Air is not dead yet! Ian A. Crossley, C. Eng.1 David Nickols, P.E., C. Eng. 1 Orren Schneider, Ph.D., P.E. 1 Michael Borsykowsky, P.E. 2 Ian A. Howkins, C. Eng. 3

Abstract Contrary to typical practice, air was selected instead of oxygen as the feed gas for ozone generation for New York City's 290-mgd Croton Water Treatment Plant. To satisfy local concerns, the plant is being constructed underground. The City Building Code prevents the use of LOX underground. To meet the needs of the 6,200-lb/day ozone facility, the only viable options to provide feed gas were pressure swing adsorption (PSA) and vacuum/pressure swing adsorption (VPSA) oxygen generation, and air. It was found that the additional space and headroom requirements and capital costs of oxygen generation equipment more than offset its operational savings compared with air feed gas preparation equipment. Introduction This paper explains the rationale that led to the unfashionable decision to use air instead of oxygen as the feed gas for ozone generation for New York City’s 290-mgd Croton Water Treatment Plant. New York City Department of Environmental Protection (NYCDEP) is currently designing a new 290-mgd water treatment plant to treat the City’s Croton water supply. The treatment process will consist of dissolved air flotation, ozonation and monomedium high-rate, deep bed filtration. The facility is designed to operate as two independent trains, each with a capacity of 145 mgd. The complete water treatment plant, clearwells and pumping stations will be located entirely below grade, beneath the golf driving range at the Mosholu Golf Course within Van Cortlandt Park, in the Borough of the Bronx, New York. To meet a federal Consent Decree, the treatment 1

Hazen and Sawyer, P.C., 498 Seventh Avenue, New York, NY 10018. New York City Department of Environmental Protection, 90-05 Horace Harding Expressway, Corona, NY 11368. 3 Metcalf & Eddy of New York, Inc., 60 East 42 Street, New York, NY 10165. 2

plant must be in operation by March 2007. The below-grade construction presents unusual design challenges, and directly affected the selection of the ozone system’s feed gas. The ozonation process is designed to provide at least 1-log inactivation of Cryptosporidium oocysts. The ozonation system will consist of two independent treatment trains, each designed for one half of the maximum flow. Each train will include air preparation followed by ozone generation; two ozone contact basins; ozone quenching facilities consisting of sodium bisulfite storage and application; and off-gas ozone destruct systems. The system is designed to provide an effective CT (ozone residual concentration times effective detention time) of about 10 mg-min/l for cold water conditions, with an effective detention time (T10 ) of 16 minutes. The major design decision was whether to use oxygen or air as the feed gas for ozone production. Using liquid oxygen (LOX) is not feasible as the New York City Building Code specifically mandates exterior locations for oxygen storage. Pressure swing adsorption (PSA) and vacuum/pressure swing adsorption (VPSA) were investigated for on-site generation of oxygen. As expected, the operating costs for these processes were favorable in comparison with air-fed ozone generation. However, it was found that the additional space and headroom requirements and capital costs of the equipment offset its operational savings compared with air feed gas preparation equipment. The Croton Water Treatment Plant (WTP) may well be the last large-scale plant to use air as the feed gas for ozonation. Ozone will be produced on site by high concentration generators (3% by mass) using dry air. The ozone generators will be sized to deliver a maximum dose of 2.4 mg/l, or 6,200 lb/day of ozone at maximum flow. There will be two 50% duty generators per train, each matched to a contact basin, with a common standby generator. Any one of the three ozone generators can feed two contactors. On-site Oxygen Generation Liquid Oxygen (LOX) The City of New York Building Code (ref. Chapter 15 Liquid Gas Storage Containers, clause §II 15-01) requires liquid oxygen storage facilities to be located outdoors. The facility is located below parkland and there is no possibility of LOX facilities being located outdoors at or near the WTP site. Other sites for a LOX facility were not available nearby. Consequently, the use of LOX is precluded. Therefore, if oxygen is to be used as a feed gas for ozone generation then it has to be generated on site. On-site generation of oxygen involves separation of the gas from the other constituents of air through a cryogenic process or a molecular sieve adsorption process.

Cryogenic Production Cryogenic oxygen production would provide 99% pure oxygen and is best suited to for demands exceeding 160,000 lb/day of oxygen or the equivalent of 10,000 lb/day of ozone (at 6% concentration). This capacity is nearly twice the maximum Croton oxygen requirements. Therefore, cryogenic on-site generation would not be economically viable owing to the oxygen demand not being great enough for this production method. Molecular Sieves If the capacity of the ozone plant justifies it, then PSA on-site generation of oxygen is a viable proposition. Molecular sieve adsorption technologies such as PSA or VPSA are best suited to oxygen production rates of up to 50,000 lb/day (3,000 lb/day ozone at 6%) and 200,000 lb/day (12,000 lb/day ozone at 6%), respectively. PSA and VPSA technologies produce oxygen at purity levels of 90 to 95%. However, they are more economically operated as part of an ozone system at 92 to 93% purity, because a small amount of nitrogen (4%) assists in the generation of ozone, while the balance as argon has no detrimental effect. The major components of molecular sieve air separation processes are: • • • • • •

Air compression; Water vapor adsorption; Carbon dioxide and hydrocarbon adsorption; Nitrogen adsorption; Switching; and Desorption.

The adsorption vessels containing the molecular sieve are usually multi-tower, with one unit adsorbing while another unit regenerates on a timed cycle, using a portion of the outlet gas and controlled by the switching mechanism. A third adsorption vessel can be used to further improve the recovery of purged oxygen. A low-pressure oxygen gas storage vessel is also used on occasions to remove pulsations and smooth the flow of gas to the ozone generators. Two on-site alternatives for oxygen generation were investigated: PSA and VPSA. There are several vendors of this type of equipment; for example, Lotepro, Air Liquide, BOC Airsep, and Praxair. For Croton, either PSA or VPSA would be appropriate. However, not all vendors are willing to offer PSA, as several believe VPSA would be the most economic process alternative and see no reason to consider PSA. PSA / VPSA Design Diversity During the investigation phase with the equipment vendors it became clear that the solutions to the production of oxygen using molecular sieves were very diverse. For example, Lotepro uses three vertically-mounted adsorption vessels, Air Liquide uses

two vertically-mounted adsorption vessels with a smoothing vessel, and Praxair uses a single vertically-mounted adsorption vessel with a very large horizontal smoothing vessel. The pipe and valve configuration used to connect the vessels and compressors is carefully designed by the vendors to ensure correct flow characteristics. This means that the layout of vessels, pipes and rotating equipment is also vendor-specific and relatively inflexible. These variations and layout limitations created a major design problem in terms of space allocation and headroom access requirements in the Croton plant. Being a multi-level, below-grade facility, the cost of increasing the building footprint is very high, in the order of $500/sq. ft. Both PSA and VPSA vessels are very tall and require access from above to allow the molecular sieve medium to be loaded. This meant that two levels are needed for the oxygen generation equipment. Had the treatment plant been located at grade, it is likely that these requirements could have been readily and economically accommodated. Levels of Redundancy The NYCDEP requires high levels of equipment redundancy in its treatment plants. These requirements led to the provision of two independent treatment trains within the facility, each capable of producing 145 mgd. Within each train there are further levels of redundancy so that no single failure can remove a train from service. The same policy was applied to the oxygen generation equipment, resulting in two sets of PSA or VPSA units for each train, each unit rated for 100% duty. Under normal circumstances, a series of LOX tanks would provide extra redundancy. LOX storage would also give the ability to reduce the size of the molecular sieves by rating them for a base load capacity and using oxygen from the LOX tanks to supplement the feed gas supply during peak loads. Neither of these strategies was available for the Croton project, and consequently capital and operational costs were increased. The maximum output of 290 mgd and maximum ozone dose of 2.44 mg/l result in a total ozone output of 6,200 lb/day. To provide a maximum 3,100 lb/day of ozone per treatment train at approximately 6% concentration, the output of oxygen needed would be about 50,000 lb/day. Normally, for ozone systems using oxygen generation, there is a optimum concentration that is determined by the balance between total equipment capital cost and electrical consumption. This optimum concentration normally falls between 6 and 8%. The lower figure was chosen as a starting point, with the intent that the optimum concentration would be determined at final design. For preliminary design, each PSA or VPSA oxygen generation unit was sized to have a capacity of 50,000 lb/day. PSA/VPSA Turndown Capability There are three sources for New York's drinking water; the Delaware, Catskill and Croton watersheds. The three sources are not filtered. The Catskill and Delaware supplies have been granted temporary, conditional filtration avoidance. The filtration plant for Croton source is now under final design, as discussed in this paper. During the periods of drought, the use of the Croton source has to be maximized. High Croton

system demands can persist for up to three months continuously during cold water conditions. Table 1 summarizes the minimum, average and maximum conditions. TABLE 1. OZONE REQUIREMENTS Plant Flow (mgd) Required (1,2) Influent Recycle Total Ozone Dose (mg/l) Minimum 90 5 95 0.78 Average 144 6 150 1.00 Maximum 290 15 305 2.44 Note: 1. Assumes 95% transfer efficiency 2. Minimum dose at minimum flow is at 4°C. Flow Condition

Ozone Requirement (lb/d) 620 1,250 6,200

The table indicates that the turndown ratio of maximum to average flow is 5:1, and the worst case ratio could be 10:1. Although ozone generators can accommodate these high ratios, the oxygen generators cannot, and would be operating at poor efficiencies during most of their operational lives. With PSA, using multiple feed compressors can mitigate this situation. However, the design of VPSA systems does not allow much better than 2:1 turndown, except by bleeding off excess gas, which would be wasteful. Supplementary Concerns for PSA/VPSA Supplementary concerns regarding on-site oxygen generation were: • • • •

Noise suppression during the venting and media regeneration of the adsorption vessels; Code implications of storage of low pressure oxygen gas in smoothing vessels underground; Venting of moist nitrogen-rich gases to atmosphere from deep within the structure of the building; and High maintenance effort needed for the valves and blowers that undergo hundreds of thousands of cycles every year.

Air Preparation For Croton WTP, air-fed ozonation offers relative simplicity, compactness and efficiency. To maximize process efficiency it is proposed that the ozone be generated at a concentration of approximately 3% by mass under maximum dose conditions. This concentration will fall with reducing ozone output because the design intent is to operate with constant output from the air compressors. Modern ozone generators are capable of operating up to 4% concentration with air as the feed gas. Like oxygen generators, there is an optimum concentration that will vary according to equipment capital and operating costs. Final design specifications will allow vendors a degree of flexibility to determine this optimum balance. A traditional large-scale air preparation system would normally consist of:

• • • • • • •

Air compression (oil-free); Cooling and condensate removal; Smoothing using an air receiver; Chilling to remove most of the moisture; Filtration and coalescing; Desiccant drying with temperature swing regeneration; and Final filtration.

During cold weather, and to save energy, it may be possible to eliminate the chilling stage if air is routed from outside the building. To achieve a dew point of better than minus 70°C (minus 95°F), it is necessary to use a combination of dryer media, such as activated alumina and molecular sieves. Temperature swing regeneration using external heating would involve periodic media change out, but this disadvantage is offset by the high operational cost penalty of using pressure swing regeneration. Using two 100% capacity air preparation streams per treatment train would reduce equipment capital cost and space requirements. However, to meet NYCDEP redundancy requirements and to improve turndown efficiency, three separate, but interconnected, 50% capacity streams of air preparation equipment are proposed for each treatment train. The interconnection would be such that an individual process element, such as a chiller, can be replaced by a common standby, or a complete replacement air preparation stream can be brought into service. The main disadvantage of using air as a feed gas, compared with oxygen, is that a higher operating cost results because of the lower efficiency of conversion of contained oxygen to ozone. Air-fed systems have also occasionally been problematic in maintaining a sufficiently low dew point in the feed gas to the generators, which has resulted in premature failure of dielectric tubes and corrosion problems within the generator. Proper design and operation of the air preparation system avoids these problems. The use of LOX eliminates these design and operational problems. Using LOX also simplifies the overall system, making for less space requirements and less maintenance. PSA/VPSA systems have higher maintenance requirements than air preparation systems. The benefits of using air-fed ozonation compared to PSA/VPSA are: • • • • • • • •

Equipment costs are lower or similar; Headroom and area requirements are lower; Electrical consumption at high turndown ratios are similar; Maintenance requirements are lower; Venting of waste gas is simpler; Gas leaks are not hazardous; Gas storage is not potentially hazardous; and Gas mixing in the contact zone is more effective.

Of these benefits, the first three were deciding factors for the Croton WTP design. Equipment Capital Cost Comparisons A comparison was made between Croton WTP ozonation requirements and the Southern Nevada Water Authority bid values for the River Mountains WTF (1), which used two 25 ton/day VPSA units to generate oxygen for ozone production. The costs are summarized in Table 2. TABLE 2: RIVER MOUNTAINS WTF BID RESULTS MAY 1997 Bid Item

River Mountains WTF Total Cost

Oxygen Separation System (based on two 25 ton/day VPSA units)

$1.95m

Ozone System / Destruct System

$1.98m

LOX System

$0.18m

Testing, Training, Start Up, etc

$0.11m

Total Equipment Capital Cost

$4.22m

Comparing the River Mountain bid results with Croton, the year 2000 approximate cost for a 6,000 lb/day ozone generation plant using four 25 ton/day VPSA units to generate oxygen would be about $6.7 million (see Table 3). A detailed comparison was not possible because a full equipment scope was not available. TABLE 3: ESTIMATED OXYGEN-FED OZONE SYSTEM COST FOR CROTON WTP BASED ON RIVER MOUNTAIN WTF BID PRICES Item

Cost

Oxygen Separation System (based on four 25 ton/d VPSA units)

$4.0 million

Ozone Generation / Destruct System

$2.0 million

Testing, Training, etc.

$0.1 million

Total Estimated Capital Cost

$6.1 million

Total Estimated Cost - Inflation Adjusted @ 3% for 3 years

$6.7 million

It is likely that the costs in Table 3 for an oxygen-fed ozone system using VPSA are conservatively low, because the bidding situation was very competitive. Budget prices supplied by an oxygen separation equipment vendor for the Croton project exceeded the figures shown in Table 3. The estimated equipment cost for an air-fed ozonation plant (equipment only) based on vendor budget prices is between $5.0 and $6.5 million. The costs included three streams of air preparation and ozone generation/destruct equipment for each of the two treatment trains. Therefore, within the limits of accuracy of the cost estimates, an oxygen-based ozonation plant would have approximately the same capital cost as an air-fed ozone plant. Space Requirement Comparisons Layout drawings were received from the oxygen-generation equipment vendors, and these were used to estimate the area required for each of the systems. Access space around the standardized vendor arrangement was added, and in instances near the compressors the cramped spaces allocated by the vendor were increased. It is likely that detailed layout to allow equipment removal would have increased these figures further. The resulting area and headroom was then used as the basis for comparison. The area developed is the summation of that required for air preparation/separation, ozone generation and ozone destruction systems. The space allocated for the air-fed system is based on the project layout drawings and is over-generous to allow easy equipment removal. A $500/sq. ft. cost factor takes into account the multi-floor average construction cost for the Croton project. With an at-grade facility this cost would be much lower. The estimated space and cost that were allocated for two trains of treatment (290 mgd) are shown in Table 4. The estimated costs take into account the extra space needed to accommodate headroom for the tallest vessels. TABLE 4: COMPARISON OF SPACE AND COST FOR FEED GAS PREPARATION AND OZONE EQUIPMENT System

Headroom Required

Area Required

Estimated Cost

PSA / Ozone

30 feet

15,400 sq. ft

$15 million

VPSA / Ozone

37 feet

28,200 sq. ft.

$28 million

Air-Fed Ozone

18 feet

18,400 sq. ft.

$9 million

Power Consumption Information was obtained from vendors relating to the electrical consumption at differing air/oxygen and ozone outputs. The information was consolidated and is presented in summarized form in Table 5. TABLE 5: ANNUALIZED ELECTRICAL COSTS System PSA / Ozone VPSA / Ozone Air-Fed Ozone

Max. Flow / Max. Dose 6,200 lb/d O3 1,756 kW $1,000,000 1,470 kW $837,000 2,285 kW $1,301,000

Average Flow / Average Dose 1,221 lb/d O3 398 kW $227,000 476 kW $271,000 508 kW $289,000

Min. Flow / Min. Dose 619 lb/d O3 232 kW $132,000 338 kW $192,000 322 kW $183,000

Note: Electricity cost estimated at 6.5 cents/kWh.

As explained earlier, the Croton WTP will be operated for the majority of its operational life at average flow and dose, which is only about 20% of maximum ozone system capacity. Under these conditions the VPSA equipment, and to a lesser extent the PSA equipment, would be operating at low efficiency. A similar penalty applies to air-fed ozonation, because the system would be operated at varying ozone-in-air concentration, and this shows clearly at minimum flow and dose conditions. From Table 5, the electrical cost savings at full output of oxygen-fed ozone generation compared to air-fed are very significant. However, at average conditions, these savings reduce dramatically to $18,000 per annum for VPSA and $62,000 per annum for PSA. In the case of minimum flow and dose conditions, VPSA is $11,000 more expensive to operate than air-fed ozonation, while PSA maintains a saving of $51,000 per annum. Total Cost of Ownership Table 6 illustrates the cost of ownership based on estimated equipment capital cost, building capital as represented by the surface area required, and the electrical operating costs at average conditions. The time period assumed for operating costs is 20 years. This is a simplified life cycle cost for comparative purposes.

TABLE 6: TOTAL COST OF OWNERSHIP System

Equipment Cost

Plan Area Cost

Electrical Cost (20 years)

Total Cost

PSA / Ozone

$6.7 m.

$15 m.

$4.5 m.

$26.2 m.

VPSA / Ozone

$6.7 m.

$28 m.

$5.4 m.

$40.1 m.

Air-Fed Ozone

$6.7 m.

$9 m.

$5.8 m.

$21.5 m.

Conclusions It can be seen from Table 6 that the lowest cost option ozone system is that using air as the feed gas. However, had the site been located at grade, the conclusion would have been very different. Equipment costs for oxygen generation could be reduced because LOX could be used and redundancy requirements could be reduced. The plan area penalty for PSA and VPSA would have reduced dramatically and the space cost per square foot would have more than halved. For a project built at grade the lowest cost option would have been PSA, followed by VPSA and then air preparation. Therefore, the Croton project can be viewed as a special case as regards the selection of air-fed in preference to oxygen-fed ozonation. References 1. "Ozone system implementation for the Southern Nevada Water Authority", by C. Bromley et al, [no source data]. Key Words Ozone, air preparation, oxygen generation, cost comparison, Croton, water treatment. Acknowledgements The authors would like to recognize the assistance provided by Fuji, Lotepro, OzoniaAir Liquide, PCI-Wedeco, and Praxair-Trailigaz during the preliminary design of the project.