AN ULTRASONIC METER TO CHARACTERIZE DEGREE OF
FOULING AND CLEANING IN REVERSE OSMOSIS FILTERS
M. Morra, L.J. Bond, G.R. Golcar
Pacific Northwest National Laboratory
P.O. Box 999
Richland, WA 99352
ABSTRACT. The development of prognostic capabilities that predict the condition and
remaining service life for key industrial systems has the potential to significantly impact
performance and the economics of operation for both current and next generation plants. This
paper describes an on-line real-time ultrasonic meter that can be used to monitor both fouling
and cleaning in reverse osmosis filters. It provides a measure for the degree of fouling. A suit
of ultrasonic transducers is mounted to operate through the filter-housing wall on a pilot-scale
service water system. A "Degree of Fouling" index is given during both fouling and cleaning
for the filters during operation for processing of saline solutions (simulated sea and brackish
waters) and solids contamination. The fouling index is transmitted to a central computer
where it is integrated in a system level prognostic algorithm.
INTRODUCTION
In many systems including nuclear plant pure water supply it is necessary to develop
on-line self-diagnostic monitoring capabilities [1]. Such systems supply critical pure water
sources for safety system and makeup requirements. The development of prognostic
capabilities that predict the condition and remaining service life for key systems has the
potential to significantly impact performance and the economics of operation for both
current and next generation plants.
Ultrasonics has been used for more than 50 years for nondestructive testing [2] and is
now an increasingly common tool used for process monitoring and characterization [3,4,5].
Corrosion, erosion and fouling are common in service water systems. As a model system
to demonstrate the potential of ultrasonic measurements in prognostic methodologies, the
monitoring of fouling and cleaning in reverse osmosis (RO) filters was selected for study.
The fouling of such filters is a process that can be studied on system operational condition
changes induced and detected in operating fouling runs between 1 and 7 days.
Fouling in RO and other membrane based filters is also a topic of interest to the
membrane and process separations communities. The single most critical problem limiting
the application of membrane processes for liquid separation is fouling. For industrial
applications the optimization of the performance of spiral wound reverse osmosis filters
used for desalination, water reclamation or other industrial chemical processing it is
necessary to quantify condition in terms of a degree of fouling, during both fouling and
cleaning. Current technologies employ indirect measures that monitor either or both
pressures or permeate flux and do not provide data during the cleaning cycle. The
CP657, Review of Quantitative Nondestructive Evaluation Vol. 22, ed. by D. O. Thompson and D. E. Chimenti
© 2003 American Institute of Physics 0-7354-0117-9/03/S20.00
1673
development
development of
of real-time
real-time measurement
measurement techniques
techniques using
using ultrasonic
ultrasonic methods
methods for
for the
the
characterization
characterization of
of both
both membrane
membrane compaction
compaction and
and fouling
fouling represented
represented aa major
major advance
advance
[6].
[6]. This
This approach
approach has
has now
now been
been used
used in
in aa number
number of
of laboratory
laboratory studies
studies on
on flat-sheet
flat-sheet
membranes
[4,7,8,9,10].
membranes [4,7,8,9,10].
This
This paper
paper describes
describes the
the development
development and
and demonstration
demonstration of
of acoustic
acoustic time
time domain
domain
reflectometry
reflectometry for
for on-line
on-line and
and real-time
real-time monitoring
monitoring of
of fouling
fouling and
and cleaning
cleaning on
on aa pilot
pilot scale
scale
service
service water
water system.
system. The
The method
method is
is able
able to
to monitor
monitor early
early stage
stage contamination,
contamination, which
which
does
does not
not result
result in
in either
either aa pressure
pressure or
or permeate
permeate flux
flux change.
change. ItIt provides
provides aa measure
measure for
for the
the
degree
of
fouling,
which
is
a
useful
prognostic
in
that
it
allows
anticipatory
operations
degree of fouling, which is a useful prognostic in that it allows anticipatory operations and
and
maintenance
maintenance (O&M)
(O&M) responses
responses to
to developing
developing system
system degradation.
degradation. The
The net
net result
result isis higher
higher
throughput
throughput and
and significantly
significantly increased
increased reliability
reliability of
of the
the effected
effected water
water dependent
dependent safety
safety
systems.
systems.
EXPERIMENTAL
EXPERIMENTAL INVESTIGATION
INVESTIGATION
A
A suit
suit of
of ultrasound
ultrasound transducers
transducers mounted
mounted to
to operate
operate through
through the
the filter-housing
filter-housing wall
wall was
was
deployed
during
the
operation
of
a
pilot-plant
scale
service
water
system
to
purify
deployed during the operation of a pilot-plant scale service water system to purify saline
saline
solutions
solutions (simulated
(simulated sea
sea and
and brackish
brackish waters)
waters) and
and also
also to
to remove
remove solids.
solids. Combinations
Combinations of
of
both
both pulse-echo
pulse-echo and
and transmission
transmission measurements
measurements are
are employed.
employed. Transducers
Transducers operate
operate with
with aa
multiplexer,
multiplexer, digitization
digitization and
and distributed
distributed signal
signal processing
processing to
to give
give feature
feature extraction
extraction that
that
forms
forms the
the bases
bases for
for an
an index
index that
that quantifies
quantifies “Degree
"Degree of
of Fouling.”
Fouling." This
This index
index can
can be
be
measured
measured during
during both
both fouling
fouling and
and cleaning,
cleaning, and
and provides
provides aa direct
direct linkage
linkage to
to the
the impurity
impurity
concentration
concentration or
or input
input stressor
stressor level
level of
of the
the fluid
fluid stream.
stream. The
The fouling
fouling index
index is
is then
then
transmitted
to
a
central
computer
where
it
is
integrated
in
a
system
level
prognostic
transmitted to a central computer where it is integrated in a system level prognostic
algorithm
algorithm [11].
[11].
Apparatus
Apparatus
Most
Most RO
RO technology
technology uses
uses aa process
process known
known as
as crossflow
crossflow that
that allows
allows the
the membrane
membrane to
to
continually
purge
impurity
accumulation
while
in
operation.
Because
of
the
effectiveness
continually purge impurity accumulation while in operation. Because of the effectiveness
of
of this
this design,
design, itit was
was necessary
necessary to
to provide
provide for
for the
the management
management and
and control
control over
over the
the filter’s
filter's
performance
performance range.
range. The
The pilot
pilot scale
scale service
service water
water system
system [1,11]
[1,11] in
in aa laboratory
laboratory setting
setting
provided
provided the
the experimental
experimental mechanism
mechanism to
to establish
establish operational
operational parameters
parameters and
and maintain
maintain the
the
performance
performance characteristics
characteristics of
of commercially
commercially available
available reverse
reverse osmosis
osmosis filters.
filters.
FIGURE
user interface.
interface.
FIGURE 1.
1. System
System graphic
graphic user
1674
FIGURE 2. Filter bank with flow direction and transducer placement. (1) Concentrate inlet flow, (2)
Concentrate outlet flow, (3) Permeate flow clean water outlet.
The pilot scale water treatment system used for the fouling experiments is a closed
flow loop system. This is shown in the user interface display in Figure 1. A central
computer provides real time operational system parametric updates of pressures, flow rates
and temperatures at key locations throughout the system. Control for motorized valves and
temperature is also provided. The system is able to render flow rates, pressures and
temperatures consistent with the filter manufacturer's specifications. The filter bank
consists of six filters that are aligned in parallel and in series, where there are two (2) sets
of three (3) parallel filters in series as shown in Figure 2.
Fouling trials were conducted separately for both the saline solution and solid
suspensions. In this closed flow loop system, fouling in the filters was induced by
gradually increasing the concentration of the solution and solid suspension in a source
water holding tank. Cleaning trials were conducted for the saline solution only.
The concentration levels of saline solution and sand suspensions were increased to a
point beyond the normal performance range of the filters. Table salt was added to the
holding tank and dissolved in water with a large electric mixer impellor. For tests
employing salt-water, a WTW Multiline P4 Universal Multimeter was used to measure
salinity. Data from the meter where provided in both weight percent salinity and micro
Siemens per cubic centimeter.
To investigate the effect of particulate fouling
diatomaceous earth and sand were added to the water in the holding tank and kept in
suspension with the motorized impellor.
20
40
60
30
Time (micro seconds)
40
50
Time {micro seconds)
(a)
(b)
FIGURE 3. Signal captured for through transmission a) covering complete transit time across filter housing
and membrane layers, b) Pulse/Echo signal with multiple reflections of filter casing and membrane layers.
Time covers range from transducer face to center bore region of filter.
1675
Ultrasonic Measurement Systems
The ultrasonic system consisted of several 1" (2.5 cm) diameter 500 kHz and 1" x 0.5"
(2.5 x 1.25-cm) 500kHz flat compressionEQuinn@mdmcorp.com wave transducers. The
frequency utilized was the highest frequency that could penetrate the fiberglass composite
filter casing and the RO membrane layers to provide acceptable signal to noise ratios
(Figure 3a - 3b). Stand off shoes were attached to the transducers to enable a more secure
attachment to the curved surface of the filter casing. Transducers were coupled to the filter
casing with both a high viscosity ultrasonic gel couplant and a thin layer of solid skin type
couplant to conform to the irregular fiberglass surface. The transducers were placed at
locations throughout the filter system layout using both pulse echo and transmission time
domain reflectometry to monitor evidence of filter fouling and membrane compaction at
various locations and to determine if fouling in filters is position dependent. The
equipment used with the suit of transducers were commercial units: Ritec square wave
broad band pulser-receiver, LeCroy digital oscilloscope, and Krautkramer and Staveley
multiplexers.
Ultrasonic measurements were taken through transmission across the entire filter and
pulse-echo mode, where the time of flight signals were limited to the distance between the
outer casing and the center bore of the RO filter. Examples of RF data are shown in Figure
3a and 3b.
The transmission signals were zoomed in to focus on the received signal's time domain
and relative amplitude in a limited window as shown in Figure 4. Each measurement
configuration was implemented with four (4) transducers or transducer pairs. The signals
for each transducer were collected individually by stepping through the multiplexers. At
each step of the fouling process signals were collected and stored on floppy disk. At the
end of each run, digitized signals were transferred to a desktop computer for display,
analysis and output.
Solute Fouling Trial
Fouling trials for the saline solution was conducted on the flow loop system with clean
filters installed and under normal operating conditions set within the manufacturer's
performance ranges. There are multiple of variables that establish this range. Some are
based on the filter diameter, length and backpressure requirements [12]. Each trail began
with system flush using pre-filtered de-ionized water where ultrasonic signals could be
verified and established as a baseline reading through the filter casing and membrane
layers with the system in operation.
400
200
-000
75
85
95
Time (micro seconds)
FIGURE 4. Waveforms shown superimposed as an expanded view of three signals received through the
filter geometry.
1676
PermeateSalinty/Flow Rate
±
|2
I 1
1
2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17
Salinity (wt%)
FIGURE 5. Salinity concentration of the permeate water flow (•). The flow rate for the permeate flow
with change in source water salinity (•).
Ultrasonic signals from the pulse-echo measurements gave data that were complex to
analyze and results were inconclusive due to effects of constructive and destructive
interference at the filter casing and membrane layer interfaces. Although there was a
detectable change, the signals were not readily intelligible.
An alternate approach using transmission measurements with transducers placed on
opposite sides of the filter casing provided a more reliable measure for filter status in real
time. Ultrasonic signals captured over time as the salt concentration was increased, show a
measurable shift in arrival time of the signals, as shown in Figure 4. There is also a
measurable change in amplitude of the received signal as the salt concentration increases.
The effective operating range of the filters used in this trial is shown in the data given
in Figure 5. There is a maximum source salinity where the filters reach their saturation
point and are no longer able to provide clean water. This occurs at approximately 12%
source water salinity, after which the permeate flow or clean-water side experiences a
dramatic increase in salt concentration that continues to rise as the source concentration is
increased. Feed flow is shown to level off at approximately 14% salt concentration. At
this point any increase in source salinity has no effect on the permeate flow rate when all
other system parameters remain the same.
The trends in the ultrasonic data over the performance range of these filters are shown
as graphs in Figures 6a and b. Relative signal amplitude rises with the increase in saltwater concentration. The time required to travel across the filter membranes is reduced,
which indicates an increase in the ultrasonic sound velocity through, the membranes and
solution system, as a function of salinity. The received signal does not show a dramatic
shift or change at the 12% concentration level when the filter has reached saturation. The
TOP
Amplitude
450
^^
Jf^^
>
F430
S
^***>sm
! 81.4
r*~
3 420
•""*••—B , ^
^•^
^^'^
H
,-^ —————————
J^+
""•
.•van
""*""
81
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17
Salinity (wt%)
1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Salinity (wt%)
(a)
(b)
FIGURE 6. Ultrasonic time-of-flight (TOF) and amplitude, a) Trend line showing increase in signal
amplitude, b) Ultrasonic (TOF) sound velocity increase as a function of source salinity.
1677
600
400
200
0
75
85
95
Time (micro seconds)
FIGURE 7. Transmission data waveforms showing TOP and amplitude trend during an example of a
particulate fouling.
ultrasonic signals from all the various positions on the filter bank were observed to behave
similarly.
In measured data the received signal does not show a dramatic shift or change at the
12% concentration level when the filter has reached saturation, and there is no visible
measure of build up or membrane compaction shown in the time domain data. The
received ultrasonic signals cover the entire range of the filter's performance and effects are
found to be reversible during cleaning. Therefore, when clean water is reintroduced in to
the system and is used to flush the filters, the ultrasonic signals return to the baseline
readings.
Solids Fouling Trials
For fouling trials measuring the effects of solids on ultrasonic signals, using the pulseecho method once again proved inconclusive. The transmission method data was found to
exhibit similar trends as in the case of the salt-water trials. The ultrasonic signals captured
over time as the solid suspension was increased, exhibited a measurable shift in arrival
time as shown in Figure 7. There is also a measurable change in amplitude of the received
signal, which is shown to decrease unlike signal amplitude for the saline fouling trials,
which increased in amplitude with the solution concentration.
The range of performance for the filters was achieved by closing the valve that controls
backpressure on the filters. The backpressure provides the differential pressure required to
maintain the filter mechanism.
Inlet Side
81.9
1 81.1
fc
"xV
Discharge Side
81.9
'PS>*-«^
%
* • "s^
I 81.3
V
£ 81. 1
^ x •»
"^v
V.
%-
•
^>m^»—•
"*~*-^-»^::
""*-*
1 2 3 4 5 6 7 8 9 101112131415161718
Fouling (pt)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Fouling (pt)
(a)
(b)
FIGURE 8. Transmission signal first arrival time, a) At the filter inlet end, and b) At the discharge end.
1678
Inlet Side
"^
| 750
3 600
\
*-«v
son1
Discharge Side
I750
*-• .
" "X,
«^B_^
"•Ntf***-*-""
~^"-"
2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17 18
Fouling (pt)
:*-«^
'
. . .
Q.
!inn
1 2 3
4
5 6
7 8
"X
*v
~s»-«
9 10 11 1.2 13 14 15 16 17 18
Fouling (pt)
(a)
(b)
FIGURE 9. Transmission signal amplitude, a) At the inlet end and b) At the discharge end.
It was observed through the range of valve operation that there is a point at which the flow
of water through the filters will level off and the filters are no longer able to provide
additional clean water flow. For our experiment, the valve was closed in steps up to 95%.
For the range of closure from 55% to 95% there was no improvement in clean water flow,
the filters are essentially fouled. It was this parameter that was used as a measure of
performance and to correlate with the ultrasonic data.
In the case of solid particle fouling, it was observed that there were differences in the
signals collected from each end of the filter, as well as at various locations for the
transducers throughout the filter bank. This is shown in the example of data given as
Figures 7 and 8. This effect may be attributed to the degree of compaction on the
membrane surface as it relates to the filter's ability to operate under cross flow self
cleaning. To identify optimal transducer location for the monitoring of fouling, data was
taken at opposite ends of the filter and compared. It was observed that the filter bank that
was first in series experienced the most severe fouling and the inlet end of these filters also
had a great build up then the outlet end.
Examples of data from particulate fouling runs are shown as Figures 8 and 9. The data
from the inlet end of the filter exhibits a more dramatic change in the signal amplitude and
sound velocity than those measured near the outlet. The filter was able to maintain the
crossflow self-cleaning process until it was brought to a fouled state by changing process
parameters, after which the ability to self-clean was limited by the additional concentration
of solids. It is also probable that under these conditions solids are forced deeper into the
filter and this has the effect of redirecting a portion of the flow to a different filter bank that
was operating in parallel. The one anomalous data point seen in Figure 8a is believed to be
due to shifting solids within the filter as the backpressure was increased.
CONCLUSION
The ability of ultrasonics to provide data for use in prognostics methodologies has been
demonstrated. It has been shown to be possible to use a non-invasive ultrasonic method to
monitor early stage contamination in real-time for the effects of fouling in reverse osmosis
filters. The method was shown to operate over the full range of operational conditions.
The fouling ultrasonic meter was shown to have the ability to monitor fouling from
solids and identify the specific location where build up and compaction occurs. This
method can be applied to monitor both the fouling process and cleaning in filters and
provide a metric for the degree of fouling.
1679
ACKNOWLEDGEMENTS
The DOE-NERI program supported this work. Pacific Northwest National Laboratory
is operated for the United States Department of Energy by Battelle under Contract DEAC06-76RLO 1830.
REFERENCES
1. Bond, L.J., Jarrell, D.B. and Gilbert, R.W., "NERI: On-line intelligent self-diagnostic
monitoring system," Trans. American Nuclear Soc., 83, 184-186 (2000).
2. Krautkramer, J. and Krautkramer, H., Ultrasonic Testing of Materials, 4th Ed. SpringerVerlag (Berlin) 1990.
3. Lynnworth, L.C., Ultrasonic measurements for process control, Academic Press
(Boston) 1989.
4. Workman, J. and 16 others, Process Analytical Chemistry, Anal. Chem., 71, 121R180R(1999).
5. Workman, J. and 6 others, Process Analytical Chemistry, Anal. Chem., 73, 2705-2718
(2001).
6. Bond, L.J., Greenberg, A.R., Mairal, A.P., Loest, G., Brewster, J.H. and Krantz, W.B.,
"Real-time nondestructive characterization of membrane compaction and Fouling, in
Review of Progress in QNDE, 14, edited by D.O. Thompson et al., Plenum (New York)
1995, pp. 1167-1173.
7. Mairal, A.P., Greenberg, A.R., Krantz, W.B. and Bond, L.J., J. Membrane Sci., 159,
158-196 (1999).
8. Mairal, A.P., Greenberg, A.R. and Krantz, W.B., Desalination, 130, 45-60 (2000).
9. Li, J., Sanderson, R. and Jacobs, E. P., J. Membrane Sci. 201, 117-29 (2002).
10. Sanderson, R., Li, J., Koen, LJ. and Lorenzen, L., J. Membrane Sci. 207, 105-117
(2002).
11. Bond, L.J., Doctor, S.R., Jarrell, D.B., and Meador, R.J., NERI: On-line intelligent
self-diagnostic monitoring for next generation nuclear plants. 2001. Report PNNL13764.
12. www.osmonics.com
1680