Está en la página 1de 14

filtration

In 2005, the first edition of AWWA Manual of Water Supply Practices M53, Microfiltration
and Ultrafiltration Membranes for Drinking Water, was published. The manual was
prepared by the AWWA Membrane Process Committee in response to the need for a
consolidated source of information about membrane filtration spanning a broad range of
relevant subject areas. Because the use of microfiltration (MF) and ultrafiltration (UF) in
potable water treatment applications has expanded rapidly since the mid-1990s, both in
terms of capacity and number of installations, a comprehensive MF/UF resource for the
water treatment industry had become imperative.

Microfiltration and ultrafiltration


membranes for drinking water
BY THE AWWA SUBCOMMITTEE
ON PERIODICAL PUBLICATIONS
OF THE MEMBRANE PROCESS
COMMITTEE

his article provides a concise and abbreviated summary of AWWA


Manual of Practice M53, Microfiltration and Ultrafiltration Membranes for Drinking Water, to serve as a quick point of reference.
For convenience, the articles organization matches that of the
manual's chapters, as follows:
water quality,
membrane science and theory,
system concepts,
microfiltration/ultrafiltration (MF/UF) manufacturers,
pilot testing,
design concepts,
operations,
system costs,
residuals characteristics and management, and
future trends.
This article is not intended to address each subject comprehensively and
thus should not be considered a stand-alone resource. More detailed information is provided in M53. Readers should refer to that manual for a more
thorough discussion of each topic.

WATER QUALITY
With nominal pore sizes of 0.1 and 0.01 m, respectively, MF and UF
membranes are designed to remove particulate matter such as turbidity and
microorganisms via a sieving mechanism, yielding finished water quality that
is superior to that achieved with conventional treatment processes. For example, MF and UF can consistently produce filtrate with turbidity below 0.1 ntu
independent of the feedwater quality. This is a level that significantly exceeds
84

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

the regulatory benchmark of the Interim Enhanced Surface Water Treatment Rule and Long Term 1 Enhanced
Surface Water Treatment Rule, which require filtered
water to be < 0.3 ntu in 95% of samples collected within
a month. In addition, regulated protozoans such as Giardia (515 m) and Cryptosporidium (35 m) can be
essentially completely removed with a fully integral MF
or UF membrane. Numerous studies have demonstrated
removals as high as 7 logs without the use of polymer or
coagulants based on experiments in which the MF/UF
influent was spiked with various protozoan cysts. Similar
studies have also demonstrated comparable removal efficiency for bacillus spores. One critical distinction between
MF and UF is the ability to reject viruses. Although UF
has been shown to consistently achieve greater than 3-log
virus reduction as a stand-alone process, MF generally
cannot remove viruses to any significant and reliable
extent without the use of pretreatment. Although some
virus reduction may be obtained with the benefit of the
cake layer that accumulates on the MF membrane surface
over a filtration cycle, this removal is highly variable and
unreliable. Tables summarizing various pathogen removal
studies for Giardia, Cryptosporidium, bacteria, and
viruses, including references, are provided in chapter 2 of
M53. Additional summaries of pathogen removal studies
are given in the report Low-pressure Membrane Filtration
for Pathogen Removal: Application, Implementation, and
Regulatory Issues (USEPA, 2001).
Because MF and UF membranes are microporous, dissolved organic and inorganic constituents cannot be
removed without the use of pretreatment processes such
as powdered activated carbon (PAC), coagulation, and/or
oxidation. As with conventional media filtration, coagulation is the most commonly applied pretreatment for control of dissolved organic carbon/disinfection by-product
(DOC/DBP) precursors and color-causing compounds. The
extent of removal is a function of DOC characteristics,
coagulant type, dose, pH, temperature, mixing time, and
mixing velocity. PAC can also be applied as MF/UF pretreatment to control DOC/DBP precursors, taste- and
odor-causing compounds, and color-causing compounds,
with removal efficiency varying with DOC concentration,
PAC dose, type of PAC material, DOC adsorbability, pH,
temperature, and contact time. The removal of iron and
manganese using MF/UF membranes requires oxidation
via aeration or the use of chemical oxidants such as permanganate, chlorine, or ozone. Arsenic, another contemporary contaminant of interest, can be removed by MF/
UF membranes only if an adsorbent or ferric iron-based
coagulant is used, particularly as applied to the more
strongly adsorbing As(V) form of arsenic.

MEMBRANE SCIENCE AND THEORY


Water transport across clean porous membranes,
expressed as flow per unit area, or flux (J), is directly
proportional to the transmembrane pressure (P, or

TMP) and inversely proportional to the absolute viscosity (), as modeled using a slightly modified form of
Darcys law (Cheryan, 1998):
J

Qtotal
P
=
A
Rm

(1)

in which Qtotal is the filtrate flow, A is the membrane


area, and Rm is the hydraulic resistance of the clean
membrane to water permeation. Note that because the
viscosity of water varies inversely with temperature,
warmer feedwater temperatures will result in either
increased flux (for operation at constant pressure) or
decreased feed pressure requirements (for operation at
constant flux); the converse effect results from colder
feedwater. Consequently, temperature is an important
consideration in the conceptual design phase, because
changes in feedwater temperature over the year may
necessitate additional membrane area to maintain the
required system capacity, depending on the maximum
TMP of the membranes, seasonal water demand, and
other site-specific factors. When considering temperature
effects on MF/UF membranes, the flux is typically normalized to a reference temperature of 20oC.
The hydraulic resistance of the membrane (Rm) can be
related to physical properties of the membrane as follows:
Rm 

8  z
 r 4 Ppore

(2)

Typical units for flux are gallons of water per square


foot of membrane area per day or liters of water per
square meter of membrane area per hour. The pore
density (P pore ) is the number of pores per unit of
membrane area, r is the pore radius,  is the tortuosity
factor, and z is the pore length. Thus the resistance to
pure water transport across a clean membrane is expected
to increase with increasing tortuosity and thickness and
with decreasing pore density and pore radius (with a
strongly influencing inverse 4th power relationship).
An important operational factor that can strongly
influence the flux is the foulants that accumulate over
a filtration cycle (reversible fouling), between cleaning
intervals (reversible fouling), and over the life of a
membrane module (irreversible fouling). This fouling
can take several forms: particulate/colloidal fouling,
organic fouling, and/or biofouling. Fouling results in
the gradual reduction in flux (for constant pressure
operation) or increase in TMP (for constant flux operation) as a consequence of adsorption or deposition of
contaminants either within the pores or on the surface
of the membrane. In Eq 1, fouling is incorporated by
expanding the resistance term to include additive factors to account for the resistance attributable to various
types of fouling (e.g., particulate, organic, biological)
as follows: Rm = R1 + R2 + R3 + . . . .

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

85

Particulate fouling increases resistance to the flow of


water across the membrane via pore blocking and the
subsequent cake layer formation over the course of a
filtration cycle. Colloidal interactions play a significant
role in particulate fouling, which is the most common
cause of the productivity decline in MF/UF membrane
processes. This fouling can typically be removed by periodic backwashing in a manner similar to that used for
conventional media filtration. Natural organic matter
(NOM), such as humic substances, has also been recognized as a primary cause of membrane fouling. Organic
fouling occurs via complex mechanisms and often
requires chemical cleaning to remove. In addition, the
organic matter enhances biological activity in a membrane system, which may increase biofouling. Generally,
biofouling occurs as a two-step process: (1) attachment
of microorganisms to a membrane surface and (2) forma-

In some cases, the resistance of polymeric membranes


has been observed to increase with large TMPs, a phenomenon known as membrane compaction. The compaction process constricts the membrane pores, reducing
MF/UF system productivity. Membrane compaction can
likewise be incorporated into Eq 1 using an additive
resistance factor.

SYSTEM CONCEPTS
The most fundamental component of an MF/UF system is the membrane material. Polyvinylidene fluoride is
common in drinking water treatment applications,
although numerous other membrane materials are used,
each with different properties as summarized in Table 1.
Important morphological properties of a membrane are
surface porosity, pore size and shape, and surface roughness. MF/UF membrane materials can be manufactured
in different geometrical configurations, which are then incorporated
into a membrane module (i.e., the
smallest distinct filter unit in an MF/
UF system). Commercially available
MF/UF membrane configurations
include hollow fiber, spiral wound,
tubular, and plate-and-frame. However, in municipal drinking water
treatment applications, almost all
proprietary MF/UF systems use hollow-fiber membranes.
MF/UF systems are generally operated in either constant flux or constant pressure modes. Under constant
pressure operation, the membrane flux will decline over
the course of a filtration cycle as the membranes foul.
Most MF/UF systems, however, are operated at constant
flux using variable-speed pumps to meet increasing pressure requirements caused by membrane fouling, which
restricts the passage of water. Operating TMPs may span
from approximately 3 psi (20 kPa) to as high as approximately 43.5 psi (300 kPa) in some cases, although more
typical values range from approximately 7 psi (50 kPa)
to 30 psi (200 kPa).

ith nominal pore sizes of 0.1 and 0.01 m, respectively,


microfiltration and ultrafiltration membranes are designed
to remove particulate matter.

tion of biofilm through cell multiplication. Subsequent


extracellular polymer generation (which constitutes the
bulk of the biofilm) also contributes to biofouling. It is
also worth noting that coagulant aid polymers used in
the conventional pretreatment process can cause rapid
and potentially irreversible fouling of MF/UF membranes.
The risk of fouling is particularly significant with cationic
polymers, given that most MF/UF membrane materials
have a negatively charged character. Thus, any pretreatment process utilizing polymer upstream of polymeric
MF/UF membranes should be carefully designed to minimize or eliminate polymer carryover.

TABLE 1

Characteristics of MF/UF membrane materials

Material

Type

Hydrophobicity

Oxidant
Tolerance

pH Range*

Polyvinylidene fluoride

MF/UF

Modified hydrophilic

Very high

211

Excellent

Low

213

Acceptable

Polypropylene

MF

Slight hydrophobic

Fouling Resistance/
Cleanability

Polyethersulfone

UF

Very hydrophilic

High

213

Very good

Polysulfone

UF

Modified hydrophilic

Moderate

213

Good

Cellulose acetate

UF

Naturally hydrophilic

Moderate

68

Good

MF/UFmicrofiltration/ultrafiltration
*General guidelines only

86

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

There are three general operating conFIGURE 1 MF/UF system configurations for encased membranes
figurations for MF/UF systems (Figure 1).
in crossflow operation (A), encased membranes in direct
With encased membranes in cross-flow
filtration (B), and submerged membranes (C)
operation (Figure l, part A), a portion of
the feed is recycled, returning a smaller
A
flow with more concentrated particulate
Recycle
Bleed
matter to the head of the filtration process.
A bleed (i.e., waste) stream may also be
used (Figure 1, part A) to control the solids
concentration in the recirculation loop, as
Filtrate
necessary. Encased membranes may also
be operated in a direct (or dead-end) filtration mode (Figure 1, part B) without recirAir or liquid
backwash
culation or bleed streams. The terms
cross-flow and direct filtration in the
context of MF/UF are understood within
Raw water Feed Prescreen Recirculation
the industry to apply only to encased mempump
pump
brane filtration systems, which operate
under positive pressure. With submerged
B
Backwash waste
(or submersible) membranes (Figure 1,
part C), a vacuum is applied to the membrane to create the pressure differential
that is the driving force for filtration.
Numerous minor operational variants for
controlling the flow of water and concenFiltrate
tration of solids within the basins may be
used in conjunction with submerged memAir or liquid
branes, as necessary, according to manubackwash
facturer recommendations.
As the membranes become fouled over
the course of operation, a combination of
Raw water Feed pump Prescreen
both backwashing and chemical cleaning
is used to remove foulants to the maxiVacuum
C
pump
mum practical extent, thus ideally restorFiltrate
ing operating parameters such as flux (for
Feed water
constant pressure operation) and TMP (for
constant flux operation) to baseline levels.
For most MF/UF systems, backwashing is
typically conducted every 3060 min for a
short duration (e.g., about 1 min) and conBleed
sists of water and air, or a combination of
both. Backwashing may also be initiated
based on a TMP set point or a predetermined volume of filtrate produced. ChemAir
ical cleaning (known as clean-in-place
[CIP]) is conducted at a point at which the
ability of routine backwashing to restore
MFmicrofiltration, UFultrafiltration
the desired system flux and/or TMP is significantly diminished. A more modest
include the type and dose of chemical(s) used, frequency,
cleaning operation of shorter duration and lower chemical
duration, temperature, rinse volume, recovery and reuse
doses may also be conducted on a routine schedule (e.g.,
of cleaning agents, and residuals neutralization and disdaily to weekly) to control fouling on an ongoing basis.
posal. A variety of chemicals may be used for cleaning
These operations are known by the synonymous terms
including detergents, acids, bases, oxidizing agents,
chemically enhanced backwash (CEB), maintenance
sequestering agents, and enzymes. Chlorine is also comwash, maintenance clean, and enhanced flux mainmonly used in the chemical cleaning process for both
tenance. Important considerations for chemical cleaning
AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

87

Summary of Design Concepts for Microfiltration/


Ultrafiltration Facilities
SITE-SPECIFIC ISSUES

CONSIDERATIONS

Raw water quality


and variability

Raw water quality and variability are primary


design considerations and should include
consideration of turbidity, total organic carbon (TOC), dissolved organic carbon (DOC),
temperature, and dissolved matter. Pretreatment processes may be required to maximize
operational efficiency depending on the levels
of these parameters.

Demand profile

Considerations are similar to those for conventional treatment plants; however, feedwater temperature must be taken into account
in determining the required membrane area,
because temperature-induced viscosity fluctuations affect production capability.

Finished water
quality goals

Membrane filtrate is typically very consistent


in particulate quality, but objectives for the
removal of TOC, DOC, taste- and odor-causing compounds, and other dissolved-phase
contaminants require pretreatment.

Redundancy and
reliability

Numerous off-line conditions should be taken


into account in planning for system redundancy, including backwashing, chemical
cleaning, plant maintenance, and extended
emergency shutdown. Sufficient storage/
equalization capacity should be provided.

Site space availability

Differing membrane filtration equipment footprints should be factored into the screening
of potential suppliers.

Future expansion
requirements

Sufficient space should be allocated to


accommodate any additional membrane units
that may be necessary for planned future
facility expansion.

Desirable operational
characteristics

Facilities are generally automated and most


have data logging and report generation programmed into the supervisory control and
data acquisition system. It is beneficial to balance the frequency of chemical cleaning with
other design parameters.

MEMBRANE-SPECIFIC ISSUES

CONSIDERATIONS

Flux

Flux serves as the basis of membrane system


design. Membrane flux and chemical cleaning
frequency are related in a nonlinear fashion.

Type of membrane units

Systems are either pressure-driven (i.e.,


encased) or vacuum-driven (i.e., submerged
or immersed) and designed in a modular
approach using one or more water production units (trains, racks).

88

disinfection and oxidation (for compatible


membrane materials) at a wide range of
doses (e.g., 22,000 mg/L).
Integrity testing is also a critical MF/UF
system feature, enabling the detection and
isolation of breaches that would otherwise
compromise the effectiveness of the membrane as a barrier for pathogens and particulate matter. Almost all commercially available
MF/UF systems designed for municipal drinking water treatment are equipped with the
ability to conduct an automated pressure
decay test using air, which directly challenges
the membrane barrier by pressurizing the
system modules and monitoring the rate of
decay over a short period of time (e.g., 5
min). Decay rates exceeding a given benchmark threshold value (specific to each site
and system) indicate an integrity breach, with
the magnitude of the response directly proportional to the magnitude of the breach.
Because these direct integrity tests do not
permit the simultaneous production of water,
they are typically only conducted once per
day on an MF/UF system. Thus, less sensitive
indirect methods are used as a complementary tool to monitor some aspects of filtrate
water quality between periodic applications
of a more sensitive direct test. Although the
indirect methods (e.g., turbidity or particle
counts) are only generally indicative of membrane integrity, they can be applied on a continuous basis during operation.

MF/UF MANUFACTURERS
There are approximately six primary
suppliers of MF/UF membranes designed
for use in municipal drinking water treatment applications that are currently active
in the North American marketplace. In most
cases, each membrane manufacturer also
builds its own proprietary system around its
membranes using a custom module design,
operational configuration, instrumentation
and control, and similar features, such that
membrane modules are not interchangeable
among different manufacturers systems.
Although most suppliers manufacture a
single type of system (i.e., MF or UF, encased
or submerged), some companies do offer
multiple product lines (e.g., MF-encased
and UF-submerged). In addition, there are
companies that manufacture MF/UF systems using membranes made by one of six
primary suppliers under licensing agreements. However, there are no current third-

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

party manufacturers of replacement membranes designed for use in other


manufacturers proprietary systems, primarily because most MF/UF facilities are
relatively new by drinking water treatment
plant standards, and thus most MF/UF
membrane modules are not yet near the
end of their useful life. The useful life of a
module is dependent on the feedwater
quality (i.e., potential for irreversible fouling) and site-specific operating scheme (i.e.,
frequency of backwashing; frequency and
intensity of cleaning operations). Accordingly, manufacturer warranties on MF/UF
membrane modules are also site-specific
but typically range from 5 to 10 years,
prorated on the basis of time in service.
Membrane manufacturers typically
provide design data for their respective
products, including materials of construction, chemical compatibilities, membrane
area, appropriate fluxes and recoveries,
cleaning strategies, and other pertinent
product parameters. Summaries of the
various commercially available MF/UF
membrane products are provided in chapter 5 of M53, including system descriptions, membrane characteristics, and a
partial list of installations.

MEMBRANE-SPECIFIC ISSUES

CONSIDERATIONS (CONTINUED)

Backwash approach

Backwashing typically uses air, water, or a


combination of both. Some systems also
employ chemicals (e.g., chlorine) as part of a
less-frequent backwash process (e.g., daily,
every two days, weekly), referred to as a
chemically enhanced backwash.

Chemical cleaning
strategy

Chemicals, such as acid, caustic, oxidants,


and/or surfactants, are used to restore the
membrane performance. Chemical cleaning
is typically performed once every one to six
months for most systems.

Process residuals

In the absence of pretreatment chemicals


(i.e., coagulants, polymers), the amount of
solids is typically much less than in a comparative conventional treatment plant. Thus,
disposal is generally considered less problematic in this context.

SYSTEM-SPECIFIC ISSUES

CONSIDERATIONS

Building layout
and orientation

Most systems are designed within buildings.


Encased systems use the slab-on-grade
approach, and submerged systems use
oncrete basins/open tanks.

Materials and
construction

Materials and type of piping are important considerations because of possible use of pulsing
during backwashing, use of cleaning chemicals, and the like. All materials should have
appropriate approval for the given application
as required by the state primacy agency (e.g.,
the National Sanitation Foundation).

Control and operational


approach

System control is typically more complex than


for a conventional water treatment plant.
Most systems use programmable logic controllers with a supervisory control and data
acquisition interface.

Pumping

Pumps associated with a membrane filtration


system can be classified by function: feed,
recirculation, backwash, chemical cleaning.
Pump selection is primarily based on size,
chemical exposure, amount of use, and cost.

Chemical cleaning
system design

Important considerations include the proper


drainage of residual solutions from piping,
local requirements for chemical handling and
containment, sizing issues, heating requirements, option to reuse chemical solutions,
and similar considerations.

Residuals disposal

Backwash and chemical cleaning waste


streams may require dechlorination and/or
pH adjustment prior to discharge.

Compressed air systems

Membrane systems may use air for process control (e.g., pneumatic valves), backwashing, or
integrity testing. The system should be designed
to produce filtered, dry, and oil-free air.

MEMBRANE APPLICATIONS
Municipal MF/UF treatment systems
were first installed in North America during the early 1990s at facilities that needed
to operate under unusual circumstances
(e.g., remote locations with minimal staffing), primarily for turbidity and microbial
removal from surface water or groundwater under the influence of surface water.
The first large-scale (> 1 mgd) membrane
filtration facility was commissioned in
1993 at Saratoga, Calif., for the San Jose
Water Company with a capacity of 3.6
mgd. The success of these early facilities in
providing cost-effective treatment, combined with increasing regulatory requirements, stimulated rapid growth in the
number of membrane filtration installations. Initially, many of these facilities were
stand-alone installations of the technology
for particulate and pathogen removal.
However, MF and UF were quickly integrated into more complex treatment
schemes as the potential applications diversified. In general, these integrated membrane systems (IMSs) can be classified into

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

89

Summary of Key Operational Aspects for MF and UF Facilities


PLANT CONTROL

CONSIDERATIONS

Basic modes

There are six basic modes of operation that may be


used by a microfiltration/ultrafiltration (MF/UF) system: filtration (F), backwash (BW), air scrub (AS),
chemically enhanced backwash (CEB), chemical
clean-in-place (CIP), and membrane integrity
verification (MIV).
Typically, F, BW, AS, CEB, and MIV are controlled
automatically based on user-selected set points.
CIP is typically selected manually and operated
semiautomatically or manually.
MF/UF systems are usually operated as constant
rate filters. Primary set points include filtered water
flow rate and frequency of periodic operations (e.g.,
BW, CEB, AS, and MIV). The operator also sets the
chemical type(s) and concentration(s) for CEB.

Control

Set points

PROCESS MONITORING CONSIDERATIONS


Long-term
performance

Unit performance

Net production
and recovery

Pretreatment

DATA COLLECTION
AND RECORD KEEPING
Commissioning
and start-up

Compare temperature-normalized results (e.g., operating pressures) after each chemical cleaning to
track system performance and cleaning effectiveness over time.
Compare units (i.e., rack, train, and similar) to determine any deviations from typical operation and performance. Check for proper operation and possible
failure of valves, which might be adversely affecting performance.
Monitor total amount of filtrate and backwash produced to determine net filtrate production and
recovery. Filtrate usage during BW, AS, CEB, CIP,
and MIV should be taken into account.
If pretreatment is used, including coagulation, flocculation, and/or sedimentation, monitor for any
excessive carryover of floc particles into the MF/UF
process. Polymer application is typically discouraged at MF/UF facilities, because some polymers
may irreversibly foul the membrane.

CONSIDERATIONS

At a minimum, record module serial numbers, installed


location, sensor calibration records, initial performance
data, and factory certifications of fiber integrity.
Regular operation Key data include feed and filtrate pressures, flows,
and turbidity; temperature; membrane integrity test
results; and calculated values of transmembrane
pressure and flux.
Maintenance
Thorough records of broken fibers, membrane
repairs, and other maintenance should be kept
throughout the life of the plant.
Integrity testing
Track membrane integrity tests over the life of the
system for each unit. Maintain detailed records of
module failures/replacements and their causes in
order to help identify future potential problems.

90

three categories: preliminary, intermediate,


and final membrane treatment.
In a preliminary membrane treatment
application, MF/UF is used as pretreatment
to remove naturally occurring particulate
matter to prevent fouling and/or reduce
chemical oxidation prior to other advanced
treatment processes such as ozone/biological
filtration, granular activated carbon (GAC),
nanofiltration, or reverse osmosis. Preliminary and intermediate membrane treatment
are similar, although in an intermediate application MF/UF is preceded by conventional
pretreatment such as coagulationflocculationsedimentation or lime softening in order
to decrease the potential for membrane fouling, remove NOM, or reduce hardness. MF/
UF is the last process (except disinfection) in
a final membrane treatment application. Preceding processes may include preoxidation,
in-line coagulation with or without flocculation, sedimentation, full conventional pretreatment, lime softening, GAC, PAC, ozonation, biological filtration, and/or media
filtration, among others. A summary of various IMS schemes and the associated target
contaminants is provided in Table 2. Several
case studies of applications utilizing different
forms of IMS treatment are included in chapter 6 of M53.

PILOT TESTING
Pilot testing is a valuable tool used to determine appropriate operating and design criteria
for membrane systems that are acceptable to
the owner, operators, engineer, regulatory
agency, and membrane suppliers. The pilottesting protocol establishes the basis for evaluating a range of reasonable operating parameters to establish design criteria while meeting
finished water quality goals. In addition, pretreatment requirements (if any), maintenance
and cleaning strategies, and cost estimates may
be developed from the data gathered during
pilot testing. Therefore, it is important that a
detailed and wellthought-out test plan be
prepared to adequately assess membrane system performance.
Before starting a pilot-testing program, it
is important to review the features of each
commercially available system and assess
their respective suitability, taking into account
project-specific criteria and costs. Project criteria that may influence the selection of one
membrane system over another include operating mode (e.g., cross-flow versus direct fil-

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

tration), disinfection goals (e.g., whether it is necessary


to use UF to achieve reliable virus removal), feedwater
chemistry/pretreatment, cleaning solutions required, site
constraints (e.g., different system footprints), and economics. The interaction between MF/UF membrane
properties and water quality is generally not well understood, making it difficult to anticipate appropriate operating conditions for a given membrane system from water
quality data alone. Thus, one of the critical objectives of
a pilot test is to determine an appropriate range of operating conditions for a given system, thereby facilitating
an estimate of the cost and feasibility of implementing
the MF/UF technology for a given site-specific application. Selecting membrane system suppliers for piloting
should be accomplished using screening criteria that are
specific enough to limit the participants to a manageable
number, based on site constraints, staffing requirements,
and budgetary considerations.
When developing a pilot-testing protocol, it is important to consider how seasonal fluctuations in water quality
will affect the operation of treatment processes. Key
parameters to consider include temperature, total organic
carbon and/or DOC, algae, turbidity, taste and odor, and
pesticides and herbicides. For most geographic locations,

TABLE 2

four-season testing is advantageous if feasible within budget and scheduling constraints. In any case, the testing
should be at least of sufficient duration to develop adequate design criteria that meet project-specific performance
objectives and determine the operating limits of the membrane system in order to minimize capital costs.
The scope of testing can be partially dictated by the
size and complexity of the proposed full-scale plant. For
example, for smaller planned facilities, conservative pilot
system operating parameters can be selected to accommodate a shorter testing period. In such cases, scaling up
the results of a more limited data set will result in a
smaller overall effect on project cost. However, for larger
facilities, the same limited data set may yield a significantly oversized system at substantially higher cost to the
project when the results are scaled up over a much greater
number of membrane racks. Consequently, for larger
systems, particularly those using pretreatment processes,
it is generally beneficial to conduct more involved piloting of longer duration.
In order to adequately evaluate the performance of
each membrane system piloted, basic operating benchmarks should be established before conducting the testing. At the least, benchmarks should be determined for

Integrated membrane system schemes and associated target water quality contaminants
Integrated Membrane System

Target Contaminant*

Preliminary Membrane Treatment System


Membrane filtration UV disinfection + hydrogen peroxide

Taste- and odor-causing compounds, VOCs, PPCPs, MTBE,


nitrosamines (e.g., NDMA)

Membrane filtration ozonebiological filtration

Taste- and odor-causing compounds, VOCs, PPCPs, MTBE,


nitrosamines (e.g., NDMA), NOM

Membrane filtration GAC

Taste- and odor-causing compounds, VOCs, PPCPs, MTBE,


nitrosamines (e.g., NDMA), NOM

Membrane filtration NF or RO

Taste- and odor-causing compounds, VOCs, PPCPs, MTBE,


nitrosamines (e.g., NDMA), NOM, hardness, TDS

Intermediate Membrane Treatment System


Conventional pretreatment membrane filtration NF or RO

Elevated or variable turbidity, NOM, hardness, TDS

Lime softening membrane filtration NF or RO

Elevated turbidity, NOM, hardness, TDS

Final Membrane Treatment System


Preoxidation membrane filtration

Iron and manganese, taste-and-odor compounds

Adsorption membrane filtration

NOM

In-line coagulation membrane filtration

NOM, arsenic

Conventional pretreatment membrane filtration

NOM, elevated turbidity

Lime softening GAC membrane filtration

Hardness, elevated turbidity, taste- and-odor-causing compounds

Lime softening conventional pretreatment


membrane filtration

Hardness, elevated turbidity, NOM, iron and manganese,


taste- and odor-causing compounds

Ozone contact flocculationPAC


membrane filtration

Elevated turbidity, NOM, taste and odor compounds, synthetic


organic chemical (pesticides)

Full conventional treatment membrane filtration

Membranes used for enhanced particulate removal (effluent polishing)

GACgranular activated carbon, MTBEmethyl tertiary butyl ether, NDMAN-nitrosodimethylamine, NFnanofiltration, NOMnatural organic matter, PAC
powdered activated carbon, PPCPpharmaceuticals and personal care products, ROreverse osmosis, TDStotal dissolved solids, UVultraviolet, VOCvolatile
organic compounds
*In addition to naturally occurring particulate and microbial matter
Including PAC and suspended-type ion exchange resin
Cristal process, Degrmont TechnologiesAquasource, Toulouse, France

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

91

the minimum acceptable system recoveries and interval


between chemical cleanings. In general, high operating
fluxes and recoveries lead to more accelerated fouling
rates and shorter chemical cleaning intervals. The target
cleaning interval should balance chemical use, system
downtime, and labor with the capital cost associated with
additional membrane units that would enable the overall
system to operate at lower fluxes, thus reducing the rate
of fouling. Membrane system recovery is the percentage
of the membrane feed that is filtered and made available
for distribution, which is decreased by cleaning, backwashing, and other intermittent system-specific processes
that require filtered water. A wide range of recoveries is
possible; however, membrane processes for potable water

dancy and reliability, site space availability, future expansion requirements, and desirable operational characteristics. Membrane-specific issues are those that are unique to
the individual membrane equipment supplier, such as the
flux, type of membrane units (i.e., encased or submerged),
backwash approach, cleaning strategy, and process residuals generation. System-specific issues are related to the
equipment selection for the facility and include building
layout and orientation, materials and construction, control
and operational approach, pumping, chemical cleaning
system design, backwash and chemical cleaning residuals
streams disposal, and compressed air systems. These issues,
summarized in the sidebar on page 88, are sorted by category along with several important considerations.
Although this sidebar highlights
some of the most significant design
issues associated with membrane
filtration systems, it is not intended
to be comprehensive. Because there
are numerous membrane treatment
equipment suppliers, there may be
significant variation in the design
features among different products,
particularly with respect to the
more nuanced aspects. Therefore,
prior to the selection of a membrane filtration system, it is important to both consider the design
issues (summarized in the sidebar) relative to specific
utility needs and to work with the suppliers to understand their respective systems as thoroughly as possible.
After selecting a system, utilities and engineers should
work closely with the selected equipment suppliers during
the design phase.

he interaction between microfiltration/ultrafiltration membranes


properties and water quality is generally not well understood, making
it difficult to anticipate appropriate operating conditions.

production typically achieve 9097%, and the target for


piloting should generally be in this range. The recovery
may be higher if a secondary MF/UF stage is used to treat
backwash water from the primary stage.
Several variables that have a significant effect on the
cost of implementing MF/UF processes can be included
in the experimental matrix, including flux, backwash
interval, concentrate bleed, and cleaning strategies. The
design flux is critical, because this parameter, along with
the recovery and downtime for intermittent operating
procedures, ultimately dictates the total membrane area
required to achieve a given production capacity. Consequently, the pilot program should test a full range of
reasonable fluxes. Other important pilot-testing activities include integrity testing and water quality monitoring. The results of the pilot testing should be compiled
in a report summarizing key findings and developing
design criteria to be used for full-scale process design.

DESIGN CONCEPTS
In general, the design approach for an MF/UF system
should balance the requirements of the utility and the
membrane equipment supplier. The design concepts applicable to most MF and UF processes can be grouped into
three categories: site-specific issues, membrane-specific
issues, and system-specific issues. Site-specific issues include
the capacity, raw water quality and variability, demand
profile of the utility, finished water quality goals, redun92

OPERATIONS
The two primary goals of membrane systems operations are maintaining productivity and maintaining filtrate
quality. Because the ability of MF/UF systems to remove
pathogens and particulate matter is independent of raw
water quality for an integral membrane barrier, it is easier
to achieve high-quality filtered water on a consistent basis
with membrane filtration than it is with conventional
media filtration. However, lower-quality feedwater does
affect the efficiency of system operation. Consequently,
relative to that for a conventional plant, a more significant
proportion of operator responsibility is dedicated to maintaining productivity. For example, declining feedwater
quality may foul the membranes more quickly and perhaps
reduce the effectiveness of routine backwashing, resulting
in more frequent chemical cleaning. In such cases, operators may be able to minimize fouling by optimizing pretreatment and/or using a more aggressive backwash process. Under more extreme conditions, operators may need
to lower the flux to control fouling, which could reduce
production capacity. A marked decline in filtered water

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

quality is generally attributable to breaches in the membrane barrier. Therefore, both direct integrity testing and
continuous indirect integrity monitoring are essential for
proper membrane system operation.
The primary aspects of MF/UF operations can be
categorized as plant control, process monitoring, and
data collection and record keeping. In order to ensure
effective operations, each of these aspects should be carefully considered and addressed in the water treatment
plant operations and maintenance plan. Important factors for each of these three aspects of operations are
summarized in the sidebar on page 90, along with key
considerations. Membrane technology is a relatively new
and rapidly changing field with various commercially
available systems and continuous introduction of new
and updated membrane products. As a result, operational
aspects of specific systems can differ significantly in some
cases. Thus, although general operations are summarized
in this article and in M53, the membrane system suppliers should be consulted for more detailed, product-specific information.

SYSTEM COSTS

ity connections for several pilot-scale systems, equipment


rental fees, and operator time, all of which have a cost that
should be taken into account.
Capital cost components of a membrane filtration
system that are atypical for conventional treatment plants
include membrane equipment, a substantial number of
valves, feed pumps (because in-plant pumping is typically
required) with variable-frequency drives, strainers, air
compression equipment (as necessary for pneumatic valves,
integrity testing, backwashing, and similar processes), and
chemical cleaning equipment (i.e., tanks, metering pumps,
and similar equipment), including facilities for spent cleaning solution neutralization. As previously noted, tanks may
also be required for submerged membrane systems if
appropriately sized existing tanks or basins (e.g., in the
case of a retrofit) are not available. Special considerations
for the operation of a membrane filtration facility include
energy, chemical cleaning, and membrane replacement.
Generalized cost curves for membrane equipment, membrane filtration facilities (including construction and installation as well as the membrane equipment), and facility
operation as a function of system capacity are provided in
Figures 2, 3, and 4, respectively. Because both the
capital and operating costs of a membrane filtration facility
vary with the flux at which the system is operated (i.e.,
fewer membrane modules are required at higher fluxes), all
three figures include curves for high, medium, and low

Equipment Cost$/gpd

Although many of the cost considerations for membrane filtration are similar to those for conventional plants,
there are some important differences. Foremost among
these differences is the proprietary nature of the membrane
equipment and the associated procurement process. Because MF/UF
equipment represents a substantial
FIGURE 2 Membrane filtration equipment costs (2003)
portion of the cost of a membrane
filtration facility, differences among
Low flux
the various proprietary systems,
Medium flux
such as appurtenant equipment,
High flux
footprint, and installation, can
have a significant effect on the
10
overall project cost. For example,
submerged membrane systems
require tanks, which may be expensive to construct for a new plant.
However, in retrofit applications in
which existing tanks can be used,
the need to construct a separate
building for an encased membrane
1
system may be more expensive in
some cases. Another differentiating
factor to consider for membrane
filtration is piloting various proprietary systems, which is generally
recommended practice in the development of an MF/UF facility but
typically not associated with conventional treatment plants. A thor0.1
0.1
1
10
ough pilot project may require
significant time, some minor conTotal Capacitymgd at 20C
struction to accommodate the util-

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

100

93

fluxes. Note that what is considered


a high, medium, or low flux
varies among different proprietary
systems. The costs cited in each figure are estimated to be accurate
within 25%. Additional details
regarding what factors are incorporated into Figures 2, 3, and 4, as
well as applicable assumptions, are
provided in M53. Costs should
always be escalated for the current
year relative to that of the original
publication of the manual.

FIGURE 3 Membrane filtration facility costs (2003)

Low flux
Medium flux
High flux

Construction Cost$/gpd

10

RESIDUALS CHARACTERISTICS
AND MANAGEMENT

0.1
0.1

10

100

Total Capacitymgd at 20C

FIGURE 4 Membrane filtration system operations and maintenance costs (2003)

High flux
Medium flux
Low flux
0.16

O&M Costs$/1,000 gal filtrate

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0.00
0.1

10

100

Capacitymgd
O&Moperations and maintenance

94

The primary residuals associated with MF/UF systems are the


waste streams generated from
backwashing and chemical cleaning operations (either as-needed
and more aggressive intermittent
CIP or more modest and routine
CEB). Encased MF/UF systems
operating in a cross-flow mode
also generate concentrated waste
streams, which are similar to
backwash residuals. With the
exception of those systems operating in cross-flow mode (or
similar variant with a bleed
stream), MF/UF systems typically
produce residuals on an intermittent basis as a result of regular
backwashing. However, for
larger systems, the combination
of multiple units backwashing at
staggered intervals may generate
a waste stream on a more continuous basis.
Backwash residual characteristics depend on the quality of the
membrane system feedwater as
well as on the recovery of the
membrane system. At typical
recoveries of 9097%, the solids
in the feed are concentrated by a
factor of 720 in the backwash
residuals. Backwash flows generally represent about 9599% of
the total residuals volume from
MF/UF systems; the remainder is
generated from the cleaning processes used to control fouling.
Waste cleaning solutions reflect
the chemicals used in the cleaning

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

process combined with the extracted foulants. The


volume of CEB waste is typically very low; for example,
a daily CEB generally represents < 0.20.4% of the
plant flow. Residuals from a CIP conducted monthly
(i.e., a typical industry benchmark interval) will normally be < 0.05% of plant flow. These volumes may be
reduced if the cleaning solutions
are recycled or reused. The characteristics of typical backwash and
chemical cleaning wastes are summarized in Table 3. Note that pH
adjustment and/or dechlorination
may be necessary prior to disposal
depending on the chemicals used
and the regulations applicable to
the method of disposal employed.
Disposal of MF/UF residuals is
generally governed by the same regulations that control
disposal of conventional treatment residuals, as summarized in Table 4. Additional state and/or local regulations
that protect sensitive ecosystems, such as cold-water fisheries or pristine watersheds, may also be applicable.
The selection of residuals disposal methods depends
on several site-specific factors: climate, availability of
land for facilities, size of the MF/UF installation, feasibility, and federal, state, and local requirements.
According to a US Bureau of Reclamationsponsored
survey of 65 MF/UF installations (Mickley, 2003), the
most common methods for backwash residuals disposal were surface water discharge (38%); sewer discharge (25%); land application, such as percolation
ponds or irrigation (22%); and recycling (14%). Treatment was provided at a small number of locations to

meet specific discharge requirements. Only one small


(0.12-mgd [0.45-ML/d]) installation in Oklahoma used
an evaporation pond. The most common methods of
CIP residuals disposal were sewer discharge (37%),
land disposal (25%), and surface water discharge
(11%). Blending, neutralization, and other treatment

he selection of residuals disposal methods depends on several


site-specific factors including climate, availability of land, and size
of the installation.

TABLE 3

were frequently provided in conjunction with these


disposal methods.
Additional detailed information regarding the management of membrane residuals is provided in an AWWA
report (AWWA Residuals Management Research Committee Subcommittee on Membrane Residuals Management, 2003).

FUTURE TRENDS
In conjunction with the development of M53, a questionnaire was issued to a limited group of professionals
with recognized MF/UF expertise in order to determine
some of the most commonly anticipated future trends
associated with the technology. The questionnaire was not
intended to be scientific but rather was meant to provide
a general outlook for MF/UF by some of the individuals

Typical characteristics of microfiltration/ultrafiltration residuals

Parameter/Characteristic

Chemically Enhanced
Backwash

Backwash

Clean-in-place

Frequency of application

3060 min

Daily to weekly

Volume of waste% of feed flow

310 (for 9097% recovery)

0.20.4

14 months
< 0.05

Chemicals commonly used

None

Citric acid, hydrochloric


acid, hypochlorite

Hydrochloric acid, citric acid,


caustic soda, hypochlorite,
surfactants

Total organic carbon

No coagulation: 12 feed

130 feed

1030 feed
NA

With coagulation: < 5 feed


Particulate concentration factor

720 feed (for 8595% recovery)

NA

pH

69

212

214*

Hypochloritemg/L

NA

101,000

5001,000

Biological oxygen demand


over 5 daysmg/L

NA

5,00010,000
(if citric acid is used)

Surfactants

NA

Low concentrations (if used)

*High pH not for cellulose acetate membranes

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

95

TABLE 4

Microfiltration/ultrafiltration residuals disposal methods and regulations

Disposal Method
Surface water discharge

Regulation

Representative Discharge Limitation

NPDES permitting under the Clean Water Act

Groundwater discharge
(only if aquifer is closely
linked to surface water)

pH: 69
TSS: < 30 mg/L plus raw water TSS
Chlorine: < 0.2 mg/L
BOD5: < 30 mg/L
Minimum 90% removal of TSS and cysts

Groundwater discharge
Percolation ponds and
leaching fields

Underground injection control


Varies with:
permitting under the Safe Drinking Water Act
Water quality in the receiving aquifer
Nature of geology

Any other methods that may


affect groundwaters
Sewer or public owned
treatment works discharge

State rules will likely apply


Industrial Pretreatment Program
under the Clean Water Act

pH: 69
TSS: < 400500 mg/L
Chlorine: < 10 mg/L
BOD5: < 400500 mg/L
Nothing that will harm infrastructure or operations or result
in NPDES violations

Landfill disposal

Resource Conservation and Recovery Act

Pass paint filter test (typically > 20% solids)


Pass toxicity test (TCLP); residuals that fail are classed as
hazardous and must be properly disposed

Land application of solids

Solid Waste Disposal Act

Limitations on cumulative loadings of metals

BOD5biological oxygen demand over 5 days, NPDESNational Pollutant Discharge Elimination System, TCLPtoxicity characteristic leaching procedure, TSS
total suspended solids

who closely track membrane filtration. Issues addressed


in the questionnaire included system capacity, standardization, new materials and modules, technology advancements, regulations, applications, and costs.
The upper benchmark for the size of membrane filtration facilities is expected to continue to get larger. Several
MF/UF facilities in the 90- to 100-mgd range are cur-

There are widely acknowledged advantages and disadvantages to increased MF/UF system standardization,
as summarized in Table 5, and thus no consensus in
regard to the prevailing trend. However, there is general
agreement that one predominant factor that could facilitate increased standardization is the emergence of a
market for replacement membranes. Although such a
market may develop in the long term, with
the majority of MF/UF installations commissioned in the past 58 years, most membrane filtration systems are still many years
away from the point at which significant
membrane module replacement would
potentially be necessary.
Future membrane materials are expected
to be more robust in terms of not only
mechanical strength but also resistance to
chemical degradation and fouling. Examples of such materials are ceramics, which
are already commercially available, and polymers with
custom-fabricated nano-engineered particles in the
matrix. Hollow fibers, by far the most common membrane material configuration, should continue to dominate the market for the foreseeable future, but innovative
fiber bundling to optimize flux, increase module surface
area, and reduce fouling is anticipated.
The most significant technology advances are expected in the area of integrity verification, including

uture membrane materials are expected to be more robust


not only in terms of mechanical strength but also resistance
to chemical degradation and fouling.

rently under design, in construction, or already in operation, with bigger facilities on the horizon. This is a trend
that will likely be driven by MF/UF retrofits of older
conventional-media filtration plants. A divergence of
characteristics between small- and large-capacity systems
is also anticipated, with smaller facilities more commonly
using encased, skid-mounted packaged systems and larger
plants more likely to adopt submerged systems, particularly in retrofit applications.
96

DECEMBER 2008 | JOURNAL AWWA 100:12 | PEER-REVIEWED | AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS

2008 American Water Works Association

TABLE 5

Summary of microfiltration/ultrafiltration standardization issues

Advantage

Disadvantage

Competitive module replacement

Reduced innovation

Lower system costs

More limited equipment options

Simplified equipment procurement


Less variable regulatory permitting
Facilitator

Impediments

Lucrative market for replacement modules

Patents

Equipment specification requirements

Current diversity of effective membrane material, backwashing/chemical


cleaning practices, and operation strategies (e.g., flux management)

the development of a direct integrity test that can be


applied on a continuous basis while the MF/UF system is producing water, overcoming the biggest limitation of current testing methods. Another potential
innovation is the use of sonic vibrations to control
membrane fouling.
It is widely anticipated that the recent LT2ESWTR
and associated Membrane Filtration Guidance Manual
(USEPA, 2005), representing the first federal regulatory framework that directly addresses membrane filtration, will result in increasing standardization of the
diverse state primacy agency policies for regulating
MF/UF. In addition, larger pathogen removal/inactivation credits could be awarded to systems that use both
MF/UF and ultraviolet light, a treatment scheme that
would also reduce chemical use and therefore DBP
formation. This combination could eventually become
the conventional treatment of the future. The applications for MF/UF should continue to diversify, particularly when used in conjunction with a variety of
pretreatment and advanced oxidation processes.
MF/UF processes are expected to become increasingly cost-competitive with media filtration, a trend
driven by superior performance, reduced production
cost (e.g., because of economies of scale), improved
design, significant increase in demand for the technology, and competition. As costs stabilize over the long
term, a well-established MF/UF market should be
more influenced by traditional factors including supply and demand, materials cost fluctuations, and inflation. At this point, for example, a system with newer
innovations, more advanced features, and/or a better
performance record should command a higher price
rather than serving as an incentive to purchase one
comparably priced system over another. In general,
for a system exceeding 20 mgd in capacity, the cost of
membrane filtration (i.e., equipment only) in 2015 is

estimated to range from $0.10 to $0.15/gpd of


installed capacity (2004 dollars), with operations and
maintenance costs between $0.05 and $0.20/kgal of
treated water (2004 dollars).
ABOUT THE AUTHORS

This article was prepared by the AWWA Subcommittee


on Periodical Publications of the Membrane Process
Committee. Primary authors were Brent Alspach
(subcommittee chair), Samer Adham, Trevor Cooke,
Paul Delphos, Jesus Garcia-Aleman, Joe Jacangelo,
Ali Karimi, Jonathan Pressman, Jim Schaefer,
and Sandeep Sethi.

REFERENCES
AWWA Residuals Management Research Committee Subcommittee
on Membrane Residuals Management, 2003. Committee Report:
Residuals Management for Low-Pressure Membranes. Jour.
AWWA, 95:6:68.
Cheryan, M., 1998. Ultrafiltration and Microfiltration Handbook. Technomic Publishing, Lancaster, Pa.
Mickley, M., 2003. Membrane Concentrate Disposal: Practices and
Regulation (2nd ed.). US Bureau of Reclamation Technical Service
Center, Denver.
USEPA (US Environmental Protection Agency), 2005. Membrane Filtration Guidance Manual. EPA 815-R-06-009. Ofce. of Water,
Washington.
USEPA, 2001. Low-pressure Membrane Filtration for Pathogen
Removal: Application, Implementation, and Regulatory Issues.
EPA 815-C-01-001. Ofce. of Water, Washington.

If you have a comment about this


article, please contact us at
journal@awwa.org.

AWWA SUBCOMMITTEE ON PERIODICAL PUBLICATIONS | 100:12 JOURNAL AWWA | PEER-REVIEWED | DECEMBER 2008

2008 American Water Works Association

97

También podría gustarte