Water-wastewater management of tapioca starch manufacturing using optimization technique

8/7/2019 Water-wastewater management of tapioca starch manufacturing using optimization technique

1/11

ScienceAsia 26 (2000) : 57-67

INTRODUCTION

The growing interest in environmental impactaddresses the most serious challenges currentlyfacing the chemical process industry. The activityso-called pollution prevention is not only the end ofthe pipe treatment, but the pollution can be preventedby earlier stages. Consequently, the holistic approachto pollution prevention presents the viewpoint ofprocess integration which main pollution preventioncan be embedded through out the sequences of unitoperations. This concept clearly performs the costeffective approach to various process objectives.

Two-main strategies are used to reduce the wasteespecially wastewater in the complex chemicalprocess. They are1:

1. Source reduction includes any in-plant actionto reduce the quantity or the toxicity of the waste atthe source. Examples include equipmentmodification, design and operational changes of theprocess, reformulation or redesign of products,substitution of raw materials, and use of environ-mentally benign chemical reactions.

2. Recycle/reuse involves the use of pollutant-

laden streams within the process. Typically,separation technologies are key elements in a recycle/

ESEARCH ARTICLER

reuse system to recover valuable materials such assolvents, metals, inorganic species, and water.

In this work, the focus is only on recycle/reuseof water through the chemical process as anintegrated system. The mathematical formulationwill be demonstrated by using a case study of starchmanufacturing which in general uses tons of waterand has a systematical processing.

METHODOLOGYAND IMPLEMENTATION

There are a lot of unit operations that requirewater to remove contaminant from the process

streams. This water in turn becomes contaminated.To utilize water throughout the system, one simplemethod is water reuse. The discharged streams fromany units can be reused in other units directly or itcan be mixed with other discharged streams and thenreused later.

This task is one kind of mass exchanged networks(MENs) problem which concerns the transfer ofmass from source (stream that contains contaminants)to sink (unit that can accept contaminants). In orderto minimize the amount of freshwater usage

throughout the process system, one can maximizethe possibility of water reuse from one operation to

Water-Wastewater Management of Tapioca StarchManufacturing Using Optimization Technique

Thongchai Srinophakuna*, Uthaiporn Suriyapraphadilokb and Suvit Tiaba Department of Chemical Engineering, Kasetsart Unitversity, Bangkok 10900, Thailand.

b Chemical Engineering Practice School, Department of Chemical Engineering, King Mongkut'sUniversity of Technology, Bangkok 10140, Thailand.

* Corresponding author. E-mail: [email protected]

Received 26 May 1999

Accepted 28 Feb 2000

ABSTRACT Water reuse is one promising method to reduce wastewater from several sources in chemicalindustry: Utilities and Processes. Moreover, water reuse can also recover valuable product leaving withwaste streams. The objective of this work is to minimize the freshwater flowrate throughout the systemand find out the optimum water network. A mass integration for segregation, mixing, reusing, anddirect recycle is set up to solve the water-wastewater problem as a whole plant concept. Dischargedwater from each unit as well as the fresh feed are considered as sources of water, whereas units that

accept water are considered as sinks. In this model, it embeds all potential configurations by allowingeach source to be segregated, mixed, allocated to other units, and returned back to the process. This setof allocation equation is then combined with the process constraints and solved as an optimizationproblem to target minimum freshwater feed into the system and design the water network simultaneously.This optimization problem is formulated as a Non-Linear Programming (NLP) and solved by the high-level modeling language GAMS (General Algebraic Modeling System). Finally, a cases study of tapiocastarch is implemented. A 13.22% reduction of freshwater usage is obtained.

KEYWORDS: wastewater minimization, water network, NLP, mass integration.


2/11

58 ScienceAsia 26 (2000)

others. In mass exchanged problem, it cannotconsider only the value of level of contaminantseg COD, BOD, TS, etc, therefore, with some streamscannot mix together. In order to embed all processconstraints, it is of interest to formulate this massintegration problem in a systematic design which

can target the minimum water flowrate and designnetwork configuration simultaneously.

El-Halwagi and Garrison2 introduced theproblem of utilization of the discharged water fromprocesses for segregation, mixing, reusing, and directrecycle. Since the aim of mass integration is at theoptimal allocation of species or streams throughoutthe process. Instead of dealing with the detailedflowsheet of the process, a global mass allocationrepresentation is presented. For each species therearesources, stream that carry that species, and sinks,

units that can accept the species. Streams leavingthe sinks become sources. Hence, sinks can also begenerators of the target species.

Also sources can be mixed together and reuse.To determine the opportunities of reuse/recycle ofeach stream, a plot of flowrate or contaminant loadversus its concentration is generated as shown inFig 1; namely a source-sink mapping diagram

In Fig 1, the shaded circles represent sources, andthe hollow circles represent sinks. Generally, therange of contaminant concentration and loading of

each sink are limited by the process constraints. Theintersection of the acceptable zone of sink and sourceprovides the acceptable zone for reuse/recycle. Anysource which lies within this zone can be directly

reuse/recycle to that sink as can be seen from sourcea to sink R in Fig 1. Referring that streams mixingis allowable to feed prior to other source, source band c can be mixed and reuse in sink R.

From the framework described in Fig 2, itcontains all potential configurations by allowing

each source to be segregated, mixed, allocated toother units, and returned back to the process.Typically, the process constraints limit the range ofcon-taminant concentration and load that each sinkcan accept. The task of optimization is to determinethe flowrate and composition of each stream whichis satisfied all those process constraints.

The concept of mass integration for segregation,mixing, reuse, and direct recycle was demonstratedby using the starch manufacturing.

TAPIOCA STARCH MANUFACTURING

Tapioca or manioc starch is obtained fromthe tuberous roots of the manioc or cassava plant,which is one of the main products in Thailand.Basically manufacturing can be divided into fourstages3:

1. Washing and peeling of the roots, raspingthem and straining the pulp with addition of water.

2. Rapid removal of the fruit water and itssoluble and replacing thus with pure water to prevent

deterioration of the pulp. This stage includessedimentation, washing of the starch in tanks, or onsettling tables, silting, and in some of the mediumand larger factories centrifuging.

Fig 2. Mass integration framework for segregation, mixing, reuse,and direct recycle.Fig 1. A source-sink mapping diagram.

Flowrateorcontaminan

tload

Concentration of contaminant

b

c

Ra

Sink

SourceSink 1

Sink 2

Sink R

Sources SourcesSinksSegregated

Sources


3/11

ScienceAsia 26 (2000) 59

3. The removal of water by draining, centri-fuging and drying.

4. Grinding, bolting and other finishingoperations.

In the tapioca starch manufacturing, processesthat consume water are washing, rasping, grinding,

screening, and separating. For washing, rasping, andgrinding process, water is used to wash the adheringdirt and added in peeling, rasping, and grindingprocess. Therefore, it does not need to use freshwaterin these processes. After that the slurry is sent toscreening process in order to separate the fiber andother insoluble particles such as protein, fat, etcThen the soluble particles are removed with waterin the separating process.

In screening and separating processes, althoughthe COD value is high in the discharged water, one

important component of those chemicals is starch.In order to recover this amount of starch, the valuabledischarged water can be reused. The criteria forselecting the streams to be reused are the COD value,amount of starch left and other physical propertiessuch as the amount of protein. Next, anotherconsideration of reusing water is which unit canaccept this water. For the starch manufacturing,water reuse is not only the reduction of freshwaterused but it can also recover the valuable starchproduct. Two major machines involved are screen

and separator. In screen machine, large solidparticles such as fiber, fat, etc are separated fromsmall particles by centrifugal force. Hence, somestarch can be lost with fiber because of the size andbranches of fiber. Whereas in separating machinestarch is separated from other lower density particlessuch as protein by centrifugal force. Because of the

limited technical supports to represent how muchwater and what amount of contaminants (includingstarch) that can affect the efficiency of the operations,this work is based on the original process flowsheetof each plant.

MATHEMATICAL MODEL

In order to set up a starch model, mass balancesmust be made by using the following assumptions:

1. There are two main components in themanioc roots: starch and water. Other componentswill be treated as one component named others.

2. Amount or value of COD in mg/l is equalto the summation of amount of starch and othersin mg/l.

3. The compositions of the starch loss in drying

unit are the same as the dry starch product.For the first assumption, it is sufficient to analyzeonly these components: starch, water, and others.For the second assumption, since COD is calculatedfrom the amount of oxidizing agent used tocompletely oxidize the chemical substances in water.For the third assumption, after the wet starchis dried in a pneumatic dryer, it is sent to cyclone.Some starch loses in the cyclone. The compositionof the lost starch is exactly the same as the dry starchproduct. Therefore, this assumption is reasonable.

To set up a model, the parameters and variablescan be summarized as shown in Fig 3.Additional assumptions are made in order to

simplify the model. They are:1. Only single contaminant is considered.2. No equilibrium relation governs the dis-

tribution of contaminant in water.

Fig 3. Flow diagram of the starch line and water line of each unit.

XOiYOi

FBYii'XOj'YOj'

XOi

YOiFROUTi

UNIT iXOiYOi

FROUTi

FOUTi

XIiYIi

FRINi

XSTOiYSTOi

STLNOUTi

XSTIiYSTI

i

STLNINi

XSj

YSj

FSTORij

FBYi'iXOiYOi


4/11


3. The efficiency of starch removal and CODremoval of each unit are constant although the inletconcentration of any streams changed.

4. The flowrates of both starch line and water-line, are the same as the original flowsheet.

Basically, the indicator used to describe the

property of wastewater is the value of COD. It actslike a single contaminant in waste stream, althoughthere are many kinds of contaminants. Hence, thefirst assumption is reasonable. Since there is no limita-tion for starch to disperse in water, in other word,there is no equilibrium relationship between them.Therefore the second assumption is also applied.

For the third assumption, since the mass balancescan be made, and the composition of each streamfor normal operating condition is known, theefficiency of starch removal and COD removal of

each unit can be calculated by

STEFFSTSTOUT STLNOUT

STSTIN STLNIN STRIN FRINi

i i

i i i i

=( )( )[ ]

( )( ) + ( )( )[ ](1)

CODEFF

CROUT FROUT

CSTIN STLNIN CRIN FRINi

i i

i i i i

=( )( )[ ]

( )( ) + ( )( )[ ](2)

The efficiency of each unit is held constantwhether the compositions of the inlet streams havechanged except in washing and drying unit. Inwashing unit, the composition of clean roots is heldconstant as in the original flowsheet. In drying unit,the composition of the starch loss and the dry starchproduct is the same.

For the last assumption, since there are notechnical supports to represent how much water andwhat amount of contaminants (including starch) thatcan affect the efficiency of the machines. That is allthe flowrates, both starch line and waterline, are the

same as the original flowsheet.

NETWORK OPTIMIZATION

The utilization of the discharged water fromprocesses for segregation, mixing, reusing, and directrecycle was formulated as a NLP problem. Thesolution of this NLP is the optimal allocation ofspecies or streams throughout the process withminimum freshwater flowrate target. Prior topresenting the mass integration and its mathematical

formulation, the following synthesis model arepresented:

Overall balance around splitter of each source R

FROUT FBY FOUT i Ri i i

i Ri =

0 (3)

Overall balance around mixer of the inlet of eachsink R

FRIN FBY FSTOR i Ri iii R

ijj S

=

0 (4)

COD balance around mixer of each source R

FRIN YI FBY YO FSTOR YS i Ri i ii ii R

ij jj S

( )( ) ( )( ) ( )( ) =

0 (5)

Starch balance around mixer of each source R

FRIN XI FBY XO FSTOR XS i Ri i ii ii R

ij jj S

( )( ) ( )( ) ( )( ) =

0 (6)

COD balance around each sink R

FRIN YI STLNIN YSTI FROUT YO

STLNOUT YSTO i R

i i i i i i

i i

( )( )+ ( )( ) ( )( )

( )( ) = 0(7)

Starch balance around each sink R

FRIN XI STLNIN XSTI FROUT XO

STLNOUT XSTO i R

i i i i i i

i i

( )( ) + ( )( ) ( )( )

( )( ) = 0(8)

Total freshwater feed

FFW FSTORiji R j S

= ,

(9)

Efficiency limit (COD)

YO CODEFF YI FRIN YSTI STLNIN

CROUT

i WASH DRY

i i

i i i i

i

= ( )( ) + ( )( )[ ]( )[ ]

*

' ',' '

(10)

Efficiency limit (starch)

XSTO STEFFXI FRIN XSTI STLNIN

STLNOUT

i WASH DRY

i i

i i i i

i

=( )( )+ ( )( )[ ]

( )[ ]

*

' ',' '

(11)

Then the optimization problem can be set up byminimizing

Z FFW = (12)subject to the Equations 3 - 11 and all other

constraints that come from the process flowsheet.

CASE STUDY

According to Fig 4. The original process, thereare four screens and three separators in this plant.The first three screens are used in the main process,while the remaining is used as a final screen to

recover starch from pulp coming from those threescreens. From Fig 4, the pulp from 'Screen 1', 'Screen


5/11


Fig 4. Flowsheet for the manufacturing of tapioca starch Plant before improving (S = % of starch, COD = COD value in mg/l).

Wash & Rasp

Fresh roots31.90 tons/hr

(S=20.1200%)

Pulp112.90 tons/hr

(S=15.7158%)

Water 71.82 tons/hr

(S=0.0622%, COD=28,829)

Clean roots27.52 tons/hr

(S=22.6628%)

Water 85.38 tons/hr

(S=13.4766%, COD=196,120)

Fiber 21.20 tons/hr

(S=14.8700%)

(S=0.1776%,

COD=40,167)

Wastewater

75.01 tons/hrPeels

Freshwater

8.31 tons/hr

Freshwater

41.52 tons/hrScreen 2

Starch milk100.01 tons/hr

(S=14.5892%)

Total fiber

93.71 tons/hrFinal Screen

Fiber45.81 tons/hr

(S=8.0200%)

Water

37.48 tons/hr

Fresh fiber8.33 tons/hr

(S=11.1800%)

Screen 3Freshwater

21.45 tons/hr

Freshwater

78.92 tons/hr

Screen 1

Grind

Fiber 27.71 tons/hr(S=17.5700%)


(S=8.5416%)


(S=7.2205%)

Separator 1Water 158.25 tons/hr

(S=0.0622%, COD=28,829 )

Fiber 3.45 tons/hr

(S=5.9107%, COD=75,205 )

Starch milk

52.49 tons/hr(S=17.9458%)


(S=32.4587%)

Wastewater 22.99 tons/hr

(S=0.0724%, COD=5,719 )


(S=34.4724%)

Freshwater

1.65 tons/hr

Freshwater

22.00 tons/hr

Water 3.33 tons/hrDewater

(S=10.1877%, COD=205,908 )

Fiber 41.35 tons/hr

Humid starch11.80 tons/hr

(S=43.9871%)

Evaporated water

4.06 tons/hrDrying

Dry starch7.43 tons/hr

(S=67.0900%)

(S=67.0900%)

Water loss 0.31 tons/hr

Separator 2

Wastewater

86.43 tons/hr

(S=0.7570%, COD=12,102)

Screen 4

Press

1.19 tons/hr

(S=7.8100%)


6/11


2', "Screen 3', and 'Screen 4' contain high value ofstarch, they are sent to the 'Final Screen' to recoverthe starch. After the pulp has been separated in the'Final Screen', filtrate, which is rich of starch, isreturned to the grinding process, while the fiber issent to 'Press' to recover water, and the 'Fresh Pulp'

is sold as a cattle food.In separating process, discharged water from the

'Separator 1' is quite high in COD, but the mainindicator of this stream is the value of Total KjeldahlNitrogen or TKN. This TKN indicates that theamount of protein of this stream is quite high. Andalso most of cyanide, which is the undesiredcomponent of the starch product, is separated in thisstage. Therefore, this discharged water is not suitableto reuse in any process except for the washing andrasping. In the other separating machines, the

property of discharged water is suitable to be reused.For starch plant A, the following sets are defined:

Ri|'Wash','Grind','Screen1','Screen2','Screen 3','Sep1',

'Screen4','Sep2','Dewater','Dry','FinalScreen','Press'=

S j 'Fresh'= { }|In order to set up an optimization program, all

the process constraints have to be embedded asfollows:

1. Any stream cannot recycle back to its process

in order to prevent the building up of con-taminants.2. Discharged water from 'Screen 1', 'Screen 2','Screen 3' and 'Screen 4' are sent to recover starch inthe 'Final Screen'.

3. Discharged water from 'Final Screen', whichis rich of starch, is returned to 'Grinding unit' only.

4. The discharged water from 'Separator 1'cannot be reused in any unit except in wash-ingprocess since this water contains high protein.

5. Water from 'Dewater' is returned to 'Screen3' only (the same as in the original flowsheet).

6. The discharged water from 'Press' can onlybe returned to 'Grinding unit' (the same as in theoriginal flowsheet).

7. Freshwater will not add to 'Grinding unit','Dewater', 'Dry', 'Final Screen', and 'Press'. Therefore,the upper bound for the inlet COD of these unitscan be set quite high in order to receive the reusewater from other units.

8. Discharged water from 'Washing unit' cannotbe reused in any other units.

9. In the starch plant, the amount of con-taminants is decreased from upstream to downstream

units. Hence, water can be reused from downstreamunit to the upstream unit. Finally, the last unit

receives only freshwater. The following streamcompositions (COD) are bounded:

0 0 00 ( ) XI 'SEP1' .

0 0 00 ( ) XI 'SEP2' .

0 0 01 ( ) XI 'Screen1' .

0 0 01 ( ) XI 'Screen2' .0 0 00 ( ) XI 'Screen3' .

0 0 013 ( ) XI 'Screen4' .

0 ( ) XI 'FinalScreen'

0 ( ) XI 'Press'

0 ( ) XI 'Wash'

0 ( ) XI 'Grind'where the unit of COD is in milligram per liter.

These process constraints which are in GAMS

input form are then added to the allocation equations(Equations 3 to 11) and are considered as Non-LinearProgramming constraints for the freshwater flowrateminimization. Fig 5 shows the solution of thisproblem.

RESULTSAND DISCUSSIONS

According to Fig 4, there are three units thatdischarge water to the biological treatment system:'Separator 1', 'Separator 2', and 'Washing unit'. Thedischarged water from 'Separator 1' and 'Washingunit' cannot be reused, since they contain high valuesof COD and TKN. Also the discharged water from'Press' is not suitable to be reused in any units exceptthe 'Grinding unit' because of its high value of totalsolid. Therefore, only one discharged stream from'Separator 2' can be segregated. This action canreduce amount of freshwater usage by 22.99 tons/hrand reduce amount of wastewater generation by22.99 tons/hr as well.

Since the Non-Linear optimization problemcannot give a unique solution, although the Non-Linear Programming is implemented. There arethree solvers in GAMS that can solve a NLPoptimization: CONOPT2, CONOPT, and MINOS5.The weak point of NLP problem is that the solutiondepends upon the starting points and the path ofsolving. In GAMS, the paths of solving dependsupon the solvers even in the same starting point.The optimal solution cannot be guaranteed, and if asolution is reached, it can only be considered as alocal optimum.

There are three alternatives presented here as

shown in Figs 5, 6 and 7, which are solved byCONOPT2, CONOPT, and MINOS5, respectively.


7/11


B

Fresh roots31.90 tons/hr

(S=20.1200%)

Water 71.82 tons/hr

(S=0.0623%, COD=29,112)

Clean roots

27.52 tons/hr

(S=22.6628%)

Water 85.38 tons/hr

(S=13.5089%, COD=197,867)

Pulp112.90 tons/hr

(S=15.7402%)

Fiber 21.20 tons/hr

(S=14.8933%)

Wash & Rasp

Grind

Screen 1Freshwater

8.31 tons/hr

Peels

1.19 tons/hr(S=7.8100%)

Waste water

75.01 tons/hr

(S=0.1777%,

COD=40,0167)

Freshwater

20.18 tons/hr

Freshwater

21.45 tons/hr

Final Screen

Press


(S=14.6118%)

Fiber

45.81 tons/hr

(S=8.0392%)

Water

Fresh fiber

8.33 tons/hr

(S=11.2068%)

Fiber 27.71 tons/hr(S=17.6158%)

Fiber 3.45 tons/hr

(S=5.9281%, COD=75,943 )


(S=7.2394%)

Starch milk

113.82 tons/hr

(S=8.5639%)

Freshwater

78.92 tons/hr

Starch milk

52.49 tons/hr(S=17.9927%)

Screen 3

Screen 2


(S=32.5477%)


(S=34.5669%)

Humid starch

11.80 tons/hr

(S=44.1077%)

Dry starch

7.43 tons/hr

(S=67.2739%)

(S=67.2739%)

Water loss 0.31 tons/hrEvaporated water

4.06 tons/hr

(S=0.7587%, COD=12,228 )

Water 3.33 tons/hr

22.00 tons/hr

Freshwater


(S=0.0623%, COD=29,112 )

Fiber 41.35 tons/hr

(S=10.2156%, COD=208,057 )


(S=0.0727%, COD=5,779 )

Dewater

Water 1.65 tons/hr

Water 21.34 tons/hr

Wastewater86.43 tons/hr

B

A

A

Screen 4

Separator 2

Drying

37.48 tons/hr

Total fiber

93.71 tons/hr

Fig 5. Flowsheet for the manufacturing of tapioca starch Plant alternative 1 after improving (S = % of starch, COD = COD value inmg/l).


8/11


A

Fresh roots

31.90 tons/hr

(S=20.1200%)

Water 71.82 tons/hr

Clean roots

27.52 tons/hr

(S=22.6628%)

Water 85.38 tons/hr

Pulp

112.90 tons/hr

(S=15.7411%)

(S=0.1776%,

COD=40,0167)

Wastewater

75.01 tons/hrPeels

1.19 tons/hr(S=7.8100%)

Freshwater

20.18 tons/hr

Freshwater

21.45 tons/hr

Freshwater

78.92 tons/hr

Freshwater

22.00 tons/hr

Water 3.33 tons/hr

(S=0.7587%, COD=12,228 )

Evaporated water

4.06 tons/hr

Dry starch

7.43 tons/hr

(S=67.2728%)

(S=67.2728%)

Water loss 0.31 tons/hr

Grind

Wash & Rasp

Fiber 21.20 tons/hr

(S=14.8992%)

Starch milk

100.01 tons/hr

(S=14.6176%)

Total fiber93.71 tons/hr

Final Screen

Fiber

45.81 tons/hr

(S=8.0399%)

PressWater

37.48 tons/hr

Fresh fiber

8.33 tons/hr

(S=11.2078%)

Fiber 27.71 tons/hr

(S=17.6155%)

Starch milk

113.82 tons/hr

(S=8.5638%)

Fiber 3.45 tons/hr

Screen 1

Screen 2

Screen 3(S=5.9280%, COD=75,938 )

Starch milk

131.82 tons/hr

(S=7.2393%)


(S=0.0623%, COD=29,110 )

Starch milk

52.49 tons/hr

(S=17.9924%)

Screen 4Fiber 41.35 tons/hr

(S=10.2155%, COD=208,043)

Wastewater

86.43 tons/hr

Water 8.31 tons/hr

Water 13.03 tons/hr

Water 1.65 tons/hr

Starch milk

16.12 tons/hr

(S=32.5471%)


(S=0.0727%, COD=5,778)Separator 2

Starch milk

15.13 tons/hr

(S=34.5663%)

Dewater

Humid starch

11.80 tons/hr

(S=44.1069%)

Drying

A

B

C

B

C

(S=13.5101%, COD=197,946)

(S=0.0623%, COD=29,106)

Fig 6. Flowsheet for the manufacturing of tapioca starch Plant alternative 2 after improving (S = % of starch, COD = COD value in

mg/l).


9/11


10/11


All of them give the same water reduction with threeapproaches. At this stage, the economical trading offmust be taken to evaluate between the freshwatersaving and the investment cost for piping andpumping in order to give the best alternative. Sincethis work is based on the original flowsheet,

especially, the flowrate of stream input and outputto any unit. It must be worth to find out the efficientrange of water flowrate and its concentration inletin order to utilize the water usage throughout thesystem. For example, the inlet flowrate of water to'Separator 1', 78.92 tons/hr, is quite high comparingto other separators, and the property of thedischarged water from this 'Separator 1' is notsuitable to be reused in any other units except inthe washing process. It is obvious that when theefficient range of water flowrate and its concentration

are known, one can reduce the amount of waterusage more effectively.Hence tapioca starch is one kind of food

manufacturing, the reuse of wastewater should beconsidered in terms of the quality of starch. Whenstarch is reused in the system, it can be degradeddepending on its age. It must be further studied tofind out how many hours it can be recycled in thesystem.

The main discussion shows the application ofmass exchanger network to retrofit the existing

process and hits the concept of cleaner technology.The technique forms a non-linear problem whichusually has many local optimal solutions. Basicallythe supplement information such as additionalinvestment cost needs to be investigated in order topolish for the best operational practice.

CONCLUSION

A mass integration for segregation, mixing,reusing, and direct recycle is generated to solve thewater-wastewater problem as a whole plant concept.Discharged water from each unit as well as the freshfeed are considered as of water, whereas units thataccept water are considered as sinks. In this model,it embeds all potential configurations by allowingeach source to be segregated, mixed, allocated toother units, and returned back to the process. Thisset of allocated equation is then combined with theprocess constraints and solved as an optimizationproblem to target the minimum freshwater feed tothe system and to design the water network simul-taneously. This optimization problem is formulated

as a Non-Linear Programming by using the high-level modeling language GAMS. Finally, a case study

of the tapioca starch plant are implemented.Freshwater usage and wastewater generation can bereduced to 13.22%.

In this work, only single contaminant is addressed.The process constraints can be taken into accountsuch as processes with water loss or gain, fixed water

flowrate constraints, etc. In principal, the solutionof non-linear optimization is not unique even withNon-Linear Programming implementation. Therefore,the expected result from this water-wastewaterminimization approach is also under this situation.There are more than one water network configura-tions that satisfy the same objective value. Hence, itmust be traded off between those alternatives forboth variable cost and fixed cost in order to achievethe best solution.

NOMENCLATURE

BOD Biological Oxygen DemandCOD Chemical Oxygen DemandTKN Total Kjeldahl NitrogenNLP Non-Linear Programming

1. Indices:i, i' unitsj fresh source ie freshwater

2. SetsR, RP unitsS freshwater

3. ParametersSTLNINi inlet flow of starch line of each unit i

in tons/hrSTLNOUTi outlet flow of starch line of each unit

i in tons/hrFRINi inlet flow of water line of each unit i

in tons/hrFROUT

i

outlet flow of water line of each uniti in tons/hr

STSTINi fraction of starch in starch line inletof each unit i in milligram per liter

STSTOUTi fraction of starch in starch line outletof each unit i in milligram per liter

STRINi fraction of starch in water line inletof each unit i in milligram per liter

STROUTi fraction of starch in water line inletof each unit i in milligram per liter

CSTINi COD inlet of starch line of each uniti in milligram per liter

CSTOUTi COD outlet of starch line of each uniti in milligram per liter


11/11


CRINi COD inlet of water line of each uniti in milligram per liter

CROUTi COD outlet of water line of each uniti in milligram per liter

STEFFi efficiency of each i to remove starchCODEFFi efficiency of each i to remove COD

XSj starch composition of fresh sources SYSj COD composition of fresh sources SIRii binary value used to indicate the

potential of water stream from unit ito be reused in unit i

UPCODi upper limit of the inlet COD of thewater line of each unit I

4. VariablesZ objective variableFFW total freshwater flowrate in tons/hr

FBYii flowrate from one unit to others i intons/hrFSTORij flowrate of stream ofj to i in tons/hrFOUTi flowrate of outlet stream of unit i to

treatment process in tons/hrXSTIi fraction of starch in starch line inlet

of each unit i in milligram per literXSTOi fraction of starch in starch line outlet

of each unit i in milligram per literXIi fraction of starch in water line inlet

of each unit i in milligram per liter

XOi fraction of starch in water line inletof each unit i in milligram per literYSTIi COD inlet of starch line of each unit

i in milligram per literYSTOi COD outlet of starch line of each unit

i in milligram per literYIi COD inlet of water line of each unit

i in milligram per literYOi COD outlet of water line of each unit

i in milligram per liter

REFERENCES

1. El-Halwagi MM (1997) Pollution Prevention through ProcessIntegration: Systematic Design Tools, 1st ed, CA, AcademicPress.

2. El-Halwagi M M and Garrison G W (1996) Synthesis of WasteInterception and Allocation Networks.AICHE J, 42:11, 3087-

99.3. Radley JA (1976) Starch Production Technology pp 1-43, pp

189-99, London, Applied Science Publishers Ltd.

Documents

Water-wastewater management of tapioca starch manufacturing using optimization technique