Journal of Membrane Science 301. High-recovery reverse osmosis desalination using intermediate chemical demineralization, страница 9

Upon concentration of the SRO retentate stream in the SRO stage, the saturation indices increased, varying in the range of 6–130, 0.05–1.6, 0.07–1.4, 0.07–1.2 for BaSO₄, CaSO₄, SiO₂, and SrSO₄, respectively, although again keeping below unity for the latter three salts for most of the testing period. The variations of the saturation indices for both the SRO feed and concentrate streams were consistent with the variation of the SCR effluent pH (Fig. 5a), which was expected given the pH dependence of the removal efficiency of the scale precursors (Fig. 3).

The PRO stage was operated to avoid membrane mineral scaling by keeping the primary RO concentrate below the membrane-scalingthresholdwiththeaidofantiscalants.Indeed, membrane scaling and fouling were not encountered in the PRO stage during pilot testing, consistent with the fact that the saturation indices for CaSO₄, SiO₂, and SrSO₄ in the PRO concentrate were below unity (Fig. 5c–e). With respect to BaSO₄ (Fig. 5b), it is noted that its crystallization kinetics is slow [3,4], while the saturation index in the PRO concentrate (17–63) was still at range in which antiscalants usage would be effective (i.e., up to saturation index of as high as ∼60 [29]). Given the above, one would surmise that scaling would not be expected in the secondary RO step if the saturation indices for the mineral salts in the SRO concentrate would be below those encountered in the PRO concentrate. SRO desalting of the ICD treated PRO concentrate at 50% water recovery (i.e., overall water recovery of 91%) resulted in saturation indices for CaSO₄, SrSO₄, and SiO₂ in the SRO concentrate that generally did not exceed saturation (Fig. 5, region (i)) and, with the exception of SiO₂, were generally below the saturation indices of the PRO concentrate, even for BaSO₄. It is noted that, during the periods ∼230–240h and ∼730–830h (Fig. 5, region (i)), the SCR effluent pH decreased below∼9(Fig.5a,region(i))andthus,duringtheaboveperiods, lower SCR removal efficiency resulted in increased saturation indicesfortheabovesaltsintheSROconcentrate.Similartrends wereobservedduringSROoperationat68%recovery(i.e.,overall water recovery of 95%; Fig. 5b–e, regions (ii)–(iv)), but with the saturation indices in the SRO concentrate being noticeably above those in the PRO concentrate for significant portion of the operational period; this was particularly noticeable during the period of 1310–1510h when the SCR effluent pH decreased well below 9 (Fig. 5a, region (ii)). The SCR performance data (Fig. 5) demonstrated that as long as the SCR effluent pH was maintained well above 9, the SCR demineralization efficiency was adequate for calcium, strontium, and barium removals so as to maintain the saturation indices for the scalants CaSO₄ and SrSO₄ below unity and for BaSO₄ below its saturation index in the PRO concentrate. This level of SCR performance was sufficient to enable SRO desalting operation free of scaling (with supplementalantiscalantdosingtoensurescalesuppression)and thus attain up to 95% overall product water recovery. It is important to recognize, however, that, even for conditions in which a given saturation index is above unity (as in the case of BaSO₄), the likelihood of surface scaling is not a certainty since mineral surface crystallization (in the bulk and onto the membrane surface) is a kinetic process [3,4,12,30–32] that is affected by pH, ionic strength, solution chemistry (including ion composition). Moreover, in the case of surface crystallization, scaling can also be impacted by the membrane surface properties and local concentration polarization [30,31].

SCR demineralization efficiency for silica removal was relatively low (30±10% removal, 95% confidence interval). The SCR effluent pH was generally at 9.6±0.3 (within a 95% confidence interval; Fig. 5a), and could not be effectively maintained at pH higher than 10. At this latter pH level, 60% or higher removal of silica would have been possible (Fig. 3). In fact, there was essentially no SiO₂ removal in the SCR during the period 1310–1510h when the SCR effluent pH was well below 9 (Fig.5a,region(ii)).Therewere,however,severalperiodsduring the pilot testing (e.g., ∼0–90h, ∼450–700h, and 2200–2350h) in which the SCR effluent pH was at the range of 10.5–11.5 such that the SiO₂ saturation index in the SRO concentrate was brought to below or to the level observed in the PRO concentrate. It is emphasized that, throughout the Phase II test period, the SiO₂ saturation index in the SRO concentrate was still maintained at a reasonably low level (<1.1) which was within the effectiveness range for membrane mineral scale suppression with a low dosage (relative to the PRO stage) antiscalant makeup. The present results, however, clearly indicate that mitigation of silica fouling/scaling in long-term operations would necessitate both careful SCR effluent pH control at sufficiently high level (10.5–11.5) and judicious choice of make-up antiscalant type and dose in the SRO stage [12].