Resources, Conservation and Recycling 49 (2006) 1–13

Re-utilisation conditions of wastewaters fromtextiles industries

Selene Maria de Arruda Guelli Ulson de Souza ∗,Aline Resmini Melo, Antonio Augusto Ulson de SouzaFederal University of Santa Catarina, Chemical Engineering Department, PO Box 476,

CEP 88040-900, Florianopolis, SC, Brazil

Received 29 October 2005; received in revised form 20 February 2006; accepted 3 March 2006Available online 17 April 2006

Abstract

The simulation is an important tool for the analysis of the possibilities in the re-utilisation amongthe steps of the industrial production process. The analysis of technical and economical viabilitycome from the definition of the strategies for direct re-utilisation of water and water re-utilisationusing regenerative processes or with dilution from industrial water supply streams. Considering thecomplexity of the contaminant matrix of textile effluents, the definition of the global parameters, suchas organic mass and colour, is essential to make the analysis procedure viable in the industrial process.This paper aims to give an efficient tool for industrial purpose, which can give a quickly predictionwhen configuration changes are suggested by industrial engineers in the continuous washing process.The new suggested conditions are introduced in the algorithm. Using the conservation balances,guaranteeing the global mass conservation, it is possible to obtain the total organic mass and waterconsumption in this new configuration and to compare with previous results to choose the best optionfor the process. A case study is presented using a computational algorithm where the best conditionswere obtained for the re-utilisation of the streams in the textile industry. The proposed simulationmodel for continuous washing processes showed to be efficient in the optimisation of process withrecycle, being possible to simulate several possibilities of reuse for a single case. The new configurationof streams has been analysed, giving higher reduction of water consumption (16.15%) and of organic

∗ Corresponding author. Tel.: +55 48 3331 9448; fax: +55 48 3331 9687.E-mail addresses: selene@enq.ufsc.br (S.M. de Arruda Guelli Ulson de Souza), alinermelo@yahoo.com.br

(A.R. Melo), augusto@enq.ufsc.br (A.A.U. de Souza).URL: http://www.enq.ufsc.br/labs/LABSIN.

0921-3449/$ – see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.resconrec.2006.03.001

2 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

mass (1.43%) compared to the previous case. The developed software is an important tool for theindustrial engineers.© 2006 Elsevier B.V. All rights reserved.

Keywords: Re-utilisation; Rationalisation; Water; Regenerative process and textile industry

1. Introduction

In Brazil, the textile sector is of great importance and has taken into account the preser-vation of hydrical resources, since the textile industry requires a high volume of waterin order to produce textiles, as discussed by Feitkenhauer and Meyer (2001). Thereforethe minimisation and reduction of water consumption on the textile processes are impor-tant goals of industries, reducing the effluents production, the environmental impact and thecosts of water supply and effluent treatment (Almato et al., 1999; Dhole et al., 1996; Majozi,2005).

Numerical simulation is a good tool to predict the best conditions to the water reusein industrial processes (Bagajewicz, 2000; Chen and Hung, 2005; Puigjaner et al., 2000).According to Smith (2000), the use of optimization by mathematical programing, togetherwith conceptual points of view, is one of the great trends in the development of integrationprocesses technology.

Textile effluents are characterised for being colourful due to the dyes that do notfix to the fibres during the dyeing process. The pollution caused to the water streamsby these compounds, lead to changes in biological cycles mainly affecting the photo-synthesis processes besides the visual pollution. Nevertheless, studies have shown thatsome dye groups, mainly azo dyes and their by-products, can be carcinogenic and/ormutagenic.

This work presents a proposal on the reuse of effluents, aiming to increase the efficiencyon water and energy consumption by minimising or recycling effluents generated in thewashing process of the textile industry. This procedure permits that a lower volume ofeffluents is generated besides lowering water consumption. A computational algorithmwas developed to determine the possibilities of effluents re-utilisation from the continuouswashing process of the textile industry. This process was chosen due to the high waterconsumption, 28.06 m3 h−1 per machine. The platform MATLAB® (2004) was chosen forsoftware development with a user-friendly interface, making easy the interaction betweenoperator and software.

In this work a case study will be presented, considering a continuous washing unit withsix boxes where a model was proposed for its optimisation.

2. Methodology

In this paper, it was decided to develop a simple algorithm, for industrial purpose,which can give a quickly prediction when configuration changes are suggested by industrialengineers in the continuous washing process.

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 3

In spite of the rigorous simulators are available for university researchers, the decisionof the authors was to develop a new algorithm of easy industrial application, includingall steps of textile process, lowering the water consumption, with higher washing processefficiency, without using commercial softwares. The MATLAB software was used only tothe treatment of the numerical data.

All results have been not calculated by the MATLAB optimizer. The stream splitterfactors have been estimated by the authors. The MATLAB software was used only to thetreatment of the numerical data.

The software developed in this work allows to the user inputting the data, trackingthe process and choosing data reports after processing. Numerical data are converted intographics for better comprehension being this visualisation an important part in the simulationbecause it makes easier the quick verification of data tendency.

Through this interface, the following input data are specified: number of units and pro-cess streams, process topology, information about the parameters of each unit, areic mass,velocity and fabric breadth, rate of washed fabric, entry parameters and cut streams and thetime interval for building graphics. For each provided datum there is a verification of itsconsistency and then it is filled into a matrix. All data are organised in a database for furtherutilisation and actualisation.

The data related to the process flowsheet are entered in the software as a matrix calledProcess Matrix, because it is an efficient and compact way for handling these data. Threeother matrixes are derived from the Process Matrix: Adjacency Matrix, Incident Matrix andStream Connection Matrix. These matrixes are utilised in the development of the algorithms.

These are the basic information of the process, which are transferred to the modularunits and these run in the sequence determined by the calculation order. The calcu-lation order is the same of the mass flow rate but if any recycle is verified then theorder is changed. If there are recycles in the process, they are identified from Tiernan(1970) algorithm. Recycles will be open by the “cut streams” where an initial estima-tive will be provided for composition and physical properties. The balance of the modularunit is carried out through iterative methods until values in the range of tolerance areobtained.

Two types of output data are available as results of simulation: graphics and results asmatrixes.

2.1. Mathematical model

The adopted mathematical models of the process units have to be as more representativeas possible; therefore the results of process simulation can be corroborated by the experi-mental results. Three types of modular units were developed for program testing: unit box,stream splitter and stream mixer.

2.1.1. Unit stream mixerThe modular unit stream mixer was developed aiming to introduce a fluid stream mixer

into the model and it was considered an adiabatic mixing process. The number of entrystreams in this unit is variable and it is specified through the Process Matrix.

4 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

The following equations were utilised for modelling:

Qf =Ni∑

i=1

Qi (1)

CODf =∑Ni

i=1Qi · CODi∑Nii=1Qi

(2)

Tf =∑Ni

i=1Qi · Ti∑Nii=1Qi

(3)

where Qf, CODf and Tf are volume flow rate, mass concentration and temperature, respec-tively, of the output stream f of the mixer. The variables Qi, CODi and Ti are volume flowrate, mass concentration and temperature, respectively, of the entry stream i of the mixerand Ni is the number of entry streams of the unit.

2.1.2. Unit stream splitterThe modular unit stream splitter was developed aiming to introduce a fluid stream splitter

into the model. In the development of this unit, the following simplifications were used:the number of output streams is variable and is specified through the Process Matrix; thefactors of the stream splitter should be provided as unit parameters through the interface(“Information of each unit”), where the sum of these parameters is 1.

The following balance equations were utilised for modelling:

Qf = ϕf Qi (4)

CODf = CODi (5)

Tf = Ti (6)

where the variable ϕf represents the factor of the stream splitter.

2.1.3. Unit boxThe modular unit box represents each tank of the continuous washer, which removes

the excess of dye not fixed to the fabric. To develop this unit, the following simplificationswere used: the output stream has the same mass concentration and temperature of the tank,therefore the unit box is considered totally mixed and the tank volume is considered constant.

The following balance equations were utilised for modelling:

Qi|Pick-up = G · v · lar · PUi

ρ(7)

Ni∑

i=1

Qi =Nf∑

f=1

Qf (8)

CODf = CODj (9)

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 5

CODj = t · ∑Nii=1(Qi · CODi) + COD0

j · V

V + t · ∑Nii=1(Qi)

(10)

Tf = Tj (11)

Tj =∑Ni

i=1(Qi · Ti)∑Nii=1(Qi)

(12)

where the variables Qi|Pick-up, G, v, lar, PUi, V and ρ are related to the water pick-up volumeflow rate, areic mass, fabric velocity, fabric breadth, water/fabric pick-up ratio, volume ofunit box and water density, respectively.

In the last equations, “i” refers to the entry streams, “f” refers to the output streams ofthe box and “j” refers to unit box. Nf is the number of output streams of the unit and COD0

j

is the initial COD of the unit box.Eq. (7) is only utilised in the streams associated to the wet fabric where the water/fabric

pick-up ratio (PUi) of each stream is provided, determining the “squeeze” effect of thefoulard or solution retention by the fabric, in a continuous process.

2.1.4. Source termsThe source terms are factors associated to the attrition of rolls in the unit box of the con-

tinuous washer. They are dimensionless process parameters of the boxes of the continuouswasher.

These source terms are used to make the model compatible with the real situation ofthe industrial process, being evaluated through the relation of the simulation data obtainedfrom the global mass balance, without the source term, and the data obtained from textileindustry.

The organic mass related to the dyeing bath inside of each box of continuous washeris function of the dye mass removed of textile fibres and of the chemicals products addedin each box of continuous washer (acids, surfactants (SDS) and softener). The attrition ofrolls has an important influence to remove the hydrolysed dye present in the fibres. Theevaluation of this individual influence in numerical values is very difficult in the industrialprocess. The influence of these parameters is considered as overall coefficient, αj, in theconception of the developed modelling.

This coefficient is evaluated using the simulation program considering the balance ofmass transfer, without the source term related to the removed dye from the textile fibres.The COD evaluation for this procedure is compared to the measured COD value obtainedfrom industrial processes.

Eq. (13) is used to calculate the source term (αj) related to the box j. The dependenceof mass flow rate is guaranteed by the term

∑Nff=1qf , which represents the total mass flow

rate of the output streams of the box j:

αj = CODindj

COD∗j

− 1 (13)

6 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

where

αj = k0j

Nf∑

f=1

qf (14)

The variable CODindj represents the COD value obtained in the industrial process in the

box j and COD∗j is the initial COD value calculated by modelling developed without the

source term. The variable k0j is the characteristic constant of box j, qf is the mass flow rateof the output stream f and αj is the source term of box j.

The parameter αj is negative in the first box, once the total COD value is introduced intothe stream 1. With that, the exceeding COD value is adjusted.

The coefficient αj assumes positive values, when the dyes are removed from fibres to thedyeing bath, thought the action of washing process. Negative values of αj indicate that theadsorption process of the chemical species, from the dyeing bath, as acids, surfactants andsoftener, is occurring in the fibres.

The parameter αj depends on the adjustments of the rolls, affinity of the dye for the fibreeven considering the dye as hydrolysed, areic mass, fabric velocity and water mass flowrate in the washed fabric.

The attrition of rolls and the addition of acid, surfactant or softener can influence thetotal COD reduction from fibres. The evaluation of this influence is very complex and is afunction of the molecular structure of the dyes and the fibres, the ionic force of surfactantsand softener, which are added in box 1.

The output streams are corrected using the source term αj obtained through Eqs. (13)and (14). The new values of the CODj (CODj|new) are obtained by

CODj|new =∑Nf

f=1(Qf · CODf )∑Nf

f=1(Qf )(15)

The COD reduction in each box j (COD reductionj) is calculated by the following equa-tion:

COD reductionj = CODj|new − t · ∑Nii=1(Qi · CODi) + COD0

j · V

V + t · ∑Nii=1(Qi)

(16)

The variable WCOFj is the mass ratio and represents the organic mass reduction of fabricper fabric mass, in the box j, which is expressed as g kg−1 of fabric. This variable WCOFj

is calculated by the following equation:

WCOFj = COD reductionj

water volume flow ratefeeding

mass flow ratedry fabric(17)

where WCOFj (g kg−1 of fabric) is the efficiency of organic mass reduction per washedfabric mass; water volume flow ratefeeding (m3 h−1) is the water pick-up volume flowrate + industrial water volume flow rate + chemicals volume flow rate + recycled water vol-ume flow rate; water pick-up volume flow rate is the water volume flow rate that enters ineach unit box transported by the fabric; industrial water volume flow rate is the water volumeflow rate that enters in each unit box, from the effluent treatment plant (ETP), usually with

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 7

Fig. 1. Flowsheet of the washing process.

COD equal to zero; chemicals volume flow rate is the volume flow rate of the surfactants,softener or acid that enters in the unit box; recycled water volume flow rate is the watervolume flow rate from another unit.

In the methodology developed for modelling and simulation of the continuous washer, theentry parameter, WCOW1, which represents the mass ratio (g kg−1 of water), was estimatedthrough the following equation:

WCOW1 = WCOFimpreg − WCOFwashed

PU1(18)

where WCOFimpreg is the organic mass reduction ratio (g kg−1 of dry fabric) in the fab-ric leaving the dyeing bath, measured through experimental analyses; WCOFwashed is theorganic mass reduction ratio (g kg−1 of dry fabric), measured from a fabric sample collectedin the output of the last box of the continuous washer; PU1 is the water/fabric pick-up ratio(kg of water kg−1 of dry fabric) from the entry stream 1.

This estimative assumes that all organic mass contained in the fabric, which has just leftthe dyeing bath, is available in the stream 1. This is assumed due to the determination ofthis COD value to be done by a fabric sample submitted to several washings.

2.2. Process description

Fig. 1 shows the flowsheet of the continuous washer utilised as a case study in this work,where each box is numerated, from 1 to 6. The streams in this process are numbered from1 to 17.

The following values were considered in this case study: areic mass = 300 g m−2;breadth = 1.36 m; fabric velocity = 80 m min−1; water mass flow rate in the washed fab-ric = 1.96 t h−1.

3. Results and discussions

Samples were taken from the fabric that left the dyeing bath and is entering the washingprocess as well as the one after the washing process. Experimental analyses of these sampleswere carried out involving several washings and the following values of mass ratio wereobtained: the WCOFimpreg, organic mass reduction ratio of the impregnated fabric from thedyeing bath, is 16.44 g kg−1 of fabric; the WCOFwashed, organic mass reduction ratio of thefabric after washing, is 6.83 g kg−1 of fabric.

8 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

Table 1Information on the washing boxes

Box Volume (L) Output COD (mg L−1) Temperature (◦C)

1 650.0 405.00 60.02 800.0 1061.00 85.03 1100.0 192.00 85.04 1000.0 128.00 75.05 1300.0 1847.00 55.06 1100.0 1735.00 55.0

Table 2Information on the entry streams in the washing process

Stream Q (m3 h−1) COD (mg L−1) Temperature (◦C)

1 1.66 11305.88 20.02 9.00 0.00 65.05 9.00 0.00 95.07 7.50 0.00 70.09 0.90 7714.00 25.0

In the steady state, samples were collected from the output streams of each unit box andthe experimental COD values were measured (mg L−1).

Table 1 shows the volume of each box as well as the industrial COD values for the outputstreams and the temperature of each box. The values of the volume flow rate, COD andtemperature for the entry streams are shown in Table 2.

The surfactants and softener are injected in the stream 1. It is observed in the stream 9that acetic acid is injected, aiming to neutralise the pH of previous strong alkaline processesand also, to facilitate the action of other auxiliary carriers.

The water/fabric pick-up ratio for each stream transported by the fabric is shown inTable 3. With this information it is possible to simulate the process and obtain the valuesfor volume flow rate, COD and temperature of each stream and each unit box.

The calculated values of the source terms for each unit box, the COD and temperaturevalues are shown in Table 4. In Table 5, it is presented the values of volume flow rate, CODand temperature of each box.

Table 3Information on the water/fabric pick-up ratio of the washing process

Pick-up stream Water/fabric pick-up ratio

1 0.853 1.004 1.006 1.008 1.00

10 1.0011 0.65

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 9

Table 4Results of the unit box in the washing process

Box Source term COD (mg L−1) Temperature (◦C)

1 −0.77 405.12 58.02 3.94 1060.34 81.83 0.17 191.85 84.74 2.22 127.88 73.05 −0.22 1845.79 57.96 −0.06 1732.69 57.9

Analysing the values of the second column in Table 4, it is possible to verify that:

• The source term of the first box is negative due to not all the entry organic mass relatedto the fabric is available in the stream 1. The estimated entry COD value was obtained inthe laboratory, where a sample of the fabric coming from the dyeing bath was taken andsubmitted to several washings, resulting in a very high COD value due to the reductionof the organic mass that was adsorbed in the fibres;

• In the unit boxes 2–4, the source term is positive. It means the attrition of the box rollsand fabric promoted the organic mass transfer from the fabric to the liquid phase;

• In box 5, the source term is negative due to the adsorption of acetic acid (with high contentof organic mass), that is utilised in this step to neutralise the pH;

• In box 6, the source term is negative due to the stream 10 to bring along some acid frombox 5.

Table 6 shows the results of the water pick-up volume flow rate, industrial water volumeflow rate, chemicals volume flow rate, recycled water volume flow rate and the efficiencyof each unit box, expressed by WCOFj.

Table 5Volume flow rate, mass concentration (COD) and temperature of each stream

Stream Q (m3 h−1) COD (mg L−1) Temperature (◦C)

1 1.66 11305.88 20.02 9.00 0.00 65.03 1.96 405.12 58.04 1.96 1060.34 81.85 9.00 0.00 95.06 1.96 191.85 84.77 7.50 0.00 70.08 1.96 127.88 73.09 0.90 7714.00 25.0

10 1.96 1845.79 57.911 1.27 1732.69 57.912 0.68 1732.69 57.913 1.58 1845.79 57.914 7.50 127.88 73.015 16.50 191.85 84.716 16.50 1060.34 81.817 8.71 405.12 58.0

10 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

Table 6Information of unit box in the washing process

Box Water pick-upvolume flow rate(m3 h−1)

Industrial watervolume flow rate(m3 h−1)

Chemicalsvolume flowrate (m3 h−1)

Recycled watervolume flow rate(m3 h−1)

WCOFj

(g kg−1 offabric)

1 1.66 9.00 0.00 0.00 2.212 1.96 0.00 0.00 16.50 7.973 1.96 9.00 0.00 7.50 0.264 1.96 7.50 0.00 0.00 0.425 1.96 0.00 0.90 0.68 −0.946 1.96 0.00 0.00 0.00 −0.11

∑= 9.81 g kg−1

Fig. 2. Flowsheet of the proposed optimised washing process.

The flowsheet of the optimised process is shown in Fig. 2. In this flowsheet, one streamsplitter (unit 8) and one mixer (unit 7) were added to the process.

Comparing Figs. 1 and 2, besides the process shown in Fig. 2 has one stream splitter andone mixer, the stream 5 had its volume flow rate reduced from 9.00 to 6.5 m3 h−1, and thestream 7 had its volume flow rate reduced from 7.50 to 6.00 m3 h−1.

All other information of the entry streams as well as the fabric was considered constant.Table 7 shows the values of the stream splitter factor of each output stream referring to

the unit stream splitter (unit 8).The simulation results obtained for this new configuration, including to the source term

values, the COD and temperature values of each box, are shown in Table 8.

Table 7Parameters of each stream splitter of the proposed optimised washing process

Stream splitter Output stream Stream splitter factor

Unit 8 17 0.3Unit 8 18 0.7

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 11

Table 8Values for new source term, COD and temperature of each unit box of the proposed optimised washing process

Box New source term COD (mg L−1) Temperature (◦C)

1 −0.77 405.12 58.02 3.34 1735.78 80.53 0.09 440.38 92.44 2.26 353.85 74.65 −0.15 1173.63 75.16 −0.02 1146.68 75.1

Table 9Results of the streams in the proposed optimised washing process

Stream Q (m3 h−1) COD (mg L−1) Temperature (◦C)

1 1.66 11305.88 20.02 9.00 0.00 65.03 1.96 405.12 58.04 1.96 1735.78 80.55 6.50 0.00 95.06 1.96 440.38 92.47 6.00 0.00 70.08 1.96 353.85 74.69 0.90 7714.00 25.0

10 1.96 1173.63 75.111 1.27 1146.68 75.112 0.68 1146.68 75.113 4.65 1173.63 75.114 6.00 353.85 74.615 6.50 440.38 92.416 12.50 398.84 84.317 3.75 398.84 84.318 8.75 398.84 84.319 8.75 1735.78 80.520 8.71 405.12 58.0

Table 10Information on the unit box of the proposed optimised washing process

Box Water pick-upvolume flowrate (m3 h−1)

Industrialwater volumeflow rate(m3 h−1)

Chemicalsvolume flowrate (m3 h−1)

Recycled watervolume flow rate(m3 h−1)

WCOFj (g kg−1 offabric)

1 1.66 9.00 0.00 0.00 2.212 1.96 0.00 0.00 8.75 7.303 1.96 6.50 0.00 0.00 0.174 1.96 6.00 0.00 0.00 1.005 1.96 0.00 0.90 3.75 −0.706 1.96 0.00 0.00 0.00 −0.03

∑= 9.95 g kg−1

12 S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13

Table 11Volume of water used per mass of dry fabric and organic mass reduction per mass of dry fabric

Process Volume of water used permass of dry fabric (L kg−1)

WCOF (organic massreduction) (g kg−1 of fabric)

Washing 13.0 9.81Optimised process 10.9 9.95

Table 9 shows the values of the volume flow rate, the COD and temperature of eachstream for this new process configuration.

The efficiency of organic mass reduction per fabric mass (WCOFj) of each box j can beanalysed through the results presented in Tables 10 and 11.

Results shown in Table 11 prove the developed software is an important tool for theprocess optimisation and the proposed optimised process gives higher reduction of waterconsumption (16.15%) and of organic mass (1.43%) compared to the previous case.

4. Conclusions

In this paper, a tool was developed aiming to optimise the reuse of aqueous streams thatare sent to the effluent treatment plant (ETP), in a continuous washing process of a textileindustry, through their direct re-utilisation, or dilution with the water from the effluenttreatment plant or, even diluting the streams themselves.

The proposed simulation model for continuous washing processes showed to be efficientin the optimisation of process with recycle, being possible to simulate several possibilitiesof reuse for different process configurations.

The new configuration of streams has been analysed, given higher reduction of organicmass (1.43%) and higher reduction of water consumption (16.15%) compared to the previ-ous case. The developed software is an important tool for the industrial engineers.

It is important to point out that many contributions can be introduced to the software tomake this tool more efficient. It is necessary to input other parameters in the optimisationcriteria such as colour, turbidity and pH, as well as parameters of cost.

Acknowledgments

The authors would like to thank FINEP/CTHIDRO for funding the project AGUATEX,Racionalizacao do Uso de Agua nos Processos da Industria Textil Catarinense, in partnershipwith the industries Buettner, Coteminas, Karsten, Menegotti and Tapajos. The authors wouldalso like to thank SENAI – Blumenau for collecting data from the textile industries, andCAPES for the scholarship.

References

Almato M, Espuna A, Puigjaner L. Optimisation of water use in batch process industries. Comput Chem Eng1999;23:1427–37.

S.M. de Arruda Guelli Ulson de Souza et al. / Resources, Conservation and Recycling 49 (2006) 1–13 13

Bagajewicz MJ. A review of recent design procedures for water network in refineries and process plants. ComputChem Eng 2000;24:2093–113.

Chen CL, Hung PS. Simultaneous synthesis of mass exchange networks for waste minimization. Comput ChemEng 2005;29:1561–76.

Dhole VR, Ramchandani N, Tainsh RA, Wasilewski M. Make your process water pay for itself. Chem Eng1996;103:100–3.

Feitkenhauer H, Meyer U. Integration of biotechnological wastewater treatment units in textile finishing facto-ries: from end of the pipe solutions to combined production and wastewater treatment units. J Biotechnol2001;89:185–92.

Majozi T. Wastewater minimization using central reusable water storage in batch plants. Comput Chem Eng2005;29:1631–46.

MATLAB®, 2004. MathWorks products, US Patent 6,857,118. http://www.mathworks.com/.Puigjaner L, Espuna A, Almato M. A software tool for helping in decision-making about water management in

batch process industries. Waste Manage 2000;20:645–9.Smith R. State of the art in process integration. Appl Therm Eng 2000;20:1337–45.Tiernan JC. An efficient search algorithm to find the elementary circuits of a graph. Comm ACM 1970;13:722.