PINEY POINT POND WATER REMEDIATION USING REVERSE OSMOSIS FINAL REPORT. Kenneth J. Jardine, William N. Futch and Dennis H.

Transcription

1

2 PINEY POINT POND WATER REMEDIATION USING REVERSE OSMOSIS FINAL REPORT Kenneth J. Jardine, William N. Futch and Dennis H. Michalski The Mosaic Co. (formerly IMC Phosphates) Vaughn V. Astley, Ph.D., (formerly IMC Phosphates, currently Dr. Phosphate, Inc.) Prepared for FLORIDA INSTITUTE OF PHOSPHATE RESEARCH 1855 West Main Street Bartow, Florida USA Project Manager: G. Michael Lloyd, Jr. FIPR Project Number: November 2005

3 DISCLAIMER The contents of this report are reproduced herein as received from the contractor. The report may have been edited as to format in conformance with the FIPR Style Manual. The opinions, findings and conclusions expressed herein are not necessarily those of the Florida Institute of Phosphate Research, nor does mention of company names or products constitute endorsement by the Florida Institute of Phosphate Research. 2005, Florida Institute of Phosphate Research

4 PERSPECTIVE While the abandonment of the Piney Point facility by its bankrupt owner, Mulberry Phosphates, has been about as bad an event as anything that could happen in the phosphate industry, it has provided a unique opportunity for full-scale testing of techniques to treat process water that could prove to be the answer to future process water problems both at operating plants and at plants that have been shut down. The problems at operating plants are not the same as those encountered at Piney Point, but it was still possible to learn a great deal from the experiences there. Reverse osmosis (R.O.) does offer an opportunity to produce very high-quality water that could be used for almost any purpose, including boiler feed water, in an operating phosphate complex. One problem that still needs to be resolved for Piney Point-type situations is what to do with the reject stream when the final volume of water is treated. Even in an operating plant, the question of how best to utilize the reject stream needs additional study. It is to be hoped that some of the ongoing studies will provide acceptable answers to these questions. G. Michael Lloyd, Jr. Research Director, Chemical Processing iii

5 ABSTRACT Phosphate plant pond water or process water can pose significant environmental risks and economic liabilities. Double liming has been the traditional treatment method for remediation of pond water. However, double liming suffers from several disadvantages, not the least of which are that the treated water often barely meets discharge criteria and that post-treatment air stripping is often required to attain acceptable ammonia concentrations. Reverse osmosis (R.O.) has long been viewed as an attractive alternative for pond water remediation. However, to be successful, the use of R.O. in the processing of pond water requires some form of pretreatment. In 2001, IMC developed an R.O. pretreatment process in anticipation of having to treat and discharge water from their Uncle Sam, Louisiana, plant. While water discharge at the Uncle Sam plant was ultimately not needed, the situation at Piney Point provided a unique opportunity to prove the process at a large scale, while aiding the State of Florida in closing the Piney Point gypsum stack. In early 2004 a large-scale plant, using IMC technology, was built at Piney Point, operated for 4½ months and produced approximately 46 million gallons of high quality water for discharge. Several process improvements were tested and implemented and real-world process economics determined. v

6 ACKNOWLEDGMENTS The authors acknowledge the participation of Mobile Process Technology of Memphis, Tennessee, for the design, construction and operation of the Piney Point demonstration plant. In particular, thanks are extended to Vann Raley, Director of Technical Services; Sean Kelly, Project Manager; and Guy Yoe, Field Service Manager. Thanks are also extended to Tony Lazanowski and our core group of highly skilled technicians: Flint Barnes, Ron Layne and Mike Melanson. vi

7 TABLE OF CONTENTS PERSPECTIVE... iii ABSTRACT...v ACKNOWLEDGMENTS... vi EXECUTIVE SUMMARY...1 INTRODUCTION...5 Background...5 Initial Process Development...6 Literature Review...8 Piney Point Project Chronology...8 PROCESS DESCRIPTION...11 Piney Point Demonstration Plant...13 RESULTS...17 Preliminary Batch Pretreatment and Small Scale R.O. Tests...17 Pilot Pretreatment and Extended Duration R.O. Tests...21 Piney Point Demonstration Plant Results...33 Demonstration Plant Construction...33 Demonstration Plant Start-Up and Phase II Milestone Attainment...36 Demonstration Plant Phase III Operation...41 Operating and Project Costs...48 CONCLUSIONS...53 REFERENCES...55 APPENDIX: U.S. Filter Treatability Test Report for Uncle Sam Pond Water vii

8 LIST OF FIGURES Figure Page 1. Block Flow Diagram of Standard Pretreatment Process and Two-Pass R. O. System Piney Point Pretreatment P&ID Piney Point Pretreatment P&ID Piney Point 2 nd Pass R.O. and Final Treatment P&ID Block Flow Diagram of Piney Point Pretreatment Pilot Plant Recovery and Flux Rate for December 11, 2003 R.O. Test Recovery and Flux Rate for December 22, 2003 R.O. Test Recovery and Flux Rate for January 13, 2004 R.O. Test Recovery and Flux Rates for January 19-23, 2004 R.O. Test Piney Point Demonstration Plant from South End Piney Point Demonstration Plant from North East End Piney Point Demonstration Plant Equipment Arrangement Florida Pines Permeate Production March Florida Pines Permeate Production February 19-June 30, viii

9 LIST OF TABLES Table Page 1. Phosphate Plant Pond Water Composition Sample Analyses for U.S. Filter Reverse Osmosis Test Preliminary Pretreatment Test Summary Preliminary 1 st Pass R.O. Test Summary Preliminary 2 nd Pass R.O. Test Summary Post-Treated R.O. Permeate Analysis Pretreatment Pilot Test Summary Dec. 8, R.O. Feed Analysis December 11, R.O. Summary Data, December 11, 2003 R.O. Test Pretreatment Pilot Test Summary Dec. 19, R.O. Summary Data, December 22, 2003 R.O. Test R.O. Summary Data, January 13, 2004 R.O. Test Pretreatment Pilot Test Summary Jan. 19, R.O. Summary Data, January 19-23, 2004 R.O. Test Piney Point Estimated Costs, Actual Costs and Accelerated Construction Related Costs Florida Pines Permeate Production February Florida Pines February 2004 Permeate Composite Analysis Florida Pines March 2004 Permeate Composite Analyses March Raw Material Consumption and Recovery April Estimated Raw Material Consumption and Recovery st Pass R.O. Operation Comparison for Dry Flocculant Test May Estimated Raw Material Consumption and Recovery June Estimated Raw Material Consumption and Recovery Florida Pines April-June Permeate Analyses Reject Sample Analyses Feed Pond Water Sample Analyses Operating and Project Cost Summary Projected Operating Costs...51 ix

10 EXECUTIVE SUMMARY Virtually all phosphate fertilizer plants have large ponds of circulating water that is recycled for use in the plants. The water in these ponds, known as pond water or process water, is an acidic mixture of many ions and compounds including weak phosphoric acid, sulfuric acid, hydrofluoric acid and hydrofluosilicic acid, with various dissolved metallic and non-metallic ions derived from the processing of phosphate rock. In operating phosphate plants a balance will normally exist between evaporation from these ponds and water input to the ponds, such that treatment and discharge of the pond water is not necessary. However, during extended periods of heavy rain or when a plant is shut down, as in the case of Piney Point, this acidic water must be treated and discharged. The traditional method of treating pond water has been lime treatment in two stages (double liming). However, double liming suffers from several disadvantages. Large settling ponds are required for the second stage solids separation. Only about 50-60% of the water ends up being discharged and this water often barely meets discharge criteria. Also, even after lime treatment air stripping is often required to reduce the ammonia concentration to acceptable levels. In January 2001, the Mulberry Corporation informed the Florida Department of Environmental Protection that financial difficulties would prevent them from providing environmental security at their Piney Point and Mulberry phosphate plants. Mulberry Corporation later abandoned the sites and the Florida DEP assumed financial responsibility for stabilizing the sites and permanently closing the phosphogypsum stacks. In May 2001 a court appointed receiver was charged with ensuring the environmental security of the sites. At about the same time IMC Phosphates, anticipating the possibility of having to treat and discharge pond water from their Uncle Sam Louisiana plant, began work on a chemical pretreatment process that would allow the pretreated pond water to be processed through reverse osmosis membranes for separation into a purified water stream (permeate) and a concentrated pond water stream (reject). Reverse osmosis had long been viewed as an attractive alternative for pond water remediation and several attempts have been made to use this technology. However, because pond water is normally a saturated solution, the early attempts usually resulted in irreversible fouling of the R.O. membranes. Also, even when the pond water was from a closed plant where rainwater had diluted it below saturation levels, an R.O. treatment could only remove a relatively small amount of purified water from the pond water before it again becomes a saturated solution. By late summer of 2001, the IMC research had resulted in the development of a process that laboratory and small batch tests indicated was capable of adequately pretreating pond water for R.O. processing. By November of 2001, however, the probability of having to treat and discharge pond water from the IMC Uncle Sam plant was no longer likely and work on the Uncle Sam project was suspended. Since the IMC pond water R.O. work had only been performed at a relatively small scale and since larger scale tests would be necessary to verify the process and establish acceptable cost estimates, IMC started discussions, in January 2002, with the 1

11 newly formed FIPR Process Water Technical Advisory Committee on the possibility of funding a pilot scale pond water R.O. treatment project. While the Piney Point pond water was significantly different from the Uncle Sam pond water, due to several years of rain water dilution, Piney Point provided a unique opportunity to prove the process at a large scale, while aiding the State of Florida in closing the Piney Point gypsum stack. During the remainder of 2002 and the first half of 2003 IMC conducted test work specifically on R.O. processing of the Piney Point pond water and continued negotiations with FIPR for partial funding of the Piney Point project. During this time period IMC also conducted discussions with various third party companies that could assist with the actual operation of the Piney Point plant and eventually negotiated an agreement with Mobile Process Technology of Memphis Tennessee to manage and operate the plant. On October 7, 2003, a contract was finally signed between the court-appointed receiver and Florida Pines (a joint venture between Pure Water Technology Holding Company, a subsidiary of IMC, and Mobile Process Technology) for water treatment at Piney Point. The contract specified a Phase I pilot program in which Florida Pines had to produce 5000 gallons of pretreated pond water suitable for R.O. processing. This was to be followed, after the completion and commissioning of the full scale plant, by a Phase II segment, during which Florida Pines had to produce and discharge 400,000 gallons per day of R.O. permeate for five consecutive days. Operation would then move to Phase III, for a total operating time of 105 days, or until June 30, At this point a contract was also in place for FIPR to supply $900,000 in capital to assist in construction of the water treatment plant that was to be reimbursed at the rate of $16.33 per thousand gallons of water discharged. The agreement with FIPR was later renegotiated such that reimbursement to FIPR amounted to approximately one third of the original $900,000. From the onset, the construction schedule was extremely aggressive and various factors caused delays. Of significant impact was that the initial plan was to use a slip stream from the existing liming facilities to provide a feed for the IMC pretreatment process. However, as a result of the barging operations, the first stage liming facilities would not be available for our use. In addition, the resultant lack of a large clarifier resulted in many process and equipment changes necessary to accommodate the significantly revised process and the clarification had now to be carried out in tanks. While some schedule slippage was due to these process changes resulting from simultaneous piloting of the pretreatment process the most significant delays were caused by vendors not being able to deliver some of the larger tanks when promised. While the original contract called for the plant to be operational by the end of December 2003, the plant did not start up until February 19, The first step of the pretreatment process used at Piney Point consisted of sulfuric acid addition to the pond water to a ph of approximately 2.5. This was done to kill algae and allow it to be removed during subsequent clarification. Caustic (NaOH solution) was then added to the pond water to raise the ph to about 7.0 and precipitate the calcium in the pond water in the form of hydroxyapatite. A flocculant was next added to the pond water slurry and it was sent to clarifiers for sludge separation. One of the compromises made at the Piney Point plant, due to space and time limitations, was the use of four 2

12 clarifiers operating in parallel for sludge separation. The clarifier overflow was then collected in the R.O. feed tank, where sulfuric acid was again added to reduce the ph to This step was necessary to increase the solubility of the calcium and other ions remaining in the pond water to keep them in solution when the water was processed through the 1 st pass R.O. unit. The water was then sent to the 1 st pass R.O., operating at about 400 psi, for separation into a reject stream, that was returned to the Piney Point pond system, and a permeate stream. The 1 st pass R.O. unit removed the majority of all contaminants from the pond water with the exception of fluorides. Fluorides were not removed in the 1 st pass because their ionic size, at acid ph, is very close to that of water. Permeate from the 1 st pass R.O. was then treated with a small amount of caustic ( lbs. NaOH/1000 gallons) to raise the ph to about 7.5 to 8.0 and fed to the 2 nd pass R.O. In the 2 nd pass R.O. fluorides are removed from the water along with most of the remaining contaminants that pass through the 1 st pass R.O. Initially, reject from the 2 nd pass R.O. was returned to the Piney Point pond system. Later it was recycled to the 1 st pass R.O. to increase overall recovery. With the exceptions of ammonia nitrogen and (occasionally) ph, permeate from the 2 nd pass R.O. met and generally exceeded all requirements for discharge. To remove the last traces of ammonia nitrogen the 2 nd passpermeate was processed through an ion exchange water softening resin. For final ph adjustment a very small amount (<1%) of well water was added to the permeate stream. From initial start-up, on February 19, 2004 until expiration of the water treatment contract on June 30, 2004, the Piney Point demonstration plant produced and discharged 46,015,799 gallons of water, with over 99.7% of the water within contract specifications (ph ; F <5 mg/l as ion; Total NH 3 <1 mg/l; Total N <2mg/L; P <0.5 mg/l; TDS <50 mg/l). The average production rate, over the entire operating period, was 368,000 gallons per day with an overall permeate recovery of 43.3% and an operating factor of 91.2%. Several process improvements and equipment up-grades were instituted during the operation period to improve permeate recovery and increase capacity. Taken as a whole the improvements increased permeate recovery from 41.7% to 44.6% and increased average permeate production from 314,000 to 412,000 gallons per day. Operating costs for the demonstration plant, during the 4 ½ months of operation, were $17.71 per thousand gallons of permeate discharged and projected future operating costs, taking into account the equipment up-grades and process improvements, were $14.81 per thousand gallons of permeate. While even the projected future R.O. operating cost for Piney Point is about twice the cost of double liming (estimated at $7.47/1000 gallons discharged), it is highly probable that operating costs could be reduced further, to about $8/1000 gallons permeate, with additional refinements to the pretreatment chemistry, adequate clarification area and the use of 600 psi 1 st pass R.O. units. Overall, the Piney Point project demonstrated that, with adequate pretreatment, R.O. based pond water treatment is possible with reasonable purified water recovery. The primary strength of the process in the high quality of water produced. The primary weakness of the process is higher cost than double liming. However, with future development, it is probable that the cost can be reduced. 3

13 INTRODUCTION BACKGROUND Phosphate plant pond water or process water is an acidic mixture of many ions and compounds. In an operating wet process phosphoric acid plant, the pond water is used for washing the calcium sulfate (gypsum) filter cake produced as a result of phosphoric acid production, cooling and scrubbing process vapors, conveying the gypsum to the stacking area and other purposes that do not require fresh water. The major acidic components of pond water are phosphoric acid, sulfuric acid, hydrofluorosilicic acid (H 2 SiF 6 ), and hydrofluoric acid (HF). Table 1 shows the major chemical components found in pond water and their concentration ranges. Table 1. Phosphate Plant Pond Water Composition. Chemical Component P SO 4 F Si (ammoniacal) N Na Mg Ca K Fe Al Cl Concentration Range ,000 ppm ppm ,000 ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm Since, in an operating phosphoric plant, the pond water is continuously recycled and used for cooling, the heat input and resulting evaporation is normally sufficient to balance the fresh water additions to the pond (rain water and other fresh water inputs from the plant). Thus remediation and discharge of the pond water is normally not required. However, there are circumstances under which treatment and discharge of the contaminated pond water is necessary. One such circumstance could be an extended period of abnormally heavy rainfall. Another such circumstance would be when the phosphoric acid plant has ceased operation ether for an extended period of time or permanently. 5

14 INITIAL PROCESS DEVELOPMENT In 1999 the IMC phosphoric acid plant in Uncle Sam, Louisiana suspended operations due to market conditions. While the suspension in operations was intended to be temporary, subsequent heavy rainfall along with reduced evaporation due to the lack of heat input to the pond made it evident that pond water remediation and discharge might soon be required. At the time the standard industry method of treating pond water for discharge was double liming. This method consists of adding a calcium compound (such as CaCO 3, Ca(OH) 2 or CaO) to the pond water, in two stages, such that the phosphate and other impurities form solid precipitates that settle and are separated from the liquid phase. During the first stage lime treatment, to a ph of about 5.5, fluorides are precipitated as CaF 2 and/or CaSiF 6. During this stage it is also thought that the hydrofluorosilicic acid present dissociates to HF and SiF 4, with the SiF 4 hydrolyzing to HF and SiO 2. Some phosphate is also precipitated at this stage as Ca 3 (PO 4 ) 2 (tri-calcium phosphate). The sludge (CaF 2, CaSiF 6, Ca3(PO 4 ) 2 and other compounds) produced at this stage is a granular, crystalline material that settles fairly rapidly and can be de-watered to about 30% solids in a gravity thickener. The sludge can be sent to the plant gypsum stack or recycled to the phosphoric acid plant, if operating, for recovery of the phosphorus. In the second stage, additional lime is added to the clear liquid from the first stage to a ph of about In this stage the remaining phosphates and fluorides are precipitated along with some sulfate and many of the metals. The sludge in this stage has poor settling and thickening properties, due to the hydroxide nature of many of the compounds, and rarely achieves more than 5%-7% solids by weight. The sludge from this stage is normally deposited in large lagoons to allow for additional de-watering. Typically, however, the clarified water from the 2 nd liming stage is still not capable of being released to the environment due to a high conductivity and the presence of ammonia. Thus, common practice is to airstrip the ammonia by spraying the water into the air and allowing it to fall back into the ponds. This process is slow and results in the ammonia being released into the environment. The water from the spray ponds is then blended with well water to reduce the conductivity before discharge. Overall, the quantity of water suitable for discharge from a double liming process is about 50%-70% of the feed volume. Reverse osmosis had been considered as an attractive alternative for pond water remediation and several attempts had been made to use this technology. However, because pond water is a saturated solution, the early attempts almost always resulted in relatively rapid, irreversible fouling of the R.O. membranes. Also, even when the pond water was from a closed plant, where rainwater had diluted it below saturation levels, an R.O. system could only remove a relatively small amount of purified water from the pond water before it again became a saturated solution. Thus, it was recognized that successful R.O. treatment of pond water would require some form of pretreatment. The Technology Development Group of IMC Phosphates began work on a pretreatment process that would allow pond water to be fed to an R.O. in the last quarter of The pretreatment research concentrated on removing only those ions that would cause solids precipitation in the pond water as it was concentrated and on keeping as 6

15 much of the phosphate as possible in solution. By about May of 2001 a potential pretreatment process had been developed and U.S. Filter was contracted to run several pilot scale R.O. tests with pretreated Uncle Sam pond water. Two 110 gallon samples of pretreated Uncle Sam pond water were prepared in May of 2001 and shipped to the U.S. Filter laboratories in Warrendale PA for the R.O. tests. The R.O. system used for the tests was a small pilot scale unit that included a 5-micron cartridge filter, a 5-HP high pressure positive displacement pump and a single seawater element (Filmtec SW 2530). The test procedure consisted of first passing the pretreated pond water through a sand filter. The filtered pond water was then fed through the R.O. system in a first pass. The ph of the first pass R.O. permeate was then adjusted to 9.0 with sodium hydroxide and this material fed through the (same) R.O. system in a second pass. The primary objective of the second pass is to reduce the fluoride concentration in the final permeate. The U.S. Filter tests demonstrated first pass permeate recoveries as high as 85% at 850 psig feed pressure while maintaining flux rates greater than 8 GFD (gallons/day/square foot of membrane area) and second pass recoveries as high as 95% at 150 psig feed pressure while maintaining flux rates greater than 20 GFD. The complete U.S. Filter test report is included as Appendix A. Table 2 shows the sample analyses for the U.S. Filter test with one of the pretreated Uncle Sam pond water samples. Table 2. Sample Analyses for U.S. Filter Reverse Osmosis Test. Parameter Raw Pond Water R.O. 1 st Pass Reject R.O. 2 nd Pass Permeate P 2 O % 5.65% 0.23 ppm CaO 0.167% 0.060% 0.1 ppm F 0.470% 0.038% 0.83 ppm SiO % 0.079% 1.0 ppm Na 2 O 0.310% 2.85% 8.1 ppm SO % 2.83% 12 ppm NH 3 - N 0.200% ppm Conductivity (µmho/cm) 19, At the conclusion of this work, preliminary design work was started on a 2MM GPD reverse-osmosis-based pond water treatment plant for the Uncle Sam, Louisiana, site. This work included developing a mass balance, estimating operating costs, developing an equipment list and estimating capital costs. This work was later suspended when normal production operations resumed at the Uncle Sam site and water balance modeling projections indicated there would be no need to treat and release pond water. However, IMC continued R.O. pretreatment research work with pond water samples from various other locations, with both higher and lower concentrations of dissolved ions. This additional work led to a better understanding of phosphate plant pond water chemistry and produced several additional pieces of information. All of this work resulted in the filing of two U.S. patent applications during February 2002 for the pond water pretreatment technology and the issue of the two patents (US 6,758,976 and 6,758,977) on July 6,

16 LITERATURE REVIEW The majority of technical literature relating to the phosphate industry gives little attention to the pond or process water system. Furthermore, when pond water is mentioned it is usually in the context of gypsum disposal (Becker 1989). Mooney, Nogueira and Thompson (1977) performed a laboratory and pilot study of the double liming process in which they identified the precipitating compounds and ph relationships for the two-stage process. This work provided some useful information in the development of our R.O. pretreatment process in terms of ph values at which fluorides have a minimum solubility in pond water. Other relevant references relating to pond water treatment or remediation consist mainly of patents concerned with variations of, or improvements to the conventional double liming process (Baumann and Bird 1970; Legal 1973; Murray and Weston 1992). PINEY POINT PROJECT CHRONOLOGY IMC began independent test work with the Piney Point pond water in January At this time the ph of the pond water sample provided was 2.86, with a fluoride concentration of 455 ppm and a calcium concentration of 582 ppm. After some initial screening tests a pretreatment process was selected and a larger sample of pond water was obtained for large-scale batch pretreatment tests and small-scale pilot R.O. tests. These larger scale batch pretreatment tests and pilot R.O. tests were conducted from July through December of Approximately 900 gallons of Piney Point pond water were pretreated and approximately 800 gallons of the pretreated water were processed through a single membrane, pilot scale, 1 st pass R.O. system. At an average feed pressure of 600 psig permeate recovery was 76% at an average flux rate of 15.4 GFD. Average R.O. feed conductivity was about 9000 µmho/cm and average permeate conductivity was 415 µmho/cm. Approximately 310 gallons of the 1 st pass R.O. permeate were then used for 2 nd pass R.O. tests. These tests were run at several feed ph values and at pressures from 200 psig to 350 psig. These tests indicated that, at optimum conditions, permeate recoveries approaching 95% could be obtained, at flux rates greater than 14 GFD. At the conclusion of this test work we could reliably produce 2 nd pass permeate with a P 2 O 5 concentration less than 0.5 mg/l and a fluoride concentrations less than 1 mg/l. However, the ammonia nitrogen concentration was still between 5 and 10 mg/l. Thus, a final test was run with about 55 gallons of the 2 nd pass permeate and a water softening resin. This test indicated that it was possible to lower the ammonia nitrogen concentration in the 2 nd pass permeate to less than 0.1 mg/l. After the conclusion of the above pilot scale work, IMC independently started some engineering work on an 800,000 GPD reverse-osmosis-based pond water treatment plant for the Piney Point site. This work consisted of developing flow sheets, P&ID s, operating and capital cost estimates and specific equipment lists, with long lead time items identified. At this time IMC also purchased four skid mounted R.O. modules for potential use at Piney Point. One of the novel features considered during the preliminary 8

17 design, both to reduce costs and speed construction time, was the use of some of the existing double liming equipment at the Piney Point site. IMC approached the newly formed FIPR Process Water Technical Advisory Committee in January 2002 on the possibility of funding a pilot scale pond water R.O. treatment project. During the next several months discussions continued with FIPR and FDEP and the scope of the project expanded from pilot scale to large demonstration scale (initially ~800,000 GPD). A financial commitment was obtained from FIPR in May 2002 and discussions began with the Piney Point receiver. At this time detailed engineering for the plant also started. During the remainder of 2002 and into the first half of 2003 discussions continued with FDEP and the Piney Point receiver while IMC explored various options to limit liability by forming a subsidiary that would joint venture with an operator. Several operating partners for the Piney Point project were explored, but by July 2003 Mobile Process Technology of Memphis Tennessee had been selected to manage and operate the plant as a joint venture partner with IMC under the name of Florida Pines LLC. The final contract, with the Piney Point receiver, for water treatment services was signed on October 7, Florida Pines started operation at Piney Point on November 6, 2003 with a 20 gpm pretreatment pilot plant. The purpose of the pilot plant was to fulfill a contractual obligation of producing 5000 gallons of pretreated pond water containing low enough fluoride and calcium concentrations to be deemed acceptable for feeding to a reverse osmosis system. This contractual obligation was achieved on November 20. Pretreatment pilot operations continued through January of 2004 to fine tune pretreatment process parameters for the full scale plant. This work was needed due to the significant changes that had occurred in the chemical composition of the Piney Point pond water since the initial testing that was conducted in 2002, and due to the lack of availability of the liming station together with its associated clarification equipment. At this time it was also determined that the full size skid mounted R.O. modules that were to be used for the 1 st pass were only capable of operating at a maximum feed pressure of 400 psig, rather than the 600 psig originally anticipated. Starting December 8, 2003, large samples of the pretreated pond water from the pilot plant were transported to the IMC Technology Development laboratory near Mulberry FL for processing through a single element pilot scale reverse osmosis system. This system was set up as the 1 st pass R.O. The objectives of this work were to obtain extended operating time with the R.O. to check for potential membrane fouling and to determine the production penalties that would result from operating the 1 st pass R.O. at a feed pressure of 400 psig rather than 600 psig. The pilot R.O was operated for approximately 10 days, with no evidence of irreversible membrane fouling. Water quality from the pilot scale R.O. was indeed excellent; with an average 1 st pass permeate conductivity of less than 515 µmho/cm. Concurrently with the pilot operations design and construction of the full scale plant was in progress. As a result of the pilot testing several pretreatment process changes were needed to accommodate the different pond water composition (compared to 9

18 2002). Thus, some design changes were required in the full scale plant causing some construction delays. From initial start-up on February 19, 2004, until contractual shut down on June 30, 2004, the Florida Pines reverse-osmosis-based water treatment plant produced million gallons of permeate, of which million gallons (99.74%) met specifications for discharge. All off-spec water was produced during the first two days of operation, with 67,500 gallons from February 19 being outside ph limits and 53,500 gallons from February 20 being discounted due an instrument malfunction. Permeate from several other days was initially contested as being out of specification in terms of the ammonia concentration. However, subsequent analyses of back-up 24 hour composite samples by an EPA accredited laboratory indicated that the ammonia concentration in all disputed samples was within the contract limit. Of the 133 days between February 19 and June 30, the Piney Point plant was scheduled to operate 124 days. Nine days of scheduled down time were taken from April 11 through April 19 to upgrade some piping and several pumps to permit increased production in the 1 st pass R.O. modules. During the 124 days of scheduled production, the operating factor of the plant was 92% with an average daily permeate production of 402.6M gallons, while operating. Average daily production after the retrofit was 412M GPD. 10

19 PROCESS DESCRIPTION The primary purpose of pretreating pond water before processing through a reverse osmosis system is to avoid solids precipitation and scale formation on and within the R.O. membranes. Previous attempts at processing pond water through a reverse osmosis system without pretreatment resulted in irreversible fouling of the membranes within hours. Phosphate plant pond water is normally saturated with respect to calcium, fluorine and silica and they are the primary components that must be removed during pretreatment. The general pretreatment methodology consists of first adding a calcium compound to the pond water to form insoluble calcium fluoride (CaF 2 ). The calcium compound is typically quick lime (CaO) or hydrated lime (Ca(OH) 2 ) and it is added as a 10% slurry to promote a more complete reaction. The quantity of calcium added can be as low as 100% of the stoichiometric quantity required by the fluoride in the pond water (neglecting the calcium initially present), but economics favor a higher calcium addition rate within the range of 110% to 130% of the stoichiometric requirement. Since a small amount of calcium phosphates also start precipitating and re-dissolving at these calcium addition levels, the solution ph is unstable and should not be used for process control. The above step assumes the initial calcium concentration in the pond water is relatively low; less than about 15% of the stoichiometric requirement (for CaF 2 ) of the fluoride in the pond water. If the initial calcium concentration of the pond water is greater than or equal to about 120% of the stoichiometric requirement of the fluoride in the pond water calcium addition may not be required (although may still be desirable for economic reasons). At initial calcium concentrations between these two levels, the calcium added is reduced by an amount proportional to the calcium initially present in the pond water. Next, a second base is added to pond water slurry to raise the ph to the value of minimum fluoride solubility. For normal pond water (with a low Ca to F ratio) this ph is approximately 5.7. For pond water with a high initial calcium concentration (relative to the fluoride), this ph can be as high as 6.4. Since it is desirable to keep as much of the P 2 O 5 as possible in solution, the second base is selected such that the phosphate salts thereof remain soluble. Typical bases are sodium hydroxide (NaOH) or ammonia (NH 3 ), with economics greatly favoring ammonia. However, when it is desirable to minimize the ammonia concentration of the final permeate from the reverse osmosis system, sodium hydroxide is preferred. The pond water slurry is then transferred to a clarifier. Sixty minutes settling time is generally sufficient, although 90 minutes is preferred. The sludge produced in the clarifier is about 8.5% or less of the feed volume at about 40% solids. The overflow from the clarifier is directed to a lined aging pond of sufficient size to provide 36 hours aging time. The purpose of the aging is to allow the silicic acid present in the liquid to decompose into hydrated silicon dioxide. 11

20 The pond water is now treated with a flocculant to remove the hydrated silicon dioxide. The best results have been obtained using a two stage flocculant system. In the first stage a cationic flocculant is added to the pond water at a dosage of about 150 ppm. Next, a collector flocculant is added at a dosage of about 8 ppm. The pond water is then directed to a settler for separation of the silica sludge. Preferably the underflow from the silica settler is centrifuged. This significantly increases the volume of pond water that can be fed forward and provides the opportunity for recovering the silica as a useful byproduct. If the initial silica concentration in the pond water is sufficiently low (~200 ppm as SiO 2 ), the aging/hydration step and the use of a flocculant for silica removal are unnecessary. In such cases the pond water can be fed directly from the sludge clarifier to the acidification step below. At this point the pond water is still saturated with respect to Ca, F and possibly SiO 2. Thus, a small quantity of sulfuric acid is added to the pond water to increase the solubility of the ions remaining in solution. Typically about 160 ppm of sulfuric acid is sufficient to decrease the ph to 3. This ph is acceptable for the R.O. membranes and is low enough to prevent precipitate formation in the R.O. reject (concentrate) stream. The pond water can now be processed through an R.O. system. To achieve acceptable fluoride concentrations in the permeate a two pass R.O. system is used. In this system the ph of the permeate from the first pass is raised to about 8-9 (via sodium hydroxide addition) and then fed to the second pass. Figure 1 shows a block flow diagram of the general pretreatment process and the two-pass R.O. system. Lime Slurry 50% NaOH Pond Water Lime Addition Caustic Addition Clarification P. W. Aging Solids 50% NaOH 1 st Stage Permeate H 2 SO 4 2 nd Pass R.O. ph Adjustment 1 st Pass R.O. ph Adjustment Hydrated Silica Separation Final Permeate Reject Reject Silica Sludge Figure 1. Block Flow Diagram of Standard Pretreatment Process and Two-Pass R. O. System. 12

21 While the above process would be applicable to pond water from most operating or recently closed phosphoric acid plants, the pond water at Piney Point presented a special case. At the start of the initial test work with the Piney Point pond water in 2002, it was evident, due to the low SiO 2 concentration in the pond water, that the hydrated silica removal step would not be required. Later, at the start of pilot pretreatment tests at Piney Point, it was determined that the fluoride concentration in the pond water had decreased significantly. Thus, lime addition would no longer be needed. However, since the ph of the pond water had also increased from an initial value of 2.86 to more than 3.4, alga growth had accelerated and a control method was needed. As a result of the on-site pilot tests the pretreatment process was modified and consisted of the following steps: (1) Sulfuric acid was added to reduce the ph to about 2.5. This effectively killed the algae. (2) Caustic was then added to increase the ph to about 7.0 and precipitate hydroxyapatite. This reduced the calcium concentration to less than 100 mg/l. (3) Flocculant was added to aid in solids settling and the solution was sent to a clarifier for solids separation. (4) The ph of the clarifier overflow was reduced to about 2.5 again by the addition of more sulfuric acid. This increased the solubility of the calcium remaining in solution and stabilized the pond water with respect to further precipitation. At this point, the pretreated pond water would be fed to the 1 st pass R.O. unit. PINEY POINT DEMONSTRATION PLANT As mentioned above, during the preliminary design phases for the Piney Point R.O. plant, use of some of the existing double liming equipment was contemplated. The main equipment that was to be used was one of the lime addition tanks and one of the large in-ground clarifiers. However, as the project progressed, it was determined that this equipment would not be available, since both clarifiers were being used for double liming. Thus, design changes were required in the demonstration plant to provide clarifiers for the pond water and hydroxyapatite solids. Also, since time and budgetary constraints precluded the field fabrication of an adequately sized single clarifier, four small clarifiers, operating in parallel, were used for the solids separation. Two of these tanks were flat bottom tanks that had previously been intended for the lime addition function. The other two tanks were cone bottom tanks purchased specifically as clarifiers. Unfortunately, all four smaller clarifiers provided only about 16% of the settling area of the single in-ground clarifier. Thus, based on the pilot plant data, it was anticipated that overall, sustainable, plant rates would be limited to about 1050 GPM feed to the clarifiers. Figures 2, 3 and 4 show piping and instrument diagrams for the R.O. demonstration plant as originally built at Piney Point. 13

22 Figure 2. Piney Point Pretreatment P&ID. 14

23 Figure 3. Piney Point 1 st Pass R.O. System P&ID. 15

24 Figure 4. Piney Point 2 nd Pass R.O. and Final Treatment P&ID. 16

25 RESULTS PRELIMINARY BATCH PRETREATMENT AND SMALL-SCALE R.O. TESTS The initial batch pretreatment and small-scale R.O. tests conducted with the Piney Point pond water in early 2002, were used primarily to determine the overall feasibility of the project and to develop initial cost estimates. Thus, the results of these tests will not be covered in any significant detail. Tables 3 and 4 show the average results from all pretreatment tests and all 1 st pass R.O. tests respectively. Most of these tests were run at similar conditions. Thus the averages present a valid picture of the results. Table 5 shows the results for the 2 nd pass R.O. tests. For the 2 nd pass R.O. tests conditions were varied significantly from one test to the next, mainly to determine the minimum ph needed for effective fluoride removal. Thus Table 5 contains a column with the range of conditions and results in addition to the averages. One factor worth noting with regard to the 1 st pass R.O. tests is that all the preliminary first pass R.O. tests were run at a membrane feed pressure of 600 psig to match the maximum allowable pressure for the brackish water membranes used. Later 1 st pass R.O. tests were run at a membrane feed pressure of 400 psig, to match the maximum allowable pressure of the large-scale R.O. modules that were used at the Piney Point demonstration plant. Based on the data, for the 600 psig and 400 psig test, it appears that the higher pressure had the effect of boosting the permeate recovery by about 10% and the permeate flux rate by about 5-6 gallons per day per square foot of membrane area (GFD). Since it was known, based on these preliminary tests and the earlier U.S. Filter R.O. tests, that even a two pass R.O. process could not reduce the nitrogen concentration (mainly NH 3 nitrogen) to the level required for discharge at Piney Point, a quick and dirty test was run with 55 gallons of the 2 nd pass permeate from the above tests and a water softener ion exchange resin. The unit used was a conventional, household type unit, containing about 1 ft 3 of resin and the permeate was pumped through the unit at about 1.5 GPM. The analysis of the final permeate is shown in Table 6. 17

26 Table 3. Preliminary Pre-Treatment Test Summary. Initial Pond Water Lime Addition Number of Batch Tests 16 Total Volume P.W. Treated (liters) 3396 P 2 O 5 Concentration (mg/l) 2947 Ca Concentration (mg/l) 554 F Concentration (mg/l) 154 SiO 2 Concentration (mg/l) 184 NH 3 -N Concentration (mg/l) 598 SO 4 Concentration (mg/l) 4696 Na Concentration (mg/l) 1077 Average Lime Addition (lbs. CaO/10 3 Gal. P.W.) 2.59 Average ph After Lime Addition 5.20 Caustic Addition (1) Average Caustic Addition (lbs 100% NaOH/10 3 Gal P.W.) Average ph After Caustic Addition 6.42 Sludge Separation Average Flocculent Dosage (ppm) 3.2 Average Settling Time (Hr) 1.5 Average Liquid Recovery (% V/V) 89.3% ph Adjustment Average Sulfuric Acid Addition (lbs. 100% H 2 SO 4 /10 3 Gal. P.W.) 8.15 Average Final ph 3.01 Average Pretreated Water Analysis P 2 O 5 (mg/l) 2111 Ca (mg/l) 151 F (mg/l) 81 SiO 2 (mg/l) 184 NH 3 -N (mg/l) 644 SO 4 (mg/l) 6543 Na (mg/l) 1655 As (mg/l) 0.42 (1) Two tests that used anhydrous ammonia in place of caustic are not included in average. 18

27 Table 4. Preliminary 1 st Pass R.O. Test Summary. Number of Tests 1 12 Total R.O. Operating Time (Hr:Min) 78:37 Average Feed Pressure (psig) 599 Average Permeate Recovery (% V/V) 79.0% Average Permeate Flux Rate (GFD) Average Permeate Conductivity (µmho/cm) 310 Average Permeate Analysis P 2 O 5 (mg/l) 15.3 Ca (mg/l) 0.8 F (mg/l) 44.3 SiO 2 (mg/l) 2.2 NH 3 -N (mg/l) 14.6 SO 4 (mg/l) 55.3 Na (mg/l) 16.6 As (mg/l) 0.01 Average Reject Analysis P 2 O 5 (mg/l) 8885 Ca (mg/l) 574 F (mg/l) 309 SiO 2 (mg/l) 671 NH 3 -N (mg/l) 2891 SO 4 (mg/l) Na (mg/l) 6770 As (mg/l) 1.61 (1) Two additional tests run with physically damaged membrane. Data not included. (2) Gallons per day per square foot of membrane area. 19

28 Table 5. Preliminary 2 nd Pass R.O. Test Summary. Number of Tests Total R.O. Operating Time (Hr:Min) 8 38:06 Average Range Lbs 100% NaOH/10 3 Gal. 1 st Pass Permeate Feed ph Feed Pressure (psig) Permeate Recovery (% V/V) 93.7% 89% - 96% Permeate Flux Rate (GFD) Permeate Conductivity (µmho/cm) Permeate Analysis P 2 O 5 (mg/l) Ca (mg/l) F (mg/l) SiO 2 (mg/l) NH 3 -N (mg/l) SO 4 (mg/l) Na (mg/l) As (mg/l) <MDL <MDL Reject Analysis P 2 O 5 (mg/l) Ca (mg/l) F (mg/l) SiO 2 (mg/l) NH 3 -N (mg/l) SO 4 (mg/l) Na (mg/l) As (mg/l) (1) Gallons per day per square foot of membrane area. 20

29 Table 6. Post-Treated R.O. Permeate Analysis. Component Result Units Method Method MDL Ammonia-Nitrogen 0.08 mg/l Arsenic mg/l Chloride 2 mg/l Chromium mg/l Copper mg/l Dissolved Calcium 0.1 mg/l Dissolved Magnesium 0.1 mg/l Dissolved Potassium mg/l Dissolved Sodium 6.5 mg/l Fluoride 2.95 mg/l Gross Alpha 1 +/- 5.4 pci/l Iron mg/l Lead mg/l Nitrate-Nitrite 0.06 mg/l ph 8.89 su Phosphorus 0.27 mg/l Selenium mg/l Silica (SiO2) 0.3 mg/l Specific Conductance 38 µmho/cm Sulfate 1 mg/l Total Dissolved Solids 30 mg/l PILOT PRETREATMENT AND EXTENDED DURATION R.O. TESTS The original pretreatment pilot plant installed at Piney Point was designed for sulfuric acid addition, lime slurry addition, caustic addition, solids settling and removal and sulfuric acid addition to the clarified pretreated pond water. It consisted of six flatbottom tanks, in series, with recirculation pumps, to provide agitation in three of them, along with several metering pumps and a final pump to return the pretreated pond water back to the Piney Point pond system. After several tests, in which it was determined that lime addition was not needed and that the settling tank was oversized, two of the tanks were removed and replaced with a smaller cone bottom settling tank with a very small flocculant mix tank just ahead of it. Figure 5 shows a block flow diagram of the pretreatment pilot plant in the final configuration. 21

30 H 2 SO 4 Raw Pond Water Tank 1 H 2 SO 4 Mix Tank Circulation, Mixing Loop (Typ.) Tank 2 Surge Tank (Former CaO Addition) 25% NaOH Tank 3 Caustic Mix Tank Pretreated Pond Water Return Tank 6 Final ph Adjustment Mix Tank H 2 SO Tank 4 Sludge Settler Sludge Solids Return 0.5% Flocculant Solution Floc. Addition Tank This tank equipped with external mixer Solids Recycle Figure 5. Block Flow Diagram of Piney Point Pretreatment Pilot Plant. Pilot plant operation at Piney Point commenced on November 6, 2003, and continued until January 19, As with most pilot operations, there was considerable refinement of equipment and process during the early tests. However, by December 8, 2003, the process configuration had been finalized and control was judged adequate to justify collecting a large sample of the pretreated water and processing it through our small-scale (~0.15 GPM) pilot R.O. system. Table 7 shows summary data for the December 8 pretreatment test. 22

31 Table 7. Pretreatment Pilot Test Summary Dec. 8, Test Date 12/8/2003 Average Pond Water Feed Rate (GPM) 18.6 Total Gallons Pond Water Fed 5425 Feed Pond water ph 3.40 H 2 SO 4 Tank Average ph 2.51 lbs. 100% H 2 SO 4 /10 3 Gal. PW 5.78 Caustic Tank Average ph 6.44 lbs. 100% NaOH/10 3 Gal. PW Average Flocculant Rate (ppm) 2.44 Average Clarifier U/F Rate (GPM) 2.45 Clarifier Loading GPM/Ft 2 (Feed) 1.93 Final ph Adjustment Tank ph 2.89 Final ph Adjustment lbs. 100% H 2 SO 4 /10 3 Gal. PW 9.79 Total Sulfuric Acid lbs. 100% H 2 SO 4 /10 3 Gal. PW Average Product Water Flow (GPM) 17.2 Total Gallons Pretreated Pond Water 4711 Pretreatment Recovery (% V/V) 87% Product Water Analysis Sp. Gr ph 2.89 Conductivity µmhos 12,200 mg/l Ca 159 mg/l P 858 mg/l F 50.8 mg/l SO mg/l Na 2070 Cost Data H 2 SO 4 98% (Tank 1) $0.10 NaOH $1.99 Flocculant $0.13 H 2 SO 4 (Tank 6) $0.14 Total ($/1000 Gal. Pretreated Water) $2.36 H 2 SO 4 cost $25/ton; NaOH cost $225/ton; Floc. cost $0.7535/lb. 23