ¼
N 1X vlocal N i¼1 i
(3)
where the summation is made over the entire set of N instants in the considered period and vlocal is the local wind speed at time i. i Note that as the current evaluation is ordinal and as the same building is supposed to be built at all sites (see the ‘third assumption’ made above), the coefficient (1/2)Cpr of Eq. (1) can be set aside, bypassing, in so doing, the problem of the pressure coefficients inaccuracy. The local wind speed vlocal was assessed by way of a i mesoscale atmospheric model referred to as FVM and developed by the Air and Soil Pollution Laboratory (LPAS) of the E´cole polytechnique fe´de´rale de Lausanne (EPFL) [7,8]. FVM took as boundary conditions data collected from the results of a complex atmospheric simulation tool referred to as LM. This latter tool used by the Swiss national weather service (MeteoSwiss) is a prediction model solving the primitive hydro-thermodynamical equations describing compressible non-hydrostatic flow in a moist atmosphere, by using the finite difference method (see, e.g. Ref. [9]). The mesoscale model FVM was applied to compute important urban effects – mainly wind speed reduction, heat island effect and pollutant transport – that are not accounted for in MeteoSwiss’ model. 3.4.2. Stack effect As was the case for wind speeds, outdoor temperatures are retrieved in FVM’s simulation results. The free-running temperature Tfr (i.e. the internal temperature of the building where no heating, cooling or ventilation is used) is calculated using a thermal simulation program, that takes into account the thermal mass of the building, the solar gains and the internal gains and obviously the external temperature. The internal building temperature is calculated starting from this freerunning temperature and by considering simultaneously comfort zones and respectively the heating, cooling or ventilation state. In the calculations undertaken here, the
1162
M. Germano / Energy and Buildings 39 (2007) 1159–1166
building of interest has been assumed to have the following characteristics. It was:
These internal and external temperatures at time i can be input in the following proposed definition of the stack effectrelated criterion: N 1X T ii T ei c2 ¼ (4) N i¼1 T ei As was the case when it came to the wind-related criterion, the coefficient righ of Eq. (2) can be set aside. 3.4.3. Noise levels As mentioned before, noise levels are retrieved from the percentage of urbanization of the site of interest. The fraction ai of rural to urban coverage has been determined as part of another project named KABA (Klimaanalyse der Region Basel, see http://www.unibas.ch/geo/mcr/Projects/KABA) whose main objective was ‘to design maps covering the trinational region of Basel (Switzerland, Germany and France)’. The noise-related criterion is proposed to have the following definition: N 1X ai N i¼1
(5)
where ai is fraction of rural to urban coverage (usually independent of time i). 3.4.4. Pollution levels Pollution concentrations result from their emissions and from their dispersion. Emissions being known, FVM can calculate their dispersion and thus their local concentrations. Nitrogen monoxide (NO) and nitrogen dioxide (NO2) have been taken as tracer gases to inform the overall pollution level. Considering these sole pollutants is quite reasonable because all the other pollutants tend to vary in the same way through time and because the current analysis is purely ordinal. The spatial distribution of these emissions has been assumed to follow the urban density distribution of the city. The pollution-related criterion is proposed to have the following definition: c4 ¼
N 1X C NOxi N i¼1
(6)
where CNOxi is the sum of the NO and of the NO2 concentrations. In addition to these criteria, the fraction of time when free cooling is possible has been computed by way of Eq. (7): qfc ¼
N 1X dfc N i¼1 i
dfc ¼
a brick fac¸ade building; south-oriented; full-time occupied.
c3 ¼
with
(7)
1 0
when T fr > T cu and T e T cu otherwise
(8)
in which Tfr is the free-running temperature, Tcu the upper limit of the comfort zone and Te is the external temperature. dfc is equal to one only if both the following conditions are fulfilled: the air is too warm indoors (in absence of heating, ventilation and cooling) (Tfr > Tcu); the outside air is cool enough to cool down the building to a comfort temperature (Te Tcu).
4. Results The method has been applied over the region of Basel and over a period when an intensive campaign of measurements has been carried out within the BUBBLE project. The BUBBLE project consisted in a large urban Planetary Boundary Layer (PBL) experiment carried out under the auspices of the European COST 715 action. Its aim was that of investigating the exchange processes occurring near the urban surface as well as the flows occurring in the upper part of the Urban Boundary Layer (UBL). The tools used were surface and remote sensing instrumentation on the one hand, and a mesoscale meteorological model on the other hand. FVM has proven to produce results that tally better with these on-site measurements than MeteoSwiss’ larger-scale model. The results hereafter show the natural ventilation potential of the Basel region during a 3-day period taking place during the mentioned measurement campaign, namely from the 25th until the 27th June 2002. The method can of course be applied over a longer period, as soon as FVM is validated over it, providing thus a more representative natural ventilation potential. The simulations undertaken with FVM discretize horizontally the Basel region in 40 40 cells, making up 1600 sites of 1 km2 each for the current method. In what follows, two (and later on eight) base sites have been arbitrarily selected among them. The 1598 remaining sites have been assessed and ranked by Qualiflex (see Section 3.3) in three classes according to whether they were ranked higher than the two base sites, lower than them or in between. This gives rise to shaded maps such as the one shown in Fig. 1 (bottom), for each of which Qualiflex has been run 1600 times (if the base sites themselves are counted). The topmost maps of figure represent the partial rankings of the 1600 sites in each of the four criteria. These rankings are the starting point of Qualiflex’s evaluation (see Section 3.3). With a view to refining the rankings, the number of base sites can be arbitrarily increased. Fig. 2 displays the same NVP assessment as earlier, this time with eight base sites, creating thus nine classes. The fraction of time when free cooling is possible, given by Eq. (7), is represented by Fig. 3.
M. Germano / Energy and Buildings 39 (2007) 1159–1166
1163
Fig. 1. Natural ventilation potential of the region of Basel from the 25th until the 27th June 2002. (Top) Partial rankings of the 1600 sites according to the four criteria: wind hours (foremost map), stack hours, noise hours and pollution hours (hindmost map). (Bottom) Global rankings from the point of view of the natural ventilation potential carried out by Qualiflex. In each map, the two base sites are represented by two squares and the built-up area by a black line. The three classes are higher NVP (light grey), intermediate (grey) and lower NVP (dark grey).
5. Results analysis The results of the natural ventilation potential assessment of the Basel region (Figs. 1 (bottom) and 2) show that the denser the city, the lower the potential for natural ventilation. This is due to the fact that wind hours, noise hours and pollution hours are obviously lower in built-up areas due to the wind speed reduction in the city and to the link of constraints with urbanization. This is a little less evident when it comes to stack hours, since the city-increased external temperature does not directly come into play. In fact, stack effect is reduced because the differences between the internal and the
external temperatures are less marked in the urban context. The fraction of time when free cooling is possible, represented by Fig. 3, is higher in the urban areas because more cooling is required and because, even so, the outside air is cool enough there. In the countryside on the other hand, the indoor air is simply already cool enough without ventilation and therefore passive cooling does not need to be applied. 6. Verification of the results An attempt is made in the current section to validate these last results, which one should be taken with an open mind
1164
M. Germano / Energy and Buildings 39 (2007) 1159–1166
ments. Measurements of ambient fine particulate matter (PM2.5), volatile organic components (VOC) compounds, carbon monoxide (CO) and so forth were undertaken inside and outside buildings. Amongst the four cities selected for this study was Basel. For this validation’s purpose, these measurements are an indirect way to assess natural ventilation by comparing internal and external concentrations of pollutants. Provided that the pollutant has no internal source and that it decays naturally, one can indeed reasonably suppose that, by high ventilation rates, its internal and external concentrations tend to level out. In a publication analyzing the results of the EXPOLIS study, Ha¨nninen et al. [11] use a so-called infiltration factor defined as F INF ¼
Fig. 2. Natural ventilation potential of the region of Basel during the IOP. Eight base sites (squares) are used here.
C ai Ca
(9)
where Cai is the concentration of ambient pollutant that has infiltrated indoors and Ca is the ambient (outdoor air) concentration. The indoor concentration Ci is the sum of Cai and of the indoor-generated concentration Cig:
insofar as the ideal conditions, i.e. the suppositions made in Section 3.1 are hardly ever fulfilled. Besides, an ideal verification should resort to ventilation rates measured in houses of a same given type, all over the studied domain and during the period of interest.
C i ¼ Cai þ Cig
6.1. The EXPOLIS study
where Ci is the total indoor concentration. This formula holds for sulphur (emitted as gaseous sulphur dioxide and oxidized to sulphate in the atmosphere) [11], which was chosen for this verification. Dockery and Spengler [12] elaborated on the mass-balance equation, assuming uniform mixing within the building and steady state conditions over the sampling period. This mass-balance equation gives the following expressions for Cai and Cig:
The EXPOLIS (Air Pollution Exposure Distributions within Adult Urban Populations in Europe) study [10] about population exposures to air pollution was chosen to support the current verification. This study focuses on working age urban populations in Europe, exposed to air pollutants in their homes, workplaces and other common urban microenviron-
(10)
In the absence of indoor-generated particles (Cig = 0), the infiltration factor can be simply written as F INF ¼
C ai ¼
Ci Ca
Pa Ca ; aþk
(11)
C ig ¼
Q Vða þ kÞ
(12)
where P is the penetration efficiency, a the air exchange rate, k the decay rate indoors, Q the source strength and V is the interior volume of the building, which are all constant under the assumed steady state conditions.The first equation of (12) and the definition of the infiltration factor give F INF ¼
Pa aþk
(13)
The air exchange rate thus reads a¼
Fig. 3. Fraction of time when free cooling is possible over the region of Basel during the IOP.
kF INF P F INF
(14)
Ha¨nninen et al. [11] assumed a value of approximately 1.0 for P and a value of 0.39 h1 for k. In order to be able to compare air exchange rates with the NVP results of this chapter, EXPOLIS sites were selected only if:
M. Germano / Energy and Buildings 39 (2007) 1159–1166 Table 4 Air exchange rates of the EXPOLIS sites Site Site Site Site Site Site
1 2 3 4 5
a (h1)
Longitude (m)
Latitude (m)
Measurements period
19.62 5.57 3.51 2.88 2.03
609,944 612,813 611,871 612,238 611,852
266,357 268,937 265,614 265,566 267,742
15–17 16–18 07–09 04–06 11–13
June 1997 July 1997 July 1997 June 1997 June 1997
the site was located in Basel; the measurement were made during the same period of the year as the period studied in the current chapter (that is 3 weeks before and 3 weeks after it); a window at least was open during the 48-h measurements. These conditions restricted the original set of 50 Basel sites to the 5 sites shown in Table 4, ordered decreasingly by air exchange rates. 6.2. Natural ventilation potential of the EXPOLIS sites There are two ways of comparing the ranking of these sites by air exchange rates and their ranking by the method described in Section 3. First, if the purpose is to validate the above NVP maps, a map as the one of Fig. 2 can be used. Fig. 4 shows a close-up of the NVP map together with the five EXPOLIS sites. As noise and pollution levels do not come into play in the EXPOLIS experiments, the weights of the noise and pollution criteria are set to zero in the multicriteria analysis. The same weights have been assigned to the wind and buoyancy criteria. The eight base sites used in Section 4 are used again here. The results are not as meaningful as could be since four sites fall in the same NVP zone. The reason for this is that the EXPOLIS study focus in locations downtown.
1165
Table 5 Rankings of the EXPOLIS sites by way of air exchange rates (first column) and by way of the method described in Section 3 (second column) Ranking by air exchange rates a (highest a in the upmost row)
Ranking by the method described in Section 3 (highest NVP in the upmost row)
Site Site Site Site Site
Site Site Site Site Site
1 2 3 4 5
1 3 4 2 5
A perhaps better and more straightforward way to make the comparison simply consists in ranking the EXPOLIS sites with the method described in Section 3. The result is shown in the second column of Table 5. 6.3. Verification results The number of EXPOLIS sites fulfilling the conditions to be accepted and their distribution throughout the FVM domain are clearly not sufficient to conduct a proper verification. Nonetheless, the map of Fig. 4 is in agreement with the ranking of the sites according to air exchange rates, since site 5, which has the lowest exchange rate, is located in a zone of lower NVP than the other sites. Moreover, Table 5 shows that, except site 2, the sites are ranked in the same order according to exchange rates and according to the method described in Section 3. 6.4. Causes of error In spite of the bad position of site 2, the results are satisfactory because the causes of error are multiple: The year of the EXPOLIS experiments (1997) is different from the year simulated by FVM (2002). Meteorological conditions may have changed from 1 year to the other. The buildings may not be identical to one another and may not be similar to the one assumed in Section 4. Some indoor-generated sulphur may have been emitted as the infiltration factor of site 5, which is greater that one suggests it. (Because F INF > 1, site 5 has in fact a negative air exchange rate. This value is obviously unphysical and should be disregarded. Despite this, one can still confidently state that site 5 has the lowest air exchange rate.) The penetration efficiency P and the decay rate k are just estimates [11] and may be different for each building. Natural ventilation may not have been exploited at best during the EXPOLIS experiments.
Fig. 4. Natural ventilation potential of the region of Basel during the IOP and location of the EXPOLIS sites. (Eight base sites are used here.)
Actually, the current verification is an indirect way of assessing the natural ventilation potential. In an ideal case, air changes measurements should have been undertaken in similar buildings during the BUBBLE experiment.
1166
M. Germano / Energy and Buildings 39 (2007) 1159–1166
7. Conclusion A semi-qualitative multicriteria analysis method has been implemented in order to assess the natural ventilation potential of the Basel (Switzerland) region. The method takes into account driving forces and constraints related to natural ventilation, in order to state whether or not a given site is appropriate for natural ventilation (and passive cooling). The results show that this natural ventilation potential, which is strictly related to airflow rate and to indoor air quality, is higher in rural areas than in urban ones. The results over the region of Basel have been validated with the help of the air pollution exposure study EXPOLIS. The validation is satisfactory, given the numerous causes of error arising. On the other hand, the fraction of time when passive cooling is possible, defined as the fraction of time when the internal building temperature can be brought down to thermal comfort by the sole means of natural ventilation, turns out to be higher in urban areas (up to 92%) and lower in rural ones (up to 25%). This arises from the fact that the building of interest (in the event an uninterruptedly occupied south-oriented brick fac¸ade building) needs less cooling in rural zones. On top of that, a further calculation of the degree hours of cooling saved by ventilation (not shown in this article) demonstrate that substantial energy savings can be made with the help of natural ventilation and that air-conditioning can be completely avoided in this region for the kind of building studied. Attention is drawn here to the fact that this method has also been implemented in a user-friendly software tool [13] (not described here) covering the whole Europe and has been the object of a validation process. These method and tool are meant for designers at the very early stages of a project of construction or of refurbishment. If, for instance, a well-ventilated building is found in a zone ranked by the method as ‘intermediate’, a designer can state with a good certainty that natural ventilation will be applicable in a building of the same type located in a zone ranked as of ‘higher potential’. It should be reminded at this point that the assessed natural ventilation potential is the potential of the site (and not, strictly speaking, that of the building). Once a site with a good potential is found, the designer’s part is to construct a building or to refurbish an existing one in a way that makes the most out of this potential. In other words, both an appropriate site and an appropriate building are necessary conditions for natural ventilation to be applied. Incidentally, this methodology is hard to validate: a proper validation would imply that buildings erected on the assessed
site do take advantage at the best of the potential of those sites, which is rarely the case in existing constructions. Acknowledgments I would like to thank most warmly Prof. Claude-Alain Roulet, my PhD supervisor, for his support, as well as Dr. Alain Clappier and Clive Muller at the Air and Soil Pollution Laboratory (LPAS) of the E´cole polytechnique fe´de´rale de Lausanne (EPFL) for allowing the use their atmospheric mesoscale model (FVM) and for their explanations. This work was partly supported by the Swiss State Secretariat for Education and Research (formerly known as the Swiss Federal Office for Education and Science). References [1] J.-M. Fu¨rbringer, Sensibilite´ de mode`les et de mesures en ae´raulique du baˆtiment a` l’aide de plans d’expe´riences, Ph.D. Thesis, Lausanne, Switzerland, E´cole polytechnique fe´de´rale de Lausanne, 1994. [2] A. Scha¨rlig, De´cider sur plusieurs crite`res. Collection Diriger l’entreprise, Presses Polytechniques et Universitaires Romandes, Lausanne, 1990. [3] A. Scha¨rlig, Pratiquer E´lectre et Prome´the´e. Collection Diriger l’entreprise, Presses Polytechniques et Universitaires Romandes, Lausanne, 1990. [4] J. Paelinck, Qualitative Multiple Criteria Analysis, Environmental Protection and Multiregional Development, Papers of the Regional Science Association, vol. 36, 1976, pp. 59–74. [5] J. Paelinck, Qualiflex, a flexible multiple-criteria method, Economic Letters 3 (1978) 193–197. [6] J. Paelinck, The Multicriteria Method QUALIFLEX: Past Experiences and Recent Developments, Working Paper 1979/15, Netherlands Economic Institute, Rotterdam, 1979. [7] A. Martilli, Development of an Urban Turbulence Parameterisation for Mesoscale Atmospheric Models, Ph.D. Thesis, E´cole polytechnique fe´de´rale de Lausanne, Lausanne, Suisse, The`se no. 2445, 2001. [8] A. Clappier, A. Martilli, P. Grossi, P. Thunis, F. Pasi, C. Krueger Bernd, B. Calpini, G. Graziani, H. van den Bergh, Effect of sea breeze on air pollution in the Greater Athens Area. Part I. Numerical simulations and field observations, Journal of Applied Meteorology 39 (2000) 546–562. [9] J. Steppeler, H.W. Bitzer, M. Minotte, L. Bonaventura, Nonhydrostatic atmospheric modeling using a z-coordinate representation, Monthly Weather Review 130 (8) (2002) 2143–2149. [10] M.J. Jantunen, O. Ha¨nninen, K. Katsouyanni, H. Knoppel, N. Ku¨nzli, E. Lebret, M. Maroni, K. Saarela, R. Sra´m, D. Zmirou, Air pollution exposure in European cities: the ‘‘EXPOLIS’’ study, Journal of Exposure Analysis and Environmental Epidemiology 8 (4) (1998) 495–518. [11] O.O. Ha¨nninen, E. Lebret, V. Ilacqua, K. Katsouyanni, N. Ku¨nzli, R.J. Sra´m, M. Jantunen, Infiltration of ambient PM2.5 and levels of indoor generated non-ETS PM2.5 in residences of four European cities, Atmospheric Environment 38 (2004) 6411–6423. [12] D.W. Dockery, J.D. Spengler, Indoor–outdoor relationships of respirable sulfates and particles, Atmospheric Environment 15 (1981) 335–343. [13] M. Germano, C. Ghiaus, C.-A. Roulet, Natural ventilation potential, in: G. Cristian, A. Francis (Eds.), Natural Ventilation in the Urban Environment. BEST (Buildings, Energy & Solar Technology) Series, Earthscan Publications Ltd., London, Sterling, VA, 2005 , , pp. 195–227 (Chapter 10).