Natural ventilation for passive cooling: measurement of discharge coefficients

E ' IEKGY A| ID BUILDI IG ELSEVIER

Energ 3 and Buildings 27 (1998) 283--292

Natural ventilation for passive cooling: measurement of discharge coefficients F. Flourentzou *, J. Van der Maas, C.-A. Roulet L~t; ~ratoire d'Ener~,ie Solaire et de Physique du B~timent (LESO-PB~, l~'+'ole Po[vte('hnique F&l~rale de l~u.~anne (t-PFLi, CH- 1015 lx~u.~'amw,Swlt. er/and

Accepted 15 August 1997

Abstract Eor the design of natural ventilation systems for passive cootmg in buildings, engineers and architects are interested in the prediction of ventilation rates as a function of position and size of the \ entitation openings. In common use, there are both simple and detailed (i.e., multizone) ventilation models which rely basically on the same Bernoulli algorithm to describe airflow through large openings+ An important source of uncertainty is related to the attribution of discharge coefficients. The present study was undertaken 1o improve our knowledge on vel,~city and discharge coefficients when measured in reJl buildings. The experiments were performed on a nalurally ventilated three-level office building where the staircase acted as exhaust chimney. In order to keep the flow pattern stable, a condition l{w air flow measurements to be reproducible, the experiments were performed on ~,indless nights where the flow was only driven by slack pressure. Air flow patterns wine traced with smoke and tracer gas. In a first set ol: experiments, air veh)cities, contraction and velocity coefficients and the position of the neutral pressure level ( NPL ) have been measured, in a see{rod set of experiments, the resulting effecti~ e area of a c~m~binatiun of two openings in series. Air flow rates derived from ,~elocity measurentents in the open doorways were found to be in agreement with the flow rates obtained with a constant injection tracer gas technique, with an uncertainty of ::~:2(Y,;{-. The velocity coefficients ¢ = ().7 :f 0.1 and jet contraction coefficients ,~:- 0.85 _+0.1 found in the experiments are sh~)wn to be in agreement with the generally accepted value oithe discharge coefficient C,t : :: q~ g = 0.6 +__0. t, giving new justification for its use in the models. Basic configurations for ventilative cooling are given to illustrate how qualitatix.e modeling used m simple models can ,.z,ive valu ible information to the designer, c~ 1998 Elsevier Science S.A. All rights reserved. K~'x ,rord,w Coeflicicnt: Velncity measurelncnl: Ventilatixe Co<+[iII~

1. Introduction Reality is infinitely complex. O u t k n o w l e d g e of the world is always finite and therefore always incomplete. The marvel is that we function quite well in the world in spite of never fully understanding it. Our tool to deal with incomplete km~wledge is modeling. A model is a small finite ! i n c o m plete) description of an infinitely c o m p l e x reality: for the propose of answering particular questions. The complexity of :~ model depends on the kind of questions we arc seeking to answer I 1 ]. Ventilation in buildings is a c o m p l e x phenomenon. A key question is which variables should be taken into account to answer a particular question. For natural ventilation in summe~, engineers and architects arc interested to k n e w if the vet~tilation rate is sufficient to extract pollutants or heat from * Corresponding author. Tel.: ~41 21 6933',63: fax: +41 2+ fl932722: e-m dl: I]ourentzou(a+leso.da.epfl.ch ~ Fel.: +41 21 6934557: fax: +41 21 6932722: ¢-mai: fouler(o' le,o epll.ch 037";-7788/98/$19.00 ~,') 1998 Elsevier Suicnt.e S.A. All rights Ecscrvcd. Pl1,,0378 7 7 8 8 ( 9 7 ) 0 0 0 4 3 I

a given space. A d o m i n a n t parameter lo answer this question is the position and size of the ventilation openings. Simple ventilation models in the pre-design phase of a project are often found to be sufficient to determine the effect o f these parameters [ 2,3]. M o r e o v e r , in a comparison of detailed and simple models of air infiltration it was found that more c o n > plex models do not generally increase the accuracy of the result 14,51. The well-known Bernoulli model offers valuable information on the air flow path, flow rate and both inlet and outlet velocities [ 6 - 9 ] . More recently, detailed ventilation models have been coupled with building thermal models, in order to evaluate the effect of cooling by natural ventilation I I
284

I-. l~'lot~r~'llt=ou el ol. /Etl{'r£n' tllld tll~ihlingv 27 t 1998) 283 292

Fig. 1. Outside view of the LESO building. ficients can be flmnd in the reference work of Idelchik [ 14]. few data are available which have been measured on real buildings. Indeed, a large number of sources deal with the c o e f f i c i e n t [ 2,3,6,9 ], but not so much with the factors determining its value. Furthermore, there is confusion between the concepts of velocity and discharge coefficient and very often, one is used in place of the other. In the absence of better information, a generally used value for the discharge coefficient of a sharp-rimmed large opening is (',~= ((i).6_+0.2)

191. In order to improve our knowledge of the actual values of flow resistance coefficients in networks, describing air flows in buildings, this paper presents new experimental results. The data has been obtained on a naturally ventilated three level office building (Fig. 1) which was available fl~r the study of basic stack ventilation configurations. The research was performed in the framework of a European research project on passive cooling of buildings [ 15,16 ] and the objective was both to validate simple ventilation algorithms and to give an experimental basis to design guidelines for nighl cooling techniques [ 17--201. The multilevel office offers many possibilities to study the influence of openings ( size and position ) on both the neutral pressure level ( N P L ) and the airflow patterns. Single sided ventilation and cross-ventilation situations have been studied. A single flow path configuralion was achieved by closing all windows and doors in the building envelope, with the exception of an exhaust on the roof and one single office window as the ventilation air inlet. Air flow patterns were traced with smoke and tracer gas. Air flov~ rates derived fl'om velocity measurements in open doorways were found to be in agreement with the flow rates obtained with a constant injection

tracer gas technique, with an uncertainty of :_+_20c,4. Overall agreement was found between the velocity measurements and theoretical models. The simple Bernoulli model is used to determine the influence of the size and position of the openings on the natural ventilation of buildings. The role of the NPL lo achieve efficient cooling is briefly disct.ssed. In the experiments, air velocities and the position of the NPL have been measured, and contraction and velocity cuel:ficients as used in the Bernoulli model have been observed, where the only driving force is stack pressure ( wind driven flow tends to be unstable, can not be modeled in a simple wa}, and has therefore been avoided in the experiments). In a second set of experiments, the resulting eflective area of a combination of two openings in series was studied.

2. Ventilation modeling In order" to clarify the confusion in the literature between discharge and velocity coefticients, the basic theory necessary, to interpret the measurements is recalled in this section. 2. I. S t a c k l,-e.vsure

Air temperature differences cause density differences, and therefore pressure differences between outdoors and indoors~ When the inside temperature of a building is higher than the outside temperature, the pressure distribution over the bui Iding height qualitatively takes the [orms shown in Fig. 2. h~ these cases, air is llowing inside through the openings below the NPL and outside through the openings above it. The

285

F. H o trent:o ¢ ~ t al / E n e r g y and Buihlinq, v 27 (1998) 28,3 292

Ap I e

b

z, o Znpl

d

NPL z,,pt ................ NPL ___

NPL-Znpl I _

To
=- -a

NPL . . . . .

Fig. 2. Distribution of stack pressure difli~rence. ~ P, and position of Ihe NPL for different opening configurations m the absence of wind,

neutral pressure level (NPL) is the height at which the interior and exterior pressures are equal [ 3]. ZNp~.is the distance of the NPL from the reference level~ In a building with unitbrm leakage distribution and in the absence of wind, the NPL is near midheight of the building (Fig. 2a). The size and position of the openings ttetermine the location of the NPL (Fig. 2b.c,d). A groundfloor opening lowers the NPL (Fig. 2b), a topfloor opening increases the height of the NPL (Fig. 2c). According to the mas;s conservation principle the air mass flow rate through the openings below the NPL equals the air mass flow rate through the openings above it (Fig. 2d ). The hydrostatic indoor/outdoor pressure difference for an enclosure is expressed as:

(/',-7)) A/*~(Z) = ( p ~ - - P i ) g ( Z - - Z N P l ~ ) = P i g ( : - - Z N P l

I

-Xp,,,~,,i = g¢)~,,zd

t2)

1

Under these conditions Bernoulli' s equation becomes:

kp~( z, ) - lp~(u~

where the ideal gas equation is used and a single zone approximation is made (no temperature stratification, constant indoor air density).

-

1

1

St,~scuT"- jz.N,, ,~' ::=0.

(4 )

Combining Eqs. ( 1 ) and (4), the air velocity in the opening can be expressed as: [

,

"'

AT

= 'Pl" 2,vl :- :~,,t./T~--7

~5)

where use is made of the approximation p~ =p~. (valid for small A TI. The velocity coefficient g' is expressed as:

2.2. Bernoulli's equation 'Fhe application of the energy conservation principle bmween the horizontal cross section of the building at the NPL level and the vertical cross section of the opening A t (Fig. 3) gives the Bernoulli equation [ 141. The surface area of the building cross section compared to the opening surface

/

NPL

Znpl~

', \,

as

) - - ' -/ -

(1)

T~,Pe

area is large, so velocity in this seclion is low and kinetic energy on the section NPL can be neglected. The pressure drop between the two sections is the sum of the local pressure drop in the opening ( coefficient ~) and the pressure drop due to friction in tile air path from A~ at NPL ( coefficient ~). They can be normali:,ed to the air velocit3, u,

Ti,Pi A~ '~¢

F~g. 3. Air flow through an opening due to slack pressure whcn I', > 7",

(6)

: v/#+ g~+ N, where N~ is the kinetic energy coefficient depending on the uniformity of the velocity distribution in the opening section, ( i s the local pressure drop coefficient for the opening and ~" is the friction pressure drop coefficient [ 14]. In a real building £includes also the effects of the pressure drop due to obstacles and direclion changes in the air flow path. ,p is determined experimentally and depends on the Reynolds number on the opening nature as well as on the obstacles in the air [tow path. It is difficult to distinguish the effect of each parameter included in ~, but an overall value can be found.

2.3. Jet fi)rrning in the opening The one-directional flow of air passing through an opening forms a jet. The stream lines are curved near the opening edge and the effective area ot' the opening is smaller than the geometrical one due to the contraction of the stream lines. In

286

k. Flou~ entzou el al. /

Apw/~tO"and Building.s" 27 ( 1998~ 283 29.2

paragraph the air flow through an external opening is visualized using smoke, showing the forming of a jet ( Fig. 7 ). The jet contraction depends on the Reynolds number and on the geometry of the edges of the opening; moreover, the dimensions of the space on both sides of the opening have an important influence on the flow contraction and therefore on the flow resistance [ 141. The effective area of the opening depends on how the je! fills the opening section. ~: is the coefficient of jet contraction (or coefficient of filling the section) and is defined as: Aelfeclive

i

m

NPL

Z~

r,o

G

(7i

A~eonlelr k 11

Fig. 4 Two openings enclosure. 2.4. A i r f l o w rate

Integration over the opening height gives, the airflow rate though the opening:

c= f w,,a:.

,s>

. i

For a single opening this integration give,,: if

1 / AT '£l: = ~g+pA 1/ g H - 7 - =

[ - -

1 1 /, H-7 AT

For two openings the air velocity can be considered constant all over the opening height if the opening is far from the NPL. The mean air velocity can be taken equal to the velocity in the center of the opening. In this case the airflow rate is:

~] is the verlical distance between the center of gravity of the bottom opening and the reterence level• For convenience the reference level can be chosen such that :.:l =: (/. For enclosures with a single rectangular vertical opening, the NPL is situated near Ihe middle of the opening height when temperature differences are small and the densities of the inlet and outlet air are ~dmilar: the error is less than 0 . 3 q tot a A Tog 10 K [ 9 ]. For more than two openings it becomes complicated to calculate the neutral level using an explicit equation and it is more con',,enient to use a numerical solution method. The software LES()COOL [ 19] inch, des a tool to determine the NPL for an enclosure with multiple large o p e , ings. The algorithm used combines the basic principle (,f AII)A [ 7,8 ] and the full algorithm describing airllow through large opening [ 22 I. 2.6

Openings in series

/

l: = A C d t / 2£'(Z--ZNpI.) T

(10)

Cd is the discharge coefficient. [t is the product of the velocity coefficient ¢ and the contraction coefficient ,:. (111

C d = ~'(~

The discharge coefficient can be determined experimentally when the air flow rate is measured directly, e.g., with a tracer gas method [ 211. The velocity coefficient can be determined similarly by measuring the air velocities.

The mass conservation principle is applied, using Eq. (10). to an air tight enclosure with two openings, one at the top and one at the bottom (Fig. 41. With Znpl, the distance of the NPL above the reference level, and H, the vertical distance between the centers ol' gravity of the two openings Eq. ( 12 ) can be derived: H

L. L; .T,2,

,:z, ,,,: E,;+ 7 = (13) When two openings on the same height = are placed in series an effective opening area, A~u, can be defined as:

2.5. Neutral pressure level

=NH =,=,

For a number of openings in series, the total pressure drop for a given mass flow, equals the sum of the partial pressure drops over each opening. Because the mean opening veloci D varies inversely with the effective opening area (continuity equation), the pressure drop over successive openings varic's inversely with the square of IhJs area 16, 13 ]. It lollows tha~,

(12)

If A, > > A ] the pressure drop over the second opening can be neglected. If A, > 1.5Ar the error of neglecting A2 is less than 11%. while if A, > 2A~. this error is less than 6cA.

3. Experiments The LESO building (Fig. 1 has offices at three level,,, with a staircase extending down to the basement and up to a

287

F. H o iret t:o ~ e; al / Ener~:v a n d Buildi,g.~ 2 7 ' 1998) 283 292

staircase. In all cases, the top opening was represented by the sunspace door at rooflevel (0.9 × 2 m, exposed to the east). The bottom opening for the first experiment (a) was the entrance door ( 0.8 × 2.3 m, exposed to the east ). For the next live experiments (Table lb-f), a ground floor window of tixed height ( 1.5 m) and variable width (0.95 m-0.2 m) was used. The window was situated in the center of the south facade. The airflow pattern is sketched with the building cross-section in Fig. 5. We report on three series of experiments, performed in September (h, e, f), November (c) and December (a, d) respectively, for which the conditions of low wind speed ( less than 0.6 m/s ) and stable indoor-outdoor temperature difference ( k T i n the range 8 to 10 K) were satisfied. T, ('Fable 1 ) is the average of four characteristic temperatures measured along the flow path between inlet and outlet. The thermal stratification along the air flow path varied between 2.5 and 5.5 K. It is noted that thermal stratification is not explicitly taken into account by the model which uses mean temperatures. The velocity coefficients measured in the experiments a to f (Table I ) are plotted in Fig. (:~. showing that there is no significant correlation between g, and the opening dimension or air velocity. Table 1 gives the values of the parameters for each experiment. The same experimental procedure was applied to two more buildings satisfying the following conditions: • Internal restrictions on the flow path are larger than two times the smaller opening • Temperature stratification is less than I K / m • Windspeed is smaller than 0.5 m/s • Building height is larger than three times the opening height in both these cases, the measured values of ~ were also 0.7 ± 0. I. confirming that this value is not specific to a particular building.

filding

Expc eri ment a

U 2

,<_...

NPL

z l

-

i

Fig, 5. Openings used liar the experiments of Section ~ l,

sunspace on rooflevel (Fig. 5). The building which is naturally ventilated through a central staircase, with openings available on all levels, is ideal for ventilation expe~riments.

3.2. Contractimz coe~cient ,: 3. I. Velocio, coelficient ~.['or large openings The contraction coefficient could be determined experimentally by visualizing the jet forming in the opening. The experimental set up is identical to the experiments of Section 3.1. The smoke is released outside the opening of the groundfloor office (003 in building sketch of Fig. 5 I. A video

The velocity coefficient was delermined in six different two-opening configurations as illustrated in Fig. 5 and Table 1. Except for the top and bottom opening, the building w.~s kept closed, including all internal doors towards the 1.0 0.9 0.8 0.7 1 0.6 0.5 0,4 0.3 0,2 0.1 0.0

r

-

a

bot.

I

q

I

a top

b bot

b top

--

. . . . .

I

-f-

I

I

I

c bot

c top

d bot

d top

e hot

J ......

----+ e top

f hot

f top

Experiment ]::i;. 6. Velocity coefficients measured m lop and bollom opening for six experimcnls. The data ( a to f, in Table I ) correspond Io decreasing bottom opening Si,'e.

288

F. l"hml entzou et al. / l:'ner~,y and Buildings 27 (1998) 283 295:!

Table I Results of experiments a to f to determine velocity coel licienls ~' Experiment Ah ( bottom ) A, (top) 7~ ( internal temp.) T~. ( external temp. ) ,q ( measured velocity ) u~ ( measured velocity ) ",p+ ( top ) u~ot ( bottoln )

m" hi: C 'C m/s m/s

+~:~,,

m

bL

-

a

h

c

d

e

l

(r

1.9 1.8 16.2 (~.0 1.3 1.4 0.73 0.75 4.4 I +,'i

1.4 1.8 22.1 13.0 1.3 1.0 0.66 0.66 6.4 7",~

1.4 I.,~ 18.9 9. I 1.4 I. I (}.68 (].fi9 6. I < I 6q

1.4 18 16.7 4.9 1.8 1.3 0.7t# ()+7t) b2 I (~

0.9 1.8 22. I 13.0 1.6 (I.6 0,68 0.63 8.7 --4q

0.3 1.8 22 I t 3.5 1.0

0.()5 n.05 11.3 0.5 (I.(!4 ().(}1 0.09 [).09 t).3

0,{+7 IO2 05'

"The velocity coeflicients ¢p~and q~+are calculated with Eqs. ( 5 ) and ( 7 ). For experiments a to f. ¿heir mean value is 0 7. Fhe theoretical standard devialitm in ~. calculated from the uncertainties m the measured parameters ( last cohmm ). yields 0.t)9. which is larger than the errur dm ived from the spread in experimcnhd values of ~ ( o- =0.06). This shows that the theolvtical ,alue for o- is pessimistic and that the ditli~rences m the experimental values are due to random enors bgq. ( I 2) tot the NPL is corrected for the building air leakage; the latter depends tm the size of the crocks aml their location with respect to the NPL. which changed for every experiment. The leakage mass flow rale. L. is ewduated by applying:

and using the experimental values ol the parameters e ( see Section 3.2 ) The correction L is given as a percentage of the llow rate Ihrough the smallest o p e m n g . Combination of Eqs. (5) and ( 15 ) yields an implicit eqtmlion in :NH fin"which value:, are given for comparison:

c,~,,t,<~' 2~:,+,+ ~,--~=,,f~

T-

c,~,,~,,:~ :~,,: --: ..... > r,

TI

+

~ :'

(1~,:

camera is placed on the opening edge, recording the flow of the smoke. Graph paper placed vertically on the w i n d o w edge allowed quantification of the contraction effect and the effective area ( 1 graduation = I c m ) . Digitized images for air velocities in the range 0.4 to 0.9 m / s are shown in Fig. 7. The .jet contraction in the opening section oscillates continuously between 2 and 4 cm independent o f tile air velocities which ranged from 0.3 to I m / s . Considering a contraction of 3_+ 1 cm, the coefficient +::. for a rectangular opening of dimensions I X 1.5 m, is 0.90_+_0.03. .?.3. O p e n h z g s in series

In the first set o f experiments, it was supposed that there was no additional resistance ( due to restrictions or obstacles) between the inlet and exhaust opening. In this second set of experiments the validity o f E q . (14) was tested. The windows of an office on the first floor ( 103 in building sketch of Fig. 8) were used as air inlets, and the area of an intermediate opening ( office door) was varied, while air flow rate measurements were performed. The effective surface area A , . , could be normalised to Ah as A~, = ,% ~.n A~, where q:~.,.~is the new velocity coel'licient after c o m b i n i n g the two openings in series Ah and A,,,,: From Eq. ( 1 4 ) , 4~,n could be derived as:

, -7-=

, -~4-

(17)

The contraction coefficients have been evaluated by visualizing the air flow path with smoke at r o o m temperature. For

the outside opening Ab. the distance between the jet and the opening edge was (3-1__ 1) cm (Fig. 7) on all four sides of the perimeter. This gives a contraction coefficient of 0 . 8 5 + 0 . 0 5 . For the intermediate opening, Am, the ~,,, i> higher (~,,,, = 0 . 9 5 ) because only two edges are sharp (the sides near the floor and near the wall do not modify the air stream lines). The results in Table 2 show good agreement between the measured velocity coeflicienls and those estimated using Eq. ( 17 !. The velocity coefficients ior the top opening have similar values as those obtained in the first set o f experiments ( Tahle 1 ). 3.4.

T r a c e r g a s nlf't~13;[ll'~'ltl(.,tl[3;

The experimental set up is the same as in Section 3.1 (Fig. 5 ) with a bottom opening of 1.4 m" and a top opening of 1.35 m 2. The inside-outside temperature difference, A 7". is 3.5 K and there is no measurable wind outside. |n the intermediate door between the ground office 003 and the staircase, a constant flow rate of 4.17 m l / s of SF6 gas is mixed naturally with the i n c o m i n g air. The concentration of SF6 measured at the top opening gives the air flow rate through the building and therefore the discharge coefficient Ca. The tracer gas apparatus CES AR and tracer gas measuring techniques are described in Rcf. 1211. Point velocity m e a s u r e m e n t s ;.It difflerent heights of the openings give the velocity o)efticient g:. Multiplied with the contraction coeflicient ,: calculated in Section 3 2 , a second discharge coefficient is obtained. Fig. 9 sho',~s good agree-

f" Flourem=ott et +J. / E , ergy and Bmldin£,s 27 (1998) 283-292

289

,]

5 7 -~/\.

:

204 2O5

103 104 003

~ :"

1 {)!! -/

. 004 005

LESO building U

'~

Z2,---+ +

[Zn~ ........... ~

~++- - N I 3 ~ L

~ ......

il _

_

I

A int(variable i[1 I1

Fig. 8. Opening o f Section 3 . 3

configuration and air flow palleln used in the uxpermlenls

Table 2

Measured v e h , c i t y cocfficienl ~ c o m p a r e d

~!

h

i

m: m: tn ~

18 07 I 9

1.8 0.7 0.9

1.8 0.7 I).72

1.8 1.4 1.9

m/, m/,,

072 1+32 074 0.66

0.64 I.I0 0.64 0.54

(}66

()57

{).64 1.01 0.62 0.50 11.53 5¢{

1.00 {).98 c).69 0.62 I).58 6{4

Experiment

Fig. 7. Video images of the airflow pattern in the window used in Experiments b-f: the air flowing in from the right is traced with smoke released outside; the two top figures are the same: the bouom ligure shows lurbulcn/ flow.

mere between the discharge coefficients evaluated by the two methods. The generally accepted value of C u = 0 6 + 0 . 1 [ 14]. is confirmed once again.

4. Design guidelines T h e NPL is a v e r y important concept for the correcl design of natural ventilation. When purely stack driven ventilation is used to cool high buildings, it is impossible to enter fresh

A~ (top opening surface area} Ah { bottom opening surface area) A,,, ( m t c r m e d i a t c opening surface area ) u, Imeasuredl I+~ < m c a s u r e d i see q top v e l o c h y c o e f f i c i e n t ) ~F I Illc+-lsul+ed boll. velocity coef. ) ~#,.,, (calculated with Eq, (17)) Ci difle+renc¢

~ ilI1 t % ( 16 )

-

I',,+

- 64/}

j

outside air through the openings above the NPL. Opening a window in a high building implies a modification of the location of the NPL and as a resuh the air flow path m the whole building can be radically changed with a change of the llow direction in several openings. To allow fresh air to enter the highest building level, the exhaust opening should be placed well above this level and

F. t:hmrert:ou et a/. /E'neC~.x a . d Buildip~:s 27 ( 19981 283 292

290

Air flow rate & Discharge coefficients 1.0 350O

0.9

3000

0.8 0.7

25~

oo. , , ., - x " - " x " "

~, [nefh]

x'x

' ' "'' x' ''; <" " '

"""""'"':'""..x--. .

. x >. .~. . , . . ~ . . . x.

2000

0.5 0.4

1500 1~0

-.

~) [ m 3 / h ]

......

Cd gas tracer measuremen

x

5~ 0

0.6

i i i- t ~ 6:40

6:45

-~ + - ~ + ~ - - ~ - + - - k - I 6:50

6:55

7:00

7:05

0.3

Cd velocity measurement

, I I ~ , +-.--t i I I - - - F - ~ 7:10

Cd

~-- ~ + - 4 - - F ~ ~ I - . ~ - ~ 7:15

0.2 0.1 0.0

7:20

Time

Fig. 9. C o m p a r i s o n o f the di~ c h a r g e coefficient m e a s u r e d b?, tracer gas and by velocity m e a s u r e m e n t s ,

(a)

to)

!= Nt

tc)

td)

Fig. 10. Four cases s h o w i n g h o w the o p e n i n g etimensio=7 and position can be m a n i p u l a t e d to obtain night coolin:g of all Ihe stories of a three story b u i l d i n g { a poor cooling o f upper level; ( b ) to ( d ) v a r i o u s wa),s Io ~mprovc cooling.

F. Fhmrentzou ct al./Energy and Bnihting,s 27 (1998) 283 202

be made as large as possible. It should be placed on the leeside of the building so that wind and stack pressure increase air flows. If the stack pressure at highest stories is low, special arrangements may correct the situation (Fig. 10). If these arrangements are impossible, mechanical ventilation could bc considered for these stories.

5. Conclusions An important source of uncertainty in both simple and detailed (i.e., multi-zone) ventilation models used to predict passive cooling is related to the attribution of discharge coefficients. The present study was undertaken to improve our knowledge on velocity and discharge coefficients when measm'ed in real buildings. The experiments were perfl)rmed on a naturally ventilated three-level office building where the slaircase acted as exhaust chinmey. In order to keep the flow pattern stable, a condition for air flow measurements to be reproducible, the experiments were perfl)rmed on windless nights where the flow was only driven by stack pressure. Air lal,;~wpatterns were traced with smoke and tracer gas. In a first set of experiments, air velocities, contraction and velocity coefficients and the position of the neulral pressure level ( N P L ) have been measured, in a second set of experiments, the resulting effective area of a combination of two openings in series. The nature of the phenomenon of natural ventilation makes the validation of models difficult whether they are simple or complex [4,5.7,8]. The precision in the experiments of this paper is of the order of 20% to 25c/,. The experiments performed in a real building to determine flow coefficients, confirm the values found in the literature obtained in laboratory experiments. The air velocity and the lracer gas air flow rate measurements conducted to determine the discharge coefficient give the same results. The velocity coefficients ~#:=0.7 ± 0. I and .jet contraction coeflicients = 0.85 + 0.1 found in the experiments are in agreement with the generally accepted value of the discharge coeflicienl C j=q~ ~ = 0 . 6 ± 0 . 1 . The results has been obtained for slightly lurbulenl flow which was driven by a steady stack pressure. In the presence ol wind the driving presst, re will be fuctuating and the flow pattern unsteady. Although it is expected that wind will increase both friction and contraction (decreasing the apparent C u value), it was flmnd that the uncertainty in the flow measurements increased so much, that wind efl'ect~.,could not bc determined. Finally, design guidelines for ventilative cooling are given to illustrate, that simple models for natural ventilation can give valuable information to the designer. The imprecision ol the model might be a minor issue if the model gives a qualitatively correct answer, especially when account is taken ot the uncertainty of input data [4]. Future research should further explore the possibilities of using qualitative' ntodeling

291

and qualitative simulation techniques to model natural ventilation.

6. List of symbols A Cu g H L ,) N

Surface area Discharge coefficient gravitational constant Distance between opening centers l,eakage mass flow ra~e Mass flow rate Kinetic energy (Coriolis)

in 2 ]

mts~'] m] kg/s] kg/s]

-1

coefficient p 7' tt t: W : Ap ~: ( ~ p q~

Pressure Temperature Air velocity Volume flow rate Opening width Height Pressure difference Jet contraction coefficient Frictional pressure drop coeflicient. Local pressure drop coefficient Air density Velocity coefficient

Pal K, °C] m/sl m~/s] In ] I11 ]

Pal ......

I

kg / m 3 I ....

Suffixes b c eff l i inl

Bottom External Effective top Internal Intermediate opening

Acknowledgements The experiments presented in this paper were performed within the framework of several research projects: a research project on Passive Cooling sponsored by the Swiss Federal Office of Energy (OFEN contract EF CO 91011 ) ; as part of the Swiss contribution to the European project PASCOOL, part of the JOUI,E 1I program, sponsored by the Swiss Federal Office of Education and Science ( OFES contract 93.(i)09).

References I ] B. Kuipcrs. Qualitative Simulation. The MIT Press, 1994. 2 ] CIBSE Guide, Air Infiltration and Natural Ventilation, Vol. A, Section A4, 1986. 3] ASHRAE Handbook--1985 l=undanlcntals, lnliltration and Ventilalion, Chap. 23, Atlanta. 1985.

292

F. l"lou:entzou et al./ Energy and Building.s 27 (1998) 2&7-292

[41 J.-M. Ftirbringer, Comparison of the accuracy el detailed and simple models of air iniiltration, 15th AIVC Conference, Buxton, UK, 1994, pp. 719-730. 15l C.-A. Roulet, J.-M. Hirbringel, R. Borchiellm . Evaluation el the muhizone air flow simulation code COMIS. I,?.~omvent '96, 5th Conference on Air Distribution in Rooms, Yokohama, Japan, July 17 19, 1996. [6] R.M. Aynsley, W Melbourne, B.J. Vickcry. Architectural Acre dynamics, Natural Ventillalion. (7hap. b. Applied Science Publ. Londtm+ 1977. [ 7 ] M. Liddament, AIDA+ an air intiltration development algorithm. Air Inlihration Review. 11/I, 1989, pp. I(}-IZ [ 8 ] L. Palmiter, Mass Flow for AIDA, AIR 11/2+ 19¢)1). [91 F. Allard, I-:. Haghighat, G. Lidbecq, R. Pelletrct, D. Bienfait. 1.. Vandaele, R. Walker, Airllow through large openings m buildings, J. Van der Maas ( Ed. i, Technical Reporl of IEA/ECB-Annex 21): Air Flow Patterns Within Buildings. EPFI,/I,ESO-PB, CH-1015 Lausanne, Switzerhmd. 1992. 1101 J.LM. Hensen, On the thermal interaction of buihling structure and heating and ventilation systcm. PhD Thesis, F.indht~veu Uni,,ersily ol Technology (FAG() ), 1991, [ 1 I ] V. Dorer, A. Weber, Multizone air flow model CO I I \ e u - l r n s y s , Trnsy', Type 57 Documentation. II'~A.ECB Annex 23 technical reporl. EMPA 175, 1994. 112] J. Van der Maas, C. A. Roulet, Night time ventilation by stack effecl. ASHRAE Transactions 97, Part 1. 1991, pp. 516- 524. [ 13] J. Van der Maas, C.-A. Roulet. Multizone cooliug inodel for calculating the potential of night lime ventilation. Prec. el the 14th AIV(' Conl;erence. Copenhagen, 1993, pp. 567-586.

14 ] I.E. ldelchik, Handbook o! Hydraulic Resistance, 3rd edn.. CRC l~res:,, FL, 1994. 151 J. Van der Maas. F. fqourentzou, J.-A. Rodriguez, P. Jaboyedolf. Passive cooling by nighl ventilation. European Conllereuce on Energy Performance and Indoor Climate m Buildings. Vol. 2, Lyon. November 1994, pp. 646-651. 16] K. Liman, F. Allard (fzds i. Ventilation and thcrmal mass subtask, Final Report, PASC()()L Research Project. EC DG Xll, Bru,,sels. 1995. 17] C. A. Roulet, J. Van del Maw,, 1:. tqourentzos. Application of passive convective cooling to buildings, Prec. of ludo,~r Air "96 7'h International Conf. on lnd~ot Air Quality and Climate. \,ol. 3, 199(~. pp. 241-247. 81 C.-A. Roule/, J. Van dcr M:uis. J:. Flourentzos, A planlling tool I,)r passive cooling ~ll buildings, Prec. of Indoor Air "9,5, 7th International Conf. on Indoor Air Quality ztlld Climate. Vol. 3. 19q'b, pp. 265-27 I. 191 F. Florentzo. J. Van der Mare, C. A. Roulet, LESOCOOI, 1.1},Thermal simulation program for building,,: a tool lkw the evaluation o1: the risk of overheating and the pmential of co(,ling through ventilation, available at I,ESO, EPF[.. ('H-]O15 l.,ausanne {see Website blip:;/ lesowww.eplt.ch/), 199(',. 121)1 Ir~,ing, Concannon, Dhargalkar. An inverse solver I'~r sizing passive ventilaliou openings, Proceedings of (71BSE National Conli~i*_'nce. Eastbourne, 1995. 1211 C.-A. Roulet, I,. Vantlaelc. Air flow pattern-, within buildings, measurement techniq~aes. "lcuhnical Note AIVC 34. Air lnfiltratiun and Ventilation Centre. Cox cntr 3 . 1991. [ 22 I J. Van der Maas. J.LM. tteuscn, A. Roos, Ventilation and energ,, llow throuBh large vetical ~)penmgs in buildings. Prec. el the 15th AI\;(" ( nnfcrence, Buxton. I!K. 10~)4.