Bloomington Water Treatment Plant Operations Manual
Filter Surveillance and Operation
June, 2002
H2O’C Engineering
2401 Tahoe Court
Columbia, MO 65203-1444
877-22-WATER
www.h2oc.com
Bloomington Water Treatment Plant Operations Manual
Filter Surveillance and Operation
Table of Contents
Introduction 3
About the Filters 3
Filter Operations 5
Monitoring 6
IEPA Turbidity Requirements 7
What to Do About High Turbidities 7
Evaluation of Filters during GAC Replacement 8
GAC Size Distribution and Characteristics 9
TOC Removal by GAC and through Plant 10
Virgin GAC Replacement 12
Sand Size Distribution and Characteristics 13
Recommended Filter Washing Protocol 15
Microbial Growth on Filtration Media 17
Particles Washed from GAC 18
Particles Washed from Sand 19
Figures
Filter Media Depths - New Filter; Old Filters 4
GAC Grain Size Distribution: New versus Old 9
TOC Removal by GAC and through Plant 11
Sand Grain Size Distribution: New versus Old 13
2
Introduction
This manual is intended to give an overview of filter operation and surveillance. It should be noted
that, as part of a water treatment system, filtration is strongly related to previous steps in the treatment
process and cannot be looked at completely independently. Ineffective pretreatment, including
coagulation and sedimentation or post-precipitation of calcium carbonate, can result in ineffective
filtration.
Filtration is often considered the particle removal process in a treatment process. However, only a
small fraction of the particles entering or produced during water treatment are actually removed in the
filters. By far, most of the particles originating in Bloomington’s lake waters are removed in the clarifiers.
The filters act largely as effluent polishers. However, as of January 2002, new federal and state
regulations have caused utilities to focus more attention on the performance of individual filters.
About the Filters
In total, there are 18 dual media (GAC/sand) filters installed at the Bloomington plant.
The twelve filters in the ‘old plant’ (Annex, 1929) were constructed in three groups of four that went
into service in 1929, 1956 and 1966, respectively. Each filter box contains 19 inches of granular activated
carbon (GAC) overlying a 12 inch layer of silica sand. Used primarily for taste-and-odor control, the GAC
is replaced with new, virgin carbon on a three year schedule. Four filters are serviced each year.
At the time of carbon replacement, the height of the sand interface is measured in each bed. Since
granular filters tend to lose 5 to 7 % of their media each year, make-up sand is added to restore the
original sand bed height of 12 inches.
3
Rated at 1.5 gpm/ft2 (a downward flow velocity of 3.8 meters per hour), the ‘old plant’ filters are
nominally operated around 0.6 gpm/ft2 (1.5 m/h) This results in a substantial empty bed contact time
(EBCT) of 19 minutes within the 19-inch layer of GAC.
The six filters in the Main Building (‘new plant’) were constructed in 1994. Each filter consists of two
separate filter boxes served by a single influent and wash water gullet. A 24 inch layer of GAC overlies a
12 inch layer of filter sand. Rated at 3 gpm/ft2 (8 m/h), these filters are normally operated around 2
gpm/ft2 (5 m/h). For 24 inches (0.61 m) of GAC, the EBCT is 7.3 minutes. GAC is replaced every 2-years.
The twelve filters in Bloomington’s Annex Building each
consist of a single 20’ by 21.75’ cell, providing an area of
19” GAC
435 square feet per filter.
A flow in the range of 0.5 to 0.6 MGD through each of 12” sand
the ‘old plant’ filters is considered optimal. This gravel
underdrain
corresponds to slightly less than one gallon per minute per
square foot, a very conservative surface loading rate. FILTER MEDIA- Old Plant
The six filters in Bloomington’s Main Building consist of
dual 10.75’ by 21.5’ cells. They provide a combined area of 24” GAC
462 square feet per filter.
A flow in the range of 1.0 to 1.3 MGD through one of 12” sand
the new plant filters is considered optimal. This gravel
underdrain
corresponds to slightly less than two gallons per minute per
square foot. FILTER MEDIA- New Plant
4
Filter Operations
Filters are run for 48 hours before being backwashed. Backwash sequences are initiated manually in
order to give the operator a chance to directly observe individual filters. While observing a filter
backwash , the operator should look for:
• air bubbles; a sign of air binding or entrainment of air in the backwater water line,
• calcium carbonate plates and chips; derived from spalling or pressure washing of
filter walls and piping.
• foreign matter in the filter; such as mudballs, cemented media, algal filaments or
mats, fibers, surface accumulations.
• uneven distribution of washwater; boils, horizontal flows, lifting of media, wall effects,
shrinkage cracks, separation of GAC and sand, hydraulic surges.
• media blowoff; carryover of light media (GAC, fine sand) into the backwash launders,
• unusual quantities of solids (unusual color) in backwash water,
• foam; an indication of the presence of organic (surface-tension-lowering) compounds.
The release of air bubbles, which has been observed by operators at Bloomington, may be controlled
by installing air release valves along the backwash water influent line or by constructing or refurbishing a
separate elevated finished water storage tank to establish gravity flow.
5
Monitoring
Even with the SCADA systems, operators need to closely monitor treatment plant process
performance, particularly in ways that computers can’t. Automated equipment, alarms and computers
have minimal powers of observation, lack judgment and are unable to respond to emergencies. They are
susceptible to many types of failure, such as electrical power surges and outages plus system component
failure. Corrosion and the build-up of solids in sampling lines also cause failure of monitoring systems
and the recording of erroneous data due to blockage and solids sloughing.
Bloomington’s filters are convenient to operate and maintain. They are all enclosed and protected
from windblown debris and sunlight. Individual filter banks are further enclosed within partitions with
windows for ready observation. Filter gallery areas are tiled for ease of cleaning and routine filter
maintenance operations.
Each individual filter is equipped with a continuous flow turbidimeter. In addition to SCADA data
acquisition systems for recording flow and head loss, turbidity is displayed visually for each individual
filter.
Similarly, backwash can be automatically controlled and monitored. Surface wash is set for 4
minutes. After surface wash is completed, high flow backwash is initiated for 6 minutes, 40 seconds.
Finally, two minutes of low-flow backwash complete the filter wash and restratification of the bed. Each
step is detailed on the SCADA control panel.
On filter refilling to operating level, a pre-programmed, graduated return-to-service, or ‘flow ramping’,
can be initiated to bring the filter back online gradually. Current filtration practice does not include flow
ramping because no significant turbidity excursions are encountered at the low rates of hydraulic filter
loading currently employed at Bloomington.
6
IEPA Turbidity Requirements
By law, filtered water turbidities on any given filter should:
• not exceed 1 ntu *
• not exceed 0.5 ntu * after the first four hours of filter operation following a backwash
* for two consecutive fifteen minute sampling intervals
What to Do About High Turbidities
If it appears that a particular filter is about to exceed IEPA turbidity requirements, take the filter out of
service until the automated results can be verified and the situation properly remedied. If time permits, a
sample of filter effluent should be taken by hand for laboratory analysis to compare with the reading given
by the continuous flow turbidimeter installed on the filter.
Continuous flow turbidimeters require frequent maintenance to avoid recording spurious turbidity
readings. The small diameter finished water sampling line leading to each turbidimeter will progressively
become coated and encrusted with attached particles, including calcium carbonate, making the line
progressively cloudier. Subsequently, due to changes in flow, temperature or disturbance of the line, the
attached particles may slough, causing an atypical reading that indicates the filter is passing particles.
7
Evaluation of Filters during GAC Replacement
Every year, under contract with Calgon Corporation, one-half of the the GAC caps on the
Bloomington water treatment plant filters are replaced with virgin (not previously used) carbon. The two-
year-old carbon at the new plant and three-year-old carbon at the old plant is removed from each of the
beds hydraulically.
The carbon/finished water slurry is ejected from the bed through flexible hoses into an empty tractor-
trailer capable of holding the GAC contents of a single filter bed. The trailer returns the used carbon to
the manufacturer for thermal reactivation and reuse for less critical applications, such as decolorizing
sugar or rum or for waste treatment processes.
The following comparative evaluations were conducted on the filter media while GAC was being
removed and replaced in Filters #13, #15, #17 of the new plant and Filters #2, #3, and #4 of the old plant
on April 23-24, 2002. Normally, the GAC in Filter #1 would also have been replaced during this cycle.
However, Filter #1 was undergoing replacement of its underdrain system at the time.
Oxygen Persistence in Sand and GAC Media
A container of filter sand removed from Filter #13 was rinsed with aerated tap water (DO 11 mg O/l;
pH 9.0). Upon standing, dissolved oxygen was found to become totally depleted. The pH decreased to
8.5. The implication of these observations is that, as a result of continuing microbial respiration, oxygen
depletion may occur in filters that remain idle for extended periods of time.
A similar test was conducted with both sand and GAC subsequently removed from Filter #4 (Annex).
Whereas DO decreased to 3.5 mg O2 /l in the GAC, it decreased to 7.5 mg O2 /l in the filter sand. The pH
had decreased to 8.0 and 8.5 in the GAC and sand, respectively. This would be consistent with
microbially-mediated oxygen depletion (respiration) which results in the production of carbon dioxide.
The lowered pH during periods of stagnation may also result in the dissolution of calcium carbonate
and magnesium hydroxide previously deposited on the filter media.
In June 2002, the Bloomington laboratory acquired a recently developed fiber optic dissolved oxygen
probe that can be inserted into the media of a filter bed. This analytical device will provide the capability
for observing dissolved oxygen at various depths within a filter both during filter operation and stagnation.
For filters which support biological growth, the degree of dissolved oxygen depletion may be a most
important measure of the necessary degree of filter backwashing.
8
GAC Size Distribution and Characteristics
From visual inspection, the GAC and sand media removed from Filters #13 (Main Building) and #4
(Annex) appeared very different from the fresh media that was being installed to replace the GAC and
augment the sand. To confirm the observations, media size distributions were determined using
Bloomington’s laboratory sieve equipment.
2.11
sieve size, mm
1.83
1.52
new GAC
1.02 old GAC
0 10 20 30 40 50 60 70 80 90 100
% passing, by weight
0.5
old GAC
0.4 new GAC
Fraction by Weight
0.3
0.2
0.1
0
<1.02 1.02-1.52 1.52-1.83 1.83-2.11 >2.11
From the sieving results obtained, it was clear that the size distribution of the GAC that had been in
use for two years had narrowed considerably. This is attributable to the abrasion of the larger GAC
granules so that their size is reduced. At the lower end of the size range, the finer GAC (<1 mm) granules
may have been washed out of the filter during backwash. The result is a more uniform GAC size
distribution. Despite the decrease in Uniformity Coefficient (U.C.), the effective size (E.S. or D10 particle
size) appeared to have remained slightly above 1 mm. Overall, the narrowing size distribution with age
makes the GAC, progressively, a more uniform and desirable filter medium.
9
In addition, there was a noticeable discoloration (graying) of the GAC that had been in service.
Normally intensely black, the granules had begun to lighten in shade. Some of the particles among the
GAC granules appeared to be solid white. These were thought to be calcium carbonate particles or chips
recruited from the filter walls and piping. To determine the extent of this recruitment, a portion of GAC
was dried, weighed and acidified to dissolve the calcium carbonate. This resulted in a 2% loss of weight.
In the future, it was decided that greater care should be taken to minimize the recruitment of these
calcium carbonate particles. Many appear to begin as large platelets dislodged from the filter surfaces
during pressure cleaning. With repeated backwash, they are broken into progressively smaller pieces.
Raking of the filter surface after pressure washing should remove most of these particles.
The carbon granules themselves showed varying degrees of white patches. When dislodged from
the carbon by shaking in a test tube, the supernatant from these granules were found to consist of large
numbers of microorganisms of diverse sizes and shapes. This is consistent with the characterization of
GAC as a biological filter since the organic nutrient adsorbed by the carbon is converted to microbial cell
mass. This cell mass is part of the turbidity removed by backwash after each filter cycle.
TOC Removal by GAC and through Plant
With the TOC equipment available in Bloomington’s laboratory, it is possible to quantify the overall
reduction in organic matter achieved by the GAC filter/adsorbers. This information should be important
for assessing the useful life of the GAC cap as well as for determining the degree of solids removal from
GAC which would maximize biological activity. Where biological processes are involved, a perfectly clean
(organism-free) filter/adsorbent surface may not be optimal.
Initial data on TOC (primarily, dissolved organic carbon, DOC) removal in Filter #15 show that the
GAC installed on April 23, 2002 initially adsorbed close to 60% of the TOC applied to the top of the filter
bed. This exceptionally high percentage of TOC removal is expected to diminish as the adsorptive
capacity of the virgin carbon is exhausted. The rate at which this will occur is not yet known.
Filter #14, which contains carbon which has been in service for one year, exhibits far less TOC
removal. Long-term data on filters with older GAC is vital to determine the degree of TOC removal
achieved by biological processes alone.
10
60
% TOC Removal by GAC-Capped Filters
50
40
New GAC - 4/23/02
30
1-year-old GAC
20
10
-10
-20
13-May 14-May 15-May 16-May 22-May 29-May 30-May 6-Jun 11-Jun
4 4
Total Organic Carbon, mg C / l
3 3
2 2
Filter 14 Infl.
Filter 15 Infl.
Reservoir Effl.
Recarbon. Effl.
Tap
Cone 1 Effl.
Cone 4 Effl.
Reservoir Infl.
Recarbon Infl.
Filter 14 Effl.
Filter 15 Effl.
Raw
1 1
77%
0 0
Overall TOC removals at the Bloomington plant (figure, left) were far higher than the 15% to 35%
required under the USEPA surface water treatment regulations. In May, 2002, most of the TOC reduction
in the plant, 56%, occurred during pretreatment (coagulation, lime softening, sedimentation). Subsequent
filtration through freshly installed GAC increased overall TOC reduction to 77%.
On June 11, 2002. a more comprehensive TOC profile through the plant (figure, right) also showed
that most of the TOC removal took place in the Claricones as a result of coagulation and lime softening.
Overall, TOC reduction was 63%.
11
Virgin GAC Replacement
After two years of service, the GAC cap is hydraulically removed from the top of the filter beds and
replaced with virgin carbon. This GAC arrives in 1,000 pound ‘totes’ that hold 40 cubic feet of the
adsorbent. Therefore, the bulk (dry) density of the GAC is 25 #/cu. ft. or 25% that of filter sand. This
lighter material stratifies in filter beds during backwash to form a discrete layer of GAC atop the filter sand.
The GAC and sand media in Bloomington’s filters are not significantly intermixed. This discrete
layering creates a sharp interface at which foreign materials and mudballs might settle and accumulate.
Over time, this accumulation could partially blind the filter in a horizontal plane resulting in reduced
filtration effectiveness, increased rate of head loss and decreased length of filter runs.
Freshly installed, the GAC will consume oxygen for some period of time since GAC is a strong
reducing agent. GAC will also continuously consume chlorine and chloramine residuals reducing these
compounds to chloride ion while the carbon itself is oxidized to carbon dioxide. This is why it is generally
considered inappropriate to prechlorinate immediately prior to filtration through a GAC-capped filter. The
consumption of disinfectant residuals also forms the basis for the claim of improved water taste when
carbon adsorbers are used for household, point-of-use treatment.
Virgin and Used GAC (above); Trailer receiving Used GAC (below)
12
Sand Size Distribution and Characteristics
Because of its comparatively large particle size (D10≈1 mm), GAC is not as effective a filter medium
as filter sand (D10≈0.5 mm). Whereas solids will accumulate within and on the surface of GAC, the finer
particles in the filter influent would be expected to be retained in the sand layer. Accordingly, the sand
layer could be critical in meeting stringent new regulatory standards on filtered water turbidity.
From the sampling of the underlying sand layer conducted during the changeover of the GAC, it is
evident that the size distribution of the filter sand has changed since it was first installed. First, the D10
size appears to have increased from approximately 0.52 to 0.58 mm. In addition, the D60 size has
increased substantially from about 0.72 to 1.05 mm.
1.02
sieve size, mm
0.76
0.66
old sand
0.51 new sand
0 10 20 30 40 50 60 70 80 90 100
% passing, by weight
0.5
old sand
0.4 new sand
Fraction by Weight
0.3
0.2
0.1
0
<0.51 0.51-0.66 0.66-0.76 0.76-1.02 >1.02
The ratio D60/D10 is used to calculate the uniformity coefficient (U.C.)of the medium, a measure of
how uniformly it is graded. The lower the U.C., the more uniform the grading and the more desirable the
material is as a filtration medium. The U.C. of the filter sand has increased from 1.38 to 1.81.
13
Appearance of New and Old Filter Sand
These increases will combine to make it increasingly difficult to attain the degree of expansion of the
sand layer which is believed to be optimal for scour and cleaning of the filter media. For washing sand
filters without the assistance surface wash, air or auxiliary scour, a 20 to 50% bed expansion is
commonly used. Since the surface wash at Bloomington does not reach into the expanded sand layer,
sand bed expansion should be in this range. The absence of well-defined stratification in the sand layer
may be an indication that this degree of expansion is not occurring.
Filter sand removed from the top... and bottom of Filter #2
Visual observation of the sand removed from the filters indicates that the increase in effective size
(E.S.) and uniformity coefficient may have occurred for the following reasons:
• Torpedo sand, which forms the base for the filter sand, was thrust upward into the filter sand layer,
14
• some larger particles (white chips) of calcium carbonate, dislodged during filter box maintenance,
had settled onto the top of the sand layer,
• a portion of the finest sand (< 0.5 mm) was absent, indicating that it may have migrated and
escaped down into the underdrain system, and finally,
• some GAC was collected in the sand sample. This was picked out as much as possible before the
sand sieve analysis was conducted. However, these larger GAC particles would tend to increase
the apparent effective size of the sand mixture.
Despite the observed changes, the sand medium remains in the range commonly utilized for filter
sand, though at the upper range. The addition of make-up sand, practiced as part of routine GAC
replacement, will lower the existing size distribution. Future additions of make-up sand may utilize still
finer grained material (E.S. 0.4 mm) to continue to restore the sand to its initial size range.
Alternately, the addition of three to four inches of a still finer (E.S. 0.2 to 0.3 mm) and denser garnet
sand (Specific Gravity 4.0 as opposed to 2.6 for silica sand) may serve to intermix with the existing media
and further increase filtration efficiency.
Ultimately, however, the regulated test of filter media particle removal effectiveness is the
measurement of turbidity. Continuous monitoring of filtered water turbidity profiles, as currently practiced
by Bloomington Water, demonstrate that filtration efficiency routinely exceeds regulatory requirements by
a significant margin
Sand in Tote and Replacement of Filter Sand - May, 2002
15
Recommended Filter Washing Protocol
While there is no defined optimum filter backwash procedure, a recommended protocol for filter
washing at Bloomington, based on conventional U.S. practice, is as follows:
• Following the lowering of the water surface level in the bed, the surface wash should
be activated and operated alone for four minutes or until any surface caking or
accumulations are broken up.
• A low-rate wash which will cause a 10% expansion of the sand layer and expand the
upper bed of GAC into the rotating arms of the surface wash should then be
conducted for 5 minutes. (It may be desirable to minimize this wash since extended
surface wash may contribute to the break-up of the friable GAC granules, thereby
creating micrometer-sized carbon ‘fines’. These carbon fines may penetrate the filter
following restart.)
• High rate wash at a nominal rate of 20 gpm/ft2 for up to five minutes should then be
used to expand the sand layer into the 20 to 50% expansion range. (At this point,
there is the greatest danger of washing out the lightest fraction of the GAC so that,
based on observation, the maximum backwash rate may be limited to that which
precludes loss of GAC.) The duration of the high-rate wash may be determined by
the operator for the seasonal (temperature) and pretreatment conditions. For
example, following the addition of polymeric coagulant aids, the backwash may be
extended to assist in removing the more tightly attached polymer.
• Finally, a low-rate wash may be employed at the end of the backwash procedure to
assist in restratifying the media.
Since some of the filter sand has increased in size and some has been lost in operation, it may be
advantageous to conduct some simple laboratory column studies using filter media removed from several
typical filters to observe the backwash rates that are, in fact, effective in achieving the desired degree of
expansion and cleaning). These filters can subsequently be used to observe the effect of polymer addition
of filter performance, head loss and removal of attached solids.
16
Microbial Growth on Filtration Media
In the absence of the application of disinfectant prior to filtration, the significant surface area provided
by the fine grained filtration media offers ample opportunity for microbial growth. Although this
phenomena has occurred in water treatment plants from earliest times, the phenomena has recently been
rediscovered as “biological treatment” or “biological filtration.”
A number of benefits can accrue to the development of a microbial community on the filter media. A
portion of the organic content of the filter influent, the ‘labile’ or biologically assimilable portion, may be
removed. Generally, this is only a small fraction of the dissolved organic carbon.
Alternately, aerobic respiration by the biological growth can lead to the loss of oxygen and anoxia
within the filter bed. Periodic sloughing of accumulations of organism call mass can lead both to turbidity
excursions and elevated numbers of organisms contributing to heterotrophic plate counts (HPC).
The unique micrographs below, taken with Bloomington’s epifluorescence microscope, illustrate the
abundance of green-fluorescing microbial filaments and stalked bacteria colonizing a GAC granule
removed from service at Bloomington. It has been speculated that the filaments (attached growth) may
assist in the attachment of particles from the filter influent to the filter media.
Bacterial filaments and colonies (green fluorescence) on GAC granule surface
17
Particles Washed from GAC
A portion of the accumulation of organisms on the biologically colonized GAC are stripped off the
granules with each succeeding filter backwash. Their removal of the heterotrophic organisms reflects the
removal of the assimilable portion of the dissolved organic carbon (DOC) achieved by the filter.
Examples of the particles and organisms removed during backwash are illustrated in the following
micrographs. These micrographs show a large and diverse group of rod-shaped bacteria (green dots),
generally embedded in clumps of detached slime (extracellular polymer) along with carbon fines (black).
Long bacterial filaments and diatoms are also present.
Detached from GAC: diatoms; bacterial rods, clumps and filaments; carbon fines
The production, and subsequent removal, of this diverse group of organisms in the GAC-capped filter
is credited with reducing microbial growths in the distribution system. These organisms have also been
credited with oxidizing haloacetic acids (HAA) formed by prechlorination prior to filtration.
Alternately, the presence of a large microbial community has spurred studies of the degradation of
water quality in filter beds when they are taken out of service for 6 to 24 hours. Oxygen is found to
decrease within the filter pore water within hours. While DOC, ammonium and nitrite ion concentrations
in the filter pore water increase, attached microbial cell mass declines under the anaerobic conditions. As
a result, water quality degradation is reported upon start-up. To avoid this deterioration and restore the
biodegradation performance of the filters, the idle filters were backwashed before being returned to
service.
18
Particles Washed from Sand
While silica sand is inert and does not provide nutrient for microbial growth, the extensive surface
provided by this fine-grained material does lends itself to the accumulation of attached microbial growth.
In many treatment systems, the influent to sand filters have been prechlorinated to prevent the
accumulation of microbial slime.
The micrographs below show organisms recovered in the rinse water from sand taken from Filter #4.
The microbial community is abundant and appears diverse. Many of the bacterial cells (orange-red rods
and filaments) are aggregated or embedded in clumps of detached slime.
It is important to note that any organisms of ‘potential health concern’ that were present in the lake
water source have largely been physically removed by coagulation and sedimentation or inactivated at
the high pH maintained during the lime softening process. Alternately, the organisms (shown in the
micrographs) that were recovered from Bloomington’s filter media are primarily new growth that has
occurred as the result of surface colonization and the utilization of the dissolved organic matter in the
filter influent. These organisms are not known to represent a human health hazard. Similar results are
observed wherever individual consumers use household treatment devices containing granular activated
carbon.
19
