Appendix
D.
Microbiological
Methods
D.
1
General
Recommendations
The
challenge
microorganism
used
to
validate
UV
reactors
should
be
cultured
and
analyzed
by
a
laboratory
staffed
by
professional
microbiologists
and
equipped
to
perform
microbiological
examinations
as
per
Standard
Methods
for
the
Examination
of
Water
and
Wastewater
(
APHA
et
al.
1998,
sections
1020­
1050).
Protocols
for
culturing
the
challenge
microorganism
and
measuring
its
concentration
should
be
based
on
published
and
peer­
reviewed
methods
and
should
be
clearly
defined
and
followed.
Measurement
of
the
concentration
of
the
challenge
microorganism
before
and
after
exposure
to
UV
light
should
be
initiated
within
24
hours
of
exposure.
If
the
challenge
microorganism
has
the
ability
to
photorepair,
exposure
of
samples
to
visible
light
should
be
kept
to
a
minimum.

Because
MS2
bacteriophage
(
MS2)
and
B.
subtilis
spores
are
commonly
being
used
as
challenge
microorganisms
for
UV
reactor
validation,
the
following
sections
describe
procedures
that
can
be
used
for
preparing
stock
solutions
of
MS2
and
B.
subtilis
spores
and
assaying
the
concentration
of
those
microorganisms
in
water
samples.
Procedures
for
preparing
stock
solutions
can
be
scaled
to
provide
the
volumes
needed
for
UV
reactor
validation.
Alternative
procedures
and
challenge
microorganisms
can
be
used
if
they
are
acceptable
to
the
State.
Section
F.
1
provides
a
rational
for
selecting
challenge
microorganisms.

D.
2
MS2
Bacteriophage
Stock
Preparation
MS2
(
ATCC
15597­
B1)
can
be
propagated
using
a
variety
of
host
bacteria
including
Escherichia
coli
C3000
(
ATCC
15597),
E.
coli
F­
amp
(
ATCC
700891),
and
others
(
Meng
and
Gerba
1996,
Oppenheimer
et
al.
1993,
NWRI/
AwwaRF
2000).
The
following
propagation
method
was
adapted
from
NWRI/
AwwaRF
(
2000):

1.
Inoculate
sterile
tryptic
soy
broth
(
TSB)
(
DIFCO,
Detroit,
Michigan)
with
host
bacterium
transferred
from
a
single
colony
grown
on
a
nutrient
agar
plate.
Incubate
the
culture
with
constant
stirring
at
35
to
37
°
C
for
18
to
24
hours.

2.
Transfer
0.5
mL
of
the
host
bacterium
culture
to
50
mL
of
fresh
TSB
and
incubate
at
35
to
37
°
C
for
4
to
6
hours
with
continuous
shaking
at
100
Hz
to
obtain
a
culture
in
its
log
growth
phase
(
approx.
3x108
cfu/
mL)
(
cfu
=
colony
forming
units).

3.
Dilute
stock
MS2
using
Tris­
buffered
saline
(
pH
7.3)
to
a
concentration
of
approximately
108
pfu/
mL
(
pfu
=
plaque
forming
units).

4.
Add
1mL
of
diluted
MS2
stock
solution
to
the
50
mL
volume
of
E.
coli
in
TSB
and
incubate
overnight
at
35
to
37
°
C.

5.
Centrifuge
the
MS2/
E.
coli
culture
at
8000
×
g
(
g
=
9.82
m/
s2)
for
10
minutes
at
4
°
C
to
remove
cellular
debris.

UV
Disinfection
Guidance
Manual
D­
1
June
2003
Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
2
June
2003
6.
Filter
the
supernatant
by
passing
it
through
a
0.45
µ
m
low
protein­
binding
filter.

7.
Assay
the
concentration
of
MS2
in
the
stock
solution
as
per
section
D.
3.

8.
Collect
and
refrigerate
the
filtrate
at
4
°
C
and
use
within
a
one­
month
period.

Propagation
should
result
in
a
highly
concentrated
stock
solution
of
essentially
monodispersed
phage
whose
UV
dose­
response
follows
first
order
kinetics
with
minimal
tailing.
Figure
D.
1
presents
the
dose­
response
of
MS2
as
reported
in
the
literature.
Over
the
range
of
REDs
demonstrated
during
validation
testing,
the
mean
dose­
response
of
the
MS2
stock
solution
should
lie
within
the
90
percent
prediction
interval
of
the
mean
response
in
Figure
D.
1.
Over
a
dose
range
of
0
to
120
mJ/
cm2,
the
predictions
intervals
may
be
defined
using
the
following
equations:

Dose
Dose
on
Inactivati
Bound
Lower
×
×
+
×
×
 
=
 
 
2
2
4
10
6
.
7
10
4
.
1
log:
Dose
Dose
on
Inactivati
Bound
Upper
×
×
+
×
×
 
=
 
 
2
2
5
10
5
.
4
10
6
.
9
log:

Figure
D.
1
UV
Dose­
Response
of
MS2
0
1
2
3
4
5
6
7
8
0
20
40
60
80
100
120
UV
Dose
(
mJ/
cm2)
Log
Inactivation
Meng
and
Gerba,
1996
Oppenheimer
et
al,
1993
Sommer
et
al,
1998
Tree
et
al,
1997
Havelaar
et
al,
1990
Nieuwstad
et
al,
1994
Mean
Mean
Response
90%
Prediction
Interval
Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
3
June
2003
D.
3
MS2
Assay
The
concentration
of
MS2
(
ATCC
15597­
B1)
in
water
samples
can
be
assayed
using
the
agar
overlay
technique
with
E
coli
(
ATCC
15597)
as
a
host
bacterium
(
Adams
1959,
Yahya
et
al.
1992,
Oppenheimer
et
al.
1993,
Meng
and
Gerba
1996).
The
following
procedure
can
be
used:

1.
Inoculate
TSB
(
Difco,
Detroit,
MI)
with
the
host
bacterium
and
incubate
at
35
to
37
°
C
for
18
to
24
hours
to
obtain
an
approximate
concentration
of
108
CFU/
mL.

2.
Transfer
1
mL
of
the
culture
to
50
mL
of
fresh
TSB
and
incubate
at
35
to
37
°
C
for
4
to
6
hours
with
continuous
shaking
at
100
Hz
to
obtain
a
culture
in
its
log
growth
phase.

3.
Obtain
serial
dilutions
of
the
MS2
sample
using
a
0.001
M
phosphate­
saline
buffer
or
TSB.

4.
Combine
and
gently
stir
1
mL
of
host
cell
solution,
0.1
mL
of
diluted
MS2
sample,
and
2
to
3
mL
of
molten
tryptic
soy
agar
(
TSA)
(
0.7
percent
agar,
45
­
48
°
C)
(
Difco,
Detroit,
MI).

5.
Pour
the
mixture
onto
solidified
TSA
(
1.5
percent
agar)
contained
within
Petri
dishes.
The
time
between
the
mixing
the
MS2
sample
with
the
E.
coli
host
and
the
plating
of
the
top
agar
layer
should
not
exceed
10
minutes.
After
plating,
the
agar
should
harden
in
10
minutes.

6.
After
the
top
agar
layer
hardens,
cover,
invert
the
Petri
dishes,
and
incubate
16
to
24
hours
at
35
to
37
°
C.

7.
Count
the
plaques
with
the
aid
of
a
colony
counter.
Plaques
are
identified
as
clear
circular
zones
1
to
10
mm
in
diameter
in
the
lawn
of
host
bacteria.

8.
Record
the
number
of
plaques
per
dish,
and
the
MS2
sample
volume
and
dilution.
If
it
is
not
possible
to
distinguish
individual
plaques
because
of
confluent
growth,
record
the
plate
counts
as
"
TNTC"
(
too
numerous
to
count).

9.
Calculate
the
MS2
concentration
in
the
water
samples:

 
 

=

i
i
V
n
ion
Concentrat
Equation
D.
1
where
ni
=
The
number
of
counts
on
each
plate
Vi
=
The
volume
of
undiluted
sample
used
with
each
plate
Example.
A
water
sample
containing
MS2
was
diluted
10,
100
and
1,000­
fold
using
a
0.1
mL
aliquot
dilution
of
the
sample
for
each.
Each
dilution
was
assayed
in
triplicate.
Plaque
Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
4
June
2003
forming
units
observed
on
the
plates
were
2,
5
and
6
for
the
1,000­
fold
diluted
sample
and
32,
40,
and
47
for
the
100­
fold
diluted
sample.
With
the
10­
fold
dilution,
plate
counts
were
too
numerous
to
count.
The
concentration
in
the
original
sample
is
calculated
as
follows:

(
)
mL
pfu
mL
mL
pfu
ion
Concentrat
/
000
,
40
100
/
3
1
.
0
1000
/
3
*
1
.
0
47
40
32
6
5
2
=
×
+
+
+
+
+
+
=

D.
4
Bacillus
Subtilis
Spore
Preparation
Bacillus
subtilis
spores
(
ATCC
6633)
can
be
propagated
using
Schaeffer's
media
(
Munakata
and
Rupert
1972,
Sommer
et
al.
1995,
DVGW
1997).
The
following
propagation
method
was
adopted
from
DVGW
(
1997):

1.
Prepare
Columbia
agar
(
Oxoid
CM
331)
as
a
1
L
solution
of
23.0
g
special
peptone
(
Oxoid
L
72),
1.0
g
starch,
5.0
g
NaCl,
and
10.0
g
agar
(
Oxoid
L
11)
in
distilled
water.
Adjust
pH
to
7.0
and
autoclave
15
minutes
at
121
º
C.

2.
Prepare
the
sporulation
media
as
a
1
L
solution
of
280
mg
MgSO4
·
H2O,
1.11
g
KCl,
3.1
mg
FeSO4
·
7H2O,
and
8.9
g
nutrient
broth
(
Oxoid
CM
67)
in
distilled
water.
Adjust
the
pH
to
7.0
and
autoclave
it
for
15
minutes
at
121
º
C.

3.
Inoculate
Columbia
agar
plates
(
Oxoid
CM
331)
with
three
smears
of
B.
subtilis
and
incubate
24
hours
at
37
º
C.

4.
Inoculate
300
mL
of
sporulation
media
with
three
colonies
collected
from
the
agar
plates.

5.
Incubate
the
sporulation
media
72
hours
at
37
º
C
on
a
shaker
operating
at
2
Hz.

6.
Sonicate
the
resulting
culture
for
10
minutes
at
50
kHz
and
10
º
C.

7.
Harvest
the
spores
by
centrifuging
80
mL
aliquots
at
5000
g
for
10
minutes
and
10
º
C.

8.
Wash
the
spores
three
times
by
resuspending
in
20
mL
of
distilled
water
and
centrifuging
at
5000
×
g
for
10
minutes
at
10
º
C.

9.
Resuspend
the
washed
spores
in
100
mL
of
0.001
M
phosphate­
saline
buffer.

10.
Inactivate
the
vegetative
B.
subtilis
by
heat
treatment
at
80
º
C
for
10
minutes.

11.
Sonicate
the
resulting
culture
for
10
minutes
at
50
kHz
and
10
º
C.

12.
Collect
the
resulting
stock
solution
and
assay
the
B.
subtilis
spore
concentration
as
per
section
D.
5.

13.
Refrigerate
the
filtrate
at
4
º
C
and
use
within
a
one­
month
period.
Sonicate
for
10
minutes
at
50
kHz
and
10
º
C
before
use.

Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
5
June
2003
Propagation
should
result
in
a
highly
concentrated
stock
solution
of
mono­
dispersed
B.
subtilis
spores
with
a
UV
dose­
response
that
follows
the
dose­
response
reported
in
the
literature
and
presented
in
Figure
D.
2.
Over
the
range
of
reduction
equivalent
doses
(
REDs)
demonstrated
during
validation
testing,
the
mean
dose­
response
of
the
B.
subtilis
stock
solution
should
lie
within
the
90
percent
prediction
interval
of
the
mean
response
provided
in
Figure
D.
2.
Over
a
dose
range
of
0
to
70
mJ/
cm2,
the
predictions
intervals
may
be
defined
using
the
following
equations:

Dose
Dose
Dose
on
Inactivati
Bound
Lower
×
×
 
×
×
+
×
×
 
=
 
 
 
2
2
3
3
5
10
3
.
5
10
7
.
2
10
0
.
2
log:
Dose
Dose
on
Inactivati
Bound
Upper
×
×
+
×
×
=
 
 
2
2
4
10
3
.
4
10
7
.
5
log:

Figure
D.
2
UV
Dose­
Response
of
Bacillus
Subtilis
Spores
­
1
0
1
2
3
4
5
6
7
0
50
100
150
UV
Dose
(
mJ/
cm2)
Log
Inactivation
DVGW,
1997
Hoyer,
2002
Chang
et
al,
1985
Sommer
et
al,
1996
Sommer
et
al,
1995
Sommer
et
al,
1998
Sommer
and
Cabaj,
1993
Sommer
and
Cabaj,
1993
Mean
90%
Prediction
Interval
of
the
Mean
Response
Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
6
June
2003
D.
5
Bacillus
Subtilis
Spore
Assay
The
concentration
of
B.
subtilis
spores
(
ATCC
6633)
in
water
samples
can
be
assayed
by
the
plate
method
using
plate
count
agar.
The
following
procedure
was
adopted
from
Deutscher
Verein
des
Gas­
und
Wasserfaches
(
DVGW)
(
1997):

1.
Prepare
plate
count
agar
(
Oxoid
CM
325)
as
a
1
L
solution
of
5.0
g
casein
peptone
(
Oxoid
L
42),
2.5
g
yeast
extract
(
Oxoid
L
21),
1.0
g
glucose,
and
9.0
g
agar
(
Oxoid
L
11)
in
distilled
water.
Adjust
pH
to
6.8
±
0.2
and
autoclave
for
15
minutes
at
121
º
C.

2.
Obtain
serial
dilutions
of
the
B.
subtilis
spore
sample
using
0.001
M
phosphate­
saline
buffer.

3.
Vacuum
filter
100
mL
of
diluted
sample
through
a
47
mm
x
0.45
µ
m
membrane
filter
(
Gelman
Sciences,
Ann
Arbor,
MI).

4.
Place
filter
onto
a
Petri
dish
containing
hardened
agar
and
cover
plates.

5.
Incubate
plates
24
±
2
hours
at
37
±
1
º
C.

6.
Count
the
number
of
colonies
formed
with
the
aid
of
a
colony
counter.

7.
Record
the
number
of
colonies
per
dish,
and
the
B.
subtilis
spore
sample
volume
and
dilution.
If
it
is
not
possible
to
distinguish
individual
colonies
because
of
confluent
growth,
record
the
plate
counts
as
"
TNTC".

8.
Calculate
the
B.
subtilis
spore
concentration
in
the
original
samples
as
cfu/
mL
using
Equation
D.
1.

D.
6
References
Adams,
M.
H.
1959.
Bacteriophage.
New
York:
Interscience
publication.

APHA/
AWWA/
WEF.
1998.
Standard
methods
for
the
examination
of
water
and
wastewater,
20th
edition.
Washington
DC:
American
Public
Health
Association,
American
Water
Works
Association,
and
Water
Environment
Federation.

DVGW.
1997.
UV
disinfection
devices
for
drinking
water
supply
 
 
Requirements
and
testing.
Bonn,
Germany:
German
Gas
and
Water
Management
Union
(
DVGW).

Havelaar,
A.
H.,
C.
C.
E.
Meulemans,
W.
M.
Pot­
Hogeboom,
and
J.
Koster.
1990.
Inactivation
of
bacteriophage
MS2
in
wastewater
effluent
with
monochromatic
and
polychromatic
ultraviolet
light.
Water
Research
24,
no.
2:
1387­
1393.

Hoyer,
O.
2002.
Personal
communication.

Proposal
Draft
Appendix
D.
Microbiological
Methods
UV
Disinfection
Guidance
Manual
D­
7
June
2003
Meng,
Q.
S.
and
C.
P.
Gerba.
1996.
Comparative
inactivation
of
enteric
adenovirus,
poliovirus
and
coliphages
by
ultraviolet
irradiation.
Wat.
Res.
30:
2665­
2668.

Munakata,
N.
and
C.
S.
Rupert.
1972.
Genetically
controlled
removal
of
"
spore
photoproduct"
from
deoxyribonucleic
acid
of
ultraviolet­
irradiated
bacillus
subtilis
spores.
J.
Bacteriol.
111:
192­
198.

Nieuwstad,
T.
J.
and
A.
H.
Havelaar.
1994.
The
kinetics
of
batch
ultraviolet
inactivation
of
bacteriophage
MS2
and
microbiological
calibration
of
an
ultraviolet
pilot
plant.
J.
Environ.
Sci.
Health
A29,
no.
5:
1993­
2007.

NWRI/
AwwaRF.
2000.
Ultraviolet
disinfection
guidelines
for
drinking
water
and
water
reuse.
National
Water
Research
Institute
and
AwwaRF
Oppenheimer,
J.
A.,
J.
E.
Hoagland,
J.­
M.
Laine,
J.
G.
Jacangelo,
and
A.
Bhamrah.
1993.
Microbial
inactivation
and
characterization
of
toxicity
and
by­
products
occurring
in
reclaimed
wastewater
disinfected
with
UV
radiation.
Water
Environment
Federation
(
WEF)
Specialty
Conference:
Planning,
design
&
operation
of
effluent
disinfection
systems.
Whippany,
New
Jersey.

Sommer,
R.
and
A.
Cabaj.
1993.
Evaluation
of
the
efficiency
of
a
UV
plant
for
drinking
water
disinfection.
Water
Science
&
Technology
27,
no.
7­
8:
357­
362.

Sommer,
R.,
A.
Cabaj,
D.
Schoenen,
J.
Gebel,
A.
Kolch,
A.
H.
Havelaar,
and
F.
M.
Schets.
1995.
Comparison
of
three
laboratory
devices
for
UV­
inactivation
of
microorganisms.
Wat.
Sci.
Tech.
31:
147­
156.

Sommer,
R.,
A.
Cabaj,
and
T.
Haider.
1996.
Microbiocidal
effect
of
reflected
UV
radiation
in
devices
for
water
disinfection.
Water
Science
&
Technology
34,
no.
5­
6:
173­
177.

Sommer,
R.,
T.
Haider,
A.
Cabaj,
W.
Pribil,
and
M.
Lhotsky.
1998.
Time
dose
reciprocity
in
UV
disinfection
of
water.
IAWQ,
Vancouver.

Tree,
J.
A.,
M.
R.
Adams,
and
D.
N.
Lees.
1997.
Virus
inactivation
during
disinfection
of
wastewater
by
chlorination
and
UV
irradiation
and
the
efficacy
of
f+
bacteriophage
as
a
'
viral
indicator'.
Water
Science
&
Technology
35:
227­
232.

Yahya,
M.
T.,
T.
M.
Straub,
and
C.
P.
Gerba.
1992.
Inactivation
of
coliphage
MS­
2
and
poliovirus
by
copper,
silver,
and
chlorine.
Canadian
Journal
of
Microbiology.
38,
no.
6:
430­
435.

Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
The
challenge
microorganism's
UV
dose­
response
should
be
measured
using
a
benchscale
device
referred
to
here
as
a
"
collimated
beam
apparatus"
(
Figure
E.
1).
The
apparatus
delivers
UV
light
to
a
microbial
suspension
usually
contained
within
a
completely
mixed
batch
reactor.
The
UV
light
enters
the
suspension
with
a
near
zero
degree
angle
of
incidence
and
is
relatively
homogenous
across
the
surface
area.
UV
dose
delivered
to
the
suspension
is
calculated
using
measurements
of
incident
UV
intensity,
exposure
time,
suspension
depth,
and
the
absorption
coefficient
of
the
suspension.
By
measuring
microbial
inactivation
in
the
suspension
as
a
function
of
UV
dose,
the
microorganism's
dose­
response
is
determined.

Figure
E.
1
Collimated
Beam
Apparatus
Low­
Pressure
Mercury
Arc
Lamp
Lamp
Enclosure
Petri
Dish
Containing
Microbial
Suspension
Magnetic
Stirrer
UV
Light
@
254
nm
Collimating
Tube
UV
Disinfection
Guidance
Manual
E­
1
June
2003
Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
2
June
2003
This
appendix
describes
the
following
components
of
collimated
beam
testing:

 
Collimated
beam
apparatus
design
and
operation
(
section
E.
1)

 
Procedure
for
irradiating
samples
using
apparatus
(
section
E.
2)

 
Calculation
of
UV
dose
delivered
by
the
apparatus
(
section
E.
3)

 
Quality
Assurance
/
Quality
Control
(
QA/
QC)
procedures
(
section
E.
4)

 
Reporting
results
(
section
E.
5)

E.
1
Apparatus
Design
and
Operation
Because
UV
dose
requirements
are
based
on
the
pathogen
inactivation
achieved
using
253.7
nm
light,
the
collimated
beam
apparatus
must
use
a
lamp
that
emits
germicidal
UV
light
only
at
254
nm
(
e.
g.,
a
low­
pressure
lamp).
To
prevent
ozone
formation,
lamps
that
emit
185
nm
light
should
not
be
used.
The
output
from
the
lamp
measured
using
a
radiometer
or
equivalent
should
not
vary
more
than
5
percent
over
the
exposure
time.
A
stable
lamp
output
can
be
obtained
by
driving
the
lamp
with
a
constant
power
source
and
maintaining
the
lamp
at
a
constant
operating
temperature.
A
voltage
regulator
may
be
used
to
obtain
a
stable
power
supply
to
the
lamps
if
the
line
voltage
is
not
sufficiently
stable.
A
stable
temperature
can
be
obtained
by
controlling
the
airflow
around
the
lamp.

The
UV
lamp
should
be
located
far
enough
above
the
surface
of
the
microbial
suspension
that
uniform
irradiance
is
obtained
across
the
sample's
surface
and
UV
light
enters
the
suspension
with
a
near
zero
degree
angle
of
incidence
(
Blatchley
1997).
A
recommended
minimum
distance
from
the
lamp
to
the
suspension
is
six
times
the
longest
distance
across
the
suspension's
surface.
In
order
to
vary
the
UV
intensity
incident
on
the
suspension,
the
distance
between
the
suspension
and
the
lamp
can
be
adjusted.

The
uniformity
of
the
intensity
field
across
the
sample's
surface
should
be
assessed
by
measuring
the
"
Petri
Factor,"
the
ratio
of
the
average
irradiance
across
the
suspension
surface
to
the
irradiance
measured
at
the
center
(
Bolton
and
Linden
2003).
The
average
irradiance
is
determined
by
averaging
radiometer
measurements
taken
at
each
point
in
a
5
mm
spaced
grid
across
an
area
defined
by
the
suspension's
surface.
If
the
radiometer's
sensing
window
is
wider
Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
3
June
2003
than
5
mm,
it
should
be
reduced
using
a
cover
slip
with
a
small
hole.
In
general,
the
collimated
beam
apparatus
should
have
a
Petri
Factor
greater
than
0.9.

The
lamp
and
the
light
path
from
the
lamp
to
the
suspension
should
be
enclosed
to
protect
the
user
from
exposure
to
UV
light.
A
box­
like
enclosure
made
of
aluminum
is
often
used.
A
length
of
pipe
is
often
used
to
enclose
the
light
path
from
the
lamp
to
the
microbial
suspension.
The
inside
surface
of
the
pipe
should
have
a
low
UV
reflectance
and
incorporate
apertures
to
improve
UV
light
collimation
(
Blatchley
1997).
A
shutter
mechanism
is
sometimes
used
to
control
the
exposure
of
the
suspension
to
UV
light.
The
exposure
times
should
be
measured
with
an
uncertainty
of
5
percent
or
less.
Exposure
times
less
than
20
seconds
are
not
recommended.

The
microbial
suspension
should
be
irradiated
in
an
open
cylindrical
container
with
a
constant
cross­
sectional
area
(
e.
g.,
Petri
dish).
The
diameter
of
the
container
should
be
smaller
than
the
diameter
of
the
light
beam
incident
on
the
container.
Sample
depth
should
be
0.5
to
2
cm.
The
material
of
the
container
should
not
adsorb
the
challenge
microorganism
enough
to
impact
its
measured
dose­
response.

Sample
volumes
irradiated
in
the
container
should
be
sufficient
for
measuring
the
challenge
microorganism's
concentration
after
irradiation.
The
microbial
suspension
should
be
mixed
using
a
stir
bar
and
a
magnetic
stirrer
at
a
rate
that
does
not
induce
vortices.
The
volume
and
diameter
of
the
stir
bar
should
be
small
relative
to
the
volume
and
depth
of
the
sample
volume.
The
irradiance
at
the
center
of
the
suspension's
surface
before
and
after
exposure
to
UV
light
should
be
measured
using
a
radiometer
calibrated
at
254
nm.
The
radiometer
calibration
should
be
National
Institute
of
Standards
and
Technology
(
NIST)
traceable
or
equivalent
with
a
known
measurement
uncertainty.
During
measurement,
the
radiometer's
calibration
plane
should
match
the
height
of
the
suspension's
surface
and
be
perpendicular
to
the
incident
UV
light.
The
calibration
plane
of
the
radiometer
should
be
specified
in
the
radiometer's
calibration
certificate.

The
depth
of
the
suspension,
including
the
stir
bar
volume,
should
be
measured
with
an
uncertainty
of
10
percent
or
less.
The
UV
absorption
coefficient
of
the
microbial
suspension
at
254
nm
should
be
measured
using
a
spectrophotometer
with
a
measurement
uncertainty
of
10
percent
or
less.
If
scattering
of
light
by
the
microorganisms
and
other
particulate
matter
within
the
suspension
is
significant,
the
UV
absorption
coefficient
should
be
measured
using
a
spectrophotometer
equipped
with
an
integrating
sphere
(
Linden
and
Darby
1998).
While
1
cm
cuvettes
are
typically
used
for
measuring
UV
absorption
coefficients,
cuvettes
with
longer
pathlengths
are
recommended
to
reduce
the
measurement
uncertainty
with
low
UV
absorbance
samples.

E.
2
Procedure
Personnel
who
perform
bench­
scale
UV
irradiation
should
be
experienced
with
the
use
and
safety
requirements
of
the
equipment.
Safety
goggles
and
latex
gloves
should
be
worn.
Skin
should
be
shielded
from
exposure
to
UV
light.
The
following
procedure
is
recommended
for
irradiating
a
water
sample
using
the
collimated
beam
apparatus:

1.
Define
the
target
UV
dose.

Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
4
June
2003
2.
Measure
the
UV
absorption
coefficient
of
the
water
sample.

3.
Place
a
known
volume
from
the
water
sample
into
a
container
and
add
a
stir
bar.

4.
Measure
the
water
depth
in
the
container.

5.
Measure
the
UV
irradiance
delivered
by
the
collimated
beam.

6.
Calculate
the
exposure
time
to
deliver
the
target
dose.

7.
Block
the
light
from
the
collimating
tube
using
a
shutter
or
equivalent.

8.
Center
the
container
containing
the
water
sample
under
the
collimating
tube.

9.
Unblock
the
light
from
the
collimating
tube
and
start
the
timer.

10.
When
the
target
exposure
time
has
elapsed,
block
the
light
from
the
collimating
tube.

11.
Remove
the
container
and
collect
the
sample
for
measurement
of
the
challenge
microorganism
concentration.
If
the
sample
is
not
assayed
immediately,
store
in
the
dark
at
4
º
C.
12.
Re­
measure
the
UV
irradiance
and
calculate
the
average
of
the
two
measurements.

13.
Using
Equation
E.
1,
calculate
the
applied
dose
using
the
measured
irradiance,
UV
absorption
coefficient,
sample
depth,
and
exposure
time.

14.
Repeat
the
procedure
for
various
target
dose
values.
The
UV
dose­
response
curve
is
a
plot
of
the
microorganism
concentration
as
a
function
of
the
applied
dose.

E.
3
Dose
Calculation
Dose
delivered
to
the
sample
is
calculated
using
Equation
E.
1:

t
ad
L
d
L
R
P
E
D
ad
f
s
)
10
ln(
)
10
1
(
)
(
)
1
(
 
+
 
=
Equation
E.
1
where
D
=
UV
dose
in
mJ/
cm2
Es
=
UV
irradiance
at
the
center
of
the
suspension's
surface
in
mW/
cm2
Pf
=
Petri
Factor
R
=
Reflectance
at
the
air­
water
interface
at
254
nm
L
=
Distance
from
lamp
centerline
to
suspension
surface
in
cm
d
=
Depth
of
the
suspension
in
cm
a
=
UV
absorption
coefficient
(
Base
10)
of
the
suspension
at
254
nm
in
cm­
1
t
=
Exposure
time
in
seconds
Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
5
June
2003
The
term
L/(
d+
L)
accounts
for
the
divergence
of
the
UV
light
from
the
collimated
beam
as
it
passes
through
the
suspension.
The
reflectance
at
the
air­
water
interface
estimated
using
Fresnel's
Law
is
0.025
given
an
index
of
refraction
of
1.000
and
1.372
for
air
and
water,
respectively.

Alternatively,
given
a
target
dose,
the
exposure
time
may
be
calculated
by
re­
arranging
Equation
E.
1
to
form
Equation
E.
2:

)
10
1
)(
1
(
)
)(
10
ln(

ad
f
s
R
L
P
E
L
d
ad
D
t
 
 
+
=
Equation
E.
2
where
variables
are
defined
as
in
Equation
E.
1
The
measurement
uncertainty
of
the
dose
delivered
by
the
collimated
beam
should
be
assessed
at
an
80
percent
confidence
interval
with
consideration
of
each
term
in
Equation
E.
1.
The
measurement
uncertainty
of
each
term
in
Equation
E.
1
can
be
determined
from
the
measurement
uncertainty
stated
for
the
instrumentation
used
to
measure
those
quantities
and
the
standard
deviation
of
repeated
measurements
made
with
that
instrumentation
(
Taylor
1997).
If
the
uncertainty
of
the
measurement
of
the
suspension
depth
and
the
UV
absorption
coefficient
is
less
than
10
percent
at
a
80
percent
confidence
level
and
the
product
ad
is
less
than
0.1,
the
uncertainty
of
the
term
(
1­
10­
ad)/
ad
can
be
assumed
as
4
percent.
This
assumption
allows
the
use
of
the
sum
of
variances
approach
to
calculate
the
uncertainty
of
the
dose
delivered
by
the
collimated
beam.

Example.
A
pipette
with
an
accuracy
of
0.2
mL
is
used
to
place
a
25
mL
microbial
sample
in
a
Petri
dish.
The
incident
irradiance
of
1.00
mW/
cm2
is
measured
using
a
radiometer.
The
uncertainty
of
the
radiometer
measurement
indicated
by
the
calibration
certificate
is
7
percent.
The
suspension
is
irradiated
for
60
seconds.
The
irradiation
time
is
monitored
using
a
stopwatch
with
an
uncertainty
of
±
1
second.
The
Petri
dish
radius,
measured
using
a
ruler
with
1
mm
graduations,
is
2.5
±
0.1
cm.
The
stir
bar
volume
is
estimated
as
1
±
0.1
mL.
The
UV
absorption
coefficient
of
the
microbial
sample
at
254
nm
is
0.050
±
0.005
cm­
1.
The
Petri
factor
of
0.90
±
0.02
is
calculated
for
the
collimated
beam
apparatus.
The
distance
from
the
lamp
to
the
surface
of
the
suspension
is
determined
using
a
ruler
as
25
±
1
cm.

The
depth
in
the
Petri
dish
is
calculated
as
the
sum
of
the
suspension
and
stir
bar
volumes
divided
by
the
area
of
the
Petri
dish.

(
)
(
)
(
)
l
Volume
Area
25
0.2
cm
1
0.1
cm
p
2.5
0.1cm
1.32
0.07cm
3
3
2
=
=
±
+
±
±
=
±
The
UV
dose
is
calculated
as:

Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
6
June
2003
(
)(
)(
)

[
(
)
(
)
]
(
)(
)
(
)
(
)(
)
(
)
(
)
D
1.00
mW/
cm
0.90
1
0.025
1
1.32cm
25cm
1
10
0.050cm
1.32
cm
ln
10
60s
46
mJ/
cm
2
0.050
cm
1.32
cm
1
2
1
=
 

+
 
=
 

 
 

Because
the
uncertainty
of
the
sample
depth
(
±
0.07
cm)
and
the
measured
UV
absorption
coefficient
(
±
0.005
cm­
1)
is
less
than
or
equal
to
10
percent
of
the
sample
depth
and
the
product
of
the
sample
depth
and
UV
absorption
coefficient
is
less
than
0.1,
the
uncertainty
of
the
term
(
1­
10­
ad)/
ad
is
assumed
as
4
percent.
The
uncertainties
of
the
terms
in
the
dose
calculation
are
as
follows:

 
Incidence
irradiance
7
percent
 
Petri
factor
2
percent
 
L/(
d+
L)
0.3
percent
 
Time
2
percent
 
(
1­
10­
ad)/
ad
4
percent
The
uncertainty
of
the
dose
calculation
is
calculated
using
the
sum
of
variances
approach
as:

Uncertainty
(
)
%
5
.
8
4
2
3
.
0
2
7
2
/
1
2
2
2
2
2
=
+
+
+
+
=

E.
4
Quality
Assurance
and
Quality
Control
QA/
QC
measures
include:

 
Designing
the
collimated
beam
apparatus
with
a
Petri
factor
greater
than
0.9
 
Selecting
instrumentation
and
methods
that
minimize
the
measurement
uncertainty
of
dose
delivery
by
the
collimated
beam
apparatus
 
Calibrating
all
radiometers
at
regular
intervals
as
recommended
by
the
manufacturer
 
Using
a
reference
radiometer
or
equivalent
method
to
regularly
check
the
measurement
accuracy
of
the
radiometer
used
to
measure
incident
irradiance
 
Ensuring
irradiance
measurements
before
and
after
exposure
to
UV
light
do
not
differ
by
more
than
5
percent
 
Ensuring
replicate
UV
inactivation
curves
do
not
differ
significantly
Proposal
Draft
Appendix
E.
Collimated
Beam
Apparatus:
Measuring
Challenge
Microorganism
UV
Dose­
Response
UV
Disinfection
Guidance
Manual
E­
7
June
2003
 
Ensuring
the
UV
dose­
response
of
the
challenge
microorganism
lies
within
expected
bounds
as
defined
by
published
dose­
response
data
E.
5
Reporting
The
following
information
should
be
documented
and
included
with
the
validation
test
report:

 
Lamp
type
 
Distance
from
the
light
source
to
the
sample
surface
 
Radiometer
make
and
model
 
Measurement
uncertainty
of
the
radiometer
and
date
of
last
calibration
 
Comparison
of
working
and
reference
radiometers
 
Volume
and
depth
of
the
microbial
suspension
 
UV
absorption
coefficient
of
the
microbial
suspension
measured
at
254
nm
 
Irradiance
measurement
before
and
after
each
irradiation
 
Petri
factor
calculations
and
results
 
Method
of
dose
determination
 
UV
dose
calculations
 
Uncertainty
assessment
E.
6
References
Blatchley,
E.
R.
1997.
Numerical
modeling
of
UV
intensity:
Application
to
collimated­
beam
reactors
and
continuous­
flow
systems.
Water
Research
31:
2205­
2218.

Bolton.
J.
and
K.
Linden.
2003.
Standardization
of
methods
for
fluence
(
UV
Dose)
determination
in
bench­
scale
UV
experiments.
J.
Environ.
Eng.
129,
no.
3:
209­
216.

Linden,
K.
G.
and
J.
L.
Darby.
1998.
UV
Disinfection
of
Marginal
Effluents:
determining
UV
Absorbance
and
Subsequent
Estimation
of
UV
Intensity.
Water
Environment
Research
70(
2).

Taylor,
J.
R.
1997.
An
introduction
to
error
analysis:
the
study
of
uncertainties
in
physical
measurements.
Sausalito,
CA:
University
Science
Books.

Proposal
Draft
UV
Disinfection
Guidance
Manual
F­
1
June
2003
Proposal
Draft
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
This
appendix
provides
background
material
for
the
validation
protocol
given
in
Appendix
C.
The
background
material
is
organized
into
the
following
six
sections.

·
Dose
delivery
by
UV
reactors.
Section
F.
1
describes
how
the
RED
of
a
challenge
microorganism
measured
during
UV
reactor
validation
is
related
to
the
capacity
of
the
UV
reactor
to
inactivate
a
target
pathogen.
This
section
describes
why
correction
factors
should
be
applied
to
the
reduction
equivalent
dose
(
RED)
of
the
challenge
microorganism
to
account
for
systematic
errors
that
arise
if
the
challenge
microorganism
is
more
resistant
to
UV
light
as
compared
to
the
target
pathogen.
The
section
concludes
by
describing
approaches
for
selecting
one
or
more
challenge
microorganisms
to
minimize
those
errors.

·
Dose
monitoring.
Section
F.
2
describes
three
approaches
whereby
measurements
of
flowrate,
UV
intensity,
and
water
UV
transmittance
(
UVT)
are
used
by
UV
reactors
to
indicate
dose
delivery.
This
section
discusses
the
importance
of
UV
intensity
sensor
placement
within
a
UV
reactor
and
provides
a
rationale
for
defining
test
conditions
to
validate
UV
reactors
using
a
given
dose
monitoring
approach.

·
UV
intensity
sensors.
Section
F.
3
describes
the
properties
of
UV
intensity
sensors,
how
those
properties
impact
the
sensor's
measurement
uncertainty,
and
how
that
measurement
uncertainty
is
used
to
define
rejection
criteria
for
using
reference
sensors
to
check
the
accuracy
of
duty
sensors.
The
section
also
discusses
how
nonuniform
lamp
aging
and
fouling
and
the
variability
in
lamp
output
affects
the
use
of
UV
intensity
sensors.

·
Polychromatic
considerations.
Section
F.
4
describes
systematic
errors
that
can
occur
with
the
validation
of
UV
reactors
equipped
with
medium­
pressure
UV
lamps.
This
section
provides
approaches
for
assessing
those
errors
and
for
defining
correction
factors
that
should
be
applied
to
validation
data.

·
Uncertainty
of
monitoring
and
dose
factors.
Section
F.
5
provides
a
rationale
for
defining
a
safety
factor
that
accounts
for
the
random
uncertainty
associated
with
UV
reactor
validation
and
monitoring.

·
Re­
validation.
Section
F.
6
discusses
how
some
changes
to
a
UV
reactor
design
made
by
a
manufacturer
would
trigger
a
need
to
re­
validate
the
UV
reactor.

F.
1
Dose
Delivery
by
UV
Reactors
Dose
delivered
to
an
individual
microorganism
passing
through
a
UV
reactor
is
defined
as
the
integral
of
UV
intensity
over
time:
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
2
June
2003
 
=
r
t
Idt
D
0
Equation
F.
1
where
D
=
Dose
delivered
to
the
microorganism
by
the
UV
reactor
I
=
UV
intensity
incident
on
the
microorganism
as
it
travels
through
the
UV
reactor
t
=
time
tr
=
Residence
time
of
the
microorganism
within
the
UV
reactor
Because
each
microorganism
passing
through
the
UV
reactor
follows
a
unique
trajectory,
each
microorganism
is
exposed
to
a
unique
dose.
For
example,
microorganisms
passing
through
the
UV
reactor
close
to
the
lamps
are
exposed
to
higher
UV
intensities
as
compared
to
microorganisms
traveling
near
the
reactor
walls
or
between
lamps.
Microorganisms
caught
in
eddies
or
dead
zones
spend
more
time
within
the
UV
reactor
as
compared
to
microorganisms
that
pass
through
the
reactor
quickly
due
to
hydraulic
short­
circuiting.
Because
each
microorganism
is
exposed
to
a
different
UV
dose,
dose
delivery
by
the
UV
reactor
is
best
described
using
a
dose
distribution,
as
opposed
to
a
single
dose
value.
A
dose
distribution
describes
the
probability
that
a
microorganism
passing
through
a
UV
reactor
will
receive
a
given
dose.
Figure
F.
1
presents
an
example
of
a
dose
distribution
for
a
UV
reactor.

Model­
based
and
experimental
approaches
have
been
identified
to
determine
the
dose
distribution
of
a
UV
reactor.
Model­
based
approaches
use
computational
fluid
dynamics
(
CFD)
to
predict
microorganism
trajectories
through
a
UV
reactor
and
hence
the
dose
delivered
to
each
microorganism.
Experimental
approaches
use
microspheres
that
undergo
a
chemical
reaction
when
exposed
to
UV
light.
The
microspheres
are
injected
upstream
of
the
UV
reactor
and
are
collected
downstream.
The
extent
of
the
UV­
induced
chemical
reaction
within
each
sphere
is
measured
and
used
to
calculate
the
dose
delivered
to
that
sphere
as
it
traveled
through
the
reactor.
While
promising,
both
model
and
experimental­
based
approaches
are
subjects
of
current
research.
Further
effort
is
necessary
to
prove
these
approaches
as
practical
and
accurate
predictors
of
UV
reactor
performance.

Dose
delivery
by
UV
reactors
is
currently
measured
using
biodosimetry
(
Qualls
and
Johnson
1983).
With
biodosimetry,
inactivation
of
a
challenge
microorganism
passed
through
the
UV
reactor
is
measured
and
related
to
a
single
dose
value
based
on
the
known
UV
dose­
response
of
that
microorganism.
This
dose
value
is
termed
the
RED.

Proposal
Draft
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
3
June
2003
Proposal
Draft
Figure
F.
1
Dose
Distribution
Delivered
by
a
UV
Reactor1
1(
Adapted
from
Chiu
et
al.
1999)

F.
1.1
Relationship
Between
RED
and
the
Dose
Distribution
The
RED
of
a
given
microorganism
depends
on
the
dose
distribution
delivered
by
the
reactor
and
the
UV
inactivation
kinetics
(
dose­
response)
of
the
challenge
microorganism
(
Cabaj
et
al.
1996).
A
general
equation
describing
this
dependence
is
Equation
F.
2:

(
)
(
)


=
=
j
1
i
i
i
D
f
p
RED
f
Equation
F.
2
where
RED
=
RED
measured
using
biodosimetry
f
=
Mathematical
function
describing
the
inactivation
kinetics
of
the
microorganism
j
=
Number
of
dose
values
in
the
dose
distribution
Di
=
ith
dose
in
the
dose
distribution
pi
=
Probability
of
occurrence
of
dose
Di
For
example,
if
the
microorganism
has
first
order
inactivation
kinetics,
the
function
f
is
shown
in
Equation
F.
3:

(
)
(
)
kD.
exp
N
D
f
N
0
­
=
=
Equation
F.
3
where
N
=
Microorganism
concentration
after
exposure
to
dose
D
N0
=
Microorganism
concentration
with
zero
UV
dose
D.
=
Applied
UV
dose
k
=
Microorganism's
first
order
inactivation
coefficient
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
4
June
2003
Proposal
Draft
Substituting
Equation
F.
3
into
F.
2
gives
the
following
equation
for
the
RED
of
a
microorganism
with
first­
order
inactivation
kinetics:

(
)









­
­
=


=
j
i
i
i
kD
p
k
RED
1
exp
ln
1
Equation
F.
4
In
equation
F.
4,
the
RED
depends
on
the
dose
distribution
of
the
UV
reactor
and
the
first
order
inactivation
coefficient
of
the
microorganism.

Figure
F.
2
presents
the
dependence
of
the
RED
on
the
first
order
inactivation
coefficient
of
the
challenge
microorganism
for
the
dose
distribution
shown
in
Figure
F.
1.
The
relation
was
calculated
using
Equation
F.
4.
As
shown,
the
RED
of
a
microorganism
with
a
small
first­
order
inactivation
coefficient
is
greater
than
the
RED
of
a
microorganism
with
a
large
first­
order
inactivation
coefficient.
Because
the
RED
depends
on
the
microorganism's
UV
inactivation
kinetics,
the
RED
of
the
challenge
microorganism
is
an
exact
measure
of
the
RED
delivered
to
a
pathogen
of
interest
only
when
the
challenge
microorganism
has
the
same
inactivation
kinetics
as
the
pathogen
(
Wright
and
Lawryshyn
2000).

Example
1.
A
UV
reactor
delivers
a
dose
distribution
represented
by
Figure
F.
1.
The
UV
reactor
is
evaluated
using
biodosimetry.
The
challenge
microorganisms
are
MS2
bacteriophage
(
MS2)
with
a
first
order
coefficient
of
0.13
cm2/
mJ
and

X174
phage
with
a
first
order
coefficient
of
1.2
cm2/
mJ.
As
shown
in
Figure
F.
2,
MS2
would
have
experienced
1.1
log
inactivation,
corresponding
to
an
RED
of
19
mJ/
cm2.
fX174
would
have
experienced
3.6
log
inactivation,
corresponding
to
an
RED
of
7.3
mJ/
cm2.
If
the
pathogen
of
interest
has
the
same
inactivation
kinetics
as

X174,
the
RED
of
MS2
would
be
2.5
times
greater
than
the
RED
delivered
to
the
pathogen,
while
the
RED
of

X174
would
be
an
exact
measure
of
the
RED
delivered
to
the
pathogen.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
5
June
2003
Proposal
Draft
Figure
F.
2
Microorganism
Log
Inactivation
and
RED
as
a
Function
of
the
Microorganism's
First
Order
Inactivation
Coefficient
for
the
UV
Reactor
Represented
in
Figure
F.
1
F.
1.2
Using
RED
to
Demonstrate
Target
Pathogen
Inactivation
If
the
UV
dose­
response
of
the
challenge
microorganism
differs
from
that
of
the
target
pathogen
and
the
dose
distribution
of
the
UV
reactor
is
not
known,
the
results
of
biodosimetry
can
only
be
used
to
estimate
the
target
pathogen
inactivation
within
a
range
bounded
by
the
inactivation
expected
assuming
ideal
and
worst­
case
hydraulics.
Figure
F.
3
provides
a
comparison
of
the
dose
distribution
of
reactors
with
ideal
and
worst­
case
hydraulics
to
a
dose
distribution
that
might
be
seen
with
a
real
reactor.

Figure
F.
3
Comparison
of
the
Dose
Distributions
of
Ideal,
Realistic,
and
Worst­
Case
UV
Reactors
A
reactor
with
ideal
hydraulics
delivers
the
same
dose
to
all
the
microorganisms
passing
through
the
reactor.
Its
dose
distribution
is
represented
by
a
single
dose.
Examples
of
a
UV
reactor
with
ideal
hydraulics
include
the
stirred
suspension
irradiated
during
the
measurement
of
UV
dose­
response
by
a
collimated
beam
device
and
a
plug
flowrate
reactor
with
complete
lateral
mixing.
In
both
cases,
the
UV
dose
delivered
is
the
product
of
the
average
UV
intensity
within
0
50
100
UV
Dose
(
mJ/
cm
2)
Probability
Ideal
0
50
100
UV
Dose
(
mJ/
cm
2)
Probability
Reality
0
80
UV
Dose
(
mJ/
cm
2)
Probability
Worst
Case
0
Infinity
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
6
June
2003
Proposal
Draft
the
reactor
and
the
residence
time.
With
an
ideal
reactor,
Equation
F.
5
shows
the
net
microbial
inactivation
achieved
by
the
reactor:

(
)
(
)
D
f
D
f
p
N
N
j
1
i
i
i
0
=
=


=
Equation
F.
5
Accordingly,
with
an
ideal
reactor,
the
RED
measured
with
a
challenge
microorganism
is
a
measure
of
the
RED
delivered
to
all
microorganisms
that
pass
through
the
reactor.
If
both
the
challenge
microorganism
and
the
pathogen
have
first
order
inactivation
kinetics,
the
log
inactivation
of
the
pathogen
is
calculated
using
Equation
F.
6:

(
)
(
)
P
p
p
D
RED
RED
k
N
N
10
0
exp
log
log
=
­
­
=









Equation
F.
6
where
log
(
N/
N0)
p
=
Log
inactivation
of
the
pathogen
kp
=
First
order
inactivation
coefficient
of
the
pathogen
RED
=
RED
observed
with
the
pathogen
D10
p
=
UV
sensitivity
of
the
pathogen
expressed
as
mJ/
cm2
per
log
The
UV
sensitivity
of
the
pathogen
is
related
to
the
first
order
inactivation
coefficient
using
Equation
F.
7:

(
)
k
k
D
30
.
2
10
ln
10
=
=
Equation
F.
7
A
UV
reactor
with
worst­
case
hydraulics
delivers
a
UV
dose
of
zero
to
all
microorganisms
passing
through
the
reactor.
However,
in
the
case
of
a
reactor
with
a
measurable
RED,
worst­
case
hydraulics
is
defined
as
infinite
dose
delivered
to
one
fraction
of
the
flowrate
and
zero
dose
delivered
to
the
other
fraction.
Under
these
conditions,
the
net
microbial
inactivation
achieved
by
the
reactor
is
calculated
according
to
Equation
F.
8:

(
)
(
)
(
)
1
2
1
j
1
i
i
i
0
p
f
p
0
f
p
D
f
p
N
N
=
°
+
=
=


=
Equation
F.
8
As
shown,
the
net
inactivation
achieved
by
the
worst­
case
UV
reactor
with
a
measurable
RED
is
constant
and
independent
of
the
inactivation
kinetics
of
the
microorganism.
With
a
worst­
case
UV
reactor,
the
measured
inactivation
is
a
measure
of
the
inactivation
that
would
occur
with
all
microorganisms
regardless
of
their
UV
sensitivity.
In
other
words,
the
log
inactivation
of
the
pathogen
is
calculated
according
to
Equation
F.
9:
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
7
June
2003
Proposal
Draft
c
p
N
N
N
N








=









0
0
log
log
Equation
F.
9
where
log
(
N/
N0)
c
=
log
inactivation
of
the
challenge
microorganism
Using
the
above
definitions
of
an
ideal
and
a
worst­
case
reactor,
the
log
inactivation
of
a
pathogen
estimated
from
biodosmetry
results
will
have
a
value
between
log(
N/
No)
c
and
RED/
Dp.
If
the
inactivation
of
the
pathogen
must
be
known
with
absolute
confidence,
the
lower
bound
of
that
range
should
be
used.
If
the
challenge
microorganism
is
more
resistant
to
UV
light
than
the
pathogen,
the
lower
bound
is
log(
N/
No)
c.
If
the
challenge
microorganism
is
less
resistant
to
UV
light
than
the
pathogen,
the
lower
bound
is
RED/
Dp.

Example
2.
A
UV
reactor
is
challenged
using
MS2
with
a
UV
sensitivity
of
18
mJ/
cm2
per
log
inactivation.
Four
log
inactivation
of
the
MS2
is
observed
corresponding
to
an
MS2
RED
of
4
¥
18
=
72
mJ/
cm2.
The
MS2
results
are
used
to
estimate
the
log
inactivation
of
two
pathogens,
one
with
a
UV
sensitivity
of
10
mJ/
cm2
per
log
inactivation
and
the
other
with
a
UV
sensitivity
of
25
mJ/
cm2
per
log
inactivation.
The
log
inactivation
of
the
first
pathogen
is
estimated
between
4.0
and
72/
10
=
7.2
log
and
the
log
inactivation
of
the
second
pathogen
is
estimated
between
72/
25
=
2.9
and
4.0
log.
The
biodosimetry
results
can
be
used
to
state
with
absolute
confidence
that
the
inactivation
of
the
first
pathogen
was
at
least
4.0
log
and
the
inactivation
of
the
second
pathogen
was
at
least
2.9
log.

Example
3.
A
UV
reactor
is
designed
for
3.0
log
Cryptosporidium
inactivation.
MS2
is
used
to
measure
the
performance
of
the
UV
reactor.
Because
MS2
is
more
resistant
to
UV
light
than
Cryptosporidium,
3.0­
log
MS2
inactivation
must
be
measured
to
state
with
absolute
confidence
that
the
reactor
achieves
3.0­
log
Cryptosporidium
inactivation.

Example
4.
A
UV
reactor
is
designed
for
two
log
adenovirus
inactivation.
Two­
log
adenovirus
inactivation
occurs
using
a
UV
dose
of
100
mJ/
cm2.
The
UV
reactor
is
validated
using
MS2.
Because
adenovirus
is
more
resistant
to
UV
light
than
MS2,
a
RED
of
100
mJ/
cm2
must
be
measured
with
MS2
to
state
with
absolute
confidence
that
the
UV
reactor
achieves
2
log
adenovirus
inactivation.

Because
UV
manufacturers
strive
to
optimize
the
hydraulic
design
of
their
UV
reactors,
using
the
worst­
case
dose
distribution
represented
in
Figure
F.
3
to
define
the
lower
bound
of
pathogen
inactivation
is
overly
conservative.
An
alternative
approach
is
to
use
the
dose
distribution
of
a
commercial
UV
reactor
that
is
representative
of
worst­
case
reactor
hydraulics.
However,
defining
a
worst­
case
commercial
UV
reactor
is
difficult
because
little
data
are
available
in
the
peer­
reviewed
UV
disinfection
literature
on
dose
distributions.
Chiu
et
al.
(
1999)
used
measured
velocity
fields
and
a
random
walk
model
to
predict
the
dose
distribution
delivered
by
a
wastewater
reactor
equipped
with
low­
pressure
(
LP)
lamps
oriented
perpendicular
to
flowrate.
The
dose
distribution
was
bimodal
due
to
a
short­
circuiting
path
along
the
reactor
walls.
Wright
and
Lawryshyn
(
2000)
compared
the
dose
distribution
of
four
reactor
designs
using
CFD­
based
dose
modeling
including
the
reactor
modeled
by
Chiu
et
al.
Based
on
this
comparison,
the
dose
distribution
developed
by
Chiu
et
al.
is
believed
to
represent
a
worst­
case
commercial
UV
reactor.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
8
June
2003
Proposal
Draft
Figure
F.
1
presents
a
dose
distribution
adapted
from
Chiu
et
al.'
s
data.
For
that
dose
distribution,
Figure
F.
2
presents
log
inactivation
and
RED
as
a
function
of
the
microorganism's
UV
sensitivity
expressed
as
a
first­
order
inactivation
coefficient.
Figure
F.
4
presents
the
same
relationship,
but
with
UV
sensitivity
expressed
as
dose
per
log
inactivation.
Using
these
figures,
the
RED
delivered
to
a
pathogen
by
a
given
UV
reactor
can
be
estimated
from
the
measured
RED
of
the
challenge
microorganism
using
Equation
F.
10:

*
*

c
p
c
p
RED
RED
RED
RED
¥
=
Equation
F.
10
where
REDP
=
RED
of
the
pathogen
estimated
for
the
UV
reactor
of
interest
REDc
=
RED
of
the
challenge
microorganism
measured
during
biodosimetry
REDp
=
RED
of
the
pathogen
estimated
from
Figure
F.
2
or
F.
4
REDc
=
RED
of
the
challenge
microorganism
estimated
from
Figures
F.
2
or
F.
4
The
RED
determined
using
Equation
F.
10
represents
the
RED
that
would
be
delivered
if
the
reactor
under
consideration
had
a
dose
distribution
representative
of
a
worst­
case
commercial
reactor.

Figure
F.
4
Microorganism
Inactivation
and
RED
as
a
Function
of
Microorganism
UV
Sensitivity
for
the
UV
Reactor
Represented
in
Figure
F.
1
Example
5.
A
UV
reactor
is
challenged
using
MS2
with
a
UV
sensitivity
of
18
mJ/
cm2
per
log
inactivation.
Four
log
inactivation
of
the
MS2
is
observed
corresponding
to
an
MS2
RED
of
4
¥
18
=
72
mJ/
cm2.
The
MS2
results
are
used
to
estimate
the
log
inactivation
of
two
pathogens,
one
with
a
UV
sensitivity
of
10
/
cm2
per
log
inactivation
and
the
other
with
a
UV
sensitivity
of
25
mJ/
cm2
per
log
inactivation.
In
Figure
F.
4,
the
RED
delivered
to
the
microorganisms
with
a
UV
sensitivity
of
10,
18,
and
25
mJ/
cm2
per
log
inactivation
is
15,
19,
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
9
June
2003
Proposal
Draft
and
21
mJ/
cm2,
respectively.
Assuming
the
UV
reactor's
performance
is
bounded
by
a
worst
case
represented
by
Figure
F.
4,
the
RED
delivered
to
the
first
pathogen
is
estimated
between
72
mJ/
cm2
and
(
72
¥
15)/
19
=
57
mJ/
cm2
and
the
RED
delivered
to
the
second
pathogen
is
estimated
between
72
and
(
72
¥
21)/
19
=
80
mJ/
cm2.
Inactivation
of
the
first
pathogen
is
estimated
between
5.7
(
57/
10)
and
7.2
(
72/
10)
log
and
inactivation
of
the
second
pathogen
is
estimated
between
2.9
(
72/
25)
and
3.2
(
80/
25)
log
inactivation.
This
range
of
inactivation
estimated
using
the
worst­
case
represented
in
Figure
F.
4
is
notably
less
than
the
range
estimated
in
Example
3
using
the
worst­
case
represented
in
Figure
F.
3.

For
regulatory
purposes,
the
lower
bound
of
the
range
of
inactivation
and
RED
estimated
for
the
pathogen
should
be
used
when
relating
challenge
microorganism
inactivation
to
target
pathogen
inactivation.
If
the
challenge
microorganism
is
more
sensitive
to
UV
light
than
the
pathogen
or
if
both
have
the
same
sensitivity,
the
RED
delivered
to
the
pathogen
should
be
estimated
using
the
RED
of
the
challenge
microorganism.
If
the
challenge
microorganism
is
more
resistant
to
UV
light
than
the
pathogen,
the
RED
delivered
to
the
pathogen
should
be
estimated
using
Equation
F.
10.

Example
6.
A
UV
reactor
is
designed
for
three
log
Cryptosporidium
inactivation.
The
dose
needed
for
3
log
Cryptosporidium
taken
from
Chapter
1
(
Table
1.4)
is
12
mJ/
cm2.
Accordingly,
the
UV
sensitivity
of
Cryptosporidium
is
defined
as
12/
3
=
4
mJ/
cm2
per
log
inactivation.
MS2
with
a
UV
sensitivity
of
18
mJ/
cm2
per
log
inactivation
is
used
to
measure
the
performance
of
the
UV
reactor.
Because
MS2
is
more
resistant
to
UV
light
than
Cryptosporidium,
Equation
F.
10
is
used
to
relate
the
RED
measured
using
MS2
to
the
dose
delivered
to
Cryptosporidium.
From
Figure
F.
4,
the
RED
delivered
to
the
microorganisms
with
a
UV
sensitivity
of
3.9
and
18
mJ/
cm2
per
log
inactivation
is
9.8
and
19,
respectively.
Thus
an
MS2
RED
of
12
¥
19/
9.8
=
23
mJ/
cm2
should
be
demonstrated
to
show
3
log
Cryptosporidium
inactivation.

Example
7.
A
UV
reactor
is
designed
for
one­
log
adenovirus
inactivation.
The
dose
needed
for
1­
log
adenovirus
taken
from
Chapter
1
(
Table
1.4)
is
58
mJ/
cm2.
MS2
with
a
UV
sensitivity
of
18
mJ/
cm2
is
used
to
measure
the
performance
of
the
UV
reactor.
Because
MS2
is
less
resistant
to
UV
light
than
adenovirus,
an
MS2
RED
of
58
mJ/
cm2
should
be
demonstrated
to
show
1­
log
adenovirus
inactivation.

The
RED
of
microorganisms
with
shoulders
and
tailing
within
the
dose­
response
curve
depends
on
the
overlap
of
the
dose
distribution
with
those
regions
(
Cabaj
et
al.
1996,
Wright
and
Lawryshyn
2000).
To
use
Figure
F.
2
to
define
safety
factors,
the
inactivation
of
the
challenge
microorganism
should
demonstrate
an
exponential
inactivation
as
a
function
of
dose
over
the
range
of
doses
in
the
dose
distribution.
This
creates
a
dilemma
if
the
dose
distribution
is
not
known.
To
avoid
this
issue,
the
dose­
response
of
an
appropriate
challenge
microorganism
should
not
demonstrate
a
shoulder
at
a
dose
beyond
50
percent
of
the
demonstrated
RED
and
should
not
demonstrate
tailing
until
one
log
inactivation
beyond
the
demonstrated
inactivation.
In
the
case
of
a
challenge
microorganism
with
a
shoulder
and
tailing
in
the
dose­
response,
the
UV
sensitivity
will
be
defined
as
the
sensitivity
over
the
region
of
exponential
inactivation
that
occurs
between
the
shoulder
and
the
onset
of
tailing.
The
shoulder
of
the
dose­
response
is
defined
by
the
intersect
of
the
exponential
region
with
the
dose
axis
(
see
Figure
F.
5).
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
10
June
2003
Proposal
Draft
Example
8.
Figure
F.
5
presents
the
measured
UV
dose­
response
of
B.
subtilis
spores.
Because
the
measured
dose­
response
has
a
shoulder
of
16.5
mJ/
cm2,
the
B.
subtilis
spores
should
only
be
used
to
demonstrate
RED
values
greater
than
or
equal
to
2
¥
16.5
=
33
mJ/
cm2.

Figure
F.
5.
UV
Dose­
Response
of
B.
subtilis
Spores
(
Adapted
from
Sommer
et
al.
1998)

The
RED
safety
factor
provides
an
incentive
to
select
a
challenge
microorganism
whose
UV
sensitivity
matches
that
of
the
target
pathogen
and
a
disincentive
for
overrating
UV
reactor
performance
by
using
challenge
microorganisms
whose
UV
sensitivity
is
much
greater
than
the
target
pathogen.

F
.1.3
Biodosimetry
Using
Two
Challenge
Microorganisms
In
order
to
provide
a
better
estimate
of
the
target
pathogen's
log
inactivation
and
RED,
two
microorganisms
with
different
UV
sensitivities
can
be
used
to
validate
UV
reactors.
The
target
pathogen's
log
inactivation
should
be
estimated
by
interpolating
the
log
inactivation
of
the
two
microorganisms
as
a
function
of
the
UV
sensitivity
defined
on
a
linear
scale
as
a
first­
order
inactivation
coefficient.
Alternatively,
the
target
pathogen's
RED
should
be
estimated
by
interpolating
the
RED
of
the
two
microorganisms
as
a
function
of
the
UV
sensitivity
defined
on
a
linear
scale
as
dose
per
log
inactivation.
If
interpolation
does
not
meet
these
provisions,
the
inactivation
of
the
pathogen
will
be
overestimated.

Example
9.
A
UV
reactor
with
a
dose
distribution
represented
in
Figure
F.
4
is
tested
using
MS2
and
fX174.
The
MS2
and
fX174
have
a
UV
sensitivity
of
18
and
2
mJ/
cm2
per
log
inactivation.
Using
biodosimetry,
1.1
and
3.6
log
inactivation
of
MS2
and
fX174
are
measured.
These
log
inactivations
correspond
to
RED
values
of
20
and
7.2
mJ/
cm2,
respectively.
The
RED
measured
with
MS2
and
fX174
is
fit
as
a
function
of
UV
sensitivity
resulting
in
the
following
equation:
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
11
June
2003
Proposal
Draft
83
.
5
731
.
0
+
¥
=
y
Sensitivit
UV
RED
This
equation
predicts
that
the
RED
delivered
to
Cryptosporidium,
defined
with
a
UV
sensitivity
of
3.9
mJ/
cm2
per
log
inactivation,
is
8.7
mJ/
cm2.

If
the
inactivation
of
the
more
UV­
sensitive
of
the
two
challenge
microorganisms
is
greater
than
the
detection
limit
of
the
assay,
interpolation
should
be
based
on
the
level
indicated
by
the
limitation.
Because
the
inactivation
of
the
UV­
sensitive
microorganism
is
underestimated,
the
interpolation
will
be
conservative
and
two­
microorganism
validation
may
not
offer
an
advantage
over
single
microorganism
validation.

Example
10.
A
UV
reactor
is
evaluated
using
MS2
and
fX174
phage.
MS2
and
fX174
are
injected
into
the
flowrate
upstream
of
the
reactor.
Influent
and
effluent
samples
are
collected
and
assayed.
The
assay
has
a
detection
limit
of
1
pfu/
mL.
The
concentrations
of
MS2
and
fX174
in
the
influent
is
determined
as
1,000,000
and
10,000
pfu/
mL,
respectively.
The
concentrations
of
MS2
and
fX174
in
the
effluent
samples
are
10,000
and
0
pfu/
mL,
respectively.
The
results
indicate
that
the
concentration
of
fX174
is
below
the
detection
limit
of
the
assay.
Accordingly,
the
log
inactivation
of
MS2
and
fX174
is
2
log
and
>
4
log,
respectively.
If
the
UV
sensitivity
of
MS2
and
fX174
are
determined
to
be
20
and
2
mJ/
cm2
per
log,
the
MS2
RED
is
40
mJ/
cm2
and
the
fX174
RED
is
>
8
mJ/
cm2.
The
following
equation
fits
the
measured
RED
as
a
function
of
UV
sensitivity:

44
.
4
77
.
1
+
¥
=
y
Sensitivit
UV
RED
This
equation
predicts
that
the
RED
delivered
to
Cryptosporidium
defined
with
a
UV
sensitivity
of
3.9
mJ/
cm2
per
log
is
11.3
mJ/
cm2.
This
compares
to
an
RED
of
20
mJ/
cm2
that
would
have
been
predicted
by
Equation
F.
10
using
the
MS2
data
alone.
In
this
case,
twomicroorganism
biodosimetry
estimated
lower
dose
delivery
to
Cryptosporidium
than
single
microorganism
biodosimetry.

In
the
past,
it
has
been
assumed
that
the
RED
measured
with
a
UV­
resistant
challenge
microorganism
can
be
used
to
demonstrate
compliance
with
a
dose
target
while
the
log
inactivation
demonstrated
with
a
UV­
sensitive
challenge
microorganism
can
be
used
to
demonstrate
compliance
to
a
log
inactivation
target.
This
approach
is
not
recommended.
It
is
not
possible
to
demonstrate
compliance
to
a
3­
log
Cryptosporidium
inactivation
by
using
UVresistant
MS2
to
show
an
RED
of
11.7
mJ/
cm2
and
using
UV­
sensitive
fX174
to
show
3­
log
inactivation.

E
xample
11.
In
Example
9,
even
though
the
RED
measured
with
MS2
was
18
mJ/
cm2
and
the
log
inactivation
measured
with
fX174
was
3.6
log,
Figure
F.
4
shows
that
Cryptosporidium,
defined
with
a
UV
sensitivity
of
3.9
mJ/
cm2
per
log,
experienced
a
log
inactivation
of
2.5
corresponding
to
an
RED
of
9.8
mJ/
cm2.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
12
June
2003
Proposal
Draft
F.
1.4
Challenge
Microorganism
Selection
Ideally,
UV
reactor
performance
should
be
validated
with
a
microorganism
whose
UV
sensitivity
matches
that
of
the
target
pathogen.
In
this
guidance
document,
the
UV
sensitivity
of
the
target
microorganisms
is
given
by
the
dose
requirements
given
in
Chapter
1
for
Cryptosporidium,
Giardia,
and
virus.
Challenge
microorganisms
currently
used
to
validate
UV
reactors
do
not
have
a
UV­
sensitivity
that
matches
the
UV­
sensitivity
of
the
target
pathogens
as
defined
in
Chapter
1.
The
UV­
resistance
of
MS2
and
B.
subtilis
spores
is
notably
greater
than
that
of
Cryptosporidium
and
Giardia,
and
notably
less
than
that
of
adenovirus.
Furthermore,
demonstrating
3
or
4­
log
virus
inactivation
with
these
challenge
microorganisms
necessitates
demonstrating
REDs
greater
than
150
mJ/
cm2.
These
REDs
correspond
to
greater
than
6­
log
inactivation
of
MS2
and
B.
subtilis
spores.
Currently,
culturing
titers
of
challenge
microorganisms
needed
to
demonstrate
greater
than
6­
log
inactivation
are
not
practical.

A
challenge
microorganism
should
have
reproducible
UV
inactivation
kinetics
over
the
dose
range
of
interest.
The
challenge
microorganism
should
be
easily
prepared
in
high
titers,
easily
enumerated
by
an
assay
based
on
microorganism
replication,
non­
pathogenic
to
humans,
and
not
harmful
to
the
environment.
If
the
challenge
microorganism
is
a
phage,
the
host
bacteria
used
to
assay
the
phage
concentration
should
not
be
pathogenic
to
humans.
MS­
2
phage,
non­
pathogenic
Escherichia
coli,
B.
subtilis
spores,
and
Saccharomyces
cerevisae
have
been
used
to
bioassay
UV
reactors
designed
to
treat
drinking
water.
Table
F.
1
summarizes
the
UV
sensitivity
of
commonly­
used
and
candidate
bioassay
microorganisms.

Table
F.
1
UV
Sensitivity
of
Bioassay
Microorganisms
and
Candidates
Dose
(
mJ/
cm2)
Reported
to
Achieve
Microorganism
1
log
2
log
3
log
4
log
Reference
MS­
2
phage
16
34
52
71
Wilson
et
al.
1992
E.
Coli
3.0
4.8
6.7
8.4
Chang
et
al.
1985
B.
subtilis
spores
28
39
50
62
Sommer
et
al.
1998
fx174
phage
2.2
5.3
7.3
11
Sommer
et
al.
1998
B40­
8
phage
12
18
23
28
Sommer
et
al.
1998
PRD­
1
phage
9.9
17
24
30
Meng
and
Gerba
1996
F.
2
Dose
Monitoring
There
are
three
approaches
currently
used
to
monitor
dose
delivery.
In
this
guidance
document,
the
terms
used
are
as
follows:

·
UV
intensity
setpoint
approach
·
UV
intensity
and
UVT
setpoint
approach
·
Calculated
dose
approach
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
13
June
2003
Proposal
Draft
With
the
UV
intensity
setpoint
approach,
dose
delivery
is
indicated
by
measured
flowrate
and
UV
intensity.
The
UV
reactor
complies
with
a
required
dose
delivery
when
the
measured
UV
intensity
is
above
an
alarm
setpoint
value
defined
as
a
function
of
flowrate
through
the
reactor.
With
this
approach,
the
UV
intensity
sensor
should
be
positioned
far
enough
from
the
lamp
that
it
provides
measurable
responses
to
changing
water
UV
absorbance
(
and
corresponding
UVT)
as
well
as
lamp
output.
With
the
UV
intensity
and
UVT
setpoint
approach,
dose
delivery
is
indicated
by
measured
flowrate,
UV
intensity,
and
UVT.
The
UV
reactor
complies
with
a
required
dose
delivery
when
the
measured
UV
intensity
and
UVT
are
above
alarm
setpoint
values,
both
defined
as
a
function
of
flowrate
through
the
reactor.
With
this
approach,
the
UV
intensity
sensor
should
be
positioned
relatively
close
to
the
lamp
so
that
it
responds
primarily
to
changing
lamp
output.
With
the
calculated
dose
approach,
dose
delivery
is
indicated
by
a
dose
value
calculated
from
measured
flowrate,
UV
intensity,
and
UVT.
The
UV
reactor
complies
with
a
required
dose
delivery
when
the
calculated
dose
is
above
an
alarm
setpoint
value.
With
this
approach,
there
are
no
requirements
for
sensor
positioning.

To
illustrate
the
UV
intensity
setpoint
approach
and
the
UV
intensity
and
UVT
setpoint
approach,
Figures
F.
6,
F.
7,
and
F.
8
present
the
relationship
between
UV
dose
and
measured
UV
intensity
for
an
annular
reactor
containing
a
single
LP
lamp.
UV
intensity
was
calculated
using
a
radial
UV
intensity
model
and
UV
dose
was
calculated
assuming
ideal
hydraulics
(
Haas
and
Sakellaropoulos
1979).
UV
intensity
and
dose
were
calculated
for
a
fixed
flowrate
of
400
gpm,
water
UVT
ranging
from
60
to
98
percent,
and
lamp
output
ranging
from
20
to
100
percent.
In
each
figure,
data
are
presented
as
plots
of
dose
versus
UV
intensity
sensor
reading
for
values
of
UVT
specified
in
the
legend.
For
each
of
those
plots,
each
point
at
a
given
UVT
represents,
in
order
of
increasing
dose,
operation
at
20,
40,
60,
80,
and
100
percent
lamp
power.
The
differences
between
these
figures
are
due
to
sensor
placement.

Figure
F.
6
Relationship
between
UV
Dose
and
Intensity
for
a
UV
Intensity
Sensor
Located
to
Give
Dose
Proportional
to
Measured
Irradiance
0
10
20
30
40
50
60
70
0
20
40
60
Sensor
(
mW/
cm2)
Dose
(
mJ/
cm2)

60%

70%

80%

85%

90%

94%

98%
S
UVT
254
nm
400
GPM
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
14
June
2003
Proposal
Draft
Figure
F.
7
Relationship
between
UV
Dose
and
Intensity
for
a
UV
Intensity
Sensor
Located
Close
to
the
Lamp
Figure
F.
8
Relationship
between
UV
Dose
and
Intensity
for
a
UV
Intensity
Sensor
Located
Far
from
the
Lamp
F.
2.1
UV
Intensity
Setpoint
Approach
Figure
F.
6
presents
the
relationship
obtained
when
the
UV
intensity
sensor
is
located
at
a
distance
from
the
lamps
where
UV
dose
is
proportional
to
measured
UV
intensity
regardless
of
the
UVT
and
lamp
output.
With
an
ideal
reactor,
this
sensor
location
occurs
where
the
measured
intensity
equals
the
average
intensity
within
the
reactor.
Because
of
the
proportional
relationship
between
dose
delivery
and
measured
intensity,
a
given
intensity
can
be
related
to
a
specific
level
of
dose
delivery.

Example
12.
The
UV
reactor
characterized
in
Figure
F.
6
is
used
in
a
disinfection
application
needing
a
UV
dose
of
20
mJ/
cm2.
At
a
flowrate
of
400
gpm,
a
UV
intensity
value
S
0
10
20
30
40
50
60
70
0
5
10
15
20
Sensor
(
mW/
cm2)
Dose
(
mJ/
cm2)

60%
70%
80%
85%
90%
94%
98%
S
UVT
254
nm
400
GPM
S'
0
10
20
30
40
50
60
70
0
20
40
60
80
100
120
140
160
Sensor
(
mW/
cm2)
Dose
(
mJ/
cm2)

60%

70%

80%

85%

90%

94%

98%
S
UVT
254
nm
400
GPM
S'
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
15
June
2003
Proposal
Draft
of
18
mW/
cm2
is
used
as
an
alarm
setpoint
to
indicate
the
UV
reactor
delivers
a
dose
of
20
mJ/
cm2.
This
alarm
setpoint
value
will
indicate
a
dose
of
20
mJ/
cm2
regardless
of
the
UVT
of
the
water
and
the
output
of
the
lamps.

Figure
F.
7
presents
the
relationship
between
dose
delivery
and
measured
UV
intensity
when
the
UV
intensity
sensor
is
placed
closer
to
the
lamp
than
the
sensor
in
Figure
F.
6.
Because
the
sensor
views
the
lamp
through
a
relatively
thin
water
layer,
the
sensor
response
to
changing
UVT
is
small
compared
to
that
in
Figure
F.
6.
Accordingly,
the
relationship
between
dose
delivery
and
measured
intensity
for
different
values
of
UVT
cannot
be
described
by
a
single
proportional
relationship.
Unlike
Figure
F.
6,
a
given
UV
intensity
is
not
related
to
a
specific
level
of
dose
delivery
but
is
related
to
a
range
of
delivered
doses.
Accordingly,
the
measured
UV
intensity
should
only
be
used
to
indicate
dose
delivery
at
the
lower
end
of
that
range,
which
occurs
under
conditions
of
maximum
lamp
power
and
reduced
UVT.

Example
13.
The
UV
reactor
characterized
in
Figure
F.
7
is
used
in
an
application
needing
a
UV
dose
of
20
mJ/
cm2.
The
UV
manufacturer
states
that
a
UV
intensity
value
S
of
80
mW/
cm2
will
indicate
a
dose
of
20
mJ/
cm2
under
design
conditions
of
85
percent
UVT
and
60
percent
lamp
output.
However,
as
shown
in
Figure
F.
7,
an
intensity
of
80
mW/
cm2
corresponds
to
a
dose
ranging
from
5
to
37
mJ/
cm2.
The
lower
end
of
this
range
occurs
with
lamp
powers
higher
that
60
percent
and
water
UVT
lower
than
85
percent.
For
a
UV
intensity
alarm
setpoint
to
ensure
a
dose
of
20
mJ/
cm2
under
all
possible
conditions
of
the
water
UVT
and
lamp
output,
a
setpoint
value
S'
of
134
mW/
cm2
should
be
chosen.

Figure
F.
8
presents
the
relationship
between
dose
delivery
and
measured
UV
intensity
when
the
UV
intensity
sensor
is
located
further
from
the
lamps
than
the
sensor
in
Figure
F.
6.
Because
the
sensor
views
the
lamp
through
a
relatively
thick
water
layer,
the
sensor
response
to
changing
water
transmittance
is
large
compared
to
that
in
Figure
F.
6.
Like
Figure
F.
7,
the
relationship
between
dose
delivery
and
measured
intensity
for
different
values
of
UVT
cannot
be
described
by
a
single
proportional
relationship.
As
such,
a
given
intensity
value
is
related
to
a
range
of
dose
values
as
opposed
to
a
single
value.
Again,
the
measured
UV
intensity
should
only
be
used
to
indicate
dose
delivery
at
the
lower
end
of
that
range.
However,
unlike
Figure
F.
7,
the
lower
end
of
the
range
occurs
under
conditions
of
reduced
lamp
power
and
maximum
UVT.

Example
14.
The
UV
reactor
characterized
in
Figure
F.
8
is
used
in
an
application
needing
a
UV
dose
of
20
mJ/
cm2.
The
UV
reactor
uses
the
UV
intensity
setpoint
approach
to
monitor
dose
delivery.
A
UV
intensity
alarm
setpoint
value
S
of
4
mW/
cm2
is
proposed
based
on
the
UV
intensity
measured
under
design
conditions
of
85
percent
UVT
and
60
percent
lamp
output.
However,
an
intensity
of
4
mW/
cm2
indicates
a
dose
ranging
from
9
to
26
mJ/
cm2.
To
indicate
a
dose
of
20
mJ/
cm2
using
the
UV
intensity
setpoint
approach,
a
setpoint
value
S'
of
8
mW/
cm2
should
be
chosen.

The
location
of
the
UV
intensity
sensor
within
a
UV
reactor
is
selected
by
the
manufacturer
of
the
UV
reactor.
If
the
UV
reactor
uses
the
UV
intensity
setpoint
approach
for
dose
monitoring,
the
UV
manufacturer
should
optimize
the
UV
intensity
sensor's
location
to
give
a
proportional
relationship
between
dose
delivery
and
measured
UV
intensity
similar
to
the
example
given
in
Figure
F.
6.
If
the
UV
manufacturer
does
not
optimize
the
UV
intensity
sensor's
location,
a
given
UV
intensity
will
correspond
to
a
range
of
UV
doses
values
as
opposed
to
a
single
value.
While
this
does
not
prevent
the
UV
reactor
from
using
the
UV
intensity
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
16
June
2003
Proposal
Draft
setpoint
approach,
the
monitoring
approach
will
not
be
as
efficient
as
with
an
optimally
located
sensor
because
the
UV
reactor
will
be
overdosing
at
many
combinations
of
UVT
and
lamp
power
that
given
rise
to
operation
at
the
setpoint.

F.
2.2
UV
Intensity
and
UVT
Setpoint
Approach
If
the
UV
intensity
sensor
is
not
at
a
location
optimal
for
the
UV
intensity
setpoint
approach,
measurements
of
UVT
can
be
used
to
provide
more
efficient
dose
monitoring.
UVT
alarm
setpoints
combined
with
UV
intensity
alarm
setpoints
can
be
used
to
indicate
dose
delivery
providing
the
UV
intensity
sensor
is
placed
relatively
close
to
the
lamp.
With
the
sensor
located
relatively
close
to
the
lamp,
dose
delivery
at
a
given
intensity
and
flowrate
decreases
with
decreasing
UVT
(
Figure
F.
7).
Accordingly,
a
UVT
alarm
setpoint
combined
with
a
UV
intensity
alarm
setpoint
provides
a
meaningful
indicator
of
dose
delivery.

Example
15.
The
UV
reactor
characterized
in
Figure
F.
7
is
used
in
an
application
needing
a
UV
dose
of
20
mJ/
cm2.
If
the
UV
reactor
used
the
UV
intensity
setpoint
approach
to
monitor
dose
delivery,
an
alarm
setpoint
S'
of
134
mW/
cm2
would
be
used
to
indicate
a
dose
delivery
of
20
mJ/
cm2.
This
approach
is
not
efficient
because
a
UV
intensity
of
134
mW/
cm2
is
associated
with
a
UV
dose
ranging
from
20
to
60
mJ/
cm2.
An
alternative
approach
for
dose
monitoring
is
to
use
the
UV
intensity
and
UVT
setpoint
approach.
Under
this
approach,
a
UV
intensity
alarm
setpoint
S
of
80
mW/
cm2
combined
with
a
UVT
alarm
setpoint
of
85
percent
will
indicate
a
dose
delivery
of
20
mJ/
cm2.
However,
the
approach
is
still
inefficient
because
UV
dose
may
range
from
20
to
38
mJ/
cm2
with
operation
of
the
reactor
at
the
setpoint
conditions.

If
the
UV
intensity
sensor
is
located
at
the
optimal
position
for
the
UV
intensity
setpoint
approach
(
Figure
F.
6),
the
UVT
reading
does
not
provide
any
additional
information
on
dose
delivery
that
is
not
provided
by
the
measured
UV
intensity.
However,
the
measured
UVT
could
be
used
to
indicate
whether
a
UV
intensity
alarm
condition
arises
from
low
UVT.

If
the
UV
intensity
sensor
is
located
too
far
from
the
lamp,
dose
delivery
at
a
given
UV
intensity
and
flowrate
increases
with
decreasing
UVT
(
Figure
F.
8).
As
such,
the
UVT
reading
cannot
be
used
as
an
alarm
setpoint
to
indicate
dose
delivery.

E
xample
16.
The
UV
reactor
characterized
in
Figure
F.
8
uses
the
UV
intensity
and
UVT
setpoint
approach
to
show
the
UV
reactor
delivers
a
dose
of
20
mJ/
cm2.
The
intensity
alarm
setpoint
is
set
to
5
mW/
cm2
and
the
UVT
alarm
setpoint
is
set
to
90
percent.
If
the
reactor
was
operating
with
a
measured
UVT
and
UV
intensity
of
85
percent
and
5
mW/
cm2,
the
delivered
dose
would
be
28
mJ/
cm2.
If
the
reactor
was
operating
with
a
UVT
and
UV
intensity
of
98
percent
and
5
mW/
cm2,
respectively,
the
delivered
dose
would
be
12
mJ/
cm2.
Thus
the
two
alarm
setpoint
values
are
not
ensuring
the
UV
reactor
complies
with
a
dose
of
20
mJ/
cm2.
To
remedy
this
problem,
the
UV
manufacturer
should
either
uses
the
UV
intensity
setpoint
approach,
move
the
UV
intensity
sensor
closer
to
the
lamps,
or
use
the
calculated
dose
approach
to
monitor
dose
delivery.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
17
June
2003
Proposal
Draft
F.
2.3
Calculated
Dose
Approach
Measurements
of
flowrate,
UV
intensity,
and
UVT
can
be
incorporated
into
theoretical,
empirical,
or
semi­
empirical
calculations
of
dose
delivery.
For
example,
the
relationships
represented
in
Figures
F.
6
to
F.
8
could
be
defined
experimentally
and
used
in
an
empirical
manner
to
calculate
dose.
Relationships
could
also
be
defined
using
advanced
modeling
approaches
and
used
to
relate
measured
intensity
to
dose
delivery
for
a
given
flowrate
and
UVT.
In
theory,
the
dose
calculation
does
not
necessitate
that
the
sensor
be
placed
at
any
one
location
within
the
reactor.
However,
if
the
sensor
placed
at
a
location
that
gives
dose
delivery
proportional
to
the
sensor
reading,
the
dose
calculation
does
not
require
UVT
as
an
input
parameter.

F.
2.4
Validating
Dose
Monitoring
The
test
conditions
used
to
validate
a
UV
reactor
should
depend
on
the
approach
used
to
monitor
dose
delivery.

If
the
UV
reactor
uses
the
UV
intensity
setpoint
approach,
the
UV
reactor
is
validated
by
measuring
the
dose
delivery
with
the
UV
intensity
adjusted
to
the
UV
intensity
alarm
setpoint
value.
The
combination
of
lamp
power
and
UVT
used
to
achieve
operation
at
the
alarm
setpoint
should
be
selected
to
capture
the
lower
end
of
the
dose
range
associated
with
the
setpoint.
If
the
UV
intensity
sensor
is
located
closer
to
the
lamp
than
the
optimal
location,
the
UV
reactor
should
be
validated
at
peak
lamp
power
and
lowered
UVT.
If
the
UV
intensity
sensor
is
located
further
from
the
lamp
than
the
optimal
location,
the
UV
reactor
should
be
validated
at
peak
UVT
and
lowered
lamp
power.
If
the
positioning
of
the
UV
intensity
sensor
relative
to
the
optimal
location
is
not
known
prior
to
validation
testing,
the
UV
reactor
should
be
validated
using
both
test
conditions.
If
the
dose
values
measured
with
both
test
conditions
are
the
same,
the
UV
intensity
sensor
is
at
the
optimal
location.

If
the
UV
reactor
uses
the
UV
intensity
and
UVT
setpoint
approach,
the
UV
reactor
is
validated
by
measuring
dose
delivery
with
the
UV
intensity
and
UVT
adjusted
to
the
alarm
setpoint
values.
Validation
should
also
confirm
that
the
UV
intensity
sensor
is
located
close
enough
to
the
lamp
that
UVT
alarm
setpoint
values
provide
a
meaningful
indicator
of
dose
delivery.
This
is
accomplished
by
showing
that
dose
delivery
decreases
with
decreasing
UVT
while
the
UV
intensity
is
held
constant
at
the
intensity
alarm
setpoint
value.
If
dose
delivery
increases
with
decreased
UVT,
the
UV
intensity
sensor
is
located
too
far
from
the
lamp
and
this
monitoring
approach
will
not
work.

If
the
reactor
uses
dose
calculations,
validation
testing
confirms
that
dose
delivery
is
greater
than
or
equal
to
the
calculated
dose.
Validation
testing
is
conducted
at
various
combinations
of
flowrate,
lamp
output,
and
UVT
that
result
in
performance
at
a
target
dose.
This
proves
the
dose
calculation
is
robust
over
the
range
of
those
variables
expected
with
operation
of
the
reactor
at
a
water
treatment
plant
(
WTP).
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
18
June
2003
Proposal
Draft
F.
3
UV
Intensity
Sensors
UV
reactors
should
be
equipped
with
at
least
one
on­
line
UV
intensity
sensor
that
measures
the
UV
intensity
at
some
point
within
the
UV
reactor.
Measurements
made
by
the
on­
line
UV
sensors
are
used
to
indicate
dose
delivery
by
the
UV
reactor.
Reference
sensors
are
used
to
check
that
the
measurements
made
by
the
on­
line
sensors
are
valid.

F.
3.1
UV
Sensor
Properties
The
UV
sensor
may
or
may
not
measure
the
UV
light
through
a
monitoring
window
that
is
separate
from
the
sensor
body.
The
monitoring
windows
should
have
a
high
UVT
over
the
spectral
response
range
of
the
UV
sensors.

The
UV
intensity
sensor
should
detect
germicidal
UV
radiation
and
produce
a
standardized
output
signal
(
e.
g.,
4
to
20
mA)
proportional
to
the
UV
irradiance
incident
on
the
sensor.
UV
intensity
sensors
should
be
calibrated
to
an
absolute
irradiance
standard
and
have
a
suitable
measurement
range,
angular
response,
spectral
response,
linearity,
and
stability
for
monitoring
and
controlling
UV
dose
delivery
by
the
UV
reactor.
An
ideal
UV
intensity
sensor
has
a
linear
response
to
incident
UV
irradiance
that
is
independent
of
water
temperature
and
does
not
degrade
with
time.
Furthermore,
the
ideal
sensor
has
a
fixed
angular
response
and
a
wavelength
response
that
mimics
the
germicidal
response
of
microorganisms.

UV
intensity
sensors
provided
by
the
manufacturer
should
be
individually
calibrated.
UV
intensity
sensors
used
to
monitor
LP
lamps
are
often
calibrated
using
the
substitution
method
(
Larason
et
al.
1998).
With
this
approach,
the
intensity
of
a
collimated
beam
of
UV
light
at
254
nm
is
measured
using
the
UV
sensor
and
compared
to
that
made
using
a
standard
measurement,
such
as
a
National
Institute
of
Standards
and
Technology
(
NIST)
traceable
sensor
or
chemical
actinometer.
The
ratio
of
the
standard
measurement
to
the
sensor
output
is
the
calibration
factor.
With
sensors
designed
to
measure
the
output
of
medium­
pressure
(
MP)
lamps,
the
sensor
can
be
either
calibrated
at
254
nm,
calibrated
as
a
function
of
wavelength,
or
calibrated
using
polychromatic
light
from
a
MP
lamp
with
a
known
spectral
output.
Regardless
of
the
approach
used,
the
calibration
should
be
traceable
to
some
absolute
measurement
standard
and
have
a
quantified
measurement
uncertainty.

Sensor
linearity
is
determined
by
comparing
the
sensor
output
as
a
function
of
incident
irradiance
to
standard
measurements
of
that
irradiance.
Sensor
temperature
response
is
determined
by
measuring
the
dependence
of
sensor
output
on
the
sensor's
operating
temperature
with
the
sensor
measuring
a
constant
irradiance.
Both
linearity
and
temperature
response
should
be
determined
over
the
range
of
irradiance
and
temperature
expected
with
the
operation
of
the
UV
reactor
at
the
WTP.
Angular
response
of
a
sensor
is
determined
by
measuring
the
dependence
of
the
sensor
output
on
the
incident
angle
of
collimated
UV
light
of
fixed
intensity.

The
spectral
response
of
a
sensor
is
determined
by
measuring
the
dependence
of
the
sensor
output
on
the
wavelength
of
monochromatic
light
of
known
irradiance
incident
on
the
sensor.
Spectral
response
is
typically
presented
as
a
plot
of
the
ratio
of
sensor
output
to
incident
irradiance
as
a
function
of
the
wavelength
of
light.
Because
it
may
be
affected
by
infrared
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
19
June
2003
Proposal
Draft
transmission
of
glass
filters
and
fluorescence
of
diffusers
that
are
part
of
the
sensor
(
Larason
and
Cromer
2001),
UV
intensity
sensor
spectral
response
should
be
evaluated
from
200
to
1000
nm.

The
long­
term
stability
of
a
UV
sensor
is
best­
determined
using
field
data
but
may
be
estimated
using
accelerated
life
cycle
testing.
The
measurement
accuracy
of
UV
sensors
can
change
over
time
with
operation
and
environmental
exposure.
Temperature
cycling,
exposure
to
UV
light,
mechanical
vibration,
and
other
factors
will
impact
the
linear,
spectral,
angular,
and
temperature
response
of
a
sensor.

The
UV
sensor
manufacturer
should
conduct
regular
testing
on
manufactured
UV
sensors
to
develop
a
database
on
sensor
properties.
While
some
sensor
properties
may
be
measured
with
each
sensor,
other
properties,
such
as
long­
term
stability,
can
only
be
measured
on
a
representative
lot
size.
The
sensor
manufacturer
should
have
available
for
inspection
the
following
information:

·
Description
of
the
properties
measured
·
Description
of
the
measurement
system
used
to
measure
each
property
·
Description
of
the
measurement
standards
used
·
Documented
uncertainty
of
each
measurement
·
Description
of
QA/
QC
procedures
used
to
ensure
the
measurements
are
traceable
·
Data
collected
over
time
that
demonstrates
that
the
properties
of
the
manufactured
sensors
meet
specifications
F
.3.2
UV
Intensity
Sensor
Measurement
Uncertainty
The
measurement
uncertainty
of
a
UV
intensity
sensor
quantifies
how
the
measurement
of
UV
intensity
made
by
the
sensor
when
mounted
on
the
UV
reactor
compares
with
the
true
value.
For
the
purposes
of
this
guidance
document,
UV
intensity
sensor
uncertainty
should
be
determined
at
a
90
percent
confidence
level
by
summing
the
uncertainty
that
arises
from
the
calibration,
linearity,
angular
and
spectral
response,
temperature
response,
and
long­
term
stability.

The
uncertainty
of
sensor
calibration
depends
on
the
uncertainty
of
the
standards
and
instrumentation
used
to
calibrate
the
sensor,
such
as
voltmeters
and
amplifiers.
Uncertainty
arises
from
linearity
and
temperature
response
because
sensor
calibration
factors,
determined
at
a
given
temperature
and
UV
irradiance,
are
used
over
a
range
of
temperatures
and
irradiances
with
operation
of
the
sensor
with
the
UV
reactor.
Uncertainty
arises
with
sensor
degradation
because
calibration
factors
are
determined
on
new
sensors.

Uncertainty
arises
with
angular
response
because
sensors,
calibrated
using
collimated
UV
light,
are
used
in
UV
reactors
to
measure
UV
light
impacting
from
different
directions.
Uncertainty
arises
with
spectral
response
because
sensors,
calibrated
at
a
fixed
wavelength,
are
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
20
June
2003
used
in
UV
reactors
equipped
with
MP
lamps.
Variability
in
spectral
and
angular
response
from
sensor
to
sensor
will
result
in
a
measurement
uncertainty
not
accounted
for
in
calibration.
The
impact
of
spectral
and
angular
response
variability
on
sensor
measurement
uncertainty
can
be
determined
either
by
calculation
or
by
measurement.
In
the
first
approach,
the
sensor
spectral
and
angular
response
measured
on
a
representative
lot
size
is
used
as
an
input
to
a
model
that
predicts
sensor
readings
in
a
UV
reactor.
The
variability
in
the
sensor
readings
predicted
by
the
model
is
used
to
define
an
uncertainty
term
that
is
included
in
the
calculation
of
sensor
uncertainty.
In
the
second
approach,
the
variability
in
measurements
made
by
a
representative
number
of
sensors
mounted
on
the
UV
reactor
is
used
to
define
the
uncertainty.

Example
17.
A
UV
sensor
manufacturer
calibrates
each
manufactured
UV
intensity
sensor
at
20
°
C
with
an
uncertainty
of
5
percent.
UV
intensity
sensor
linearity,
temperature
response,
angular
response,
and
spectral
response
is
evaluated
on
every
tenth
sensor
manufactured.
Linearity
ranges
from
1
to
3
percent
over
the
measurement
range
of
the
sensor.
Temperature
response
ranges
from
0.1
to
0.2
percent
per
C
°
,
or
an
uncertainty
of
4
percent
from
0
to
40
°
C.
Models
predict
that
the
variability
in
angular
and
spectral
response
from
sensor
to
sensor
will
cause
uncertainties
of
8
and
4
percent,
respectively.
An
evaluation
of
sensors
returned
from
the
field
indicates
that
the
long­
term
drift
over
a
one­
year
period
is
10
percent.
The
measurement
uncertainty
of
the
sensors
is
calculated
as
the
square
root
of
the
sum
of
the
squares
of
the
individual
uncertainties
as
per:

percent
15
10
4
8
4
3
5
M
2
2
2
2
2
2
=
+
+
+
+
+
=
ty
uncertain
easurement
F.
3.3
On­
line
and
Reference
UV
Intensity
Sensors
Degradation
in
UV
intensity
sensor
performance
can
lead
to
significant
under­
or
overestimations
of
dose
delivery
by
the
UV
reactor's
on­
line
monitoring
system.
To
prevent
underdosing,
the
measurement
uncertainty
of
the
UV
intensity
sensors
should
be
incorporated
as
a
safety
factor
into
the
sizing
and
operation
of
a
UV
installation
and
the
performance
of
the
online
sensor
should
be
regularly
checked
by
use
of
a
reference
sensor.
Measurements
made
by
the
on­
line
and
reference
sensor
should
meet
the
following
equation:

[
]
2
/
1
2
Duty
2
Ref
Ref
Duty
 
 
100
1
I
I
+
 
×
 

  
 

 

  
 
 
Equation
F.
11
where
IRef
=
Intensity
measured
with
the
reference
sensor
(
W/
m2)
IDuty
=
Intensity
measured
with
the
duty
sensor
(
W/
m2)
 Ref
=
Measurement
Uncertainty
of
the
reference
UV
intensity
sensor
(%)
 Duty
=
Measurement
Uncertainty
of
the
duty
UV
intensity
sensor
(%)

If
this
condition
is
not
met,
the
cause
for
the
discrepancy
should
be
determined
and
resolved.
Typically,
the
discrepancy
indicates
degradation
of
the
on­
line
sensor
that
necessitates
recalibration
or
replacement.

Proposal
Draft
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
21
June
2003
Proposal
Draft
Example
18.
A
UV
reactor
uses
on­
line
sensors
with
an
uncertainty
of
15
percent.
A
reference
sensor
with
an
uncertainty
of
5
percent
is
used
to
check
the
on­
line
sensors
when
the
UV
reactor
is
operating
at
the
WTP.
Measurements
made
by
the
on­
line
sensors
are
considered
out
of
spec
when:

[
]
%
16
5
15
100
1
I
I
1/
2
2
2
Ref
Duty
=
+
£
¥








­

F.
3.4
Positioning
of
UV
Intensity
Sensors
While
the
UV
output
along
the
length
and
around
the
circumference
of
a
new
UV
lamp
will
be
relatively
uniform,
this
may
not
be
true
with
aged
or
fouled
lamps.
Sputtering
of
electrode
material
leads
to
deposits
on
the
inside
of
the
lamp
sleeve
within
2
or
3
inches
from
the
electrode.
Discoloration
of
the
lamp
sleeve
with
lamp
aging
varies
along
the
length
of
the
lamp.
Sleeve
fouling
varies
spatially
both
along
the
length
and
circumference
of
the
lamp
sleeve
(
Lin
et
al.
1999).

If
lamps
experience
non­
uniform
aging
along
their
length,
the
UV
intensity
sensor
should
be
located
to
monitor
the
section
along
the
lamp
that
experienced
the
greatest
decrease
in
UV
output
with
aging.
The
sensor
should
not
be
located
to
monitor
the
section
that
experiences
the
least
decrease
in
UV
output.

F.
3.5
Number
of
UV
Intensity
Sensors
Variability
in
UV
output
from
lamp
to
lamp
impacts
both
dose
delivery
and
monitoring.
A
lamp
with
a
lower
output
will
deliver
lower
doses
to
microorganisms
passing
in
its
vicinity,
thereby
shifting
the
dose
distribution
to
lower
values
and
reducing
the
net
performance
of
the
reactor.
The
shift
in
the
dose
distribution
will
be
more
pronounced
with
a
reactor
with
fewer
lamps.
Because
the
dose
distribution
is
affected,
the
impact
on
net
performance
will
be
greater
with
a
more
UV­
sensitive
microorganism.
If
the
number
of
UV
intensity
sensors
is
less
than
the
number
of
lamps
and
the
sensors
monitor
those
lamps
with
the
highest
output,
the
monitoring
system
will
overestimate
dose
delivery
by
the
UV
reactor.

The
monitoring
strategy
used
to
ensure
that
UV
dose
delivery
meets
regulatory
targets
should
account
for
the
variability
of
UV
output
from
lamp­
to­
lamp.
If
each
lamp
in
the
reactor
is
monitored
by
a
UV
intensity
sensor,
dose
delivery
compliance
should
be
based
on
the
lowest
lamp
output,
unless
an
accepted
and
validated
dose
calculation
methodology
can
account
for
lamp­
to­
lamp
variability.
If
the
number
of
sensors
used
is
less
than
the
number
of
lamps,
either
the
lamp
with
the
lowest
output
should
be
monitored
and
used
for
dose
compliance,
or
the
setpoint
used
for
dose
delivery
compliance
should
include
a
safety
factor
to
account
for
lamp­
tolamp
variability.

E
xample
19.
A
UV
reactor
installed
at
a
WTP
is
equipped
with
four
lamps
and
two
UV
intensity
sensors.
Because
of
variability
in
lamp
output,
the
UV
intensity
5
cm
from
each
lamp
is
15,
10,
8,
and
20
mW/
cm2,
respectively.
If
one
sensor
monitors
the
first
lamp
and
the
second
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
22
June
2003
Proposal
Draft
monitors
the
forth
lamp,
the
monitoring
system
will
over­
estimate
the
dose
delivery
by
the
UV
reactor
because
microorganisms
passing
by
the
second
and
third
lamps
will
receive
lower
doses
than
the
microorganisms
passing
by
the
first
and
fourth
lamps.

During
UV
reactor
validation,
variability
in
UV
output
from
lamp
to
lamp
should
not
cause
the
UV
reactor
to
be
overrated.
If
the
number
of
sensors
is
less
than
the
number
of
lamps,
the
sensors
should
be
monitoring
the
lamps
with
the
lowest
output.
If
UV
intensity
sensors
record
different
values
during
validation,
intensity
setpoints
and
calculations
should
be
based
on
the
lowest
values
recorded.

Example
20.
A
UV
reactor
undergoing
validation
is
equipped
with
four
lamps
and
two
sensors.
Dose
delivery
is
monitored
using
the
UV
intensity
setpoint
approach.
Because
of
variability
in
lamp
output,
the
UV
intensity
5
cm
from
each
lamp
is
10,
15,
8,
and
12
mW/
cm2,
respectively.
To
ensure
validation
results
are
meaningful,
the
sensors
should
be
monitoring
the
first
and
third
lamps.

F.
4
Polychromatic
Considerations
With
UV
reactors
equipped
with
LP
or
low
pressure
high
output
(
LPHO)
lamps,
dose
delivery
and
monitoring
occurs
at
a
single
wavelength
of
254
nm.
With
UV
reactors
equipped
with
MP
lamps,
dose
delivery
and
monitoring
involves
a
response
to
multiple
wavelengths.
Dose
delivery
is
an
integrated
response
to
UV
light
from
200
to
320
nm.
The
output
from
the
UV
intensity
sensor
is
an
integrated
response
to
UV
light
over
wavelengths
spanning
the
sensor's
spectral
response.
UV
absorbance
monitors
typically
measure
UV
absorbance
at
a
single
wavelength
of
254
nm.
If
the
spectral
properties
of
the
UV
reactor
that
influence
dose
delivery
and
monitoring
during
operation
of
the
UV
installation
at
a
WTP
are
the
same
as
the
spectral
properties
during
validation,
then
the
same
dose
delivery
and
monitoring
characterized
during
validation
will
occur
at
the
WTP.
However,
if
the
spectral
properties
are
different,
dose
delivery
and
monitoring
at
the
WTP
will
differ
from
dose
delivery
and
monitoring
measured
during
validation.
The
following
spectral
properties
may
differ:

·
Action
spectra
of
the
challenge
microorganism
used
during
validation
and
the
target
pathogen
·
Spectral
UV
absorbance
of
the
water
during
validation
and
at
the
WTP
·
UV
output
of
the
lamps
during
validation
and
at
the
WTP
·
UVT
of
the
lamp
sleeves
during
validation
and
at
the
WTP
Safety
factors
should
be
applied
to
the
validation
data
for
polychromatic
UV
reactors
if
spectral
differences
will
lead
to
under
dosing
at
the
WTP.
This
section
describes
approaches
for
assessing
the
impact
of
differences
in
spectral
properties
and
deriving
those
safety
factors.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
23
June
2003
Proposal
Draft
F.
4.1
Action
Spectra
The
dependence
of
microorganism
inactivation
kinetics
on
UV
wavelength
may
be
described
using
an
action
spectrum
­
the
UV
inactivation
sensitivity
as
a
function
of
wavelength
(
Figure
F.
9).
Ideally,
the
action
spectrum
of
the
challenge
microorganism
used
to
validate
a
polychromatic
UV
reactor
would
either
match
that
of
the
target
microorganism
or
provide
a
conservative
estimate
of
inactivation.

Figure
F.
9
Action
Spectra
for
Various
Microorganisms1
1
(
Adapted
from
Rauth
1965)

The
impact
of
various
action
spectra
on
UV
dose
delivery
may
be
estimated
by
calculating
the
germicidal
lamp
output
using
Equation
F.
12:

(
)
(
)


=
D
=
320
200
l
l
l
l
G
P
P
G
Equation
F.
12
where
PG
=
Germicidal
output
of
the
MP
lamp
(
W/
cm)

=
Wavelength
(
nm)
P(

)
=
Lamp
output
(
W/
nm)
measured
over
1
nm
increments
at
wavelength

G(

)
=
Relative
UV
sensitivity
of
the
microorganism
at
wavelength

Dl
=
1
nm
increment
Using
the
action
spectra
published
for
fourteen
microorganisms
(
Rauth
1965,
Cabaj
et
al.
2002,
Linden
2001),
Table
F.
2
presents
the
germicidal
lamp
output
calculated
for
a
MP
lamp
and
the
ratio
of
that
output
to
that
of
Cryptosporidium.
A
ratio
greater
than
one
indicates
that
the
action
spectra
of
the
microorganism
favors
greater
inactivation
than
the
action
spectra
of
Cryptosporidium.
If
a
challenge
microorganism
with
a
ratio
greater
than
one
is
used
to
validate
a
0.0
0.5
1.0
1.5
2.0
2.5
225
235
245
255
265
275
285
295
305
Wavelength
(
nm)
Action
relative
to
254
nm
Herpes
simplex
MS2,
R­
17,
fr,
7­
S
oX­
174
T2
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
24
June
2003
Proposal
Draft
MP
reactor
for
Cryptosporidium
inactivation,
the
ratio
should
be
used
as
a
correction
factor
to
account
for
the
greater
inactivation
of
the
challenge
that
arises
from
the
differences
in
action
spectra.
In
the
case
of
MS2
and
B.
subtilis,
the
ratio
is
close
to
one
and
the
correction
is
small.
However,
based
on
the
data
in
Table
F.
2,
a
correction
factor
of
1.16
would
be
needed
with
UV
reactors
equipped
with
MP
lamps
if
fX174
was
used
to
show
Cryptosporidium
inactivation.

Table
F.
2
Germicidal
Output
Delivered
to
14
Microorganisms
by
a
MP
Lamp
Microorganism
Type
/
Nucleic
acid
(
SS
=
Single
Strand,
DS
=
Double
Strand)
Germicidal
Output
(
W/
cm)
Germicidal
Output
Relative
to
Cryptosporidium
Cryptosporidium
oocysts
DS
DNA
5.64
1.00
MS­
2,
R­
17,
fr,
7­
S
Phage
/
SS
RNA
5.78
1.04
B.
subtilis
spores
DS
DNA
5.58
0.99

X174
Phage
/
DS
DNA
6.53
1.16
Reovirus­
3
Animal
virus
/
DS
RNA
7.46
1.32
Polyoma
Animal
virus
/
DS
DNA
6.74
1.18
T2
Phage
/
DS
DNA
6.05
1.07
VSV
Animal
virus
/
RNA
5.53
0.99
Vaccinia
Animal
virus
/
DS
DNA
5.46
0.98
EMC
Animal
virus
/
SS
RNA
5.98
1.07
Herpes
simplex
Human
virus
/
DS
DNA
7.00
1.26
The
germicidal
output
of
the
MP
lamp
calculated
using
the
action
spectra
of
B.
subtilis
spores
and
MS2
is
equal
to
or
less
than
that
of
most
of
the
14
microorganisms
listed
in
Table
F.
2.
It
is
thus
reasonable
to
assume
that
these
microorganisms
are
acceptable
as
challenge
microorganisms
for
many
pathogens
whose
action
spectrum
is
not
known,
like
adenovirus
and
Giardia.
However,
if
an
alternative
challenge
microorganism
is
to
be
used,
its
action
spectra
should
be
assessed
for
suitability.

As
an
alternate
approach
to
measuring
the
action
spectrum
and
using
Equation
F.
12,
the
correction
factor
can
also
be
estimated
by
comparing
the
dose­
response
of
the
challenge
microorganism
to
that
of
MS2
measured
with
a
LP
and
MP
lamp.
The
correction
factor
would
be
defined
as:

MS2
MP
LP
Challenge
LP
MP
k
k
k
k
1.04
r
Facto
Safety
















=
Equation
F.
13
where
kMP
=
Slope
of
the
dose­
response
measure
with
the
MP
collimated
beam
(
cm2/
mJ)
kLP
=
Slope
of
the
dose­
response
measure
with
the
LP
collimated
beam
(
cm2/
mJ)
1.04
=
Germicidal
output
of
MS2
relative
to
Cryptosporidium,
from
Table
F.
2
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
25
June
2003
Proposal
Draft
The
correction
factor
that
accounts
for
differences
in
the
action
spectra
is
not
the
same
correction
factor
that
accounts
for
differences
in
the
UV
sensitivity
described
in
section
F.
1.2.
The
correction
factor
described
in
section
F.
1.2
applies
to
all
UV
reactors
regardless
of
lamp
type.
The
correction
factor
described
in
this
section
is
applicable
to
MP
reactors.
It
should
be
used
in
addition
to
the
correction
factor
described
in
section
F.
1.2.

F
.4.2
Water
Absorption
During
UV
reactor
validation,
a
UV­
absorbing
chemical
is
added
to
the
water
passing
through
the
reactor
in
order
to
simulate
high
UV
absorbance
events
that
could
occur
at
the
WTP.
UV­
absorbing
chemicals
that
have
been
used
to
validate
UV
reactors
include
sodium
thiosulfate,
fluorescein,
coffee,
tea,
and
parahydroxybenzoic
acid.
Ideally,
the
spectral
absorption
of
the
water
used
to
validate
UV
reactors
equipped
with
MP
lamps
should
match
the
spectral
absorption
of
the
water
at
the
WTP
over
the
wavelength
range
associated
with
dose
delivery
and
monitoring
(
Figure
F.
10).

Figure
F.
11
compares
the
UV
absorbance
spectra
of
coffee
and
lignin
sulphonate
to
that
of
two
drinking
water
sources
(
Water
A
and
Water
B).
For
a
given
UVT
at
254
nm,
the
UV
absorption
at
wavelengths
above
and
below
254
nm
is
greater
with
coffee,
tea,
and
lignin
sulphonate
than
with
the
drinking
water
sources.
If
those
chemicals
are
used
during
validation
of
a
MP
reactor,
the
RED
and
UV
intensity
measured
at
a
given
flowrate,
lamp
output,
and
water
UVT
will
be
lower
during
validation
than
at
the
WTP.

Figure
F.
10
Spectral
UV
Absorption
of
Water
at
Various
WTPs
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
26
June
2003
Proposal
Draft
Figure
F.
11
Comparison
of
the
UV
Absorbance
Spectrum
of
Additives
used
during
UV
Reactor
Validation
to
the
UV
Absorbance
of
Two
Finished
Waters
The
impact
of
the
difference
in
the
UV
absorbance
spectra
on
the
measured
intensity
will
depend
on
sensor
placement
relative
to
the
lamps.
If
the
sensor
is
located
close
to
the
lamps,
the
sensor
reading
during
validation
will
be
only
slightly
lower
than
the
reading
at
the
WTP.
Accordingly,
for
a
given
sensor
reading,
flowrate,
and
water
UVT,
the
RED
delivered
at
the
WTP
will
be
greater
than
the
RED
measured
during
validation.
However,
if
the
sensor
is
placed
far
enough
from
the
lamp,
the
UV
intensity
measured
during
validation
will
be
much
lower
than
the
reading
at
the
WTP.
As
such,
for
a
given
sensor
reading,
flowrate,
and
water
UVT,
the
RED
delivered
at
the
WTP
will
be
less
than
the
RED
measured
during
validation.
If
the
UV
intensity
sensor's
spectral
response
mimics
the
microorganism's
action
spectra
and
the
sensor
is
located
at
a
position
where
the
dose
delivery
is
proportional
to
the
sensor
reading,
the
RED
delivered
at
the
WTP
will
equal
the
RED
measured
during
validation,
even
with
the
differences
in
the
UV
absorbance
spectra
shown
in
Figure
F.
11
(
Wright
et
al.
2002).
However,
this
relationship
will
not
hold
true
if
the
sensor's
spectral
response
deviates
sufficiently
from
the
microorganism's
action
spectra.

Modeling
approaches
can
be
used
to
predict
and
compare
the
RED
and
UV
intensity
sensor
readings
obtained
during
validation
to
those
expected
at
a
WTP.
The
modeling
approach
can
be
used
to
define
correction
factors
applicable
to
validation
results
to
ensure
dose
monitoring
provides
valid
measurements
at
the
WTP.
UV
intensity
readings
should
be
predicted
using
polychromatic
intensity
models
that
factor
in
the
spectral
and
angular
response
of
the
sensor.
While
RED
predictions
could
be
obtained
using
CFD­
based
dose
modeling
approaches,
ideal
dose
delivery
models
should
be
used
to
provide
conservative
correction
factors.
The
ideal
dose
delivery
model
is
conservative
because
the
sensor
location
within
a
reactor
where
the
dose
delivery
is
proportional
to
sensor
reading
is
predicted
to
occur
closer
to
the
lamp
with
the
ideal
model
than
with
a
CFD­
based
dose
delivery
model
(
Wright
et
al.
2002).
As
such,
the
transition
to
a
correction
factor
greater
than
one
occurs
with
closer
sensor­
to­
lamp
distance
with
the
ideal
dose
delivery
model
than
with
the
CFD­
based
dose
delivery
model.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
27
June
2003
Proposal
Draft
Table
F.
3
provides
predictions
of
dose
delivery
and
sensor
measurements
for
an
ideal
annular
reactor.
The
reactor
consists
of
a
cylinder
with
an
18.81­
cm
radius
and
a
length
greater
than
the
arc
length
of
the
lamp.
The
reactor
is
equipped
with
a
single
MP
lamp
oriented
along
the
central
axis
of
the
cylinder
(
i.
e.,
at
a
radius
of
0
cm).
The
lamp
is
housed
in
a
lamp
sleeve
with
a
radius
of
3.81
cm.
The
spectral
output
of
the
lamp
is
given
in
Figure
F.
12.
The
spectral
UV
absorbances
used
in
the
model
are
provided
in
Figure
F.
11.
UV
intensity
was
modeled
using
a
polychromatic
radial
intensity
model
and
the
dose
was
calculated
as
the
product
of
the
average
germicidal
intensity
and
the
hydraulic
residence
time
as
per
the
following
equation:

(
)
(
)
(
)
(
)
(
)
(
)

=








­
­
­
=
320
200
1
exp
l
l
a
l
a
l
l
l
e
wl
e
q
arc
Q
r
T
G
L
P
D
Equation
F.
14
where
D
=
Dose
delivered
by
the
reactor
(
mJ/
cm2)
Larc
=
Arc
length
of
the
lamp
(
cm)
Tq(
l)
=
Lamp
sleeve
UVT
ae(
l)
=
Naperian
UV
absorbance
rwl
=
Reactor
water
layer,
defined
as
the
radial
distance
from
the
sleeve
to
the
reactor
wall
(
cm)
Q
=
Flowrate
through
the
reactor
(
cm3/
s)

UV
intensity
sensor
measurements
were
modeled
at
different
lamp­
to­
sensor
distances
for
sensors
with
the
spectral
response
shown
in
Figure
F.
13
as
per
Equation
F.
15:

(
)
(
)
(
)
(
)(
)
(
)


=
­
­
=
400
200
2
exp
l
p
l
a
l
l
l
r
r
r
T
S
P
I
S
e
q
Equation
F.
15
where
I
=
Intensity
measured
by
the
sensor
S(
l)
=
Sensor
spectral
response
normalized
to
unity
at
254
nm
r
=
Distance
from
the
sensor
to
the
lamp
(
cm)
rS
=
Lamp
sleeve
outer
radius
(
cm)
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
28
June
2003
Proposal
Draft
Table
F.
3
Dose
and
UV
Intensity
Sensor
Measurements
Modeled
for
a
MP
Annular
Reactor
Performance
Parameters
Water
A
Water
B
Coffee
Lignin
Sulphonate
MS2
RED
(
mJ/
cm2)
72
67
60
61
Sensor
Water
Layer
(
cm)
1
Measured
UV
Intensity
(
254
nm
equivalent
mW/
cm2)
2.0
269
256
238
245
5.0
136
122
101
110
10
59.7
48.7
31.6
40.4
15
31.9
23.6
11.7
18.2
SiC
20
19.2
13.0
4.77
9.34
2.0
112
107
103
104
5.0
48.0
44.8
10.9
11.8
10
15.3
13.7
3.46
4.06
15
5.02
4.97
1.18
1.51
Filtered
SiC
20
2.42
1.95
0.410
0.593
1
Water
layer
is
defined
as
the
distance
between
the
lamp
sleeve
and
the
UV
intensity
sensor.

Figure
F.
12
UV
Output
of
a
MP
Lamp
Table
F.
3
presents
the
MS2
RED
and
sensor
measurements
predicted
for
the
annular
reactor
operating
at
a
flowrate
of
200
gpm,
a
water
UVT
of
85
percent
at
254
nm,
and
100
percent
lamp
power.
As
expected,
the
dose
delivered
with
coffee
and
lignin
sulphonate
for
a
given
flowrate,
water
UVT,
and
lamp
power
was
less
than
the
dose
delivered
with
both
WTP
waters.

For
a
given
sensor
reading,
flowrate,
and
UVT,
Table
F.
4
presents
the
ratio
of
the
dose
measured
during
validation
to
the
dose
delivered
at
the
WTP
calculated
using
the
data
from
Table
F.
3.
A
ratio
greater
than
one
indicates
that
the
dose
measured
during
validation
will
be
greater
than
the
dose
delivered
at
the
WTP.
As
expected,
the
ratio
is
less
than
one
with
the
UV
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
29
June
2003
Proposal
Draft
intensity
sensor
located
close
to
the
lamps
and
greater
than
one
with
the
UV
intensity
sensor
located
far
from
the
lamps.
For
a
given
sensor
position,
the
ratio
with
lignin
sulphonate
is
closer
to
one
than
the
ratio
with
coffee
indicated
that
lignin
sulphonate
better
matches
the
UV
absorption
spectra
of
WTP
waters.
The
ratio
is
also
closer
to
one
with
a
germicidal
sensor
spectral
response
compared
to
the
non­
germicidal
response.
This
indicates
that
validation
results
with
a
germicidal
sensor
are
more
representative
of
performance
at
a
WTP
than
validation
results
with
a
non­
germicidal
sensor.

Table
F.
4
Impact
of
Water
UV
Absorbance
on
the
UV
Intensity
Sensor
Value
Associated
with
a
Given
UV
Dose
Delivery
Ratio
of
Dose
Delivered
During
Validation
to
Dose
Delivered
at
the
WTP
for
a
Given
Sensor
Reading
UV
Sensor
Water
Layer
(
cm)
Coffee
to
Water
A
Coffee
to
Water
B
Lignin
Sulphonate
to
Water
A
Lignin
Sulphonate
to
Water
B
2
0.93
0.96
0.93
0.95
5
1.12
1.09
1.04
1.01
10
1.56
1.37
1.25
1.10
15
2.25
1.80
1.48
1.19
SiC
20
3.34
2.44
1.74
1.27
2
0.89
0.93
0.91
0.94
5
0.98
0.99
0.98
0.99
10
1.16
1.12
1.09
1.06
15
1.39
1.29
1.21
1.12
Filtered
SiC
20
1.70
1.48
1.35
1.18
For
the
germicidal
sensor,
Figure
F.
13
presents
the
ratio
of
the
dose
expected
with
coffee
to
the
dose
expected
with
finished
Water
A
as
a
function
of
sensor
position
and
water
UVT.
With
the
sensor
located
close
to
the
lamp,
the
ratio
is
less
than
one
over
a
wide
range
of
water
UVT
values.
However,
the
ratio
increases
above
one
with
increased
sensor­
to­
lamp
water
layer
and,
for
the
most
part,
increases
with
decreased
UVT.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
30
June
2003
Figure
F.
13
Comparison
of
Dose
Expected
with
Coffee
as
a
UV
Absorber
to
Dose
Expected
with
WTP
Water
for
a
MP
Reactor
Equipped
with
a
Germicidal
Sensor
For
a
given
UV
reactor
equipped
with
MP
lamps,
the
impact
of
differences
in
the
spectral
UV
absorbance
between
validation
and
operation
at
a
WTP
should
be
evaluated
and
used
to
establish
correction
factors.
The
correction
factor
is
calculated
for
a
given
flowrate,
sensor
reading,
and
UVT,
as
the
ratio
of
the
dose
expected
during
validation
to
the
dose
expected
at
the
WTP.
If
the
ratio
is
less
than
one,
no
correction
factor
is
needed.

F.
4.3
Spectral
Shifts
Spectral
shifts
in
the
UV
output
of
MP
lamps
may
occur
as
MP
lamps
age.
Spectral
shifts
in
the
UVT
of
light
through
lamp
sleeves
may
occur
as
sleeves
age
and
undergo
internal
and
external
fouling.
Spectral
shifts
in
the
UVT
of
sensor
windows
may
occur
with
window
fouling.
Spectral
shifts
associated
with
the
lamp­
sleeve
assembly
will
impact
both
dose
delivery
and
monitoring,
while
spectral
shifts
associated
with
window
fouling
will
impact
monitoring
only.

Figure
F.
14
presents
reported
data
on
the
spectral
shift
in
MP
lamp
output
and
lamp
sleeve
UVT
experienced
with
aging.
Figure
F.
15
presents
data
comparing
the
UVT
of
clean
and
fouled
lamp
sleeves.
In
both
cases,
aging
and
fouling
have
reduced
the
output
of
lowwavelength
UV
light
from
the
lamp/
sleeve
assembly
more
than
the
output
of
higher
wavelength
UV
light.
The
impact
of
lamp
and
sleeve
aging
and
sleeve
fouling
can
be
assessed
by
validation
testing.
Alternatively,
the
impact
can
be
modeled
and
used
to
define
a
correction
factor
applicable
to
validation
results
generated
using
new
lamps.

Proposal
Draft
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
31
June
2003
Figure
F.
14
Spectral
Shifts
in
the
MP
Lamp
Output
and
Lamp
Sleeve
UVT
Reported
with
Aging1
1
Adapted
from
Phillips
1983
and
Kawar
et
al.
1998.

Figure
F.
15
Comparison
of
the
UVT
of
New
and
Fouled
Lamp
Sleeves
For
a
measured
flowrate,
water
UVT,
and
UV
intensity,
Figures
F.
16,
F.
17,
and
F.
18
provide
the
ratio
of
the
dose
delivered
with
new
lamps
and
sleeves
to
the
dose
delivered
with
aged
lamps,
aged
sleeves,
and
fouled
sleeves,
respectively.
In
each
figure,
the
dose
ratio
is
presented
as
a
function
of
water
UVT
and
sensor­
to­
lamp
water
layer
for
two
different
sensors.
One
sensor
had
a
SiC
spectral
response
while
the
other
had
a
germicidal
response.
Dose
and
UV
Proposal
Draft
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
32
June
2003
Proposal
Draft
response.
Dose
and
UV
intensity
values
were
predicted
using
Equations
F.
14
and
F.
15
applied
to
the
annular
reactor
described
in
section
F.
4.2.

Figure
F.
16
Comparison
of
Dose
Delivered
by
a
MP
Reactor
with
New
and
Aged
Type
214
Lamp
Sleeves
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
33
June
2003
Proposal
Draft
Figure
F.
17
Comparison
of
Dose
Delivered
by
a
MP
Reactor
with
New
and
Aged
Lamps
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
34
June
2003
Proposal
Draft
Figure
F.
18
Comparison
of
Dose
Delivered
by
a
MP
Reactor
with
New
Type
214
Lamp
Sleeves
to
Fouled
Sleeves
In
each
figure
(
Figures
F.
16
to
F.
18),
the
dose
ratio
increases
with
decreased
water
UVT
and
increased
sensor­
to­
lamp
distance.
The
ratio
is
closer
to
one
with
germicidal
sensors
compared
with
sensors
with
a
SiC
spectral
response.

For
a
given
UV
reactor
equipped
with
MP
lamps,
the
impact
of
spectral
shifts
in
lamp
output
and
sleeve
UVT
should
be
evaluated
and
used
to
establish
correction
factors.
The
correction
factor
is
calculated,
for
a
given
flowrate,
sensor
reading,
and
UVT,
as
the
ratio
of
the
dose
expected
with
and
without
the
spectral
shift
expected
with
operation
of
the
UV
reactor
at
the
WTP.
If
the
ratio
is
less
than
one,
no
correction
factor
is
needed.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
35
June
2003
Proposal
Draft
Spectral
shifts
associated
with
lamp
and
sleeve
aging
can
be
avoided
by
regular
replacement
of
those
components.
Spectral
shifts
arising
from
fouling
on
external
surfaces
of
lamp
sleeves
and
sensor
windows
can
be
minimized
with
good
cleaning
practices.
However,
fouling
can
also
occur
on
internal
surfaces
of
lamp
sleeves
and
sensor
windows.

F.
5
Uncertainty
of
Dose
Monitoring
and
Safety
Factors
UV
installations
should
be
sized
and
operated
in
a
manner
that
accounts
for
the
measurement
uncertainty
associated
with
dose
delivery
monitoring.
The
objective
of
dose
delivery
monitoring
is
to
indicate
the
level
of
inactivation
of
the
target
pathogen.
Safety
factors
applied
to
UV
installations
that
account
for
measurement
uncertainty
should
be
chosen
to
ensure
that
UV
reactors
meet
inactivation
targets
at
a
90­
percent
confidence
level.
A
90
percent
confidence
level
is
consistent
with
the
confidence
level
used
to
define
dose
values
for
Cryptosporidium,
Giardia,
and
virus
in
Chapter
1.

F.
5.1
Analytical
Foundation
for
Defining
Uncertainty
This
section
derives
a
measurement
equation
for
UV
dose
monitoring.
This
equation
is
used
in
this
guidance
document
as
the
analytical
foundation
for
defining
the
uncertainty
of
dose
monitoring.

Consider
a
UV
installation
operating
at
a
WTP.
Assuming
first
order
kinetics,
the
log
inactivation
of
a
target
pathogen
achieved
by
the
UV
reactor
at
some
point
in
time
can
be
expressed
using
Equation
F.
16:

log
N
RED
D
p
p
p
=
10
Equation
F.
16
where
log
Np
=
Log
inactivation
of
the
pathogen
REDp
=
RED
of
the
pathogen
(
mJ/
cm2)
D10p
=
UV
sensitivity
of
the
pathogen
(
mJ/
cm2
per
log
inactivation)

If
the
UV
reactor
delivers
a
dose
distribution,
the
log
inactivation
of
the
pathogen
is
related
to
the
inactivation
of
a
challenge
microorganism
using
Equation
F.
17:

p
c
RED
p
D
RED
B
N
10
log
=
Equation
F.
17
where
REDc
=
RED
of
the
challenge
microorganism
(
mJ/
cm2)
BRED
=
Ratio
of
the
RED
of
the
pathogen
to
that
of
the
challenge
microorganism
Assuming
the
challenge
microorganism
RED
is
proportional
to
the
measured
UV
intensity,
log
inactivation
of
the
pathogen
can
be
expressed
according
to
Equation
F.
18:
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
36
June
2003
Proposal
Draft
p
RED
p
D
I
B
N
10
log
a
=
Equation
F.
18
where
I
=
UV
intensity
measured
at
the
WTP
(
mW/
cm2)
a
=
Constant
relating
challenge
microorganism
inactivation
to
measured
intensity
(
J/
W)

The
constant
k
is
determined
during
validation
as
the
ratio
of
the
measured
RED
of
the
challenge
microorganism
to
the
measured
intensity.
Assuming
that
inactivation
is
proportional
to
flowrate,
Equation
F.
19
can
be
used:

Q
Q
I
I
D
RED
B
N
v
v
p
cv
RED
p
10
log
=
Equation
F.
19
where
REDcv
=
RED
of
the
challenge
microorganism
measured
during
validation
Iv
=
UV
intensity
measured
during
validation
Qv
=
Flowrate
measured
during
validation
(
mgd)
Q
=
Flowrate
measured
at
the
WTP
(
mgd)

If
spectral
properties
such
as
lamp
output,
sleeve
UVT,
and
water
UV
absorbance
during
validation
differ
from
those
during
operation
of
the
UV
installation
at
the
WTP,
Equation
F.
19
is
expressed
as
Equation
F.
21:

Poly
v
v
p
cv
REF
p
B
Q
Q
I
I
D
RED
B
N
10
log
=
Equation
F.
20
where
variables
are
defined
as
in
Equation
F.
19
The
term
BPoly
is
the
ratio
of
challenge
microorganism
RED
expected
at
the
WTP
to
the
challenge
microorganism
RED
expected
during
validation
for
the
same
conditions
of
flowrate,
water
UVT,
and
UV
intensity.

Assuming
the
dose­
response
of
the
challenge
microorganism
follows
first
order
kinetics,
the
challenge
microorganism
RED
during
validation
is
calculated
using
the
log
inactivation
of
the
challenge
microorganism
measured
through
the
reactor
as
per
Equation
F.
21:

cv
ef
in
c
cv
N
N
D
RED






=
log
10
Equation
F.
21
where
D10c
=
UV
sensitivity
of
the
challenge
microorganism
(
mJ/
cm2
per
log
inactivation)
Nin
=
Challenge
microorganism
concentration
measured
at
the
reactor
influent
Nef
=
Challenge
microorganism
concentration
measured
at
the
reactor
effluent
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
37
June
2003
Proposal
Draft
The
UV
sensitivity
of
the
challenge
microorganism
can
be
calculated
according
to
Equation
F.
22
from
the
UV
dose­
response
measured
using
the
collimated
beam
apparatus:

i
D
D
CB
c
log
10
=
Equation
F.
22
where
DCB
=
Dose
delivered
by
the
collimated
beam
apparatus
log
i
=
Log
inactivation
of
the
challenge
microorganism
observed
with
dose
DCB
The
dose
delivered
by
the
collimated
beam
apparatus
is
defined
by
Equation
E.
1
(
section
E.
3).
Substituting
Equations
F.
21
and
F.
22,
and
E.
1
into
Equation
F.
20
gives
the
measurement
equation
for
dose
monitoring
using
the
UV
intensity
setpoint
approach:

)
10
ln(
)
1
1
(
log
)
10
1
)(
1
(
log
log
10
al
L
i
D
R
P
E
Q
Q
I
I
N
N
B
B
N
p
al
f
s
v
v
cv
ef
in
Poly
RED
p
+
­
­









=
­

Equation
F.
23
F.
5.2
Calculating
Total
Uncertainty
Errors
in
dose
monitoring
can
be
classified
as
either
biases
or
random
uncertainties.

Biases
are
systematic
errors
that
favor
either
an
over
or
under
estimation
of
dose
delivery.
A
bias
error
will
occur
with
dose
monitoring
if
the
monitoring
approach
does
not
account
for
differences
in
the
RED
measured
with
the
challenge
microorganism
and
the
RED
delivered
to
the
target
pathogen.
A
bias
error
will
also
occur
if
the
monitoring
approach
does
not
account
for
differences
between
the
spectral
properties
of
the
UV
reactor
that
impact
dose
delivery
and
monitoring
during
validation
and
those
properties
during
operation
of
the
UV
reactor
at
the
WTP.
A
bias
error
will
occur
if
the
radiometer,
UV
intensity
sensor,
flowmeter,
or
UVT
monitor
used
during
validation
always
reads
either
high
or
low.
Bias
errors
should
be
accounted
for
using
correction
factors.
The
approaches
for
defining
correction
factors
to
account
for
bias
errors
represented
by
the
terms
BRED
and
BPoly
in
the
measurement
equation
are
provided
in
Sections
F.
1
and
F.
4,
respectively.

Random
uncertainty
is
associated
with
every
term
in
the
measurement
equation
(
Equation
F.
23).
If
the
measurement
equation
consists
of
linear
relationships
of
independent
variables
whose
random
uncertainty
is
normally
distributed,
standard
approaches
can
be
used
to
calculate
the
uncertainty
of
the
measured
variable
from
the
uncertainty
of
each
term
in
the
measurement
equation.
For
example,
if
the
measurement
equation
is
y
=
x1
+
x2
or
y
=
x1
­
x2,
the
uncertainty
of
y
due
to
the
uncertainty
of
x1
and
x2
is
calculated
using
Equation
F.
24:
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
38
June
2003
Proposal
Draft
(
)
2
1
2
2
2
1
s
s
s
+
=
Equation
F.
24
where
s
=
Uncertainty
of
y
in
absolute
units
s1
=
Uncertainty
of
x1
in
absolute
units
s2
=
Uncertainty
of
x2
in
absolute
units
On
the
other
hand,
if
the
measurement
equation
is
y
=
x1
¥
x2
or
y
=
x1
/
x2,
the
uncertainty
of
y
due
to
the
uncertainty
of
x1
and
x2
is
calculated
using
Equation
F.
25:

(
)
2
1
2
2
2
1
s
s
s
+
=
Equation
F.
25
where
s
=
Uncertainty
of
y
in
percent
s1
=
Uncertainty
of
x1
in
percent
s2
=
Uncertainty
of
x2
in
percent
If
the
measurement
equation
involves
non­
linear
relations
like
y=
x1
exp(
x2),
Monte
Carlo
approaches
should
be
used
to
define
the
uncertainty
of
y.

Determining
the
random
uncertainty
of
a
measured
quantity
requires
making
assumptions
about
the
statistical
distribution
of
measurements.
If
the
distribution
is
normal,
the
uncertainty
is
calculated
as
the
product
of
the
sample
standard
deviation
and
the
t­
statistic.
If
the
number
of
samples
is
high,
the
t­
statistic
can
be
approximated
by
the
z­
statistic.
If
the
standard
deviation
of
the
population
is
known,
the
uncertainty
is
calculated
as
the
product
of
the
population
standard
deviation
and
the
z­
statistic.
T
and
z­
statistics
are
often
given
in
the
appendices
of
statistics
texts.
The
NIST
provides
recommendations
for
specifying
the
uncertainty
for
quantities
that
are
not
normally
distributed.

Table
F.
5
defines
an
approach
for
estimating
the
uncertainties
of
each
term
in
the
measurement
Equation
F.
23.
The
total
random
uncertainty
of
dose
monitoring
can
be
estimated
by
summing
the
uncertainties
associated
with
each
term
in
Equation
F.
23
using
the
above
stated
rules.
Assuming
the
terms
BRED
and
BPoly
are
the
only
bias
terms,
a
safety
factor
for
dose
monitoring
can
be
defined
according
to
Equation
F.
26:

(
)
e
B
B
SF
Poly
RED
+
¥
¥
=
1
Equation
F.
26
where
e
=
Total
random
uncertainty
associated
with
the
measurement
equation.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
39
June
2003
Proposal
Draft
Table
F.
5
Terms
Used
to
Define
the
Uncertainty
of
Dose
Monitoring
Term
Assumption
BRED
and
Babs
No
term
used
if
values
are
selected
as
safety
factors
as
described
in
Sections
F.
1
and
F.
4.
If
terms
are
calculated,
use
uncertainty
of
model
predictions
to
define
uncertainty
of
these
terms.

I
and
Iv
UV
intensity
measurement
uncertainty
is
often
defined
by
the
UV
intensity
sensor
manufacturer.
If
a
reference
sensor
is
used
to
check
the
uncertainty
of
a
duty
sensor,
the
uncertainty
of
the
duty
sensor
should
be
defined
as
the
rejection
criteria
used
to
determine
if
the
on­
line
sensor
is
out
of
tolerance.
See
Equation
F.
11.
Q
and
Qv
Use
measurement
uncertainty
defined
by
flowmeter
manufacturer
D10p
Accounted
for
in
dose
targets
provided
in
Chapter
1
Log(
Nin/
Nef)
Calculated
as
a
confidence
interval
using
standard
deviation
and
Student's
t­
statistic
associated
with
samples
collected
during
validation.
See
Equation
C.
7
DCB
Calculated
as
a
confidence
interval
using
the
measurement
uncertainties
of
the
terms
in
Equation
C.
2.
See
Appendix
E
and
Equation
C.
8.
Log(
i)
Use
confidence
interval
of
challenge
dose­
response.
See
sections
C.
4.9.7
and
C.
4.9.8
The
safety
factor
defines
the
relationship
between
the
dose
targets
provided
in
Chapter
1
and
the
RED
that
should
be
delivered
by
the
UV
reactor
at
the
WTP.

F.
6
Re­
validation
If
the
design
of
a
validated
UV
reactor
changes,
the
UV
reactor
should
be
re­
validated
if
the
design
change
significantly
impacts
dose
delivery
or
monitoring.
Dose
delivery
and
sensor
modeling
can
be
used
to
assess
the
impact
of
the
design
change
and
justify
the
need,
or
lack
of
need,
for
re­
validation.
This
section
discusses
UV
reactor
modifications
and
provides
guidance
on
the
need
for
re­
validation.

F.
6.1
Lamp
Assembly
Design
changes
to
the
lamp
assembly
include
changes
made
by
the
lamp
manufacturer
to
the
lamp,
selection
of
a
new
lamp
type
by
the
UV
manufacturer,
and
changes
made
by
the
UV
manufacturer
to
the
components
associated
with
the
lamp
assembly.
The
relationship
between
dose
delivery
and
monitoring
may
be
impacted
by
any
design
change
involving
modifications
to
the
following
components:

·
Lamp
arc
length
·
Any
reflectors,
connectors,
and
spacers
used
at
the
lamp
ends
·
Lamp
envelope
diameter
·
Lamp
envelope
UVT
from
185
nm
to
400
nm
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
40
June
2003
Proposal
Draft
·
Mercury
content
of
the
lamp
·
Argon
content
of
the
lamp
The
lamp's
arc
length
and
the
use
of
components
at
the
ends
of
the
lamps
(
like
reflectors,
spacers,
and
connectors)
impact
the
UV
intensity
field
in
the
region
near
the
lamp
ends.
Design
changes
to
these
components
could
impact
dose
delivery,
especially
if
the
lamps
are
oriented
perpendicular
to
flowrate.
Design
changes
could
also
impact
UV
intensity
sensor
measurements
if
the
lamp
ends
are
within
the
viewing
angle
of
the
sensors.
Dose
delivery
and
UV
intensity
sensor
modeling
can
be
used
to
assess
the
impacts
on
changing
the
lamp
arc
length
or
components
used
at
the
lamp
ends.
If
the
impacts
are
considered
significant,
the
reactor
should
be
re­
validated.

With
LP
lamps,
the
UV­
emitting
plasma
occupies
the
space
within
the
lamp
envelope.
With
MP
lamps,
the
plasma
forms
a
narrow
arc
that
occupies
a
portion
of
the
space
within
the
lamp
envelope.
In
the
presence
of
electromagnetic
fields,
the
plasma
within
a
MP
lamp
can
be
displaced
off
center
within
the
lamp.
The
diameter
of
a
plasma
centered
within
the
lamp
envelope
should
have
a
small
impact
on
the
UV
intensity
field
and
dose
delivery
(
Bolton
2000).
However,
displacement
of
the
plasma
off­
center
within
the
envelope
could
impact
the
intensity
field
and
dose
delivery.
The
reactor
should
be
re­
validated
if
design
changes
to
the
lamp
diameter
significantly
impact
the
intensity
field.

The
UVT
of
the
lamp
envelope
will
impact
the
UV
output
of
both
LP
and
MP
lamps.
With
LP
lamps,
envelope
material
can
be
selected
to
allow
or
prevent
LP
lamps
from
emitting
UV
light
at
185
nm.
While
UV
light
at
185
nm
has
a
negligible
impact
on
dose
delivery
and
UV
intensity
sensor
measurements
because
of
the
high
UV
absorbance
of
water
at
this
wavelength,
185
nm
light
may
promote
the
formation
of
ozone
within
the
lamp
sleeve.
Ozone
will
absorb
UV
light
at
254
nm
and
lower
the
output
from
the
lamp.
Ozone
could
degrade
components
within
the
lamp
assembly
leading
to
internal
sleeve
fouling.
Typically,
LP
lamps
are
selected
with
envelopes
that
prevent
output
at
185
nm.

With
MP
lamps,
the
envelope
material
has
a
significant
impact
on
the
intensity
of
UV
light
emitted
below
260
nm.
Lamp
envelope
material
can
be
selected
to
eliminate
or
maximize
UV
output
at
lower
wavelengths.
Since
envelope
transmittance
decreases
with
increased
temperature,
the
UVT
of
the
envelope
of
a
MP
lamp
should
be
assessed
at
the
operating
temperature
of
the
lamp.
Dose
delivery
and
UV
intensity
sensor
modeling
can
be
used
to
assess
the
impacts
of
changing
lamp
material
and
justify
the
need
for
re­
validation.

LP
lamps
typically
operate
near
40
º
C
with
a
relatively
low
mercury
vapor
pressure
that
promotes
UV
output
at
254
nm.
Because
the
amount
of
mercury
added
to
the
lamp
is
well
in
excess
of
the
amount
that
enters
the
vapor
state
during
lamp
operation,
the
UV
output
of
a
LP
lamp
is
independent
of
the
mercury
dose
added
to
the
lamp
during
lamp
manufacture.
On
the
other
hand,
MP
lamps
operate
at
a
high
temperature,
near
600
º
C,
with
all
of
the
added
mercury
in
the
vapor
phase.
As
such,
the
mercury
vapor
pressure
is
dependent
on
the
mercury
dose
and
the
lamp
operating
temperature.
The
vapor
pressure
influences
the
fraction
of
mercury
that
is
ionized
or
excited
to
higher
energy
states,
and
hence
the
spectral
output
of
the
MP
lamp.
Table
F.
6
presents
the
calculated
impact
of
mercury
dose
on
the
germicidal
output
and
measured
intensity
from
a
MP
lamp
operating
with
an
electrical
input
of
70
W/
cm.
The
results
suggest
that
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
41
June
2003
Proposal
Draft
a
change
in
mercury
dose
has
no
impact
on
the
relationship
between
dose
delivery
and
monitoring
with
germicidal
sensors
and
a
small
impact
on
the
relationship
with
SiC
sensors.

Table
F.
6
Impact
of
the
Mercury
Dose
on
the
Relationship
Between
Germicidal
Output
and
Measured
Output
of
a
MP
Lamp1
UV
Output
(
W/
cm)
Weighted
by
Ratios
Mercury
Dose
(
mg/
cm)
MS2
Action
SiC
Sensor
Filtered
SiC
Sensor
SiC:
MS2
Filtered
SiC:
MS2
4.8
6.88
11.4
6.82
1.65
0.990
8
6.53
10.3
6.46
1.58
0.989
10.1
7.10
10.8
7.06
1.52
0.993
1
Adapted
from
lamp
output
data
from
248
to
400
nm
provided
by
Phillips
(
1983).

F.
6.2
Ballasts
Modifications
to
lamp
ballasts
include
changing
the
operating
voltage,
current,
frequency,
and
waveform.
With
LP
lamps,
modifications
will
impact
the
amount
of
UV
generated
by
the
lamp,
but
will
not
impact
the
relationship
between
dose
delivery
and
UV
intensity
measurements.
With
MP
lamps
and
some
LPHO
lamps,
changes
in
lamp
operating
temperature
and
mercury
pressure
caused
by
changes
in
ballast
power
will
impact
the
spectral
distribution
of
emitted
light.
Table
F.
7
presents
the
impact
of
changing
the
input
power
from
48
to
92
W/
cm
on
the
germicidal
output
and
measured
intensity
from
a
MP
lamp
dosed
with
4.8
mg/
cm
of
mercury.
The
results
suggest
a
change
in
lamp
operating
power
has
no
impact
on
the
relationship
between
dose
delivery
and
monitoring
with
germicidal
sensors
and
a
small
impact
with
SiC
sensors.

Table
F.
7
Impact
of
Operating
Power
on
the
Relationship
Between
Germicidal
Output
and
Measured
Output
of
a
MP
Lamp1
UV
Output
(
W/
cm)
Weighted
by
Ratios
Lamp
Input
Power
(
W/
cm)
MS2
Action
SiC
Sensor
Filtered
SiC
Sensor
SiC:
MS2
Filtered
SiC:
MS2
48
4.13
7.01
4.08
1.70
0.99
70
6.86
11.3
6.78
1.66
0.99
92
9.29
15.2
9.14
1.65
0.98
1
Adapted
from
lamp
output
data
from
248
to
400
nm
provided
by
Phillips
(
1983)
for
a
MP
lamp
dosed
with
4.8
mg/
cm
Hg.

F.
6.3
Lamp
Sleeves
Design
changes
to
the
lamp
sleeves
include
changing
the
sleeve
diameter,
thickness,
and
material.
Changing
the
sleeve
diameter
may
impact
the
hydraulics
through
the
reactor,
the
measurement
of
UV
intensity,
and
the
optimal
placement
of
UV
intensity
sensors
relative
to
the
lamp.
Changing
the
thickness
and
material
of
the
lamp
sleeve
will
impact
the
spectral
UVT,
thereby
impacting
both
dose
delivery
and
UV
intensity
measurements.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
42
June
2003
Proposal
Draft
Dose
delivery
and
UV
intensity
sensor
modeling
may
be
used
to
assess
the
impact
of
lamp
sleeve
design
changes.
Figure
F.
19
provides
the
ratio
of
dose
delivered
with
a
standard
sleeve
to
dose
delivered
with
an
"
ozone­
free"
sleeve
for
a
given
sensor
reading
as
a
function
of
water
UVT,
sensor­
to­
lamp
distance,
and
sensor
spectral
response.
Dose
and
UV
intensity
values
were
predicted
using
Equations
F.
14
and
F.
15
applied
to
the
annular
reactor
described
in
section
F.
4.2.
Sleeve
UVT
is
provided
in
Figure
F.
20.
The
results
show
that
a
design
change
from
a
regular
sleeve
to
an
ozone­
free
sleeve
described
in
Figure
F.
20
would
have
a
small
impact
on
the
relationship
between
dose
delivery
and
UV
intensity
sensor
readings
with
a
SiC
sensor
and
a
negligible
impact
with
a
germicidal
sensor.
Modeling
can
also
be
used
to
show
that
the
dose
delivery
at
a
given
lamp
output,
water
UVT,
and
flowrate
would
be
approximately
10
percent
greater
with
the
standard
sleeve
than
with
the
ozone­
free
sleeve.
If
models
indicate
the
sleeve
design
change
causes
a
significant
impact
on
dose
delivery
and
monitoring,
the
UV
reactor
should
be
re­
validated.

Figure
F.
19
Ratios
of
Dose
Delivered
with
Standard
Sleeve
to
Dose
Delivered
with
"
Ozone­
Free"
Sleeves
by
an
Annular
Reactor
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
43
June
2003
Proposal
Draft
Figure
F.
20
UVT
of
Standard
and
"
Ozone­
Free"
Quartz
Assuming
Air­
Quartz
and
Quartz­
Water
Interfaces
F.
6.4
Reactor
and
Component
Dimensions
Modifications
to
the
wetted
dimensions
and
positioning
of
the
components
within
the
reactor
will
impact
the
reactor
hydraulics
and
dose
delivery.
Modifications
could
also
impact
the
intensity
field
within
the
reactor
and
the
measurement
of
UV
intensity.
Modifications
include
changes
to
the
dimensions
of
the
reactor,
inlet
piping,
exit
piping,
baffles,
lamp
sleeves,
wipers,
and
UV
intensity
sensors.
The
impact
of
such
modifications
on
dose
delivery
and
UV
intensity
measurements
can
be
insignificant
or
large.
Addition
of
a
baffle
plate
will
likely
have
a
large
impact
on
dose
delivery
and
a
small
impact
on
measured
UV
intensity,
while
changing
the
position
of
a
UV
intensity
sensor
will
likely
have
a
small
impact
on
dose
and
a
large
impact
on
measured
UV
intensity.
Dose
delivery
and
UV
intensity
modeling
may
be
used
to
assess
the
impacts
of
these
modifications.
If
the
impacts
are
significant,
the
reactor
should
be
re­
validated.

F.
6.5
UV
Intensity
Sensors
Modifications
to
the
UV
intensity
sensors
include
changes
made
by
the
sensor
manufacturer
to
the
sensor,
changes
by
the
UV
manufacturer
to
the
sensor
housing
and
associated
optical
components,
and
changes
by
the
UV
manufacturer
to
the
number
and
positioning
of
the
sensors
within
the
reactor.

Changes
to
the
semi­
conductor
and
optical
components
within
the
UV
intensity
sensor
could
impact
the
sensor's
spectral
response,
linearity,
angular
response,
and
temperature
stability.
Changes
to
those
properties
could
impact
the
sensor's
measurement
uncertainty.
If
the
new
measurement
uncertainty
is
quantified,
it
should
be
used
to
define
a
new
safety
factor
for
the
UV
reactor.
If
the
angular
response
or
spectral
response
of
the
sensor
changes,
measurements
supported
by
calculations
should
be
used
to
evaluate
the
impact
of
the
change
on
dose
delivery
monitoring.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
44
June
2003
Proposal
Draft
Changes
to
the
measuring
window
of
the
UV
intensity
sensor
include
dimensional
and
material
changes.
Changes
may
impact
the
UVT
of
the
window
and
the
detection
angle.
Measurements
supported
by
calculations
should
be
used
to
evaluate
the
impact
of
the
change
on
dose
delivery
monitoring.

Modifications
to
the
positioning
of
the
UV
intensity
sensor
within
the
reactor
could
disturb
the
flowrate
and
impact
dose
delivery.
If
the
impact
on
dose
delivery
is
negligible,
measurements
supported
by
calculations
may
be
used
to
compare
measured
UV
intensity
at
the
two
positions
and
modify
the
dose
monitoring
approach
without
the
need
for
re­
validation.

Addition
of
UV
intensity
sensors
to
the
reactor
could
disturb
the
flowrate
through
the
UV
reactor
and
impact
dose
delivery.
If
sensors
are
added,
they
should
be
positioned
relative
to
the
lamps
in
a
similar
manner
as
the
other
sensors.
For
example,
if
one
sensor
is
positioned
to
view
two
lamps
through
a
5­
cm
water
layer,
then
all
added
sensor
should
view
two
lamps
through
a
5
cm
water
layer.

F.
7
References
Bolton,
J.
R.
2000.
Calculation
of
ultraviolet
fluence
rate
distributions
in
an
annular
reactor:
significance
of
refraction
and
reflection.
Water
Research
34:
3315­
3324.

Cabaj,
A.,
R.
Sommer,
and
D.
Schoenen.
1996.
Biodosimetry:
model
calculations
for
U.
V.
water
disinfection
devices
with
regard
to
dose
distributions.
Water
Research.
30,
no.
4:
1003­
1009.

Cabaj,
A.,
R.
Sommer,
W.
Pribil,
and
T.
Haider.
2002.
The
spectral
UV
sensitivity
of
microorgnisms
used
in
biodosimetry.
Water
Science
and
Technology:
Water
Supply,
2(
3):
175­
181.

Chang,
J.
C.
H.,
S.
F.
Osoff,
D.
C.
Lobe,
M.
H.
Dorfman,
C.
M.
Dumais,
R.
G.
Qualls,
and
J.
D.
Johnson.
1985.
UV
inactivation
of
pathogenic
and
indicator
microorganisms.
Applied
and
Environmental
Microbiology
49(
6):
1361­
1365.

Chiu,
K.­
P.,
D.
A.
Lyn,
P.
Savoye,
and
E.
R.
Blatchley.
1999.
Effect
of
UV
system
modifications
on
disinfection
performance.
Journal
of
Environmental
Engineering
125(
5):
459­
469.

Haas,
C.
N.
and
G.
P.
Sakellaropoulos.
1979.
Rational
analysis
of
ultraviolet
disinfection.
National
Conference
on
Environmental
Engineering,
Proc.
ASCE
Specialty
Conf.,
San
Francisco,
CA,
July
9­
11.

Kawar,
K.,
J.
Jenkins,
B.
Srikanth,
and
A.
Shurtleff.
1998.
Ultraviolet
light
deterioration
of
the
light
transmittance
of
quartz
sleeves
due
to
continuous
exposure
to
UV
radiation.
Ultrapure
Water
October,
67­
71.

Larason,
T.
C.,
S.
S.
Bruce,
and
A.
C.
Parr.
1998.
Spectroradiometric
Detector
Measurements.
NIST
Special
Publication
250­
41.
U.
S.
Government
Printing
Office,
Washington.
Appendix
F.
Background
to
the
UV
Reactor
Validation
Protocol
UV
Disinfection
Guidance
Manual
F­
45
June
2003
Proposal
Draft
Larason
T.
C.
and
C.
L.
Cromer.
2001.
Source
of
error
in
UV
radiation
measurements.
Journal
of
Research
of
the
National
Institute
of
Standards
and
Technology
106,
no.
4:
649­
656.

Lin,
L.,
C.
T.
Johnston,
E.
R.
Blatchley.
1999.
Inorganic
fouling
at
quartz:
water
interfaces
in
ultraviolet
photoreactors
I:
chemical
characterization.
Water
Research
33:
3321­
3329.

Linden
K.
G.,
G.
Shin,
and
M.
D.
Sobsey.
2001.
Comparative
effectiveness
of
UV
wavelengths
for
the
inactivation
of
Cryptosporidium
parvum
oocysts
in
water.
Water
Science
&
Technology
43(
12):
171­
174.

Meng,
Q.
S.
and
C.
P.
Gerba.
1996.
Comparative
inactivation
of
enteric
adenovirus,
poliovirus
and
coliphages
by
ultraviolet
irradiation.
Wat.
Res.
30,
no.
11:
2665­
2668.

Phillips,
R.
1983.
Sources
and
Applications
of
Ultraviolet
Radiation.
New
York:
Academic
Press,

Qualls,
R.
G.
and
J.
D.
Johnson.
1983.
Bioassay
and
dose
measurement
in
UV
disinfection.
Applied
and
Environmental
Microbiology
45,
no.
3:
872­
877.

Rauth,
A.
M.
1965.
The
physical
state
of
viral
nucleic
acid
and
the
sensitivity
of
viruses
to
ultraviolet
light.
Biophysical
Journal
5:
257­
273.

Sommer,
R.,
T.
Haider,
A.
Cabaj,
W.
Pribil,
and
M.
Lhotsky.
1998.
Time
dose
reciprocity
in
UV
disinfection
of
water.
IAWQ
Vancouver
1998
Poster.

Wilson,
B.
R.,
P.
F.
Roessler,
E.
Van
Dellen,
M.
Abbaszadegan,
and
C.
P.
Gerba.
1992
Coliphage
MS­
2
as
a
UV
water
disinfection
efficacy
test
surrogate
for
bacterial
and
viral
pathogens.
Proceedings
of
the
Water
Quality
Technology
Conference,
Nov.
15­
19,
1992,
Toronto,
219­
235.

Wright,
H.
B.
and
Y.
A.
Lawryshyn.
2000.
An
assessment
of
the
bioassay
concept
for
UV
reactor
validation.
Disinfection
2000:
Disinfection
of
Wastes
in
the
New
Millenium,
New
Orleans,
Louisiana,
March
15­
18,
2000;
Water
Environment
Federation,
Alexandria,
Virginia
Wright,
H.
B.,
E.
Mackey,
and
P.
White.
2002.
UV
Disinfection
Compliance
Monitoring
for
Drinking
Water
Applications.
Proceedings
of
the
Water
Quality
technology
Conference
and
Exhibition,
Seattle,
November
10
­
14,
2002.
Appendix
G.
Issues
for
Unfiltered
Systems
Unfiltered
systems
are
utilities
that
use
surface
water
sources
and
meet
the
filtration
avoidance
criteria
of
the
Surface
Water
Treatment
Rule
(
SWTR)
(
40
CFR
141.71).
The
Long
Term
2
Enhanced
Surface
Water
Treatment
Rule
(
LT2ESWTR)
requires
unfiltered
systems
to
meet
overall
disinfection
requirements
(
i.
e.,
Cryptosporidium,
Giardia,
and
virus
inactivation)
using
a
minimum
of
two
disinfectants
(
40
CFR
141.721(
d)).
The
information
presented
in
this
manual
is
focused
on
post­
filtration
applications
of
UV
disinfection;
however,
the
information
is
also
relevant
to
UV
disinfection
of
unfiltered
supplies.
In
addition,
the
UV
dose
requirements
presented
in
section
1.3.1.3
are
applicable
to
both
filtered
water
and
water
supplies
that
meet
the
regulatory
requirements
for
filtration
avoidance
(
40
CFR
141.729(
d)).
This
appendix
identifies
issues
that
are
specific
to
unfiltered
applications
of
UV
disinfection.
The
following
issues
are
of
particular
interest
to
unfiltered
supplies
because
they
make
applying
UV
disinfection
different
from
post­
filter
locations:

 
Water
quality
(
especially
particle
content)

 
Debris
 
Ozone
residual
(
when
ozone
is
applied
prior
to
UV
disinfection)

 
Off­
specification
requirements
recommendations
G.
1
Water
Quality
Differences
in
the
quantity
and
nature
of
particles
in
unfiltered
surface
water
supplies
are
the
most
pertinent
distinction
between
post­
filtration
and
unfiltered
supply
water
qualities.
Typically,
the
turbidity
in
unfiltered
surface
waters
is
less
than
1
nephelometric
turbidity
units
(
NTU).
However,
the
SWTR
allows
turbidity
up
to
5
NTU
immediately
prior
to
the
first
point
of
disinfection
application
(
40
CFR
141.71).
Several
studies
have
examined
the
effects
of
turbidity
up
to
10
NTU
on
UV
disinfection,
including
changes
in
UV
absorbance
measurements
made
with
a
spectrophotometer
and
inactivation
of
microorganisms.

Particles
in
water
absorb
and
scatter
UV
light
to
varying
degrees
based
on
size
and
composition.
Particles
impact
the
disinfection
process
in
two
distinct
manners:

1.
Particles
can
decrease
the
UV
transmittance
(
UVT)
of
water
and
thereby
impact
UV
dose
delivery
(
section
A.
4.1.2).

2.
Particle
association
can
shield
microorganisms
from
UV
light,
thereby
changing
the
characteristics
of
the
UV
dose­
response
curve
(
section
A.
2.6.5).

Christensen
and
Linden
(
2001)
concluded
that
the
light
scattering
and
changes
in
absorbance
caused
by
turbidity
up
to
10
NTU
can
be
accounted
for
when
calculating
UV
dose
in
collimated
beam
testing
provided
that
the
ultraviolet
absorbance
at
254
nanometers
(
A254)
of
the
sample
is
measured
according
to
a
modified
version
of
Standard
Method
5910B
(
i.
e.,
without
UV
Disinfection
Guidance
Manual
G­
1
June
2003
Proposal
Draft
Appendix
G.
Issues
for
Unfiltered
Systems
UV
Disinfection
Guidance
Manual
G­
2
June
2003
0.45
µ
m
filtration).
Direct
reading
spectrophotometers,
the
most
common
type
of
spectrophotometer,
may
overestimate
the
A254
of
water
with
turbidity
greater
than
3
NTU,
resulting
in
an
overly
conservative
UV
dose
calculation
(
Christensen
and
Linden
2002).
To
reduce
this
overestimation,
an
integrating
sphere
can
be
installed
in
a
direct­
reading
spectrophotometer
that
will
provide
accurate
A254
measurements.
Regardless
of
the
type
of
spectrophotometer
used,
the
effects
of
increased
absorbance
due
to
particles
can
be
accounted
for
in
the
A254
measurement,
which
can
then
be
used
to
determine
the
design
UVT.
If
an
appropriate
design
UVT
is
used,
the
UV
reactor
will
be
able
to
respond
to
changes
in
UVT
that
arise
due
to
particles.

Particles
and
microorganisms
in
a
water
sample
are
either
dispersed
or
aggregated
together.
Studies
have
demonstrated
that
dispersed
coliform
bacteria
in
wastewater
are
easier
to
disinfect
than
aggregated
bacteria
(
Parker
and
Darby
1995).
To
date,
research
examining
the
effects
of
particles
in
drinking
water
on
UV
disinfection
has
been
performed
with
seeded
organisms
and
particles.
It
is
unknown
at
this
time
how
well
these
studies
represent
naturally
occurring
microorganism
and
particle
interactions.
However,
since
the
concentration
of
microorganisms
in
unfiltered
sources
is
typically
below
detectable
limits,
methods
to
examine
this
phenomenon
directly
(
without
seeding)
do
not
currently
exist.
Consequently,
seeded
drinking
water
studies
can
only
suggest
the
impact
of
turbidity
on
dose­
response
as
it
relates
to
the
impact
of
UV
light
scattering
by
particles
rather
than
particle­
association
or
clumping
of
microorganisms.

Recent
research
has
shown
that
particles
present
in
supplies
meeting
regulatory
requirements
for
unfiltered
drinking
water
do
not
impact
the
UV
inactivation
of
seeded
microorganisms.
Passantino
and
Malley
(
2001)
reported
that
for
unfiltered
surface
waters,
turbidity
up
to
7
NTU
does
not
affect
the
inactivation
of
seeded
male
specific­
2
bacteriophage
(
MS2)
in
bench­
scale,
batch,
collimated
beam
testing.
In
this
study,
turbidity
was
increased
by
adding
natural
sediment
to
waters
collected
from
unfiltered
water
supplies.
Therefore,
naturally
occurring
interactions
between
particles
and
microorganisms
could
not
be
evaluated.
In
another
study,
batch
(
bench­
scale)
and
continuous­
flow
(
pilot­
scale)
studies
showed
that
turbidity
ranging
from
0.65
to
7
NTU
does
not
affect
the
UV
dose
necessary
per
log
inactivation
of
seeded
MS2,
Giardia
muris,
or
Cryptosporidium
parvum
in
unfiltered
waters
(
Oppenheimer
et
al.
2002).
Womba
et
al.
(
2002)
evaluated
the
impact
of
turbidity
on
UV
inactivation
of
MS2
at
the
bench­
and
pilot­
scale.
They
found
that
on
the
bench­
scale,
when
the
impact
of
turbidity
was
accounted
for
in
the
UV
dose
determination,
the
inactivation
of
MS2
was
not
affected
by
turbidity.
However,
in
this
study
on
the
pilot­
scale,
because
the
lamp
intensity
and
flowrate
(
and
therefore
residence
time
in
the
reactor)
remained
constant,
the
effects
of
turbidity
were
not
accounted
for
in
the
reactor
control
strategy.
Therefore,
the
reduction
equivalent
dose
(
RED)
observed
decreased
as
turbidity
increased.

Unfiltered
supplies
are
also
susceptible
to
algal
blooms.
Womba
et
al.
(
2002)
monitored
algae
levels
in
an
unfiltered
supply
reservoir
for
over
one
year
and
found
that
algal
counts
were
typically
below
30,000
cells/
mL;
however
one
algae
event
had
a
higher
level
of
nearly
300,000
cells/
mL.
Although
not
regulated,
the
presence
of
algae
may
interfere
with
the
UV
disinfection
process.
Womba
et
al.
(
2002)
and
Passantino
and
Malley
(
2001)
examined
the
effects
of
algae
on
UV
disinfection
of
MS2
at
the
bench­
scale
in
batch,
collimated
beam
testing.
Both
studies
found
that
up
to
algal
counts
up
to
70,000
cells/
mL
and
42,000
cells/
mL,
respectively,
do
not
affect
the
inactivation
of
MS2.

Proposal
Draft
Appendix
G.
Issues
for
Unfiltered
Systems
UV
Disinfection
Guidance
Manual
G­
3
June
2003
G.
2
Debris
Relative
to
post­
filter
applications
of
UV
disinfection,
there
may
be
greater
opportunity
for
debris
to
be
present
in
the
influent
to
UV
reactors
in
unfiltered
applications.
Debris
entering
the
UV
reactor
with
sufficient
momentum
could
cause
lamp
sleeve
and
lamp
breakage.
The
mass
and
size
of
an
object
that
might
cause
damage
is
installation­
specific
and
depends
on
UV
reactor
configuration
(
e.
g.,
horizontal
versus
vertical
reactor
orientation)
and
water
velocity
through
the
reactor.
As
such,
designs
should
incorporate
features
that
prevent
potentially
damaging
objects
from
entering
the
system;
the
optimal
approach
is
site­
specific.
Such
features
could
include
screens,
baffles,
or
low
velocity
collection
areas.
Another
option
is
to
install
the
UV
reactors
vertically
with
the
inlet
closest
to
the
ground,
following
a
low
velocity
zone.
This
arrangement
will
decrease
the
momentum
of
larger
debris
and
reduce
the
risk
of
lamp
breakage.
The
effects
of
lamp
breakage
and
methods
of
minimizing
it
are
discussed
in
Appendix
N.

G.
3
Ozone
Impacts
on
Absorbance
Some
utilities
using
an
unfiltered
source
may
consider
applying
ozone
in
addition
to
UV
disinfection.
There
are
a
number
of
benefits
associated
with
this
process
combination,
including
addressing
multi­
barrier
disinfection
requirements.
Additionally,
if
ozone
is
added
prior
to
UV
disinfection,
the
A254
of
the
water
can
be
reduced
as
shown
in
Figure
G.
1,
thereby
improving
the
efficiency
of
UV
disinfection.

Figure
G.
1
Impact
Of
Pre­
Ozonation
On
A254
(
Malley
2002).

W
a
v
e
le
n
g
th
(
n
m
)
2
0
0
2
2
0
2
4
0
2
6
0
2
8
0
3
0
0
3
2
0
UV
Absorbance
(
cm­
1)

0
.0
0
.1
0
.2
0
.3
0
.4
0
.5
P
re
o
z
o
n
a
te
d
U
V
In
flu
e
n
t
(
n
o
d
e
te
c
ta
b
le
o
z
o
n
e
re
s
id
u
a
l)
U
V
I
n
f
lu
e
n
t
­
N
o
o
z
o
n
e
It
should
be
noted,
however,
that
ozone
is
a
strong
UV
absorber
with
a
molar
absorbance
value
of
0.0677
L/
mg/
cm
at
254
nm.
Figure
G.
2
illustrates
the
impact
of
ozone
concentration
on
UVT1
for
three
baseline
transmittance
values.
If
ozone
is
applied
prior
to
UV
reactors
and
residual
ozone
enters
the
UV
reactor,
the
increase
in
UV
absorbance
due
to
ozone
residual
can
be
 
=
1
UVT
254
10
100
(%)
A
Proposal
Draft
Appendix
G.
Issues
for
Unfiltered
Systems
UV
Disinfection
Guidance
Manual
G­
4
June
2003
significant
and
should
be
considered
when
determining
the
design
UVT.
To
address
this
issue,
utilities
can
monitor
the
ozone
residual
and
add
an
ozone­
reducing
chemical
to
maintain
the
ozone
residual
below
a
specified
setpoint
value
(
e.
g.,
0.1
mg/
L).
There
are
several
chemicals
that
can
be
used
to
quench
ozone;
however,
some
chemicals
(
such
as
sodium
thiosulfate)
have
a
high
molar
absorbance
value
(
as
shown
in
Table
A.
5,
section
A.
4.1.3),
and
thus
have
the
potential
to
decrease
the
UVT.
Such
chemicals
should
not
be
used
prior
to
UV
disinfection.
Sulfite
has
a
lower
molar
absorbance
value
and
is
therefore
an
acceptable
chemical
to
quench
ozone
residual.
The
impact
of
water
treatment
chemicals
on
UV
absorbance
can
be
assessed
by
preparing
solutions
of
various
concentration
and
measuring
their
UV
absorbance
using
a
standard
spectrophotometer
(
Bolton
et
al.
2001).

Figure
G.
2.
Impact
of
Ozone
Residual
on
UVT
(
adapted
from
Cushing
et
al.
2001)

60%
70%
80%
90%
100%

0.00
0.25
0.50
0.75
1.00
Aqueous
Ozone
Concentration
(
mg/
L)
UVT
(
254
nm;
10
mm
pathlength)

Initial
UVT
=
80%
Initial
UVT
=
87%
Initial
UVT
=
95%

In
at
least
some
cases,
the
increase
in
UVT
resulting
from
ozone
addition
will
improve
overall
UV
disinfection
effectiveness
provided
that
any
remaining
ozone
residual
is
adequately
controlled.
Each
utility
should
explore
the
sequence
of
disinfectants
that
best
fits
their
sitespecific
objectives
and
constraints.

G.
4
Off­
specification
Requirements
Off­
specification
is
when
the
UV
reactor
is
operating
outside
of
its
validated
limits.
UV
installations
should
be
designed
with
process
monitoring
and
control
components
(
e.
g.,
alarms,
shut­
off
valves)
to
prevent
water
from
entering
the
distribution
system
when
a
UV
reactor
is
operating
outside
of
validated
conditions.
Unfiltered
systems
that
use
UV
disinfection
to
meet
the
Cryptosporidium
treatment
requirement
of
the
LT2ESWTR
must
demonstrate
that
at
least
95
Proposal
Draft
Appendix
G.
Issues
for
Unfiltered
Systems
UV
Disinfection
Guidance
Manual
G­
5
June
2003
percent
of
the
water
delivered
to
the
public
during
each
month
is
treated
by
UV
reactors
operating
within
validated
limits
(
i.
e.,
operating
conditions
that
have
been
validated
to
achieve
the
necessary
log
inactivation)
(
40
CFR
141.721(
c)).
Failure
to
demonstrate
this
will
result
in
a
treatment
technique
violation.

The
UV
reactors
are
off­
specification
when
any
of
the
following
conditions
occur
(
40
CFR
141.729(
d)):

 
The
flow,
UV
intensity,
or
lamp
status
is
outside
of
the
validated
range.

 
The
UVT
or
UV
intensity
is
outside
of
the
validated
range
(
if
the
UV
intensity
and
UVT
setpoint
approach
is
used
(
section
3.1.5))

 
The
calculated
dose
is
outside
of
the
validated
range
at
a
given
flow
(
if
the
calculated
dose
approach
is
used
(
section
3.1.5))

 
All
UV
lamps
in
all
UV
reactors
are
off
because
of
a
power
interruption
or
power
quality
problem
(
as
discussed
in
section
3.1.3.3),
and
water
is
flowing
through
the
reactors.

More
information
on
off­
specification
is
in
section
3.1.3,
and
compliance
information
is
in
section
5.4.1.

G.
5
References
Bolton,
J.
R.,
M.
I.
Stefan,
R.
S.
Cushing,
and
E.
Mackey.
2001.
Importance
of
water
absorbance/
transmittance
on
the
efficiency
of
ultraviolet
disinfection
reactors.
Proceedings
of
the
IUVA
1st
International
Congress,
June
14­
16,
Washington,
D.
C.

Christensen,
J.
and
K.
Linden.
2001.
Ultraviolet
disinfection
of
unfiltered
drinking
water:
particle
impacts.
Proceedings
of
the
IUVA
1st
International
Congress,
June
14­
16,
Washington,
D.
C.

Christensen,
J.
and
K.
Linden.
2002.
New
findings
regarding
the
impacts
of
suspended
particles
on
UV
disinfection
of
drinking
water.
Proceedings
of
the
AWWA
Annual
Conference,
June
16­
20,
New
Orleans,
L.
A.

Cushing,
R.
S.,
E.
D.
Mackey,
J.
R.
Bolton,
and
M.
I.
Stefan.
2001.
Impact
of
common
water
treatment
chemicals
on
UV
disinfection.
Proceedings
of
the
AWWA
Annual
Conference
and
Exposition,
June
17­
21,
Washington
DC.

Malley,
J.
P.
2002.
Historical
perspective
of
UV
use.
Presented
at
the
AWWA
Water
Quality
Technology
Conference,
November
10­
14,
Seattle,
W.
A.

Proposal
Draft
Appendix
G.
Issues
for
Unfiltered
Systems
UV
Disinfection
Guidance
Manual
G­
6
June
2003
Oppenheimer
J.,
T.
Gillogly,
G.
Stolarik,
and
R.
Ward.
2002.
Comparing
the
efficiency
of
low
and
medium
pressure
UV
light
for
inactivating
Giardia
muris
and
Cryptosporidium
parvum
in
waters
with
low
and
high
levels
of
turbidity.
Proceedings
of
the
AWWA
Annual
Conference,
June
16­
20,
New
Orleans,
L.
A.

Parker,
J.
A.
and
J.
L.
Darby.
1995.
Particle­
associated
coliform
in
secondary
effluents:
shielding
from
ultraviolet
light
disinfection.
Water
Environment
Research
67:
1065­
1075.

Passantino,
L.
and
J.
P.
Malley.
2001.
Impacts
of
turbidity
and
algal
content
of
unfiltered
drinking
water
supplies
on
the
ultraviolet
disinfection
process.
Proceedings
of
the
AWWA
Annual
Conference
and
Exposition,
June
17­
21,
Washington
D.
C.

Womba
P.,
W.
Bellamy,
J.
Malley,
and
C.
Douglas.
2002.
UV
disinfection
and
disinfection
byproduct
characteristics
of
an
unfiltered
water
supply,
Project
2747.
Denver,
C.
O.:
AwwaRF
periodic
report.

Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
The
UV
installation
design,
operation,
and
maintenance
principles
presented
in
Chapters
3
and
5
of
this
manual
are
focused
on
the
use
of
UV
reactors
to
disinfect
filtered
surface
water.
Most
of
the
information
presented
in
those
chapters
is
also
applicable
to
ground
water
systems.
Additional
ground
water­
specific
regulatory
requirements
and
recommendations,
site
issues,
hydraulic
issues,
and
water
quality
issues
that
affect
design
and
operation
are
discussed
in
this
appendix.

H.
1
Ground
Water
Systems
Background
Regulations
should
be
reviewed
to
determine
the
goals
and
requirements
for
disinfection.
Existing
treatment
processes
and
distribution
system
parameters
should
also
be
analyzed
before
selecting
a
strategy
for
integrating
UV
disinfection
into
the
system.

H.
1.1
Regulatory
Background
Currently,
federal
regulations
do
not
require
ground
water
systems
to
provide
primary
or
secondary
disinfection
unless
the
water
is
a
ground
water
source
under
the
direct
influence
of
surface
water
(
GWUDI).
However,
some
States
require
ground
water
systems
to
maintain
a
residual
disinfectant
in
the
distribution
system.
In
addition,
ground
water
systems
are
required
to
meet
the
requirements
of
the
Total
Coliform
Rule
(
TCR)
(
54
FR
27544)
and
the
Stage
1
Disinfection
and
Disinfection
Byproducts
Rule
(
DBPR)
(
63
FR
69390)
and
are
expected
to
be
affected
by
the
upcoming
Stage
2
DBPR.

The
upcoming
Ground
Water
Rule,
as
proposed,
would
require
some
ground
water
systems
to
provide
4­
log
removal
or
inactivation
of
viruses.
Systems
with
significant
deficiencies,
as
determined
by
States
during
sanitary
surveys,
and
systems
that
detect
fecal
indicators
in
their
source
water
will
be
affected.
These
systems
will
be
required
to
correct
any
deficiencies,
provide
water
from
an
alternative
source,
or
install
treatment
that
provides
4­
log
removal
or
inactivation
of
viruses.

Ground
water
supplies
that
are
designated
as
GWUDI,
as
defined
in
the
Surface
Water
Treatment
Rule
(
SWTR),
40
CFR
Part
141.2,
are
classified
as
Subpart
H
Systems
and
must
meet
the
same
regulatory
requirements
as
surface
water
systems.
GWUDI
systems
often
use
many
of
the
same
treatment
strategies
as
surface
water
systems,
including
filtration.
The
issues
involved
with
implementing
UV
disinfection
at
filtered
water
utilities
(
including
GWUDI)
are
discussed
in
detail
in
Chapters
1
through
5
of
this
manual.
GWUDI
systems
are
subject
to
the
Stage
1
DBPR
and
would
be
subject
to
the
upcoming
Stage
2
DBPR
and
the
Long
Term
2
Enhanced
Surface
Water
Treatment
Rule
(
LT2ESWTR).
Both
of
these
regulations
are
summarized
in
section
1.3.

UV
Disinfection
Guidance
Manual
H­
1
June
2003
Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
2
June
2003
H.
1.2
Typical
Ground
Water
System
Design
Most
ground
water
systems
operate
by
cycling
ground
water
pumps
on
and
off
in
response
to
demand
or
storage
capacity.
Because
significant
treatment
is
not
usually
necessary
beyond
secondary
disinfection,
ground
water
systems
typically
do
not
have
a
single,
centralized
treatment
system.
Many
ground
water
systems
pump
to
storage
(
e.
g.,
hydropneumatic
tank,
elevated
storage
tank),
but
some
may
discharge
directly
to
the
distribution
system.
A
production
well
typically
consists
of
a
well
pump,
and
may
contain
a
chlorinator
(
for
secondary
disinfection)
and
corrosion
control
equipment
(
for
Lead
and
Copper
Rule
compliance),
air
release
valves,
vacuum
relief
valves,
and
other
ancillary
equipment
necessary
for
well
operation
(
Figure
H.
1).

Figure
H.
1.
Typical
Ground
Water
Well
Site
Layout
UV
reactors
may
be
installed
at
each
well
in
a
production
system.
If
multiple
wells
are
located
in
the
same
area,
centralizing
the
flow
through
a
common
header
minimizes
the
number
of
UV
reactors
needed
and
possibly
reduces
the
project
cost.
In
addition,
treatment
for
other
aesthetic
issues
(
e.
g.,
removal
of
iron
and
manganese
or
stripping
of
sulfides)
may
be
more
effectively
accomplished
with
centralized
treatment.
An
engineering
cost
analysis
should
be
conducted
to
compare
centralized
treatment
with
the
installation
of
individual
reactors
at
each
well.

H.
2
Water
Quality
Issues
Although
ground
water
typically
exhibits
small
variations
in
water
quality,
specific
parameters
need
to
be
analyzed
when
planning
for
a
ground
water
system.
As
with
surface
water
systems,
UV
absorbance
at
254
nm
(
A254)
and
the
corresponding
UV
transmittance
(
UVT)
is
the
most
important
parameter
when
designing
a
UV
installation
because
it
affects
the
UV
reactor
size.
In
addition,
many
naturally
occurring
constituents
present
in
ground
water
(
e.
g.,
calcium,
iron,
manganese,
aluminum,
chloride,
carbonate,
sulfide,
and
phosphate)
are
capable
of
fouling
the
lamp
sleeves
in
UV
reactors,
and
these
constituents
should
be
monitored.
The
potential
for
Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
3
June
2003
fouling
is
greater
with
medium
pressure
(
MP)
reactors
than
low
pressure
(
LP)
and
low
pressure
high
output
(
LPHO)
reactors
because
MP
lamps
operate
at
higher
temperatures
(
section
2.4.2).
Mechanical
wipers
are
often
effective
at
removing
fouling
on
the
lamp
sleeves.
In
situations
where
the
ground
water
has
detectable
levels
of
iron
and
manganese,
chlorination
prior
to
the
UV
reactors
may
cause
increased
fouling
or
staining,
necessitating
chemical
cleaning
(
Malley
et
al.
2001).
A
complete
discussion
of
the
relevant
water
quality
parameters
and
the
determination
of
their
design
values
is
presented
in
section
3.1.3.1.

With
ground
water
systems,
it
is
common
for
one
or
more
wells
to
be
taken
out
of
service
for
an
extended
period
due
to
fluctuations
in
water
demand,
ground
water
quality,
operational
problems,
or
other
planned
and
unplanned
circumstances.
Toivanen
(
2000)
reported
that
the
lamp
sleeves
and
internal
surfaces
of
the
UV
reactors
became
fouled
when
the
UV
reactors
were
out­
of­
service
and
full
of
water.
The
amount
of
time
it
takes
to
foul
the
UV
reactor
while
off­
line
is
site­
specific
and
depends
on
the
water
quality.
At
a
minimum,
it
is
recommended
that
the
reactors
be
drained
if
the
UV
reactor
is
off­
line
for
more
than
one
week;
however,
the
appropriate
period
for
this
could
be
shorter
or
longer
depending
on
the
water
quality.
If
the
UV
reactor
will
be
off­
line
for
an
extended
period
of
time
(
longer
than
30
days),
it
is
recommended
that
the
reactor
be
cleaned
prior
to
re­
starting
the
UV
reactor.
Routine
shutdown
and
start­
up
procedures
are
discussed
in
section
5.2.3.

H.
3
Off­
Specification
Issues
UV
reactors
must
be
validated
as
discussed
in
Chapter
4
and
operated
within
the
conditions
determined
during
validation.
When
a
utility
is
operating
outside
of
the
validated
limits,
the
utility
is
operating
"
off­
specification."

LT2ESWTR
includes
requirements
limiting
off­
specification
for
compliance
with
the
LT2ESWTR
for
unfiltered
supplies
(
40
CFR
141.721(
c)(
2));
however,
the
rule
does
not
state
an
off­
specification
requirement
for
filtered
systems
or
ground
water
systems.
States
may
develop
statewide
or
site­
specific
requirements
off­
specification
requirements
for
ground
water
systems.

There
are
two
ways
that
a
ground
water
system
could
be
operating
off­
specification.
First,
off­
specification
can
occur
when
the
flow,
UVT,
or
UV
intensity
is
outside
of
the
validated
conditions.
Second,
UV
lamps
can
lose
arc
if
a
voltage
fluctuation,
power
quality
anomaly,
or
a
power
interruption
occurs.
LP
lamps
generally
can
return
to
full
operating
status
within
15
seconds
after
power
is
restored.
However,
LPHO
and
MP
UV
lamps
exhibit
restart
times
between
4
and
10
minutes
if
power
is
interrupted
(
Cotton
et
al.
2002).
During
these
restart
times,
the
water
being
distributed
is
inadequately
disinfected
and
is
considered
off­
specification.

H.
3.1
Power
Quality
Assessment
A
power
quality
assessment
at
each
well
site
should
be
performed
to
determine
if
power
quality
might
cause
off­
specification
operation.
In
addition,
the
reliability
of
commercial
power
at
remote
sites
may
be
less
than
that
of
more
populated
areas.
Backup
power
or
an
uninterruptible
power
supply
(
UPS)
may
be
needed,
depending
on
the
findings
of
the
power
Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
4
June
2003
quality
assessment.
If
backup
power
is
already
available
for
the
well
pumps,
then
the
backup
power
supply
should
be
assessed
to
determine
if
sufficient
output
is
available
for
the
UV
reactor.
However,
UPS
may
also
be
needed
if
there
are
frequent
power
quality
problems.
For
systems
that
have
storage
following
UV
disinfection,
it
may
be
possible
to
isolate
the
UV
reactor
and
rely
on
stored
water
to
meet
demand
during
periods
of
power
failure.
Power
quality
assessments
are
discussed
in
more
detail
in
section
3.1.3.3.

H.
3.2
Well
Pump
Cycling
Well
pumps
may
regularly
cycle
on
and
off
in
response
to
changes
in
distribution
system
pressure,
causing
the
UV
reactor
to
also
be
cycled.
Frequent
lamp
cycling
reduces
lamp
life.
Manufacturers
recommend
that
whenever
possible
lamps
remain
energized
for
a
minimum
of
6
hours
(
Dinkloh
2001).
In
addition,
the
warm­
up
time
when
the
UV
reactor
is
coming
on­
line
is
considered
off­
specification
until
the
UV
intensity
sensor
reading
reaches
the
validated
value
if
water
is
flowing
to
the
distribution
system.

Depending
on
its
current
operation
and
direction
from
the
State,
the
utility
may
need
to
consider
changing
the
well
pump
cycling
strategy
or
incorporate
UV
reactor
controls
to
reduce
off­
specification
time
and
to
meet
the
needs
of
the
distribution
system.
The
utility
should
discuss
its
proposed
operating
strategy
with
the
UV
manufacturer
to
ensure
it
is
appropriate
for
the
selected
UV
reactors.
While
there
may
be
any
number
of
operating
strategies
that
a
utility
could
use,
two
operational
strategies
that
could
be
incorporated
to
sequence
the
well
pumping
with
the
operation
of
the
UV
reactor
are
presented
below.

The
first
strategy
is
to
incorporate
a
delay
that
prevents
the
well
pump
from
starting
until
the
UV
reactor
reaches
its
validated
UV
intensity
sensor
setpoint
(
i.
e.,
no
flow
through
the
UV
reactor).
Under
this
control
strategy
there
will
be
a
period
when
the
UV
reactor
will
be
"
on"
but
no
flow
will
be
passing
through
it.
This
control
strategy
is
only
effective
when
LP
or
LPHO
reactors
are
used
because
their
lamps
can
operate
for
up
to
1
hour
under
no­
flow
conditions
(
Dinkloh
2001)
without
overheating.
However,
MP
UV
reactors
may
heat
the
water
above
the
safe
operating
temperature
of
50
degrees
Celsius
in
1
to
15
minutes,
causing
the
reactor's
internal
safety
devices
to
shut
the
reactor
off
(
Miller
2001).
As
such,
this
control
strategy
may
be
infeasible
for
MP
reactors
unless
they
incorporate
a
low
flow
waste
line
that
allows
water
to
circulate
through
the
reactor
in
order
for
MP
lamps
to
reach
the
validated
UV
intensity
sensor
setpoint
without
overheating.
The
UV
manufacturer
should
be
contacted
to
confirm
that
this
operational
strategy
is
feasible
with
or
without
the
waste
line.

The
second
strategy
is
to
provide
a
system
of
automated
valves
that
diverts
the
UV
reactor
discharge
away
from
the
distribution
system
until
the
reactor
reaches
its
validated
UV
intensity
sensor
setpoint.
Then
the
automated
valves
are
repositioned
to
direct
the
water
from
the
UV
reactor
to
the
distribution
system.
This
strategy
delivers
the
off­
specification
water
to
"
waste"
until
the
validated
UV
intensity
sensor
setpoint
is
reached.
This
ensures
that
sufficient
cooling
water
will
flow
through
MP
reactors
to
prevent
overheating
and
reactor
shutdown.
For
this
strategy,
the
utility
needs
to
develop
an
approach
for
managing
the
water
that
is
wasted
during
reactor
warm­
up.
The
water
may
be
wasted
to
a
sanitary
sewer,
storm
sewer,
on­
site
disposal
or
drainage
system,
or
temporary
storage
tank.
The
utility
should
coordinate
the
discharge
location
with
the
State
and
other
involved
parties.

Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
5
June
2003
Both
operational
strategies
introduce
a
lag
between
the
time
when
the
pump
is
initiated
and
the
time
when
water
is
introduced
into
the
distribution
system.
Because
of
this,
existing
controls
may
need
to
be
adjusted
to
avoid
insufficient
system
pressure
or
storage
during
periods
of
UV
reactor
warm­
up.
This
will
be
particularly
important
for
those
ground
water
systems
that
have
frequent
on­
off
cycles
or
limited
storage.

H.
4
Well
Location
Issues
Ground
water
production
wells
are
sited
in
a
variety
of
locations,
ranging
from
urban
areas
to
remote
installations.
The
well
location
will
affect
the
design
and
operation
of
the
UV
installation,
especially
if
there
is
limited
space.

H.
4.1
Design
Considerations
As
discussed
in
section
3.3.5.2,
the
UV
reactor
should
be
installed
within
a
building
or
underground
vault
if
possible
to
facilitate
maintenance
and
protect
sensitive
equipment.
The
need
for
enclosure
of
the
UV
installation
will
ultimately
be
based
on
the
manufacturer's
recommendations,
local
regulatory
and
code
requirements,
environmental
conditions,
and
sitespecific
constraints.
Site
security
should
be
appropriate
to
prevent
tampering
with
the
equipment
and
water
supply
and
to
protect
people
from
injury
(
e.
g.,
electrocution).

Well
sites,
particularly
in
urban
areas,
may
be
spatially
constrained
by
adjacent
development.
As
a
result,
the
amount
of
exposed
pipe
and
available
area
for
locating
equipment
may
be
limited.
In
these
cases,
it
may
be
necessary
to
modify
the
pump
discharge
piping
to
accommodate
a
UV
reactor.
When
constructing
a
UV
installation
in
an
extremely
confined
location,
the
designer
must
consider
the
area
necessary
for
operation
and
maintenance
and
the
area
needed
for
installation
(
e.
g.,
staging
areas,
personnel,
and
equipment
access).
In
addition,
the
inlet
and
outlet
piping
should
meet
the
criteria
listed
in
section
3.3.1.1
as
compared
to
the
validated
inlet
and
outlet
piping.

UV
reactors
are
susceptible
to
damage
by
suspended
sand
particles
or
other
debris
that
may
be
present
in
a
ground
water
supply
and
pass
through
the
well
screens.
Therefore,
it
is
important
to
determine
if
sand,
grit,
or
fines
are
present
in
a
well
supply
and
if
it
is
necessary
to
install
a
sand/
debris
trap
or
removal
equipment
prior
to
UV
disinfection.
Particles
flowing
through
the
UV
reactor
may
scratch
the
lamp
sleeves,
cause
the
sleeve
wiping
mechanisms
to
jam,
or
shield
pathogens
from
UV
light,
thereby
decreasing
the
UV
disinfection
effectiveness.
In
addition,
larger
particles
could
break
the
lamp
sleeves
and
lamps
(
see
Appendix
N
for
lamp
breakage
issues).

H.
4.2
Operational
Issues
Because
most
well
sites
are
not
continuously
staffed,
UV
installations
may
need
sufficient
automation
to
allow
remote
monitoring
and
operation.
Controls
and
alarms
should
be
designed
to
ensure
that
real­
time
operational
and
monitoring
data
are
communicated
to
the
Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
6
June
2003
operators.
These
factors
also
emphasize
the
importance
of
a
power
quality
assessment
and
the
design
of
alarms,
monitoring
capabilities,
and
backup
power
facilities.

Disposal
of
the
chemicals
used
to
clean
the
UV
reactors
may
be
an
issue
if
an
on­
line
chemical
cleaning
(
OCC)
system
is
used
(
section
2.4.5).
If
sewer
connections
or
other
standard
means
of
disposal
are
not
available,
then
chemical
waste
will
need
to
be
transported
off­
site
for
disposal
or
handled
on­
site.
Utilities
should
consult
with
chemical
suppliers
and
the
State
when
developing
disposal
strategies.

H.
5
Hydraulic
Issues
The
hydraulic
issues
associated
with
ground
water
systems
include
high
operating
pressures,
piping
configuration,
air
entrainment,
and
the
potential
of
water
hammer
and
surge
events.

Many
well
pumps
discharge
directly
to
the
distribution
system
or
to
elevated
or
pressurized
storage;
therefore,
the
discharge
will
often
be
at
system
pressure.
The
UV
reactor
design
may
need
to
be
modified
to
accommodate
these
higher
distribution
system
pressures.

The
actual
inlet
and
outlet
hydraulics
of
the
UV
reactor
should
be
designed
to
match
the
validated
hydraulics
as
discussed
in
section
3.3.1.
Space
is
often
limited
with
ground
water
installations
so
valves,
flow
meters,
or
other
appurtenances
may
be
directly
upstream
or
downstream
of
the
UV
reactor.
Consequently,
these
site
constraints
may
need
to
be
considered
in
determining
how
the
UV
reactor
should
be
validated.
Detailed
discussions
of
UV
installation
layout
and
validation
are
given
in
section
3.3.5
and
Chapter
4,
respectively.

UV
reactors
should
be
flooded
at
all
times
because
air
binding
can
interfere
with
the
UV
disinfection
process
or
cause
the
lamps
to
overheat.
UV
reactors
should
be
located
downstream
of
any
existing
or
planned
air
removal
equipment
(
if
necessary).
Otherwise,
the
UV
installation
design
should
include
a
means
for
automatically
releasing
air
prior
to
the
UV
reactor.
The
UV
reactor
may
have
integral
air
release
valves
or
valve
ports,
or
air
release
valves
can
be
installed
in
the
inlet
and
outlet
piping.

Pressure
surge
events
(
water
hammer)
near
the
UV
reactor
may
be
more
likely
with
ground
water
systems
than
surface
water
systems
because
of
the
UV
reactor's
proximity
to
the
well
pumps.
Surge
events
can
cause
positive
or
negative
pressure
transients
in
the
well
discharge
piping.
Negative
pressures
as
small
as
­
1.5
psi
may
cause
the
lamp
sleeves
to
break
(
Dinkloh
2001).
A
surge
analysis
is
recommended
to
determine
if
surge
protection
is
necessary.
Many
well
sites
and
distribution
systems
are
already
equipped
with
surge
control
tanks
to
dampen
surge
effects.
These
tanks
may
provide
sufficient
protection
for
the
UV
reactors,
depending
on
their
location
relative
to
the
UV
reactors.

Other
surge
control
devices,
such
as
air/
vacuum
release
valves,
may
rely
on
the
introduction
of
air
into
the
system
to
mitigate
surge.
As
discussed
previously,
the
presence
of
air
can
negatively
affect
the
performance
of
the
UV
reactors.
Air/
vacuum
valves
should
only
be
used
if
surge
tanks
are
not
an
option
and
the
design
can
eliminate
the
air
prior
to
the
UV
reactor
(
e.
g.,
install
the
valve
in
a
section
of
pipe
at
a
higher
elevation
than
the
UV
reactor).

Proposal
Draft
Appendix
H.
Issues
for
Ground
Water
Systems
UV
Disinfection
Guidance
Manual
H­
7
June
2003
H.
6
References
Cotton,
C.
A.,
R.
S.
Cushing,
and
D.
M.
Owen.
2002.
The
impact
of
the
draft
UV
disinfection
requirements
on
UV
facility
design
and
operation.
Proceedings
of
the
AWWA
Annual
Conference,
June
16­
20,
New
Orleans,
LA.

Dinkloh,
L.
2001.
Wedeco­
Ideal
Horizons.
Telephone
conversations
and
email
correspondence
by
Ben
Hauck,
Malcolm
Pirnie
Inc.,
regarding
UV
reactors.
April
25
and
August
10.

Malley,
J.
P.,
B.
A.
Petri,
G.
L.
Hunter,
D.
Moran,
M.
Nadeau,
and
J.
Leach.
2001.
Full­
scale
implementation
of
UV
in
groundwater
disinfection
systems.
Denver,
CO:
AwwaRF
Final
Report.

Miller,
A.
2001.
Trojan
Technologies.
Telephone
conversation
by
Ben
Hauck,
Malcolm
Pirnie
Inc.,
regarding
UV
reactors.
August
10.

Toivanen,
E.
2000.
Experiences
with
UV
disinfection
at
Helsinki
water.
IUVA
News
2,
no.
6:
4­
8.

Proposal
Draft
Appendix
I.
Issues
for
Small
Systems
The
objectives
of
this
appendix
are
to
highlight
the
issues
that
small
systems
face
when
considering
UV
disinfection
and
to
reference
the
more
detailed
discussion
of
these
issues
in
this
manual.

To
be
classified
as
a
public
water
system,
a
utility
must
provide
water
to
a
minimum
of
15
service
connections
or
serve
at
least
25
people
for
at
least
60
days
per
year
(
40
CFR
141.2).
For
the
purpose
of
this
appendix,
the
term
small
system
includes
those
utilities
serving
fewer
than
10,000
people
or
having
a
daily
production
rate
of
less
than
approximately
1.0
mgd.
Most
of
the
information
regarding
UV
disinfection
in
Chapters
2,
3,
4,
and
5
is
valid
for
both
small
and
large
systems.

I.
1
Is
UV
Disinfection
Applicable
to
Small
Systems?

UV
disinfection
is
applicable
to
small
systems
and
may
be
attractive
for
the
following
reasons:

 
It
is
a
relatively
low
cost
technology
for
the
inactivation
of
Cryptosporidium
(
Cotton
et
al.
2001).

 
Chemical
use
is
little
to
none.

 
Operation
is
relatively
simple
and
maintenance
is
low.

 
Space
needs
are
small.

Two
types
of
UV
reactors
can
potentially
be
used
by
small
systems,
conventional
and
point­
of­
entry
(
POE)
devices.
Conventional
UV
reactors
are
manufactured
for
a
wide
range
of
flows
(
e.
g.,
from
20
gallons
per
minute
(
gpm)
up
to
40
mgd)
and
are
described
in
section
2.4.
POE
units
are
small
UV
reactors
that
are
installed
at
the
service
connection
of
the
customer.
POE
units
contain
the
same
components
as
conventional
low­
pressure
(
LP)
installations
but
are
more
compact.
They
are
primarily
intended
for
use
at
individual
properties
and
may
be
more
suitable
for
utilities
with
a
limited
number
of
service
connections.
POE
units
are
required
to
be
owned,
controlled,
and
maintained
by
the
utility
or
by
a
person
under
contract
with
the
utility
to
facilitate
proper
operation
and
maintenance
and
compliance
with
the
treatment
requirements
(
Safe
Drinking
Water
Act
(
SDWA)
Section
1412(
b)(
4)(
E)).
The
use
of
POE
units
may
result
in
higher
total
costs
when
compared
to
a
centralized,
conventional
UV
installation
for
all
but
the
smallest
water
utilities.

I.
2
What
Information
is
Necessary
to
Assess
the
Feasibility
of
UV
Disinfection?

Small
systems
generally
need
the
same
information
to
assess
UV
disinfection
as
larger
systems.
Chapter
3
describes
the
planning
and
design
process
for
a
UV
installation
in
a
UV
Disinfection
Guidance
Manual
I­
1
June
2003
Proposal
Draft
Appendix
I.
Issues
for
Small
Systems
UV
Disinfection
Guidance
Manual
I­
2
June
2003
conventional
plant
and
discusses
each
of
the
elements
that
should
be
considered.
In
general,
the
utility
should
answer
the
following
questions
when
assessing
the
suitability
of
UV
disinfection:

 
What
are
the
disinfection
goals
and
can
UV
disinfection
be
used
to
meet
these
goals?
(
section
3.1.1)

 
What
are
the
minimum,
average,
and
maximum
flowrates
that
the
UV
reactors
will
need
to
treat?
(
section
3.1.3.2)

 
What
is
the
design
UV
absorbance
at
254
nm
(
A254)
and
corresponding
design
UV
transmittance
(
UVT)?
What
is
the
fouling
potential
of
the
water
supply?
What
is
the
potential
for
process
upset
or
variability
in
water
quality?
Do
any
of
the
existing
processes
have
the
potential
to
interfere
with
the
performance
of
the
UV
reactors?
(
section
3.1.3.1)

 
Can
the
UV
reactors
be
incorporated
into
the
existing
hydraulic
profile?
If
not,
can
the
existing
operations
be
modified
to
accommodate
the
UV
reactors,
or
does
intermediate
pumping
need
to
be
installed?
(
section
3.1.6.1)

 
Can
the
UV
installation
be
incorporated
into
the
existing
facility
layout?
Does
a
building
need
to
be
constructed
to
house
the
UV
reactors?
(
section
3.1.6.2)

 
Is
the
quality
and
reliability
of
the
electrical
power
supply
adequate?
Does
a
backup
power
supply
or
other
supplemental
electrical
equipment
need
to
be
installed?
(
sections
3.1.3.3
and
3.3.4)

 
How
should
the
UV
reactors
be
controlled?
What
level
of
automation
and
operational
complexity
is
appropriate?
Does
the
potential
for
power
savings
justify
using
a
more
complex
operating
strategy?
Is
the
existing
operations
staff
sufficient?
(
section
3.3.3)

 
Is
the
number
of
UV
reactors
installed
appropriate
to
efficiently
respond
to
the
anticipated
range
of
flowrates?
Does
the
UV
installation
have
the
capability
to
be
expanded
to
meet
future
increases
in
demand?
Is
there
sufficient
redundancy
to
allow
operating
flexibility
and
to
meet
the
disinfection
goals
under
the
operating
scenarios?
(
section
3.1.3.2)

 
How
should
the
UV
reactors
be
procured?
(
section
3.2)

 
Do
the
characteristics
of
the
proposed
UV
application
(
e.
g.,
flowrate,
UVT,
UV
intensity)
differ
from
those
under
which
a
selected
UV
reactor
was
validated?
If
so,
should
the
selected
equipment
be
validated
on­
site
or
off­
site
under
characteristics
that
match
those
of
the
intended
installation?
(
Chapter
4
and
section
3.1.4.2)

 
What
is
the
capital
cost
of
the
UV
installation?
What
are
the
operating
costs
associated
with
a
UV
installation?
(
section
3.1.7)

 
What
is
the
cost
of
the
UV
installation
as
compared
to
other
disinfection
alternatives?

Proposal
Draft
Appendix
I.
Issues
for
Small
Systems
UV
Disinfection
Guidance
Manual
I­
3
June
2003
 
Is
there
a
cost
benefit
to
using
POE
units
as
opposed
to
a
centralized
UV
installation?
If
so,
how
should
the
utility
administer
the
POE
units?
Is
some
form
of
access
agreement
or
water
use
ordinance
necessary
to
allow
administration
of
the
POE
units?

I.
3
Do
the
UV
Reactors
Need
to
be
Housed
in
a
Building?

If
possible,
the
UV
reactors
should
be
constructed
within
a
building
to
facilitate
maintenance
and
protect
the
UV
reactors.
Nonetheless,
the
need
for
enclosing
the
UV
reactors
will
ultimately
be
based
on
manufacturers'
recommendations,
State
requirements,
and
environmental
conditions.
Although
some
current
UV
installations
do
not
have
a
building
(
e.
g.,
Hanovia
facility
in
Australia),
local
building
and
electrical
codes
may
necessitate
a
building
or
other
protection
for
the
electrical
equipment.
Regardless
of
whether
the
UV
reactors
are
enclosed,
site
security
is
important
to
prevent
tampering
with
the
equipment
and
water
supply
and
to
protect
people
from
injury
(
e.
g.,
electrocution).
Section
3.3.4
discusses
the
electrical
equipment
issues
that
should
be
considered
during
the
planning
and
design
of
a
UV
installation.

I.
4
Do
the
Components
of
a
Small
System
Differ
from
Larger
UV
Reactors?

The
main
components
of
a
UV
reactor
(
including
the
necessary
instrumentation
and
controls)
do
not
differ
between
large
and
small
systems.
Components
of
the
UV
reactor
may
include
the
UV
lamps,
lamp
sleeves,
UV
intensity
sensors,
ballasts,
and
cleaning
mechanisms,
which
are
described
in
section
2.4.

Full­
scale
drinking
water
applications
generally
use
LP,
low­
pressure
high­
output
(
LPHO),
or
medium­
pressure
(
MP)
lamps.
Small
systems
may
find
reactors
that
use
LP
or
LPHO
lamps
more
economical
because
they
convert
power
into
germicidal
wavelengths
of
UV
light
more
efficiently
than
MP
lamps.
Additionally,
LP
lamps
typically
have
a
longer
useful
life
than
either
MP
or
LPHO
lamps.
For
small
systems,
UV
reactors
with
LP
lamps
are
likely
to
represent
the
most
cost­
effective
disinfection
solution.
For
systems
that
serve
near
10,000
people
or
treat
near
1
mgd,
more
consideration
should
be
given
to
LPHO
or
MP
lamps.
An
additional
discussion
of
the
different
lamp
types
is
given
in
section
2.4.2.

I.
5
What
are
the
Power
Needs?

The
power
needs
depend
on
the
manufacturer.
Common
manufacturers'
power
needs
are
as
follows:

 
POE
UV
units
 
120V/
60Hz/
1
phase
 
Conventional
LP
reactors
 
120/
208V/
60Hz/
3
phase
 
Conventional
LPHO
reactors
 
480V/
60Hz/
3
phase
 
Conventional
MP
reactors
 
480V/
60Hz/
3
phase
Proposal
Draft
Appendix
I.
Issues
for
Small
Systems
UV
Disinfection
Guidance
Manual
I­
4
June
2003
Backup
power
may
be
necessary,
depending
on
the
type
of
installation
that
is
selected,
the
power
quality
at
the
installation
site,
and
the
regulatory
requirements
for
the
installation.
However,
backup
power
for
small
systems
may
not
be
necessary
as
some
small
systems
can
accommodate
a
shutdown
for
longer
periods
because
there
is
sufficient
storage
to
meet
demand.
Additional
detail
on
the
need
for
backup
power
and
the
factors
that
should
be
considered
when
assessing
the
power
supply
are
discussed
in
section
3.1.3.3.

I.
6
Do
Small
UV
Reactors
Need
to
be
Validated?

All
UV
reactors,
including
POE
units,
are
required
to
be
validated
(
40
CFR
141.729(
d)).
Small
systems
will
probably
purchase
UV
reactors
that
have
been
validated
by
the
manufacturer.
UV
intensity
sensor
operating
setpoints
(
and
potentially
UVT
setpoints)
are
established
at
specific
flowrates
during
validation
testing.
These
are
the
setpoints
that
the
systems
are
required
to
operate
within
to
receive
inactivation
credit.
For
many
small
UV
reactors
and
nearly
all
POE
units,
UV
reactor
control
will
be
limited
to
"
on"
and
"
off"
with
UV
reactor
shutdown
under
specific
critical
alarm
conditions.
Chapter
4
discusses
the
UV
reactor
validation
requirements,
and
Chapter
5
describes
operating
requirements.

I.
7
How
are
UV
Reactors
Monitored?

Monitoring
UV
reactors
(
conventional
and
POE)
is
required
to
ensure
that
the
UV
reactors
are
operating
within
the
validated
range
(
40
CFR
141.729(
d)).
Parameters
that
must
be
monitored
include
flowrate,
UV
intensity
sensor
readings,
and
UVT
(
if
it
is
part
of
the
control
strategy)
(
40
CFR
141.729(
d)).
POE
units
should
be
equipped
with
mechanical
warnings
to
ensure
that
customers
are
automatically
notified
of
operational
problems.
Additional
detail
on
monitoring
requirements
is
provided
in
section
5.4.1.

I.
8
Can
the
UV
Reactors
be
Operated
Remotely?

UV
reactors
can
be
operated
remotely
if
the
monitoring
components
provide
a
4­
20
mA
analog
output
signal
and
are
integrated
into
a
control
strategy.
Even
though
UV
reactors
can
be
operated
remotely,
routine
inspections
and
on­
site
maintenance
will
be
necessary
to
confirm
that
the
UV
reactor
is
operating
properly.
Provisions
for
hydraulic
and
electrical
lockout
should
be
provided
to
enable
local
isolation
and
lockout
for
maintenance.
Section
3.3.3
provides
additional
detail
on
control
strategies
for
centralized
UV
installations,
and
section
5.3
discusses
operations
and
maintenance
needs.

If
the
utility
uses
POE
units,
it
may
be
beneficial
to
telemeter
all
alarm
conditions
to
a
central
location
to
facilitate
administration
and
maintenance
of
the
POE
units.
However,
incorporating
this
remote
capability
will
likely
increase
the
cost
of
the
UV
installation.

Proposal
Draft
Appendix
I.
Issues
for
Small
Systems
UV
Disinfection
Guidance
Manual
I­
5
June
2003
I.
9
How
Much
Maintenance
is
Needed?

Maintenance
is
generally
limited
but
will
vary
depending
on
the
manufacturer
and
the
specific
application.
Maintenance
may
include
the
following
activities:

 
Periodic
calibration
verification
of
UV
intensity
sensors,
UVT
meters,
or
flowmeters
 
Periodic
replacement
of
UV
intensity
sensors,
UVT
meters,
or
flowmeters
(
if
applicable),
depending
on
calibration
or
age
of
the
equipment
 
Lamp
sleeve
and
reactor
cleaning
 
Replacement
of
UV
lamps
and
other
components
 
Maintenance
of
other
operating
components
and
the
electrical
systems
Operators
should
be
trained
by
the
UV
manufacturer
on
the
proper
operation
and
maintenance
of
the
UV
reactors.
The
utility
should
consider
contracting
trained
service
personnel
to
maintain
the
UV
reactors
if
this
is
not
possible.
Additional
detail
on
operations
and
maintenance
is
given
in
section
5.3.

I.
10
References
Cotton,
C.
A.,
D.
M.
Owen,
G.
C.
Cline,
and
T.
P.
Brodeur.
2001.
UV
disinfection
costs
for
inactivating
Cryptosporidium.
Journal
of
the
American
Water
Works
Association
93,
no.
2:
67­
74.

Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
In
some
cases,
pilot­
or
demonstration­
scale
testing
may
be
warranted
to
aid
in
selection
of
design
criteria.
For
example,
long­
term
UV
unit
performance
will
be
impacted
by
lamp
aging
and
sleeve
fouling.
With
increased
use,
lamp
output
decreases
due
to
deposition
of
inorganic
material
on
the
outside
and
inside
of
the
sleeve
(
i.
e.,
"
fouling").
Fouling
reduces
the
transmittance
of
the
lamp
energy
to
the
water.
Over
time,
these
phenomena
will
contribute
to
a
reduction
in
UV
dose.
The
effect
of
these
parameters
should
be
incorporated
into
the
UV
reactor
design.
A
"
lamp
aging
factor"
and
a
"
fouling
factor"
are
usually
specified
by
the
design
engineer.
Pilot
testing
can
provide
useful
information
for
the
development
of
these
factors.

This
appendix
discusses
when
pilot
or
demonstration
tests
may
be
needed
and
the
types
of
tests
that
may
be
performed
on
UV
disinfection
systems.
The
purpose(
s)
of
pilot
and
demonstration
testing
is
to
establish
or
confirm
system
design
factors,
test
system
reliability,
and
evaluate
operation
and
maintenance
(
O&
M)
needs.
The
tests
described
herein
may
be
performed
individually
or
in
parallel.
Validation
of
reactor
microbial
inactivation
performance
is
addressed
separately
in
Appendix
C
(
Validation
Protocol).

J.
1
When
Is
Pilot
or
Demonstration
Testing
Needed?

Pilot
and
demonstration
tests
can
be
used
to
meet
the
following
three
goals:

1.
Assess
the
impact
of
unusual
water
quality
conditions
(
e.
g.,
high
calcium
or
iron
concentrations).

2.
Improve
estimation
of
safety
factors
for
large
water
systems
for
which
such
an
investment
can
yield
a
high
return
in
reduced
life
cycle
costs.

3.
Gain
first­
hand
experience
with
operating
and
maintaining
a
UV
installation.

A
UV
disinfection
system
should
be
designed
with
some
knowledge
of
the
likely
fouling
potential
of
the
water
and
lamp­
aging
characteristics
to
ensure
the
system
operates
as
intended.
If
the
design
and
the
operation
protocol
do
not
properly
account
for
the
effects
of
lamp
aging
and
sleeve
fouling,
the
system
may
go
into
alarm
frequently
(
indicating
under
dosing).

While
pilot
or
demonstration
testing
may
be
warranted
in
some
cases,
it
is
becoming
less
necessary
as
more
performance
and
fouling
information
is
developed.
The
need
for
pilot
or
demonstration
testing
should
be
carefully
considered
in
light
of
the
pre­
existing
data
available
on
both
system
performance
and
water
quality
effects
on
sleeve
fouling.
Pilot
or
demonstration
testing
may
be
used
to
gain
operational
experience
or
primacy
agency
acceptance,
as
discussed
in
the
following
sections.

Microbiological
challenge
tests
are
not
recommended
during
pilot
studies
because
inactivation
efficiency
in
a
pilot
system
may
not
be
indicative
of
full­
scale
performance.
However,
UV
reactor
validation
bioassays
could
be
conducted
as
part
of
a
full­
scale
UV
Disinfection
Guidance
Manual
J­
1
June
2003
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
2
June
2003
demonstration
test
if
on­
site
testing
is
planned.
Appendix
C
presents
a
detailed
discussion
of
UV
reactor
performance
validation.

Because
UV
disinfection
is
a
relatively
new
drinking
water
treatment
technology
in
the
United
States,
State
regulatory
agency
acceptance
may
depend
in
part
on
the
confidence
in
the
technology
gained
through
pilot­
and
demonstration­
scale
studies.
Identifying
previous
studies
of
similar
scope
that
provide
background
and
precedents
may
also
be
helpful
in
gaining
acceptance
of
its
planned
use
(
see,
for
example,
et
al.
Mackey
2001).

If
a
utility
chooses
to
or
is
required
to
conduct
pilot
or
demonstration
tests,
the
primacy
agency
should
understand
the
objective(
s)
of
the
test(
s)
and
the
methodologies
used.
It
is
recommended
that
the
primacy
agency
be
contacted
before
testing
and
involved
throughout
the
pilot
and/
or
demonstration
testing.
Identifying
and
resolving
State
regulatory
agency
concerns
when
planning
testing
can
help
produce
a
more
useful
dataset.
Additionally,
it
may
be
helpful
to
include
the
State
in
interim
briefings
on
progress
and
results,
and
to
give
them
a
final
report
after
completing
the
testing.

J.
1.1
Water
Quality
Impacts
Extensive
data
have
been
generated
from
pilot­
scale
testing
on
waters
of
low
to
moderate
hardness
and
iron
content
(
Mackey
et
al.
2001,
Mackey
and
Cushing
2003).
At
total
hardness
and
calcium
levels
below
140
mg/
L
and
low
iron
(
less
than
0.1
mg/
L),
standard
cleaning
protocols
and
wiper
frequencies
(
one
sweep
every
15
minutes
to
an
hour)
were
more
than
adequate
to
deal
with
the
impact
of
sleeve
fouling
at
all
sites
tested.
At
sites
with
hardness
or
iron
in
the
feed
water
that
exceed
these
levels,
it
may
be
advantageous
to
evaluate
fouling
rates
on
a
site­
specific
or
worst
case
basis
via
pilot
or
demonstration
testing
to
identify
how
best
to
keep
the
sleeves
clean.

J.
1.2
Lamp
Fouling
Factors
for
Large
Systems
In
UV
reactor
design,
a
lamp
aging
factor
of
0.7
is
commonly
used,
as
discussed
in
section
3.1.3.1
of
the
Manual.
For
larger
systems,
it
may
be
economical
to
pilot
or
demonstration
test
lamp
aging
to
provide
data
for
selecting
lamp
aging
and
low­
dose­
alarm
design
factors
that
will
best
balance
operational
costs
(
how
many
hours
one
wants
to
be
able
to
operate
a
lamp
before
replacing
it)
with
capital
costs
(
the
size
of
the
system
needed
based
on
end­
of­
lamp­
life).
Lamp
aging
factors
may
also
be
obtained
from
a
certified
lamp
age
testing
program
performed
by
equipment
or
lamp
manufacturers.
A
lower
lamp
aging
factor
means
the
utility
will
have
less
frequent
lamp
replacements,
but
may
require
a
larger
system
to
ensure
compliance
at
all
times.

J.
1.3
Gaining
Operational
Experience
Due
to
the
small
number
of
U.
S.
drinking
water
UV
installations,
very
few
United
States
operators
have
experience
with
UV
disinfection
systems.
It
may
be
beneficial
for
a
facility's
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
3
June
2003
staff
to
obtain
operational
experience
with
UV
disinfection
systems
prior
to
selecting
and
implementing
UV
disinfection.
If
a
utility
staff
becomes
familiar
with
the
operational
aspects
of
UV
disinfection,
that
staff
will
be
able
to
provide
feedback
input
on
the
UV
installation
design.
In
addition,
operational
experience
can
help
facility
managers
to
determine
the
staffing/
training
needs
and
help
maintenance
staff
understand
and
plan
for
the
maintenance
needs
of
the
system
(
e.
g.,
time
to
change
lamps
and
calibrate
UV
intensity
sensors
and
perform
manual
cleaning).

On­
site
testing
is
site­
specific
depends
on
the
needs
and
preferences
of
the
utility.
Methods
by
which
facility
staff
can
gain
operational
experience
(
besides
on­
site
testing)
include:
site
visits
and
partnerships
with
systems
already
using
UV
disinfection;
conversations
and
visits
with
manufacturers
and
attendance
of
seminars;
and
on­
site
training
programs
(
a
detailed
discussion
of
training
programs
is
provided
in
section
5.7.2).

J.
2
Pilot­
Versus
Demonstration­
Scale
Testing
Table
J.
1
presents
a
comparison
of
the
advantages
and
disadvantages
associated
with
pilot­
scale
and
demonstration­
scale
testing.
Pilot­
scale
testing
involves
operating
a
smaller
version
of
a
full­
sized
UV
disinfection
reactor.
It
may
or
may
not
include
all
the
components
of
the
full­
sized
system.
Demonstration­
scale
testing
is
essentially
pilot
testing
of
a
full­
scale
UV
disinfection
reactor.

Table
J.
1
Comparison
of
Pilot
and
Demonstration
Testing
Testing
Method
Advantages
Disadvantages
Pilot­
scale
 
Smaller
footprint
needed
for
UV
reactors
 
Less­
expensive
installation
and
operation
 
Operators
gain
O&
M
experience
 
High
flexibility
in
placement
of
equipment
 
Lesser
volumes
of
water
to
dispose
 
Design
conditions
for
UV
disinfection
systems
may
not
scaleup
to
full­
scale
systems
 
In
rare
cases
it
may
be
advisable
to
use
pilot­
scale
treatment
process
equipment
(
filters,
clarifiers,
etc.)
to
simulate
operational
conditions
(
e.
g.,
upstream
ozone
process)

Demonstration­
Scale
 
Confidence
in
long­
term
operation
of
the
UV
unit
due
to
the
representative
scale
at
which
results
are
obtained
 
Scale­
up
factors
need
not
be
developed
 
Operators
gain
operations
and
maintenance
experience
on
a
fullscale
system.
 
Approval
from
the
primacy
agency
may
be
required
to
conduct
a
demonstration
study
 
Demonstration
setups
are
not
as
flexible
as
pilot
studies
for
operational
experimentation
 
Installation
and
operation
more
expensive
than
pilot
scale
 
Greater
volumes
of
water
to
dispose.

Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
4
June
2003
J.
3
Documenting
the
Test
Reactor
For
a
given
test,
should
the
properties
of
the
components
that
may
influence
the
final
outcome
of
the
test
should
be
identified
and
recorded.
That
record
may
later
be
used
to
confirm
that
key
components
of
installed
UV
reactors
match
those
of
the
systems
tested.
Table
J.
2
lists
the
components
of
a
UV
disinfection
system
that
should
be
documented
and
compared
between
testing
and
the
final
design.

Table
J.
2
Key
Components
Associated
with
UV
Reactor
Pilot­
Scale
and
Demonstration
Scale
Testing
Test
Components
to
Document
Operational
Experience
Controls,
alarms,
cleaning
mechanisms,
operation,
maintenance.
Fouling
Assessment
Lamps,
sleeves,
ballasts,
power
settings,
UV
intensity
sensor
windows,
flow
velocity.
Head
loss
Assessment
(
demonstration­
scale
only)
Reactor
and
wetted
components,
inlet/
outlet
conditions.
Ballast
Performance
Lamps,
sleeves,
ballast,
power
settings,
operation.
Cleaning
Mechanism
Performance
Lamps,
sleeves,
ballasts,
power
settings,
ballast
operation,
UV
intensity
sensor
windows
(
if
wiper
used),
cleaning
mechanism,
cleaning
solutions,
wiper
maintenance
and
operation.
Lamp
Aging/
Failure
Lamps,
sleeves,
ballasts,
power
settings,
ballast
operation,
cleaning
mechanism,
cleaning
solutions,
wiper
maintenance
and
operation.
Sleeve
Breakage
Sleeves,
cleaning
mechanisms,
flow
velocity,
water
hammer.
Controls/
Alarms
Lamps,
sleeves,
ballasts,
UV
intensity
sensors,
cleaning
mechanisms,
controls,
operation.

J.
4
Testing
Objectives
Pilot/
demonstration
testing
may
be
used
to
gain
information
on
a
specific
UV
reactor,
a
specific
water
treatment
plant
(
WTP)
site,
or
a
combination
of
the
two.
Common
test
objectives
include
the
following
topics:

 
The
long­
term
performance
and
failure
modes
of
the
lamps
 
The
efficacy
of
cleaning
mechanisms
for
lamp
sleeves
and
UV
intensity
sensor
windows
 
The
stability
of
UV
intensity
and
UVT
monitors
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
5
June
2003
 
The
reliability
of
controls
and
alarm
systems
 
The
ease
of
lamp
and
UV
intensity
sensor
replacement,
the
use
of
reference
sensors,
and
the
maintenance
of
cleaning
devices
and
solutions
 
The
rate
of
fouling
on
lamp
sleeves
and
UV
intensity
sensor
windows
 
The
most
appropriate
cleaning
method
 
The
head
loss
across
the
reactor
at
various
flow
rates
(
demonstration­
scale
only)

 
The
impact
on
other
unit
operations
at
the
WTP
The
information
obtained
during
pilot
and
demonstration
testing
should
be
applicable
to
the
final
UV
disinfection
system
installed
at
the
WTP.
Accordingly,
the
equipment
tested
should
be
representative
of
the
UV
disinfection
system
that
will
be
installed.
Specific
elements
of
a
pilot/
demo­
scale
system
that
should
be
identical
include
the
UV
intensity
sensors,
lamp
and
sleeve
type,
power
system,
cleaning
system,
cleaning
frequency,
and
water
quality.
For
example,
lamp­
aging
data
on
a
3
kW
25
cm
medium
pressure
(
MP)
lamp
driven
by
an
electromagnetic
ballast
cannot
be
used
to
predict
the
aging
expected
with
a
10
kW
50
cm
MP
lamp
driven
by
a
transformer.

J.
5
Testing
Protocols
This
section
describes
the
major
elements
and
benefits
of
a
range
of
pilot
and
demonstration
testing
protocols
to
investigate
sleeve
fouling
and
cleaning,
lamp
aging,
head
loss
and
alarms
and
controls.

J.
5.1
Assessing
Fouling
A
fouling
assessment
can
be
conducted
to
answer
the
following
questions:

 
How
fast
do
the
lamps
foul?

 
How
does
water
quality
affect
fouling?

 
What
lamp
fouling
factor
should
be
specified?

W
 
hat
lamp
cleaning
interval
is
required?

W
 
hat
sleeve
replacement
interval
is
required?

H
 
ow
do
lamp/
reactor
configurations
affect
fouling?

Is
 
fouling
of
the
UV
intensity
sensor
window(
s)
significant
and
how
should
it
be
addressed?

Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
6
June
2003
Fouling
may
occur
on
the
inner
and
outer
surfaces
of
the
lamp
sleeves,
the
internal
surfaces
of
the
reactor,
and
UV
intensity
sensor
windows.
Lamp
sleeve
fouling
may
have
an
impact
d
to
on
of
d
for
assessing
sleeve
fouling
similar
to
that
employed
by
Lin
et
al.
(
1999b),
new
lamp
can
be
placed
inside
the
fouled
sleeve
and
ignited.
Ultraviolet
absorbance
at
254
nanom
chemical
cleaning
should
restore
the
sleeve
A254
to
very
near
that
of
a
new,
clean
leeve.
If
not,
manually
clean
the
inside
of
the
sleeve
and
measure
A254.
If
it
is
still
low,
the
sleeve
s
a
fouling,
care
should
be
taken
to
ensure
that
the
results
scalep
to
full­
scale
applications.
Some
differences
in
system
geometry
may
lead
to
erroneous
conclus
leeve
f
e
UV
intensity
sensor
windows,
clean
the
sensor
monitoring
indows
with
phosphoric
or
citric
acid
at
varying
time
intervals
and
record
the
change
in
sensor
reading
.5.2
Evaluating
Cleaning
Systems
ethods
can
be
performed
to
answer
the
following
uestions:

Does
a
particular
cleaning
protocol
work
for
the
UV
reactor
application?
on
dose
delivery
and
cleaning
requirements.
Sensor
window
fouling
may
have
an
important
impact
on
assessing
dose
delivery
(
e.
g.,
the
sensor
will
not
be
able
to
accurately
measure
lamp
intensity).
Fouling
on
the
wetted
surfaces
of
a
UV
reactor
has
been
attribute
precipitation
of
compounds
whose
solubility
decreases
as
temperature
increases,
precipitati
compounds
with
low
solubility,
and
deposition
of
particles
by
gravity
settling
and
turbulenceinduced
impaction
(
Lin
et
al.
1999a).
More
detailed
discussion
on
fouling
is
provided
in
sections
3.1.3.1
and
A.
4.1.4.

In
one
metho
a
eters
(
A254)
is
measured
by
a
calibrated
radiometer
and
compared
to
a
similar
measurement
made
using
a
new,
clean
sleeve.
The
ratio
of
these
two
measurements
(
UV
light
passing
through
the
fouled
sleeve
to
that
passing
through
the
new
sleeve)
is
the
lamp
sleevefouling
factor.

Manual,
s
material
has
likely
degraded.
If
A254
cannot
be
recovered,
further
testing
may
be
used
to
identify
a
proper
sleeve
replacement
interval.
The
lamp
sleeve­
fouling
factor
can
be
plotted
a
function
of
time.
Worst­
case
results
can
be
analyzed
to
determine
cleaning
requirements
and
fouling
factors
for
design
purposes.

When
assessing
lamp
sleeve
u
ions
based
on
pilot
data
alone.
For
instance,
in
parallel
flow
reactors,
fouling
has
been
found
to
be
uneven
along
the
length
of
the
lamps
(
Lin
et
al.
1999a).
If
the
lamp
and
lamp
s
geometry
(
e.
g.,
length
or
diameter)
of
the
pilot
unit
is
very
different
from
the
full­
scale
system,
the
fouling
that
will
occur
in
the
full­
scale
plant
may
be
markedly
different
from
expectations
based
on
pilot­
scale
data.
The
lamp
lengths
will
be
very
different
and
end­
effects
may
be
more
pronounced
(
i.
e.,
the
blackened
lamp
ends
of
an
aged
lamp
will
comprise
a
greater
percentage
o
the
total
length
of
the
lamp).

To
assess
fouling
on
th
w
s.
It
is
expected
that
the
rate
of
fouling
on
the
lamps
will
be
greater
than
the
rate
of
fouling
on
the
sensor
windows
due
to
elevated
lamp
temperature.

J
An
evaluation
of
system
cleaning
m
q
 

Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
7
June
2003
 
What
is
the
long
term
effectiveness
of
the
cleaning
method?

 
What
cleaning
frequency
is
required
for
each
method
considered?

nts
that
accumu
te
on
the
lamp
sleeves
and
UV
intensity
sensor
windows.
Lamp
sleeve
cleaning
method
ning),
ipers
g
.5.2.1
Assessing
Lamp
Sleeve
Cleaning
Protocols
mp
sleeves,
and
the
sults
b
lowest
r
at
the
minimum
flow
rate
and
operate
the
lamps
at
maxim
ith
systems
using
mechanical
or
physicochemical
wipers,
an
unwiped
sleeve
se
this
ata
to
optimize
the
cleaning
frequency.
Sensor
windows
should
be
manually
cleaned
before
measur
heduled
sleeve
cleaning
cycle,
remove
the
lamp
sleeves
and
assess
the
sleeve
A254
for
low
pressure
(
LP)
lamps
and
absorbance
from
20
hen
manual,
chemical
cleaning
is
recommended,
remove
the
sleeves
and
measure
the
sleeve
A254
before
and
after
cleaning
the
outer
surfaces.
If
the
new
sleeve
transm
ntify
.5.2.2
Assessing
UV
Intensity
Sensor
Window
Cleaning
Protocols
actor
e
cond
maxim
leaning
Various
cleaning
methods
can
be
used
to
periodically
remove
the
foula
la
s
include
off­
line
chemical
cleaning
(
OCC)
(
off­
line
manual
or
mechanized
clea
on­
line
mechanical
cleaning
(
OMC)
(
e.
g.,
brushes
or
rings),
and
on­
line
physicochemical
w
(
acid
solution
in
a
wiper
collar).
Sensor
window
cleaning
methods
also
include
manual
cleanin
and
mechanical
wipers.

J
A
sleeve
cleaning
assessment
should
be
performed
on
at
least
four
la
e
used
to
identify
a
sleeve
fouling
design
factor
for
sizing
the
UV
reactor
based
on
the
re
individual
sleeve­
fouling
factor
observed.
This
will
help
ensure
proper
dose
delivery
for
the
entire
life
of
the
sleeve.
One
method
for
assessing
lamp
sleeve
cleaning
needs
is
detailed
below:

Pass
water
through
the
reacto
um
power.
W
can
be
used
as
a
control
to
verify
that
fouling
is
occurring.
The
manufacturer's
recommendations
regarding
the
maintenance
of
the
cleaning
device
should
be
followed.

Record
the
UV
intensity
sensor
readings
before
and
after
the
cleaning
cycle
and
u
d
ements
to
ensure
only
lamp
sleeve
fouling
is
affecting
the
sensor
values.
If
possible,
check
all
UV
intensity
sensor
readings
with
a
reference
sensor.

At
regular
time
intervals
and
immediately
prior
to
the
sc
0
­
400
nm
for
MP.
The
non­
destructive
method
of
Lin
et
al.
(
1999b)
may
be
used
as
discussed
in
section
J.
5.1.

After
6
months,
or
w
ittance
is
not
restored
by
the
cleaning,
it
is
likely
that
the
sleeve
material
has
fouled
internally
or
permanently
degraded.
Further
monitoring
and
testing
may
be
necessary
to
ide
the
proper
sleeve
replacement
interval.

J
To
assess
fouling
of
the
UV
intensity
sensor
window,
one
alternative
is
to
operate
the
under
th
itions
suggested
in
section
J.
5.2.1
(
i.
e.,
minimum
water
flow
rate,
re
um
lamp
power).
After
6
months,
or
a
time
interval
suggested
by
the
manufacturer,
a
chemical
cleaning
of
the
monitoring
sensor
window
could
be
performed.
Alternatively,
c
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
8
June
2003
could
be
performed
when
the
sensor
reading
falls
to
a
minimum
value
suggested
by
the
manufacturer.
The
A254
of
the
window
should
be
measured
before
and
after
cleaning.
A
sensor
window
cleaning
frequency
can
then
be
estimated
as
discussed
in
section
3.1.3.1.

J.
5.3
Assessing
Head
Loss
ld
be
performed
for
demonstration­
scale
(
full­
scale)
systems
verify
that
head
loss
constraints
at
the
final
install
station
will
not
be
exceeded.
Head
loss
data
from
pi
ion
J.
1:
Head
loss
assessments
shou
to
lot
scale
units
should
not
be
used
to
estimate
head
loss
in
a
full­
scale
system.

The
head
loss,
 
H,
through
a
UV
reactor
may
be
calculated
according
to
Equat
g
Kv2
Equation
J.
1
H
2
=

where
=
Head
loss
coefficient
(
unitless)
for
the
UV
reactor
=
Water
velocity
(
m/
s)
through
the
reactor
,
the
UV
unit
can
be
installed
with
strumentation
to
measure
pressure
loss
across
the
reactor
(
including
baffles
and
specialized
inlet/
ou
res
due
to
um
and
ma
imum
flow
rates,
these
measured
head
loss
values
can
be
plotted
as
a
function
of
the
square
J.
5.4
amp
Aging
and
Failure
e
conducted
to
answer
the
following
questions:

e?

?

ds
of
hours.
The
germicidal
output
of
the
lam
will
decline
during
this
period
(
Phillips
1983).
In
MP
systems,
UV
lamp
aging
can
also
result
i
 

K
v
g
=
Gravitational
constant
(
9.8
m/
s2)

To
assess
head
loss
through
a
UV
reactor
in
tlet
piping).
Since
the
head
loss
coefficient
will
be
higher
at
lower
temperatu
decreased
water
viscosity,
it
may
be
desirable,
if
feasible,
to
measure
head
loss
at
the
lowest
water
temperature
expected
at
the
UV
reactor
installation
to
assess
the
worst­
case
condition.

If
head
loss
is
measured
at
various
flow
rates
through
the
reactor,
including
the
minim
x
of
the
calculated
water
velocity
through
the
reactor
to
determine
a
head
loss
coefficient.

L
A
lamp
aging
evaluation
can
b
 
What
is
the
actual
operating
lamp
life?

 
How
does
lamp
output
degrade
over
tim
 
What
lamp
aging
factor
should
be
specified
The
service
life
of
a
UV
lamp
extends
for
thousan
p
n
a
change
in
the
spectral
output
over
time.
With
polychromatic
(
MP)
UV
lamps,
lower
wavelengths
will
likely
decline
at
a
faster
rate
than
will
higher
wavelengths.
The
rate
and
manner
in
which
a
lamp
ages
is
lamp­
and
operation­
specific.
A
detailed
discussion
of
lamp
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
9
June
2003
aging
is
presented
in
section
A.
3.1.6.
Lamp
output
will
decrease
over
time
as
a
function
of
lamp
hours
in
operation,
the
number
of
on/
off
cycles,
and
the
power
applied
per
unit
(
lamp)
length.

La
the
mp
aging
tests
should
be
designed
to
assess
the
reduction
and
variance
in
lamp
germicidal
output
over
time
under
defined
worst­
case
operating
conditions.
Lamp
age
testing
may
us
igned
e
batch,
lamp
assembly,
electrical
haracteristics
of
the
ballasts,
heat
transfer
from
the
lamps
to
the
water,
and
operation
of
the
lamps.

clude
electrical
power
delivered
to
the
ballast,
electrical
power
delivered
to
the
lamp,
and
water
temperature.
If
UV
intensity
sensors
alone
monito
radation
of
the
lamp
assembly,
including
electrodes
and
seals,
and
any
darkening
of
the
lamp
 
any
fouling
on
the
internal
surfaces
of
the
lamp
sleeves;

ballast
operation
(
e.
g.,
power
setting),
heat
transfer
(
e.
g.,
lamp
sleeves),
and
environment
(
water
measured
using
one
of
the
following:
a
radiometer
ter
 
rkening
on
the
lamp).
y
be
 

n
be
used
to
identify
operational
sues
and
provide
operational
guidance.
The
output
of
the
lamps
measured
under
fixed
operati
and
e
either
a
pilot/
demonstration­
scale
UV
reactor
installed
at
a
WTP
or
a
test
bed
des
to
emulate
the
reactor
(
i.
e.,
identical
power
supply).
It
is
strongly
recommended
that
all
tests
b
done
with
the
lamps
housed
in
the
sleeves
and
powered
by
the
ballasts
that
will
be
used
in
the
final
application.
It
is
best
if
the
lamp
sleeves
are
maintained
free
of
external
foulant
during
aging
tests,
in
a
manner
similar
to
that
of
the
final
application.

Factors
to
consider
in
designing
the
test(
s)
include
lamp
c
Since
lamps
will
be
manufactured
in
batches,
it
is
recommended
that
lamps
from
several
different
lots
be
evaluated.
During
demonstration
and
pilot
scale
testing,
the
lamps
should
be
operated
in
a
manner
and
in
an
environment
that
reflects
conditions
expected
when
the
UV
disinfection
system
is
installed
at
a
WTP.

Parameters
to
monitor
over
time
in
r
lamp
output,
it
is
recommended
that
the
A254
of
the
water
also
be
measured.

During
testing,
it
is
recommended
that
the
following
analyses
be
considered:

 
Visually
inspect
the
lamps
at
regular
intervals
to
document
any
visible
deg
envelope;

Document
 
Measure
the
germicidal
output
of
the
lamp
under
fixed
conditions
of
temperature
and
transmission).
 
One
measurement
should
be
made
with
the
lamps
aged
100
hours
("
new").
 
The
germicidal
output
may
be
equipped
with
a
germicidal
filter;
a
reference
UV
intensity
sensor
or
radiome
from
200
to
400
nm;
or
by
bioassay.
The
output
from
various
positions
along
the
lamp
may
be
measured
based
on
visual
inspection
(
i.
e.,
the
pattern
of
da
 
If
lamp
power
is
variable,
lamp
output
as
a
function
of
lamp
power
setting
ma
measured.
Assess
the
output
from
lamps
of
different
lots.

Pilot/
demo­
scale
test
data
and
visual
inspections
ca
is
ng
conditions
can
be
plotted
over
time
and
fit
to
provide
mean
expected
performance
prediction
intervals
(
e.
g.,
90th,
95th,
and
99th
percentiles)
to
estimate
the
range
of
performance
in
Proposal
Draft
Appendix
J.
Pilot­
Scale
and
Demonstration­
Scale
Testing
UV
Disinfection
Guidance
Manual
J­
10
June
2003
lamp
intensity
at
different
lamp
ages.
In
a
robust
system,
all
the
lamps
will
age
in
a
similar
manner.
If
lamps
age
differently
than
expected,
the
results
will
affect
dose
delivery
and
UV
intensity
sensor
measurements.
This
data
can
be
used
to
assess
a
proper
end­
of­
lamp­
life.

J.
5.5
valuating
Controls
and
Alarms
ls
and
alarms
should
be
conducted
to
verify
their
erformance
and
to
gain
familiarity
with
alarm/
control
response
procedures.
A
test
plan
is
typicall
educing
ts,
or
ified
.6
References
and
E.
R.
Blatchley.
1999a.
Inorganic
fouling
at
Quartz:
Water
interfaces
in
ultraviolet
photoreactors
 
I
Chemical
characterization.
Wat.
Res.
33,
no
15:

Lin,
L.­
S.,
C.
T.
Johnston,
and
E.
R.
Blatchley.
1999b.
Inorganic
fouling
at
Quartz:
Water
interfaces
in
ultraviolet
photoreactors
 
II
Temporal
and
spatial
distribution.
Wat.
Res.
33,

Mackey
,
and
G.
F.
Crozes.
2001.
Practical
aspects
of
UV
disinfection.
Denver,
CO:
AWWA
Research
Foundation
and
AWWA.

Mackey
Bridging
pilot­
scale
testing
to
full­
scale
design
of
UV
disinfection
systems.
Denver,
CO:
AWWA
Research
Foundation
Phillips
rces
and
application
of
ultraviolet
radiation.
Academic
Press:
New
York.
E
An
evaluation
of
the
UV
reactor
contro
p
y
organized
prior
to
testing,
describing
the
control
function
to
be
tested,
the
test
procedure,
and
the
expected
response.
Faults
may
be
induced
or
simulated.
Low­
dose
conditions
may
be
simulated
by
reducing
lamp
power,
increasing
the
A254
of
the
water,
r
or
increasing
flow
beyond
the
validation
limits
of
the
reactor,
turning
off
lamps
or
ballas
disconnecting
sensors.
Valve
failure,
high
temperature,
and
ground
fault
interrupts
may
be
induced
or
simulated.
Simulating
faults
may
require
disconnecting
components
of
the
UV
disinfection
system
or
using
modified
electronics.
Accordingly,
qualified
personnel
as
ident
by
the
manufacturer
of
the
UV
disinfection
system
should
undertake
these
simulations.
All
operational
functions
can
be
verified,
including
startup
and
shutdown
sequences
and
cleaning
cycles.
Dose
pacing
if
used,
may
be
verified
by
monitoring
lamp
power
settings
and
dose
compliance
as
flow
rate,
A254,
and
lamp
output
are
varied.

J
Lin,
L.­
S.,
C.
T.
Johnston,

3321­
3329.

no
15:
3330­
3338.

,
E.
D.,
R.
S.
Cushing
,
E.
D.
and
R.
S.
Cushing.
Publication
anticipated
in
2003.

and
AWWA.

,
R.
1983.
Sou
Proposal
Draft
