Subscriber access provided by Karolinska Institutet, University Library
Characterization of Natural and Affected Environments
Water Distribution System Failure Risks with Increasing Temperatures Emily N. Bondank, Mikhail V. Chester, and Benjamin L. Ruddell Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b01591 • Publication Date (Web): 09 Aug 2018 Downloaded from http://pubs.acs.org on August 13, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 27
Environmental Science & Technology
1
WATER DISTRIBUTION SYSTEM FAILURE RISKS WITH
2
INCREASING TEMPERATURES
3
Emily N. Bondank1*,
[email protected], 480-285-8327
4
Mikhail V. Chester1
5
Benjamin L. Ruddell2
6
1
7
Avenue, Tempe, AZ 85281
8
2
9
Knoles Drive, PO Box 5693, Flagstaff, AZ 86011
Civil, Environmental, and Sustainable Engineering, Arizona State University, 660 S College
School of Informatics, Computing, and Cyber Systems, Northern Arizona University, 1295 S.
10 11
* Author to whom correspondence should be addressed
12
ABSTRACT
13
In the coming decades, ambient temperature
14
increase from climate change threatens to reduce not
15
only the availability of water, but also the operational
16
reliability of engineered water systems. Relatively little
17
is known about how temperature stress can increasingly
18
cause hardware components to fail, quality to be
19
affected, and service outages to occur. Changes to the
20
estimated-time- to-failure of major water system hardware and the probability of quality non-
21
compliance were estimated for a modern potable water system that experiences hot summer
22
temperatures, similar to Phoenix, Arizona and Las Vegas, Nevada. A fault tree model was
23
developed to estimate the probability that consequential service outages in quantity and quality
24
will occur. Component failures are projected to have a percent increase of 10 - 101% in scenarios
25
where peak summer temperature has increased from 36 to 44o C, which create the conditions for
26
service outages to have a percent increase of 7 - 91%. Increased service outages due to multiple
27
pumping unit failures and water quality non-compliances are the most notable concerns for water
28
utilities. The most effective strategies to prevent temperature-related component failure should 1 ACS Paragon Plus Environment
Environmental Science & Technology
Page 2 of 27
29
focus on maintaining sufficient chlorine residual and cooling pumping unit motors and
30
electronics.
31
INTRODUCTION
32
Civil infrastructure systems are vital for delivering resources, providing protection, and
33
facilitating most urban activities. Typically, these systems are designed to last for long periods of
34
time, often on the order of decades, and some systems persist for over a century.1 Today,
35
infrastructure operational limits are designed based on historical climate conditions,1 and global
36
climate models project that these conditions will likely be more frequently exceeded in the
37
future.1 Consequently, the predictions of infrastructure reliability from historical climate data
38
may over-predict the lifespans and reliability under future conditions. One particular climate
39
change-related hazard is global temperature rise, and in regions that already have hot climates,
40
further increases in temperature may pose serious risks for people and the infrastructure upon
41
which they rely.2 In the US, of particular concern is the Southwest region, where limited water
42
supplies coupled with further increases in temperature may pose major challenges. For example,
43
the National Climate Association reports that the Southwestern US “regional annual average
44
temperatures are projected to rise by 1.4-3.0oC by mid-century and by 3.0-5.3oC by end-of-
45
century with continued growth in global emissions (A2 emissions scenario), and with the greatest
46
increases being in the summer and fall”.3 If urban densification occurs, it has the potential to
47
cause even greater temperature increases via urban heat island.4 In hot climates, it is reported that
48
infrastructure components occasionally fail from high summertime temperatures because of
49
overheating and increasing rates of undesirable chemical reactions.5,6 These failures can lead to
50
service outages when there is not enough redundancy or emergency response.7 Thus, increased
51
temperatures have the potential to increase the probability of outages if design, operation, and
52
management practices remain the same. An infrastructure of particular concern is that of potable
53
water distribution, where heat may result in failures that lead to disruptions of quantity and
54
quality.7
55
Water distribution systems are particularly critical to the economic health of cities and
56
regions, especially in hot conditions. The delivery of safe and sufficient water to residents and
57
commercial establishments is vital to almost all residential, commercial, industrial, and public
58
operations, and the overall reliability is dependent upon both the availability of the resource, the 2 ACS Paragon Plus Environment
Page 3 of 27
Environmental Science & Technology
59
reliability of the infrastructure, and the quality of the water. It is especially important that water
60
systems remain reliable as temperatures rise because in addition to greater consumption by
61
individuals,8 the viability of many services may also require increased consumption. The
62
electricity generation and agricultural industries in particular may need increasing amounts of
63
water in a hotter future.9–11 For individuals, heat exposure can also cause a variety of health
64
issues that would be significantly exacerbated without access to clean water.12 Additionally,
65
research has shown that from increased evaporation and decreased snowmelt, the amount of
66
fresh water available to some regions will diminish with increasing global temperatures.3 When
67
the availability, infrastructural reliability, and quality are all stressed by rising temperatures,
68
there could be a significantly greater threat of provisional inadequacies and consequential human
69
health and economic losses.
70
The reliability of water systems is already challenging to maintain under current temperature
71
conditions. Utilities do not always implement asset management programs to track the failure
72
rates of their components and strategically plan for preventative maintenance to decrease service
73
outages.13 Instead, components are typically operated to failure and utilities rely on their ability
74
to quickly respond and repair, as preventative maintenance budgets are usually constrained.14,15
75
There remains a question as to whether this strategy will continue to work under future
76
conditions. Some utilities have identified potential threats of increased temperatures, but very
77
few have formally included climate change in their design process.5 The American Society of
78
Civil Engineers notes that even when scenarios of climate change are explored, there will be “a
79
tradeoff between the cost of increasing the system reliability and the potential cost and
80
consequences of potential failure”.1 Utilities therefore need information to help them prioritize
81
decisions in planning, design, maintenance, and operations.
82
Given the potential for increasing temperatures, utilities will need to know how their systems
83
may be affected and where efforts should be most focused to prevent and prepare for changes in
84
their system. More specifically, they will need to know how to prepare to mitigate failures ahead
85
of time and how to prepare for the effects of failures.16 This study strives to answer two
86
questions in the context of seasonally hot cities where temperature increases risk of failure: (1)
87
What is the projected increase in risk of potable water service loss with increases in maximum
88
summertime temperatures? and (2) Where should municipal water utility management strategies
89
be focused to mitigate increases in risk? To address these questions, the functionality of water 3 ACS Paragon Plus Environment
Environmental Science & Technology
Page 4 of 27
90
system components and water quality are analyzed under temperature conditions characteristic of
91
Phoenix, Arizona and Las Vegas, Nevada, considering climate change. An exploration of the
92
temperature-related sensitivities and consequential risk is valuable for understanding issues that
93
may affect other cities in the future.
94
METHODOLOGY
95
To aid water utilities in understanding where their systems will be more vulnerable and
96
how they can strategically mitigate temperature-related service interruptions, the following
97
approach was used: i) potential temperature-induced sensitivities were identified for physical
98
components and aspects of water quality; ii) quantitative relationships between failure and
99
temperature exposure were identified; iii) ranges and distributions of major operating conditions
100
impacting degradation were identified; iv) average maximum daily summertime temperature
101
projection data were processed for Phoenix, Arizona and Las Vegas, Nevada; vi) Monte Carlo
102
simulations were used to perform calculations of failure metrics for each temperature scenario
103
given operating conditions; vi) probability distribution functions were fitted to Monte Carlo
104
outputs; vii) projections of failure using failure rates for physical components and probabilities
105
of water quality non- compliance were calculated from probability distribution functions; and,
106
ix) fault trees were created to estimate how individual component failures and quality non-
107
compliances could propagate to service outages given different operational scenarios. The
108
resulting component projections of failure, percent increases in failure, and expected number of
109
service outages under scenarios of the possible temperature exposure in the next 30 years
110
considering best and worst case operating conditions are used to make recommendations for
111
focused management strategies. A process flow diagram of the methodology is shown in SI
112
Figure S1.
113
Quantifying Temperature-Related Exposure and Degradation
114
The physical components and aspects of water quality that have been shown to be
115
sensitive to temperature were first identified through literature review to assess how operational
116
failures might increase. Temperature sensitivities affect component wear and the rate of chemical
117
reactions, the latter leading to the potential for increased pipe degradation and corrosion or
118
quality non-compliance. Temperature-sensitivities are found in the motors17–20 and 4 ACS Paragon Plus Environment
Page 5 of 27
Environmental Science & Technology
119
electronics21,22 used in pumping units, thermoplastic23 and metal pipes24, and in the chemical
120
processes in the water from decay of the disinfectant residual25 and increase in disinfection
121
byproduct production (DBP)26,27 (SI Table S1). While high temperature is known to affect water
122
demand8, soil expansion28, material stress in pipes from temperature change29,30, the health of
123
system operators31, nitrification rates32–35, and the growth of harmful bacteria like
124
Mycobacterium Avium Complex36,37 and Legionella,38,39 these effects were excluded from
125
analysis because of lack of basic data and quantitative relationships.
126
There is ample evidence that summer temperatures contribute to degradation of water
127
system components and quality. From experience of operation, the Las Vegas Valley Water
128
District has found that their cooling systems for pumping units may be inadequate for higher
129
summer temperatures and that thermoplastic pipes fail more frequently in the desert heat.5,40
130
Observational studies of rates of corrosion and DBP production over year-long periods show that
131
rates are higher during summer when the water is warmer.24,26,27,41 Additionally, an experimental
132
study of chlorine residual decay in water samples also shows that reaction rates increase when
133
samples are subjected to warmer water temperatures under periods of about a day.25
134
Studies of environmental and public health hazards have found that the impact of a
135
hazard depends on “the concentration, amount or intensity of a particular agent that reaches a
136
target system in terms of its duration, frequency, and intensity”.42 It is therefore assumed that
137
physical components and aspects of water quality would have varying levels of degradation as a
138
function of their durations and magnitudes of exposure. While the empirical studies of water
139
quality aspects and corrosion rates show that a summertime period or shorter is enough to cause
140
the reported change in reaction rates, the exact duration of exposure that it takes for temperatures
141
above rated thresholds to cause degradation to the physical components (i.e. PVC pipes, motors,
142
and electronics) at the reported rates is unknown. We speculate that because degradation is
143
cumulative for physical components, cumulative exposure duration might be important. No
144
empirical data or models were identified to establish a useable relationship between cumulative
145
exposure and degradation rate for these components, so we propose a theoretical relationship that
146
is derived in SI Section S2. However, we are not able to employ this cumulative temperature
147
model in the current paper owing to a lack of empirical data on the effects of cumulative
148
exposure. Fortunately, there is some information available on the relationship between maximum
149
temperature exposure and component degradation rate. One source states that reported motor 5 ACS Paragon Plus Environment
Environmental Science & Technology
Page 6 of 27
150
degradation rates apply “even if the overheating was only temporary”.19 We therefore interpret
151
published degradation rates for PVC pipes, motors, and electronics as being appropriate to
152
characterize degradation from brief but repeated exposure of water systems to peak temperatures
153
over the components’ service lives. When water systems are deployed in the field, they are
154
exposed to exactly this type of pattern of “temporary”, but repeated, maximum temperatures; this
155
exposure happens almost every summer afternoon. The hottest afternoons in U.S. cities tend to
156
occur in June, July, and August, which is when temperatures sometimes exceed safe operating
157
temperature ratings of components.
158
Since both physical components and aspects of water quality are affected by high
159
summertime temperatures, this study models failure due to predicted averages of maximum daily
160
summertime temperatures between June and August, during the hottest three hours of each
161
summer day. This model has a cumulative-peak-exposure interpretation because a component in
162
our model is exposed to this peak temperature for roughly 270 hours at least once and at most
163
every year (the average temperature of the hottest 3 hours each day43 for 3 months, referred to as
164
“3x3” in this paper). If it is the case, however, that degradation rates of physical components
165
increase every time the component experiences much smaller durations of exposure (e.g. during
166
a minute, hour, or week), then the model underestimates projections of failure. If it is the case
167
that the degradation occurs only given years of continuous exposure that are longer in duration
168
than the 3x3 window time frame, the model overestimates failures. With empirical daily or
169
monthly failure data, future studies could determine the specific duration of exposure needed to
170
cause these rates of degradation to physical components.
171 172
Urban Water System Case Study
173
Given their large populations, hot environments, and modern infrastructure, a potable
174
water system with characteristic conditions of the cities of Phoenix, Arizona and Las Vegas,
175
Nevada was modeled as a case study. These metro areas are two of the largest and fastest
176
growing regions in the Southwestern US, and experience some of the highest temperatures of
177
metro area across the US.44,45,47 They both have populations of around 5 million people and
178
experience 35 – 40oC average daily summer temperatures.44,45 The largest water utilities in each
179
metro region service around 1.5 million customers each.46,47 Since studies on temperature affects 6 ACS Paragon Plus Environment
Page 7 of 27
Environmental Science & Technology
180
to operation are sparse, the failure metric equations and ranges of operational characteristics
181
from literature were used from systems around the world that do not necessarily represent the
182
Las Vegas and Phoenix case studies. Therefore, the temperature data and number of components
183
are the aspects of the system that are characterized by the case studies. A comparison of the
184
relevant water utility characteristics is shown in SI Table S2.
185
Climate projections show significant increases in temperature in Phoenix and Las Vegas
186
into the future. Phoenix and Las Vegas projections of the averages of maximum daily summer
187
temperatures (3x3) are shown in SI Figure S3. Temperature projections were processed from
188
CMIP5 12x12 km gridded data from Global Climate Model (GCM) projections of representative
189
concentration pathway (RCP) scenarios 2.6, 6, 4.5, and 8.5.2 The 3x3 maximum air temperature
190
in Phoenix is projected to increase to at least 40°C in 2020 and at most 44°C in 2050 (including
191
the GCM’s standard deviation of temperature). In Las Vegas it is projected to increase to at least
192
36°C in 2020 and at most 41°C in 2050. The temperature range used to force the failure model
193
considers the total range of both cities with GCM uncertainty, i.e., 36 – 44oC at 1oC intervals.
194 195 196
Modeling Increases in Component and Water Quality Failure Failure can be measured as events where reliability requirements are not met.15 Motors
197
and electronics within pumping units failing to operate, pipes breaking and not delivering water
198
at required pressures, chlorine residual concentrations below regulation, and DBP concentrations
199
above regulation each constitute failures. Estimated-time-to-failure (ETTF), which represents the
200
lifetime of components in units of years, and chemical concentration therefore represent “failure
201
metrics” that indicate the potential for failure of components and the aspects of water quality,
202
respectively. With temperature sensitivities identified, a review was conducted to identify the
203
quantitative relationships between temperature and component failure and chemical reaction
204
rates to characterize failure metrics. The ETTF parameter was calculated when the quantitative
205
relationship between exposure temperature and lifespan of component was given and was used to
206
estimate effects of overheating on motors and electronics, degradation of polyvinyl chloride
207
(PVC) pipes, and corrosion of iron pipes. Chemical concentration represents the failure metric of
208
chlorine residual decay and the production of DBPs.
7 ACS Paragon Plus Environment
Environmental Science & Technology
209
Page 8 of 27
An increase in ambient temperature threatens overheating of motors and electronics that
210
are vital to the operation of the pumping units. Motors can overheat from the combined
211
dissipated heat from motor windings and the ambient temperature surrounding the motor,
212
causing destruction of the insulation which can lead to burnt stator windings.17–20 It is reported as
213
a rule of thumb in the industry that for every 10°C increase in the operating temperature over the
214
capacity of the insulation--155°C for class F motors -- the lifespan decreases by one-half (SI
215
Section S4.1.1).17–20 Conversely, the lifespan increases by one-half for every 10°C decrease in
216
the operating temperature below the capacity of the insulation. 18 The electronic controls that are
217
used for pump operation are also sensitive to the combination of dissipated heat and the outside
218
temperature within their enclosures, and for every 10°C rise in enclosure temperature above
219
40oC, the lifespan of the electronics decreases by one-half (SI Section S4.1.2).21,22 It was
220
assumed that electronics have the same property as motors when temperatures are below their
221
capacity. With these relationships, we estimate the ETTFs that result from exposure to average
222
peak summertime temperature scenarios.
223
Water temperatures in the distribution system rise in response to increases in ambient
224
temperatures. The relationship between water and air temperatures was modeled using the
225
empirically derived linear regression equations from a study of temperatures at water treatment
226
plant outlets in Japan.48 Coefficients of the regression range from 0.52 – 0.89 oC in water / oC in
227
air and the constants range from 1.88 – 7.89 oC in water. Despite some novel research on the
228
topic, 48,49 predicting water temperature within distribution pipes is challenging due to a lack of
229
empirical data. With high water temperatures, thermoplastic pipes can experience overbearing
230
pressures, with the greatest effects on PVC.23 The derating of the PVC pipe is linear with
231
increasing water temperatures (SI section S4.1.3). Iron pipes are sensitive to indirect effects of
232
water temperature through internal corrosion.24 The relationship between corrosion rate and
233
water temperature over a year-long period has been reported in the literature for the cast iron
234
pipe material, so this is the type of iron pipe modeled. Temperature may have a similar effect on
235
the corrosion rate of ductile iron and steel pipes, though no relationship was found in literature.
236
Corrosion rates for cast iron pipes are reported to be empirically different for water distribution
237
systems (WDS) and water treatment plants (WTP) (SI Section S4.1.4). 24 Corrosion rates and
238
pipe age were used to calculate pit depth and then to the calculate the ETTF from the remaining
239
life of the pipe according to Randall-Smith et al.50 8 ACS Paragon Plus Environment
Page 9 of 27
Environmental Science & Technology
240
Water quality is also affected by an increase in water temperature. Summertime
241
temperatures increase the reaction rates between the organics and disinfectants in water, thereby
242
increasing the formation of the cancerous DBP, total trihalomethane (TTHM), according to an
243
empirical study on seasonal drinking water quality in Istanbul City, Turkey, where water is
244
supplied through surface water and is treated through “aeration, prechlorination, coagulation,
245
flocculation-sedimentation, filtration, and postchlorination” (SI Section S4.1.5).26 The
246
concentration of another DBP, total haloacetic acid (THAA), is directly dependent upon the
247
concentration of TTHMs and seasonal temperatures as well, according to an empirical study of
248
three drinking water systems in the United Kingdom which represent a range of source water
249
conditions – “upland surface water, a lowland surface water, and groundwater” with standard
250
treatment mechanisms: aeration, filtration, coagulation, sedimentation, and chlorination (SI
251
Section S4.1.6)27.
252
The other type of temperature-related quality concern is that of chlorine residual decay as
253
water travels to the consumer. The final chlorine concentration was calculated based on the
254
initial concentration of chlorine from dosage at a water treatment plant, the minimum allowable
255
concentration of chlorine, and the chlorine decay constant (SI Section S4.1.7). The decay
256
constant’s relationship with temperature experienced over day-long periods was taken from an
257
experimental study with samples from two water distribution systems in Birmingham, Alabama,
258
by Hua et al.25 Table 1 shows the equations used to calculate the failure metrics for each
259
component and aspect of water quality, though a more detailed discussion can be found in the SI
260
section S4.
261 262 263 264 265 266 267 268 269 270
Table 1: Failure Metric Equations. More detailed equations are shown in SI section S4. T= temperature rise above component threshold , Tw = water temperature, ETTF (T/Tw) = estimated-time-to-failure after air or water exposure [years], C(Tw) = chemical concentration after water temperature exposure, MTTFT1 = historical mean-time-to-failure [years], rd = lifespan degradation fraction per 10oC above capacity [no units], t = age of pipe [years], tw = water age, Pi = internal pit depth [cm], ∂ = pipe wall thickness [cm], C0 = initial chlorine concentration [mg/L], TOC = total organic carbon [mg/L], Cl2 = chlorine dosage [mg/L], SUVA = specific UV absorbance [l/mg*m], Br = bromide [mg/L], ResT = water age [h], Season = season of year, numerically expressed as: 1 for spring, 1.46 for summer, 1.31 for autumn, and 1.01 for winter. Component/Aspect of Water Quality ETTF (T or Tw) = Motors and Electronics ( ∗ ܨܶܶܯ1 − ்) ݎ/ଵ ்ଵ ௗ
Failure Metrics C(Tw) =
Source 17–22
9 ACS Paragon Plus Environment
Environmental Science & Technology
Page 10 of 27
23
PVC Pipes ்ܨܶܶܯଵ ∗ (−0.0123ܶ௪ + 1.293) ݐ Iron Pipes ߜ 0.5(ݐ0.0774 ∗ ܶ௪ − 0.1073) + ܲ Chlorine Residual Concentration TTHM Concentration THAA Concentration
24,50
ܥ ݁
.ହ బ.బఴరభೢ ି ௧ೢ
25
11.967(TOC)0.398*Tw 0.158*Cl20.702
26
0.99(TTHM)0.64*(Cl2)0.15*SUVA0.09*(Br+0.005)-0.12*(ResT+5)0.07*Season
27
271 272
While temperature contributes to failure, the effects of operating characteristics can also be
273
significant. SI Table S6 shows the characteristics used to calculate the failure metrics. Examples
274
include the range in water temperature vs. air temperature regression coefficients, MTTFT1 and
275
heat dissipation of motors and electronics, and the age of water in pipelines. Estimates of the
276
possible ranges and likely distributions of these characteristics were identified through literature
277
review and were modeled as uniform or lognormal distributions to account for their uncertainty.
278
Uniform distributions were used when there was no information available about the relative
279
likelihood of certain values over others within the reported range. Lognormal distributions were
280
used when there was evidence for a skewed distribution present in real water distribution
281
systems.
282
Monte Carlo simulations were then performed to calculate and characterize probability
283
distributions of the failure metrics. When performing the calculation of failure metrics for each
284
temperature, the distributions of operational characteristics were sampled 5,000 times. To
285
estimate the variation in failure metrics under different regimes of operating characteristics, the
286
ranges of operational characteristics were divided into lower and upper halves (representing best-
287
a worst- case operating conditions) and Monte Carlo simulations were run over both halves
288
separately. The terms best- and worst- cases within the reported literature are not determined to
289
be optimal and sub-optimal respectively from independent sources. Distributions of “failure
290
metrics” were created for each average maximum summertime temperature and operational
291
scenario by performing Monte Carlo simulations for each 1oC increase in the range 36-44oC,
292
resulting in similar histogram outputs that are just shifted to higher values of failure metrics with
293
increasing temperature. The output failure metric values from the simulations were fit into
294
probability distribution functions using Anderson Darling Statistics, probability plots,
295
visualization and judgement about distributions that best fit the process underlying the data. 51 10 ACS Paragon Plus Environment
Page 11 of 27
Environmental Science & Technology
296
Weibull distributions were fitted to the physical components in pumping units and pipes because
297
the form most accurately represents degradation with age and it also produced reasonable fits, as
298
shown in figures S4-S11 .52 These distributions characterize the histograms of component
299
failures from a population of components overtime, and time represents ages of the components
300
– starting at zero.52 Motors and PVC pipes were the only components that had outputs that were
301
statistically equal to the Weibull at the 10% significance level. The output distributions for
302
chlorine residual concentration were fitted as exponential distributions based on best fit and the
303
need to be consistent across operating scenarios. The output distributions for DBP concentration
304
were fitted as Gamma distributions based on best fit. Output data and their fit distributions and
305
probability plots are shown in figures S4-S20. Parameters of the distributions and Anderson-
306
Darling Statistics are shown in SI Tables S8 and S9.
307
The projection of each component failure and type of water non-compliance was calculated
308
through either hazard functions or integration of the distributions characterized by the failure
309
metrics. The ETTF distributions of physical components were used to calculate components’
310
annual failure rates with the hazard function of the Weibull distribution (units: % failed/year),
311
which characterizes their failure behavior starting in the next instant of their lives, given that they
312
had already lived a certain number of years. 53 The time period of the rate was then converted to
313
be during the next summertime period of its life instead of a full year --as characterized by a
314
period of 270 hours out of the total 8760 hours in a year (SI Section S4.3.1).53 It is necessary to
315
calculate failure rate after the components have operated up until a non-zero age because there
316
would be very little likelihood of their failure given any temperature exposure if they were brand
317
new. The DBP formation distributions were used to calculate the probability that a concentration
318
from the distribution would be above the EPA regulated threshold of 80 µg/L and 60 µg/L for
319
THMs and THAAs respectively54 for each sampling station in the network for any point in time
320
that the water temperature scenario is experienced, as shown in SI section S4.3.2. Similarly, the
321
chlorine residual distributions were used to calculate the probability that a concentration from a
322
sampling station would be undetectable and therefore non-compliant with EPA regulation for
323
any point in time that the water temperature scenario is experienced.55 A figure of the overall
324
failure metric distribution formation and probability calculation methodology is shown in Figure
325
1.
11 ACS Paragon Plus Environment
Environmental Science & Technology
Page 12 of 27
326 327 328 329 330 331
Figure 1: Methods of projection of component-level failure calculations. For physical components, the ETTF distribution is shifted to the left, which effectively shifts to the left through Monte Carlo analysis under increasing temperature scenarios. For disinfection byproducts, the failure metric, chemical concentration (C) is shifted to the right, and for chlorine residual is shifted to the left. Shaded regions represent integrated areas under the curves that are used to calculate failure rate and probability of failure.
332
Modeling Failure Cascades to Service Outages
333
Component probabilities of failure and failure rates were propagated to probabilities of
334
systemic failure using a reliability engineering fault tree analysis to assess the impact of
335
component hierarchies, points of redundancy, and presence of back-up systems on the likelihood
336
of systemic failure.53 Water distribution systemic failure is defined as the pressure, flow, or
337
quality falling below specified values at one or more nodes in the network. The three temperature
338
induced pathways to service outages are pumping station outages, pipe breaks, and water quality
339
non-compliance, and are shown as individual fault trees in Figure 2. Pipe breaks and pumping
340
station outages can both directly cause failures in the water distribution system (WDS) and
341
indirectly cause failures in the WDS due to failures of water treatment plants (WTP). The fault
342
trees highlight which component failures lead to service outages. To calculate systemic
343
probabilities of failure, physical components were assumed to have series behavior (meaning that
344
if any one component fails, the sub-system fails), and failure rates were propagated from a
345
component level to a system level according to the system reliability equations that state that
346
individual component or sub-system failure rates are summed in a series system (SI Section
347
S4).53 The use of the hazard function to calculate component failure rate allows for the
348
characterization of failure behavior of all components during the same instant in time (when they
349
are all at the certain ages in the scenario). To calculate the probability of water quality non-
350
compliance, the probability of either a non-compliance from TTHM or chlorine residual decay 12 ACS Paragon Plus Environment
Page 13 of 27
Environmental Science & Technology
351
occurring was found through the union of their probabilities (SI Section S5).53 Expected
352
occurrences of service outages were calculated from these propagated failure rates and
353
probabilities, along with the estimates of the number of components. Scenarios of numbers of
354
available redundancies of components were explored in the calculation of pumping unit failure
355
rate leading to pumping station failure rate, water treatment or water distribution failure rate, and
356
ultimately to distribution system failure rate (SI Tables S4 & S5). These values are relative to the
357
average number of pipes, pumping stations, and quality sampling sites in Phoenix and Las Vegas
358
(SI Table S5). An important type of redundancy is stored water in tanks throughout the network
359
that can offset the pressure and flow lost during a pipe and pump station break and provide a
360
location for re-dosing the distribution system with chlorine disinfectant. Without estimating
361
water volumes at different times, water distribution tank storage was assumed to be 100% likely
362
to be inadequate for stopping a pumping station outage or pipe break from causing some
363
magnitude of water outage. A more specific representation of the availability of water storage
364
would require knowing temporal and spatial information of the system structure and operations
365
over time.
366 367 368 369 370
Figure 2: Fault Tree Diagrams showing three trees of systemic failure leading to a possible water service outage: 1) pump station outage, 2) pipe break, and 3) water quality non-compliance. The boxes represent failure events and their different colors represent hierarchies of failure (i.e. sub-component, component, intermediate systemic, ultimate systemic). It was assumed that if one out of two pumping units, pumping
13 ACS Paragon Plus Environment
Environmental Science & Technology
371 372 373
Page 14 of 27
stations, and water treatment plants failed, demand would not adequately be met and failure would propagate as detailed in SI Table S4.
It is possible that these individual pathways to outages could also happen
374
simultaneously—leading to potentially longer and more severe supply and quality outages. For
375
example, low pressures in the system from pipe breaks or pumping station outages not only
376
independently cause quantity supply outages and seepage of contamination, but when coupled
377
with pre-existing quality issues, pose a human health threat.56 If both physical components and
378
quality fail in the same time frame, the customers could experience longer periods of low
379
pressures and durations of contamination because of the potential difficulty of rerouting and
380
flushing water.15,56 The probabilities of simultaneous occurrences of different types of water
381
outages were thus also calculated, using standard probability law of simultaneous events for the
382
expected values of service losses from physical component failure rates and water quality non-
383
compliance probabilities (SI Section S5). Simultaneous probability was estimated through
384
viewing the conditional probability portion of the physical component failure rate as stand-alone
385
under the certain time interval of 270 hours.
386
RESULTS
387
Projected Increase in Component Risk
388
Over the full range of possible temperatures in all RCPs and including GCM uncertainty,
389
the increase in component probability of failure from 2020 to 2050 ranges from 10-101% for the
390
Phoenix and Las Vegas-characteristic utility. The average probabilities of failure and failure
391
rates (between best- and worst- operating conditions) are the highest for chlorine residual and
392
pumping stations, pumping units, motors and electronics (within pumping units). This holds
393
across all temperatures, as shown in Figure 3. Detailed results are shown in SI Table S10.
394 395
14 ACS Paragon Plus Environment
Page 15 of 27
396 397 398 399 400 401 402
Environmental Science & Technology
Figure 3: Average Component and Water Quality Projection of Failure: Component failure rates and probabilities of failure are shown as a function of temperature. Temperature ranges for Phoenix and Las Vegas are plotted on the abscissa. Chlorine residual, and pumping units are the components with the highest probability of failure. Pumping stations, inadequate chlorine residual, and motors are projected to have the greatest percent increases in failure.
The components that pose the greatest threat to reliability are those that have both the
403
highest probability or rate of failure and the greatest percent increase in these values between the
404
2020 and 2050 scenarios. Thus, the most concerning aspect is water quality non-compliances due
405
to the decay of chlorine residual and TTHM production. Chlorine residual decay will have the
406
largest probability of failure and will also have a large percent increase with increasing
407
temperatures. The relationship between inadequate chlorine residual and temperature is
408
exponential, and therefore the percent increase in failure rates between 2020 and 2050 will be
409
53% ± 36% on average. There is no validation available for the historical frequency of non-
410
compliances, though there is evidence that non-compliances have been non-zero. Annual water
411
quality reports show that out of the last 5 years in the City of Phoenix, there were 4 years where
412
concentrations fell below the minimum residual concentration at least once sometime during the
413
year.57–61 TTHM production has the second highest probability of failure but has a lower percent
414
increase of 17% ± 10% on average. It is unlikely however that the non-compliance thresholds of
415
the other form of DBP, THAA, will be exceeded under this scenario of operational practices.
416
This is because the calculated concentrations of THAA at different temperatures were only
417
around 25% of the regulated levels. Though the chance of occurrence is high, it should be noted
418
that there are no reports of DBP violations in the past in either city in recent years.46,62
419 420
The failure processes of second-most concern are those of electronics, motors, pumping units and pumping stations. Pumping station failure rates are projected to have a percent increase 15 ACS Paragon Plus Environment
Environmental Science & Technology
Page 16 of 27
421
of 76% ± 15%, depending on operating conditions. The historical average failure rate of motors
422
across a variety of industries in the US is on average 3-12% every year under current temperature
423
conditions.17 Extending the 270-hr failure rates to yearly values provides a value to check
424
validity with historical data. The estimates of annual failure rates under all temperature scenarios
425
from 36oC (8.6% ± 5.6%) to 44oC (26% ± 16%) fall within this historical range.
426
Pipe failures show a less significant increase in threat to utility reliability in the future
427
compared to other component failures. Between the two types of degradation, corrosion causes a
428
greater increase in probability of failure than does the degradation of thermoplastic pipes, though
429
probability of failure of PVC pipes from degradation will be consistently greater than probability
430
of failure of iron pipes from corrosion. The corrosion process associated with iron pipes causes
431
the probability of failure to have a percent increase of 52% ± 12% in the average WDS, and of
432
76% ± 8% in the average WTP. The PVC pipe failure rate has a percent increase of 10% ± 0.2%
433
in the average WDS. The historical failure rates of polyethylene pipes, pipes with similar
434
degradation rates to PVC pipes, in Las Vegas in 2005 was 2.2% (679 breaks out of 25,000 PE
435
pipes) in the summer and 6.5% over the year (1623 breaks out of 25,000 PE pipes).5,40,63 The
436
PVC pipe failure rate estimate from the model given the average peak summer temperature of
437
2005 (44oC)64 in Las Vegas is 0.2% ± 0.13% in the summer and 6.5% ± 4.3% over the year, so is
438
consistent with historical data. Current annual iron pipe break rates are reported to be 6% on
439
average in the United States.47 For WTP iron pipes, the estimates of annual failure rates under all
440
temperature scenarios from 36oC (0.32% ± 0.32%) to 44oC (0.84% ± 0.84%) fall below this
441
historical range. Additionally, WDS iron pipe failure rates are negligible under all temperatures
442
(0.005% ± 0.005% per year at 44 oC). Underestimates in modeled versus observed failure rates
443
are in part due to the fact that there are multiple modes of pipe breaks in reality - longitudinal,
444
circumferential, corrosion through hole, split bell/bell shear, and joint failure28 - that are not
445
accounted for in the model. In the results, the lifetime of the pipe is only modeled from an
446
increase in corrosion and degradation. While the Weibull distribution can predict other forms of
447
aging, it cannot account for the random breaks due to factors like inadequate bedding support
448
and live loads caused by traffic that are independent of climate effects.28 Moreover, results of
449
pipe failure rates show that utilities should expect the high rate of PVC pipes to slightly increase,
450
but should also anticipate an increase in WTP iron pipe breaks with increases in temperature.
451 16 ACS Paragon Plus Environment
Page 17 of 27
452 453
Environmental Science & Technology
Projected Increase in Service Outages The probability of service outages increases as a result of propagated component failures.
454
Individual water outages and water quality non-compliances are projected to have a percent
455
increase within the 7-91% range and simultaneous water outages and water quality non-
456
compliances are projected to have a percent increase within the 23-123% range, depending on
457
the type of event and how the system is operated. Figure 4 shows the resulting expected
458
occurrence of water outage and quality failure when all combinations of systemic probability are
459
considered.
460 461
462 463 464 465 466 467 468 469 470
Figure 4: Isolated and Simultaneous Service Outages. Dashed lines represent Las Vegas temperature ranges from 2020 – 2050 while dotted lines represent temperature ranges in Phoenix. Venn diagrams in the first column show which type of systemic failure was analyzed. The ranges of failures due to changing operating conditions are shown in colored bands to characterize opportunities for risk mitigation. All best-case operating conditions together contribute to the lower range of expected number of failures and all worst-case conditions together contribute to the upper range of expected number of failures. The large range shown from the bands suggests that the expected number of failures is sensitive to changes in operating conditions.
Results show that the components that have the highest probability of failure and failure
471
rates and that increase the most in failures - namely water quality non-compliances and pumping
472
station outages - directly contribute to the greatest increase in service outages. The type of
473
isolated water outage that is projected to increase the most is that from pump station failures. The
474
expected number of water outages from pump station failures has a percent increase of 76% ± 17 ACS Paragon Plus Environment
Environmental Science & Technology
Page 18 of 27
475
15%. The second highest percent increase in water outages is from quality non-compliances of
476
17% ± 0.4%, and lastly, water outage percent increase from pipe breaks is estimated to be 10% ±
477
3%. Unfortunately, there was no identified available historical data for how many simultaneous
478
outages currently occur currently for direct comparison.
479
Because the individual modes of water outage that increase the most are also more likely
480
to happen simultaneously with other types of service outages, the simultaneous service outages
481
that will have the greatest percent increase are those that include pump station outages and water
482
quality non-compliances. This means that though utilities are used to responding to frequent pipe
483
breaks, the increasing simultaneous occurrence of pipe breaks with pumping station outages
484
and/or water quality non-compliances could cause outages that are increasingly difficult to
485
recover from.
486
Priority Maintenance Strategies
487
Instances of water quality non-compliance and pumping station failures are also the failure
488
modes that have the greatest potential for being prevented, and as such should be prioritized for
489
failure prevention. The large sensitivity in failures for all events that are caused by water quality
490
non-compliances and pumping station failures (as shown in the colored bands in Figure 4) show
491
that changing the operating conditions for these components would be the most effective way to
492
reduce failures. The trends in failure shift from linear to exponential between operating
493
conditions. Specifically, the results show that there is a maximum 59% absolute reduction in
494
probability of inadequate chlorine residual failure when operators inject the higher range of dose
495
of chlorine at the entrance point to the WDS, and make sure to maintain a low water age
496
throughout the system. It should be noted, however, that injecting the maximum chlorine dosage
497
and allowing for more organic carbon causes a 64% absolute increase in probability of non-
498
compliances from TTHM production. Therefore, both residual chlorine and DBP concentrations
499
should be monitored carefully under any chlorine dosing strategy. Pumping station failure rates
500
are also sensitive to operating conditions as there is a maximum 15% absolute reduction in the
501
pumping station failure rate when motors and electronics dissipate low amounts of heat, and
502
there are cooling devices implemented that reduce operating temperatures. The expected value of
503
pipe failures from corrosion and degradation is not very sensitive to changes in characteristics
18 ACS Paragon Plus Environment
Page 19 of 27
Environmental Science & Technology
504
like pipe diameter and normal operating pressure (1% absolute reduction), and thus could not be
505
easily decreased through changing these characteristics.
506
MODEL UNCERTAINTY
507
Uncertainties associated with water temperature in underground pipes, the time it takes
508
for degradation to occur, availability of water storage, component failures from waterhammer,
509
and projections of future trends in model parameters should all be considered when assessing
510
applicability of the risk projections to a specific water utility. These characteristics were
511
necessarily inserted into the model as assumptions, but their variability would have an effect on
512
the output probabilities of failure. A 10% change to degradation rates results in a percent change
513
of 0-22% of motor and electronic failures depending on the temperature and operating conditions
514
scenario. Specific values are shown in SI Section S7. The addition of this variance would make
515
pumping stations more comparable to all other components in terms of percent increase in
516
failure. Assuming that there is a 50% chance (instead of 100% chance) of an inadequate amount
517
of water storage decreases the likelihood of outage from pipe break by 50% and an outage from
518
pumping station failure by 50%. Additionally, if quantitative relationships to describe the effect
519
of waterhammer from one component failure causing another became available, the frequency of
520
pipe and pumping unit failure might increase. Lastly, it is hard to know what the resulting risk
521
will be when normal structural and operational characteristics change over time from urban
522
expansion, transformative designs, etc.
523
DISCUSSION
524
The study is a critical first step towards helping utilities prepare for climate change and
525
extreme events by identifying and characterizing the aspects of the system and chain of failure
526
events most sensitive to heat. The results are modeled to show how discrete increases in
527
temperature increase chances of failure. As framed, the results provide a directionally reliable
528
estimate telling us that component and service failures will increase with increasing
529
temperatures. While estimates of risk contain uncertainty, comparisons between the estimated
530
increase in risk for each temperature-sensitive aspect of the system are nevertheless valuable for
531
prioritizing where strategies should be focused. 19 ACS Paragon Plus Environment
Environmental Science & Technology
532
Page 20 of 27
The results also help identify several critical areas for future research and data collection.
533
This model can be most directly applied to warm regions around the world with modern
534
centralized water distribution infrastructure, (e.g. the US Sun Belt, Middle East, North Africa,
535
and South Asia), where a large and growing fraction of humanity lives and where a large fraction
536
of global 21st century water infrastructure investment will occur.65 Even in colder regions the
537
model may be useful, with necessary further work because an increase in annual and summer
538
temperature may benefit the system operation by helping to prevent pipes from freezing and
539
accelerating the rates of beneficial chemical reactions needed for water treatment. 66
540
Additionally, a network model of components and iterative simulations through time would be
541
beneficial to capture locations of component failures leading to different resulting magnitudes in
542
outages, and the accumulation of degradation overtime by exploring the alternative assumption
543
that shorter time periods (rather than the 3x3 duration) cause the reported degradation rates.
544
More modeling work, combined with city-specific and component-specific engineering data
545
quantifying heat-induced failure rates is necessary to more precisely quantify heat-induced
546
service failure risk in particular WDSs. An analysis of relative costs of different suites of
547
preparative actions and their consequences would also be valuable, but this requires data on
548
preventative maintenance, repair, response, lost consumer use, and capital improvement costs for
549
each type of failure. Furthermore, the methods of projecting risks of future external threats could
550
be expanded to include other threats, including flooding and wildfires. It could also be used for
551
the projection of risk in other infrastructure systems like transportation and electricity, which
552
also have temperature and other climate change event sensitivities.6,67
553
Considering the possibility of other increasingly frequent extreme weather events,
554
utilities should recognize that improved response times coupled with the capacity for agile and
555
flexible resources use (including money and equipment) will be critical. The water community
556
should also work to phase out vulnerabilities by improving system design (e.g. increasing
557
pumping unit insulation capacity, reducing water temperature in cooling towers, or adopting
558
“smart” booster chlorination and network sensors5,68), and by improving training and institutional
559
response capacity. Ultimately, in times when maintenance and response actions are severely
560
constrained by budgets, it is of the utmost importance to identify and prioritize strategies by
561
utilizing information on likely mechanisms of failure.
20 ACS Paragon Plus Environment
Page 21 of 27
Environmental Science & Technology
562
SUPPORTING INFORMATION
563
Methodology process flow, temperature sensitivities and exposure, potable water treatment and
564
distribution system characteristics, number of water system components, redundancy scenarios,
565
number of modeled opportunities for systemic failure, modeled operating characteristics, failure
566
metric calculation method, probability distribution creation method, projection of failure
567
calculation method, system failure laws, failure percent increase results, and model uncertainty.
568
ACKNOWLEDGEMENTS
569
This material is based upon work supported by the National Science Foundation under Grant No.
570
1360509.
571
REFERENCES
572
(1)
573 574
Olsen, R. Adapting Infrastructure and Civil Engineering Practice to a Changing Climate; American Society of Civil Engineers, 2015.
(2)
Downscaled CMIP3 an CMIP5 Climate and Hydrology Projections: Release of
575
Downscaled CMIP5 Climate Projections, Comparison with Preceding Information, and
576
Summary of User Needs. Prepared by the U.S. Department of the Interior, Bureau of
577
Reclamation, 2013.
578
(3)
Garfin, G.; Franco, G.; Blanco, H.; Comrie, A.; Gonzalez, P.; Piechota, T.; Smyth, R.;
579
Waskom, R. Ch. 20: Southwest. In Climate Change Impacts in the United States: The
580
Third National Climate Assessment; U.S. Global Research Program, 462-486, 2014; pp
581
462–486.
582
(4)
583 584
U.S. Environmental Protection Agency Office of Atmospheric Programs, 2012. (5)
585 586
Reducing Urban Heat Islands: Compendium of Strategies - Urban Heat Island Basics.
Heyn, K.; Winsor, W. Climate Risks to Water Utility Built Assets and Infrastructure. Portland Water Bureau, City of Portland, Oregon, 2015.
(6)
Coughlin, K.; Goldman, C. Physical Impacts of Climate Change on the Western US
587
Electricity System: A Scoping Study; Ernest Orlando Lawrence Berkeley National
588
Laboratory; 2008.
589
(7)
Mays, L. W. Reliability Analysis of Water Distribution Systems; American Society of 21 ACS Paragon Plus Environment
Environmental Science & Technology
590 591
Civil Engineers: New York, NY, 1989. (8)
Guhathakurta, S.; Gober, P. The Impact of the Phoenix Urban Heat Island on Residential
592
Water Use. J. Am. Plan. Assoc. 2007, 73 (3), 317–329 DOI:
593
10.1080/01944360708977980.
594
(9)
Bartos, M. D.; Chester, M. V. The Conservation Nexus: Valuing Interdependent Water
595
and Energy Savings in Arizona. Environ. Sci. Technol. 2014, 48, 2139−2149 DOI
596
10.1021/es4033343 |
597
Page 22 of 27
(10)
Hwang, H.; Lansey, K. ARVIN : ARizona Value INtegrated Water , Energy , and
598
Agriculture Planning Model ARVIN : ARizona Value INtegrated Optimization
599
Model.Presentation.[Online] 2016. https://wrrc.arizona.edu/events/brown-bag/brown-bag-
600
seminar-arizona-value-integrated-food-energy-water-arvin-few
601
(11)
Berardy, A.; Chester, M. V. Climate Change Vulnerability in the Food, Energy, and Water
602
Nexus: Concerns for Agricultural Production in Arizona and Its Urban Export Supply.
603
Environ. Res. Lett. 2017, 12 (3) DOI: 10.1042/BJ20101136>.
604
(12)
Eisenman, D. P.; Wilhalme, H.; Tseng, C.; Chester, M.; English, P.; Pincetl, S.; Fraser, A.;
605
Vangala, S.; Dhaliwal, S. K. Health & Place Heat Death Associations with the Built
606
Environment , Social Vulnerability and Their Interactions with Rising Temperature.
607
Health Place. 2016, 41, 89–99 DOI: 10.1016/j.healthplace.2016.08.007.
608
(13) Rubin, S. A Call for Reliability Standards. J. / Am. Water Work. Assoc. 2011, 103 (1), 22–
609 610
24. (14)
611 612
Water Work. Assoc. 2011, 37. (15)
613 614
AWWA. Buried No Longer: Confronting America’s Water Infrastructure Challenge. Am.
Mays, L. W. Water Distribution System Handbook; Mays, L. W., Ed.; McGraw-Hill Professional Publishing, 2000.
(16)
Park, J.; Seager, T. P.; Rao, P. S. C.; Convertino, M.; Linkov, I. Integrating Risk and
615
Resilience Approaches to Catastrophe Management in Engineering Systems. Risk Anal.
616
2013, 33 (3), 356–367 DOI: 10.1111/j.1539-6924.2012.01885.x.
617
(17)
618 619 620
Banerjee, A.; Tiwari, A.; Vico, J.; Wester, C. Motor Protection Principles. 2008 61st Annu. Conf. Prot. Relay Eng. 2008, 215–231 DOI: 10.1109/CPRE.2008.4515057.
(18)
Cowern, E. The Hot Issue of Motor Temperature Ratings. EC M Electr. Constr. Maint. 2000, 99 (9), 104–106. 22 ACS Paragon Plus Environment
Page 23 of 27
621
Environmental Science & Technology
(19)
Csanyi, E. Heat as One of the Most Common Cause of Motor Failure Service Life.
622
Electrical Engineering Portal [Online]. 2015, pp 1–4. http://www.idc-
623
online.com/technical_references/pdfs/electrical_engineering/Heat_as_one_of_the_most_c
624
ommon_cause_of_motor_failure.pdf
625
(20)
Basics for Practical Operation Motor Starting. Rockwell Automation 1998, p 46.
626
(21)
Thermal Management: Heat Dissipation in Electrical Enclosures; Hoffman, 2017.
627
(22)
Kreeley, B.; Coulton, S. Increasing the Lifespan and Reliability of Electrical
628 629
Components; Kooltronic, Inc. pp 1–5, 2016. (23)
630 631
Whittle, A.; Stahmer, M. Temperature Derating of PVC Pipes for Pressure Applications; Plastics Industry Pipe Association of Australia Limited, 2005.
(24)
Volk, C.; Dundore, E.; Schiermann, J.; Lechevallier, M. Practical Evaluation of Iron
632
Corrosion Control in a Drinking Water Distribution System. Water Res. 2000, 34 (6),
633
1967–1974 DOI: 10.1016/S0043-1354(99)00342-5.
634
(25)
Hua, F.; West, J. R.; Barker, R. A.; Forster, C. F. Modelling of Chlorine Decay in
635
Municipal Water Supplies. Water Res. 1999, 33 (12), 2735–2746 DOI: 10.1016/S0043-
636
1354(98)00519-3.
637
(26)
Toroz, I.; Uyak, V. Seasonal Variations of Trihalomethanes ( THMs ) in Water
638
Distribution Networks of Istanbul City. Elsevier Desalination. 2005, 176, 127–141 DOI:
639
10.1016/j.desal.2004.11.008.
640
(27)
Graham, N. J. D.; Collins, C. D.; Nieuwenhuijsen, M.; Templeton, M. R. The Formation
641
and Occurrence of Haloacetic Acids in Drinking Water; Imperial College London,
642
London UK, 2009.
643
(28)
Hu, Y.; Wang, D.; Cossitt, K.; Chowdhury, R. AC Pipe in North America : Inventory ,
644
Breakage , and Working Environments. J. Pipeline Syst. Eng. Pract. 2010, 1 (November),
645
156–172 DOI: 10.1061/(ASCE)PS. l949-1204.0000064.
646
(29)
647 648
1994, 120 (2), 312–321. (30)
649 650
Habibian, A. Effect of Temperature Changes on Water-Main Breaks. J. Transp. Eng.,
Rajani, B.; Zhan, C.; Kuraoka, S. Pipe-Soil Interaction Analysis of Jointed Water Mains. Can. Geotech. J. 1996, 33, 393–404.
(31)
Operator’s Companion, 10th ed.; USA Blue Book, 2013.
23 ACS Paragon Plus Environment
Environmental Science & Technology
651
(32)
Pintar, K. D. M.; Slawson, R. M. Effect of Temperature and Disinfection Strategies on
652
Ammonia-Oxidizing Bacteria in a Bench-Scale Drinking Water Distribution System.
653
Water Res. 2003, 37 (8), 1805–1817 DOI: 10.1016/S0043-1354(02)00538-9.
654
(33)
655
Page 24 of 27
Wild, H. E.; Sawyer, C. N.; McMahon, T. C. Factors Affecting Nitrification Kinetics. Water Pollut. Control Fed. 1971, 43, 10.
656
(34)
Nitrification; American Water Works Association, 2002.
657
(35)
Lipponen, M. T. T.; Suutari, M. H.; Martikainen, P. J. Occurrence of Nitrifying Bacteria
658
and Nitrification in Finnish Drinking Water Distribution Systems. Water Res. 2002, 36
659
(17), 4319–4329 DOI: 10.1016/S0043-1354(02)00169-0.
660
(36)
Torvinen, E.; Suomalainen, S.; Lehtola, M. J.; Ilkka, T.; Zacheus, O.; Paulin, L.; Katila,
661
M.; Martikainen, P. J.; Miettinen, I. T. Mycobacteria in Water and Loose Deposits of
662
Drinking Water Distribution Systems in Finland Mycobacteria in Water and Loose
663
Deposits of Drinking Water Distribution Systems in Finland. Appl. Environ. Microbiol.
664
2004, 70 (4), 1973–1981 DOI: 10.1128/AEM.70.4.1973.
665
(37)
Archuleta, R.; Mullens, P.; Primm, T. P. The Relationship of Temperature to Desiccation
666
and Starvation Tolerance of the Mycobacterium Avium Complex. Arch. Microbiol. 2002,
667
178 (4), 311–314 DOI: 10.1007/s00203-002-0455-x.
668
(38)
Wadowsky, R. M.; Wolford, R.; McNamara, A. N.; Yee, R. B. Effect of Temperature Ph
669
and Oxygen Level on the Multiplication of Naturally Occurring Legionella-Pneumophila
670
in Potable Water. Appl. Environ. Microbiol. 1985, 49 (5), 1197–1205.
671
(39)
Rogers, J.; Dowsett, A. B.; Dennis, P. J.; Lee, J. V; Keevil ’, A. C. W. Influence of
672
Temperature and Plumbing Material Selection on Biofilm Formation and Growth of
673
Legionella Pneumophila in a Model Potable Water System Containing Complex
674
Microbial Flora. Appl. Environ. Microbiol. 1994, 60 (5), 1585–1592.
675
(40)
Ralph, L.; Heyn, K.; Brooks, K.; Adams, A. What You Need to Know About Climate
676
Risks to Water Utility Infrastructure and Assets. In AWWA Webinar Program. November
677
30, 2016.
678
(41)
Chowdhury, S.; Champagne, P.; McLellan, P. J. Models for Predicting Disinfection
679
Byproduct (DBP) Formation in Drinking Waters: A Chronological Review. Sci. Total
680
Environ. 2009, 407 (14), 4189–4206 DOI: 10.1016/j.scitotenv.2009.04.006. 24 ACS Paragon Plus Environment
Page 25 of 27
681
Environmental Science & Technology
(42)
Christensen, F. M.; Andersen, O.; Duijm, N. J.; Harremoës, P. Risk Terminology - A
682
Platform for Common Understanding and Better Communication. J. Hazard. Mater.
683
2003, 103 (3), 181–203 DOI: 10.1016/S0304-3894(03)00039-6.
684 685
(43)
686 687
https://www.timeanddate.com/weather/usa/phoenix/historic (accessed Jan 1, 2017). (44)
688 689
US Census Bureau Population and Housing Unit Estimates; https://www.census.gov/programs-surveys/popest.html
(45)
690 691
Past Weather in Phoenix, Arizona, USA – Yesterday and Las 2 Weeks;
US Climate Data- Phoenix; http://www.usclimatedata.com/climate/phoenix/arizona/united-states/usaz0166
(46)
692
City of Phoenix Water Services; https://www.phoenix.gov/waterservices/waterquality
693
(47) Operating and Capital Budget Plan: Fiscal Year 2015 - 2016; Las Vegas, NV, 2015.
694
(48)
Sakaue, K., Kamata, M., Iwamoto, S., & Nimiya, H. The prediction method of water
695
temperature in distribution pipes. In CIB W062 26th International Symposium on Water
696
Supply and Drainage for Buildings, CIB W62 International Symposium, Rio de Janeiro,
697
Brazil, Sept. 18-20 2000, in-house publishing: Rotterdam, Netherlands, 2000; Session
698
B4.
699
(49)
Blokker, E. J. M.; Pieterse-Quirijns, E. J. Modeling Temperature in the Drinking Water
700
Distribution System. J. Am. Water Work. Assoc. 2013, 105 (1), E19–E28 DOI:
701
10.5942/jawwa.2013.105.0011.
702
(50)
703 704
Randall-Smith, M. Guidance Manual for the Structural Condition Assessments of Trunk Mains; WRC, 1992.
(51)
705
Palisade Website; http://www.palisade.com/risk/
706
(52)
Tobias, P. A. Applied Reliability, 3rd ed.; CRC Press LLC, 1986.
707
(53)
Henley, E. J.; Kumamoto, H. Reliability Engineering and Risk Assessment, Vol. 193.;
708 709 710
Prentice-Hall: Englewood Cliffs, NJ, 1981. (54)
National Primary Drinking Water Regulations; United States Environmental Protection Agency, 2013. 141–142 DOI: EPA 816-F-09-004.
25 ACS Paragon Plus Environment
Environmental Science & Technology
711
(55)
712 713
Page 26 of 27
National Primary Drinking Water Regulations: Filtration and Disinfection; United States Environmental Protection Agency, 2013.
(56)
714
Water Transmission and Distribution for Water Treatment Plant Operators; Florida Rural Water Association: Florida, US, 2005.
715
(57)
2013 Annual Water Quality Report; City of Phoenix, 2013.
716
(58)
2014 Water Quality Report; City of Phoenix, 2014.
717
(59)
2015 Water Quality Report; City of Phoenix, 2015.
718
(60)
2012 Annual Water Quality Report; City of Phoenix, 2012.
719
(61)
2011 Annual Water Quality Report; City of Phoenix, 2011.
720
(62)
Operating and Capital Budget 2017; Las Vegas Valley Water Authority, 2016.
721
(63)
US Climate Data – Las Vegas;
722
http://www.usclimatedata.com/climate/las-vegas/nevada/united-states/usnv0049
723
(64)
2016 Water Quality Report; Las Vegas, NV, 2016.
724
(65)
The United Nations World Water Development Report 2015: Water for a Sustainable
725
World; The United Nations. Paris, France, 2015.
726
(66)
Kim, Y. H. Coagulants and Flocculants: Theory and Practice; Tall Oaks Pub., 1995.
727
(67)
Climate Change and Transportation; Transportation Research Board, 2012.
728
(68)
Allen, M.; Preis, A.; Iobal, M.; Srirangarajan, S.; Lim, H.; Girod, L.; Whittle, A.; Whittle,
729
A.; Girod, L.; Preis, A.; Allen, M.; Lim, H.; Iqbal, M.; Srirangarajan, S.; Fu, C.; Wong,
730
K.; Goldsmith, D. Managing the Water Distribution Network with a Smart Water Grid.
731
Smart Water, 2016, 1 (1), 4 DOI: 10.1186/s40713-016-0004-4.
26 ACS Paragon Plus Environment
Page 27 of 27
Environmental Science & Technology
47x34mm (300 x 300 DPI)
ACS Paragon Plus Environment