2 5 5 . 1

8 3 A P

>

'"PLICATION OF THE SLOW SAND FILTRATION PROCESS TO THE

TREATMENT OF SMALL TOWN WATER SUPPLIES

BY

B.A. BOWLES, W . M . DREW and G. HIRTH,

STATE RIVERS AND WATER SUPPLY COMMISSION OF VICTORIA.

SUMMARY The history of slow sand filters and the design considerations for apilot study are briefly described together with details of the construction andcommissioning of a pilot plant at Yarrawonga, Victoria (Study Phase I). Pre-liminary results from the first operational phase (Study Phase II), includingoperational characteristics and the effect of filtration on chemical, bacterio-logical and biological parameters are also presented. The future operation ofthe plant (Study Phase III) is discussed. The results obtained so far indicatethat, with a suitable form of prefiltration, slow sand filters are capable ofproducing a water of good quality in all respects from a raw water supplycontaining high concentrations of algae.

1. INTRODUCTION 2. HISTORICAL

Slow sand filters are used extensivelyin a number of countries such as theUnited Kingdom, the Netherlands, otherparts of Europe and some developingcountries, for the removal of micro-organisms and to improve the generalphysico-chemical properties of watersintended for domestic supplies.

In developed countries, slow sandfiltration is used only after rapidsand or microstrainer filtration;however, in many developing countriesit provides a single-step treatmentfor raw waters with turbidity notexceeding 50 NTU.

Because of the simplicity of the pro-cess and its potentially low costs ofoperation and maintenance, a pilotstudy using four small experimentalfilters is being undertaken at Yarra-wonga, Victoria, by the Ministry ofWater Resources, to evaluate theapplication of the process to Victor-ian town water supplies.

The origins of slow sand filtration arelost in the dim mists of time, but asfar as is known, the sand filter des-igned and built by John Gibb in Paisley,Scotland, in 1804, was the first suchfilter for the treatment of water foruse by the public (Baker, 1949) .Following from improvements on Gibb'sdesign by Robert Thorn in 1827, andJames Simpson in 1829, by 1851 theadvantages of sand filtration were soevident, that Parliament required thefiltration of all river water suppliedto London (Huisman & Wood, 1974) .

A most convincing proof of the effect-iveness of slow sand filtration wasprovided in 1892, by the experience oftwo neighbouring cities, Hamburg andAltona, which both drew their drinkingwater from the River Elbe, the formerdelivering it untreated, while thelatter filtered the whole of its supply.When the river water became infectedwith cholera organisms, Hamburg sufferedfrom a cholera epidemic which caused

28-1

ILine Strainer 1mm Holes

Sampling point—•

iFor FilUnq with filtered

after acraping IAir- Inlet

I

Raw WaterIn

'

Float ValveV notch outlet

weir

I-Outlet

Large Filters -tOmrn Float valves 2Srnm David .ShepherdSmall Filters 25 mm Floch valves ID mm Emal

ooC\J

Overflow

Fig. 1: Plan of Pilot Plant

more than 7 500 deaths, while Altonaescaped almost unscathed, notwith-standing the fact that Altona's waterintake was downstream of Hamburg.Subsequent waterborne epidemics inmany parts of the world confirmed thisexperience. In every case, infectionwas almost entirely confined to peopledrinking unfiltered water (Hazen,1895; Huisman & Wood, 1974).

At that time, modern methods of disin-fection were quite unknown, so thatthe protection afforded the publicagainst waterborne pathogens wasalmost entirely dependent on theefficient operation of the sandfilters. Following the introductionof chlorination during the early1900's, and the development ofchemical coagulation followed by thesmaller "rapid" sand filters (about100 times as fast as the slow fil-ters) , the slow sand filters tendedto fall from favour.

As an example of those still surviving,some four-fifths of London's drinkingwater is still treated by slow sandfiltration, although nowadays, thesand filtration is third in a four-stage process of purification :-

(i) long-term storage in largereservoirs,

(ii) rapid sand filtration ormicrostraining,

(iii) slow sand filtration, and(iv) disinfection by chlorination.

During the last decade or two, therehas been a revival of interest in slowsand filters, especially for the treat-ment of water for small communities,and in developing countries wherelocal materials can be used and wherethe necessary skills for construction,operation and maintenance usually liewithin the resources of the localcommunity. The success of many ofthese installations, especially inconjunction with the public-healthsafeguard provided by post-chlorin-ation, has renewed consideration ofthe suitability and economic viabilityof slow sand filtration for small townwater supplies in Victoria.

3. SITE SELECTION AND DESIGN OFPILOT PLANT

As community expectations rise, and asdwindling supplies of natural waterare increasingly affected by develop-ment of one sort or another, there isa growing need for water treatment fortown water supplies throughout theState. In particular, towns in thenorthern part of Victoria, especiallythose drawing their raw water from theRiver Murray, usually experience a runof highly coloured and turbid waterduring the winter-spring period.

It was therefore proposed that thepilot plant be located on the Murray,within a few hours' drive of Melbourne,preferably beside an existing convent-ional treatment plant. Such a sitewas selected at Yarrawonga, beside LakeMulwala, an on-river storage on theMurray. An additional factor in siteselection was the work already done byDr. Bowles, a member of the projectteam, for her doctoral thesis"Phytoplankton Populations of the RiverThames" which includes detailed comp-arisons with levels of phytoplanktonin the River Murray (Bowles, 1978).

.At the same time, Caldwell ConnellEngineers Pty. Ltd., were engaged as"desk" consultants for the project, anda number of eminent biologists,scientists and engineers were invitedto join a Project Reference Panel tooverview the design and progress ofthe study.

It was agreed that four separate filterunits should be provided (two large,two small), they should be circular, ofconcrete, readily transportable, andthe two large units wide enough toreceive sunlight directly onto the sandlayer for a few hours each day, even inmid-winter, at that latitude of 36°south. The general arrangement of thefour open-topped filters (with provis-ion for roofing later on if required),together with two covered tanks at thesouthern end (one for raw water and onefor filtered water) and a partly buriedtransfer tank at the northern end, isshown in Fig. 1.

28-3

Two of the filter units, A and B,contain 1 metre of washed river sandapproximating to the WHO specificationfor slow sand filters (Huisman & Wood,1974 and Van Dijk and Oomen, 1978),while filters C and D contain a coarsesand washed and graded from the samesource. The sand gradings are shownin Table I.

TABLE I

Sieve No.

8 mm4 -

2 "

1.2 mm

600 microns

300

150

75

Fine SandFiltersA & B

% passing

-

-

-

100

83

32

8

2.4

Coarse SandFiltersC & D

% passing

100

97

65

36

6

I

I

0.5

The concept of "effective size" ofthe filter sand, introduced by Hazenas long ago as 1892 (Hazen, 1895), isthe size of grain such that 10% byweight of the particles are smallerthan itself, and 90% larger thanitself. This value cannot be measuredbut is the diameter on the mechanicalanalysis graph corresponding to the10% passing line, and is best referredto as the d 0 value. Similarly, thed60 value is read off at the 60% line.These values for the fine and coarsesands, compared with values suggestedin WHO publications, are given inTable II.

The shape factor for the sand grainshas been assessed as 0.85 (worn) forthe fine sand, and 0.75 (angular) forthe coarse sand, on a scale whichranges from 1.00 (spherical) through0.9 (rounded) to 0.65 (broken).

The total height of each filter is2.75 m, having an initial water depthof 1 m, filter sand 1 m, and gradedgravel to a depth of 450 mm, surround-ing slotted underdrains (Fig. 2).

TABLE II

Parameter

diod60

WHO Suggestion(Huisman & Wood,

1974)

0.15 to 0.35 mm

0.2 "1.0

Fine Sand(FiltersA & B)

0.16 mm

0.45 "

Coarse Sand(FiltersC & D)

0.70 mm

1.8

U =

Porosity(% voids)

1.5 " 3.0 2.8

39%

2.6

33%

U = Uniformity coefficientc

28—4

SECTION A-A

Wobet-

Sand

Slotted Pipe*

Figure 2: Diagramatic section through a slow sand filter

Plastic polyethylene pipe withscrewed hand-tight connections is usedthroughout to facilitate changeoversto different modes of operation.Pumps and piping are designed forfiltering rates up to 0.5 m3/m2.hr(i.e. 0.5 m/hr), but to date the flowrate has been fixed at 0.1 m/hr forthe fine filters and up to 0.15 m/hrfor the coarse filters.

Because the site is flat, constanthead for the filters is provided bycontinuous pumping to a float valvein each filter. The flow rate iscontrolled by an outlet valve, lead-ing to a hump which maintains waterlevel above the sand (in the eventof a power failure), and preventsnegative heads under the filter skin,thus avoiding air binding in the

28-5

filter sand. In addition to a V-notchweir at each outlet, all flows intoeach filter are metered through ordin-ary domestic meters.

The area of each of the two largetanks is 7 m2, and the two small ones3 m2, each giving discharges of 11.7L/min and 5.0 L/min respectively at adownward filtering rate of 0.1 m/hr.Each filter has only two manometertappings, one in the supernatant waterand one in the gravel underdrain.

4. COMMISSIONING (PHASE I)

The plant was commissioned in mid-January 1982, when raw water was rundirectly on to all four filters inparallel (i.e. A and B, fine sand andC and D, coarse sand) at a controlledoutflow rate of 0.1 m/hr. On sitetesting was performed for colour,turbidity, electrical conductivity, pHand dissolved oxygen on the raw andproduct waters. Laboratory algalcounts were made on samples taken fromthe same locations.

It was found that although the colour,turbidity and algal counts were notunusual for that time of year (i.e.mid-summer) for Lake Mulwala water,the fine filters (A and B) blockedwithin a week. On the other hand thecoarse filters (C and D) ran forapproximately four weeks before block-ing. Scraping of the fine filters ona weekly basis enabled them to runfor a total time of five weeks.

The commissioning phase (designatedPhase I) was terminated after fiveweeks because of the unacceptablyshort runs achieved for the finefilters. It was obvious that someform of pretreatment was necessaryto achieve adequate run times.

Some typical results for Phase Iwere: -

TABLE III

Filters Colour(ApparentPt/Co)

Turbidity(JTU)

A & B(fine sand)

Raw waterFilteredwater

C & D(coarse sand)

Raw waterFilteredwater

90, 80

5, 5

29, 17

1.5, 2.5

80, 100

10, 10

20, 19

2.0, 1.6

With the experience gained duringcommissioning it was decided to makealterations to the flow of waterthrough the plant and to use thecoarse sand filters (i.e. C and D) asprefilters for the fine sand filters(A and B).

This change to the flow pattern(see Fig. 3) and the intended mode ofoperation led to a new phase of oper-ation (designated phase II) whicheffectively became the first experi-mental phase.

5. EXPERIMENTAL (PHASES II and III)

5.1 phase II Operation of Filters

This phase began in late February,1982, when raw water pumped from theraw-water tank R was directed firstto coarse filters C and D in parallel,then by gravity to the transfer tankT, pumped back to the filtered watertank F, and finally to the finefilters A and B in parallel.

This mode of operation, with filtersC and D acting as prefilters, enabledthree runs, each of 10 weeks, to be

28-6

InFl

Figure 3: Direction of flow through the plant in Phase II

achieved for the fine filters A and B.During each of these periods ofoperation (i.e. 10 weeks) the coarsefilters had to be scraped at intervalsof between two and six weeks. At thecommencement of Phase II filters A andB were given a 'ripening'period ofseveral weeks at 0.05 m/hr, beforeincreasing to 0.1 m/hr, while C and Dhad to be run at 0.15 m/hr or greater,to ensure that the transfer pumps

always had adequate water while theplant was essentially unattended forup to 5 or 6 days per week.

Phase II ceased at the end ofSeptember 1982, after 31 weeks inwhich three separate runs were com-pleted with fine sand filters (A andB). The results obtained arediscussed in Section 6.

if*

fc

il.

28-7

5.2 Sampling Procedures

5.2.1 physico-chemical measurements

Samples were collected throughoutthe plant of:-

raw waterfiltered water from coarsefilters C and Dfiltered water from finefilters A and B.

In addition, head loss measurementswere made on each filter. The param-eters measured were:-

colourturbidity

. PHdissolved oxygenelectrical conductivity.

5.2.2 Bacteriological measurements

Standard bacteriological techniqueswere used to collect samples from theoutlets of both fine sand filters(i.e. A and B) during the threefilter runs of Phase II. At thecommencement of the study, the inletsto the filters were sampled at thefloat valve inflow, but owing todifficulties with collection, thesepoints were replaced by samplingpoints on the rising pipes of theinlets. Still later, they were re-placed by a single sample on the out-let of the tank used to store thewater produced by the coarse sandfilters. Samples were collectedweekly, except on those four occas-ions when the plant was not fullyoperational (e.g. during scraping orbecause of a power failure).

The following bacteriologicalparameters were routinely measured:-

. E.colicoliforms

. Aeromonas hydrophila

. Pseudomonas aeruginosa

. 22 and 37°C plate counttotal background population.

The results of coliform and Aeromonascounts in filter A are presented inthis paper (see Section 6.2).

5.2.3 Biological measurements

Biological changes in the sand itselfand also in the water produced by thefilters were both investigated.

The algal, protozoan and invertebratepopulations of the sand were investig-ated using a "micro-core" technique.Cores, 10 mm diameter and filled withclean sand, were planted in thefilters after scraping and success-ively withdrawn during the course ofa filter run. At the end of each run,additional large sand cores (40 mmdiameter) were also collected byforcing a core barrel into the filterbed.

Water samples from various parts ofthe pilot plant were collected atweekly intervals for algal determin-ations and particle size/frequencydistribution measurements.

The following parameters were measuredon the sand cores:-

algal numbersprotozoa concentrationsrotifer concentrationsconcentrations of otherinvertebrates.

On water samples, measurements weremade on the following:-

algal numbersinvertebrate numbersparticle numbers and sizes (usinga Coulter Counter Model ZB witha 260 ym orifice).

5.3 Phase III

In this phase (current at the time ofpreparing this paper) changes to thesampling and testing programme havebeen initiated to optimise data coll-ection and to concentrate on thoseparameters of most significance. A

28-8

swimming-pool filter with automaticbackwash has been introduced to pro-vide an alternative to using thecoarse filters for prefiltration. Theflow diagram (Fig. 4) incorporatingthe swimming pool filter, indicatesthat this filter has been introducedwhilst still retaining as a controlthe coarse filters to supply one ofthe fine filters. Further phases willhopefully include a horizontal pre-filtration arrangement and possiblysome additional alteration to flowpaths throughout the plant. At thisstage, however, only those stepsactually taken to modify the samplingprogramme and the plant operation willbe discussed. Future plans will bediscussed in Section 8.

5.3.1 Physico-chemical programmefor Phase III

Electrical conductivity and pH, showvery little variation throughout theplant and are not being measuredintensively in this phase. However,if significant changes are made inthe plant operation, raw water qualitychanges, or if new sand is introduced,then intensive monitoring is to bereintroduced.

Colour and turbidity are the mostsignificant measurements as theirreduction is a direct measure of theeffectiveness of the plant. Thesemeasurements continue to be madethroughout the plant and relativelyintensively.

Inflow To Wasb<

i1

Figure 4: Direction of flow through the plant in Phase III.

28-9

Dissolved oxygen concentration, unlessmeasured intensively over 24 hourperiods, does not indicate diurnalcycles of production by algae and up-take by the aquatic biota of thefilters. During phase III, intensivemeasurements of these changes arebeing made on the water above the sandand in the outflow. As dissolvedoxygen concentration in the outflowingwater is a good indicator of health ofthe organisms in the filters, regularmeasurements will be maintained at theoutflows of the fine filters.

5.3.2 Bacteriological programme forPhase III

In accordance with the flow arrange-ments to the fine filters, the bacter-iology of the inputs and outlets ofthese filters, and the two holdingtanks F and T are being investigated.The parameters currently being measur-ed are E.oo1iy coliforms, Aeromonas,Pseudomonas and a 37°C plate count.Further, if future results indicatelittle differences between the holdingtanks and the respective inlets to thefilters, sampling of the inlets willbe discontinued.

5.3.3 Biological programme forPhase III

Algal numbers are being determinedusing a rapid counting technique onthe inflows, supernatant water abovethe sand, and in the outflows of thefine filters. These counts will beused to establish the input loadingsfrom both the supplied water and fromalgal growth, as well as the effici-ency of the filters with respect toremoving algae. Algal, counts aresupplemented by particle size/frequ-ency measurements obtained by Coultercount. This latter technique permitsthe exact size, volume and concentr-ation in the respective samples to bemeasured.

Estimates of the density of algal matson the surface of the sand in the two

fine filters are to be made by meas-urements of dry weight and loss onignition at 500°C at the end of eachfilter run. The depth of penetrationof organic material into both thecoarse and fine sand filters is beingestimated by the analysis of sandcores collected at the end of eachrun.

Build up of protozoa, algae and smallinvertebrates in the fine sand fil-ters are being investigated usingmicro-cores planted in the filters atthe beginning of each run and with-drawn and analysed successively dur-ing the run period.

6. RESULTS AND DISCUSSION (Phase II)

The results presented in this sectionrelate mainly to Phase II; Phase Iresults having been considered underSection 4, and Phase III results arenot available at the time of writing.

6.1 physico-chemical Results

6.1.1 Temperature

As expected, water temperatures areconsistent with the seasonal changes.During this phase, the warmest waterwas found in March (24-26°C) and thecoolest in July (5-6°C) . Figure 5 isa plot of temperature data for filterA and is typical of the other 'filters;the outlet water is frequently coolerthan the supernatant water by 1 to 2°C.

6.1.2 Dissolved oxygen

The inlet and outlet of both fine andcoarse sand filters have had verysimilar dissolved oxygen concentrat-ions throughout this phase. On mostoccasions, over 10 ppm dissolvedoxygen was found. The single dailymeasurements of dissolved oxygen werenot sufficient to indicate anydiurnal changes in concentrationwhich might be expected to occur as aresult of algal production on thesurface of the filters.

28—10

25-

20-

15 -

10-

5 -

Outlet vs/obc,r

Supernohanb

5 l 5 £ 5 5 l 5 C 5 0 l 5 t 6 5 1 5 2 5 5 *Mar. Apr May Jun. Jul.

8 Ift E5 5 B UAug.

Figure 5: Temperature changes in the supernatant and outlet water of filter.

The outlet water of the coarse filterswas consistently of a lower dissolvedoxygen concentration than the raw(inlet) water and the greatest differ-ences occurred when the head loss wasat its highest.

6.1.3 pH and electrical conductivity

In general, the pH changed by only0.2-0.4 pH units as a result of fil-tration. Since June 1982, the pH ofthe outlet water of the coarse sandfilters has been lower than the raw(inlet) water, while the reverseeffect occurred in the fine sandfilters.

The high pH values experienced inSeptember in the fine sand filterscould be the result of algal product-ion.

Filtration has had very little effecton the electrical conductivity of thewater which has remained within 60-100 yS/cm throughout Phase II.

6.1.4 Colour

All colour measurements during PhaseII were made without filtration orcentrifugation and are therefore moretruly termed 'apparent colour1. Thecoarse filters caused a decrease in

28—11

Pt/Co units RlberA InletOutlet

80-

60_

Z0-

„-,5 15 M

Mar.

* X*V W_._.

5 15 15 5Apr.

15 25 5 IS Z5May Jun.

* fr « _ » _ . M^_4

S IS Z5Jul.

.Xk s is &

Aug.

\xw.._/—

— i — i i i i — i —» 8 IS 25

Sept.

Figure 6: Changes in colour caused by coarse and fine sand filtration.

water colour from between 10-80 Pt-Counits to values of 5-10 pt-Co unitsin most cases. A further decrease inwater colour from 5-10 pt-Co units to5 Pt-Co units resulted from fine sandfiltration. The results from filtersD and A are shown in Figure 6.

6.1.5 Turbidity

Turbidities ranging from 3-28 NTU inthe raw water were decreased to about1 JTU by the coarse filters, on mostoccasions (Fig. 7). After the

initial maturation period in March-April when outlet turbidities werehigher than those of the inlet, thefine sand filters decreased turbidityfrom 1 JTU to about 0.5 JTU (Fig. 7).

6.1.6 Head loss

The coarse filters, which havereceived raw water during this phase,have had only very short "runs"before the head loss seriouslyaffected the flow rates and thefilters had to be scraped (Fig. 8).

28—12

SB-

20-

16 -

8-

4 -

NTU

_ _ _ _ _ Outlet

o

4-

Z-

o

— — -- \,__ — .•"_-. — —— \---_ j.^/~- *>/-•- -'• ~-

Rlt«j Alt-.l*it

U-*-^-^ _ _ _ _ . Outleh

3 to 15 £5 3 15 25 5 15 U 3 15 £5 5 15 2& 5 15 Z5 5 15 25Mar. Apr. a"A Jun. Jul. Aug. Sept.

Figure 7: Changes in turbidity caused by coarse and fine sand filtration

Initially the removal ofwas sufficient to reduceto near zero, but towardPhase II, 2 cm had to bethe coarse filters. Theon the coarse filter, of

1 cm of sandhead lossthe end ofremoved fromlongest runfive weeks

duration, occurred in May-June andcorresponded to a period of lowturbidity in the inlet water. Onother occasions, especially towardthe end of this phase, the coarsefilters have required scraping everytwo weeks.

28-13

Metres

0-6-

o-

Q."

0-3-

0.2-

0-1-

Filter

0.6-

0-5-

0-H-

0-3-

0-2.-

O-1-

Filter A

5 IS £5 5 10 £9 fi (5 £fi 5 IS £S 5 IB £5 5 15 ZS 5 16 £6Apr. f*1*3 Jun. Jul. Aug.

Figure 8: Changes in head loss in coarse and fine sand filters.

Very low turbidity in the pre-filt-ered inlet water from filters C and Dhas allowed the fine filters (A and B)to run for 10 weeks duration. FilterA consistently produced an expon-ential response to head loss withtime, the rapid increase in head lossoccurring some 40 days after filt-ration began (Fig. 8).

6.2 Bacteriological Results

The results of coliforra and Aeromonascounts in filter A (fine sand) havebeen used as the basis for thefollowing discussion.

Coliform numbers were used as a meansof indicating bacterial removal by

28-14

4:

f

1

3.

I

Fig. 9: The Bacteriology of Fine Sand Filter

28—15

the filters as E.coli was very rarelypresent, even in the raw water,during Phase II. Except in initialstages of Phase II, coliform numbersinvariably decreased after filtrationthrough fine sand. In June, July andAugust (corresponding to weeks 19 to32 in Pig. 9) there was least re-duction in numbers; presumably thiswas because water temperatures werelow and grazing protozoa populationswere at their least active. SinceSeptember, the removal of coliformshas been more effective.

Aeromonas hydrophila is an aquaticbacterium which occasionally causesan ulcerative disease in fish, andhas, very rarely, been known toinfect man. As Aeromonas is a nat-ural component of the freshwaterbacterial flora, it was doubtfulwhether filtration could cause areduction in numbers,- and it seemedpossible that numbers might evenincrease as a result of growth withinthe filters. However, the results sofar indicate that Aeromonas was re-moved by filtration, although not,perhaps, as effectively as coliforms(Fig. 9).

6.3 Biological Results

Sand samples

Melosira granulata which was abundantin the raw water during winter andspring was mainly removed by thecoarse sand filters. The skin whichformed on the fine sand filtersduring winter consisted mainly of theMelosira which had penetrated thecoarse sand, and two other diatoms,Navicula sp. and Nitzschia acieularis.In spring as the temperature andsolar radiation increased, a thickmat of the green filamentous algaZygnema sp. developed on filter A,and to a lesser extent on filter B.The Zygnema mat extended as longstreamers into the water above thesand and formed dense patches on thewater surface. Despite this, there

were no indications that the surfaceof the sand had been lifted by thealgal filaments.

When the fine filters were drainedfor scraping and had partially dried,the algal skin compacted to a thick-ness of 2 mm. This skin could be"rolled up" and removed.

Sand cores illustrated the develop-ment of a sand flora and fauna whichwas most abundant just below the skinsurface, still abundant at a depth of10 mm, but decreased rapidly over adepth of 80 mm. Below 80 mm very feworganisms were present. The top10 mm contained, as well as Zygnema}two motile diatoms, Navioula sp. andNitzschia acieularis which are verycommonly found in sand filters andare capable of moving through thesand. The presence of many protozoanspecies was also recorded; smallflagellates being most abundant andciliated species were also commonbut amoeboid protozoa were rare. Theflora and fauna of the fine sandfilters resembled in many respectsthe typical slow sand filter populat-ions found in Thames Water Authorityfilters (Bowles personal communicat-ion) except that in the Englishsituation the skin is formed byMelosira varians in spring and thegreen algal Cladophora during summer.

Water samples

Algae were uncommon in the outflowfrom the fine sand filters duringthis period, being predominantlyrestricted to small flagellates andNitzschia acieularis. This speciesis a fine needle shape organism knownfor its filter penetrating capabilit-ies (Thames Water Authority, unpub-lished) . Nitzsehia acieularisnumbers commonly ranged between 0 and20 cells/mL in the outflows of thefine sand filters. Total particlenumbers in the filtered water werealso low, being very similar to adistilled water background.

28-1(5

7. CONCLUSIONS

The following conclusions have beendrawn from the commissioning phase(Phase I) and the first experimentalphase (Phase II) of this study.

The water quality results fromPhase II of the study are encouraging.With the aid of roughing filters ithas proved possible to produce a highquality treated water from the finesand filters A and B with a runlength of ten weeks between cleaningperiods. Simple scraping of the finesand filters was sufficient to re-store head losses to their initiallevel.

The fine sand filters A and B havedeveloped a characteristic slow sandfilter flora and fauna and have oper-ated as true biological filters. Theremoval of particles, includingbacteria and algae would appear tooccur within the top 80 mm of the sandbed. Despite high concentrations ofalgae in the raw water in the latterpart of Phase II, there have been veryfew algae in the treated water and itsoverall particle content has also beenlow. No taste and odour problems haveoccurred.

The quality of water produced byfine sand filters was therefore sat-isfactory in all respects. In a fullscale operation, the main purpose ofchlorination would be to ensuredisinfection throughout the retic-ulation system, rather than treatmentof the filtered water. Owing to thelow level of bacteria and otherparticulate material in the treatedwater it is expected that chlorin-ation costs could be low, fullchlorination easily achieved, andtaste and odour problems resultingfrom the chlorination of algaeminimal.

Additional work is required todetermine the most effective meansof prefiltering water prior totreatment by fine sand filtration.

8. FUTURE WORK

Having established broad operationalcriteria for slow sand filters underthe conditions found at Yarrawonga,it is proposed that Phase III andsubsequent phases will be used toassess various pre-filtration options,together with establishing the upperlimits for flow rate through the finefilter beds.

Although there were insufficientnumbers of E.coli for observations onthe effects of fine sand filters to bemade, the removal of coliforms andother natural aquatic bacteria indic-ated that, had E.coli been present,its numbers would certainly have beendecreased as a result of filtration.

The physico-chemical quality ofthe treated water was good, itsapparent colour being 5 Pt/Co unitsand its turbidity 1.5 to 2.5 JTU.These levels were measured at thedetection limits of the field instru-mentation. However, due to lack ofrain, the ability of the filters totreat highly coloured or turbidwater has not yet been tested.

During this phase, fine sand filter Bwill be used as a control for experi-ments conducted on fine sand filter A.Additionally, if the pretreatmentsection of the plant has sufficientcapacity, filter D will be resandedwith fine sand and also be usedexperimentally. The effectiveness ofshading the filters as a means ofdecreasing surface algal growth andincreasing run length (hence decreas-ing cleaning costs in a full scaleoperation), and of the relationshipbetween high rate filtration and headloss on shaded and unshaded filterswill be investigated. In addition,attempts will be made to make pre-liminary economic comparisons withmore conventional treatment plants,

28-17

although this will be difficultbecause of the small scale of thecurrent pilot plant.

The study is scheduled to be concludedat the end of 1983 (on-site experi-mental work concluding mid-1983) withthe cost of the study estimated to be$200 000 approximately by the time ofcompletion.

ACKNOWLE DGEMENTS

The authors wish to thank Mr. J. Mann,Director of Water Resources, Ministryof Water Resources and Water Supply,Victoria for his interest in, and forarranging the funding for the project,and R. Goes, S. Phillips, D. Welsh andA. McNeill for providing engineeringand analytical information. In addit-ion, the authors are grateful to theState Rivers and Water Supply Commiss-ion, Victoria for permission toprepare this paper.

REFERENCES

BOOK

MONOGRAPH

THESIS

BAKER, M.N. (1949). The Quest for Pure Water,The American Water Works Association, Inc.

HAZEN, A. (1895). The Filtration of Public Water-Supplies.John Wiley and Sons.

HUISMAN, L. & WOOD, W.E. (1974). Slow Sand Filtration.World Health Organisation, Geneva.

VAN DIJK, J.C. and OOMEN, J.H.C.M. (1978). Slow SandFiltration for Community Water Supply in Developing Countries.WHO International Reference Centre, Technical Paper No. 11 .

BOWLES, B.A. (1978). Phytoplankton Populations of the RiverThames. Ph.D. Thesis, London University.

28-18