WATER QUALITY MONITORING
IN
DISTRIBUTION SYSTEMS
by
Nina I. McClelland
National Sanitation Foundation
Ann Arbor, Michigan 48106
and
K. H. Mancy
The University of Michigan
Ann Arbor, Michigan 48109
Contract No. 68-03-0043
Project Officer
James M. Symons
Water Supply Research Division
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environ-
mental Research Laboratory, U.S. Environmental Protection
Agency, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor
does mention of trade names or commercial products constitute
endorsement or recommendation for use.
-------
FOREWORD
The Environmental Protection Agency was created because
of increasing public and government concern about the dangers
of pollution to the health and welfare of the American people.
Noxious air, foul water, and spoiled land are tragic testi-
mony to the deterioration of our natural environment. The
complexity of that environment and the interplay between its
components require a concentrated and integrated attack on
the problem.
Research and development is that necessary first step
in problem solution and it involves defining the problem,
measuring its impact, and searching for solutions. The Muni-
cipal Environmental Research Laboratory develops new and im-
proved technology and systems for the prevention, treatment,
and management of wastewater and solid and hazardous waste
pollutant discharges from municipal and community sources,
for the preservation and treatment of public drinking water
supplies, and to minimize the adverse economic, social,
health, and aesthetic effects of pollution. This publication
is one of the products of that research; a most vital communi-
cations link between the researcher and the user community.
Concern for protecting the public health by assuring
the quality of public drinking water supplies is apparent with
passage of the Safe Drinking Water Act (Public Law 93-523) ,
signed by the President on December 16, 1974. This is the
first federal act dealing in depth with providing safe
drinking water for public use. Through this project, the
feasibility of measuring levels of drinking water contaminants
at tkz con^ame^.'-4 tap has been demonstrated. Subsequent use
of the mobile water quality monitoring laboratory will per-
mit EPA to assure compliance with many of the maximum levels
for inorganic chemicals proposed as national interim primary
drinking water standards (CFR Vol. 40, No. 51—Friday, March
14, 1975).
Francis T. Mayo, Director
Municipal Environmental Research
Laboratory
-------
ABSTRACT
A mobile laboratory with 18 integrated, computer con-
trolled parametric systems for monitoring potable water
quality in distribution systems was developed and field
evaluated at ten locations in four United States cities:
Chicago, Illinois; Ann Arbor and Detroit, Michigan; and
Philadelphia, Pennsylvania. Temperature, conductivity, pH,
chloride, dissolved oxygen, free and total residual chlorine,
turbidity, corrosion rate, free and total fluorides, alka-
linity, hardness, nitrate, copper, cadmium, lead, and cal-
cium carbonate deposition rate are measured using commercially
available and newly developed sensor systems.
IV
-------
CONTENTS
Foreword
Abstract
List of Figures
List of Tables
Acknowledgments
Sections
I Conclusions 1
II Recommendations 4
III Introduction 5
General 5
Study Aims and Objectives 5
Study Plan 6
IV Prototype Monitor 8
General 8
Theory 8
Performance Characteristics 8
SRM-Housed Systems 14
General 14
Electronics 15
Recorder 18
Sensors 18
Interpretation of Hardness and Nitrate Data 19
Commercial Systems 24
Free and Total Residual Chlorine 24
General 24
Theory 24
Experimental 27
Results and Discussion 31
Turbidity 33
General 33
Experimental 33
Corrosion Rate 33
General 33
Theory 36
Experimental 39
Results and Discussion 40
v
-------
Page
Other Systems 41
Calcium Carbonate Deposition Test 41
General 41
Theory 41
Experimental 50
Results and Discussion 54
Field Experience 59
Alkalinity 62
General 62
Theory 62
Experimental 70
Results and Discussion 72
Special Case Study 75
Free and Total Fluorides 75
General 75
Theory 76
Experimental 77
Results and Discussion 78
Automated System 87
Continuous Fluoride Monitor 106
Field Studies 111
Trace Metals 121
General 121
Theory 123
Experimental 126
Results and Discussion 130
Special Study 130
V Mobile Laboratory 140
General 140
Physical Description 140
Instrumentation 141
Organization 141
Computer System 142
Field Studies 149
VI References 168
VII Project Publications 172
VIII Appendices 174
VI
-------
FIGURES
No. Page
1 NSF/EPA Mobile Water Quality Monitoring Laboratory 7
2 Schematic Arrangement of Schneider Robot Monitor 14
3 Schematic Diagram of SRM Flow Cell Cubicle, Top View 16
4 Schematic Illustration of Sensor Positions and Over- 17
all Arrangement of SRM-Housed Systems
5 Typical Strip Chart Recorder Output of Hardness Data 21
6 Conductivity Correction Nomogram 23
7 Distribution of Hypochlorous Acid and Hypochlorite 26
Ion in Water at Different pH Values and Tempera-
tures (2)
8 Schematic Illustration of Residual Chlorine Analyzer 29
9 Schematic Illustration of Cl2 Analyzer 30
10 Free and Total Residual Chlorine Variations at Gulf 32
Oil, Philadelphia
11 Schematic Illustration of Turbidity System 34
12 Schematic Illustration of Turbidimeter in the Mobile 35
Laboratory
13 Schematic Illustration of Corrosion Rate Monitor in 40
the Mobile Laboratory
14 Fluid Flow in the Rotating Electrode System 45
15 Concentration Profile at Electrode Surface under 47
Steady State Conditions, in the Absence of CaCOs
Film
16 Oxygen Current - Time Curves during CaC03 Deposition 49
17 Stability Index Monitor for Saturation Equilibrium 51
Measurement
18 PIR Rotator - Electrode System 53
19 CCDT System in the Mobile Laboratory 53
20 Current versus Potential Sweep 55
21 Calcium Carbonate Deposition Test 56
22 Correlation of CCDT Slope with Ryzner Stability Index 57
vii
-------
No.
23
24
25
26
27
28
29
30
CCDT versus Ryzner Stability Index
Effect of Caustic Feeder Cycling at Chicago CWFP
on CCDT and pH Monitored at Site No. 1, Chicago/
Lehigh
[H ] versus C032 formality, given applying Equa-
tion 98
Experimental and Theoretical Relationships between
Alkalinity and pH
Schematic Illustration of Alkalinity Monitor in
the Mobile Laboratory
Good Quality Recording of Continuous Alkalinity
Analysis
Alkalinity Calibration Curve
Characterization of Fluorides and Preparation of
Page
61
63
67
68
71
73
74
79
Buffers
31 Effect of pH on Fluoride Ion-Selective Electrode 80
Measurement
32 Calibration Curves of Different Levels of pH 82
33 Effect of pH on Fluoride Ion Electrode Sensitivity 85
34 Complexometric Titrations of Fluoride with Aluminum 86
35 Aluminum Complexing Effect and Fluoride Recovery by 89
Different Masking Agents
36 Rate of Fluoride Recovery with CDTA Masking Agent 90
37 Rate of Fluoride Recovery with Citrate Masking 91
Agent
38 Schematic of Automated Sample Analysis for Free and 92
Total Fluoride
39 Effect of Pump Surges on Reference Electrode 95
40 Effect of Flow Rate versus Response Time 97
41 Recorder Output for Calibration of Free and Total 98
Fluoride (5 minutes)
42 Recorder Output for Calibration of Free and Total 99
Fluoride (2 minutes)
43 Recorder Output for Calibration of Free and Total 100
Fluoride (1 minute)
44 Calibration Curves for Automated Fluoride Electrode 102
Measurement
45 Reproducibility of Fluoride Measurement in Absence 104
of Complexing Cations
viii
-------
No. Page
46 Reproducibility of Fluoride Measurement (1.0 mg/1) 105
in Presence of Aluminum
47 Schematic Illustration of Free and Total Fluorides 110
Monitor in the Mobile Laboratory
48 Fluoride System Output including a Calibration 112
Series
49 Typical Calibration Curves for Fluorides, Prepared 113
from Output in Figure 48
50 Special Fluoride Study at Chicago/Lehigh Station 118
51 Map Identifying Relative Locations of Monitoring 119
Sites in Chicago
52 Output from Chicago Fluoride Monitor, recorded at 120
Chicago/Greenleaf
53 Special Fluoride Study at Calumet Harbor 122
54 Plate-Strip Sequence for Differential Anodic 125
Stripping Voltammetry
55 Schematic Diagram of DASV Flow Cell used in Mobile 128
Laboratory System
56 Schematic Illustration of Trace Metals Monitor in 129
the Mobile Laboratory
57 Typical Voltammograms of a Standard Solution and 131
Tap Water Analyses
58 Sweep Rate Response 132
59 Household Trace Metals Survey, Cadmium 134
60 Household Trace Metals Survey, Lead 135
61 Household Trace Metals Survey, Copper 136
62 Schematic Illustration of Right Side Interior of 143
Mobile Laboratory
63 Schematic Illustration of Left Side Interior of 144
Mobile Laboratory
64 Photograph Showing Overview of Mobile Laboratory 145
Interior
65 Relationship of Computer System to Analytical 147
Sensor Systems
66 Computer System Software Relationships 148
67 Flowchart of INIT 150
68 Flowchart of NSFC 151
69 Flowchart of PUN 153
70 Flowchart of BELL 154
IX
-------
No. Page
71 Flowchart of SMPL 155
72 Flowchart of FLOW 156
73 Flowchart of DAQ 157
74 Flowchart of Subroutine A (DAQ) 158
75 Flowchart of DATA 159
76 Normal Data Flow 162
77 Data from Main Flushing at Chicago/Calumet Harbor 165
78 Relative Locations of Philadelphia Monitoring Sites 166
79 [H+] as a Function of C032~ and C12 Formality 185
80 Experimental and Theoretical Relationships between 186
Alkalinity and Potentials
81 AE versus [Clal in Alkalinity Determination 187
82 Potential versus Alkalinity Calibration Series 191
x
-------
TABLES
No. Page
1 Sensors in the Prototype Monitor 9
2 Electrochemical Sensors for Water Quality Monitoring 10
3 Performance Characteristics of Mobile Laboratory 12
Sensor Systems
4 Summary of Data from Analyses of Test Solutions 58
5 Results of Field Studies 60
6 [H ] versus Alkalinity, applying Equation 98 69
7 Experimental and Theoretical Relationships between 70
Alkalinity and pH
8 OH~ Selectivity Coefficients (K) 81
9 Shift in Intercept of Calibration Curves at Low pH 84
Values
10 Aluminum Complexing Effect and Fluoride Recovery by 88
Different Masking Agents
11 Iron Effect and Fluoride Recovery by Masking Agents 93
12 Reproducibility of Automated Measurements at Differ- 106
ent Fluoride Levels in Presence and Absence of
Aluminum
13 Fluoride and Iron Determinations in Water Samples 107
(mg/1) from Jackson, Michigan, 2067
14 Characterization of Fluoride in Water Samples from 108
Wyoming, Michigan, by the Manual and Automated
Electrode Methods
15 Statistical Data Related to Manual and Automated 109
Fluoride Electrode Measurements in Water Samples
from Wyoming, Michigan
16 Characterization of Fluoride in Water Samples from 115
Ann Arbor, Michigan (April 6, 1971)
17 Chronological Characterization of Fluoride in Water 116
Distribution System of Ann Arbor, Michigan
18 Summary of Free and Total Fluoride Data (Chicago/ 117
Lehigh Station)
19 Electrode Deterioration 133
xi
-------
No. Page
20 Electrode Response After Regeneration 133
21 Trace Metals Survey Sites 137
22 Summary of Raw Data from Household Trace Metals 138
Survey
23 [H+] as a Function of C032~ and C12 Formality 188
24 Experimental and Theoretical Relationships between 189
Alkalinity and Potentials
25 Theoretical Relationship between Alkalinity and 189
Potential for Solutions with and without Free
Residual C12
26 Changes in Potential from Addition of 1.0 mg/1 190
Free Residual C12
27 Effect of Thiosulfate 192
-------
ACKNOWLEDGMENTS
Sincere appreciation is expressed to the following persons who
provided input to the project:
Gordon G. Robeck and James M. Symons, Ph.D., Director and Chief,
Water Supply Research Division, respectively, who served as proj-
ect officers. Through their efforts, continuation support was
made available to equip the mobile laboratory and computerize
monitoring operations. The merits of adding practical applica-
tion to analytical feasibility are readily apparent. The overall
guidance and support provided by both Mr. Robeck and Dr. Symons
are greatly appreciated.
Further guidance and considerable constructive criticism were
provided by members of the Advisory Committee. Their assistance,
both individually and collectively, is also very much appreciated.
Commercially available instrumentation with potential applica-
bility for the proposed monitoring system was evaluated early
in the project. The cooperation of manufacturers who generously
consigned equipment for this purpose is gratefully acknowledged:
Aqua Test Corporation
Capital Controls Company, Inc.
Hach Chemical Company
Schneider Instrument Company
During the various field assignments, personnel at each respec-
tive local water utility provided liason with NSF/EPA staff oper-
ating the mobile laboratory. Each of these persons responded
willingly to needs of the permanent crew, and demonstrated en-
thusiastic support for the monitoring effort. Virtually every
request was granted. Thanks is expressed on behalf of the entire
project staff for their untiring assistance: Harvey Mieske, Ann
Arbor; Richard A. Pavia, N. J. Davoust, Charles Halter, Ben F.
Willey, and Green Whitney, Chicago; Carmen F. Guarino, Charles E.
Vickerman, Joseph V. Radziul, Alan Hess, Charles Pierce, and
David Gotshall, Philadelphia; and Gerald Remus and Albert Shannon,
Ph.D., Detroit.
To assure smooth transfer of operations from NSF to EPA, three
persons in EPA were assigned responsibility for operating the
mobile laboratory: Marvin C. Gardels, Ph.D., Robert Thurnau,
and Daniel F. Watkins. The EPA crew visited Ann Arbor to get
xiii
-------
acquainted with NSF project staff, become familiar with systems
in the mobile laboratory, and participate in the final meeting of
the Advisory Committee. During the monitor's month-long visit
to Philadelphia, the EPA crew rotated for one-week periods and
accompanied the NSF operator to and from Philadelphia. EPA parti-
cipation was scheduled to provide each operator with experience
in shutting down, moving, and starting up in a new location. It
was reassuring to observe the enthusiasm of the EPA crew for the
task ahead. After more than four years of effort, the mobile
laboratory and each of its systems were, themselves, like part of
the project staff. A great deal of pride and effort went into
their development. It is gratifying to know that future opera-
tions are in the hands of competent, concerned professionals.
Two weeks were scheduled to install - and develop software for -
the on-board mini computer. The experts said it could not be
done, but three dedicated people proved them wrong: William B.
Everett, NSF; Charles A. Khuen, Ph.D. and James Seydel, Ph.D.,
The University of Michigan. These people met, assigned tasks,
and agreed that the task was virtually impossible in the time
available to them, then proceeded to ignore the clock and complete
their assignments on schedule. For their initiative and persis-
tence, sincere thanks to each of them for a job very well done.
Finally, the staff for this project worked tirelessly to achieve
each of the expressed objectives. Their loyalty and dedication
to the project through each phase of effort was indeed commend-
able. Special thanks are expressed to John R. Adams, who devoted
many extra hours to operating the mobile laboratory on its
various field assignments; Robert R. Wood; Warren K. Schimpff,
Ph.D.; Thomas L. Schwenk, M.D.; Diane Daniel and Sue Buske, proj-
ect secretaries, Nina I. McClelland, Ph.D. was project director,
and K. H. Mancy, Ph.D. and Dr. McClelland, coprincipal investiga-
tors.
xiv
-------
SECTION I
CONCLUSIONS
It is analytically feasible, using automated electrochemical
techniques, to detect and measure changes in quality which do
occur in potable water during transmission through distribution
systems. A fully operational prototype drinking water quality
monitor, with 18 integrated, computer controlled, parametric
systems installed in a mobile laboratory was delivered by the
National Sanitation Foundation to the National Environmental
Re'search Center, Cincinnati, Ohio upon successful completion of
the project, "Water Quality Monitoring in Distribution Systems."
All systems were evaluated under actual field conditions by
operating the mobile laboratory for two and a half months at
remote sites on transmission lines in Ann Arbor and Detroit,
Michigan; Chicago, Illinois; and Philadelphia, Pennsylvania.
The reliability and applicability of on-board systems in the
mobile laboratory for detecting and measuring significant changes
in quality at remote monitoring sites along distribution networks
was clearly established. Because of relatively infrequent
sampling schedules and time lapse between sampling and laboratory
analysis, automated, onsite measurements are potentially more
meaningful than data acquired from perimeter surveys. Widespread
use of the NSF/EPA mobile laboratory and other similar systems
could be expected to contribute significantly to diagnostic
activities and preventive quality assurance programs consistent
with surveillance and enforcement requirements of the Safe Drinking
Water Act.
Correlation of data acquired by the mobile laboratory with measure-
ments recorded at the water purification plant provides for cal-
culation of residence times in the distribution system at periods
of varying demand, indicates the need for flushing at or near
deadends in the system, and assists the treatment plant operator
in evaluating the need for-and effect of-changes in treatment
plant operation.
Equipment design specifications developed early in the project were
oriented principally to laboratory use. Computer control and
mobile laboratory application became objectives only after the
analytical feasibility of individual parametric systems was clearly
established. Specific observations and analysis of data acquired
during field operations suggest the following refinements in
planning second generation equipment intended for field use; e.g.,
-------
1. Computer capabilities can- and should- be greatly
expanded. At the least, programs should be
developed for better utilization of the automatic
sampler; e.g., sampler operates when predetermined
levels of specified parameters are exceeded; and
on-board electronic data processing capability
including computer graphics should be developed.
In addition, either a cassette or magnetic tape
system should be provided in lieu of the paper
tape option purchased for this project. Selection
was made on the basis of economy at the time of
purchase; however, paper tape output is extremely
difficult to handle in the field. Cassettes are
attractive for ease in mailing from remote loca-
tions for additional processing on a larger com-
puter away from the mobile laboratory.
2. The original DASV system was designed for labora-
tory operation. Its complex arrangement with
switching/timing capability is unnecessary with
the mini computer available. A commercial po-
tentiostat is entirely adequate for DASV mea-
surements , operational control should be a
function of the computer. In addition, pH ad-
justment capability should be provided for the
DASV monitor to extend its capability for mea-
suring additional metals.
A single Technicon pump could be used for all
systems requiring reagent addition; i.e., DASV,
alkalinity, and fluorides. Recurrent bubble
entrapment problems in the fluoride systems
should be eliminated. Further study of solution
ground, peristaltic pump noise, and the role of
organics with respect to the alkalinity monitor
is indicated. Prepotentiometer electronics
might be added to filter the signal.
3. Liquid junction ion-selective electrodes; i.e.,
hardness and nitrate should not be used when
advanced electrode technology provides more re-
liable electrodes. Although it is unlikely that
solid state probes will be feasible in the near
future, improved bodies with better electrical
shielding are promised by the manufacturers.
Use of an electrode other than the divalent
cation electrode for measuring hardness, and
placing hardness outside of an SRM system should
be considered. The overall usefulness of nitrate
measurements in distribution system monitoring
should be reviewed. It might be desirable to
retain a nitrate electrode "for use as needed"
and substitute routine monitoring of sodium or
calcium.
2
-------
4. Use of an electrode multiplexer should be con-
sidered. One multiplexer channel could be de-
dicated to bench measurements and calibrations
while other channels are used for Technicon
systems measuring special interest parameters.
Linking multiplexer, potentiometer, and computer
would reduce the number of potentiometers re-
quired in a mobile laboratory.
5. Better provision for external grounding, re-
placing the plastic adapter for intake water
supply, and acquiring a pump for use in low
line pressure areas in the field are recommended
changes in the NSF/EPA mobile laboratory. Addi-
tional airconditioning capability is also a
desirable improvement.
In developing a "second generation" mobile water quality monitor-
ing laboratory, considerable savings in time and cost can be
anticipated as a result of experience with this study.
-------
SECTION II
RECOMMENDATIONS
1. The NSF/EPA mobile laboratory should be placed in regular
service and made available to communities where water quality
deterioration in the distribution system is known to occur.
For systems in which the quality of water at the tap could be
improved by modifying treatment plant operation, the mobile
laboratory is an excellent diagnostic tool. It can be used
to identify changes and, with continued monitoring, to assure
the effectiveness of selected corrective measures.
2. The NSF/EPA mobile water quality monitoring laboratory should
be made available to states for support of their new respon-
sibilities, defined by the federal Safe Drinking Water Act
(Public Law 93-523); and used routinely by EPA for public
drinking water supply surveillance activities.
3. Additional mobile water quality monitoring laboratories should
be constructed. This would provide for operation of two or
more systems on the same transmission line as well as wide-
spread, simultaneous monitoring capability. Federal, state,
district, and large local water utility resources should be
committed to this objective.
-------
SECTION III
INTRODUCTION
General
The primary objective of any drinking water utility is to pro-
vide an adequate supply of high quality water to its users.
The quality of water leaving the treatment plant is assured by
routine analytical procedures at the local utility and periodic
surveillance by state water supply agency personnel.
It is generally assumed, however, that quality of a treated
water supply changes during transportation and delivery to the
user's tap. These changes may occur as a result of: (1) physi-
cochemical or biochemical transformations in the municipal- or
household-pipeline; i.e., spontaneous changes in the water in-
dependent of pipeline characteristics; or (2) physicochemical
interactions of the water with the municipal- or household-pipe-
line. Examples of these effects include "unstable" water which
deposits calcium carbonate (CaCOs) film as a function of its
residence time in the piping system (a physicochemical trans-
formation in the water itself), and uptake by the water of avail-
able metals from the piping system (a physicochemical inter-
action of water with the pipeline).
Regardless of the type of change which occurs during distribution,
quality at tke, U.A&SI'& tap must be assured. In 1965 the Research
Committee of the American Water Works Association (AWWA) recom-
mended that instrumentation for measuring corrosion and stability
be developed for distribution system monitoring applications.
The research study, "Water Quality Monitoring in Distribution
Systems," undertaken by the National Sanitation Foundation
under grant and contract support from the U.S. Environmental
Protection Agency, was a direct result of this recommendation.
Study Aims and Objectives
Principal objectives of the NSF study were to develop basic
design criteria and operational specifications for a continuous
monitoring system which could measure changes in water quality
characteristics in distribution systems and provide for analyti-
cal quality control of water purification processes. Specific
aims included: (1) establishing the analytical feasibility of
commercially available sensor systems for potable water quality
-------
monitoring applications, (2) developing new sensor systems as
required, (3) assembling a prototype monitor, (4) establishing
performance characteristics for each integrated parametric
system in the prototype monitor, (5) installing the monitor in
a mobile laboratory, and (6) operating the mobile laboratory
at selected sites under actual field conditions.
The NSF/EPA mobile water quality monitoring laboratory, shown
in Figure 1, was designed, constructed, and successfully oper-
ated at the municipal water treatment plant in Ann Arbor,
Michigan; four sites in metropolitan Chicago; a remote distri-
bution system site in Detroit; and four sites in Philadelphia.
It was delivered to EPA/NERC, Cincinnati, Ohio in October 1973
for operation by EPA in future distribution system monitoring
and research assignments.
Study Plan
Eleven persons with expertise in water utility management served
as an advisory committee to establish priorities and provide
overall guidance to the project staff. (The advisory committee
membership is listed as Appendix A.) Parameters, selected by
the advisory committee for inclusion in the prototype monitoring
system, included temperature, conductivity, pH, chloride, dis-
solved oxygen, free and total residual chlorine, turbidity,
corrosion rate, free and total fluorides, alkalinity, hardness,
nitrate, copper, cadmium, lead, and calcium carbonate deposi-
tion rate.
Consistent with project aims and objectives, commercially avail-
able sensor systems were loaned to the project by their respec-
tive manufacturers. The analytical feasibility of including
these systems in the prototype monitor was carefully evaluated
in the laboratory. No effort was made to modify existing-or
develop new-sensors when commercially available systems were
shown to be applicable.
-------
o
-P
03
J-l
o
X!
tn
C
-H
M
O
-P
•H
£
O
g
-P
•H
O1
(U
-P
0)
rH
•H
X!
O
i=
<
ft
W
\
fci
01
0)
-------
SECTION IV
PROTOTYPE MONITOR
General
Eighteen parametric sensors are included in the prototype moni-
toring system developed during this study. Nine of these are
commercially available and were used without modification.
(See Table 1.) Five of the nine - temperature, conductivity,
pH, chloride, and dissolved oxygen - are original equipment
with the Schneider Robot Monitor(RM-25), an integrated instru-
ment system designed for monitoring the quality of water in
lakes and streams. These five sensors were shown to be highly
reliable with no requirements for special timing sequence or
preconditioning of the sample stream.
Other commercial sensors used without modification include free
and total residual chlorine analyzers from Capital Controls
(Models 871 and 872, respectively), a Hach CR low range light
scatter turbidimeter (Model 1720), and an instantaneous corrosion
rate monitor by Petrolite Corporation.
Orion ion-selective electrodes for fluoride, hardness, and
nitrate are included in the prototype monitor. The fluoride
system, which provides for both free and total fluoride measure-
ments, requires preconditioning of the sample stream. Sensors
for hardness and nitrate, though housed in Schneider modules,
required development of support electronics and a flow interrupt
timing/switching device.
Theory
Basic theoretical relationships describing electrochemical
sensors for a number of parameters in the prototype monitor are
summarized in Table 2. Theoretical considerations for other
systems are described in greater detail in respective sections
of the report.
Performance Characteristics
Performance characteristics of each sensor in the prototype
system are described in Table 3. Terms are defined as follows:
-------
PS
o
EH
H
!S
O
a
w
EH
O
EH
O
A
CM
H
ffi
EH
CO
PS
o
CO
sz
w
co
cu
CO CU
CU rH 0
SH rd tn G
•H --H C CU
3 O -H 2
D1 CU g (J1
cu CU-H cu
PS CO EH CO
•d
cu
c
0
•H
CU 4-1
H -H
CU 1 ^d
g CU G
fd SH O
co CM u
rH
cu fd cu
•H -rl rH
•W O X!
•H SH fd g
T3 CU >H CU
O g-H 4->
g g fd co
G 0 > >i
!D u <; co
cu
pi
!>i
EH
J_|
O
CO
G
CU
CO
SH
CU
4J
OJ
g
rd
SH
fd
CU
^~>.
J_l
o
4->
CO
•H
g
CU
X!
EH
CU
(-1
rj
•P
rd
rM
cu
Cu
g
(U
EH
V
rH
rH
(U
O
J>^
4->
•H
•H
4->
O
rj
'c
o
o
u
1
<
K*1
4->
•H
>
•rH
4->
u
E3
'd
G
o
o
%.
cu
•g
M
4J
o
cu
H
w
CO
CO
fd
rH
o
K
cu
^i.
cu
J>
•H
4->
U
CU
rH
CU
pj
O
H
(U
4-)
rd CU
•P T3
CO O
}_j
•d -P
•H 0
rH CU
O rH
CO W
cu
•H
SH
O
rH
XI
a
^.
cu
•g
SH
4->
u
cu
rH
W
U
•H
SH
4J
1
g
fd
4->
rH
o
*d
cu
£>
•H G
O CU
co tn
CO >i
•H X
Q O
^v
s.
rH
H
0
CJ
O
•H
C
fd
£>
rH
fd
a
rH
fd
rj
•d
•H
CO CU
CU G
PH 'H
^
CU O
CU
rH
fd
o
rH
(d
rj
t3
•H
CO
CU CU
•H
rH >H
fd o
4-J H
O X!
H U
s>.
!H
cu
4J
Q)
g
o
rH
CU
X!
Cu
cu
!>i
4J
•H
•d
X!
r)
H
^v
CU
o
G
fd
4J
4j
•H
g
'd
^
£
o
•H
4-1 (U
N O1
•H -H
LJ r~*
^^ M
rd X!
rH O
O CU
CM EH
CU
4->
fd
Pi
C
O
•H
CO
O
SH
0
U
^v
CU
J>
-H
4->
O
CU
tH
CU
CO
G
0
H
CU
4J
rd cu
-p -d
CO O
^
'd 4-*
•H 0
rH CU
O rH
CO W
CU
T)
•H
J_l
0
rH
pL|
cu
CU
PM
^>.
CU
J>
•H
4->
U
CU
rH
CU
CO
G
O
H
CU
4-)
fd cu
•P n3
CO O
r<
-d -P
•H o
rH Q)
O rH
CO W
CU
•H
^_|
0
rH
Pn
rH
rd
4->
0
EH
^>
fj
O
•H
4J
rrj
c
•rH
g
O
u
u
•rH
M
4-> CU
CU T3
g o
G JH
0 4J
•H 0
4J CU
G rH
CU H
4->
O W
CM Cu
^i
•P
•H
G
•H
rH
fd
X
rH
.
^
G CU
O ^
H O
J^
G 4-1
0 0
•H Q)
4-> rH
O H
C
rj Q)
I'D >
•H
T3 4->
•H rj
3 CU
•H CU
rl CO
CO
CO
CU
CH
•d
fd
ffi
"^V
G CU
o fd
H O
J_J
G 4->
O 0
•H CU
4-> rH
U W
G
3 CU
rD >
•H
13 4-1
•H O
3 CU
CT'rH
•H CU
i-H CO
CU
fd
rM
•H
&
^.
^
tn
G
-H
Cu
Cu
•H
J-l
4-j
CO
o
•H — *
•d >
O CO
G <
r< Q
H
fd >i
•H ^ [
4-> 4-1
G CU
CU g
(j C
^1 J-H
cu fd
MH 4-1
lii i
•H O
Q >
g
3
•H
g
fd
U
%.
•v
J>
CO
fZ£
Q
•d
rd
0)
V
%.
^>
CO
f^J
Q
j^
CU
Cu
0
u
tn
G
•H
4J
fd cu
[ \ r^
o o
fyj SH
4J
0 D
•H CU
4J rH
fd W
4-)
CO U
O CO
•H -H
4-> Q
C
cu tn
4J G
O -H
CO
CU
EH
G
CU O
•P-H
g fd -P
3 G-H
•H O CO
O XI O
rH SH D.
rd rd CU
O O Q
-------
Table 2. ELECTROCHEMICAL SENSORS FOR WATER QUALITY MONITORING
Sensor Type
Conduc tome trie
Po ten tiome trie
- Glass electrode
(e.g. , pH)
- Membrane electrodes
(e.g., ion-selective
electrodes)
- cationic
- anionic
Voltammetric Membrane
Electrodes
Equation
L = Kc I Ci Xi zi (1)
RT f 2i/Zil
E = K + £i In a. + KTa x D (2)
m zt L i J J J
pH = -log aR+ (3)
pM+ = -log aM+ (4)
pA = -log aA~ (5)
^ = [z FAPm E] a02 (6)
where:
L
Ke
C.
z. =
E =
m
F =
K. =
D
id =
A =
b =
specific conductance
cell constant
ionic concentration
ionic equivalent conductance
ionic valency
measured electrode potential
Faraday constant
selectivity coefficient
diffusion current
electrode surface area
membrane permeability coefficient
membrane thickness
10
-------
(1) V
-------
CO
E
LU
CO
CC
LU
CO
CC
o
oc
o
CO
LLJ
_J
OQ
o
u.
o
CO
o
CO
CC
LU
u
<
CC
<
X
o
LU
u
CC
o
LI-
CC
01
Q.
CO
CO
Long Term Stability,
Recalibration and
Maintainance Requiremsnts
JC
en +-
y c u
c w H-
o
h- Q
6
'o
O)
ol
u.
c
O)
*U (A
£: * t:
** a»
> ~
tj ~~
u
IT.
•ST
"aT
O)
c
re
CC
C
o
OJ ~
C) £
a j
^
ai
t;
£
These liquid junction ion se-
lective electrodes have been
shown to have stable poten-
tials which permit continuous
use for periods up to seven
days without recalibration;
c ~ ^S
•o-aji.! £
!2.H IB' c 3
o °^_ o •£ C
111 5 IE
2 3 ? ? c 3
c S jc E .2 ^
0 = o£ 3 E
o>
Eo
sy
"-T °
Ol 1
E o >
" i S"
O S in — . to q
ro to ro ^J i—
u II ii M n
CM CM CM p4 tM
C a> 3 •- ra
N U- U Z U
cT 'SO
T3 CO
ro c ro
S oo
0} . | '^
c °^ O CT
O « *— W
c * O
o 5v>
2 r o
1
i
however, due to the shape of
the electrode tip, twice daily
attention is required to insure
that bubbles have not formed
on the tips. Complete rejuven-
ation of the probes must be
performed every 30 days.
Z
CM
CM
•o
(U
o jj
w C
I 8.
fe E
£ O
1- U
q
z
CM O
0 °- ™
O CM 10°
0 O i
X O O
VI
Q, _
^ ^^
u P
eu *"
0 §
O . —
"ot
E
*•
0)
C
Z
7 days; No known main-
tainance
Z
CM
CM
T3
0)
L_ +J
2?
E a
fc E
j; o
H 0
^
I"
CM
+ 1
2
O
,11
I
O
Ol
E
§
I
a*
E
q
to
I
U
7 days; No known main-
tainance I
to
Z
Z
^
Z
u.
0
o
+ 1
c
o
Z
u.
0
o
CM
A
.
<;
z
£
3
Q.
E
u
7 days; Internal flow chan-
nel and electrode surfaces
must be cleaned monthly
u
"£ ._
E .E
Q. 3 ~3* CJI
llfil
N
in
•D
i- *j
S 5
I Q.
£ E
f O
H O
§
o
X
S
o
u
I
*T
0)
c
o
Z
5r
o
o o5
o o£
CM «t o E
o o £
.
<;
z
>.
*>
C
•o
c
o
o
7 days; No known main-
tainance
"i
Z
CM
CM
•o
Q>
||
IS.
uE
JC O
I-O
w
'c
3
CM
q
o
+1
ra
Z
CM
CM
a
.
E
-C O
HO
Ol
E
o
+ 1
c
0
Z
CM«t_
AAl
Ol
E
d
T)
>£
as
.- X
QO
Daily checking of calibra-
tions by comparison with
„
1"s1
Ol *^
"" ^_> <
o "i "t;
if!
a°]
— ^ tj
O. \t\ ^
ro -5*1
in « c
•o
k. -4-t
OS
•A C
II
a> E
£ O
1-0
"Si
E
q
+f
c
o
Z
o o <
fr
00 (
'is
E
5
d
n
3
S
at **
v^
ito
required. Reagent reservoirs
must be refilled every 7 days
>
i
i^
: « c
Sg-o
i "^
: 13
) >T3
j J3 n
H
3 C
>•£ o
:|1
CI
3 E
Ii
E
2
d
3
3
oce
5.2
KO
7 days; Periodic cleaning of
photocell windows maintains
sensitivity. Light fouling of
the windows does not effect
accuracy within the 7 day
calibration period
Z
Z
^J
Z
D
^
3
">
q
o
i
3
12
-------
c
o
u
Long Term Stability,
Recalibration and
Maintainance Requirements
x;
"en*-
u c o
01
HQ
.5
'M
'o
0)
£
M
Selectivity Coefficieni
or
Interferences
S
at
O»
c
IX
c
0
t5 4-*
v •-
Q 3
a)
•s
E
1
j
T3 Ol ^."t?
Daily introduction of stan-
dards maintains accuracy an
demonstrates integrity of th
flow system. Electrodes are
known to be stable over ver;
long periods of time. Reager
reservoirs must be refilled
every ten days
.c
en _
£^
a)
3
" ?
"• .2 E, cT 0
*; « .5 c s-
u •*•
:o S g
-sic?
i£ 2 i- ~ t
c- £
* O ** w
"J CO _ 2 JC
+ .5
"Si
N
0
+ 1
: c o-^
ir.8 ® ° !n 2
O i^ O • O '
1, 0 „ 0 n 0
i i " i "
o o o
1
o.
I
o
"en
E
o
o
2
o
IL
£
"Si
E
o
1
o
1
•8
Reagent reserviors must be
refilled every four days. Re-
calibration must be perform
for each new set of reagents
f I
£ ° »•£
£-a-£ QC
0 *J ^ CO .2
1 .1 = ^
O "O O -^"O
_ Q u .Q ra
Z
c
o
c
c
3
to
o
u
n
n =
+ | S
E
i-n • S
Residual chlorine inte
feres, but is eliminate.
by addition of thiosul
fate to the sample stn
to
c ^.O
S. °*^
^ 0^
en op >
O "TO CT
U> U CM S
^
Dl
o
>.
'§
15
f
Standards are analyzed auto
matically every six hours.
DASV electrodes must be
replated daily
V
at
3
£2S-
trt j
0 "
*-o
-~ (o y A ra
IS
??
11;
ro -o -*-
in S a
Z
q
z
(A
4>
C.
Trace organics can in-
terfere by giving false
peaks. Accurate iden-
tification of peak po-
tential indicates the
oresence of interferen
=^
I"
f
o
5i
oi
E
O
o.
o
1
I
•o
fl>
u
=5.
F
I
_
01
E
i
o
•o
s
d
E
I
,_
E
1
o
w
a
5
'S w
Each analysis is discrete.
Electrode must be rejuvenat
between each run by cleanir
with dilute nitric acid
Z
.EC ,
C j2 ,O i^. o
"co 0) '£ ^ 13 c
™ *" > w C O
C
o
c
JC
c
D
<;
z
Unknown
Z
z
§
01 C
SS .
No calibration or maintaina
requirements. Internal elect
ic standard demonstrates in>
strument integrity.
Z
ill*.
Hill
w
3 w n O
13 c u '5
» ° .E °
| ._ ^ 0 a
H > -« .E 2
'o
^ 0)
in ?
+t 2
0>
C
o
o
•a >• ^"
n o Q- 2
•S **" ° a
oflll
z
1
IA
O
o
u
o •«
I I
2 f
o ».
-Q o
o 13
E ~
s I
i s
i ?
I c
E 5
S £
1
I
S
a
5
f §
£ «
o .a
s c
•i •-
>- c
11 I
t
1
a
3
a
a.
* ~
c c
E 3
I sl
i il i
q "i
z z
— CM
13
-------
SRM-HOUSED SYSTEMS
General
The Schneider Robot Monitor (SRM) is a modular instrument system
with three stacked sliding drawer cubicles (Figure 2). Sensors,
flow cells, flow controlled valves (except the flow interrupt
assembly for hardness and nitrate), inlet reservoirs, an outlet
header, all interconnecting tubing and fittings, a splash pro-
tected terminal board for sensor connections, and a high pres-
sure header and tubing designed for automatic sensor cleaning
are housed in the bottom cubicle. (To facilitate visual inspec-
tion of the entire flow system, all tubing is transparent plastic,
and flow cells, reservoirs and sample headers, machined from
solid blocks of Plexiglas, are polished to be transparent and
clear. There are no metal plumbing components in contact with
the sample stream.)
SCHNEIDER
ROBOT
MONITOR
FROM
COMPUTE*
MUITI POINT
RECORDER
MODULE
ELECTRONICS
MODULE
SENSOR
FLOW CELL
MODULE
TO COMPUTER
DRAIN
Figure 2. Schematic arrangement of Schneider Robot Monitor.
14
-------
Flow through the cubicle is illustrated in Figure 3. The sample
stream enters the cubicle through the inlet reservoir and is
distributed evenly to all flow cells. To ensure adequate mixing
and to facilitate cleaning, flow cells are funnel-shaped at the
inside bottom. They are capped at the top by neoprene stoppers
through which the sensor assemblies are inserted. Each flow
cell is provided with an air bleeder hole to prevent air locking.
The effluent from all flow cells is collected by the outlet
header and discharged to a drain through a common effluent line.
Each individual SRM-housed sensor except temperature is installed
in a separate flow cell. For convenience in calibrating the
dissolved oxygen electrode, the temperature sensor is installed
as part of the DO electrode assembly. In the prototype system,
six flow cells are engaged and two positions open, as shown
schematically in Figure 4.
Electronics
All sensor systems are wired to their respective electronic
circuitry (analyzers) through the electrode terminal board in
the flow cell cubicle. This board consists of 37 barrier strip
terminals and four steatite insulated terminals for coax cable
tie points. Individual sensor assemblies are provided with
appropriate leads or terminals attached for lead connection.
The eight plug-in type modular electronic analyzers are housed
in the middle cubicle. The panel-rack assembly is slide mounted
as two separate drawers for easy access. The panel through
which power is supplied to the analyzers and a test signal
supply for the recorder are located in the analyzer cubicle.
A solid state operational amplifier is the active element in
each analyzer system. This type of circuitry is particularly
useful for stable amplification of small DC voltages and per-
mits the use of ancillary items for temperature compensation,
linearization of sensor characteristics, etc. A built-in cali-
bration check and moisture sealed panel meter are provided with
each analyzer module. Panel meters have individually calibrated
hand drafted scales with an accuracy of ;+0.5 percent.
Outputs from the electrometers are used as inputs to a second
stage operational amplifier. Offset is introduced in the
second stage to give a proper zero scale, and provision is made
for handling either polarity or measuring electrode input.
Temperature compensation with a wide range of adjustment is
accomplished with a thermistor in the input or feedback circuits
of a third stage operational amplifier.
15
-------
Out
Sensor Cleaning
Network
n
Outlet
Header
Flow Control Valve
Inlet Header
Flow Cells
Front
To flow cell
From flow cell
Downward flow
Note: 4 to 8 flow
cells may be
used in the
cubicle.
Figure 3. Schematic diagram of the SRM flow cell cubicle,
top view.
16
-------
s S
liljlii I
j I i i 11 [_ J^yM -±.4=f $
Of
o
h-
Z
0
H-
o
co
O
Of
Of
111
Q
111
z
I
u
«/>
r
L
i
L
r. r
$
^
u.
( l_- .. - -v|r- - ^£ -J<«
1 ^--^.-.^
•^ -Tp r^
!_~~ . _x"__ i r
1
r"^~ 1
i"
1
1
1 ' i
j
i 1 1
i 4 1
L._, L. L_,
1 I
_l 1
1 I '
in ' ^
C I
0) '_
g1
1
^H CN
c
(U 1 'r,
Oj I — °
0 *
\
L
Ul _j Ul
2 $
._Zj 1 i fi
1 r1
ji fO
[ J j 1 tn •
T T f * ^ F
! j i |jj 1 -H dJ
I U """1 ' § 1 -H M
r1 1 | 1 *i S^1
iii i Si **
III 1 7 U (1)
o l o tn
t; i co
1-r. "J , 0) ^
i j i in I tn i
-J J ' J «w §
1^1 ow
1 ' ' (-•
III on
j i f 4 --4
'—, ' ! ! ^a
1 -P e
._ _J i tn QJ
1 ! si
I.I rH (0
1 | .J SJ
r~ • oo V-i
i s a"
Q 1" 2 ^ . .
" •- O — (0 rH
I u •- S (0
fl) SJ
1 -• X! Q)
S r^ n
ro *" "^ *
I Q U, ^*
a. "~~ _
< « ^
I O ^
-------
Recorder
Sensor output from the SRM-housed systems is displayed (in
addition to the analyzer panel meters) on an Esterline Angus
24-point servo recorder (Model E1124E), located in the upper
cubicle. The recorder has a 50 millivolt range and is used
with 60 seconds per point printing speed and one inch per hour
chart drive.
Sensors
The five original SRM sensors - temperature, conductivity, pH,
dissolved oxygen, and chloride - are operated and maintained in
conformance with the manufacturer's instructions contained in
manuals supplied with the mobile laboratory. They sense the
species of interest, generally without interference from any
other species which may be present in the test water. No rea-
gents are added for their normal operation, but standard solu-
tions are required weekly for calibration. In field operation,
no unusual problems were observed with sensors for temperature,
pH, and chloride. Very occasional erratic readout or discontin-
uities in strip chart data from the conductivity system were
caused by bubble entrapment on the sensor surface. The dis-
solved oxygen sensor is flow sensitive. Short-term noncumulative
drift in output was observed occasionally when flows dropped to
critically low levels. In general, one gpm flow at 20 psi is
maintained past the electrode. Long-term cumulative drifts in
calibration indicate degeneration of electrode response and the
need for rejuvenation.
Unique problems were addressed in developing operational systems
for hardness and nitrate. Like the DO membrane electrode, the
hardness and nitrate probes are flow sensitive; however, they
produce stable output only under quiescent conditions. They must
be c.cmp£e.£e£t/ isolated from hydrodynamic effects of the sample
stream. A flow interrupt system, including chassis, power sup-
ply, and relay for switching AC voltage to a three-way solenoid
valve was installed in the mobile laboratory. Test water passing
through the valve goes either to waste (during interruption of
flow) or to exchange the sample in the hardness and nitrate flow
cells. The relay power assembly is activated by digital output
from the computer through the program "FLOW," which sets a digital
output bit high, waits 15 minutes, sets the bit off, waits 45
minutes, then begins the cycle again; i.e., test water flows
through the sample cell for 15 minutes, then the flow cell is
quiescent and the sample completely isolated from test water
flowing to other sample cells for 45 minutes. During this in-
terval, electrode response equilibrates and valid measurements
of hardness and nitrate are recorded just prior to the time when
samples are exchanged by new flow into the cells.
18
-------
In addition to being flow sensitive, hardness and nitrate
electrodes are, as a result of their design and construction,
extremely subject to bubble entrapment and accumulation of
static electricity. Bubbles are easily trapped in the concave
bottom configuration of Orion liquid junction electrodes. To
overcome this effect, box-shaped bubble traps constructed from
heavy gauge, coarse-mesh stainless steel screen are attached
to the hardness and nitrate electrodes by a plastic ring which
fits tightly over the outside of the electrode body. The bottom
of the trap is covered with a flat sheet of plastic to keep
bubbles from reaching the electrode from the bottom, and the
entire trap assembly is filled with glass beads into which the
electrode is seated. The glass beads effectively prevent any
movement of bubbles to the electrode while allowing completely
free movement of test solution to the sensing surface.
An alternate electrode design (marketed by Corning) was also
evaluated. Its flat bottom surface was an attractive feature,
but exceptionally high impedance made it totally impractical for
use in the prototype system. Although the Orion 92 series
liquid junction electrodes have a tendency to accumulate static
electrical charge, a new design with lower impedance and im-
proved shielding has been announced. It is strongly recommended
that, when they are available, the feasibility of substituting
new liquid junction series electrodes in the mobile laboratory
be evaluated by EPA.
The nitrate liquid junction ion-selective electrode is expected
to give valid analytical_results at nitrate activity somewhat
less than ten mg/1 of NO~; however, chloride and bicarbonate ions
interfere when they exceed the nitrate concentration by a factor
of ten. The nitrate probe is calibrated over two decades of
nitrate activity; i.e., 1.0 mg/1 at the bottom of the scale,
ten mg/1 at midscale, and 100 mg/1 of NO! activity at full scale.
Most United States tap waters contain less than 1.0 mg/1, and
normal response of the electrode is near the bottom of the scale.
As nitrate concentration in the test water increases, electrode
response is proportional. If the nitrate concentration should
increase to a dangerous level (e.g., 45 mg/1), the electrode
can be expected to track the change with precision. Because of
the logarithmic response of the probe, good precision is attain-
able at very low as well as high levels.
Preparation of reagents used with SRM-housed systems is described
in Appendix B.
Interpretation of Hardness and Nitrate Data
In interpreting hardness and nitrate data, activity coefficients
and complexation are important considerations. Physical charac-
teristics of the liquid junction ion-selective electrode make it
19
-------
difficult to design a sample flow cell without imposing pres-
sure problems on the electrode; thus, it is impossible to add
ionic strength buffers and masking agents vis-a-vis the fluoride
monitoring system. (Addition of an ionic strength buffer would
provide for measuring standards and sample at constant ionic
strength giving direct measurement of ionic concentration;
masking agents would free bound Ca2 and Mg2 from sulfate and
bicarbonate for direct sensing by the electrode.) Electrode
measurements are adjusted by applying constant factors, deter-
mined experimentally.
Two corrections must be applied: one for the activity coeffi-
cient and the second, for that percentage of the total amount
of Ca2 and Mg2 which may be complexed. If the two factors are
constant over the monitoring period, the equation for the cor-
rection of free uncomplexed species activity to total concentra-
tion is:
,+ (M' + )F
[M2+]T = L_ (7)
fM2+ «M2 +
where:
(M2 ) = the monitored value
fM2+ = the partition coefficient or the
fraction of the divalent metal ion
concentration which is unbound
aM2+ = the average activity coefficient
for divalent metal ions in the
monitor stream
Another factor to be considered in operating the hardness moni-
toring system is the effect of the flow interrupt device, which
alternately causes flow/no flow in the system. Figure 5 is an
example of typical strip chart recorder output for hardness data.
Measurement points on half of the output are connected to better
illustrate the nature of electrode response to the flow inter-
rupt system. During the flow portion of the cycle, response of
the electrode is erratic or noisy; when flow is stopped, the
signal becomes quiet and quite stable. Hardness is measured
after the signal has stabilized.
On the strip chart recording, shown in Figure 5, the divalent
cation electrode is calibrated to read from 3.16 x lO^M CaCl2
to 3.16 x 10"3M CaCl2 over the 12 inch scale. This places
1.0 x 10~3M CaCl2, or 100 mg/1 of equivalent CaC03/ at exactly
midscale (60.0 recorder chart units). To properly analyze the
data, electrode and EDTA titration data are compared to determine
the fraction of hardness which is free to be sensed by the
electrode. The complete procedure for converting strip chart
recorder output for free divalent cation activity to total
hardness concentration is as follows:
20
-------
Figure 5.' Typical strip chart recorder output of' hardness*data.
21
-------
1. Determine the free divalent cation
activity by comparing chart recorder
output to a calibration curve.
2. Use the nomogram provided by the
Orion Company (Figure 6) to convert
activity to concentration measurements
a. Determine free activity
b. Determine conductivity
c. Place a straightedge along
a line through the free hardness
activity and the measured speci-
fic conductance and read the per-
cent correction for total hardness
concentration where the straight-
edge crosses the third line.
d. Correct the free hardness activity
to free hardness concentration:
[Me2*] . . , = (Me2 + ), + C. (Me2 + )c (8)
electrode free f free
3. Determine the fraction of the total hard-
ness concentration which is complexed and
unavailable to be sensed by the divalent
cation electrode.
[Me ]electrode f . .
[Me2+]EDTA
4. Determine the total hardness concentration:
, ,e , ,,nx
free fv f ree \ (10)
total ~| f J
Example:
1. Free divalent cation activity = 100 mg/1
equiv CaCOs .
2. a. Free activity = 100 mg/1 equiv CaCOs.
b. Conductivity = 320 yu/cm.
c. Cf = 5.5 percent or 0.055.
d. [Me2+] = (100 mg/l)+0.55(100 mg/1)
= 100 mg/1 +5.5 mg/1
= 106 mg/1
f
r =
electrode 106 mg/1
- = - •" — =
[Me ]EDTA
22
-------
30
1000
900
800
700
600
500
400
500
ZOO
100
90
80
70
60
50
40
30
20
10
8
6
4
2
1
total hardness
— activity
ppm CaCO3
—
-
-
-
-
-
—
-^
— *-.
specific
conductivity
micromhos (25°C)
10000--
900O--
8000--
7000--
60OO--
50OO--
4OOO--
3000--
2000--
1000-
900-
800-
700-
^600-
400-
300-
200-
100-
50-
10-
I-
% correction
for total hardness
concentration
50
48
46
44
42
4O
38
36
34
32
3O
28
26
24
22
20
18
16
14
12
' 10
8
6
4
2
0
Figure 6. Conductivity correction nomogram.
23
-------
4 rMe2+l = ! (100 mg/l)+.055(100 mg/1)
total ( 0.82
= 129 mg/1 of equiv CaCOa
Nitrate measurements are not expected to require the degree of
correction described for the hardness system. Nitrate is not
found in a complexed state and its concentration usually need
not be known with great accuracy, unless it is present near
the maximum permissible level specified in the Drinking Water
Standards (1). Unfortuantely, NSF cannot present any actual
field data obtained from the nitrate sensing system; i.e., the
mobile laboratory was not operated in an area where nitrate
activity exceeded the detection limit of the sensor.
COMMERCIAL SYSTEMS
FREE AND TOTAL RESIDUAL CHLORINE
General
Residual chlorine is the single most important parameter in
drinking water quality. Its presence in a finished water is
generally considered to indicate that disinfection is complete,
and the water is safe to drink.
At the water treatment plant, chlorine application is highly
mechanized and carefully controlled. It must be added to the
treated water at a level sufficient to accomplish disinfection
but not sufficient to incite taste and odor complaints from the
consumer. A total residual of 1.0 mg/1 in water leaving the
treatment plant is generally assumed to meet these objectives.
In traveling through the distribution system, it is entirely
possible that the level and type of chlorine residual will
change. These changes may occur as a function of other parameters
which characterize the quality of the water itself, or as a re-
sult of conditions within the transmission pipeline. It is
therefore desirable, if not necessary, that residual chlorine
levels be monitored in the distribution system.
Theory
Regardless of the form in which chlorine is added, it reacts
with the water to form hypochlorous acid or hypochlorite ion:
C12 + H20 *==* HOC1 + H+ + Cl~ (11)
gaseous hypo-
chlorine chlorous
acid
24
-------
Ca(OCl)2 + H20 S5=5 Ca2 + + 2 OC1~ + H20 (12)
calcium hypo-
hypo- chlorite
chlorite ion
NaOCl + HZ0 ^=5= Na+ + OC1~ + H20 (13)
sodium
hypo-
chlorite
The equilibrium condition of HOC1 and OC1 is described by
Equation 14:
HOC1 ^==B OC1~ + H+ (14)
K. = 2.5 x 10~8
i
Thus, the ratio of HOC1 and OC1 present following disinfection
is determined by pH of the treated water:
log [OC1"] = PH + log K U5)
[HOC1]
The effect of pH on [HOC1]: [OC1~] species distribution is illus-
trated graphically in Figure 7. From this figure, it is apparent
that a mixture of the two species exists in most United States
drinking water supplies. In either form, it is referred to as
"free residual chlorine."
The term "total residual chlorine" includes both free and
"combined" forms. The most common combined chlorines are mono-
chloramine and dichloramine, found in the presence of ammonia:
NH3 + HOCl -> NH2 Cl + H20 (16)
mono-
chlor-
amine
NH2C1 + HOCl -> NHC12 + H20 (17)
dichlor-
amine
Trichloramine occurs in tap water only infrequently; e.g., with
extreme pH or breakpoint chlorination:
NH2C1 + HOCl -> NC13 + H20 (18)
tri-
chlor-
amine
25
-------
90
10 11
100
Figure 7. Distribution of hypochlorous acid and hypochlorite
ion in water at different pH values and tempera-
tures (2) .
The rate at which the various forms of combined chlorine achieve
disinfection is significantly less than that of free chlorine.
Nine analytical methods for measuring residual chlorine are
described in Standard Methods (3). One of these, the amperometric
technique, is used in the mobile laboratory in both the free
and total residual chlorine analyzers, and in the manual titrator
used to standardize the analyzers.
In the amperometric procedure, a galvanic cell is developed;
i.e., a current flows between immersed electrodes when a free
oxyhalogen acid; e.g., HOCl (hypochlorous acid), is present in
the test water. Null point amperometric titration describes the
principle of this method. Residual chlorine is quantitatively
removed from solution by reaction with a titrated reducing agent;
e.g., phenylarsine oxide. As reducing agent is added, current
resulting from the presence of HOCl in the test water is reduced.
Titrant is added until there is no further decrease in cell
current. The titrant required to reach this end point (i.e.,
no further current decrease) is directly proportional to the
level of residual chlorine present in the water.
26
-------
Equation 19 describes the principle of electrolytic reduction of
an oxyhalogen acid (HOX) at an inert cathode in the cell:
Cu°|Cu2+||x~JHOX, Au (or Pt) (19)
(or Cu°|Cu2+[|C1~|HOC1, Au (or Pt)) (20)
Half reactions for Equation 20 include:
Anodic: Cu° = Cu2+ + 2e (21)
E° = -0.337 volts
a
Cathodic: HOC1 + H+ + 2e = Cl~ + 2H2O (22)
E° = 1.28 volts
c
Combining Equations 21 and 22, the overall reaction is:
Cu° + HOC1 + H+ = Cl~ + Cu2+ + H20 (23)
and the standard cell emf will be,
E° = E° + E° = 0.94 volts (24)
CG-L X 3. C
Consequently, the cell potential at equilibrium, at 25°C, will
be' 2 +
E = 0.94 + 0.03 log -^ ] [C1 3 (25)
Celi [HOC1][H+]
Thus, cell emf is a function of [Cu2], [HOC1], and [H+]. The
potential of the large surface area (nonpolarizable) copper anode
is a constant (as [Cu2 ] in the bulk of the test solution approaches
zero) and is determined by the rate of dissolution of copper from
its surface. In practice, [H ] is kept constant by the addition
of pH 4.5 sodium acetate/acetic acid buffer. Cell emf, then, is
determined by - and varies with - the level of HOC1 in the testwater,
Experimental
Presently there are at least four amperometric-type residual
chlorine analyzers which are commercially available. In two of
the four systems, potential is measured across an extremely high
resistance (rather than current flowing in the galvanic cell, mea-
sured across a low, fixed resistance). Three of the four systems
use gold/copper electrode couples; the fourth, a platinum/copper
couple. The gold or platinum electrode serves simply as an inert
surface at which the HOX is reduced; thus, the theory of operation
of each of the four instruments is essentially the same. Cell
27
-------
geometry for each of the systems includes a. large surface area
copper cylinder (the anode), surrounding a wire or rod of the
noble metal (cathode).
Cleanliness of the anode is vitally important. Fouling causes
surface area variations and large drifts in potential over short
time intervals. As a result, electrode cleaning devices are
included with each instrument. One of the systems has the cell
filled with plastic balls. A motorized arm drives the balls in
the cell. Two of the units are filled with plastic pellets
which abrade a rotating electrode. In the fourth unit, grit
in the sample cleans the electrodes, agitated by velocity of
the flowing sample stream.
It is also important to maintain constant ionic strength in
the sample stream. Significant changes in total ionic strength
(measured as total dissolved solids or as conductivity) result
in HOX activity coefficient variations without changes in analy-
tical concentration. The high strength sodium acetate/acetic
acid buffer adjusts ionic strength in addition to controlling pH.
Signal variations resulting from temperature effects on cell emf
require compensation. A typical galvanic residual chlorine
analyzer is illustrated schematically in Figure 8. In this
figure, R2 is a variable resistance in parallel with the sensing
electrode. It can be manually adjustable (assuming that tempera-
ture of the sample stream remains constant between calibrations),
or a temperature dependent transducer (thermistor) can be used
in which resistance varies automatically with changes in temper-
ature.
Total residual chlorine is measured by adding potassium iodide
to free combined residuals:
NH2C1 + I~-*—r NH2I + Cl" (26)
mono- mono-
chlor- iod-
amine amine
NHC12 + 2I~^=?NHI2 + 2C1~ (27)
dichlor-
amine
Because iodamines are highly unstable in aqueous solutions, the
reactions continue to produce hypoiodous acid, an oxyhalogen
acid which is sensed by the galvanic cell:
NH2I + H2O^=^NH3 + HOI (28)
hypo-
iodous
acid
NHI2 + H2Oi=;NH2I + HOI (29)
28
-------
MEASUREMENT CIRCUITRY
POTENTIOMETRY
HIGH R, R4
r
I6 LOWR, R3
A/WV-
H-)
CU
GALVANOMETRY
AU
TEMPERATURE COMPENSATION
-AUTOMATIC (THERMISTOR)
-MANUAL
CALIBRATION RESISTANCE
Figure 8. Schematic illustration of residual chlorine
analyzer.
29
-------
Two Capital Controls residual chlorine analyzers, Models 871
for free residual and 872 for total residual, and a Wallace and
Tiernan (W&T) amperometric titrator, Series A-790, are used in
the mobile laboratory. The Capital Controls analyzers are wall
mounted galvanic units, operated continuously. The electrodes
are cleaned by 200 polyvinylchloride (PVC) balls rotating in
the test stream. A schematic illustration of these units is
shown in Figure 9. The manual W&T titrator is used for daily
calibration of the continuous analyzers.
Each continuous analyzer has a gold/copper couple. Chlorine is
the active electrolyte and produces a current measured with a
microammeter. Reagents are introduced to the analytical cell
by a cam feeder which operates every ten seconds. Reagent
reservoirs are two liter bottles with sufficient capacity for
approximately seven days. In both units, a buffer addition
adjusts pH of the test water to 4.5, and adjusts ionic strength.
The free chlorine analyzer responds directly to free chlorine
in the sample stream. A second reagent, potassium iodide, is
added in the total chlorine analyzer as a masking agent to con-
vert combined to free forms which can be sensed by the electrode,
Preparation of reagents used for monitoring residual chlorine
is described in Appendix B.
ROTATING
STRIKER
CYLINDRICAL
COPPER
ELECTRODE
DRAIN
SAMPLE
INLET
SCREEN
ELECTRIC
SIGNAL
TO
MONITOR
OVERFLOW
TO DRAIN
GOLD
ELECTRODE
Figure 9. Schematic illustration of C12 analyzer,
30
-------
Results and Discussion
During more than three and a half months of continuous field
operation and up to three years of intermittent operation in the
laboratory, the residual chlorine analyzers operated with com-
plete satisfaction. Maintenance was limited to routine weekly
preparation of buffer and masking agent reagents and daily cali-
bration.
In Chicago, the first two monitoring sites were on heavily used
distribution mains where residence time was normally on the
order of 15 to 20 hours, and total chlorine residuals, consis-
tently in the range of 0.7 to 0.9 mg/1. Each of these sites was
approximately ten miles from the filtration plant. At the
third and fourth monitoring sites, Des Plaines and Calumet
Harbor, the unit was located far out, near the ends of the distri-
bution systems. At Des Plaines, the water had been transported
nearly 20 miles through main transmission lines and the residence
time was unknown. Free chlorine averaged 0.14 mg/1 and total
chlorine, 0.28 mg/1. At Calumet Harbor, the monitoring site was
only eight miles from the water treatment plant, but near the
end of a large main used mostly for industrial fire flows. As
a result, residence time in the main was approximately 35 hours.
Free chlorine at this site averaged 0.24 mg/1 and total, 0.40
mg/1 over a nine-day period of observation.
Changes in residual chlorine levels during a main flush at the
Calumet Harbor site are described on pages 164 and 165.
At Philadelphia International Airport, there was opportunity to
evaluate analyzer response to ammoniated samples. Free chlorine
ranged from "not detectable" to 0.1 mg/1, and was generally
<0.05 mg/1. With no apparent difficulty, the free chlorine
analyzer differentiated between free chlorine and loosely com-
bined chloramines. (At the operating pH; i.e., 4.5, the pos-
sibility of converting some combined forms to free chlorine
was questioned (4).) This observation was confirmed by ampero-
metric titration at pH seven.
At the fourth location in Philadelphia, the Gulf Oil Refinery,
sudden variations in chlorine residual were observed. A portion
of the recorder output from this site is shown in Figure 10.
The capacity of these analyzers to detect sudden changes in
both increasing and decreasing concentrations is clear from the
output, confirmed by titration at the time of occurrence. The
variations were attributed to the mixing of water from two dif-
ferent storage locations.
31
-------
iTxtrations
support these
data
Titrations
support these -
data __
~T> ~-j" "
L_4_J__u
—1_
T
1 i , : . , ,
f~~~ 1 Total
5-1 Residual —
i Chlorine i
11—i—f—- -
I y- I
1"D*1—
Figure 10.
Free and total residual chlorine variations
at Gulf Oil, Philadelphia.
32
-------
TURBIDITY
General
Turbidity is defined as, "an expression of the optical property
of a sample which causes light to be scattered and absorbed
rather than transmitted in straight lines through the sample"
(5). It is caused by the presence of suspended matter in the
sample stream. The American Water Works Association has developed
a goal of 0.1 Jackson Turbidity Units (JTU) for drinking water
(6); the EPA standard is five JTU or less (1).
Experimental
The Hach CR low range turbidimeter, Model 1720, continuous flow
nephelometer is used to measure turbidity in the mobile labora-
tory. The measurement is accomplished by passing a strong beam
of light through the sample. Turbidity, present as fine parti-
cles, scatters a portion of the light beam which is measured by
two photocells submerged in the sample. If the sample is free
of turbidity, no light is scattered and no light reaches the
measuring cells; conversely, the presence of turbidity in the
sample results in light being scattered. Some of this light
falls on the photocells, and is registered on the readout meter.
The meter is calibrated with a reflectance rod of 5.0 JTU.
Schematic illustrations of the turbidity system and the turbidi-
meter in the mobile laboratory are shown in Figures 11 and 12.
This system was essentially trouble free throughout its use in
the NSF project. Data acquired routinely during field assign-
ments are reported in the respective reports of these activities
(7,8,9) .
CORROSION RATE
General
Corrosion is a serious-or potentially serious-problem in most
potable water distribution systems. Taste, odor, and appearance
problems resulting from corrosion of transmission lines may make
water unacceptable to the consumer, and corrosion induced pipe-
line perforations may permit back siphonage of contaminated
groundwater into the system.
A major investment of the water supply industry is the distri-
bution network. Continuous effort to produce a water with no
tendency toward pipeline corrosion is a principal treatment
objective.
33
-------
REMOVABLE HEAD
(CONTAINING ENTIRE
ELECTRICAL ASSEMBLY)
TO DRAIN
SAMPLE DRAIN
TO MASTER INDICATOR
113V-COCY IN
HOTOCELLS
PARTICLES OF SUSPENDED
MATTER REFLECT LIGHT
WHICH IS MEASURED BY
THE PHOTOCELLS.
STANDARD
REFLECTANCE
ROD (SHOWN IN PLACE)
DRAIN PLUG
SAMPLE INLET
Figure 11. Schematic illustration of turbidity system.
34
-------
*
o
-U
m
n
o
X!
m
'"I
X
Q
i
(U
X!
-p
G
•H
(1)
-P
(U
g
•H
T3
-H
XI
M
o
-H
-p
(0
o
—I
r-t
rH
•H
O
•H
-P
(0
e
0)
X!
O
CO
O
GO
3
o.
O
O
CN
H
(U
Cn
35
-------
Theory
Electrode potentials, expressed as intensity measurements of
reversible equilibrium conditions, and the Butler equation for
electrochemical kinetics, are basic considerations in describing
the current-potential relationships associated with corrosion.
It is assumed that equilibrium conditions are maintained in the
thermodynamic development of electrode potentials; i.e., there
are both charge balance and mass balance in electrode reactions
for a metal (electrode) in a solution of its ions:
M fe^ M2+ + 2e~ (30)
Thus, at equilibrium, the same velocity occurs regardless of
the direction of the reaction. The electrode potential is
determined by the metal ion activity, expressed by the Nernstian
relationship:
RT [0x]
E = E° + — In —2 (31)
2F [Red]
where:
E = measured electrode potential
E° = standard electrode potential
In corrosion, however, the anodic and cathodic reactions are
usually different. Although there is charge balance, there is
also mass flux. The anodic mass flux is, in fact, corrosion.
A typical corrosion cell is shown in the following series of
equations:
Anodic: Me -> Me2+ + 2e~ (32)
Cathodic: 2H+ + 2e~ + H2 (33)
or
1/2O2 + H20 + 2e •+ 40H (34)
Thus, there is a net mass flux, hydrogen gas evolves from the
system, and corrosion products are formed from metal dissolution.
The metal is not at equilibrium potential (as it is in Equation
30). It exists in transient or steady state with its potential
a function of the rate determining step in the reaction. If the
rate determining step is the anodic process, the potential of
the metal shifts in a positive direction; if it is cathodic,
excess electrons will accumulate on the metal, causing a negative
shift in potential. Because of the kinetic effect and the
possibility that another cathodic reactant may be available, a
corroding metal surface in neutral solutions rarely acquires
equilibrium potential values.
36
-------
In addition to the energetics of corrosion; i.e., the tendency
of a reaction to proceed as measured by its thermodynamic pro-
perties, reaction kinetics are vitally important considerations.
The basic equation for electrochemical kinetics is expressed as:
RT • (35)
where :
i = polarization current density
ip = exchange current density
a° = charge transfer coefficient
n = polarization overvoltage
The overvoltage is defined by:
n = E. - E (36)
i eq
where :
E. = electrode potential with an impressed
1 current, i, flowing through the cell
E = electrode equilibrium potential (with
^ no external current flowing through the
cell)
The exchange current, i , represents the spontaneous rates at
which the anodic and cathodic reactions are proceeding. Because
these are not identical for corrosion conditions, i indicates
the rate of metal dissolution for the anodic reaction and the
rate of the cathodic reaction, whatever it might be.
Thus, i is an exact measure of the rate of reaction if the true
current°density relationship is known. It is difficult to mea-
sure because true surface area is not easy to determine. It
is usually known within a factor of two, so use of o.ppo.A.en;t
surface area in calculating i does not invalidate the results.
The effect of major current excursions on naturally corroding
metal surfaces is important in measuring corrosion rate by
polarization techniques. Once an electrode has been used for a
complete polarization curve, its surface has changed irreversibly
"from its original freely corroding condition. If, however, only
small increments of current or potential are applied to the
electrode surface , the perturbation to the naturally occurring
corrosion rate is small and might not affect the long term
corrosion rate of the sample being measured.
Thus, the overvoltage of a reversible electrode reaction is a
linear function of the applied current density for values of
overvoltage only slightly different from the reversible potential.
37
-------
This relationship can be developed from Equation 31, using, for
small values of x, the approximation:
x
(37)
For small values of overvoltage, n/ Equation 35 can be rewritten
as:
2*nl - fn- (1-a)%
RT J L RT
(38)
. F_
""0 RT
T
I di
L
l
RT
i
~
i
(39)
(40)
n+o -0
From the empirical Tafel relationship:
n = a + b log i
(41)
where:
a and b = constants for a particular system
i = i + i
p a c
Equation 40 becomes:
'dn 1
b b
a c
2.3 i
(42)
where:
subscript "a" = anodic Tafel slope for b
subscript "c" = cathodic Tafel slope for b
The differential dri/di is the polarization resistance, expressed
as ohms. In corrosion measurements, the term i (the exchange
current for reversible systems) is replaced by icorr- The cor-
rosion current, icorr/ per unit area represents the rate of the
corrosion reaction and is directly proportional to weight loss
(the classical method for measuring corrosion).
Simplifying Equation 42:
AE _
Ai 2.3(b.
babc
V
(43)
38
-------
or,
b b
log AE = lQg - a_^ -- lQg . (44)
AI 2.3(b + b ) corr
a c
Plotting AE/AI versus log i produces a slope equal to -1.
When the cathodic reaction is controlled by diffusion, b is
approximated by infinity, and Equation 44 expressed as:
(45)
AI 2.3i
corr
or
= . (46)
C0rr 2.3 AE
Experimental
A corrosion rate instrument, capable of monitoring the instan-
taneous corrosion rate of metals in their use-environment, is
installed in the mobile laboratory (Figure 13). This instrument,
manufactured by the Petrolite Division of Petreco Corporation, uses
a "polarization of adjusted known surface area admittance" techni-
que with three identical electrodes machined from the material of
interest. In the measurement cycle, a current is imposed between
the working electrode and an auxiliary counter electrode sufficient
to polarize the test electrode ten millivolts with respect to
the reference (third) electrode. The magnitude of the current
required to effect the ten mv polarization is directly propor-
tional to the corrosion rate, and is interpreted as mils per year
of surface corrosion. In use, this instrument is set up with
three active stations. Station one is an internal calibration
device used to monitor correct operation of the instrument.
Stations two and three are used for actual corrosion rate mea-
surements. For station two, a set of freshly polished electrodes
is installed in a flow cell, and corrosion rate measurements are
initiated immediately. Data obtained with these electrodes over
several days or weeks of monitoring are used to provide informa-
tion about the rate of attainment of corrosion equilibrium in
the test medium. For comparison, a second set of electrodes is
sent ahead to a new monitoring site for advance exposure (passi-
vated) to the test water. When the laboratory arrives, these
electrodes are installed in the test cell to provide comparative
equilibrium corrosion rate measurements.
39
-------
CORROSION RATE MONITOR
DRAIN
TO COMPUTER
MEASURING
ELECTRONICS
STRIP- CHART
RECORDER
SENSOR
SWITCHING
RELAYS
SENSOR
FLOW
CELLS
FROM COMPUTER
TAP
WATER
SAMPLE
Figure 13. Schematic illustration of corrosion rate monitor in
the mobile laboratory.
Results and Discussion
Mechanically, the corrosion rate monitor proved to be a trouble-
free system in Chicago; however, significance of the Des Plaines
data is questionable, because the electrodes were not passivated
in the test waters.
The time required for passivation is a function of corrosion and
deposition characteristics of the test water. In Grand Rapids,
Michigan, for example, where water is heavily scale forming,
passivation is accomplished in a matter of hours. In Ann Arbor,
where water is corrosive, months are required.
New (repolished) electrodes were used at three of the four
Chicago area test sites. Moving the van on a weekly basis (three
of the four locations) precluded passivating and acquiring long-
term data with the same electrode systems. In preparing for the
Philadelphia trip, new electrodes were sent ahead for passiva-
tion in the water in which they were to be used. Because the
corrosion rate monitor is a four channel system, passivated and
new electrodes can be used simultaneously to develop both kinetic
and long-term corrosion rate data.
Unfortunately, the Philadelphia trip produced no meaningful cor-
rosion rate data. The Chicago data were meaningful; i.e., daily
mean corrosion rates at the first location ranged from 2.2 to
3.0 mils per year, and at the second location, 2.0 to 2.2 mils
40
-------
per year. These sites are served by water from the Central Water
Filtration Plant (CWFP), where caustic is added for corrosion
control. In Des Plaines, the daily mean ranged from 3.8 to 6.8
mils per year, reflecting different mix ratios of Chicago CWFP
and Des Plaines treated supplies. At Calumet Harbor, the daily
mean was 5.0 to 7.6 mils per year for water treated at Chicago's
South Water Filtration Plant (SWFP) where no corrosion inhibitors
are added in the treatment process.
These observations suggest that the Petrolite instrument has
good potential for application in the water supply industry;
however, improvement in its ruggedness and mechanical integrity
are recommended.
OTHER SYSTEMS
CALCIUM CARBONATE DEPOSITION TEST
General
Calcium carbonate (CaC03) deposition is an important consideration
in water utility operation. Excessive deposition in the trans-
mission pipeline reduces carrying capacity; conversely, corrosion
may occur when no protective film is formed. Either event re-
sults in serious economic losses to the community; thus, a prac-
tical objective of the water treatment operator is to produce
a finished water which will deposit a light protective film in
the distribution network.
Theory
Water which is either over or undersaturated with respect to
CaC03 tends to undergo chemical change and approach a state of
equilibrium. Thermodynamically , this tendency can be expressed
in terms of change of free energy (AG) , as shown in the following
reactions :
CaCO3 ,v + Ca2+ + CO32~
AG = RT ln -
K
s
CaC03/ x + H+ + Ca2+ + 2HCQ3
AG = RT m [HCOa"] (48)
Ks[H+]
41
-------
CaC03, v + H2C03* + Ca2+ + 2HC03
AG = RT in [Ca2+][HCQ3-]2K2 (4g)
KjK [H2C03*]
s
/
where :
KI = first dissociation constant for H2C03*
K2 = second dissociation constant for H2C03*
K = solubility product of CaC03
[H2C03*f = [H2C03]+[C02]
For oversaturated waters, AG is negative; for undersaturated
waters, AG is positive.
It is common practice to characterize the degree of departure
from calcium carbonate equilibrium conditions by calculating
the hydrogen ion activity at which, without change in total
alkalinity and calcium content, a water would be in equilibrium
with solid calcium carbonate (3); i.e.,
+ [Ca2+]act[HC°3~]act
[H+] = - Set - act (5Q)
q
The H activity of the test water is correlated with a hypothe-
tical value for H . When [H ] > [H ] , the water is aggres-
sive with respect to CaC03; when [H ] eq< [H ] , the water
will deposit CaC03. meas eq
Langelier (10) proposed using the difference between measured pH
and calculated equilibrium pH as a measure of "stability;" i.e.,
the tendency of a water to be scale forming or corrosive. He
defined the term "saturation index" (SI) as follows:
SI = pH - pH (51)
^ meas ^ eq
where pH is defined from Equation 50 as:
- log [HCO,'] (52)
The concentration of HC03 can be calculated from analytically
measurable parameters; e.g.:
[Alk] - Kw/rH+i + [H+]
[HC03~] = - - - I J - (53)
2K
[H+])
42
-------
and
[HC03~] = XiCT (54)
where:
C = the analytical concentration for total
carbonates
and
*r I HCO 3 I / r- r- \
Xi = (5~>)
CT
Within the pH range of many natural waters (7.5 to 9.0), Xi is
close to unity and [HC03~] - CT. Below pH 8.3,
[HC03~] = [Alk]T (56)
Larson and Buswell (11) added an ionic strength correction (u):
pH = log(K /K2) - log [Ca2+] . - log[Alk]_ + 9.30
6Q S ciC"C J.
2.5 (5?)
5.5y + 5.3/^7+1
Accordingly, from Equation 51, a positive saturation index indi-
cates a scale forming water; a negative index indicates a scale
dissolving water.
Another parameter, the "stability index" (S) , proposed by
Ryzner (12) , refers also to equilibrium pH:
S = 2PHeg - PH (58)
According to Ryzner, whose work was principally empirical and
based on weight changes of coupons in a test water, waters with
a stability index of <1 . 5 are scale forming; waters with indices
>7.5 are increasingly scale dissolving.
Calculation of the calcium carbonate stability index, using
either Equation 51 or 58, has been widely applied in the water
treatment industry; however, there are a number of limitations
to .this procedure. The calculated index provides information
about the extent of departure from equilibrium only; i.e. , the
directional tendency of a water to deposit or dissolve a calcium
carbonate film. Kinetic aspects of water stabilization or rates
of attainment of equilibrium are not considered; thus, the rate
43
-------
and amount of calcium carbonate deposited or dissolved cannot
be predicted. It has been demonstrated experimentally that,
in the absence of crystallization nuclei, oversaturated solu-
tions of calcium carbonate may remain oversaturated for years
(13,14,15). In addition, the presence of Mg2 , polyphosphates,
or dissolved organic matter (present in most water supplies) may
significantly influence the rate of calcium carbonate precipi-
tation or dissolution. This effect is not necessarily pH
dependent.
7 +
The effect of Mg on the rate of calcium carbonate nucleation
was studied by Pytkowicz (13). These studies indicate that
magnesium-free artificial seawater yields much shorter times of
nucleation (in order of minutes) than natural seawaters (in
the order of several hours). Enrichment of water samples with
magnesium was found to significantly decrease the rate of cal-
cium carbonate nucleation. This was explained as a result of
formation of magnesium carbonate complexes, which are highly
stable with higher alkalinities. These complexes are very
effective in inhibiting calcium carbonate crystal formation.
Polyphosphates and trace organics interfere with the rate of
calcium carbonate deposition because of their tendency to form
complexes. This causes a sequestering effect which keeps the
calcium ions in solution. This effect can be demonstrated by
comparing results of Ca2 determinations using a calcium ion-
selective electrode and the EDTA titration technique. The former
measures free Ca2 and the latter, total Ca2 concentration.
If EDTA titrations are greater than ion-selective electrode
measurements, all of the calcium present in the water supply is
not in the free ionic form. It is distributed between the free
form and a fraction which exists as solubilized complexes,
generally organo- or phospho- in nature.
McClanahan (16) studied the role of dissolved oxygen and the
mechanism of CaCOs film formation with respect to cast iron
corrosion using a three electrode system, the rotating ring
disc electrode.
In the iron corrosion process, the metal undergoes anodic dis-
solution at iron starved regions (Equation 59), and oxygen is
cathodically reduced to hydroxide at oxygen rich regions
(Equation 60). The occurrence of hydroxide raises the pH in
the vicinity of the corroding metal surface and may cause the
ion product of CaCOs to surpass the solubility limits, depositing
CaCOs on the corroding metal surface:
Fe, N -*- Fe2+, , + 2e~ (59)
(s) (aq)
1/2 O2 + H20 + 2e~ -> 2OH~ (60)
2Ca2+ + 2HC03~ + 20H~ -> 2CaC03 / v + 2H2O (61)
\ s)
44
-------
Thus, it is clear that a water which is under saturated with
CaCOs may deposit CaCOs regardless of its Langlier Index; i.e.,
in theory, any pipe material capable of supplying electrons may
cause CaC03 to deposit from waters which contain both calcium
and carbonate ions .
Rotating R-cng
The rotating ring disc electrode is a voltammetric system with
rigorously defined hydrodynamic transport. Consider a flat disc
rotating with uniform angular velocity about an axis perpendicu-
lar to the plane of the electrode. Because of the centrifugal
force acting upon it, the layer of adjacent solution adheres to
the disc and acquires a radial flow. The radial flow displace-
ment is accompanied by an axial flow from the bulk of the test
solution to the surface of the electrode ( 17) . Figure 14 is a
schematic illustration of fluid flow in the rotating electrode
system.
electrode
v
X
Figure 14. Fluid flow in the rotating electrode system.
Under constant rates of rotation, the thickness of the hydro-
dynamic boundary layer is a function of the angular velocity
of the rotating disc (w) and the kinematic viscosity of the test
solution (v) (18); thus, mass transport to the electrode surface
depends solely on the rate of rotation of the disc electrode.
45
-------
It is assumed, based on the Nernst diffusion layer theory, that
a major part of the concentration gradient occurs across a
stagnant liquid layer, the Nernst diffusion layer (S )(19).
The hydrodynamic boundary layer (SH) can be equal to or greater
than S .
n
The rate of mass transport; e.g., molecular oxygen in CCDT, in
the x direction, normal to the electrode surface, can be ex-
pressed as follows:
i = D!^-vxi (62)
where:
C = dissolved oxygen concentration
t = time
D = oxygen diffusivity coefficient
vx = axial liquid velocity
Under steady state conditions:
i£ = 0 (63)
at
and
vx9£=Di!£ (64)
9x ax2
If the rotating disc electrode is maintained at a potential at
which the following reaction will proceed:
02 + 2H2O + 4e~ + 40H~ (65)
the concentration profile shown in Figure 15 will prevail in
the vicinity of the electrode.
Because electrolytic reduction of molecular oxygen at the elec-
trode surface (Equation 65) is more rapid than oxygen mass trans-
port, the concentration of oxygen at the surface of the electrode
approaches zero (C = C ^ ). In addition, a linear concentration
gradient is assumed to exist within the Nernst diffusion layer
(Figure 15).
The current flowing in the electrode is proportional to the con-
centration gradient at the electrode surface ,.
(£)
X+o.
46
-------
'B
(X)
CB ~
Assumed linear
concentration gradient
Distance Normal to the Electrode Surface (x)
Figure 15. Concentration profile at electrode surface under
steady state conditions, in the absence of CaC03
film.
and can be expressed as:
where:
(66)
X->o
i = diffusion current
n = number of electrons transferred per mole
of reactant
F = the Faraday constant
D = diffusion coefficient
C = concentration of electroactive species
X = distance to the electrode surface
Under steady state conditions:
/3CN
C -C
B o
(67)
N
47
-------
where :
CR = bulk concentration
C = concentration at the electrode surface
6 = diffusion layer thickness
If C =0, the limiting current is:
CR
i0 = nFAD -*£ (68)
* °N
The thickness of the diffusion layer can be estimated from (64) :
6N = kD1/3oT1/2v1/6 (69)
Combining Equations 68 and 69:
i = K nFAD2//3co1//2v~1//6C0 (70)
B
Thus, according to Equation 70, the current is limited by dif-
fusion in an inter facial Nernst layer, the thickness of which
is solely dependent on the angular velocity of the rotating disc.
The theory of rotating ring disc electrodes is discussed in much
greater detail in the literature (18,20,21,22).
Ca£.Cxcam Carbonate, f-iim fotimat-ion
The steady state current for the electrolytic reduction of dis-
solved oxygen at the surface of the rotating disc electrode re-
mains constant with respect to time, as shown in Curve 1, Figure
16. This is true if P , temperature, and the speed of rota-
tion are constant, and 2the test solution contains no Ca2 . The
oxygen limiting current decreases with time in the presence of
Ca2 , and the rate of decrease is directly proportional to the
concentration of Ca2 , as shown in Curves 2 through 5, Figure 16.
This decrease in current is attributed to the formation of CaC03
film on the surface of the electrode, according to the following
reactions :
02 + 2H20 + 4e~ -*• 40H~ (71)
OH~ + HC03~*=^ C032~ (72)
Ca2+ + C032~^=5=CaC03 + (73)
Formation of CaC03 film on the electrode starts at the center of
the disc and grows radially. This observation is compatible with
the hydrodynamic regime prevailing at the surface of the elec-
trode. The rate of film growth is proportional to the change of
oxygen diffusion current with time. Curves 2 through 5 (Figure 16)
48
-------
o
•
ro
fi
O
•H
4->
•H
W
O
O
U
(0
U
en
M
3
O
g
•H
U)
0)
o
(IJ
g
•rH
I
-p
cu
S-l
u
tn
X
O
OJ
49
-------
exhibit an initial transient change in current followed by a
steady state or linear decrease, and finally the current reaches
a minimum limiting value. At minimum current values, the elec-
trode surface is assumed to be completely covered by CaC03 film.
Acidification or addition of a sequestering agent to the test
solution solubilizes the CaC03 film, returning the current to
its original value. The equation for diffusion current in the
presence of CaCOa film is:
i = nFA(Pm/6F)CB (74)
where :
P^ = coefficient of CaC03 film permeability to
molecular oxygen
6-r, = thickness of CaCOs film
r
It is interesting to note that after the CaCOs film is formed,
both P and 6 are independent of system hydrodynamics. Permea-
bility of the film to molecular oxygen is the limiting step in
transfer of oxygen to the electrode surface, and P depends only
on CaCO3 crystallinity . m
Experimental
Co-iumn
In 1939, Enslow proposed a "continuous stability indicator"
( 23, 24) . A stream of water was slowly but continuously passed
through a column of powdered chalk. The pH and titratable
alkalinity were measured before and after contact in the column.
Changes in these parameters were related to the corrosivity or
scale forming capacity of the test water. A contact time of
two hours was considered sufficient for the water to be in com-
plete equilibrium with calcium carbonate.
Early in the NSF project, an apparatus similar to Enslow "s column
was constructed from a 24-inch length of three-inch ID plastic
pipe. (See Figure 17.) Test water was recirculated through the
column using a peristaltic pump. The column was packed either
with marble (calcium carbonate) chips or powdered calcium car-
bonate. Ion-selective electrodes were used to measure influent
and effluent hardness. Changes in pH , even with low flows and
high recirculation rates, were very small, never approaching
the hypothetical pH . Changes in hardness were negligible. It
soon became apparenf^that very long contact times were required
for the water to come into equilibrium with calcium carbonate.
A second study was designed to demonstrate the rate of attaining
calcium carbonate equilibrium in various waters in contact with
different forms of calcium carbonate. In a batch system, waters
50
-------
I
CD
G
•H
S
CD
-P
CD
Q
W
CD
dJ
G
(U
CD
CD
CQ
CU
O
G
CD
H
CD
•rH -P
Q G
CD
X
CD
T3
G
H
•H
•H
-P
CO
CU
H
O
-P
•rH
G
UH
U)
a
•rH
.£
U
CD
•H
£3
O
O
rd
\
•H
MH
O
CD
M
3 O
CO 4-)
CD rH
> CD
CD -P
•H fO
U
(0 rH g
O 3
o m -H
-P rH
CD ,Q
G g -H
O -rH rH
•rH -p -rH
-P 3
(0 -P D1
rH U CD
3 fd
CJ 4J G
rH G O
•H O -H
O U -P
CD (0
in -P SH
G 3
M CD -P
O -H fO
fe U CO
d)
-P
G
CD
e
CD
n
3
03
m
0)
g
-H
SH
X)
•H
r-H
•H
3
G1
CD
o
-H
-p
m
n
3
-p
<0
to
!H
O
VH
O
-p
•H
G
O
X
0)
-P
-H
-P
cn
cu
tn
•H
51
-------
of defined chemical characteristics were mixed at constant rates
with marble chips or pure powdered calcium carbonate. Tempera-
ture of the test samples was controlled in a water bath at 25
+0.2°C. Hardness and pH were measured with respect to time using
a glass pH electrode and an Orion divalent cation selective
electrode. Attainment of equilibrium was shown to be asymptotic,
and pH changes did not closely coincide with changes in [Ca2*] .
The time required to reach reasonably steady values of pH and
[Ca2 ] was in the order of 30 to 60 hours.
Catc-ium Carbonate. VzpoA
-------
Disc Contact
Motor Pulley
Electrode
Pulley
Motor
Ring
Contact
Electrode
Base
Figure 18. PIR rotator - electrode system.
CALCIUM CARBONATE
DEPOSITION TEST
STRIP-
CHART
RECORDER
POTENTIOSTAT /
CURRENT MEASURING
BRIDGE
TO COMPUTER
DRAIN
Figure 19. CCDT system in the mobile laboratory.
53
-------
is controlled by mixing C02 with air - from which C02 has been
scrubbed by bubbling through a solution of concentrated NaOH -
in a Matheson 665 gas proportioner, and monitoring pH with a
glass electrode system inserted in the sample cell.
When the test is initiated, 100 ml of the test water is immed-
iately withdrawn from the sample cell for analysis of hardness,
calcium, alkalinity, and pH.
Rotational speed of the electrode is 1600 rpm. This speed was
selected as a compromise between the needs for maintaining con-
stant hydrodynamic conditions at the surface of the electrode
and mixing in the bulk of the solution, and the turbulence asso-
ciated with high rotational velocities. At 1600 rpm, there is
little turbulence in the test solution. Speeds greater than
1600 rpm provide no additional increase in diffusion current
attributable to changes in mass transport of oxygen to the
electrode surface.
The potential at which the cathode is controlled, -0.9 V versus
S.C.E., is the potential at which the smallest change in current
is observed over the largest region of change in potential. A
recording of current as a function of potential sweep, used to
determine the operating potential, is shown in Figure 20. The
test is continued for a period sufficient to establish the
linear portion of the current decay curve. CCDT is reported as
the slope of the linear portion of the curve, in microampc per
minute.
During field operation of the mobile laboratory, CCDT was mea-
sured in a covered continuous flow cell in which the test water
entered at the bottom and overflowed through a port at the top.
pH control is not required in the continuous flow system.
Results and Discussion
Plots of diffusion current versus time for a series of formulated
water samples are shown in Figure 21. Chemical characteristics
of each of these samples are shown in Table 4. Curves 1 and 5
in Figure 21 were obtained from waters with Ryzner Stability
Indices (S)=6.92 and 6.95 and CCDT values of 10.08 and 11.32
yamps/min, respectively. These waters are expected to be scale
forming. Curve 2 is from a water with S=8.76 and CCDT=0.8 yamps/
min, a water expected to be corrosive to CaCOs. Curves 6, 7, and
8 are from waters with moderate scale forming characteristics, as
shown by their significant values of CCDT.
CCDT versus S for each of the formulated water samples is plotted
in Figure 22. The utility of CCDT as a rapid technique for mea-
suring scale forming tendency is clearly demonstrated by the
high correlation of these data.
54
-------
800
700
Reduction of oxygen at
rotating gold electrode
coated with a thin CaCO.
film.
Same as A, but in
purged water
.'- \ —: - t—,
- -
_ i T - r
-J-l "T'T
L t „!
1 _J _L i _4_ ;_
"
-h-M—I
- T-j-- r- -,—-(•-- -—,—-
• :iT4: .rriripirt::
:rrrz:±r±r
.::•--::_.
^1.0 -0.8 -0.6 -0.4 -0.2
Potential (volts vs. S.C.E.)
Figure 20. Current versus potential sweep.
0.0
55
-------
600
w
-------
C!
-H
E
en
a,
u
-H
s
QJ
Qi
O
rH
CO
EH
Q
U
U
100
10
0.1
Increasing scale
formation tendency
6.5
7.0 7.5 8.0 8.5
Ryzner Stability Index
9.0
Figure 22. Correlation of CCDT slope with Ryzner stability
index.
57
-------
CO
z
g
i-
01
CO
111
CO
o
cc
>•
QC
<
D
CO
o>
3
C
IS ol
(/>
.ti co C":
co c o
0^-^,0
|-§"
E _
D —
•Jj O)
S ra CO
^^8
i^<^
c
O *~t
•^ CM
o a>
O £ o>
0) LL
CO
q
o
CM
OJ
CD
in
CO
5
CM
OJ
CM
CM
CO
o
OJ
in
in
00
co
CM
CM
^
CO
d
CO
CO
s
9
CM
in
CO
CO
r~-
in
o
CM
OJ
8
CM
CM
S S S 5
r- CN ^ ^
in o o oo
O) Is"1- CN t—
CD r-- oo co
§ £ § !2
? ? 9 9
OJ CM OJ O
oj CM ^j- ^r
r^ co co co
«- CO CO CM
CO O O OJ
oo co co r^
q •* «ct CM
in ^-' d in
CO CO <* CO
in o CM CM
OJ r-~ in oj
CD P~ CO •*
Tt CM CO CO
1^ OJ CO CM
in co r~ co
58
-------
The effect of pH and levels of calcium and bicarbonate ions on
CCDT was evaluated with solutions of Ca(HCO3)2 at pH=6.5, 7.5,
and 8.3. In each of these solutions, bicarbonate concentration
was four times greater than the levels of calcium. The solutions
were prepared by mixing CaClz and NaHCOa, and bubbling with a
mixture of oxygen, nitrogen, and carbon dioxide. Bubbling was
continued throughout the test to maintain a constant pH. Re-
sults of these analyses, plotted in Figures 21 and 22, demon-
strate that relationships do exist between CCDT and pH, Ca2 ,
and HC03~ levels in a test water.
Tap water samples, brought to the laboratory from a number of
Michigan cities as well as those included in the mobile labora-
tory field assignments, showed a wide variety of scaling char-
acteristics, measured by CCDT. These data with pertinent
chemical characteristics of each test water are shown in Table 5.
CCDT values for each of the 18 samples in Table 5 are plotted
versus S as semilogarithmic coordinates in Figure 23. The num-
bers circled in Figure 23 coincide with sample numbers assigned
in Table 5. Dotted lines in Figure 23 separate the grid into
regions representing varying degrees of scale formation. With
CCDT values >3, waters are known to form films which reduce
carrying capacities of transmission lines. From CCDT >0.75 but
<3.0, films may be deposited which will, in time, affect carrying
capacities; however, small amounts of inhibitors added to the
water will significantly reduce the problem. Waters with CCDT
values between 0.2 and 0.75 can be expected to produce thin hard
films of CaCOa which will effectively protect against corrosion.
The most apparent departure between observed and predicted
scaling using the Ryzner Index occurred at pH >9; i.e., lime
softened waters designated 4, 6, 8, and 9, and a Zeolite softened
water, sample 1. The Zeolite treated water has high pH and high
alkalinity; thus, S was low although calcium hardness was re-
duced to 15 mg/1. This water was not scale forming and no CCDT
slope was observed, but S ~ 6.6 predicted excessive scale forma-
tion. The occurrence of S values overpredicting scale formation;
i.e., S indicates excessive scaling when none is observed, is
not uncommon for finished waters with high pH.
Field Experience
During field assignment of the mobile laboratory in the metro-
politan Chicago area, unique opportunity was provided to evalu-
ate the responsiveness of CCDT to water quality changes in the
distribution system. Caustic is added routinely at the Central
Water Filtration Plant (CWFP) for corrosion control. For the
24-hour period beginning at 1200 hours on May 30, 1973, caustic
feeders were taken out of service. The mobile laboratory was
operating at a four year old fire station, 10.5 miles from the
CWFP. Two continuous flow CCDT measurements were made each day.
59
-------
CO
UJ
O
D
CO
Q
_l
UJ
O
CO
CO
UJ
oc
in
.2
to
PO4
Addition
C/J
cr
to
t-
Q
O
0
v>
T
a
I
a
>
Total
Alkalini
E
'«
CD
c
0)
CO
£
D
•o
c3
(A
-5 £
o •§
i ^-
1— CO
I
z
00
LO
co
*—
o
T—
+
CD
c
o
z
O)
LO
Is-
§
00
CN
CM
CO
O
m
CO
00
CO
I
"5
o
CO
•
00
o
00
8
?
CD
0
z
5
00
•*
r-.
CD
CO
rj-
CO
o
CO
o
00
CN
I
I
<
1
z
LO
CO
CO
o
9
CO
LO
LO
LO
CO
LO
•3"
00
§
r~
CO
o
LO
O
I
+j"
'o
S
D
>
t~~
LO
r~.
c^
T
CD
O
Z
5)
00
LO
CM
O
CN
CO
LO
CO
CO
o
CN
O
Ypsilanti, Ml
z
§
CO
s
9
CN
CM
^
1^-
CO
8
CO
CO
en
CN
T
CO
o
CO
CO
i
w
~o
'OL
CO
IT
TJ
§
a
z
r—
en
r~
CD
o
T
r-
00
o>
CO
LO
o
o
CO
CO
LO
CO
CN
o
LO
CN
i
c
CO
1
r--
r—
en
r~-
LO
r-
?
CO
C
o
CN
r~-
CO
r-
•
^c
'vt
C
2
z
CN
CM
r^
'S-
^-
?
o
LO
r~
CO
CO
r-.
o
LO
r~~
CN
CO
CO
CO
en
en
o
CO
en
I
a.
1-
1
<
1
Z
00
en
r~.
3
9
g
^
>*
co
Is-
o
CO
r-
CN
CO
§
o
o
8
Petersburg, Ml
z
§
00
8
9
s
CD
CO
00
en
0
00
CO
o
o
CO
CO
o
8
*
5T
u_
o
_i
o"
I
6
z
en
00
00
co
<*
9
CM
CO
T—
•«t
CO
CO
O)
r-.
o
00
•<*
CO
o
Oi
*
sr
LL
1
_l
o"
O)
8
6
z
r-.
00
CN
O
?
CN
CN
CO
r^
CO
LO
r-~
00
en
•^•
r»-
CO
CO
o
CN
*
<
Q_
(0*
*
0}
-o
CD
£
z
CO
^•
r-
00
o
9
LO
•*
00
CO
r~
8
r-
CN
CN
CN
00
r~
CO
CO
0
5
00
o'
Z
C/5
>
Z
LO
LO
t~-
CN
O
?
8
r-
LD
r~
en
LO
r~.
CO
en
CO
CO
CN
o
en
a
E
8
co
£
£
E
to ^ •*
C CO §
S "a
-^ <+-
CO
CO
CO
CO
CN
?
s
CN
r*.
00
co
rv
co
8
R
5
o
o
LO
i
CD"
_c
8
60
-------
10.0
-p
w
C
O
w
O
a
QJ
P
0)
-P
O
•H
U
U
1.0
0.1
Excessive
u Moderate
Slight
No observable CCDT slope
O
_J
I
6.5
7.0 7.5 8.0 8.5
Ryzner Stability Index
9.0
Figure 23. CCDT versus Ryzner stability index,
61
-------
Arrival of the distributed caustic-free water at the mobile lab
was marked by changes in observed levels of pH and CCDT. Mean
CCDT for the previous day (May 29) was 0.52. The lowest daily
mean CCDT, 0.26, occurred on May 31. On June 3, a mean CCDT of
0.47 indicated that water with caustic had returned to the area.
At CCDT =: 0.50, light hard protective film formation is assured.
The low mean CCDT (0.26) indicated that more corrosive water
was present in the lines. These data are plotted in Figure 24.
In extensive laboratory and field applications of CCDT, this
test was shown repeatedly to be capable of accurately predicting,
without respect to the metallic corrosion reaction, whether or
not films of CaCOa would form on the surface of distribution
system pipelines.
ALKALINITY
General
An automated potentiometric technique for measuring total alka-
linity in tap water was developed in this study. The method
was fully evaluated in the laboratory and used successfully in
the mobile laboratory during field assignments. As a result
of apparent faulty operation on Cincinnati tap water which has
high levels of residual chlorine, detailed theoretical relation-
ships were developed and the procedure slightly modified.
Theory
A linear relationship exists between the resultant pH of a
suitably selected phthalate buffer solution (KHP) and the total
carbonate formality of a test water. Theoretical aspects of
this relationship are described using an example test water.
Assume the alkalinity of the test water is in the range 30 to
60 mg/1 as CaC03 (3(10)'" to 6(10)~" F C032 ). It has been
shown experimentally that five parts of the test water and one
part stock phthalate buffer produce the linear relationship be-
tween resultant pH and sample alkalinity. Initial formalities
of the buffer are:
KHP = 0.050
HC1 = 0.019
and of the test water:
Na2C03 = 6/5 C
and final formalities for the mixture (buffer:sample = 1.5) are:
KHP = 8.33(10)~3
62
-------
13
ro
•
CO
CN
•
CO
00
o
•
00
r
X
X
I
X
\
X —
r
X
X
v
"" X
\
— X
EH
Q
U
U
CN
cu
a
fd
a
I
— x
\
(Ti
O
00
O
m
•
o
01
•
o
CN
•
O
-P _
-P
3
O
fd
M S
0) O =
-P SH co
•H o
a 3
a) o
g o
f
I
oo
o
0)
fi
(N
O
(D
fi
O
4J
•H
a
o
g
Ou
EH
P
O
O
ti
O
eu
o
o
en
m
o
•H
^1
u
QJ
1?
•H
ro
K*1
fd
jg
o
oo
^
fd
S
CTl
CN
£•1
fd
-P
rd
tn
0) -H •
S rH 45
•H U tn
EH >i-H
O 45
CD
M ^
0) \
t5 O
0) Cn
(U fd
4-1 O
-H
0 45
•H O
-P
CD -
J3 rH
nj
O •
O
4H 3
O
0)
-P -P
0 -H
(U CO
M-l -P
U U Q EH
CN
•H
Cn
63
-------
HC1 = 3.17(10)~3
Na2C03 = C
[H ] can be calculated as a function of C, applying the following
series of known relationships:
Kl = [H ] [HP ] = 1.3(10)~3 (75)
[H2P]
2 —
= [H ] [P - L ~6
2
K2 = - = 3.9(10)~6 (76)
[HP ]
K3 = [H+] [HC°3"] = 4.6(10)- (77)
[H2C03]
K, = [H ] [C°32 ] = 4.4(10)~11 (78)
[HC03~]
KW = [H+] [OH~] = 10~14 (79)
[Cl~] = 3.17 (10)~3 (80)
[K+] = 8.33(10)~3 (81)
[Na+] = 2 C (82)
[H2C03] + [HC03~] + [C032~] = C (83)
[H2P] + [HP~] + [P2~] = 8.33(10)~3 (84)
~]+ 2[P2~] + [HC03"]+ 2[C032~] (85)
By assuming [CO32~] = 0, [P2 ] = 0, and [OH ] = 0, derivations can
be stated as follows:
from Equation 77:
+
[H2C03] = -tS-L [HC03~] (86)
K3
from Equation 75:
+
[H2P] = JJLJ. [Hp~] (87)
from Equations 83 and 86:
[HC03~]
(88)
K3
64
-------
and
[HC0
4.6(10) 3 C
[IT]+ 4.6(10) 3
from Equations 84 and 87:
(89)
[HP ]
+ 1
K
= 8.33(10)
(90)
and
from
[HP ] =
1.083(10) 5
(91)
Equations 79, 81, 82, 85, 89, and 91:
[H+] + 2 C + 8.33(10)~3= 3.17(10)~3
1.083(10)"5
[H+]+ 1.3(10) 3
4.6(10)"3C
[H+]+ 4.6(10)~3
(92)
then:
L.C.D. =
[H+] 2 + 5.9(10)~3
+ 4.6(10)"3
5.98(10)~6
(93)
where:
L.C.D. E lowest common denominator
+ 2C[H+]2 + 5.16(10)~3[H+]2 + 5 . 9 (10) ~ 3 [H+] 2 + 1. 18 (10) ~2C [
+ 3.044(10)~5 [H+] + 5.98(10)~6 [H+] + 1.196(10)~5C
+ 3.086(10)~8- 1.083(10)~5 [H+] - 4.982(10)~8 - 4.6(10)~3 C [
(94)
- 5.98(10)~6C = 0
+] 3 + J1.106(10)~2+ 2C\ [H+]2 + J2.
- 1.896(10)~8 + 5.98(10)~6C = 0
559(10)"5 + 7.2(10)~3C
5.98(10)~6C = 0
neglect the [H ]3 term and solve with quadratic equation:
(95)
65
-------
2.559(10)~5-7.2(10) 3C + / J2 . 559 (10) ~5+7 . 2 (10) ~3 c}2
-4 J1.106(10)~2+ 2C(
j-1.896(10)~8 + 5.98 (10)~6C|
2.212(10)
-2
4C
(96)
-2.559(10)"5 -7.2 (10)"3 C + / 6 . 548 (10) ~ a °+5 .184 (10)~5 C2
+3.685(10)"7 C +8.388 (10)"10
-2.646UO)"7 C +1.517UO)"7 C
-4.784 (10) 5 C
2.212 (10) 2 + 4C
(97)
-2.559 (10)"5 - 7.2 (10)~3 C + /4 (10)"6 C 2 +2.556 (10) 7 C
+1.489 (10)"9
2.212(10)" 2+ 4C
•(98)
The relationship between hydrogen ion concentration and carbonate
formality given in Equation 98 is shown to be linear by the data
in Table 6 and Figure 25. This linearity extends to the rela-
tionship between resultant pH and carbonate formality, as shown
in Figure 26. In addition, results of an actual experiment are
compared with theoretical predictions in Figure 26. These data
are shown in Table 7.
Thus, when the initial pH and total carbonate formality of a test
water are known, concentrations of the species of interest can
be calculated as follows:
H] [HC03-] = 4.6(10)-3.
[H2C03]
•[H2C03J =
K-
[HC03~]
K, =
[C032"]
[HC03~]
[HC03~] (100)
66
-------
o
00
o
o
m
in
in
o
•
m
CO
•H
-P
•H
H
(0
e
M
O
O
U
en
ffi
m
CM
(U
67
-------
O
00
-P
•H
a
•H
iH
05
O
cn
C
O
•rH
-P
ai
-------
[H2C03] + [HC03~] + [C032~] = C
Substitution of Equations 99 and 100 into 101 gives:
K2 )
(101)
[HC03~] =
C +
Kl
(102)
[HC03~] =
Ki [H+]C
[H+]2
KiK2
Table 6. [H+] VERSUS ALKALINITY, APPLYING EQUATION 98
Alkalinity
(mg/1 as CaCO3)
24
36
48
60
72
84
[H+]
(molarity)
5.33(10)'^
5.07 (lO)'1*
4.82(10)"^
4.57(10)"^
4.34(10)"'*
4.11(10)~'t
*(C03 formality) x 10 = Alkalinity in
mg/1 as CaCO3
Linear regression:
[H+] = {57.9133 - 0.237515*ALK)(10)~5
69
-------
Table 7. EXPERIMENTAL AND THEORETICAL RELATIONSHIPS
BETWEEN ALKALINITY AND pH
Alkalinity
(mg/1 as CaC03)
• 24
36
48
60
72
84
pH
Experimental
3.303
3.347
3.391
3.434
3.476
3.519
Theoretical
3.273
3.295
3.317
3.340
3.362
3.386
Experimental
The sensing element in the alkalinity monitor is a combination
glass pH/reference electrode with a ground glass collar. The
electrode fits directly into a glass flow cell. Instrumentation
includes a Corning Model 101 digital pH/mv meter; Honeywell
Electronik 194 multi-speed, multi-range servo recorder for
signal conditioning; and an analog output to the computer. A
Technicon II peristaltic pump is used for reagent/sample propor-
tioning. A Valcor series SV-72 three-way miniature dri solenoid
valve switches between baseline and sample solutions. AC line
voltage is switched to the valve by computer controlled relay.
A schematic illustration of the system is shown in Figure 27.
The Corning Model 101 is especially suited for use in the alka-
linity system because of its exceptionally quiet operation and
its provision for a solution ground. The alkalinity measurement
is read from approximately 20 millivolts variation in a signal
with a total magnitude of only 200 millivolts; thus, the need
for quiet operation is apparent. The peristaltic action of the
pump controlling flow through the system produces an electrical
signal. This peristaltic noise is reduced by a solution ground
placed in the sample line. Many ground configurations were
studied. The most successful provides for contact with the
solution at two points - in the influent sample line, just before
the pump; and after the debubbler, just before the sensor cell.
These two grounds are connected together and terminate at the
solution ground terminal in the pH meter.
70
-------
u. Z
tu -_
z 2
, fc-
ui 3
vi -"
< O
a -
u. 3
3 O
a in
tu
i-
(n
<
s
o
*-
0
z
ui x
£ u
s ±
< i
VI VI
O
Z
•V ^
™ X
2! i
a
3
N
-i „
8?
DC
O
t/i
Z
UI
it
i — — ~~ — 1 ^
M
1 °
^ ,„ ,,.] _ 1 1
" i m
r I >• M
2*2 s 5
^ ^ *•• P*X
i — 1
Q)
„ , i
QC r^
< 5
P>
o g
Z *=i
•» r^
? -H
h-
° a. 0
0 * -P
« 3 -rH
a. o. f->
S 0
g
4-1
•H
.— t
•H
i — 1
fO
rH
O
C!
O
•rH
-P
^ TO
UJ <_,
u -P
^-i
z
o o
h- U
x
u
r o
« U
O
•H
-P
(S
e
0)
X!
O
CN
0)
Cn
-rH
En
71
-------
Because of the acid nature of solutions used in the system, it
is not uncommon for the ground wire to become coated with cor-
rosion products and lose contact with the solution. Whenever
signal quality degenerates, the grounds are simply replaced.
The system operates on a 26-minute cycle controlled by the com-
puter. At initiation of the cycle, the relay is switched to
the "on" position and the three-way valve is connected to the
baseline solution. For the next 13 minutes, baseline solution
is drawn into the system. Air bubbles are introduced beyond the
pump to separate the sample stream into discrete samples. The
buffer is added and a three-inch glass mixing coil provides
mixing and reaction time before the sample reaches the electrode
system. The sample stream is debubbled just before it enters
the sensor flow cell. The bubble stream goes to waste under
siphon while the remainder of the sample is drawn past the
sensors, then to waste. At the end of the 13-minute period, the
computer sets the relay to "off" and the three-way valve is open
to the test water stream. A complete cycle of 26 minutes is
required to read both baseline and test water samples.
Preparation of reagents used with the alkalinity monitor is in-
cluded in Appendix B.
Results and Discussion
A good quality recording of "continuous" alkalinity analyses is
shown in Figure 28. The analytical region of the curve is from
the tip of "drop" to the center of the equilibrated signal from
the sample. Greatest precision is achieved by measuring peak
height immediately after initial leveling off. The computer be-
gins its measurement after switching from baseline to sample and
continues to seek equilibrated signal for a period of six minutes,
In Figure 28, the height of the measured peak is 27.0 chart units
(C.U.). Measurement of peak height to the nearest 0.5 C.U. and
reading alkalinity from a calibration curve produces data with
an accuracy of +2 mg/1, essentially the same as that which is
stated in Standard Methods (3) for potentiometric titration of
alkalinity.
A calibration series of five or six standards is run for each
combination of newly prepared baseline or buffer solution. Peak
heights in C.U. are plotted on the abscissa and concentration
on the ordinate. A typical calibration curve is shown in Figure
29. To ensure accuracy of the data, the standards are checked
by potentiometric titration and the continuous output compared
with a potentiometric titration of a tap water sample.
72
-------
-H
in
(0
-p
•H
ooeo
(0
tn
3
O
3
C
•H
•4-1
fi
O
O
IH
O
tn
0
•H
T)
H
O
O
0)
4J
•H
CP
O
O
O
oo
CN
0)
tn
•H
73
-------
24
20
16
w
4J
-H
-P
M
^u
£ 12
Cn
a;
(13
(1)
Oi
10 20 30 40 50 60
Alkalinity (mg/1 equivalent calcium carbonate)
Figure 29. Alkalinity calibration curve.
74
-------
Special Case Study
Following delivery of the mobile laboratory to the EPA/NERC
facility in Cincinnati, hookup was made to monitor Cincinnati
tap water. For some time, the alkalinity system did not produce
data which was equivalent to that reported by NSF during most
periods of field assignment; i.e., the basic relationship be-
tween final pH and alkalinity could not be reproduced. Titra-
tions with quiescent solutions allowed to equilibrate, stirred
solutions at equilibrium, and stirred solutions read at constant
time intervals produced apparently randomly scattered data points.
High levels of free residual chlorine were thought to be contri-
buting to this anomalous behavior. The phenomenon had been ob-
served on two previous occasions - once when the laboratory was
located immediately adjacent to the water treatment plant in
Ann Arbor, Michigan, and once in the Calumet Harbor area of
metropolitan Chicago.
To quantify the role of free residual chlorine in potentiometric
measurement of alkalinity, an experiment was proposed and re-
sults of the experiment calculated theoretically. The protocol
and related calculations are presented in detail in Appendix C.
Experimental data closely approximated the theoretical results
and established the anomaly could not be attributed to the pre-
sence of chlorine alone. The apparent interference was elimin-
ated by adding thiosulfate (five mg/1) to the buffer solution,
suggesting that chlorine may have contributed to the problem,
perhaps synergistically with an unidentified organic present in
the test water. The problem has not recurred since EPA substi-
tuted a combination electrode with a calomel reference for the
silver chloride reference-combination electrode in the original
equipment.
FREE AND TOTAL FLUORIDES
General
Fluorides are added to treated water supplies as sodium fluosili-
cate (NazSiFe), commonly referred to as sodium silica fluoride;
fluosilicic acid (H2SiF6); or sodium fluoride (NaF), each of which
dissociates to release free fluoride ions (F~). Prior to this
study, it was generally assumed that fluorides reached the con-
sumer in this form; i.e., as free ions, and that those which were
added during treatment were not different from those which may be
present in natural waters. This assumption is invalid. In most
chemically treated finished waters, fluorides are complexed with
polyvalent cations, such as aluminum or iron, which are commonly
added as coagulants or may be present as natural constituents of
the raw water.
In a survey of public water supplies in the 100 Largest Cities
in the United States (25), residual aluminum and iron in distri-
75
-------
bution systems were reported to be as high as 1.5 and 1.7 mg/1,
respectively.
Fluoride is an important parameter of drinking water quality.
In high concentrations it may have a toxic effect; at low levels
it is beneficial; i.e., an effective preventer of dental caries
(26,27,28,29). Chemical reactions between F~ and tooth enamel
are postulated as follows:
or
Cai o (POO e (OH) 2 + 2F -> Cai0(P04)6F2
hydroxylapatite fluorapatite
20H
Caio(POn)e(OH)2 + 20F
hydroxylapatite
8H+ +
10CaF2 +
calcium
fluoride
2H20
(103)
(104)
Reaction 103 is generally favored at low levels of fluoride; i.e.,
in fluoridated waters; reaction 104 occurs in the presence of
high concentrations of F~; e.g., dentifrices (30).
The recommended level for fluoride in drinking water is less
than 1.7 mg/1 (1); thus, analytical detection methods must be
accurate at this level. In the NSF project, the potentiometric
ion-selective electrode technique was studied in detail, and
automated for measuring both free and total fluorides in the
mobile laboratory.
Theory
The potentiometric membrane electrode selective for F~ can be
described schematically as follows:
Internal
Reference
Electrode
Ag/AgCl
Internal
Reference [XF]
Solution
AgCl(s)
Membrane
LaF'
External
Solution [YF]
Test
Solution
External
Reference
Electrode
Saturated
Calomel
Electrode
(S.C.E.)
where:
vertical lines represent boundaries or liquid
junctions between components of the cell.
The basis for ion-selective electrode measurements is the Nernst
equation:
aj
(105)
E = E _ M in
m ° ZF
ap- + I *a±
i
76
-------
where:
E = measured potential
E = a constant equal to the sum of the potentials
° at the Ag/AgCl electrode, the saturated calo-
mel electrode, the liquid junction potential
between the test solution and the reference
electrode, and the potential across the mem-
brane when F~ activity in the test solution
is = 1.0
R = gas content
T = absolute temperature
z = number of electrons involved in the electrode
reaction
ap- = activity of F~
a. = activity of ith interfering ion
K1 = selectivity coefficient of interfering ion
with respect to F~.
Activity (a) is related to concentration (c) as shown in the
following empirical expression:
a = ac (106)
where:
a = activity coefficient
At 25°C, the slope of the line described by Equation 105 is
approximately 60 mv; however, in practice measurements are
usually read from calibration curves. In calibrating ion-selec-
tive electrodes, changes in total ionic strength of the test
solution must be considered. With the fluoride electrode, OH"
is the only species which interferes with F~ measurements. The
effect is apparent when the level of OH~ is greater than one-
tenth the level of F~ present in the test water (31,32)._ Other_
anions commonly found in drinking water; i.e., Cl , SOi,2 , HCOs ,
N03 , and P0i*3~ do not interfere even when they are present at
levels 1,000 times the level of F~ (31,33,34,35,36).
It is important to differentiate between OH interference and+
the complexing effect of certain cations; e.g., H , Al3 , Fe3
(31,34) . Fluoride combined with these cations cannot be sensed
by the electrode, but this effect is different from the inter-
ference of OH~.
Experimental
Equipment used in studying and optimizing the fluoride ion-selec-
tive electrode measurement technique included an Orion digital
pH/mv meter, Model 801; Orion solid state fluoride ion-selective
electrodes, Model 94-09; saturated calomel reference electrodes;
77
-------
and polyethylene containers to minimize loss of F by adsorption
onto glassware. A Corning digital electrometer, Model 101, was
substituted for the Orion 801 early in the study. The Corning
meter proved to be a much more reliable instrument, and with its
activity mode, appreciably simpler to use. Calibration was set
so the upper limit read "100" for 5 x 10~5M F~ and the lower
limit, "20" for 10"5M F~; thus, percent recovery of F~ in the
test solution could be read directly.
In studying the effect of pH, a stock solution of sodium fluoride
was prepared from which suitable dilutions were made. pH was
varied by addition of strong acids and bases, and ionic strength
was maintained at 0.5M by addition of appropriate volumes of
concentrated sodium nitrate (NaN03) solution. Potentials were
measured after equilibrium was established. At low fluoride and
high pH levels, more than 15 minutes were required to reach
equilibrium.
Complexemetrie titration techniques were used to characterize
aluminum-fluoride complexes. pH was controlled by an acetate
buffer. Solubility of the complexes was suppressed with ethanol.
In measuring total fluorides, a masking agent is added to complex
Al3 more strongly than F~, liberating bound fluoride for sensing
by the electrode. The relative efficiencies of four masking
agents was studied: EDTA (disodium salt of ethylenediaminetetra-
acetic acid), CDTA (cyclohexanediaminetetraacetic acid), oxalate,
and citrate.
Equipment in the free and total fluoride monitoring system in-
stalled in the mobile laboratory included the Orion digital
pH/mv meter, Model 801 (for total fluorides) and a Leeds and
Northrup expanded scale pH/mv meter, Model 7415-EO (for free F~);
two Orion solid state fluoride ion-selective electrodes,
Model 94-09; Ag/AgCl (total) and calomel (free) reference elec-
trodes; an Esterline Angus dual pen strip chart recorder; a
Technicon Pump II for reagent/sample proportioning; and Valcor
three-way miniature dri solenoid valves, series SV-72, for
switching between baseline and standard solutions. AC line
voltage was switched to the valves by computer controlled relays.
A schematic procedure for measuring free and total fluorides with
ion-selective electrodes is shown in Figure 30. Preparation of
reagents is described in Appendix B.
Results and Discussion
pH E&6e.ct
The effect of pH on fluoride electrode measurements is shown in
Figure 31. The OH~ interference effect is apparent when the
78
-------
SAMPLE
Free Fluoride, A
- Add free fluoride buffer
(FFB)* to sample, 1:1
- Potentiometric measurement
Total Fluoride, B
- Add total fluoride buffer
(TFB)+ to sample, 1:1
- Potentiometric measurement
Complexed Fluoride, C = B-A
Preparation of Buffers:
To about 700 ml double distilled water add:
*Free fluoride buffer,
FFB
Total fluoride buffer,
TFB
Acetic acid
Sodium nitrate
Sodium citrate
5.7 ml (0.1 M)
84.99 g (1.0 M)
None
11.4 ml (0.2 M)
59.5 g (0.7 M)
58.8 g (0.2 M)
Adjust the pH to 5.2 with concentrated solution of sodium
hydroxide, and the volume to 1.0 liter.
Figure 30. Characterization of fluorides and preparation of buffers
79
-------
250
11
13
Figure 31. Effect of pH on fluoride ion-selective electrode
measurement.
80
-------
curves in Figure 31 show a nonlinear relationship and shift to-
ward more negative potentials. Because F~ and OH~ have similar
changes and ionic radii (31), the electrode senses both (but to
varying degrees); thus, a false increase in apparent F~ activity
may occur at high levels of OH~.
The effect of H-F complexation is also apparent in Figure 31.
Hydrogen ions combine with fluoride ions to form hydrogen fluoride
(HF) and bifluoride ions (HF2~), neither of which are sensed by
the electrode (37,38).
According to the literature (31,32), the fluoride electrode can
be used for measurements in drinking water at pH five to eight;
however, data from this study indicate a much more restrictive
pH range (Figure 31). At 10~6M F~, the optimum range is 5.0 to
5.4; thus, pH 5.2 is used in the NSF fluoride monitor and is
recommended for all ion-selective electrode measurements of
fluorides in water supplies.
The selectivity coefficient (K) of OH~ with respect to F~ can be
calculated as follows:
_ _
E = E - In [(F ) + K(OH )]
O r
(107)
Using the data from Figure 31, estimated values for K are shown
in Table 8. Note that K increases as the level of F~ decreases;
i.e., the interference effect of OH~ is more predominant at lower
concentrations of F~. At levels of F~ commonly found in drinking
waters, values of K are on the order of 0.70. By calculation,
using Equation 107, it can be shown that at 10~SM F~, the OH~ in-
terference effect is significant only at pH > 5.4.
A further illustration of OH~ interference is shown in Figure 32
in which calibration curves at various levels of pH are plotted
as electrode potential versus concentration of F~. Nonlinear
electrode response is greater as fluoride concentration decreases
and pH increases. At low pH, electrode response follows the
classical Nernstian relationship. Linearity was extended to
at pH values of two and three. The parallel shift in
10~6M F
Table 8. OH SELECTIVITY COEFFICIENTS (K)
F~ Concentration
10~6M - lO^M
10~"M - 10~3M
10~3M - 10~2M
K
0.70
0.07
0.05
81
-------
-50
0.019
SO
g
"c
100
0>
1
o
150
200
250
- pH 12
pH 2
10
0.19
1.9
19.0
r6 |0-5 |0-4 !0-3
Ruoride Concentration, Cj
mg/l-
190
IO'2M
Figure 32. Calibration curves of different levels of pH,
82
-------
the intercept of the calibration curves below pH five can be
.ained in terms of H complexation (37):
Ki(HF) v 0 Q (108)
pl\ i — ^ . .7
explained in terms of H complexat
(F~) (H+) -»- KI (HF) R _ 2
(HF) (F") -*• K2(HF2~) K = o 77 (109)
where:
KI = first dissociation constant
K2 = second dissociation constant
and:
where:
CT = 2(HF2~) + (HF) + (F-) (110)
C = total concentration of fluoride
For solutions of fluoride more dilute than 0.05M, (HF2 ) can be
neglected and Equation 107 rewritten as (39):
CT = (HF) + (F~) (111)
Combining Equations 108 and 111 and solving for F~:
(F-) = CT ^ — (112)
Then, substituting Equation 112 into the Nernst equation and
rearranging:
•prn V" T3T1
E = E - — In £•* — In C (113)
0 F K! + (H+) F X
where the term - RT/F In KI/KI + (H+), the "shift term," is con-
stant for given pH and temperature and independent of fluoride
concentration. At pH > 5, (H ) is relatively small, the ratio
KI/KI + (H ) approaches 1.0, and the shift term = 0; i.e., no
shift in intercept occurs. At pH < 5, (H ) is large, and the
shift term is positive and increases with decreasing pH. Experi-
mentally determined values for shift in intercept at 25°C cor-
relate closely with calculated values, as shown in Table 9.
83
-------
Table 9. SHIFT IN INTERCEPT OF CALIBRATION CURVES AT LOW pH
VALUES
pH Range
2-3
3-5
Calculated
36 . mv
14. mv
Experimental
40. mv
15. mv
It follows, then, that at constant pH, H complex formation is
constant at all levels of fluoride and has no effect on the
slope of the calibration curve; i.e., at constant pH, changes
in the slope of a calibration curve developed within sensitivity
limits of the electrode are the result of OH interference only.
The effect of pH on electrode sensitivity is shown in Figure 33.
Vsiva.le.nt Cat-ion
2
2
Both calcium (Ca2 ) and magnesium (Mg2 ) ions form insoluble or
sparingly soluble salts of fluoride. High levels of these
cations (145 mg/1 Ca2 and 120 mg/1 Mg2 as CaC03), the principle
hardness ions, are commonly found in distributed finished waters
(23). Even in much higher concentrations than those reported
for treated waters, neither Ca2 or Mg2 showed any effect on
fluoride electrode measurements. Recovery of 5_ x 10~5M F~ was
complete in the presence of 10~2M Ca2 and Mg2 ; i.e., >400 mg/1
Ca2+ and >240 mg/1 Mg2+. At these levels, any CaF2 or MgF2
which is formed is well below their respective solubility pro-
ducts; i.e., K = 10.4 and 8.15, respectively.
s
Cat-ion
Trivalent cations; e.g., A13+ and Fe3 , rapidly complex with
fluoride. The extent of complexation is a function of pH and
relative levels of F~ and complexing species in the test solution.
Water which is treated by alum coagulation may contain from 3.3
to 1,500 mg/1 aluminum (25). Over the period March 1970 to
February 1971, aluminum levels in Ann Arbor, Michigan tap water
varied from <20 to 500 mg/1, measured by emission spectroscopy .
The Al-F complex was characterized by potentiometric (ion-selec-
tive electrode) titration. Results are shown in Figure 34.
Addition of acetate buffer (Curve II) to maintain pH 5 . 2 signifi-
cantly improved end point detection. With no pH control (Curve I) ,
addition of A13+ titrant lowered the pH of the test solution re-
sulting in further H + F~ complexation. Addition of ethanol
(Curve III) also significantly improved the titration. Both ethanol
and buffer added to the test solution (Curve IV) produced the
84
-------
20
840
i
<3
60
7
pH
13
Figure 33. Effect of pH on fluoride ion electrode sensitivity.
85
-------
-200
-160 -
Titration of 10.0 ml of O.IM NaF,
brought up to 70.0 ml total volume
by the addition of:
I. 60.0 ml of redistilled water
I.
IE.
10.0 mi of I.OM acetate buffer
+ 60O ml distilled water
50.0 ml absolute ethanol
+ 10.0 ml distilled water
10.0 ml acetate buffer
iSO.O mi ethanoi
234
ml of O.IM AICI3 Titrant
Figure 34. Complexometric titrations of fluoride with aluminum.
86
-------
optimum condition for titration. Stoichiometric calculation of
the complex coordination number using the data from Figure 34
identified the complex as (AlF6)3~.
A number of masking agents have been proposed for potentiometric
measurement of total fluorides (40,41,42). Frant and Ross (40)
added citrate to total ionic strength adjustment buffer (TISAB);
Crosby, et al (41) changed the composition of TISAB and added
EDTA; and Harwood (42) used CDTA. Four separate masking agents
were studied early in the NSF project: oxalate, citrate, EDTA,
and CDTA. Solutions of each were made from corresponding
equivalent weights. Increments of aluminum were added to standard
solutions of 5 x 10~5M or 1.0 mg/1 F~. When complexation was
complete, a masking agent was added and percent recovery of F~
calculated. Electrode measurements were made in the presence of
l.OM acetate buffer (pH = 5.2) with constant ionic strength
(l.OM NaN03). Results are shown in Table 10.
EDTA, CDTA, oxalic and citric acids have respective stability
constants with aluminum as follows: 16.1(43), 17.6(43), 14.6 (44),
and 20.0 (45). EDTA and oxalate did not yield high recovery
rates. Despite its relatively high stability constant, Al-EDTA
complex formation is too slow for EDTA to be an effective masking
agent in the measurement of fluoride. Heating or standing over-
night showed no improvement. The solubility of oxalate in TISAB
buffer solution was limited. At 6.0 meq/1, CDTA yielded the
highest recovery, but higher levels of CDTA had no further effect
on recovery rates (Figure '35). At higher concentrations of any
masking agent, citrate yielded highest recoveries. As shown in
Table 10 and Figure 36, 60 meq/1 citrate provided good recoveries.
At 600 meq/1 citrate (0.2M), 100 percent recovery at 0.5 mg/1
A13+ occurred after 6.5 minutes; at 0.1 mg/1 Al3^ recovery was
complete after three minutes. (See Figure 37.) Averaged addi-
tive effects with no synergism or potentiation were observed
when two or more of the masking agents were mixed.
Iron frequently occurs in drinking water in higher concentrations
than aluminum; i.e., levels from two to 1,700 mg/1 are reported
in the 100 Largest Cities in the United States (25)• The effect
of Fe3+ - F~ complex formation was also studied. Results, shown
in Table 11, indicate that iron does not form stable complexes
with fluoride. Nearly 100 percent recovery occurred with acetate
buffer alone. At Fe3"^ >2.0 mg/1, competitive complexing effects
were observed. Complete F~ recovery at very high concentrations
of Fe3+ occurred with citrate and, to a lesser extent, with CDTA.
Automated System
A schematic illustration of the automated free and total fluoride
measurement system is shown in Figure 38. This system was oper-
ated in the laboratory to establish performance characteristics;
87
-------
Table 10. ALUMINUM COMPLEXING EFFECT AND FLUORIDE RECOVERY
BY DIFFERENT MASKING AGENTS
Aluminum
Concentration
Acetate
Buffer
EDTA
Oxalate
CDTA
Citrate
meq/l
mg/l
0.1 Wl
1.0M
meq/l
6.
30.
60.
6.
30.
6.
30.
60.
6.
30.
60.
% Recovery of Fluoride
0.0
0.0
100
100
101
100
100
100
102
102
101
100
101
101
102
0.006
0.054
94
95
98
97
97
99
100
101
100
100
100
100
101
0.012
0.108
90
92
94
92
90
96
97
99
97
97
100
100
101
0.024
0.216
80
83
88
86
85
94
96
97
95
94
99
100
100
0.06
0.54
56
62
81
80
79
89
93
95
91
89
95
100
101
0.12
1.08
28
48
69
69
68
85
91
92
87
85
92
99
100
0.6
5.4
5
15
24
25
27
54
82
86
85
82
67
89
95
1.2
10.8
3
10
15
16
15
32
67
73
73
73
53
83
91
Buffered masking agents added to standard solutions containing 5 x
(0.95 mg/l), and variable aluminum concentration
fluoride,
-------
Al3* mg/l
3.6 5.4 72
0.4
0.6
Al3* meq/l
0.8
1.0
10.8
1.2
I Acetate buffer, l.OM
II EDTA, 60 meq/l
III Oxalate, 30 meq/l
IV CDTA, 60 meq/l
V Citrate, 60 meq/l
VI Citrate, 200 meq/l
Figure 35.
Aluminum complexing effect and fluoride recovery
by different masking agents.
89
-------
100
90
o
01
•s
80
5
o
o
o
a:
70
60
• 6 meq/l
• 30 meq/i
60 meq/l
8 12 16
Time in Minutes
20
24
Standard solutions contain 1.0 mg/1 fluoride and
0.5 mg/1 aluminum.
Figure 36. Rate of fluoride recovery with CDTA masking agent.
90
-------
100
£90
*k_
o
tZ
M-
o
>,80
S
o
o
70
60
0
8 12
Time in Minutes
16
20
24
Standard solutions contain 1.0 mg/1 fluoride and
0.5 mg/1 aluminum.
Figure 37. Rate of fluoride recovery with citrate masking
agent.
91
-------
«
Jj
2
T3
|E
jZ "5
<»>co
a>
it
t>
in
a>
•o
oE
j»iqqr»q»o—•**
£
In
Is
o Q} fy ao
<^ o o o
S -
E
S
a.
_a<
o.
I
CU
T5
•H
M
O
-P
o
-P
i
rH
(0
c
m
cu
H
&
(C
OJ
-P
(0
g
o
-P
o
u
-H
-P
(0
g
CL)
^
0
00
ro
0)
-H
CM
92
-------
CO
z
LLJ
a
I
>-
03
cc
LU
8
LU
CC
LJJ
a
E
o
O
<
O
LJJ
01
Z
O
cc
•8
o
C
"S
L_
§
8
0)
OC
*
^
CN
CM
r~
(O
O
CM
O
CO
o
0
s
o
d
CN
o
0
CO
o
o
o
o
o
—
g
E
2
8
5
a
CM
CM
£
-
CO
CM
CM
CM
«-
LD
O
CM
CM
0
«~
0
o
o
O)
C
o
L.
§
<3
s
un
O)
oo
O)
O)
CD
8
o
o
0
o
o
o
1 1
<2 .
to j;
si
< OQ
§
O)
O)
§
o
o
^
o
o
8
o
o
T~
8
T—
o
o
8
<
8
r~ O> O
a> o> o
888
» *^ T~
O O O
o o o
o o o
o o o
t T™" '
o o o
o o o
^
o o o
o o o
888
888
T— r— r—
o o o
o o o
R S
o>
!;:
•(-»
O
93
-------
i.e., buffer to sample ratio, flow rates, mixing times, etc.
required for design of the "continuous" fluoride monitor. Flow
rates for each reagent/sample stream are expressed numerically
in the schematic. Standard solutions and samples are placed in
individual sampler cups. Automatic sampling is initiated after
stable baselines are established for reagents flowing through
the two flow cells. The sample- or standard- is withdrawn,
segmented with air, and mixed with buffer in a mixing coil.- The
sample stream is debubbled; i.e., the segmented air and a portion
of the test solution are pumped to waste. Residual test solu-
tion is passed through the flow cell for sensing by the electrode.
Sensed potentials for standards or samples from each electrode
system appear as output peaks from the recorder, one each for
free and total fluoride. Calibration curves are used to read
the respective peak heights as fluoride concentrations. Complexed
fluoride is reported as the difference between total and free.
The Nernstian semilogarithmic relationship has an important effect
on automated measurements; i.e., it precludes the use of zero
concentration fluoride as baseline solution in the measurement
system. A fluoride standard solution must be used as the wash
between samples and standards. A concentration of 0.1 mg/1 F~
is used for the wash/baseline solution. This level of F~ is
well above the lower limit of detection for the electrode, is
in the linear range of electrode response, and is a level at
which the electrode response rate is relatively fast. Fluoride
levels higher than the baseline concentration; i.e., F~ >0.1 mg/1
appear as positive peaks (above the baseline), and F~ <0.1 mg/1
appear as negative peaks (below the baseline).
E^e.c.t oft Pump on Re^e^ence Etic.tn.ode,
Early in the study, electrode potentials fluctuated in regular
rhythmic sequence with surges of the proportioning pump. This
effect is shown in Figure '39. These fluctuations were assumed
to be related to variation in junction potential of the reference
electrode, resulting from movement of the solution boundary at
the porous electrode junction. The effect was overcome by
shifting the location of the pump to the debubbling line ahead
of the flow cell, and placing the reference electrode slightly
downstream after the fluoride sensing electrode. With these two
changes, the surging effect of the pump was suppressed and a
smooth curve obtained as recorder output (Figure 39,) .
ftu.£ 6e.fi: Sample, Rat
-------
jl
o
fc
%
_*J
u>
8-S
V
ii
^)i \
ttuj L
•o
o
o
-------
water is 1.0 mg/1 and the buffer to sample ratio is 1:1, 1:2, or
1:5, the final concentration of fluoride in the sample stream is
0.5, 0.67, or 0.83 mg/1, respectively. In addition, simple
dilution might lower the level of fluoride beyond minimum detec-
tion of the electrode or the linear range of electrode response.
A buffer to sample ratio of 1:5 was used throughout this study.
F£ow Rate.*
Early in the study, a net flow rate of 0.95 ml/minute through
the flow cell was used. Electrode response was significantly
improved when the flow rate was increased to 2.04 ml/minute.
This effect is shown in Figure 40 . Higher flow rates resulted
in no further improvement .
Coils are used to mix reagents with test waters in the fluoride
system, and to provide sufficient lag time to allow reaction
kinetics to proceed to completion. A two-minute coil is adequate
for the free fluoride system, but a longer coil is required for
the masking agent to react with Al-F complexes in the total
fluoride system. At A13+ = 1.5 mg/1, recovery of fluoride at
one mg/1 is 92.5 percent complete with a six-minute coil, and
100 percent complete with a nine-minute coil. A 12-minute coil
is used for higher levels of complexes in the test water.
Conc.
-------
time-*
Sample and Wash Time: A
B
Flow Rate: I
II
3 minutes
2 minutes
2.04 ml/minute
0.95 ml/minute
Figure 40. Effect of flow rate versus response time,
97
-------
1
1
/
/
.X
^.rf»
\
\
___y
ILL
- • — -77
.— _i/
J
t
" i^
v^
^J)
\s:
-Ji
c c~
* o> T
•o -§ y
J 1 L
Li. u.
"5
§^
C" ^
li. — o
o> o -^
E I
& ^c.
0
c
QJ
0)
n •
m x5
u
m ro
o a)
O 0)
•H -P
-p 3
to a
i-l -H
rQ g
•H
rH LO
rt
0 -
0)
n e
O -H
m -P
-P X!
3 W
QJ rO
d
O 'd
M fO
OJ
T3 OJ
M rH
O DJ
u e
0) (0
Pi CO
•
H
•^r
Q)
M
3
iji
•H
98
-------
o> 0
."2 -o
II
"o o*
.o .»-.
O
d
-P
O
-P
r—
t
E
cd
0)
3
(d G
M -H
^3 g
-rH
•H CN
(d
o -
-------
A. Calibration
Total Fluoride
Free Fluoride
B. Reproducibility (1.0 mg/! F)
Total Fluoride
Free Fluoride
no Al
0.5 rr-g/i Al
Figure 43.
Recorder output for calibration of free and total
fluoride. Sample and wash time, 1 minute each.
100
-------
show calibration curves made at five, two, and one minute
sample/wash times, respectively. At one minute, reproducibility
was poor, with or without aluminum present (Figure 43) . A
sample/wash time of two minutes each; i.e., 15 samples/hour, was
highly reproducible and adequate for automation.
Recorder outputs for the calibration procedure with sample/wash
times of two minutes each are shown in Figure 42. Fluoride
standards with increments from 0.1 to 2.0 mg/1 were used. Each
standard is represented by two peaks , one for free and one for
total fluoride. Calibration curves drawn from corresponding
peak heights versus fluoride concentrations are shown in Figure
44. A linear relationship was shown to exist for samples to
200 mg/1 or >10~2M F~ , but this level is much greater than any
level expected to occur in drinking water measurements.
Precision of the automated system was evaluated by calculating
the deviation (S) , in mv, from peak heights and determining the
corresponding S in concentration from the following relationships :
and:
where:
= EQ - 2.3 p- log Ai (114)
E2 = EQ - 2.3 p log A2 (115)
EI = potential of fluoride concentration Ci
A! = activity of fluoride concentration Ci
E2 = potential of fluoride concentration C2
A2 = activity of fluoride concentration C2
Ei - E2 = 2.3 P. (log C2 - log d) (116)
r
AE = log C2 - log Ci (117)
slope
Thus, with the potential difference resulting from fluoride mea-
surements, the slope, and original concentration (Ci) known, C2
and AC can be calculated; and the difference in concentration
(AC ) at any other fluoride level (C ) resulting from a potential
variation (AE ) can be expressed as:
101
-------
I I I I I I I
Free Fluoride
Total Fluoride
i i i i i i i
0.2 0.5 1.0
Ruoride Concentrations Img/t)
2.0
Figure 44. Calibration curves for automated fluoride
electrode measurement.
102
-------
Acx • 57 cx AEx
i.e., at 25°C, F~ = one mg/1, and the slope is - 60 mv, the
effect of a one mv potential variation is:
-^ = log C2 - 0 (119)
or:
C2 = 1.04 mg/1 (120)
and:
AC = 1.04 - 1.0 = 0.04 mg/1 (121)
Reproducibility of fluoride measurements at various levels of
fluoride and no complexing agents is shown in Figure 45. The
reproducibility with fluoride of 1.0 mg/1 and various levels of
aluminum is shown in Figure 46. Standard deviations and coeffi-
cients of variation for peak heights in Figures 45 and 46 are
shown in Table 12. The high precision of the method is clear
from the low standard deviations and coefficients of variation.
The concentration effect ranged from coefficients of 0.17 to
0.63 percent with a mean of 0.37 percent in the absence of
aluminum, and from 0.35 to 0.99 percent (1.0 mg/1 F + 1.5 mg/1
Al) with a mean of 0.67 percent with aluminum present. The peak
height coefficient of variation is generally smaller than the
concentration coefficient, but increases as peak height decreases;
thus, concentration coefficients are better measurements of
precision.
ACC.UAO.CI/
Accuracy is measured by mean error (E), which is_defined as the
difference between the actual mean of all data (X) and the true
mean (T.V.); i.e.,
E = X - T.V. (122)
and:
R.E. = XT~VT'V- x 100 (123)
where:
R.E. = relative error
103
-------
Total Fluoride
Free Fluoride
' «
/I //I ,'/1 /I ''A
\l
fi i\
VJ
1.5 mg/l Fluoride
B.
Total Fluoride
Free Fluoride
1.0 mg/l Fluoride
C.
Total Fluoride
Free Fluoride
VI VI
0.6 mg/l Fluoride
Figure 45. Reproducibility of fluoride measurement in absence
of complexing cations.
104
-------
A.
~— Total Fluoride
Free Fluoride
VJ
' I /
i /
i i
i i
i i
i i
\ i \
\ i *
O.I mg/l Aluminum
B.
Total Fluoride
Free Fluoride
/i A *
.V
/I
(
i
/-
i
i
/i
/
'
r
\
f
i
r
\\
i
r
i
f
\
I
/'
'
/-
i'
i
i
i
i
\!
v
V V« v T-'
\J NJ VJ
V!
0.5 mg/l Aluminum
C.
Total Fluoride
Free Fluoride
/•i /
i ;
\!
'*
\
'
N!
!
/ '
/
1
i
i
i
i
I
j
j
i
\
\
\
/
\
\
/
\
\
\
i i
i
i
i
i
i
i
t
1.5 mg/l Aluminum
Figure 46. Reproducibility of fluoride measurement (1.0 mg/l)
in presence of aluminum.
105
-------
Table 12. REPRODUCIBILITY OF AUTOMATED MEASUREMENTS AT DIFFERENT
FLUORIDE LEVELS IN PRESENCE AND ABSENCE OF ALUMINUM
Original
Conc.(mg/l)
F + Al
1.5 + 0.0
1.0 + 0.0
0.6 + 0.0
1.0 + 0.1
1.0 + 0.5
1.0 + 1.5
Fluoride
Form
Free
Total
Free
Total
Free
Total
Free
Total
Free
Total
Free
Total
Fluoride Measured
Mean Value
Peak
Height
(Chart
division)
47.97
54.59
38.28
43.30
25.99
28.18
36.52
43.61
24.58
42.9
-9.26
43.32
F.Conc.
(mg/l)
1.50
1.50
1.00
1.00
.60
.60
.92
1.02
.58
.98
.12
1.00
Standard
Deviation
Peak
Height
.065
.158
.101
.074
.030
.118
.188
.185
.104
.119
.247
.178
F.Conc.
.004
.009
.004
.003
.001
.003
.008
.007
.002
.005
.001
.005
Coefficient
Variation
Peak
Height
(%)
.14
.29
.26
.17
.12
.42
.51
.42
.42
.28
2.66
.411
F.Conc.
(%)
.27
.63
.40
.30
.17
.47
.75
.73
.35
.50
.99
.71
In establishing the accuracy of the fluoride automated system,
data were correlated with SPADNS data for the same samples, ex-
pressed as T.V. Actual data obtained for test waters from two
Michigan cities using both SPADNS and the automated technique
are shown in Tables 13, 14, and 15 (pages 107, 108, and 109,
respectively).
Continuous Fluoride Monitor
Figure 47 is a schematic illustration of the complete fluoride
monitoring system installed in the mobile laboratory. The
system operates on a 26-minute cycle, controlled by the computer,
At initiation of the cycle, the relay is switched to the "on"
position and the three-way valve is connected through to the
baseline solution. For the next 13 minutes, baseline solution
is drawn into the system. Ahead of the pump, a Y-connector
separates the sample and sends it to both the free and total
106
-------
r-
•
Z
o
o
DC
D)
CO
UJ
QC
01
(A
Z
g
oc
UJ
H
LU
Q
Z
O
cc.
LU
Q
cc
o
_i
U.
_£
J2
CO
£ §1
o ~
s
CD
Q. — —— — —
CO
1
C c c-
.2 '| ^>
1 !~~
UJ
.
03 -T3
0) ^ 0 ^
i § g E
LL [i]
Location
1 1
m
"'
CO
q
o
in
00
CN
d
CN
CN
d
CN
CN
d
L_
£
re
re
"3
5
-
in
,_'
CO
o
d
^f ^3 r*^ ^f
CN CN CN C53
T-' ^ d d
O o> r- T-
co q co in
d i-^ d d
in »- o oo
CN T-_ CO ^;
d i-' d d
in o en r»
CN T- CN Tfr
d i-' d d
a- <->
3 o c
Q. Q. re
* I I .
i s| 5 a
« « § ^ •
^ ^ a o g5
^ ^ in S m
S £&i J= «
re re o ~
EL ZZ ^ S
CN co Tt in
CO 00 O CO
q o> q CN
co *— o o
co q r-; «-;
d i-' d ^
8 8 R ^
d «- d *-
O3 CD O «-
CM O> P- «-
d d d --1
•
Q.
co • o
O '~
<% Z - cu .2
^> .ti .i c
UJ O u. D
CO P- 00 O)
CN * CN
CO X CO
d-d
«— o> in
oo r^ r^
odd
OS 00 CO
p** p*» p**
odd
O) O) CM
r^ i*1""" r**«
odd
CD
(/>
3
O
-C c
o
•s >. s
c ^ oo
5 ? =
S re j?
OC _l CO
O «- CN
107
-------
01
X
h-
CQ
z
<
g
i
o
o
5
o
DC
CO
LU
r? to
S8
DC I
LU H
I- LU
LU
sg
ttui
gsi
-2
S5
N Q
S?
LU <
X <
o S
_QJ
JD
CO
c ^
o 01
± £
1 I
E £
D —
*
"P 01
£ S
OJ O
c3 <
D
U_
OJ
03
LL 2
CO
.22 o>
y- ^
to 3
00 z
c
o
1
1 § ' ^ "I 88 ' '
T P P P
I CN |0 || Og ||
|| CN CN ^r «- ooco coin
CDOl CDCM TfOO in^t
CNCN »-CO «-CM r-CN
mCN coco inm »— in inin
,--<- oo oo »-o oo
«- «- r^co -stco coco ^rco
«- «- oo oo oo oo
CN»- incN cooo oo^- r--co
^— r~ Is** r^ oo to oo r*** oo r**1
T-T- 0)10 t^CN CMCD I^O
T- ^- r^p*". oor^- or-* oo oo
^ OQ | | | | | | [|
|l co^- LOCO i>*oo O)O
s <-•
5 S ^ |
5 t '5 2
ro o> C £
•- o S) c
c -3 5 E
CU (O 3 O 03
g Q. I C3 CD
II II II
II II II
O ^" O) 00 CN 00
co r**" r^ co CD ^f
i- CN T- CM i- CN
Sco in co CD in
o o o o o
co co co CD in in
o o o o o o
in i- co in ' in CM
o> r^ r^ *— oo o
oo r^ oo co oo P^
< CD 2
y b bt"
1 LO
q
1 0
O3 CO
r~ co
«- CO
oo r>.
o o
CO CO
o o
CO CM
r«« o
oo r^
1 1
r^. oo
£
O)
o
Z
O5
CM
CM
I
in
*
8
o.
I
CO
O3
CN
CN
_
a
E
to
U.
OI
c
01
E
0}
(U
T3
O
D
C
(O
E
CD
.C
E
o
•g
108
-------
Table 15. SOME STATISTICAL DATA RELATED TO MANUAL AND AUTOMATED
FLUORIDE ELECTRODE MEASUREMENTS IN WATER SAMPLES FROM
WYOMING, MICHIGAN
Parameter
Mean, x *
Std. Deviation, S
Correlation coefficient, r
Degrees of freedom, d.f.
t— value
Statistical difference
Free— F Measurement
Manual
0.737 mg/l
0.2367
Automated
0.704 mg/l
0.2252
0.9931
34
0.4436
N.S.t
Total -F Measurement
Manual
0.952 mg/l
0.3065
Automated
0.955 mg/l
0.3034
0.9989
34
0.0240
N.S.
*AII the 18 samples were included.
'Not significant - no difference could be established.
fluoride systems. Beyond the pump, an air bubble is introduced
to separate the sample stream into many discrete samples and the
appropriate buffer reagent is added. Next, a mixing coil pro-
vides mixing and reaction time before sample reaches the sensors,
Mixing coil length is especially important in areas where free
fluoride is significantly different than total. The sample
stream is debubbled just before it enters the sensor flow cells.
The bubble stream goes to waste under siphon while the remainder
of the sample is drawn by a fourth pump line past the sensors,
then to waste. At the end of the 13-minute period, the computer
sets the relay to "off" and the three-way valve is open to the
tap water sample stream* A complete cycle of 26 minutes is re-
quired to read both baseline and tap water samples. Prepared
reagents are sufficient for ten days to two weeks, depending on
system downtime.
Fluoride sensor flow cells are constructed from glass, not
plastics. As new cells are required, they are aged (placed in
service for 24 hours to reach equilibrium with fluorides in the
test stream) to prevent loss of fluoride from adsorption. Other
glass components; e.g., mixing coils, are similarly aged.
Normal daily maintenance of the fluoride system includes careful
examination of the recorder output to assure that the signal is
of high quality. Bubbles which become trapped in the flow cells
and on the surface of the electrode sensing element are the most
frequent problem. They are removed by briefly pinching off the
109
-------
„. 2
z o
5 "
•• ?f —..,
• VI
VI
Q
— 0
" ""
Ui O U.
S 2 S
u. u. *
Ul
E .
< 0 «•
i-aS:
O -i 3
£ U. «
Ul
»-
VI
<
O
t-
j *i .? •< •* «.
Ul
Q
ee
O
3
_i
u.
_i
<
h-
0
»-
Q
Z
<
Ul
Ul
DC
U.
(9
Z
Z
h-
i!
tt 3
x a.
•t
ui
«i
«•
«
a
«
Ul
0
>
X
v.
X
&
1 STRIP-CHART
•ECOIDER
•
4
M
^
0
u
O
3
>
i
•
111
»-
Ul
X
<
i
X
g
c,
c
~
111
u
ct
o
VI
Z
lil
VI
1
1
1
1
\
)
t*
m
<
N,
i
i
1
(
I
1
1
- -/
~|
)
M
•<
^ M
? 2
•
3
•
IU
a
^
^
u
J
K
9
A
Z
u
A
> X
Z u,
^ H-
Z ui
«- S
[
STRIP-CHART
RECORDiR
S « .
^ " >
5 * <
M V> >
1
1
1
>-
<
^
IU
•1
X
Ul
1-
3
u.
X
0
u
O
U.
•c
IU
_ _ ^_ ^
~ - ^— -J
a.
S
o
u
O
»-
a
-H
-o
-P
-H
c
o
e
in
a)
T3
-H
0)
(D
M
o
O >i
•H H
-P O
(0 -P
M ai
•P M
0) O
•H
(D
O rH
-H -H
-P X5
nJ O
e e
0)
x; Q)
O ,£
en -P
0)
tn
-H
110
-------
flow cell outlet and releasing it quickly, causing a flow surge
in the cell. Persistent bubbles are dislodged by removing the
electrode from the flow cell and replacing it immediately. The
millivolt meter is in the standby mode during this operation.
Standard solutions used for calibration of the strip chart and
computer output are 0.2, 0.5, and 1.0 mg/1 fluoride in distilled
water. Peak heights are measured in chart units (or mv) and
plotted on semilogarithmic graph paper, with peak height on
the linear scale and concentration on the log scale. Figure 48
is an example of fluoride system output and includes a calibra-
tion series. Figure 49 is the calibration curve prepared from
this recording.
[F ] (mg/1)
0.1
0.2
0.5
1.0
Peak Height (C.U.)
Free
0.0
25.0
53.5
78.0
Total
0.0
17.0
36.5
52.5
Fluoride concentrations in tap water are determined by measuring
peak heights and comparing them with the calibration charts. It
is important to note that the three or more points describing
the calibration curve may not be linear, particularly as elec-
trodes age. It is not uncommon to observe a decreasing slope
with decreasing concentrations of fluoride.
Field Studies
The fluoride system was used to acquire data from several special
field studies. Samples from Jackson, Wyoming, and Ann Arbor,
Michigan were analyzed in the laboratory prior to installing the
fluoride monitor in the mobile laboratory.
Jackson, M'i.efo't.gan
Raw water for the City of Jackson, Michigan is drawn from deep
wells (380 to 400 feet) in rock. Treatment includes chlorination
and fluoridation only. Fluoride is added as sodium silica
fluoride. Polyethylene bottles were used to collect 12 grab
samples from the well field, the water treatment plant, and a
number of sites along a principal distribution system pipeline.
Fluoride in these samples was measured in the NSF laboratory by
ion-selective electrode and SPADNS; and iron, by phenanthroline.
Results are shown in Table 13. Iron levels ranged from 0.27 to
111
-------
Cfl
0)
•H
S-l
0)
c
o
•H
-P
X!
•H
O
(0
tn
C
•H
u
c
•H
-P
3
O4
•4->
3
O
o;
-P
tn
cu
T5
•H
M
O
00
OJ
Cn
-H
112
-------
60
50
40
-P
•H
c
3
u
-P
x:
tn
0)
20
10
Free Fluoride
80
60
40
Total Fluoride
0 .1 .2 .3 .4 .5 1.0 .2 .5 1.0
Fluoride Concentration (mg/1)
Figure 49. Typical calibration curves for fluorides, prepared
from output in Figure 48.
113
-------
2.0 mg/1, with a mean of 1.12 mg/1. (The laundry, sample no. 11,
apparently used iron removal procedures to protect against stains.
This sample was excluded from the mean.) Comparison of the data
for free versus total measured fluorides indicated no signifi-
cant complex formation at these levels of iron. This is consis-
tent with previous laboratory observations. The wide variations
in fluoride levels were attributed to inefficient mixing as the
water was fluoridated. Sodium silica fluoride is mixed as a.
paste and fed to the water in pulses, creating a nonhomogeneous
mixture of fluoride and treated water.
The fluoride data acquired by ion-selective electrode and that
from SPADNS were highly correlated; i.e., correlation coefficient
equal to 0.994 and student's t equal to 0.1975, indicating there
was no statistically significant difference between the two
methods.
Wyoming, M-ick-igan
The population of Wyoming is 78,000. Raw water is drawn from
380 to 400 feet deep wells in rock. Treatment includes alum
flocculation, filtration, phosphate addition for corrosion con-
trol, chlorination, and fluoridation. Both grab and weekly
composite samples were analyzed in the laboratory. Results are
shown in Table 14.
Levels of complexed fluoride were clearly significant in the
Wyoming treated water supply. In the distribution system, they
ranged from 14.8 to 33.3 percent of the total fluoride. Key
samples were also analyzed for aluminum and iron by emission
spectroscopy (Table 14). There was close correlation between
aluminum levels and fluoride complexation (correlation coeffi-
cient = 0.96). The data from manual and automated measurement
of fluorides in Wyoming samples are compared statistically in
Table 15. Again, the two methods were not significantly different,
Ann
Ann Arbor has a population of 84,000 and takes 80 percent of its
raw water supply from the Huron River, and 20 percent from wells
in drift, 35 to 56 feet deep. Treatment includes alum floccula-
tion, lime softening, filtration, phosphate addition for corro-
sion control, chlorination, and-f luoridation. Data from 15 loca-
tions sampled on April 6, 1971 are shown in Table 16. Most of
the Ann Arbor area is represented in three samples. Aluminum in
these samples was <0.02 mg/1; and iron, <0.1 mg/1. No complexing
effects occur at these levels; i.e., free and total fluoride
levels are essentially the same. It is further suggested by the
data in Table 16, that fluoride levels are similar throughout the
distribution system; i.e., they are not affected by length of
pipeline or main residence times.
114
-------
Table 16. CHARACTERIZATION OF FLUORIDE IN WATER SAMPLES FROM
ANN ARBOR, MICHIGAN (APRIL 6,1971)
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Location
River raw water
Well raw water
Plant finished effluent
Shell station, 1251 N. Maple
Fire station, 2130 W. Huron
Standard station, 1336 S. Main
Lakewood School, 344 Gray Lake
Residency, 2250 S. Seventh
Water Department Distribution
Center, S. Industrial Highway
Main fire station, 219 E. Huron
Northside School, 91 2 Barton
Bolgos Farms Store,
3601 Plymouth Road
Concordia College, 4090 Geddes
Mary D. Mitchell School,
Pittsview Drive
Newport Elementary School,
2776 Newport Road
Free Fluoride
(mg/l)
.14
.30
.96
.94
.95
.96
.95
.98
.98
.95
.93
.96
.96
.96
.98
Total Fluoride
(mg/l)
.14
.30
.97
.95
.96
.96
.94
.98
.98
.96
.93
.94
.97
.97
.99
Average values of fluoride analyses from different locations in
Ann Arbor at various times during the period 1970 to 1972 are
shown in Table 17. During 1970, fluoridation was inadequate.
Fluoride concentrations were considerably below the optimum level.
115
-------
Table 17. CHRONOLOGICAL CHARACTERIZATION OF FLUORIDE IN WATER
DISTRIBUTION SYSTEM OF ANN ARBOR, MICHIGAN
Date
July 28, 1972
April 3, 1972
January 20, 1971
September 16, 1971
April 6, 1971
March 22, 1971
March 9, 1971
November 6, 1970
October 4, 1970
October 15, 1970
October 5, 1970
Free Fluoride*
(mg/1)
0.98
0.98
1.08
1.11
0.96
0.97
0.94
0.43
0.59
0.29
0.49
Total Fluoride*
(mg/1)
.99
1.00
1.10
1.11
0.97
0.98
0.94
0.50
0.64
0.33
0.53
Complexed
Fluoride
Percent
NS
NS
NS
NS
NS
NS
NS
14.0
7.8
9.1
7.5
*Values presented are mean values of >10 samples
Aluminum analyses showed variations from <0.02 to 0.5 mg/1, and
iron varied from <0.015 to 0.9 mg/1 during that period. Fluori-
dation has improved since early 1971. Levels now approach
1.0 mg/1, the optimum level. Aluminum is generally below
0.05 mg/1; and iron, below 0.3 mg/1.
Ck^icago, ILL-ino-ib
Raw water for the City of Chicago is entirely of surface origin.
It is drawn from Lake Michigan and treated in one of two plants,
the Central- or the South- Water Filtration Plants. The combined
rated pumping capacity from these two plants is nearly three
billion gallons per day. Distribution is made through more than
4,000 miles of tunnels and mains to all of Chicago and 72 subur-
ban communities. Treatment at both plants includes alum-lime
flocculation, high rate filtration, chlorination, and fluorida-
tion. Caustic for corrosion control is added at the Central
Plant.
During its first month's assignment in the Chicago area, the
mobile laboratory was located at a fire station, 10.5 miles from
the Central Plant. Average daily mean, high, and low free and
total fluoride levels recorded on coincident days of normal opera-
116
-------
tion are summarized in Table 18. Although aluminum is not mea-
sured in the mobile laboratory, the water treatment plant labora-
tory reported a level of 0.16 mg/1 in routine quarterly distri-
bution system samples from the north district. It is apparent
that complexed fluorides are important in the Chicago water
supply.
Table 18. SUMMARY OF FREE AND TOTAL FLUORIDE DATA
(CHICAGO/LEHIGH STATION)
Total Fluoride, (mg/1)
High
1.10
Low
.66
Average
Daily
Mean
.90
n
15
Free Fluoride, (mg/1)
High
.90
Low
.47
Average
Daily
Mean
.71
n
15
Beginning at 1200 hours on May 25, 1973, fluoride feeders at the
Central Plant were taken out of service. The purpose of this
operational change was to observe the capability of the fluoride
monitor in the NSF/EPA mobile laboratory to detect the change,
and to determine residence times in the distribution system.
Results are reflected by a reduction in total fluoride levels
recorded from 1200 to 2000 hours.
This study was repeated on May 29th and 30th with fluoride feeders
cycled in- and out-of-service two times at four—hour intervals.
The first out-of-service period was from midnight May 28th to
0400 hours May 29th. Results are plotted separately in Figure
50. The data show that transmission time to the Lehigh location
during periods of low demand is 17 +_ 1 hours. This is apparent
from the first recorded drop in total fluoride. When the feeders
were placed back in service, recovery was incomplete because of
the low demand period and mixing in the distribution system.
Also as a result of these factors, the second decrease was signi-
ficantly lower than the first, and lasted until demand increased
during the morning hours.
The City of Chicago operates a high purity aluminum electrode
fluoride monitor (Fisher Porter) at another fire station 8.5
miles from the Central Filtration Plant. The relative locations
of these stations are shown in Figure 51. Data recorded by the
Chicago monitor during the May 29th-May 30th period are shown in
Figure 52. The'actual levels of fluoride are not expressed on
the recorder output, but it is interesting to note the relative
117
-------
S/I
s/o
S/I
o
H
w
t-»
i
O
O
-l OJ
0) U
in -H
009T ° »
0 C c
0-H o
QQfrT w w '^
H H (^
0) 0) 4J
TJ TJ U)
0) 0)
(U 0) ,£
H II -?
CO CO r"
OOOT O H
0080
0090
QOfrO
0020
OOfrZ
OOZZ
0002
008T
0080
o
Cn
O
•H
U
-P
CO
o* o en
CD n
2 . cu
u
-------
?-) II ^^ 1 ''•-'l.n.—, ux J-
Figure 51.
Map identifying relative locations of monitoring
sites in Chicago.
119
-------
cr>
Ul
(0
C
OJ
0)
H
O
O
tn
(0
O
-H
J3
U
0)
tJ
M
O
<1J
C
O
0)
T3
•H
M
O
O
•H
A
U
-P
3
CU
4J
3
O
CN
ID
(U
01
•H
120
-------
times at which peaks were recorded by the Chicago monitor versus
those by the NSF monitor:
1st minimum Crest 2nd minimum
Chicago Monitor 1650 1915 2250 (May 29)
NSF Monitor 2000 2200 0600 (May 30)
(Reference to Figure 51 indicates more direct transmission to
the Chicago monitor than to the NSF mobile laboratory.)
A subsequent fluoride detection study occurred at the mobile
laboratory's fourth location in the Chicago area, Calumet, which
receives its water from the South Water Filtration Plant. Be-
ginning at 1000 hours on June 25, 1973, fluoride feeders were
interrupted twice, with two consecutive on-off cycles. These
data are plotted in Figure 53.. Note that times of recorded
changes at the South Plant outlet are shown in this figure.
The first change at the treatment plant outfall was detected
5-1/4 hours after the feeders were taken out of service. The
first minimum level was recorded at the plant outlet 9-3/4 hours
after the feeders were out-of-service. This level was recorded
by the NSF monitor after 31 hours.
TRACE METALS
General
Trace metals in public water supplies are of special concern be-
cause of their potential long-term chronic effects on the con-
sumer. Surface and ground supplies vary in types and quantities
of metals constituents but, in general, both are significantly
lower than EPA Interim Primary Drinking Water Standards (1) .
Metals are not removed by conventional water treatment processes.
In addition to their occurrence in raw waters, they may be intro-
duced to treated waters as impurities in chemicals used in the
treatment process, or as products of corrosion in municipal dis-
tribution systems or individual household plumbing.
Simple analytical methods for detecting and measuring trace
metals in water supplies are not generally available. Except
for neutron activation analysis (NAA) and anodic stripping voltam-
metry (ASV), direct measurement of trace metals cannot be made
without concentration of the test solution; i.e., separation, di-
gestion with acid, and concentration, which frequently modifies
the physicochemical characteristics of the species of interest.
Although activation analysis offers the high sensitivity required
for trace analysis, its use is limited by availability of reactor
facilities. Further, it is an elemental analysis procedure that
121
-------
5/1-
s/o-
s/i-
w
•
*
.
0
•
•
• ©
\
•
• ©
•
• 0
•
•*
D
• °0«—
l^«—
4> O) ^ —
S / o-1 — v 1 1 > 1 1 1 1 1 1
00tr2
"0002
"008T
"0091
"OOfT
"002T
"0001
"0080
"OOfrO
"0020
"OOt-2
"0022
"0002
^008T
"OOOT
"0020
"008T
OOOT
o
•H
Vj to
0) O
O Q)
3 C
O-M
(fl 01
M M
0) t
m
S
t
.1
CN
*5
ftf
S3
-P
0)
u
-p
(tf
-P
tn
0)
•O
-H
M
O
O
(U
ro
Q)
-H
Cn
122
-------
provides no information about oxidation state or degree and type
of complexation.
The analytical feasibility of anodic stripping voltammetry
for trace metals analysis in natural and wastewaters has been
established (46). In addition to its high sensitivity, simpli-
city, and portability, ASV has the unique capability of charac-
terizing free versus complexed trace metals.
Theory
ASV consists of: (1) a preconcentration (preelectrolysis, catho-
dic deposition, or plating) step, in which the metal ion (or ions)
of interest is reduced by controlled potential electrolysis and
deposited as an amalgam on the surface of a composite mercury
graphite electrode (CMGE), and (2) anodic dissolution (reverse
electrolysis, or stripping), in which the metal is oxidized and
returned to solution. ASV is, therefore, a nondestructive
method of analysis.
The plating step is carried out under reproducible conditions so
that concentration onto the electrode is quantitative (total
plating), or a reproducible fraction of the desired component is
deposited from solution (partial plating). By controlling the
potential during this process, the more easily electrolyzed con-
stituents can be separated; thus, the plating step is achieved
by applying to the electrodes a potential which is cathodic of
the polarographic half-wave potential by approximately 0.3 to 0.4
volts (where the current has its limiting value), and maintaining
this potential for a given period of time under reproducible con-
ditions of stirring, type, and area of the microelectrode, and
composition of the medium. After a short rest period (which
allows the solution to become quiescent, the stripping process
is initiated.
The stripping process is performed by a voltammetric scanning
procedure which produces a response proportional to the amount
of material deposited. The resulting "stripping voltammogram"
produces peaks, the heights of which are proportional to the con-
centrations of corresponding electroactive metal ions, and the
potentials, a qualitative indication of the nature of the species
present in the solution. Thus, the important characteristics of
the voltammogram are: the heights of the peaks (peak current,
i , in microamperes), peak width at 1/2 i (W 1/2 in volts or
mSllivolts), and peak potential (E , in vBlts). These charac-
teristics are affected by the type"of microelectrode used, and
by the rate of voltage change (sweep rate, v, millivolts per
second) employed in the stripping process.
Three electrodes are used in conventional ASV: reference
(Ag/AgCl), counter (a platinum wire), and test or working (mer-
123
-------
cury-graphite) electrodes. Test electrodes are made by plating
a thin film of mercury onto a polished, wax impregnated graphite
rod. Because oxygen is an interference in trace level analyses,
the flow cell has provision for nitrogen deaeration. The analy-
tical curve (voltammogram) includes peaks for background current
as well as for metals being measured.
Differential ASV (DASV) is a four electrode system. A second
working electrode is used to subtract extraneous background
signals from the voltammogram, significantly improving the sensi-
tivity of the technique. The fourth electrode is identical to
the working electrode except that it is excluded from the plating
step. During the stripping procedure, the background current
signal from the fourth electrode is subtracted from the analytical
current signal producing only current peaks for the metals of
interest. The special DASV plating-stripping sequence is dia-
grammed in Figure 54.
In the voltammogram, the peak value of the current is proportional
to the concentration of the reduced species, the area and thick-
ness of the electrode, and the potential scan rate, as shown in
Equation 124:
i = (ZFA1 v)CR (124)
where :
i = peak current
•L = number of electrons transferred
F = the Faraday
A = electrode surface area
1 = thickness of mercury film
d> = 2!
* RT
v = potential scan rate
e = Naperian logarithm
C^ = concentration of reduced species
K.
Film thickness is determined by controlled electrolysis of the
mercury onto a graphite electrode. The value of i varies nearly
proportionally to the scan rate for very thin mercery films;
however, it varies significantly with film thickness at higher
sweep rates. For lower values of one, i varies only slightly
with film thickness. P
Equation 125 expresses peak position:
(125)
124
-------
-P
CD
T
*4
CD
o
w -—
« '5
w C
-8 s
03
SI
s
• M
H
-P
iH
O
•H
PU
a
•H
S-l
-P
en
o
•H
t3
o
£
ftf
•H
-P
C
CD
M
0)
m
m
-rH
5-1
o
a)
u
C
QJ
CD
tn
•H
!-l
-P
W
I
CD
-P
-H
125
-------
where:
E = peak potential
6 = Nernst diffusion layer
D = diffusion constant
E = standard potentia-l
E , then, is a function of scan rate, film thickness, and the
Nfernst diffusion layer. This theory does not apply for fast
scan rates with very thick films. Peak width, which is important
in resolution, varies only slightly with scan rate when the
mercury thickness is small, and varies slightly with thickness
with a slow scan rate.
Experimental
Equipment items in the trace metals monitor in the mobile labora-
tory include a custom built, four electrode potentiostat/timer
(P/T) unit with automatic and manual modes; Technicon 16-channel
proportioning pump and long jacketed mixing coil; Varian recorder,
Model G2500; two Valcor miniature three-way valves; a customized
Plexiglas flow cell; and four electrodes, Ag/AgCl reference,
platinum (Pt) counter, and two graphite working electrodes.
In the automatic mode, all functions are controlled by the timer.
There are six variable timed functions, five of which are necessary
for one complete cycle:
1. Warm up (200 to 2,000 sec.)
Allows time for system to warm up and sample
to travel between its point of origin and the
flow cell. (This function is used only when
system is turned on.)
2. Plating primary electrode (100 to 1,000 sec.)
Switches primary electrode to plating poten-
tial and initiates nitrogen mixing.
3. Plating differential electrode (10 to 100 sec.)
Switches differential electrode to plating
potential and stops nitrogen mixing.
4. Preelectrolysis procedures (1 to 10 sec.)
Starts recorder and lowers pen.
5. Analytical step (2 to 20 sec.)
Starts potential sweep on both electrodes
simultaneously. (Sweep stops automatically.)
6. Wash period (30 to 300 sec.)
Starts nitrogen mixing, allowing cell to be
flushed with new sample prior to analysis.
(System switches back to function 2.)
In the mobile laboratory, the time for one complete DASV cycle
is 13 minutes. Plating potentials for each function are set at
126
-------
a digital panel meter on the front panel of the control unit. A
digital counter counts the number of analytical cycles which have
occurred and, at some preset number of cycles, switches a valve
at the beginning of function six, pumping standard instead of
sample through tha cell.
Sample is delivered to the flow cell, acid added, and standard
pumped with the Technicon proportioning pump. Nitric acid is
mixed with the test water in a ratio sufficient to produce a
resultant pH of 2.5. (Metals exist in water in three forms:
free ions, labile complexes, and nonlabile complexes. ASV does
not measure the nonlabile complexes unless the sample is acid-
ified to either break down the nonlabile metal complexes or
exchange the H for a metal ion. Total metals are measured only
with acidification.)
A ten-inch Varian recorder with remote start/stop, remote pen
lift, event marker, and a Z-fold paper tray is used with the DASV
system. (An event marker is available and could be used to re-
cord the beginning and end of a standardization sequence.)
A water jacketed mixing coil maintains constant sample tempera-
ture and aids in the control of outgassing. Bubbles formed by
outgassing are removed through a debubbler before the sample is
acidified.
The flow of nitrogen is initiated by one of the Valcor valves in
the system; the second valve switches from sample to standard
in the standardization sequence.
A schematic illustration of the Plexiglas DASV flow cell in the
mobile laboratory is shown in Figure 55. The internal volume of
the cell is ten ml. Test electrodes are inserted through the
sides of the cell, directly opposite each other, and mounted
with epoxy to 1/4-inch hose conversion fittings for easy removal
for servicing. Reference and counter electrodes, also directly
opposite, are inserted through the front and back of the cell.
The test solution is pumped in through the bottom and overflows
through a port in the stopper at the top. Nitrogen enters
through a small diameter polyethylene tube inserted in the bottom.
Working electrodes are made from 1-1/2 inch lengths of 5/16 inch
diameter wax impregnated graphite rods. The end of each is
polished for working electrode surface. Complete instructions
for electrode preparation are included in the "Operation and
Maintenance Manual," provided with the mobile laboratory (47).
A schematic illustration of the total DASV system is shown in
Figure 56. Before plating is initiated, both working electrodes
are at rest potential and the solution is mixed. To initiate
plating, the analytical electrode is adjusted to plating poten-
tial; the differential electrode remains at rest potential. When
127
-------
To Waste
Rubber
Stopper
Ag/AgCl Reference
Electrode
Graphite
Electrode
Pt Counter
Electrode
Threaded
Electrode
Holder
Graphite
Electrode
Sample
Inlet
Plexiglas Flow
Cell Body
Nitrogen Mixing
Gas Inlet
Figure 55.
Schematic diagram of DASV flow cell used in
the mobile laboratory.
128
-------
Ill
OK _j
III ft.
«• 5 *
2* 2
o
Q
Z
<
w»
Q
^ «
Ul < Q
i!|
*n *n >
<
1
\
i
! {
, o E
? ^ 5
..1
u
<
O
z
*-
o ,
o s
« 3
tL. O.
X^
/ « - «.
'| i 5 3 > «
- 1 1 2t < 5
« i «. u. • > _ 5
i , ... , *
i
1
i I
£ i ' *
o j 1 J os..
z o 1 ; ^ - J 2
S =: S I | z" £ 5
i i i ;
i i
ii i
1 ; ;
• i
1 1 1
1 ! ^ a
O
z
UJ 0
it o s
t 2 O
Q < >
Z
O IK
« -1
>- 3
U O
" i
m <
± O
O
4-)
fO
M
O
(U
H
•rH
ft
i
Q)
XI
4-)
M
O
-P
•rH
C
i
in
-P
0)
S
QJ
O
(0
M
-P
c!
o
-H
-P
(0
L-.
U
-H
4->
rtj
a
0)
XJ
O
CO
in
OJ
tr>
•H
129
-------
plating is completed, mixing is stopped and the differential
electrode adjusted to plating potential. After a short waiting
period (to attain solution quiescence and electrical stability),
the stripping step is initiated by simultaneous application of
the voltage sweep to both electrodes. The sweep is stopped at
a predetermined potential and rest potential is again applied to
both electrodes. Typical DASV voltammograms for standards and
a tap water sample are shown in Figure 57. Preparation of rea-
gents for DASV analyses is detailed in Appendix B.
Results and Discussion
Performance characteristics of the DASV system were determined
experimentally. Calibration curves from standard solutions of
5, 10, and 15 ppb of each of the metals of interest (Cd, Pb,
and Cu), run in triplicate, established'a linear response for
each of the metals. The plating time response curve for discrete
samples is S-shaped with a limiting value representing total
plating of a metal. In the continuous flow system, the plating
time is linear as a result of the continuous flow of fresh
sample; i.e., total plating cannot occur.
Continued use of working electrodes results in reduced signals
for each of the measured metals. Periodic standardization is
required to observe changes in electrode sensitivity. Data in
Table 19 relate to a tap water sample spiked with five mg/1 each
of Cd, Pb, and Cu. The sample was cycled through the flow cell
continuously for a period of 18 hours. Although electrode re-
sponse decreases with time (Table 19), the rate of decrease is
relatively constant; thus, results can be read from a calibra-
tion curve corrected for elapsed time. Periodic rejuvenation
of the electrodes can be accomplished by a short period of
plating additional mercury, as shown in Table 20. These data
are from a pair of electrodes used continuously for alternate
standard/sample cycling over a 48-hour period. The electrodes
were rejuvenated by short mercury plating periods after each
24 hours of use.
Electrode response to variations in sweep rate is shown in
Figure 58. Rapid sweep rates produce higher current response;
thus, 162.5 mv/sec. was used for much of the DASV acquired data.
Special Study
A special study --of the effect of residence times in household
plumbing on metals uptake in drinking water was undertaken during
this project. NSF employees were asked to bring two samples to
the laboratory each day for five consecutive days. One sample
was drawn from the tap immediately upon arising each morning;
i.e., before any other water was drawn from the tap. The second
130
-------
Tap Water
Analyses
Standards
0.005 mg/1
Cd. Pb, and Cu
Figure 57. Typical voltairanograms of a" standard" solution and tap
water analyses.
131
-------
24 -
10 ppb lead
5 min plate
10 ml sample
20 -
I 16
•rH
-------
Table 19. ELECTRODE DETERIORATION
Run
1
10
20
30
40
Cd ACd
33.0
6.0
27.0
4.0
23.0
5.0
18.0
4.0
14.0
Pb APb
44.0
1.0
43.0
1.0
42.0
0.5
41.5
0.5
41.0
Cu ACu
94.0
6.0
88.0
8.5
79.5
8.5
71.0
8.0
63.0
Table 20. ELECTRODE RESPONSE AFTER REGENERATION
(10 minute plating times; 10 ppb each
metal)
Start
24 hrs.
48 hrs.
Cd
23
24.5
27.0
Pb
22
25.5
27.0
Cu
50.5
54.0
55.0
sample was drawn from the same tap after the water had run contin-
uously for three minutes. Samples were collected in acid-washed
polyethylene bottles provided by the laboratory. For the analysis,
ten ml of sample was placed in the flow cell and acidified to
pH 2.5. Plating time was five minutes. Household systems in-
cluded in the study are described in Table 21. Results are sum-
marized in Table 22 and plotted in Figures 59., 60, and 61.
Regardless of the sample or characteristics of the system from
which it was obtained, levels of Cd were insignificant (Figure 59).
None of the metals was found at a significant level in samples
from galvanized plumbing. Relatively high levels of both Pb
and Cu were shown to accumulate with overnight residence in cop-
per plumbing systems, but these levels were quickly reduced with
flushing. (Figures 60 and 61.)
This study should be repeated in cities with more aggressive
drinking water supplies. Ann Arbor water has an unusually high
133
-------
CD
12
•S
3
S
O.
O,
«o
e
•H
O
0)
CO
TJ
O
O
1
B
I
4)
«4
• i^
en
M
3
SD 3
'S .S
S S
> A
o
-P
0)
Q)
U
O
£1
0)
cn
CTl
in
a)
tn
•H
qdd
134
-------
I I I
immnniiiiiin
~o
0
0)
tfi
'£*
o
03
13
o
in
CM
4)
en
a
I £
to
QJ
CM
"•^
0)
rt
-P
(U
e
0)
o
O
A
0)
0
X
0)
•H
IT)
CM
qdd
135
-------
S-l
3
tn
w
i-H
(0
-p
i
OJ
0
(0
1
I
in
rt
o in
CO fM
•fe-\v
I
00 (O
a)
en
^
O
S3
vo
Q)
S-l
-H
fci
qdd
136
-------
Table 21. TRACE METALS SURVEY SITES
Site No.
Description
Approximate
Building Age
(yrs)
Plumbing
1 Single family home
2 Single family home
3 Single family home
4 2nd story apartment in 2 unit dwelling
5 2 story town house in new complex
6 Single family home
7 Commercial Building
8 Single family home
9 Single family home
10 Single family home
11 Single family home
12 2nd floor apartment in new apartment complex
13 Sub-basement in old University of Ml laboratory
14 Basement in new University of Ml laboratory
15 6th story in new University of Ml laboratory
10
13
17
35
2
49
10
20
12
20
1.5
1
30
2
2
Galvanized iron
Galvanized iron
Galvanized iron
Galvanized iron
Copper
Copper installed 2 years ago
Copper
Galvanized iron
Copper
Galvanized iron
Copper
Copper
Galvanized iron-copper mixed
Copper
Copper
137
-------
Table 22. SUMMARY OF RAW DATA FROM HOUSEHOLD TRACE METALS SURVEY
GALVANIZED IRON PLUMBING
Site No.
1
2
3
4
8
10
Average
Cd ppb
O.R.*
0.16
0.11
0.12
0.06
0.5
0.29
0.2
3 +
0.08
0.08
0.08
0.05
0.11
0.04
0.07
Pb ppb
O.R.
2.68
2.84
1.24
1.4
1.35
1.84
1.9
3
1.75
1.46
1.15
1.2
0.5
0.48
1.1
Cu ppb
O.R.
0.64
2.78
0.35
0.22
1.65
1.58
1.2
3
0.42
0.39
0.45
0.14
0.54
0.26
0.4
COPPER PLUMBING
Site No.
5
6
7
9
11
12
13
14
15
Average
Cd p
O.R. i
0.02
—
0.25
—
0.01
—
0.26
—
0.06
pb
3
0.02
—
0.04
—
0.01
—
0.03
—
0.01
Pb ppb
O.R.
3.22
1.35
4.28
11.5
4.18
3.5
18.9
5.25
23.4
8.4
3
1.86
1.08
0.92
1.0
0.47
0.83
0.58
0.78
2.3
1.1
Cu ppb
O.R.
33.9
28.0
27.5
9.2
32.7
10.2
6.64
25.6
21.0
21.6
3
5.9
7.0
1.97
0.62
3.4
1.2
0.38
3.6
1.8
2.5
*O.R. = Overnight Residual
3+ = 3 minutes later
138
-------
pH (>_10.0 during the week of the study) . A comparable study with
tap water pH <_ seven is recommended. Plastics piping systems
should be included in the study.
139
-------
SECTION V
MOBILE LABORATORY
General
In the final phase of the NSF water quality monitoring project,
the prototype monitor; i.e., all instrumentation and ancillary
equipment, was installed in a mobile laboratory and tested in
actual field operation. Important design criteria included
ruggedness of instrument mountings and overall laboratory porta-
bility. Ideally, the laboratory had to be easily and quickly
movable and capable of being installed virtually anywhere on a
potable water distribution network.
Physical Description
A heavy duty, intercity CMC delivery van with 4.9 meter long by
two meter high load space was selected to house the NSF/EPA
mobile water quality monitoring laboratory. The van is designed
to carry heavy loads over long distances. It rides on six wheels
(dual rear wheels) and includes heavy duty suspension to protect
the delicate instruments against extreme road hazards. Its load
space is sufficiently large to provide adequate benchtop work
space, and sufficiently high to permit persons of average height
to work in comfort. All cabinetry, electrical, and plumbing
installations were customized to NSF specifications by a local
general contractor. Cabinetry was constructed from plastic
laminated heavy composition board, typical of that which is
found in household kitchens and benchtops, covered with "matte
white" formica.
The sink unit contains a closed cabinet and an electric heater
with thermostatically controlled blower. An additional base-
board heater is provided on the opposite side, at the opposite
end of the van. For cooling, a 12,000 BTU ARA mobile home air-
conditioner is mounted in the ceiling.
External power (100 amps, 220v, single phase) is brought to the
van over a four conductor, 100 feet long by one inch diameter
cable which weighs slightly over 100 pounds. When it is con-
nected to the van, the cable splits at the circuit breaker box
to two, llOv circuits. The two main circuits are further divided
into nine smaller circuits for lights, airconditioner, left wall
outlets, right wall front and right wall rear outlets, heaters,
140
-------
hot water heater, computer, and Schneider Robot Monitor systems
No overloads were observed during six months of mobile labora-
tory operation.
NOTE: A ptitnctpat Au.st.Qe.-type. powe.fi tie.qu.tie.me.nt tA e.x.e,fite.d by
tke. kot wa.te.si kaa.te.ti. Tkt& i.te.m tt> not c.onA-Lde,sie.d e.A
tn any ^utasie. psioje,c.t& oft tkt* k-ind. One, ku.ndfie.d amp
womtd not be, sie.qu.tsLe.d
-------
Figures 62 and 63 are schematic plan views of the equipped van
interior. In Figure 62, "reagent reservoirs" contain, from left
to right, alkalinity baseline solution, alkalinity buffer, free
fluoride buffer, total fluoride buffer with masking agent, and
fluoride baseline solution. Sensor flow cells and mixing coils
for free and total fluoride and alkalinity are mounted on the
platform to the left of the pump.
The flow cell, mixing coil, and run/by-pass switching values for
DASV are mounted on the H-type rack near the center in Figure 63.
The rotator, electronics module, and recorder for CCDT, and the
corrosion rate monitor are not fixed in place. For travel,
these instruments are stored on thick pads on the floor. A
photograph showing an overview of the interior arrangement of
the mobile laboratory is shown in Figure 64.
Computer System
All sensor systems in the mobile laboratory are controlled by an
on-board mini computer system which includes a Texas Instruments
digital computer, Model 960-A; Texas Instruments Silent 700/30
teleprinter; combination high speed paper tape reader perforator;
and a Computer Products wide range analog to digital (A/D) con-
verter, Model RTP7480. The computer contains 16K (16,384 words)
of semiconductor memory, expandable internally to 32K and, with
the addition of an external chassis, to 64K. An internal timer;
16 input/output digital switching board wired for interrupt; and
an internal communications register unit (CRU) expansion chassis,
required for five to 20 peripheral connections, are options used
with the computer. Seven CRU connections are used in the mobile
laboratory.
The gain of the wide range A/D converter is software programmable
and output resolution is 13 bits (gain setting/4095) for full
scale input voltage ranges of +2, 5, 10 mv ... +10.24v. Input
from parametric systems to the converter, through Belden 8451
shielded twisted pairs plus ground cable, is made by connection
to multiplexer boards mounted in the converter. Up to 16 eight-
channel multiplexer cards can be installed internally - and 48
externally - to the converter chassis, for a total of 512 analog
inputs. Two multiplexer cards (16 channels) are installed in
the mobile laboratory.
The high speed paper tape reader/perforator reads 300 and punches
75 characters per minute. The Silent 700/30 teleprinter writes
on heat sensitive paper, operating at printing speeds up to 30
characters per second. Its quiet operation makes it ideally
suited for the mobile laboratory application.
The physical relationship of the Texas Instruments (TI) 960-A
computer system to sensor equipment on the mobile water quality
142
-------
«
UI
»-
UI
X
>
2
*^
X
a
TOTAL
UI
UI
m
c
• c
RESIDUAL
CHLORIN
RESIDUAL
CHLORINE
w>
•c
o
at
UI
VI
UI
•c
UI
O
<
UI
K
^
X
a
>•
H-
ALKALIN
SCHNEIDER
ANALYZE!
ANALYZER
"2C
^
?c
C
•c
1 RECORDI
A
/
t 0
0 ^
• z
i i
a,
5
3
a.
..
I =
I <
VI
ui m
FLUORIC
RECORDI
ROBOT
f
I
<
(
f
(
(
..
MONITOR
WHEEL
at
UI
^
<
ui
Z
IW
U
<
v>
p
<
<
«l
-^
UI
$
O
-P
a)
rH
•H
X!
o
4-1
0
M
o
•H
i-l
0)
-p
c
•H
w
.p
X!
tJ>
•rH
o
o
-H
-p
Hi
n
-p
w
•H
O
-P
6
0)
u
CO
0)
i-l
143
-------
ED
PE
PUN
HIGH S
PAPER
READER
COM
ANG
A/D
CONV
32
u o
u £
si
O w^
.
o
111
0
3
§ i i
S
.
n
o
4J
(0
(-1
O
X!
QJ
H
•H
X)
§
o
•H
n
0)
•4J
c
-H
-------
o
•H
M
0)
-P
G
•H
O
-P
m
H
QJ
X!
O
e
4-1
o
•H
M
<1J
O
•H
^
O
Cn
O
-P
O
A
CU
Q)
•H
PM
145
-------
monitoring laboratory van is shown in overall view in Figure 65.
Digital input from the anodic stripping voltammetry equipment
is used to trigger the initiation of a data sampling cycle.
Within the data sampling cycle, appropriate digital outputs are
sent to the corrosion, alkalinity, and free and total fluoride
measuring equipment. Independent programs operating from the
internal interval timer in the computer send digital output
signals to the sampler and control the flow through the nitrate
and hardness cells, located in the Schneider Robot Monitor.
In the data sampling cycle, the internal interval timer controls
the timing for reading the analog signals from each of the
different sensor instrumentation packages. The data acquisition
program in the computer determines the gain to be used in the
analog-to-digital converter (ADC) in reading each analog input
channel.
Within the computer, all program operations are under the control
of the Texas Instruments supervisor program PAM (Process Auto-
mation. Monitor).
Eight worker tasks have been installed to operate in the computer
under the control of PAM. Interrelations between the worker
tasks, described as follows, are shown in Figure 66.
1. NSFC is a program containing utility routines,
such as a decimal core dump routine. It is
built to facilitate adding additional options.
After NSFC is called into execution (either
by Control-X or NSFC in response to OP?), a
two-character code is used to select the op-
tion desired for execution. NSFC must be
assigned Task ID 30.
2. BELL is a simple program which rings the bell
on the teleprinter repeatedly until the opera-
tor halts the program. Its purpose is to
attract the operator's attention.
3. INIT initializes output storage area, checks
PUN, starts flow control, periodic sampling,
and data acquisition. It is used for start-
up when the system is initially loaded. (By
changing the operation of the ST option in
NSFC to bid task 32-INIT- instead of task
34-DAQ, the need to start system operation
through job control can be eliminated.)
4. DAQ is the basic data acquisition program
which reads the various analog input channels.
It is initiated by an interrupt signal which
indicates that DASV is ready to input data.
During the next 13-minute interval, DAQ takes
readings from each sensor system except alka-
146
-------
n
w
g
Q)
4-1
W
>1
W
O
in
o
•H
-P
a
fl
m
o
-P
e
0)
4J
W
>!
U)
M
0)
cu
g
o
u
4-1
o
a,
-H
O
-H
-P
fC
rH
Q)
m
\i>
a)
K
3
Cn
-H
147
-------
o
<
O
z
z ^
u O >-
u. U 3
Z
O
« -
O 5
I >
5 «
•* IM
00.
O =>
z *
O4
•H
X!
cn
•I
o
O
•H
4J
QJ
(U
S-l
td
O
w
e
OJ
tn
s-i
OJ
-p
=1
cu
e
o
u
QJ
148
-------
Unity and corrosion rate, and switches
the fluoride systems from baseline to
sample. During the second 13-minute
period, it takes readings from all sensors,
including alkalinity and corrosion rate.
At the end of the full 26-minute cycle, it
waits for another interrupt signal from
DASV to begin the cycle again.
5. DATA obtains data from the buffer area of
DAQ, processes the data as needed, and
stores the results in the output storage
area.
6. FLOW controls the flow of water to the
nitrate and hardness cells in the Schneider
Robot Monitor.
7. SMPL controls the sampler to collect hourly
samples of the water flowing through the
mobile laboratory.
8. PUN outputs data from the output storage
area and clears that area for fresh data.
Sufficient space is available in the allo-
cated storage area to retain up to two
hours of data. DATA unsuspends PUN whenever
more data is put into the output storage area.
Currently, PUN outputs data and clears the
area each time it is unsuspended; however,
PUN can be modified to count the number of
times it has been unsuspended since the
last output of data, outputting data at
greater intervals.
Flow charts of each of the worker tasks are shown in Figures 67
through 75. The flow of information through the system is des-
cribed in Figure 76.
A detailed description of the operating system, DAQ, an operator
manual for the computer system on-board the NSF mobile laboratory,
and a paper entitled, "Computer Acquisition of Data from Paper
Tape Output of the TI960-A," are included with the "Operation
and Maintenance Manual" delivered to EPA with the mobile labora-
tory ( 47) .
Field Studies
To evaluate the performance of all on-board systems under actual
field conditions, the mobile laboratory was operated for one
week in Ann Arbor (away from NSF), two months in the metropolitan
Chicago area, and one week in Detroit, Michigan during April,
May, June, and early July, 1973. It returned to NSF in July
for installation and programming of the computer, then visited
four sites in Philadelphia, Pennsylvania during the month of
149
-------
f START A
CLEAI OUTPUT
DATA STORAGE
AREA
BtO TASK 40
(PUN )
WAIT A
BIT
UNSUSPENO
TASK 40
• 10 TASK 39
ISMPl)
PRINT TIME
MESSAGE
BID TASK 38
(FLOW)
BID TASK 34
IDAQ)
T)ID TASK 40\ NO
.EXIST ?
PRINT ERROR
MESSAGE
Figure 67. Flowchart of INIT,
150
-------
( START )
It
w
1
WRITE
'MODE '
+
READ TWO
CHARACTERS
KD
WRITE
'BAD ENTRY'
WRITE TWO *
SPACESIEAD FOUR
CHARACTERS
STARTING WITH
FIRST NUMBER
SEND DECIMAL
ASCII OF TEN
CORE LOCATIONS
PER LINE UNTIL
LAST LOCATION
IS% SECOND
NUMBER
SEND NUMBER
OF POINTS
SENT
Figure 68.. Flowchart of NSFC.
151
-------
V
•ID TASK 34
(OAQ )
1
C END )
SIT STOP FLAG
(WOtO 'M ' >
• IT 19
1
WilTI
'TIRMINATf
I
( »• )
V
•ID TASK 1
IPfBUO)
1
( •» )
Figure 68 (continued)
152
-------
[ START A
III
WRITE
HEADER
UNSUSPENDED
SET PARAMETER
COUNTER
GET LENGTH
4
^
NEXT
PARAMETER
A
GET TIME AND
VALUE, WRITE
RECORD
DECREMENT
LENGTH
NO
I
SUSPEND
SELF
Figure 69. Flowchart of PUN
153
-------
[ START J
SEND "BELL"
TO PRINTER
YES
CLEAR
STPFLG
WAIT A BIT
Figure 70. Flowchart of BELL
154
-------
( START )
SET DIGITAL
OUTPUT BIT 3 : 1
WASTE SOME
TIME
SET DIGITAL
OUTPUT BIT 3 = 0
WAIT 5 MINS.
12 TIMES
Figure 71. Flowchart of SMPL.
155
-------
START
SET DIGITAL
OUTPUT BIT 4 = 1
WAIT 5 MINS.
3 TIMES
SET DIGITAL
OUTPUT BIT 4:0
WAIT 5 MINS.
9 TIMES
Figure 72. Flowchart of FLOW.
156
-------
SET DO-3 OFF
SAMPLE, AVERAGE, AND STORE
AD15
SET DO-3 ON
SAMPLE, AVERAGE, AND STORE
AD 16
SET DO -? OFF
WAIT ONE MINUTE,CALL A
WAIT TWO MINUTES
SET DO -1 ON
WAIT FOUR MINUTES
WAIT 30 SECONDS
CALL SUBROUTINE A
GO TO START
Figure 73. Flowchart of DAQ.
157
-------
SAMPLE, AVERAGE,AND STORE
CHANNEL ADI THRU AD 11
•ID DATA PROCESSING TASK
RETURN
Figure 74. Flowchart of Subroutine A (DAQJt
158
-------
GET FLAGWORD, DATE/TIME , AND
ELEVEN READINGS FROM DAQ
SET POINTER
FOR NEXT
PARAMETER
GET DATE /TIME AND FlUORIDE
READINGS FROM DAO
STORE DATE/TIME
AND READING IN
OUTPUT STORAGE
SET POINTER
FOR NEXT
PARAMETER
Figure 75. Flowchart of DATA,
159
-------
GET DATE/TIME AND FIRST 1*
READINGS FROM DAQ
•UMPl 'GET DATA POINTER ' TO
• EGINNING OF CADMIUM WINDOW
LOOK
FOR MINIMUM
DONE
WITH 16
READINGS IN
UFFER ?
DONE
£ CADMIUM
WINDOW
GET 16 READINGS
FROM DAO
DONE
WITH LEAD
WINDOW
7
OfT 16 READINGS
FROM DAQ
DON!
WITH COPPER
WINDOW
STORE DATE /TIME,
CADMIUM, LEAD,AND
COPPER IN OUTPUT STORAGE
GET DATE/TIME. CVCl I NO.
AND READING FROM
DAQ
STORE DATE/TIME AND
READING IN OUTPUT STORAGE
Figure 75 (Continued)
160
-------
GET DATE /TIME AND 16
READINGS FROM DAQ
FINI
O
UNSUSPEND PUN
GET NEXT 16 READINGS
FROM DAQ
AVERAGE LAST 32
READINGS
READING :
"GROUND DROP" -
AVERAGE OF LAST
32
STORE DATE/TIME AND
READING IN OUTPUT STORAGE
Figure 75 (Continued)
161
-------
Z a. o.
3 < <
0.0,1-
o
< ^
O vn
0 O
O
iu O
>- U
o
z
u
O
fd
-P
"» O uu
< u «
O
<
Q
r-
0)
Cn
-H
5
162
-------
September. Acquisition of the computer was fully justified
during the Chicago assignments. One man-day for every data-day
is a conservative estimate of the time required to transpose
parametric data from analog strip chart output to numbers which
could be analyzed.
The first stop was the Ann Arbor municipal water treatment plant
where the principal objective was to gain proficiency with
getting the parametric systems on-line and calibrated, and
disassembling for the next move. The van was parked outside,
adjacent to the filter building. Although the visit was brief
and operators inexperienced, some meaningful data were collected,
and the troubleshooting experience was valuable.
The first field assignment away from Ann Arbor was in metropoli-
tan Chicago. N. J. Davoust, Engineer, Chicago Division of Water
Purification, and T. E. Larson, Head, Chemistry Section, Illinois
State Water Survey, enthusiastically supported the project from
its inception and served as members of the advisory committee.
The entire Chicago assignment was coordinated through N. J.
Davoust.
The trip from Ann Arbor to Chicago was made without incident.
The van went directly to the first monitoring site, where it
was parked inside a four year old fire station in the northwest
part of the city. (Engine Company 79 at 6424 N. Lehigh) , 10.5
miles from the Central Water Filtration Plant. All service lines
to the van were copper. To evaluate long-term performance char-
acteristics, especially sensor stability, the mobile laboratory
remained at this site for one month.
NOTE: Ve.ta4.l.e.d ne.pon.tt> o £ data c.ot£e.c.tid at a.LL JLoc.at4.onA
pu.bt4.&he.d Ae.pa/iate.-ty and pfLOvtdtd to the. age.ncy th.tiou.Qh
the. ^4.e.td tn.4.pt> w
-------
The third stop was in the City of Des Plaines, adjacent to
northwest Chicago and O'Hare International Airport, within five
miles of the Lehigh (first) location. The van was located at
the Public Works Building (1111 Campground Road), where distri-
buted water was normally a mixture of 75 percent purchased
Chicago water (from the Central Filtration Plant) and 25 percent
Des Plaines treated well water. The mix was altered to 1:1 to
demonstrate the ability of on-board systems in the mobile labora-
tory to detect subtle changes in the distribution system. Their
responsiveness was reliably demonstrated.
The fourth and final monitoring site on the Chicago trip was a
nine year old fire station (Engine Company 80 at Doty Avenue and
127th Street) in the Calumet Harbor area, 41 miles from the
previous location in Des Plaines, and 6.1 miles from the South
Water Filtration Plant. Water arrived at this location through
a one-mile section of six inch transmission line. The fire
station was situated on the extreme dead end of the transmission
line. The cast iron service line to the building was connected
to the mobile laboratory through a few feet of galvanized pipe.
Beginning at 1000 hours on June 29th, the main was flushed by
opening all connections in the fire station. At 1115 hours,
changes in measured levels of turbidity, free and total residual
chlorine, temperature, and pH were appreciable, as shown in
Figure 77. Relatively high levels of residual chlorine persisted
throughout the remaining period of observation.
Each of the last three sites in the Chicago area was visited for
just over a week. Moving between them was accomplished with a
minimum of downtime. Water and wastewater connections at a new
station were made by the laboratory operator; electric service
was provided by a city electrician. With normal operation, all
systems were on-line and calibrated in little more than two hours
after arriving at a new location.
J. V. Radziul, Chief of Research and Development, Philadelphia
Water Department and member of the project advisory committee,
coordinated activities and planning for the Philadelphia trip.
During the month of September, the mobile laboratory visited
four sites in the City of Philadelphia, carefully selected to pro-
file water quality from the various sources and treatment facil-
ities providing water to Philadelphia. The locations, identified
in Figure 78, include the Oak Lane Reservoir in the northern sec-
tion of the city (site no. 1), Philadelphia International Airport
in the southern section of the city (site no. 2), centrally lo-
cated City Hall (site no. 3), and Gulf Oil Refinery, less than
one mile from site no. 2, but with a different source of treated
water (site no. 4).
Raw water for Philadelphia is entirely of surface origin but
drawn from two sources, the Schuylkill River and the Delaware
River. Water from the Delaware is treated at the Torresdale
164
-------
2.2 -
2.0 -
1.8 -
1.6 -
1.4 -
JTU 1.2-
1.0 -
0.80-
0.60-
0.40-
0.20-
TURBIDITY
• • • •
TOTAL RESIDUAL CHLORINE
0.55-
0.45- • •
mg/1 0.40-
0. 35-
0.30- _
0.25-
63.0- 4
62.5- •
61.5- • ^ • • *
61.0- • • •
60.5-
7.85-
7.70- • • • *
o' 1 ol i3 I o' ' o I ol 1 o 1 ol 1 o 1 o
O tl^O (DOOOOOOO
OOCOJJCM •<* VO 00 O CM Tl-
O -H r-l (8 i-( r-\ rH r-t CM CSJ CM
It. 5 June 29— »
H C
9 • •
• « 9
TEMPERATURE
•
pH
1 o ol I ol ol
o o o o
04 ^s* ^^ oo
o o o o
June 30 — >
Figure 77. Data from main flushing at Chicago, Calumet Harbor
165
-------
TORRES DALC WflTt
OUrCIM LOME WAT IT R.
MIXED WITH roaaesoflLE
w
WATER SUPPLY
MA(W TRANSMISSION LINES
CITY OF PHILADELPHIA
Figure 78. Relative locations of Philadelphia monitoring
sites.
166
-------
Filtration Plant, third largest in the world (282 mgd), and de-
livered to most of the eastern half of the city. Schuylkill
water is treated at Queen Lane (120 mgd) or Belmont (78 mgd)
Filtration Plants, and distributed principally to western sec-
tions of the city. Each of the treatment facilities is highly
automated. Distribution is controlled by a "load control"
facility, a central point to which all data related to demand
are transmitted by microwave. All water is treated by prechlor-
ination, presedimentation, alum-lime coagulation, rapid and slow
mixing, sedimentation, rapid sand filtration, chlorination, and
fluoridation. In addition, Belmont water is ammoniated to im-
prove its taste and odor characteristics. Finished water is
stored in three large reservoirs, Roxborough, Oak Lane (site
no. 1), and East Park, and elevated tanks at various locations
in the city.
The water at site no. 1 was treated at Torresdale and stored in
the reservoir. Test water at the monitor was taken from the
reservoir or from the main filling the reservoir. At the air-
port, site no. 2, water was from the Belmont filtration plant
exclusively. Torresdale, or Queen Lane, or a mixture of the
two waters was monitored at City Hall. At the Gulf Oil Refinery,
water was delivered principally from Queen Lane, with some mixing
from Torresdale possible.
During the first week in Philadelphia, little reliable data was
collected. A record heat wave created virtually inoperable
conditions for the operator and many of the sensor systems.
Additional problems resulted from inadequate external grounding
of the laboratory.
A malfunction in the computer power supply caused repeated but
temporary "failures" throughout the Philadelphia assignment.
Numerous calls by TI service personnel resulted in little satis-
faction to the NSF staff. In fact, a general comment regarding
failure of TI representatives to adequately diagnose and correct
frequent problems with the on-board computer system is in order.
A typical notation from the operator's log for site no. 3 cites
a malfunction in one of the A/D multiplexer cards which was not
replaced for three days after the TI field engineer was notified
of the problem. In virtually every period of computer induced
downtime, the NSF operator or an NSF computer specialist - not
the TI field engineer - was responsible for diagnosing the
failure. Although TI hardware offers many advantages for mobile
laboratory application, it is unlikely that this computer -system
would be selected for any future NSF project applications.
167
-------
SECTION VI
REFERENCES
1. Environmental Protection Agency. Interim Primary Drinking
Water Standards. Federal Register. £0(51), March 14, 1975.
2. Fair, G. M., J. C. Geyer, and D. A. Okun. Water and Waste-
water Engineering. John Wiley & Sons, Inc., New York, 1968.
p. 17-31.
3. Standard Methods for the Examination of Water & Wastewater.
13th Edition. American Public Health Association, American
Water Works Association, and Water Pollution Control Feder-
ation, New York, 1971.
4. McClelland, N. I. Water Quality Monitoring in Distribution
Systems. National Sanitation Foundation, Ann Arbor, Michigan,
May 1971. p. 169.
5. Turbidimeters. Second Revised Edition. Hach Chemical Com-
pany, Ames, 1973.
6. Quality Goals for Potable Water. Journal American Water
Works Association, September 1973. p. 62.
7. Water Quality Monitoring in Metropolitan Chicago: Project
Report. National Sanitation Foundation, Ann Arbor, Michigan,
1973.
8. Water Quality Monitoring in Detroit: Data Presentation.
National Sanitation Foundation, Ann Arbor, Michigan, 1973.
9. Water Quality Monitoring in Metropolitan Philadelphia: Proj-
ect Report. National Sanitation Foundation, Ann Arbor,
Michigan, 1974.
10. Langelier, M. A. The Analytical Control of Anti-Corrosion
Water Treatment. Journal American Water Works Association,
28:1500, 1936.
11. Larson, T. E. and A. M. Buswell. Calcium Carbonate Saturation
Index and Alkalinity Interpretations. Journal American Water
Works Association, 34:1664, 1942.
168
-------
12. Ryzner, J. W. A New Index for Determining the Amount of
Calcium Carbonate Scale Formed by a Water. Journal American
Water Works Association, 36:472, 1944.
13. Pytkowicz, R. M. Rates of Inorganic Calcium Carbonate Nu-
cleation. Journal Geology, 73:196, 1965.
14. Weyl, P. K. Solution Kinetics of Calcite. Journal Geology,
66:163, 1958.
15. Weyl, P. K. The Carbonate Saturometer. Journal Geology,
69:32, 1961.
16. McClanahan, M. A. Mechanism of Cast Iron Corrosion Inhibi-
tion by Calcium Carbonate Deposition in Water Distribution
Systems. Doctoral dissertation, The University of Michigan,
Ann Arbor, 1968.
17. Schlicting, H. Boundary Layer Theory. McGraw-Hill Book Co.,
1960. p. 83.
18. Levich, V. G. Physicochemical Hydrodynamics. Prentice Hall,
Inc., 1962.
19. Nernst, W. Z. Physical Chemistry, 47:52, 1904.
20. Riddiford, A. C. Advances in Electrochemistry and Electro-
chemical Engineering. 4:47, 1966.
21. Napp, D. T. Doctoral dissertation, University of Minne-
sota, Minneapolis, 1967.
22. Johnson, D. C. Doctoral dissertation, University of Minne-
sota, Minneapolis, 1967.
23. Enslow, L. H. The Continuous Stability Indicator. Water and
Sewerage Works, 107, March 1939.
24. Enslow, L. H. The Continuous Stability Indicator and the
Langelier Index. Water and Sewerage Works, 283, July 1939.
25. Durfor, C. N. and E. Becker. Public Water Supplies of the
100 Largest Cities in the United States. 1962. U.S. Geo-
logical Survey Water Supply Paper 1812. U.S. Government
Printing Office, Washington, D.C., 1964.
26. Zipkin, I. and F. J. McClure. Fluoride Drinking Waters.
U.S. Department of Education and Welfare Public Health
Service Publication, 825:483, 1962.
27. Hein, J. W., D. E. Gardner, and G. B. Haydon. Preliminary
Investigations of the Effect of Sodium Monofluorophosphate
169
-------
on Salivary Acid Production and Hydroxyapatite Solubility.
Journal Dentist Research, 30:466, 1951.
28. Cooley, W. E. Reaction of Tin (II) and Fluoride Ions with
Etched Enamel. Journal Dentist Research, 40:1199, 1961.
29. Brudevold, F., et al. Uptake of Tin and Fluoride by Intact
Enamel. Journal American Dentist Association, 53:159, 1956.
30. Cooley, W. E. Applied Research in the Development of Anti-
caries Dentifrices. Journal Chemical Education, 47:177, 1970.
31. Frant, M. S. and J. W. Ross. Electrode for Sensing Fluoride
Ion Activity in Solutions. Science, 154:1553, 1966.
32. Orion Instruction Manual, Fluoride Ion Activity Electrode,
Model 94-09. Second Edition. Orion Research Inc., Cambridge,
Massachusetts, 1967.
33. Bock, R. and F. Z. Strecker. Direkte Electrometrische
Bestimmung des Fluorid-ions. Analytical Chemistry (Germany),
234:322, 1968.
34. Harwood, J. E. The Use of an Ion Selective Electrode for
Routine Fluoride Analysis on Water Samples. Journal Water
Research, 3:273, 1969.
35. Mesmer, R. E. Lanthanum Fluoride Electrode Response in
Aqueous Chloride Media. Analytical Chemistry, 40:443, 1968.
36. Rechnitz, G. A. Ion Selective Electrodes. Chemistry and
Engineering News, 45 (25):146, 1967.
37. Scrinivasan, K. and G. A. Rechnitz. Activity Measurement
with a Fluoride-Selective Membrane Electrode. Analytical
Chemistry, 40:509, 1968.
38. Ciavatta, L. Hydrolysis of Fluoride Ion, F~, in 3 M Na
(CIO.*)" and 3 M K+(C1~, F~) Media. Arkiv Kemi (Sweden),
21:129, 1963.
39. Butler, J. N. Ionic Equilibrium, A Mathematical Approach.
Addison-Wesley Publishing Co. Inc., 1964. Chapter 5-1,
p. 115.
40. Frant, M. S. and J. W. Ross. Use of a Total Ionic Strength
Adjustment Buffer for Electrode Determination of Fluoride in
Water Supplies. Analytical Chemistry, 40:1169, 1968.
41. Crosby, N. T., A. L. Dennis, and J. G. Stevens. An Evaluation
of Some Methods for the Determination of Fluoride in Potable
Waters and Other Aqueous Solution. Analyst, 93:643, 1968.
170
-------
42. Kelada, N. P. Electrochemical Characterization of Free &
Complexed Fluorides in Drinking Water and Effects of Aluminum
& Iron on Fluoride Incorporation into Tooth Enamel. Doctoral
dissertation, University of Michigan, Ann Arbor, 1972.
43. Schwarzenback, G., R. Gut, and G. Anderegg, 117. Komplexone
XXV. Die Polarographische untersuchung von austausch-
gleichgewichten. Neue daten der piltungskonstanten von metall
komplexen der Athylendiamin-tetraessigsaure und der 1,2-
Diamin-cyclohexan-tetraessigsaure. Helv. Chim. Acta (Germany),
37:937, 1954.
44. Lacroix, S. Etude de quelques Complexes et Composes peu
Solubles des Ions. Ann. Chim. (France), 4:5, 1949.
45. Bertin-Batsch, C. Etude par des Methodes variees de quelques
Complexes Organiques de 1'Ion Ferrique. Ann. Chim (France),
7:481, 1952.
46. Peterson, T. L., D. 0. Brant, and K. H. Mancy. Characteriza-
tion of Trace Metals in Municipal Water Supplies by Thin
Layer Anodic Stripping Voltammetry. University of Michigan,
Ann Arbor. (Presented at 160th National Meeting of the
American Chemical Society. Chicago, September 1970.)
47. Operation and Maintenance Manual for Mobile Water Quality
Monitoring Laboratory, National Sanitation Foundation, Ann
Arbor, Michigan, 1973.
171
-------
SECTION VII
PROJECT PUBLICATIONS
1* Committee Report, AWWA Corrosion and Stability Committee.
"Water Quality Determination in Distribution Systems." (Pre-
sented at the 90th Annual Conference, American Water Works
Association, Washington, D.C., 1970. Journal American Water
Works Association, 795, 1972.
2. Kelada, N. P., N. I. McClelland, and K. H. Mancy. Character-
ization of Fluorides in Water Supplies by Selective Ion
Electrodes. (Presented at 160th National Meeting, American
Chemical Society, Chicago, Illinois, 1970.)
3. McClelland, N. I. and K. H. Mancy. Electrochemical Character-
ization of Drinking Water. (Presented at the 141st Meeting,
Electrochemical Society, Houston, Texas, 1972.)
4. McClelland, N. I. and K. H. Mancy. Water Quality Monitoring
in Distribution Systems. (Presented at the 92nd Annual Con-
ference, American Water Works Association, Chicago, Illinois,
1972.)
5. Schimpff, W. K., N. I. McClelland, K. H. Mancy, and H. E.
Allen. An Automated Differential Anodic Stripping Technique
for Monitoring Trace Metals in Distribution Systems. (Pre-
sented at the 92nd Annual Conference, American Water Works
Association, Chicago, Illinois, 1972.)
6. McClelland, N. I. Monitoring of Municipal Drinking Waters:
Process to Consumer. (Presented at the 15th Eastern Analyti-
cal Symposium, New York, New York, 1973.)
7. McClelland, N. I. Monitoring of Drinking Water Quality in
Distribution Systems. (Presented at the 2nd Joint Conference
on Sensing of Environmental Pollutants, Washington, D.C.,
1973.)
8. McClelland, N. I., J. R. Adams, R. R. Wood, and K. H. Mancy.
Application of Monitoring Technology (for Assuring) Drinking
Water Quality. (Presented at the 167th National Meeting,
American Chemical Society, Los Angeles, California, 1974.)
9. Adams, J. R., R. R. Wood, and N. I. McClelland. Drinking
Water Quality: Application of Monitoring Technology. (Pre-
sented at the 6th Central Regional Meeting, American Chemical
Society, Detroit, Michigan, 1974.)
10. McClelland, N. I. and K. H. Mancy. Water Quality Surveillance
in Distribution Systems. (Presented at 94th Annual Conference,
172
-------
American Water Works Association, Boston, Massachusetts, 1974.)
11. Symons, J. M., M. C. Gardels, K. H. Mancy, and N. I. McClelland,
Continuous Distribution System Monitoring to Study the Effects
of Hardness and Other Water Quality Parameters. (Presented
at 94th Annual Conference, American Water Works Association,
Boston, Massachusetts, 1974.)
12. McClelland, N. I. Experiences in Monitoring Residual Chlorine.
(Presented at 94th Annual Conference, American Water Works
Association, Boston, Massachusetts, 1974.)
173
-------
SECTION VIII
APPENDICES
Page
A. Technical Advisory Committee 175
B. Preparation of Reagents 176
C. Special Alkalinity Case Study 181
174
-------
APPENDIX A
TECHNICAL ADVISORY COMMITTEE
Harry A. Faber, Chairman (1969-1972)
88 Main Street
Whitesboro, NY 13492
N. J. Davoust, Chairman (1973)
Engineer of Water Purification
Central Water Filtration Plant
1000 E. Ohio Street
Chicago, IL 60611
T. E. Larson
Head, Chemistry Section
Illinois State Water Survey
Box 232
Urbana, IL 61801
Blucher A. Poole
Consultant
5839 Brockton Drive
Indianapolis, IN 46220
Joseph V. Radziul
Chief, Research & Development
Water Department
1270 Municipal Services Bldg.
Philadelphia, PA 19107
Verdun Randolph
Associate Director
Consumer Health Protection
Illinois Department of Public Health
535 West Jefferson Street
Springfield, IL 62761
Gordon G. Robeck
Director
Water Supply Research Division
U.S. Environmental Protection
Agency
Municipal Environmental
Research Laboratory
Cincinnati, OH 45268
Elroy F. Spitzer
Director
American Water Works Assoc.
Research Foundation
6666 W. Quincy Avenue
Denver, CO 80235
James M. Symons
Chief, Physical and Chemical
Removal Branch
Water Supply Research Division
U.S. Environmental Protection
Agency
Municipal Environmental
Research Laboratory
Cincinnati, OH 45268
T. L. VanderVelde
Deputy Chief
Bureau of Environmental and
Occupational Health
Michigan Department of Public
Health
Lansing, MI 48914
Richard Woodward
Vice President
Camp, Dresser, & McKee
One Center Plaza
Boston, MA 02108
175
-------
APPENDIX B
PREPARATION OF REAGENTS
SRM-HOUSED SYSTEMS
Standard solutions are required for weekly calibration of SRM-
housed sensor systems. No other reagents are used for routine
operation of these systems. Standard solutions are prepared as
follows:
HARDNESS
b
Calcium chloride stock solution
Dissolve 11.099 g CaCl2 (anhydrous) in approximately
one liter of distilled water. This solution should
be standardized against EDTA and adjusted to contain
lO^'M CaCl2 concentration.
Standard solutions
To prepare standard solution, dilute 13.0 ml stock
solution to 1,000 ml with distilled water, using a
volummetric flask. This solution contains 10 (~ ^M
Ca2+ activity. Dilute 1.075 ml stock solution to
1,000 ml. This solution contains 10 (-^M Ca2+ activity.
Separate each of the standards into two 500 ml wide -
mouth jars. Label one jar of each standard "wash,"
and the other, "standard."
NITRATE
a. Potassium nitrate stock solution
Dissolve 1.6308 g KN03 in exactly 1.0 liter of dis-
tilled water.
b. Standard solutions
To prepare standard solutions, dilute 104.2 ml stock
solution to 1,000 ml with distilled water, using a
volummetric flask. This solution contains 100 mg/1
NO3~ activity. Dilute 10.2 ml stock solution to
1,000 ml. This solution contains 10 mg/1 N0s~ activity.
176
-------
Dilute 1.0 ml stock solution to 1,000 ml. This
solution contains 1.0 mg/1 NO3~ activity. Separate
each into "wash" and "standard" solutions.
CHLORIDE
a. Sodium chloride stock solution
Dissolve 1.6482 g NaCl (dried at 140°C) in chloride
free water and dilute to 1.0 liter. This solution
contains 1.0 mg Cl~/ml.
b. Standard solutions
To prepare standard solutions, dilute 100 ml stock
solution to 1,000 ml with chloride free water, using
a volummetric flask. This solution contains 100 mg/1
Cl . Dilute 10 ml stock solution to 1,000 ml. This
solution contains 10 mg/1 Cl~. Separate each into
"wash" and "standard" solutions.
CONDUCTIVITY
a. Standard reference potassium chloride solution
Dissolve 745.6 mg anhydrous KCl in freshly boiled
distilled-deionized water and dilute to 1,000 ml
at 25°C. This solution has a specific conductance
of 1,413 umhos/cm at 25°C. This solution is used
to calibrate the laboratory conductivity meter, used
to prepare standards of 1,000 and 100 ymhos/cm.
b. Standard solutions
Add KCl solution to distilled water at 25°C to pre-
pare standard solutions of 1,000 and 100 ymhos/cm,
respectively. Separate each into "wash" and "standard"
solutions.
pH
a. pH 6.86 standard buffer
(0.025 M potassium dihydrogen phosphate +0.025 M
disodium hydrogen phosphate)
Dissolve 1.179 g KHaPO^ and 4.30 g NazHPO^ in dis-
tilled water and dilute to 1,000 ml. This solution
is pH 6.86 at 25°C.
b. pH 9.18 standard buffer
(0.01 M borax)
177
-------
Dissolve 3.80 g Na2B^O?'10 H20 in distilled water
and dilute to 1,000 ml. This solution is pH 9.18
at 25°C.
RESIDUAL CHLORINE
a. Buffer solution
(Sodium acetate-acetic acid buffer)
Fill the buffer dispensing bottle approximately one-
fourth full with distilled water. Add 855 ml glacial
HN03. Add 460 g CH3COONa«3 H2O with vigorous mixing
and stir till completely dissolved. Fill bottle to
top with distilled water. This solution is used with
free- and total- residual chlorine analyzers.
b. Potassium iodide solution
Dissolve 20 g KI in distilled water and dilute to
approximately two liters in the KI dispensing bottle
(total chlorine analyzer only).
ALKALINITY
a. Buffer solution
(Potassium hydrogen phthalate buffer)
Dissolve 91.80 g potassium hydrogen phthalate
(HOCOCelUCOOK) in approximately eight liters of dis-
tilled water. Dilute 13.4 ml concentrated hydro-
chloric acid (HC1) to 800 ml with distilled water.
Mix the two solutions, add 0.0635 g sodium thiosulfata
(Na2S203), and bring up to nine liters. Adjust the
pH to 3.1 using concentrated HC1.
b. Baseline solution
(1) Sodium carbonate stock solution
Dissolve 0.477 g Na2C03 in C02 free distilled
water and dilute to 1.0 liter.
(2) To prepare working baseline solution, dilute
200 ml stock solution to 9.0 liters with C02
free distilled water. This solution is equi-
valent to 10 mg/1 CaC03.
c. Standard solutions
(1) Sodium carbonate stock solution
Dissolve 1.0590 g Na2C03 in C02 free distilled
water and dilute to 1.0 liter. Store in tightly
stoppered glass bottle. Make up fresh stock
solution monthly.
178
-------
(2) To prepare standard solutions, dilute 1.0, 2.0,
3.0, ... 9.0 ml stock solution to 100 ml each.
These solutions contain 10, 20, 30, ... 90 mg/1
equivalent CaCO3, respectively. Fresh standard
solutions must be prepared for each calibration
series.
FLUORIDES
a. Buffer solutions
(1) Free fluoride
Dissolve 850 g sodium nitrate (NaN03) in approxi-
mately nine liters distilled water. Add 57 ml
glacial acetic acid (CH3COOH) and mix. Bring
volume up to 10.0 liters. Adjust to pH 5.2 with
concentrated sodium hydroxide (NaOH) solution.
(2) Total fluoride
Dissolve 595 g NaN03 and 588 g sodium citrate
(Na3C6H507*2 H20) in approximately nine liters
distilled water. Add 115 ml CH3COOH and mix.
Bring volume up to 10.0 liters. Adjust to
pH 5.2 with NaOH.
b. Baseline solution
Dilute 20.0 ml 100 mg/1 F~ Harleco standard to
20.0 liters.
c. Standard solutions
Dilute 2.0, 5.0, and 10.0 ml 100 mg/1 F~ Harleco
standard to 1,000 ml each. These solutions contain
0.2, 0.5, and 1.0 mg/1 F~~, respectively. Store in
plastic bottles.
TRACE METALS ANALYZER (DASV for Cd, Pb, and Cu)
a. pH adjustment solution
The strength of the pH adjustment solution will vary
with the pH of the test water. Add concentrated
nitric acid (HNOs) d>iopu)-Lt> e. to distilled water to the
predetermined level at which, when added to test
water in the appropriate ratio, will result in a
final pH of 2.5.
b. Standard solution
(1) Place 1.0 ml 1,000 mg/1 Cd Harleco standard,
1.0 ml 1,000 mg/1 Pb Harleco standard, and
1.0 ml 1,000 mg/1 Cu Harleco standard in a
100 ml volummetric flask. Bring to volume
179
-------
with distilled water and mix. This solution
contains 10 mg/1 each of Cd, Pb, and Cu.
Transfer to plastic bottle for storage.
(2) Dilute 1.0 ml of stock solution to 2.0 liters
in the quartz reservoir, using test water as
the diluent. (The known addition technique
is used for standardization in trace analysis.)
This solution contains 0.05 mg each of Cd, Pb,
and Cu.
c. Mercury plating solution
Dissolve one drop (approximately 0.05 ml) of reagent
grade mercury (Hg) in concentrated HNOs (as little
as possible). When solution is complete, dilute to
approximately 100 ml with distilled water. Store
in a glass stoppered bottle. Inject 2.0 ml of this
solution into the flow cell filled with distilled
water to plate (rejuvenate) the electrodes.
180
-------
APPENDIX C
SPECIAL ALKALINITY CASE STUDY
Experimental Procedure
1. Prepare 0.050 F KHP and 0.019 F HCl
standard phthalate buffer.
2. Prepare alkalinity standards from
Na2CO3 in distilled water.
3. Prepare additional alkalinity standards
using distilled with known concentra-
tions of chlorine, prepared from gaseous
chlorine.
4. Mix five parts of each alkalinity
standard with one part buffer and
determine resultant pH.
5. Plot H versus alkalinity, developing
a series of curves from the standards
with varying concentrations of chlorine.
Theoretical Results
In the presence of free residual chlorine, initial formalities
for the buffer are:
KHP = 0.050
HCl = 0.019
and for the test water:
Na2C03 = 6/5 C
C12 = 6/5 E
Final formalities at buffer:sample = 1:5 are:
HHP = 8.33(10)~3
HCl = 3.17(10)~3
Na2C03 = C
C12 = E
Known relationships are the same as those expressed in Equations
75 through 79 and 81 through 84. Equation 80 is rewritten as:
[Cl~] = 3.17(10)~3 + [HOC1] + [OC1~] (126)
181
-------
and Equation 85:
[H+] 4- [Na+] 4- [K+] = [OH~] 4- [Cl"] 4
4- [HC03~] 4- 2[C032~] 4- [OC1~] (127)
[Na+] + [K+] = [OH~] + [Cl ] 4- [HP ] + 2[P2~]
Other known reactions include:
K5 = [HQC1] [H+] [el"] = 4.2(1(J)-* (128)
[C12]
K6 = ] = 3.4(10)8 (129)
[HOC1]
= [H ] [OC1 ~8
and
2[C12] 4- [HOC1] 4- [OCl"] 4- [Cl"]= 2E 4- 3.17(10)~3 (130)
Assuming [C032~] = [P2~] = [OH~] = [C12] = [OCl"] = 0, and that
only the equilibria expressed in Equations 131 and 132 occur for
free residual chlorine:
C12 4- H20 -> HOC1 4- H+ 4- Cl" (131)
HOC1 4- H20 -* H30+ 4- OCl" (132)
then derivations shown in Equations 86 through 98 and those in
Equations 133 and 134 can be expressed:
[Cl"] = 3.17(10)~3 4- E (133)
[HOC1] = E (134)
Also, from Equations 80 and 86:
FH+1 ( \
[C12] = -kg-J- ^3.17(10) 3 + [HOC1] > [HOC1] (135)
From Equations 79, 81, 82, 85, 95, and 98:
[H+] 4- 2C 4- 8.33(10)~3 = 3.17(10)~3 4- E 4- -
[H'] 4-1. 3 (10)
4.6(10)~3C
•«• l" •» . A ^-/^*S\ 3
(
[HT]+4.6(10)
L.C.D. = [H+]2 + 5.9(10)~3[H+] + 5.98(10)~6 (137)
182
-------
1.083 (10) ~5 4.6 (10) 3C
— ij ~ . — .
IH+]3 + 2CIH+]2 + 5.16(10)^ [H+]2 - E[H+]2 + 5.9(10)~3 [H+]2
+1.18(10)~2C IH+] + 3.044 (10)~5 IH+] - 5.9 (10)"3 E[H+]
+5.98(10)"6 IH+] + 1.196(10)~5C + 3.086(10)~8 - 5.98(10)~6E
-1.083(10)~5 IH+] - 4.982(10)~8 - 4 . 6 (10) ~3 C IH+]
-5.98 (lO)""6 C = 0 (139)
2 + 2C - E | [H+]2 + J2.559(10)~5 + 7.2(10)~3C
5.98(10)~6C - 5.98(10)~6E} = 0
assuming [HT]3 = 0.
-2.559(10)"5
+3 = (140)
+ 5.9(10)~3E /J2.559(10)~5+7.2(10)~3C
- 5.9(10)~3E(2
+ / - 4 { 1.106(10)"2+ 2C - E)
{-1.896 (10)~8 + 5.98(10)~6C
-5.98(10)~6E
2
2.212(10) + 4C - 2E
(141)
-2.559(10) 5 - 7.2(10) 3C
+ 5.9(10)"3E / 4(10)"6C2+2.556(10)"7C
+ / +1.489(10)~9 +1.089(10)~5E2
, \ -1.132(10)"7E -1.32(10)"5CE
j"*"! = I
2.212(10)~2 + 4C - 2E (142)
183
-------
The family of curves obtained by application of Equation 142 are
shown in Table 23 and Figure 79.
Although alkalinity has been shown to be linear with respect to
pH when no free chlorine is present, it is important to note
that the alkalinity monitor reads potential in mv, not pH. To
convert [H ] in the theoretical results to potential, the Nernst
equation is applied:
E = E° + 0.059 log [H+] (143)
or
E = E° - (0.059)pH (144)
where:
E° = a function of reference electrode potential
and E is the measured potential
E° cannot be predicted theoretically. If it is assumed to be
equal to zero, then:
E = -(0.059)pH (145)
Data from Table 23 are expressed as potentials in Table 24 and
Figure 80. As shown in Figure 80, experimental and theoretical
results for alkalinity solutions with no free chlorine residual
vary only about three to seven mv at alkalinity equal to 30 to
70 mg/1 as CaC03. This slight variation is likely attributable
to experimental error and instrument limitations.
In Table 25, data from Table 23 for C12 = 1.0 and 2.0(10)~5F
(approximately 0.7 and 1.4 mg/1 free residual chlorine) are ex-
pressed as potentials and compared with theoretical results for
C12 = 0. The greatest variation in potential associated with a
specific level of alkalinity and a change from C12 = 0 to
C12 = 1.4 mg/1 is 0.3 mv. This small change in potential cannot
be observed with the alkalinity monitor.
Experimental Results
Experimental data obtained by NSF closely approximated the theore-
tical results. NaOCl was substituted for gaseous C12 in the pro-
cedure described on page 181. Alkalinity standards included 30,
40, 50, 60, 70, and 80 mg/1 as CaCO3, prepared from Na2C03•
Changes in volume with addition of NaOCl to the standards were
negligible and disregarded. (The greatest addition was 0.1 ml
to 60 ml of standard, or 0.167 percent change in volume.) The
result of simple regression applied to the data is shown in
Figure 81.
184
-------
o
o
o
00
o
CN
•P
•H
o
4-1
CM
H
U
O
O I
nj
O
m
0)
LO
•
in
o
•
m
m
m
ro
ro
OT
185
-------
o
oo
o
in
o
O
n
O
U
(d
rH
0)
u
•H
4->
OJ
!-l
O
(D
X!
rd tn
td -H
4-1 4-)
G C
CU (D
6 4J
-H O
!H O<
0)
m
o
00
O
O
CM
O
cn
(AUI)
186
-------
co
0s-
a E
• r-,.
s s
o-. o
c
o
•H
C -P
o 3
•H rH
-P O
3 Cfi
O
W
C
•H
m
a
ro
01
0)
rH
X!
fO
EH
in in
*) o
U; ii ; ii;
e 8 £
•H O
O
OJ OJ
rH fd
U !3
o o
OOC=OOO
O O O
e e s
CO 0-- O
fd fc
e e
M S-l
O O
pr i Pr i
187
-------
o
•
o
CO U
CN
I
00
I
C
O
-H
-P
o
IT)
o
ro
-------
Table 24. EXPERIMENTAL AND THEORETICAL RELATIONSHIPS
BETWEEN ALKALINITY AND POTENTIALS
Alkalinity
(mg/1 as CaC03)
24
36
48
60
72
84
Potentials (mv)
Experimental
-194.9
-197.5
-200.1
-202.6
-205.3
-207.6
Theoretical
-193.1
-194.4
-195.7
-197.1
-198.4
-199.8
Table 25. THEORETICAL RELATIONSHIP BETWEEN ALKALINITY AND
POTENTIAL FOR SOLUTIONS WITH AND WITHOUT FREE
RESIDUAL C12
Alkalinity
(mg/1 as CaC03)
24
36
48
50
72
84
Potential (mv)
No C12
-193.1
-194.4
-195.7
-197.1
-198.4
-199.8
0.7 mg/1
-193.0
-194.3
-195.6
-196.9
-198.3
-199.7
1.4 mg/1
-192.9
-194.2
-195.5
-196.8
-198.2
-199.5
189
-------
Assuming that the NaOCl was made stoichiometrically according to
the reaction:
2NaOH
C1
2Na
Cl
OC1
H2O
(146)
i.e., there was no excess NaOH in the NaOCl, the net effect of
substituting NaOCl for gaseous C12 is a change in sign of the
quantity representing C12 in Equation 146; i.e., a cation (Na )
is added in lieu of an anion (Cl ). The predicted slope is
approximately -0.18 mv per mg/1 free C12. This value is slightly
larger than the experimental value obtained in the NSF laboratory
(.07/.18), but it is 28 times smaller than the value obtained
by EPA. (See Table 26.)
Table 26. CHANGES IN POTENTIAL FROM ADDITION
OF 1.0 mg/1 FREE RESIDUAL C12
Slope =(as mv/mg/1)
Experimental
NSF
-0.07
EPA
-5
Theoretical
-0.18
Despite the theoretical relationships which were established and
closely related data acquired experimentally by NSF, anomalous
results continued to occur with Cincinnati tap water monitoring.
Although it does not seem likely that the observed effect can be
attributed only to the presence of free chlorine residual,
chlorine is very likely involved, perhaps synergistically with
a complex organic material.
Additional experiments were undertaken to determine the effect
of adding a reducing agent to the alkalinity system. Thiosulfate
(five mg/1) added to the buffer solution did not interfere with
potentiometric measurement of alkalinity and it eliminated the
interference effect. (Data are shown in Table 27 and Figure 82.)
As a result, it is now recommended as standard procedure that
S20a2 be added to pH 3.1 phthalate buffer solution used in all
applications of the potentiometric alkalinity monitoring system.
190
-------
M
Q)
-P
ttf O
-P (0
U
-P
•H
rd
DO
0)
-P
fll
H
-P
•H
O
in
O
O
(0
u
w
to
tn
g
-P
•H
C
-H
rH
o
n
O
rH
(N
IT)
O
Cvl
O
O
CM
en
(U
-H
M
(D
O
-H
-P
O
>i
4J
•H
C
-H
rH
fO
y,
rH
(0
W
3
en
H
(D
•rH
-P
c
0)
-P
o
CN
oo
CD
M
(AUI)
191
-------
LLJ
<
CO
O
LL
O
O
LU
LL
LL
LU
CM
"*>
j=
CO
'•*-«
c
£
O
Q.
CN
CN
co
O
CM
to
-f-
«—
CM
O
CD
Q.
E
CO
to
CM
O
+
CD
Q.
E
CO
to
CN
CO
O
CM
to
+
CD
Q.
E
&
CD
Q.
E
c3
"co
>8
lo
CO VI
S/ CO
—
~^
O>
_§
in
| CO 1 | 1
8
O
1 g 1 1 1
CN
CD O CN
CM O3 | | 00
r- 0 O
CM CM CN
CM O CM O If)
CN 00 CM 00 r^
«- o o en o
CM CN CM ^ CM
O O O O % O5
CN
JN
O
o>
E
CM
CO
O
CM
CO
to
c
O
'5
Tl
-a
CD
i_
CO
O.
CD
C
O
O
192
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO. 2.
EPA-600/2-77-074
..TITLE AND SUBTITLE
WATER QUALITY MONITORING IN
DISTRIBUTION SYSTEMS
/. AUTHOR(S)
Nina I. McClelland
K . H . Mancy
P. PERFORMING ORGANIZATION NAME AND ADDRESS
National Sanitation Foundation
Ann Arbor, Michigan 48106
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory — Cin. ,OH
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
March 1977 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
ICB047;ROAP 21AQB ;TASK011
11. CONTRACT/GRANT NO.
68-03-0043
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A mobile laboratory with 18 integrated, computer controlled parametric
systems for monitoring potable water quality in distribution systems
was developed and field evaluated at ten locations in four United
States cities: Chicago, Illinois; Ann Arbor and Detroit, Michigan;
and Philadelphia, Pennsylvania. Temperature, conductivity, pH,
chloride, dissolved oxygen, free and total residual chlorine, turbidity
corrosion rate, free and total fluorides, alkalinity, hardness,
nitrate, copper, cadmium, lead, and calcium carbonate deposition rate
are measured using commercially available and newly developed sensor
systems.
7. KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Water quality
Chemical analysis
Monitors
Data acquisition
Distribution systems
Potable Water
8. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
b. IDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASSIFIED
c. COSATI Field/Group
7 B
13 B
21. NO. OF PAGES
207
22. PRICE
PA'Form 2220-1 (9-73)
193
*US GOVERNMENT PRINTING OFFICE 1977-757-056/5576
-------
HAY -. 1 1977
. ,,'! V
0) in ~-~ •-» .^
O- .
8 I 8- A
?•
o
o
CD 0
o c
o 3.
?r -y
\
O
0)
2.
(0
o ^
CD -J3
Ooo
o O
II-
^ ?L
w
-+>
o q
? B
o' &
sf
%
Q
m
Z
n
X 2
£ 2
2 3
-( f>
o j-
* •£
lo
D m
s q
Is
•D Z
1 >
3 o
2. m
z
o
c
(fl
m
Z
<
33 -0
o o
z
2
m
Q
m -( m
^ > >
> r z
co TJ D
S ^ -n
01 O m
H m
m (/)
O
TJ
>
------- |