October 2009
NSF 09/28/EPADWCTR
EPA/600/R-09/124
Environmental Technology
Verification Report
Removal of Inorganic, Microbial, and
Particulate Contaminants from a Fresh
Surface Water
Village Marine Tec.
Expeditionary Unit Water Purifier,
Generation 1
Prepared by
NSF International
Under a Cooperative Agreement with
U.S. Environmental Protection Agency
-------
This page is intentionally blank
-------
THE ENVIRONMENTAL TECHNOLOGY VERIFICATION
PROGRAM
U.S. Environmental Protection Agency
NSF International
ETV Joint Verification Statement
TECHNOLOGY TYPE: ULTRAFILTRATION AND REVERSE OSMOSIS
APPLICATION:
PRODUCT NAME:
VENDOR:
ADDRESS:
PHONE:
EMAIL:
REMOVAL OF CHEMICAL AND MICROBIAL
CONTAMINANTS FROM A SURFACE DRINKING WATER
SOURCE
EXPEDITIONARY UNIT WATER PURIFIER (EUWP),
GENERATION 1
VILLAGE MARINE TEC.
2000 W. 135TH ST.
GARDENA, CA 90249
310-516-9911
SALES@VILLAGEMARINE.COM
NSF International (NSF) manages the Drinking Water Systems (DWS) Center under the U.S.
Environmental Protection Agency's (EPA) Environmental Technology Verification (ETV) Program. The
DWS Center evaluated the performance of the Village Marine Tec. Generation 1 Expeditionary Unit
Water Purifier (EUWP). The EUWP, designed under U.S. Military specifications for civilian use,
employs ultrafiltration (UF) and reverse osmosis (RO) to produce drinking water from a variety of
sources. This document provides the verification test results for the EUWP system evaluated at a fresh
surface water site at Selfridge Air National Guard Base in Michigan.
EPA created the ETV Program to facilitate the deployment of innovative or improved environmental
technologies through performance verification and dissemination of information. The goal of the ETV
Program is to further environmental protection by accelerating the acceptance and use of improved and
more cost-effective technologies. ETV seeks to achieve this goal by providing high-quality, peer-
reviewed data on technology performance to those involved in the design, distribution, permitting,
purchase, and use of environmental technologies.
ETV works in partnership with recognized standards and testing organizations, stakeholder groups
(consisting of buyers, vendor organizations, and permitters), and with the voluntary participation of
individual technology developers. The program evaluates the performance of innovative technologies by
developing test plans that are responsive to the needs of stakeholders, conducting field or laboratory tests
(as appropriate), collecting and analyzing data, and preparing peer-reviewed reports. All evaluations are
conducted in accordance with rigorous quality assurance protocols to ensure that data of known and
adequate quality are generated and that the results are defensible.
NSF 09/28/EPADWCTR
The accompanying notice is an integral part of this verification statement.
VS-i
October 2009
-------
PRODUCT DESCRIPTION
The following technology description was provided by the manufacturer and has not been verified.
The EUWP was developed to treat challenging water sources with variable turbidity, chemical
contamination, and very high total dissolved solids (TDS) including seawater, during emergency
situations when other water treatment facilities are incapacitated. The EUWP components include feed
pumps, a UF pretreatment system, a one or two pass RO desalination system with an energy recovery
device, storage tanks, and product pumps. It has chemical feed systems for optional pretreatment
coagulation and post treatment chlorination. Clean-in-place systems (CIP) are included with the UF and
RO skids. During this verification test, coagulation pretreatment was employed, but chlorination was not
evaluated.
Design specifications indicate that the UF system alone has a production capacity up to 250,000 gallons
per day (gpd) from a fresh water source with up to 500 mg/L TDS and a temperature of 25°C. The
combined UF and RO system is designed to produce from 98,000 gpd up to 162,000 gpd, depending on
the TDS of the source water and the recovery settings of the RO process.
VERIFICATION TEST DESCRIPTION
Test Site
The testing site was Lake St. Clair at Selfridge Air National Guard Base in Michigan. The source water
for testing was raw lake water. Initial characterization samples of raw lake water were collected in August
2006, and again in May 2007 for the second round of testing. Highlights of the source water
characterization are presented in Table VS-i. The measured concentrations of regulated metals,
phosphorus, nitrite, and nitrate are not shown here, but are presented in the final report, because they are
either below the laboratory reporting limit or below the limit in the EPA National Primary Drinking
Water Regulations (NPDWR) limit.
Table VS-i. Lake St. Clair Raw Water Characterization Data
Sample Date
Parameter 08/16/06 05/31/07
Total Organic Carbon (TOC, mg/L) 2.9 NM1
UV Light Absorbance at 254 nanometers (UV254, Abs) 0.0668 NM
Total Suspended Solids (TSS, mg/L) <5 <2
TDS (mg/L) 130 140
Alkalinity (mg/L as CaCO3) 70 86
Total Hardness (mg/L as CaCO3) 95 110
Total Silica (mg/L as SiO2) 1.1 1.1
Specific Conductance (nmhos/cm) NM 250
Cryptosporidium (oocysts/L) <1 NM
Giardia (cysts/L) <1 NM
Heterotrophic Plate Count (HPC, CFU/mL) 500 NM
Total Coliforms (CFU/100 mL) 291 NM
Bacillus Endospores (CFU/100 mL) NM 689
(1) NM = not measured
Methods and Procedures
Initial testing of the EUWP was conducted in September and October of 2006 by the U.S Army Tank-
Automotive Research, Development, and Engineering Center (TARDEC), with assistance from the U.S.
Bureau of Reclamation (USER). Immediately prior to the ETV test, the initial UF pressure decay tests
indicated that pressure was being lost at a higher than desirable rate. The problem was investigated, and
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-ii
-------
was found to be the o-ring seals between the membrane modules and filtrate collection tubes. As a
temporary fix, polytetrafluoroethylene (Teflon®) thread sealing tape was wrapped around the o-rings to
increase the seal surface between the o-rings and membrane cartridges, and the test proceeded. After
testing was complete, the UF performance data indicated that the temporary fix did not maintain sufficient
membrane integrity. Therefore, a second test employing only the UF system was conducted in July and
August of 2007 after permanent repairs were made. Issues concerning the seal problems and subsequent
repairs are discussed in the ETV verification report.
The testing activities followed a test/quality assurance plan (TQAP) prepared specifically for the project.
The TQAP was developed in accordance with the ETV Protocols EPA/NSF Protocol for Equipment
Verification Testing for Removal of Inorganic Constituents - April 2002, and the EPA/NSF Protocol for
Equipment Verification Testing for Physical Removal of Microbiological and Paniculate Contaminants -
September 2005.
The 2006 verification test began on September 25, and ran for the planned 30 day test period, ending on
October 25. The UF system was operated each day on semi-continuous basis, automatically shutting
down when the RO feed tank was full. A typical operating day for the UF system was 15-17 hours (h) in
duration. The RO system was setup to operate continuously, and typically ran 22 to 24 h per day. The RO
system was shutdown periodically for various maintenance activities, or when alarms occurred and shut
the system down. When alarms and shutdown occurred during unattended operation at night, the entire
system would remain shutdown until an operator arrived in the morning.
The 2007 UF system retest was conducted from July 30 to August 24. The retest was stopped short of 30
days because the intent of the test as stated in the ETV test protocol - operation until a membrane
cleaning was needed - was met. During the retest, the UF system was in operation an average of 14 h per
day, not including down time for backwashes, cleanings, and other maintenance activities.
Flow, pressure, conductivity, and temperature recordings were collected twice per day when possible to
quantify membrane flux, specific flux, flux decline, and recovery. Turbidity and pH readings were also
recorded twice per day. The UF skid included in-line particle counters which recorded particle counts
every five minutes. Pressure decay tests were conducted daily on the UF system to verify membrane
integrity. Once per week samples were collected from the UF and RO process streams for analysis of
alkalinity, hardness, total silica, TDS, TOC, TSS, UV254, HPC (2006 test only), and total coliforms (2006
test only). For the 2007 test, Bacillus endospores were substituted for HPC and total coliforms.
VERIFICATION OF PERFORMANCE - 2006 TEST
Finished Water Quality
The UF system reduced the turbidity from a mean of 4.77 Nephelometric Turbidity Units (NTU) in the
feed water to a mean of 0.14 NTU in the UF filtrate. The UF system reduced the turbidity of the feed
water by a mean value of 95.9%. All filtrate turbidity measurements were below the NPDWR of 1 NTU.
The second NPDWR criterion for turbidity is that 95% of the daily samples in any month must be <0.3
NTU. Only one filtrate turbidity measurement out of 58 was above 0.3 NTU: 0.47 NTU on October 5.
Therefore, the EUWP UF system met the second NPDWR turbidity requirement, as 98% of the turbidity
measurements were <0.3 NTU.
The RO membranes provided additional turbidity removal, resulting in a mean turbidity of 0.09 NTU
from the permeate grab samples. The maximum measured RO permeate turbidity was 0.18 NTU. In
general, the RO system provided an additional turbidity reduction in the range of 40% to 66%.
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-iii
-------
The UF system showed only a minor reduction in organic material as measured by the TOC data. The UF
feed TOC concentrations ranged from 2.1 to 2.7 mg/L, and the UF filtrate levels were typically only 0.1
to 0.4 mg/L lower. These data indicate that most of the organic material, as measured by TOC, was
dissolved in the feed water. The RO system reduced the permeate TOC to below the detection limit of 0.1
mg/L.
The RO system also reduced the dissolved ions in the water, as measured by conductivity, with a mean
percent reduction of 99.4%. The mean conductivity of the RO permeate was 1.8 microSiemens per
centimeter ((iS/cm) compared to a mean RO feed conductivity of 287 (iS/cm. The maximum measured
permeate conductivity was 4.9 (iS/cm. Hardness, alkalinity, TDS, and total silica were all removed to
below the detection limit in the RO permeate.
UF andRO Membrane Integrity
Daily pressure decay tests were used to document UF membrane integrity, and HPC and total coliforms
were measured in the UF feed and filtrate as a microbial membrane integrity indicator. The in-line
particle counters provided an additional measurement of membrane integrity, and the capability of the
system to remove particulate and microbial contaminants.
As discussed in the Methods and Procedures section, prior to the 2006 test TARDEC and USER
discovered that the seals between the UF elements and membrane module housings were not as tight as
desired. After the problem was temporarily fixed, the pressure decay rate was measured as 0.37 pounds
per square inch, gauge (psig) per minute (min). While this was higher than desired, there was no critical
pressure decay rate to achieve, so the test proceeded. The mean daily pressure decay rate for the test was
0.29 psig/min, with a maximum observed decay rate of 0.43 psig/min.
While the turbidity data indicated that the UF system performed satisfactorily, the microbiological data
showed higher than expected UF filtrate counts. The UF feed geometric mean HPC count was 2810
CFU/mL, and the filtrate geometric mean HPC count was 1670 CFU/mL. Mean total coliform counts
were not calculated because only five sets of samples were collected. The UF feed total coliform counts
ranged from 41 to 532 CFU/100 mL, while the filtrate counts ranged from 11 to 94 CFU/100 mL. High
numbers of HPC and total coliforms were also found in the RO permeate. The mean RO permeate HPC
count was 247 CFU/mL and the RO permeate total coliform counts ranged from <1 to 95 CFU/100 mL.
This phenomenon has been observed in other published membrane studies, but it was beyond the scope of
this study to determine whether the observed HPC and total coliform levels were breaching the
membrane, or were a result of microbial contamination and growth downstream of the UF and RO
membranes from previous field tests of the EUWP.
There is no reportable particle count data for the 2006 test because after the test was completed it was
discovered that the particle counters had been improperly calibrated.
Direct integrity measurements of the RO system were performed prior to the start of the verification test,
and again at the end of the test. A dye marker test was conducted, where a food-grade dye was added to
the RO feed water, and UV absorbance levels were compared among the feed, permeate, and concentrate
streams over a ten minute period. For the pre-verification test, the dye rejection rate was 99.6%, while
that for the post-verification dye test was 99.8%. As with the UF pressure decay tests, there was no
critical rejection level.
UF System Operation
UF process operations data for the 2006 test are presented in Table VS-ii. The intake flow is defined as
the source water pumped into the UF feed water tank. The mean UF feed water flow rate of 246 gallons
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-iv
-------
per minute (gpm) was below the design feed flow rate of 259 gpm specified for the system. The UF water
recovery was 89.5% based on the mean feed water and filtrate flow rates. The UF system only operated
15 h per day, on average, but the 220 gpm mean filtrate flow corresponds to a 24-h production rate of
316,800 gallons (gal). The UF system target production rate was 250,000 gpd (not including backwash
water). The backwash process used about 900 gal of UF filtrate per event, and a backwash was conducted
every 30 minutes. For 24 h of operation, a total of 43,200 gal of UF filtrate would be used for
backwashes. Subtracting the backwash water from the calculated daily UF filtrate production results in
273,600 gpd of UF product water, which was above the performance goal of 250,000 gpd.
Table VS-ii. 2006 Test UF Operations Productivity Data
Parameter
Standard
Count Mean Median Minimum Maximum Deviation
95%
Confidence
Interval
UF Operation per day (h)
Intake Flow (gpm)
Feed Flow (gpm)
Filtrate Flow (gpm)
Retentate Flow (gpm)
Backwash Flow (gpm)
Feed Pressure (psig)
Retentate Pressure (psig)
Filtrate Temperature (0F:)
31
58
59
59
59
59
59
59
15.0
298
246
220
26
21
19
52
17.2
299
248
222
26
Estimated
21
19
52
3.4
278
175
149
21
at 900 :
12
10
43
21.5
302
268
243
31
4.85
3.34
16.0
16.1
1.81
±1.71
±0.86
±4.07
±4.10
±0.46
gal per backwash cycle
33
31
60
4.26
4.20
5.16
±1.09
±1.07
±1.32
(1) °F = degrees Fahrenheit
A chemical coagulant (ferric chloride) was not used at the beginning of the verification test. At the start of
the test on September 25, the trans-membrane pressure (TMP) was 11 psig. However, it quickly rose to
26 psig on September 29. As the TMP rose, the specific flux declined from 3.56 gallons per square foot
per day (gfd)/psig on September 25 to 1.38 gfd/psig on September 29. It was evident that a coagulant
should be used to attempt to lengthen the time between UF cleanings. The UF system was shut down on
September 30 and cleaned. The CIP was successful as the specific flux rose to 3.52 gfd/psig. Ferric
chloride was injected to the feed water upstream of the UF feed tank from September 29 through the end
of the test. The addition of the coagulant improved performance, and the system was able to maintain
filtrate production with the TMP below 20 psig until the last two days of the test. The specific flux varied
between 3.0 and 4.5 gfd/psig from September 29 to October 18, and then it dropped down to 2.46
gfd/psig on October 19. From October 19 to the end of the test on October 25, it ranged from
approximately 1.5 to 3.0 gfd/psig.
RO System Operation
The RO process operations data for the 2006 test are presented in Table VS-iii. The mean RO permeate
flows of 53 gpm for Array 1 and 21 gpm for Array 2 yield a mean total permeate production of 74 gpm.
The mean feed water flow of 107 gpm for Array 1 and 53 gpm for Array 2 were below the target feed
rates of 116 gpm and 58 gpm, respectively. The recovery for Array 1 was 49.5%, (design target 50%) and
the recovery for Array 2 was 39.6% (design target 48%).
Over the 30-day verification test, the RO feed water totalizer showed 5,382,670 gal of water fed to the
RO unit. At an average recovery of 47% (prorated between Array 1 at 49.5% and Array 2 at 39.6%), the
total volume of permeate produced was approximately 2,530,000 gal or an average of 84,330 gpd over the
entire test period. The target flowrate fell short of the goal of producing 100,000 gpd of finished water.
The RO system maintained a steady permeate flow rate for both arrays throughout the verification test.
The feed pressure was increased over the duration of the test to maintain feed water flow rates. The Array
NSF 09/28/EPADWCTR
The accompanying notice is an integral part of this verification statement.
VS-v
October 2009
-------
1 feed pressure increased from 387 psig on September 25 to a maximum of 539 psig on October 24. The
concentrate pressure from Array 1 was used by the energy recovery device to increase feed water pressure
for Array 2. Based on the small pressure loss from the transfer of pressure between the Array 1
concentrate and the Array 2 feed water, the energy recovery device worked properly during the test.
Table VS-iii. RO System Operations Productivity Data for 2006 Test
95%
Standard Confidence
Parameter Count Mean Median Minimum Maximum Deviation Interval
Array 1 Feed Flow (gpm) 59 107 107 104 110 1.38 ±0.35
Array 1 Permeate Flow (gpm) 59 53 53 44 56 2.0 ± 0.50
Array 1 Concentrate Flow (gpm) 59 54 54 48 62 2.4 ±0.61
Array 2 Feed Flow (gpm) 59 53 52 49 59 2.3 ±0.60
Array 2 Permeate Flow (gpm) 59 21 21 19 24 1.1 ±0.27
Array 2 Concentrate Flow (gpm) 59 32 31 27 37 2.3 ±0.58
Array 1 Feed Pressure (psig) 59 444 428 374 539 45.9 ±11.7
Array 1 Concentrate Pressure (psig) 59 346 330 286 419 40.5 ±10.3
Array 2 Feed Pressure (psig) 59 345 327 284 436 42.5 ±10.8
Array 2 Concentrate Pressure (psig) 59 255 238 204 325 35.2 ± 8.98
Array 1 and 2 Combined Permeate 59 28 27 15 39 4.6 ±1.2
Pressure (psig)
The specific flux calculations show that the RO membranes were slowly being fouled during operation.
Over the 30-day test, the specific flux dropped by approximately 31% for Array 1, from 0.050 to 0.035
gfd/psig and 26% for Array 2, from 0.054 to 0.040 gfd/psig. The RO system was chemically cleaned on
October 6 using a citric acid low pH solution. The specific flux just before the start of the cleaning was
0.043 gfd/psig, and the cleaning increased the specific flux to 0.047 gfd/psig. Given the slow but steady
trend of decreasing specific flux, an anti-sealant was fed to the RO system beginning on October 12. This
chemical feed continued through the end of the verification test.
VERIFICATION OF PERFORMANCE - 2007 UF SYSTEM RETEST
The 2007 retest was conducted from July 31 to August 24. Prior to starting the retest, each membrane
cartridge was individually integrity tested, and several were found to have broken fibers that required
plugging. This is a typical practice prior to installation of hollow-fiber membrane modules. After
plugging these fibers, each cartridge was again pressure tested. The results showed that 15 of the 16
modules were acceptable, so TARDEC and USER decided to operate the UF system with only 15
membranes. After completion of the individual module pressure decay tests and repairs, the full system
pressure decay rate was 0.025 psig/min. This value was more than ten times lower than the mean value of
0.29 psig/min obtained during the 2006 verification test. This indicated that the repairs made to the UF
system following the 2006 test were providing belter membrane module pressure-hold capability.
Finished Water Quality
For the 2007 retest, the UF system reduced the turbidity from a mean of 2.3 NTU in the feed water to a
mean of 0.14 NTU in the UF filtrate. Despite the UF system integrity issues during the 2006 test, the
2006 mean filtrate turbidity was the same as for the 2007 test. Turbidity in the feed water was reduced by
a mean value of 92.5%. There were two spikes in the feed water turbidity - on August 6, and from August
20 to 22. Both spikes were likely caused by rain events on these days. These feed water turbidity spikes
did cause small increases in the filtrate turbidity, but only one measurement - 0.51 NTU on August 22 -
was above 0.3 NTU. Therefore, the UF system also met the NPDWR turbidity requirements during the
2007 test.
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-vi
-------
UF Membrane Integrity
Pressure decay tests were again conducted daily for the 2007 UF system retest. The observed pressure
decay rates were 5-10 times lower than those from the 2006 test, with a mean value of 0.025 psig/min.
These direct integrity test results were indicative of membrane modules with no significant observable
breaches.
The mean 2 to 3 (im particle count for the feed water was 13,376/10 mL. The range of 2 to 3 um particle
counts for the feed water was 1 to 39,418/10 mL. The filtrate had a mean particle count in the 2 to 3 um
size of 112/10 mL with a median of 55/10 mL and a range of 0 to 13,908/10 mL. However, the maximum
particle count of 13,908 may not be indicative of typical performance. The UF system went through a
backflush cycle every half-hour, and during these backflushes the particle counts were still being
recorded. Consequently, the filtrate particle count data included numerous spikes. The backflushes were
not time-stamped, so the spikes due to backflushes could not be identified with certainty and removed
from the data set. As evidenced by the low mean and median filtrate counts, most of the counts were less
than 200/10 mL. The UF system reduced the 2 to 3 um particles by a mean value of 2.21 log 10.
The mean 3 to 5 um particle count for the feed water was 24,634/10 mL. The range of 3 to 5 um particle
counts for the feed water was 0 to 91,595/10 mL. The filtrate had a mean 3 to 5 um particle count of
157/10 mL with a median of 77/10 mL and a range of 0 to 14,059/10 mL. As with the 2 to 3 um
maximum count, the 3 to 5 um maximum count of 14,059 may not be indicative of UF performance due
to particle count data being collected during the backflushes. The UF system reduced the 3 to 5 um
particles by a mean value of 2.33 Iogi0.
The geometric mean UF feed Bacillus endospore count was 1,562 CFU/100 mL, with range of 862 to
7,420 CFU/100 mL. The mean filtrate endospore count was 203 CFU/100 mL, with a range of 78 to 996
CFU/100 mL. The mean log reduction was 0.88 logic with a range of 0.07 to 1.74 logio for the feed and
filtrate sample pairs. This was a lower reduction than predicted based on the observed pressure decay
rates and the particle count data. To explore the concern of membrane module integrity further, additional
studies were conducted on selected modules from this UF skid. Results from these additional studies
conducted at the NSF testing facility in Ann Arbor, MI, are not presented in this verification report. The
following reference report provides separate ETV verification testing results for the laboratory challenge
study of selected EUWP UF modules: "Removal of Microbial Contaminants in Drinking Water: Koch
Membrane Systems, Inc. Targa® 10-48-35-PMC™ Ultrafiltration Membrane, as Used in the Village
Marine Tec. Expeditionary Unit Water Purifier", EPA/600/R-09/075, http://www.epa.gov/etv.
UF System Operation
The 2007 UF system retest operations data are presented in Table VS-iv. With only 15 modules in
operation, the mean feed and filtrate flow rates of 232 gpm and 206 gpm, respectively, were lower than
those for the 2006 test. Based on the mean flow rates, the mean water recovery for the UF system was
88.8%. The 206 gpm mean filtrate flow corresponds to a 24-h production rate of 296,640 gpd. Subtracting
the backwash water from the calculated daily filtrate production results in 253,440 gpd of UF product
water, which is still above the design UF production of 250,000 gpd, despite being short one module.
Actual UF filtrate production was tracked using the RO feed totalizer. The total filtrate produced (not
including backwash water) was 3,551,000 gal over 350.1 h of operation. This yields a mean useable UF
filtrate production of 242,500 gpd. If the filtrate water used for backwashing the system is added (595,730
gal) to this production volume, then the mean total filtrate production is 283,200 gpd.
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-vii
-------
Table VS-iv. UF System Operations Productivity Data for 2007 Test
95%
Standard Confidence
Parameter Count Mean Median Minimum Maximum Deviation Interval
UF Operation per day (h)
Intake Flow (gpm)
Feed Flow (gpm)
Filtrate Flow (gpm)
Retentate Flow (gpm)
Backwash Flow (gpm)
Feed Pressure (psig)
Retentate Pressure (psig)
Filtrate Temperature (°F)
25
44
45
45
44
45
45
45
13.8
288
232
206
26
Not
24
22
74
14.3
296
237
212
26
measured
25
23
75
4.0
235
174
148
25
- approximately
13
11
62
21.5
303
271
245
28
900 gal
32
31
84
4.
16
19
19
0.
6
.2
.7
.6
7
±1
±4.
.8
.8
±5.7
5.7
±0.
.2
per backwash
5.
5.
5.
9
8
3
±1
±1
±1.
.7
.7
.6
From August 2 through 7, the feed water pressure needed to be increased every day to maintain the target
filtrate flow rate. During this time, TMP increased from 7 to 17 psig. On August 7, the UF system was
shutdown for a chemical cleaning, and put back into service on August 9. The TMP did not drop as a
result of the cleaning, but instead further increased up to 22 psig on August 12. Therefore, the feed
pressure was increased to 30 psig in order to maintain water flow rates. The UF system was again
shutdown and a second chemical cleaning performed on August 13. This cleaning dropped the TMP down
to 16 psig. The feed water pressure was increased again to over 30 psig on August 14 and TMP increased
accordingly. A decision was made to operate the UF system at the higher feed water pressure and TMP,
since these pressures were still within the design specification and operating specification for the unit.
The UF feed pressure remained steady for several days and was actually lower during the last week of the
test. TMP remained fairly steady at around 20 psig for the duration of the test.
As the TMP increased, the specific flux declined. The CIP was successful in stabilizing the drop in
specific flux, but did not result in returning the membrane to the specific flux attained at the beginning of
the test. The specific flux at the start of the test on July 30 was 4.62 gfd/psig. The specific flux dropped to
1.78 gfd/psig on August 7, then remained between 1.12 and 2.18 gfd/psig for the remainder of the test.
Ferric chloride was also used as a coagulant during the retest. During the initial test runs for the retest, jar
tests showed a ferric chloride dose of 1 mg/L as Fe should be the target feed rate. This feed rate was
maintained until the rapid increase in TMP and drop in specific flux occurred. After the chemical cleaning
on August 7 and 8, the ferric chloride feed rate was increased to 2 mg/L as Fe. Subsequent jar tests
suggested that with the low source water turbidity, the ferric chloride feed should actually be decreased.
The ferric chloride feed was shut off on August 10 and remained off until the CIP was required on August
13. The rapid loss of flux and rise in TMP indicated that the coagulant should be used in the system, but
at a lower dose than used at the start of the test. The ferric chloride feed was set at 0.2 mL/min (0.02 mg/L
as Fe) and continued at that rate for the remainder of the test.
QUALITY ASSURANCE/QUALITY CONTROL (QA/QC)
NSF provided technical and quality assurance oversight of the verification testing as described in the
verification report, including a review of 100% of the data. NSF QA personnel also conducted a technical
systems audit during testing to ensure the testing was in compliance with the test plan. One important
finding was that the particle count data from the 2006 test was incorrect due to improper calibration of the
particle counters. The particle counters were calibrated properly for the 2007 retest, so only the particle
count data from the 2007 test is reported.
A complete description of the QA/QC procedures is provided in the verification report.
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-viii
-------
Original signed by Sally Gutierrez 11/24/09 Original signed by Robert Ferguson 12/14/09
Sally Gutierrez Date Robert Ferguson Date
Director Vice President
National Risk Management Research Laboratory Water Systems
Office of Research and Development NSF International
United States Environmental Protection Agency
NOTICE: Verifications are based on evaluation of technology performance under specific,
predetermined criteria and the appropriate quality assurance procedures. EPA and NSF make no
expressed or implied warranties as to the performance of the technology and do not certify that a
technology will always operate as verified. The end-user is solely responsible for complying with
any and all applicable federal, state, and local requirements. Mention of corporate names, trade
names, or commercial products does not constitute endorsement or recommendation for use of
specific products. This report is not an NSF Certification of the specific product mentioned
herein.
Availability of Supporting Documents
Copies of the test protocol, the verification statement, and the verification report (NSF
report # NSF 09/28/EPADWCTR) are available from the following sources:
1. ETV Drinking Water Systems Center Manager (order hard copy)
NSF International
P.O. Box 130140
Ann Arbor, Michigan 48113-0140
2. Electronic PDF copy
NSF web site: http://www.nsf.org/info/etv
EPA web site: http://www.epa.gov/etv
NSF 09/28/EPADWCTR The accompanying notice is an integral part of this verification statement. October 2009
VS-ix
-------
This page is intentionally blank
-------
October 2009
Environmental Technology Verification Report
Removal of Inorganic, Microbial, and Particulate Contaminants from a Fresh
Surface Water
Village Marine Tec.
Expeditionary Unit Water Purifier, Generation 1
Prepared by:
Michael Blumenstein and C. Bruce Bartley NSF International, Ann Arbor, MI 48105
Dale Scherger, Scherger Associates, Ann Arbor, MI 48105
Michelle Chapman, United Stated Bureau of Reclamation, Denver, CO 80225
Jeffrey Q. Adams, Project Officer, U.S. Environmental Protection Agency, Cincinnati, OH
45268
Under a cooperative agreement with the U.S. Environmental Protection Agency
-------
Notice
The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded and managed, or partially funded and collaborated in, the research described herein. It
has been subjected to the Agency's peer and administrative review and has been approved for
publication. Any opinions expressed in this report are those of the author (s) and do not
necessarily reflect the views of the Agency, therefore, no official endorsement should be inferred.
Any mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
11
-------
Foreword
The EPA is charged by Congress with protecting the nation's air, water, and land resources.
Under a mandate of national environmental laws, the Agency strives to formulate and implement
actions leading to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, the EPA's Office of Research and
Development provides data and science support that can be used to solve environmental
problems and to build the scientific knowledge base needed to manage our ecological resources
wisely, to understand how pollutants affect our health, and to prevent or reduce environmental
risks.
The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of six environmental technology centers.
Information about each of these centers can be found on the internet at http://www.epa.gov/etv.
Under a cooperative agreement, NSF International has received EPA funding to plan, coordinate,
and conduct technology verification studies for the ETV "Drinking Water Systems Center" and
report the results to the community at large. The DWS Center has targeted drinking water
concerns such as arsenic reduction, microbiological contaminants, particulate removal,
disinfection by-products, radionuclides, and numerous chemical contaminants. Information
concerning specific environmental technology areas can be found on the internet at
http://www.epa.gov/nrmrl/std/etv/verifications.html.
in
-------
Table of Contents
Verification Statement VS-i
Notice ii
Foreword iii
List of Figures vi
List of Tables vii
Appendices viii
Abbreviations and Acronyms ix
Acknowledgements xi
Chapter 1 Introduction 1
1.1 ETV Purpose and Program Operation 1
1.2 Testing Participants and Responsibilities 1
1.2.2 NSF International 2
1.2.3 ONR 3
1.2.4 TARDEC 3
1.2.5 USER 3
1.2.6 Village Marine Tec 4
1.3 Verification Testing Site 4
1.3.1 Source Water Description and Feed Water Quality 5
Chapter 2 Equipment Capabilities and Description 6
2.1 Equipment Capabilities 6
2.2 General System Description 7
2.3 Concept of Treatment Processes 9
2.3.1 UF Pretreatment/Suspended Solids Filtration 9
2.3.2 RO Desalination 10
2.4 Detailed System Description 10
2.4.1 Raw Water Intake 13
2.4.2 UF System Description 13
2.4.3 RO System 19
2.5 General Requirements and Limitations 27
2.6 Waste Generation and Permits 29
2.6.1 UFCIP 29
2.6.2 ROCIP 29
2.6.3 RO Concentrate 30
2.6.4 UF Backwash andRetentate 30
2.6.5 Discharge Permits 30
2.7 Discussion of the Operator Requirements 30
Chapter 3 Methods and Procedures 32
3.1 Introduction 32
3.2 Quantitative and Qualitative Evaluation Criteria 32
3.3 Key Treated Water Quality and Operational Parameters 33
3.4 Operations and Maintenance 34
3.5 Field Operations 34
3.6 Overview of ETV Testing Plan 34
3.6.1 Task A: Characterization of Feed Water 35
IV
-------
3.6.2 Task B: Equipment Installation, Initial Test Runs, and Initial System Integrity
Tests 35
3.6.3 Task C: Verification Test 35
3.7 Task A: Characterization of Feed Water 36
3.8 Task B: Equipment Installation, Initial Test Runs, and Initial System Integrity Tests 36
3.9 Task C: Verification Testing 36
3.9.1 Task Cl: Membrane Flux and Operation 37
3.9.2 TaskC2: Cleaning Efficiency 42
3.9.3 TaskC3: Finished Water Quality 43
3.9.4 Task C4: Membrane Integrity Testing 44
3.9.5 Task C5: Data Handling Protocol 45
3.9.6 Task C6: Quality Assurance Project Plan 45
Chapter 4 Results and Discussion 52
4.1 2006 EUWP Test 53
4.1.1 Task A: Raw Water Characterization 53
4.1.2 TaskB: Equipment Installation and Initial Test Runs 53
4.1.3 TaskC: Verification Test 55
4.1.4 2006 Chemical Consumption 91
4.2 2007 EUWP Retest 92
4.2.1 Task A: Raw Water Characterization 92
4.2.2 TaskB: Equipment Install and Initial Test Runs 92
4.2.3 TaskC: 2007 Verification Retest 93
4.2.4 2007 Chemical Consumption 116
4.3 Quality Assurance/Quality Control 117
4.3.1 Introduction 117
4.3.2 Documentation 117
4.3.3 Quality Audits 118
4.3.4 Test Procedure QA/QC 118
4.3.5 Sample Handling 118
4.3.6 Physical and Chemical Analytical Methods QA/QC 119
4.3.7 Microbiology Laboratory QA/QC 119
4.3.8 Documentation 119
4.3.9 Data Review 119
4.3.10 Data Quality Indicators 120
-------
List of Figures
Figure 1-1. Photo of concrete pad used for EUWP testing 4
Figure 2-1. Process component diagram 8
Figure 2-2. Koch UF hollow fiber modules, a single fiber, and the process flow through the
module 9
Figure 2-3. EUWP system process schematic 11
Figure 2-4. Schematic of typical EUWP layout 12
Figure 2-5. Photo of raw water intake and intake strainer 13
Figure 2-6. Photo of the UF skid 15
Figure 2-7. Photo of the UF cartridges mounted in theUF skid 15
Figure 2-8. Piping and instrumentation diagram of UF skid 18
Figure 2-9. Photo of the RO skid 19
Figure 2-10. Photo of the RO skid membrane vessels 20
Figure 2-11. Vessel arrangement schematic 20
Figure 2-12. Membrane arrangement schematic 21
Figure 2-13. P&ID of RO skid 25
Figure 2-14. PX pressure exchanger 26
Figure 4-1. UF system filtrate production for 2006 test 57
Figure 4-2. Plot of UF system flow rates for 2006 test 59
Figure 4-3. Plot of UF system feed and retentate pressures for 2006 test 60
Figure 4-4. Plot of UF system IMP for 2006 test 60
Figure 4-5. UF system specific flux calculations for 2006 test 61
Figure 4-6. Loss of specific flux overtime for 2006 test 61
Figure 4-7. UF Power consumption per hour of operation for 2006 test 62
Figure 4-8. RO system flow rates for 2006 test 65
Figure 4-9. RO system operating pressures for 2006 test 65
Figure 4-10. RO percent recoveries for 2006 test 66
Figure 4-11. RO system specific flux for 2006 test 67
Figure 4-12. RO power use per hour of operation for 2006 test 68
Figure 4-13. UF feed water turbidity for 2006 test 72
Figure 4-14. UF filtrate water turbidity for 2006 test 73
Figure 4-15. RO conductivity results for 2006 test 75
Figure 4-16. Pressure decay overtime for the 2006 test 86
Figure 4-17. UF filtrate production for the 2007 retest 95
Figure 4-18. UF system flow rates for 2007 retest 96
Figure 4-19. UF system feed and retentate pressures for 2007 retest 96
Figure 4-20. UF system IMP for 2007 retest 97
Figure 4-21. UF system specific flux for 2007 retest 98
Figure 4-22. Loss of specific flux overtime for 2007 retest 99
Figure 4-23. UF Power consumption per hour of operation for 2007 retest 100
Figure 4-24. UF feed turbidity for the 2007 retest 103
Figure 4-25. UF filtrate turbidity for the 2007 retest 104
Figure 4-26. UF feed and filtrate in-line turbidity readings for the 2007 retest 105
Figure 4-27. Pressure Decay Results for the 2007 Retest 110
Figure 4-28. Bacillus endospores results for the 2007 retest 112
VI
-------
Figure 4-29. 2-3 jim particle counts for the 2007 retest 113
Figure 4-30. 3-5 jim particle counts for the 2007 retest 114
List of Tables
Table 2-1. Koch Membrane Systems Targa 10-48-35-PMC Cartridge Specifications 14
Table 2-2. UF Skid Statistics 14
Table 2-3. RO System Membrane Element Characteristics 21
Table 2-4. RO Skid Statistics 22
Table 2-5. EUWP Site Considerations and Dimensions 27
Table 2-6. Equipment Limitations 28
Table 2-7. Membrane Limitations 29
Table 3-1. Key Treated Water Quality Parameters 33
Table 3-2. Water Quality and Operational Parameters Measured Online 33
Table 3-3. Operational Parameter Sampling Locations 38
Table 3-4. Key Operating Parameters 38
Table 3-5. Operational Data Plots Appearing in Chapter 4 38
Table 3-6. Water Quality Sampling Schedule 44
Table 3-7. On-Site Analytical Equpment QA Activities 46
Table 3-8. On-Site Data Generation QC Activities 47
Table 3-9. Analytical Methods for Laboratory Analyses 49
Table 3-10. Completeness Requirements 51
Table 4-1. Initial Characterization Sampling Results 54
Table 4-2. UF Operational Measurement Statistics for 2006 Test 56
Table 4-3. RO System Operational Measurement Statistics for 2006 Test 63
Table 4-4. UF Cleaning Solution TOC Results 70
Table 4-5. Turbidity Results for 2006 Test - Hand-Held Meter 73
Table 4-6. Conductivity Results for 2006 Test for In-Line Meter 76
Table 4-7. pH results for 2006 Test - In line Meter 78
Table 4-8. Temperature Data for 2006 Test - In line Meter 79
Table 4-8. Temperature Data for 2006 Test-In line Meter (continued) 80
Table 4-9 Other UF System Water Quality Data for 2006 Test 82
Table 4-10. Other RO System Water Quality Data for 2006 Test 83
Table 4-11. Pressure Decay Data for the 2006 Test 85
Table 4-12. HPC Results for the 2006 Test 88
Table 4-13. Total Coliform Results for the 2006 Test 88
Table 4-14. RO Dye Test Results -September 23, 2006 89
Table 4-15. RO Dye Test Results-October 24, 2006 90
Table 4-16. UF Membrane Integrity Indicators for October 2006 91
Table 4-17. UF System Operational Measurement Statistics for 2007 Retest 94
Table 4-18. UF Cleaning Solution TOC Results for 2007 Retest 101
Table 4-19. Turbidity Data for the 2007 Retest - Hand-Held Meter 102
Table 4-20. In-Line Turbidity Measurement Statistics for the 2007 Retest 105
Table 4-21. Conductivity, pH, and Temperature Data for the 2007 Retest 106
vn
-------
Table 4-22. Other Water Quality Data for the 2007 Retest 107
Table 4-23. Pressure Decay Results for UF System for the 2007 Retest 109
Table 4-24. Bacillus Endospores - 2007 Retest Ill
Table 4-25. UF Membrane Integrity Indicators for August 19, 2007 to August 23, 2007 116
Appendices
Appendix A - Operation and Maintenance Manual
Appendix B - Field Logbooks, Field Log Sheets, Field Calibration Records
Appendix C - NSF Laboratory Data Reports and Sample Chain of Custody Forms
Appendix D - Spreadsheets
Vlll
-------
Abbreviations and Acronyms
ANGB Air National Guard Base
BOD biochemical oxygen demand
°C degrees Celcius
CPU colony-forming unit
CI Confidence Interval
CIP clean-in-place
cm centimeter
DF2 diesel fuel, grade 2
DFA diesel fuel, arctic grade
DQO data quality objectives
DWSC Drinking Water Systems Center
EPA United States Environmental Protection Agency
ETV Environmental Technology Verification
EUWP Expeditionary Unit Water Purifier
°F degrees Fahrenheit
FRP fiberglass reinforced plastic
ft foot (feet)
FTO field testing organization
gal gallons
gfd gallons per foot per day
gpd gallons per day
gpm gallons per minute
h hour
FIPC Heterotrophic plate count
in inch
JP8 j et propellent 8 (j et fuel)
kgal kilogallon
kW kilowatt
kWh kilowatt hour
L liter
Ib pound
LT2ESWTR Long Term 2 Enhanced Surface Water Treatment Rule
m meter
mg milligram
mL milliliter
mS millisiemens
NBC nuclear, biological, and chemical
ND not detectible
NDP net driving pressure
NIST National Institute of Standards and Technology
NM not measured
NPDWR National Primary Drinking Water Regulations
NRMRL National Risk Management Research Laboratory
NSF NSF International (previously known as the National Sanitation Foundation)
IX
-------
NSWCCD United States Naval Surface Warfare Center - Carderock Division
NTU Nephelometric turbidity units
O&M operations and maintenance
ONR Office of Naval Research
ORD Office of Research and Development
P&ID piping and instrumentation diagram
PE performance evaluation
PLC programmable logic controller
ppm parts per million
psi pounds per square inch
psig pounds per square inch, gauge
PVC polyvinyl chloride
PX pressure exchanger
QA/QC quality assurance/quality control
QAPP quality assurance project plan
RO reverse osmosis
RPD relative percent difference
SDI silt density index
SM Standard Methods for the Examination of Water and Wastewater
SNL Sandia National Laboratories
TARDEC United States Army Tank-Automotive Research, Development, and Engineering
Center
TDS total dissolved solids
TOC total organic carbon
TQAP test/quality assurance plan
TQG tactical quiet generator
TSS total suspended solids
TMP trans-membrane pressure
UF ultrafiltration
USER United States Bureau of Reclamation
UV254 ultra violet absorbance at 254 nanometers
VOC volatile organic compounds
|im micron
|iS microSiemens
-------
Acknowledgements
The U.S. Army Tank-Automotive Research, Development, and Engineering Center (TARDEC)
was the main field testing organization. TARDEC was supported by the U.S. Bureau of
Reclamation (USER). TARDEC and USER were responsible for all elements of the tests,
including operation of the equipment, collection of samples, instrument calibration, and data
collection.
This verification report was authored by Mr. Michael Blumenstein and Mr. C. Bruce Bartley of
the NSF International ETV Drinking Water Systems Center (DWSC), Mr. Dale Scherger of
Scherger Associates (3017 Rumsey Drive, Ann Arbor, MI 48105), and Ms. Michelle Chapman
of USER. The verification report was based on the project test/quality assurance plan authored
by DWSC, USER, and TARDEC.
The laboratory selected for the analytical work was:
NSF International Chemistry Laboratory
789 N. Dixboro Road
Ann Arbor, Michigan 48105
Contact: Mr. Kurt Kneen
The manufacturer of the EUWP was:
Village Marine Tec.
2000 W. 13 5th St.
Gardena, CA 90249
Phone: 310-516-9911
The TARDEC engineers responsible for the field tests were Mr. Bob Shalewitz, Ms. Lori
Bolster, and Mr. Jeremy Walker. The USER support staff included Ms. Michelle Chapman, Mr.
Daniel Gonzales, and Mr. Steve Dundorf.
The NSF DWSC project manager was Mr. Michael Blumenstein. The DWSC is managed by
Mr. C. Bruce Bartley. Ms. Kristie Wilhelm of the DWSC provided valuable assistance with
report preparation.
The authors would like to thank Mr. James Weise of the State of Alaska Department of
Environmental Conservation, and Mr. Craig Patterson of EPA for their reviews of the
verification report.
XI
-------
This page is intentionally blank
-------
Chapter 1
Introduction
1.1 ETV Purpose and Program Operation
The U.S. Environmental Protection Agency (EPA) has created the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative or improved
environmental technologies through performance verification and dissemination of information.
The goal of the ETV Program is to further environmental protection by accelerating the
acceptance and use of improved and more cost-effective technologies. ETV seeks to achieve this
goal by providing high-quality, peer-reviewed data on technology performance to those involved
in the design, distribution, permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized standards and testing organizations; with stakeholder
groups consisting of buyers, vendor organizations, and permitters; and with the full participation
of individual technology developers. The program evaluates the performance of innovative
technologies by developing test plans responsive to the needs of stakeholders, conducting field
demonstrations, collecting and analyzing data, and preparing peer-reviewed reports. All
evaluations are conducted in accordance with rigorous quality assurance protocols to ensure that
data of known and adequate quality are generated and that the results are defensible.
The EPA has partnered with NSF International (NSF) under the ETV Drinking Water Systems
Center (DWSC) to verify the performance of water supply technologies that serve both small and
large communities. A goal of verification testing is to enhance and facilitate the acceptance of
scalable drinking water treatment equipment by state drinking water regulatory officials and
consulting engineers, while reducing the need for testing of equipment at each location where the
equipment's use is contemplated. NSF meets this goal by working with manufacturers and NSF-
qualified Field Testing Organizations (FTO) to conduct verification testing under the approved
protocols. It is important to note that verification of the equipment does not mean the equipment
is "certified" by NSF or "accepted" by EPA. Rather, it recognizes that the performance of the
equipment has been determined and verified by these organizations for those conditions tested by
the FTO.
The DWSC recently evaluated the performance of the Village Marine Tec. Generation 1
Expeditionary Unit Water Purifier (EUWP). The EUWP, developed for the U.S. Military, uses
ultrafiltration (UF) and reverse osmosis (RO) to produce drinking water from a variety of
different sources. This document provides the verification test results for the EUWP system at a
fresh surface water site in Selfridge Air National Guard Base (ANGB) in Michigan.
1.2 Testing Participants and Responsibilities
EUWP design, construction, and testing was overseen by a federal multi-agency team composed
of representatives from Office of Naval Research (ONR); Army Tank-Automotive Research,
Development, and Engineering Center (TARDEC); Naval Surface Warfare Command -
Carderock Division (NSWCCD); United States Department of Interior Bureau of Reclamation
(USER); and Sandia National Laboratories (SNL). The manufacturer, Village Marine Tec., was
1
-------
contracted to design and build the EUWP to the team's Generation 1 specifications using 2004
state-of-the-art technology.
The organizations involved in the verification testing project were:
• EPA
• NSF
• ONR
• TARDEC
• USER
• Village Marine Tec.
The following is a brief description of all of the ETV participants and their roles and
responsibilities.
1.2.1 EPA
EPA, through its Office of Research and Development (ORD), has financially supported and
collaborated with NSF under Cooperative Agreements R-82833301 and CR833980. This
verification effort was supported by the DWSC operating under the ETV Program. This
document has been peer-reviewed, reviewed by EPA, and recommended for public release.
1.2.2 NSF International
NSF is an independent, not-for-profit testing and certification organization dedicated to public
health and safety and to the protection of the environment. Founded in 1946 and located in Ann
Arbor, Michigan, NSF has been instrumental in the development of consensus standards for the
protection of public health and the environment. NSF also provides testing and certification
services to ensure products bearing the NSF Name, Logo and/or Mark meet those standards. The
EPA partnered with NSF to verify the performance of drinking water treatment systems through
the EPA's ETV Program.
NSF authored the test plan and test report. NSF also served as the analytical laboratory for all
water quality parameters not measured in the field. NSF also provided technical oversight during
testing and conducted an audit of the field testing activities.
Contact Information:
NSF International
789 N. Dixboro Road
Ann Arbor, Michigan 48105
Contact: Mr. Bruce Bartley, Project Manager
Phone: (734) 769-8010
Fax: (734) 769-0109
Email: bartley@nsf.org
-------
1.2.3 ONR
The U.S. Navy ONR provided oversight of the EUWP development program which involved
developing high productivity water treatment units for land and shipboard military and civilian
emergency preparedness applications. ONR also provided funding for the EUWP ETV testing
project.
Contact Information:
Office of Naval Research
Logistics Thrust Program
Operations Technology Division
800 N. Quincy St.
Arlington, VA 22217
Contact: Major Alan Stocks
Phone: 703-696-2561
Email: stocksa@onr.navy.mil
1.2.4 TARDEC
TARDEC served as the FTO for this verification. TARDEC was responsible for all on-site
testing activities, including operation of the test equipment, collection of samples, measurement
of water quality parameters, calibration and check of instrumentation, and operational data
collection.
Contact Information:
U.S. Army TARDEC/RDECOM
AMSRD-TAR-D/210, MS 110
6501 E. Eleven Mile Road
Warren, MI 48397
Contact: Mr. Bob Shalewitz, TARDEC EUWP Program Manager
Phone: 586-574-4128
Email: bob.shalewitz@us.army.mil
1.2.5 USER
USER functioned as a co-FTO, providing field operations support, and technical support for
equipment operation.
Contact Information:
U.S. Bureau of Reclamation
Denver Federal Center (D-8230)
P.O. Box 25007
Denver, CO 80225
Contact: Ms. Michelle Chapman
Phone: 303-445-2264
Email: mchapman@do.usbr.gov
-------
1.2.6 Village Marine Tec.
The EUWP manufacturer was Village Marine Tec. The manufacturer was responsible for
supplying a field-ready treatment system equipped with all necessary components, including
instrumentation and controls, and an operation and maintenance (O&M) manual. The
manufacturer was responsible for providing logistical and technical support, as needed, as well
as technical assistance to the FTO during operation and monitoring of the equipment undergoing
field verification testing.
Contact Information:
Village Marine Tec.
2000 W. 13 5th St.
Gardena, CA 90249
Phone:310-516-9911
Email: sales@villagemarine.com
1.3 Verification Testing Site
The EUWP verification testing occurred at Selfridge ANGB at 127 Wing Public Affairs Office,
29423 George Avenue, Selfridge, MI 48045-5290. Selfridge ANGB is located in southeastern
Michigan, 30 miles northeast of Detroit on the shore of Lake St. Clair, with an elevation of 580
feet (ft).
The EUWP was situated on a cement pad a few yards from Lake St. Clair (Figure 1-1). The raw
water for testing was drawn from the inlet at the left in the photo.
Figure 1-1. Photo of concrete pad used for EUWP testing.
-------
1.3.1 Source Water Description and Feed Water Quality
Raw water from Lake St. Clair was used for ETV testing. Approximately 430 square miles (mi2)
(1,114 square kilometers (km2)) in area, the lake is part of the Great Lakes system. The lake,
along with the St. Clair River and Detroit River, provides the connection between Lake Huron to
the north and Lake Erie to the south. It is a shallow lake with an average depth of about 10 ft (3
m) and a maximum natural depth of 21 ft (6.4 m) (Wikipedia, 2006).
-------
Chapter 2
Equipment Capabilities and Description
The EUWP was designed to meet purified water needs in areas with challenging water sources of
very high total dissolved solids (TDS), turbidity, or hazardous contamination during emergency
situations when other water treatment facilities are incapacitated. The system uses UF and RO to
produce potable water. It is not intended to meet general municipal water treatment needs in a
cost effective manner. The design requirements - to produce 100,000 gallons per day (gpd) and
be C-130 transportable - forced the use of lightweight durable materials, such as titanium, that
are more costly and would not usually be required for municipal water treatment. The
requirements to treat source water with up to 60,000 milligrams per liter (mg/L) TDS and ensure
removal of nuclear, biological, and chemical (NBC) contaminants to a safe limit, drove the
design to two parallel arrays - with a 2nd permeate pass resulting in a maximum of 65% recovery.
Most municipal water treatment systems can easily attain much higher recovery levels. The
EUWP is also intended as a demonstration of the state-of-the-art of desalination for emergency
situations.
Key innovations applied in the EUWP are:
• High flux UF membrane cartridges;
• Innovative staging of RO membrane modules; and
• Small system energy recovery to pressurize a parallel array.
The EUWP was developed to meet the following objectives:
• Develop a high capacity drinking water purification unit to provide strategic water
production capability with a focus on peacekeeping, humanitarian aid, and disaster relief
missions that the military frequently supports.
• Further the state of desalination technology with a view toward reduced operational costs,
size, and weight; improved reliability; and verifying emerging technologies.
2.1 Equipment Capabilities
The objective of this verification test was to document the ability of the EUWP to meet the
following performance criteria:
The EUWP is capable of producing 100,000 gpd of water meeting EPA's National
Primary Drinking Water Regulations (NPDWR) from raw Lake St. Clair water based on
contaminants found in the source water during the initial water characterization phase of
EW testing.
The EUWP is intended to meet purified water needs in areas with challenging water sources of
very high TDS, turbidity, or hazardous contamination during emergency situations when other
water treatment facilities are incapacitated. The unit was designed to meet or exceed Tri-Service
Field Water Quality Standards for short-term consumption by healthy adults. However, the
technology used is capable of exceeding the EPA NPDWR.
-------
The EUWP, using the UF system only, can produce 250,000 gpd of potable water from a fresh
water source with up to 500 mg/L TDS and a temperature of 77 Fahrenheit (°F) (25 Celsius, °C),
provided that contaminants not removed by UF are not present in the source water. Using the UF
and RO system, it is designed to produce from 98,000 gpd up to 162,000 gpd depending on the
TDS of the source water and the recovery settings of the RO system. Production is decreased to
125,000 gpd (50% recovery) for higher TDS waters. It can also produce 98,000 gpd from a NBC
contaminated source with up to 45,000 mg/L TDS. NBC contaminant removal and seawater
desalination were not verified as part of this ETV testing at Selfridge ANGB.
2.2 General System Description
• Equipment name: Expeditionary Unit Water Purifier (EUWP)
• Model number: Generation 1
• Manufacturer: Village Marine Tec., 2000 W. 135th St., Gardena, CA 90249, (310) 324-
4156.
• Power requirements: 480 volts, 250 Amp, 60 hertz, 3-phase electrical, or two 60 kilowatt
(kW) diesel Tactical Quiet Generators (TQG).
- UF Requirements - 125 amps maximum
- RO Requirements - 125 amps maximum
The EUWP is composed of feed pumps, a UF pretreatment system, a 1 or 2 pass RO desalination
system with energy recovery, storage tanks, and product pumps (Figure 2-1). It has chemical
feed systems for pretreatment and post treatment. Clean-in-place systems (CIP) are included with
the skids.
-------
UF Product Tank
RO Skid
—
200 urn
Figure 2-1. Process component diagram.
UF Skid Waste
L
-©—>
200 urn
Waste UF Hollow Fiber
i Membranes
2nd Pass
—>
Waste
Feed Tank
RO Product Tank
-©—+•
-------
2.3 Concept of Treatment Processes
2.3.1 UF Pretreatment/Suspended Solids Filtration
UF is a low-pressure (5-90 pounds per square inch, gauge (psig)) membrane process that
separates particulates based on size exclusion. The UF process retains oils, particulate matter,
bacteria, and suspended solids that contribute to turbidity and a high silt density index (SDI).
Feed water to RO systems should have turbidity less than 0.1 Nephelometric Turbidity Units
(NTU) and a SDI less than 3. UF membranes pass water, dissolved salts, and most dissolved
organic compounds. UF pore sizes range from 0.002 to 0.1 micron (um), and the molecular
weight cutoff ranges from 1,000 to 500,000. Koch Membrane Systems Targa-10 hollow fiber
UF membranes are used in the EUWP. Water flows from the inside of the fiber to the outside
causing suspended solids to collect on the inside of the fiber. Periodically, the system must
be vigorously backwashed to remove this material from the system. Figure 2-2 shows
example UF cartridges, a single fiber, and the flow pattern used in this system.
The key operating parameters for a UF system are the instantaneous flux and the overall
productivity taking into account the volume required for backwash. Generally, the higher the
instantaneous flux, the more often backwashing will be required. There is an optimum flux
where overall productivity is maximized, called the critical flux. For municipal systems, it is
economical to operate the system at the critical flux. The EUWP is an emergency supply
system with extreme weight restrictions to enable transport. The weight restrictions drove
design of the UF system to operate at a maximum flux with more frequent backwashes.
Process
Feed Flow
Hollow
Fiber Membranes
Figure 2-2. Koch UF hollow fiber modules, a single fiber, and the process flow through
the module.
-------
2.3.2 RO Desalination
Dissolved salts and larger molecular weight organic molecules can be removed by RO.
Osmosis is a naturally occurring phenomenon in which pure water is transported down a
chemical potential gradient across a semi-permeable membrane from a low concentration
solution to a high concentration solution. One measure of the chemical potential is the
osmotic pressure. Osmotic pressure is dependant on the concentration of ions and dissolved
compounds. It can be measured by pressurizing the concentrated solution until osmotic
induced flow stops. If this pressure is exceeded, then osmotic flow reverses from
concentrated solution to the dilute solution.
RO is a moderate to high-pressure (80 - 1,200 psig) membrane separation process. The
membranes in the EUWP are spiral wound with up to seven modules in a vessel. They are
operated under cross-flow conditions at a pressure above the osmotic pressure of the bulk
solution, plus additional pressure to overcome resistance of the modules. Water passing
through the RO membrane is called permeate, and the concentrated discharge stream is called
concentrate.
The separation model is of solution and diffusion of material through the polymer of the
membrane. Dissolved salts are transported very slowly compared to water and other un-
charged molecules. Uncharged molecules may be rejected based on size exclusion,
depending on their mass and geometry.
2.4 Detailed System Description
This section provides a detailed system description. See the system operation manual in
Appendix A for further details about the system and operation. Note that the system was
designed and manufactured prior to promulgation of the final EPA Long Term 2 Enchanced
Surface Water Treatment Rule (LT2ESWTR). The EUWP, as tested, was not designed to
comply with the LT2ESWTR indirect integrity monitoring requirement that calls for the
system to shut down pending a direct integrity test, if two consecutive turbidity readings
exceed 0.15 NTU. The EUWP does have in-line turbidity meters to monitor the feed and
filtrate streams for the UF skid, but the programmable logic controller (PLC) was not
programmed to automatically shut down the system, if necessary. The RO system has an in-
line turbidity meter for the RO permeate process stream. The RO system also includes in-line
conductivity meters to monitor performance. The system process schematic and detailed
layout are shown in Figures 2-3 and 2-4, respectively.
10
-------
Concentrate/
Waste
Figure 2-3. EUWP system process schematic.
11
-------
Figure 2-4. Schematic of typical EUWP layout.
12
-------
2.4.1 Raw Water Intake
Raw Lake St. Clair water was drawn from the inlet shown at the left of Figure 1-2. An intake
strainer was used to keep large pieces of debris from being drawn up. A close-up photo of the
intake location and intake strainer is shown as Figure 2-5. Before the raw water reached the UF
feed tank, ferric chloride was injected as a coagulant, and the water was strained again through
two Amiad Filtrate Systems model TAF-750 filters operating in parallel. The filters are equipped
with 200 |im weave wire screens. The strainers did not remove any ferric chloride floe, since
there was not enough time for particles larger than 200 |im to form between the injection point
and the strainer. The 3,000 gallon (gal) UF feed tank provides at least 12 minutes (min) of
retention time for floe formation.
Figure 2-5. Photo of raw water intake and intake strainer.
®
2.4.2 UF System Description
The UF membranes used in the EUWP are model TARGA 1 0-48-3 5-PMC, manufactured by
Koch Membrane Systems. The UF cartridge specifications are presented in Table 2-1. The UF
membranes are configured in two trains of eight cartridges each, all of which are operated in
parallel. The membranes are operated such that 10% of the feed flow exits the cartridges as
13
-------
retentate. Statistics of the UF skid are presented in Table 2-2. Photos of the UF skid are shown in
Figure 2-6 and Figure 2-7.
Table 2-1. Koch Membrane Systems Targa 10-48-35-PMC Cartridge Specifications
Parameter
Value
Nominal molecular weight cut-off
Maximum recommended flow (per cartridge)
Maximum pressure
Maximum transmembrane pressure (TMP)
Maximum backflush TMP
Inner fiber diameter
Membrane area
Cartridge diameter
Cartridge length
100,000
32.2 gpm(1)
45 psig
30 psig
20 psig
0.035 in(2)
554 ft2(3)
10.75 in
48 in
(1) gallons per minute
(2) inch(es)
(2) square feet
Table 2-2. UF Skid Statistics
Parameter
Value
Production capacity
Maximum applied pressure
Maximum TMP
Water temperature range
Turbidity range
Dimensions
Weight
Basic metals
Operating ambient temperature range
Storage and transport air temperature
range
Relative humidity:
Maximum slope of unit when deployed
for operation
Power source requirement
Fuel type
Fuel capacity (60 KW Generator)
250,000 gpd
45 psig
30 psig
34-104 °F
0-150 NTU
20'Lx8'Hx8'W
15,500 Ibs dry, fully paced out for deployment, less fuel
UF System Piping: Fiberglass, Titanium, Nylon
Air System Piping: Nylon Tubing
32°F-120 °F
32°F-120 °F
3%-95%
5 degrees side to side, 2 degrees end to end
60 kW Generator (self contained) or power grid connection
consisting of 480 volts, 125 amps. UF system and external
pumping power requirements are 2.1 kWh/kgal(1)
DF2 (Diesel Fuel, Grade 2)
DFA (Diesel Fuel, Arctic Grade)
JP8 (jet propellent 8)
43 gal
(l)Kilowatt-hours per kilogallon
14
-------
Figure 2-6. Photo of the UF skid.
Figure 2-7. Photo of the UF cartridges mounted in the UF skid.
15
-------
2.4.2.1 UF System Operation
The following is a basic description of the flow path and functional description of the UF system
in normal operation for an open surface water source. The operation manual provides a full
description of UF operation. Figure 2-8 is a piping and instrumentation diagram (P&ID) of the
UF system.
1. Pump #1 (PI) brings water through the intake strainer #1 (ST1) (if an open intake is used)
to the UF skid. Before entering the UF feed tank, water is strained (ST2) again to 200
microns on the UF skid. The strainers serve to eliminate debris that would clog the
membrane fibers. Water exits strainer #2 and is stored in the UF feed tank (TK2) which
serves as a break tank between the feed water supply and the UF feed.
2. If necessary, ferric chloride coagulant from Chemical Pump #1 (CP1) can be added to the
feed stream before entering ST2 to enhance UF performance. The decision to use ferric
chloride is site-specific, based on the raw water quality, if known, and/or the results of a
jar test.
3. Pump #3 (P3) moves water from TK2 to the UF membranes.
4. The UF filtrate flows to tank #3 (TK3). TK3 acts as a break tank between the UF skid
and the RO skid and a back flush reservoir for the UF skid.
5. Pump #5 (P5) pumps water from TK3 to the RO skid or directly through the disinfection
system (CL1 - calcium hypochlorite) to the distribution system when RO is not required.
The disinfection system will not be used for this verification.
2.4.2.2 UF CIP Procedure
The UF system must be cleaned when the TMP drop exceeds 35 psig after a normal backflush
cycle. This cleaning cycle is required approximately every 30 days, depending on the water
source. The CIP procedure typically uses citric acid as the low pH cleaning agent, and sodium
hydroxide as the high pH cleaning agent. Note that different cleaning agents may need to be used
for certain foulants.
If system operation requires the use of ferric chloride as a coagulant, then a low pH clean must
be performed first, followed by a high pH clean. If ferric chloride is not being used, then a high
pH clean must be performed first, then a low pH clean. Ferric chloride was used during testing at
Selfridge ANGB. The following is a basic description of the flow path and functional description
of the UF system in normal operation. The operation manual provides a full description of UF
operation, including an operational summary described below.
1. Prior to CIP, perform a fresh back flush.
2. Following backwash, set up system for UF normal mode of operation. Activate UF drain
mode on the screen.
3. Wait for the system to drain.
4. Connect the hose from the CIP tank to the system.
5. Touch the CIP button on the screen. Select CIP Mode ON. The PLC will automatically
move the pneumatically operated valves to the correct positions.
6. Enable heaters to maintain CIP solution to between 96 and 100°F.
7. Turn tank mixer on using CIP display screen
8. Add the appropriate amount of chemical to achieve the desired pH.
16
-------
9. Check the pH of the mixture in tank 4 at sample port V22 every 15 min. Use citric acid to
lower the pH to 3 or use sodium hydroxide to raise the pH to 11.
10. With high pH only, add an appropriate amount of calcium hypochlorite.
11. Start CIP by touching the CIP button at the top left of the CIP screen then start to pump
the solution using P3.
12. Allow the chemical to circulate through the selected array for 20 to 30 min.
13. Let the system soak for several hours after recirculation if needed to remove tough
fouling.
14. Repeat recirculation with the desired chemicals.
15. Following chemical recirculation, rinse the system as necessary with clean water.
17
-------
Figure 2-8. Piping and instrumentation diagram of UF skid.
18
-------
2.4.3 RO System
The RO skid is shown below in Figures 2-9 and 2-10.
The RO system has the capability to operate in single-pass or double-pass mode if necessary (the
double-pass mode was not used for this ETV test). The first pass of the RO system consists of a
unique combination of moderate rejection/high productivity and high rejection/moderate
productivity membranes. The first pass is composed of two parallel arrays (Figure 2-11). The
first array is fed by the high-pressure pump and has two parallel trains with two four-element
vessels each (Vessels 1, 2, 3, and 4 in Figure 2-11). The energy from the brine of this array is
used to pressurize feed water via a pressure exchanger energy recovery device to feed a second
array consisting of a single train of two four-element vessels (Vessels 5 and 6 in Figure 2-11).
The second pass RO system consists of a 2—4 array, where a second high-pressure pump boosts
permeate pressure from the first pass feeding two parallel four-element vessels (Vessels 7 and 8
in Figure 2-11). The brine from these vessels then feeds one additional four-element vessel
(Vessel 9 in Figure 2-11).
The RO design incorporates an internally staged RO element configuration on the first pass
(Figure 2-12). This configuration consists of two Dow Chemical Company FILMTEC™ SW30-
HR LE-400 elements, followed by two FILMTEC SW30-XLE400 elements, which are in turn
followed by four FILMTEC SW30-HR-12000 ultra-low-energy experimental membranes. All
membranes are polyamide thin-film composite type. The second pass RO system uses AquaPro
LE-8040UP membrane elements. Table 2-3 provides performance data for the elements used in
the system.
Figure 2-9. Photo of the RO skid.
19
-------
Figure 2-10. Photo of the RO skid membrane vessels.
-<
1 — >
*l 1
^l 1
>x)}>l 1
•^
r1 1
Uj 1
R
I 2
I 2
I 2
2
2
I 3
I 3
I 3
3
3
I 4
I 4
I 4
4
1
4
L
h
h
h
•
-*! s
-*! s
-*! s
6
7
Permeate
6
6
7
7
8 h
8 h
8 h
—^ Concentrate
^
-§hN
-i^
-*l 1
2
3
4
|-
*-
*.
Numbers indicate pressure vessels
Figure 2-11. Vessel arrangement schematic.
20
-------
© 0
© O
0
© ©
X1 | X1 | X2 | X2
X1 | X1 | X2 | X2
X3 | X3 | X3 | X3
X3 | X3 | X3 | X3
X1 X1 X2 X2
X3 X3 X3 X3
, *1 X4 | X4 | X4 | X4 | — h
ijr
' | X4 | X4 | X4 | X4 | — ^
l__
— *\ X4 | X4 | X4 | X4 | »
_fc
Numbers indicate pressure vessels
Figure 2-12. Membrane arrangement schematic.
Table 2-3. RO System Membrane Element Characteristics
Vessel
1st Pass
2,3,5
1st Pass
2,3,5
1st Pass
1,4,6
2nd Pass
7,8,9
Product
FILMTEC SW30-HR LE-
400
FILMTEC SW30-
XLE-400
FILMTEC SW30-HR
-12000 (experimental)
AquaPro LE-8040UP *
Designator
XI
X2
X3
X4
Nominal Active
Surface Area
ft2 (m2)
380 (35)
400 (37)
400 (37)
400 (37)
Permeate
Flowrate gpd
(m3/d)
6000 (26)
9000 (34)
12,000 (45)
10,200 (38)
Stabilized Salt
Rejection
(%)
99.8
99.7
99.7
99.7
: Toray membrane assembled by AquaPro /Village Marine.
21
-------
2.4.3.1 RO skid statistics
Table 2-4 presents statistics of the RO skid.
Table 2-4. RO Skid Statistics
Parameter
Value
Production capacity
Water temperature range
Dimensions
Weight
Basic metals
Operating ambient temperature range
Storage and transport air temperature Range
Relative humidity
Maximum slope of unit when deployed for
operation
Power source requirement
Fuel type (if using RO pump engine)*
Fuel capacity (if using RO pump engine)*
~ 125,000 gpd for single pass on surface water above 25,000 mg/L
TDS or groundwater above 2500 mg/L TDS
-162,000 gpd for other lower TDS waters
-98,000 gpd in double pass mode
34-104 °F
20'Lx8'Hx8'W
15,500 Ibs dry, fully paced out for deployment, less fuel
High Pressure Piping: Titanium
Production Piping: 316L Stainless Steel and fiberglass reinforced
plastic (FRP)
32°F-120 °F
32°F-120 °F
3%-95%
No Restrictions
Power for all but high-pressure pump is supplied from UF skid.
HP pump requirements are 480 Volts and 125 Amps. The
operational power use is 7.4 kWh/kgal for the RO system only.
DF2, DFA, JP8
60 gal
* Electric RO pump was used for ETV testing.
2.4.3.2 RO System Operation
The following is a basic description of the flow path and functional description of the RO system
in normal operation. The RO system has the capacity to operate in either a one or two pass mode.
The second pass is only used if sufficient treatment is not achieved with the first pass (especially
for NBC contamination). The operation manual provides a full description of RO operation.
Figure 2-13 is a P&ID of the RO system.
1. The UF filtrate is supplied to the RO 1st pass through P5 from TK3.
2. The RO 1st pass includes two arrays. The RO feed water (from the UF filtrate) flows into
vessels 2 and 3 (PV2, PV3). The concentrate from vessels 2 and 3 flow into vessels 1
and 4 (PV1, PV4), respectively. The combined concentrate from vessels 1 and 4 flows
through the energy recovery device, which boosts raw water pressure and feeds vessel 5
(PV5) of the second array. The concentrate from PV5 flows into vessel 6 (PV6). High
pressure pump #6 (P6) supplies pressure for the 1st pass 1st and 2nd arrays and the
pressure exchanger #8 (P8) supplies pressure for the 1st pass 3rd array.
3. Sodium metabisulfite from chemical pump #2 (CP2) and tank #7 (TK7) can be added
after P5 to remove chlorine, if necessary. Free chlorine can damage RO membranes. The
22
-------
maximum allowable chlorine level is membrane specific with the minimum chlorine
tolerance being non-detect.
4. Anti-sealant from chemical pump #3 (CP3) and tank #8 (TK8) is added after P5 to
minimize RO membrane scaling.
5. P6 increases the pressure to the required 1st pass 1st array operating pressure (800-1,200
psi depending on water conditions).
6. Concentrate from the 1st pass 1st array flows through the pressure exchanger P8. P8
exchanges energy from the high pressure, high salinity 1st pass concentrate to the lower
pressure, lower salinity UF filtrate feed water. The UF filtrate pressurized by P8 flows
into the 2nd array.
7. Pressure control valves #5, #6, and #7 (PCV5, PCV6, PCV7) are used to adjust pressure
within the RO 1st pass piping. When PCV5 is fully open, P8 is bypassed. When
restricted, PCV5 provides backpressure for P6.
8. As PCV6 is restricted, water is forced through P8.
9. When open, PCV7 prevents P8 overflow during start up. When restricted, it provides
additional backpressure for P6.
10. Second pass operation is optional and will not be verified in this testing. During NBC
operations or when the 1st pass permeate quality does not meet requirements, the 2nd pass
is required.
11. The 2nd pass has one array with 12 membranes (PV7, PV8, PV9). The 1st pass permeate
feeds the 2nd pass. If the raw water source does not contain NBC, concentrate from the
2nd pass (which is lower concentration because 2nd pass feed is 1st pass permeate) is
recycled back to the raw water source to reduce the salinity of the inlet water.
12. Sodium hydroxide from chemical pump #4 (CP4) is added at the 2nd pass inlet to adjust
pH to improve the rejection of certain contaminants that are ionized at high pH such as
Boron.
13. Pump #7 (P7) pressurizes the 1st pass permeate. Pressure control valve #8 (PCV8)
provides the backpressure for pump #7 (P7).
14. The 1st pass permeate is monitored by and displayed on conductivity sensors #1 and #2
(CS1, CS2), which determine if the permeate purity meets requirements. Permeate
salinity is affected by temperature, TDS, and age of the RO membranes. If the permeate
purity does not meet requirements, CS1 de-energizes solenoid valve #1, which then
dumps the undesirable permeate back to the feed water source. If the permeate purity
meets requirements, CS2 activates solenoid valve #1, allowing the handle on the dump
valve to be latched, causing the high purity permeate to flow from the RO skid to the
product water storage tanks. This diversion feature is disabled during 2nd pass operation.
15. Prior to distribution, RO permeate flows through the calcium hypochlorite disinfection
system to the product water storage tanks. This system will not be operated during this
test phase.
2.4.3.3 RO CIP Procedure
The RO elements should be cleaned whenever the temperature corrected product water output
drops by 10 to 15% from the initial baseline established at the beginning of operation or from the
expected output. The RO elements should also be cleaned when the TDS level of the product
water exceeds 500 mg/L. Prior to cleaning the membranes, verify that any reduction in product
output is not the result of a corresponding variation in raw water inlet temperature or salinity by
23
-------
normalizing the data to a set of initial conditions. The following is a summarization of the
operating instructions from the operations manual:
1. Set RO system in normal operation mode. Verify that valves are in the correct startup
position. Make sure that the system output is being discharged to waste.
2. Select RO clean mode on main display screen.
3. Fill tank #4 with about 300 gal of fresh, un-chlorinated water to within 12 in of the top.
4. If ferric chloride is used in the system, perform the low pH adjustment first. If ferric
chloride is not used, perform high pH adjustment first. (ETV note: ferric chloride was
used during ETV test.)
5. Dissolve the appropriate amount of alkaline detergent or citric acid in a bucket of water.
6. Check the pH of the mixture in tank #4 and adjust as needed. Use citric acid to lower pH
to 3 or use sodium hydroxide to raise the pH to 11.
7. Start P5 and allow chemical solution to circulate for three minutes. Check and adjust pH
as needed.
8. Allow the cleaning solution to circulate for 15 min.
9. Touch "RO Clean" on the screen. Then touch "Enable RO Clean."
10. Allow system to soak for 1 to 15 hours (h).
11. After soaking for the desired length of time, re-circulate the cleaning solution for 30 min.
12. Drain system and dispose of cleaning agents.
13. Repeat above steps for each desired chemical solution.
14. Rinse the RO system with fresh water.
24
-------
X -=^_xi_J__xi L_xj_r
^ 3i~ X4 T X4 !~ X4 ]~ X4 ~}|> X BRINE
— 1 H L 1 J — It 1 GPM; 13-17
RO PRODUCT
WATER OUTLET TO
PRODUCT WATER
CHLORINATION INLET
GPM:
64-102 (1ST PASS ONLY)
50-69 (W/ 2ND PASS)
UF+RO 1ST/2ND PASS MODE
Figure 2-13. P&ID of RO skid.
25
-------
2.4.3.4 Pressure Exchanger
RO is an inherently power intensive process. Historically, energy from the high-pressure brine
was wasted through the utilization of a control valve to control the process. Today, several
systems are available to recover the energy contained in the high-pressure brine to help offset the
energy required. The EUWP uses the PX® Pressure Exchanger® (Model 90S) from Energy
Recovery, Inc (Figure 2-14). The PX operates on the principle of positive displacement to allow
incoming raw water to be pressurized by direct contact with the concentrate from a high-pressure
membrane system. It uses a cylindrical rotor with longitudinal ducts parallel to its axis to transfer
the pressure energy from the concentrate stream to the feed stream. The rotor fits into a ceramic
sleeve between two ceramic end covers with precise clearances that, when filled with high-
pressure water, create an almost frictionless hydrodynamic bearing. At any given time, half of
the rotor ducts are exposed to the high-pressure stream and half of the ducts are exposed to the
low-pressure stream. As the rotor turns, the energy is transferred to the low-pressure stream,
pushing the feed water on to the booster pump. This type of energy device has been shown to be
90% efficient in transferring energy. During operation in Alamogordo, New Mexico, the average
observed efficiency of the energy recovery device was 78 ± 8 %.
In a typical system, the pressurized feed water from the PX goes to a booster pump, which
restores the pressure lost in the exchange and feeds a second RO vessel. However, the EUWP
utilizes a parallel pass 1 train operation at approximately 10% lower pressure than the train
operating directly off the high pressure pump. PX dimensions are 24 in long x 6.5 in diameter.
Wetted materials are duplex stainless steel, ceramics, polyvinyl chloride (PVC), and fiberglass
reinforced plastic (FRP).
Feedwater | | Refect fluid | J Liqyid piston
Rotation
High-pressure feed water
going to 2nd parallel 1st pass
High-pressure concentrate or
reject fluid from reverse osmosis membranes
High pressure
Sealed area
Low pressure
H
Low-pressure feeclwater
Inlet from brackish supply
pump
PX—Pressure Exchanger™
Low-pressure concentrate or
reject flu id to drain
http://www.energv-recoverv.com/pdf/PX45SPX70SPX90S.pdf
Figure 2-14. PX pressure exchanger.
26
-------
2.5 General Requirements and Limitations
Table 2-5 lists the general environmental requirements for setup and operation of the EUWP.
Table 2-5. EUWP Site Considerations and Dimensions
Site Considerations
Site Dimensions
Drive-in access for on-road equipment
Work area required for equipment maneuvering and
setup
Fairly smooth, level, and clear ground surface
Cleared path to water source
Work area elevation above pump #1
Elevation/distance of pump #1 above the water source
Distance of pump #1 from inlet strainer #1 in water
source
Water depth from the inlet strainer #1 to the bottom of
the raw water source
Distance of distribution tanks from EUWP
Distance of distribution tanks from adjacent
distribution tank
Distance of distribution pump #9 from tee adaptors
Cleaning waste storage tank
At least 10 ft wide
At least 75 ft x 100 ft
Grade not to exceed 5° side to side and 2° end to end for
UF configured platform or skid. No restriction for the
RO skid. Ensure the elevation of tank #3 is equal to or
higher than the UF skid (higher is better).
Wide enough to move equipment
Maximum 25 feet vertical and 100 feet horizontal
Maximum 15 feet vertical and 50 feet horizontal
Maximum 50 feet
3 feet minimum; 5 feet or more preferred
Limited by hose length. Check hoses to determine
distance.
Limited by hose length. Check hoses to determine
distance.
Limited by hose length. Check hoses to determine
distance.
Less than 50 feet from the waste out connection
The EUWP was designed to be transported by air using a C-130 aircraft, or by land using any
number of commercial and military haul transporters. The skids have forklift pockets that allow
handling with an appropriately sized forklift.
Volume and type of consumables are site-specific depending on raw source water quality. As
recommended by the membrane manufacturer, calcium hypochlorite, citric acid, or sodium
hydroxide may be required to perform a CIP. Also as recommended by the membrane
manufacturer, citric acid, sodium hydroxide, and/or a membrane detergent may be required to
perform an RO cleaning. Depending on the raw water source quality, chemical additions may be
needed for protection of the membranes during operation. Ferric chloride may be added at the
UF skid to prevent clogging of the membranes by natural organic matter or high suspended
solids in the feed water. Antiscalant and/or sodium meta-bisulphite may be added at the RO skid
to prevent scaling and remove chlorine present in the feed water; and sodium hydroxide may be
added to raise the pH to aid rejection of constituents during the 2nd pass. Calcium hypochlorite in
granular or tablet form containing 65-70% free chlorine may be added prior to filtrate or
permeate storage as a disinfectant (this did not occur as part of this ETV test). Table 2-6 covers
equipment limitations and Table 2-7 presents membrane limitations.
27
-------
Table 2-6. Equipment Limitations
Svstem
Inlet Pump #1
Strainer
Parameter
Suction head (maximum)
Differential pressure (maximum) before manual backwash
Backpressure required for strainer auto flushing
25ft
7 psig
35 psig
Value
UF
Pretreatment requirements
Feed pressure (maximum)
Ambient temperature range
Water temperature range
Control air pressure
Damaging chemicals
TMP (maximum) before CIP required
Pressure surges
UF Membranes Stagnation time (maximum) before preservation required
with 1,000 - 5,000 mg/L sodium bisulfite (see operations
manual for details)
(see Table 2-1 for more details)
200 um strainer
45 psig
32 - 120°F
34 _104°F
60 psig
Grease, Oil, Silicon
35 psig
Minimize by operating valves slowly
14 days (somewhat temperature
dependent)
UF CIP Water Turbidity
Iron
Manganese
Aluminum
Reactive silica
Colloidal silica
Total silica
Calcium sulfate
Calcium carbonate
Microbiological
SDI
pH range
Maximum feed pressure
Maximum air pressure
Temperature range
Filtered
All water must be free of particulate matter
< 1.0 MTU
O.05 mg/L
O.05 mg/L
<0.5 mg/L
ND(1)
ND
<10 mg/L
< saturated at 50°C (122°F)
< saturated
no living or dead material
<3.0
1.5-13
45 psig
15 psig
32°F to 120°F
500 um prior to entering UF
RO
Water temperature range
Maximum SDI
Operating ambient temperature range
Storage and transport air temperature range
Relative humidity
Pretreatment requirements
Maximum operating concentrate pressure after backpressure
valve
Maximum operating permeate pressure (maximum)
2nd pass inlet pressure (maximum)
RO high pressure pump #6 maximum speed
RO high pressure pump #6 minimum inlet pressure
Stagnation time (maximum) before preservation required
34 _104°F
5 (membrane dependent)
32°F to 120°F
32°F to 120°F
3% to 95%
UF or 200 um strainer on RO skid
200 psig
100 psig
300 psig
600 revolutions per minute (RPM)
30 psig
1 week (somewhat temperature
RO Membranes (see Table 2-7 for details)
(1) Non-detect
28
-------
Table 2-7. Membrane Limitations
s
t s Si
a.
S
s
S
'B
ft
Membrane
ll
l!
'3
3
u
E
-------
2.6.3 RO Concentrate
The RO concentrate was blended with the RO permeate, UF backwash, and UF retentate, and the
resulting mixture was discharged into Lake St. Clair.
2.6.4 UF Backwash and Retentate
The UF system automatically initiates a backwash every 30 min to remove captured material
from the membrane surface. Each backwash cycle consists of backflushing the membranes with
UF filtrate for a short period followed by a forward "fast flush" using feed water. In addition to
the backwash, the UF system also discharges a continuous retentate stream.
Both waste streams exited the system using a common discharge line that was routed to a storage
tank. The contents of the storage tank were discharged to Lake St. Clair when the tank was full.
2.6.5 Discharge Permits
TARDEC obtained a permit from the Michigan Department of Environmental Quality to
discharge the mixture of UF backwash, UF retentate, RO concentrate, and RO permeate back
into Lake St. Clair.
2.7 Discussion of the Operator Requirements
The following information on operator requirements is supplied by the manufacturer for
informational purposes only. A team of four water treatment specialists, with proper site
validation, layout planning and using a 10,000-lb forklift, should be able to have the EUWP
setup and producing potable water within eight hours. Depending on the distribution connection
requirements and availability of the connections, distribution of the produced potable water may
take longer.
Except for periodic O&M and data collection, once set up and operational, the EUWP is capable
of operating unattended. Staffing requirements are based on the O&M or data collection efforts
being performed. Due to the use of high pressure, electricity or diesel, and chemicals, O&M on
the equipment and piping should be performed by a minimum of two persons. Data collection
requires only one person.
The EUWP requires a skilled operator familiar with water treatment processes, equipment, and
concepts to perform O&M and collect data. A skilled operator could meet any of a variety of
requirements as discussed below. Operation of the EUWP should be performed by an individual
with similar experience, knowledge, or training as provided within these programs.
A U.S. military water treatment specialist (classified as skill level 4 through 1) supervises or
performs installation, operation of water purification equipment, water storage, and distribution
operations and activities.
30
-------
The minimum skill level 4 requires the specialist to:
• Assist in water reconnaissance, site preparation, and setup of water treatment activity;
• Operate and maintain water treatment equipment;
• Receive, issue, and store potable water; and
• Perform water quality analysis testing and verification.
Although remote operation is not available, the EUWP can be monitored remotely 24 h per day
by use of the water system management tool, WaterEye™ (www.watereye.com). WaterEye
provides timely, critical operations monitoring information utilizing colored indicators to either
confirm system status or alert potential problems. In addition, WaterEye can assist with
managing daily, monthly, and yearly compliance requirements by monitoring compliance data
and automatically creating reports. WaterEye maintains a database of monitored instrument
readings, which are read every 15 min and uploaded to their server every 30 min. Alarm
conditions are immediately uploaded for response. WaterEye can also display/store information
calculated from uploaded instrument readings. Data must be either uploaded directly from the
PLC on the EUWP or be able to be calculated from that data.
-------
Chapter 3
Methods and Procedures
3.1 Introduction
The full EUWP system was tested at Selfridge ANGB during September and October of 2006.
Immediately prior to the test, TARDEC and USER discovered that the UF system seals between
the housings and membrane modules were not as tight as desired. The problem was temporarily
fixed, and NSF allowed the test to proceed because a future verification test at Port Hueneme,
CA, would verify UF seal integrity after the problem was permanently fixed. After testing was
complete, NSF reviewed the UF performance data and concluded that the temporary fix was not
sufficient. Therefore, a second test of the UF system only was required, after permanent repairs
were made. The UF system retest was conducted in July and August of 2007. See Section 4.3.1
for further discussion about the seal problem and how it was fixed.
3.2 Quantitative and Qualitative Evaluation Criteria
The objectives of the verification test were to evaluate equipment in the following areas:
• The actual results obtained by the equipment as operated under the conditions at the test
site;
• The impacts on performance of any variations in feed water quality or process variation;
• The logistical, human, and other resources necessary to operate the equipment; and
• The reliability, ruggedness, ranges of usefulness, and ease of operation of the equipment.
There are three main components of the EUWP that were evaluated at the same time: the UF
system, the RO system, and the energy recovery system. All three components must function
successfully to meet the performance objectives.
To address these objectives, the verification test employed the quantitative and qualitative factors
listed below.
Qualitative factor:
• Waste discharge requirements.
Quantitative factors:
• Water quality data;
• Physical operations data - flow, membrane flux, recovery, and pressure;
• Power usage;
• Chemical usage;
• Waste stream generation; and
• Operating cycle length.
32
-------
3.3 Key Treated Water Quality and Operational Parameters
Treated product water must meet EPA NPDWR, and should meet EPA secondary Standards
whenever possible. As discussed in Section 2.1, the objective of this ETV verification was to
demonstrate that the EUWP can provide water that meets the requirements of the EPA NPDWR.
As such, a list of key treated water parameters was developed based on the EPA regulations, and
other water quality parameters of interest. Regulated contaminants not present in raw water
samples analyzed during the characterization of feed water task were not included in the list. The
final list is presented in Table 3-1.
Table 3-1. Key Treated Water Quality Parameters
Parameter
pH Total Silica
Temperature TDS
Conductivity Total Organic Carbon (TOC)
Tuibidity Total Suspended Solids (TSS)
Particle Counts Ultraviolet Light Absorbance at 254 nm (UV254)
Alkalinity Heterotrophic Plate Count (HPC) Bacteria (2006 test)
Hardness Total Coliforms (2006 test)
Bacillus Endospores (2007 test)
A portion of the water quality and operational parameters were measured continuously via online
instrumentation, as listed in Table 3-2.
Table 3-2. Water Quality and Operational Parameters Measured Online
Membrane
Flow
Pressure
Conductivity
Temperature
Turbidity
Particle count
•a
fli
3
ta
P
X
X
X
X
2
1
1
«3
PH
ta
P
X
X
0)
1
—«
ta
ta
P
X
X
X
X
^ a. ^ «
g IS g fc
•g ^| B aJ
ta ^ „ ^ g «s g
O O § O S3 O S3
& & u etf &H etf &H
X X X X
XX X
X
OJ
"S
s
O
Q-
o
X
33
-------
3.4 Operations and Maintenance
Village Marine Tec. provided an operations and maintenance manual for the EUWP, which is
included in Appendix A. The ETV test protocols call for review of the manual in regards to the
ability of the user to successfully operate the system armed with only the information in the
manual. An objective review of the manual by the field operators was not possible, because they
already had intimate knowledge of the EUWP prior to the test. Therefore, a review is not
included in this report.
The following aspects of operability are addressed in Chapters 2 and 4, and in the appendices:
• Fluctuation of flow rates and pressures through unit (the time interval at which resetting
is needed);
• Presence of devices to aid the operator with flow control adjustment;
• Availability of pressure measurement;
• Measurement of raw water rate of flow;
• Pace of chemical feed with raw water; and
• Operation of the PLC control system.
3.5 Field Operations
Acting as the FTO, TARDEC conducted the testing of the EUWP as described below. TARDEC
and USER field personnel performed field analytical work using field laboratory equipment and
procedures for pH, temperature, conductivity, and turbidity. NSF performed water quality
analytical work for samples not analyzed on-site. Field staff were on site each day to operate the
system and collect water quality data during the verification test.
The test plan called for the EUWP to be operated 24 hours per day, seven days per week,
excluding regular backwashes and cleaning periods. However, this was not the case for much of
the test period due to various alarms that shut the system down during the night when field
personnel were not present.
3.6 Overview of ETV Testing Plan
A test/quality assurance plan (TQAP) was prepared for the EUWP verification test in accordance
with the ETV Protocols EPA/NSF Protocol for Equipment Verification Testing for Removal of
Inorganic Constituents - April 2002, and the EPA/NSF Protocol for Equipment Verification
Testing for Physical Removal of Microbiological and Particulate Contaminants - September
2005. The TQAP divided the work into three main tasks (A, B, C) with Task C, the verification
test itself, divided into six subtasks.
These tasks are:
Task A: Characterization of Feed Water
Task B: Equipment Installation, Initial Test Runs, and Initial System Integrity Tests
Task C: Verification Test
Task Cl: Membrane Flux and Recovery
Task C2: Cleaning Efficiency
Task C3: Finished Water Quality
34
-------
Task C4: Membrane Module Integrity
Task C5: Data Handling Protocol
Task C6: Quality Assurance/Quality Control (QA/QC)
The TQAP, which included a Quality Assurance Project Plan (QAPP), specified procedures to be
used to ensure the accurate documentation of both water quality and equipment performance. An
overview of each task is provided below with detailed information on testing procedures
presented in later sections.
3.6.1 Task A: Characterization of Feed Water
The objective of this initial operations task was to obtain a chemical, biological, and physical
characterization of the feed water prior to testing.
3.6.2 Task B: Equipment Installation, Initial Test Runs, and Initial System Integrity Tests
The objective of this initial operations task was to evaluate equipment operation and determine
the treatment conditions that resulted in effective treatment of the feed water. This task was
considered shakedown testing and was carried out prior to performing Task C.
3.6.3 Task C: Verification Test
The verification test itself consisted of six tasks described as follows:
3.6.3.1 Task Cl: Mem brane Flux and Recovery
Task Cl evaluated membrane operation and entailed quantification of membrane flux decline
rates and product water recoveries. The rates of flux decline demonstrate membrane performance
at the specific operating conditions established during Task B.
3.6.3.2 Task C2: Cleaning Efficiency
An important aspect of membrane operation is the restoration of membrane productivity after
membrane flux decline has occurred. The objective of this task was to evaluate the efficiency of
the membrane cleaning procedure. The fraction of specific flux restored following a chemical
cleaning and after successive filter runs was determined.
3.6.3.3 Task C3: Finished Water Quality
The objective of this task was to evaluate the quality of water produced by the EUWP. Treated
water quality was evaluated in relation to feed water quality and operational conditions. The
monitored water quality parameters included the following: pH, temperature, conductivity,
alkalinity, total hardness, total silica, TDS, turbidity, particle concentrations, TSS, TOC, and
ultraviolet light absorbance at a 254 nm wavelength (UV254). Also, total coliforms and HPC
bacteria were measured during the 2006 test, and Bacillus endospores were measured during the
2007 test. The switch to Bacillus endospores was made because there were too many HPC and
total coliforms present on the treated water sides of both the UF and RO skids, likely from the
existing populations of bacteria. It was hoped that there would be fewer Bacillus endospores
already present on the treated water sides of the skids, so that monitoring their populations would
provide a better indicator of microbial removal.
35
-------
3.6.3.4 Task C4: Membrane Module Integrity
The objective of this task was to demonstrate the methodology for monitoring membrane
integrity and to verify the integrity of membrane modules.
3.6.3.5 Task C5: Data Handling Protocol
The objective of this task was to establish an effective field protocol for data management at the
field operations site and for data transmission between TARDEC and NSF.
3.6.3.6 Task C6: Quality Assurance and Quality Control
An important aspect of verification testing is the protocol developed for QA/QC. The objective
of this task was to assure accurate measurement of operational and water quality parameters
during membrane equipment verification testing.
3.7 Task A: Characterization of Feed Water
The objective of this task was to determine the chemical, biological, and physical characteristics
of the feed water. Grab samples were collected from Lake St. Clair at the test site in August 2006
for water quality analysis. To evaluate the Lake St. Clair water for organic chemicals, a sample
was analyzed for the volatile organic compounds (VOC) included under EPA Method 502.2. It
was assumed that any organic compounds present in the water column at significant levels would
be VOCs. Any more hydrophobic organic chemicals, such as pesticides, would not be present
dissolved in the water column in significant quantities, but rather would be adsorbed onto
suspended organic particles.
3.8 Task B: Equipment Installation, Initial Test Runs, and Initial System Integrity Tests
The objective of this task was to properly install the equipment and begin equipment operation,
then evaluate operation and determine whether the operating conditions resulted in effective
treatment of the water. In this task, a preliminary assessment of the treatment performance of the
equipment was made. This task was considered a shakedown testing period and was completed
before Task C. This task also included pressure decay testing of the UF membranes, and dye
removal testing of the RO system. See Section 3.9.4.1 for further discussion about these two
tests.
3.9 Task C: Verification Testing
The verification test was started on September 25, 2006 and ran for the planned 30 day test
period, ending on October 25, 2006. The UF system was operated each day on semi-continuous
basis, automatically shutting down when the RO feed tank was full. A typical operating day for
the UF system was 15 to 17 h in duration. The RO system was setup to operate continuously.
After the first three days of the test, when the system was shutdown at night, the RO system
typically ran 22 to 24 h each day. The RO system did shutdown periodically for various
maintenance activities or when alarms occurred and shut the system down. When alarms and
shutdown occurred during unattended operation at night, the entire system would remain
shutdown until an operator arrived in the morning.
36
-------
The UF system retest was conducted from July 30 to August 24, 2007. The retest was stopped
short of 30 days because the intent of the test as stated in the ETV test protocol - operation to a
membrane cleaning - was met. During the retest, the UF system was operated an average of 14 h
per day, not including down time for backwashes, cleanings, and other maintenance activities.
The TQAP describes six tasks to be performed to achieve a successful verification test. Each of
these tasks is described in detail in this section.
3.9.1 Task Cl: Membrane Flux and Operation
The purpose of this task was to evaluate membrane flux during extended operation to
demonstrate membrane performance. The objectives of this task were to demonstrate the feed
water recovery achieved by the membrane equipment, and the rate of flux decline observed over
extended membrane operation. Flow, pressure, conductivity, and temperature data were collected
daily in order to quantify the loss of productivity in terms of specific flux decline.
3.9.1.1 Work Plan
Twice per day - in the morning and afternoon - the operator checked the flowrates and recovery
and made adjustments as necessary to put the system on target. Thirty minutes after resetting
target flow and recovery, the operator recorded the appropriate water quality and operational
data, as outlined in Table 3-3. The set points for key operating parameters are listed in Table 3-4.
Chemical usage was monitored by recording the concentration and tank level on a daily basis.
3.9.1.2 Evaluation Criteria
Completion of this task involved quantification of membrane flux decline rates and product
water recoveries. Summaries of the data collected for Task 1 are presented in tabular format in
Chapter 4 for both the RO and UF systems.
The plots listed in Table 3-5 are also presented in Chapter 4 to illustrate equipment operation for
Task 1. Note that all plots are of the parameter over time.
37
-------
Table 3-3. Operational Parameter Sampling Locations
^_
S5
Parameter
Flow X
Pressure
Conductivity
Temperature
Power usage
Operating hours
•a
%
ta
P
X
X
X
S -a
aj es cS
* a fe T=
i s -g "S
•3 <3 cs a>
tM P^ P3 ^
te. te. te. O
P P P P4
XXX
XX X
X
X X
C4 4»
* a
O |
X
X
X
OJ
1 1
* a
•s S
^H U
0 §
rt u
X
X
X
X
!°
^fi
ta
P
X
X
1
^fl
O
X
X
Table 3-4. Key Operating Parameters
Parameter
Set Point
UF feed flow (gpm)
UF recovery (%)
RO feed flow 1st pass array 1 (gpm)
RO feed flow 1st pass array 1 (gpm)
RO recovery levels (%)
259
90
116
58
50 (1st array) and 48 (2nd array)
Table 3-5. Operational Data Plots Appearing in Chapter 4
UF Skid
ROSkid
Filtrate Production
Flow Rates
Operating Pressures
Trans-Membrane Pressures
Specific Flux
Loss of Specific Flux
Power Consumption
Flow Rates
Percent Recovery
Operating Pressures
Specific Flux
Power Consumption
3.9.1.3 Equations
UF System
The following are the definitions and equations used for the verification report for the UF
system:
Filtrate: Treated water produced by the UF process.
Retentate: The water rejected by the UF system
Feed water. The water introduced to the membrane elements after all chemical additions.
38
-------
Raw water. The source water supply.
Membrane flux: The average flux across the UF membrane surface calculated by dividing the
flow rate of filtrate by the surface area of the membrane. Membrane flux is calculated as
follows:
s
where:
Jt = filtrate flux at time t (gallons per square foot per day (gfd))
Qp = filtrate flow (gpd)
S = membrane surface area (ft2)
Temperature Adjustment for Flux Calculation: Temperature corrections to 20°C for filtrate flux
and specific flux are made to correct for the variation of water viscosity with temperature. The
following empirically derived equation was used to provide temperature corrections for specific
flux calculations:
J,=-
where:
Jt
QP
s
T
S
filtrate flux at time t (gfd)
filtrate flow (gpd)
membrane surface area (ft2)
temperature of the feed water (°C)
Transmembrane Pressure: The pressure across the membrane, equal to the average feed water
pressure on the membrane (average of inlet pressure and outlet pressure) minus the filtrate
(permeate) pressure:
TMP =
-P.
where:
IMP
Pf
PC
Pr,
transmembrane pressure (psig)
inlet pressure to the feed side of the membrane (psig)
outlet pressure on the retentate side of the membrane (psig)
filtrate pressure on the treated water side of the membrane (psig)
Specific flux: The filtrate flux that has been normalized for the TMP. The equation used for
calculation of specific flux is given by the formula provided below. Specific flux is usually
discussed with use of flux values that have been temperature-adjusted to 20°C per equation
above:
where:
IMP
Jt
IMP
Transmembrane pressure across the membrane (psig)
filtrate flux at time t (gfd) (temperature-corrected flux values were employed)
39
-------
Jtm = specific flux at time t (gfd/psig)
RO System
Permeate: Water produced by the RO membrane process.
Feed Water: Water introduced to the membrane element.
Concentrate: Water rejected by the RO membrane system.
Permeate Flux: The average permeate flux is the flow of permeate divided by the surface area of
the membrane. Permeate flux is calculated according the following formula:
-4
where:
Jt = permeate flux at time t (gpd))
Qp = permeate flow (gpd)
S = membrane surface area (ft2)
Temperature Adjustment for Flux Calculation: Temperature corrections to 25 °C for permeate
flux and specific flux were made to correct for the variation of water viscosity with temperature.
The following empirically-derived equation was used to provide temperature corrections for
specific flux calculations:
Q xe-°-0239-(r-25)
J((at25°C) = ^ _
where:
Jt = permeate flux at time t (gfd)
Qp = permeate flow (gpd)
S = membrane surface area (ft2)
T = temperature of the feed water (°C)
Net Driving Pressure: For this test, a temperature conversion chart provided by the
manufacturer was used for all temperature correction. Net Driving Pressure (NDP) is the total
average pressure available to force water through the membrane into the permeate stream. Net
driving pressure is calculated according to the following formula:
where:
NDP = net driving pressure for solvent transport across the membrane (psig)
Pf = feed water pressure to the feed side of the membrane (psig)
Pc = concentrate pressure on the concentrate side of the membrane (psig)
Pp = permeate pressure on the treated water side of the membrane (psig)
A;r = osmotic pressure (psig)
40
-------
Osmotic Pressure Gradient: The term osmotic pressure gradient refers to the difference in
osmotic pressure generated across the membrane barrier as a result of different concentrations of
dissolved salts. The following equation provides an estimate of the osmotic pressure across the
semi-permeable membrane through generic use of the difference in TDS concentrations on either
side of the membrane:
A;r =
TDSc)
-TDSr
0.6 psi
~
where:
A;r
TDSf
TDSC
= osmotic pressure (psig)
= feed water TDS concentration (mg/L)
= concentrate TDS concentration (mg/L)
TDSp = permeate TDS concentration (mg/L)
Note that the different proportions of monovalent and multivalent ions composing the TDS will
influence the actual osmotic pressure, with lower unit pressures resulting from multivalent
species. The osmotic pressure ratio of 1 psig per 100 mg/L is based upon TDS largely composed
of sodium chloride or other monovalent ions. In contrast, for TDS composed of multivalent ions,
the ratio is closer to 0.5 psig per 100 mg/L TDS. Osmotic pressure was estimated using the ionic
strength of the feed and concentrate based on the weekly data for cations and anions (Ca, Mg,
Na, K, Li, Cl, SC>4, HCOs). The ratio of 1 psig per 100 mg/L TDS gave a much higher osmotic
pressure and the ratio of 0.5 psig per 100 mg/L TDS gave a lower osmotic pressure. It was
determined that the equation for TDS using a factor 0.6 psig per 100 mg/L TDS most closely
approximates the osmotic pressure calculated based on the ionic strength data available for this
water.
Specific Flux: The term specific flux is used to refer to permeate flux that has been normalized
for the net driving pressure. The equation used for calculation of specific flux is given by the
formula provided below. Specific flux is usually calculated with use of flux values that have
been temperature-adjusted to 25 °C:
J ~
Jtm ~
NDP
where:
Jtm
NDP
Jt
specific flux (gfd/psig)
net driving pressure for solvent transport across the membrane (psig)
permeate flux at time t (gfd). Temperature-corrected flux values should be
employed.
41
-------
Water Recovery: The recovery of feed water as permeate water is given as the ratio of permeate
flow to feed water flow:
% System Recovery = 1 00
^ /
where:
Qf = feed water flow to the membrane (gpm)
Qp = permeate flow (gpm)
Loss of Original Specific Flux:
Percent Loss = 100-1 -- -
where:
Jso = specific flux (gfd/psig) at time zero point of membrane testing.
Js = specific flux (gfd/psig) at time T of membrane testing.
Solute Rejection: Solute rejection is controlled by a number of operational variables that must be
reported at the time of water sample collection. Bulk rejection of a targeted inorganic chemical
contaminant may be calculated by the following equation:
C -C
f v
Percent Solute Rejection = 100 •
cf
where:
Cf = feed water concentration of specific constituent (mg/L)
Cp = permeate concentration of specific constituent (mg/L).
3.9.2 Task C2: Cleaning Efficiency
An important aspect of membrane operation is the restoration of membrane productivity after
specific flux decline has occurred. The effectiveness of chemical cleaning to restore membrane
productivity was evaluated.
3.9.2.1 Work Plan
The manufacturer specified that the UF cleaning procedure should be executed when the TMP
drop exceeds 35 psig, even after a backwash. The manufacturer specified that the RO system be
cleaned when there is a 10 to 15% decrease in normalized permeate flowrate, 15% increase in
TMP drop or permeate TDS concentration.
Flow, pressure, and temperature data were recorded immediately before the system was shut
down for cleaning and immediately upon return to membrane operation after cleaning procedure
was complete.
42
-------
Two primary indicators of cleaning efficiency and restoration of membrane productivity were
examined in this task:
• Immediate recovery of membrane productivity (% recovery of specific flux); and
• Long term maintenance of specific flux over an equivalent time period.
The pH, temperature, conductivity, and TOC of each cleaning solution were measured after the
cleaning. Flow, pressure, and temperature data were also collected during the cleaning
procedure. Following the cleaning procedure, the specific membrane flux was calculated at the
same operating conditions used prior to the cleaning. This value was compared to the pre-
cleaning specific flux to determine the efficiency of the cleaning procedure. See Section 2.4.2.2
for the UF cleaning procedure, Section 2.4.3.3 for the RO cleaning procedure, and also the
User's Manual (Appendix A) for details on the cleaning procedures employed.
3.9.2.2 Evaluation Criteria
The outputs for this task are post-cleaning flux recoveries, and the cleaning efficacy indicators
described above (including flow, pressure, and temperature data).
3.9.3 Task C3: Finished Water Quality
The objective of this task was to assess the ability of the membrane equipment (both UF and RO)
to meet the water quality goals specified by the manufacturer.
3.9.3.1 Work Plan
The water quality parameters in Table 3-6 were measured as indicated during the testing period.
To the extent possible, scheduled on-site analyses for each sampling point were performed on
water samples collected at the same time as the samples shipped off site.
In addition to manual sample collection for the water quality parameters listed in Table 3-6, in-
line particle counters recorded particle counts for the UF feed and UF filtrate streams every five
minutes. This data was only available for the 2007 UF test due to incorrect calibration of the
particle counters for the 2006 test. Note that particle count data is not presented in the water
quality discussion of Chapter 4, but rather in the membrane integrity section, since the primary
purpose of the particle counters is to serve as a monitor of membrane integrity.
43
-------
Table 3-6. Water Quality Sampling Schedule
Parameter
On-site Measurement
pH
Temperature
Conductivity
Turbidity
"8
4>
£
ta
P
D
D
D
D
OJ
CS
fe
ta
P
D
D
D
D
"8
-*^
a
2
OJ
Pi
ta
P
c§
Js
^s
u
sS
CO
ta
P
e
'3
^ =
o o
U |
P &
E
E
"C
fe
O
D
D
D
D
c§ oj
&H "g
^H g
o S
Pi &H
D
D
D
D
- -s
3 2
* a
•s o>
^H U
^
0 §
pi u
D
D
D
D
Laboratory Measurements
TOC
UV254
TSS
TDS, dissolved (<0.45 urn)
Alkalinity, total (as CaCO3)
Hardness, total (as CaCO3)
Silica, total (as SiO2)
HPC (2006 only)
Total Conforms (2006 only)
W
W
W
W
W
W
W
D
W
W
W
W
W
W
W
W
D
W
W E
W W
D D
W W
W
W
W
W
W
W
W
D
W
W
W
W
W
W
W
W
D
W
W
W
W
W
W
W
W
D
W
D = twice daily; exception is HPC, which was collected once daily, Monday through Thursday.
E = at every cleaning event
W = weekly
3.9.3.2 Evaluation Criteria
All water quality data generated during the test periods is presented in a tabular format in
Chapter 4. In addition, the UF feed and filtrate turbidity data, and the RO conductivity data is
presented in a graphical format.
3.9.4 Task C4: Membrane Integrity Testing
The objective of this task is to demonstrate the methodology to be employed for direct integrity
testing and indirect integrity monitoring of the RO and UF membrane elements. Direct testing
and indirect monitoring methods were used together to provide consistent and sensitive
evaluation of membrane system integrity.
3.9.4.1 Direct Integrity Testing:
The direct integrity testing method employed on the UF system was a pressure decay test, similar
to that described in ASTM International Standard D6908 - Standard Practice for Integrity
Testing of Water Filtration Membrane Systems. A pressure decay test was performed during
Task B to establish a baseline pressure decay rate for the UF system. During testing, the pressure
decay test was performed daily. The pressure decay test was also performed after each UF
system cleaning.
44
-------
The direct integrity test method employed on the RO system was a marker test with food-grade
dye. The marker dye test was conducted prior to the start of the ETV test, and after the RO
cleaning at the end of testing. The dye used was FD&C Food-grade Dye #40, Allura Red. The
concentration of this dye was measured in the RO feed and RO permeate to determine the level
of dye rejection by the RO system.
3.9.4.2 Continuous Indirect Integrity Monitoring:
Continuous indirect integrity monitoring methods were employed on both the UF and RO
systems. Turbidity was monitored continuously on the UF feed, UF filtrate, and RO permeate. In
addition to turbidity monitoring, particle counts were continuously monitored on the UF system,
and specific conductance was continuously monitored on the RO membrane unit. Turbidity
readings were recorded every fifteen minutes, while particle counts were recorded every five
minutes, and conductivity readings were recorded hourly. Results of the direct integrity tests, and
indirect integrity monitoring are presented in Chapter 4.
3.9.5 Task C5: Data Handling Protocol
The objectives of this task were to: 1) establish an effective structure for the recording and
transmission of test field test data, such that TARDEC provided sufficient and reliable data; and
2) develop an effective and accurate statistical analysis of the data.
3.9.5.7 Work Plan
The EUWP test system was equipped with a computer monitoring system. Some of the required
measurements (see Table 3-2) were recorded automatically by the automated system. The
remaining required measurements were recorded by hand by the field operator on-site. The data
was recorded onto specially prepared bench sheets. Miscellaneous operational notes were
recorded in a data logbook with numbered pages (Appendix B). All errors were crossed out with
one line, and the error was initialed and dated. Completed pages were signed, dated, and
numbered by the individual responsible for the entries.
The database for the project was set up in the form of custom-designed spreadsheets. A
spreadsheet containing the operational data, including calculations, was developed by USER. A
spreadsheet containing the water quality data was developed by NSF. Following data entry,
100% of the data in the spreadsheets was checked against the numbers on the field log sheets or
laboratory analysis outputs.
3.9.6 Task C6: Quality Assurance Project Plan
QA/QC of the operation of the equipment and the measured water quality parameters was
maintained through a QAPP, as described in this section.
3.9.6.1 Experimental Objectives
The objective of this task was to maintain strict QA/QC methods and procedures during the
verification test. This included maintaining instrument calibration and operation within the
ranges specified by the manufacturer.
45
-------
The elements of the QAPP for this verification test included:
• work plan;
• QA/QC verifications;
• data correctness;
• calculation of indicators of data quality; and
• corrective action plan
3.9.6.2 Work Plan
A routine daily walk-through during testing was conducted to verify that each piece of
equipment or instrumentation was operating properly. Chemical addition rates and receiving
stream flowrates were checked to verify that they flowed at the expected rates. Values recorded
by the automated data acquisition program were checked daily against those displayed on the
instrument displays and those measured on-site.
3.9.6.3 QA/QC Verifications
Tables 3-7 and 3-8 give the on-site QA and on-site QC activities, respectively, for the
verification test. NSF Laboratory analytical QA and QC activities followed those specified in the
NSF Laboratory Quality Assurance Manual.
Table 3-7. On-Site Analytical Equpment QA Activities
Equipment
Action Required
Initial Flowmeters - electronic
Turbidimeter - in-line (1720E)
Turbidimeter - in-line (FilterTrak)
Particle counter - in-line
UV spectrophotometer
Verified calibration volumetrically
Provided factory calibration certificate
Provided factory calibration certificate
Provided factory calibration certificate
Provided factory calibration certificate
Daily Chemical feed pump
Turbidimeter - in-line
pH meter - portable
Turbidimeter - in-line
Particle counters - in-line
Volumetrically checked flowrate
Verified with portable turbidimeter
3-point calibration (4,7,10)
Volumetrically checked flowrate
Volumetrically checked flowrate
Weekly Rotameters
UF filtrate flow
Particle counter - in-line
Temperature - portable
Turbidimeter - portable
Conductivity meter - portable
Inspected for buildup of algae, salt, etc.
Verified volumetrically
Cleaned sensors
Verified calibration with NIST-certified thermometer
Calibrated using <0.1, 20, 100, and 800 NTU standards
Calibrated at 2 points
Every Two
Weeks
Flowmeters - electronic
Verified calibration volumetrically
Prior to Tubing
Test Particle counter - in-line
Turbidimeter - in-line (1720E)
Turbidimeter - in-line (FilterTrak)
Checked condition, checked for leaks
Factory calibration
Cleaned and calibrated using 20 NTU standard
Cleaned and calibrated using 0.8 NTU standard
46
-------
Table 3-8. On-Site Data Generation QC Activities
Item Action Required
Daily Data Reviewed system performance data since previous day
Weekly Data Compared field and lab water quality results when available
3. 9. 6. 4 Data Correctness
There are five indicators of data quality that were used for this verification test:
• representativeness;
• statistical uncertainty;
• precision;
• accuracy; and
• completeness.
These five indicators are discussed in detail in the sections that follow.
3.9.6.4.1 Representativeness
Representativeness of the data for this verification was ensured by executing consistent sample
collection and data collection procedures, including:
• Consistency of sample locations;
• Timing of sample collection;
• Analytical methods; and
• Sampling procedures, sample preservation, packaging, and transport.
3. 9.6.4.1.1 On-Site Analytical Methods
The analytical methods for on-site monitoring of raw and treated water quality are described
below.
Analyses for pH were performed according to Standard Method 4500-H+ B using a Myron L
Ultrameter II Model 6P. Three-point calibration (using pH 4, 7, and 10 buffer solutions) was
performed daily.
Temperature
Readings for temperature were conducted in accordance with Standard Method 2550 using a
Myron L Ultrameter II Model 6P. A calibration check was performed weekly with a NIST-
traceable thermometer.
Turbidity
Turbidity was measured at all sampling points using a hand-held turbidimeter. In addition, in-
line turbidimeters were used for measurement of UF feed and filtrate. All measurements were
conducted according to Standard Method 2130 B.
47
-------
Hand-held Turbidimeters: A Hach 21 OOP Portable Turbidimeter (range 0 to 1000 NTU) was
used to measure the turbidity of the appropriate grab samples. The turbidimeter was calibrated
weekly using formazin turbidity standards of <0.1, 20, 100, and 800 NTU.
In-Line Turbidimeters: In-line Hach turbidimeters were used for measurement of turbidity in the
feed (Hach 1720 E - Low Range) and UF filtrate water (Hach FilterTrak 660). The Hach 1720E
has a range from 0 to 100 NTU and uses a 20 NTU calibration standard. The Hach FilterTrak has
a range from 0.005 to 5.00 NTU and uses a 0.8 NTU calibration standard. These turbidimeters
were calibrated at the start of the test. In-line readings were periodically compared to the
readings from the hand-held turbidimeter. If the comparison suggested inaccurate readings, the
in-line turbidimeter was recalibrated. A volumetric check on the sample flowrate was performed
daily.
Conductivity
Analyses for conductivity were performed according to manufacturer's instructions using a
Myron L Ultrameter II Model 6P. A two-point calibration was performed weekly.
Particle Count
In-line particle counters were employed for measurement of particle concentrations in UF
membrane unit feed and filtrate waters. The Hach 2200 PCX in-line particle sensor is able to
measure particles with a range of 2 um to 750 um in up to 32 user-defined bins. The particle
counters were calibrated by the manufacturer prior to the ETV test.
3.9.6.4.1.2 Sample Collection, Shipment, and Storage for Laboratory Analyses
Samples were collected in bottles prepared by NSF and shipped to the test site. All samples were
preserved, if required, according to the proper analytical method. Bottles for parameters
requiring preservation were shipped to the test site containing the preservative. All samples were
kept on ice in coolers and shipped overnight to NSF. Chain of custody forms accompanied all
samples. No travel blanks were required during Task C testing because no organic chemical
analyses were required. All samples were analyzed within the allowable hold time.
3.9.6.4.1.3 Laboratory Analytical Methods
A comprehensive list of laboratory analytical methods used can be found in Table 3-9. TDS from
the lab analysis was correlated to conductivity for calculation of normalized permeate flow and
rejection trends over time. TDS was used to calculate the osmotic pressure gradient needed for
net driving pressure calculations.
48
-------
Table 3-9. Analytical Methods for Laboratory Analyses
Parameter
TOC
UV254
TSS
TDS
Alkalinity, total (as CaCO3)
Hardness, total (as CaCO3)
Silica, total (as SiO2)
HPC
Total coliforms
Method
SM(1)5310C
SM 5910B
EPA 160.2
SM 2540C
EPA 3 10.2
SM 2340C
EPA 200.7
SM9215B
SM 9222B
NSF Reporting
Limit
O.lmg/L
0.000 Abs/cm(2)
5mg/L
5mg/L
5mg/L
2mg/L
0.1 mg/L
1 CFU/mL
1 CFU/lOOmL
Hold Time
28 days
2 days
7 days
7 days
14 days
180 days
28 days
24 hours
24 hours
Sample Container
4-40 mL glass
1 L plastic
1 L plastic
1 L plastic
1 L plastic
125 mL plastic
125 mL plastic
125 mL plastic
125 mL plastic
Sample
Preservation
H2SO4 to pH<2
none
none
none
none
HNO3 to pH<2
none
none
none
(1) SM=Standard Methods for the Examination of Water and Wastewater
(2) Abs/cm = UV absorbance per centimeter
3.9.6.4.2 Statistical Uncertainty
For the water quality parameters monitored, 95% confidence intervals were calculated for data
sets of eight values or more. The following equation was used for confidence interval
calculation:
Confidence Interval = x± [tn-i,i .(0/2) x (S/Vn)]
where:
x = sample mean
S = sample standard deviation
n = number of independent measurements included in the data set
t = Student's t distribution value with n-1 degrees of freedom
a = significance level, defined for 95% confidence as: 1 - 0.95 = 0.05
According to the 95% confidence interval approach, the a term is defined to have the value of
0.05, thus simplifying the equation for the 95% confidence interval in the following manner:
95% Confidence Interval = x + [tn-i,o.975 x (S/Vn)].
3.9.6.4.3 A ccuracy
The accuracy of on-site analytical equipment was periodically verified according to the schedule
in Table 3-7. The calibration records for the analytical equipment were recorded on bench sheets
(Appendix B). All calibrations were performed at the frequency required. All calibration data
were within the specified QC objectives on all days analyses were performed.
Accuracy for the laboratory analyses was quantified as the percent recovery of a parameter in a
sample to which a known quantity of that parameter was added. The following equation was
used to calculate accuracy:
where:
Percent Recovery = 100 x [(Xknown - Xmeasured) +
n = known concentration of measured parameter
Xmeasured = measured concentration of parameter
49
-------
Accuracy also incorporates calibration procedures and use of certified standards to ensure the
calibration curves and references for analysis are near the "true value." Accuracy of analytical
readings was measured through the use of spiked samples and lab control samples.
The NSF Laboratory Quality Assurance Manual establishes the frequency of spike sample
analyses at 10% of the samples analyzed. Laboratory control samples are also run at a frequency
of 10%. The recovery limits specified for the parameters in this verification were 70-130% for
laboratory-fortified samples and 85-115% for laboratory control samples. The NSF QA
department reviewed the laboratory records and found all analyses for all sample groups were
within the QC requirements for recovery. Calibration requirements were also achieved for all
analyses.
As an additional check on accuracy, performance evaluation (PE) samples were purchased and
sent to the field technicians for analysis.
3.9.6.4.4 Precision
Precision refers to the degree of mutual agreement among individual measurements and provides
an estimate of random error. To quantify precision, the relative percent difference (RPD) of
duplicate analyses was calculated. RPD was measured by use of the following equation:
RPD=
:200
where:
Sl = sample analysis result; and
^ = sample duplicate analysis result.
Acceptable analytical precision for the verification test was set at an RPD of 30%. Field
duplicates were collected at a frequency of 1 out of every 10 samples for each parameter, to
incorporate both sampling and analytical variation to measure overall precision against this
objective. In addition, the NSF Laboratory also conducted laboratory duplicate measurements at
10% frequency of samples analyzed. The laboratory precision for the methods selected was
tighter than the 30% overall requirement, generally set at 20% based on the standard NSF
Chemistry Laboratory method performance.
3.9.6.4.5 Comple teness
Completeness refers to the amount of valid, acceptable data collected from a measurement
process compared to the amount expected to be obtained. Completeness was quantified
according to the following equation:
%C = (V/T)X100
where:
%C = percent completeness
V = number of measurements judged valid
T = total number of measurements
50
-------
The completeness objective for data generated during this verification test was based on the
number of samples collected and analyzed for each parameter and/or method. Table 3-10
presents the completeness requirements based on the sampling frequency spelled out in the
test/QA plan.
Table 3-10. Completeness Requirements
Number of Samples per Parameter and/or Method Percent Completeness
0-10 80%
11-50 90%
> 50 95%
3.9.6.5 Operation and Maintenance
The EUWP was operated and maintained according to limits stated in Chapter 2 and the EUWP
Operation and Maintenance Manual.
51
-------
Chapter 4
Results and Discussion
This chapter presents a summary of the water quality and operating data collected during
the ETV test. Operating data are presented to describe the flow rates, volume of treated
water produced, backwash volumes and frequency, pressure differential across the UF
and RO skids, and related operating information. Water quality data are presented for the
key parameters listed in Table 3-1. Information on membrane integrity testing, pressure
decay testing, and additional work performed during the 2007 retest of the UF
membranes is also included in this Chapter. QA/QC information, as described by the
QAPP (Section 3.9.6) for this verification test, is presented at the end of the chapter.
Two test events occurred: a test of the full EUWP system in 2006, and a retest of the UF
skid in 2007. For the 2006 test, the EUWP system was delivered to Selfridge ANGB at
the end of August 2006. Shakedown testing was conducted in September. The 30-day
verification test began on September 25, 2006, and ended on October 25, 2006.
During the equipment installation and initial test runs phase of the 2006 verification test,
TARDEC and USER discovered that the UF system seals between the housings and
membrane modules were not as tight as desired. The pressure decay tests indicated that
system was not meeting the expected performance requirement and could allow leakage
that would result in lower removals of biological agents. The problem was temporarily
fixed with polytetrafluoroethylene (PTFE) tape, and a subsequent pressure decay test
yielded a pressure decay rate that satisfied TARDEC and USER. NSF allowed the test to
proceed with the temporary fix because a future verification test of a second EUWP
system at Port Hueneme, CA, would verify UF seal integrity after the problem was
permanently fixed. See Sections 4.1.2 and 4.1.3.4 for the pressure decay test data and
further discussion.
After reviewing the UF performance data from the 2006 test, NSF concluded that the
temporary fix was not sufficient, and the UF system should be retested after the seal
problem was permanently fixed. The cause was determined to be deformation of the
nylon end caps and filtrate tube adapters, perhaps from contact with chlorine during
cleaning, or from exposure to sunlight during operation. New filtrate tube adapters were
fabricated, and the end caps were re-machined. Also, thicker o-rings were used between
the filtrate tube adapters and end caps. The EUWP was shipped back to Selfridge ANGB
from storage in Alamogordo, New Mexico in July of 2007 for the UF retest. The test of
the repaired UF system was performed from July 30 to August 24, 2007.
The results of the 2006 ETV test and the additional 2007 retest, are presented herein.
Both the UF and RO skids were operated for the 2007 retest, but ETV test data was only
collected from the UF system. Note that the 2007 retest was stopped short of 30 days
because the intent of the test as stated in the ETV test protocol - to operate until a
membrane cleaning was conducted - was met.
52
-------
4.1 2006 EUWP Test
4.1.1 Task A: Raw Water Characterization
Two sets of grab samples were collected in August 2006 to characterize the raw water
supply, and to determine if any regulated metals or VOCs were present and should be
included in the final sampling plan. The results of these analyses are presented in Table
4-1. Based on these results, no metals or VOCs were added to the sampling plan.
4.1.2 Task B: Equipment Installation and Initial Test Runs
The objective of this task was to evaluate equipment operation and determine whether the
operating conditions result in effective treatment of the water. In this task, a preliminary
assessment of the treatment performance of the equipment was made. This task is
considered a shakedown testing period and was completed before the start of the
verification test.
The unit plumbing, electrical hook-ups, and pumping of raw water to the UF feed tank
were completed on September 12. The initial test runs and shakedown period took place
between September 13 and the beginning of the official ETV test on September 25.
During this period, all sensors were calibrated, communications were established with the
particle counters and turbidimeters, and the PLC was operated to check that programming
and data collection were operating properly. Inline turbidimeters were calibrated and,
based on these results, the manufacturer was called to calibrate the filtrate inline unit.
Handheld analyzers were calibrated and checked and colorimetric methods were tested. It
was determined that ferric chloride coagulation would not be necessary to keep the UF
system running smoothly. Subsequently, after the test started, UF membrane fouling
issues resulted in ferric chloride being added as a coagulant to lengthen run times
between chemical cleanings (CIP).
A pressure decay test for the UF system was an important part of the initial test runs to
verify that the UF membranes and the connections were properly sealed. Pressure decay
tests were performed on September 13, 14 and 15. These tests showed that pressure was
being lost at a higher than desirable rate. An investigation of the problem revealed that
the o-ring seals between the membrane modules and filtrate collection tubes were
unsatisfactory. To confirm that only the o-rings were the cause of the high pressure decay
rates, each membrane cartridge was integrity tested individually using an "air bubble"
test. Each membrane cartridge was submerged in a trough of water and air pressure
applied to check for bubbles emerging from the ends of individual UF fibers. These tests
indicated that no fibers were compromised, based on the field logs noting each membrane
as a "pass". PTFE tape was wrapped around the o-rings in question to increase the seal
surface between the o-rings and the membrane cartridges, and the cartridges were re-
installed in the UF system. This reduced the pressure decay rate to 0.37 psig/min, and the
permeate turbidity from 0.3 NTU to 0.08 NTU.
53
-------
Table 4-1. Initial Characterization Sampling Results
Parameter
Sample Date
08/02/06 08/16/06 05/31/07
Parameter
Sample
Date
08/16/06
TOC (mg/L) 2.9
UV254 (Abs) 0.0668
TSS (mg/L) ND(5) ND(2)
TDS (mg/L) 130 140
Alkalinity (mg/L CaCO3) 70 86
Total Hardness (mg/L as CaCO3) 95 110
Nitrate (mg/L of N) ND(0.05) 0.25
Nitrite (mg/L of N) ND(0.02) ND(0.02) ND(0.02)
Total Silica (mg/L SiO2) 1.1 1.1
Orthophosphate (mg/L P) ND(0.02)
Specific Conductance (umhos/cm) 250
Cryptosporidium (oocysts/L) <1
Giardia (cysts/L) <1
HPC (CFU/mL) 500
Total Coliforms (CFU/100 mL) 291
Bacillus Endospores 689
Regulated Metals by EPA 200.8 (all |ag/L)
Antimony ND(0.6) ND(0.5)
Total Arsenic ND(1) 1 ND(2)
Barium 18 20 17
Beryllium ND(0.5) ND(0.5)
Cadmium ND(0.3) ND(0.3) ND(0.2)
Chromium ND(1) ND(1) ND(1)
Copper ND(2) 2 2
Lead ND(1) ND(1) ND(1)
Mercury ND(0.2) ND(0.2) ND(0.2)
Selenium ND(4) ND(4) ND(2)
Thallium ND(0.2) ND(0.2) ND(0.2)
Unregulated VOC's by EPA 502.2 (all |ag/L)
Dichlorodifluoromethane ND(0.5)
Chloromethane ND(0.5)
Vinyl Chloride ND(0.5)
Bromomethane ND(0.5)
Chloroethane ND(0.5)
Trichlorofluoromethane ND(0.5)
Trichlorotrifluoroethane ND(0.5)
1,1-Dichloroethylene ND(0.5)
Methylene Chloride ND(0.5)
trans-l,2-Dichloroethylene ND(0.5)
1,1-Dichloroethane ND(0.5)
2,2-Dichloropropane ND(0.5)
cis-l,2-Dichloroethylene ND(0.5)
Chloroform ND(0.5)
Bromochloromethane ND(0.5)
1,1,1-Trichloroethane ND(0.5)
1,1-Dichloropropene ND(0.5)
Carbon Tetrachloride ND(0.5)
Unregulated VOC's by EPA 502.2
(continued)
1,2-Dichloroethane ND(0.5)
Trichloroethylene ND(0.5)
1,2-Dichloropropane ND(0.5)
Bromodichloromethane ND(0.5)
Dibromomethane ND(0.5)
cis-l,3-Dichloropropene ND(0.5)
trans-l,3-Dichloropropene ND(0.5)
1,1,2-Trichloroethane ND(0.5)
1,3-Dichloropropane ND(0.5)
Tetrachloroethylene ND(0.5)
Chlorodibromomethane ND(0.5)
Chlorobenzene ND(0.5)
1,1,1,2-Tetrachloroethane ND(0.5)
Bromoform ND(0.5)
1,1,2,2-Tetrachloroethane ND(0.5)
1,2,3-Trichloropropane ND(0.5)
1,3-Dichlorobenzene ND(0.5)
1,4-Dichlorobenzene ND(0.5)
1,2-Dichlorobenzene ND(0.5)
Carbon Disulfide ND(1)
Methyl-tert-Butyl Ether ND(0.5)
Methyl Ethyl Ketone ND(0.5)
Methyl Isobutyl Ketone ND(0.5)
Toluene ND(0.5)
Ethyl Benzene ND(0.5)
m+p-Xylenes ND(1)
o-Xylene ND(0.5)
Styrene ND(0.5)
Isopropylbenzene ND(0.5)
n-Propylbenzene ND(0.5)
Bromobenzene ND(0.5)
2-Chlorotoluene ND(0.5)
4-Chlorotoluene ND(0.5)
1,3,5-Trimethylbenzene ND(0.5)
tert-Butylbenzene ND(0.5)
1,2,4-Trimethylbenzene ND(0.5)
sec-Butylbenzene ND(0.5)
p-Isopropyltoluene ND(0.5)
1,2,3-Trimethylbenzene ND(0.5)
n-Butylbenzene ND(0.5)
1,2,4-Trichlorobenzene ND(0.5)
Hexachlorobutadiene ND(0.5)
1,2,3-Trichlorobenzene ND(0.5)
Napthalene ND(0.5)
Benzene ND(0.5)
Total Trihalomethanes ND(0.5)
54
-------
The pressure decay rate was higher than TARDEC and USER desired, but NSF and EPA
deemed it acceptable to start the verification test because previous laboratory challenge
tests on a smaller version of the Koch UF cartridge demonstrated bacteria removal of six
logio or greater, with a corresponding pressure decay rate of 0.29 psig/min (NSF 2006).
Over the course of the test, the daily pressure decay test rates ranged from 0.20 to 0.43
psig/min, with a mean of 0.29 psig/min, so the daily pressure decay test results did not
raise any alarms that there were continuing membrane integrity issues.
The RO system was dye tested on September 23, 2006. This test showed a rejection rate
of 99.5%. The inline conductivity meters were also monitored at the start of operation to
confirm the rejection rate of the RO membranes. See Section 4.1.3.4 for further
discussion.
TARDEC and USER were satisfied with the performance of the EUWP, as indicated by
the outputs of the in-line particle counters and turbidimeters for the UF system, and the
conductivity meters for the RO system (data not shown), so testing began on September
25.
4.1.3 Task C: Verification Test
The 2006 verification test was started on September 25 and ran for the planned 30 day
test period, ending on October 25. The UF system was operated each day on a semi-
continuous basis, automatically shutting down when the RO feed tank was full. A typical
operating day for the UF system was 15 to 17 h in duration. The RO system was setup to
operate continuously. After the first three days of the test, when the system was shutdown
at night, the RO system typically ran 22-24 h each day. The RO system did shutdown
periodically for various maintenance activities or when alarms occurred and shut the
system down. When alarms and shutdown occurred during unattended operation at night,
the entire system would remain shutdown until an operator arrived in the morning.
The on-site operators collected operating data and on-site water quality samples twice per
day in accordance with the test plan schedule. The following sections present the
operating data and water quality data.
4.1.3.1 Task Cl: Membrane Flux and Operation
The purpose of this task was to evaluate system performance during operation. The
objectives of this task were to demonstrate the appropriate operational conditions for the
system, the feed water recovery achieved by the UF and RO membranes, and the rate of
flux decline observed over the operation period.
Operational data were collected and on-site water quality measurements were made twice
per day throughout both test periods, except for days when the UF was being cleaned and
therefore not operating for a portion, or all, of the day. The complete data set can be
found in Appendix B.
55
-------
As discussed in Section 4.1.2, during the 2006 test the UF system experienced membrane
seal integrity issues. The seal problems could affect the quality of the treated water, but
as the data will show the microscopic leaks around the seals did not impact the collection
of operational measurements, such as flow rates, pressure differentials, temperature, and
calculated specific flux. Filtrate from the UF system was the feed water for the RO
system during this first test.
4.1.3.1.1 UF System
The UF operational statistics for the 2006 test are presented in Table 4-2. The UF skid
does not have a filtrate flow meter or filtrate pressure gauge. Therefore, the total filtrate
flow rate was calculated as the UF feed water flow rate minus the UF retentate flow rate.
The intake flow is the intake from the source water into the UF feed water tank. The
intake pump is technically not part of the UF skid, but the intake flow is included here as
part of the overall UF treatment process. The intake pump ran at a higher flow rate than
the UF system to ensure that the UF feed water tank always contained sufficient water to
operate the UF system.
Table 4-2. UF Operational Measurement Statistics for 2006 Test
95%
Standard Confidence
Parameter Count Mean Median Minimum Maximum Deviation Interval (CI)
UF Operation per day (h)
Intake Flow (gpm)
Feed Flow (gpm)
Filtrate Flow (gpm)
Retentate Flow (gpm)
Backwash Flow (gpm)
Feed Pressure (psig)
Retentate Pressure (psig)
Filtrate Temperature (°F)
31
58
59
59
59
59
59
59
15.0
298
246
220
26
21
19
52
17.2
299
248
222
26
Estimated
21
19
52
3.4
278(1)
175
149
21
at 900
12
10
43
21.5
302
268
243
31
4.85
3.34
16.0
16.1
1.81
±1.71
±0.86
±4.07
±4.10
±0.46
gal per backwash cycle
33
31
60
4.2.6
4.20
5.16
±1.09
±1.07
±1.32
(1) Intake flow of 181 gpm recorded the morning of October 5. This reading is considered an outlier and
has not been included in the statistics because the intake strainer was partially clogged. The afternoon
reading on October 5 after the cleaning was 298 gpm.
The mean UF feed water flow rate of 246 gpm was somewhat below the design feed flow
rate of 259 gpm specified for the system (See Table 3-2). The mean filtrate flow rate of
220 gpm corresponds to a flow rate of 13.8 gpm for each of the 16 UF membrane
modules. Using these mean flow numbers, the UF water recovery was 89.5% based on
the mean feed water and filtrate flow rates. While the UF system did not operate a full
twenty four hours per day during this test, the 220 gpm mean filtrate flow would
correspond to a 24-h production rate of 316,800 gal. This filtrate production volume
includes water used for backwashes.
The stated UF production rate is 250,000 gpd (not including backwash water). The
backwash process uses 900 gal of UF filtrate per event, and a backwash is conducted
56
-------
every 30 min. For 24 h of operation, 48 backwashes would be conducted, using a total of
43,200 gal of UF filtrate. Subtracting this backwash volume from the calculated 24-h
volume of 316,800 gal leaves 273,600 gal of UF product water, which is above the
specified 250,000 gpd.
The EUWP included a totalizer to track the hours of UF system operation. The daily
hours of operation varied widely, from 3.4 to 21.5, depending on the volume of filtrate
required for the RO system and downtime for various maintenance activities. The UF
system was operated an average of 15.0 h per day. There were six days of operation of
less than 10 h. Excluding these days would increase the mean hours of operation to 17.1 h
(range of 12.1 to 21.5 h).
UF filtrate production was also tracked using the RO feed totalizer. This production
volume was the actual filtrate used for the RO feed water and thus does not include the
filtrate used for backwash waste. Figure 4-1 shows the cumulative water production for
the UF system for the duration of the 2006 verification test. The UF filtrate volume
produced includes both the RO feed and the filtrate used for backwashes. Because of the
wide range of operational hours per day, the UF production was also calculated based on
gal produced per hour of operation. The mean UF production per hour was 11.8
kilogallons (kgal), with a range oil.61 to 14.7 kgal/hr.
7000
6000
» 5000
A
=
1 4000
=
•a
2
a.
S 3000
1= 2000
1000
0
-UF Filtrate
-ROFeed
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06
Date
Figure 4-1. UF system filtrate production for 2006 test.
10/25/06
57
-------
Figure 4-2 shows the UF system flow rates over the duration of the 2006 verification test.
The retentate flow rate remained steady throughout the test. The feed water flow rate and
filtrate flow rate dropped over time as the membrane became fouled with solids and TMP
increased. Figure 4-3 shows the feed and retentate pressures during the test and Figure 4-
4 shows the calculated TMP results. These three figures clearly depict the impact of
solids build up on the UF membranes during the first few days of operation and again
during the last week of operation.
The increase in TMP on September 30 coupled with the decrease in flow rate indicated
that the membranes required a CIP, as the normal backwash cycle was not sufficiently
cleaning the UF membranes. The system was shutdown on September 30 and a CIP
initiated. After the cleaning was completed, flow rates and TMP returned to normal
ranges and similar to the values measured at the beginning of the test.
A chemical coagulant (ferric chloride) was not used at the beginning of the verification
test. However, after the fairly rapid increase of TMP occurred, it was evident that a
coagulant should be used to attempt to lengthen the time between required CIP events.
Ferric chloride was fed to the feed water upstream of the UF membranes beginning on
September 29, 2006 and continued for the duration of the test. A purchased ferric
chloride solution with a concentration of 33-36 % ferric chloride (12% as Fe) was fed to
the intake water at a feed rate of 4 ml/min (0.06 gal/hr) to give an iron dose of 0.42 mg/L
as Fe. The addition of the coagulant improved performance and the system was able to
maintain filtrate production and TMP below 20 psig until the last two days of the test. If
the verification test were to have continued beyond thirty days, a CIP would have been
necessary to maintain flow rate and lower the TMP.
Figure 4-5 shows the specific flux calculated for the UF system during the test. The
impact of solids buildup on the system is clear prior to the CIP performed on September
30. The CIP was successful as the specific flux was actually higher after the cleaning than
at the beginning of the test. This may have been due to some solids buildup on the
membranes during the shakedown and startup period.
Figure 4-6 shows the loss (or gain) of specific flux over the duration of the verification
test. The change in specific flux is calculated by comparing the specific flux on a given
day to the value calculated at the start of the test. This type of data shows the impact of
cleaning and backwash by comparing any given days specific flux to the start of the test.
58
-------
300
250
200
E
o.
M
01
03
«
o
tu
150
100
50
System cleaned
on 9/30
• Feed -•- Filtrate -*- Retentate
0
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-2. Plot of UF system flow rates for 2006 test.
System cleaned
on 9/30
- Feed Pressure -•- Retentate Pressure (psig)
0
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
59
-------
Figure 4-3. Plot of UF system feed and retentate pressures for 2006 test.
System Cleaned
on 9/30
0
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06
Date
Figure 4-4. Plot of UF system TMP for 2006 test.
10/25/06
60
-------
5.00
0.00
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-5. UF system specific flux calculations for 2006 test.
09/25/06
09/30/06
10/05/06
10/10/06
Date
10/15/06
10/20/06
10/25/06
Figure 4-6. Loss of specific flux over time for 2006 test.
61
-------
The power use for the UF system was monitored by a power meter that was separate from
that for the RO high pressure pump. The UF power meter does include the power use by
ancillary equipment on the RO skid. This meter provided power data for the UF system.
The operators recorded power readings twice daily. The power data was then combined
with the hours of UF system operation to calculate the power used per hour of operation.
The mean power consumption was 39 kiloWatt-hours (kWh) per hour of operation with a
median value of 39 kWh per hour of operation. Figure 4-7 shows the power consumption
per hour of operation during the test. The spike in power use on October 6 and 7 occurred
at the same time that the RO system was being cleaned and the RO power use dropped. It
is not known why this occurred, but the UF power meter also includes the ancillary
systems on the RO skid.
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-7. UF Power consumption per hour of operation for 2006 test.
4.1.3.1.2 RO System
The RO operational statistics for the 2006 test are presented in Table 4-3. The RO system
has flow meters and pressure gauges to monitor the feed water, concentrate and permeate
for Array 1. However, during the test the concentrate flow meter was not functioning
properly. Therefore, the concentrate flow rate reported in Table 4-3 is calculated as the
62
-------
difference between the feed water flow rate and the permeate flow rate. Array 2 has flow
meters for the permeate and concentrate, and gauges to monitor pressure for the feed
water, permeate, and concentrate. The feed flow rate for Array 2 was calculated by
adding the permeate and concentrate flow rates. The UF system supplied all of the feed
water for the RO system.
Table 4-3. RO System Operational Measurement Statistics for 2006 Test
Standard
Parameter
Array 1 Feed Flow (gpm)
Array 1 Permeate Flow (gpm)
Array 1 Concentrate Flow (gpm)
Array 2 Feed Flow (gpm)
Array 2 Permeate Flow (gpm)
Array 2 Concentrate Flow (gpm)
Array 1 Feed Pressure (psig)
Array 1 Concentrate Pressure (psig)
Array 2 Feed Pressure (psig)
Array 2 Concentrate Pressure (psig)
Array 1 and 2 Combined Permeate
Pressure (psig)
Count
59
59
59
59
59
59
59
59
59
59
59
Mean
107
53
54
53
21
32
444
346
345
255
28
Median
107
53
54
52
21
31
428
330
327
238
27
Minimum
104
44
48
49
19
27
374
286
284
204
15
Maximum
110
56
62
59
24
37
539
419
436
325
39
Deviation
1.38
2.0
2.4
2.3
1.1
2.3
45.9
40.5
42.5
35.2
4.6
95% C I
±0.35
±0.50
±0.61
±0.60
±0.27
±0.58
±11.7
±10.3
±10.8
±8.98
±1.2
The RO system operated continuously during the verification test, except when alarms
shut the unit down during unattended operation over night, or when maintenance was
required on the system. During the first three days of testing the RO unit was shutdown
overnight due to a miscommunication with the operators. The RO system operated
greater than 20 h on 21 of the 31 test days (Day 0 through Day 30), and greater than 10 h
on 27 days of the 31 test days. The mean operating hours were 19.4 h per day with a
median of 22 h per day. The maximum operating hours were 24 h and the minimum was
6 h. The 95% confidence interval shows expected operating hours of 17-21 h per day. If
the first three days of operation are removed from the calculations, the 95% confidence
interval is 18-22 h of operation per day. These operating hours give a utilization rate of
75 to 92%.
63
-------
The mean RO permeate flows of 53 gpm for Array 1 and 21 gpm for Array 2 yield a
mean total permeate production of 74 gpm, which is below the design permeate rate of
112.5 gpm (162,000 gpd) specified in Table 2-4 for low TDS waters. The mean feed
water flow of 107 gpm for Array 1 and 53 gpm for Array 2 were below the target feed
rates specified in Table 3-2. However, the actual recovery for Array 1 was 49.5%, which
is close to the design target of 50%. The recovery for Array 2 was 39.6%, which is lower
than the target of 48% specified for the unit. Over the 30-day verification test, accounting
for downtime for maintenance, alarms, and other shutdowns, the RO feed water totalizer
showed that 5,382,670 gal of water was fed to the RO unit. At an average recovery of
47% (prorated between Array 1 at 49.5% and Array 2 at 39.6%), the total volume of
permeate produced was approximately 2,530,000 gal or an average of 84,330 gpd over
the entire test period. This falls short of the goal of demonstrating production of 100,000
gpd of finished water.
The RO system maintained a steady permeate flow rate for both arrays throughout the
verification test. Figure 4-8 shows the daily flow rates for feed water, permeate, and
concentrate for both arrays. Figure 4-9 shows the feed water and concentrate pressures
for both arrays. Feed water pressure was increased over the duration of the test in order to
maintain feed water flow rates. The concentrate pressure from Array 1 was used by the
energy recovery device to increase feed water pressure for Array 2. These pressures were
similar throughout the test. The Array 1 concentrate pressure had a mean value of 346 psi
with a median value of 330 psi. The Array 2 feed pressure had a mean value of 345 psi
with a median value of 327 psi. The 95% confidence interval for the Array 1 concentrate
pressure was 336 to 357 psi, and the 95 % confidence interval for the Array 2 feed water
pressure was 335 to 356 psi. Based on the small pressure loss from the transfer of
pressure between the Array 1 concentrate and the Array 2 feed water, the energy recovery
device worked properly during the test, and in an efficient manner.
Figure 4-10 shows the percent recoveries achieved by the RO system. Recoveries, as
measured by the flow rate of the permeate divided by the feed water flow rate were
consistent throughout the test. The recoveries for Array 2 were lower than for Array 1.
64
-------
/06 9/30/06
—•—Array 1 Feed Flow
-»— Array 2 Feed Flow
10/5/06 10/10/06 10/15
Date
-•-Array 1 Permeate Flow
-H- Array 2 Permeate Flow
06 10/20/06 10/25/Of
-*- Array 1 Concentrate Flow
-*- Array 2 Concentrate Flow
Figure 4-8. RO system flow rates for 2006 test.
600
500
Array 2 Feed Pressure
Array 2 Concentrate Pressure
Array 1 Feed Pressure
Array 1 Concentrate Pressure
0
9/25/06
9/30/06 10/5/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-9. RO system operating pressures for 2006 test.
65
-------
60%
50%
40%
30%
20%
10%
Array 1 Recovery — •— Array 2 Recovery
09/25/06
09/30/06
10/05/06
10/10/06
Date
10/15/06
10/20/06
10/25/06
Figure 4-10. RO percent recoveries for 2006 test.
A common method of evaluating RO membrane performance is to calculate the specific
flux, which adjusts the permeate flux based on NDP. The calculation of NDP that was
used in the determination of specific flux included the calculation of osmotic pressure. A
correlation between TDS and conductivity was calculated based on the weekly TDS data.
This correlation was then used with the daily conductivity data to estimate TDS on a
daily basis and calculate osmotic pressure. The equation for the line determined for this
correlation is y(TDS) = 0.6014x(conductivity).
Figure 4-11 shows the specific flux for the two RO system arrays based on NDP and
adjusted to a temperature of 25 °C. The trend shown by these data clearly indicate that the
RO membranes were slowly being fouled as would be expected. The specific flux
dropped by approximately 31% for Array 1 and 26% for Array 2 over the 30-day test.
While the membranes were still functioning at the end of the test, it could be projected
that the membranes would have required cleaning sometime in the next 30 to 60 days,
based on the trend in the specific flux and the corresponding trend in increasing feed
pressures (to maintain flow rate).
66
-------
0.06
0.05
s« 0.02
0.01 -
0.00
- Specific Flux Array 1 -•- Specific Flux Array 2
9/25/06 9/30/06 10/5/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-11. RO system specific flux for 2006 test.
The RO system was chemical cleaned on October 6 using a citric acid low pH solution.
Two days earlier the RO system had been shutdown by the PLC due to a high pressure
differential across the system. The next day the RO system was flushed for two hours
with fresh water and then placed back into operation. The test plan requires that the RO
membranes be cleaned at least once during, or at the end of the test period. Therefore, it
was decided to clean the RO, even though the pressure differential appeared acceptable
and the specific flux was only slightly lower than on the first day of operation (0.046
gfd/psi versus 0.050 gfd/psi). The specific flux just before the start of the cleaning was
0.0429 gfd/psi. The cleaning did increase the specific flux to 0.047 gfd/psi on the
afternoon of the next day October 7.
Given the slow but steady trend of decreasing specific flux, see Figure 4-11, an anti-
sealant was fed to the RO system beginning on October 12. Anti-sealants can help reduce
surface buildup on RO membranes and slow the loss of flux across the membranes and
the need for higher pressures to maintain permeate flow. The ONDEO (Nalco)
PermaTreat® PC-191 anti-sealant was fed at a rate of approximately 7.5 mL/min to
achieve an anti-sealant concentration in the RO feed water of 4 mg/L. This chemical feed
continued through the end of the verification test.
The RO system power consumption was monitored twice daily during the verification
test. The RO system had a separate power meter that was read by the operators and
67
-------
recorded in the logbook. The power consumption in kWh per hour of RO operation was
calculated by dividing the power use for each time period by the hours of RO operation,
as monitored by the high pressure pump operating meter. The mean power consumption
was 26 kWh per hour of operation with minimum of 15 kWh per hour of operation and a
maximum of 31 kWh per hour of operation. The median power use was 26 kWh per hour
of RO operation. Figure 4-12 shows the power use over the duration of the verification
test.
10/15/06 10/20/06
09/25/06 09/30/06 10/05/06 10/10/06
Date
Figure 4-12. RO power use per hour of operation for 2006 test.
10/25/06
4.1.3.2 Task C2: Cleaning Efficiency
An important aspect of membrane operation is the ability to achieve long run times
between chemical cleanings (maintain up time and minimize chemical use) and to restore
membrane production after flux decline due to buildup of solids on the membrane and in
the membrane pores. The objective of this task was to evaluate the membrane cleaning
procedures and determine the fraction of specific flux restored following chemical
cleaning.
4.1.3.2.1 UF Backwash and Cleaning Frequency and Performance
The UF system is designed to be backwashed automatically after every 30 min of
operation. The backwash is designed to remove solids that have accumulated on and
within the membrane. Frequent effective backwashes provide restoration of water
68
-------
production and lengthen the time until chemical cleaning is required. The automatic
backwash system reverses the flow through the membrane to remove material
accumulated on the membrane surface, and then a fast forward flow flush is performed to
clear the membrane. The system uses UF filtrate water for the backwash cycle.
The automatic backwash system functioned properly during the verification test. The
automatic cycle initiated on schedule once every 30-min, as programmed, and the entire
process was automated. The backwash cycle counter tracked the number of backwashes
performed during the test. The backwash system used 900 gal of filtrate for each
backwash cycle. Based on the number of backwashes performed and the flow rates
achieved in the verification test, the backwash system used approximately 13-14% of the
filtrate produced by the UF system.
Based on vendor experience it was expected that the UF system would require chemical
cleaning about every 30 days. However, at the start of the verification test, the solids
buildup occurred much quicker and the first CIP was performed on September 30, on the
6th day of operation. At the time the unit was brought off line for cleaning, the TMP had
increased to 25 psig from 11 psig at the start of the test. The specific flux had decreased
to 1.38 gfd/psig from the starting value of 3.56 gfd/psig. The CIP was successful as the
specific flux was 3.52 gfd/psig after the cleaning (99% recovery of specific flux) and the
specific flux actually was somewhat higher after the cleaning than at the beginning of the
test (see Figure 4-5).
The UF system ran for an additional 25 days and the automatic backwash system, in
conjunction with the addition of ferric chloride as a coagulant (dose of 0.4 mg/L as Fe),
was able to maintain the system at TMP below 20 psig and a specific flux in the 2 to 3
gfd/psig range. The UF was not chemically cleaned again during the test, but would
probably have needed a CIP within the 30 days estimated in the system specification.
Figure 4-6 shows the loss (or gain) of specific flux over the duration of the verification
test. The change in specific flux was calculated by comparing the specific flux on a given
day to the value calculated at the start of the test. This type of data shows the impact of
cleaning and backwash by comparing any given day's specific flux to the start of the test.
As can be seen, there was a steady loss of specific flux at the beginning of the test, but
the CIP on September 30 was successful and actually resulted in the UF system having a
higher specific flux after the cleaning as compared to the start of the test.
The UF CIP procedure (Section 2.4.1.2) uses three chemicals, citric acid for the low pH
cleaning and sodium hydroxide and calcium hypochlorite for the high pH cleaning. The
amount of citric acid and sodium hydroxide needed to make a pH 3 or pH 11 cleaning
solution will vary based on the water used for the making the cleaning solution and the
concentration of acid or base used. For this test, 8 cups (64 ounces dry) of citric acid was
added to the 300 gal of water in the CIP tank. At a density of 1.665 grams/mL, 8 cups is
approximately 3144 grams or 6.9 Ibs. This gave a cleaning solution pH of 2.98 to 3.08.
For the high pH solution, 1.1 L of 0.5% by weight sodium hydroxide was added to the
69
-------
300 gal of water in the CIP tank. This resulted in a cleaning solution pH of 11.00-11.63.
In addition 300 gram of calcium hypochlorite was added to the high pH solution.
The UF cleaning solution was heated in the CIP tank with the low pH solution ranging
from 35 to 39 °C and the high pH solution 32 to 37 °C. Each bank of modules was
cleaned with each solution for 20 to 30 min.
4.1.3.2.2 RO Cleaning Frequency and Performance
The RO system was cleaned on October 6 using an acid solution. It is not clear that the
RO required chemical cleaning at this point in the verification test. An increasing
pressure differential and decreasing specific flux were the basis for the cleaning. There
was only a slight increase in the specific flux after the cleaning. Figure 4-11 showed the
specific flux for the two RO system arrays based on NDP and adjusted to a temperature
of 25 °C. The trend shown by these data indicate that the RO membranes were slowly
being fouled after the October 6, 2006 CIP, as would be expected. The specific flux
dropped by approximately 31% for Array 1 and 26% for Array 2 over the 30-day test.
While the membranes were still functioning at the end of the test, they probably would
have required a chemical cleaning sometime in the next 30 to 60 days based on the trend
in the specific flux and the corresponding trend in increasing feed pressures (to maintain
flow rate).
The RO cleaning was performed with an acid solution. Citric acid was added to the 300-
gallon CIP tank to achieve a pH in the range of 3.75 to 3.96. The specific amount of citric
acid added was not recorded, but based on the UF CIP data, it can be estimated that
approximately 4 to 6 Ibs of citric acid was used to reach this pH. The system was
circulated for approximately two hours and then allowed to soak overnight. The cleaning
solution was circulated again for two hours and then the RO system was put back into
normal operation.
4.1.3.2.3 Total Organic Carbon Results for UF Cleaning Solutions
Samples of the cleaning solution for the UF system CIP were collected from one cleaning
period. These samples were analyzed for TOC as specified in the ETV Protocol and the
Test Plan. The TOC results for the September 30, 2006 UF system cleaning solutions are
presented in Table 4-4. The TOC was higher in the low pH solution. The used cleaning
solution was acceptable for discharge to the sanitary sewer system at Selfridge ANB and
was discharged to the sewer system after each cleaning cycle.
Table 4-4. UF Cleaning Solution TOC Results
Sample Date TOC (mg/L)
Low pH solution 9/30/06 400
High pH solution 9/30/06 64
70
-------
4.1.3.3 Task C3: 2006 Water Quality Results
The primary objective of this task was to assess the ability of the membrane equipment to
meet the water quality goals, which were established as producing water that meets EPA
National Drinking Water Regulations. Several water quality parameters were selected as
indicator parameters to demonstrate the performance of the UF and RO membranes.
Turbidity and conductivity were selected as two key parameters, as turbidity removal by
the system would indicate the ability to remove particulate related contaminants, and a
reduction in conductivity (indicator of TDS content) would show the ability of the RO
system to remove dissolved contaminants. Both turbidity and conductivity were
measured with in-line meters in the EUWP, and were also measured with portable
equipment on site, at least twice per day. In addition, pH and temperature were measured
at least twice per day. Other water quality parameters were monitored by collecting
samples on a weekly basis. These parameters included TOC, TDS, TSS, Alkalinity,
Hardness, Silica, and UV254 absorbance. This section presents the water quality results
for the 2006 verification test of both the UF and RO systems. Data on the bacteriological
samples and integrity testing are presented later in a separate section of this report.
Figures 4-13 and 4-14 present the grab sample turbidity readings for the UF feed and UF
filtrate over the duration of the test. Table 4-5 lists all of the grab sample turbidity
readings and the summary statistics for the verification test. As can be seen, the UF
system reduced the turbidity from a mean of 4.77 NTU in the feed water to a mean of
0.14 NTU in the UF filtrate. The 95% confidence level for the grab sample turbidity
readings shows that filtrate turbidity can be expected to be in the range of 0.12 to 0.16
NTU. The UF system reduced the turbidity of the feed water by a mean value of 95.9%,
with a median reduction of 96.4%
All filtrate turbidity measurements were below the NPDWR of 1 NTU. The second
NPDWR criteria for turbidity is that 95% of the daily samples in any month must be <0.3
NTU. Only one filtrate turbidity measurement out of 58 was above 0.3 NTU: 0.47 NTU
on October 5. Therefore, the EUWP UF system met the second NPDWR turbidity
requirement, as 98% of the turbidity measurements were <0.3 NTU.
The feed water turbidity also spiked on October 5, up to 40.3 NTU. This was likely due
to a rain event that occurred at the test site on October 3 and 4. The Lake St. Clair inlet
from which the raw water was drawn has a stormwater runoff outfall, so discharge of the
runoff into the inlet could have caused the spike in turbidity.
As discussed in Sections 2.4 and 3.9.4.2, the EUWP also includes in-line turbidity meters
which measure the turbidity every 15 min as a means of monitoring membrane integrity.
There were numerous instances where two or more consecutive turbidity measurements
were above the LT2ESWTR action level of 0.15 NTU for shutting the system down and
performing a direct integrity test. However, as discussed in Section 2.4, this did not occur
because the EUWP is not compliant with the LT2ESWTR indirect integrity monitoring
requirements. The in-line turbidity data was logged onto a laptop computer, but the
71
-------
computer was not connected to the EUWP for the purpose of shutting down the system if
necessary.
Also, the in-line turbidimeters were not shut off when the UF system was down for
cleaning or other maintenance activities. This could give false high readings, as was the
case when the system was shut down for a chemical cleaning. A second issue with the in-
line turbidity data is that the date and time of the data logger were not reset prior to the
start of the verification test. Therefore, it is not possible to correlate the turbidity data
with the dates and times that the system was shut down for cleaning or maintenance.
Because of incorrect date and time stamp for the turbidity readings, the data is not
presented in this report.
The RO system had an additional impact on the turbidity levels with the RO permeate
grab samples having a mean turbidity of 0.09 NTU. The maximum measured RO
permeate turbidity was 0.18 NTU. This represents a further reduction in the range of 40%
to 66% through the RO system.
09/25/06
09/30/06
10/05/06
10/10/06
Date
10/15/06
10/20/06
10/25/06
Figure 4-13. UF feed water turbidity for 2006 test.
72
-------
0.50
0.45
9/25/06
9/30/06
10/5/06
10/10/06
Date
10/15/06
10/20/06
10/25/06
Figure 4-14. UF filtrate water turbidity for 2006 test.
Table 4-5. Turbidity Results for 2006 Test - Hand-Held Meter
Date
9/25/2006
9/25/2006
9/26/2006
9/26/2006
9/27/2006
9/27/2006
9/28/2006
9/28/2006
9/29/2006
9/29/2006
9/30/2006
10/1/2006
10/1/2006
10/2/2006
10/2/2006
10/3/2006
10/3/2006
10/4/2006
10/4/2006
10/5/2006
10/5/2006
UF Feed
2.13
2.12
2.31
2.12
1.73
2.54
1.24
3.65
2.67
1.64
1.76
1.97
1.99
2.47
3.43
2.76
2.17
2.93
8.25
40.3
20.0
Turbidity (NTU)
RO 1st Pass
UF Filtrate RO Feed Permeate
0.18
0.11
0.16
0.14
0.09
0.17
0.09
0.14
0.02
0.10
0.28
0.03
0.11
0.09
0.11
0.15
0.13
0.17
0.11
0.47
0.30
0.16
0.13
0.10
0.13
0.11
0.17
0.10
0.11
0.05
0.08
0.54
0.11
0.15
0.14
0.12
0.16
0.11
0.11
0.14
0.41
0.19
0.08
0.13
0.08
0.13
0.09
0.10
0.07
0.17
0.07
0.06
0.07
0.08
0.09
0.07
0.18
0.09
0.12
0.11
0.14
0.09
0.12
RO
Concentrate
0.22
0.47
0.20
0.77
0.30
0.09
0.16
0.29
0.14
0.12
0.58
0.10
0.12
0.24
0.25
0.12
0.12
0.15
0.14
0.17
0.16
UF%
Reduction
91.5
94.8
93.1
93.4
94.8
93.3
92.7
96.2
99.3
93.9
84.1
98.5
94.5
96.4
96.8
94.6
94.0
94.2
98.7
98.8
98.5
73
-------
Table 4-5. Turbidity Results for 2006 Test - Hand-Held Meter (continued)
Turbidity (NTU)
Date
10/6/2006
10/7/2006
10/7/2006
10/8/2006
10/8/2006
10/9/2006
10/9/2006
10/10/2006
10/11/2006
10/11/2006
10/12/2006
10/12/2006
10/13/2006
10/13/2006
10/14/2006
10/14/2006
10/15/2006
10/15/2006
10/16/2006
10/16/2006
10/17/2006
10/17/2006
10/18/2006
10/18/2006
10/19/2006
10/19/2006
10/20/2006
10/20/2006
10/21/2006
10/21/2006
10/22/2006
10/22/2006
10/23/2006
10/23/2006
10/24/2006
10/24/2006
10/25/2006
10/25/2006
Mean:
Median:
Minimum:
Maximum:
Std. Dev.:
95% CI:
UF Feed
9.08
5.65
5.06
6.18
4.60
3.16
5.85
4.38
4.94
5.31
5.86
4.35
3.5
2.75
2.26
2.16
1.93
2.09
2.19
1.63
14.9
11.9
4.13
4.39
3.39
2.94
3.16
2.45
2.21
2.37
5.43
3.69
3.92
4.09
3.33
8.35
3.53
4.35
4.77
3.33
1.24
40.3
5.72
1.46
UF Filtrate
0.10
0.12
0.12
0.11
0.15
0.10
0.11
0.14
0.10
0.11
0.13
0.14
0.13
0.11
0.12
0.10
0.01
0.20
0.09
0.19
0.14
0.24
0.12
0.14
0.08
0.13
0.11
0.1
0.09
NM(1)
0.13
0.13
0.1
0.16
0.12
0.07
0.26
0.19
0.14
0.12
0.01
0.47
0.07
0.02
RO Feed
0.15
0.13
0.10
0.13
0.11
0.15
0.13
0.13
0.11
0.15
0.10
0.16
0.12
0.11
0.12
0.11
0.13
0.14
0.09
0.11
0.25
0.30
0.09
0.24
0.14
0.08
0.12
0.09
0.11
0.12
0.09
0.16
0.12
0.13
0.12
0.63
0.16
0.17
0.15
0.13
0.05
0.63
0.10
0.03
RO 1st Pass
Permeate
0.10
0.09
0.10
0.09
0.12
0.09
0.08
0.14
0.10
0.08
0.09
0.08
0.13
0.10
0.08
0.01
0.08
0.10
0.08
0.13
0.10
0.06
0.10
0.08
0.06
0.09
0.08
0.04
0.07
0.09
0.02
0.1
0.1
0.05
0.08
0.07
0.12
0.09
0.09
0.09
0.01
0.18
0.03
0.008
RO
Concentrate
0.09
0.56
0.29
0.10
0.38
0.15
0.09
0.34
0.13
0.14
0.13
0.37
0.16
0.26
0.13
0.14
0.20
0.22
0.1
0.31
0.22
0.54
0.54
0.21
0.32
0.28
0.14
0.13
0.38
0.61
0.05
0.33
0.23
0.12
0.14
0.70
0.14
0.3
0.25
0.2
0.05
0.77
0.16
0.04
UF%
Reduction
98.9
97.9
97.6
98.2
96.7
96.8
98.1
96.8
98.0
97.9
97.8
96.8
96.3
96.0
94.7
95.4
99.5
90.4
95.9
88.3
99.1
98.0
97.1
96.8
97.6
95.6
96.5
95.9
95.9
—
97.6
96.5
97.4
96.1
96.4
99.2
92.6
95.6
95.9
96.4
84.1
99.5
2.77
0.71
(1) not measured
74
-------
Figure 4-15 presents the conductivity data for the RO system over the duration of the test.
Table 4-6 shows the conductivity results for the UF and RO systems and the summary
statistics for the verification test. The RO system reduced the dissolved ions in the water,
as measured by conductivity. The mean conductivity in the RO permeate was 1.8
microSiemens per centimeter (|iS/cm) compared to the mean conductivity in the RO feed
water of 287 jiS/cm. The median feed water conductivity was 289 jiS/cm and the median
RO permeate conductivity was 1.6 |iS/cm. The RO unit reduced the conductivity
(indicator of dissolved salts rejection) by a mean value of 99.4 %. The direct
measurement of TDS, data shown later in Table 4-9, shows that the total dissolved solids
concentration in the RO permeate was always below the detection limit of 5 mg/L.
20
18
16
RO Feed at Strainer (uS/cm)
RO Concentrate (uS/cm)
RO 1st Pass Permeate (uS/cm)
r.
-a
tu
/-v 01
s «
^ i
cc e
e s
^§
73
e
o
U
09/25/06 09/30/06 10/05/06 10/10/06 10/15/06 10/20/06 10/25/06
Date
Figure 4-15. RO conductivity results for 2006 test.
75
-------
Table 4-6. Conductivity Results for 2006 Test for In-Line Meter
Date
9/25/2006
9/25/2006
9/26/2006
9/26/2006
9/27/2006
9/27/2006
9/28/2006
9/28/2006
9/29/2006
9/29/2006
9/30/2006
10/1/2006
10/1/2006
10/2/2006
10/2/2006
10/3/2006
10/3/2006
10/4/2006
10/4/2006
10/5/2006
10/5/2006
10/6/2006
10/7/2006
10/7/2006
10/8/2006
10/8/2006
10/9/2006
10/9/2006
10/10/2006
10/11/2006
10/11/2006
10/12/2006
10/12/2006
10/13/2006
10/13/2006
10/14/2006
10/14/2006
10/15/2006
10/15/2006
10/16/2006
10/16/2006
10/17/2006
10/17/2006
10/18/2006
10/18/2006
10/19/2006
10/19/2006
10/20/2006
10/20/2006
10/21/2006
UF Feed
252
271
262
258
253
284
270
260
289
311
265
334
291
288
310
316
304
305
306
275
265
282
238
266
250
278
277
296
306
290
295
271
275
295
277
273
264
251
305
268
353
294
293
290
283
297
297
322
324
299
Conductivity (uS/cm)
RO 1st Pass
UF Filtrate RO Feed Permeate
248
262
267
258
253
294
269
260
282
271
264
335
291
288
289
313
301
305
306
274
265
281
237
265
250
271
277
297
306
290
296
271
274
294
279
272
265
251
298
268
273
294
293
291
284
296
296
321
323
298
247 1.6
262 1.6
263 2.0
258 1.4
252 1.3
281 4.2
270 1.3
261 1.2
282 1.5
271 1.2
184 1.2
339 2.1
291 3.4
288
289
314
302
305
306
275 L
265
279
236 :
265
250
267
277
296
.4
.4
.7
.3
.5
.5
\.l
A
.8
..0
.6
.5
o
.J
.6
.7
308 4.9
289 2.0
296 1.6
271 1.7
275 1.4
295 1.6
279 1.4
272 2.0
262 1.5
250 1.6
298 1.8
268 1.4
273 1.3
294 1.6
291 1.6
290 1.8
285 1.3
293 1.9
298 1.6
319 1.7
316 1.4
298 1.5
RO
Concentrate
804
859
846
839
815
920
857
826
874
538
368
672
604
610
601
654
631
586
588
518
495
554
452
511
481
516
535
571
591
553
574
508
536
567
535
518
501
477
550
502
529
570
554
567
559
563
587
624
620
594
RO%
Rejection
99.3
99.4
99.2
99.5
99.5
98.5
99.5
99.6
99.5
99.6
99.3
99.4
98.8
99.5
99.5
99.5
99.6
99.5
99.5
98.3
99.5
99.3
99.1
99.4
99.4
99.5
99.4
99.4
98.4
99.3
99.5
99.4
99.5
99.5
99.5
99.3
99.4
99.4
99.4
99.5
99.5
99.5
99.5
99.4
99.6
99.4
99.5
99.5
99.6
99.5
76
-------
Table 4-6. Conductivity Results for 2006 Test for In-Line Meter (continued)
Conductivity (jig/LS/cm)
Date
10/21/2006
10/22/2006
10/22/2006
10/23/2006
10/23/2006
10/24/2006
10/24/2006
10/25/2006
10/25/2006
Mean:
Median:
Minimum:
Maximum:
Std. Dev.:
95% CI:
UF Feed
335
316
338
298
320
330
305
353
316
291
291
238
353
26
6.7
UF Filtrate
335
313
338
298
318
330
307
350
315
288
290
237
350
25
6.4
RO Feed
335
316
333
300
318
330
308
353
320
287
289
184
353
29
7.3
RO 1st Pass
Permeate
1.9
2.1
2.3
2.1
2.1
2.3
1.4
2.0
2.7
1.8
1.6
1.2
4.9
0.8
0.2
RO
Concentrate
658
625
661
586
627
640
581
666
608
609
586
368
920
117
29.9
RO%
Rejection
99.4
99.4
99.3
99.3
99.3
99.3
99.5
99.4
99.2
99.3
99.4
98.3
99.6
0.25
0.065
Tables 4-7 and 4-8 present the pH and temperature data collected from the UF and RO
systems. pH was steady over the test period and the UF system had little impact on the
pH of the water. The RO system did lower the pH in the permeate. This is expected, as
the constituents that contribute to hardness and alkalinity (dissolved species) are rejected
by the membrane. The resultant permeate has less buffering capacity and will tend to
have a lower pH. As shown later in Table 4-10, hardness was reduced to <2 mg/L as
CaCOs and alkalinity was reduced to <5 mg/L as CaCOs. The pH of the permeate ranged
from 5.2 to 9.0.
The UF and RO systems had no effect on the temperature of the water as it passed
through. Water temperature in the lake feed water at the beginning of the test was in the
16 °C to 18 °C range and dropped during the test to 8 to 9 °C by the end of the test in late
October. This is typical in the northern climate. Temperature variation and impact on
membrane operating production (flux and specific flux) were accounted for in the
operating section by standardizing the data to either 20 °C or 25 °C, as described in
Section 3.9.1.3. The temperature data in Table 4-8 served as the basis for the temperature
adjustment calculations.
77
-------
Table 4-7. pH results for 2006 Test - In line Meter
Date
9/25/2006
9/25/2006
9/26/2006
9/26/2006
9/27/2006
9/27/2006
9/28/2006
9/28/2006
9/29/2006
9/29/2006
9/30/2006
10/1/2006
10/1/2006
10/2/2006
10/2/2006
10/3/2006
10/3/2006
10/4/2006
10/4/2006
10/5/2006
10/5/2006
10/6/2006
10/7/2006
10/7/2006
10/8/2006
10/8/2006
10/9/2006
10/9/2006
10/10/2006
10/11/2006
10/11/2006
10/12/2006
10/12/2006
10/13/2006
10/13/2006
10/14/2006
10/14/2006
10/15/2006
10/15/2006
10/16/2006
10/16/2006
10/17/2006
10/17/2006
10/18/2006
10/18/2006
10/19/2006
10/19/2006
10/20/2006
10/20/2006
10/21/2006
UF Feed
7.9
7.6
7.2
8.2
8.6
8.3
8.4
8.8
7.4
8.7
8.3
7.7
8.8
7.9
8.0
7.5
8.2
7.9
8.1
7.6
7.8
7.7
7.9
7.5
7.7
8.0
8.0
7.8
7.9
7.8
8.0
8.0
8.2
8.1
7.8
7.8
8.1
7.8
7.4
7.5
7.6
7.9
7.8
7.4
8.1
7.8
7.9
8.0
8.4
8.0
UF Filtrate
7.5
8.3
7.8
8.4
8.5
8.4
8.5
8.8
7.7
8.5
8.4
7.8
8.8
8.0
8.4
7.8
8.5
7.9
8.1
7.9
8.1
8.0
7.9
7.9
7.9
8.2
8.1
7.7
8.0
7.8
8.1
7.9
8.1
7.7
8.1
7.4
8.1
7.3
8.2
8.0
7.8
7.7
8.0
7.8
8.3
7.6
7.6
7.7
8.3
8.0
pH
RO Feed
8.1
8.4
8.0
8.5
8.4
8.5
8.5
8.8
8.0
8.6
8.5
7.4
8.6
7.8
8.1
7.5
7.8
7.5
7.9
7.6
8.0
7.7
7.7
7.8
7.8
8.1
7.7
8.0
7.8
7.8
8.1
7.5
8.1
7.8
7.9
7.9
8.2
7.9
8.1
8.0
8.2
8.0
8.0
7.9
8.3
8.0
7.9
8.0
8.3
8.0
RO 1st Pass
Permeate
6.3
8.0
6.2
8.2
7.6
5.9
7.8
8.3
6.6
8.0
9.0
6.4
5.7
6.9
8.3
5.5
8.0
8.1
7.3
6.9
8.4
7.5
5.8
6.3
6.6
8.0
6.2
7.0
7.8
6.6
8.3
5.9
5.4
7.4
8.7
7.6
5.4
7.0
6.3
6.2
8.1
6.4
6.6
5.7
5.7
6.7
6.8
6.5
7.5
6.8
RO Cone.
8.0
8.3
8.1
8.3
8.3
8.3
8.3
8.5
8.2
8.5
8.5
7.9
8.7
8.1
8.2
7.8
8.2
8.0
8.1
8.0
8.2
8.0
7.9
8.0
8.0
8.2
8.0
8.0
8.1
7.8
8.0
8.0
8.1
8.7
8.3
7.9
8.1
8.0
8.1
8.1
8.2
8.0
8.1
8.0
8.3
7.9
8.0
7.9
8.2
8.0
78
-------
Table 4-7. pH results for 2006 Test - In line Meter
Date
10/21/2006
10/22/2006
10/22/2006
10/23/2006
10/23/2006
10/24/2006
10/24/2006
10/25/2006
10/25/2006
Mean:
Median:
Minimum:
Maximum:
Std. Dev.:
95% CI:
UF Feed
7.6
7.7
7.9
7.6
7.6
7.3
8.2
7.4
7.3
7.9
7.9
7.2
8.8
0.36
0.09
Table 4-8. Temperature
pH
UF Filtrate RO Feed
7.8
7.6
7.8
7.5
7.3
7.6
8.1
7.7
7.2
8.0
7.9
7.2
8.8
0.36
0.09
Data for 2006
7.8
7.8
7.3
7.7
7.7
7.7
8.1
7.4
7.4
8.0
8.0
7.3
8.8
0.33
0.09
Test - In line
RO 1st Pass
Permeate
5.9
5.9
5.2
5.7
5.5
5.9
7.3
6.1
5.7
6.8
6.6
5.2
9.0
0.99
0.25
Meter
RO Cone.
7.8
7.9
7.9
7.7
7.7
7.7
8.3
7.8
7.5
8.1
8.0
7.5
8.7
0.23
0.06
Temperature (°C)
Date
9/25/2006
9/25/2006
9/26/2006
9/26/2006
9/27/2006
9/27/2006
9/28/2006
9/28/2006
9/29/2006
9/29/2006
9/30/2006
10/1/2006
10/1/2006
10/2/2006
10/2/2006
10/3/2006
10/3/2006
10/4/2006
10/4/2006
10/5/2006
10/5/2006
10/6/2006
10/7/2006
10/7/2006
10/8/2006
10/8/2006
UF Feed
16.4
17.7
16.4
17.3
17.1
18.2
15.4
17.1
15.2
15.9
14.5
16.4
16.0
15.3
16.4
16.1
17.4
17.0
16.6
13.6
14.5
14.4
15.7
14.7
13.5
15.2
UF Filtrate
16.3
17.6
16.3
17.1
17.0
18.2
15.7
16.6
15.2
15.7
14.7
16.5
15.9
15.1
16.0
15.8
17.1
16.9
16.5
13.5
14.1
14.2
15.1
14.4
13.3
15.2
RO Feed
16.1
17.5
16.0
16.9
16.6
17.7
15.9
16.5
14.9
15.8
15.1
15.5
16.0
15.3
15.9
15.9
17.1
16.9
16.8
13.5
14.3
14.7
14.4
14.4
13.4
14.9
RO Permeate
16.6
18.3
16.7
18.2
17.0
18.9
16.2
17.2
15.4
16.1
15.3
16.6
16.7
15.5
16.6
16.3
17.9
17.4
16.7
13.5
14.7
15.4
17.1
14.7
13.5
15.6
RO Cone.
17.2
18.5
17.0
18.0
17.6
18.7
17.0
17.7
15.9
16.5
15.9
16.0
16.6
16.0
16.6
16.7
17.7
17.3
17.4
14.1
14.7
14.6
14.7
14.9
14.0
15.4
79
-------
Table 4-8. Temperature Data for 2006 Test - In line Meter (continued)
Temperature (°C)
Date
10/9/2006
10/9/2006
10/10/2006
10/11/2006
10/11/2006
10/12/2006
10/12/2006
10/13/2006
10/13/2006
10/14/2006
10/14/2006
10/15/2006
10/15/2006
10/16/2006
10/16/2006
10/17/2006
10/17/2006
10/18/2006
10/18/2006
10/19/2006
10/19/2006
10/20/2006
10/20/2006
10/21/2006
10/21/2006
10/22/2006
10/22/2006
10/23/2006
10/23/2006
10/24/2006
10/24/2006
10/25/2006
10/25/2006
Mean:
Median:
Minimum:
Maximum:
Std. Dev.:
95% CI:
UF Feed
13.9
14.0
15.4
14.9
15.7
12.4
12.2
10.4
10.7
9.5
10.3
10.1
10.9
9.7
11.0
10.5
11.9
11.6
12.5
11.6
12.0
11.2
11.6
10.8
11.4
11.2
11.0
9.1
9.8
8.0
9.1
8.6
8.4
13.3
13.6
8.0
18.2
2.9
0.7
UF Filtrate
14.0
13.8
15.2
14.8
15.4
12.3
12.4
10.8
10.3
10.0
10.2
10.1
10.4
9.7
10.5
10.5
11.5
11.5
12.2
11.5
12.1
11.3
11.5
10.7
11.3
11.4
11.0
9.1
10.5
7.9
8.5
8.1
8.5
13.2
13.5
7.9
18.2
2.8
0.7
RO Feed
14.0
14.1
15.3
15.0
15.5
13.1
12.6
10.3
9.9
9.7
10.4
9.3
10.5
9.8
10.5
10.7
11.7
11.4
12.0
11.7
11.9
11.2
11.7
11.0
11.5
11.2
11.2
9.5
9.6
7.9
8.5
8.2
8.5
13.2
13.5
7.9
17.7
2.8
0.7
RO Permeate
14.1
14.1
15.4
15.6
16.1
12.5
11.8
10.1
10.3
9.9
10.9
10.3
10.8
10.2
10.9
11.0
12.6
11.9
12.6
11.8
12.1
11.3
11.8
11.0
11.8
11.4
10.8
9.3
9.9
8.1
8.4
8.1
8.7
13.6
13.5
8.1
18.9
3.0
0.8
RO Cone.
14.7
14.9
15.8
15.6
16.1
13.8
13.2
10.9
10.9
10.3
11.2
10.2
11.2
10.5
11.4
11.4
13.2
12.2
12.6
12.5
12.6
11.9
12.7
11.9
12.6
12.0
11.9
10.5
10.6
8.8
9.3
8.9
9.2
13.9
14.1
8.8
18.7
2.8
0.7
Tables 4-9 and 4-10 present the other water quality data collected on a weekly basis
during the verification test. The UF system removed suspended material as shown by the
TSS in the filtrate always being below the detection limit of 2 mg/L. These data support
the daily turbidity results showing over 95% reduction in turbidity. The UF system did
not change the other water quality parameters, as would be expected. These other
parameters, such as hardness, alkalinity, TDS, etc. primarily represent dissolved
inorganic constituents that are not removed by physical filtration.
80
-------
The RO system did remove many of the inorganic dissolved species as shown by the
results in Table 4-10 for the RO permeate. Hardness, alkalinity, TDS, and total silica
were all removed to below the detection limit in the permeate water. The RO concentrate
increased in concentration for these parameters above the feed water levels, as expected.
These data support the daily conductivity measurements that showed a significant
reduction in dissolved salts during the test. The RO membranes, at these operating
conditions, rejected the dissolved salts present in the feed water throughout the test.
The UF system showed only a minor reduction in organic material as measured by the
TOC data. The feed water, filtrate, and retentate all showed TOC concentrations within a
range of 2.1 to 2.7 mg/L. The filtrate typically showed a 0.1 to 0.4 mg/L reduction in
TOC compared to the feed water, and the retentate occasionally showed an increase in
TOC of 0.1 mg/L compared to the feed water. These data would indicate that most of the
organic material, as measured by TOC, was dissolved in the feed water. The RO system
had a major impact on the TOC levels, reducing the TOC in the permeate to below the
detection limit of 0.1 mg/L. As in the case of the dissolved salts, the RO membranes
rejected dissolved organic material, as measured by TOC, and reduced the TOC
concentration by greater than 95%.
81
-------
Table 4-9 Other UF System Water Quality Data for 2006 Test
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
UF Feed
<2
3
6
14
o
5
TSS (mg/L)
Filtrate Retentate
ND (2) <2
ND(2) NM
ND(2) 8
ND (2) 16
ND(2) 4
Backwash
5
23
23
320
50
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
TOC (mg/L)
UF Feed Filtrate
2.3 2.1
2.7 2.3
2.5 2.1
2.6 2.2
2.4 2.7
Retentate
2.4
2.7
2.6
2.6
2.4
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
Hardness
UF Feed
110
120
130
120
140
(mg/L as CaCO3)
Filtrate
110
120
120
120
140
Alkalinity (mg/L as CaCO3)
UF Feed Filtrate
86
93
97
94
100
90
92
96
95
100
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
TDS (mg/L)
UF Feed Filtrate
140 150
170 170
200 180
180 170
180 190
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
Total
UF Feed
0.45
0.2
1.1
3.1
2.1
Silica (mg/L)
Filtrate
0.2
0.2
0.7
1.0
1.3
UV254
UF Feed
0.0617
0.1063
0.1169
0.1542
0.0760
(absorbance/cm)
Filtrate
0.0319
0.0198
0.0332
0.0500
0.0512
82
-------
Table 4-10. Other RO System Water Quality Data for 2006 Test
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
RO Feed
<2
<2
<2
<2
<2
TSS (mg/L)
Permeate
<2
<2
<2
<2
<2
Concentrate
<5
<2
<2
<2
<2
Discharge
<2
<2
6
11
3
Date
TOC (mg/L)
RO Feed Permeate
Concentrate
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
2.1
2.3
2.1
2.2
2.4
0.1
7.6
5.2
4.6
4.2
4.5
Date
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
Hardness (mg/L as
RO Feed Permeate
110
120
120
120
140
<2
<2
<2
<2
<2
CaCO3)
Concentrate
410
280
260
250
260
Alkalinity (mg/L as
RO Feed Permeate
87
93
97
94
100
<5
<5
<5
<5
<5
CaCO3)
Concentrate
310
200
200
180
190
Date
TDS (mg/L)
Date
RO Feed
Permeate
Concentrate
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
150
170
180
180
180
<5
<5
<5
<5
<5
510
370
370
330
330
Total Silica (mg/L) UV254 (absorbance/cm)
RO Feed Permeate Concentrate RO Feed Permeate Concentrate
9/26/06
10/03/06
10/10/06
10/17/06
10/24/06
O.2
O.2
0.8
0.7
1.3
O.2
O.2
O.2
O.2
O.2
0.98
O.2
1.4
1.3
2.4
0.0320
0.0534
0.0377
0.0459
0.0432
O.OOOO
O.OOOO
O.OOOO
NM
O.OOOO
0.1462
0.0980
NM
0.0858
0.0833
4.1.3.4 Task C4: 2006Membrane Module Integrity
The objective of this task was to demonstrate methodology for direct integrity testing and
indirect integrity monitoring of the UF and RO membranes. Pressure decay tests were
used to document UF membrane integrity, and dye marker tests were used for the RO
membrane system.
83
-------
As discussed in Section 4, the initial UF pressure test on September 13, 2006 showed that
pressure was being lost at a higher than desirable rate. The problem was investigated, and
was found to be the o-ring seals between the membrane modules and filtrate collection
tubes. As a temporary fix, PTFE tape was wrapped around the o-rings to increase the seal
surface between the o-rings and membrane cartridges.
The pre-verification RO system dye reduction test was conducted on September 23. The
dye test showed that based on absorbance readings the membranes were rejecting the dye
as expected, and the dye was not leaking by any seals or through the membranes. RO
feed water absorbance was 2.860, and RO permeate was in the range of 0.013 to 0.015
over a 10-min run, which equates to a rejection of 99.5%. This dye test is considered a
direct measurement of RO membrane integrity. The in-line conductivity meters were also
monitored at the start of the test to confirm the rejection rate of the RO membranes.
4.1.3.4.1 UF System Pressure Decay andMicrobial Reduction Results
Pressure decay tests were performed each operating day during the verification test.
Table 4-11 presents the pressure decay data, and Figure 4-16 shows the pressure decay
data in a graphical format.
After the UF seal problem was temporarily fixed, a pressure decay test was conducted on
September 22. The data for this test is also presented in Table 4-11. The pressure decay
rate from this test was higher than desirable, but NSF and EPA allowed the test to
proceed. The pressure decay on the first day of the official test period was 1.14 psig/min,
which was almost four times higher than the 0.37 psig/min obtained on September 22.
After the test was completed, the technician found that the air hose was leaking, so these
initial data were not representative of the actual conditions. An air leak occurred again on
October 9, when the UF Array 1 retentate valve was not completely closed. Excluding the
pressure decay rates measured on September 25 and October 9, the pressure decay results
were fairly consistent with a mean value of 0.29 psig/min and a median value of 0.28
psig/min. The highest pressure decay rate measured was 0.43 psig/min.
84
-------
Table 4-11. Pressure Decay Data for the 2006 Test
Date
09/22/06
09/25/06
09/26/06
09/27/06
09/28/06
09/29/06
10/01/06
10/02/06
10/03/06
10/04/06
10/05/06
10/06/06
10/07/06
10/08/06
10/09/06
10/10/06
10/11/06
10/12/06
10/13/06
10/14/06
10/15/06
10/16/06
10/17/06
10/18/06
10/19/06
10/20/06
10/21/06
10/22/06
10/23/06
10/24/06
10/25/06
0
Min.
20.0
20.0
20.0
20.0
20.0
20.1
20.2
20.0
20.0
20.0
20.0
20.0
20.0
19.9
20.0
20.1
20.0
20.3
21.0
20.0
20.0
20.0
20.0
20.0
20.1
20.0
20.0
20.1
20.0
20.0
20.2
2
Min.
19
19
19
19
19
19
19
19
19
19
19
19
19
19
18
19
19
19
20
19
19
19
19
19
19
19
19
19
19
19
19
.5
.2
.6
.6
.7
.8
.9
.5
.3
.8
.7
.7
.7
.2
.0
.8
.8
.9
.4
.6
.7
.6
.7
.8
.7
.6
.6
.8
.6
.6
.8
4
Min.
19.0
18.5
19.4
19.2
19.3
19.3
19.5
19.0
19.0
19.5
19.3
19.4
19.4
18.9
16.2
19.5
19.5
19.5
19.8
19.2
19.3
19.2
19.3
19.4
19.3
19.2
19.3
19.6
19.2
19.2
19.3
6
Min.
18.5
17.8
19.1
18.9
19.0
19.0
19.2
18.5
18.8
19.2
19.0
19.4
19.2
18.5
14.7
19.2
19.3
19.1
19.0
18.8
19.0
18.9
19.0
19.2
19.0
18.8
19.0
19.4
18.7
18.7
18.8
8
Min.
18.2
17.3
18.8
18.6
18.6
18.6
18.8
18.0
18.6
19.0
18.7
18.9
18.9
18.3
13.3
18.9
19.0
18.7
18.5
18.4
18.7
18.5
18.8
18.9
18.6
18.4
18.6
19.2
18.3
18.4
18.4
10 12
Decay
Min. Min. (psig/min)
17
14
18
18
18
18
18
17
18
18
18
18
18
18
12
18
18
18
18
18
18
18
18
18
18
18
18
18
18
18
18
.7 17.
.7 12.
.5 18.
.4 18.
.4 18.
.3 18.
.5 18.
.6 17.
.5 18.
.8 18.
.4 18.
.6 18.
.7 18.
4
0
4
1
1
0
2
3
4
5
1
4
4
.0 17.7
.5 11.
.5 18.
4
o
J
.8 18.6
.4 18.
.2 18.
0
0
.1 17.7
.4 18.
.2 18.
.4 18.
.6 18.
.2 18.
.1 17.
.4 18.
.9 18.
.0 17.
.0 17.
.0 17.
Mean
Median
Maximum
Minimum
0
0
2
4
0
7
1
6
6
6
7
(i)
(i)
(i)
(i)
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.37
.14
.23
.27
.27
.30
.29
.39
.23
.21
.27
.23
.23
.31
.23
.26
.20
.33
.43
.33
.29
.29
.26
.23
.30
.33
.27
.21
.34
.34
.36
.29
.28
.43
.20
(1) The statistics do not include the September 22, September 25, and October 9 data. See above for further
discussion.
85
-------
22.0
20.0
—*—13-Sep-06
-A— 22-Sep-06
-*- 22-Sep-06
—I—26-Sep-06
-28-Sep-06
—D—l-Oct-06
3-Oct-06
-•— 5-Oct-06
7-Oct-06
—*— 9-Oct-06
-A—ll-Oct-06
-*-13-Oct-06
—I— 15-Oct-06
17-0ct-06
-B-19-Oct-06
-H-21-Oct-06
-»-23-Oct-06
25-Oct-06
• 22-Sep-06
- 22-Sep-06
•25-Sep-06
•27-Sep-06
•29-Sep-06
- 2-Oct-06
4-Oct-06
- 6-Oct-06
8-Oct-06
10-Oct-06
12-Oct-06
14-Oct-06
•16-Oct-06
•18-0ct-06
• 20-Oct-06
- 22-Oct-06
• 24-Oct-06
0246
Elapsed Time (min)
Figure 4-16. Pressure decay over time for the 2006 test.
10
12
86
-------
The pressure decay tests showed that the integrity of the membranes and seals remained
steady over the verification test. However, the mean value of 0.29 psig/min was on the
high side of the values expected for this system and higher than desired for best removal
and control of bacteria, virus, and other microbial agents. Previous ETV lab testing of the
Koch UF membranes had shown pressure decay rates in the range of 0.11 to 0.29
psig/min. The ETV report, Removal of Microbial Contaminants in Drinking Water Koch
Membrane Systems, Inc. HF-82-35-PMPW™ Ultrafiltration Membrane, September
2006, reported pressure decay test results from laboratory testing of the same UF fibers
used in the EUWP, but not the same end cap design. The report stated: "The pressure
decay rate for Cartridge 1 was measured to be 0.11 psig/min. The measured pressure
decay rates for Cartridge 2 were 0.14 and 0.29 psig/min. Koch Membrane Systems
provided an estimated severed fiber pressure decay rate of 2.1 psig/min for the HF-82-35-
PMPW membrane, so the measured decay rates for Cartridge 2 are not indicative of a
breach in membrane integrity. Also, the air bubble leak-check tests did not indicate that
any membrane fibers were compromised during testing."
The verification statement for the FIF-82-35-PMPW membrane reported: "The UF
cartridges were challenged with approximately 5 logic of the bacteriophage viruses fr and
MS2, 7 to 8 logic of the bacteria Brevundimonas diminuta, and 5.7 logic of live
Cryptosporidium parvum oocysts. The membranes removed a minimum of 4.8 logic of
the viruses, 6.0 logic of B. diminuta, and 5.7 logic of C. parvum." Therefore, it was
expected that the UF system would achieve significant log reductions of microbial agents
at the operating conditions found in the field test.
However, the HPC and total coliform data collected during the verification test did not
show significant reduction. The data, shown in Tables 4-12 and 4-13 indicate that total
coliform was reduced by at most 70% (0.7 logio) and HPC by at most 40% (0.4 logic) on
a mean basis in the UF filtrate. In fact HPC and total coliform were actually higher in the
filtrate compared to the feed water on occasions. The UF feed geometric mean HPC
count was 2810 CFU/mL, and the filtrate geometric mean HPC count was 1670 CFU/mL.
Mean total coliform counts were not calculated because only five sets of samples were
collected. The UF feed total coliform counts ranged from 41 to 532 CFU/100 mL, while
the filtrate counts ranged from 11 to 94 CFU/100 mL. High numbers of HPC and total
coliforms were also found in the RO permeate. The mean RO permeate HPC count was
247 CFU/mL and the RO permeate total coliform counts ranged from <1 to 95 CFU/100
mL. This phenomenon has been observed in other published membrane studies, but it
was beyond the scope of this study to determine whether the observed HPC and total
coliform levels were breaching the membrane, or were a result of microbial
contamination and growth downstream of the UF and RO membranes from previous field
tests of the EUWP.
87
-------
Table 4-12. HPC Results for the 2006 Test
HPC (CFU/mL)
Date
09/25/06
09/26/06
09/27/06
09/28/06
10/02/06
10/03/06
10/04/06
10/05/06
10/09/06
10/10/06
10/11/06
10/16/06
10/17/06
10/18/06
10/19/06
10/23/06
10/24/06
10/25/06
Geometric
Mean:
Maximum:
Minimum:
UF Feed
14,400
14,800
6,700
8,200
7,300
8,200
10,900
13,600
1,470
2,220
10,200
400
970
690
530
560
680
640
2,810
14,800
400
UF Filtrate
7,800
78,000
7,400
14,100
905
1,840
1,160
5,200
2,380
810
13,9000
700
300
330
260
180
70
260
1,670
139,000
70
UF
Retentate
8,900
1,230
7,700
283
1,250
1,460
1,150
4,700
520
3,200
210,000
270
1,170
1,190
280
200
440
590
1,400
210,000
200
UF
Backwash
17,500
15,600
23,700
8,900
5,300
10,300
8,200
8,500
9,500
12,000
218,000
900
9,100
7,100
980
1,630
3,100
430
6,670
218,000
430
RO Feed
15,600
9,500
9,500
6,900
3,500
1,430
1,400
8,700
3,100
1,920
189,000
420
500
730
500
330
210
630
2,250
189,000
210
RO Permeate
314
99,000
131
3,800
346
13
9
270
10
260
172,000
10
630
<10
<10
<10
54
30
247
172,000
9
RO
Concentrate
24,700
67,000
15,700
33,200
5,300
4,000
911
7,800
5,000
2,370
1,760
1,770
560
800
1,030
760
290
930
3,030
67,000
290
Table 4-13. Total Coliform Results for the 2006 Test
Date
UF Feed UF Filtrate
Total Coliform (CFU/lOOmL)
UF UF
Retentate Backwash RO Feed
RO
RO Permeate Concentrate
9/26/2006
10/3/2006
10/10/2006
10/17/2006
10/24/2006
67
154
173
41
532
75
55
94
30
11
78
3
9
7
540
240
<1
16
7
724
93
29
71
68
13
95
<1
35
55
<1
247
66
97
27
70
4.1.3.4.2 Particle Count Data
After completion of the 2006 ETV test, it was discovered that the particle counters had
been improperly calibrated, so the particle count data from this test is not presented here.
4.1.3.4.3 RO System Dye Test Results
Direct integrity measurements of the RO system were performed prior to the start of the
verification test and again at the end of the test. The test method was the dye marker test
where a food grade dye is added to the RO feed water, and feed, permeate, and
88
-------
concentrate UV absorbance values are measured. The RO permeate samples should have
low absorbance if the membranes are in good condition and the seals are tight.
The dye test results are shown on Tables 4-14 and 4-15. As can be seen, the RO system
showed good integrity at the beginning and end of the verification test. For the September
23, 2006 dye test prior to the start of the verification test, the RO feed water absorbance
was 2.860, and the mean Array 1 RO permeate was 0.012 over a 10 min run, yielding a
rejection rate of 99.6%. The Array 2 RO permeate absorbance was 0.011, also yielding a
rejection rate of 99.6%. At the end of the verification test, the rejection was slightly better
with an Array 1 rejection of 99.8% and an Array 2 rejection of 99.9%. These data show
that the RO membrane and sealing system maintained integrity throughout the test.
Table 4-14. RO Dye Test Results - September 23,2006
Time (min)
0
1
2
3
4
5
6
7
8
9
10
Mean:
Median:
Maximum:
Minimum:
RO Feed
Water
(absorbance)
y
o 12
T3
-------
Table 4-15. RO Dye Test Results - October 24, 2006
Time (min)
0
1
2
3
4
5
6
7
8
9
10
Mean:
Median:
Maximum:
Minimum:
RO Feed
Water
(absorbance)
g
o li~
a u
to 'S
C^ '
y o
•?? -i
O
-------
become apparent over a longer time frame, as was observed during the month-long
course of testing.
Table 4-16. UF Membrane Integrity Indicators for October 2006
Date
10/4
10/4
10/5
10/5
10/6
Time(1)
10:00
17:00
09:00
15:40
14:15(2)
Average ± 95%
Confidence
Interval(3):
Feed
Turbidity
(NTU)
2.93
8.25
40.3
20.0
9.08
4.77 ± 1.46
Filtrate
Turbidity
(NTU)
0.17
0.11
0.47
0.30
0.10
0.14 ±0.02
Feed
HPC
(CFU/m
L)
NM
10,900
13,600
NM
NM
2,810
Filtrate
HPC
(CFU/mL)
NM
1,160
5,200
NM
NM
1,669
TMP
(psig)
10
11
11
12
12
14 ±1
Pressure
Decay
(psig/min)
0.21
0.27
0.23
0.29 ±0.02
(1) Time of turbidity and TMP readings as part of the twice per day on-site data collection.
(2) Operational and water quality measurements were only recorded once on October 6.
(3) Averages are for all data over the course of the test, not just the data presented here.
4.1.4 2006 Chemical Consumption
The verification test in 2006 was started without the use of a coagulant. Following the
fairly rapid decrease in UF specific flux, it was determined that the addition of ferric
chloride as a coagulant aid could improve the UF operation. Beginning on October 2,
ferric chloride was added to the intake water stream prior to the UF system. The ferric
chloride solution had a concentration of 12 % as Fe. Feed rate varied from 0 to 1.46 gpd.
The feed rate average was 4 mL/min or 0.063 gal of ferric chloride per operating hour.
This represents a coagulant dosage of approximately 0.4 mg/L as Fe in the intake water.
At this coagulant dose, the system would use 106 mg of Fe per 1000 gal of intake water
or approximately 0.23 Ibs of Fe per one million gal of intake water.
The RO system was designed to have a scale inhibitor added if needed. During the 2006
verification test it was determined after 19 days of operation that the addition of a scale
inhibitor might improve and/or lengthen RO run time before chemical cleaning was
needed. ONDEO (Nalco) PermaTreat® PC-191 anti-sealant was made by using 3.34 L of
product to make 15 L of feed solution. The feed solution was fed at a rate of
approximately 7.5 mL/min to achieve an anti-sealant concentration in the RO feed water
of 4 mg/L or 0.11 gal per operating hour. This dose rate translates to approximately 1 gal
of concentrated inhibitor per 36,000 gal of RO feed water.
The chemicals needed for the UF CIP were citric acid, sodium hydroxide, and calcium
hypochlorite. Citric acid was used to lower the pH of the cleaning solution for the low pH
cleaning cycle, and sodium hydroxide was used for the high pH cleaning cycle. The
91
-------
calcium hypochlorite provided chlorine to help kill any biological growth on the
membranes to help oxidize organic material.
The chemicals specified to be used for the RO chemical cleaning were citric acid and an
alkaline detergent. Only the acid cleaning was performed during the ETV test. Citric acid
was used to lower the pH of the cleaning solution for the low pH cleaning cycle, and the
alkaline detergent if it had been used would have been used for the high pH cleaning
cycle. The actual amount of acid or base needed to lower or raise the pH of the water
used for the CIP solution will depend on the local water chemistry. For this test, tap water
was used for the cleaning solution.
The first CIP for the UF system used 8 cups (64 ounces dry vol.) of citric acid added to
the 300 gal of water in the CIP tank. This is approximately 6.9 Ibs. of citric acid. This
gave a cleaning solution pH of 2.98 to 3.08. For the high pH solution, 1.1 L of 0.5%
sodium hydroxide was added to the 300 gal of water in the CIP tank. This resulted in a
cleaning solution pH of 11.00-11.63. In addition 300 grams of calcium hypochlorite was
added to the high pH solution.
For the RO cleaning, citric acid was added to the 300 gal CIP tank to achieve a pH in the
range of 3.75 to 3.96. The specific volume of citric acid was not recorded, but based on
the UF CIP data, it can be estimated that approximately 4 to 6 Ibs. of citric acid was used
in the 300 gal tank to reach this pH.
4.2 2007 EUWP Retest
4.2.1 Task A: Raw Water Characterization
Two sets of grab samples were collected in August 2006 to characterize the raw water
supply, and to determine if any regulated metals or VOCs were present and should be
included in the final sampling plan. The results of these analyses are presented in Table
4-1. Based on these results, no metals or VOCs were added to the sampling plan.
4.2.2 Task B: Equipment Install and Initial Test Runs
A retest of the UF system was scheduled for July and August 2007, due to the problems
with the seals and integrity of the UF system during the initial ETV test in 2006. The
EUWP unit was delivered to Selfridge ANGB in July 2007 and the unit was prepared for
the retest. The emphasis during this start-up period was on UF membrane integrity as
measured by pressure decay tests on individual membranes and on the system as a whole.
The unit plumbing, electrical hook-ups, and pumping of raw water to the UF feed tank
were completed in the same manner as for the first test. The retest was designed to test
the UF system only, so no monitoring was planned or performed for the RO system.
However, due to the system design, the RO feed water flow meter was needed to monitor
the UF filtrate production and the RO pump was needed to move the filtrate from the
intermediate holding tank. Therefore, the RO system was operated during the retest, but
only to obtain flow data for the UF filtrate and to discharge water from the skid.
92
-------
It was determined during the first test that ferric chloride coagulation was necessary to
keep the UF system running smoothly. Therefore, the ferric chloride feed system and
tanks were setup and prepared for operation. For the 2007 test, ferric chloride was used
right from the start of the retest.
On July 18, 2007, the integrity of each UF membrane was individually checked by the
measurement of pressure decay over a 10-min period. Thirteen of the sixteen membranes
showed pressure decay in the range of 0.01 psig/min to 0.1 psig/min. Two membranes
had high decay rates. Inspection of the membranes determined that there was one broken
fiber in each membrane. The fibers were plugged and the membranes retested with both
now showing pressure decays of less than 0.15 psig/min. These membranes were
determined to be useable. The third leaking membrane had eight broken fibers and an end
cap seal problem. Because of the large number of broken fibers, this membrane was
removed from the system and it was decided to run the test with 15 membranes instead of
the normal 16 membranes.
Full UF system pressure decay tests were performed on six days between July 21 and
July 27. The full system pressure decay ranged from 0.02 to 0.1 psi/min. These results
showed that the seal problems encountered in the 2006 test had been resolved and that
fiber plugging for the two membranes was successful.
The start-up operation particle count and turbidity data (not shown) demonstrated that the
UF system was functioning properly, so testing proceeded.
4.2.3 Task C: 2007 Verification Retest
The 2007 verification retest of the repaired UF system was started on July 30 and ended
on August 24, 2007. The 2007 retest was stopped short of 30 days because the intent of
the test as stated in the ETV test protocol - to operate until a membrane cleaning was
conducted - was met. Both the UF and RO skids were operated for the 2007 retest, but
ETV test data was only collected from the UF system.
The on-site operators collected operating data and on-site water quality samples twice per
day in accordance with the test plan schedule. The following sections present the 2007
retest operating data and water quality data.
4.2.3.1 Task Cl: Mem brane Flux and Operation
The purpose of this task was to evaluate system performance during operation. The
objectives of this task were to demonstrate the appropriate operational conditions for the
system, the feed water recovery achieved by the UF membrane, and the rate of flux
decline observed over the operation period.
Operational data were collected and on-site water quality measurements were made twice
per day throughout both test periods, except for days when the UF was being cleaned and
therefore not operating for a portion, or all, of the day. The RO system was not monitored
93
-------
during the retest. The data were summarized for presentation and discussion in this
section. The complete data set can be found in Appendix B.
The UF retest operational statistics are presented in Table 4-17. Feed water flow rate and
retentate flow rate were measured directly by flow meters. The filtrate flow rate was
calculated as the UF feed water flow rate minus the UF retentate flow rate. The intake
flow was the intake from the source water into the UF feed water tank. The intake pump
ran at a higher flow rate than the UF system to ensure that the UF feed water tank always
contained sufficient water to operate the UF system.
Table 4-17. UF System Operational Measurement Statistics for 2007 Retest
Parameter
Standard
Count Mean Median Minimum Maximum Deviation 95% CI
UF Operation per day (hr)
Intake Flow (gpm)
Feed Flow (gpm)
Filtrate Flow (gpm)
Retentate Flow (gpm)
Backwash Flow (gpm)
Feed Pressure (psig)
Retentate Pressure (psig)
Filtrate Temperature (°F)
25
44
45
45
44
45
45
45
13.8 14.3
288 296
232 237
206 212
26 26
Not measured -
24 25
22 23
74 75
4.0 21.5
235 303
174 271
148 245
25 28
approximately 900 j
13 32
11 31
62 84
4.6
16.2
19.7
19.6
0.7
jal per backwash
5.9
5.8
5.3
±1.8
±4.8
±5.7
5.7
±0.2
±1.7
±1.7
±1.6
As discussed in Section 4.2.2, the UF system was operated with only 15 modules during
the 2007 test. The mean feed water flow rate of 232 gpm was similar to the 2006 test
(mean of 246 gpm), and was somewhat below the design feed flow rate of 259 gpm
specified in Table 3-2. The mean filtrate flow of 206 gpm was also lower than the 2006
test (mean of 220 gpm). Based on the mean flow rates, the mean water recovery for the
UF system was 88.8%, which is close to the 2006 recovery of 89.5%. The 206 gpm mean
filtrate flow corresponds to a 24-h production rate of 296,640 gpd. This filtrate
production rate includes water used for backwashes. The stated UF design production
rate is 250,000 gpd (not including backwash water). The backwash process used 900 gal
of UF filtrate per event, and a backwash is conducted every 30 min. Therefore, for 24 h
of operation, 48 backwashes would be conducted using a total of 43,200 gal of UF
filtrate. Subtracting this volume from the calculated daily filtrate production volume of
296,640 gal leaves 253,440 gal of UF product water, which is similar to the design
production volume 250,000 gpd.
The EUWP includes a totalizer to monitor the hours of UF system operation. The hours
of operation during the retest varied widely, from 4 to 21.5 h. The retest was designed for
the UF system to operate for a similar number of hours as the 2006 test. Typically the
unit was operated over a 16-h period with downtime due to maintenance and cleaning.
The mean hours of operation for the 25-day test were 13.8 h (2006 test mean was 15 h)
with a median of 14.3 h.
94
-------
Total UF filtrate production was also tracked using the RO feed totalizer. The total
filtrate produced was 3,551,000 gal over 350.1 h of operation. This yields a mean useable
UF filtrate production per hour of operation of 10.1 kgal. If the filtrate water used for
backwashing the system is added (595,730 gal) to this production volume, then the mean
total filtrate production per hour of operation is approximately 11.8 kgal. Figure 4-17
shows the cumulative filtrate production over the duration of the 2007 retest.
4000
07/30/07 08/04/07 08/09/07 08/14/07 08/19/07 08/24/07
Date
Figure 4-17. UF filtrate production for the 2007 retest.
Figure 4-18 shows the UF system flow rates over the duration of the retest. The retentate
flow rate remained steady through the test. Figure 4-19 shows the feed and retentate
pressures during the test and Figure 4-20 shows the calculated TMP results. After the first
four days of operation (July 30 - August 2), the feed water pressure was increased in
order to maintain the target flow rates for feed water and filtrate. TMP increased from
<10 psig to 17 psig. Therefore, the UF system was shutdown for CIP. The system was
cleaned on August 8 and put back into service. The TMP did not drop back to original
operating conditions as expected, but did decrease slightly after several hours of
operation from a high of 19 psig to a low of 16 psig.
95
-------
300
250
System cleaned
on 8/8 and 8/13
Feed -*- Filtrate -•— Retentate
0
07/30/07 08/04/07 08/09/07 08/14/07
Date
Figure 4-18. UF system flow rates for 2007 retest.
08/19/07
08/24/07
System cleaned
on 8/8 and 8/13
07/30/07
08/04/07
08/09/07 08/14/07
Date
08/19/07
08/24/07
Figure 4-19. UF system feed and retentate pressures for 2007 retest.
96
-------
30
25
20
15
10
0
07/30/07
System cleaned
on 8/8 and 8/13
08/04/07
08/09/07 08/14/07
Date
08/19/07
08/24/07
Figure 4-20. UF system TMP for 2007 retest.
The TMP increased again on August 12 to 19 psig and feed water pressure exceeded 30
psig to in order to maintain water flow rates. The UF was again shutdown and a chemical
cleaning performed on August 13. This cleaning dropped the TMP back to 16 psig, but
again the TMP was not as low as when the system was started on July 27. The feed water
pressure increased again to over 30 psig on August 14 and TMP increased accordingly. It
was decided to continue to operate the UF at the higher feed water pressure and TMP, as
these pressures were still within the design specification and operating specification for
the unit. As can be seen in Figure 4-19 and 4-20, the UF feed pressure remained steady
for several days and was actually lower for the last week of the test. TMP remained fairly
steady for the duration of the test.
Figure 4-21 shows the specific flux calculated for the UF system during the retest. The
impact of solids buildup or some type of change in membrane filtrate capacity on the
system is clear prior to the CIP performed on August 8. The CIP was successful in
stabilizing the drop in specific flux (and increasing TMP discussed above), but did not
result in returning the membrane to the specific flux attained at the beginning of the test.
Following the second cleaning on August 13, the specific flux continued to drop for the
next three days and then actually started to increase slightly over the remaining ten days
of the test.
Figure 4-22 shows the loss of specific flux over the duration of the retest. The loss of
specific flux is calculated by comparing the specific flux on a given day to the value
97
-------
calculated at the start of the test. This type of data shows the impact of cleaning and
backwash by comparing any given days specific flux to the start of the test. As can be
seen, there was a steady loss of specific flux at the beginning of the test, but the CIP on
August 8 and again in August 13 stabilized the loss of specific flux.
A chemical coagulant (ferric chloride) was used during the retest. During the initial test
runs during the setup for the retest, jar tests showed a ferric chloride dose of 1 mg/L as Fe
(a 12% Fe solution was fed at 10 mL/min for an intake flow rate of 300 gpm) should be
the target feed rate. This feed rate was maintained until the rapid increase in TMP and
drop in specific flux occurred. After the chemical cleaning on August 7 and 8, the ferric
chloride feed rate was increased to 20 mL/min at the target intake flow rate of 300 gpm,
yielding a dose rate of 2 mg/L as Fe. Subsequent jar test suggested that with the low
turbidity in the source water the ferric chloride feed should actually be decreased. The
ferric chloride feed was shut off on August 10 and remained off until the CIP was
required on August 13. The rapid loss of flux and rise in TMP indicated that the
coagulant should be used in the system, but at a lower dose than used at the start of the
test. The ferric chloride feed was set at 0.2 mL/min (0.02 mg/L as Fe) and continued at
that rate for the remainder of the test.
System cleaned
on 8/8 and 8/13
07/30/07
08/04/07
08/09/07
08/14/07
08/19/07
08/24/07
Date
Figure 4-21. UF system specific flux for 2007 retest.
98
-------
System cleaned
on 8/8 and 8/13
07/30/07
08/04/07
08/09/07
08/14/07
08/19/07
08/24/07
Date
Figure 4-22. Loss of specific flux over time for 2007 retest.
The power use for the UF system was monitored by a power meter that was separate from
the RO system. This provided power data specific to the UF system. Twice daily power
readings were recorded by the operators. The power data was then combined with the
hours of UF system operation to calculate the power used per hour of operation. The
mean power consumption was 37 kWh per hour of operation with a median value of 37
kWh per hour of operation. The initial ETV test in 2006 had a mean power consumption
of 39 kWh per hour of operation. Figure 4-23 shows the power consumption per hour of
operation during the 2007 test.
99
-------
0
07/30/07
08/04/07
08/09/07
08/14/07
08/19/07
08/24/07
Date
Figure 4-23. UF Power consumption per hour of operation for 2007 retest.
4.2.3.2 Task C2: Cleaning Efficiency
The objective of this task was to evaluate the membrane cleaning procedures and
determine the fraction of specific flux restored following chemical cleaning.
4.2.3.2.1 UF Backwash and Cleaning Frequency and Performance
The automatic backwash system was operated on a 30-min cycle with 900 gal of filtrate
used per backwash cycle, which was the same schedule and volume used for the initial
verification test. The backwash volume represented approximately 15% of the filtrate
produced during the 2007 retest, which was similar to the 2006 test when approximately
14% of the filtrate produced was used for backwash.
The TMP began to build quickly at the start of the retest and the specific flux dropped
from 4.62 gfd/psig to 1.78 gfd/psig over the first seven days of the test.
A CIP was performed on August 8. Following the CIP, the measured specific flux was
1.71. and 1.78 gfd/psig on the next two readings. The CIP was not successful in restoring
the membrane to the original specific flux of 4.62 gfd/psig. The TMP was in the range of
18-19 psig, which was just below the recommended 20 psig that would indicate a CIP
was needed. The decision was made to continue operating and monitoring the change in
TMP and specific flux. As discussed above, the ferric chloride feed was turned off on
August 10.
100
-------
On August 12, after four more days of operation, the IMP increased to 22 psig and the
specific flux dropped to 1.27 gfd/psig. A second chemical cleaning was performed on
August 13 in an attempt to restore the membrane productivity and lower the TMP. The
CIP was only partially successful, as the specific flux after the cleaning was 1.93 gfd/psig
and the TMP was lowered to 16 psig. This represents a minimal recovery of specific flux.
After the August 13 cleaning, the ferric chloride feed was turned back on, but at a lower
dose than used previously.
It is not known why the chemical cleaning was successful in the initial verification test,
but was not able to restore the membranes to the original conditions during the retest. As
shown in Figures 4-20, 4-21, and 4-22, the UF system did stabilize after the second
cleaning and specific flux and TMP remained constant or actually increased slightly
during operation after the second cleaning.
The amount of CIP chemical used, the pH of the solutions, and the temperatures were not
recorded during this retest. Operators have indicated that the same procedures and
chemicals were used, but this cannot be verified due to the lack of written records.
4.2.3.2.2 Total Organic Carbon Results for UF Cleaning Solutions
Samples of the cleaning solution for the UF system CIP were collected from two cleaning
periods. These samples were analyzed for TOC as specified in the ETV Protocol and the
Test Plan. The TOC results for the August 8, 2007 and August 13, 2007 UF system
cleaning solutions are presented in Table 4-18. Note that no low pH solution sample was
collected for the August 13, 2007 cleaning. The TOC was higher in the low pH solution.
The used cleaning solution was acceptable for discharge to the sanitary sewer system at
Selfridge ANB and was discharged to the sewer system after each cleaning cycle.
Table 4-18. UF Cleaning Solution TOC Results for 2007 Retest
Sample Date TOC (mg/L)
Low pH solution 8/08/07 750
High pH solution 8/08/07 160
High pH solution 8/13/07 48
4.2.3.3 Task C3: 2007 Water Quality Results
The primary objective of this task was to assess the ability of the membrane equipment to
meet the water quality goals, which were established as producing water that meets EPA
National Drinking Water Regulations. This section presents the water quality results for
the 2007 verification retest. Data on the bacteriological samples and integrity testing are
presented later in a separate section of this report.
Table 4-19 shows the daily turbidity results and the summary statistics for the verification
test. Figures 4-24 and 4-25 present the grab sample turbidity readings for the UF feed and
101
-------
Table 4-19. Turbidity Data for the 2007 Retest - Hand-Held Meter
Turbidity
Date UF Feed
07/30/07 0.7
07/30/07 .5
07/31/07 .6
07/31/07 .7
08/01/07 .3
08/01/07 .6
08/02/07 .4
08/02/07 .4
08/03/07 .4
08/03/07 .3
08/04/07 .3
08/04/07 .5
08/05/07 .1
08/05/07 .3
08/06/07 4.6
08/06/07 2.6
08/09/07
08/09/07
08/10/07
08/10/07
08/11/07
08/11/07
08/12/07
08/13/07
08/14/07
08/15/07
08/15/07
08/16/07
08/16/07
08/17/07
08/17/07
08/18/07
08/18/07
08/19/07
08/19/07
.6
.3
.2
.4
.2
.3
.5
.8
.7
.4
.6
.3
.7
.4
.3
.3
.4
.1
.4
08/20/07 8.3
08/20/07 9.0
08/21/07 5.5
08/21/07 5.2
08/22/07 3.0
08/22/07 4.9
08/23/07 3.3
08/23/07 3.4
08/24/07 2.7
08/24/07 1.7
Mean: 2.3
Median: 1.5
Minimum: 0.7
Maximum: 9.0
Std. Dev.: 1.8
95% CI: ±0.5
(NTU)
UF Filtrate
0.01
0.01
O.01
0.10
0.10
0.20
0.20
0.10
0.10
0.10
0.07
0.12
0.08
0.09
0.12
0.23
0.09
0.08
0.09
0.15
0.12
0.16
0.17
0.12
0.16
0.11
0.17
0.14
0.14
0.16
0.17
0.14
0.15
0.13
0.13
0.15
0.18
0.17
0.24
0.20
0.51
0.20
0.23
0.10
0.25
0.14
0.14
0.01
0.51
0.08
±0.02
UF % Reduction
98.5
99.3
99.4
94.1
92.3
87.5
85.7
92.9
92.9
92.3
94.7
92.1
92.9
93.2
97.4
91.1
94.3
93.9
92.7
89.1
89.7
88.1
88.9
93.3
90.8
95.4
89.4
93.8
91.8
88.2
86.5
89.1
89.3
87.7
91.0
98.2
98.0
96.9
95.4
93.2
89.6
93.9
93.2
96.3
85.5
92.5
92.9
85.5
100
3.8
±1.1
102
-------
filtrate over the duration of the retest. Note that there is no data for August 7 and 8
because the UF system was shut down for cleaning. The UF system reduced the turbidity
from a mean of 2.3 NTU in the feed water to a mean of 0.14 NTU in the UF filtrate. The
filtrate turbidity 95% confidence interval is 0.12 NTU to 0.16 NTU. Note that despite the
UF system integrity issues during the 2006 test, as discussed in Chapter 4, the 2006 mean
filtrate turbidity and 95% confidence interval were the same as for the 2007 test.
Turbidity in the feed water was reduced by a mean value of 92.5%, with a median
reduction of 92.9% through the UF system. There were two spikes in the feed water
turbidity - on August 6, and from August 20 to 22. Both spikes were likely caused by
rain events on these days. These feed water turbidity spikes did cause small increases in
the filtrate turbidity, but only one measurement - 0.51 NTU on August 22 - was above
0.3 NTU. Therefore, the UF system also met the NPDWR turbidity requirements during
the 2007 retest.
o
07/27/07
08/01/07
08/06/07
08/11/07
Date
08/16/07
08/21/07
08/26/07
Figure 4-24. UF feed turbidity for the 2007 retest.
103
-------
07/27/07
08/01/07
08/06/07
08/11/07
Date
08/16/07
08/21/07
08/26/07
Figure 4-25. UF filtrate turbidity for the 2007 retest.
The in-line turbidimeter readings for the 2007 retest can be used to evaluate water quality
because the date and time were properly set. However, the in-line turbidity readings only
span from July 30 to August 21. The computer used to log the data crashed on August 21,
and was not replaced for the last three days of the test. Figure 4-26 graphically shows the
UF feed and filtrate in-line turbidity readings. Note that there are two y-axes in the graph,
one for the feed and one for the filtrate. Also note that there are two large gaps in the
data, corresponding to UF system cleanings, and a few smaller gaps when the system
automatically shut down overnight. The first cleaning was on August 7 and 8, and the
second cleaning was on August 13. The summary statistics for the UF feed and UF
filtrate in-line turbidity measurements are shown in Table 4-20. The mean UF filtrate
turbidity is 0.019 NTU, and the maximum recorded turbidity is 0.236 NTU. At no point
did two consecutive measurements exceed the 0.15 NTU value that would have required
the system to be taken off-line for a direct integrity test.
104
-------
14 -i
0.25
£3
H
z
73
I
=
0.00
7/30/2007
8/4/2007
8/9/2007
8/14/2007
8/19/2007
Date
Figure 4-26. UF feed and filtrate in-line turbidity readings for the 2007 retest.
Table 4-20. In-Line Turbidity Measurement Statistics for the 2007 Retest
Mean:
Median:
Minimum:
Maximum:
Count:
Std. Dev.:
95% CI:
UF Feed (NTU)
1.89
1.29
0.433
13.61
1568
1.75
±0.087
UF Filtrate (NTU)
0.019
0.017
0.004
0.236
1568
0.011
±0.001
Table 4-21 shows the conductivity results for the UF and the summary statistics for the
retest. Because the RO system reduced the dissolved ions in the water in the 2006 test, it
was not monitored during this retest. As expected, there was no change in conductivity
levels in the UF treated water.
Table 4-21 also presents the pH and temperature data collected from the UF system. The
pH in the filtrate ranged from 7.3 to 9.0 and was in the same range as the feed water.
105
-------
Table 4-21. Conductivity, pH, and Temperature Data for the 2007 Retest
Date
07/30/07
07/30/07
07/31/07
07/31/07
08/01/07
08/01/07
08/02/07
08/02/07
08/03/07
08/03/07
08/04/07
08/04/07
08/05/07
08/05/07
08/06/07
08/06/07
08/09/07
08/09/07
08/10/07
08/10/07
08/1 1/07
08/1 1/07
08/12/07
08/13/07
08/14/07
08/15/07
08/15/07
08/16/07
08/16/07
08/17/07
08/17/07
08/18/07
08/18/07
08/19/07
08/19/07
08/20/07
08/20/07
08/21/07
08/21/07
08/22/07
08/22/07
08/23/07
08/23/07
08/24/07
08/24/07
Mean:
Median:
Minimum:
Maximum:
Std. Dev.:
95% CI:
UF Feed
8.15
9.15
8.08
8.55
8.54
8.89
8.40
9.01
8.71
8.98
8.19
9.48
8.33
8.15
7.61
8.63
7.99
7.99
7.56
8.35
7.34
8.67
7.85
7.92
8.17
7.68
8.23
7.60
8.47
7.55
8.05
7.58
7.86
7.65
7.56
7.02
9.13
8.76
8.85
8.04
8.52
7.69
8.17
7.85
8.76
8.22
8.17
7.02
9.48
0.55
0.16
PH
UF Filtrate
8.22
8.94
7.87
8.46
8.12
8.65
7.94
8.82
8.27
8.90
7.67
9.02
8.30
7.52
7.47
7.91
7.31
7.49
7.37
8.36
7.67
8.69
7.64
7.94
8.24
7.45
7.86
7.36
7.91
7.47
7.79
7.52
7.77
7.43
7.45
7.31
8.05
8.01
8.29
7.68
7.53
7.49
7.71
7.53
7.79
7.92
7.79
7.31
9.02
0.48
0.14
Conductivity
UF Feed
271
294
300
312
308
325
340
297
310
295
332
250
302
368
238
253
260
294
337
329
336
329
345
337
355
388
361
385
348
383
414
403
413
428
446
387
214
259
397
365
312
434
426
296
285
335
332
214
446
57
17
(uS/cm)
UF Filtrate
275
296
299
298
310
317
312
299
313
295
334
256
305
372
239
248
265
296
340
327
330
330
349
336
348
384
368
386
356
389
401
399
405
429
447
156
213
263
357
368
304
436
427
299
290
328
327
156
447
61
18
Temperature
UF Feed UF
25.1
27.5
32.2
34.2
26.3
36.8
28.1
33.4
29.1
29.7
26.6
27.8
24.3
24.1
23.7
29.4
25.8
25.5
25.2
28.9
25.3
29.2
27.3
24.1
27.3
27.0
24.8
27.4
26.8
24.7
27.4
23.4
22.1
20.7
19.6
17.3
18.2
19.0
22.1
21.5
25.1
21.7
NM
25.3
27.3
25.9
25.7
17.3
36.8
4.0
1.2
(°C)
Filtrate
26.2
27.7
29.1
30.1
27.0
30.1
28.0
30.5
28.9
30.8
27.4
27.5
24.5
24.8
23.8
27.2
25.6
25.4
25.3
27.2
25.0
27.7
26.7
24.3
26.9
25.9
25.0
25.4
26.6
24.9
25.9
22.9
22.7
19.8
20.1
17.6
18.3
18.8
20.9
20.8
22.6
21.7
27.8
24.6
25.7
25.2
25.6
17.6
30.8
3.3
1.0
106
-------
The UF treatment had no effect on the temperature of the water as it passed through the
system. Water temperature in the lake feed water at the beginning of the test was in the
25 to 30 °C range and dropped during the test to 20 to 25 °C by the end of the test in late
August. Temperature variation and impact on membrane operating production (flux and
specific flux) were accounted for in the operating section by standardizing the data to 20
°C, as described in Section 3.9.1.3. The temperature data in Table 4-21 served as the basis
for the temperature adjustment calculations.
Table 4-22 presents the other water quality data collected on a weekly basis during the
retest. The UF system removed suspended material as shown by the TSS concentration in
the filtrate, which was always below the detection limit of 2 mg/L. These data support the
daily turbidity results showing over 92% reduction in turbidity. The 2006 test results also
showed the TSS in the filtrate was always less than 2 mg/L. The UF system did not
change the other water quality parameters, as would be expected. These other parameters,
such as hardness, alkalinity, TDS, etc. primarily represent dissolved inorganic
constituents that are not removed by physical filtration.
The UF system showed minor reduction in organic material as measured by TOC. The
feed water, filtrate, and retentate all showed TOC concentrations within a range of 2.0 to
3.4 mg/L. The filtrate typically showed a 0.0 to 0.8 mg/L reduction in TOC compared to
the feed water, and the retentate occasionally showed an increase in TOC of 0.5 mg/L
compared to the feed water. These data would indicate that most of the organic material,
as measured by the TOC test, was dissolved in the feed water.
Table 4-22. Other Water Quality Data for the 2007 Retest
Date
UF Feed
TSS (mg/L)
Filtrate Retentate Backwash
7/30/07
8/10/07
8/15/07
8/22/07
3
ND(2)
ND(2)
3
ND(2)
ND(2)
ND(2)
ND(2)
6
10
ND(2)
10
20
50
ND(4)
40
Date
7/30/07
8/10/07
8/15/07
8/22/07
UF Feed
2
2
3
2
8
8
0
9
TOC (mg/L)
Filtrate Retentate
2.4
2.0
3.1
2.5
2.8
3.3
2.7
3.4
Backwash
NM
NM
3.1
2.6
Date
7/30/07
8/10/07
8/15/07
8/22/07
Hardness (mg/L
UF Feed
95
88
110
74
as CaCO3)
Filtrate
95
87
110
72
Alkalinity
Feed Water
69
150
210
70
(mg/L as CaCO3)
Filtrate
66
140
210
63
107
-------
Table 4-22. Other Water Quality Data for the 2007 Retest (continued)
Date
7/30/07
8/10/07
8/15/07
8/22/07
TDS (mg/L)
UF Feed Filtrate Backwash
Date
7/30/07
8/10/07
8/15/07
8/22/07
170
150
210
200
UV254
UF Feed
NM
0.08
0.0904
0.141
160
140
210
170
NM
NM
NM
190
(absorbance/mL)
Filtrate
0.051
0.0439
0.0611
0.090
NM - not measured
4.2.3.4 Task C4: 2007Membrane Module Integrity
The objective of this task was to demonstrate methodology for direct integrity testing and
indirect integrity monitoring of the UF membranes. Pressure decay tests were used to
document UF membrane integrity.
4.2.3.4.1 UF System Pressure Decay andMicrobialReduction Results -2007 Retest
As discussed in Chapter 4, following the 2006 test, the UF seal integrity problem was
addressed, and a retest was scheduled for 2007.
Prior to starting the retest, each membrane cartridge was individually tested as discussed
in Section 4.2.2, and several were found to have broken fibers that required plugging.
After plugging these fibers, each cartridge was again pressure tested. The results showed
that 15 of the 16 modules were acceptable, so TARDEC and USER decided to run the
test with only 15 membranes. A full system pressure decay test was run on July 23, and
the pressure decay rate was calculated as 0.025 psig/min. This value was more than ten
times lower than the mean value of 0.29 psig/min obtained during the 2006 verification
test.
For the 2007 test, pressure decay tests were again conducted daily. The results of these
tests are shown in Table 4-23. The pressure decay rates were 5-10 times lower than those
from the 2006 test, with a mean value of 0.025 psig/min and a median value of 0.017
psig/min. These pressure decay rates were also lower than those measured during the
laboratory tests. Figure 4-27 shows the pressure decay results graphically.
108
-------
Table 4-23. Pressure Decay Results for UF System for the 2007 Retest
Date
7/30/2007
7/31/2007
8/1/2007
8/2/2007
8/3/2007
8/4/2007
8/5/2007
8/6/2007
8/9/2007
8/10/2007
8/11/2007
8/12/2007
8/14/2007
8/15/2007
8/16/2007
8/17/2007
8/18/2007
8/19/2007
8/20/2007
8/21/2007
8/22/2007
8/23/2007
8/24/2007
(1) NM = not
0
15.1
15.1
15.2
15.2
15.2
15.1
15.0
15.1
15.1
15.1
15.0
15.1
15.1
15.1
15.1
15.1
15.0
15.1
15.1
15.0
15.1
15.1
15.1
0.5
15.0
15.1
15.1
15.1
15.1
15.1
14.9
15.1
15.1
15.1
14.9
15.0
15.1
15.1
15.0
15.1
15.0
15.1
15.0
15.0
15.1
15.1
15.1
1.0
15.0
15.1
15.0
15.1
15.1
15.1
14.9
15.1
15.1
15.1
14.9
15.0
15.0
15.0
15.0
15.1
14.9
15.1
15.0
14.9
15.1
15.1
15.0
1.5
14.9
15.1
14.9
15.1
15.1
15.1
14.9
15.1
15.1
15.0
14.9
14.9
14.9
15.0
15.0
15.0
14.9
15.0
14.9
14.9
15.0
15.1
15.0
2.0
14.9
15.1
14.9
15.1
15.1
15.1
14.9
15.1
15.0
15.0
14.9
14.9
14.9
14.9
14.9
15.0
14.9
14.9
14.9
14.9
15.0
15.1
14.9
2.5
14.9
15.1
14.8
15.1
15.1
15.0
14.9
15.0
15.0
14.9
14.9
14.9
14.9
14.9
14.9
15.0
14.9
14.9
14.8
14.9
14.9
15.1
14.9
3.0
14.9
15.1
14.7
15.1
15.0
15.0
14.9
15.0
14.9
14.9
14.9
14.9
14.9
14.9
14.9
15.0
14.9
14.9
14.7
14.9
14.9
15.1
14.9
3.5
14.9
15.1
14.7
15.1
15.0
15.0
14.9
15.0
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.7
14.9
14.9
15.1
14.9
4.0
14.9
15.1
14.6
15.1
15.0
15.0
14.9
15.0
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.9
14.7
14.7
14.9
14.9
15.0
14.9
5.0
14.9
15.1
14.5
15.1
14.9
14.9
14.9
14.9
14.9
14.9
14.8
14.9
14.9
14.9
14.9
14.9
14.9
14.7
14.6
14.9
14.9
15.0
14.9
6.0
14.9
15.1
14.5
15.1
14.9
14.9
14.9
14.9
14.9
14.9
14.8
14.9
14.8
14.9
14.9
14.9
14.9
14.6
14.5
14.9
14.8
15.0
14.9
7.0
14.9
15.1
14.4
15.1
14.9
14.9
14.9
14.9
14.9
14.9
14.7
14.9
14.7
14.8
14.9
14.9
14.9
14.6
14.5
14.9
14.8
15.0
14.9
8.0
14.9
15.1
14.4
15.1
14.8
14.9
14.9
14.9
14.9
14.9
14.7
14.9
14.7
14.7
14.9
14.9
14.9
14.5
14.5
14.9
14.7
15.0
14.9
9.0
14.9
15.1
14.4
15.1
14.7
14.9
14.9
14.9
14.9
14.8
14.7
14.9
14.7
14.7
14.9
14.9
14.9
14.5
14.4
14.9
14.7
15.0
14.9
10.0
14.9
15.1
14.3
15.1
14.7
14.9
14.9
14.9
14.9
14.8
14.7
14.9
14.7
14.7
14.9
14.9
14.9
14.4
14.4
14.9
14.7
15.0
14.9
measured
11.0 12.0
14.9 NM(1)
15.1 NM
14.3 14.3
15.1 15.1
14.7 14.7
14.9 14.9
14.9 14.9
14.9 14.9
14.8 14.8
14.7 14.7
14.7 14.7
14.9 14.9
14.7 14.6
14.7 14.7
14.9 14.9
14.9 14.9
14.9 14.9
14.4 14.3
14.4 14.3
14.9 14.9
14.7 14.6
15.0 15.0
14.9 14.9
Minimum
Maximum
Mean
Median
95% Confidence Interval
Decay
(psig/min)
0.018
0.000
0.075
0.008
0.042
0.017
0.008
0.017
0.025
0.033
0.025
0.017
0.033
0.033
0.017
0.017
0.008
0.058
0.058
0.008
0.033
0.008
0.017
0.000
0.075
0.025
0.017
± 0.008
109
-------
15.2
15.1
14.3
14.2
0
246
Elapsed Time (min)
Figure 4-27. Pressure Decay Results for the 2007 Retest.
10
12
—•— 7/30/2007
-A-7/31/2007
—X— 8/1/2007
-*- 8/2/2007
-•— 8/3/2007
—I— 8/4/2007
—— 8/5/2007
—•— 8/6/2007
—0— 8/9/2007
-•-8/10/2007
—A—8/11/2007
-X-8/12/2007
—X—8/14/2007
-•-8/15/2007
—1—8/16/2007
8/17/2007
8/18/2007
—•—8/19/2007
-•- 8/20/2007
-A-8/21/2007
—X— 8/22/2007
-*- 8/23/2007
—•— 8/24/2007
110
-------
For the 2007 test, Bacillus endospores were substituted for HPC and total coliform. This change
was made for the following reasons:
• Expected higher concentrations would allow better measurement for log reductions
(note actual concentrations were not as high as expected);
• Good size range to represent bacteria and challenge the membrane (0.8 micron x 1.8
micron);
• Almost no potential for growth in piping and equipment as compared to HPC;
• Less sensitive to temperature during shipment to laboratory; and
• Can be easily used in laboratory tests for comparison.
The Bacillus endospores data are shown in Table 4-24. Figure 4-28 shows the results over the
duration of the retest. The log reductions measured had a mean value of 0.88 logio with a range
of 0.07 to 1.74 logic. These observed reductions were lower than predicted from an integral UF
membrane, and the prior lab challenge data discussed in Section 4.1.3.4.1. It is possible that there
was endospore contamination on the filtrate side of the membranes, given the previous
membrane integrity problems, or that there were continuing membrane integrity issues.
Table 4-24. Bacillus Endospores - 2007 Retest
Bacillus Endospores (CFU/100 mL)
Date
7/30/2007
7/31/2007
8/1/2007
8/2/2007
8/6/2007
8/9/2007
8/9/2007
8/14/2007
8/15/2007
8/15/2007
8/16/2007
8/16/2008
8/20/2008
8/21/2008
8/22/2008
8/23/2008
Geometric Mean:
Minimum:
Maximum:
Feed Water
1,224
874
880
862
1,262
1,276
1,540
1,159
1,504
1,063
1,269
1,258
6,360
7,420
1,691
2,122
1,562
862
7,420
UF Filtrate
165
217
206
142
184
134
144
996
413
231
323
228
359
134
98
78
203
78
996
Log Reduction
0.87
0.61
0.63
0.78
0.84
0.98
1.03
0.07
0.56
0.66
0.59
0.74
1.25
1.74
1.24
1.43
0.88
0.07
1.74
UF Retentate
1,124
769
1,021
833
1,704
2,024
,725
,308
,690
,106
,787
,700
7,820
9,120
3,514
4,199
1,823
769
9,120
UF Backwash
644
NM
NM
NM
NM
NM
6,954
4,010
NM
NM
NM
NM
NM
10,660
NM
NM
NA
NA
NA
111
-------
10000
1000
100
•g
10
* * *
* Raw Endospores • Filtrate Endospores
1
07/30/07
08/04/07
08/09/07 08/14/07
Date
08/19/07
08/24/07
Figure 4-28. Bacillus endospores results for the 2007 retest.
4.2.3.4.2 UF System Particle Count Data - 2007 Retest
The in-line particle counters were calibrated properly for the 2007 retest, so the data is reported
herein. These counters measured the particle counts in the UF feed and UF filtrate every five
minutes, and stored the data for transfer to a personal computer. Particle count data can be
helpful in evaluating the integrity and performance of membrane systems and in predicting the
reduction/rejection of microbial contaminants.
The particle count data was condensed from five-minute increments to one-hour averages for
graphical presentation. The data were separated to provide information on various size ranges
(e.g. 2-3 |im, 3-5 jim), as these sizes correspond to the sizes of various microbial contaminants of
interest in drinking water, such as Cryptosporidium (3 to 5 jim).
Figure 4-29 displays the feed and filtrate 2-3 jim particle counts, while Figure 4-30 shows the 3-
5 |im particle counts.
112
-------
Some notes about these figures and the particle count data presented:
• The y-axis of both graphs is in logarithmic scale.
• There is no particle count data for the first two days, and last three days of testing. The
particle count data supplied by the field operators begins at 8:30 a.m. on August 1. The
data ends at 3:35 p.m. on August 21 because the computer logging the data crashed.
• As with the 2007 turbidity data, there is no data for August 7 and 8, and from the
afternoon of August 12 to the afternoon of August 13, because the UF system was shut
down for cleaning.
• There were numerous other smaller gaps in the data, representing when the UF system
was shut down for the daily pressure decay tests, or other reasons.
• There were numerous single time point spikes that were likely due to the automatic
backwashes executed every half hour.
100000
10000
o
Si-
's
o
u
1000
» Filtrate Feed
:•!•• .• 1 -i .• •• A. V * •• .
•'•.-. £•* /:v • ,flr«F.- • • A- •'.
'.' ".-l""-sV """." :x^ r-' *'."\
10
1
08/01/07
08/06/07
08/11/07
Date
08/16/07
08/21/07
Figure 4-29. 2-3 um particle counts for the 2007 retest.
113
-------
100000
10000
s
1000
e
o
U
•3 100
« Filtrate Feed
(V-
• './-J :•• *:
• .-W .- •
•'•:.".;. W-'V-£ •' -. ••• %-j 1
:.v;;v%y, : ^.-t-;; • -: \.-;
Asyl\"*f"\ ."i
.* .*. 'Sf. •- .• - . • .
• • *• «
08/01/07
08/06/07
08/11/07
Date
08/16/07
08/21/07
Figure 4-30. 3-5 um particle counts for the 2007 retest.
The mean 2-3 jim particle count for the feed water was 13,376/lOmL with a median value of
13,395/lOmL. The range of particle counts for the feed water was 0-39,418/10mL. The filtrate
had a mean particle count in the 2-3 jim size of 112/lOmL with a median of 55/10mL and a
range of 0-13,908/10mL. Note that these statistics are based on the individual counts, not the
hourly averages calculated for the graphs.
The UF system goes through a backflush cycle every half-hour. During these backflushes, the
particle counters are still working, so the raw particle count data includes numerous spikes in the
counts, presumably because the counts were taken during backflushes. Therefore, the maximum
filtrate particle count of 13,908 may not be indicative of the performance of the UF system under
operation. As evidenced by the low mean filtrate count, the vast majority of the counts were less
than 200/lOmL. Of 3,408 2-3 |im counts used for this analysis, only 371 were >200/10mL, 67
were >500/10mL, and 25 were >l,000/10mL. The UF system reduced the 2-3 jim particles by a
mean value of 2.21 log 10 with a median reduction value of 2.38 logio. However, there were
many instances where the log reduction was negative because the feed count was lower than the
filtrate count. If these negative log reductions are removed from the average log reduction
calculations, the mean increases to 2.30 logic, and the median is 2.39 logic.
The mean 3-5 |im particle count for the feed water was 24,634/lOmL with a median value of
23,605/lOmL. The range of particle counts for the feed water was from 0-91,595/10mL. The
114
-------
filtrate had a mean 3-5 jim particle count of 157/lOmL with a median of 77/10mL and a range of
0-14,059/10mL. As with the 2-3 jim data, these statistics are based on the individual
measurements, not the hourly averages. Of the 3,395 3-5 jim counts, 670 were >200/10mL, 127
were >500/10mL, and 45 were >l,000/10mL. As with the 2-3 jim maximum count, the 3-5 jim
maximum count of 14,059 may not be indicative of UF performance due to particle count data
being collected during the backflushes. The UF system reduced the 3-5 jim particles by a mean
value of 2.33 logic with a median reduction value of 2.52 logic. If the negative log reduction
calculations are removed from the calculations, the mean is 2.45 logic, and the median is 2.54
logio
As can be seen, the UF system reduced the particle count in these size ranges. The observed
reduction of paniculate matter supports the pressure decay test data in showing that the UF
system maintained integrity throughout the test. Further, a 2.4 logic reduction would tend to
predict a similar or larger reduction in equal and larger size microbial contaminants. Combined
with the pressure decay test, these results would tend to support that the UF system should give
2-3 logic control of these contaminants.
Unfortunately, the direct measurement of Bacillus endospores does not, and could not confirm
these indicator tests of UF system performance for microbial contaminants.
4.2.3.4.3 Correlation of Membrane Integrity Indicators
As discussed with the presentation of the turbidity data for the 2007 retest in Section 4.2.3.3.1,
although the UF system membrane integrity issues from the 2006 test were resolved, the 2007
hand-held turbidimeter data did not indicate improved membrane performance. Both the 2006
and 2007 tests had mean UF filtrate turbidities of 0.14 NTU, with 95% confidence intervals of ±
0.02. However, this may just be an issue of the sensitivity of the meter.
For both the 2006 and 2007 tests, the highest UF filtrate turbidity readings were associated with
increased turbidity in the UF feed water due to rain events. The turbidity, particle count, bacteria,
and pressure decay test data collected during these events can be compared to look for any
correlations.
A rain event occurred during the 2007 retest on August 20 to 22. The feed water turbidity did not
increase as high during this rain event, but examining this event is useful because there is particle
count data and inline turbidity data, although this data is only available through the afternoon of
August 21 due to the crash of the data logging computer.
Table 4-25 presents the turbidity data from this time period, and the corresponding Bacillus
endospores data, pressure decay test results, and recorded transmembrane pressure readings.
Also included for comparison are the averages for each parameter over the course of the test. The
feedwater turbidity peaked at 9.0 NTU on August 20, but the filtrate turbidity peaked on the
afternoon of August 22. Unfortunately the in-line turbidimeters were off-line by this point, so
these readings cannot be correlated with the in-line turbidity readings at the same time. On
August 20 and 21, the Bacillus endospore and 2-3 jim particle counts were also significantly
above average. The August 20 filtrate endospore count was significantly higher than the average
for the test, but on August 21, the filtrate endospore count was below average. The filtrate 2-3
115
-------
|im particle counts on August 20 and 21 do not indicate any membrane integrity issues. The
August 20 pressure decay rate was significantly above the average, but it was then significantly
below average on the 21st. The TMP spiked up to 26 psig on the morning of August 20, but then
it was back down to 21 psig in the early evening. The filtrate turbidity reading of 0.51 NTU on
August 22 was significantly above the mean for the test. However, the Bacillus endospores and
pressure decay test data for August 22 do not indicate any membrane integrity issues. The
weekly water chemistry samples were collected on August 22, the results of which are displayed
in Table 4-22. The filtrate TOC was 2.5 mg/L, which is in the middle of the range of filtrate
values for the test. The feed and filtrate UV254 absorbance results were 0.141 and 0.090
respectively, both of which were the highest values measured during the test.
Table 4-25. UF Membrane Integrity Indicators for August 19, 2007 to August 23, 2007
In-Line Bacillus 2-3 jim
Benchtop Turbidity'2' Endospores Particle Counts'2' Pressure
Turbidity (NTU) (NTU) (CFU/100 mL) (#/mL) Decay TMP
Date Time'1' Feed Filtrate Feed Filtrate Feed Filtrate Feed Filtrate (psig/min) (psig)
8/19 18:50
8/20 09:54
8/20 18:37
8/21 09:40
8/21 17:07
8/22 09:50
8/22 17:05
8/23 09:15
Average ± 95%
Confidence
Interval^:
1.4
8.3
9.0
5.5
5.2
3.0
4.9
3.3
2.3 ±0.5
0.13
0.15
0.18
0.17
0.24
0.20
0.51
0.20
0.14±
0.02
1.04
6.97
8.31
4.88
4.39
NM
NM
NM
1.89 ±
0.087
0.015
0.017
0.019
0.017
0.016
NM
NM
NM
0.019 ±
0.001
NM
6360
7420
1691
2122
1,562 ±
1,113
NM
359
134
98
78
203 ±
125
1,192
2,712
3,405
3,633
3,627
NM
NM
NM
1,338 ±
25.3
2.1
8.6
0.4
2.2
2.2
NM
NM
NM
11.2±
1.3
0.058
0.058
0.008
0.033
0.008
0.025 ±
0.008
19
26
21
20
20
20
20
18
16 ± 1.5
(1) Time of turbidity and TMP readings as part of the twice per day on-site data collection.
(2) Representative turbidity and particle count recordings from time closest to the time in column two.
(3) Averages are for all data over the course of the test, not just the data presented here.
4.2.4 2007 Chemical Consumption
During the 2007 retest, ferric chloride (12% Fe) was fed to the UF intake water at an initial rate
of 10 ml/min or approximately 0.16 gal per operating hour. This dosing rate gave a coagulant
concentration of approximately 1 mg/L as Fe in the intake water. After the August 6-9 period
when TMP was increasing quickly, it was initially decided to double the feed rate and then,
based on jar tests, to shutoff the ferric chloride feed. Subsequently, on August 15, after several
jar tests, it was shown that a lower dose of ferric chloride would be effective. The ferric chloride
was turned back on at a rate of 0.2 ml/min or 0.003 gal/h, which gave an iron dose rate of
approximately 0.02 mg/L. The amount of iron used at the low dose rate would correspond to
approximately 5 mg of Fe per 1000 gal of intake water and at the high dose rate to approximately
265 mg per 1000 gal of intake water.
The chemicals needed for the UF CIP were citric acid, sodium hydroxide, and calcium
hypochlorite. Citric acid was used to lower the pH of the cleaning solution for the low pH
116
-------
cleaning cycle, and sodium hydroxide was used for the high pH cleaning cycle. The calcium
hypochlorite provided chlorine to help kill any biological growth on the membranes to help
oxidize organic material.
The chemicals specified to be used for the RO chemical cleaning were citric acid and an alkaline
detergent. Only the acid cleaning was performed during the ETV test. Citric acid was used to
lower the pH of the cleaning solution for the low pH cleaning cycle, and the alkaline detergent if
it had been used would have been used for the high pH cleaning cycle. The actual amount of acid
or base needed to lower or raise the pH of the water used for the CIP solution will depend on the
local water chemistry. For this test, tap water was used for the cleaning solution.
During the 2007 UF system retest, the membranes were cleaned twice using the same basic
procedures and chemicals as for the 2006 chemical cleaning according to the operators.
Unfortunately, the actual amount of citric acid added to the CIP tank was not recorded for either
cleaning event in 2007, so the exact amount used is not known.
4.3 Quality Assurance/Quality Control
4.3.1 Introduction
An important aspect of verification testing is the QA/QC procedures and requirements. As
described in Task C6 of the methods and procedures (Section 3.9.6), a structured QAPP was
implemented to ensure the quality of collected data. Careful adherence to the procedures ensured
that the data presented in this report were of sound quality, defensible, and representative of the
equipment performance. The primary areas of evaluation were representativeness, accuracy,
precision, and completeness.
4.3.2 Documentation
The field technicians recorded on-site data and calculations in a field logbook and on specially
prepared field log sheets. The operating logbook included calibration records for the field
equipment used for on-site analyses. Copies of the logbook, the daily data log sheets, and
calibration log sheets are in Appendix B.
Data from the on-site laboratory and data log sheets were entered into Excel spreadsheets. These
spreadsheets were used to calculate various statistics (average, mean, standard deviation, etc.).
NSF DWSC staff checked 100% of the data entered into the spreadsheets to confirm the
information was correct. The spreadsheets are presented in Appendix D.
Samples collected and delivered to the NSF Laboratory for analysis were tracked using chain-of-
custody forms. Each sample was assigned a location name, date, and time of collection. The
laboratory reported the analytical results using the NSF Chemistry Laboratory management
system reports. These reports were received and reviewed by NSF DWSC staff. These laboratory
data were entered into the data spreadsheets, corrected, and verified in the same manner as the
field data. Lab reports and chain-of-custody records are included in Appendix C.
117
-------
4.3.3 Quality Audits
The NSF QA officer performed an on-site audit on September 26, 2006, which was Day 2 of
testing. The audit focused on review of the field procedures, including the collection of operating
data and performance of on-site analytical methods. The TQAP requirements and QAPP were
used as the basis for the audit. The NSF QA officer prepared an audit report. All deficiencies
were corrected immediately. The QA officer did not conduct an audit for the 2007 test. Rather,
DWSC staff visited the test site to verify continued compliance with the corrective action
requests from the 2006 audit.
The NSF QA Department reviewed the Chemistry Laboratory analytical results for adherence to
the QA requirements for calibration, precision, and accuracy detailed in the project QAPP and
for compliance with the laboratory quality assurance requirements. The laboratory raw data
records (run logs, bench sheets, calibrations records, etc.) are maintained at NSF and are
available for review.
4.3.4 Test Procedure QA/QC
The USER and TARDEC staff conducted the field monitoring, measurements, and sample
collection and handling in accordance with the EPA-approved TQAP created specifically for this
verification. NSF testing laboratory staff conducted the chemical and microbiological analyses
by following the TQAP. NSF QA Department staff performed audits during testing to ensure the
proper procedures were followed. The audit yielded no significant findings.
4.3.5 Sample Handling
All samples analyzed by the NSF Chemistry and Microbiology Laboratories were labeled with
unique identification numbers. These identification numbers appear in the NSF laboratory
reports for the tests. All samples were analyzed within allowable holding times.
However, some microbiological samples from the 2007 test did not meet the holding temperature
established in the TQAP. All samples were refrigerated immediately after collection, then
shipped in insulated shipping containers with ice or cold packs. At the beginning of the 2007 test,
ice packs were used to cool the samples, and it was found that they were not sufficient to cool the
samples properly. The starting temperature of the samples, and ambient air temperatures were
sufficiently high that the samples did not reach the proper holding temperature by the time
(within 24 h) they reached the NSF Laboratory. About halfway through the test, the field
technicians switched to ice, and this sufficiently cooled the samples.
The potential effect of exceeding the holding temperature depends on whether the
microorganisms were in a vegetative, spore or cyst state. For microorganisms in a vegetative
state such as coliform and HPC, the warm temperatures could increase growth and potentially
create a bias towards higher than expected counts. However, higher holding temperatures would
not bias counts of microorganisms that are in a spore or cyst state such as Bacillus endospores,
Giardia or Cryptosporidium, as other environmental conditions must be created to stimulate their
growth. Since the only microbiological parameter for the 2007 test was Bacillus endospores, the
high temperatures likely did not bias the endospore counts.
118
-------
4.3.6 Physical and Chemical Analytical Methods QA/QC
The calibrations of all NSF laboratory analytical instruments and the analyses of all parameters
complied with the QA/QC provisions of the NSF Laboratories Quality Assurance Manual.
Bench top field instruments that measured turbidity, pH meter, temperature and specific
conductance (conductivity) were calibrated in accordance with the data quality objectives (DQO)
in the TQAP during 2006 and 2007, with one exception. Temperature was not calibrated with a
NIST-certified precision thermometer during the first week of sampling in 2006.
In-line field meters for particle counts and turbidity measurements were factory calibrated, and
certificates were provided as required in the TQAP. However, in 2006, the incorrect calibration
certificate data for bin voltages was entered into the software program for the particle counters.
This resulted in rendering the particle count data inaccurate and not meeting the DQO. This was
a critical parameter for the ETV test, as particle count data were used as a key indicator of
membrane integrity, and to indicate real time removal of particles similar in size to pathogenic
microorganisms. The NSF QA department and DWSC manager concluded that because of this,
the UF system needed to be retested in 2007.
4.3.7 Microbiology Laboratory QA/QC
4.3.7.1 Growth Media Positive Controls
All media were checked for sterility and positive growth response when prepared and when used
for microorganism enumeration. The media was discarded if growth occurred on the sterility
check media, or if there was an absence of growth in the positive response check.
4.3.7.2 Negative Controls
For each sample batch processed, an unused membrane filter and a blank with 100 mL of sterile
buffered deionized water filtered through the membrane were also placed onto the appropriate
media and incubated with the samples as negative controls. No growth was observed on any
blanks.
4.3.8 Documentation
All laboratory activities were documented using specially prepared laboratory bench sheets and
NSF laboratory reports. Data from the bench sheets and laboratory reports were entered into
Excel spreadsheets. These spreadsheets were used to calculate average feeds and filtrates, and
logio reductions for each challenge. One hundred percent of the data entered into the
spreadsheets was checked by a reviewer to confirm all data and calculations were correct.
4.3.9 Data Review
NSF QA/QC staff reviewed the raw data records for compliance with QA/QC requirements. NSF
ETV staff checked at least 10% of the data in the NSF laboratory reports against the lab bench
sheets.
119
-------
4.3.10 Data Quality Indicators
The quality of data generated for this ETV was established through four indicators of data
quality: representativeness, accuracy, precision, and completeness.
4.3.10.1 Representativeness
Representativeness refers to the degree to which the data accurately and precisely represent the
expected performance of the EUWP system under conditions expected for use in an emergency
response situation, or theater of war. The EUWP was operated similar to conditions of
deployment in an emergency. As stated in Chapter 2, the raw water source was a fresh surface
water, representing a possible application for the EUWP during deployment. Two other ETV
reports considered the EUWP performance when using sea water and secondary waste water as
its feed.
Representativeness was ensured by consistent execution of the test protocol and TQAP for the
test, including timing of sample collection, sampling procedures, and sample preservation.
Representativeness was also ensured by using each analytical method at its optimum capability
to provide results that represent the most accurate and precise measurement it is capable of
achieving.
4.3.10.2 A ccuracy
Accuracy was quantified as the percent recovery of the parameter in a sample of known quantity.
Accuracy was measured through use of both matrix spikes of a known quantity and certified
standards during calibration of an instrument. For chemical analyses performed by the NSF
Laboratory, certified QC standards and/or matrix spikes were run with each batch of samples.
The percent recoveries of all matrix spikes and standards were within the allowable limits for all
analytical methods.
For physical and chemical analyses performed in the field, PE samples for pH and turbidity were
run once during the testing period. The reported values for pH and turbidity were within the
acceptable range for the PE samples.
4.3.10.3 Precision
Precision refers to the degree of mutual agreement among individual measurements and provides
an estimate of random error. One sample per batch was analyzed in duplicate for the NSF
Laboratory measurements. For field measurements, one process stream was analyzed in
duplicate every day. Precision of duplicate analyses was measured through RPD.
All RPD were within the allowable limit of 30 percent for each parameter with the following
exceptions:
• During the 2006 test, one conductivity measurement of RO permeate and four
measurements of turbidity exceeded 30 percent RPD. In all cases the samples were
measured very close to the instrument's limit of detection. Under such circumstances,
RPD may be greater than 30 percent, which is expected and acceptable.
• During the 2007 retest, two measurements for turbidity exceeded 30 percent RPD, and
again those measurements were at the limit of detection of the instrument.
120
-------
4.3.10.4 Completeness
Completeness is the proportion of valid, acceptable data generated using each method as
compared to the requirements of the test/QA plan. The completeness objective for data generated
during verification testing is based on the number of samples collected and analyzed for each
parameter and/or method, as presented in Table 3-10.
All planned water chemistry samples were collected and analyzed. All scheduled microbiological
samples were collected and analyzed with acceptable results.
On three out of 31 days of testing in 2006, measurements for pH, temperature, conductivity, and
turbidity were made only once per day rather than twice per day. This gave a completeness of
95% for these parameters, which met the goal of 90% in the TQAP.
During the 2007 test, measurements for pH, temperature, conductivity, and turbidity were
collected once on one day, which resulted in a completeness of 98% for these parameters. The
DQO of the TQAP was met for completeness.
121
-------
References
EPA and NSF International (2002). EPA/NSF Protocol for Equipment Verification Testing for
Removal of Inorganic Constituents. NSF International.
EPA and NSF International (2005). EPA/NSF Protocol for Equipment Verification Testing for
Physical Removal of Microbiological and Paniculate Contaminants.
NSF International (2006). Removal of Microbial Contaminants in Drinking Water, Koch
Membrane Systems, Inc. HF-82-35-PMPW™ Ultrafiltration Membrane. EPA/600/R-06/100.
122
------- |