Вы находитесь на странице: 1из 8

Carbon Dioxide and Dissolved Oxygen

Technical Paper Removal From Makeup Water By Gas

Tr a n s f e r M e m b r a n e s

S.H. Macklin, Northeast Utilities

W. E . H a a s , I o n i c s , I n c o r p o r a t e d
W . S . M i l l e r, I o n i c s , I n c o r p o r a t e d

Summary: New membrane processes for water • Hot water steam injection/stripping
treatment are continually being sought after for
• Oxidation/reduction reaction chemicals.
their simple and continuous operation. A new
• Ion exchange processes.
approach for removing dissolved gases from
make-up supplies utilizes hydrophobic hollow Removing dissolved gases from water with Gas
fiber membranes. These membranes have many Transfer Membranes (GTM®) is a proven, low
characteristics that enhance or replace other flow rate laboratory technology. Recent
degassing methods. This paper reports carbon advances in GTM® design now make them
dioxide and dissolved oxygen removal feasible for high flow rate industrial systems2, 3.
efficiencies for a 100 gpm gas transfer Using these systems alone or in conjunction with
membrane unit installed in a nuclear power other membrane processes offer new or
plant’s make-up system. improved options for water treatment.

Introduction This paper briefly describes the theory behind

GTM® units and how they complement reverse
Several dissolved gases concern the water
osmosis (RO) and catalytic oxygen removal
treatment industry. The most common are
technology. We also present carbon dioxide
oxygen, carbon dioxide and hydrogen sulfide.
(CO2) and oxygen (O2) removal test data from
These gases exist naturally in many water
a makeup water treatment system at a
supplies. Removing, adding or controlling the
nuclear utility.
dissolved concentration of these gases depend
on the water treatment user’s pure water GTM® Principles
requirements. For instance, controlling carbon
dioxide concentration is important for the The GTM® tested, utilizes hollow fiber

beverage industry. Municipalities remove hydrophobic polypropylene membranes. This

hydrogen sulfide gas, improving drinking water membrane is gas permeable and water

taste. And, utilities remove oxygen from boiler impermeable. The hollow fiber configuration,

feed makeup to prevent corrosion. Removing with the gas phase inside the fiber, offers high

these gases prior to, or during ion exchange surface area for maximum contact time. The

demineralization, can have a significant impact patented Liqui-Cel ® design4,5 tested includes

on operating performance and efficiency. The an internal baffle to promote turbulent flow in and

conventional water treatment arsenal available around the hollow fibers.

for dissolved gas removal includes1:

Each thread-like fiber has a 300 micron outer

• Aeration (open and forced draft). diameter with 0.05 micron gas permeable pores.

• Cold water vacuum deaeration.

As characteristic with membrane processes,

three streams are identifiable: Station is located in Waterford, CT. The station
consist of three independent units. Unit 1 is a
• Feed stream.
General Electric 650 MW boiling water reactor
• Process effluent.
(BWR) design, commissioned in 1970. Unit 2
• Concentrate waste stream.
and 3 are pressurized water reactors (PWR). Unit
2 is a Combustion Engineering 850 MW design
As illustrated in Figure 1, the tube interior commissioned in 1975. Unit 3 is a Westinghouse
operates under vacuum, nitrogen sweep gas or 1150 MW design, commissioned in 1986.
both in combination. Dalton’s Law of Partial
Pressure and Henry’s law of Gas Solubility1, 6 The plant’s makeup water is deoxygenated and

define the mechanisms that drive dissolved gas demineralized by a service contract company.

removal by GTM®. In short, reducing the gas Unit 1 and 2 share the same makeup water

concentration over the water allows the treatment system and operates independently

dissolved gases to expand into the hydrophobic from Unit 3’s makeup system. We selected Unit

hollow fiber and be swept away. Reducing the 3’s makeup system for the test site based on its

CO2 and O2 gas concentration is accomplished design and Northeast Utilities’ long history for

by applying a vacuum, using inert nitrogen supporting and testing new technology7, 8.

sweep gas or combining both.

Makeup System Design. The raw water
Test Site supply, analysis shown in Table 1, originates from
Lake Konomoc and is treated by the city of New
Northeast Utilities Millstone Nuclear Power
London. Unit 3’s 200 gpm makeup water
treatment system is shown in Figure 2 with a 100
gpm GTM® unit incorporated. A by-pass is
Figure 1 Liqui-Cel ® GTM unit
utilized during 200 gpm peak demand periods.
Vacuum and/or Sweep
Sweep Gas Gas With the GTM®, the make up system utilizes

Distribution Baffle Cartridge

three separate membrane processes. Each
membrane unit has a separate operating
purpose that compliments the other. The
Aqueous Aqueous makeup system, as originally designed, includes
Stream Stream

Hollow Fiber Collection

ultrafiltration (UF) followed by reverse osmosis
Membrane Tube
(RO). The makeup water is then deoxygenated
and polished by ion exchange demineralization.

Figure 2 Unit 3 Makeup Water Treatment System

Ion Exchange
New London
297 gpm






200 gpm
N2 To Storage with
Nitrogen Blanket
Carbon Catalyzed
Reject Reject

Table 1 Millstone Municipal Raw Water Supply Table 2 RO Permeate Quality
Cations Anions TDS as Ions 5 - 7 mg/l
(mg/l as CaCO3) (mg/l as CaCO3)
Conductivity: 5 - 13 µmhos
Ca+2 8 - 30 HCO3-1 6 - 13 HCO3-1 < 1.0 mg/l
Mg+2 3-5 Cl-1 6 - 14 CO2 as CO2: 6 - 12 mg/l
Na+1 8 - 16 SO4-2 2 - 19 O2 as O2: 6 - 8 mg/l
K+1 1-2 NO3-1 0-2 TOC as C: 0.2 - 0.4 mg/l
TDS as 45 - 65 Conductivity: 50 - 90 pH: 6.1 - 8.1
Ions: mg/l µmhos
CO2 as 0.75 - 1.5
O2 as O2: 6 - 8 mg/l
CO2: mg/l
SiO2 as higher hardness and alkaline feeds, acid
SiO2: 2 - 5 mg/l TOC as C: 2 - 6 mg/l
pretreatment reduces the Langlier Saturation
pH: 6.1 - 8.1 Turbidity: 0.01 - 2.0 NTU
Index (LSI)9 in the RO reject stream, preventing
calcium carbonate membrane scaling. Although,
Ultrafiltration. The UF utilizes an 80,000
injecting acid minimizes CA membrane
molecular weight cut off polysulfone membrane
hydrolysis and reduces the scaling potential, it
in a hollow fiber configuration. The system,
creates free carbon dioxide, refer to Eq. (1).
operating between 85% to 90% recovery,
effectively pretreats the city supply to minimize
Every 1 ppm H2SO4 added to the RO feed
colloidal fouling on the down stream RO
Eq. ( 1 )
membranes. The UF membranes are chemically
cleaned on a daily to monthly basis depending
on the city water quality. Cleaning requirement is
determined by effluent silt density index greater
than five, increasing pressure drop or flux
increases the CO2 concentration by 0.89 ppm.
Reverse osmosis membranes do not remove
dissolved gases. Therefore, the RO permeate
Reverse Osmosis. The RO utilizes cellulose
contains equal amounts of CO2 as generated in
acetate (CA) membranes in a spiral wound
the feed water, adding undue ion exchange
configuration. Cellulose acetate membranes, on
loading on the down stream demineralizer10.
this water supply, have successfully
Conventionally, CO2 is removed by forced draft
demonstrated nine years operating experience
aeration or vacuum deaerators1.
without a single cleaning.

This membrane process reduces the total

Deoxygenation. The RO permeate is
dissolved solids (TDS) loading on the down
deoxygenated to less than 3 ppb dissolved
stream ion exchange demineralizer by 85% to
oxygen utilizing a patented DEOX®
95%. A typical permeate analysis is tabulated in
process11, 12. The process, catalyzed by
Table 2. Chemical pretreatment to the RO
activated carbon, reacts dissolved oxygen with
includes chlorine for biological control and
hydrazine producing nitrogen and water.
sulfuric acid to prevent CA membrane hydrolysis.
Hydrazine over feed and carbon leachables are

removed by the down stream ion exchange feed pH, temperature and vacuum/nitrogen
demineralizer. combinations. Over 2.5 million gallons were
treated during the GTM® test. As illustrated in
Figure 3, oxygen removal ranged between 78%

Table 3 Makeup Water Quality and 95%. Carbon dioxide removal, illustrated in

Conductivity: 0.056 µmhos Figure 4, ranged from 65% to 95%.

Silica as SiO2: < 5 ppb
Na+ as ion: < 0.2 ppb
Cl- as ion: < 0.1 ppb
O2 as O2: < 3 ppb Figure 3 Percent Oxygen Removal
TOC as C: < 10 ppb

Demineralization. Demineralization and

removal of carbon leachables is accomplished
by a three step ion exchange process employing,
strong acid cation, strong base anion and
polishing mixed bed resin. The activated carbon
for the DEOX process and the ion exchange
demineralizer are mobile mounted for off site
regeneration. The final effluent quality is
tabulated in Table 3.

Gas Transfer Membrane. The GTM is piped

in between the RO and the DEOX process. RO Figure 4 Percent Carbon Dioxide Removal

pretreatment is recommended to prevent fouling

by colloids and large particles that might
otherwise become entrapped in the GTM®’s
hollow fibers. Originally designed, the makeup
system removes the CO2 generated in the RO
process only by the strong base anion exchange
resin in the demineralization process.

Selecting Unit 3’s makeup system for the test site

was based on the GTM®’s CO2 and O2 removal
capabilities. The test objectives for the study are:
CO2 Removal. The GTM® influent and effluent
• Carbon dioxide reduction.
CO2 concentration at this site is below the
• Dissolved oxygen reduction.
detectable 10 ppm wet chemistry method
• Reduce hydrazine usage. employed. The CO2 data presented here is
calculated using standard methods13 and
Results weak acid chemistry. Based on stoichiometric
acid feed, the maximum calculated CO2
During a four month period approximately two
concentration feeding the GTM® unit is 6 to 12
hundred data points were collected at varying

ppm. Based on the 65% to 95% CO2 removal pH probes were installed to measure the
efficiency, this equates to 0.3 ppm to 4.2 ppm pH difference across the GTM®. This allowed
GTM® effluent CO2 concentration. Carbon for a reliable in-line method to monitor CO2
dioxide concentration depends on the removal performance.
chemical relationship between alkalinity and pH,
Figure 5 illustrates that CO2 removal efficiency
refer to Eq. (7)10.
is linearly related to feed pH. At lower pH,
Eq. ( 7 )
with correspondingly higher feed CO2
concentrations, percent removal exceeds 85%.
This data, when extrapolated, shows that
optimum operating pH for CO2 removal is 4.7
(free mineral acidity) where all alkalinity is
converted to CO2. The data was collected at 21°
Based on this relationship, Equation (7) is to 22° C and vacuum at 27” Hg.
solved for [CO2], providing a method where by
carbon dioxide concentration is calculated, Figure 5 Carbon Dioxide Removal pH Effect
refer to Eq. (8).

Eq. ( 8 )
% Removal

The GTM®’s dissolved gas removal efficiency,

expressed as a percentage, is defined by
Equation (9).

Eq. ( 9 )

Figure 6 demonstrates CO2 removal efficiency

versus temperature. Carbon dioxide removal

Assuming the alkalinity concentration does not improves as the feed water temperature

change through the GTM® unit, combining increases. As the water temperature approaches

Equation (8) and (9) calculates removal efficiency its boiling point, gas removal approaches 100%.

based on the influent (INF) and effluent (EFF) pH As water temperature approaches its freezing

change, refer to Eq. (10). point, gas removal approaches 0%. This data
was collected at influent pH 5.0 to 5.5 and
Eq. ( 10 ) vacuum at 27” Hg.

Figure 7 illustrates the CO2 removal efficiency

while employing nitrogen sweep gas at 0.5
Based on these equations, pre and post in-line SCFM, with and without vacuum at 27” Hg. The

data indicates CO2 removal is at its lowest when During the test dissolved oxygen data was
using nitrogen and vacuum in combination. collected at varying temperatures and while
However, CO2 removal efficiency is at its highest applying nitrogen sweep gas with and without
when using nitrogen alone. The poor efficiency vacuum. The GTM® consistently reduced the
using nitrogen in combination with vacuum can dissolved oxygen below 1.1 ppm. The minimum
be due to nitrogen impurity, air in-leakage and dissolved oxygen concentration recorded was
difficulty controlling feed pH. 350 ppb utilizing nitrogen and vacuum. Utilizing
vacuum only, dissolved oxygen concentration
ranged between 900 to 700 ppb. In addition to
Figure 6 Carbon Dioxide Removal
Temperature Effect varying conditions, the GTM®’s effect on the
hydrazine usage for the down stream
deoxygenation process was investigated.
Figure 8 illustrates the logarithmic temperature
% Removal

effect on the GTM®’s dissolved oxygen removal.

This information is reported in a semi-log format
so that the relationship can be shown linearly. As
expected, dissolved oxygen removal efficiency
improves with increasing temperature. Over a
13° to 25° C temperature range, the GTM®’s
effluent dissolved O2 decreased from 1.3 ppm to
0.6 ppm respectively.

Figure 7 Carbon Dioxide Removal Figure 8 Dissolved Oxygen Removal

Nitrogen and Vacuum Effect
ppm D.O. ( mg/l )
% Removal

Unlike CO2, nitrogen sweep gas with and without

vacuum, demonstrated more influence on O2
O2 Removal. Oxygen solubility is not removal. Combining vacuum with 0.5 scfm
dependent on pH changes. Hence, oxygen nitrogen, improved dissolved O2 removal as
removal is easier to understand and accomplish compared to the singular use of either by 4% to
than CO2. 11%. Using 0.5 scfm nitrogen independently,

shown in Figure 9, slightly improved overall O2
Figure 10 Hydrazine Dosage
removal compared to vacuum alone.

lbs/k Gallons Processed ( x102 )

Figure 9 Oxygen Removal
% Removal

is suggested. GTM® and catalytic oxygen

removal systems have site specific evaluated
Hydrazine. Catalyzed oxygen removal costs. Working in combination, single pass
processes reduce dissolved oxygen by greater GTM® with a catalytic oxygen removal system
than 99.9%. The patented deoxygenation down stream significantly reduces operating and
system permanently installed in the plants equipment cost for both processes.
make-up system achieves less than 3 ppb
dissolved oxygen. Installing GTM®, prior to Conclusion
the catalyzed oxygen removal process,
dropped hydrazine consumption 7 to 10 fold This study provides insight into dissolved oxygen

without sacrificing dissolved O2 effluent and low level carbon dioxide removal utilizing

quality. The data illustrated in Figure 10 a new membrane process. GTM® technology

reveals a substantially reduced hydrazine offers several advantages over conventional

requirement. Combining these processes dissolved gas removal methods. GTM®’s are

includes several advantages: comparably smaller, modular in design and

offer lower capital cost with no chemical
• Minimizes hydrazine handling. requirements.
• Reduces hydrazine consumption.
• Reduces TDS loading on down stream Operating several GTM® units in series with
demineralizer due to excess hydrazine over and without nitrogen sweep gas is a subject
for future reports. Furthermore, higher CO2
• Provides efficient means to produce < 3 ppb
dissolved O2 utilizing a two step process. feed concentrations warrant investigation.

• Provides system redundancy. And, additional testing is required for a proper

• Does not increase operation requirements. cost analysis versus achievable effluent
quality. The data presented here provides a
In order to achieve less than 3 ppb dissolved O2 strong foundation for GTM® technology that
with GTM® alone, multiple units in series, can be used to design future makeup water
combined with vacuum and nitrogen sweep gas treatment systems.

This paper was presented at IWC-95-41 8. T.F. Burns, C. Booth, “Utrafiltration
Pretreatment of Reverse Osmosis Make-up

1. S.B. Applebaum, Demineralization by Ion Philadelphia, PA, Mar., 1988.

Exchange: In Water Treatment and Chemical

9. Annual Book of ASTM Standards, vol 11.02,
Processing of Others Liquids, (San Diego, CA:
1995: Method D3739.
Academic Press, 1968), pp 108 - 129.

10. R.L. Reitz, “A Computer Program for

2. F. Wiesler, R. Sodaro, “Degassing of Water
Evaluation of Water Chemistry and Scale
Using Novel Membrane Technology”,
Prevention in RO Desalination Systems”,
ULTRAPURE WATER EXPO 95, Philadelphia, PA,
ULTRAPURE WATER, 6(1), Jan./Feb., 1989.
May 1995.

3. P. D’Angelo, “Oxygen Removal: Theory and 11. R.C. Dickerson, W.S. Miller, U.S. Patent 4,

Potential Use of Deoxygenation Membranes in 556, 492, “Deoxygenation Process”, Dec. 3,

the Utility Industry”, ULTRAPURE WATER, 12 (5), 1985.

Jul./Aug., 1995.
12. W.S. Miller, “Oxygen Removal by Catalyzed

4. R. Prasad, C. Runkle, H. Shuey, U.S. Patent Carbon Beds”, EPRI Condensate Polishing

5,264,171, “Method of Making Spiral-Wound Workshop, Richmond, VA, Oct. 31, 1995.

Hollow Fiber Membrane Fabric Cartridges and

13. Standard Methods for the Examination of
Modules Having Flow Directing Baffles”, Nov. 23,
Water and Wastewater, 17th Edition, APHA,
Washington, DC, 4500-CO2 A-D.

5. R. Prasad, C. Runkle, H. Shuey, U.S. Patent

5,352,361, “Spiral-Wound Hollow Fiber
Membrane Fabric Cartridges and Modules
Having Flow -Directing Baffles”, Feb. 16, 1993.

6. T.L. Brown, H.E. LeMay, Chemistry: The

Central Science, 2nd edition, (New Jersey:
Prentice-Hall, 1981) pp 245-270 and pp 345-375.

7. W.S. Miller, V.W. Jones, R.H. Langer, “TOC

Removal From Make-up Water at Millstone
Nuclear Generating Station”, 47th International
Water Conference, Pittsburgh, PA, Oct., 1986.

Ionics, Incorporated © 2004 Ionics, Incorporated. All rights reserved worldwide. The information contained in this publication is believed to be
65 Grove Street accurate and reliable, but is not to be construed as implying any warranty or guarantee of performance.
Watertown, MA 02472

TP-513 E-US 1204 formerly ECOLO 507