Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Supporting Information for

Sodium Hydroxide Production from Seawater


Desalination Brine: Process Design and Energy
efficiency

Fengmin Dua, David E. M. Warsingera, Tamanna I. Urmia, Gregory P. Thiela, Amit Kumara, John H Lienhard Va*

a
Rohsenow Kendall Heat Transfer Laboratory, Department of Mechanical Engineering Massachusetts Institute of
Technology, 77 Massachusetts Avenue, Cambridge MA 02139-4307 USA

*
Corresponding Author lienhard@mit.edu

The supporting information gives Aspen modeling details of the brine-to-caustic process. Firstly, modeling tools
available in Aspen are briefly introduced. Then, detailed modeling procedures of the membrane electrolyzer are
given, including the validation of the developed model against a set of reference plant data in the literature.
Dependencies of the electrolyzer performance on its modeling parameters are shown in a set of parametric
studies. Afterwards, the Aspen models of pre- and post-treatment components involved in the brine-to-caustic
process are elaborated. At last, mass flows, temperatures and salt concentrations at each stage of the process
are summarized, which should help to reproduce the study.
Additionally, we discuss briefly how the thermodynamic least work is defined and how it is evaluated in Aspen.
24 Pages, 7 Tables, 6 Figures

S-1
Short Introduction to Aspen Plus
Aspen Plus provides various features for construction and study of models. Some of them involved in this work
are briefly introduced in this section.

Blocks
The following blocks (components) provided by Aspen Plus are implemented in our modeling. Their functions
used are briefly introduced in the following:

• Mixer: allows for two or more feed streams to be combined;

• FSplit (Splitter): allows the feed stream to be separated into two or more fractions with same
composition and properties;

• Sep (Separator): allows the feed stream to be separated to multiple fractions based on user-specified
splits for each individual species;

• Flash2 (Flash): allows vapor-liquid separation by specified temperature, pressure or heat duty;

• Heater: models one side of a heat exchanger by changing the thermodynamic condition (temperature,
pressure, vapor fraction) of the feed stream. Heat stream can be specified as well;

• HeatX (Heat exchanger): models two-stream heat exchanger;

• MHeatX: models multiple-stream heat exchanger;

• RStoic (Stoichiometric Reactor): models reactor with known extent or conversion;

• RGibbs (Gibbs Reactor): models equilibrium reactor. Aspen calculates the product composition by
minimizing the Gibbs free energy;

• Pump: models pump;

• Compr: models compressor;

• Valve: models valve, i.e. for throttling;

• CFuge: models the filtration.

S-2
Streams
Streams either connect Blocks together or is used as feed or product of the flowsheet. Material, heat or work
streams can be specified in Aspen plus, whereas only material streams are involved in this work. The
composition, flow rate and thermodynamic properties have to be provided for feed material streams.1 Other
streams are calculated by Aspen based on the Blocks that they connect to.

Internal recycling streams (i.e. the recirculating brine in Figure 1) are known as tear streams in Aspen. Aspen
automatically iterates to solve for flows and compositions of these streams. However, initial guess of the
composition and properties of the tear streams may be given by user to enhance the numerical stability of the
model.

Tools for simulation and analysis


• Design Specification (DS)

A DS sets a variable in a simulation which would be calculated elsewise. Aspen internally adjusts a
user-specified input parameter to meet this requirement.1 For example, the requirement is the pH
after an acidifier (mixer), and the requirement can be met by varying the acid stream into the mixer.

• Sensitivity Analysis

Sensitivity analysis varies an input parameter and calculates how other user-specified variables
depend on this parameter.

Modeling details

Membrane electrolyzer in detail

Cathode chamber and membrane transport


The transport through ion-exchange membrane is closely related to the production on the cathode side (red
part in Figure 2, main paper), as the sodium ions that pass through the membrane become part of the product
caustic. Along with sodium ions, water is transported through the ion-exchange membrane.2 The molar ratio of

S-3
water/sodium transported is called the water transport number, denoted by 𝑓𝑓. Transport of other ions are of
very small extent2 and thus neglected.

On the cathode, hydrogen and hydroxide ions are produced. Due to lack of cathodic side reactions, hydrogen
produced by the cell is stochiometric and obeys Faraday’s law:

2𝑛𝑛̇ H2
𝑛𝑛̇ H2 = 𝑛𝑛̇ H2 ,id ⇒ 1 = (S-1)
(𝐼𝐼/𝐹𝐹)

Hereby, 𝐼𝐼 is the current and 𝐹𝐹 the Faraday’s constant (96,485 C/mol).

The ideal stoichiometric production of NaOH is:

𝑛𝑛̇ NaOH,id
1= (S-2)
(𝐼𝐼/𝐹𝐹)

However, a fraction of the produced hydroxide ions undesirably migrates back to the anolyte through the
membrane. This hydroxide back-migration reduces sodium transport through the membrane and the production
of caustic soda. The cathode current efficiency 𝜂𝜂, which is defined by the following,2 takes this effect into
account:

∆𝑛𝑛̇ NaOH ∆𝑛𝑛̇ NaOH 𝑛𝑛̇ Na+


𝜂𝜂 = = = (S-3)
𝑛𝑛̇ NaOH,id (𝐼𝐼/𝐹𝐹) (𝐼𝐼/𝐹𝐹)

∆𝑛𝑛̇ NaOH is the caustic soda molar stream produced in the cell. This equals to the sodium molar flow
transported through the membrane, 𝑛𝑛̇ Na+ .

Anode chamber

The reactions occurring are illustrated in Figure S1. Ideally, a Faraday of electricity (1 mol electrons) produces
0.5 mol gaseous chlorine. In real cells, however, this amount is reduced by oxygen production (side reaction 1 in
Figure S1) and dissolution of chlorine into the anolyte (side reaction 2 in Figure S1). We take this into account by
introducing the anode current efficiencies.

S-4
Side reaction 1

Side reaction 2

Figure S1. Reaction network occurring in the electrolyzer anode chamber. The main reaction is chlorine
oxidation (leading to Cl2). The primary side reactions are water oxidation (1) and chlorine dissolution (2) in the
anolyte solution. The blue color indicates reactions that occur on the anode and are irreversible. The green color
indicates equilibrium reactions in the anolyte solution.

The process chlorine current efficiency, 𝜉𝜉 P , is defined by equation (S-4), derived from the literature:2

𝑛𝑛̇ Cl2 ,total 2𝑛𝑛̇ Cl2 ,total ∆𝑛𝑛̇ NaCl


𝜉𝜉 P = = = (S-4)
𝑛𝑛̇ Cl2 ,id (𝐼𝐼/𝐹𝐹) (𝐼𝐼/𝐹𝐹)

The chlorine current efficiency, also known as the anode current efficiency2, 𝜉𝜉, is defined by equation (S-5):

𝑛𝑛̇ Cl2 ,g 2𝑛𝑛̇ Cl2 ,g


𝜉𝜉 = = (S-5)
𝑛𝑛̇ Cl2 ,id (𝐼𝐼/𝐹𝐹)

where 𝑛𝑛̇ Cl2 ,total is the total chlorine flow produced on the anode while 𝑛𝑛̇ Cl2 ,g is the gaseous chlorine flow.
𝑛𝑛̇ Cl2 ,id denotes the chlorine production of an ideal cell. The quantity 𝜉𝜉 P therefore accounts for side reaction 1
(water oxidation) whereas 𝜉𝜉 considers both side reactions 1 and 2 (water oxidation and chlorine dissolving).

The oxygen production can be calculated based on 𝜉𝜉 P :

1 𝐼𝐼 1 𝐼𝐼
𝑛𝑛̇ O2 = � − 2𝑛𝑛̇ Cl2 ,total � = �1 − 𝜉𝜉 P � ⋅ (S-6)
4 𝐹𝐹 4 𝐹𝐹

S-5
The NaCl conversion, 𝑋𝑋NaCl, is the fraction of feed NaCl that reacts in the electrolysis cell and is a primary
performance parameter. Equation (S-7) shows the relationship between this quantity and the electricity
consumed:

∆𝑛𝑛̇ NaCl 2𝑛𝑛̇ Cl2 ,total 𝜉𝜉 P 𝐼𝐼


𝑋𝑋NaCl = = = ⋅ (S-7)
𝑛𝑛̇ NaCl,F 𝑛𝑛̇ NaCl,F 𝑛𝑛̇ NaCl,F 𝐹𝐹

where 𝑛𝑛̇ NaCl,F is the molar flow of NaCl in the electrolyzer feed, and ∆𝑛𝑛̇ NaCl is the molar amount of NaCl that
reacts in the electrolyzer per time unit.

The process chlorine efficiency 𝜉𝜉 P plays a key role in both equations (S-6) and (S-7), linking the NaCl
conversion and the oxygen production to the current consumed. For this reason, 𝜉𝜉 P is used instead of 𝜉𝜉 (in
contrast to the literature2 as the “anode current efficiency” in the electrolyzer modeling.

Modeling of membrane electrolyzer in Aspen Plus


Several modeling approaches of membrane electrolyzer are found in the literature.3-6 While some investigated
global performance of the membrane cell,4,5 other focused on details i.e. reaction kinetics and transport
phenomena.3,6 As only steady-state solution is of interest here, we use a different modeling approach for the
electrolyzer, namely by combination of available unit operations (Blocks) in Aspen. Without the modeling of
reaction kinetics and transport phenomena, this simulation approach still allows the implementation of
important process parameters (i.e. current efficiencies, water transport number) and the determination of
material flows inside the cell in detail, while taking chemical and phase equilibrium into consideration.

The resulting Aspen flowsheet is illustrated in Figure S2.

S-6
Figure S2. Membrane chlor-alkali electrolysis cell model in Aspen Plus.

The used unit operations and their major tasks are listed in Table S1.

Table S1. Aspen units and their major tasks in the membrane electrolysis cell.

Aspen block Corresponding cell part Major task of the unit

Mixer ACIDIF Acidifier Adjust the feed brine pH

RStoic R-AN Anode Reaction Cl– to Cl2 and H2O to O2/H+

RGibbs REX Anolyte Equilibrium gaseous Cl2 with anolyte

Sep MEMB Membrane Separate Na+ and H2O from anolyte -> catholyte

RStoic R-CA Cathode Reaction H2O to H2/OH–

Sep S-BACK Membrane Separate OH– from catholyte -> anolyte

Flash2 FLASH Catholyte Separate gaseous hydrogen from catholyte

- Split concentrated caustic into product and recycle


FSplit CAUS-SPL
stream

Mixer CAUS-MIX - Mix the recycle stream with deionized water

S-7
Modeling parameters used are listed in Table 1 (main paper).

The Design Specs (DS) used in the model are summarized below:

Table S2. Aspen Design Specs in membrane electrolysis cell.

Aspen Design Spec Expression Vary

𝑛𝑛̇ Cl2
Anode efficiency = 𝜉𝜉 P = 96% 𝑛𝑛̇ H2
𝑛𝑛̇ H2

𝑛𝑛̇ Na+
Cathode efficiency = 𝜂𝜂 = 94% 𝑛𝑛̇ Na+
𝑛𝑛̇ H2

Caustic concentration in 𝑤𝑤NaOH,in = 30.3% Split ratio

Caustic concentration out 𝑤𝑤NaOH,out = 32% Dilution water flow

O2 production 𝑛𝑛̇ H2 − 𝑛𝑛̇ Cl2 = 2𝑛𝑛̇ O2 𝑛𝑛̇ O2

𝑛𝑛̇ H2 O
Water transport number = 𝑓𝑓 = 4.25 𝑛𝑛̇ H2 O
𝑛𝑛̇ Na+

Hydroxide back-migration 2𝑛𝑛̇ H2 − 𝑛𝑛̇ Na+ = 𝑛𝑛̇ OH− 𝑛𝑛̇ OH−

Validation of membrane electrolyzer model


The results of the membrane electrolyzer model are validated against a set of reference plant data given in
the Handbook of chlor-alkali technology.2 Model parameters and their values for the validation are listed in
Table S3. Since all model parameters except the mass flows in the feed are independent of mass, the model is
scalable. For this comparison, all flows in the reference plant are scaled down by a factor of 1000.

Table S3. Parameters used in model validation of the membrane electrolyzer. The quantity 𝜉𝜉 P affects the
chlorine/oxygen ratio, 𝜂𝜂 the hydroxide back-migration, and 𝑓𝑓 the water transport through the membrane. The
fraction of feed NaCl that is converted in the cell is 𝑋𝑋NaCl.

Symbol Description Value Used Unit Source

𝑇𝑇cell Temperature 88 °C Reference2

S-8
𝑝𝑝An Pressure anolyte 1.09 bar Reference2

𝑝𝑝Ca Pressure catholyte 1.05 bar Reference2

𝑚𝑚̇NaCl,F NaCl flow in the feed 111.612 kg/h Reference2

𝑚𝑚̇H2 O,F Water flow in the feed 335.684 kg/h Reference2

𝑚𝑚̇HCl,F HCl flow in the feed 0.0272 kg/h Reference2

𝑚𝑚̇Na2 SO4 ,F Na2SO4 flow in the feed 2.400 kg/h Reference2

𝑚𝑚̇NaClO3 ,F NaClO3 flow in the feed 3.720 kg/h Reference2

𝑤𝑤NaOH,P Product NaOH concentration 32 wt% Reference2

𝑤𝑤NaOH,F Recycle NaOH concentration 30.3 wt% Reference2

𝜂𝜂 Cathode current efficiency 94 % Reference2

𝜉𝜉 P Anode current efficiency 95.5 % Fitted

𝑓𝑓 Water transport number 4.25 - Reference2

𝑋𝑋NaCl NaCl conversion 44 % Fitted

Table S4. Comparison of model with data from the reference plant2 scaled down 1000 times.

System characteristic Model Reference Unit Relative difference (%)

Anolyte Flow 336.8 335.7 kg/h +0.3

Anolyte Concentration 18.75 18.73 wt% +0.1

Anolyte dissolved Cl2* 1.819 1.002 kg/h +81.5

Chlorine in anode gas 27.99 29.06 kg/h –3.7

Oxygen in anode gas 0.315 0.3093 kg/h +1.8

Water in anode gas 6.875 8.117 kg/h –15.3

Water flow through membrane 63.35 62.00 kg/h +2.2

S-9
Hydrogen in cathode gas 0.883 0.875 kg/h +0.9

Water in cathode gas 3.444 3.500 kg/h –1.6

Caustic production 103.4 104.2 kg/h –0.8

*Anolyte dissolved Cl2 counts equivalent amounts of ClO– (1 mol equivalent to 1 mol Cl2) and ClO3– (1 mol
equivalent to 3 mol Cl2).

The results are shown in Table S4. Most of the relative differences are under ±5%. The electrolyzer model in
Aspen is therefore capable of predicting the stream properties from and to the membrane cell under given
parameters. Additionally, the validation confirms the scalability of the model.

However, two items in Table S4 have high error: the Cl2 dissolved in the anolyte and the water in the anode-
side vapor. The latter does not affect the system performance and is thus tolerated. The former is presumably
caused by the assumption that Cl2 is in complete chemical equilibrium with the anolyte (side reaction 2 in Figure
S1). In a real system, the ideal equilibria may not be reached due to finite reaction rates. Modeling of the
reaction kinetics is beyond the detail level of this system-level modeling study. The influence of this error on the
system process performance would be simply the overestimation of the dechlorination chemical dosage.
Another possible reason for the deviation between simulation and real values is the inaccuracies in the
thermodynamic property functions, implemented in the Aspen ENRTL model.

Parametric study of the membrane electrolyzer model


This section gives an overview of the dependency of membrane cell performance on its parameters, introduced
in Table 1 (main paper) and SI section “Modeling of membrane electrolyzer in Aspen Plus”. All parametric
studies of this section are based on the validation case, introduced in the last section.

NaCl conversion in electrolyzer


NaCl conversion is the fraction of NaCl reacted in the cell. Due to stoichiometry of the electrolysis reaction,
the NaCl conversion directly relates to the current consumption in the cell (see equation (S-7)). In the parametric
study, the NaCl conversion is varied from 0.4 to 0.5. Figure S3 shows product (NaOH and Cl2) productivity at the
anode compared to the NaCl conversion: the relationship is linear.

S-10
Figure S3. Productivity of the electrolyzer (production rate of NaOH and gaseous Cl2) with respect to NaCl
conversion. The stability limit (dotted line) on anolyte concentration is 200 g/L NaCl (= 18.2 wt% at 90 °C, by
Aspen).2 Other parameters remain as the validation case (see Table S3). Although cell productivity increases with
NaCl conversion, in the given case, a >49% conversion would lead to unstable operation (red zone).

Higher conversion (shown in green, right axis) values are desirable but are limited by instability at low anolyte
NaCl concentration (unstable below 18.2 wt%), limiting conversion to no more than 49%. In the process chain,
the NaCl conversion is adjusted by a DS so that a 19 wt% anolyte concentration is reached (see Table 1 in main
paper). This way, a relative high cell productivity is achieved with a safe margin from unstable operating
conditions.

Anode and cathode current efficiencies in electrolyzer

Current efficiencies play a key role in terms of energy consumption and cell productivity of the electrolyzer
and require a detailed parametric study. Varying the cathode current efficiency 𝜂𝜂 between 90% and 100% yields
Figure S4.

S-11
Figure S4. Impact of cathode current efficiency on NaOH and gaseous Cl2 production as well as anolyte pH,
anode current efficiency 95.5 % constant, other model parameters as Table S3. The pH declines rapidly at
current efficiencies above 95%, risking system damage.

Figure S4 shows the cell productivity as well as the pH of the anolyte solution with respect to the cathode
current efficiency 𝜂𝜂. NaOH production increases linearly with an increasing efficiency. This can be deduced from
equation (S-3).

At low cathode current efficiencies (𝜂𝜂 < 95%), chlorine gas production is not at maximum. This is due to the
back-migrated hydroxide ions which react with the chlorine. As the hydroxide ions are always fully depleted by
chlorine, the pH value of the anolyte does not go beyond 4.3. At high cathode current efficiencies (𝜂𝜂 > 95%),
however, the pH of the anolyte goes down to almost 1, which is undesirable in membrane cells as the ion-
exchange membrane may be protonated.2 It is therefore crucial to monitor and control the pH at the anolyte by
adjusting the HCl dosage in the brine acidification. Literature2 suggests that, at 96% cathode current efficiency or
higher, very little or no acid is needed. This agrees well with our result.

S-12
The results of varying the anode current efficiency 𝜉𝜉 P are illustrated in Figure S5. While varying 𝜉𝜉 P , the total
consumed electric current 𝐼𝐼 is kept at constant. The NaCl conversion is adjusted accordingly, based on equation
(S-7).

Figure S5. Impact of anode current efficiency on production of NaOH and gaseous Cl2 as well as the anolyte pH
under constant current consumption, cathode efficiency 94% constant, other parameters as Table S3.

It can be concluded that caustic soda production under constant current consumption is independent of 𝜉𝜉 P .
Chlorine gas production as well as the anolyte pH reach a plateau at 94% anode current efficiency and do not
increase further. The transition occurs exactly when both current efficiencies are equal:

𝜂𝜂 = 𝜉𝜉 P (S-8)

This transition is also present in Figure S4 where the anolyte pH and chlorine production largely change when
𝜂𝜂 reaches about 95%. Note in that sensitivity study, the anode current efficiency is 95.5%.

S-13
This fact can be explained by comparing the protons produced on the anode and the hydroxides leaked back
to the anode. If anode (𝜉𝜉 P ) and cathode efficiencies (𝜂𝜂) are equal, the protons and hydroxides exactly neutralize.

𝐼𝐼 𝐼𝐼
𝑛𝑛̇ OH− = (1 − 𝜂𝜂) ⋅ ; 𝑛𝑛̇ H+ = 4𝑛𝑛̇ O2 = �1 − 𝜉𝜉 P � ⋅ (S-9)
𝐹𝐹 𝐹𝐹

If 𝜂𝜂 is higher, protons will be in an excess amount which will bring the anolyte pH down. If 𝜉𝜉 P is higher,
hydroxides will remain and react with chlorine, reducing the gas production amount.

It can be concluded that high anode and the cathode current efficiency is beneficial for chlorine and caustic
soda production, respectively. However, both efficiencies ought to have close values from each other. If 𝜂𝜂 is too
low, the chlorine produced will be wasted in the cell and the dechlorination costs will increase. If 𝜉𝜉 P is too low,
the anolyte pH will be very low and hard to control.

Given the results of the sensitivity study, the feed brine pH in the final system-level process is set at 3.
Although lower feed pH brings improvement in the anode current efficiency2, a pH under 2 is not tolerable for
the cell.2 As pH of the anolyte is strongly dependent on the relative value of both current efficiencies which are
not controllable, a safe distance to the limit should be kept. The chosen brine pH is slightly lower than that in the
validation case. A lower feed brine pH leads to a higher anode current efficiency2 and an estimated value of 96%
compared to 95.5% in the validation case is applied (see Table 1 of main paper).

Varying feed NaCl concentration into the electrolyzer

The feed NaCl concentration is an important parameter in the electrolyzer as it determines the cell
productivity and the energy consumption from the concentration steps earlier. Its impact on the cell itself is
investigated in this section.

In the sensitivity study, feed NaCl concentration is varied by changing the water flow from Table S3. NaCl flow
and other parameters are kept constant. The change in water flow matches a NaCl solution of 25 wt% to 27 wt%
(slightly oversatured, fed with solid NaCl). The interval is chosen according to the feed concentration
requirement of 290-310 g/L NaCl.7

Apparently, a higher feed NaCl concentration would result in a higher anolyte concentration, providing that
the current consumed is kept constant. As the maximum NaCl conversion is majorly dependent on the lower
limit of the anolyte NaCl concentration, a higher feed NaCl concentration makes higher conversion possible. The

S-14
maximum cell productivity (as dry NaOH) at maximum conversion (anolyte at 200 g/L, limit of stable operation)
is compared in Figure S6 where it is shown that the cell productivity is higher with feed NaCl concentration.

Figure S6. NaOH production rate (dry) at various feed brine concentration, maximum conversion (anolyte
concentration 200 g/L), model parameters as Table S3.

Modeling of other components in Aspen Plus

Summary of components

Table S5 lists the individual pre- and post-treatment components with their main roles in the process chain.

Table S5. Summary of components and their major tasks in the process chain (Figure 1).

Component Category Major Task

Remove some of Ca, Mg; remove most of


Nanofiltration (NF) Primary purification
sulfate

S-15
Electrodialysis (ED) Primary concentration Concentrate up to 20 wt% NaCl

Evaporation or Mechanical
Final concentration Concentrate up to NaCl saturation
Vapor Compression (MVC)

Chemical Softening Secondary purification Remove Ca, Mg to ppm-level

Ion-Exchange (IX) Final purification Remove Ca, Mg to ppb-level

Remove free chlorine produced in the


Dechlorinator Post-treatment
electrolyzer

Brine acidifier

Brine acidifier is the pretreatment step immediately before the membrane electrolyzer. As ion-exchange
works best at elevated pH,2 the acidifying should be done only after the IX step. The acidification by adding HCl
increases the anode current efficiency of the cell2 and decomposes carbonate and hydroxide left in the chemical
softening stage. The optimal pH after the acidification is 2-5.2

The acidifier is modeled by two Mixers. HCl is dosed to the brine to decompose the carbonate and hydroxide
left in the chemical softening stage in the first Mixer. The amount of acid is set to be stoichiometric. In the
second Mixer (implemented in the electrolyzer model), the acid is used to adjust the brine pH. The amount of
HCl is set by a Design Spec to obtain the desired pH.

Ion-exchange

Cation ion-exchange (IX) is the final purification step before the electrolyzer and is used to remove Ca2+/Mg2+
as well as other possible cation impurities from the feed brine. The ion-exchange resin for brine purification
must specifically be highly selective to Ca2+/Mg2+ ions so that the desired Na+ ions in the brine are not removed.
In this study, the resins are modeled with the properties of the resin AmberliteTM IRC747 from Dow.8

The basic reaction for IX columns is:2

M 2+ + 2R− Na+ �⎯⎯� (R− )2 M 2+ + 2Na+ (S-R1)

where M 2+ can be any divalent ions (Ca, Mg, Sr, Ba, etc.). The species R− denotes the polymer bone of the
resin with a negative charge. The resin needs to be regenerated back to full capacity after it becomes (near-
)saturated with impurities. HCl and NaOH are consumed to regenerate the resin.2,8

S-16
The IX column is modeled as a continuous process like other components in Aspen. Resin and regenerant (HCl,
NaOH) amount needed are estimated based on technical datasheets from the supplier.8 Following assumptions
are made:

• Ion-exchange reactions have 100% conversion. This leads to a zero-hardness level in the electrolyzer
feed. In the real system, there might be a ppb-level of hardness leakage. However, this difference
does not affect modeling results in the electrolysis cell.

• The regeneration of resin is complete.

• Temperature of brine does not change in IX stage.

Chemical softening

The chemical softening stage includes dosage of soda ash (Na2CO3) and caustic soda. Two major precipitation
reactions are present for the removal Ca2+ and Mg2+:

Ca2+ + CO2−
3 �⎯� CaCO3 (s) (S-R2)

Mg 2+ + 2OH − �⎯� Mg(OH)2 (s) (S-R3)

The model in Aspen uses two separate Mixers and a CFuge (Filter) in Aspen. Typically, a certain over-dosage of
the chemicals is necessary to achieve a low remaining concentration of both Ca2+ and Mg2+. We apply the values
given by reference2, which are 0.6 g/L NaOH and 0.3 g/L Na2CO3. In Aspen, mass flows of both precipitants are
controlled by two design specs (DS), varying concentration of carbonate and hydroxide in the effluent of the
chemical softeners. The precipitate is separated from the brine by a perfect filtration.

Evaporation

A steam-driven evaporator is one of the possibilities to concentrate the ED concentrate to saturation. The
evaporator is modeled by using a Heater and a Flash2 in Aspen. In the Heater, the brine is isobarically heated to
a vapor-liquid-mixture whereas in the Flash2, both phases are separated adiabatically and isobarically. A DS
adjusts the vapor fraction at the outlet of the Heater to achieve the desired brine outlet concentration (set as 27
wt%).

A preheater (as HeatX, two-stream heat exchanger) is implemented to preheat the ED concentrate before
mixing it with the recycle stream. The hot side of the preheater is the hot brine at the outlet of the Evaporator.
The output brine is cooled to 60 °C, temperature for chemical softening.

S-17
Mechanical vapor compression

MVC is the alternative for the final concentration step.

The MVC process is modeled by a Preheater (MHeatX), a boiler (HeatX and Flash2), a compressor (Compr), a
Pump and a Valve. The feed stream is throttled in the Valve to a pressure of 0.3 bar, then preheated (to 65 °C)
and sent to the boiler. A DS (Design Spec) controls the vapor fraction after the boiler so that the desired brine
concentration is achieved (27 wt%). The vapor-liquid-mixture is separated by the adiabatic Flash2 and the vapor
gets compressed in the Compr. The compressed vapor is used to heat the boiler. The condensed vapor (hot
water) and the hot brine are both put in the hot side of the preheater. The brine is cooled to a fixed 60 °C for the
chemical softening stage. A DS controls the compression ratio in the compressor to ensure that the hot water is
cooled exactly to 60 °C in the preheater. The cooled brine is brought back to ambient pressure by a Pump block
and fed in the chemical softening stage.

Electrodialysis

The eletrodialysis process (ED) is modeled as a black-box model in Aspen. The model was an improved
adaptation of a blackbox ED model developed by Nayar et al.,9 using the same equations for calculating energy
consumption and costs as Nayar et al.9 However, it is an improvement over Nayar et al.’s blackbox model in that
the effect of water transport across the ED membrane was also captured without the need for discretizing the
stack in to smaller computational units. The Aspen model was further fine-tuned to approximate the results
from a more sophisticated ED model developed by Nayar et al.10 that simulated the performance across the
length of an ED stack.

The blackbox Aspen model assumed that there’s only transfer of monovalent ions and pure water across the
ED membrane. The effect of pure water transport was captured by fixing the ratio of mass flow rate between
the concentrate outlet and inlet to be 4:1. This ratio was based on the simulation results from Nayar et al.10
which accounted for water transport across the membrane and was in alignment with information we gathered
from the industry.

The energy consumption is calculated from the amount of salt transported by the current while accounting for
a current utilization factor of 0.7 and a voltage of 0.3 V, similar to Nayar et al..9

𝑛𝑛̇ Na+,ED
𝑊𝑊̇ED = 𝑈𝑈ED ⋅ 𝐼𝐼ED = 𝑈𝑈ED ⋅ 𝐹𝐹 ⋅ (S-10)
𝜂𝜂ED

S-18
where 𝑈𝑈ED = 0.3 V is the voltage of the ED stack, 𝜂𝜂ED = 0.7 the current utilization factor, which describes
how much the total current effectively contributes to ion transport through ED membrane. 𝑛𝑛̇ Na+,ED is the molar
amount of sodium ions transported through ED membrane.

Nanofiltration
The major role of nanofiltration is the removal of sulfate ions. In addition, NF removes a fraction of hardness
ions (Ca2+ and Mg2+). However, NF also partially removes Na+ and Cl– ions.

ROSA 9 from Dow11 is used to model the NF system. The NF 270-400 membrane12 is chosen due to its low NaCl
rejection and high active surface area. Based on the permeate flow needed for the ED process, the feed flow is
calculated based on minimum retentate flow for the given membrane unit (provided by ROSA 9). The water
recovery can then be determined as the ratio permeate/feed. A constant density is assumed during the
calculation.

ROSA 9 calculates the required feed pressure for the separation. Additionally, rejections of individual ions are
obtained from the software, which are defined by:

𝑐𝑐i,P
𝑅𝑅i = ⋅ 100 % (S-11)
𝑐𝑐i,F

Where 𝑅𝑅i is the rejection of the ion i, and 𝑐𝑐i,P and 𝑐𝑐i,F refer to the concentration of the ion i in the permeate
and the feed, respectively. The rejection of Br– is not calculated by ROSA. Instead, it is assumed that Br– behaves
like Cl–. The rejections as well as the water recovery ratio are implemented in Aspen as various DS, controlling a
Sep block which represents the NF membrane.

Dechlorination

Dechlorination is used to remove free chlorine from the depleted brine produced by the membrane
electrolyzer. Its first step includes adding HCl to shift the chlorine equilibrium (Figure S1) towards gaseous
chlorine (As the chemical equilibrium shown in Figure S1 produces protons, addition of acids suppresses the
chlorine dissolution while addition of bases enhances it.) and separating the chlorine gas, e.g., with air
stripping.2 The second step is adding reducing agents (i.e., sodium bisulfite NaHSO3) to remove the remaining
chlorine. The reaction can be written as:

HSO− − + −
3 + Cl2 + H2 O �⎯⎯� 2Cl + 2H + HSO4 (S-R4)

S-19
In the Aspen modeling, dechlorination units are simplified to two Mixers, one with HCl and the other with
NaHSO3. A local chemical reaction system is implemented including all the global reactions (salt dissociation,
precipitation) and the new redox reaction (S-R4). Aspen calculates the chemical equilibrium based on all the
reactions in the local system.

The HCl amount is controlled by a DS setting the pH of the mixture. The NaHSO3 amount is adjusted to the
stoichiometric amount of chlorine after the primary dechlorination. No over-dose is needed here due to the
complete redox reaction.

Others

• Purge: The purge is implemented by a FSplit component which separates the stream to purge and
recycled stream based on the user input split ratio.

• Brine heater: A Heater is used to heat the brine to the desired 88 °C feed temperature of the
membrane electrolyzer.

• Brine pump: A Pump is used to bring the brine from ambient (1 bar) to the anode side pressure in the
membrane electrolyzer (1.09 bar).

Summary of modeling parameters


Table S6 lists modeling parameters not included in the main text.

Table S6. Summary of detailed modeling parameters of the final system-level process not shown in the main
text.

Component Parameter Value Unit Source/Rationale

Mass flux increase in C


Eletrodialysis 4 - SI section “Electrodialysis”
channel
(C: concentrate, Voltage 0.3 V SI section “Electrodialysis”
D: diluate)
Current efficiency 70 % SI section “Electrodialysis”

NaOH concentration 250 g/L Reference13


Chemical
NaOH over-dosage 0.3 g/L Reference2
Softening
Na2CO3 concentration 37 g/L Reference13

S-20
Na2CO3 over-dosage 0.6 g/L Reference2

HCl amount (as dry) 9.25 g/h Reference8 + Calculation/Estimation

HCl concentration 6 wt% Reference8


IX
NaOH amount (as dry) 8.22 g/h Reference8 + Calculation/Estimation

NaOH concentration 4 wt% Reference8

HCl concentration 37 wt% Commercial concentrated HCl


Acidifier
HCl amount 1.356 kg/h Acidify feed brine to a pH of 3

NaHSO3 amount 4.32 kg/h Stoichiometry to free chlorine


Dechlorination
NaHSO3 concentration 38 wt% Common strength2

Summary of process outputs


Detailed operating conditions (mass flow rate, temperature, salt mass fraction) at all stages of the process are
given here, which will be helpful for reproducibility of this study. Table S7 summarizes the flows in each stage of
the system-level process and some of their properties.

Table S7. Summary flows of the final system-level process.

Stream Mass flow Temperature Solution concentration

[kg/h] [°C] [wt%]

NF feed 17500 25 °C 7.0

NF retentate 12198 26 °C 7.5

NF permeate = ED feed 5302 26 °C 5.7

ED diluate 4580 26 °C 3.5

ED concentrate 722 26 °C 20.0

Recycle stream 265 90 °C 19.3

Evaporator/MVC feed 987 76 °C (after preheater) 19.9

Evaporator/MVC vapor 258 108 °C Pure water

Evaporator/MVC brine 729 60 °C 27.0


= Chemical softening feed

S-21
Chemical softening NaOH 9.09 25 °C 20.5

Chemical softening Na2CO3 16.2 25 °C 3.6

Precipitate 0.733 59 °C Solid mixture

Chemical softening out 753 59 °C 26.2


= IX feed

IX HCl 0.154 25 °C 6

IX NaOH 0.206 25 °C 4

IX out = Acidifier feed 753 59 °C 26.2

Acidifier HCl 1.36 25 °C 37

Acidifier out 755 88 °C 26.2


=Electrolyzer feed

Electrolyzer Cl2 56.4 88 °C As pure gas

Electrolyzer H2 1.79 88 °C As pure gas

Electrolyzer NaOH 208 88 °C 32

Electrolyzer depleted brine 528 88 °C 19.0


= Dechlorination feed

Dechlorination NaHSO3 4.32 25 °C 38

Purge 265 90 °C 19.3

The scalability of the model allows the application of other NaOH production rates.

Least Work
Energy use is a dominant cost and environmental concern for brine concentration and chlor-alkali electrolysis.
For improving energy efficiency, the energy used must be compared to the thermodynamic least work, i.e., the
minimum energy possible for a given process.

The least work for an individual component in the system is given by:14,15

S-22
𝑊𝑊̇least = Ξ̇out − Ξ̇in = � 𝑚𝑚̇(ℎ − 𝑇𝑇∞ 𝑠𝑠) − � 𝑚𝑚̇(ℎ − 𝑇𝑇∞ 𝑠𝑠) (S-12)
out in

where Ξ̇out and Ξ̇in are the exergy (available work) of the streams passing through the outlet and the inlet, 𝑇𝑇∞
refers to environmental temperature (298.15 K), and ℎ and 𝑠𝑠 are the mass-specific enthalpy and entropy of a
stream than can be obtained from the Aspen model.

References
(1) Aspen Technology Inc, Aspen plus: Aspen plus user guide. Technical Report Aspen Technology Inc., 2000.

(2) O’Brien, T. F.; Bommaraju, T. V.; Hine, F. Handbook of Chlor-Alkali Technology. Springer, 2005.

(3) Byrne, P.; Bosander, P.; Parhammar, O.; Fontes, E. A primary, secondary and pseudo-tertiary
mathematical model of a chlor-alkali membrane cell. J. Appl. Electrochem. 2000, 30 (12), 1361–1367.

(4) McCluney, S.; Van Zee, J. An optimization analysis of a diaphragm cell/evaporator system for NaOH
production. J. Electrochem. Soc. 1989, 136 (9), 2556–2564.

(5) Jalali, A.; Mohammadi, F.; Ashrafizadeh, S. Effects of process conditions on cell voltage, current
efficiency and voltage balance of a chlor-alkali membrane cell. Desalination 2009, 237 (1-3), 126–139.

(6) Leah, R.; Brandon, N.; Vesovic, V.; Kelsall, G. Numerical modeling of the mass transport and chemistry of
a simplified membrane-divided chlor-alkali reactor. J. Electrochem. Soc. 2000, 147 (11), 4173–4183.

(7) Bommaraju, T. V.; Lüke, B.; O’Brien, T. F.; Blackburn, M. C. Chlorine. Kirk-Othmer encyclopedia of
chemical technology, 2004.

(8) Dow Water & Process Solutions, AmberliteTM selective resins for brine purification in the chlor-alkali
industry, Technical Report Dow Water & Process Solutions.

(9) Nayar, K. G.; Fernandes, J.; McGovern, R. K.; Dominguez, K. P.; Al-Anzi, B., Lienhard, J. H. Costs and
energy needs of RO-ED hybrid systems for zero brine discharge seawater desalination. Int. Desalin. Assoc. World
Congr. 2017, Sao Paulo, Brazil.

S-23
(10) Nayar, K.; McGovern, R.; Fernandes, J.; Al-Anzi, B.; Lienhard, J. H. On the costs and energy benefits of
hybridizing RO and ED systems for salt production. Desalination 2017, in Manuscript.

(11) Dow Water & Process Solutions, Design Software: ROSA (Reverse Osmosis System Analysis).
http://www.dow.com/en-us/water-and-process-solutions/resources/design-software (revised on 2017.08.29).

(12) DOW FILMTEC Membranes, Product data sheet of DOW FILMTECTM NF270-400/34i element. Technical
Report DOW FILMTEC Membranes, 2015.

(13) Garriga, S. C. Valorization of brines in the chlor-alkali industry. Integration of precipitation and
membrane processes. PhD thesis, Universitat Politècnica de Catalunya, 2011.

(14) Lienhard, J. H.; Mistry, K. H.; Sharqawy, M. H.; Thiel, G. P. Thermodynamics, Exergy, and Energy
Efficiency in Desalination Systems. In Desalination Sustainability: A Technical, Socioeconomic, and Environmental
Approach, Chpt. 4, H. A. Arafat, ed. Elsevier Publishing Co., 2017.

(15) Mistry, K. H.; McGovern, R. K.; Thiel, G. P.; Summers, E. K.; Zubair, S. M.; Lienhard, J. H. Entropy
generation analysis of desalination technologies. Entropy 2011, 13 (10), 1829–1864.

S-24

You might also like