United States
Environmental Protection
Agency
Municipal Environmental
Research Laboratory
.Cincinnati OH 45268
Research and Development
EPA-600/S2-84-160 Dec. 1984
Project Summary
Ultraviolet Disinfection of a
Secondary Effluent:
Measurement of Dose and
Effects of Filtration
J. Donald Johnson and Robert G. Quails
Ultraviolet (UV) disinfection of
wastewater secondary effluent was
investigated in a two-phase study to
develop methods for measuring UV
dose and to determine the effects of
filtration on UV disinfection. The first
phase of this study involved a pilot plant
study comparing filtration, water
quality parameters, and two
reactors. The pilot plant study led to
laboratory experiments involving: (1)
the development of a method for in situ
measurement of intensity using a
calibrated bioassay, (2) experimental
verification of a method for calculating
intensities, (3) evaluation of the role of
lamp spacing in dose efficiency, and (4)
simulation of UV disinfection in
continuous flow.
A bioassay method was developed to
measure average dose rate (i.e.,
intensity) within a UV reactor. The
survival of Bacillus subtilis spores was
determined as a function of UV dose to
calibrate the sensitivity of the spores.
Spores were added to unknown
systems, and the survival used to
determine the average dose rate. A
modification was used for flowthrough
reactors in which spores were injected
as a spike and collected at a known time
from injection.
A point-source summation method
for calculation of dose rate was verified
by bioassay measurements in a simple
cylinder. This calculation method was
also applied to multiple lamp reactors.
Spectrophotometric measurements
significantly overestimated the UV
absorbance in wastewater because of
scattering. A method to correct for
scattering was tested. A method for
simulating survival in complex flow-
through reactors was presented, and a
simulation of our pilot plant runs
corresponded reasonably well with the
observed survival. Mixed media
filtration significantly improved disin-
fection in pilot plant experiments. A
laboratory experiment showed that a
relatively small number of coliforms
were protected inside particles, but
they were the factor limiting
disinfection at -3 or -4 logs of survival.
This Project Summary was developed
by EPA's Municipal Environmental
Research Laboratory. Cincinnati. 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
Environmental problems associated
wit'h chlorination have prompted
research into alternatives for disinfection
of wastewater effluents. Residuals and
byproducts of chlorination can be toxic to
aquatic life in receiving waters, and they
may form carcinogenic byproducts. In
addition, chlorination is less effective in
killing viruses, spores, and cysts than in
killing bacteria. One disinfection process
that would not be expected to produce
undesirable byproducts is ultraviolet (UV)
light.
-------
The U.S. Environmental Protection
Agency (EPA) has funded several pilot- or
full-scale investigations of UV
disinfection of wastewater. Though these
studies have generally met disinfection
goals successfully, comparisons of
results have been limited because no
direct method exists for measuring UV
doses, nor has there been a substantiated
method for calculating doses in the com-
plicated geometries of a practical reactor.
Lack of such measurement methods has
also prevented the controlled evaluation
of variables such as UVabsorbanceofthe
water, filtration, reactor design, and the
varying sensitivities of different
organisms.
This study was initiated to develop
methods for measuring UV dose and to
determine the effects of filtration on UV
disinfection. The first phase involved a
pilot plant study comparing: (1) the effects
of mixed-media filtration, (2) the effects of
randomly varying water quality param-
eters, and (3) the effects of different lamp
spacing in two UV disinfection reactors.
Experience from the pilot plant study led
us to the second laboratory experimental
phase involving: (1) development of a
method for in situ measurement of dose
rate using a calibrated bioassay, (2) ex-
perimental verification of a method for
calculating dose rates, (3) separation of
the effects of absorbed and scattered UV
light relative to spectrophotometer meas-
urements, (4) evaluation of the role of
lamp spacing in dose efficiency, and (5)
simulation of UV disinfection in
flowthrough reactors.
The following are several problems that
have occurred when estimating doses in
previous studies of UV disinfection:
1. UV radiometer detectors measure
intensity on a planar surface and
thus do not correctly measure the
three-dimensional intensity (i.e.,
dose rate) to which a cell may be
exposed near a long, tubular lamp.
2. A UV radiometer detector positioned
in the wall of a disinfection reactor
cannot be used to estimate the
average dose rate (intensity) within
the entire reactor.
3. Wastewater contains particles that
scatter UV light so that spectropho-
tometers tend to overestimate UV
absorbance.
4. Equations have been used that
incorrectly calculate the dose rate
near a tubular lamp in an absorbing
solution.
5. In flowthrough systems, the distri-
bution of exposure times is not
simply related to volume and flow
rate.
Bioassay Method for
Measurement of Dose Rate
or Average Intensity
A bioassay method was developed to
measure average dose rate in flowthrough
reactors. Dose is defined as:
Dose = (dose rate) (exposure time) (1)
or, in units:
mW-sec/cm2 = (mW/cm2) (sec) (2)
The term "dose rate" has been used
instead of the more familiar "intensity"
because of the ambiguities in the UV
literature in definitions of intensity. The
survival (NS/N0) of organisms is a
function of dose:
Ns/N0 = fn (dose)
(3)
where N0 and Ns are the density of orga-
nisms before and after irradiation,
respectively. Equations 1 and 3 imply that
dose rate and exposure time may be
varied reciprocally to obtain the same
survival.
The survival of Bacillus subtilis spores
was determined as a function of the UV
dose to calibrate the sensitivity of the
spores. Since dose rate (as measured by a
radiometer) was only applicable in a
col II ma ted beam, the spores were
exposed for varying periods of time to a
collimated beam of UV light in a stirred
petri dish. The dose rate at the surface of
the suspension was measured. Since
fluid depth and absorbance were
minimal, the dose could be calculated
based on the measured dose rate and
the exposure time. In cases where
absorbance was significant, the average
dose rate was calculated using an
integration of Beer's law over the fluid
depth. Calibration curves of log survival
versus dose were constructed (Figure 1)
and found to be quite reproducible over
several months. The dose rate may be
determined in an unknown system by (1)
determining the survival (NS/N0); (2)
reading the dose corresponding to the
observed survival using the calibration
curve (Figure 1); and (3) using the known
exposure time in Equation 1 to calculate
average dose rate.
Separation of Effects of UV
Absorbance and Scattering
Calculating the average UV dose rate
requires an absorbance measurement.
Wastewater effluents contain particles
that may scatter as well as absorb the UV
light. Bioassay experiments showed that i
scattered UV light was still effective for '
killing bacteria. Since the usual spectre-
photometric measurements do not
separate scattering and absorbance, a
way was needed to separate the two. An
established method using a frosted
cuvette for both the blank and sample
allowed a correction for most of the
-2
I -3
-5
a A
_L
16 24
Dose (mW-sec/cm2)
32
40
Figure 1.
Log survival of Bacillus subtilis spores versus U V dose in a collimated beam of known
dose rate. Different symbols represent five separate runs. Data from doses of 10 to
30.5 mW-sec/cm* appear linear and fit regression line Y = 0.167X+1.01 (r = 0.98).
-------
scatter. A piece of oil-saturated paper
placed on the cuvette face may also be
used.
This technique was tested against a
bioassay method to separate absorbance
and scattering. A sample of tertiary
effluent (14 NTU turbidity) was filtered
through a 0.45-ju.m filter. Suspensions of
intermediate turbidity were made by
mixing portions of the filtered and
unfiltered sample. Thus the soluble
absorbing component was held constant,
and the particulate component varied.
Samples were spiked with Bacillus
spores and irradiated in a petri dish in the
collimated beam apparatus. The average
dose rate in the suspension was assayed.
With the integrated form of Beer's law, a
determination was made of the
absorbance that would yield the observed
assayed dose rate. The assayed
absorbance for the suspensions of
varying particulate content is shown as a
function of the spectrophotometric ab-
sorbance in Figure 2. The difference
between the spectrophotometric absorb-
ance and the assayed absorbance was
the scattering component. The soluble
absorbance, particulate absorbance, and
scattering were, respectively, 47%, 41%,
and 12% of the spectrophotometric ab-
sorbance. The frosted cuvette method
showed a slightly lower scattering
component. The scattering component
was estimated to have averaged 9% in the
pilot plant studies. The soluble absorb-
ance was 60% to 80% of the spectropho-
tometric absorbance in most of the
secondary effluents measured.
Calculation of Dose Rate
Common radiometer detectors cannot
be used to measure dose rate near a
.40
.30
8
I
3
.20
.10
I
Depth (cm)
o 2
n 3
A 4
\
\
.10 .20
Spectrophotometric Absorbance
.30
.40
Figure 2.
Spectrophotometric absorbance versus absorbance measured by the bioassay
method for a Chapel Hill tertiary effluent sample. The soluble UV absorbance was
kept constant, and the particulate concentration varied by diluting the unfiltered
14-NTU samples with the filtered0.07-NTU sample. The solid line would represent
an exact correspondence between the two methods. The dotted line is a regression
through the data points. The soluble and particulate absorbance and scatter com-
ponents of the spectrophotometric absorbance of the unfiltered sample are indicated.
tubular lamp because they measure
energy flux on a plane surface. Light
received at angles other than 90° to the
surface of the detector is attenuated,
since the surface of the detector inter-
cepts fewer of the rays. Biological cells in
motion in a solution, however, present a
three-dimensional target, and they
respond to the three-dimensional dose
rate from all angles within a disinfection
reactor.
To calculate the UV dose rate at a point
near a tubular lamp in an absorbing
solution, we need a nuclear engineering
equation called the pointsource
summation (PSS) calculation. This
equation assumes that a line segment
source can be treated as the sum of a
number of point sources. We can consid-
er a cylindrical coordinate system around
a line segment light source surrounded
by a quartz sleeve (Figure 3). The linear
source of UV output OPT is divided into
point sources, each of which has strength
S (units in Watts):
S = OPT/N
(4)
The dose rate at a point IIR,zc) result-
ing from one point source (Z L) can then be
treated as the product of the spherical
spreading times the attenuation resulting
from absorbance over a definite path
length (P-P,):
4,r(R2 + Z'LC)
(5)
where a is the absorbance of the medium
and the other geometry is shown in
Figure 3. The total dose rate at point
'(R,zc)is (he sum of the contributions of
each point source (at each ZL) over the
source length (ZLN):
'(R,ZC)
UR.ZC!
The use of this calculation requires two
measurements: absorbance of the water,
and the lamp UV output. Output was
measured by integrating the dose rate
measurements over a spherical surface
centered on the lamp centroid (Figure 4).
By placing the radiometer detector far
from the lamp (190 cm), the rays were
almost parallel, and the dose rate could
be properly measured. A string and a
protractor were attached below the lamp
-------
centroid, and the detector was rotated
around the 90° arc described by the string
at radius r (190 cm). Output was
calculated as follows: (1) Dose rate (I) was
measured at angles 6 between 0° and
180°. The outer surface area of a slice of
the sphere of radius r with arc length d0 is
2-rra dd (Figure 4), where a — r sinf. The
energy leaving the surface of the slice is I
2-r-a dt). The energy leaving the surface of
the sphere is 2-rrryo I s\r\8 d#. The I sin#
values were plotted as a function of 6
(radians), and the area under the curve
fromO to Trwas measured gravi metrically.
To test the PSS calculation, the
calculated average dose rate inside a
cylinder was compared with that
measured by the spore bioassay. The PSS
calculation was used in a computer
program to average the dose rates over
the volume of a cylinder around a lamp.
This procedure was carried out for a
series of cylinders of varying radii and for
fluids of different absorbances.
Suspensions of spores were exposed
for a fixed time to UV light inside the
cylindrical apparatus shown in Figure 5.
A movable paper tube was located
between the lamp and quartz sleeve so
that the lamp could be warmed up and an
exact exposure made. The suspensions
were well-stirred. Fulvic acid was added
as a natural UV absorber. The survival of
the spores was measured, and the
assayed average dose rate was deter-
mined as putlined previously.
The PSS calculations were generally
verified by the bioassay measurements.
Figure 6 compares the calculated PSS
curves (solid lines) and the bioassay data
(data points). The correspondence was
fairly good both for cylinders of different
radii and for fluids of different absorb-
ances. But the calculated values tend to
be a few percent higher than the bioassay
measurements in the smaller cylinders.
The stirring device produced interference
in very thin cylinders. We also performed
the same experiment using spores spiked
in a secondary effluent, and the PSS.
calculations were within 10% of the
bioassay dose rates. The calculation
methods that had been used in some
previous studies were also applied to
these cylinders, and those methods gave
results that differed greatly from our
experimental average dose rates.
Practical UV reactors are flowthrough
systems and have a distribution of
exposure times. To use the bioassay of
dose rate in a flowthreugh system, a way
was needed to determine a definite
exposure time. To do this, the spores
were used in a manner analogous to a
Zc
ZLC
Lamp Quartz
Wall
Figure 3. Cylindrical reactor geometry for point source summation calculation.
tracer injection study. A flowthrough tube
surrounding a UV lamp was used to
demonstrate this method. Spores were
injected into the flowstream of water at
the entrance to the tube, and the outflow
fractions were collected in a rotating
sampling tray as a function of time from
injection. The injection was performed
with the light on and repeated with the
light off. The distribution of unirradiated
spores reflected the residence time distri-
bution (RTD). The survival (NS/NQ) was
calculated for each flow fraction
separately by comparing spore densities
in the corresponding irradiated and
unirradiated fractions at a given time from
injection. The average dose rate was then
determined for each fraction by finding
the corresponding dose from the calibra-
tion curve and dividing by the time from
injection. A modification of the spore
injection bioassay may be used to
measure average dose rate in full-scale
reactors.
Figure 4. Method for determining output
of a tubular lamp.
-------
Ultraviolet
Lamp
Quartz Tube —»
Plexiglass
Cylinder
(f
Sliding Black
Paper Tube
Between Lamp
and Quartz
Tube
Stirring
Device
Figure 5. Cylindrical batch irradiation
apparatus.
The assayed average dose rates within
the flowthrough tubes corresponded well
with the calculations of the PSS model
(Figure 6, injection experiments). The
distribution of unirradiated and irradiated
viable spores also showed that most of
the surviving spores were those that
emerged from the tube before the
average residence time (RT). This result
illustrates the drastic effect that non-plug
flow can have on the disinfection
efficiency.
Calculation of Dose Rate in
Multiple Lamp Reactors
To calculate the average dose rate in
multiple lamp reactors, the following
method was used:
(1) Dose rate at each point was
considered to be the sum of the
7 2 3 4 5
Radius of Cylinder (cm)
Figure 6. Average dose rate (i.e., intensity) within a cylinder of radius, R. The solid lines arethe
curves calculated by the point source summation for indicated absorbances. The
0.24 absorbance line omitted for clarity. Data points represent bioassayed average
intensity within the cylinders of various sizes. Data points for 1.32- and 1.59-cm
radii were obtained from flowthrough tubes rather than by batch.
contributions from each lamp
calculated by the PSS model.
(2) Dose rate was mapped at each
point on a grid on cross-sections of
the reactor.
(3) Dose rates were averaged over the
cross-sections and along the length
of the reactor.
UV lamps transmit little of the UV light
coming from adjacent tubes, so it was
necessary to make calculations that took
this shadowing into account. Our
calculations made the following simpli-
fications: that reflection from the reactor
walls was negligible under actual
operating conditions, and that reflection
and refraction by the quartz sleeves were
negligible. The average dose rate
calculations were performed by
FORTRAN computer programs that: (1)
proceeded from point to point on a
representative cross-section of the
reactor, (2) excluded the point if it lay
within a quartz sleeve, (3) considered
contributions from each lamp, (4)
excluded the contribution from a lamp if it
was blocked by another lamp, and (5)
called the PSS calculation as a subroutine.
Divergent views exist on the design of
UV reactors. Some are based on improper
equations or conventional wisdom rather
than on calculation or experiment
because of the lack of adequate methods
for measuring or calculating UV dose
Our models can be useful for research
and development of reactor design. We
contrasted the efficiency of the different
schemes of lamp spacing. Any surface or
object that absorbs UV energy (e.g .walls,
baffles, other lamps), besides the
unavoidable absorbance of the water
itself, reduces the efficient use of the UV
energy The product of dose rate times
reactor volume was shown to be a factor
that is directly proportional to the
effectiveness of the unit at treating
volumes of water under ideal flow
conditions. This factor isolates the
effectiveness of the dose rate regime
from the effects of flow dispersion and
can be used to compare reactors of
different lamp spacings and volume. At a
given flow rate and number of lamps, a
close lamp spacing gives a higher
average dose rate but a shorter RT
-------
Calculations in this study showed how
the distance the light was allowed to
penetrate before encountering an
obstruction affects the efficiency of light
use. Figure 7 illustrates the product of
dose rate and volume in cylinders of
radius R around a UV lamp. The point at
which the lines level out is the radius at
which no significant UV light penetrates
in the cylindrical geometry. For an
absorbance representative of secondary
effluent (0.16), walls or other
obstructions within 5 cm can absorb a
significant amount of available UV light.
Two reactors used in the pilot plant exper-
iments were compared based on the
products of their dose rate and volumes.
The reactor with lamps placed close to
one another and the walls, the Pure
Water Systems (PWS) unit, had an
average dose rate almost twice as high as
the other reactor (Aquaf ine). But the PWS
reactor had a much smaller volume (and
RT), so the dose rate and volume products
were almost equal. The PWS reactor did
use a greater lamp wattage, however.
The term "dose rate and volume
efficiency" (dose rate and volume pro-
duct/input wattage) was used to
compare the efficiency of the use of lamp
wattage. The PWS was much less
efficient because of the proximity of trie
lamps to each other and to the walls with
the resulting absorption of the light.
The dose rate and volume product does
not consider the effects of non-ideal flow.
Though the dose rate and volume product
of the two reactors were nearly equal, the
PWS reactor gave 0.6 to 2.1 logs greater
survival of fecal coliforms than did the
Aquafine at the same flow rate because
of severe short circuiting of flow m the
PWS reactor. Thus the effects of flow
dispersion must be considered separately
from the dose rate regime in determining
the ultimate disinfection efficiency.
This study also use simulation of a full-
scale reactor operated in northwest
Bergen County, New Jersey, to show the
effect of varying lamp spacing on the UV
light use efficiency and to provide an
analysis of the relative costs.
Simulation of Dose and
Disinfection in Flowthrough
Reactors
The second factor in calculating dose is
exposure time, which can lead to as much
error in calculations as dose rate.
Flowthrough reactors have a distribution
of RT. Neither the RT calculated from flow
rate and volume nor the average RT
determined from dye studies can be used
20
16
1
I 12
1
It 6
u"
Radius (cm)'
Figure 7. Effectiveness of various fluid depths in cylinders of radius R around UV lamps.
Calculated values of the product of average dose rate in a cylinder of radius R
times the volume of that cylinder are shown versus the outer radius of the cylinder
for fluids of absorbances 0,0.16 and 0.32.
to predict the average survival Since
survival is not linearly related to dose, the
average dose is insufficient to predict the
average survival over the RTD, but the
survivor density must be calculated for
each flow fraction and then summed.
Equations 7 through 10 show how the
density of survivors (Ns) may be predicted
from the following data.
1. Coliform density in influent (No),
2. Average dose rate (DR), either
measured or calculated,
3. RTD, and
4. Dose survival curve (determined
accurately, e.g., in collimated beam
apparatus).
For an aliquot of volume Vt entering the
reactor at time to, the aliquot will exit in n
fractions of volumes Vj at times tj. Survival
in each fraction is some nonlinear func-
tion (fn) of dose.
Ns
.7- = fn (dose)
•MO
(7)
The dose for the ith fraction = (DR) (8)
<''> Ns
Survival in the ith fraction =-rr-L-(9)
= fn[(DR)(t,J] NOJ
The average density of survivors, Ns, is:
NS=N0 SVj (fn [(PR) (t,)]) (10)
Data from a dye study on the RTD may be
put in a form to use in these equations.
The area under a curve of dye
concentration versus time is set equal to
Vt (and may be thought of as a 1-ml
aliquot entering the reactor). Then,
Vj = (A t) (relative dye concentration)
V
(11)
For a computer simulation of average
survival, the RTD and dose-survival curve
data pairs were fed into arrays, and
intermediate values needed in Equation
10 were generated by linear interpolation.
As an example of simulated survival in
a flowthrough reactor, runs with the
Aquafine reactor were simulated and
compared with the observed survival in
the pilot plant experiments The input
data listed above were necessary. The
average dose rate was that calculated by
the PSS model for two levels of applied
-------
voltage. The RTD was measured with dye
injection and adjusted to a higher flow
rate. Methods did not exist at the time of
the pilot plant runs to determine an accu-
rate dose-survival curve, so one was
determined some time later for a sample
from the same site.
The average log survival predicted by
the simulation corresponded reasonably
well with that observed in the pilot plant
runs (Table 1). Some deviation might be
expected, since the dose-survival curve
was based on one sample taken at a later
date. Further research should involve
simulation using data obtained simulta-
neously with full reactor runs.
Table 1. Actual Versus Simulated Survi-
val (S) of Total Conforms in a
Sandy Creek 2° Effluent
Average Pilot
Lamp Intensity Simulated Plant
Voltage fmW/cm2! Log S Log S
60
128
6.2
9.7
-3.26
-3.61
-3.29 (±.13)
-3.69 (±.16)
The simulation makes it possible to use
another method for bioassay of dose
rate. If bioassay spores are allowed to
flow continuously through a reactor, the
dose rate cannot be measured because
there is a distribution of RT. But if the RTD
is known from a dye study, the simulation
may be run with various values for dose
rate until the simulated average survival
matches the observed average survival by
trial and error. On large reactors, the
injection method of bioassay would
probably be easier, however.
Simulation takes into account the
factors of the dose rate and volume
characteristics as well as the effects of
flow dispersion and sensitivity of the
target organisms. Thus simulation can be
a useful tool for research and develop-
ment of reactor design. For example, it
can be used to find optimum lamp
configurations and tradeoffs with flow
dispersion. Or it can be used to predict
the design parameters needed for a
specific situation so that costly over-
design is not necessary. The predicted
survival of a standard coliform sample at
a given flow rate may be used to compare
a number of different reactors. The
simulations may also be used to prepare
empirical curves of predicted survival
versus flow rate, operating voltage, water
quality, etc., for a particular installation
as a guide to continuous operation.
Protection of Cells Inside
Particles and Effects of Filtration
In our pilot plant experiments, an ex-
tended aeration secondary effluent was
subjected to mixed-media filtration. Both
filtered and unfiltered effluents were
subjected to UV disinfection in two
UV reactors at two different flow rates
and two levels of applied lamp voltage.
The filtered effluents showed significantly
better disinfection for both total coliform
and fecal coliform shown in Table 2. Total
coliform log survival was 0.33 to 0.79
logs lower in the filter treatments. The
effect of filtration on UV absorbance was
small and did not account for the
disinfection differences. The differences
in suspended solids, turbidity, and
UV absorbance indicate that the filtration
tended to remove the larger particles that
had relatively little effect on the
absorbance. Average dose-rate calcula-
tions and simulation supported the idea
that the filtration effect was not due to the
lower absorbance. The conclusion was
that a relatively small number of difficult
to disinfect coliforms was protected
inside particles, but that these tended to
be removed by filtration.
A laboratory experiment was performed
to support the hypothesis on the effects of
particle protection. The dose-survival
curves were determined for an unfiltered
effluent sample, and the same sample
passed through a 70-//m and an 8-//m
pore size filter. Since coliforms are about
1 -2 /um in size, the 8-//m filter allowed
Table 2.
Flow
Inactivation Shown as Mean
-Log Survival of Fecal Coliforms
in Unfiltered and Filtered
Secondary Effluent
-Log Survival of
Fecal Coliforms
Rate (L/s) Voltage Unfiltered Filtered
4.92 60 3.08 (.20)* 3.88 (.19)
128 3.41 (.23) 4.17 (.18)
2.27 60 3.91 (.23) 4.291.17)
128 3.47 (.28) 3.92 (.24)
*Standard deviations of logs are shown in
parentheses.
only single cells or very small aggregates
to pass. The survival curve of this fraction
(Figure 8) shows disinfection continuing
beyond -4.5 logs of survival, where
survivors were undetectable. Curves for
the 70-fjm filtered and unfiltered samples
tend to level out after -2 or -3 logs of
survival. The coliforms not passing the
8-fjm filter were extremely resistant to
UV. Since the curves were similar until
fewer than about 10 of the coliforms (or
1%) were surviving, the protected coli-
forms appeared to be a small minority, but
they became the limiting factor to disin-
fection at levels needed to meet legal
standards.
Other Pilot Plant Results
The Aquafine reactor met the disinfec-
tion goal of 200 MPN/100 ml in every
,-f
-2
!-*
I
o Filtered through 8fj filter
A Filtered through 70/u filter
• Unfiltered
_L
J
8 12
Dose (mW-sec/cm2)
16
20
Figure 8. Effect of filtration on survival of total coliforms in Sandy Creek effluent. Each point
represents one separate exposure. Points with arrows indicate limits of detectability
for exposures in which nd survivors were found.
-• U S GOVERNMENT PRINTING OFFICE 1985- 559-111/10747
-------
case, but the PWS reactor did not
because of the poor quality of residence
time or short circuiting of flow. Changes
in applied lamp voltage and flow rate
produce relatively small changes in
survival because (as can be seen from the
dose-survival curve m Figure 8, for
example) the dose-survival curves level
out at -3 or -4 logs of survival. Stepwise
multiple regression of randomly varying
water quality parameters on log survival
of coliforms showed no consistent
correlations. This lack of correlation was
probably due to the relatively small
variation in UVabsorbanceandthe lack of
response to dose increases at -3 to -4 logs
of survival. The spectrophotometric
absorbance was predicted well by
coliform densities or (if these were not
considered) by COD, turbidity, and
suspended solids together.
Recommendations
The following methods should be used
to compare different UV .reactors. The
effects of dose rate and reactor volume
must be evaluated separately from the
effects of non-ideal flow. The dose rate
and volume product is the best measure
to compare reactors of different size and
similar flow characteristics at a given
absorbance, as this product is proportion-
al to the effectiveness of the reactor
under ideal flow conditions. In addition,
the RTD should always be reported. No
single measure exists that can
simultaneously consider the dose rate
and volume product and the effects of
non-ideal flow. But comparison may best
be made by simulating the average
survival using a standard dose survival
curve for coliforms such as that reported
here. Reactors of different sizes could be
compared by reporting the flow rate
necessary to achieve a -3 log survival (for
example) using the standard curve.
The following procedures should be
used in UV pilot or demonstration plant
research so that results may be general-
ized within and between studies. Use of a
UV radiometer to determine dose rate
should be limited to a collimated beam. A
radiometer detector situated in a reactor
wall cannot be used to estimate average
dose rate without an accurate empirical
determination. A newly installed reactor
may undergo a series of calibration runs
to prepare accurate dose data for
continuous use. Accurate tracer studies
should be done to determine the RTD over
the range of flow rates used. Accurate
measurements of lamp output as a
function of temperature should be
performed. Injection bioassay measure-
ments of dose rate should be made at
different absorbances. The PSS calcula-
tions may be compared with the bioassav
and used for interpolation. For continu-
ous monitoring of average dose rate, one
may use empirical curves of average dose
rate versus relative radiometer readings
at several points in the reactor for
different absorbances. Dose-survival
curves on effluent samples should be
determined accurately in a collimated
beam for comparison with other studies.
Curves of average survival versus
average dose under conditions of non-
ideal flow can apply only to a particular
situation.
If the disinfection of single coliform
cells in wastewater under ideal flow
conditions is considered ideal efficiency,
then the results of this report show the
following to be the chief factors limiting
ideal efficiency in practice: (1) protection
of cells inside particles, (2) flow
dispersion and poor mixing across dose-
rate gradients, and (3) shadowing and
absorption of UV light within a reactor.
The full report was submitted in
fulfillment of EPA Grant No. R-804770 by
the University of North Carolina at Chapel
Hill under the sponsorship of the U,S.
Environmental Protection Agency.
J. D. Johnson and R. C. Quails are with Environmental Science and Engineering
Department, University of North Carolina, Chapel Hi/I, NC 27514.
A. D. Venosa is the EPA Project Officer (see below).
The complete report, entitled "Ultraviolet Disinfection of a Secondary Effluent:
Measurement of Dose and Effects of Filtration," (Order No. PB 85-114 023;
Cost: $14.50, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE 8. FEES P/
EPA
PERMIT No G-35
Official Business
Penalty for Private Use $300
» I . '- V.
•* . V ;_
j - r t i
------- |