Technical Paper Oxygen Removal by Catalyzed Carbon Beds Author: William S. Miller EPRI Condensate Polishing Workshop, October 29, 30 and 31, 1985, Richmond, Virginia Introduction This paper will discuss a chemical process for removing dissolved oxygen from either make-up water supplies or condensate. The process involves the addition of hydrazine to the oxygenated influent, then passing the mixture through activated carbon media to catalyze the reaction, followed by a suitable ion exchange bed downstream to remove any carbon impurities or hydrazine overfeed. The reaction is stoichiometric and rapid even at cold (35°F [1°C]) influent temperatures and consistently produces an effluent analyzing less than 10 ppb dissolved oxygen from influents containing 12 to 14 ppm dissolved oxygen. The process is already proven commercially. Since May 1983, numerous utility and industrial users have used this process to produce over 500 million gallons (2 million m3) of deoxygenated water in a variety of application modes. Hydrazine Reaction The reaction of hydrazine and oxygen can be expressed:1 N2H4 (Hydrazine) + 02 → (Oxygen) N2 + 2H20 (Nitrogen) (Water) Since the reaction products are inert nitrogen gas and water, hydrazine has a unique advantage over other chemical oxygen scavengers such as sodium sulfite or isoascorbic acid, which actually add dissolved solids to the boiler feed water. However, the reaction kinetics for the hydrazine and oxygen reaction are poor and prohibitively slow at 35°F to 80°F (1°C to 12°C) range common to make-up water supplies. For instance, an 11 fold excess of hydrazine will only reduce the dissolved oxygen from 9 to 7 ppm over a two hour period at 70°F (21°C). Even large excesses of catalyzed hydrazine, which contains metal or organometal additives, do not sufficiently increase reaction rates for low temperature applications. For the above example, using an 11 fold excess of catalyzed hydrazine still requires a reaction time of 10 to 30 minutes at 21°C (70°F) while allowing extremely large excesses of hydrazine and catalyst to remain in the boiler feed water.3 Activated Carbon Catalysis Considering the relative sluggish reaction rates just discussed, the rapid rate of the hydrazine - oxygen reaction when catalyzed by an activated carbon bed is astonishing. At 35°F (1°C) reactions are still completed in seconds rather than minutes or hours. Production of 10 ppb dissolved oxygen levels is attained immediately on start-up and can often be maintained for a period of several hours after shutdown of the hydrazine feed. This “flywheel effect” is most useful for applications designed to maintain low dissolved oxygen levels in storage tanks or during boiler layups. Equipment requirements are minimal. A chemical feed pump, a pressure vessel to contain either granular or powdered carbon, followed by a downstream ion exchange unit is all that is required. We have not found the need for a special grade of carbon. To date, most of our commercial experience is with a coal-based carbon. We have used products from three different manufacturers Find a contact near you by visiting www.ge.com/water and clicking on “Contact Us”. * Trademark of General Electric Company; may be registered in one or more countries. ©2009, General Electric Company. All rights reserved. TP1052EN.doc Jun-09 with equal success. Laboratory tests on wood based carbons also showed satisfactory deoxygenation capability. Process Kinetics To study the time required for complete reaction, pilot tests where conducted with fiberglass pressure tanks containing 1 cubic foot of 12 x 40 mesh coal based carbon. A feed of 12 ppm hydrazine was added to a 51°F (10.5°C) influent water supply containing about 9 to 10 ppm dissolved oxygen and passed downflow through the carbon column at various flow rates. Both Norfolk and Virginia City water (240 Mmhos) and a two-step D.I. influent (10 mmho) were tested. The effluent was analyzed by a direct reading in-line Rexnord Model #340-0 dissolved oxygen analyzer, backed up by colorimetric tests using Chemets manufactured by Chemetrics. Results are summarized in Table 1. Table 1: Deoxygenation Tests Dissolved O2- Carbon Effluent obtained. This contradicts actual commercial experience, which has presented no problems in attaining less than 10 ppb dissolved oxygen effluent quality in the operating range of pH 5.5 to 10. Carbon Leachables One of the few limitations for the hydrazine - carbon deoxygenation process is the need to provide downstream ion exchange treatment to remove impurities leached from the carbon during the process and any overfeed of hydrazine. Theoretically, there is no need to overfeed hydrazine as the carbon catalyst will allow stoichiometric operation. From a practical view, it is advisable to feed 10-20% excess to allow for subtle changes in influent flow rate, dissolved oxygen content, and hydrazine pumping inaccuracies. Hydrazine is effectively removed from solution to non-detectable concentrations (<5 ppb) by a hydrogen form strong acid cation resin. The selectivity of the sodium form of the resin is insufficient to obtain a commercially acceptable capacity for hydrazine. When considering the carbon catalyzed process one must not overlook the potential for effluent contamination due to the carbon media. In addition to carbon fines, several dissolved ionic salts will be leached into solution. The ionic leachables from coal based carbon is shown in Table 2: Table 2: Coal Based Carbon Leachables As shown, the reaction rates and attainable effluent quality are markedly improved over the previously stated rates for non-carbon catalyzed process. All full-scale installations have obtained the effluent qualities predicted even when starting with raw lake or river water at 35°F (1°C). While the results in Table 1 were confirmed with a minimum of three repeats, I still find the difference in kinetic rates between the city supply versus the D.I. supply difficult to explain. Since the effluent qualities are attainable within five minutes of startup, the difference cannot be explained by overloading organics on the carbon with city water. The use of two-step demineralized water does introduce a pH variable, i.e. pH 8.5 for two-step versus pH 7.3 for the city supply. A paper presented at the 1985 Liberty Bell Corrosion Course suggests better kinetics at higher pH, but the data also shows that below pH 9, effluent qualities below 100 ppb were not Page 2 These leachables could come either from the ash content of the carbon or perhaps from the water used to initially wet, transfer, and backwash the carbon. The concentration of the leachables in the carbon effluent decreases slowly with time but even after 64 bed volumes there are still significant levels of impurities being added to the effluent. The data presented in Table 3 was obtained by passing 0.1 Mmho water containing 8 ppm dissolved oxygen through a bed of coal based carbon at a flow rate of 1 GPM/FT3. TP1052EN Table 3: Carbon Leachables vs. Time (0.004 m3/h) deionized water analyzing 860 ppb TOC through a 1 ft3 (0.03 m3) bed of virgin activated carbon. The deionized water was prepared by taking an alum coagulated municipal surface supply analyzing 3,000 ppb TOC and passing it through a “tired” mixed bed whose anion component was about 30% low in salt splitting capacity. Table 4: TOC Removal by Activated Carbon ppb TOC Since the deionized influent analyzed only 5 ppb silica, the amount of silica leachable is very significant and requires a downstream strong base anion resin in the hydroxide form for removal. Additional leaching tests with a deionized feed containing 8-10 ppm hydrazine did not alter the type or concentration of leachables. Surprisingly, start/stop operation did not produce an exaggerated concentration of leachables upon restart, but simply maintained the smooth downward trend shown in Table 3. Most importantly, a properly regenerated mixed bed downstream of the carbon column has no trouble in removing any of the carbon leachables and has often provided an upgrade of D.I. influent water quality. Actual operating data obtained over a one-year period from a 500 gpm (2 m3/h) installation is shown in Table 5. The influent was a coagulated and filtered municipal surface supply which was deionized by strong acid cation, strong base, type I, gel anion, activated carbon with hydrazine feed for deoxygenation, followed by mixed bed polishing. Table 5: TOC Removal — System Profile ppb TOC TOC Removal The removal of organics by activated carbon is well documented in the literature. Factors such as the chemical structure of the organic compound, concentration, temperature, pH, carbon pore and particle size can affect adsorptive capability. For the general category of organics, sometimes referred to as “naturally occurring organics” commonly found in make-up water supplies, activated carbon is only effective for removing 30% to 70% of the total influent TOC. In fact, if one performs TOC profiles across surface water supplied demineralizing systems which contain an activated carbon column, one would find that most of the organic removal is attained by a type I strong base anion resin. The introduction of hydrazine has not shown any benefit in reducing the organic removal capability of activated carbon compared to a carbon column without a hydrazine feed. Tables 4 and 5 present some TOC removal data for a coal base activated carbon having a surface area 960 to 1200 ft2/0.04oz. (800 to 1000 m2/g). The results in Table 4 were obtained by passing 1 gpm TP1052EN During this time period over 100 million gallons (400,000 m3) were processed through 360 ft3 (10m3) of activated carbon. To engineer and install a system to produce parts per billion levels of dissolved oxygen makes little sense unless these levels can be maintained while the water is being stored. Storage tanks open to atmosphere will not maintain low oxygen levels in the stored water. The common approach to the problem is to engineer either a floating bladder or nitrogen blanket system on the storage vessel. Both systems have been found acceptable for maintaining less than 10 ppb dissolved oxygen levels. To prevent air ingress or nitrogen loss through a storage tank overflow pipe, an overflow “wier box” or “u-bend” full and overflowed periodically has been used successfully. Page 3 Should the nitrogen blanket be “lost”, it could take an extended time for storage tank dissolved oxygen levels to return to normal. The data in Table 6 are from a 500,000 gallon (2000 m3) storage tank designed with a top fill and bottom drain. Table 6: Storage Tank Dissolved Oxygen The major process features include the following: 1. Low temperature operation (1°C) 2. Effective on both raw and demineralized supplies 3. Immediate and consistent effluent quality below 10 ppb O2 4. No downstream impurities are added 5. Flywheel effect prevents immediate quality degradation upon loss of chemical feed Potential applications include deoxygenation of make-up supplies, maintenance of low dissolved oxygen levels during system lay-ups and in storage tanks, and full flow deoxygenation of condensate. A patent has been allowed for the process and issuance is scheduled for December 3, 1985. During the period this data was obtained, the daily draw from the tank was only slightly less than the gallons added which averaged 400,000 gpd (1500 m3/day). The average quality being fed to the tank was 1 to 3 ppb dissolved oxygen. Addendum Table 7: Deoxygenation Experience As stated previously, most of the low level dissolved oxygen measurements were made with an in-line Rexnord dissolved oxygen analyzer. Some interesting differences in analyses have been obtained depending how the analyzer is installed. If the analyzer, which requires a constant 50 ml/min flow, is fed by plastic tubing the recorded oxygen levels will be higher than if hard piping is used. In one case, braided nylon tubing was used as the feed line to the analyzer and a range of 5 to 10 ppb dissolved oxygen was indicated. With no other changes being made except to run stainless steel tubing to the instrument, the indicated oxygen values immediately and consistently dropped to 0.5 to 2 ppb dissolved oxygen. This observation, which is discussed in the Rexnord operating manual, is apparently due to the ability of plastic tubing to breathe, allowing air to contaminate the sample. References 1. Hercules Bulletin WMC-l23, Boiler Oxygen Corrosion. 2. Hydrazine and Its Derivatives, Schmidt, E.W., pg. 829, John Wiley & Sons, New York, NY, 1984. 3. IBID, pg. 828 4. Carmen, C., Oxygen Reduction Via Activated Carbon Precoat, 23rd Annual Liberty Bell Corrosion Course 4, Sept. 1985. Summary The process for removing dissolved oxygen by addition of hydrazine to the influent and passing the mixture through activated carbon followed by downstream ion exchange treatment is a simple, effective, and commercially available process. Page 4 TP1052EN
© Copyright 2026 Paperzz