Linlin  Yang      Advisor:  Ignacio  E.  Grossmann  

 Department  of  Chemical  Engineering  

Carnegie  Mellon  University    

March  11,  2012    

“Water  is  the  fastest  growing  market  at  the  moment,  with  a  size  of  $500  billion  globally.”  “If  nothing  is  done,  there  will  be  a  40  percent  gap  between  supply  and  demand  by  2030.”    

2  

Water  and  energy  are  important  resources  in  the  process  industries    

3  Ahmetovic  &  Grossmann  

Conven>onal  water  network  

Freshwater  

Boiler  Feedwater  treatment  

Steam  System  

Wastewater  Boiler  Blowdown  

Steam  

Condensate    Losses  

Boiler  

Water-­‐using  unit  1  

Water-­‐using  unit  2  

Water-­‐using  unit  3  

Raw  Water  Raw  Water    Treatment  

Wastewater  

Cooling    Tower  

Cooling  Tower  Blowdown  

Water  Loss  by  Evapora2on  

Discharge  

Wastewater    Treatment  

Storm  Water  

Process  uses  

Other  Uses  (Housekeeping)  

Wastewater  

No  regenera>on  

reuse  

No  reuse  

Process  Unit  

Contaminants  

Process  Unit  

Contaminants  

»  Integrated  water  network  with  reuse,  recycle,  and  regeneraLon  schemes  »  superstructure  is  formulated  using  a  nonconvex  NLP  model    

4  Karuppiah  &  Grossmann  (2006);  Ahmetovic  &  Grossmann  (2010)  

Superstructure  based  water  network  design  

Freshwater   Discharge  

Treatment  Unit  

Contaminants  

       SpliTer  

Mixer  

Treatment  Unit  

Contaminants  

5  

Freshwater  targe>ng  formula>on  

outinkj

kpj

ij

iin

pout

iout

pin

kinout

ij

kj

insi

ik

outmi

fwifw

ij

ikj

k

outmi

ik

fw

pkpiPUpjCFLCFpiPUpPFpkPUpPF

sksiSUsjCC

skSUsFF

mkMUmjCFCFCF

mkMUmFF

FZ

out

in

in

∈∈∈∀∀=+∈∈∀=

∈∈∀=

∈∈∀∈∀∀=

∈∈∀=

∈∈∀∀+≥

∈∈∀=

=

∑

∑

∑

∈

∈

∈

,,,,,,,

,

,,)(

,s.t.

min

,

max,max,

SpliUers  mass  balances  

Process    unit  mass  balances  

Mixer  mass  balances  

This  formula>on  provides  target  for  a  network  consists  of  a  set  of  water-­‐using  process  units  using  linear  constraints  

(LP)  

AssumpLon:  for  some  contaminant  j  that  reaches  its  concentraLon  upper  bound  at  a  given  unit,  it  also  reaches  the  upper  bound  at  all  other  process  units  from  which  reuse  streams  have  non-­‐zero  flowrate  

Goal:  determine  minimum  freshwater  consump>on    

Use  heat  and  water  network  formulaLon  (MINLP  model)    to  obtain  network  structure  

6  Bogataj  &  Bagajewicz  (2007)  

Heat-­‐integrated  WN  reported  in  the  literature  

749  conLnuous  variables  115  binary  variables  

7  

Extension:  heat-­‐integrated  water  network  

SUssCjspTtTSUssHispTT

ttcf

TTCFQQc

TT

TTCF

TTt

TTtcfQ

outminjs

pout

inis

p

injs

outjsjsjssCjs

outis

inisisissHisH

poutis

pinisisissHis

mpin

js

mpout

jsjsjssCjsH

out

out

out

out

∈∀∪∈=∀Δ+=

∈∀∪∈=∀=

−Σ−

−Σ+=

−−

−Σ−

Δ−−−

Δ−−Σ≥

∪∈

∪∈

∪∈

∪∈

)(

)(

)(

}],0max{

)},0[max{

}](,0max{

)}(,0[max{

PU1,  T1  

PU2,  T2  

Freshwater  Tfw  

Discharge  Tdis  

All  black  streams  can  par>cipate  in  heat  integra>on  

outin

kj

kpj

ij

iin

pout

iout

pin

kinout

ij

kj

insi

ikout

mifw

ifw

ij

ikj

k

outmi

ik

pkpiPUpjCFLCF

piPUpPFpkPUpPF

sksiSUsjCC

skSUsFFmkMUmj

CFCFCF

mkMUmFF

out

in

in

∈∈∈∀∀=+

∈∈∀=

∈∈∀=

∈∈∀∈∀∀=

∈∈∀=

∈∈∀∀

+≥

∈∈∀=

∑

∑

∑

∈

∈

∈

,,,

,,,,

,

,,

)(

,

,

max,max,

fwfwCCHH FcQcQc ++=φ.minObjec>ve  fcn    

Water  targe>ng  (LP)   Heat  targe>ng  (LP)  

8  

Revisit:  heat-­‐integrated  water  network  u>lity  targe>ng  

Freshwater  Tfw  =  20°C      

Discharge  Tdis  =  30°C      

PU1          T1  =  40°C  

PU3  T3  =  75°C  

PU2          T2  =  100°C  

PU4          T4  =  50°C  

Parameter  

CHU  ($/kW  a)   260   THUin  (°C)   126  

CCU  ($/kW  a)   150   THUout  (°C)   126  

CFW  ($/t)   2.5   TCUin  (°C)   15  

HRAT  (°C)   10   TCUout  (°C)   20  

Use  heat  and  water  targe>ng  formula>on:  

Minimum  hea>ng  u>lity:  3767  kW  Minimum  cooling  u>lity  :  No  cooling  u>lity  required  Minimum  freshwater  consump>on:  324  ton/h  

Same  result  as  network  approach  

206  conLnuous  variables  229  constraints  

9  

Simultaneous  op>miza>on  strategy  

HEAT  TARGETING  

PROCESS  FLOWSHEET  

WATER  TARGETING  

Cold  streams  

 

Hot  streams  

 

MUC*    

*MUC  –  minimum  uLlity  consumpLon    

Hot  uLlity    

Cold  uLlity    

MUC*    

Water  streams  

 

Wastewater  

Freshwater  

ULlity  networks  

 

PROCESS  STRUCTURE  

WN  STRUCTURE   HEN  STRUCTURE  

PROCESS  FLOWSHEET  WITH  HEN  AND  WN  

VvUuXxFvg

QQugvuxgvuxh

FcQcQcvuxF

fwWN

CHHEN

P

fwfwCUj

jC

jC

HUi

iH

iH

∈∈∈

≤

≤

≤

=

+++= ∑∑∈∈

,,

0),(

0),,( 0),,(

0),,( s.t.

),,(.min φ

10  

Simultaneous  op>miza>on:  methanol  synthesis  from  syngas    

+  cooling  cycle  +  boiler  loop    

Freshwater  requirement  

Hea>ng  requirement  

Cooling  requirement  

Duran  &  Grossmann  (1987)  

SEQUENTIAL   SIMULTANEOUS  Profit                                            (1000  $/yr)   62,695   73,416  Investment  cost      (1000  $)   1,891   1,174  OperaLng  parameters      

electricity  (KW)   6.59   1.84  freshwater  (kg/s)   36.43   29.25  

heaLng  uLlity  (109  KJ/yr)   0.293   0  cooling  uLlity  (109  KJ/yr)   67.3   72.7  

Steam  generated  (109  kJ/yr)   2448   1965  overall  conversion   0.68   0.88  

Material  flowrate  (106  kmol/yr)      feedstock   48.04   37.13  product     10.89   10.89  

11  

Sequen>al  vs.  simultaneous  result  comparison  

17%  improvement  Solved  with  BARON  9    

12  R.  Karuppiah,  A.  Peschel,  I.  E.  Grossmann,  M.  Marln,  W.  son,  and  L.  Zullo,  “Energy  opLmizaLon  for  the  design  of  corn-­‐based  ethanol  plants,”  AIChE  Journal,  vol.  54,  no.  6,  2008,  pp.  1499–1525.  

Example  2:  Bioethanol  produc>on  

Water  Consump>on  and  genera>on  

Fermenta>on  

Solid  Removal  

Water/EtOH  Separa>on  

Pretreatment  

13  

Water  network  superstructure  

Cjin,max  (ppm)   TSS   TDS   ORG  

Boiler  loop   2   100   10  

Cooling  cycle   10   500   10  

1-­‐Βjt  

Screens   95%   0   0  

Reverse  osmosis   0   90%   0  

Anaerobic  tank   0   0   99%  

PU  1  

Cooling  requirement  (Qc)  

Cooling  Tower  

Freshwater  

blowdown  

boiler  

HeaLng  requirement  (Qs)  

Steam  loss  Steam  

blowdown  

Coldwater  outlet  

Warmwater  inlet  

Condensate  

Feedwater  

PU  3  

Screens  

ANA  

Reverse  osmosis  

PU  2  

Formula>on  ˃  Dew  point  equaLon    -­‐  condenser  

temperature        ˃  Bubble  point  equaLon  -­‐  feed  and  reboiler  

temperature  ˃  Fenske  equaLon  -­‐  #  of  trays  ˃  Watson's  equaLon  –  heat  of  vaporizaLon  ˃  Mass  balance  ˃  Energy  balance  

   Assump>ons  ˃  Constant  relaLve  volaLlity  ˃  Ideal  soluLon  ˃  Water  is  the  only  component  contribuLng  

to  heat  of  vaporizaLon    ˃  Temperature  change  due  to  pumps  is  

negligible   14  

Mul>effect  columns  

Feed

Distillate

Bottom

1 atm Feed

LP column

HP column

Distillate

Bottom

Result  

NLP  solver:  CONOPT  3  MINLP  solver:  BARON  9  GAMS  23.7  

15  

Reboiler  duty  reduced  by  ~36%  by  with  mul>effect  column  

Even  though  the  objecLve  funcLon  did  not  improve  using  simultaneous  method,  we  can  see  that  the  soluLon  Lme  did  not  increase  drasLcally    

No  integra>on  

Sequen>al  single  column  

Sequen>al  w/  mul>effect  

Simultaneous  w/  Mul>effect  

Cost  (MM$/yr)   14.91   11.77   8.57   8.57  

Cooling  water  use  (kg/s)   2895.6   1998.3   1127.3   1124.8  

Freshwater  use  (kg/s)   40.8   127.6   90.0   90.0  

Steam  use  (kg/s)   35.1   28.3   21.2   21.3  

CPU(s)   387   387   470   563  

#  eqns   2,232   2,232   3,213   5,221  

#  cont  var   2,921   2,921   3,914   5,392  

Cooling    water  

U>lity  integra>on  –  power,  water,  &  heat    

blowdown  

Makeup    water  

H1-C1

H1-C2

H2-C1

H2-C2

H1-C1

H1-C2

H2-C1

H2-C2

Stage 1 Stage 2

Temperature location k=1

Temperature location k=2

Temperature location k=3

s1

s2

W1

W2

C1

C2

H1

H2

tC1,1 tC1,2 tC1,3

tC2,1 tC2,2 tC2,3

tH1,1 tH1,2 tH1,3

tH2,1 tH2,2 tH2,3

CoolersHeatersblowdown  

blowdown  

Condensate  return  

HP          MP          LP  

Warm  water    return  

Raw  Water  

SiO2  Removal  

discharge  

Condensate  loss  

Cooling  Tower  

Process  Load  (HEN)                          

Sand  Filter  

Reverse  Osmosis  

Water  Treatment                    

Scrubber  

ULlity  System  

16  

fwfws

ssst d

stdtur

d

extd

fixext

st d

std

fixtur

st

stbb

st

stb

fixb FcFcWcYcYcFcYc ++++++= ∑∑∑∑∑∑∑∑ varvarφ

U>lity  system    §           Logical  constraints    §           Demand  constraints  §           Power  balances  §           Mass  balances  

17  

Problem  statement    

ULlity  system  • Existence  of  boiler  • Existence  of  turbine  • Back  pressure  turbine  • ExtracLon  turbine  (addiLonal  cost  $20,000)  

• Flowsheet  power  demand  (7500kW)  • 70%  condensate  return  

HEN  • 2  hot  streams/  2  cold  streams  • Inlet  and  outlet  temperature  can  vary  within  +/-­‐  10  K    

• Heat  capacity  flowrate  can  vary  within  20%  

• Two  treams  have  assigned  costs  • Hot  uLlity  -­‐  HP,  MP,  and  LP  steam  • Cold  uLlity  -­‐  cooling  water  

WN  • HP  boiler  has  more  stringent  feedwater  requirement    

• HP  boiler/MP  boiler  have  different  blowdown  rates  

• RO  consumes  electricity    • Raw  water  needs  treatment  • TSS,  TDS,  GAS  present  in  freshwater  • Discharge  limit  imposed  

Water  network    §  Mass  balances  §  Power  demand  constraint  

Boiler  cost   Turbine  cost   freshwater    cost  

Flowsheet    stream  cost  

Objec>ve  func>on  

Mul>ple  hot  u>lity  targe>ng    (Duran  &  Grossmann)    §  HeaLng  uLliLes  targets  §  Cooling  uLlity  target  

18  

Result  

Sequen>al   Simultaneous  Cost  (1000  $  /  yr)   884.2   641.5  

U>lity  

HP  boiler  flowrate  (kg/s)     Yes   17.66   Yes   18.20  

MP  boiler  flowrate  (kg/s)     No   No  

Power  demand  external  (kW)   HP  à  LP   7500   ExtracLon   7500  

Reverse  osmosis  power  demand  (kW)     MP  à  LP   62.0   MP  à  LP   63.89  

HEN  U>lity  (kW)  

Cooling   1463.8   751.1  

HP  steam   3820.2   5727.2  

MP  steam   13628.2   21065.7  

LP  steam   4743.4   19110.2  

Fcp,H1  (kW/K)   48   32  

Fcp,C2  (kW/K)   144   216  

WN  flowrate  (kg/s)  

Freshwater   7.26   6.47  

Sand  filter   7.2   6.4  

Reverse  osmosis   5.6   5.8  

Scrubber   2.4    1.2  

»  Developed  LP  formulaLons  for  targeLng  minimum  freshwater  consumpLon  for  a  set  of  water-­‐using  process  units  under  a  specific  condiLon    

»  Extended  the  water  targeLng  formulaLon  to  nonisothermal  water  network  

»  Targe>ng  method  can  be  used  to  improve  objec>ve  func>on  and  computa>onal  effort  under  the  simultaneous  approach  for  flowsheet  op>miza>on  

»  The  interac>on  among  power  use,  heat  use,  and  water  use  can  be  exploited  to  achieve  beUer  flowsheet  design  

 

19  

Conclusion  

Thank  you!