亚利桑那州图森市运营着一座150 mgd臭氧直接过滤厂，该厂于1993年投入使用，旨在处理来自科洛拉多河的地表水，并通过亚利桑那州中部项目（CAP）输送。盖层水的特点是具有中等浊度（2-3 NTU）和总有机碳（2-3 mg/L），以及高硬度（碳酸钙为200-250 mg/L）和总溶解固体（600-700 mg/L）。1995年，该市和美国复垦局（BOR）开始研究提高CAP水质量的方法，以满足当地有关将成品CAP水质量与该市优质历史水源深层地下水相匹配的要求。 这要求盖水满足1995年《水消费者保护法》规定的以下参数:TOC和lt0。4 mg/L、TDS和lt210 mg/L以及硬度和lt83 mg/L碳酸钙。Tucson/BOR的研究表明，满足这些要求的最具成本效益的方法是在适当的初步处理后使用膜，即反渗透（RO）或纳滤（NF）。NF和RO使用分子质量截止值为100至500道尔顿的半渗透、无孔膜。RO主要用于减少TDS（脱矿），而NF用于软化。 两者都能很好地去除TOC和消毒副产物（DBP）前体。虽然也能达到良好的浑浊度和去除病原体的效果，但这两种方法都不能作为这些污染物的主要屏障。NF和RO几乎完全用于处理地下水，通常含有低水平的颗粒物和生物物质。当用于处理地表水（如帽水）时，控制膜污染是这些工艺成本效益应用的主要挑战。地表水含有高浓度的颗粒物，而非- 可生物降解和可生物降解的天然有机物（分别为NBOM和BOM），可导致NF/RO膜快速且通常不可逆的污染。膜污染导致产水量减少，在某些情况下，产品水（渗透）质量降低，因此需要频繁清洗和过早更换元件。污垢可由物理、生物和化学污染物引起，通常是混合污染。物理污垢可通过给水颗粒物含量的排出（澄清和/或过滤）来控制。 通过控制细菌和BOM水平，可以将生物污染降至最低。通过优化膜工艺上游的化学澄清，可以将NBOM和可氧化金属的化学污染降至最低。本研究的目的是评估两种基于生物的初步处理工艺（慢砂过滤（SSF）和生物优化快速砂过滤（BORSF））和一种物理工艺（微滤（MF））对帽水中不同类别NF/RO膜剂的控制效果。 NF/RO对每个预处理出水的操作提供了一种方法，以确定每种污染物类别对NF/RO膜性能影响的相对贡献。本文报告了2000年试验期间预处理和NF/RO中试系统的结果。包括3个参考文献、表格和图表。

The City of Tucson, Arizona operates a 150-mgd ozone, direct filtration plant,commissioned in 1993 and designed to treat surface water originating from theColorado River and delivered via the Central Arizona Project (CAP). CAP water ischaracterized by moderate turbidity (2-3 NTU) and total organic carbon (2-3mg/L), and high hardness (200-250 mg/L as calcium carbonate) and total dissolvedsolids (600-700 mg/L). In 1995, the City and the U.S. Bureau of Reclamation (BOR)began studying methods to enhance the quality of CAP water to meet localrequirements related to matching finished CAP water quality with that of theCity's high quality and historical water source, deep groundwater. This requiredthat the CAP water meet the following parameters as stipulated in the 1995 WaterConsumer Protection Act: TOC <0.4 mg/L, TDS <210 mg/L and hardness <83 mg/L ascalcium carbonate. The Tucson/BOR study indicated that the most cost effectivemethod for meeting such requirements would be with membranes, either reverseosmosis (RO) or nanofiltration (NF), following appropriate preliminary treatment.NF and RO use semi-permeable, non-porous membranes with molecular mass cutoffs of100 to 500 Daltons. RO is designed primarily for TDS reduction (demineralization)while NF is employed for softening. Both provide excellent removal of TOC anddisinfection byproduct (DBP) precursors. Good turbidity and pathogen removal isalso achieved, however neither is designed to serve as a primary barrier to thesecontaminants. NF and RO have been used almost exclusively for the treatment ofgroundwaters, typically having low levels of particulate and biological matter.When used to treat surface waters, such as CAP water, control of membrane foulingis the major challenge for the cost-effective application of these processes.Surface waters contain high concentrations of particles, non-biodegradable andbiodegradable natural organic matter (NBOM and BOM, respectively) that can causerapid and often irreversible fouling of the NF/RO membranes. Membrane foulingcauses reduced water production and, in some cases, decreased product water(permeate) quality, thereby necessitating frequent cleaning and premature elementreplacement. Fouling can be caused by physical, biological and chemicalcontaminants, usually in combination. Physical fouling can be controlled byreduction in particulate content of the feed water (clarification and/orfiltration). Biological fouling can be minimized by controlling both bacterialand BOM levels. Chemical fouling, both from NBOM and oxidizable metals, can beminimized through optimized chemical clarification upstream of the membraneprocess. The purpose of this study is to evaluate the effectiveness of twobiological-based preliminary treatment processes (slow sand filtration (SSF) andbiologically optimized rapid sand filtration (BORSF)) and one physical process(microfiltration (MF)) for the control of the different classes of NF/RO membranefoulants in CAP water. Operation of NF/RO on the effluent from each pretreatmentprovides a means to determine the relative contribution of each foulant class toperformance impacts on the NF/RO membranes. This paper reports on the results ofthe pretreatment and NF/RO pilot systems during testing conducted in 2000. Includes 3 references, tables, figures.