United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Cincinnati OH 45268
•¥
Research and Development
EPA-600/S2-81-080 July 1981
Project Summary
Evaluation of
Maintenance for Fugitive
VOC Emissions Control
G J. Langley and R. G. Wetherold
The U.S. EPA Office of Air Quality
Planning and Standards (OAQPS) has
the responsibility for formulating
regulations for the control of fugitive
emissions of volatile organic com-
pounds (VOC). "Fugitive emissions"
generally refer to the diffuse release of
vaporized hydrocarbon or other organic
compounds. Fugitive emissions origi-
nate from equipment leaks as well as
large and/or diffuse sources. The study
reported here was undertaken by the
Office of Research and Development
to assist OAQPS in the development
of regulations.
The project was designed to quantify
the effectiveness of routine (on-line)
maintenance in the reduction of fugi-
tive VOC emissions from in-line valves.
An overall emission reduction of ap-
proximately 70% was achieved by
tightening the bolts on the valve
packing gland. This level of control
was sustained for up to about six
months. The rates of leak occurrence
and recurrence were also evaluated,
as well as the time required to conduct
the on-line maintenance.
This Project Summary was devel-
oped by EPA's Industrial Environ-
mental Research Laboratory, Cin-
cinnati, OH, to announce key findings
of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
The U.S. EPA Office of Air Quality
Planning and Standards (OAQPS) is
currently in the process of formulating
regulations for the control of fugitive
emissions of volatile organic compounds
(VOC). The study reported here was
undertaken by the Office of Research
and Development to assist OAQPS in
this effort. The project was intended to
develop data to determine the effective-
ness of routine (on-line) maintenance in
the reduction of fugitive VOC emissions
from in-line valves. The overall effec-
tiveness of an inspection/maintenance
program was examined by studying:
• immediate emission reduction due
to maintenance,
• the propagation of leaks after
maintenance, and
• the rate at which new leaks occur,
for both pumps and valves.
This study was conducted by the
Radian Corporation (Austin, Texas)
under contract to EPA (Contract No. 68-
03-2776-04). The project began in 1979
when the scope and technical approach
were developed through several meet-
ings with the Chemical Manufacturers
Association and the Texas Chemical
Council, as well as with individual
chemical companies. The field work
was completed in 1980.
The experimental design called for
the study of three types of organic
chemical production units at each of
-------
two locations. The processes chosen
were: (1) ethylene production, (2) cumene
production from benzene and propylene;
and (3) vinyl acetate production. The
processes represented a wide range of
conditions found in organic chemical
manufacturing plants. Ethylene was
chosen because typically these units
are large and widespread, operate with
a wide range of process conditions, and
handle very volatile materials. Cumene
was of interest because this type of unit
(one using the reaction of benzene and
propylene) handles a hazardous air
pollutant, benzene. Production of vinyl
acetate from the reaction of ethylene
and acetic acid was chosen because
some of the process streams are corro-
sive.
The sampling and analytical proce-
dures are described in the Project
Report. The details of the methods and
statistical techniques for data analysis
can also be found in the Project Report.
The sampling techniques included both
"screening" and the actual measure-
ment of hydrocarbon emission rates.
Screening was done with a Century
Systems Corporation OVA-108 and a
Bacharach TLV Sniffer. The valves were
screened by traversing 360 degrees
around the stem seal and the seam
where the packing gland merges with
the valve bonnet. The point of maximum
concentration was identified. The OVA-
108 was used before the TLV to quickly
identify the area of maximum concen-
tration because of its faster response
time. The sample probes were placed as
close to the maximum leak as possible.
The recorded screening value was the
highest reading obtained twice during
an interval of about one minute. Pumps
were screened at the outer shaft seals
First Visit to a Unit
by completely traversing 360 degree:
with the OVA to locate the point 01
maximum concentration. Occasionally
a 12-inch Teflon extension was added tc
the OVA probe in order to extend pas)
safety screens on vertical pumps.
Selected valves and pumps were
sampled to determine the massemission
rate using the flow-through method
described in the Project Report. The
general sampling procedure was: (1) the
source was screened with the OVA-108
and TLV Sniffer and the values and time
of day were recorded; (2) the source was
tented with Mylar® and duct tape and
sampled; (3) the tent was removed and
the source was rescreened as in the first
step, above. The sequence of screening
and sampling are illustrated in Figure 1.
Valve maintenance consisted of
tightening packing glands while moni-
toring the leak. The term "directed
/
c
1
f?\
nitial
~)VA Reading
Leak Rate
Measurement
OVA Reading OV>
1 I
f")\ (^\
1 Reading C
Directed
Maintenance
VA Reading C
\ \
r7\ ici
VA Reading
Leak Rate
Measurement
OVA Reading
\ ,
ft\
time — 0
Second Visit (and Third Visit) to a Unit
OVA Reading
1
(z)
Leak Rate
Measurement
OVA Reading
1
®
7 Initial before maintenance OVA reading.
2 After tenting, before maintenance OVA reading.
3 Before maintenance OVA reading—the screening value obtained immediately before maintenance.
4 After maintenance, 1st OVA reading—the screening value obtained immediately after maintenance.
5 Before tenting, after maintenance OVA reading.
6 After tenting, after maintenance OVA reading.
7 Before tenting OVA reading.
8 After tenting OVA reading.
Figure 1. Sequence of emissions measurements and screening values.
2
-------
maintenance" refers to this procedure
when a hydrocarbon detector is used
during the maintenance activity. The
leak is monitored with the instrument
until no further reduction of screening
values is observed or until the valve
stem rotation starts to be restricted. The
type of maintenance personnel per-
forming the repairs depended upon the
type of valves that were to be maintained.
Control valves required instrument
personnel experienced with the process
unit and with the precautions necessary
to safely maintain operations while
repairing control valves. Block valves
were maintained by regular mainte-
nance personnel such as pipefitters or
boilermakers.
The maintenance procedures gener-
ally consisted of first screening the
valve with the OVA-108 and recording
the value. The packing gland nuts were
then tightened a little at a time while
monitoring the leak with the OVA.
Tightening was continued to the point of
either minimizing the leak, causing the
stem movement to tighten or grab, or
reaching the bottom of the packing
bolts. The valve was then operated, if
the process permitted it, and rescreened.
If the leak remained or worsened, the
packing was further tightened until the
limits described above were reached. In
no case was the packing tightened such
that the operation of the valve was
impaired.
Certain valves could not be maintained
due to their locations in the process
stream. These were in critical service
where sticking, jamming, or breaking of
the valve might precipitate a unit shut-
down.
Results and Conclusions
Three aspects of the effect of valve
maintenance on fugitive emissions
were studied:
• the immediate effect based on
measured leak rates,
• the long term effect based on
measured leak rates, and
• the immediate effect based on
screening values.
Analysis of the immediate effect of
maintenance using measured leak rates
produced an overall estimate of 71.3%
reduction in fugitive emissions (95%
confidence limits of 54% to 88%) im-
mediately after maintenance. This
estimate is the weighted percent reduc-
tion (WPR), calculated by:
WPR =
(f Mass Emissions
Before Ma int.
-FMass Emissions
After Maint.)
FMass Emissions
Before Maint.
where m is the number of valves
maintained.
Paired observations of measured leak
rates were available for 1 55 attempts at
maintenance. Weighted percent reduc-
tions were calculated for various
groupings of the 1 55 attempts at main-
tenance, and are given in Table 1. A
graphical presentation is given in Figure
2. Since none of the WPR estimates are
statistically different for any of the
groupings, the overall estimate is the
most appropriate estimate for application
in other organic chemical units.
It was also of interest to investigate
the change in the WPR estimates for
varying screening action levels. When
only valves with the immediately before
maintenance screening values > 10,000
ppmv were considered, the WPR esti-
mate decreased slightly to 70.1% with
the 95% confidence interval of 46% to
95%. This estimate is almost identical to
the overall estimate of 71.3%. The WPR
estimate for those sources where the
before maintenance OVA reading was
<10,000 was 82.4% with an approxi-
mate 95% confidence interval of 57% to
100%. These two WPR estimates are
not statistically different.
Table 1. Immediate Effect of Valve Maintenance
All Cases
Cumene Units
Ethylene Units
Vinyl Acetate Units
Gas Service Valves
Liquid Service Valves
Number of
Attempts at
Maintenance
155
54
69
32
71
84
Weighted
Percent Reduction 95% Confidence
(WPR) Limits for WPR
71.3
81.6
56.6
72.9
84.5
42.0
(54, 88)
(67. 96)
(22, 91)
(34, 100)
(74, 95)
(0.4, 84)
wo
90
t 70
3
(g 60
S 50
a
£ 40
| 30
I 20
;o
o
i
All Valves \ Ethylene \ Gas Service
Cumene Vinyl Acetate Liquid Service
Bracketed intervals are 95% confidence intervals.
Figure 2. Immediate effect of valve maintenance.
3
-------
Later sampling of the maintained
sources to study the long-term effect of
maintenance indicated that the re-
duction estimates obtained for im-
mediate effects of maintenance held for
the length of the study(up to six months).
To put the long-term effect of main-
tenance into perspective, it is helpful to
compare the emissions from the main-
tained valves to those from a control
group of unmaintained valves over a
period of time. Figure 3 is a graphic
display of this comparison. The major
conclusion that can be drawn from
Figure 3 is that the immediate effect of
maintenance discussed previously was
sustained for the duration of the project.
The minor changes in the control group
and the maintenance group after the
initial sampling visit (during which the
maintenance occurred) are not statisti-
cally significant.
The immediate effect of maintenance
can also be viewed in terms of screening
values only. This can be used to evaluate
a leak detection and repair strategy,
where a leak is defined by a screening
value greater than some specified value.
For example, given a definition of a leak
as a screening value greater than or
equal to 10,000 ppmv, the effectiveness
of maintenance was evaluated. Analysis
of the immediate effect of maintenance
based on screening data and using
>10,000 ppmv as the definition of a leak
produced an estimate of about 30%
reduction in the number of leaking
sources as a result of the maintenance.
This indicates that about 70% of the
leaking sources could not be repaired
(where repaired is defined as screening
OO.OOOppmv after maintenance).
However, it should be pointed out that
even though the screening values were
reduced to below 10,000 ppmv for only
30% of the valves, this corresponded to
a 70% reduction in mass emissions.
To study the recurrence of leaks after
maintenance, data from the 155 main-
tenance attempts were examined. For
this analysis, only those valves which
screened >10,000 ppmv immediately
before maintenance and screened
<10,000 ppmv immediately after main-
tenance were considered as having a
potential to recur. This eliminated all but
28 cases from the analysis. Of these 28
cases, eight were seen to recur (i.e.,
screen >10,000 ppmv at some time
following the after-maintenance screen-
ing). Of the eight valves whose leaks
recurred, four recurred within a few
days after maintenance. The other four
The sample size for the maintained group is 43
The sample size for the control group is 13
Controls
0.128
0 70-1
0.09-
<^>
jt o.os -
30.07-
Q
^ 0.06-
^0.05-
§ 0.04-
<5
0.03-
' 0.02-
0.01 -
0.00
0.034
Maintained
Upper 95%
Confidence t
Interval
A verage
0.064
liiiiiii
ii
••
ii
ii
••
••
ii
••
ii
••
ii
ii
0.053
Initial
Samples
After
Maintenance
Median Median Median Median
Day Day Day Day
73 68 211 151
After Maintenance After Maintenance
Figure 3. Long-term effect of maintenance vs. control group.
recurrences were spread over the study
period (up to 7 months). Because of the
two distinct groupings of recurrences
over time, a mixed-model was used in
estimating the recurrence rate. This
mixed model consists of a uniform
distribution for recurrence times within
five days after repair and an exponential
distribution for recurrence times greater
than five days after repair.
A graphical presentation of the
modeled percentages for recurrence
along with an approximate 95% confi-
dence region is given in Figure 4. The
empirical distribution function (actual
data) is indicated by the dotted line.
In Table 2, 30-day, 90-day, and 180-
day recurrence rate estimates are given
along with their approximate 95% con-
fidence limits.
The rate of occurrence of leaks was
studied using pumps and valves that
initially screened at <10,000 ppmv and
were not maintained during the project.
An exponential model was used to
approximate the actual distribution of
the time to first occurrence of a leak
(screening value >10,000 ppmv). This
model is widely used to summarize data
similar in nature to leak occurrences if
the assumption can be made that the
occurrence rate remains constant. A
major advantage of the exponential
model over other statistical distribution
models is that a single parameter fully
and completely describes a given expo-
nential distribution. The results for
various groupings of equipment are
given in the Project Report. In compar-
ison to recurrence rates, the occurrence
rates are much lower. Also, pump seals
have a statistically significant higher
rate of occurrence than valves.
Example plots of the cumulative
distribution functions are shown in
Figure 5. The predicted occurrence rate
for periods up to eight months can be
obtained directly from these curves. The
fact that the curves are not straight lines
is a consequence of the effectively
decreasing population size, since
sources which begin leaking are no
longer included in the population. It
should be kept in mind that the under-
lying occurrence rates are always as-
sumed to be. constant, however. As a
check on the model, the observed and
predicted occurrence rates were com-
pared. This is shown in Figure 6.
Finally, to aid in assessing the costs of
valve maintenance, the total time (in
minutes) associated with maintenance
was recorded each time that a series of
valves was maintained. The maintenance
time ranged from 3.7 minutes per valve
to 28.7 minutes per valve with an
4
-------
70 -
240
Note. Dotted Line is actual data.
Dashed lines indicate a 95% confidence region.
Figure 4. Recurrence rate estimate vs. empirical distribution function.
Table 2.
Valve Leak Recurrence Rate Estimates
Recurrence Rate Estimate
95% Confidence Limits on the
Recurrence Rate Estimates
30-day
90 -day
180-day
17.2%
23.9%
32.9%
(5,37)
(7,48)
(10,61)
average of 9.6 minutes per valve (95%
confidence interval for average: 8.6 to
10.6 minutes per valve). These data
indicate that ten minutes per valve
would be a reasonable maintenance
time to use m assessing costs of valve
maintenance.
The activities included in maintenance
for this study were restricted to tight-
ening packing gland bolts to compress
the packing material around the valve
stem and seat while the valve was in
service. This operation is a simple on-
line maintenance procedure.
Other on-line maintenance procedures
could have been used and are currently
practiced in industrial plants. Although
some of these other methods may be
more time consuming, they have been
demonstrated as effective. Some valves
have lubricated packing and are equip-
ped with fittings to inject lubricant into
the packing gland while the valve is in
service. There are also valves equipped
with backseating capabilities which
allow replacement of the packing with-
out dismantling the valve. Also available
are commercial leak sealing services
which can inject sealant into a valve to
seal the leak while the valve is in
service. Finally, some process units
have piping configurations which allow
bypass or isolation of a valve for re-
packing while it is in place, even though
it is not on-line.
-------
Occurrence Rate Estimate for All Valves
30 60 90 120 150 180 210 240
50
40
3
$30
c
§ 20
10
Occurrence Rate Estimate for A/1 Pumps
0
30 60
90 120 150 180. 210 240
Days
Note: Dashed lines indicate 95% confidence region.
Figure 5. Overall occurrence rate estimates.
-------
All Valves—Model vs. Empirical CDF
25
90 120 150 180 210 240
All Pumps—Model vs. Empirical CDF
0
30
60
120 150
Days
Note: Dotted line is plot of actual data.
Figure 6. Occurrence rate estimate vs. empirical CDF,
180
210
240
G. J. Langley and Ft. G. Wetherold are with Radian Corporation, Austin, TX 78766.
Robert C. Weber is the EPA Project Officer (see below).
The complete report, entitled "Evaluation of Maintenance for Fugitive VOC
Emissions Control," {Order No. PB 81-206 005; Cost: $18.50, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Cincinnati, OH 45268
6 U.S. GOVERNMENT PRINTING OFFICE, 1981 -757-012/7230
-------
United States Center for Environmental Research
Environmental Protection Information
Agency Cmcmnat, OH 45268
Agency
EPA 335
Official Business
Penalty for Private Use $300
RETURN POSTAGE GUARANTEED
Third-Class
Bulk Rate
ST
TL
-------
&EPA
United States
Environmental Protection
Agency
Research and Development
r,
.-*
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600/S2-81-077,078:079 July 1981
Project Summary
Removing Trace Organics
From Drinking Water Using
Activated Carbon and
Polymeric Adsorbents
C. S. Oulman, V. L. Snoeyink, J. T. O'Connor, and M. J. Taras
"Bench-Scale Evaluation of Resins
and Activated Carbons for Water
Purification." by V.L. Snoeyink, W.A.
Chudyk, D.O. Beckmann, P.M. Boening,
and T.J. Temperly. In the first of a
three-volume study, adsorption iso-
therms and bench-scale column studies
were used to com pare the performance
of five types of commercially available
activated carbons and four types of
resins for removing humic acids, f ulvic
acids, 2-methylisoborneol (MIB), and
chloroform from water. For adsorbing
humic materials, some of the activated
carbons and the weak base phenol-
formaldehyde resins performed satis-
factorily. The same activated carbons
provided satisfactory removal of MIB,
although the capacity was reduced
somewhat in the presence of humic
acid. The carbonaceous resin and one
of the activated carbons has about the
same capacity for chloroform removal
at concentrations under 0.5 mg/L.
The presence of 10 mg/L of humic
acid had little effect on their capacity
for adsorbing chloroform.
"The Removal of Trace Organics
from Drinking Water Using Activated
Carbon and Polymeric Adsorbents,"
by J.T. O'Connor, D. Badorek, and L.
Thiem. In the second volume, a pilot
plant was operated at the Kansas City,
Missouri, Water Treatment Plant to
study adsorption as a means of remov-
ing trihalomethanes (TTHM) and total
organic carbon (TOC) from drinking
water. The pilot plant consisted of 15
columns, 15 cm (6 in.) in diameter,
and each containing about a 0.9-m (3-
ft) depth of a granular adsorbent.
Granular activated carbons and poly-
meric adsorbents were compared in
four extended tests conducted over
periods of 183, 111, 65, and 129
days. The pilot-plant studies demon-
strated the effects of regeneration,
variations in trace organic concentra-
tion, and depth of adsorbent on trace
organic removal including effluent
concentration and adsorption capacity.
"Trace Organics Removal Using
Activated Carbon and Polymeric Ad-
sorbents," by C.S. Oulman. In the
third volume, a survey was made to
determine the trace organic matter in
raw and treated water from 14 water
utilities across the United States.
Monthly analyses were made for TTHM
and TOC. Analyses were made on
carbon/resin adsorbable ether extracts
from each utility for a number of
indicator compounds and for bacterial
mutagenicity as measured by the
Ames test. The results of the water
quality survey indicated that most of
the water utilities are able to produce
an acceptable finished water with
conventional treatment methods. In
those places where additional treat-
ment is needed for trace organics
removal, activated carbon will probably
-------
be the more versatile adsorbent to use.
The results from this effort have been
published by Glatz, et al. in the Journal
American Water Works Association.
70(8):465-468, 1978.
EPA did not participate in this por-
tion of the overall project, but some of
the results have been included in the
full report. This third volume also
contains an executive summary of
results obtained from the first and
second volumes. Only the bench-scale
activities (first volume) and the pilot-
scale adsorption studies (second vol-
ume) are discussed in this Project
Summary.
This Project Summary was devel-
oped by EPA's Municipal Environ-
mental Research Laboratory, Cincin-
nati, OH, Jo announce key findings of
the research project that is fully docu-
mented in three separate reports fsee
Project Report ordering information at
back).
Introduction
In 1975, the American Water Works
Association (AWWA) listed 15 high-
priority problems needing study. One
problem was called "Reliable Screening
Tests and Techniques for Determining
an Evaluation of Organics in Drinking
Water," and another was "Method for
Removal of Organics in Drinking Water."
Likewise, the U.S. Environmental Pro-
tection Agency (EPA) was concerned
that many organic compounds were not
being removed by conventional water
treatment practice; and further, that
chlorine used for disinfection was
shown to produce halogenated by-
products. A cooperative research effort
was initiated to (1) examine the oc-
currence of trace organics in drinking
water; and (2) evaluate the efficiency of
their removal by a "broad spectrum"
adsorbent, granular activated carbon,
and more selective adsorbents—poly-
meric resins.
The project was divided into three
parts and reported separately. Bench-
scale studies (first volume) were con-
ducted at the University of Illinois to
select the adsorbents for pilot-scale
column use. In the second volume,
adsorption columns were designed,
fabricated, and operated by personnel
from the University of Missouri—Co-
lumbia and located at the Kansas City,
Missouri, water treatment plant The
third volume of the project was a survey
of trace organics in 14 different water
utilities across the United States. That
work was supported by the AWWA
Research Foundation and its partici-
pating members and conducted by the
Ames Laboratory at Iowa State Uni-
versity. The results and conclusions
from the first and second volumes will
be discussed individually in this Project
Summary.
A major objective of the overall study
was an evaluation of the removal of
taste-and-odor-producing compounds
as measured by threshold odor number
(TON). The influent TON values were,
however, consistently low, which made
clear-cut evaluations of the removal of
odor-producing compounds inconclusive.
Bench-Scale Studies
(first volume)
Bench-scale studies were made to
determine which of the various com-
mercially available adsorbents should
be used in the side-by-side comparison
pilot-scale study of activated carbon and
polymeric adsorbents. Adsorption iso-
therms, using water-containing chloro-
form, humic and f ulvic acids, and MIB as
solutes, were determined on five types
of activated carbons and four different
resins.
The macroporous, phenol-formalde-
hyde, weak-base resin had a high
capacity for humic substances and
could be regenerated with sodium
hydroxide but was not able to remove
the earthy-musty odor compound, MIB.
The styrene-divinyl benzene resin did
not adsorb humic substances, but it did
have some capacity for MIB. The acrylic
and carbonaceous resins did not adsorb
humic materials or MIB, butthecarbon-
aceous resin had an excellent capacity
for chloroform. The activated carbons
could remove the humic substances and
the MIB but had a relatively small
capacity for chloroform.
A number of adsorbents were recom-
mended for use in the pilot-plant tests,
based on the adsorption isotherms and
the mini-column studies. Initially, one
carbon, Westvaco Nuchar® WVG,* was
recommended for a side-by-side com-
parison with polymeric adsorbents
because of its good capacity for both
humic substances (27.6 mg/g**) and
MIB (112.6 mg/g). Later, other carbons
were selected for inclusion in the pilbt
plant tests.
No one polymeric adsorbent could be(
recommended as having the capacity to
remove organic matter in such a wide
range of molecular weights as did any of
the activated carbons. Therefore, two
materials were selected to be used in
tandem — an adsorbent for high molecu-
lar weight compounds such as humic
acids and an adsorbent for low molecu-
lar weight compounds such as MIB and
chloroform. Diamond Shamrock ES-
561 was recommended for the humic
acid removal application because of its
reasonably low swelling properties and
fairly high capacity for humic acids in
the neutral pH range. This was not one
of the polymeric adsorbents evaluated
in the bench-scale tests, but its proper-
ties are similar to one of the phenol-
formaldehyde resins tested. Rohm and
Haas Ambersorb® XE340 was the other
polymeric adsorbent tested; it has a high
capacity (18.2 mg/g) for chloroform
removal. In the later phases of the pilot-
plant experimentation, however, addi-
tional resins were selected for study.
Pilot-Scale Adsorption Studies
(second volume)
Results
Fifteen glass columns, each 1 5 cm in
diameter and containing approximately
0.9 m of adsorbent, were located at the
Kansas City, Missouri, water treatment
plant and operated in a post-filtration
mode. Table 1 indicates the way the
various columns were loaded during
each phase of the study. The average
concentrations of TTHM and TOC were
42 fjg/L and 2.5 mg/L, respectively.
Phase I
Over an initial period (Phase I) of 133
days, a 0.9-m bed of bituminous-base
granular activated carbon (Nuchar®
WVG) removed 70%* of the influent
TTHM. Approximately 2.7 m of Nuchar®
WVG were required to remove 99% of
the influent TTHM. A 0.9-m-deep bed of
a carbonaceous resin Ambersorb® XE-
340 provided 98% TTHM removal during
the same period. Steaming the Nuchar®
WVG increased TTHM removal to 87%,
whereas steaming the weak-base anion
exchange resin (Diamond Shamrock
ES-561) had no significant measurable
effect on an initially low removal ef-
ficiency. The periodic steaming of the
'Mention of commercial products does not con-
stitute approval or endorsement by EPA.
"Isotherm capacities reported for an equilibrium
concentration of 1,000 /jg/L.
*AII removal efficiencies are based on influent and
effluent concentrations averaged over the project
phase period
-------
Table 1. Identification of Adsorbents Utilized in Kansas City, Missouri Pilot
Plant Evaluation of Removal of Organic Substances from Drinking
Water
Column
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Phase 1
1(1 33 days)
Feb.-Aug. '77
WVGM
WVGM
WVGM
ES-561M
ES-561M
XE-340M**
—
—
—
—
—
—
—
XE-340M
EX-561M**
Phase II
11(111 days)
Aug. -Dec. '77
WVGM
WVGM
WVGM
ES-561 nm
ES-561, 1R)
XE-340nn)**
HD-1030M
HD-1030M
HD-1030M
LCKM
C-THMM
ROWO.SM
Sand
WVGM**
ES-561 am**
Phase III
III (65 days)
Mar-June '78
WVGnm
WVGnm
WVGnm
XE-340vm**
IRA-904M
WVGM
HD-1030nm
HD-1030nm
HD-1030nm
A-162M
WVGM
HD-1030M
Sand
LCKM
WVGM
Phase IV
IV (129 days)
July-Nov. '78
WVG(2m
WVGun
WVG(2m
XE-340l3m*
IRA-904nm*
WVGM*
HD- 1030 urn
HD-1030(2n
HD-1030(2m
WVGnm
WVGM
HD-1030M
HD-1030M
WVGM
WVGM**
Empty Bed Contact Time:
[11.2 min @ 2 gpm/sf
*2.2 min @ 10 gpm/sf
WVG Bituminous Base Carbon - Westvaco
HD-1030 Lignite Base Carbon - ICI
LCK Petroleum Base Carbon - Union Carbide
C- THM Bituminous Base Carbon developed by Calgon for enhanced removal
of trihalomethanes.
ROW 0.8 Extruded Peat Base Carbon - American Nor it
ES-561 Weak Base Anion Exchange Resin - Diamond Shamrock
XE-340 Carbonaceous Resin - Rohm and Haas
IRA-904 Strong Base Anion Exchange Resin - Rohm and Haas
A-162 Strong Base Anion Exchange Resin - Diamond Shamrock
[**Adsorbents subjected to steaming
(v) Virgin adsorbent
(1R) Once regenerated
(2R) Twice regenerated
(3R) Thrice regenerated
Nuchar® WVG reduced the number of
microorganisms recovered from the
adsorbent from 89,000 to 5,800 colo-
nies/gram.
Throughout Phase I, the influent con-
centration of TTHM steadily increased
from less than 10 /jg/L (February) to
over 80 fjg/L TTHM (June) as influent
water temperature increased. An im-
portant outgrowth of the present study
was the establishment of the seasonal
pattern of TTHM formation in the finished
water at Kansas City. This facilitated
subsequent decisions as to when virgin
and regenerated adsorbents should be
placed in service. TOC measurements
were included in the sampling protocol
near the end of Phase I.
Phase II
During Phase II, all 15 columns were in
operation; this permitted the perfor-
mance of the carbons made from bitu-
minous coal, lignite, peat, and petroleum
to be compared. TTHM levels had in-
creased to a peak of approximately 200
fjg/L in the late summer, establishing a
pattern that was to be repeated in the
following year. This maximum TTHM
level provided a more significant chal-
lenge for the adsorbent than was present
in Phase I. Over the 111 days of Phase II
operation, the TTHM removals were
comparable to those observed during
the first 111 days of Phase I with the
following removals: bituminous, 80%;
lignite, 83%; petroleum, 82%; bitumi-
nous base carbon enhanced for TTHM
removal, 85%; and extruded peat base
carbon, 68%. Steaming of the Nuchar®
WVG column again increased TTHM
removal. The steamed column removed
92% of the TTHM over a period of 111
days, exactly equaling the percent
removal observed over the first 111 -day
period of Phase I.
Once again, the Ambersorb® XE-340
was effective in removing 90% of the
influent TTHM. Because the Diamond
Shamrock ES-561 continued to be
erratic and generally ineffective, it was
eliminated from further pilot-plant
testing. The 0.9-m-deep beds of granu-
lar activated carbon removals of TOC
were bituminous, 51%; lignite, 37%;
petroleum, 30%; bituminous carbon
enhanced for TTHM removal, 19%; and
extruded peat base carbon, 45%. It
became evident that there were far
greater differences in carbon perform-
ances with TOC than with the removal
of the small amounts of TTHM in the
influent. Moreover, the Calgon carbon
(Filtrasorb® C) developed to enhance
TTHM removal did achieve superior
removal of TTHM but at the expense of
reduced TOC removal capability. Ap-
parently, the pore size distribution that
results in more effective TTHM removal
retards the removal of a range of other,
larger adsorbates. Steaming of the
Nuchar® WVG column appeared to in-
crease TOC removal modestly to 56%.
Perhaps most significant was that the
2.7-m depth of Nuchar® WVG was in-
capable of removing more than 75% of
the influent TOC at the 5 m/hr (2
gpm/ft2) application rate.
Neither Diamond Shamrock ES-561
nor Ambersorb® XE-340 showed any
significant TOC removal. At the end of
Phase II, the Nuchar® WVG and Hydro-
darco® 1030 were returned to the
respective manufacturers for thermal
reactivation to prepare for a study of the
effect of reactivation on adsorbent per-
formance (Phase 111).
-------
Phase III
Reactivating the carbons restored their
virgin adsorption capacity, as measured
by Iodine Number and Decolorizing Index.
During Phase II of the study, comparison
of TOC removal indicated similar
performance between 0.9-m-deep beds
of Nuchar® WVG, whether it was a
once-reactivated (60%), virgin (62%), or
regularly backwashed virgin (61 %)
product. Once-reactivated and virgin
lignite base carbon (Hydrodarco® 1030)
removed 43% and 51% of the TOC,
respectively, over the period of Phase III.
No steaming was done during Phase
III. Since the influent concentrations of
TTHM were generally less than 5 /jg/L,
little information on TTHM removal was
obtained. Thus, Phase III was terminated
after only 65 days so that Phase IV could
be initiated immediately before the time
when the maximum TTHM influent
concentration was expected, based on
the previous year's seasonal pattern.
The Ambersorb® XE-340 was again
ineffective in TOC removal during Phase
III, whereas a 0.9-m-deep bed of a
strong-base anion exchange resin (Rohm
and Haas Amberlite® IRA-904) appeared
to remove roughly one-third of the
influent TOC. Another bed of a different
strong-base anion exchange resin (Dia-
mond Shamrock A-162) removed 47%
of the influent TOC. On a weight basis.
Diamond Shamrock A-162 was only
one-half as effective as Nuchar® WVG
activated carbon.
The Ambersorb® XE-340 was placed
first in the series of three columns to
observe the effect of high molecular
weight components of TOC on the ad-
sorption of TTHM by the resin. Laboratory
studies had indicated that high molecu-
lar weight organic substances might be
irreversibly adsorbed to the carbonace-
ous resin leading to "fouling" and loss
of TTHM removal capability. Since this
resin is still under development and
evaluation, the manufacturer was un-
certain of the appropriate regeneration
procedure. After subsequent EPA stu-
dies, a far more vigorous steaming
procedure is now being recommended
for the Ambersorb® XE-340 than was
recommended at the time of the present
study.
Large numbers of microorganisms
were dislodged from the adsorbents at
the end of Phase III in June 1978. The
inert sand media and the resins harbored
few microorganisms whereas the acti-
vated carbon supported significant
growth, particularly in those columns
that had the most TOC removed. Back-
washing of a column containing virgin
Nuchar® WVG resulted in more than an
order of magnitude reduction in bacterial
count. For example, 850,000 colonies/
gram were found on the activated
carbon in the undisturbed column,
compared with 52,000 colonies/gram
on the backwashed activated carbon,
indicating the effectiveness of back-
wash in controlling accumulations of
organisms.
Phase IV
Nuchar® WVG and Hydrodarco® 1030
were both reactivated to levels beyond
their virgin capacities by the manu-
facturers in preparation for Phase IV.
This resulted in the following TTHM
removals by 0.9-m-deep beds of Nuchar®
WVG: twice-reactivated, 75%; once-
reactivated, 67%; virgin, 66%; and virgin
(replicated), 67%. The regularly steamed
column of virgin Nuchar® WVG removed
96% of the influent TTHM overthe 129-
day period. The twice-reactivated, once-
reactivated, and virgin Hydrodarco®
removed 77%, 72%, and 62% of the
influent TTHM. Again, the reactivation
of this activated carbon beyond its virgin
capacity resulted in slightly enhanced
TTHM removal.
TOC removals were marginal, as
before. The 0.9-m-deep beds of Nuchar®
WVG removals were twice-reactivated,
55%; once-reactivated, 50%; virgin,
56%; and virgin (replicated), 53%. Steam-
ing increased the TOC removal of the
virgin Nuchar® WVG to 63%. Approxi-
mately 2.7 m of the twice-reactivated
Nuchar® WVG removed a total of 78%
of the TOC from fairly constant influent
levels of 2 mg/L. The 0.9-m-deep beds
of Hydrodarco® 1030, twice-reactivated,
once-reactivated, and virgin removed
39%, 38%, and 36% of the influent TOC.
Approximately 2.7 m of the Hydrodarco®
1030 removed just under 58% of the
TOC.
The Ambersorb® XE-340 was regen-
erated after each of the first three
phases with 1 -1 /2-bed volumes of low
pressure (12 psig) steam. Despite this
procedure, the performance of the
Ambersorb® XE-340 declined. In Phase
IV, at the higher (20 m/hr) application
rate, Ambersorb® XE-340 removed only
54% of the influent TTHM and 5% of the
influent TOC. As observed a year earlier,
the cool (November) water temperatures
suppressed organism growth on all of
the adsorbents.
Overall, the influent adsorbate con-|
centrations and column removals were
consistent between replicates, in suc-
cessive phases of operation, and with
successive regenerations. Granular
activated carbon exhibited the potential
for prolonged removal of both TTHM and
TOC.
Conclusions
1. In studies using pilot plant adsorp-
tion columns to adsorb halogenated
organic substances from softened,
filtered Kansas City water, acti-
vated cabon and carbonaceous
resin were able to remove TTHM
for extended periods. Conversely,
strong and weak-base anion ex-
change resins were not able to
remove the TTHM. Onlythe granu-
lar activated carbons were effective
in removing significant amounts
of the TOC present. Periodic steam-
ing of the activated carbon columns
reduced bacterial growth and en-
hanced TTHM and TOC removal.
Regular backwashing of the acti-
vated carbon columns was also
effective in reducing the accumu-
lation of bacterial growth.
2- Little or no difference was observed
in removals obtained by twice-
reactivated, once-reactivated, and
virgin activated carbons, indicating
that calcium carbonate deposits
from lime-softened water did not
coat the adsorbent and impair its
adsorption capacity.
The full three-volume report was
submitted in fulfillment of Grant No. R-
804433 by the University of Illinois,
University of Missouri—Columbia, and
Iowa State University, under the spon-
sorship of the U.S. Environmental Pro-
tection Agency.
4
-------
C. S. Oulman, V. L Snoeyink, J, T. O'Connor, and M. J. Taras are with Iowa State
University, Ames, IA 50010; University of Illinois, Urbana, 1L 61801; Univer-
sity of Missouri, Columbia, MO 65211; and A WWA Research Foundation,
Denver, CO 80235, respectively.
Thomas Love, Jr. is the EPA Project Officer (see below).
The complete reports, entitled:
"The Removal of Trace Organics from Drinking Water Using Activated Carbon
and Polymeric Adsorbents," (Order No. PB 81-196 768; Cost:$11.00)
"Bench-Scale Evaluation nf Resins and Activated Carbons for Water Purifica-
tion," (Order No. PB 81-196 776; Cost: $8.00)
"Trace Organics Removal Using Activated Carbon and Polymeric Adsorbents,"
(Order No. PB 81-196 784; Cost: $8.00)
The above reports will be available only from: (prices are subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati. OH 45268
i US GOVERNMENT PRINTING OFFICE 1981-757-012/7ZOZ
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use S300
RETURN POSTAGE GUARANTEED
------- |