Department of Civil and Structural Engineering,Faculty of Engineering and Built Environment,Universiti Kebangsaan Malaysia, Bandar Baru Bangi43600,Malaysia

Ensuringwater security by utilizing roof-harvested rainwater and lakewater treated w ith a low-cost integrated adsorption-fi ltration system

Riffat Shaheed,Wan Hanna MeliniWan Mohtar*,Ahmed El-Shafie

Department of Civil and Structural Engineering,Faculty of Engineering and Built Environment,Universiti Kebangsaan Malaysia, Bandar Baru Bangi43600,Malaysia

Abstract

Drinking water is supplied through a centralized water supply system and may notbe accessed by communities in rural areas of Malaysia. This study investigated the performance of a low-cost,self-prepared combined activated carbon and sand fi ltration(CACSF)system for roofharvested rainwater and lakewater for potable use.Activated carbon was self-prepared using locally sourced coconut shell and was activated using commonly available salt rather than a high-tech procedure that requiresa chem ical reagent.The fi ltration chamberwascomprised of local, readily available sand.The experimentswere conducted w ith varying antecedentdry intervals(ADIs)of up to 15 d and lakewaterw ith varying initial chemical oxygen demand(COD)concentration.The CACSF system managed to produce effluents complying w ith the drinking water standards for the parameters pH,dissolved oxygen(DO),biochem ical oxygen demand(BOD5),COD,total suspended solids(TSS),and ammonia nitrogen(NH3-N).The CACSF system successfully decreased the population of Escherichia coli(E.coli)in the influents to less than 30CFU/m L.Samplesw ith a higher population of E.coli(that is,greater than 30CFU/m L)did notshow 100%removal.The system also showed high potential as an alternative for treated drinking water for roof-harvested rainwater and class II lake water.

?2017 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http:// creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:Low-cost activated carbon；Integrated adsorption-sand fi ltration；Roof-harvested rainwater；Lakewater；Water security

1.Introduction

Water security is defined as reliable and continuous access to safe drinking water for health,livelihood,and development (Grey and Sadoff,2007).In order to ensure water security, theremustbe access to safe and sufficientdrinkingwateratan affordable cost tomeet basic needs,which include sanitation and hygiene.The United Nationshasestimated that1.2 billion people do notdrink safewater and at least746m illion people still do not have access to safe drinking water(World Bank, 2014).

Themostcommon sourcesofwaterused fordrinkingwater supply and irrigation are surface water and ground water. Between the two only a smallamount isaccessible to humans, since most of the surface water is locked in glaciers,snow caps,and ice(Gleick and Palaniappan,2010).The ecosystem is experiencing increasing pressure due to anthropogenic activities,such as urbanization,agriculture,industry,and infrastructure development.Climate change and population grow th have also strongly impacted the ecosystem(Sukereman etal., 2013).Since the lastcentury,the use ofwater has increased by more than two times relative to population grow th.By 2025, water w ithdrawal is predicted to increase by 50%in developing countriesand 18%in developed countries.Assuch,it is predicted that almost 800 m illion peoplem ight not have access to treated water and face absolute water scarcity.It is further predicted that seven billion people from 60 countries w ill face water crisis in the year 2050(WWDR,2003)andfeeding a population of nine billion people in 2050 would require 50%more water than the amount currently used (World Bank,2014).

Malaysia,being blessed w ith an annual rainfall of 2900mm,has faced a series ofwater crises over the last three decades(WWDR,2003).A lthough the majority of the Malaysian population has access to the public water supply system,particularly in the urban areas,themismanagementof water resourceshasexacerbated thewater crisis(Chan,2012). This crisis w ill be aggravated by increasing population,as Malaysia isexpected to have a population of 43m illion people in 2050 and the consumptivewater demand isalso expected to rise to 18.2 billion cubicmeters(MNREM,2012).Tom itigate the future challenges of rapid econom ic development and increasing population,the Malaysian water-related authorities aim at securing water resources in order to ensure that a sufficient amount of water is available to meet the demands of both human society and ecosystems.Thewater security action plan entails the reduction of twomajor elements:consumption and non-revenuewater(NRW)(World Bank,2014).The plan is to reduce per capita consumption in urban areas from 230 L per capita per day(LPD)in 2010 to 150 LPD in 2050.The principles of integrated water resourcesmanagement(IWRM) have been incorporated in the five-year development plan since the Eighth Malaysia Plan.One of the alternatives for water resources in the naturalwater policy is the utilization of rainwater.

Tropical regions such as Malaysia receive rainfall throughout the year and thismakes the idea of utilizing rainwater as a resource an attractive option.The conventional water supply network is considered economically unfeasible, particularly in areas w ith limited access to the public water supply system.The current treatment system relies heavily on surfacewater,which is regarded asawater resource for treated drinking water.However,uncontrolled anthropogenic activities have degraded the quality of surface water and have caused prolonged drought,which in turn adds stress on production demand.As such,rainwater is considered an attractive option in the effort tomeetwater demand since itusually does not contain high levelsof contam inants and thus can be easily treated on-site(Meera and Ahammed,2006；Che-Ani et al., 2009).Preliminary risk analysis suggests that rainwater is of good quality in general and contains only low levels of pathogens(Dillaha and Zolan,1985).Although groundwater could be an attractive source of water,it is not seen as a popular option in Malaysia due to the lim ited quantity and below-par quality,except in the northeastern region.

Theoretically,rainwater is relatively free from impurities but becomes contam inated by pollutants in the atmosphere during precipitation.In general,the presence and the concentrations of organic,inorganic,physical,and biological impurities depend on several factors,such as roof characteristics,meteorological factors,location of the roof,hydrological aspects,chemical properties of the substance,and storage material(Meera and Ahammed,2006；Despins et al.,2009). Furthermore,the quality of rainwaterm ightdeteriorate during harvesting and storage due to w ind-blown dirt,leaves,fecal droppings,and contam inants present in the catchment area. Themost common impurities(or physico-chem ical qualities) investigated in harvested rainwater are m icrobes(heterotrophic plate count and coliform),organic content(carbon and nitrogen),heavymetal(Hg,Pb,Cu,Fe,Mn,Zn,Cu,and Ni) dust,fine particles(turbidity and solids),and ions(Ca,Mg, Na,K,nitrates,and sulfates).A few studieshave reported that harvested rainwater often does notmeet the m icrobiological drinkingwaterquality standard sincemostof the contam inants are fecal coliforms(Escherichia coli,E.coli)from animal origin(Yaziz et al.,1989；Handia et al.,2003；Vialle et al., 2011).

It has been determ ined that roofmaterial has a significant effect on the quality of rainwater；rainwater of the highest quality is harvested using steel roofs,followed by roofs of asphalt shingles,galvanized iron,and finally concrete tile roofs(Yaziz et al.,1989；Despins et al.,2009).The location where rainwater is harvested is also important in determ ining the quality of water collected,in that the water collected in rural catchment areas is often of higher quality compared to that collected close to industrial areas(Despins et al.,2009).

The presence of heavy metals in rainwater is often given particular attention due to their toxicity；thesemetals cannot be chem ically transformed if their concentrations exceed the threshold lim it and they cannot be easily removed w ithout complex treatment(Davis etal.,2001).In general,the quality of rainwater in Malaysia is good because the amounts of Hg, Pb,Cu,Fe,Mn,Zn,Cu,and Ni are below the perm issible limits(Seong and Sapari,2003).The concentrationsof Pb and Hg are higher in rainwater collected between 1990 and 1997, but the amounthas decreased to an insignificant levelsince the complete ban of leaded petrol in 1997.The levels of Na,Mg, F,Cl,,andwere found tomeet the drinking water standard(Seong and Sapari,2003).

Several studies have been conducted to determ ine the quality of rainwater and the results have shown that rainwater harvested far from highwaysand industrialareasdoesnothave high levels of contam ination and atmospheric pollutants (Yaziz etal.,1989；Thomas and Greene,1993；Appan,1997；Ayers et al.,2002；Gould,1999).Themost common collectionmechanism,particularly for domestic purposes,is a roof. In Malaysia,it is recommended that the fi rst 3mm of rainfall be the fi rst flush since this has been found to be sufficient to ensure high-quality rainwater(Yaziz et al.,1989).However, the antecedent dry intervals(ADIs)are also an important factor and the depth of the fi rst flush should be increased for rainwater harvested after an ADIof 15 d(Shaheed and Wan Mohtar,2014).

The parameter of utmost concern in harvested rainwater is the presence of m icrobial pathogens,particularly the fecal bacteria present in animal droppings.The collected rainwater could pose a risk to human health if it is consumed untreated (Ahmed et al.,2008,2011).However,the focus of this study was not the utilization of untreated harvested rainwater. Instead,this study focused on providing a low-costalternative of a locally produced treatmentsystem that iseasy tomaintain for potable use.

The high quality of collected rainwatermeans that it could potentially be used as potable water with minimal treatment, particularly in placeswhere a centralized water supply system is notavailable or during water crisis.This study attempted to investigate the effectiveness of a low-cost,self-prepared combined activated carbon and sand fi ltration(CACSF) system for roof-harvested rainwater.Successful implementation of this approach provides an alternative for rural communities that lack proper equipment and trained personnel. The assessment of the system is extended to surface water.

2.M ethodology

The low-cost CACSF transparent fiber glass fi lter was constructed from an experimental rig consisting of four inner chambers.The schematic diagram for the chambers is shown in Fig.1(the diagram isnotdrawn to scale).The fi rstchamber, labeled A,functions as a diffuser and has slightly larger dimensions.The next three chambers,labeled B,C,and D,have the same dimensionsof13.3 cm×13.3 cm.Each chamberhas its own drain outlet at the end of the chamber that facilitates cleaning and the discharge of water.The fi rst level of treatment in Chamber B is adsorption by self-prepared activated carbon.Next,Chamber C,which contains fine sand,servesas the fi ltration chamber.To facilitate the flow through themedia, the bottom of chambers B and C are fi lled w ith gravelbetween 6mm and 2 cm in height,and a thin layerof coarse sand is laid over them.Influents from chambers B and C are fed from the bottom,and after fi ltration the effluents pass over the weirs and flow into the collector Chamber D before being discharged.The higher position of the outlet(1.5 cm higher than the weirs)maintains a steady water level in all chambers and therefore keeps themedia under water.

The self-prepared activated carbon wasmade from locally sourced coconutshell.An 11 cm-high pile of activated carbon was placed in Chamber B.This heightwas chosen so that it would be the same height as the sand；the justification for choosing thisheightw illbe discussed later.Itshould be noted that the production of activated carbon from coconutshellshas been commercialized on a large scale.However,in this study, the activated carbon was locally produced using low levels of technology by utilizing the commonly available salt as an activation reagent.Systematic tests were conducted and the optimum setup,which produced amaximum diameter pore of 17.3μm,was pyrolyzed at 700°C for 1.5 h.Fig.2 shows the developed pore walls under optimum conditions,where the orderly structuralskeleton is visible.The porouswallwaswell structured,and increasing the temperature to 800°C disintegrated the wall.An optimum size(between 1 and 2.36 mm)and the details of optim izing self-made activated carbon are discussed in Shaheed et al.(2015).

A fter adsorption,the influent made its way through the down-flow fi ltration process.Tomake use of locally available sources,the sand used in this chamber was collected from natural streams.The sand was dried and then sieved for particle sizesof0.15mm and 1mm for fine sand and coarse sand, respectively.Gravel,coarse sand,and fine sand were washed separately and put in the oven for a complete dried sediment m ixture.The depth of sand media was 11 cm,which allowed for an appropriate biological zone of between 5 and 10 cm (CAWST,2009).Chamber C was fi lled w ith fine sand w ith an effective diameter(d10)of 0.15-0.30 mm and a uniformity coefficient(d60/d10)of 1.5-3.0,whered60andd10represent the sediment diameters w ith 60%and 10%of the sediment passing,respectively.Fine sand was carefully placed to prevent the segregation of particles and to ensure that particles werewelldistributedw ithin the chambers.The flow rateof the fi ltration process was set at 0.417 L/m in(2.5 L/h),which complied w ith the CAWST(2009)specifications,and thus the experimental retention time was 87 m in.

Rainwater sampleswere collected from the roof of a cabin room,located close to the campus in Bandar Baru Bangi,a district in Selangor 30 km away from the capital Kuala Lumpur.The structure was covered with a corrugated galvanized iron sheet rooftop.This rooftop material is ideal for collecting rainwater as it provides complete washability of pollutants by a fi rst flush.The fi rst 3 mm of rainwater were regarded as the fi rst flush as setby the standard guidelinesand in keeping w ith the findings of previous studies(Yaziz etal., 1989；Coombes et al.,2000；Martinson and Thomas,2005). In total 11 rainwater samples were collected and prelim inary analysis showed that rainwater for ADIs of 1 and 2 d were consistent w ith the acceptable lim its of drinking waterguidelines.As such,only rainwater samples w ith an ADI greater than 2 d were tested using the treatment system.

Fig.1.Experimental reactor.

Fig.2.Scanning electron m icroscopy(SEM)image of activated carbon pyrolyzed at 700°C for 1.5 h.

In assessing the performance of the CACSF system,the water samples treated were not lim ited to rainwater,but also included samples from surfacewater.Six samples from a lake were collected after rain and fed into the fi lter.The lake receivessurfacewater runoff from the surrounding areaand acts as a retention pond before discharging water into the nearest river.The samples were collected between the months of February and March,a relatively dry season(for a tropical country)where the rainfall intensity is between 7 and 43mm. A typical rainfall during this period lasts for about 30m in to 1 h,often as a drizzle and not a heavy downpour.Approximately 30 L of surface water was extracted 6 h after the rainfall to allow for the settlementof suspended soil particles. Samples of one influent and two effluents from the activated carbon and fine sandmediawere collected after fi ltration.The water quality parameters are categorized as stipulated by the Department of Environment of Malaysia(DOE,2010).

The two different effluents from chambers B(activated carbon)and C(sand)were collected to assess the performance of each treatment process.To avoid confusion,the effluents from chambers B and Cw illhereafterbe referred to as EB and EC,respectively.The influents,EB,and EC were analyzed in terms of seven parameters,namely pH,DO,BOD5,COD, NH3-N,TSS,andE.coli.As discussed earlier,although other impurities are often measured,this study did not investigate the concentration of heavy metals(based on the findings reported by Seong and Sapari(2003)).Furthermore,as the study area issurrounded bymature trees,and the sampleswere kept in a plastic storage tank,the turbidity and totalorganic carbon have been consistently found to be less than 2 NTU and 10mg/L,respectively(Despins et al.,2009).

To ensure consistent measurement,the value for each parameter was obtained from an averaged value of three samples.Table 1 describes the instruments used to test each quality parameterassessed in thisstudy.The numberofE.coliwas determ ined by plating 1 m L of sample water on a 3M Petrifi lm?after a 48-h incubation at 36°C.

Effluents weremeasured and compared w ith the drinking standard guidelines set by the M inistry of Health of Malaysia (MOH,2000),the Departmentof the EnvironmentofMalaysia (DOE,2010),as well as the World Health Organization (WHO,1993).

The performance of the fi ltermedia is expressed as a percentage reduction from influent and EB for activated carbon, and influentand EC for sandmedia.The percentage reduction data were then used to calculate the average percentage reduction by taking into consideration the non-zero values.To check the dispersion of values w ithin the same set of percentage reduction values,the mean of standard error(MSE) was determined,which is calculated as follows:

whereσdenotes the standard deviation andNis the total population.Thismeans that theMSEisexpected to be smaller for a larger sample size.It gives a rough estimate of the interval inwhich themean value of population is likely to fall.It represents the degree of precision w ith which the sample statistic represents the population parameter.

3.Results and discussion

This section starts w ith a discussion of the quality of collected rainwater and surface water from a lake.The water quality parameters(pH,DO,BOD5,COD,NH3-N,TSS,andE.coli)for rainwater w ith an ADImore than 2 d and surface water samples are presented in Table 2.

The data show that,compared to the practiced standard values for water quality parameters,the collected rainwater samples satisfy all parameters except forE.coli.As such,the harvested rainwater could be considered to have good quality. The quality of surfacewater for two groups showed a distinct difference:the parameters for the fi rstgroup(i.e.,SW 1,SW 2, SW 3,and SW 4)were beyond class IIIstandards,but the parameters for the second group(SW 5 and SW 6)met the class II standards,according to the classification by the Departmentof Environment of Malaysia(DOE,2010).

The efficiency of each processwas evaluated based on the seven water parameters,and w ill be discussed separately for rainwater and surfacewater samples.It should be noted that, as the values for rainwater influent are already below the recommended values,the evaluation of removal efficiency is purely for academ ic purposes(to test the CACSF system)and is focused on the removal ofE.coli.Attention is also paid to the removalefficiency of the CACSF system for the treatment of surfacewater.

Table 1Standard methods and instruments used for testing various parameters.

Table 2Water quality parameters for rainwater and surfacewater samples.

3.1.pH change

The pH values for the influent of rainwater and surface water samplesare presented w ith the values for EB and EC in Table 3.In general,the pH values increased consistently throughout the CACSF treatment.However,the increment is only w ithin a small range and liesbelow the permissible limit for the drinking water standard.

The data show thatalleffluents,and almostall influents,are w ithin theacceptable lim it.A pH increasehasbeen shown to be independentof rawmaterialswhetherornottheactivated carbon isacid-washed.ACprepared from bitum inousorsub-bituminous coal,coconut,wood,orpeatshowedan increaseinpHwhenused inwater(Fameretal.,1996).Itisbelieved thatthepH increaseis a function of the AC surface where the released protons from hydrolysispositively charged the carbon surface.The anions in wateror the hydroxide ions from the initialhydrolysisofwater then neutralized the recently positively charged carbon surface.

3.2.DO reduction

The concentrations of DO in rainwater and surface water influents,EB,and EC are given in Table 4.The concentrations of DO decreased after fi ltration through the CACSF system. However,the decrease was not significant,and the remaining DO levels in the effluentswere stillwellabove the acceptable level setby theMOH.The profi le is consistent for all samples and the ADIs of rainwater samples have no effect on the behavior of DO.

The decrease in the DO concentration in rainwater effluents was due to the adsorption of oxygen by the activated carbon media.Since theMSEvalueswere 1.48 and 0.31,the population represented precise data for the tested parameters.Even after adsorption and sand fi ltration,all effluents showed acceptable concentration levels of DO,above the standard value of 7.0mg/L.

With regard to surface water,a different scenario was observed and the decrease in DO concentration is significant. The percentage of reduction by activated carbon adsorption wasbetween 29%and 39%,w ith an average value of 33%and anMSEvalue of 1.33.Sand fi ltration,on the other hand,removed a largeramountof DO,w ith an averagepercentageof reduction of 42%and anMSEvalue of 0.87.The reduction in DO concentration for surfacewater was higher than for rainwater for both adsorption(33%compared to 7.5%)and fi ltration(42%compared to 1.5%)on average.The range of DO reduction recorded in this study was congruent w ith the results obtained by Gimba and Turoti(2006),where it was found to be46.2%and 38.5%,and theactivated carbon used in theirstudywasprepared by activating coconutshellw ith FeCl3and CaCl2.A lthough the values obtained in this study were slightly lower than the values obtained by Gimba and Turoti (2006),the performance of our low-tech activated carbon is comparable to thehigh-tech activated carbon,which requiresa chemical reagent,skilled personnel,and high-tech equipment.

Table 3pH values for influents and effluents in CACSF system.

Table 4Measured DO concentrations for rainwater and surface water influents and effl uents.

Previous research has found that little oxygen is consumed during fi ltration,w ith the oxygen concentration usually being depletedw ithin the schmutzdecke(Buzunis,1995).In the case ofabio-sand fi lter,Kennedy etal.(2012)found that,onaverage, biofi lm could reduce up to 60%of DO in the initial influent. Thisstudy assumed that there isno schmutzdecke formation or that it isnotactive in the up-flow fi lter(namely,a fi lterw ithout any schmutzdecke),and thus the adsorption of oxygen by the sandmedia ismuch less,between 0.05 and 0.22mg/L.

3.3.BOD5reduction

Themeasured BOD5levels in both rainwater and surface water influents and effluents are given in Table 5.The DOE has set a value of less than 1mg/L for the BOD5concentration.However,the MOH and WHO have not set any healthbased guideline values for the BOD5concentration(MOH, 2000；WHO,1993).Environment Canada(EC,1979)reported that drinking water sources should have a BOD5concentration lower than 3 mg/L and water w ith a BOD5concentration lower than 4mg/L is considered to be of good quality.The concentration of BOD5is related to that of DO and,assuggested by De(2003),am inimum DO concentration of 2-7 mg/L should be retained after the degradation of oxidizable organicmatter.In Table 5,theMSEfor rainwater samples is2.13w ith activated carbon adsorption and 2.04w ith sand fi ltration,and theMSEfor surfacewater samples is 0.49 w ith activated carbon adsorption and 0.67 w ith sand fi ltration.

Although there are two sets of standards(DOE and EC) available,this study adopted the BOD5concentration(less than 1 mg/L)set by the DOE in the assessment of the performance of the CACSF system.

Table 5 shows that the values of BOD5in rainwater influents are relatively low,w ith the highest value being 2.77mg/L.However,itshould be noted that the fi ltermedia is capable of removing organicmatter from the rainwater(48% by activated carbon and 11%by sand,on average).On the other hand,the values of BOD5in surface water influents varied between 1.82 and 5.30mg/L.The CACSF system was able to elim inate organicmatter and reduced BOD5values by 59%on average through both adsorption and fi ltration.The 43%BOD5reduction efficiency by the sandmedia(0.15mm) of the CACSF system(i.e.,39%-42%)from surface water is sim ilar to the values obtained by Abudi(2011).It should be noted that he used a 0.35-mm sand particle,which ismuch larger than whatwas used in this study.

A fter undergoing sand fi ltration,four rainwater(RW 6, RW 7,RW 8,and RW 11)and two surfacewater(SW 5 and SW 6) effluentswere found tomeet the standard setby the DOE.The values for all effluents were less than 2 mg/L and could be considered to have the potential tomeet the standard setby the DOE.Even though rainwater effluentsw ith an ADI less than 9 d did notquitemeet the DOE standard,the values obtained werew ithin 1mg/L,the value set by the DOE.This indicates that,even w ith a typically higher pollutant concentration(due to long dry days),the CACSF system has the potential to treat water for use as an alternative potablewater source.

In thecaseof theup-flow fi ltersused in thisstudy,theaverage efficiency of BOD5removalby sand fi ltration for rainwaterwas 11%.Itisbelieved that the performance of the system could be improved by increasing the depth of the sandmedia.

Table 5Measured BOD5concentrations for rainwater and surfacewater influents and effluents.

3.4.COD reduction

COD was not detected in rainwater influents and this is believed to be due to the lower concentration than the quantification limits.Therefore,the discussion in this subsection is in reference to surfacewater.

Surface water samples have relatively high COD concentrations and the measured COD concentrations in influents and effluents after both adsorption and fi ltration are given in Table 6.It can be noted thatsamples SW 5 and SW 6meet the recommended COD concentration(10mg/L)for raw water set by the DOE.

The data show that the CACSF system can completely remove COD.The adsorption process itself successfully reduced COD concentration by 85%-90%and continuing the fi ltration process resulted in a 100%removal of COD from surfacewater.TheMSEvaluesare 1.03 and zero for activatedcarbon adsorption and sand fi ltration,respectively,and as such,the tested parameter is deemed accurate.The performance of the self-prepared activated carbon was better than thatof the high-tech activated carbonmade from coconutshell and rice husk,in which COD reductions of 46%-71%and 45%-73%were obtained,respectively(Mohan et al.,2008). Further reduction after fi ltration is consistent w ith previous findings,in which a 75%-100%COD reduction was reported when sand fi ltration was used(Palmateer et al.,1999；Amy et al.,2006；CAWST,2009；Logsdon et al.,2002).Thus,in terms of COD concentration,the effluent from the CACSF system hasmet the drinking water standard.

3.5.TSS reduction

Themeasured TSS concentrations in rainwater and surface water influentsand effluents after adsorption and fi ltration are given in Table 7.In general,both rainwater and surfacewater have relatively low TSS concentrations,less than 50 mg/L. The limit for TSS is considered to be 25 mg/L as set by the DOE.It should be noted that,except for SW 2,all samples show values below the permissible limit.For short ADIs,thecollected rainwater has typical values less than the standard value for drinking water.In Table 7,theMSEfor rainwater samples is7.07w ith activated carbon adsorption and zerow ith sand fi ltration,and theMSEfor surfacewater samples is 7.48 w ith activated carbon adsorption and 6.37 w ith sand fi ltration.

Table 6Measured COD concentrations for surfacewater influents and effluents.

Table 7Measured TSS concentrations for rainwater and surface water influents and effluents.

The results show that fi ltration through activated carbon and fine sand decreased the TSS concentration.Themeasured TSS concentration for rainwater ranged between 1.33 and 10.66 mg/L,and all samples had values well below the perm issible lim it.The activated carbon removed a significant amountof TSS from influentsand the TSS foreffluents ranged between 0 and 3.84 mg/L.Fine sand fi ltration removed even more TSS and,except for SW 1 and SW 2,all effluentswere free of TSS.Treatmentw ith the CACSF system could bring the level of TSS in the rainwater effluent to zero,and this is independentof the ADI range discussed in this study.

Since WHO(1993)did not set any standard for TSS in drinking water and the DOE has set the lim it at 25 mg/L, surfacewaterw ith a TSS concentration lower than 25mg/L is considered to be of good quality.However,as the priority of this study was to exam ine the performance of the CACSF system,the follow ing discussion w ill only focus on samples w ith TSS concentrations higher than 5 mg/L.The TSS concentrations for both types of influent decreased after activated carbon adsorption and sand fi ltration.Theactivated carbon and the sand removed 39%-100%(w ith an average of 62.6%)and 73%-100%(w ith an average of 95.4%)TSS,respectively.

A low concentration of TSS in influents makes their removal easier.Palmateer et al.(1999)reported that fine sand fi ltration produced a 100%reduction of TSS from lakewater.

3.6.NH3-N reduction

Themeasured NH3-N concentrations in the influents and effluents after both activated carbon adsorption and sand fi ltration are shown in Table 8.The acceptable lim it for NH3-N in this study was 1.5mg/L,as setby the MOH.Themeasured NH3-N concentrations from collected rainwater influents ranged between 0.05 and 0.81mg/L,and all rainwatersamples showed valueswellbelow theallowable lim itand are therefore considered to be of good quality.As such,theoretically,the removalof NH3-N using this treatmentsystem isnota priority. However,the performance of activated carbon was exam ined and the results showed that it is possible to reduce NH3-N by asmuch as 93%.

As can be seen in Table 8,the NH3-N valuesmeasured in the surface water influents have the possibility of exceeding the recommended raw water quality(set at 1.5 mg/L by the MOH).The removal of NH3-N from surface water is not as efficient as it is for rainwater,w ith an average reduction of only 45.7%.TheMSEfor rainwater samples is 2.57 w ith activated carbon adsorption and 8.16 w ith sand fi ltration,and theMSEfor surface water samples is 6.34 w ith activated carbon adsorption and 3.50 w ith sand fi ltration.When the NH3-N concentration exceeds 1.4 mg/L,adsorption w ith activated carbon does not guarantee a quality suitable for drinking.Further treatment w ith sand fi ltration managed toreduce the NH3-N to a level below the perm issible lim it even for the highest NH3-N concentrationmeasured in this study.

Table 8Measured NH3-N concentrations for rainwaterand surfacewater influentsand effluents.

3.7.E.coli removal

The fecal coliformE.colishould notbe present in effluent, as strictly stipulated by the drinking water guidelines(MOH, 2000).This study investigated the efficiency of the low-cost CACSF system in removingE.coli.As previously mentioned,samples RW 6 and RW 7(w ith an ADI of 3 d) did not have anyE.coliand were not analyzed.TheE.colipopulation in both influents and effluents is presented in Table 9.The table shows that the measuredE.coliin rainwater influent ranged from 2 to 9CFU/m L,and theE.coliincreased when the ADIs were longer.The activated carbon was able to completely removeE.colifrom samplesw ith an initial population of less than 9 CFU/m L.E.coliwas still present in sample RW 11,which was collected after an ADIof 15 d,although there was a 75%removal.This indicates that the developed micro-pore size in the self-prepared activated carbon is capable of trapping the fecal coliform.

As expected,the population ofE.coliwas significantly higher in surfacewater than in rainwater and ranged from 23 to 119CFU/m L.TheE.colipopulationwasaveragely reduced by 98%through activated carbon adsorption and was then further reduced by 5%by the sandmedia.With a98%average removal efficiency from rainwater and surface water,and as theMSEwas as low as 6.25,the low-tech activated carbon showed prom ising potential in removingE.coli.Sand fi ltration isconsidered a polishingmechanism to further trapE.colias the incoming influent from the activated carbon already has a lowE.colipopulation.

Sand fi ltration is an approved system forE.coliremoval；several studieshave reported 96.5%removalefficiency through the use of down-flow bio-sand fi lters(Buzunis,1995；Baumgartner,2006).Table9 clearly shows that rainwater fi ltrationusing theCACSFsystem couldbring theeffluentE.colilevel to zero even for an ADIof up to 11 d.For samplesw ith longer ADIs,it is assumed that either a greater fi rst flush volume for rainwateroran increasein theheightof thesand layerisrequired.

Table 9MeasuredE.colipopulation for rainwater and surface water influents and effluents.

The data showed that the system worked well when the concentration ofE.coliin the influent was less than 30 CFU/m L.For samples w ith less than 105 CFU/m L ofE.coli,the system worked relatively well even thoughE.coliwas detected in the effluent(i.e.,1 CFU/m L).It is assumed that for influent w ith anE.colipopulation greater than 120 CFU/m L the current system is not yet able to produce a desirable potable quality effluent.However,the system is expected to be able to completely removeE.coliif the retention time is increased.Commercially produced activated carbon reducedE.coliby 75.21%w ithin a 1-h contact time and reducedE.coliby 100%after 3 h(Karnib et al.,2013).

The second levelof treatmentw ith up-flow sand fi ltration is suitable for the removalofE.coli.However,it ispredicted that down-flow fi ltration is themore suitablemethod for achieving higher reduction efficiency.The results obtained for reducingE.coliusing the down-flow sand media varied between researchers,w ith the percentage of removal ranging from 80% to 100%(Manz etal.,1993；Karnib etal.,2013).Williamsand Eng(1987)observed thatat50mm below the sand surface,theE.colireduction was90%,and at20 cm the reduction reached 99.5%.Increasing the depth of the present up-flow sand fi ltration w ill increase the retention time and this m ight consequently improve the performance of the system.

In general,the CACSF system is able to remove 92%-100%ofE.coli,and lower removal efficiency was observed when theE.colipopulation was greater than 100 CFU/m L.

4.Conclusions

This study investigated the performance of a low-cost,selfprepared combined activated carbon and sand fi ltration(CACSF)system in treating roof-harvested rainwaterandwater collected from a lake.In general,harvested rainwater was of good quality and,except forE.coliand NH3-N,the parameter values were significantly below the drinking water standard guidelines.Thewater samples from the lakemeteither class II or class IIIstandards,and thewater needed treatment tomake it potable.Overall,the integrated CACSF system managed to treat the harvested rainwater and lakewater to place itw ithin the perm issible lim it setby the drinking water standard for all parameters,except the bacteriaE.coli.The performance of the self-prepared activated carbon was found to be w ithin the acceptable rangeof commercially availableactivated carbon or high-tech bio-waste activated carbon.

The CACSF system was able to remove the entireE.colipopulation forsamplesw ithapopulationof lessthan31CFU/m L. The present setup however could not produce effluent that meets the drinking water standard for samplesw ith anE.colipopulation of more than 30 CFU/m L.The CACSF system showed satisfactory performance in treating class II surface water but needs to be improved to treat class III surface water.In general,the system performed satisfactorily and offers an attractive option to rural communities,or,during water crisis,a way of providing an alternative water source through the treatmentof harvested rainwater(for short ADIs) and lakewater.However,longer ADIsand abstraction of lake water require a greater height of activated carbon and sand layers(in the chambers)to ensure a higher quality of effluent.

Abudi,Z.N.,2011.The effect of sand fi lter characteristics on removal efficiency of organicmatter from grey water.Al-Qadisiya J.Eng.Sci.4(2), 143-155.

Ahmed,W.,Huygens,F.,Goonetilleke,A.,Gardner,T.,2008.Real-time PCR detection of pathogenic m icroorganisms in roof-harvested rainwater in Southeast Queensland,Australia.Appl.Environ.M icrobiol.74(17), 5490-5496.http://dx.doi.org/10.1128/AEM.00331-08.

Ahmed,W.,Gardner,T.,Toze,S.,2011.M icrobiological quality of roofharvested rainwater and health risks:A review.J.Environ.Qual.40(1), 13-21.http://dx.doi.org/10.2134/jeq2010.0345.

Amy,G.,Carlson,K.,Collins,M.R.,Drewes,J.,Gruenheid,S.,Jekel,M., 2006.Integrated comparison of biofi ltration in engineered versus natural systems.In:Gimbel,R.,Graham,N.,Collins,M.R.,eds.,Recent Progress in Slow Sand and A lternative Biofi ltration Processes,1,London,pp.3-11.

Appan,A.,1997.Roof water collection systems in some Southeast Asian countries:Status and water quality levels.J.R.Soc.Health 117(5), 319-323.http://dx.doi.org/10.1177/146642409711700510.

Ayers,G.P.,Peng,L.C.,Gillett,R.W.,Fook,L.S.,2002.Rainwater composition and acidity at five sites in Malaysia,in 1996.Water Air Soil Pollut. 133(1),15-30.http://dx.doi.org/10.1023/A:1012967614759.

Baumgartner,J.,2006.The Effect of User Behavior on the Performance of Two HouseholdWater Filtration Systems.M.S.Dissertation.University of Harvard,Boston.

Buzunis,B.J.,1995.Interm ittently Operated Slow Sand Filtration:ANew Water TreatmentProcess.M.S.Dissertation.University of Calgary,Calgary.

Centre for A ffordable Water and Sanitation Technology(CAWST),2009. Biosand Filter Manual,Design,Construction,Installation,Operation and Maintenance.CAWST,Calgary.

Chan,N.W.,2012.Managing urban rivers and water quality in Malaysia for sustainable water resources.Int.J.Water Resour.Dev.28(2),343-354. http://dx.doi.org/10.1080/07900627.2012.668643.

Che-Ani,A.I.,Shaari,N.,Sairi,A.,Zain,M.F.M.,Tahir,M.M.,2009.Rainwater harvesting as an alternative water supply in the future.Eur.J.Sci. Res.34(1),132-140.

Coombes,P.J.,Kuczera,G.,Kalma,J.D.,Dunstan,H.,2000.Rainwaterquality from roofs,tanksand hotwatersystemsat figtree place.In:Proceedingsof the 3rd International Hydrology and Water Resources Symposium,Perth.

Davis,A.P.,Shokouhian,M.,Ni,S.,2001.Loading estimatesof lead,copper, cadm ium,and zinc in urban runoff from specifi c sources.Chemosphere 44(5),997-1009.http://dx.doi.org/10.1016/S0045-6535(00)00561-0.

De,A.K.,2003.Environmental Chem istry,fi fth ed.New Age International Publisher,New Delhi.

Department of Environment of Malaysia(DOE),2010.Environmental Requirements:A Guide for Investors,eleventh ed.Department of Environment,Putrajaya.

Despins,C.,Farahbakhsh,K.,Leidl,C.,2009.Assessmentof rainwaterquality from rainwaterharvesting systemsinOntario,Canada.J.WaterSupply Res. Technol.AQUA 58(2),117-134.http://dx.doi.org/10.2166/aqua.2009.013.

Dillaha III,T.A.,Zolan,W.J.,1985.Rainwater catchment water quality in M icronesia.Water Res.19(6),741-746.http://dx.doi.org/10.1016/0043-1354(85)90121-6.

Environment Canada(EC),1979.Analytical Methods Manual.Inland water Directorate.Water Quality Branch,Ottawa.

Famer,R.W.,Dussert,B.W.,Kovacic,S.L.,1996.Improved granular activated carbon for the stabilization ofwastewater pH.Am.Chem.Soc.Div.Fuel Chem.41(1),456-460.

Gimba,C.E.,Turoti,M.,2006.Adsorption efficiency of coconut shell-based activated carbons on colour of molasses,oils,dissolved oxygen and related parameters from industrial effluent.Sci.World J.1(1),21-26. http://dx.doi.org/10.4314/sw j.v1i1.51692.

Gleick,P.H.,Palaniappan,M.,2010.Peak water lim its to freshwater w ithdrawal and use.Proc.Natl.Acad.Sci.107(25),11155-11162.http:// dx.doi.org/10.1073/pnas.1004812107.

Gould,J.,1999.Is rainwater safe to drink?A review of recent findings.In: Proceedingsof the9th InternationalRainwaterCistern Systems,Brazil,p.9.

Grey,D.,Sadoff,C.W.,2007.Sink orsw im?Water security forgrow th and development.Water Policy 9(6),545-571.http://dx.doi.org/10.2166/wp.2007.021.

Handia,L.,Tembo,J.M.,Mwiindwa,C.,2003.Potential of rainwater harvesting in urban Zambia.Phys.Chem.Earth 28(20-27),893-896.http:// dx.doi.org/10.1016/j.pce.2003.08.016.

Karnib,M.,Holail,H.,Olama,Z.,Kabbani,A.,Hines,M.,2013.The antibacterial activity of activated carbon,silver,silver impregnated activated carbon and silica sand nanoparticles against pathogenicE.coliBL21.Int. J.Curr.M icrobiol.App.Sci.2(4),20-30.

Kennedy,T.J.,Hernandez,E.A.,Morse,A.N.,Anderson,T.A.,2012.Hydraulic loading rate effect on removal rates in a BioSand fi lter:A pilot study of three conditions.WaterAirSoilPollut.223(7),4527-4537.http:// dx.doi.org/10.1007/s11270-012-1215-4.

Logsdon,G.S.,Kohne,R.,Abel,S.,LaBonde,S.,2002.Slow sand fi ltration for small water systems.J.Environ.Eng.Sci.1(5),339-348.http:// dx.doi.org/10.1139/s02-025.

Manz,D.H.,Buzunis,B.J.,Morales,C.,1993.Household Water Supply and Testing Project.Final Report on Assessment of Bio-sand Filter in Nicaragua.University of Calgary,Calgary.

Martinson,D.B.,Thomas,T.H.,2005.Quantifying the fi rst flush phenomenon: Effects of fi rst-flush on water yield and quality.In:Proceedings of 12th International Conference on Rain Water Catchment Systems,New Delhi.

Meera,V.,Ahammed,M.M.,2006.Water quality of rooftop rainwater harvesting systems:A review.J.Water Supply Res.Technol.55(4),257-268.

M inistry of Health of Malaysia(MOH),2000.National Drinking Water Quality Standards by Engineering Services Division.MOH,Putrajaya.

M inistry of Natural Resources and Environment,Malaysia(MNREM), 2012.National Water Resources Policy.http://www.nre.gov.my/msmy/PustakaMedia/Penerbitan/Dasar%20Sumber%20Air%20Negara.pdf [Retrieved 18 July 2014].

Mohan,D.,Singh,K.P.,Singh,V.K.,2008.Wastewater treatment using low costactivated carbonsderived from agriculturalbyproducts:A case study. J.Hazard.Mater 152(3),1045-1053.http://dx.doi.org/10.1016/ j.jhazmat.2007.07.079.

Palmateer,G.,Manz,D.,Jurkovic,A.,M clnnis,R.,Unger,S.,Kwan,K.K., Dutka,B.J.,1999.Toxicant and parasite challenge of Manz intermittent slow sand fi lter.Environ.Toxicol.14(2),217-225.http://dx.doi.org/ 10.1002/(SICI)1522-7278(199905)14:23.0.CO；2-L.

Seong,Y.C.,Sapari,N.,2003.Rainwater quality in Peninsular Malaysia.In: 11th InternationalRainwater CatchmentSystemsConference,Mexico City.

Shaheed,R.,Wan Mohtar,W.H.M.,2014.Potential of using rainwater for potable purpose in Malaysia w ith varying antecedent dry intervals.J. Teknol.72(1),57-61.

Shaheed,R.,Azhari,C.H.,Ahsan,A.,Wan Mohtar,W.H.M.,2015.Production and characterisation of low-tech activated carbon from coconut shell.J. Hydrol.Environ.Res.3(1),6-14.

Sukereman,A.S.,Suratman,R.,Hussin,K.,2013.The issues towards sustainability of water resource management and the current governance practice in managing water crisis in Malaysia.In:Proceedings of the International Conference on Social Science Research,ICSSR 2013,Penang, pp.301-319.

Thomas,P.R.,Greene,G.R.,1993.Rainwater quality from different roof catchments.Water Sci.Technol.28(3-5),291-299.

Vialle,C.,Sablayrolles,C.,Lovera,M.,Jacob,S.,Huau,M.C.,Montrejaud-Vignoles,M.,2011.Monitoring of water quality from roof runoff:Interpretation using multivariate analysis.Water Res.45(12),3765-3775. http://dx.doi.org/10.1016/j.watres.2011.04.029.

Williams,P.G.,Eng,B.,1987.A study of bacteria reduction by slow sand fi ltration.In:IWPCBiennialConference,PortElizabeth.National Institute forwater Research,Pretoria.

World Bank,2014.Water:Overview.http://www.worldbank.org/en/topic/ water/overview[Retrieved 18 July 2014].

World Health Organization(WHO),1993.Guidelines for Drinking Water Quality.World Health Organization,Geneva.

World Water Development Report(WWDR),2003.Water for People,Water for Life.United Nations Educational,Scientific and Cultural Organization (UNESCO)and Berghahn Books,Barcelona.

Yaziz,M.I.,Gunting,H.,Sapari,N.,Ghazali,A.W.,1989.Variations in rainwaterquality from roof catchments.Water Res.23(6),761-765.http:// dx.doi.org/10.1016/0043-1354(89)90211-X.

Received 22 July 2016；accepted 5March 2017 Available online 31 May 2017

This work was supported by the Universiti Kebangsaan Malaysia Grant (Grant No.GUP-2014-077).

*Corresponding author.

E-mail address:hanna@ukm.edu.my(Wan Hanna MeliniWan Mohtar).

Peer review under responsibility of Hohai University.

http://dx.doi.org/10.1016/j.wse.2017.05.002

1674-2370/?2017 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http:// creativecommons.org/licenses/by-nc-nd/4.0/).