You are on page 1of 12

Journal of Membrane Science 210 (2002) 91102

Desalination of whey by electrodialysis and ion exchange resins:


analysis of both processes with regard to sustainability by
calculating their cumulative energy demand
Michael Greiter a,b , Senad Novalin a, , Martin Wendland b ,
Klaus-Dieter Kulbe a , Johann Fischer b
a Institute of Food Technology, University of Agricultural Sciences, Muthgasse 18, A-1190 Wien, Austria
b Institute of Agricultural, Environmental and Energy Engineering, University of Agricultural Sciences, Muthgasse 107, A-1190 Wien, Austria
Received 16 May 2002; received in revised form 8 December 2002; accepted 8 December 2002

Abstract
Desalination of whey is an important process in food technology and can be done either by electrodialysis (ED) or by ion
exchange (IE) resins. The aim of the present paper is an assessment of the cumulative energy demand of both processes for a
technical plant including production of the regeneration agents for the ion exchangers and treatment of waste water. A flow
of 45 m3 per day of nanofiltered whey, which is three times concentrated and partially desalted, is assumed as feed for the
plant. Data for the assessment was obtained from own laboratory experiments to be reported here, as well as from commercial
plants. Applying IE, the final whey product is desalted to a degree of 99%. The IE process yields 3.7 m3 of waste water with
36.3 kg ash and an organic charge of 26 kg chemical oxygen demand (COD)/m3 feed whey. The energy demand of IE amounts
to 0.15 kWh for pumping, 25.33 kWh for the production of the regenerants, and 9.75 kWh for the reduction of the organic
charge/m3 whey. With ED, the final whey product is desalted to only 90%. ED yields 1.25 m3 of waste water with 8.1 kg ash
and an organic charge of 8.4 kg COD/m3 whey. The energy demand of ED amounts to 4.2 kWh for pumping, 5.38 kWh for
the electric current through the ED cells, and 3.16 kWh for the reduction of the organic charge/m3 whey.
2002 Elsevier Science B.V. All rights reserved.
Keywords: Ion exchange; Electrodialysis; Desalination; Sustainability; Energy demand

1. Introduction dialysis (ED). These processes differ in their mode


and possibilities of operation.
In technical processes, mineral salts and other ions If competing technical processes are available for
often have to be removed from solutions. Examples a given task in industry, the selection of a certain
are the demineralisation of sea water, boiler feed water process is in essence done on the basis of current
and aqueous solutions in the food and chemical economic reasons, whilst sustainability still plays a
industry. Typical processes are reverse osmosis (RO), minor role for many industries. Sustainability implies,
nanofiltration (NF), ion exchange (IE), and electro- in brief, small cumulative energy demands and ma-
terial cycles which are closed in themselves as far as
Corresponding author. Tel.: +43-1-36006-6288; possible. We expect, however, that the requirements
fax: +43-1-36006-6251. for the sustainability of processes will increase greatly
E-mail address: nov@edv2.boku.ac.at (S. Novalin). and will have a strong impact on production costs in

0376-7388/02/$ see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 6 - 7 3 8 8 ( 0 2 ) 0 0 3 7 8 - 2
92 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

the future. Hence, investigations on the sustainability nanofiltered, three times concentrated whey per day,
of competing processes have to be performed. was done.
The processes to be investigated here with respect For the theory of IE and IE membranes, the reader
to their sustainability come from the food industry and is referred to Helfferich [6] and Dorfner [7]. The fun-
concern the desalination of whey. In the EU, several damentals of ED are discussed, e.g. by Strathmann
million tonnes of whey originate from cheese produc- [8] and Rautenbach [9]. Applications of IE, ED and
tion per year. Desalination is necessary for further use NF in the case of demineralisation of whey are found
of whey as a raw material for other valuable products elsewhere [25,10,11].
such as baby food or for the recovery of proteins.
Processes which can be used for the desalination of
whey are ion exchange (IE) or electrodialysis (ED). 2. Materials and methods
These processes have already been investigated for
more than 20 years (e.g. [15]), mainly with regard 2.1. Whey
to the economic framework of that time. From the
point of view of sustainability, both processes require, Nanofiltered, three times concentrated sweet whey
however, quite large cumulative energy demands and with a total solid (TS) content of 1820% (here and
produce large amounts of waste water, which have in following: % denotes wt.%), coming from a dairy,
not yet been fully assessed. was used. During the nanofiltration the ash content
The cumulative energy demand comprises the is reduced by 30% and more [12]. In particular, the
energy demand of the whole process chain. Thus, for monovalent ions (Cl , K+ and Na+ ) are removed
its determination, one also needs to take into account, to approximately 70%. Table 1 shows the difference
besides the energy consumption of the ED and IE between nanofiltered three times concentrated whey
processes themselves, the energy consumption of the and whey which was three times concentrated by
production of regenerants in the case of IE and of the evaporation. The subsequent desalination by ED or
waste-water treatment of both processes. Therefore, it IE in our laboratory experiments was performed in
is necessary to have a reliable data base of the process batch runs using 1.5 l of whey per run.
parameters, the direct process energy, the regener-
ant consumption and the waste-water amounts and 2.2. Electrodialysis (ED)
contents for both processes. Data has been collected
from literature and industrial plants but did not prove ED is an electrochemical separation process in
sufficient, especially concerning the waste-water pro- which ion permeable membranes are used to remove
duction and contents. ions from solution [68]. Membrane cells are formed,
Hence, experiments were carried out with an IE with each cell bounded by a cationic permeable mem-
and an ED laboratory scale unit in order to validate
and extend existing sources. Data concerning the level
Table 1
of desalination, the amount of regenerants, the energy
Content of ions in a three times concentrated sweet whey (TS
needed and the amount and contents of the waste 1820%) after evaporation (Chaveron [13]) and after nanofiltration
water was collected. In the case of IE, all these data (own values)
could be compared with data from an industrial plant.
Ion Concentration (mg/l)
In the case of ED, a rough comparison to an industrial
plant, which demineralises to a level of 50%, was Vaporised whey NF whey
done. A critical comparison of our experimental data, Na+ 1120 725
literature data and the data of the industrial plants K+ 4490 2590
yields a reliable and sufficiently complete and con- Ca2+ 900 895
Mg2+ 220 225
sistent data set for an ED and an IE process which is
Cl 3400 835
presented here and can be used in an estimation of the PO4 3500 2935
cumulative energy demand. Finally, a scale-up of this
Sum 13630 8200
data to a plant with the capacity to desalinate 45 m3
M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102 93

brane on one side and an anionic permeable mem- membranes next to the anode. Every membrane has
brane on the other. The cells of the diluate (whey) and a total surface of 0.0078 m2 (0.06 m 0.13 m) and
concentrate (brine) solution are arranged alternately. an effective membrane surface of 0.0037 m2 . Hence,
Under the influence of an electric field provided by the total effective membrane surface is 0.0296 m2
dc voltage, cations of the whey migrate towards and for each membrane type. The ion exchange mem-
through the cation exchange membranes and into the branes Neosepta CMX and AMX are produced by
concentrate solution. Similarly, the anions of the whey the Tokuyama Corporation (Tokyo, Japan). The dis-
migrate towards and through the anion exchange tance between the membranes is 0.5 103 m. The
membranes and into the concentrate solution. As a flow velocities were about 0.08 m/s. Delbeke [14] and
result, the concentration of cations and anions in the Hiraoka et al. [15] recommended a flow velocity of
whey falls while their concentration in the concen- about 0.1 m/s for whey concentrated by evaporation.
trate solutions rises. To obtain the desalination level The temperature of the solutions was between 24
of 90%, it is necessary to recycle the whey through and 28 C. By diluting the concentrate solution, the
the stack a number of times. value of conductivity of approximately 10 mS/cm is
Fig. 1 shows the ED laboratory unit schematically maintained constant during the whole process. This
and the arrangement of the different measuring in- conductivity approximately corresponds to the data
struments. The stack used during the experiments has of a commercial plant.
eight cell pairs and two additional cation exchange First, the limiting current density (LCD) was
determined by the method of Cowan and Brown [16],
yielding a value of 17 mA/cm2 . The LCD [8] is the cur-
rent density at which the ion concentration at the sur-
faces of the cation and/or anion exchange membranes
in the cells with the depleted solution will become
zero. The LCD is proportional to the ion concentration
in the diluate and the mass transfer coefficient, which
is determined mainly by the cell geometry and the feed
solution flow velocity. If in ED the LCD is exceeded,
the process efficiency will be drastically diminished
because of the increasing electrical resistance of the
solution and because of water splitting, which leads
to both pH changes and additional operational prob-
lems. To avoid these problems, the current density
was adjusted by the voltage to 13.5 mA/cm2 at the
beginning of a batch run, which corresponds to about
80% of the LCD and yields a total current of 0.5 A.
Thereafter, the voltage was kept constant. The level
of desalination can be estimated by the decrease in
conductivity.
During the run, the following parameters were reg-
istered every 10 min: the current (I), the inner voltage
(Ui ), the outer voltage (Uo ), the temperature (T), the
conductivity (C) and pH-value (pH) of the diluate
and concentrate solution. The values of the flux of
all three solutions and the conductivity and pH-value
of the electrode rinse solution were measured at
the beginning and at the end of the run. The pres-
sure is more or less constant with 0.05 MPa excess
Fig. 1. Set-up of the laboratory scale ED unit. pressure.
94 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

As is common in industrial practice, a cleaning


procedure with the following steps was done after four
runs of 5 h each:
1. water rinsing (3 100 ml, 10 min each);
2. acid washing (HNO3 0.5% solution; 200 ml,
30 min);
3. water rinsing (3 100 ml, 10 min each);
4. alkaline washing (NaOH 0.5% solution; 200 ml,
30 min);
5. water rinsing (1 200 ml and 3 100 ml, 10 min
each).
The procedure took a total of 2 h and 40 min. The
waste water of the electrodialysis process consists of
the total volume of the cleaning solutions (1500 ml),
plus the volume of the concentrate solutions of the
four corresponding runs. Besides the volume of the
waste water, its COD-value and its salt content were
also analysed.

2.3. Ion exchange (IE)

IE is a fixed-bed separation technology using ion


exchange resins [6,7]. These resins have a large num-
ber of firmly attached bonds on their surfaces, which
can reversibly absorb one type of ions. The physical
form of the resins is a small sphere with a diameter
between 0.4 and 0.8 mm. A cation exchange resin
captures positively charged ions and an anion ex-
change resin negatively charged ions. In fresh resin,
the cations attached to the cationic resin are H+ , those Fig. 2. Set-up of the laboratory scale IE unit.
attached to the anionic resin OH . Passing raw whey
over a bed of cation exchange resin would result in The laboratory scale IE unit shown in Fig. 2 has
exchange of the cations in the whey against H+ ions. two columns. One column is filled with the strong acid
Subsequent passage of this whey through the anion cation exchange resin Lewatit S1428 and the other
exchange resin would absorb anions, displacing the with the weak alkaline anion exchange resin S3428,
OH ions. As a result, desalted whey exits the anion both available from Bayer AG (Leverkusen/Germany).
exchange resin column. The columns from Kronlab (Sinsheim/Germany) have
The capacity of exchangeable ions is limited. The a height of 1 m and an inner diameter of 0.025 m. The
amount of whey that can be desalted by the given regenerants are HCl (5% solution) and NaOH (4%
quantities of resins can be calculated. The same ap- solution).
plies to the required amount of water and regenerants In accordance to data from a commercial ion ex-
(acid in the case of cation exchange resin, alkali in change plant, a capacity of 3.75 l whey/l cation
the case of anion exchange resin). These solutions exchange resin and a capacity of 4.15 l whey/l anion
are sufficiently concentrated to remove the absorbed exchange resin can be expected. For regeneration,
cations and anions and replace them by H+ and 65 g HCl (pure)/l cation exchange resin and 70 g
OH ions, regenerating the resins to their previous NaOH (pure)/l anion exchange resin can be expected.
state. This data was used to estimate the amounts of whey
M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102 95

treated in and regenerants necessary for the labora- columns are collected and represent the waste water.
tory scale IE unit. The data is in good agreement with The COD-value, the salt content and volume of the
Delaneys data [2]. The bed volume of the resins in waste water were analysed. It is possible to control
both columns of the laboratory scale unit was 0.4 l. the procedure by measuring the conductivity and the
The following desalination and regeneration proce- pH-value of the solutions after passing the column.
dure was optimised in this work regarding the min- Fig. 2 also shows the arrangement of the measuring
imisation of the waste-water amount and contents. instruments.
The procedure is the same for both columns, but the
amount of the regenerants and washing water differs. 2.4. Analysis of the salt content
1. The whey is desalted at a flow rate of 3 m/h.
In the beginning and at the end of a run, samples
2. The column is rinsed by water obtained as reversed
were taken from the original NF whey and the de-
osmosis (RO) permeate. Its volume is 1 bed vol-
salted whey in the case of IE and from the diluate and
ume, the flow rate 3 m/h. After half of the volume
concentrate solution in the case of ED. The collected
has passed the column, the procedure is stopped.
samples were deep-frozen (30 C) and analysed
3. Up-flow rinsing with RO permeate, its volume is
later. To remove proteins, the samples were boiled
1.5 bed volumes and the flow rate 15 m/h.
for 5 min and centrifuged (12 000 rpm) before injec-
4. Up-flow regeneration: cation exchange column
tion. An ion chromatograph Dionex DX-120 (Dionex,
1.275 l HCl (5%)/l resin, anion exchange column
USA) was used for the analysis, following a standard
1.705 l NaOH (4%)/l resin; flow rate: 5 m/h.
procedure. Table 2 shows the apparatus and materi-
5. Up-flow rinsing with RO permeate: cation exch-
als. The analysis is evaluated by the Peaknet Control
ange column with 2.5 bed volumes, anion exchange
software (Dionex, USA).
column with 4 bed volumes; flow rate: 5 m/h.
The average flow rate is 4.2 m/h and in good 2.5. Analysis of the COD-value
agreement with the value of the commercial plant,
which is approximately 4 m/h. The rinsing water The collected waste-water samples were deep-
(ref. point 2) is added to the desalted whey. Apart frozen (30 C) and analysed later. In the case of ED,
from the diluted whey, all other solutions passing the there are two types of waste water: the concentrate

Table 2
Apparatus and materials of the ion and COD analysis
Apparatus/materials Type/working temperature/flow rate Company

Ion chromatograph Dionex DX-120 Dionex (USA)


Cation column Dionex IonPac CS-12A; 25 C Dionex (USA)
Anion column Dionex IonPac AS-14; 25 C Dionex (USA)
Cation suppressor Dionex Suppressor CSRS Dionex (USA)
Anion suppressor Dionex Suppressor ASRS Dionex (USA)
Conductivity detector Dionex Detection Stabiliser DS4-1 Dionex (USA)
Cation elution solution Methanesulfonic acid (puriss.); 1.2 ml/min Fluka (Germany)
Anion elution solution Sodium hydrogen carbonate (p.a.)/sodium Merck (Germany)
carbonate anhydrous (p.a.); 1.2 ml/min
Standard solutions (Na, K, Ca, Mg, Certipur (1000 mg/l) Merck (Germany)
NH4 , Cl, PO4 , SO4 )
COD analyzer Spectroquant Nova 60 Merck (Germany)
Test kit COD cell test 14540 (10150 mg/l) Merck (Germany)
Test kit COD cell test 14541 (1001500 mg/l) Merck (Germany)
Test kit COD cell test 14691 (3003500 mg/l) Merck (Germany)
Thermoreactor Thermoreactor TR 200; 148 C Merck (Germany)
96 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

solution and the waste water of the cleaning proce- an arrow, represents the LCD. This value of 0.626 A
dure. Each of these was analysed separately. In the and the effective membrane surface (0.0037 m2 ) yield
case of IE, there are three solutions which were anal- the LCD of approximately 17 mA/cm2 . Eighty per-
ysed: the waste water after the cation desalination step, cent of this value are the 13.5 mA/cm2 which were
the waste water after the anion desalination step, and adjusted by the voltage at the beginning of the runs.
the final waste water after mixing both. Table 2 lists Kessler [17] recommended that the value of the cur-
the apparatus and the test kits, which were used in rent density should be between 10 and 15 mA/cm2 .
the COD analysis. The evaluation of the test kits was In contrast to IE, ED cannot desalt up to 99%. In
done photometrically with the Spectroquant Nova 60 this work, a 90% desalination degree was obtained.
(Merck, Germany). The runs lasted between 4 and 4.5 h each. To con-
trol the level of desalination, the conductivity of the
diluate solution was measured online. Fig. 4 shows
3. Experimental results the correlation between the decrease in conductivity
and the decrease of the concentration of ions (molar
3.1. Electrodialysis equivalent) of a trial during which samples were taken
after 0, 60, 120, 180 and 270 min.
Fig. 3 shows the estimation of the LCD by the The waste water of ED is composed of the waste wa-
method of Cowan and Brown [16] from our experi- ter of the cleaning procedure and the concentrate solu-
mental results. In order to maintain constant process tion, which has a larger volume, a higher COD-value
parameters, particularly equal concentrations, only and a higher salt content. Table 3 gives an overview of
one tank was used for both diluate and concentrate the waste-water composition. It has to be mentioned,
solutions. On the end-membranes (cationic exchange that the cleaning procedure is conducted after four runs
membranes), two platinum wires were placed to mea- and thus the amount of waste water from cleaning had
sure the potential (Ui in Fig. 1) across the membranes to be divided among the runs proportionately.
excluding the electrodes. The voltage was increased by The energy demand of ED is composed of the en-
defined steps and the corresponding current (I) noted. ergy demand of the separation of ions and the energy
The electrical membrane resistance (Rm ) was calcu- demand of the pumps, which was not analysed for the
lated by the relation Rm = Ui /I and is plotted against laboratory scale unit because of the unrealistic ratio
the reciprocal value of the current. The data can be between the size of the stack and that of the pump
represented by two straight lines, which intersect at and pipe system. Concerning the separation of ions,
1.6 A1 . The intersection point, which is marked by energy is used for the transport of ions through the

Fig. 3. Determination of the LCD with the method of Cowan and Brown.
M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102 97

Fig. 4. Decrease of conductivity and corresponding decrease of the molar equivalent in percent referred to the values of the original NF whey.

solutions and the membranes and for the overcoming the theoretical current (Ith ) and the practical current
of the resistance of the electrodes. The resistance of (Ipr ) have to be calculated. The theoretical current
the electrodes was not considered in this case, because was determined from the cation analysis. A total
its influence is too high with regard to the size of of 0.208 molar equivalent charges correspond to the
the stack: the laboratory stack only has 8 cell pairs average value of cations, removed from the whey and
whereas a commercial stack consists of as many as 200 re-found in the concentrate solution. The density of
and more cell pairs. The voltage drop for a cell pair charge flux J = 4.5 104 mol/m2 s results there-
was determined from the inner voltage and amounted from. The theoretical current is calculated with the
to 0.85 V, the resistance per cell pair increases during formula
a run and the measurements gave an average value
of 3.8 . Including the resistance of the two cation Ith = FJAeff , (2)
exchange membranes next to the anode, the energy whereby F is the Faraday constant [96 485.3 As mol1 ]
demand for desalting 1.5 l whey is 26.5 kJ. and Aeff the effective membrane surface [0.0037 m2 ].
To obtain the current efficiency by the formula The practical current is obtained by the measured
Ith charges (3518 As), divided by the time. According to
= , (1)
Ipr formula (1) = 0.7 results.

Table 3
Composition of waste water of ED and IE laboratory scale unit (LU) and IE commercial plant (CP)a
Process Waste water Volume (ml) COD (g) Ash (g) pH

Electrodialysis (LU) Cleaning procedure 250 0.5 0.4


Concentrate solution 1000 6.5 7.7
Whole waste water 1250 7 8.1 6
Ion exchange (LU) Cation desalination 1600 5
Anion desalination 2100 21
Whole waste water 3700 26 36.3 56
Ion exchange (CP) Whole waste water 3400 20
a The values are referred to 1 l original NF whey.
98 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

Fig. 5. The plots represent the performance of the IE cation exchange resin. The pH-value and the conductivity are plotted as a function of
the throughput volumes measured in units of bed volumes. Plot A: down-flow desalination of whey. Plot B: up-flow rinsing and regeneration.

3.2. Ion exchange conductivity mark the breakthrough and the moment
of initiation of desalted whey collection. Finally, the
The cleaning and regeneration procedure was opti- whey displaced by the rinsing solution in the column
mised in this work, as described in Section 2.3. The is diluted with half of the bed volume of rinsing so-
parameters, which are characteristic for the runs, are lution. As a result, the total solids content decreases
the pH-value and the conductivity of the solution leav- from 1819% to about 15% after processing through
ing the column. Fig. 5 shows the down-flow desalting both columns.
step in Plot A for the cation exchange and the up-flow In Plot B, the first rinsing with a flow rate of 15 m/h
rinsing and regeneration step in Plot B. The pH-value should rinse macro molecules on top of the column.
and the conductivity are plotted against the volume of According to Herv [18], rinsing and regeneration are
the solutions, which have passed the column and are made up-flow, above all to economise regenerants use.
measured in bed volumes. Herv points out that in the down-flow system strong
In Plot A, the breakthrough of whey and starting acid cation exchange resins need 300% molar equiva-
point of rinsing are marked with an arrow. The sud- lent HCl for regeneration relative to capacity, whereas
den decrease of pH-value and the sudden increase in weak alkaline anion exchange resins need just 150%
M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102 99

molar equivalent NaOH. He further states that 40% of 4. Scale-up and comparison
regenerants can be economised performing an up-flow
regeneration. In this work, the data of regenerants from The experimental results were used for a scale-up to
a commercial plant were used. The consumption of re- commercial plants with a capacity to desalinate 45 m3
generants corresponds to 200220% of the capacity of nanofiltered, three times concentrated whey per day.
each resin. As mentioned above, the anion exchange An overview of the data, which are calculated in the
resin would not need as much regenerant as the cation next two chapters, is given in Table 4.
exchange resin. Legal regulations, however, demand a
final pH-value of the waste water to be between 6.5 4.1. Electrodialysis
and 9. Mixing the waste water of both regeneration
steps therefore makes almost equal amounts of re- The working day of a typical commercial ED plant
generants necessary to have a neutralising effect. The is composed of four runs (5 h each) and one cleaning
last rinsing is performed to displace the ions, which procedure (4 h). The 45 m3 volume of whey is divided
were exchanged by the regenerants. In Plot B, the into 4 11.25 m3 , which corresponds to 2.25 m3 /h.
large peak in the conductivity shows this effect. This The laboratory scale unit desalted 1.5 l of whey to
peak is larger (wider) in the anion exchange, where 90% with 0.0296 m2 membrane surface in 260 min,
one needs more solution to obtain the same rinsing corresponding to 0.0117 m3 /m2 h. This leads to a
effect. required membrane surface of 192.4 m2 . The total
The composition of the waste water from IE is listed membrane surface is 212 m2 (192.4 m2 + 10% se-
in Table 3. The differences between the COD-value curity). The membranes have an effective surface of
and the waste-water volume of the laboratory scale 0.4 m2 , so 530 cation and 530 anion exchange mem-
unit and commercial plant can be seen. It is also branes have to be used. As a result, there are 530 cell
pointed out that besides 8.8 g ash, which represents pairs, which are split into two stacks.
the exchanged ions of 1 l whey, additional 27.5 g NaCl One stack has 265 cell pairs and the necessary volt-
get into the waste water because of the applied regen- age per cell pair is 0.85 V (see Table 2) and 225.25 V
erants. Table 3 also shows the difference between the per stack. The average current density of the laboratory
volume and COD-value of the waste water from cation runs was 6.095 mA/cm2 or 60.95 A/m2 . The current
and anion desalination. of the commercial plant is then 24.38 A related

Table 4
Overview of data for an ED and IE commercial plant, which have the capacity to desalinate 45 m3 of NF whey per daya
Process Parameter Surface/mass/volume Energy demand (kWh)

Electrodialysis Membrane surface 212 m2


Separation of ions 242
Pumps 189
Reduction of COD 142
Waste water 56 m3
Ash 365 kg
Ion exchange Cation exchange resin 4 m3
Anion exchange resin 3.6 m3
HCl (pure) 780 kg 1140b
NaOH (pure) 840 kg 1140b
Pump 6.6
Reduction of COD 439
Waste water 167 m3
Ash 1634 kg
a This data is the result of a scale-up using experimental and industrial data.
b The value refers to the combined energy demand for the production of HCl and NaOH.
100 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

to 0.4 m2 effective membrane surface. This corre- 15 m3 whey has to be desalted per run. The necessary
sponds to an electric power of 5491.6 W. According to amount of resins is calculated with the data from a
industrial experience, the resistance of the electrodes commercial ion exchange plant and results in 4 m3
is taken into account adding 10% to the calculated cation exchange resin and 3.6 m3 anion exchange
energy demand, so the power increases to 6040.8 W resin.
per stack or 12 081.6 W for both stacks. Because A total of 260 kg pure HCl and 252 kg pure NaOH
the desalination runs 20 h per day, the total energy are needed for regeneration alone. This would re-
demand is 241.6 kWh. sult in a surplus of acid in the waste water. At the
The power of the pumps is calculated by: end of regeneration, the waste water should have
P = 100 p V P 1 (kW) (3) a pH-value between 6.5 and 9. Therefore, almost
equal molar amounts of regenerants have to be used
whereby p is the pressure drop [bar], V the volume to effect conforming neutralisation. So, 780 kg pure
flow (m3 /s), and p the pump efficiency assumed at HCl or 1765 l HCl (37%) and 840 kg pure NaOH or
0.7. In agreement with realistic assumptions, the pres- 1100 l NaOH (50%) are needed for regeneration for
sure drop is supposed to be 3 bar. The volume flow of three runs. Hartz and Hoppe [21] report, that an elec-
diluate and concentrate solutions is 0.016 103 m3 /s trolysis process with membrane technology, which
per cell (corresponding to 0.08 m/s flow velocity is the state-of-the-art technology by Krupp Uhde
and a cross-section of 2 104 m2 ). The concen- (Germany), has an energy demand of 2000 kWh for
trate and the diluate solution pump have to transport the production of 1 tonne NaOH. This value includes
8.48 103 m3 /s each (for two stacks with 256 cells). the co-production of H2 and Cl2 . The synthesis of
Inserting the values in the formula (3), the power of HCl (with 1 mol H2 and 1 mol Cl2 ) has an enthalpy of
each pump is 3.634 kW and the energy demand of a reaction of 184 kJ. This energy, which theoretically
whole day is 87.2 kWh. A volume flow of electrode could be recovered, is taken into account. This yields
rinse solution of 1.38 103 m3 /s is assumed, so the an upper limit for the energy demand of approxi-
power of the pump is 0.595 kW and the energy de- mately 1140 kWh for the production of regenerants
mand for 24 h is 14.3 kWh. As a result, total energy necessary for 1 day.
demand for all three pumps is 188.7 kWh. The volume flow during the desalination of 1.5 h
The COD-value of the laboratory scale ED unit in is composed of the 15 m3 whey and the 2 m3 rins-
Table 3 corresponds to a process without pole reversal. ing water, which is used to displace the whey in the
But in industry, pole reversal is the usual method and columns. During the regeneration of 5.5 h, the volume
works fully automated. The rinsing in between the flow is calculated with 15 m3 whey multiplied with
reversals is very effective, so a low diluting and a high 3.7, which is the ratio of waste water to whey. The
cleaning effect can be assumed. To get realistic values, total volume flow is 2.2 103 m3 /s. In agreement
20% were added to the COD-value of the laboratory with industrial experience, a pressure drop of 1 bar is
scale ED unit. This results in 8.4 g COD/l whey or supposed. Using formula (3), the power is 0.31 kW
378 kg COD/45 m3 whey. A conventional reduction of and the energy demand for 21 h is 6.6 kWh.
1 kg COD has an energy demand between 0.25 and The COD-value of 26 g/l whey (Table 3) yields to
0.5 kWh [19]. A total of 0.375 kWh for 1 kg COD were 1170 kg COD/45 m3 . The energy demand for reduc-
assumed, which results in a total energy demand for tion of this COD amount is 439 kWh. The volume
reduction of 378 kg COD of 142 kWh. of IE waste water is 3.7 l/l whey and 166.5 m3 /45 m3
The volume of ED waste water is 1.25 l/l whey or whey.
56 m3 /45 m3 whey.

4.2. Ion exchange 5. Discussion

The working day of a typical commercial IE plant If the separation process is considered, ED as
is composed of three runs, which last 7 h each (1.5 h an electrochemical process utilises substantially
desalination and 5.5 h regeneration [20]). A total of more energy than IE. With regard to sustainability,
M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102 101

a calculation of the cumulative energy demand is Acknowledgements


recommended. Therefore, the energy analysis was
extended to include the production of the regener- The present work was supported by Forschungss-
ants and the waste-water treatment. The production timulationsprogramm des Rektors, Universitt fr
of regenerants is achieved by electrolysis. This pro- Bodenkultur (University of Agricultural Sciences),
cess spends 2000 kWh of energy for the production Vienna, Austria.
of 1 ton NaOH. In the context of sustainability, one
problem is how to evaluate the possibilities of using
process by-products, for example, in this case hydro- References
gen and chlorine. Chlorine is a raw material in the
[1] D.W. Houldsworth, Demineralization of whey by means of
plastics industry, NaOH being a cheap by-product of
ion exchange and electrodialysis, J. Soc. Dairy Technol. 33
chlorine production. HCl is produced using chlorine (1980) 45.
and the remaining hydrogen has a high energy poten- [2] R.A.M. Delaney, Demineralization of whey, Aust. J. Dairy
tial. In this work the enthalpy of reaction of the HCl Technol. 31 (1976) 12.
synthesis with hydrogen and chlorine is taken into [3] M. Evans, Electrodialysis and ion exchange as deminera-
lisation methods in dairy processing, in: R. Hansen (Ed.),
consideration, because the calculation of the energy
Evaporation, Membrane Filtration and Spray Drying in
demand is thus done directly with the products of the Milk Powder and Cheese Production, North European Dairy
electrolysis and gives an upper limit for the energy Journal, Copenhagen, Denmark, 1985, pp. 5580.
demand if regenerants were taken into account. [4] W. Ulrich, Exemple dutilisation du lactosrum dans
In the above, a key area of study is also the waste lindustrie alimentaire, Schweiz. Milchwirtsch. Forschung 5
(1976) 99.
water and its composition. Obviously IE not only pro-
[5] B.M. Ennis, J.J. Higgins, Demineralization of deproteinated
duces three times more waste water and COD than sweet wheyan economic study, N. Z. J. Dairy Sci. Technol.
ED does, but also 44.5 times more salt (expressed in 16 (1982) 27.
mass). This is of particular interest, as the European [6] F. Helfferich, Ionenaustauscher, Verlag Chemie GmbH,
Union is going to introduce taxes for the waste water Weinheim/Bergstrae, 1959.
[7] K. Dorfner, Ion Exchangers, Walter de Gruyter, New York,
related to the real costs of waste-water treatment by 1991.
2010. The energy demand for the reduction of COD [8] H. Strathmann, Electrodialysis, in: K. Sirkar, W.S. Ho (Eds.),
represents only one part of the total energy demand Membrane Handbook, Van Nostrand Reinhold, New York,
for waste-water treatment. 1992, pp. 219262.
Taking into account [15], the following conclusion [9] R. Rautenbach, Membranverfahren, Springer, Berlin, 1997.
[10] R. de Boer, T. Robbertsen, Electrodialysis and ion exchange
can be made regarding sustainability: NF should be process: the case of milk whey, in: C. Cantarelli, C. Peri
used as a pre-treatment step, as whey is concentrated (Eds.), Progress in Food Engineering, Forster-Verlag AG,
and partly demineralised simultaneously. For further Ksnacht, Switzerland, 1983, pp. 393403.
desalination, ED is to be preferred. First, the cumu- [11] H.C. van der Horst, J.M.K. Timmer, T. Robertson, J. Leenders,
lative energy demand of ED is significantly lower. Use of nanofiltration for concentration and demineralization
in the dairy industry: model for mass transport, J. Membr.
Second, both the effluent volume and the total salt Sci. 104 (1995) 205.
discharge are lower as well. [12] T. Hutson, Nanofiltration and ion exchange for the deminera-
On the other hand, the ammonium bicarbonate pro- lization of whey, in: Proceedings of the 2nd International
cess or SMR process [22,23] is an example that IE Whey Conference on Whey, Chicago, 1997, International
process could be improved by recycling regenerants. Dairy Federation, Brussels, Belgium, 1998, pp. 8892.
[13] M. Chaveron, J.J. Sihver, H. Duperrex, Demineralization of
Furthermore, a desalination of 99% is easily obtained Whey, US Patent 4,138,501 (1979).
with IE. [14] R. Delbeke, La dminralisation par lectrodialyse du
Additional aspects, such as transportation of regen- lactosrum doux de fromagerie, Le Lait 541/542 (1975) 76.
erants, energy demand for the production of resins, [15] Y. Hiraoka, K. Itoh, S. Taneya, Demineralization of cheese
membranes etc. should strictly also be analysed. whey and skimmed milk by electrodialysis with ion exchange
membranes, Milchwissenschaft 34 (7) (1979) 397.
Obviously only first steps were done in the com- [16] D.A. Cowan, J.H. Brown, Effect of turbulence on limiting
parison of these technologies with regard to their current in electrodialysis cells, Ind. Eng. Chem. 51 (12) (1959)
sustainability. 1445.
102 M. Greiter et al. / Journal of Membrane Science 210 (2002) 91102

[17] H.G. Kessler, Lebensmittel-und Bioverfahrenstechnik: International Dairy Federation, Brussels, Belgium, 1998,
Molkereitechnologie, Verlag A. Kessler, Freising, 1996. pp. 9399.
[18] D. Herv, Whey treatment by ion-exchange resins, Proc. [21] H.J. Hartz, C.F. Hoppe, in: Proceedings of the 10th Krupp
Biochem. 9 (2) (1974) 16. Uhde Chlorine Symposium on Krupp Uhdes Progress in
[19] ATV-Handbuch: Biologische und weitergehende Abwa- Chlorine Business, Dortmund, Germany, 68 May 1998.
sserreinigung, Verlag Ernst & Sohn, Berlin, Germany, [22] H.B. Jnsson, Method and Apparatus for Desalination of
1997. Whey, US Patent 4,159,350 (1979).
[20] F. Rousset, P. Reboux, Nanofiltration and ion exchange for [23] H.B. Jnsson, L.-E. Olsson, The SMR processa new
the demineralization of whey, in: Proceedings of the 2nd ion exchange process to demineralize cheese whey,
International Whey Conference on Whey, Chicago, 1997, Milchwissenschaft 36 (1981) 8.

You might also like