外文翻译饮用水处理通过用金属膜制成的凝固微滤来去除天然有机物质大学论文.doc

外文资料
Removal of natural organic matter (NOM) in drinking water treatment by coagulation–microfiltration using metal membranes
Torove Leiknes∗, Hallvard Ødegaard, Håvard Myklebust

Department of Hydraulic and Environmental Engineering, NTNU, Norwegian University of Science and Technology, S.P. Andersensvei 5, N-7491 Trondheim, Norway

Abstract

Drinking water sources in Norway are characterized by high concentrations of natural organic matter (NOM), low pH, low alkalinity and low turbidity. The removal of NOM is therefore in many cases a general requirement in producing potable water. Drinking water treatment plants are commonly designed with coagulation direct filtration or with NF spiral wound membrane processes. This study has investigated the feasibility and potential of using inorganic metal microfiltration membranes in a submerged membrane configuration with coagulation pre-treatment for drinking water production. Variations in operating modes and conditions were tested, from dead-end operation to semi sequencing batch operation using air scouring and backwashing cycles for membrane cleaning and fouling control. Fluxes around 180 LMH at trans membrane pressures below 0.3 bar where achieved over production cycles in excess of 50 h. Treatment efficiencies in general showed >95% colour removal, ∼85% UV removal, 65–75% TOC removal and <0.2 NTU turbidity and non-detectable suspended solids in the permeate. The initial results show that MF metal membranes is an interesting alternative to sand filtration in coagulation/direct filtration for treating drinking water.
Keywords: Natural organic matter、Coagulation–microfiltration、 Metal membranes

1. Introduction

About 90% of Norwegian drinking water supplies are from surface water sources, generally from lakes which typically have very low turbidity, alkalinity and hardness but high colour resulting from natural organic matter (NOM). One of the major problems of using surface water sources in northern climates is high content of NOM and total organic carbon (TOC). Removal of NOM is required since coloured water is unattractive to consumers, results in colouring of clothes during washing, can cause odor and taste, increases corrosion and biofilm growth in the distribution network, and is a precursor to the formation of disinfection by-products (DBP) when water is disinfected. Halogenated compounds resulting from chlorination of drinking water containing concentrations of NOM has been a major concern since their discovery in the early 1970s as some of the chlorination by-products are carcinogenic. Drinking water sources in Norway are commonly described as having high colour, low pH and low alkalinity where typical values are given in Table 1. The removal of NOM is therefore a major treatment requirement in the production of potable water where concentrations typically in the 30–80 mg/L Pt true colour range are reduced to less than 10 mg/L Pt.
The most common drinking water treatment plant designs in Norway are based on coagulation and direct filtration or nanofiltration (NF) membrane filtration processes [13]. Coagulation direct filtration plants (enhanced coagulation) are still the dominant treatment plant design option.
In the last 10–15 years membrane processes based on nanofiltration (NF) using spiral would module configurations have been success- fully used in Norway for removing NOM, and approximately 100 membrane plants are in operation today. The NF membrane plants are commonly designed to operate with a constant flux of ∼17L m−2 h−1 (LMH) at a trans membrane pressure (TMP) of 3–6 bar with a water recovery of ∼70%. Some of the disadvantages of the NF spiral wound membrane systems used are a relatively low recovery, high energy consumptions due to the operating pressures, and fouling by the NOM and sub-micron particulates resulting in the need for a daily cleaning procedure in addition to the periodic maintenance cleaning procedures [13,14]. In a recent survey about experiences with different treatment plant types, both the operators and owners of treatment plants using membranes were generally very satisfied with using membrane technology. However, the survey also indicated an interest in alternative membrane treatment plant designs that were more energy efficient and which could reduce the necessary cleaning frequency by efficient fouling control. Two approaches can be followed to achieve this; using various pre-treatment options of the raw water prior to the NF filtration units or by using different types of membranes , membrane modules and operating options. Studies using microfiltration (MF) and ultrafiltration (UF) membranes as well as alternative membrane module designs (hollow fiber cross-flow modules and submerged modules) combined with pre-treatment by coagulation to reduce and control fouling have been reported [1,4–7,9–11]. They all demonstrate the advantages and ben- efits of combining coagulation pre-treatment with membrane filtration when UF and MF membranes are used.
Membranes in drinking water treatment are commonly based on spiral wound systems or cross-flow hollow fiber/tubular systems. These membrane processes are pres- sure driven membrane modules and mounted in different array designs to optimize the process. Energy costs required to pressurize the membrane vessels and maintain high enough fluid cross-flow velocities often is a substantial component of these systems. Submerged membrane designs offer a new approach both to the membrane module design and low pressure operating conditions which can be beneficial for the overall energy requirements to operate a process. The submerged membrane process design combined with coagulation pre-treatment was chosen for this study as an alternative treatment process to coagulation direct-flirtation for the removal of NOM. Inorganic metal membranes were also chosen as the membrane is both chemically and physically robust, allowing for alternative cleaning strategies for fouling control compared to what is feasible with polymeric membranes.
The objective of this study has been to investigate the feasibility and potential of inorganic MF metal membranes combined with coagulation for the treatment of drinking water from highly coloured surface water. A low pressure sub- merged membrane module configuration was chosen combined with the coagulation pre-treatment. The metal mem- branes have been supplied by Hitachi Metals Ltd., Japan.
2. Experimental

2.1. Production of raw water

All the experiments in this study were conducted with feed water having a colour of 50 mg/L Pt at pH 7 which is typical and representative for Norwegian raw water sources. The feed water to the membrane reactor was prepared using a NOM concentrate from a full-scale ion exchange treatment plant by mixing the concentrate into tap water to make up the desired composition. Analysis of the reconstructed water showed that the feed water is representative of the natural water source. Reconstructed feed water was chosen for the study to maintain the same initial conditions for all experiments conducted such that the performance of the process under varying operating conditions could be evaluated and compared. Hydrochloric acid (HCl) was used for pH adjustment and control to ensure optimal pH of 6.3 ± 0.2 for the coagulation step. The reconstructed raw water with a colour of 50 mg/L Pt had a DOC concentration of 6.1 ± 0.25 mg/L C and a UV254 -absorbance of 31.1 ± 1.1 m−1 .
The coagulant used was a polyaluminium chloride (PAX-16), aqueous solution from Kemira Chemicals AS. Preliminary coagulation tests were first conducted in jar-tests to find the optimum pH and coagulant dosage necessary to remove the NOM. Dosages of 2, 3, 4 and 5 mg/L Al were tested at the optimal pH of 6.3 ± 0.1 to determine the colour removal. Results revealed that a specific aluminium dosage of 5 mg/L Al removed 94% of true colour, 87% of UV-absorbing compounds, and 71% of DOC [9]. The removal of colour did not increase much from a dosage of 4–5 mg/L Al, however, the removal of DOC did in- crease as well as the Zeta potential of the particles formed. A dosage of 5 mg/L Al was therefore chosen as the preferred coagulant amount. The Zeta potential of the particles formed increased from around −22 to +5 mV with increasing dosage. This increase in the Zeta potential was also considered beneficial compared to the lower dosages which gave negative Zeta potential values. However, the average Zeta potential of the particles in the membrane reactor were measured to be around −7.75 ± 4.19 mV. The lower value found may be due different conditions in the membrane reactor compared to the jar-tests such as effects of sludge concentration, hydraulic and flocculation conditions, how- ever, the value measured is close to a neutral charge which is beneficial for the aggregate formation. All experiments were therefore conducted with a coagulant dose of 5 mg/L Al producing a feed water to the membrane reactor with a pH of around 6.3 ± 0.2.
Flocculation of the feed water was done using a pipe flocculator to maintain a rapid development of the micro-flocs. The pipe flocculator was designed with a hydraulic retention time (HRT) of 30 s and a hydraulic gradient G of 400 s−1 . The suspended solids concentration in the feed water after coagulation/flocculation with a coagulant dose of 5 mg/L Al was around 25 mg/L SS.

2.2. Membrane module specification

The metal membranes provided by Hitachi Metal Ltd. are made as sheets. Each sheet is constructed by sintering metal powder in a support layer to form the membrane. The nominal pore size of the membrane has been characterized using both the bubble point method and a particle size exclusion analysis. These methods determined the membrane having a nominal pore size of 0.95 and 0.2 m, respectively. As such the membrane can be classified as an “open” mi crofiltration membrane, however, the particle size exclusion giving a pore size of 0.2m is most likely more representative of the membrane characteristics due to the structure of the membrane [3].
The membrane module was designed and built as a plate and frame system using a sandwich construction where an aluminium frame was designed to hold two membrane sheets on each side with a support layer inside [15]. The frame measurements were; height 430 mm, length 270 mm, width 10 mm, giving an effective membrane surface area of 0.1596 m2 per module. For the initial investigation in this study only one membrane module was immersed in the membrane reactor and the total membrane area used in this study was therefore 0.1596 m2.

2.3. Experimental configuration

The membrane reactor is a rectangular tank (h = 80 cm, w = 27 cm, l = 30 cm) where the membrane module was positioned approximately 15 cm above the bottom. An arrangement for sludge extraction and sludge sampling was made in the bottom of the tank. A sampling point was also installed in the middle of the tank to extract representative samples of the concentrate in the membrane reactor. The permeate was extracted using a low pressure vacuum pump and stored in a permeate reservoir for backwashing. A maximum TMP of 0.5 bar was set as a limit for operating the vacuum pump and the condition at which an extensive cleaning of the membrane module was necessary. An air blower and aeration device for coarse bubble aeration was installed to enable air scouring for fouling control and membrane cleaning routines.

2.4. Experimental analysis

The performance of the membrane module was deter- mined by measuring the transmembrane pressure (TMP) for constant flux operation. The development of TMP for different fluxes was measured continuously using an online pres- sure transducer connected to a data acquisition system from National Instruments, Field Point (FP1000 with FP-AI-110 analogue input), in combination with the LabVIEW 6.1 data acquisition and analysis program. The water temperature was also logged continuously with a temperature transducer. Water flow rates were measured manually with rota meters on the respective lines of flow. Fouling rates for the membrane performance were calculated as rate of permeability decline, expressed as normalized flux divided by the trans- membrane pressure (L m−2 h−2 bar−1 ).
The water treatment efficiencies were measured by analyzing the removal of colour, TOC, UV254 -absorbance, turbidity and suspended solids. Samples from the feed water stream, the concentrate in the membrane reactor, and in the permeate stream were analyzed. Analysis protocol followed Norwegian Standards. True colour and UV-absorption were determined with a Hitachi U-3000 UV–vis spectrophotometer. Colour was determined by measuring the absorbance of a sample at 410 nm in a 5 cm cell. UV absorption was determined at 254 nm using a 1 cm quartz cell. Dissolved organic carbon (DOC) was determined by catalytic wet oxidation (Tekmar Dohrmann Apollo 9000). Raw water samples were passed through a 0.45 m Sartorius cellulose nitrate filter prior to analysis to remove particulate matter. Zeta-potential was analyzed by laser Doppler velocimetry (Coulter 440SX). Residual aluminium in permeate samples was also analyzed by inductively coupled plasma-mass spectrometry (HR-ICP-MS).

2.5. Membrane cleaning procedure

The membrane was cleaned between each experiment and the cleaning procedure consisted of combining phys- ical and chemical procedures. The membrane reactor was first drained and filled with clean water. The direction of permeate was reversed to backwash the membrane combined with vigorous air scouring for a period. Mechanical cleaning of the membrane was first employed by gently brushing the membrane surface with a soft brush before rinsing. The membrane module was then soaked in a hypochlorite solution (200 mg/L) for a couple of hours to remove organics that may have adsorbed to the metal surface, followed by soaking in a weak citric acid solution to remove any inorganic fouling. The efficiency of the cleaning procedure was checked by repeating clean water flux tests and comparing results with the initial clean water flux evaluation.

3. Conclusions

The initial results show that a MF process with coagulation pre-treatment using metal membranes has a great potential for drinking water treatment. Coagulation pre-treatment with polyaluminium chloride (PAX-16) of raw water with a colour of 50 mg/L Pt revealed that a specific aluminium dosage of 5 mg/L Al removed >95% of true colour, ∼87% of UV-absorbing compounds, and 65–75% of DOC. A consistent high permeate quality was achieved for all experiments irrespective of operating modes investigated. Initial studies with dead-end operation and variations of backwashing and air scouring showed that membrane fouling was reversible and primarily cake formation. This cake layer was easily re- moved when the membrane was cleaned extensively and the