lable at ScienceDirect

Journal of Cleaner Production 112 (2016) 3152e3163

Contents lists avai

Journal of Cleaner Production

journal homepage: www.elsevier .com/locate/ jc lepro

Evaluation of ultrafiltration and conventional water treatmentsystems for sustainable development: an industrial scale case study

Chun Ming Chew a, Mohamed Kheireddine Aroua a, *, Mohd Azlan Hussain a,Wan Mohd Zamri Wan Ismail b

a Department of Chemical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysiab Air Kelantan Sdn. Bhd., Bangunan PMBK, Jalan Kuala Krai, 15050 Kota Bahru, Kelantan, Malaysia

a r t i c l e i n f o

Article history:Received 21 December 2014Received in revised form28 September 2015Accepted 8 October 2015Available online 19 October 2015

Keywords:UltrafiltrationDrinking waterMedia filtrationSustainable development

* Corresponding author. Tel.: þ60 3 7967 4615; faxE-mail address: mk_aroua@um.edu.my (M.K. Arou

http://dx.doi.org/10.1016/j.jclepro.2015.10.0370959-6526/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Sustainable developments in water treatment systems are challenges in the 21st century. More indus-trial-scale drinking water treatment plants are using ultrafiltration (UF) membrane systems. Conven-tional media filtration is the mainstream treatment process for these plants especially in developingcountries. Evaluations of both industrial-scale UF and conventional drinking water treatment systemshave been carried out in this study. These treatment systems are evaluated based on 5 aspects which arecapital expenditure, operational expenditure, maintenance cost, treated water quality and water losses. Acase study water treatment plant in Malaysia which encompassed both the UF and conventional mediafiltration systems are used to elucidate this comparison study. River water source is fed as surface rawwater to both systems. The UF system has exhibited consistent filtrate quality regardless of the fluctu-ation of raw water quality. Sludge discharged from the UF backwash remains the same characteristics asthe feed water except with higher concentration of solids content. However, sludge from the conven-tional system contains high concentration of Aluminium residual originating from the coagulant. The UFsystem caused higher water losses compared to the conventional system. As for operation and main-tenance expenditures, the conventional systems are more economical. Sensitivity analyses have beencarried out on the capital expenditure and operational expenditure. Precaution measures have beentaken to ensure all data collected are relevant and accurate. More than 12 months of treatment systemsoperational data are collected, compiled and analysed to substantiate the results. This study intends tohighlight the commercial and environmental sustainability of both systems. The major contribution andnovelty of this work is that it provides useful reference to the decision makers and stakeholders on theselection of treatment process for industrial-scale drinking water facilities to accommodate their currentand future requirements.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Drinking water sources primarily come from freshwater inflowsuch as surface river water (Davies and Mazumder, 2003). Mostdrinking water treatment plants use conventional media filtrationprocess to physically remove particles as final polishing(Zouboulis et al., 2007). Water treatment technology advance-ment has allowed more effective alternative processes in large-scale. However, the main drawbacks of these alternative treat-ment processes compared with the conventional system is the

: þ60 3 7967 5206.a).

differing infrastructure and operational costs that require furtherevaluation (Emelko et al., 2011).

Conventional potable or drinking water treatment systemconsists of coagulationeflocculation, sedimentation and finalpolishing by media filtration. In this treatment system, coagulantis dosed and mixed with the raw water to allow the formation offine/large flocs. These flocs characteristics are key points for thesolid/liquid separation process in the subsequent treatment sys-tem (Zhao et al., 2014). The flocs characteristics depend on thecoagulant type/dosage, pH and temperature of the water(Mitrouli et al., 2008). Based on pilot-scale experiments, thefiltration time through the media filter and flocs characteristicssignificantly affect the quality of the filtrate water turbidity.Therefore maintaining filtrate quality in media filtration system is

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3153

a challenging task due to the constantly changing raw watercharacteristics.

Ultrafiltration (UF) is one of the most widely used alternativetreatment process in potable water production (Mierzwa et al.,2012). In the past two decade, UF has been one of the mostimportant technology advances inwater treatment systems (Huanget al., 2008). The smaller floor space required by UF systemscompared to media filtration systems are attractive to manystakeholders (Vedavyasan, 2007). Conventional media filtrationsystems are based on deep bed filtration process while the UFoperates on the surface or cake filtration principle. UF is capable ofreplacing the whole conventional potable water treatment processof sedimentation and media filtration system due to its ability toremove colloid, particles, bacteria and viruses from water (Tianet al., 2013b). The key feature of UF is its capability to control thepermeate quality by careful selection of the membrane materialand pore size (Arkhangelsky et al., 2012). However the majordrawback of UF systems in large-scale application is membranefouling which is tedious to control (Shirazi et al., 2010; Wang et al.,2014; El-Abbassi et al., 2014; Filloux et al., 2012; Xiarchos et al.,2003).

In general, UF is considered a more expensive technologycompare with the conventional media filtration systems (Mass�eet al., 2011). Detailed study has been carried out by researchers(Xiao et al., 2012) on industrial-scale UF membrane water treat-ment plant at Nantong, China. This study has shown that the UFsystem is capable to produce very stable filtrate quality with theabsence of coliform bacteria and 99% turbidity removal efficiency.The major operational problem in membrane system is the issueof fouling which incurred extra capital, labour, chemical and en-ergy cost to the industry (Barello et al., 2014). Similar study wasalso conducted on conventional media filtration water treatmentplant in South Africa (Makungo et al., 2011). Final water qualityresults indicated only partial compliance with recommendedturbidity guidelines. Due to the different raw water sources andlocations of both industrial-scale treatment plants, comparisonsbetween the two treatment systems are obscured and difficult tojustify.

Sustainable development is defined as integrated approaches toensure social and economic prosperity that are able to addressenvironmental sustainability with waste minimization (Khaliliet al., 2015). Businesses around the world are undertaking drasticchanges in their production strategies by incorporating the conceptof sustainable development (Sen et al., 2015). Many case studies(Nunes et al., 2014; Alkaya and Demirer, 2014; Liu et al., 2014;Handley et al., 2002) have been carried out to demonstrate sus-tainability in commercial production, environmental and wastemanagement. Stakeholders of water treatment plants consist ofmulti-facet backgrounds in technical, marketing, business admin-istration, quality assurance and skilled-workers. Categorized com-parison on commercial and environmental information is morecommon for ease of comprehension and to make decisions basedon their current requirements.

Prior work on practical performance analysis (Chew et al., 2015)was carried out on industrial-scale UF water treatment plant withno comparisons made against conventional systems. Evaluation oncommon design and operational issues such as UF feed waterquality/quantity, required membrane surface area, selection ofchemical for membrane cleaning, fouling of membrane, efficiencyof backwash and electricity consumption are elaborated withoutmuch emphasis on the commercial and environmental sustain-ability of UF systems in the case study. Another industrial-scale casestudy was carried at the Barcelona Metropolitan Area (Raich-Montiu et al., 2014) to evaluate membrane technologies mainlyreverse osmosis and reverse electro dialysis on organoleptic quality

of drinking water. In this case study, the quality of the filtrate fromthe membrane systems were evaluated extensively but withoutanalysis on the commercial aspects. These evaluations are useful fortechnical personnel such as engineers and technologist but mightbe too overwhelming for management level decision makers. Theobjective of this study is to evaluate and highlight comparisons ofthe industrial-scale UF against conventional media filtration sys-tems for potable water production. A total of 5 categories wereselected which encompass capital expenditure, operationalexpenditure, maintenance cost, treated water quality and waterlosses. These data are based on an industrial-scale potable watertreatment plant which operates both UF and conventional mediafiltration systems using the same river water source. The analysis ofindustrial-scale systems provide better perspective on actualoperational issues than typical laboratory researches and experi-ments to further substantiate previous findings. Besides commer-cial evaluations, environmental impacts on the 2 systems are alsopresented in this case study. The arguments presented from thesedata provide useful comparisons in terms of commercial andenvironmental sustainability. This study will provide essential in-formation to the decision makers and stakeholders on the selectionof treatment processes for industrial-scale potable water facilitieswith emphasis on sustainability.

Major obstacle in this work is to consistently ensure all thecollected data are relevant and accurate. Specific training and in-structions were given to the treatment plant operators to ensure alldaily samples analyses are conducted in accordance to the standardoperation procedures (SOP). Analysis equipment is calibrated toensure data accuracy. Consistent operational data recording andcompilation were carried out from 2013 to 2014 by competentoperation personnel. Gathering reliable data for any industrial-scale studies are a challenging task due to various uncertainties(Grassini et al., 2015). These obstacles have been overcome by thestrong support and co-operation given by the treatment plantowner in this case study.

2. Water treatment plant background and evaluationmethodology

This case-study water treatment plant was constructed inKelantan, Malaysia since more than 30 years ago to producedrinking water to its surrounding township. Fig. 1 shows the blockdiagram of the processes in the treatment plant. Initially this watertreatment plant utilizes the conventional media filtration system asits sole treatment process. Surface river water is channelled to thewater treatment plant for natural aeration process through acascading aerator. Aluminium based coagulant is dosed into theraw water and mixed through a series of baffle plates for thecoagulationeflocculation process to produce dense flocs. The flocsare then allowed to settle in two sedimentation clarifiers in parallel.Each of the clarifiers provide surface loading rate of 2.81 m/h,which falls between the recommended loading rate of 0.71e3.3 m/h (Kawamura, 2000). The clarified water is channelled to four unitsof gravity media filters which consist of fine sand measuring0.7e1.4 mm in diameter. Each of the filters have filtration rate of7.0 m/h which is below the maximum recommended filtration rateof 10.0 m/h for a full-scale drinking water treatment plant gravityfilter (Zouboulis et al., 2007). Chlorine gas is dosed into the filtrateto maintain the free residual chlorine of between 1.5 and 2.0 mg/Lbefore supplying to the consumer. This conventional treatmentsystem is capable of supplying up to 11,000 m3/d of potable waterto its surrounding areas. Fig. 2 shows the detailed block diagram ofthe conventional media filtration system.

In 2012, an UF membrane system was constructed in the watertreatment plant using the same surface river water as the feed

Fig. 1. Block diagram of processes in water treatment plant.

Fig. 2. Detailed block diagram of conventional media filtration system.

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e31633154

water. The UF system construction has been completed and com-mences operation on February 2013. Pre-treatment of the UF sys-tem consists of aeration of the feed water through a cascadingaerator and an extended aerator before proceeding for large par-ticulate screening by four units of pressurized sand filters withfiltration rate of 15 m/h to reduce the number of particles in thefiltrate (Kawamura, 2000). The pre-treated water is then pumped

into 120 units of UF membrane modules arranged in parallel (Chewet al., 2015). All the UF membranes are manufactured by IngeGmbH, Germany with modified Polyethersulfone (mPES) material.Sodium hypochlorite (NaOCl) is dosed in the UF permeate withsimilar free residual chlorine as the conventional media filtrationsystem for disinfection before supplying to the consumer. This UFmembrane treatment system is designed to supply up to a

Fig. 3. Detailed block diagram of UF system.

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3155

maximum of 14,000 m3/d of drinking water. Fig. 3 shows thedetailed block diagram of the UF system.

Previous study has concluded that natural organic matter(NOM) in various water sources is a major concern in water treat-ment processes (Joseph et al., 2012). This is mainly because thedifferent type and quantity of NOM affects the required dosage ofcoagulant to achieve optimum clarification results. The presence ofNOM also acts as foulant in UF membrane. Feed water character-istics play an important role in determining the operating condi-tions of treatment processes. In this case study, both the mediafiltration and UF systems are using the same source of surface riverwater as feed water. The characteristic of the feed water to bothsystems are exactly the same and comparisons made are morerelevant.

Operation and analysis data of bothwater treatment systems areevaluated and compiled into tables and graphs. Water sampleanalysis are carried out in-house within the water treatment plantown laboratory. Turbidity of water samples are measured usingHACH Laboratory Turbidimeter Model 2100N with EnvironmentalProtection Agency (EPA)eMethod 180.1. NOMhas been reported tohave direct correlation with dissolved organic carbon (DOC) insurface water samples (Kennedy et al., 2005). DOC is analysed usingHACH DR 2800 Spectrophotometer under Method 10129 for TotalOrganic Carbon (TOC) analysis procedures with samples passingthrough a 0.45 mm filter. Chemical oxygen demand (COD) of sam-ples are measured using reactor digestion method whileAluminium residuals are measured using Method 8012 with Alu-minon powder pillows. All these analysis methods are documentedin the operation manual of the DR 2800 Spectrophotometer.

3. Results and discussion

The operation and analysis data of the treatment plant iscompiled and presented in tables and graphs based on the indus-trial-scale water treatment plant operating conditions from 2013 to2014. These data provides useful comparison on the UF and con-ventional systems to elucidate the 5 categories in capital expen-diture, operational expenditure, maintenance expenses, treatedwater quality and water losses.

3.1. Capital expenditure

Capital expenditure (CAPEX) is the expenses to acquire fixedassets of the water treatment system which include the civilstructure, mechanical equipment, electrical control panels and

other necessary hardware to fulfil the treatment process re-quirements. The media filtration system was constructed morethan 30 years ago in 1980 at an approximate cost of USD 1,067,000.Based on yearly inflation rate published by “The World Bank” forMalaysia (TWB, 2014), the estimated cost of the conventional me-dia filtration system in the year 2013 would require USD 2,760,726to construct. Fig. 4 shows the yearly estimated increase in con-struction cost of the 11,000 m3/d conventional media filtrationsystem up till 2013. The construction cost for the UF system in 2013is much higher at USD 3,710,000. This is in accordance to literaturereport that capital cost of membrane system normally exceeds thatof conventional system by a significant amount (Pearce, 2007). Itwas also reported that UF system require less space/land comparedto conventional system. Table 1 shows the comparison in cost ofconstruction and land requirement for both water treatmentsystems.

After normalization of the cost of construction and landrequirement with the treatment capacity, construction cost of theUF system is only 5.6% higher than the conventional systemwhile itrequire 69.6% lesser land space to construct. The fact that UFmembrane price has been decreasing and smaller land re-quirements have made this treatment process very affordable to beimplemented in large-scale (Tian et al., 2013a). As most developedcities are scarce on land space, the cost to procure smaller piece ofland for the UF treatment system might eventually yield muchlower CAPEX than the conventional system. It is important tomentioned that the land values are dependent on location as urbanland would fetch far higher value than rural lands (Liu et al., 2015).Sensitivity analyses of land values against the CAPEX for both sys-tems are carried out and indicated in Fig. 5.

In this case study, the CAPEX consists of both the cost of con-struction and land. Sensitivity analysis can be effectively performedin order to identify the key parameters influencing certain evalu-ation aspects (Igos et al., 2014). Based on the construction cost/m3

and the required land shown in Table 1, the CAPEX to construct a14 million litres a day (14MLD) and 11 million litres a day (11MLD)conventional (media) and UF systems are shown in Fig. 5. Thisfigure indicates the increase in CAPEX when the land value variesfrom 10 to 300 USD/m2. It is estimated that when the land valueexceeds 170 USD/m2, the CAPEX for the UF system is much lowerthan the conventional system for both the 14MLD and 11MLDsystems. This analysis indicates that in urban area where the landvaluemight exceed 170USD/m2, it is very feasible to choose UF overthe conventional system due to the smaller land required whichtranslates to lower CAPEX.

USD 1,067,000 (Year 1980)

USD 2,760,726(Year 2013)

0

500,000

1,000,000

1,500,000

2,000,000

2,500,000

3,000,000

1980 2983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013

Estim

ated

con

stru

ctio

n co

st (U

SD)

Year

Fig. 4. Estimation of construction cost for conventional system based on yearly inflation rate.

Table 1Cost of construction and land requirement for both treatment systems.

Media filtration system UF system

Cost of construction in 2013 USD 2,760,726 (estimated) USD 3,710,000 (actual)Treatment capacity 11,000 m3/d 14,000 m3/dLand requirement 1376 m2 528 m2

Cost/m3 capacity USD 251 USD 265Land/m3 capacity 0.125 m2 0.038 m2

Fig. 5. Sensitivity analysis of various land value on CAPEX.

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e31633156

3.2. Operational expenditure

Operational expenditure (OPEX) is the expenses which arerelated to the operation of the treatment plants to produce thedesired drinking water quality and quantity. These are the cost of

the resources used to maintain the drinking water output from thefacilities. The major operational cost can be mainly categorized aschemical and electricity cost. In this case study treatment plant,chemicals are used as coagulant for the coagulationeflocculationprocess, treated water disinfection and cleaning of the UF

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3157

membrane. Figs. 2 and 3 show the dosing points of the coagulantand disinfectant chemicals.

It has been proven from research findings (Zahrim and Hilal,2013) that after coagulationeflocculation, media filtration iscapable of removing floc by attaching them to the media grain. Thistreatment plant is utilizing aluminium chlorohydrate (ACH) ascoagulant. ACH has been proven in laboratory experiments torequire far less dosage of between 60% and 70% than alum toremove organics for conventional water treatment (Wang et al.,2008). It has also shown that alkalinity consumption of ACH isalso less than alum in coagulated suspension. The pH of the ACHcoagulated suspension only marginally decreases to less than 0.5.Further pH adjustment to increase the alkalinity and pH is notnecessary for the final treated water. Chlorine (Cl2) gas is used asdisinfectant for this treatment system. Chlorine has been utilizedworldwide as disinfectant residual to counteract microbialcontamination and proliferation in potable water supply systems(Monteiro et al., 2014). The chlorine residual in the final treatedwater is maintained between 1.5 and 2.0 mg/L before supplying outfrom the treatment plant.

Pre-treatment of the UF system consists of air oxidation andscreening of the feed water by pressurized sand filters. The UFsystem does not require any coagulant prior to filtration. It has beenreported (Xiao et al., 2012) that coagulation have limited effect oneliminating organic foulants for UF systems. Large particles in thefeed water are removed by the pressurized sand filters prior toentering the UF membrane. Sodium hypochlorite (NaOCl) or liquidchlorine is used as the treated water disinfectant on the UF filtratewhile sodium hydroxide (NaOH) and hydrochloric acid (HCl) areused as membrane cleaning chemicals. NaOH and HCl have beenwidely used in UF membrane cleaning particularly in potable watertreatment plant due to its lower cost (Regula et al., 2014; Kumarand Pal, 2013).

Cl2 gas and NaOCl are both used as a disinfection agent in thefiltrate water. Cl2 gas in bulk quantity being the cheaper optionsbetween the 2 disinfectants have made it the most commonly useddisinfectant in water treatment plants (Kim et al., 2002). NaOClproduces free chlorine in water and can be handled without highexpertise and with minimal hazard as compared to chlorine gas(Garcia-Villanova et al., 2010). Both Cl2 gas and NaOCl have beenproven to be good disinfectant against virus and bacteria in potablewater (Kim et al., 2002). It has also been reported that NaOCl andCl2 gas produce harmful disinfection by-products known as tri-halomethanes (THM) which might affect human health and requireoptimum dosage to reduce it. The facilities requirements to dose Cl2gas into the filtered water are more complicated such as vacuumsystem, high pressure pump, regulator and lead gasket to preventgas leakages. Handling and dosing of NaOCl in liquid form is muchsimpler with just dosing pumps and storage tanks. Table 2 showsthe coagulants and disinfectants selection justifications. Each ofthese chemicals have their own advantages and disadvantages.

Table 2Selection justifications for coagulants and disinfectants.

Advantages

CoagulantACH Require less dosage than alum

Insignificant pH changes due to low alkalinity consumptioAlum Lower price per unit

Commonly availableDisinfectantChlorine gas Lower price per unit

Good disinfectant propertiesNaOCl (liquid) Ease of handling

Good disinfectant properties

Owners of the water treatment plants decide on the most appro-priate chemicals to suit their current needs. Table 3 shows thechemical used for both the media filtration and UF systems. Theconsumption of NaOCl and Cl2 are excluded in Table 3 as the se-lections of these 2 chemicals have no relation to the treatmentprocesses under this study.

The two treatment processes (conventional media filtration andUF systems) require different units of rotating mechanical equip-ment such as pumps, air blowers and compressors for operation.Table 4 summarizes the major rotating mechanical equipment ofboth systems. The UF system is very energy/electricity intensivecompared to the conventional system. It has been reported(Bonnelye et al., 2008) that with acceptable raw water quality, UFsystems are more costly than conventional media filtration sys-tems. The operational expenditure (OPEX) is much higher for UFsystems compared with conventional media filtration as it involveboth chemical and electricity cost as shown in Tables 3 and 4.Overall the UF system (USD 35.40) requires 3 times more opera-tional cost than media filtration system (USD 11.06) to produce1000 m3 of treated water. It is interesting to highlight that thechemical cost for UF system is only 43% of the chemical cost formedia filtration system as no coagulant is required to producefiltrate which is of better quality elaborated in the following sec-tions. The higher oxidation requirement (air blower), large partic-ulate screening by pressurized sand filter (pressurized filter feedpump) and higher feed pressure (UF feed pump) in Table 4 are themajor electricity consumption in UF system. Most of the processesinmedia filtration system utilize gravitational forces such asmixingin coagulationeflocculation, sedimentation clarification and filtra-tion which does not require any electricity. Other ancillary equip-ment electricity consumption for building services (eg. buildinglighting and air-conditioner) and treated water pumps to the res-ervoirs have been excluded in Table 4 since these electricity con-sumption are unrelated to both treatment processes. Both systemsare located at the same water treatment plant and the same groupsof operators are attending to both systems. The manpower opera-tion of the treatment plant is divided into 3 shifts with 2 operatorsfor each shifts. Based on the high level of automation in the UFsystem, the operators actually spend less time on the daily opera-tion of the UF systems compared to the conventional system. Arough estimated proportion of the time spent by the 2 operators are25% for UF system and 75% for conventional system. It is a regula-tory requirement in most water treatment plants that there shouldbe at least 2 individuals presence on each shifts for security reasons.Even though the UF systems may only require 1 individual for theoperation in theory, the general practice would not permit that.Assumptions made that the operators' wages for both systemswould not have a significant impact on the comparison of OPEX forboth systems are justified.

Sensitivity analysis has been carried out to determine the pricevariation of chemicals and electricity cost on the OPEX for both the

Disadvantages

nHigher price per unitNot commonly availableRequire much higher dosageSignificant pH drop and require chemical for post pH adjustment

Difficulty in handling and very hazardousProduce by-products (THM)Higher price per unitProduce by-products (THM)

Table 3Chemical dosage and cost for water treatment systems.

Media filtration system Ultrafiltration system

Average dosage Cost per 1000 m3 treated water Average dosage Cost per 1000 m3 treated water

Coagulant (ACH) 11.0 mg/L USD 9.57 e e

Sodium hydroxide (48%) e e Membrane cleaning USD 2.66Hydrochloric acid (33%) e e Membrane cleaning USD 1.46

Total cost of chemical per 1000 m3 oftreated water produced

USD 9.57 USD 4.12

Table 4Electricity consumed and cost for water treatment systems.

Media filtration system(Power consumed)

Ultrafiltration system(Power consumed)

Chlorine booster pump 4.0 kWh e

Backwash pump 0.2 kWh 2.5 kWhAir blower 0.1 kWh 32.0 kWhUF feed pump e 55.0 kWhPressurized filter feed pump e 30.0 kWhAir compressor e 0.2 kWhNaOCl/ACH dosing pump 0.2 kWh 0.3 kWhTotal electricity consumed in 1 h 4.5 kWh (To produce 458 m3

of treated water)120.0 kWh (To produce 583 m3

of treated water)Total electricity consumed per 1000 m3 of treated

water produced9.8 kWh 205.8 kWh

Total cost of electricity per 1000 m3 of treated water produced USD 1.49 USD 31.28

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e31633158

system in Fig. 6. The variations are based on the 4 assumptionslisted below:-

Category A e Chemical and electricity prices are same as statedin Tables 3 and 4.Category B e Electricity price is reduced by 50% and chemicalprice remains.Category C e Chemical price is increased by 100% and electricityprice remains.Category D e Electricity price is reduced by 50% and chemicalprice is increase by 100%

In all the categories in Fig. 6, it shows that the OPEX of theconventional system (media) is lower than the UF system. It isinteresting to highlight that in “Category D” when the electricityprice is reduced by 50% and the chemical price is increased by 100%,

Fig. 6. Sensitivity an

the margin between the 2 systems are the closest. This implies thatby reducing the cost of electricity production by using renewableenergy sources such as solar/wind might eventually makes UFsystem very feasible. Chemical costs are prone to global inflationand for the past 10 years, the prices of coagulant/membranecleaning chemicals in Malaysia are on the rise. In the near futurewhen electricity productions by renewable energy resources havebecome more accessible and the chemical cost continues to in-crease due to global inflation, the OPEX of UF system might even-tually become comparable to the conventional system.

3.3. Maintenance expenses

Maintenance expenses are the cost incurred to keep the treat-ment system's components in good working condition. The factthat UF system requires more mechanical rotating equipment

alysis of OPEX.

Table 5Maintenance cost for both systems.

Media filtration system Ultrafiltration system

Rotating equipment greasing and lubrication USD 100/year USD 200/yearCleaning of sludge in clarifier and filter tanks USD 1000/year e

Chemical cleaning of UF membrane e USD 1500/yearPiping system leakages repair and automation components maintenance USD 500/year USD 3500/year

Total cost of maintenance in a year USD 1600 USD 5200

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3159

entails higher maintenance cost as well. Based on the first year ofthe UF system operation, estimated maintenance cost comparisonsare shown in Table 5. It has been mentioned that UF systems aremore costly in both construction cost and OPEX, similar trend isalso observed in the maintenance expenses. Due to the higheroperating pressure of the UF system, the maintenance and repair ofpipe leakages aremore significant. Themaintenance cost is also oneof the factor which contributed to the higher price of the UF system(Bonnelye et al., 2008).

The UF system is designed to be fully automated with minimaloperators' intervention while the conventional media filtrationsystem require operators' manual adjustment on coagulant dosageand filter media backwash. The operators of this treatment plantactually spend far less time on the daily operation of the UF systemcomparedwith the conventional system. Hourly rawwater samplesare taken and analysed to ensure proper coagulant dosage for theconventional system. Higher level of automation in the UF systemrequires higher maintenance cost for the automation componentsas indicated in Table 5.

3.4. Treated water quality

Pilot-scale experiments have been carried out by researchers inparallel with both conventional media filtration and UF systemwith raw seawater (Guastalli et al., 2013). It has been proven thatthe UF showed excellent and high stability in the good quality of itspermeate. The same result was observed in this case study indus-trial-scale plant. Fig. 7 shows the final treated water quality of bothsystems and the raw water characteristics in the treatment plant.The turbidity of the conventional system's filtrate fluctuatewith theincrease of the raw water turbidity, while the UF system produceconsistent filtrate quality which is below 1.0 Nephelometric

0

2

4

6

8

10

12

14

16

18

20

UF

and

med

ia fi

lter f

iltra

te tu

rbid

ity (N

TU)

Date

Fig. 7. Turbidity of UF filtrate, med

Turbidity Units (NTU) all the time. Prior to the media filtration, thecoagulationeflocculation process plays an important role to ach-ieve the desired water quality. The coagulant dosage is dependenton the raw water quality in real-time and changes drastically dur-ing heavy rain conditions which cause high turbidity in the rawwater (Wu and Lo, 2010).

In this case study, UF system has exhibited its ability toconstantly produce low turbidity (high quality) of filtrate in com-parison to media filtration. The fluctuation of the raw waterturbidity does not seems to effect the quality of the UF permeate.No coagulant is required prior to the UF filtration system. Theaverage UF filtrate recorded in May 2013 was 0.37 NTU while themedia filter filtrate produced an average of 1.47 NTU. Disinfectantswere added into both filtrates when the turbidity values wererecorded. The local authority requirement for final treated waterturbidity is below 5 NTU.

The removal of DOCwith conventional media filtration system isdeemed to be unsatisfactory compared to pressure-driven mem-brane processes (Kabsch-Korbutowicz, 2006). Table 6 shows theDOC, COD and pH analysis from the surface raw water, mediafiltration and UF filtrates from this treatment plant. DOC has beenreported to be precursor to the formation of disinfection by-prod-ucts (Cool et al., 2014) which are linked to various health relateddiseases. The DOC in the surface raw water sample is only 3.6 mg/Lwhich is 50% lower than the global mean of 7.24 mg/L for surfacewater reported in literature (Cool et al., 2014). Pilot plant studiescarried out by other researchers (Guastalli et al., 2013) have proventhat the DOC removal efficiency of media filtration is lower than UFsystem, similar results are obtained in Table 6. COD is not a majorproblem to most of the Malaysia's potable water treatment plant(Hasan et al., 2011). High concentrations of COD lead to thedepletion of dissolved oxygen and the development of septic

0

10

20

30

40

50

60

Raw

wat

er tu

rbid

ity (N

TU)

Media filter

UF

Raw

ia filter filtrate and raw water.

Table 6DOC, COD and pH of water samples in May 2013.

Surface raw water Media filtration system Ultrafiltration system

DOC (mg/L) 3.6 2.1 0.7COD (mg/L) 8.0 2.0 5.0pH 7.1 6.7 6.9

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e31633160

condition. The COD of the surface raw water is 8 mg/L which fallswithin the recommended level of below 10 mg/L stipulated by theMinistry of Health, Malaysia. COD level of the filtrate from bothconventional (2 mg/L) and UF (5 mg/L) systems are very muchlower than reported laboratory drinkingwater COD level of 18mg/L(Sarkar et al., 2007). There is also a significant drop in pH on thefiltrate from the media filtration system due to the hydrolysis re-action of the ACHwhich produces Hþ as indicated in Eqn (1). The UFfiltrate has veryminor variation in pHwith the surface rawwater asno coagulant is added prior to filtration.

Al2ðOHÞ5Cl/Al2ðOHÞþ5 þ Cl� þ H2O/2AlðOHÞ3 þ Hþ þ Cl�

(1)

3.5. Water losses

Water losses are mainly due to sludge discharge, cleaning thefilter media and UF membrane through backwash process. Table 7shows the water losses of both systems in May 2013 which corre-late the same sampling period of the water quality in Fig. 7.

It has been reported in laboratory-scale studies that water lossesof UF membrane can be as high as 13.3% (Bai et al., 2013) to allowmore frequent sludge discharge interval and backwash to alleviatemembrane fouling. The fluctuating rawwater turbidity as indicatedin Fig. 7 has high tendency to cause membrane fouling. The higherwater losses are deemed required to thoroughly clean the mem-brane to an acceptable level for continuous operation andminimizemembrane fouling.

The media filtration system water losses are within the recom-mended level of less than 7% recommended by local authority.Major water losses of 4% of the total raw water occurred in theclarifiers due to the high suspended solids sedimentations after thecoagulationeflocculation process. Aluminium sludge amount inpotable water treatment plants are expected to be 1%e5%depending on the raw water quality (Tantawy, 2015). The remain-ingwater losses are through the backwash of themedia filter whichfall within the recommended 2%e3% reported in the literature(Kawamura, 2000).

It is interesting to highlight that even though the amount ofwater losses is higher in the UF system, the sludge composition isquite similar to the feed water except in a more concentrated formthan the solid contents of the surface raw water as no coagulant isadded prior to the filtration process. In contrast, the sludge dis-charged from the media filtration system contains high

Table 7Water losses for both systems in May 2013.

Raw water into system (average)Treated water produced (average)Water losses through sedimentation clarifiers (sludge discharge)Water losses through media filter/membrane backwashPercentage water lossesEstimated dry sludge generation (80% dryness)Aluminium residual in sludge

composition of Aluminium residual (58 mg/L) originating from thecoagulant as indicated in Table 7. Aluminium has been linked to theoccurrence of Alzheimer's diseases, children mental retardationand common effects of heavymetals accumulation (Tantawy, 2015).The sludge discharged from the UF system has lesser environ-mental impact than the sludge from the conventional system.

Further disposal of Aluminium sludge from this treatment plantin a more environmental friendly method is in progress and it shallincurred additional cost on the OPEX. Literature has reported thatAluminium sludge disposal can cost up to USD 130 per ton(Dassanayake et al., 2015). The monthly estimated Aluminiumsludge discharged from the conventional system comes up to 148ton in Table 7 which translates to an additional USD 1.85/m3 oftreated water produced. Sludge produced from the UF system canbe discharged at the river's downstream as there is no chemicaladded and it only differs in higher suspended solids content thanthe upstream river water.

4. Discussions

It should be emphasized that all the chemical, electricity, capitalexpenditure and maintenance cost are based on local Malaysiaevaluation in 2013 and 2014. Other regions of theworld would havemuch different costing due to currency exchange rate but the sametrend would be observed as all these commodities require energyintensive manufacturing and thus relying heavily on global market(Mercure and Salas, 2013). Sensitivity analyses on CAPEX and OPEXwere carried out and illustrated in Figs. 5 and 6. These analyseshave indicated that under certain circumstances (eg. land valueexceeding USD 170/m2, increase of chemical cost, decrease inelectricity production cost), the UF system might be more feasiblethan the conventional system. These comparisons were made ingeneral to elucidate the industrial-scalewater treatment systems toreflect the agreement and disagreement with the laboratory/pilot-scale experiments carried out by other researchers.

Sustainable development should encompass both commercialand environmental impacts of the water treatment systems whileensuring high quality of treated water outputs. UF has exhibitedconsistency in producing high quality filtrate without the use ofcoagulant. Even though the water losses of the UF system is higherthan conventional system, the sludge discharged duringmembranebackwash remains free of heavymetal residual from coagulant withonly higher solid concentration than the feed water. These arenatural sludges which pose very minimal environmental impact.The amount of membrane cleaning chemical used are small

Media filtration system Ultrafiltration system

11,152 m3/d 7420 m3/d10,413 m3/d 6600 m3/d4.0% e

2.6% 11.1%6.6% 11.1%148 ton/month 164 ton/month58 mg/L Not detected

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3161

compared to the amount of backwash water discharged. Electricityutilization rate of 0.21 kWh/m3 of UF filtrate is in close approxi-mation to literature report of 0.20 kWh/m3 (Pearce, 2008). The UFprocess is highly automated with very minimal operators' inter-vention on a small footprint for the facilities.

The conventional media filtration system requires less capital,operational and maintenance expenditure compared to UF sys-tems. However, it occupies larger land area for the facilities andnecessitates more manpower for the operation. The constructioncost for media filtration system is at USD 251/m3 of capacity whilethe UF system is USD 265/m3 of capacity. UF system requires muchsmaller land to construct which is 0.038 m2/m3 of capacity incomparison with the conventional system of 0.125 m2/m3 of ca-pacity. Report from literature shows agreement in these findingson cost of construction and land requirement (Pearce, 2007). OPEXfor the conventional system is at USD 11.06/1000 m3 of treatedwater while the OPEX of the UF system is 3 times higher at USD35.40/1000 m3 of treated water. This is due to the more energyintensive process of the UF system. Maintenance expenses of theconventional system are at USD 1600/year and the UF system atUSD 5200/year. The UF system requires much more energyintensive components such as pumps to provide the pressure formembrane filtration and results in higher maintenance cost. Boththe OPEX and maintenance expenses of the UF system is deemedhigher than conventional system in literature (Bonnelye et al.,2008).

The treated water quality of conventional system is highlydependent on the fluctuation of the feed water characteristic andthe response of the operators on the coagulant dosage. Sludgedischarged from the clarifiers and media filters are contaminatedwith heavy metal residual originates from the coagulant. Moreresources are required to treat these sludges to reduce the envi-ronmental impact. The treatment of the sludge is beyond thescope of this study. Capital expenditure (CAPEX), operationalexpenditure (OPEX) and maintenance expenses data of most in-dustrial-scale water treatment plants are commonly recorded formanagement review. These data provides very straight forwardindication on the commercial aspects of the treatment plants.Other relevant commercial aspects could be depreciation/appre-ciation of assets, profit losses due to treatment plant downtimeand water demand/supply. Table 8 lists all the possible commercialaspect of the treatment plants and justification on selection of the3 aspects (CAPEX, OPEX and maintenance expenses). Most in-dustrial-scale water treatment plants record water quality andwater losses data on daily basis for their internal evaluation. Otherrelevant data such as carbon emissions and sludge quantity/composition even though are useful but rarely kept proper records

Table 8Possible evaluation aspects and their selection justifications.

Possible aspects Justifications

Commercial aspectsCapital expenditure (CAPEX) Good indication on treaOperational expenditure (OPEX) Good indication on longMaintenance expenses Good indication on longDepreciation/appreciation of asset Fair indication of profitProfit losses due to treatment plant downtime Today's modern treatm

time, the plant is forcedPlant downtime is almo

Water demand/supply Water demand normallshall determine the via

Quality/quantity aspectsTreated water quality Good indication on treaWater losses Good indication on treaCarbon emissions Seldom recorded as it inSludge quantity/composition Seldom recorded as it in

by plant operators. Table 8 shows the possible quality/quantityaspects and justifications on the 2 selected aspects (treated waterquality and water losses). The selections of these 5 aspects aredeemed widely accepted and straight forward approach for com-parisons. Table 9 summarized all the results and discussion inthese 5 aspects of study. These aspects have been adopted becauseof its wide acceptance on engineering projects to represent com-mercial and environmental feasibility (Cheng, 2014; Zaman andLee, 2015; Dahal et al., 2015; Molinos-Senante et al., 2013; Brit-ton et al., 2013).

Evaluations and analyses about water infrastructure expansionhave long term implications because the water infrastructure sys-tems would likely be in place for decades (Qi and Chang, 2013;L�opez-Rold�an et al., 2015). Even though UF systems are currentlyan “expensive” technology for water treatment plants, it seems toprovide promising sustainability for the future. Sensitivity analyseson CAPEX and OPEX have shown commercial viability on UF sys-tems when global inflationwill cause land value and chemical priceto increase. The utilization of renewable energy would yield lowerelectricity production cost and further substantiate the lower OPEXfor UF systems. Further improvements are necessary to reduceelectricity consumption of UF systems to make it less energyintensive. Simultaneous electricity generation from renewablesources in water treatment system should be explored to reduceOPEX (Asghar et al., 2015). More robust facilities design should beimplemented to reduce downtime and maintenance cost. Themany benefits of UF offered to the consumers and to the environ-ment are worth the considerations of water treatment plantsstakeholders.

5. Conclusions

The primary purpose of this study is to evaluate the sustain-ability of industrial-scale UF and conventional potable watertreatment systems in terms of commercial and environmentalimpacts. Five categories have been compared to elucidate the sys-tems in these areas. The comparisons between both systems indi-cate higher CAPEX, OPEX, maintenance expenses and water lossesfor UF system. However, sensitivities analyses have been conductedon circumstanceswhereby the UF systemmight eventually bemorecommercially viable than conventional systems. These include highprices of land value, chemicals and lower electricity production costby using renewable energy sources. Advantages of the UF systemare exhibited through its stability to produce consistently goodquality of filtrate, smaller land requirement, non-toxic sludgedischarge and highly automated process with less manpowerrequired. The fluctuating surface water quality pose a great

tment plant feasibility to suit the allocated budgetterm commercial sustainability and to determine the end product pricingterm liability to sustain operation

/loss and require extensive market research and evaluationent plants are in normal operation for 99% of the time. Less than 1% of theto shut down for maintenance, replacement of spare parts or power failure.st negligible.y increases yearly as population grows. The operation cost and maintenance costbility of the treatment plant based on the current demand.

tment plant performance and recorded daily by operatorstment plant performance and recorded daily by operatorsvolves high level of operator competencyvolves high end instrumentation and equipment for recording

Table 9Summarized comparison of conventional filtration and UF systems.

No. Comparison aspect Media filtration UF system Literature report Comments

1. Capital expenditure Construction cost at USD 251/m3 of capacity. Landrequirement is 0.125 m2/m3 ofcapacity.

Construction cost at USD 265/m3 of capacity. Landrequirement is 0.038 m2/m3 ofcapacity.

Cost of construction for UFsystem is higher but requireless space to construct (Pearce,2007).

The construction cost of UFsystem is only 5.6% higher thanthe media filtration system,while the land requirement is69.6% less. In most urban areawhere land is scarce andexpensive, the UF system mightbe eventually cheaper in capitalexpenditure

2. Operational expenditure Chemical cost at USD 9.57/1000 m3 of treated water.Electricity cost at USD 1.49/1000 m3 of treated water.Electricity at 9.8 kWh/1000 m3

of treated water.

Chemical cost at USD 4.12/1000 m3 of treated water.Electricity cost at USD 31.28/1000 m3 of treated water.Electricity at 205.8 kWh/1000 m3 of treated water.

With acceptable quality of rawwater, UF system is much morecostly than conventionalsystem (Bonnelye et al., 2008).

Chemical cost for UF system isonly 43% of conventionalsystem. The higher operationalcost contributed by electricityutilization due to variousenergy intensive and highoperating pressure processes.

3. Maintenance expenses Maintenance expenses at USD1600/year

Maintenance expenses at USD5200/year

Due to various mechanicalequipment in UF system, thecost of maintenance is muchhigher as well (Bonnelye et al.,2008).

Operators spend less time ondaily operation of the fullyautomated UF system comparewith the manual conventionalsystem. The higher level ofautomation and operationpressure entails highermaintenance cost.

4. Treated water quality Less than 1 NTU irrespective ofraw water turbidity. Nocoagulant required. DOC, CODand pH of filtrate are incompliance.

Fluctuate between 1 and 4 NTUdepending on raw waterturbidity and adjustment ofcoagulant dosage is required.DOC, COD and pH are incompliance.

UF system has been proven toproduce stable good quality offiltrate (Guastalli et al., 2013).Coagulant dosage is required tobe adjusted to suit the changingquality of raw water in mediafiltration (Wu and Lo, 2010).

Fluctuating raw waterconditions require frequentcoagulant dosage adjustment inconventional system to ensurecompliance of quality filtrate.UF system is capable to producefar better filtrate quality evenwithout any coagulant dosage.

5. Water losses 6.6%.Sludge contains 58 mg/L of Alresidual

11.1%.No Al residual is detected insludge

UF system water losses can beas high as 13.3% to ensureproper cleaning of themembrane (Bai et al., 2013).Water losses of 4% (Tantawy,2015) at clarifiers and 2.6%(Kawamura, 2000) for filtermedia backwash is acceptable

Even though the water lossesfor the UF system is higher thanthe conventional system, thecomposition of sludge for theformer has less environmentalimpact due to the absence ofaluminium from the coagulant.

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e31633162

challenge for conventional media filtration to produce good qualityof filtrate. Precise coagulant dosage is deemed necessary to producedense flocs for conventional sedimentation and media filtration.Sludge discharged is substantially polluted with heavy metal re-sidual from the coagulant. UF system provides promising sustain-ability with no coagulant required for high quality filtrate. Theseanalyses and figures will have wide and applicable impact to thefuture direction of industrial-scale potable water treatment plantstakeholders. In this work, only one particular case study has beenhighlighted due to limited industrial-scale UF water treatmentplants available in Malaysia. Case studies on similar industrial-scalepotable water treatment plants which encompassed both UF andconventional systems using the same raw water source from otherregions of the world is highly recommended for more substantialjustifications.

Acknowledgement

This research was funded by University of Malaya PostgraduateResearch Grant, PPP (Project No.: PG041-2013B). The authorswould like to express gratitude to the Centre for Separation Sci-ence and Technology (CSST), University of Malaya for providingfacilities and technical support in this research. The kind assis-tance from Air Kelantan Sdn. Bhd. and Techkem Water Sdn. Bhd. toprovide actual plant data for this research paper is highlyappreciated.

References

Alkaya, E., Demirer, G.N., 2014. Sustainable textile production: a case study from awoven fabric manufacturing mill in Turkey. J. Clean. Prod. 65, 595e603.

Arkhangelsky, E., Duek, A., Gitis, V., 2012. Maximal pore size in UF membranes.J. Membr. Sci. 394e395, 89e97.

Asghar, A., Abdul Raman, A.A., Wan Daud, W.M.A., 2015. Advanced oxidation pro-cesses for in-situ production of hydrogen peroxide/hydroxyl radical for textilewastewater treatment: a review. J. Clean. Prod. 87, 826e838.

Bai, L., Qu, F., Liang, H., Ma, J., Chang, H., Wang, M., Li, G., 2013. Membrane foulingduring ultrafiltration (UF) of surface water: effects of sludge discharge interval(SDI). Desalination 319, 18e24.

Barello, M., Manca, D., Patel, R., Mujtaba, I.M., 2014. Neural network based corre-lation for estimating water permeability constant in RO desalination processunder fouling. Desalination 345, 101e111.

Bonnelye, V., Guey, L., Castillo, J.D., 2008. UF/MF as RO pre-treatment: the realbenefit. Desalination 222, 59e65.

Britton, T.C., Stewart, R.A., O'Halloran, K.R., 2013. Smart metering: enabler for rapidand effective post meter leakage identification and water loss management.J. Clean. Prod. 54, 166e176.

Cheng, Y.-M., 2014. An exploration into cost-influencing factors on constructionprojects. Int. J. Proj. Manag. 32, 850e860.

Chew, C.M., Aroua, M.K., Hussain, M.A., Ismail, W.M.Z.W., 2015. Practical perfor-mance analysis of an industrial-scale ultrafiltration membrane water treatmentplant. J. Taiwan Inst. Chem. Eng. 46, 132e139.

Cool, G., Lebel, A., Sadiq, R., Rodriguez, M.J., 2014. Impact of catchment geophysicalcharacteristics and climate on the regional variability of dissolved organiccarbon (DOC) in surface water. Sci. Total Environ. 490, 947e956.

Dahal, K., Al-Arfaj, K., Paudyal, K., 2015. Modelling generator maintenance sched-uling costs in deregulated power markets. Eur. J. Oper. Res. 240, 551e561.

Dassanayake, K.B., Jayasinghe, G.Y., Surapaneni, A., Hetherington, C., 2015. A reviewon alum sludge reuse with special reference to agricultural applications andfuture challenges. Waste Manag. 38, 321e335.

C.M. Chew et al. / Journal of Cleaner Production 112 (2016) 3152e3163 3163

Davies, J.-M., Mazumder, A., 2003. Health and environmental policy issues inCanada: the role of watershed management in sustaining clean drinking waterquality at surface sources. J. Environ. Manag. 68, 273e286.

El-Abbassi, A., Kiai, H., Raiti, J., Hafidi, A., 2014. Application of ultrafiltration for oliveprocessing wastewaters treatment. J. Clean. Prod. 65, 432e438.

Emelko, M.B., Silins, U., Bladon, K.D., Stone, M., 2011. Implications of land distur-bance on drinking water treatability in a changing climate: demonstrating theneed for “source water supply and protection” strategies. Water Res. 45,461e472.

Filloux, E., Gallard, H., Croue, J.P., 2012. Identification of effluent organic matterfractions responsible for low-pressure membrane fouling. Water Res. 46,5531e5540.

Garcia-Villanova, R.J., Oliveira Dantas Leite, M.V., Hernandez Hierro, J.M., de CastroAlfageme, S., Garcia Hernandez, C., 2010. Occurrence of bromate, chlorite andchlorate in drinking waters disinfected with hypochlorite reagents. Tracingtheir origins. Sci. Total Environ. 408, 2616e2620.

Grassini, P., van Bussel, L.G.J., Van Wart, J., Wolf, J., Claessens, L., Yang, H.,Boogaard, H., de Groot, H., van Ittersum, M.K., Cassman, K.G., 2015. How good isgood enough? Data requirements for reliable crop yield simulations and yield-gap analysis. Field Crop Res. 177, 49e63.

Guastalli, A.R., Simon, F.X., Penru, Y., Kerchove, A. d, Llorens, J., Baig, S., 2013.Comparison of DMF and UF pre-treatment for particulate material and dis-solved organic matter removal in SWRO desalination. Desalination 322,144e150.

Handley, C., Brandon, N.P., van der Vorst, R., 2002. Impact of the European Unionvehicle waste directive on end-of-life options for polymer electrolyte fuel cells.J. Power Source 106, 344e352.

Hasan, H.A., Abdullah, S.R.S., Kamarudin, S.K., Kofli, N.T., 2011. Response surfacemethodology for optimization of simultaneous COD, NH4

þeN and Mn2þ removalfrom drinking water by biological aerated filter. Desalination 275, 50e61.

Huang, H., Young, T.A., Jacangelo, J.G., 2008. Unified membrane fouling index forlow pressure membrane filtration of natural waters: principles and method-ology. Environ. Sci. Technol. 42, 714e720.

Igos, E., Dalle, A., Tiruta-Barna, L., Benetto, E., Baudin, I., Mery, Y., 2014. Life cycleassessment of water treatment: what is the contribution of infrastructure andoperation at unit process level? J. Clean. Prod. 65, 424e431.

Joseph, L., Flora, J.R.V., Park, Y.-G., Badawy, M., Saleh, H., Yoon, Y., 2012. Removal ofnatural organic matter from potential drinking water sources by combinedcoagulation and adsorption using carbon nanomaterials. Sep. Purif. Technol. 95,64e72.

Kabsch-Korbutowicz, M., 2006. Impact of pre-coagulation on ultrafiltration processperformance. Desalination 194, 232e238.

Kawamura, S., 2000. Integrated Design and Operation of Water Treatment Facilities.J.W. Sons, New York.

Kennedy, M.D., Chun, H.K., Quintanilla Yangali, V.A., Heijman, B.G.J., Schippers, J.C.,2005. Natural organic matter (NOM) fouling of ultrafiltration membranes:fractionation of NOM in surface water and characterisation by LC-OCD. Desa-lination 178, 73e83.

Khalili, N.R., Duecker, S., Ashton, W., Chavez, F., 2015. From cleaner production tosustainable development: the role of academia. J. Clean. Prod. 96, 30e43.

Kim, B.R., Anderson, J.E., Mueller, S.A., Gaines, W.A., Kendall, A.M., 2002. Literaturereviewdefficacy of various disinfectants against Legionella in water systems.Water Res. 36, 4433e4444.

Kumar, R., Pal, P., 2013. Turning hazardous waste into value-added products: pro-duction and characterization of struvite from ammoniacal waste with newapproaches. J. Clean. Prod. 43, 59e70.

Liu, M., Woudstra, T., Promes, E.J.O., Restrepo, S.Y.G., Aravind, P.V., 2014. Systemdevelopment and self-sustainability analysis for upgrading human waste topower. Energy 68, 377e384.

Liu, Y., Luo, T., Liu, Z., Kong, X., Li, J., Tan, R., 2015. A comparative analysis of urbanand rural construction land use change and driving forces: implications forurbanerural coordination development in Wuhan, Central China. Habitat Int.47, 113e125.

L�opez-Rold�an, R., Platikanov, S., Martín-Alonso, J., Tauler, R., Gonz�alez, S.,Cortina, J.L., 2015. Integration of ultravioletevisible spectral and physico-chemical data in chemometrics analysis for improved discrimination of watersources and blends for application to the complex drinking water distributionnetwork of Barcelona. J. Clean. Prod. http://dx.doi.org/10.1016/j.jclepro.2015.1006.1074.

Makungo, R., Odiyo, J.O., Tshidzumba, N., 2011. Performance of small water treat-ment plants: the case study of Mutshedzi water treatment plant. Phys. Chem.Earth Parts A/B/C 36, 1151e1158.

Mass�e, A., Thi, H.N., Legentilhomme, P., Jaouen, P., 2011. Dead-end and tangentialultrafiltration of natural salted water: influence of operating parameters onspecific energy consumption. J. Membr. Sci. 380, 192e198.

Mercure, J.-F., Salas, P., 2013. On the global economic potentials and marginal costsof non-renewable resources and the price of energy commodities. Energy Policy63, 469e483.

Mierzwa, J.C., da Silva, M.C.C., Veras, L.R.V., Subtil, E.L., Rodrigues, R., Li, T.,Landenberger, K.R., 2012. Enhancing spiral-wound ultrafiltration performancefor direct drinking water treatment through operational procedures improve-ment: a feasible option for the Sao Paulo Metropolitan Region. Desalination 307,68e75.

Mitrouli, S.T., Yiantsios, S.G., Karabelas, A.J., Mitrakas, M., Follesdal, M., Kjolseth, P.A.,2008. Pretreatment for desalination of seawater from an open intake by dual-media filtration: pilot testing and comparison of two different media. Desali-nation 222, 24e37.

Molinos-Senante, M., Hern�andez-Sancho, F., Sala-Garrido, R., Cirelli, G., 2013. Eco-nomic feasibility study for intensive and extensive wastewater treatmentconsidering greenhouse gases emissions. J. Environ. Manag. 123, 98e104.

Monteiro, L., Figueiredo, D., Dias, S., Freitas, R., Covas, D., Menaia, J., Coelho, S.T.,2014. Modelling of chlorine decay in drinking water supply systems usingEPANET MSX. Procedia Eng. 70, 1192e1200.

Nunes, B., Bennett, D., Marques Júnior, S., 2014. Sustainable agricultural production:an investigation in Brazilian semi-arid livestock farms. J. Clean. Prod. 64,414e425.

Pearce, G.K., 2007. The case for UF/MF pretreatment to RO in seawater applications.Desalination 203, 286e295.

Pearce, G.K., 2008. UF/MF pre-treatment to RO in seawater and wastewater reuseapplications: a comparison of energy costs. Desalination 222, 66e73.

Qi, C., Chang, N.-B., 2013. Integrated carbon footprint and cost evaluation of adrinking water infrastructure system for screening expansion alternatives.J. Clean. Prod. 60, 170e181.

Raich-Montiu, J., Barios, J., Garcia, V., Medina, M.E., Valero, F., Devesa, R., Cortina, J.L.,2014. Integrating membrane technologies and blending options in water pro-duction and distribution systems to improve organoleptic properties. The caseof the Barcelona Metropolitan Area. J. Clean. Prod. 69, 250e259.

Regula, C., Carretier, E., Wyart, Y., Gesan-Guiziou, G., Vincent, A., Boudot, D.,Moulin, P., 2014. Chemical cleaning/disinfection and ageing of organic UFmembranes: a review. Water Res. 56, 325e365.

Sarkar, B., Venkateswralu, N., Rao, R.N., Bhattacharjee, C., Kale, V., 2007. Treatmentof pesticide contaminated surface water for production of potable water by acoagulation-adsorption-nanofiltration approach. Desalination 212, 129e140.

Sen, P., Roy, M., Pal, P., 2015. Exploring role of environmental proactivity in financialperformance of manufacturing enterprises: a structural modelling approach.J. Clean. Prod. http://dx.doi.org/10.1016/j.jclepro.2015.1005.1076.

Shirazi, S., Lin, C.-J., Chen, D., 2010. Inorganic fouling of pressure-driven membraneprocesses d a critical review. Desalination 250, 236e248.

Tantawy, M.A., 2015. Characterization and pozzolanic properties of calcined alumsludge. Mater. Res. Bull. 61, 415e421.

Tian, J.-y., Ernst, M., Cui, F., Jekel, M., 2013a. Effect of particle size and concentrationon the synergistic UF membrane fouling by particles and NOM fractions.J. Membr. Sci. 446, 1e9.

Tian, J.Y., Ernst, M., Cui, F., Jekel, M., 2013b. Correlations of relevant membranefoulants with UF membrane fouling in different waters. Water Res. 47,1218e1228.

TWB, 2014. The World Bank. Inflation, Consumer Prices (Annual %). Retrieved 11-10-2014, from data.worldbank.org/indicator/FP.CPI.TOTL.2G.

Vedavyasan, C.V., 2007. Pretreatment trends d an overview. Desalination 203,296e299.

Wang, J., Cui, Z., Jia, H., Zhang, H., 2014. The effect of fiber length on non-uniformand hysteresis phenomenon in hollow fiber membrane backflushing. Desali-nation 337, 98e108.

Wang, J., Guan, J., Santiwong, S.R., Waite, T.D., 2008. Characterization of floc size andstructure under different monomer and polymer coagulants on microfiltrationmembrane fouling. J. Membr. Sci. 321, 132e138.

Wu, G.-D., Lo, S.-L., 2010. Effects of data normalization and inherent-factor on de-cision of optimal coagulant dosage in water treatment by artificial neuralnetwork. Expert Syst. Appl. 37, 4974e4983.

Xiao, P., Xiao, F., Wang, D.-S., Qin, T., He, S.-P., 2012. Investigation of organic foulantsbehavior on hollow-fiber UF membranes in a drinking water treatment plant.Sep. Purif. Technol. 95, 109e117.

Xiarchos, I., Doulia, D., Gekas, V., Tragardh, G., 2003. Polymeric ultrafiltrationmembranes and surfactants. Sep. Purif. Technol. 32, 215e278.

Zahrim, A.Y., Hilal, N., 2013. Treatment of highly concentrated dye solution bycoagulation/flocculation-sand filtration and nanofiltration. Water Resour. Ind. 3,23e34.

Zaman, M., Lee, J.H., 2015. Optimization of the various modes of flexible operationfor post-combustion CO2 capture plant. Comput. Chem. Eng. 75, 14e27.

Zhao, Y.X., Gao, B.Y., Zhang, G.Z., Phuntsho, S., Shon, H.K., 2014. Coagulation by ti-tanium tetrachloride for fulvis acid removal: factors influencing coagulationefficiency and floc characteristics. Desalination 335, 70e77.

Zouboulis, A., Traskas, G., Samaras, P., 2007. Comparison of single and dual mediafiltration in a full-scale drinking water treatment plant. Desalination 213,334e342.