Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Article
Just Accepted
Just Accepted manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides Just Accepted as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. Just Accepted manuscripts
appear in full in PDF format accompanied by an HTML abstract. Just Accepted manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI). Just Accepted is an optional service offered
to authors. Therefore, the Just Accepted Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the Just
Accepted Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these Just Accepted manuscripts.
Page 1 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Corresponding author: Tel: +27 11 717 7384; Fax: +27 82 456 1500
Email address: Thokozani.Majozi@wits.ac.za;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 2 of 36
holistic water network superstructure for water and energy minimisation. This is done
through an integrated water network that is open for direct reuse, recycle, and regeneration
reuse/recycle for sustainable cost-effective water and energy usage.
Water pinch techniques and mathematical based optimization techniques are the two main
approaches used for optimal synthesis of water networks in process industry. Wang and
Smith
1,2
presented the seminal work in water pinch analysis which sets to target the reuse,
1,
tabular and heuristic methods for water network synthesis. The insight-based techniques do
not involve computational algorithms in generating solutions. They do require significant
problem simplifications and assumptions and are inherently limited to mass transfer based
operations 3-5.
Takama, et al. 6 presented a superstructure optimisation approach for water minimisation in a
petrochemical refinery based on a fixed mass load. The work of Takama, et al.
extended by several other researchers
7-11
was later
respective water network superstructure. Notable among them are the work of Quesada
Quesada and Grossmann
12
10
13
who
present algorithms to obtain feasible solution in cases of the more complex nonlinear
programming (NLP) and mixed integer nonlinear programming (MINLP) problems that are
in many cases a challenge to solve. Mathematical optimization allows treatment of water
network synthesis problems in their full complexity by considering representative costfunctions, multiple contaminants, and various topological constraints
6, 14-17
. Mathematical
optimization approaches also have an added advantage of simulating the water network into a
desired network structure and operational condition 18.
Tan, et al.
regenerator which allows for possible reuse/recycle. The work of Tan, et al. 5 considered the
blackbox approach which uses linear cost functions for the membrane regenerators. This
does not give an accurate cost representation of the water network. Khor, et al.
19
addressed
the gap on the work of Tan, et al. 5 by developing a detailed model representation for water
network regeneration synthesis using a MINLP optimization framework. The work of Khor,
et al.
19
was, however, limited to a single regenerator with a fixed design. In a more recent
20
Page 3 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
treatment technologies capable of treating all major contaminants. The work focused on unit
specific short cut cost functions in order to gain detailed understanding of trade-offs between
removal efficiency of treatment units and the cost of the units, as well as the impact on the
unit design. To date all the work on water and membrane regeneration has been focusing on
water minimizing and cost functions of the water networks. No effort has been made on the
simultaneous synthesis of the membrane regeneration units and water network for water and
energy minimization.
The membrane technologies adopted in this work involves electrodialysis (ED) and reverses
osmosis (RO). Electrodialysis, in principle, is based on the electromigration of ions through
cation and anion exchange permselective membranes by means of electric current
21-23
Industrial applications of electrodialysis include brackish water desalination, boiler feed and
process water and wastewater treatment 22.
Reverse osmosis (RO) is a pressure driven membrane separation process that selectively
allows the passage of one or more species through the membrane unit. Industrial applications
of the reverse osmosis unit include its use for municipal and industrial water and wastewater
treatment. It has also gained large industrial favor as partitioning regenerator to enhance
water quality for reuse/recycle
24
5, 26, 27
representation of the membrane systems. There is, therefore, a need for a technique that
would cater for energy minimization through detailed synthesis of membrane regeneration
systems in order to obtain optimal variables that affect the operation and economics of the
regenerator unit.
The main objective of this work is to develop a systematic technique that would
simultaneously optimize energy and water consumption within a water network
superstructure. The technique is based on an integrated water and membrane regeneration
network superstructure. The membrane regenerators in this representation are ED and RO.
The choice of regenerators is motivated by the increasing demand and use of membrane
technology for water treatment in the process industry and also the varying potential of
3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 4 of 36
19
The
superstructure in this work incorporates multiple regenerators which are open for parallel and
series connection as well as recycle and reuse of both permeate and reject streams from
regenerators.
The fixed flowrate approach adopted in this work considers water using
4
Page 5 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
processes as sources and sinks that generate or consume a fixed amount of water respectively.
Total fixed flowrate is adopted because it presents a general representation of both mass
transfer and non-mass transfer-based water using operations 27.
ED
RO
Sources
Sinks
J=FW
I=WW
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Q de
Q fe
Permeate
and
reject
to
regenerators
Page 6 of 36
QFed
ee
ED
Q Ped
Q ge
Q Re d
Permeate
and
reject
to
regenerators
Qdr
Permeate
and
reject
to
regenerators
Q fr
Q er
Q Fro
RO
Q Pr o
Q gr
QRro
Q1ed
Q Jed
Sources
Q1x
1
J=FW
Q ids
Q ies
Q1ro
Q Jro
Q 1a, i
QJx
Q i fs
Q igs
Q iz
Sink
i
Q Ja , i
j J
iI
(1)
It should be noted that the wastewater sink, WW, is considered as the final sink.
3.2. Mass and concentration balances for regeneration units
Flow and load balances are conducted for all streams entering the regenerator units. The
streams from sources, permeate and reject streams of both regenerators, are open to
reuse/recycle depending on the component concentration limits of the regeneration units.
Constraints (2) and (3) are water balances around the mixer preceding the ED and RO units
respectively. Where Q Fro and Q Fed
are the total flowrates into the ED and RO units
respectively. Here Q dr and Q er ,
is permeate and reject flowrates from the ED unit into the RO
unit. Whereas, Q fr and Q gr ,
are permeate and reject flowrates from the RO unit into the RO
unit.
ro
dr
er
fr
gr
Fro
Qj + Q + Q + Q + Q = Q
(2)
jJ
Page 7 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ed
de
ee
fe
ge
Fed
Qj + Q + Q + Q + Q = Q
(3)
jJ
Similarly, Q de and Q ee ,
and Q fe and Q ge are flowrates from permeate and reject streams of
ED and RO units into the ED unit.
Based on the varying tolerance of contaminants of the ED and RO units, a contaminant
balance for the regenerator feed is conducted based on the maximum contaminant each
regenerator can take. Hence the corresponding contaminant balance around the mixer of each
regenerator unit is represented by constraints (4) and (5) for ED and RO respectively.
CUe
CUr
ed x
de Ped
+ Q eeC Red + Q feC Pro + Q geC Rro
Qj Cj + Q C
jJ
Fed
ro x
dr Ped
+ Q er C Red + Q fr C Pro + Q gr C Rro
Qj C j + Q C
j J
Fro
(4)
(5)
It should be noted that, C Ue and C Ur are the maximum allowable contaminant into the ED
and RO units respectively. C xj is the concentration from any source j. Whereas C Ped and
C Re d ,
and C Pr o and C Rro are concentrations of contaminants for both permeate and reject
stream of ED and RO units.
3.3. Mass and concentration balances for permeate and reject streams
Based on Figure 2, water balances for permeate and reject streams of the ED unit are
represented in constraints (6) and (7). The flowrates from permeate and reject streams into
sinks are represented by Qids and Qies .
Q Ped = Qids + Q dr + Q de
(6)
Q Re d = Qies + Q er + Q ee
(7)
iI
iI
Similarly, the water balances for permeate and reject streams of the RO unit are represented
by constraints (8) and (9). Where Qi fs and Qigs
are permeate and reject flowrates into sinks.
Q Pr o = Qi fs + Q fr + Q fe
(8)
iI
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 8 of 36
Q Rro = Qigs + Q gr + Q ge
iI
(9)
(10)
jJ
It should be noted, however, that the last source represents the freshwater source, FW. The
maximum load into any sink, i, should not exceed the maximum allowable contaminant
concentrations CiU ,
into that sink. This constraint is expressed by constraint (11).
a
ds Ped
+ Qies C Red + Qi fs C Pro + Qigs C Rro
Q j ,i C j + Qi C
jJ
Qiz
CiU
(11)
Q FedC Fed
Q d C Ped
Q d C Fed
Q ped C Ped
Qr
Qm
r
c
Q C
fc
Q c Cc
Q crC cr
Figure 3: Schematic of a single stage ED plant
Q RedC Red
Page 9 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In order to formulate the model for the regeneration unit, assumptions are made to describe
the feed solution properties, hardware dimensions and operating conditions (Lee et al., 2002).
The assumptions are:
(i)
The diluate and concentrate cells are geometrically similar with identical flow
patterns.
(ii)
The flowrate in the diluate and concentrate compartments are equal and uniform
Qd = Qc .
(iii) The fluid considered is Newtonian, i.e. the viscosity remains constant.
(iv) The unit is operated in a co-current flow.
(v)
During operation the current should not exceed the limiting current density
(vi) Water transport across the membrane is negligible compared to the flowrate in the
diluate and concentrate streams.
(vii) Membrane thickness is negligible.
(viii) The concentration of the salt species is calculated using molar equivalents.
From the diagram Q d is the diluate flowrate, Q c is the concentrate stream flowrate, Q cr is the
concentrate stream recycle flowrate, Q m
is the flowrate from the feed stream that mixed with
the concentrate stream to balance the flowrates. Whereas, C fc
is the feed concentration for
the concentrate stream.
Physical parameters and necessary variables are incorporated to obtain an MINLP model.
The total annualised cost TACe , comprises of the electric current through the ED unit, stack
design considerations, desalination energy requirement, pumping energy requirement and
material balances and is formulated based on the following constraints.
3.5.1 Electric Current
The required electrical power through an ED unit is based on Faradays law which relates to
the driving force that is required to transfer electrons from one stream to another in the ED
unit as related in Lee, et al. 28. It also relates to the degree of desalination C , the flowrate of
the diluate stream and the number of cell pairs N in the stack as represented in constraint (12).
is the current utilisation, F is the faradays constant which is required for the total current
required to drive electrons from one stream to the other. The electrochemical valences of the
ionic contaminants are represented by z.
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 10 of 36
Q d FC z
I=
N
(12)
The degree of desalination is measured by the concentration difference between the diluate
and concentrate streams across the ED unit as defined by the mass balance in constraint (13).
C = C Fed C Ped = C c C fc
(13)
The limiting current density is determined by the mass transfer coefficient for the transport of
ions across the membrane surface. This is difficult to determine theoretically, since it is a
function of solution flow velocity, concentration and stack and spacer configuration Lee, et
al. 28. Therefore, the limiting current density is determined experimentally for a certain flow
velocity, concentration and stack configuration as shown in constrain (14).
I prac = a LCDC d ( u )b
LCD
(14)
A safety factor, , within the range of 0.7 to 0.9, is used to adjust the practical limiting
current density which is dependent on the flow pattern. Constants a and b are determined by
measuring the limiting current density under different flow conditions.
3.5.2 Stack design considerations
Efficient operation of an ED unit is dependent on the membrane area for a given feed
solution, current density, number of cell pairs as well as the production rate
23
. Spacers are
used to enhance mixing and attain uniform flow through the ED unit and also to separate and
support the membranes. However, they reduce the volume of available cell, hence decreasing
the flowrate. A safety factor, , is included to cater for the corrections as shown in constraint
(15). Here is the cell thickness, w is the diluate cell width and v is the linear flow velocity.
Q Ped = N w v
(15)
Membrane area is one of the design characteristics that determine the rate of desalination
within an ED unit. The rate of desalination increases with the exposure of feed water on the
membrane area. The presence of spacers reduces the available area for current due to shadow
effect. As a result the practically required membrane area is larger than the theoretically
10
Page 11 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
required area. A correction factor, , is introduced to account for this effect as can be seen in
constraint (16).
A =
Ped
d
C
C AC prac
+
1
+
I
C c
(16)
From constraint (16), the parameter AC is the area resistance of the anion and cation
exchange membranes, is the equivalent electrical conductivity of the solution.
3.5.3 Energy requirement
The energy required for the operation of an ED system is made of two components. The
energy required for the transfer of ions from the solution across membrane material and the
energy required for pumping the feed solution through the ED unit. The rate of consumption
of either form of energy is dependent, among other factors, on the concentration of feed
solution and the available membrane area to the feed solution.
Direct energy required in an ED unit is dependent on the voltage and current applied across
the unit. The voltage drop across an ED unit is a result of resistance and potentials due to
solutions of different salt concentration
29
ions with membrane matrix and water molecules. There is also energy loss due to electrode
processes in the terminal compartments, although the energy loss due to resistance is much
greater
23
Membranes should be closely arranged in order to reduce energy losses due to resistance of
the cell pair unit as a result of salt transfer
29
C c C Fed
ln
Ped
C N Q z F
C Ped C fc + AC
U=
A
C
11
(17)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 12 of 36
The voltage across an ED unit is, therefore, related to the total power, PER , required to
produce a product capacity, Q Ped . The specific energy required for desalination is
represented by constraint (18).
Pow
Q Ped
E spec =
(18)
The pumping energy is the energy consumed in pumping the feed water into the ED unit.
Since the membrane compartment is arranged in a rectangular form, the flows through the
diluate and concentrate streams are considered to be passing through a rectangular channel.
The geometry of the pipe is considered to be that of rectangular channel with a pressure drop
P =
12 v L
d2
(19)
The pumping energy is, therefore, calculated based on the pressure drop as shown in
constraint (20). Here is a conversion factor available in literature, p is the pumping
efficiency23.
E pump =
P
p
(20)
Q Fed = Q Ped + Q Re d
(21)
Q Fed = Q d + Q m
(22)
12
Page 13 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Q d = Q Ped + Q r
(23)
Q c = Q m + Q cr
(24)
Q Re d = Q c + Q r Q cr
(25)
Corresponding load balances are conducted across the ED unit in order to obtain the species
balance in the streams and to demonstrate that the amount of contaminants removed from the
diluate stream equals the amount of contaminants accumulated in the concentrate stream, for
the case where Qc=Qd.
(26)
Q r C Ped + Q c C c = Q cr C cr + Q Re d C Re d
(27)
Q mC Fed + Q cr C cr = Q c C fc
(28)
Q d C Fed + Q c C fc = Q d C Ped + Q c C c
(29)
The liquid recovery rate, r, is the amount of product water that is directed to the recycle
concentrate stream in order to reduce its salinity and is shown in constraint (30). It is required
in order to avoid water transport due to osmosis across the membranes 23.
r=
Q Ped
Qd
(30)
The purge concentrate is replaced by an amount of the less concentrated feed, and is
represented by the mixing ratio, m, in constraint (31).
Qd
m = Fed
Q
(31)
13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 14 of 36
Based on the formulation and physical design of the ED unit it is assumed that both the
diluate and concentrate maintain a constant flowrate. This is done in order to avoid strain on
the membrane material.
The cost function of the ED regeneration unit which is expressed as the TACe comprises of
the capital and operational costs. The capital cost consists of the costs associated with the
purchase of pumping equipment, and the establishment of the plant. The operating cost is the
costs incurred due to day to day running of the plant over a specified time. It is associated
with the electrical energy costs due to pumping of feed water into the system and
desalination. Both costs are incorporated into a single function called TACe which is
synthesised to obtain optimal design parameters as shown in constraint (32). Here K mb is the
capital cost of membrane, t max is the maximum equipment life, AOT is the annual operating
time of the plant, K el
represents the cost of the electrical power.
TAC e =
K mb A
+ AOT K el Q Ped E Pump + E spec
max
t
(32)
25
19
Q FroC Fro P F
Feed
RO unit
Reject
Permeate
Pump
Q Pr o
Q Rro
C Pr o
C Rro
PP
PR
14
Turbine
Page 15 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
D
constraint (33). Where 2 M
K
D
N solute = 2 M
K
C S
(33)
Water flux relates to rate at which water permeates through RO unit. It is directly related to
the temperature and pressure as well as RO module dimensions and water properties as
shown in constraint (34). Here AP is the water permeability constant, P is the pressure drop
across the unit, F is the osmotic pressure on the feed side of the RO unit, and is a
constant that represents the design features of the hollow fibre module.
N water = AP P Fro
CS
C
(34)
15
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 16 of 36
C Fro + C Rro
CS =
2
(35)
PR and PP
are the feed, retentate and permeate pressures of the RO unit respectively.
P =
( PF + PR )
PP
2
(36)
The pressure on the shell side Pshell ,
of the RO hollow-fibre module is represented by the
pressure difference between the feed side and the reject side of the RO unit.
Pshell = PF PR
(37)
Substituting the shell side pressure drop into the pressure drop across the reverse osmosis unit
gives rise to the feed pressure applied to the RON.
PF = P + shell + PP
2
(38)
16
Page 17 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pow( turb ) =
Pow( pump ) =
(39)
(40)
. In this formulation the osmotic pressure on the permeate side is neglected since the
concentration is assumed to be significantly lower. The osmotic pressure on the retentate side
RO is adopted from the formulation of Saif, et al. 32 and is shown in constraint (41).
ro = OS C Fro
(41)
The osmotic pressure coefficient OS ,
ranges between osmotic pressure on the feed side and
average solute concentration on the feed side. The average concentration on the feed side is
reformulated based on the concentration on the permeate side as adopted from Khor, et al. 19.
The pressures across the membrane material and membrane area are important parameters
that determine the performance of the RON. Here S C
is the solute permeability coefficient.
C Fro =
C Pr o AP [P ro ]
SC
(42)
32
membrane, the osmotic pressure on the reject side and the number of modules present in the
RON according to constraint (43). The parameter S m is the membrane area per module.
Q Pro = N m Sm AP [P ro ]
(43)
Constraint (43) is then reformulated based on the average solute concentration on the feed
side to cater for the number of RO modules in the RON.
17
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Q Pr o
Nm =
AP Sm P OS C Fro
Page 18 of 36
(44)
The cost function of the RON comprises of variables and physical parameters of the reverse
osmosis membrane unit. It consists of the annualised fixed capital cost of turbine, pump,
membrane modules as well as the operating costs for pump and pretreatment of chemicals.
The operating revenue through energy recovery by the turbine at the retentate side is also
incorporated to supplement the cost of energy usage.
0.65
pump
0.43
(45)
Let B = {a, ro,ed , ds, dr, de, es, er, ee, fs, fr , fe, gs, gr, ge}
(46)
Based on constraint (46), constraints (47) (50) is introduced, which satisfy all possible
bounds for minimum or maximum flowrates that can be used to govern the existence of
piping interconnections within the water network superstructure. It can also be used to control
the structural features of the design.
b
Q bLj ,i Y jb,i Q bj ,i Q bU
j ,i Y j ,i
b
Q bLj Y jb Q bj Q bU
j Yj
b B j J i I
b B j J
18
(47)
(48)
Page 19 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
b B i I
(49)
QbLY b Qb QbU Y b
b B
(50)
K bj ,i
pQ bj ,i
+ qY jb,i
3600 v
D bj ,i
(51)
The subscripts j and i represents interconnections between any sources and sinks respectively.
Similarly, K bj , K ib and K b in constraint (52), are costs for interpipping connections from
source to regenerators, permeate and reject streams of regenerators to sinks and permeate and
reject streams into regeneration units. Where Y jb,i , Y jb , Yib and Y b are the binary variables for
existence of piping connections respectively.
a
b
K +
Kj
j ,i
Obj = min jJ iI
b{ro ,ed }, jJ
AF
b
b
Ki +
K
{
}
{
}
b
ds
,
es
,
fs
,
gs
,
i
I
b
B
\
ro
,
ed
,
a
,
ds
,
es
,
fs
,
gs
19
(52)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1-norm
Manhattan
Page 20 of 36
considered
for
all
piping
interconnections. All pipes are assumed to be of the same material properties and as a result
the carbon steel pipes parameters of p and q are adopted for the piping costs. v is the stream
flow velocity and AF is the annualisation factor adopted from Chew, et al. 18 which is used to
annualise the piping cost. The resulting mathematical model is an MINLP. The nonlinear
terms are due to the presence of bilinear terms in mass balance equations and power terms in
the cost functions of regeneration units. The MINLP model was solved using GAMS 24.2
using the general purpose global optimisation solver BARON.
4. Illustrative Example
The developed mathematical model is applied to a pulp and paper case study adopted from
Chew, et al.
18
. The choice of the case study is motivated by the high amount of ionic
components produced by the pulp and paper industry. More so, the pulp and paper industry
involves miscible phase networks which consist of aqueous systems where streams lose their
identities through the mixing process hence the case study is suitable for a fixed flowrate
method adopted for this work.
Table 1: Basic data for water sources and sinks
Sources, j
j
Sinks, i
Flowrate
Concentration
(ton/h)
(mg/L)
2.07
89.4
3.26
34.0
0.34
272
0.34
84.0
0.024
18.3
1.34
50.0
7.22
36.0
7.22
6.3
WW
FW
Flowrate
Max. concentration
(ton/h)
(mg/L)
600
Table 1 shows the basic data for the plant water network which comprises of five water
sources including the freshwater source, and five water sinks including the wastewater sink.
Tables 2 and 3 show process data for ED and RO unit, whereas Table 4 shows the economic
data for detailed design of both regeneration units.
20
Page 21 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Value
Current utilization,
0.9
Cell width, w
0.42 m
10.5 m2/kep
Equivalent conductance,
9.65x107 As/keq
Faradays constant, F
Electrochemical valence, z
25000
LCD
0.5
0.0007 m2
Volume factor,
0.8
Velocity in pipes, v
1 m/s
Shadow factor,
0.7
5 years
Safety factor,
0.7
Pump efficiency,
0.7
0.7
27.2
0.12 $/kWh
mb
150 $/m2
21
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 22 of 36
Value
0.4 atm
4.083x10-4 atm
1.82x10-8 m/s
9.84x10-4 kg/(m.s)
Water viscosity,
5.573x10-8 m/(s.atm)
1 atm
1 atm
Design parameter,
0.694
0.7
0.7
180 m2
0.7
6.5 $/(year.W0.65)
18.4 $/(year.W0.43)
2300 $/year
Value
8760 h
0.12 $/kWh
3x10-5 $/kg
Annualization factor, AF
0.23
FW
0.001 $/kg
0.001 $/kg
7200
2500
22
Page 23 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The case study was applied to three different scenarios in order to ascertain the benefits of
incorporating detailed network of the membrane regeneration units in the constraints of the
water network.
(i) Scenario 1 considered a model of the water network without regeneration.
(ii) Scenario 2 considered a model of water network with regeneration units based on the
blackbox approach.
(iii) Scenario 3 considered detailed synthesis of the regeneration units incorporated into
the overall water network objective function.
For scenario 2 and 3, the optimisation was conducted for both fixed and variable removal
ratios. For comparison, in both cases the removal ratio had a fixed value of 0.7.
The optimal results for all scenarios are presented in Table 5. The results showed a reduction
in freshwater consumption, wastewater generation as well as the total annualised water
network cost as compared to the scenario without regenerations. The results of scenario 3 are
displayed in Figures 8 and 9 respectively. Scenario 3 showed significant reduction in water
network cost as well as freshwater consumption and wastewater reduction as compared to the
direct water network model without regeneration. However, there was an increase in the total
water network cost compared to the scenario 2. This is as a result of scenario 3 being a true
representation of the total water network as it incorporates a detailed design of the membrane
regeneration units and gives an accurate expression of the regeneration cost compared to the
linear cost function of scenario 2 which uses the blackbox method. Apart from presenting
accurate cost of water regeneration units, the detailed model presented in this work also gives
optimal design configurations of the membrane regeneration units for minimum energy
usage.
23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 24 of 36
Scenario 2
Scenario3
Fixed RR
Variable RR
Fixed RR
Variable RR
11.42
9.83
11.64
10.3
37.5
46.3
36.4
43.7
8.89
7.3
9.11
7.75
43.7
53.7
42.3
50.9
1170000
597000
508000
634000
626000
0.06
687.50
2764.20
865
16709.78
18.30
(ton/h)
% of freshwater
savings
Wastewater
15.79
generated (ton/h)
% of wastewater
saved
Total cost of
water network($)
CPU time (s)
From the results in Table 5 it is evident that variable removal ratio in scenario 3 presents the
optimal configuration, since the model is allowed to choose the performance parameters of
the membrane regenerators. The results also show that incorporating multiple membrane
regenerators with different performance and inlet and outlet contaminant limits in a water
network can lead to an optimal use of freshwater. The inlet contaminant limits were set at
different levels in order to give membrane regenerators varying options for contaminant
treatment. The model that incorporates variable removal ratio in scenario 3 proved to be the
optimal result for the case study as represented in Figure 9. The configuration showed
regeneration reuse, regeneration recycle within the water network between the regenerators.
This resulted in 43.7 % reduction in freshwater consumption, 50.9% reduction in wastewater
generation and 46 % savings in the total annualised water network cost as compared to
scenario 1. Tables 6 and 7 show the design results of the ED and RO units, respectively. In
particular, the optimum removal ratios for both units are different from the fixed removal
ratio of 0.7. The results also show the detailed design parameters that could be used in
designing the optimum units in terms of energy and water use. The statistics of the model for
all the cases are shown in Table 8.
24
Page 25 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Value
50
A (m2)
54.4
L (m)
0.8
I (A)
12.5
v (m/s)
0.01
U (V)
30.2
Espec (J/s)
0.02
Epump (J/s)
0.004
P (kPa)
16.3
QFed (t/h)
1.03
RRe
0.8
Value
10
PF (kPa)
5.73x105
PR (kPa)
5.33x105
Pshell (kPa)
4.52x105
ro(kPa)
1.63
Pturb (J/s)
3615.97
Ppump (J/s)
301.66
QFro (t//h)
0.72
RRr
0.65
25
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 26 of 36
Scenario2
Fixed
Variable
RR
RR
Scenario3
Fixed RR
Variable
RR
No. of constraints
31
232
232
276
276
68
185
187
222
224
25
67
67
69
69
Tolerance
0.001
0.001
0.001
0.001
Table 8 shows characteristics of all the models associated with this work. It can be seen that
the number of integer and binary variables increases with the detail model. This makes the
model highly nonlinear as a result it becomes a mixed integer nonlinear programming
(MINLP). MINLP models are inherently difficult to solve hence the CPU time for the
detailed model is high compared to the base model and blackbox models as can be seen in
Table 6.
0.62
1
2
0.34
1.45
0.02
3
4
0 . 34
2 . 62
1.34
7.22
6.78
FW
26
4x10-3
7.22
4
WW
Page 27 of 36
6x10-3
0.04
0.34
ED
3
0.
0
RO
0.06
0
0.3
0.4
2
0.
6
0.3
1
1
0.
0.01
2
3
0.024
0 . 33
1.3
8
2
2.88
1.32
0 . 04
7.22
6.76
FW
WW
Figure 6: Optimum water network configuration for scenario 2 (fixed removal ratio)
0.02
0.07
1.01
0.05
64
0.
0.
34
0.9
4
RO
1.2
0.34
33
0.
0.
53
ED
0
0.5 .30
0
2
3
4
FW
03
0.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1.73
0.024
0.40
2.00
2.90
5.20
4
3.88
WW
Figure 7: Optimum water network configuration for scenario 2 (variable removal ratio)
27
6x10-3
0.02
0.33
ED
1
0.3
0.
63
RO
0
0.3
0.4
0.3
0
12
0.
0.34
2
0.024
1
2
2 . 65
1 . 44
1.32
7.22
6.80
FW
WW
0.45
Figure 8: Optimum water network configuration for scenario 3 (fixed removal ratio)
0.03
0.04
0.58
ED
3
0.
0
RO
0.7
2
0.3
3
8
0.3
2
0.
0.
63
0
0.6
1
1
0.
2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 28 of 36
2
3
0.024
0 . 66
0 . 01
0.04
2
2.60
1.4
4
1.08
6.62
5.80
FW
WW
Figure 9: Optimum water network configurations for scenario 3 (variable removal ratio)
28
Page 29 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5. Conclusion
This work has addressed the synthesis of a multi-membrane regeneration water network by
proposing a MINLP optimisation model that incorporates detailed mechanistic models of an
ED and RO units within a water network. The developed water network allows for direct
reuse/recycle and regeneration reuse/recycle. The developed MINLP model is applied to a
pulp and paper industry case study and solved using GAMS/BARON. The results showed
that setting the removal ratio as variable yields optimal configuration as compared to the
fixed removal ratio. The model also showed that incorporating a detailed model of the
regeneration unit into the overall model yields accurate expression of the regeneration cost as
compared to the black-box model. It also gives optimal operating variables of the
regeneration units for minimal energy and water usage. The results in this work demonstrate
that integrating detailed models of regeneration units in a water network superstructure
results in significant reduction in water consumption as well as wastewater reduction in the
process industry. The optimal design of the membrane regeneration units also aids by given
investors accurate cost representation for minimal energy usage. The complexity of the model
results in high computational time. However, this is cannot be deemed a serious limitation in
the context of synthesis and design, since this particular problem is solved once prior to
detailed design. This is unlike in scheduling models where the optimization problem is solved
very frequently over short time horizons. It is noteworthy that the proposed model can be
extended to include multiple type regenerators. Hence the formulation in this work offers a
scope for future work to incorporate multiple ED, RO, ultrafiltration, nanofiltration units, etc.
6. Nomenclature
6.1 Sets
J = {j|j = water source}
I = {i|i = water sink}
B ={b|b = flowrates}
6.2 Parameters
a LCD
b LCD
current density
F
29
Faraday constant
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Electrochemical
Sm
valence
Page 30 of 36
K el
K mb
Viscosity of water
cost
RR e
K tr
Conversion factor
AP
Water permeability
unit
RR r
unit
coefficient
AOT
Number of years
K Chem
LR e
LR r
Q jx
treatment
C jx
Contaminant co from
K FW
K elect
K mod
source j
Q iz
Flowrate of sink i
Maximum allowable
C iU
membrane module
K Pump
treatment of chemicals
WW
contaminant
Cost coefficient of
pump
Cost coefficient of
Volume factor
turbine
Shadow factor
D2 M K
Safety factor
SC
Solute permeability
Water viscosity
coefficient
Design parameter
OS
Osmotic pressure
Current utilisation
PF
Feed pressure
Cell thickness
PP
Permeate pressure of
RON
K turb
p, q
ED unit
pump
steel piping
30
Page 31 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
turb
ED unit
RO unit
AC
Q Re d
Membrane resistance
unit
Q Fro
Equivalent conductance
Q Pr o
Velocity of fluid in
Q Rro
Manhattan distance
Qids
Manhattan distance
regeneration units
Q de
Manhattan distance
ED unit
from source to
Dib
Reject flowrate of RO
unit
D bj
Permeate flowrate of
RO unit
pipes
D bj ,i
Reject flowrate of ED
Q dr
Manhattan distance
RO unit
reject streams to ED
Qies
and RO units
Q ee
FW
Freshwater flowrate
WW
Wastewater flowrate
ED unit
Q aj ,i
Q roj
Q er
to sink i
RO unit
to RO unit
Q edj
Q Fed
Q Ped
Qi fs
to ED unit
sink i
Q fe
unit
Permeate flowrate of
ED unit
31
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Q fr
of ED unit
RO unit
Qigs
Cc
contaminant in the
concentrate stream
C cr
ED unit
CUr
contaminant into RO
unit
C Ue
contaminant into ED
contaminant co into ED
unit
Qd
Concentration of
Qc
unit
Concentration of
Qm
Qr
unit
Concentration of
Q cr
Concentration of
Membrane area of ED
unit
permeate stream of RO
Electric current
unit
Velocity of stream in
Concentration of
ED unit
contaminant co in the
reject stream of RO unit
C
fc
Feed concentration in
Pressure drop
I prac
Practical current
32
Concentrate stream
recycle flowrate
contaminant co in the
Rro
Recycle stream
flowrate of ED unit
contaminant co in the
Pr o
Mixing flowrate of ED
unit
permeate stream of ED
C
Concentrate stream
flowrate of ED unit
contaminant co in the
Re d
contaminant co into RO
C Ped
Maximum allowable
Concentration of
unit
C Fro
Maximum allowable
C Fed
Concentration of
Q gr
Concentration of
Q ge
Page 32 of 36
Page 33 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
density
E pump
Pumping energy
Nm
E spec
Specific energy
K bj ,i
osmosis modules
N
K ib
units
1 Existence of piping
interconnection between
b
y j ,i =
source and sink,
0 Otherwise
1 Existence of piping
interconnection between
b
y j =
source and regeneration units,
0 Otherwise
Kb
Pow
Pow( turb )
Power of turbine
Pow( pump )
Power of pump
Cs
Concentration on the
1 Existence of piping
interconnection between
b
y i =
permeate and reject and sinks,
0 Otherwise
1 Existence of piping
interconnection between
b
y =
permeate and reject into units,
0 Otherwise
shell side
N solute
Solute flux
N water
Water flux
TAC e
TAC r
ro
Osmotic pressure at
retentate side of RO
unit
Osmotic pressure on
the feed side
33
Number Electrodialysis
cell pairs
source to sinks
K bj
Number of reverse
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 34 of 36
7. Acknowledgement.
The authors would like to thank the Water Research Commission (WRC) and National
Research Foundation (NRF) for funding this work under the NRF/DST Chair in Sustainable
Process Engineering at the University of the Witwatersrand, South Africa
8. References
1.
Wang, Y.; Smith, R., Wastewater minimisation. Chemical Engineering Science 1994,
49, (7), 981-1006.
2.
Wang, Y.-P.; Smith, R., Design of distributed effluent treatment systems. Chemical
Engineering Science 1994, 49, (18), 3127-3145.
3.
Manan, Z. A.; Tan, Y. L.; Foo, D. C. Y., Targeting the minimum water flow rate
using water cascade analysis technique. AIChE Journal 2004, 50, (12), 3169-3183.
4.
Bandyopadhyay, S.; Cormos, C.-C., Water management in process industries
incorporating regeneration and recycle through a single treatment unit. Industrial &
Engineering Chemistry Research 2008, 47, (4), 1111-1119.
5.
Tan, R. R.; Ng, D. K.; Foo, D. C.; Aviso, K. B., A superstructure model for the
synthesis of single-contaminant water networks with partitioning regenerators. Process Safety
and Environmental Protection 2009, 87, (3), 197-205.
6.
Takama, N.; Kuriyama, T.; Shiroko, K.; Umeda, T., Optimal planning of water
allocation in industry. Journal of Chemical Engineering of Japan 1980, 13, (6), 478-483.
7.
Ahmetovi, E.; Grossmann, I. E., Strategies for the global optimization of integrated
process water networks. Computer aided chemical engineering 2010, 28, 901-906.
8.
Doyle, S.; Smith, R., Targeting water reuse with multiple contaminants. Process
safety and environmental protection 1997, 75, (3), 181-189.
9.
Huang, C.-H.; Chang, C.-T.; Ling, H.-C.; Chang, C.-C., A mathematical
programming model for water usage and treatment network design. Industrial & Engineering
Chemistry Research 1999, 38, (7), 2666-2679.
10.
Karuppiah, R.; Grossmann, I. E., Global optimization for the synthesis of integrated
water systems in chemical processes. Computers & Chemical Engineering 2006, 30, (4), 650673.
11.
Rossiter, A.; Nath, R., Wastewater minimization using nonlinear programming. Waste
Minimisation through Process Design. New York: McGraw Hill 1995.
12.
Quesada, I.; Grossmann, I. E., Global optimization of bilinear process networks with
multicomponent flows. Computers & Chemical Engineering 1995, 19, (12), 1219-1242.
34
Page 35 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
13.
Faria, D. C.; Bagajewicz, M. J., On the appropriate modeling of process plant water
systems. AIChE journal 2010, 56, (3), 668-689.
14.
Savelski, M. J.; Bagajewicz, M. J., On the optimality conditions of water utilization
systems in process plants with single contaminants. Chemical Engineering Science 2000, 55,
(21), 5035-5048.
15.
Bagajewicz, M.; Savelski, M., On the use of linear models for the design of water
utilization systems in process plants with a single contaminant. Chemical engineering
research and design 2001, 79, (5), 600-610.
16.
Lee, S.; Grossmann, I. E., Global optimization of nonlinear generalized disjunctive
programming with bilinear equality constraints: applications to process networks. Computers
& chemical engineering 2003, 27, (11), 1557-1575.
17.
Gunaratnam, M.; Alva-Argaez, A.; Kokossis, A.; Kim, J.-K.; Smith, R., Automated
design of total water systems. Industrial & Engineering Chemistry Research 2005, 44, (3),
588-599.
18.
Chew, I. M. L.; Tan, R.; Ng, D. K. S.; Foo, D. C. Y.; Majozi, T.; Gouws, J., Synthesis
of direct and indirect interplant water network. Industrial & Engineering Chemistry Research
2008, 47, (23), 9485-9496.
19.
Khor, C. S.; Foo, D. C.; El-Halwagi, M. M.; Tan, R. R.; Shah, N., A superstructure
optimization approach for membrane separation-based water regeneration network synthesis
with detailed nonlinear mechanistic reverse osmosis model. Industrial & Engineering
Chemistry Research 2011, 50, (23), 13444-13456.
20.
Yang, L.; Salcedo-Diaz, R.; Grossmann, I. E., Water Network Optimization with
Wastewater Regeneration Models. Industrial & Engineering Chemistry Research 2014, 53,
(45), 17680-17695.
21.
Korngold, E., Electrodialysis unit: Optimization and calculation of energy
requirement. Desalination 1982, 40, (1), 171-179.
22.
Strathmann, H., Electrodialysis, a mature technology with a multitude of new
applications. Desalination 2010, 264, (3), 268-288.
23.
Tsiakis, P.; Papageorgiou, L. G., Optimal design of an electrodialysis brackish water
desalination plant. Desalination 2005, 173, (2), 173-186.
24.
Garud, R.; Kore, S.; Kore, V.; Kulkarni, G., A short review on process and
applications of reverse osmosis. Universal journal of Environmental research and technology
2011, 1, (2).
25.
El-Halwagi, M., Sustainable design through process integration. Massachusetts:
Elsevier Inc 2012.
26.
Alva-Argez, A.; Kokossis, A. C.; Smith, R., Wastewater minimisation of industrial
systems using an integrated approach. Computers & Chemical Engineering 1998, 22, S741S744.
35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 36 of 36
27.
Khor, C. S.; Chachuat, B.; Shah, N., A superstructure optimization approach for water
network synthesis with membrane separation-based regenerators. Computers & Chemical
Engineering 2012, 42, 48-63.
28.
Lee, H.-J.; Sarfert, F.; Strathmann, H.; Moon, S.-H., Designing of an electrodialysis
desalination plant. Desalination 2002, 142, (3), 267-286.
29.
Strathmann, H.; Krol, J.; Rapp, H.-J.; Eigenberger, G., Limiting current density and
water dissociation in bipolar membranes. Journal of Membrane Science 1997, 125, (1), 123142.
30.
Dandavati, M. S.; Doshi, M. R.; Gill, W. N., Hollow fiber reverse osmosis:
experiments and analysis of radial flow systems. Chemical Engineering Science 1975, 30,
(8), 877-886.
31.
Evangelista, F., Improved Graphical-Analysis Method for the Design of ReverseOsmosis Plants. Industrial Engineering Chemical Process 1986, 25, (86), 366-375.
32.
Saif, Y.; Elkamel, A.; Pritzker, M., Optimal design of reverse-osmosis networks for
wastewater treatment. Chemical Engineering and Processing: Process Intensification 2008,
47, (12), 2163-2174.
36