ASSESMENT OF MFI AS SEAWATER PRETREATMENT DESIGN TOOL AT CARBONERAS DEMONSTRATION PLANT

Transcription

1 ASSESMENT OF MFI AS SEAWATER PRETREATMENT DESIGN TOOL AT CARBONERAS DEMONSTRATION PLANT Authors: Presenter: García, Miguel; Sanz, Joan; Carulla, Carme; Nebot, Enrique; Ortega, Juan M.; Casañas, Antonio; Lubian, Luis M.; Quevedo, Noelia García Hernández, Miguel Project Engineer AcuaMed - Spain Abstract This paper presents the application of modified fouling index (MFI) to study RO pretreatment and RO feed quality for a Mediterranean seawater open intake. To reach this objective a demonstration plant was operated according to AcuaMed guidelines at Carboneras desalination plant (Spain) for eleven months. Raw water SDI ranged from 4.5 to 25 %/min in this period of time. RO pretreatment combined several process units: ballasted flocculation, direct filtration, ferric chloride coagulation, filtration and in-series filtration. MFI measurements obtained were evaluated from different pretreatment configurations using probability distribution at 95 th percentile level. According to the membrane manufacturer MFI was 1 s/l 2 as target because it could be considered sufficient to control colloidal and particulate fouling; and 4 s/l 2 as maximum level. MFI and SDI were measured at the same time by using an automatic device. SDI and MFI results had been compared for the best RO water quality (MFI close to 1 s/l 2 before cartridge filters) in order to establish some correlation between both fouling control parameters. RO feed MFI showed in this range lower variance than SDI. Wor World Congress/Perth Convention and Exhibition Centre (PCEC), Perth, Western Australia September 4-9, 2011

2 I. INTRODUCTION Modified Fouling Index (MFI), developed by J.C. Schippers and J.Verdow in 1980, allows applying an onsite test directly linked to colloidal concentration and the resistance of the cake formed on the 0.45 micron cellulose esters filter. The MFI is a more accurate index than the SDI to predict the tendency of water to foul RO systems (Boerlage, 2007). Applying MFI to reverse osmosis design has been published on membrane manufacturers guidelines that deal with feed water quality as a component to ensure good operation of the membrane system. This paper presents the results of the MFI measurements carried out for the study of seawater pretreatment at the demonstration plant installed at Carboneras desalination plant (Almeria, Mediterranean sea, Spain) owned by AcuaMed. Different seawater pretreatment configurations were tested at the demo plant (with a maximum pretreatment capacity of 110 m 3 /h). These tests were carried out using water coming from Carboneras desalination plant open intake from February to November After the testing and data sampling period, the treatment and analysis of that data built up this study. The design of the pretreatment line resulted from combining all different unitary procedures. It is to say: in-line coagulation, direct filtration, ballasted flocculation, gravity filtration, and under pressure filtration in one or two stages. Throughout the pilot process, the chlorination-dechlorination process was not applied to raw water. The demonstration plant was also provided with a reverse osmosis system counting seven 8-inch membrane elements. The RO system was fed with pretreated water. The evolution of the reverse osmosis system was monitored by means of the normalization of operation data. MFI measurements were carried out at the inlet and outlet of each single unitary process including raw water during the whole operating stage at Carboneras demonstration plant. A prototype device was developed to measure the huge amount of MFI results. This prototype allowed the simultaneous measuring of SDI and MFI for their later comparative analysis. The results given by the MFI were complemented with the study of other parameters related to colloidal matter and seawater microbiology since they may affect the MFI measurements. All data was studied in order to determine the performance of each pretreatment unit as well as raw water quality during the four seasons of the year. The main objective of the demonstration project was to define the most reliable pretreatment configuration under any raw water quality conditions using MFI as a design tool. 1.1 Carboneras Desalination Plant description Carboneras desalination plant is located near Almeria, in the southeast of the Iberian peninsule (Figure 1); one of the driest areas of Spain with a rainfall average of around 200 mm/year [1]. The plant has a capacity of 42 hm 3 /year, and it is the core of the desalinated water distribution network of eastern Almería. This network supplies desalinated water for different uses: water for irrigation at Campo de Níjar; drinking water for the municipalities of the Valle del Almanzora, Níjar and Carboneras; and water for industrial use in the municipality of Carboneras. -2-

3 The desalination plant will be capable of supplying desalinated water to the downtown area and east of the province, when distribution network projects in progress are finished. The distribution network interconnects Carboneras and Bajo Almanzora (20 hm 3 /year) desalination plants. Figure 1: Location of the plant. The desalination plant at Carboneras has a nominal capacity of 120,000 m 3 /day, and MVA of apparent power transformers. The seawater open intake consists of two lines of 2,000 mm of diameter located at a depth of 15 m inside the port. Seawater enters the plant by gravity. After passing through a bar screen cleaners (5 cm wide between each bar), water flows through the sand trap channels (6 x 70 m in length and 6 x 8 m 2 section) to the seawater pumps aspiration reservoir. From this point on, the plant is divided in two lines of 60,000 m 3 /day production (Figure 2). Figure 2: Plant distribution of Carboneras desalination plant. The 2 x 7 seawater pumps (980 m 3 /h and 140 kw) pump the seawater to the pretreatment. The chemical pretreatment consist of sodium chloride, sulphuric acid, coagulant, bisulphite and dispersant dosing. Physical pretreatment comprises 2 x 22 silica sand filters (1.2 and 0.9 mm) and 2 x 6 horizontal cartridge filters (20 absolute µm) batteries. After being fitted in the pretreatment, the seawater is pumped through the high pressure pumps (6 + 1 pumps per line of 980 m 3 /h and 1,400 kw) at 69 bars. The plant has 6 racks of 10,000 m 3 /d production capacity per line. The Pelton turbines are the energy recovery devices. -3-

4 High-pressure trains consist of the high pressure pump, the electric motor and Pelton turbine sharing the same axis. Brine is discharged together with the cooling stream of the power plant located next to the desalination plant. The dilution ratio (100% capacity of the desalination plant) is 1: 20. As there is no blending in the distribution network system, the permeate needs to be pos-treated in order to protect materials of the distribution network and to make it suitable for its different uses. The objective is to improve water stability and aggressiveness. The process consists of calcium bicarbonate formation by addition of carbonic gas and lime slurry. The product water is disinfected for water supply purposes. From the product water reservoir in the plant, water is pumped to different destinations where it is required for irrigation, supply or industrial use (Figure 3). Figure 3: Nowadays (left) and future (right) water distribution areas (grey colored) of Carboneras desalination plant. 1.2 Raw water quality The main seawater quality parameters at this point in the Mediterranean during 2006 are shown on Table 1. Temperature, ph and conductivity were registered by the instrumentation of the full scale plant located in the seawater treatment line. TSS was determined in the plant laboratory. Table 1: Average values of raw water at Carboneras desalination plant in ph Cond. TOC TSS T SDI RW ms/cm mg/l mg/l ºC %/min Max Min < Average Std. dev

5 The raw water SDI, measured at the sand filters inlet, was from 3 to 6 %/min in normal conditions at Carboneras desalination plant during 2006, as shown in Figure 4. TOC was analyzed by an external laboratory from samples taken close to the open intake (Figure 5). Figure 4: SDI and temperature in 2006 at full scale plant water inlet. Figure 5: Carboneras seawater open intake. Seawater SDI measured at the demonstration plant presented higher values than the full scale plant because the seawater intake of the pumps of the full-scale and demonstration plants were located in different points in the aspiration reservoir (Figure 6). -5-

6 Figure 6: Seawater SDI in 2006 at demonstration plant inlet (N=165). Natural organic matter analysis by LSEC-OCD (Liquid Size Exclusion Chromatography-Organic Carbon Detector) system (Leparc et al., 2007) showed the following distribution with TOC value of 0.9 mg C/L: Fraction P (polysaccharides, colloids, and proteins): 5% Fraction HS+BB (humins, humic acids, fulvic acids, and humic substances hydrolisates): 44% Fraction A+N (low molar mass organic acids, and amphiphilics): 17% Fraction LMM (low molar mass neutrals): 38% The composition of the phytoplankton in the surroundings of the open intake and in the sand trap channels was almost the same. There was no difference between open sea and aspiration reservoir values. It was the typical Mediterranean composition with an abundant presence of diatoms (Chaetoceros spp., Stephanopyxis palmeriana, Pseudo-nitzschia sp.), and dinoflagellates (Protoperidinium sp.), as well as diverse zooplankton (copepods, tintinnids). The analysis of the nano and picoplankton by flow cytometry at aspiration reservoir showed two kinds of population: Synechococcus and Prochlorophytes or old Cyanophytes. Particulate matter retained on MFI filters was studied by scanning electron microscopy. All raw water samples presented coccolithophorids abundance with diatoms fragments (Figure 7). -6-

7 1.3 Demonstration plant Figure 7: SEM micrograph of particulate matter (coccolithophorids). The tests were carried out at a demonstration scale plant (Figure 8 and 9). The demonstration plant was fed with seawater pumped from the aspiration reservoir of the full-scale plant, once water had crossed sand trap channels. Water speed at the sand trap channels during the testing period was from to m/s, as a function of the variable production at the full-scale plant. The purpose for the demonstration plant according to AcuaMed, was to obtain and maintain a stable value of SDI<4 during the process by a physic-chemical treatment. The goal was to obtain an SDI below 4 before entering the membrane. As an internal objective, authors proposed to achieve an SDI<3 (Quevedo et al., 2011) The treatment line includes a ballasted flocculation (Actiflo), serial filtration followed by a 7 membrane reverse osmosis rack at the end of the process. Ferric chloride was used as coagulant in the coagulation tank and Hydrex 3551 (a non-ionic polyacrylamide with low charge anionic: less than 3%) as a flocculant. Figure 8: Diagram of the demonstration plant. -7-

8 Figure 9: Demonstration plant overview. The seawater at Carboneras was characterized by low turbidity (<1 NTU), low total suspended solids (<2 mg/l) and typical Mediterranean microbiological characteristics. Under these conditions it is possible to reach a SDI<3 with a 90 th percentile of reliability using certain configurations. From the 28 initial tests considered, only 7 were able to obtain an SDI<3 %/min with a single filter and 13 tests with two filters. All tests that reached a SDI<3 %/min presented this physical units configuration: Actiflo/ coagulation tank + Filter 1 + Filter 2. This configuration, accompanied by 3 mg/l of FeCl 3 and 0.05 mg/l of flocculant and 1mg/L of polydadmac in filter 1, gives satisfying results maintaining a SDI below 3 %/min during the process. These results have been previously published (Quevedo et al., 2011). II. OBJECTIVES AND METHODOLOGY The main objective of this paper was to study the MFI decreasing value by an enhanced physicochemical treatment at the demonstration scale plant from January 2006 to November The investigation on MFI has essentially consisted of verifying the effect of a physicochemical treatment on the seawater quality in order to achieve a MFI value bellow 4 s/l 2 with 90 th percentile statistic reliability before RO membranes. A second objective was also proposed: to reach a MFI<1 s/l 2 at 90 th percentile of reliability. These values were selected according to the membrane manufacturer guidelines (FilmTec, 2009). Others objectives were the biological growth control and the analysis of the contribution of each operation unit. Five parameters have been considered in the development of the behavior of each one of the tests carried out in order to study and analyze their impact in the global process: configuration (operation units), chemical dosage, filtration rate, ph and raw water characteristics. The term "configuration" is referred to those elements or physical units that constitute the process. The concept of "test" refers to the chemical dosage that has been tested in the process. There are several tests for each configuration. It is called "measure" to the results obtained for each sample. -8-

9 The combinations of physical elements were: Configuration 1 : Ballasted flocculation + Sand filter 1 + Sand filter 2 Configuration 2 : Direct filtration (Sand filter 1 + Sand filter 2) Configuration 3 : Coagulation tank + Sand filter 1 + Sand filter 2 The assessment criteria were: - Any test having at least 90 th percentile of reliability was valid (interval confidence 95 th percentile), and - The list of trials considered are those that meet MFI <1 s/l 2 or MFI<4 s/l 2. Statistical analysis applied was performed using Minitab14 software. The microbiological strategy was to eliminate continuous chlorination/dechlorination processes. No membrane biofouling was found in the RO membranes at the demonstration scale plant during the 11 month testing period. This strategy was also implemented at the full scale plant. As a consequence of the elimination of the continuous chlorination dosing, a decrease in the membrane cleaning frequency was reported. In order to simultaneously measure a big number of MFI and SDI values at the demonstration plant, a specific device built for this project was used (Figure 10). This device included a PLC to automatically register the weight of the water being filtered each 30 seconds (according to MFI method). MFI and SDI were measured using Millipore filter type HAWP with a pore of 0.45 µm. MFI measurements were developed according to Mr. Schippers J.C., and Verdouw J. (1980) methodology at a constant pressure of 2.1 bar. Figure 10: SDI & MFI measurement device. III. RESULTS AND DISCUSSION The three configurations were studied in three groups, attending to raw water SDI values. These SDI ranges for raw water were 4<SDI<7.1, 7.1<SDI<13.5, and 13.5<SDI<

10 3.1 First raw water SDI group In the first raw water SDI group (4<SDI<7.1), it was possible to obtain MFI lower than 1 at 90 th percentile using ballasted flocculation with one filtration step or coagulation tank with one filtration step (Table 2). The second filtration step only improved those results in one configuration test. Direct filtration did achieve MFI lower than 1 s/l 2. Table 2: MFI values 90 th percentile obtained with raw water SDI between 4 and Second raw water SDI group In the second raw water SDI group (7.1<SDI<13.5), it was possible to obtain MFI lower than 1 s/l 2 at 90 th percentile using ballasted flocculation with one filtration step or coagulation tank with one filtration step (Table 3). Second filtration step only improved these results in one configuration test. Table 3: MFI values 90 th percentile obtained with raw water SDI between 7.1 and Third raw water SDI group In the third raw water SDI group (13.5<SDI<25), it was possible to obtain MFI lower than 1 s/l 2 at 90 th percentile using coagulation tank with one filtration step or direct filtration with one or two filtration step (Table 4). Ballasted flocculation combined with filtration achieved MFI values lower than 4 s/l 2 but higher than 1 s/l 2. Table 4: MFI values 90 th percentile obtained with raw water SDI between 13.5 and

11 RO feed SDI and MFI values were compared using SDI&MFI automatic device. There was no correlation between these fouling indexes when fitted line plot was calculated (Figure 11). R-Sq was only for MFI values close to 1 s/l 2. Also, statistical summary applied to SDI and MFI measurements when MFI was close to 1 s/l 2 showed differences on variance for each fouling index (Figure 12 and 13). Figure 11: MFI 0.45 and SDI fitted line plot for RO feed (N=74). Figure 12: Statistical summary of RO feed SDI measurements when MFI is close to 1 s/l 2. Figure 13: Statistical summary of RO feed MFI measurements when MFI is close to 1 s/l

12 IV. CONCLUSION The seawater of Carboneras is characterized by low turbidity (<1 NTU), low suspended solids (<2 mg/l) and typical Mediterranean microbiological characteristics. With high seawater SDI values (up to 25), it was possible to reach an MFI <1 s/l 2 with a 90 th percentile of reliability using a coagulation tank followed by filtration step or using a two-step filtration. No correlation between RO feed SDI and MFI values was obtained when MFI is close to 1 s/l 2, according to the target value by the membrane manufacturer. MFI measurements were consistent for the best RO feed water quality, but SDI measurements presented a higher variance. MFI target from membrane manufacturer lower or equal to 1 s/l 2 is equivalent to SDI target (lower or equal 3 %/min), but it was not possible to find an equivalence value one to one for the same MFI and SDI sample measurement. V. ACKNOWLEDGEMENTS The authors want to express their appreciation to Mr. Adrián Baltanás, Mr. Fernando Troyano, and Mr. José Alonso from AcuaMed for their support to conduct this demonstration project at Carboneras desalination plant. The authors also want to thank the technical staff of AcuaMed for all the support as well as the assistance and services offered during the demonstration project. The authors also want to thank AcuaMed for their financial assistance under the contract AT/02/06. University of Cadiz, Anjou Recherche (Veolia Environnement), CSIC and other organizations directly collaborated with this study. Finally, the authors want to acknowledge the staff of the technical and engineering divisions of VWSI for their help in setting, operating and supervising the demonstration plant. VI. REFERENCES Books: Dow Water Solutions & Process Solutions (2010) FilmTec Reverse Osmosis Membranes Technical Manual. Available on: Papers: Boerlage S.F.E. (2007) Understanding the SDI and modified fouling indices (MFI 0.45 and MFI UF ). Proceedings of IDA Congress MP07-143, Maspalomas, Spain. Leparc J, Schrotter J.C., Rapenne S., Croué J.P., Lebaron P., Lafon D., Gaid K. (2007) Use of advanced analytical tools for monitoring performance of seawater pretreatment processes. Proceedings of IDA Congress MP07-124, Maspalomas, Spain. Quevedo N., Sanz J., Ocen C., Lobo A., Temprano J., and Tejero I. (2011) Reverse osmosis pretreatment alternatives: Demonstration plant in the seawater desalination plant in Carboneras, Spain. Desalination 265, Schippers J.C., and Verdouw J. (1980) The modified fouling index, a method of determining the fouling characteristics of water. Desalination, Webs: [1]