Optimisation of combined coagulation and microfiltration for water treatment

Abstract

The effect of upstream coagulant dosing for full-flow microfiltration of an upland-reservoir water has been investigated. The process, run under conditions of constant flux and pH and based on a ferric salt, is compared with a published study of another full-flow process based on alum dosing and operated at constant pressure and coagulant concentration. The current study includes data for the residual deposit remaining following backflushing by reverse flow. Results are presented in terms of the specific-cake resistance $\text{(R′}{}_{\mathrm{<mtext>c</mtext>}}\text{,}\text{m}{}^{\mathrm{-2}}\text{)}$ as a function of pH or coagulant dose. Reasonable correlation with classical cake filtration theory was obtained, such that R′_c was assumed to be independent of run time and cake thickness. The following trends have been noted:

• The optimum pH for the alum-based system appears to be between 7.5 and 8 on the basis of cake resistance.

• The effect of coagulant dose between 18 and 71 μM Fe³⁺ is much more significant than a change in pH between 5 and 9 for the alum system: a 53-fold increase in specific flux compared with a 7-fold increase with reference to the limiting R′_c values at pH 4.8 and 7.7.

• A low coagulant dose (0.018 mM, 1.0 mg l⁻¹ Fe³⁺) appears to have a slightly detrimental effect on downstream microfiltration operation. The low coagulant doses apparently cause incomplete aggregation of colloidal particles such that internal fouling of the membrane takes place.

• The residual (cleaning cycle) deposit resistance followed roughly the same trend as the backflush cycle-cake resistance with coagulant concentration, but with a much reduced value (about 16 times lower, on average).

• An optimum coagulant dose of 0.055 mM (3.1 mg/l) Fe³⁺ can be identified on the basis of operational cost based on coagulant cost and cake resistance, all other aspects of the system being substantially unchanged. It is concluded that coagulation with downstream microfiltration offers a cost-effective means of removing natural organic matter, achieving a THMFP removal of around 80% at the optimum dose.

Introduction

The combination of membrane microfiltration (MF) with upstream coagulant dosing is known to yield significant flux enhancement over microfiltration alone for the treatment of surface waters laden with natural organic matter (NOM) (Ben Aı̈m et al., 1988; Lahoussine-Turcard et al., 1992; Magara et al., 1998; Wiesner et al., 1989). This process is important in suppressing trihalomethane (THM) formation from surface waters. The use of membrane separation as opposed to conventional clarification techniques permits a much reduced flocculation time—in the order of 15–60 s—and thus a more compact plant. This arises from the sub-micron pore size of the filtration membranes, thereby requiring flocs to grow only to around 2 μm in size.

Whilst recent work published in this area is quite extensive, data on real systems is relatively scarce and the effect of coagulant dosing on backflush cycle efficiency in particular has received little attention. Moreover, little has been published on the comparative efficacy of the candidate ferric and aluminium-based inorganic coagulants, and no cost-benefit analysis has been published pertaining to the coagulation–microfiltration system.

Results are presented both for ex situ coagulation trials based on particle size measurement and coagulation-full flow MF pilot-scale trials based on an upland water from the Huntington reservoir in the North West of England (Table 1). Removal of NOM, represented by the total organic carbon (TOC) data, from this water is critical for meeting water quality regulations pertaining to THMs, these being disinfection byproducts generated as a result of chlorination of residual organic matter.