The use of regeneration profiles as a tool to optimise the

performance of demineralisation water treatment plants

Casper JH du P Strydom* and C Frederik Schutte

Water Utilisation Division, Department of Chemical Engineering, University of Pretoria, South Africa

Abstract

Demineralised water is used as make-up water to compensate for losses in the condensate-boiler feed-water stream at power stations
and other steam raising plants. Demineralisation plants are designed to produce pure water that contains virtually no impurities.
Ion exchange is invariably used for demineralisation. Effective regeneration of exhausted resin is an important aspect to ensure
optimal performance of the ion exchange process. Regeneration profiles were used in this investigation to determine the
effectiveness of regeneration and to optimise the regeneration process.

Keywords: demineralisation, regeneration profiles, optimised regeneration

Introduction
TABLE 1
The make-up water for modern high-pressure fossil-fuelled boilers Recommended water quality limits for boiler
has to be of extremely high quality with virtually complete removal make-up water
of salts and gases. Impure water may cause failure of a turbine or
boiler that will result in large financial losses amounting to millions Parameter Units Limit
of Rand per day. The make-up water quality depends on the design
and unit operations incorporated in a treatment system. A number Turbidity (NTU) 0.2
of processes may be used in the pretreatment of raw water, but Specific conductivity(25ºC) (µS/cm) 0.1
invariably the final stage is an ion-exchange demineralisation Sodium (as Na+) (µg/kg) 2
process (Modern Power Station Practice, 1992; Drew, 1994). This Silica (as SiO2) (µg/kg) 10
article focuses on the use of regeneration profiles to optimise ion- Chloride (as Cl-) (µg/kg) 2
exchange regeneration and to increase run lengths, thereby reduc- Sulphate (as SO42-) (µg/kg) 2
ing water production costs. TOC (as C) (µg/kg) 300

Boiler make-up quality Source: Eskom, 2001

Boiler make-up water is the product water from a demineralisation

plant that is fed into the boiler drum to compensate for water losses.
The quality of the boiler make-up water determines the boiler trated regenerant solution, restoring the resin bed to the ionic form
operating efficiency (Harfst, 1993). The Eskom chemistry standard that is again useful to the process (Frederick, 1996).
for make-up water for drum boilers operating at 17 MPa and above
is listed in Table 1. Demineralisation consists of the following steps:

Ion-exchange process Cation exchange

The cation exchanger contains two types of cation resins, namely
Ion exchange is a process whereby pretreated water is stripped from strong-acid exchange resins (SAC) and weak-acid exchange resins
unwanted cations and anions to give a product water of specific (WAC). The SAC can split neutral salts, i.e. remove non-carbonate
quality. Ion exchange comprises the reversible exchange of ions hardness, while WAC can only remove carbonate hardness from
between a solid (resin) and a liquid in which there is no substantial the water. On exhaustion the resins are restored to the original state
change in the structure of the solid. At the start of a service cycle by regenerating the WAC downwards with a 0.8 to 1.5% H2SO4
the predominant ionic form of the resins is either the hydrogen solution, while the SAC is regenerated upwards with a 5% H2SO4
(cation) or hydroxyl (anion) form. The hydrogen and hydroxyl ions solution.
are exchanged respectively for unwanted cations and anions in the
feed water. Exhausted resins are regenerated with a highly concen- Degasification
The raw water entering the cation exchangers contains alkalinity
* To whom all correspondence should be addressed. (bicarbonate and carbonate ions) that decomposes into carbon
! +2717 612 6203; fax: +2717 612 6336; dioxide (CO2) and water. Forced draft degasifiers are used to
e-mail: [email protected] remove the CO2 to reduce the load on the anion exchanger (Meyers,
Received 7 February 2003; accepted in revised form 8 July 2003. 1996).

Available on website http://www.wrc.org.za ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 411
 TABLE 2
Regeneration procedure applied at Matla Power Station

Cation unit Anion unit

Step Step description Duration Step Step description Duration

(min) (min)

01 Backwash top 8 01(a) Backwash weak base 8

02 Settle 2 02(a) Drain weak base 15
03 Forced settle 4 01(b) Drain strong base 15
04 Establish flows 3 02(b) Pre-injection strong base
05(a) 0.5/2.0% acid injection - 80 03(a) 4% NaOH injection 50
top and bottom strong base
05(b) 0.8/3.0% acid injection - 20 03(b) NaOH injection strong 30
top and bottom and weak base
06 Rinse top and bottom 80 04 Slow rinse/pump flush 75
07 Downward rinse 30 05 Fast rinse strong and 45
weak base
08 Backwash top 3 06 Drain weak base 15
Refill strong base 14
07 Refill weak base 10
08 Fast rinse weak base 20

Source: Foster Wheeler Delta Cochrane, 1978

Anion exchange • Effluent samples taken and analysed at various points in the
The weak-base exchange resins (WBA) and strong-base exchange process at different time intervals during regeneration.
resins (SBA) are held in two separate exchangers. The strong-base • Analysis results plotted to determine trends.
and weak-base anion exchangers are regenerated in the upward and
downward direction with 4% NaOH at 49°C respectively. The Cation and anion resin regeneration can be profiled separately,
weak-base anion exchanger is regenerated with used regenerant plotting the various parameters analysed as a function of time.
from the strong-base unit.
Study objectives
Mixed bed exchanger
The mixed bed exchanger is filled with SAC and SBA resins that The efficiency of regeneration is generally evaluated in terms of the
“polish” the effluent of the strong-base anion exchanger. The SAC restoration of the exchange capacity of the resins. This routine
and SBA are regenerated with 6% H2SO4 and 4% NaOH at 49°C procedure does not give any specific information about the effi-
respectively. ciency of the individual regeneration steps and inefficiencies may
therefore go undetected. The purpose of this study was to evaluate
Optimisation of regeneration the use of regeneration profiles as a tool to optimise the regenera-
tion process.
The regeneration process can be optimised by analysing concentra-
tion profiles of the different ions and substances in the regenerant Regeneration studies
streams. A regeneration profile is a valuable tool to assess the
efficiency and efficacy of the different regeneration steps and the Background information
run length of the demineraliser. It is a graphical representation
produced by analysing samples taken during regeneration at vari- The investigations were carried out at the demineralisation plant of
ous time intervals. Regeneration profiles are useful as a tool to the Eskom Matla Power Station. The plant consists of three parallel
answer questions such as: treatment trains, each producing a maximum flow of 220 m3/h with
a total hydraulic capacity of 660 m3/h. The regeneration procedure
• Are the backwash steps long enough to remove all colloids? applied at Matla Power Station is illustrated in Table 2 and includes
• Are the correct acid and caustic strengths used in the injection a description of process steps and the time required for each step.
phases? The mixed bed regeneration is excluded from the table because the
• Are the rinse times in line with design values? profile was only done on the cation and anion resin regeneration.
• What is the condition of the ion-exchange resins in use?
Methodology
To develop a regeneration profile the following has to be done:
The following procedure was followed in creating regeneration
• Ion-exchange resin samples evaluated before and after regen- profiles:
eration.

412 ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 Available on website http://www.wrc.org.za
 TABLE 3 TABLE 4
Chemical analysis on the various regeneration Chemical analysis of demineraliser effluent before
processes regeneration of both

Cation regeneration Anion regeneration Parameter Units Cation inlet Cation inlet
2 July 2002 24 July 2002
Acid strength on the top and Caustic strength on injection
bottom acid injection (expressed steps (expressed in Calcium hardness mg/l 14 14
in percentage) percentage) Magnesium hardness mg/l 21 24
Total hardness mg/l 35 38
Acid strength on the effluent Caustic strength on the Sodium mg/l 7.4 8.7
(expressed in percentage) effluent (expressed in Potassium mg/l 2 2.7
percentage) Chloride mg/l 10 10.2
Silica µg/l 350 410
Turbidity (expressed as NTU) Turbidity (expressed as NTU) Sulphate mg/l 7.6 7.5

Electric cond.(expressed in Electric cond.(expressed in

mS/cm) at 25ºC mS/cm) at 25ºC
The cation resin regeneration changes are:
Total hardness (expressed as Sulphate (expressed as SO42-)
For the 1st acid injection step:
mg/l CaCO3)
• The acid strength of 0.5% top injection/2.0% bottom injection
was changed to 0.7% top injection/2.0% bottom injection.
Calcium hardness (expressed as Chloride (expressed as Cl-)
• The injection time was decreased from 80 to 70 min.
mg/l CaCO3)
For the 2nd acid injection step:
Magnesium hardness (expressed Silica (expressed as SiO2)
• The acid strength of 0.8% top injection/3.0% bottom injection
as mg/l CaCO3)
was changed to 0.7% top injection/4.0% bottom injection.
• The injection time was increased from 20 to 30 min.
Sodium (expressed as Na+) Colour (indications for
organics) (expressed as
Downward rinse: The rinsing time was increased from 30 to 35 min.
percentage) – own technique
The anion resin regeneration changes are:
Potassium (expressed as K+)
On the 4% caustic injection to the strong base step:
• The strong base resin caustic injection time was increased from
• The demineralisation process is operated to maximum capac- 50 to 60 min.
ity. In this case the maximum capacity was determined by run- • From the 40th to the 50th min in the injection step additional
length and was set at 12 000 m3. The run-length was based on samples were taken every 2 min for silica analysis.
conductivity and a safety margin was allowed.
• Water samples for the regeneration profiles were taken at On the weak base caustic injection step:
exhaustion of the demineralisation run. • The injection time was increased from 30 to 40 min.
• Samples were taken at predetermined time intervals to carry out
analysis on the different parameters. The time intervals varied Results
from 1 to 2 min at the beginning of each regeneration step to 5
min later in the process. Demineralisation plant chemical analysis
• Chemical analysis was done according to Standard Methods,
1995. Table 4 shows the analyses of the inlet to the cation exchanger
• The analysis results were plotted as individual graphs. before the first regeneration run (2 July 2002) and before the second
• Regeneration profiles were analysed to identify possible short- regeneration run incorporating changes to optimise the process (24
comings in the regeneration processes. July 2002).
• Regeneration procedures were then adjusted to eliminate short- Only cation and anion regeneration profiles were determined
comings. for the most important process steps. Steps 1, 2, and 3 of the cation
regeneration and Steps 1(a), 2(a), and 1(b) of the anion regenera-
The analyses done to determine the profiles are given in Table 3. tion were omitted (refer to Table 2). The profiles started with the
After making changes to the regeneration processes to elimi- H2SO4 injection for the cation and NaOH injection for the anion
nate shortcomings, a second profile was compiled to assess im- because the major changes take place during these steps. The cation
provements to the regeneration processes. Based on the findings and anion regeneration profiles done on 02 July 2002, are given in
from the first regeneration run, the following changes were intro- Figs. 1 to 3 and those done on 24 July 2002 are given in Figs. 4 to
duced to optimise the process steps. The reasons for the changes are 6. Table 3 gives the analysis results. The species loaded onto the
discussed in the section on results and discussion. resins during the service runs and removed during the regenerations
are given in Table 5.

Available on website http://www.wrc.org.za ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 413
 TABLE 5
Species loaded onto and removed from the resins during 1st and 2nd regeneration
First regeneration
Volume treated m3 Tot H* Ca H* Mg H* Sodium Potass Chloride Sulphate Silica
l as
mg/l l as
mg/l las
mg/l l as
mg/l l as
mg/l mg/ll as mg/ll as mg/ll as
CaCO3 CaCO3 CaCO3 Na+ K+ Cl- SO42- SiO2

10704 35 14 21 7.4 2 10 7.6 0.35

kg loaded 374.64 149.86 224.78 79.21 21.41 107.04 81.35 3.75
Removed 333.68 131.48 202.20 70.81 18.92 103.19 65.47 8.73
%removed 89.1 87.7 90.0 89.4 88.4 96.41 80.5 233.1
Second regeneration
Volume treated m3 Tot H* Ca H* Mg H* Sodium Potass Chloride Sulphate Silica
l as
mg/l l as
mg/l las
mg/l l as
mg/l l as
mg/l mg/ll as mg/ll as mg/ll as
CaCO3 CaCO3 CaCO3 Na+ K+ Cl- SO42- SiO2

10340 38 14 24 8.7 2.7 10.2 7.5 0.41

kg loaded 392.92 144.76 248.16 89.96 27.92 105.47 77.55 4.24
Removed 360.24 131.22 229.01 81.12 25.84 94.86 64.76 6.50
% removed 91.68 90.65 92.28 90.18 92.6 89.9 83.5 153.3

* Note: H – Hardness

Discussion Calcium, magnesium, and total hardness removal

Table 5 shows that more calcium and magnesium ions were loaded
Cation regeneration on to the resin during the 2nd service run and also that more ions
were removed during regeneration. This can be ascribed to the
Table 6 shows the top and bottom acid injection strength during the higher average top acid strength for the 2nd regeneration of 0.6%,
two cation regeneration runs. against 0.5% for the 1st regeneration. Figures 1 and 4 show that the
For the 1st regeneration the top acid remained at 0.5% through- top acid injection strength for both regenerations was never ob-
out the 1st and 2nd injection steps. It was supposed to remain at 0.5% tained and was, on average, lower than anticipated. This had the
for the 1st 80 min and then increased to 0.8% for the last 20 min effect of insufficient removal of calcium and mgnesium ions. The
injection step according to prescribed procedures. The bottom acid dotted line on the total hardness graphs illustrates an assumption of
remained at 2.5% for the entire 100 min but was supposed to what the removal would have been if the top acid strengths were
increase from 2% (1st injection, 80 min) to 3% (2nd injection, 20 correct.
min).
For the 2nd regeneration the top acid remained at 0.6% for the Sodium and potassium removal
entire 100 min. It was supposed to remain at 0.7% for the duration The SAC resin removes sodium and potassium ions. From Table 5
of the two injection steps. The bottom acid had an average strength it can be seen that the removal percentage for sodium virtually
of 2.6% during the 1st injection step of 70 min, but should have been remained the same for both regenerations at about 90%. However,
2%. For the 2nd injection step of 30 min the average strength was there was a marked difference in the case of potassium. During the
3.1%, which should have been 4%. For optimisation purposes the second regeneration 92.6% was removed compared to 88.4%
top acid strength was changed from 0.5% (1st injection) and 0.8% during the 1st regeneration. Although the SAC resin has greater
(2nd injection) to 0.7% for the entire 100 min injection time. The preference for sodium ions than for potassium ions, the preference
bottom acid strength was changed from 2% (1st injection) and 3% for sodium release during the service run before potassium could
(2nd injection) to 2% (1st injection) and 4% (2nd injection). The time clearly be seen on the two individual graphs. Unlike calcium and
for the 1st injection was decreased from 80 to 70 min and increased magnesium the bulk of the sodium and potassium ions was not
for the 2nd injection from 20 to 30 min for a total of 100 min injection removed during the acid injection phases but rather during the
time. “Rinse Top and Bottom” step. The reason for this was that the
The reason for these changes was to remove as many of the bottom acid strength was never controlled at the specification
hardness salts exchanged by the WAC resin as quickly as possible values (see Table 6). The 2nd injection bottom acid strength for both
before the rinse steps. The acid strength increases had to be done regenerations on average was lower than what was expected, that
manually during the 2nd regeneration by either turning up the acid is 2.5% against 3% (1st regeneration) and 3.1% against 4% (2nd
dosing pump stroke or reducing the dilution water flow. The regeneration).
existing system could not ensure automatic control of increased
acid concentration. This caused the acid strengths to fluctuate. In Conductivity removal
both regenerations the effluent acid strengths started to increase in In order to reduce the conductivity of the effluent, the time for the
the 55th min of the injection steps, indicating that most of the downward rinse step was increased from 30 (1st regeneration) to
regeneration process was completed. This, however, was only true 35 min (2nd regeneration). During the first regeneration the conduc-
for calcium and magnesium. tivity was 1 950 µS/cm after 30 min, whereas during the second

414 ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 Available on website http://www.wrc.org.za
 Downward
regeneration after 35 min it was 1 550 Acid injection Top & Bottom Rinse Top & Bottom Rinse
8000 50
µS/cm. In both cases it should have 7500
1 st Acid inj 0.5 / 2.0 %

been below 200 µS/cm. From the 7000

2 nd Acid inj 0.8 / 3.0 % 45

above it is obvious that more tests are 6500 40

needed to determine the correct rinse 6000
5500 35
times.

Concentration (mg/l)
5000

NTU, mS/cm or %
30
4500
Anion regeneration 4000 25
3500
20
The anion regeneration is done on the 3000

weak base and strong base resins in 2500 15

2000
two separate vessels. Results are there- 10
1500
fore discussed separately (Table 7).
1000
5
500
First regeneration: Figs. 2 and 0 0
3 (2 July 2002) 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210
Time (minutes)
The NaOH strength for the strong
Tot Hard mg/l as CaCO3 Ca Hard mg/l as CaCO3 Mg Hard mg/l as CaCO3 Na mg/l as Na
base regeneration average met the K m g/l as K Top % Str Inl (X10) Bottom % Str Inl (X10) Effluent %Str (X10)
specification of 4.0%, whereas the Turb (NTU) Conduct. mS/cm

NaOH strength for the weak base

Figure 1
regeneration was lower than the speci-
CATION 1 REGEN PROFILE - Major steps only - 2 July 2002
fied value of 4.0%. Referring to Figs.
1(b) and 1(c) it was evident that the 1200
4% Caustic injection Caustic inj. Slow rinse Fast rinse
170
160
injection times were too short be- 1100
150
cause a large amount of ions were 1000 140
removed during the rinse steps in- 130
900
stead of during the NaOH injection 120
step. 800
110
Concentration (mg/l)

700 100

mS/cm or %
Second regeneration: Figs. 5 600
90

and 6 (24 July 2002) 80

500 70
The injection times for the strong
60
base regeneration steps were increased 400
50
from 50 to 60 min (1st injection) and 300 40
30 to 40 min (2nd injection). For the 200 30
weak base regeneration the injection 20
100
time was increased from 30 to 40 min. 10

The reason for this decision was to 0 0

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
remove the majority of the ions dur- Time (minutes)
ing the injection phase. SB Cl m g/l (/10) SB SiO2 m g/l SB OH % Inl (x10) SB OH % eff (x10) SB K25 mS/cm

Chloride removal Figure 2

Table 5 shows that more chloride ANION 1 REGEN PROFILE - Strong Base - Major steps only - 2 July
ions were loaded on to the anion
1200 170
resins during the 1st service run than Caustic inj. Slow rinse Fast rinse
160
1100
on to the 2nd. Also, more chloride was 150
removed from the resin during the 1st 1000 140

regeneration although the injection 900

130

times were increased by a total of 20 800

120
110
min. Table 8 shows that when com-
Concentration (mg/l)

700 100
mS/cm or %

paring the two regenerations, the SBA 90

resin removed more chlorides during 600
80
the 1st regeneration than the 2nd. The 500 70

opposite happened with the WBA 400

60

resin, more chlorides were removed 50

300 40
during the 2nd regeneration.
200 30
This shows that the 2nd weak base 20
regeneration benefited more from the 100
10
10 min increase in injection time than 0 0
the strong base with reference to % 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
Time (minutes)
removal vs. loading. This confirms
W B SO4 m g/l (/100) W B Cl m g/l (/10) W B SiO2 W B OH % inl (x10) W B OH % eff (x10) W B K25 W B Col% (/10)
the theory that chlorides are predomi-
nantly exchanged by the weak base Figure 3
resins. ANION 1 REGEN PROFILE - Weak Base - Major steps only - 2 July 2002

Available on website http://www.wrc.org.za ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 415
 D ow n w ard

8 000
A cid injection T op & B o ttom R inse T op & B otto m R in se
50
Silica removal
1 st A cid in j 0.7/ 2.0
7 500
2 n d A cid in j 0 .7 / 4.0
Table 5 shows that more silica was
45
7 000 %
removed than was loaded on the
6 500 40 resin. When comparing Figs. 2 and 3
6 000
35
and Figs. 5 and 6 it appears that more
5 500

5 000
silica was loaded from the strong
Concentration (mg/l)

NTU, mS/cm or %
30
4 500 base onto the weak base than was
4 000 25 removed by the weak base during the
3 500
20
rinse steps. When comparing loaded
3 000
figures with removal figures, as in-
2 500 15
2 000
dicated in Table 5, it is evident that
1 500 10 the analysis could not be correct.
1 000
5
Silica was determined colorimetri-
500 cally on a spectrophotometer at wave-
0
0 10 20 30 40 50 60 70 80 90 1 00 110 1 20 130 140 15 0 160 17 0 180 1 90 20 0 2 10 22 0
0
length 810 nm. The following may
T im e (m in u tes ) be reasons why the analysis failed to
T o t H a rd m g/l as C aC O 3
T o p % S tr In l (x10 )
C a H a rd m g/l as C a C O 3
B ottom % S tr In l (x10 )
M g H ard m g/l a s C aC O 3
E fflue nt % S tr (x10)
N a m g/l as N a
T u rb (N T U )
K m g/l a s K +
C ond uct m S /cm
give the correct concentrations:

Figure 4 • The sample was too alkaline in

CATION 1 REGEN PROFILE - Major steps only - 24 July 2002 the case of the strong base regen-
eration and inadequate neutrali-
1200 170
4% Caustic injection Caustic inj. Slow rinse Fast rinse
sation was done. In the case of
160
1100 the weak base regeneration the
150
1000 140
sample was not as alkaline as in
900 130 the case of the strong base.
120 • The dilution factor was too large,
800
110 causing a significant error in the
Concentration (mg/l)

700 100 calculated values.

mS/cm or %
90
600 • Precipitation of silica before
80
500
analysis could have taken place.
70
60
400
50 Silica determinations in high alka-
300
40 line solutions should be further in-
200 30 vestigated for future regeneration
100
20 profile compilation.
10
0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 Sulphate removal
Time (minutes) Table 5 indicates that more sulphates
SB Cl mg/l (/10) SB SiO2 mg/l SB OH % inl (X10) SB OH % eff (X10) SB K25 (mS/cm) were removed during the 2nd regen-
eration. This can be ascribed to the
Figure 5 longer injection times during the 2nd
ANION 1 REGEN PROFILE - Strong Base Major steps only - 24 July 2002 regeneration.

1200 170 Colour

Caustic inj. Slow rinse Fast rinse
1100
160 During the regeneration process a
150
1000
change in colour of the weak base
140
130
effluent was evident. This colour
900
120 change from clear to a dark toffee
800
110 colour occurred during the initial
Concentration (mg/l)

700 100 stages of the fast rinse step due to the

mS/cm or %

600
90
release of organic material from the
80
500
ion-exchange resins.
70
60
To determine organic material
400
50 of high concentrations, is very com-
300 40 plex and difficult. Methods like Oxy-
200 30 gen Absorbed (4 h method) and To-
20
100 tal Organic Carbon (TOC) is avail-
10
0 0
able but in this case it was impracti-
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 cal to use it, because of the high
Time (minutes)
concentrations encountered. It was
WB SO4 mg/l (/100) WB Cl mg/l (/10) WB SiO2 WB OH % inl (X10) WB OH % eff (X10) WB K25 WB Col% (/10) decided to use the colour method to
distinguish between different con-
Figure 6 centrations. The absorption taken up
ANION 1 REGEN PROFILE - Weak Base Major steps only - 24 July 2002

416 ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 Available on website http://www.wrc.org.za
 TABLE 6
Top and bottom acid injection strengths during cation regeneration

1st Regeneration 2nd Regeneration

Top Bottom Time Top Bottom Time

Acid % Acid % Min Acid % Acid % Min

1st Injection 0.5 2.5 80 0.6 2.6 70

Specification 0.5 2.0 80 0.7 2.0 70

2nd Injection 0.5 2.5 20 0.6 3.1 30

Specification 0.8 2.0 20 0.7 4.0 30

TABLE 7
NaOH injection strengths during 1st and 2nd regenerations

1st Regeneration 2nd Regeneration

NaOH % Time Min NaOH % Time Min

Injection to strong base only 3.4-4.2 50 2.8-4.5 60

Specification 4.0 50 4.0 60

Injection to strong base and weak base*

Specification 4.04.0 3030 4.44.0 4040

Injection to strong base and weak base** 2.9-3.6 30 3.8-4.4 40

Specification 4.0 30 4.0 40

by each sample was measured on a spectrophotometer at wave-

length 425 nm. A graph, absorbance on the y-axis against percent- TABLE 8
age on the x-axis, was drawn up. Absorbance of 1.0 was taken as Chloride removal by strong and weak base resins
100% and a straight line graph was obtained. The concentration of during the two regenerations
each sample was read from the graph. In cases where the sample had
a colour greater than the absorbance of 1.0 the sample was diluted. Total Strong Weak
More organic material was removed by the WBA resins during the chloride base base
2nd regeneration than during the 1st. This increase was most prob- removal chloride chloride
removal removal
ably due to the longer injection times during the 2nd regeneration
(refer Table 5).
First regeneration 103.19 kg 85.83 kg 17.36 kg
Conductivity
The conductivity appeared very erratic during the fast rinse step Second regeneration 94.86 kg 71.31 kg 23.55 kg
(Figs. 2 and 5). This was due to changes in flow direction through
the weak base resin.
Increase acid injection strengths
Recommendations For the 2nd regeneration the acid strengths for both top and bottom
injections were adjusted from 0.5%/2.0% (1st injection) and
The research indicated a number of changes that could be made to 0.8%/3.0% (2nd injection) to 0.7%/2.0% and 0.7/4.0% (see Table 6)
the plant and process to improve the overall regeneration process respectively. It is recommended that the adjusted acid strengths be
and/or to obtain longer run lengths. These are: used since it is evident from Table 5 that more species were
removed during the 2nd regeneration.
Cation regeneration
Decrease the time for the 1st acid injection and
Modify the existing top and bottom acid injection step increase it for the 2nd
To control the acid strengths an automated system should be It is recommended that the time for the 1st acid injection be reduced
installed. During the project it was proved that manual acid from 80 to 70 min and the time for the 2nd acid injection be increased
operations were risky and control was poor. This can result in from 20 to 30 min. The total injection time should remain at 100
calcium sulphate precipitation on the cation resins. It is also min. The aim of this change is to remove more species from the
recommended that demineralised water instead of filtered water be resins during the regeneration step and not allowing this to happen
used as dilution water for both top and bottom acid injections. in the rinse steps.

Available on website http://www.wrc.org.za ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 417
Anion regeneration Acknowledgments

Increase the strong base NaOH injection time in both The authors acknowledge the support and data provided by the
injection steps Utilities Department at the Matla Power Station owned by Eskom
It is recommended that the injection time be increased from 50 to Pty (Ltd) and to Mpho Manamela who assisted in the laboratory
60 min for the strong base regeneration, and from 30 to 40 min for work.
the strong base/weak base parallel injection. The total injection
time for the strong base was increased from 80 to 100 min. The References
reason for this change was to remove as much silica and chloride
as possible during the injection steps and not during the rinse steps. DREW (1994) Principles of Industrial Water Treatment. Drew Industrial
Division, Boonton, New Jersey.
Conclusions ESKOM (2001) Chemistry Standard for Coal Fired Units with Drum
Boilers Operating at 17 MPa and Above. Eskom, Johannesburg.
FOSTER WHEELER DELTA COCHRANE (1978) Operation Manual for
Conclusions that can be made from the study of the regeneration Foster Wheeler Delta Cochrane Process Water Treatment Complex for
profiles are that longer injection times for both the cation and anion Matla South Power Station – Vol. 2. Eskom, Johannesburg
regenerations are needed for better removal of most of the species. FREDERICK K (1996) Countercurrent regeneration: Principles and Ap-
The study indicated that a large portion of species removal is plications. Ultrapure Water (Jul/Aug) 53-56.
obtained from the rinse steps. HARFST WF (1993) Improved feedwater quality enhances boiler opera-
Regeneration profiles are the ‘fingerprints’ of the deminerali- tion. Ultrapure Water (Feb) 34-37.
sation process. By analysing the fingerprints, inefficiencies can be MEYERS P (1996) Demineraliser design considerations – Part 2. Ultrapure
Water (Oct) 17-22.
eliminated and unnecessary costs and time delays can be avoided.
MODERN POWER STATION PRACTICE (3rd edn.) (1992) Chemistry
The studies showed that regeneration profiles can be used effec- and Metallurgy – Vol. E. Pergamon Press, New York. 37-43.
tively to identify plant and process problems and therefore to STANDARD METHODS (1995) Standard Methods for the Examination
optimise the process. of Water and Wastewater (19th edn.) Joint Ed. Board: Eaton AD,
It is recommended that regeneration profiles be done on an Clesceri LS and Greenberg AE. APHA-AWWA-WPCF. United Book
annual basis on all demineralisation trains. As experience grows it Press, Inc., Baltimore, Maryland.
might not be necessary to analyse for the full extent of the
regeneration process and for all parameters.

418 ISSN 0378-4738 = Water SA Vol. 29 No. 4 October 2003 Available on website http://www.wrc.org.za