Reduction of Boron Concentration in Water Produced by a Reverse Osmosis Sea water Desalination Unit

Transcription

1 Reduction of Boron Concentration in Water Produced by a Reverse Osmosis Sea water Desalination Unit Koh-ichi Fukunaga, Masahiko Matsukata, Korekazu Ueyama and Shoji Kimura Dept. of Chem. Eng., Osaka University, Machikaneyama, Toyonaka 560, Japan Quality of water from RO sea water desalination plants can satisfy almost all regulation items of drinking water quality regulations, except a concentration of boron, which should be lower than 0.2 mg/l. Although this is not a compulsory item now, but it may become so in near future. Thus it is necessary to know rejection ability of various RO membranes against boron and to establish necessary measures to cope with this requirement. In this research, transport parameters of boron permeation were determined using boron rejection data of some commercial membranes. Membranes used were both high and low pressure types, which may be necessary for a two stage process. A process design computer program was set up and was run to estimate boron rejections of a RO unit using transport parameters. Results show that by increasing NRe and the recycle ratio boron concentration can be reduced, but, generally it is difficult to reduce below 0.2 mg/l unless ph at the second stage is increased over 9.9. Key words : boron reduction/reverse osmosis/seawater desalination 1. Intoduction In recent years, a number of reverse osmosis (RO) sea water desalination plants are gradually increasing by the general recongnition of an energy-saving nature of the process and its stable membrane performance against TDS rejection for long term operations. In the mean time requirement for a drinking water quality are becoming more severe due to pollution of water resorces by various causes. Thus a number of items of drinking water quality regulations have been increased in 1992 by the Ministry of Health and Welfare, Japan. Quality of water from RO sea water desalination plants can satisfy almost all reguration items, except a concentration of boron, which should be lower than 0.2 mg/l. Although this is not a compulsory item now, but it may become so in near future. Thus it is necessry to know rejection ability of various RO membranes against boron and to establish necessry measures to cope with this requirement. In this research, boron rejection data of some commercial membranes were obtained using small test cells, and they were used to determine transport parameters of boron permeation. Membranes used were both high and low pressure types, which may be nece-

2 Fukunaga EMatsukata EUeyama EKimura : Reduction of Boron Concentration in Water Produced by a Reverse Osmosis Sea water Desalination Unit sary to be used for a two stage process. Then a process design computer program for boron removal was set up and it was run to estimate boron rejections of RO unit using transport parameters determined above under various operating conditions. Since this kind of information is needed urgently, the preliminary results are reported here. The extensive results will be reported in the near future. When ƒð is close to 1, Eq. (3) is transformed as follows, while Eqs. (4) and (5) are transformed as follows. 2. Transport Equations The basic transport equations used for general membrane permeation are derived by Kedem and Katcalsky based on the irreversible thermodynamics1) as follows. where Lp, ƒð, w, express, pure water permeability, reflection coeffecient and solute permeability, respectively. C is a logmean of a Comparing above equations, numerically B is very close to P, when a is close to 1. So in such a case Eqs. (4) and (5) are useful for RO data analysis and can be used to estimate RO membrane behaviors4). Finally, solute concentration at a membrane surface, CM, can be estimated from that in a bulk, CB, taking into account of the effect of concentration polarization, by the following equation 3), which includes a mass transfer coefficient, k. solute concentration at a membrane surface, CM, and that in a permeate, Cp. Usually the difference of these values are large and it may not be appropriate to use the average value. To solve this problem Spiegler and Kedem2) derived a following equation by integrating a differential form of Eq. (2) along a membrane thickness, 3. Experimental Batch type test cells, each of which can be fitted a 75 mm ƒó test membrane piece. Concentration polarization effect was regulated by a magnetic stirrer bar. Mass transfer coefficients of this cell was determined by where P=wRT. Another transport equations, called a solution-diffusion type, have been used for RO data analysis3), which are given as follows. using the velocity variation method 5). RO membranes used were a high pressure type, NTR-70SWC and a low pressure one, NTR-759HR, supplied by Nittoh Denko Corp. Since boron tends to dissociate to boric ion at large ph (pka : 9.246) ), and the latter

3 Table 1 Transport coefficient of Boron 4. Results Fig, 1 A 2-stage process flow diagram form may be rejected more by RO membranes, boron rejection was measured at ph range from 6.5 to 9.9. Pressure range tested was from 2 to 6 MPa for NTR-70SWC, and from 1 to 2.6 MPa for NTR-759HR. 3.5% NaC1 aqueous solutions, which contained 4.5 mga boron were used for high pressure Lp was determined by pure water permeation experiments. P and u for boron were determined as follows. R values were calculated using measured Cp values taken at constant ph under various pressures, and CM values obtained using Eq. (8). Then plotting R against 1/JV, P and a were determined by the curve-fitting method to fit to Eq. (3). Transport parameters thus determined are summarized at Table 1, where it is seen that values of these membranes are very close to 1. Therefore hereafter the solution-diffusion type equations, Eqs. (4) and (5), are assumed valid to predict behaviors of these membranes. It becomes also clear that P values become small at large ph as estimated above. membranes and solutions, which contained only 1 mg/l boron were used for low pressure membranes. Initially boron concentration in the feed was varied to know its effect on transport parameters, but it become clear there was no such effect in the concentration range from 0.5 to 10 mg/l. Boron concentrations were measured by using ICP. Temperature of test cells were kept at 25 Ž Design scheme 5. Process design Using transport parameters obtained above, behaviors of a RO sea water desalination plant against boron removal were simulated using a computer. Since it is clear that a single stage process can not reduce boron concentration below 0.2 mg/l from the boron removal data so far, it is necessary to add a second stage for boron re-

4 Fukunaga EMatsukata EUeyama EKimura : Reduction of Boron Concentration in Water Produced by a Reverse Osmosis Sea water Desalination Unit moval. This scheme is shown in Fig. 1, where a brine flow from the second stage is recycled to the feed sea water to recover the permeate from the first stage. Although the amout of the recycle flow (FR) was varied, the brine flow from the first stage (B1 in Fig. 1) is kept at 60% and the permeate flow from the second stage (P2 in Fig. 1) is kept at 40% of the amount of the feed flow (F in Fig. 1). This means the total recovery of the system is kept at 40% Assumptions used in the design Assumptions used are as follows. (1) Permeation flux is determined by a operating pressure and the osmotic pressure of NaCl solution. (2) Rejection of NaCl at the first stage is 100%. (3) For boron rejection calculation, Eqs. (4), (5) and P values listed in Table 1 are used. Fig. 2 A simulation scheme of a spiral wound type module (Mass balance of the ith cell) (4) A spiral wound type module is assumed, and the flow velocity through every module is assumed constant. That means a flow width reduces continuously by the increase of permeate recovery, not like a Christmas tree type. (5) For estimating mass transfer coefficients, a following equation referred in the litereture7) for the spiral wound type element is used. A hydraulic diameter, dh, used in this calculation was set to 2d, where d is a width of a flow channel and is assumed to be 0.7 mm. (6) Since a friction factor of real spiral elements is not clear for us, we neglected the pressure drop inside elements Design procedures The spiral wound module is assumed to be a continuous longitudinal flow channel divided to N cells along the feed flow direction. Usually N was set numerically equal to a recovery of permeate of each stage expressed in %. A feed velocity in the channel is kept at constant value, U. Membranes are located at both side of the channel of both area is designated width is d, and channel and a sum as a. The channel area is A, which reduces along the flow direction, since some of the feed is permeated through a membrane with flux Jv as shown in Fig. 2. Table 2 Conditions of simulation caluculation

5 bo- Fig. 4 Effect of recycle ratio on permeate ron concentration (NRe of both stages is 250) Fig. 3 Effect of Reynolds number on (a) permeate boron concentration and (b) flux of the first stage (recycle ratio is 0) 5. 4 Simuration results Conditions of the simulation calculation are summarized in Table 2. Calculations were performed at various feed flow rates, which are expressed by NRe, and at various recycle ratio, which is defined as follows. Recycle ratio= (FR/F) x 100( % ) Fig. 3 shows the effect of NRe on the flux and the permeate boron concentration at stage 1, which shows that the effect of concentration polarization can not be neglected in the spiral type module. Fig. 4 shows the effect of the recycle ratio on the permeate boron concentration. These figures show that by increasing NRe and/or recycle ratio, boron concentration becomes small, and can be reduced below 0.2 mg/l, when the ph value of the second stage is kept at 9.9. Useally, however, membrane life may become shorter, when it is used at large ph. Also, since ph value of the first stage is usually kept at 6.5 to prevent scale formation, both the feed and the reject from the second stage need ph control to keep ph of the second stage at 9.9. Considering above facts it is difficult to adopt such a system. 6. Conclusion Transport parameters of boron permeation were determined for both high and low pressure types of commercial RO membranes. Results showed that reflection coef-

6 Fukunaga EMatsukata EUeyama EKimura : Reduction of Boron Concentration in Water Produced by a Reverse Osmosis Sea water Desalination Unit ficients were very close to 1 and the solution-diffusion type transport equations were valid for predicting behaviors of these membranes against boron removal. Using these results simulation calculations were performed at various NRe and the recycle ratio. Results show that by increasing these values boron concentration can be reduced, but, generally it is difficult to reduce it below 0.2 mg/l unless ph at the second stage is increased over 9.9. Since ph at the first stage is usually run at 6.5, both the feed and the reject of the second stage need ph control, which is not so convenient. Acknowledgement Authors are grateful to Nittoh Denko Corp. for supplying us RO membrane samples. Lietrature cited 1) 0. Kedem and A. Katchalsky : Biochim. et Biophysic Acta, 279, 229 (1958) 2) K. S. Spiegler and 0. Kedem : Desalination, 1, 311, (1966) 3) S. Kimura and S. Sourirajan : AIChEJ, 13, (1967) 4) S. Kimura : Desalinaiton, 100, 77 (1995) 5) S. Nakao & S. Kimura : J. Chem. Eng. Japan, 14, 32 (1981) 6) Kagakubinran 3rd ed., Kisohenn II -338 (1984) 7) G. Schock & A. Miguel : Desalination, 64, 339 (1987)