Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands
The Co l l e c t i o n Ef f i c i e n c y of a Ma s s i v e Fog Co l l e c t o r ROBERT S. SCHEMENAUER and PAUL I. JOE Atmospheric Environment Service, 4905 Duf[erin Street, Downsview, Ontario M3H 5T4 (Canada) (Received December 29, 1988; accepted after revision May 22, 1989 ) ABSTRACT Schemenauer, R.S. and Joe, P.I., 1989. The collection efficiency of a massive fog collector. Atmos. Res., 24: 53-69. Very large (48 m 2) fog-water collectors are being used on the coastal mountains in northern Chile to generate water. The microphysical characteristics of the high elevation fog (camanchaca) have been examined and the collection efficiency of the collectors measured. The camanchaca exhibits characteristics of clouds, reflecting its source as a marine stratocumulus deck. Droplet mean volume diameters (MVD) in ten cases ranged from 10.8 to 15.3 #m. Droplet concentrations were typically 400 cm -z with fog liquid water contents ranging from 0.22 to 0.73 g m -3. The large fog-water collectors consist of a double layer of mesh made from a 1-mm wide flat polypropylene ribbon. The theoretical collection efficiencies of a 1-mm wide ribbon, for droplets with the observed MVD, at wind speeds from 2 to 8 m s -1, are 75 to 95%. The field measurements of the collection efficiency of the mesh at the centerline of a large collector gave values of ~ 66% (3.5-6.5 m s-l ; 11 #m MVD). This is in good agreement with the theoretical value for a single ribbon once the areal coverage of the mesh is taken into account. At lower windspeeds, the mea- sured collection efficiencies dropped to ~ 26% (1.9 m s- 1; 15/gin MVD). A simple parameteriza- tion of the mesh collection efficiency allowed some properties of meshes to be examined, e.g. the mesh shows a marked decrease in droplet collection as the ribbon width is increased while main- taining a constant percentage areal coverage. The measured water output from the large collector was 2.9 times lower than predicted using the measured amount of water removed at the centerline and the wind speed 6 m upstream. This implies a large-collector efficiency of only ~ 20%. This low value may result from a lowering of wind speed as the fog approaches the mesh, a reduced collection efficiency away from the center- line, and water losses in the system. RESUME Dans les montagnes c6ti~res du nord du Chili, on se sert, pour obtenir de l'eau, d'immenses capteurs d'eau de brouillard (48 m2). On a examin~ les caract~ristiques microphysiques du brouil- lard de forte altitude (camanchaca) et mesur~ l'efficacit~ de captage. Le camanchaca pr~sente des caract~ristiques de nuages, en refl~tant la source sous forme de plate-forme de stratocumulus marin. 0169-8095/89/$03.50 1989 Elsevier Science Publishers B.V. 54 R.S. SCHEMENAUER AND P.I. JOE Dans dix cas, le diam~tre de la masse moyenne (DMM) des gouttelettes s'est situ6 entre 10,8 et 15,3/~m. La concentration type des gouttelettes dtait de 400/cm 3, la teneur en eau du brouillard allant de 0,22 h 0,73 g/m 3. Les grands capteurs d'eau de brouillard consistent en une double couche de filet en ruban plat de polypropyl~ne de 1 mm de large. La capacit~ th~orique de captage d' un ruban de 1 mm de large, pour des gouttelettes du DMM observe, par un vent soufflant de 2 h 8 m/s, est de 75 h 95%. Sur le terrain, les mesures de la capacitd de captage du filet h la ligne mddiane d' un grand capteur ont donn~ des valeurs d' environ 66% (3,5 h 6,5 m/s; DMM de 11/~m). Une fois qu'on tient compte de la couverture surfacique du filet, ce chiffre correspond bien h la valeur th6orique affdrente h un seul ruban. A des vitesses plus basses du vent, la capacit$ mesurde de captage est tomb~e h environ 26% ( 1,9 m/s, DMM de 15/Ira). Le simple ~tablissement des param~tres de la capacitd de captage du filet a permis d'examiner certaines propridtds du filet. Par exemple, le filet recueille beaucoup moins de gouttelettes quand la largeur du ruban s'accro]t, la couverture surfacique en pourcentage ~tant constante. La production mesur~e d'eau du grand capteur a 6t~ 2,9 fois plus basse que la production prdvue d'apr~s la quantit~ mesur~e d'eau enlev~e h la ligne m~diane et h la vitesse du vent h 6 men amont. Le grand capteur aurait donc une capacitd de captage de seulement 20% environ. Cette basse valeur r~sulte peut-~tre de la baisse de la vitesse du vent quand le brouillard s'approche du filet, d' une baisse de la capacitd de captage quand on s'dloigne de la ligne mddiane, et des pertes d'eau de l'installation. INTRODUCTION The demand for fresh water is currently a major political, social and eco- nomic issue in t he world. Predictions are t hat t he problems will continue to grow more serious as populations increase and conventional water supplies are depleted or contaminated. Faced with a growing demand and a depleting sup- ply, we have to be prepared to explore unconvent i onal sources of water. The high elevation coastal fogs along t he west coast of Sout h America are one such source. A high-pressure area is present in t he Pacific Ocean off t he west coast of Sout h America t hroughout the year. The trade wind inversion produced by subsidence in t he anticyclone is found along t he coasts of Peru and nort hern Chile at heights t hat gradually decrease towards t he south. Typical heights are in t he 600 m to 1200 m range. The inversion caps t he vertical devel opment of t he extensive fields of marine stratocumulus found over t he ocean ( Schemen- auer et al., 1988). The cloud decks are typically 100 m to 400 m thick and do not produce rain, t hough occasional drizzle is experienced. The stratocumulus decks are blown onshore by a prevailing southwest wind along t he nort hern coast of Chile. Where t he coastal mount ai ns are at an appropriate height, they intercept t he clouds resulting in persistent periods of fog. These high elevation fogs are called camanchacas. Kerfoot (1968), Goodman (1982) and Schemenauer (1986) have reviewed t he literature on fog water collection by vegetation and small collectors. They concluded t hat in certain areas t he interception of fog water can provide an i mport ant i nput to t he ecosystem. A history of such work in nort hern Chile led THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 55 to the est abl i shment in 1987 of The Camanchaca Project. It is a combined research and operational proj ect designed to investigate t he meteorological conditions leading to t he formation of t he camanchaca, t he microphysical properties of t he camanchaca, t he opt i mum collector design and construction, and the delivery of water to a village of 330 people 6 km away. The mai n fund- ing agency is t he Int ernat i onal Development Research Centre in Ottawa, Can- ada. The scientific participants are t he University of Chile, t he Pontifical Catholic University of Chile and t he Atmospheric Envi ronment Service of En- vi ronment Canada. The collector and pipeline construction is supervised and carried out by t he Corporacion Nacional Forestal in t he 4th Region. The choice of opt i mum sampling locations is made by conducting prelimi- nary experiments on relative fog collection using small collectors (Schemen- auer et al., 1987; Cereceda et al., 1988). Thi s has resulted in t he siting of fifty large 48-m 2 collectors (atrapanieblas) and t he generation of substantial amount s of water (Schemenauer and Cereceda, 1988; Schemenauer, 1988). This paper will describe t he results of in-situ measurement s designed to deter- mine t he collection efficiency of t he atrapanieblas. Knowledge of t he collection efficiencies is essential for optimizing collector design and for minimizing water costs. THE FIELD SITE The field site is 60 km nort h of t he city of La Serena in north-central Chile. The mai n experimental location is on a ridge at 780 m (2926' S 71 15' W). The ridge extends in a nor t h- sout h direction for about 5 km and is flanked on either end by l l 00- m mount ai ns. It is 6 km from t he coast of t he Pacific Ocean where a small fishing-village, Chungungo, is located. Meteorological stations have operated continuously since November 1987 at 780 m and 720 m recording a st andard set of meteorological paramet ers as well as t he flowrates from t he collectors. During t he 2-week intensive field pro- grams in 1987 and 1988, cont i nuous meteorological data were also collected at elevations of 30 m and 1100 m and frequent radiosonde ascents were made. I NSTRUMENTATI ON Each of t he 50 atrapanieblas is 12 m long and 4 m high. The base of t he mesh varies from 1 m to 2 m above ground depending on t he undul at i ons of t he terrain. The actual collector studied in this paper is illustrated in Fig. 1. It is on t he crest of t he ridge with a few eucalyptus trees at a distance of 25 m or more on t he downwind side. The collecting material is a double layer of black polypropylene mesh (Fig. 2) t hat is made in Chile. The mesh is a triangular weave of a flat fiber about 1 mm wide and 0.1 mm thick. The fiber is woven into a mesh with a pore size of about 1 cm. The double layer of mesh can cover 56 R.S. SCHEMENAUER AND P.I. JOE Fig. 1. A 12 4 m 2 fog-water collector on t he ridge at El Tofo, Chile. The PMS F SSP probes are in t he cent er and t he meteorological t ower is on t he right. up to ~ 70% of t he surface area of t he collector depending on how t he fibers overlap. A complete set of meteorological measurement s is made 6 m in front of t he atrapaniebla at t he centerline height (3.5 m). Measurement s of t he characteristics of t he fog-droplet sizes and concentra- tions were made with two Particle Measuring Systems Forward Scattering Spect romet er Probes (FSSPs). Bot h FSSPs were equipped with aspirators to pull the droplets t hrough t he measuring section at a const ant 25 m s - 1 One FSSP was mount ed immediately in front of t he atrapaniebla but moved 0.5 m horizontally off center. The second FSSP was mount ed behi nd t he atrapanie- bla but 0.5 m horizontally off center in t he other direction. Measurement s were made at 3 heights from j ust below t he center of t he atrapaniebla to j ust above, with each position separated by 0.5 m. Both FSSPs underwent electronic checks and bead calibrations before, during and after t he two-week field proj ects and in addition were operated side by side at times to compare t he spectra. The droplet sizes and concentrations were recorded each second. Normally the probes were operated with a nomi nal channel width of 2/ l m for each of the 15 channels. Occasionally this was changed to nomi nal channel widths of 0.5, 1 or 3/~m to examine parts of t he droplet spect rum in more detail. THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 57 Fig. 2. The Rashel l mesh as used i n t he Camanchaca Proj ect. A double layer is shown. The fiber is 1 mm wide and 0.1 mm t hi ck. DROPLET- SI ZE DI STRI BUTI ONS The camanchaca droplet-size distributions almost always had a single peak with a mode in the 10 to 14 ttm range. Maximum droplet sizes only rarely exceeded 30 ttm and droplets of this size were contributing little to the fog liquid water content (LWC). Table I presents the droplet mean volume diameters (MVD) and concen- trations from the front FSSP for ten cases. The MVD varied from 10.8 to 15.3 ttm reflecting different cloud base heights or updraft velocities and thus differ- ent growth times for the droplets. The droplet concentrations on the other hand, except for the 1987 case, remain fairly close to 400 cm -3. Examples of the droplet spectra are shown in Figs. 3 and 4. Fig. 3 shows the spectra from the front and back FSSPs for the 12 November 1987 case. The droplet concen- trations are dramatically reduced on the rear side of the mesh for all sizes except the very largest. Few large droplets occurred either in front of or behind the mesh. The average concentration in front of the mesh was 231 cm -3 and behind the mesh 73 cm- ~. Fig. 4 is a similar plot for the first case of 9 November 1988. The reduction in droplet concentrations is not so marked as in the pre- vious case but it is similar in t hat the largest reductions take place in the mid- 58 R.S. SCHEMENAUER AND P.I. JOE TABLE I The collection efficiency of t he mesh for 10 sets of measurement s near the center of the collector; t he sample duration (t), liquid water cont ent (LWC), efficiency (Era), droplet mean volume diameter (MVD), droplet concent rat i on and wind speed are given Date Position t LWC (g m -a) Em Front spectrum Wi nd number (s) (To) ( ms -1) front back MVD (zm) conc. (cm -3) 12 Nov. 87 3 20 0.31 0.098 69 12 231 6.5 4 Nov. 88 3 100 0.31 0.099 68 11.5 383 3.5 4 Nov. 88 3 100 0.22 0.071 67 10.8 301 3.5 4 Nov. 88 3 100 0.32 0.10 68 11.1 406 3.5 9 Nov. 88 4 300 0.68 0.50 27 14.4 477 1.9 9 Nov. 88 4 300 0.72 0.45 37 14.6 435 2.6 9 Nov. 88 4 300 0.73 0.46 36 14.9 419 2.6 9 Nov. 88 5 300 0.66 0.38 43 14.6 408 3.2 9 Nov. 88 5 300 0.73 0.36 51 15.3 384 3.1 9 Nov. 88 5 300 0.68 0.31 55 15.2 366 3.4 f , - 0 i i i i i 1 ~ i 4 6 8 1 0 ~.2 i 4 t 6 1 8 2 0 O r o p l e t D i a m e t e e {pm} " 0 m 2 2 2 4 2 6 2 ( ] 3 0 Fig. 3. The fog droplet-size distribution in front of t he mesh (squares) and behind the mesh (crosses) on 12 November 1987. THE COL L E CT I ON EF F I CI ENCY OF A MASSIVE FOG COLLECTOR 59 2 1 . 9- i . 8 - i , 7- t . 5- 1 . 5 - 1 . 4 - ~ 1.3- ~ 1.2- p-- ~ 0 . 9 0.7 ~ 0.5 ~-1 0.5 0.4 0.3 (?.2 0. t i 2 s ~o ~4 ~a ~ ~6 30 Oroplet Diameter ( p m) Fig. 4. The fog droplet-size distribution in front of the mesh (squares) and behind the mesh (crosses) on 9 November 1988. si ze ranges. The average dropl et c onc e nt r at i on on t he f ront F S S P was 477 cm - 3 and on t he rear F S S P 291 c m - 3. FOG LIQUID WATER CONTENT The f og LWC me as u r e me nt s i n f ront of and b e hi nd t he at rapani ebl a are gi ven i n Tabl e I. The i nc omi ng c amanc hac a had LWC val ues f rom 0. 22 t o 0. 73 g m - 3 for t he t e n cases. The s e val ues are t ypi cal of t he l ower or mi d- l evel s of c u mu l u s ( S c he me nau e r and Isaac, 198 4) and are cons i derabl y above what J i u s t o ( 198 1) report s for mari ne or c ont i ne nt al s urf ace- bas ed fogs. Thi s i s reasonabl e, gi ven t hat t he camanchaca resul t s f rom mari ne s t rat ocumul us cl oud decks bei ng advect ed over t he ridge by t he sea breeze. The f act t hat t he LWC i n t he c amanc hac a can be 0. 7 g m - 3 or hi gher i s of maj or i mport ance i n est ab- l i s hi ng t he wat er avai l abi l i t y on t he mou nt ai n. LWC val ues t hi s hi gh or hi gher are s upport ed by adi abat i c LWC cal cul at i ons on days wi t h l ow ( 1000 mb ) and warm ( 20 C) cl oud bases. Fig. 5. s hows how t he LWC was di s t ri but ed as a f unc t i on of si ze for t he case of 12 No v e mb e r 1987. It i s very hi ghl y concent rat ed near t he peak i n t he drop- l et s pect rum around 12/ ~m. Dropl et s < 8 / ~m and > 18 / ~m di amet er were con- 6 0 R.S. SCHEMENAUER AND P.I. JOE 0 12 0 11 0 1 0 0 9 " ~ 0 08 E ~ " 0 0 7 ( q 0 05 I E 0 0 5 - 0 . 0 4 - r J " ~ 0 0 3 - J 0 . 0 2 - 0 01 0 ,,~ 2 T I I I I I I I I T ~" " r ~" 4 6 8 10 12 14 ~6 ~B 20 22 24 26 28 30 D r o p l e t D i a m e t e r {pm) Fi g. 5. The di s t ri but i on o f l i qui d wat er c ont e nt i n f ront o f t he me s h as a f u nc t i on o f dropl et di amet er for t he case o f Fi g. 3. tri buti ng very little to t he camanchaca LWC i n t hi s case. Thi s clearly i ndi cates t hat any fog-water collector i n use at t he si te wi l l have to be particularly effi- ci ent i n removi ng droplets in the 10 to 16 # m di ameter range. ME A S U R E D ME S H EFFI CI ENCY The effi ci ency (Era) of the doubl e layer of nyl on mes h in removi ng t he ca- manchaca LWC is present ed i n Tabl e I for each of t he 10 cases. Em is si mpl y the difference bet ween t he front and back LWC val ues expressed as a percent- age of t he front LWC. There is considerable vari ati on i n t he values, from 26% to 69%. It shoul d be clearly not ed t hese val ues refer to t he overall measured effi ci ency of t he mes h and not to t he value for a si ngl e ribbon or a si ngl e droplet size. The most l i kel y meteorol ogi cal variable to be i nf l uenci ng t he col l ecti on effi ci ency is t he wi nd speed, and i n Fig. 6 t he col l ecti on effi ci enci es for the first four cases in Tabl e I are pl otted versus t hi s parameter. It appears t hat there is onl y a sl i ght dependency on wi nd speed, from 3.5 to 6.5 m s -1, for t he l l - / l m MVD droplets. The si x cases of 9 November 1988 from Tabl e I are pl otted i n Fig. 7. For wi nd speeds of 1.9 to 3.4 m s -1, and 15-/~m MVD droplets, there is a strong dependency of col l ecti on effi ci ency on wi nd speed. The col l ecti on ef- fi ci enci es for t he l l - # m MVD case are probably good to + 5% si nce a side by THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 61 1 0 0 - 9 0 - 8 o - ~ 70- .- IIJ . , . 4 6 0 - ~ 50- C o 4 0 - " ~ 3 0 - ID ' ~ 20- 10- 0 T I I T - - [ M - - 2 4 6 Wind Speed (ms - 1 ) 11 pm Fi g. 6. The meas ured col l ect i on e f f i c i e nc y o f t he me s h at t he cent er of t he large col l ect or as a f unct i on o f wi nd s peed for t he four cas es f rom Tabl e I wi t h dropl et MV D of ~ 11 pm. 1 O0 90 I ~ 80 70 C "'~ 60 2 ~- 5o t - O 40 .r,4 .IJ ~ 3o ~ 2o 10 0 1 5 prn i i i i i [ i 2 4 6 Wind Speed (ms - 1 ) Fi g. 7. The meas ured col l ect i on e f f i c i e nc y o f t he me s h at t he cent er o f t he large col l ect or as a f unct i on o f wi nd s peed for t he s i x cas es f rom Tabl e I wi t h dropl et MV D of ~ 15 Mm. side intercomparison of the two probes was done under these conditions. Thi s was not possible for the 15 pm case and thus the least squares fit to the data poi nts could ultimately have to be shi fted _+ 10% when another data set is available. The linear regression equations for the two sets of data are: 62 R.S. SCHEMENAUER AND P.L JOE Em =6 6 . 3 + 0 . 4 1 ws; N=4 ; MV D = l l / ~m ( 1) Em = - 6 . 9 + 1 7 . 3 ws; r = 0.94; N=6; MVD = 15/ira ( 2) where Em is the collection efficiency in percent and w~ the wind speed in meters per second. The equations are only valid for the wind-speed ranges noted above. The col l ecti on-effi ci ency calculations in eqs. 1 and 2 are valid for a range of wind speeds that is representative of the site. Fig. 8 shows a typical diurnal variation in wind speed. The sea breeze is very strong in the afternoon with typical peak values of 5 to 8 m s - 1. At night the air is someti mes calm but more typically the winds are 1 to 2 m s - 1 from the east. In the case shown, the three peaks in wind speeds between 00h30 and 09h30 were associated wi th shifts in wi nd direction from east to west. The measured centerline mesh efficiencies are reasonably high and indicate a strong dependence on wi nd speed at low speeds. It is not clear i f there is a dependence on droplet size since the data were collected in different wind- speed ranges. Thi s will be discussed further below. The maxi mum achievable large-collector effi ci ency is given by the percentage area covered by the ribbon i f the ribbon i t sel f has a collection effi ci ency of 1. In the case of the atrapan- iebla, thi s value is < 76%, since each layer of mesh can cover a maxi mum of ~ 38%. The actual field values are not far of f thi s theoretical maximum. Thi s implies that the ribbons themsel ves indeed have a high collection efficiency. The field data support the concept of siting the collectors in locations with higher wi nd speeds and, as we will see below, the theoretical calculations result I 0 o " 0 ( 3 . C 8 - 7- 5- 5- 4 - 3 - 2 - 0 1 i i i i i i f t 4 3 0 t 6 3 0 t 8 3 0 2 0 3 0 2 2 3 0 30 2 3 0 4 3 0 6 3 0 8 3 0 t 0 3 0 Time i t 2 3 0 1 4 3 0 Fig. 8. The diurnal variation of wi nd speed on the ridge at E1 Tofo starting at 14h30 local t i me on 12 November 1987 and endi ng approximately 15h30 on 13 November 1987. THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 63 in higher collection efficiencies for larger droplet sizes. Thi s implies working on t he crest of a ridgeline or in a not ch in t he ridgeline and at altitudes well above the base of t he cloud deck generating t he fog. COLLECTOR FLOWRATES Using t he collector area (48 m2), t he measured amount of LWC removed from Table I, and t he wind speed, it is possible to calculate t he expected out put from t he atrapaniebla. Thi s is compared in Table II to t he measured out put for seven cases. The three cases on 4 November 1988 have been omi t t ed be- cause t he fog event was j ust starting and t he out put flow had not stabilized. In every case in Table II t he calculated flowrate is higher t han t he measured flowrate from t he atrapaniebla. On average, t he calculated value is 2.9 times too high. It is probable t hat the upst ream wind speeds overestimate t he speed at t he surface of t he mesh. It is also possible t hat the collection efficiencies across t he mesh are lower t han at the centerline. The wind-speed values used in Table II come from measurement s for t he same time periods as for t he LWC data but they were made 6 m upst ream of t he atrapaniebla. Some reduction in wind speed would be expected as t he air approaches t he atrapaniebla. This will result from t he obstruction to t he mai n flow and t he drag of t he mesh. There will also be changes in t he angle of attack of t he wind over t he surface of t he mesh. Until detailed wind measurement s can be made in t he vicinity of an atrapaniebla, and more extensive collection-efficiency measurement s are available, the empirical factor (2.9) will have to be used to characterize the difference between measured and calculated flowrates. Thi s does not affect eqns. 1 and 2 which correlate the measured collection efficiency to t he up- stream wind-speed measurements. Water losses in t he collection system are felt to be low for the atrapaniebla TABLE II A compari son of t he calculated and measured flowrates from t he 48-m 2 at r apani ebl a Dat e LWC removed W. Fl owrat e ( gm -3) ( ms -1) calc. meas. (cm a s -1) (cm 3 s -1 ) 12 Nov. 87 0.21 6.5 65.5 18.5 9 Nov. 88 0.18 1.9 16.4 5.1 9 Nov. 88 0.27 2.6 33.7 12.1 9 Nov. 88 0.27 2.6 33.7 17.1 9 Nov. 88 0.28 3.2 43.0 24.2 9 Nov. 88 0.37 3.1 55.1 14.7 9 Nov. 88 0.37 3.4 50.4 20.7 64 R.S. SCHEMENAUER AND P.I. JOE studied. Cert ai nl y no more t han ~ 10% of t he collected wat er could be lost due to design problems or leaks in t he t rough and pipes. Thi s t herefore cannot be t he expl anat i on for t he large difference not ed above bet ween t he calculated and measured values. Similarly, an error in t he fl owmet er measurement s ( + 5% ) or wi nd speed ( + 5% ) cannot explain t he difference. THEORETICAL CONSIDERATIONS In this section, t he comput ed collection efficiencies for ribbons from Lang- mui r and Blodgett (1946) are applied to det ermi ne t he effectiveness of t he collectors. The t heory assumes pot ent i al flow about an isolated ribbon and considers only inertial impaction. Meshes and filters are frequent l y viewed as an assemblage of i nt eract i ng but separate collectors (Spielman, 1977). How- ever, t he description of t hese i nt eract i ng flow fields around such assemblages and t he paramet eri zat i on of t hei r effects have not been properly addressed. First we exami ne t he collection of individual ribbons. Fig. 9 shows t he col- lection efficiency of a ribbon for wi nd speeds of 2 and 8 m s- 1 and for droplet sizes of 11 and 15 Hm as a funct i on of ribbon width. The results are what one would expect from inertial i mpact i on (Langmui r and Blodgett, 1946). As t he ribbon width increases, wi nd speed (droplet velocity) decreases or droplet size decreases, all of which decrease t he droplet inertia, t he efficiency decreases and t he droplet t ends to move wi t h t he air flow around t he ribbon. A 1-mm wide ribbon, however, collects 11- and 15-Hm di amet er droplets very efficiently (90 to 95% ) when t he wi nd speed is 8 m s -1. 1.0 0 . 9 0 . 8 0 . 7 U 0 . 6 C L I 0.5 0.4 0. 3 0 . 2 - 1 o 1 1 p m , S m s 1 - 1 5 p r o , 2 r n s - 1 0 . 1 o 1 1 p r o , 2 m s - 1 0 , 0 i [ [ r ~ i I l I 2 4 6 8 1 0 Ri bb0 n Width ( m m ) Fig. 9. The collection efficiency of a flat ribbon for wind speeds of 2 and 8 m s -1 and for droplet sizes of 11 and 15 Hm as a function of collector width (after Langmuir and Blodgett, 1946). THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 6 5 In Fig. 10, t he col l ecti on effi ci ency is mul ti pl i ed by t he wi dt h of t he ribbon to comput e t he amount of water t hat a si ngl e ribbon mi ght be expect ed to be able to collect. Even t hough t he effi ci ency drops of f wi t h ribbon wi dth, t he col l ecti on area i ncreases faster wi t h t he result t hat water col l ecti on i ncreases wi t h i ncreasi ng ribbon wi dth. Note, however, t hat beyond a certain wi dth, t he col l ecti on approaches a limit. Fig. 10 shows that, i f you coul d onl y put up one ribbon, a 10-ram wi de ribbon woul d generate more water t han a 1- mm wi de ribbon i n the fog condi t i ons on El Tofo. The effect is more pronounced at hi gh wi nd speeds ( 8 m s - 1) t han l ow wi nd speeds (2 m s -1 ) and for larger droplets ( 15/ ~m) t han smaller drop- l ets ( 11/~m ). If you put t he same ar ea of ribbon in a fog t hough, t en 1-mm wi de ribbons woul d generate more water t han one 10- mm wi de ribbon. Fig. 10 hel ps expl ai n why eucal yptus trees, wi t h l ong l eaves 1 to 2 cm wide, are good fog water collectors on E1 Tofo. The trees reach hei ght s of 15 m or more and t he l eaves are exposed to fog droplets travel l i ng wi t h hi gher wi nd speeds t han are measured near t he surface. Thi s is a regime in whi ch wi de collectors perform effectively. The comput at i on can be ext ended to consi der mul ti pl e ribbons in a mesh. The confi gurati on of t he ribbons formi ng t he mesh is not speci fi ed and there- fore t he results are onl y grossly correct. The mesh coul d be a si ngl e layer or t wo layers t hat are t ouchi ng. An assembl age of ribbons is assumed whi ch forms some sort of matrix; t hi s assembl age is parameteri zed by the fraction of the collector cross- secti onal area t hat is obscured by the col l ecti ng surface. Fig. 11 shows t he results for a mes h t hat covers 50% of t he collector' s cross-secti onal 7 ~ 6 15 Jam, 8 m s - 1 !:= o 11 t ~ m, 8 m s - 1 t - - + 15 p m , 2 r n s - 1 / * ~ 4-1 a 11 } J m, 2 m s - 1 5 C 0 4 I - I . , - I r r 3 X > , 2 - ( D .,-q ~2 . , - t 1 0 2 4 6 8 ~0 Ri b b o n Wi d t h (m m ] Fi g. 10. The product o f t he col l ect i on e f f i c i e nc y ( E) and t he ri bbon wi d t h ( D ) as a f unct i on o f ri bbon wi d t h for t wo dropl et s i z e s and t wo wi nd speeds. 66 R.S. SCHEMENAUER AND P.I. JOE 0.6 0,5 0. 4 Z X O 0. 3 x 0.2 0.1 + 15 p rn , 8 m s-l-11 D 1 I p r n . 8 m s _ . z~ 15 p ro. 2 m s . o !1 ~m, 2 m s 2 4 6 8 Ri bbon Wi dt h {mm} 10 Fig. 11. The product of the collection efficiency (E), the ribbon width (D) and the number of ribbons (N) (to maintain 50% areal coverage) as a function of ribbon width. area. A single st rand of 1-mm ribbon in a 1-cm triangular configuration (such as the Rashell mesh in Chile) obscures about 38% of t he area. A double layer would obscure < 76% of t he area since some overlap of strands is inevitable. Therefore, 50% coverage is not an unreasonable assumption. Fig. 11 shows t hat if the number of ribbons is such t hat an areal coverage of 50% is main- tained, t hen t he amount of water collected by a mesh decreases with increasing ribbon size. Therefore, many small ribbons in a mesh are more effective t han a few large ribbons for collecting water. As t he collector size (ribbon width) gets very small, t he collection efficiency approaches 1 and N gets very large. The mesh with 50% coverage will t hen remove 50% of the fog droplets ap- proaching it. Though, for a ribbon width equal to zero, t he collection would be zero. For t he case of an atrapaniebla with a 50% coverage of l -ram wide fibers, Fig. 11 predicts t hat t he mesh will collect ~ 11% more 15-/1m diameter droplets t han ll-/~m diameter droplets when t he wind is 2 m s - ' . One should be very careful in applying this figure, in t hat t he ribbon geom- etry within the mesh and t he detailed flow fields at t he ribbon have been ig- nored. The parameterization of droplet collection by meshes is, however, an i mport ant subject and will receive cont i nued attention. Fig. 12 is an adaptation of t he collection efficiency versus ribbon-width data in Fig. 9. In Fig. 12, t he collection efficiency of a l -ram wide ribbon, for 11-/~m diameter droplets, is shown as a function of wind speed for two cases. A very simple way of t hi nki ng of t he large collector {atrapaniebla) is as a cross-sec- tional area which has a certain fraction covered by 1-mm ribbons. The product THE COLLECTI ON EF F I CI ENCY OF A MASSI VE F OG COLLECTOR 67 1. 0 O g t 0 . 8 0 0 . 7 r J l ~ 0 . 6 f . - 0 . 5 X ~ , 0. 4 U C 0 3 U " ~ 0, 2 LI.I 0.1 E x O . 7 * E x O . 5 0.Q i , ~ r r r r i i @ 2 4 6 ~ 10 Wind Speed (m s - ~ ) F i g. 12. The product of ribbon collection efficiency and mesh areal coverage (50% and 70% ) a s a function of wind speed. of t he theoreti cal ri bbon col l ecti on effi ci ency ( E) and t he fraction of the large collector covered ( A) shoul d therefore give an approxi mat i on of the effi ci ency of t he large collector ( ELc) . Thi s is s hown i n Fig. 12 for 70% and 50% coverage. ELC rises rapidly from 0 to ~ 2 m s - 1 and t hen levels of f wi t h onl y a smal l increase i n val ue from ~ 3 to 10 m s - 1. The latter part of t he curve l ooks very si mi l ar to t he measured val ues of Em for l l - / Lm MVD droplets in Fig. 6. The measured val ues of Em for 15/ i ra MVD droplets at l ow wi nd speeds (Fig. 7) fall of f rapidly as t he speed decreases, i n qual i tati ve agreement wi t h Fig. 12. But t he drop in measured E~ occurs at hi gher wi nd speeds t han is expected. As a first approxi mati on, neverthel ess, one can t hi nk of t he center of t he atra- pani ebl a act i ng as a set of l - ram wi de ri bbons coveri ng 70% of t he cross-sec- t i onal area. The actual percentage area covered (Fig. 2) is probably l ess t han 70% but it is di ffi cul t to est i mat e si nce t he double layer of mes h is flexible and forms a non- pl anar collector whi ch bends i n moderate wi nds. The percentage area covered may i n fact be di fferent i n di fferent parts of t he atrapani ebl a and at di fferent wi nd speeds. It is encouragi ng t hat t he t rends of t he measured mes h results are consi st ent wi t h t he theoreti cal ribbon results and t hat the parameteri zed mes h calcula- t i ons provide some i nsi ght. However, more studi es, ei ther at t he site or i n a wi nd t unnel , are needed to better parameterize t he col l ecti on of fog water by t he mes h in order to opti mi ze t he collector design. 68 R.S. SCHEMENAUER AND P.I. JOE CONCLUSI ONS The camanchaca (high elevation fog) along t he nort h-cent ral coast of Chile exhibits characteristics similar to small cumulus and stratocumulus. Measured liquid-water cont ent s were 0.22 to 0.73 g m -3, substantially above t hat found in surface-based fogs. Droplet concentrations were typically 400 cm -3 again more characteristic of clouds t han of surface-based fogs. In t he t en cases stud- ied, droplet MVD ranged from 10.8 to 15.3/lm. Maxi mum droplet sizes rarely exceeded 30/lm. The collection efficiency at t he center of a 48-m 2 polypropylene mesh fog- water collector has been demonst rat ed to be strongly dependent on bot h wind speed and droplet mean volume diameter. Values as high as 65 to 70% were measured for l l - / l m MVD droplets when the wind speeds were between 3.5 and 6.5 m s- 1. Use of the measured centerline collection efficiency and t he wind speed mea- sured 6 m upstream, results in a calculated large-collector out put t hat is 2.9 times higher t han t he measured output. Thi s implies t hat t he actual average wind speed at t he mesh surface may be lower by a factor of 3 due to blockage effects of the large collector or t hat t he collection efficiencies across t he mesh are lower t han at t he centerline. The implied efficiency of t he collector as a whole is about 20%, i.e. it removes about 20% of t he fog water approaching it. Theoretical calculations of t he collection efficiency of an isolated 1-mm wide flat ribbon provide results which agree with t he measured field values in dem- onst rat i ng higher collection efficiencies for larger droplets and higher wind speeds. Numerical modelling of t he collection efficiency of a non-planar, two- layered, densely packed mesh is an extremely difficult task. However, a simple parameterization provided insight into t he role fiber width plays in t he collec- tion efficiency of a mesh with a const ant percentage areal coverage. Ultimately, more extensive field measurements, coupled with wi nd-t unnel data, should enable one to better parameterize t he collection efficiency of t he mesh and possibly to improve t he efficiency of t he collectors. The ul t i mat e goal of The Camanchaca Proj ect is to generate large quantities of fog water for use in coastal villages in Chile. As we have seen in this paper, the out put of t he collectors is going to vary depending on t he wind-speed re- gime and the droplet sizes and fog LWC at each site. Atrapanieblas such as those being used at t he El Tofo site may be applicable at most other locations with similar meteorological conditions, however, operations in extreme con- ditions such as very high or low wind speeds may require modifications to t he meshes. ACKNOWLEDGEMENTS The authors would like to acknowledge t he t remendous amount of work done by Mohammed Wasey both at t he field site and subsequently with t he data THE COLLECTION EFFICIENCY OF A MASSIVE FOG COLLECTOR 69 analysis. Walter Strapp also helped significantly with the FSSP data interpre- tation. Our t hanks go to Carol Sguigna for the preparation of the manuscri pt and Carol Wi nst on for her work on the diagrams. We owe a debt of gratitude to Guido Soto and Waldo Canto of CONAF for supervising the constructi on of the atrapanieblas and to UNESCO for providing funds for site mai ntenance and security. We would like to t hank Humberto Fuenzalida of the Uni versi ty of Chile for the design of the tower on whi ch the measurements were made and both Humberto Fuenzalida, and Pilar Cereceda of the Ponti fi cal Catholic Uni - versity of Chile for many valuable discussions during the course of the project. One of the authors ( RSS) would like to acknowledge the travel support pro- vided by the International Devel opment Research Centre ( IDRC) in Ottawa. The measurements described in thi s report were obtained during the 1987 and 1988 Camanchaca Project i ntensi ve field periods. The primary fundi ng agency for the Camanchaca Project is IDRC, Ottawa, Canada. REFERENCES Cereceda-Troncoso, P., Schemenauer, R.S. and Carvajal-Rojas, N., 1988. Factores topograficos que determinan la distribucion de las neblinas costeras en E1 Tofo (IV Region de Coquimbo - Chile). X Congr. Nac. de Geografia, 27-29 April, Santiago, pp. B1-B6. Goodman, J., 1982. Water potential from advection fog. Dept. Meteorol., San Jose State Univ., Calif., Tech. Rep., 1, 45 pp. Jiusto, J.E., 1981. Fog structure. In: P.V. Hobbs and A. Deepak (Editors), Clouds: Their Forma- tion Optical Properties and Effects. Academic Press, London, pp. 187-235. Kerfoot, O., 1968. Mist precipitation on vegetation. For. Abstr., 29: 8-20. Langmuir, I. and Blodgett, K,B., 1946. A mathematical investigation of water droplet trajectories. Army Air Force Tech. Rep. No. 5418, Washington, D.C., 68 pp. Schemenauer, R.S., 1986. Acidic deposition to forests: the 1985 Chemistry of High Elevation Fog (CHEF) project. Atmos. Ocean, 24: 303-328. Schemenauer, R.S., 1988. Fog water to quench a desert' s thirst. WMO Bull., 37: 281-286. Schemenauer, R.S. and Cereceda, P., 1988. The collection of fog water in Chile for use in coastal villages. Proc. VIth IWRA World Cong. on Water Resources, Vol. II, May 29-June 3, Ottawa, pp. 660-669. Schemenauer, R.S. and Isaac, G.A., 1984. The importance of cloud top lifetime in the description of natural cloud characteristics. J. Climate Appl. Meteorol., 23: 267-279. Schemenauer, R.S., Cereceda, P. and Carvajal, N., 1987. Measurements of fog water deposition and their relationships to terrain features. J. Climate Appl. Meteorol., 26: 1285-1291. Schemenauer, R.S., Fuenzalida, H. and Cereceda, P., 1988. A neglected water resource: the ca- manchaca of South America. Bull. Am. Meteorol. Soc., 69:138-147. Spielman, L.A., 1977. Particle capture from low speed laminar flows. Ann. Rev. Fluid Mech., 9: 297-319.