assessing the hydraulic transient performance of water and wastewater systems · pdf...
TRANSCRIPT
ASSESSING THE HYDRAULIC TRANSIENT PERFORMANCE OF WATER AND
WASTEWATER SYSTEMS USING FIELD AND NUMERICAL MODELING DATA
by
Djordje Radulj
A thesis submitted in conformity with the requirements for the degree of Master of Applied Science
Graduate Department of Civil Engineering University of Toronto
© Copyright by Djordje Radulj 2010
ii
Assessing the Hydraulic Transient Performance of Water and
Wastewater Systems Using Field and Numerical Modeling Data
Djordje Radulj
Master of Applied Science
Graduate Department of Civil Engineering
University of Toronto
2010
Abstract
A large proportion of water and wastewater systems have traditionally been analyzed and designed
without the consideration of the nature, risk, and potential consequence of hydraulic transients.
Recent advancements in numerical hydraulic modeling have spawned a specialty hydraulic field
based on numerical transient analysis. The current practice within this field often lacks physical
understanding and can be misguided by both the current knowledge, technology based limitations,
and by the sole reliance on numerical models. This thesis aims to provide insights into some of the
shortcomings of current practice and to develop the importance and application of field data based
confirmations. The thesis examines the advances in the current field oriented technology for
recording transient pressures, and provides examples and insights on how this data can be used both
in conjunction with numerical modeling and on its own as a first step to a proposed frequency based
transient risk assessment methodology. The thesis establishes definitions and a preliminary
methodology for a Transient Risk Index.
iii
Acknowledgments
I have always appreciated a good and direct summary, and therefore please accept the following
form of acknowledgement to the individuals that have contributed to the completion of this work:
• Prof. Bryan Karney – for the opportunity, trust, time, technical guidance, and overall
flexibility.
• Prof. Barry Adams – for the quick, efficient and constructive review, and the initial pre-
thesis thought process inspiration.
• Direct and Indirect Contributors and Reviewers: Bryon Singh, George Illaszewicz,
Andrew O’Connor, Arash Alkozai, and Ahmad Malekpour – for all your contributions
(large and small) and overall support that made this a lot easier and more efficient.
• Fabian Papa – for the support, resources, and overall guidance and direction.
• My Family: Petar, Vesna and Sonja Radulj (and others…) – for your support behind the
scenes that made all of this significantly easier and obviously…possible.
• Brittany Dupak – for the constant reminders to just get it done; and for everything else in
life and this world.
Thank you all!
iv
Table of Contents
Abstract .............................................................................................................................................................. ii
Acknowledgments .......................................................................................................................................... iii
Table of Contents ........................................................................................................................................... iv
List of Tables .................................................................................................................................................. vii
List of Figures ............................................................................................................................................... viii
Abbreviations .................................................................................................................................................... x
Chapter 1 Introduction ............................................................................................................................. 1
1.1 Background and Context ..................................................................................................................... 1
1.2 Significance of Topic ........................................................................................................................... 1
1.3 Organization .......................................................................................................................................... 2
Chapter 2 Overview of Hydraulic Transients ....................................................................................... 4
2.1 Definition and Overview ..................................................................................................................... 4
2.2 Type of Systems .................................................................................................................................... 6
2.2.1 Water Supply and Transmission Systems ............................................................................ 6
2.2.2 Water Distribution Networks ............................................................................................... 8
2.2.3 Wastewater Forcemain Systems.......................................................................................... 10
2.3 Risks Due to Transients .................................................................................................................... 10
2.3.1 Background ............................................................................................................................ 10
2.3.2 Traditional Risk Analysis ..................................................................................................... 12
2.4 Transient Protection Options and/or Equipment ........................................................................ 14
2.5 Hydraulic Transient Analysis Approaches ...................................................................................... 22
2.5.1 Do-Nothing ........................................................................................................................... 22
2.5.2 Rule-of-Thumb and Standard Convention ....................................................................... 23
2.5.3 Numerical Modeling ............................................................................................................. 25
2.5.4 Field Data and Investigation ............................................................................................... 28
2.5.5 Hybrid ..................................................................................................................................... 29
2.6 Summary .............................................................................................................................................. 30
Chapter 3 Current Analysis Shortcomings .......................................................................................... 31
3.1 Steady State Models ............................................................................................................................ 31
3.2 Average and Maximum Day Demands ........................................................................................... 34
3.3 Current Design Approach ................................................................................................................. 38
v
3.3.1 Steady State Design Loads ................................................................................................... 39
3.3.2 Transient Modeling Approach ............................................................................................ 40
3.3.3 Wastewater Design Considerations .................................................................................... 44
3.4 Lack of Confirmation and Monitoring ............................................................................................ 47
3.4.1 Confirmation, Validation and Calibration ......................................................................... 48
3.4.2 Performance Monitoring ..................................................................................................... 50
3.5 Summary .............................................................................................................................................. 52
Chapter 4 Water Quality Issues ............................................................................................................. 53
4.1 Background ......................................................................................................................................... 53
4.2 Important Pathogens ......................................................................................................................... 55
4.3 Case Study and Discussion ............................................................................................................... 56
4.4 Summary .............................................................................................................................................. 59
Chapter 5 Advances in Field Work ...................................................................................................... 62
5.1 Traditional Field Work Approach .................................................................................................... 62
5.2 Modern Pressure Monitors and Comparison ................................................................................. 63
5.3 Pipetech TP-1 Transient Pressure Monitor .................................................................................... 72
5.4 Sample Data ........................................................................................................................................ 79
5.4.1 Long-Term Histories ............................................................................................................ 80
5.4.2 Individual Events .................................................................................................................. 87
5.4.3 Unique Events ....................................................................................................................... 92
5.5 Sample Statistics .................................................................................................................................. 96
5.5.1 Event Summaries .................................................................................................................. 96
5.5.2 General Event Statistics .................................................................................................... 100
5.6 Summary ........................................................................................................................................... 105
Chapter 6 Validation of Numerical Models ..................................................................................... 106
6.1 Case Study I: Region of Peel .......................................................................................................... 106
6.1.1 Background ......................................................................................................................... 106
6.1.2 Analysis ................................................................................................................................ 107
6.1.3 Discussion ........................................................................................................................... 113
6.2 Case Study II: Region of Durham ................................................................................................. 115
6.2.1 Background ......................................................................................................................... 115
6.2.2 Analysis ................................................................................................................................ 116
vi
6.3 Case Study III: Mexico City ........................................................................................................... 120
6.3.1 Background ......................................................................................................................... 120
6.3.2 Analysis ................................................................................................................................ 120
6.4 Case Study IV: Wastewater Forcemain ........................................................................................ 123
6.4.1 Background ......................................................................................................................... 123
6.4.2 Analysis ................................................................................................................................ 124
6.5 Summary ........................................................................................................................................... 128
Chapter 7 Transient Risk Index ......................................................................................................... 129
7.1 Purpose and Background ............................................................................................................... 129
7.1.1 The Nature of System Failure .......................................................................................... 129
7.1.2 Pressure Class Design for Isolated Transient Events ................................................... 130
7.1.3 Pressure Class Design for Cyclic Loading ...................................................................... 131
7.1.4 Importance of Risk Assessment ...................................................................................... 132
7.2 Requirements, Parameters, and Definitions ................................................................................ 133
7.2.1 System and External Influences ....................................................................................... 133
7.2.2 Single Transient Event Parameters ................................................................................. 135
7.2.3 Multiple Transient Event Parameters ............................................................................. 137
7.3 Methodology .................................................................................................................................... 140
7.4 Sample Results ................................................................................................................................. 143
7.5 Discussion ........................................................................................................................................ 144
7.6 Summary ........................................................................................................................................... 147
Chapter 8 Summary and Conclusions ............................................................................................... 148
References .................................................................................................................................................... 151
vii
List of Tables
Table 4-1: Air Valve Chamber Water Sampling Test Results ................................................................... 58
Table 5-1: Pressure Monitor Comparison Summary ................................................................................. 64
Table 5-2: Pressure Monitor Comparison Summary (cont’d) .................................................................. 65
Table 5-3: Transient Pressure Monitor Ranking for Long-Term Applications ..................................... 71
Table 5-4: Sample TP-1 Pressure Record Settings ..................................................................................... 77
Table 5-5: Unplanned Transient Event Summary – La Caldera PS (Mexico City) ............................... 98
Table 5-6: Transient Event Pressure Summary – Tlahuac/Mixquic Well Systems (Mexico City) ...... 99
Table 5-7: Detailed Event Summary – Region of Peel System .............................................................. 100
Table 5-8: General Short-Term Statistics – Tlahuac & Mixquic Wells (Mexico City) ........................ 101
Table 5-9: General Long-Term Statistics – La Caldera PS (Mexico City) ............................................ 101
Table 5-10: General Long-Term Statistics – Region of Peel System .................................................... 102
Table 5-11: Detailed Long-Term Statistics – Lakeview WTP (Region of Peel) .................................. 103
Table 7-1: Sample Pipe Pressure Class Properties (Mielke, 2004) ......................................................... 131
Table 7-2: Fatigue Load Factors for PE Pipe (PIPA, 2002) ................................................................... 132
Table 7-3: Single Transient Event Parameter Definitions ...................................................................... 136
viii
List of Figures
Figure 2-1: Sample Water Transmission System (Mexico City) .................................................................. 7
Figure 2-2: Sample Water Distribution Network (Region of York) ........................................................... 9
Figure 2-3: Traditional Transient Analysis Solution Approaches ............................................................ 27
Figure 3-1: Wave Velocity v. Air Content (Wylie and Streeter, 1993) ..................................................... 46
Figure 3-2: Sample Vapour Pressure Sensitivity Analysis for a Wastewater System ............................. 47
Figure 4-1: Air Valve Chamber Water Sampling Locations ..................................................................... 57
Figure 4-2: Air Valve Chamber Water Sampling Test Result Comparison ............................................ 58
Figure 4-3: Comparison of Total Coliform Concentrations ..................................................................... 60
Figure 4-4: Comparison of E. coli Concentrations .................................................................................... 60
Figure 4-5: Comparison of Enterococci Concentrations .......................................................................... 61
Figure 4-6: Comparison of Clostridium Perfringens Concentrations ..................................................... 61
Figure 5-1: TP-1 Transient Pressure Monitor Components ..................................................................... 73
Figure 5-2: Typical TP-1 Installation in the Field ...................................................................................... 73
Figure 5-3: Sample Calibration Pressure Profile ......................................................................................... 75
Figure 5-4: TP-1 Pressure Monitoring Equipment Settings ..................................................................... 76
Figure 5-5: Pressure v. Time Plots for Different TP-1 Settings ............................................................... 78
Figure 5-6: Pressure v. Time Plots for the Same Event but with Different Time Scales ..................... 79
Figure 5-7: Sample Transient Pressure History – Hanlan PS (Region of Peel) ..................................... 81
Figure 5-8: Sample Transient Pressure History – Lakeview WTP (Region of Peel) ............................. 82
Figure 5-9: Sample Transient Pressure History – Glen Cairn PS (City of Ottawa) .............................. 82
Figure 5-10: Sample Transient Pressure History – La Caldera PS (Mexico City) – 2 Months ............ 83
Figure 5-11: Sample Transient Pressure History – La Caldera PS (Mexico City) – 2 Days ................. 84
Figure 5-12: Sample Transient Pressure History – Mixquic Well No. 13 (Mexico City) ...................... 84
Figure 5-13: Sample Transient Pressure History – Ajax WSP (Region of Durham) – 3 Months ....... 86
Figure 5-14: Sample Transient Pressure History – Ajax WSP (Region of Durham) – 3 Days ............ 86
Figure 5-15: Typical Pump Start-Up – Ajax WSP (Region of Durham)................................................. 88
Figure 5-16: Typical Pump Shut-Off – Ajax WSP (Region of Durham) ................................................ 88
Figure 5-17: Typical Pump Shut-Off at High Point – La Caldera Pipeline (Mexico City) ................... 89
Figure 5-18: Typical Pump Switch – Harwood PS (Region of Durham) ............................................... 89
Figure 5-19: Power Failure Event without Upsurge – Ajax WSP (Region of Durham) ...................... 91
Figure 5-20: Power Failure Event with Upsurge – Hanlan PS (Region of Peel) ................................... 91
ix
Figure 5-21: Multi Pump Shut-Off, From 2 to 1 to 0 Pumps – La Caldera PS (Mexico City) ............ 92
Figure 5-22: Complete System Shutdown – La Caldera PS (Mexico City) ............................................. 93
Figure 5-23: Single Well Shutdown – Mixquic Well No. 4 (Mexico City) .............................................. 93
Figure 5-24: Progressive Check Valve Failure – La Caldera PS (Mexico City) ...................................... 95
Figure 5-25: Distribution System PRV Adjustment – North Richmond Hill PS (Region of York) .. 95
Figure 5-26: Pump Control Logic Change – Montreal Road PS (City of Ottawa) ............................... 96
Figure 5-27: Sample Cumulative Pressure Distribution – Lakeview WTP (Region of Peel) ............. 104
Figure 6-1: Model v. Field Event Pressure Validation – Streetsville PS (Region of Peel) .................. 110
Figure 6-2: Model v. Field Event Pressure Validation – MV900 Chamber 8 (Region of Peel) ........ 110
Figure 6-3: Model v. Field Event Pressure Validation – Hanlan PS (Region of Peel) ........................ 111
Figure 6-4: Model v. Field Event Pressure Validation – BS1500 Chamber 8 (Region of Peel) ........ 111
Figure 6-5: Model v. Field Event Pressure Validation for 2 Pumps – Hanlan PS (Region of Peel) . 112
Figure 6-6: Model v. Field Event Pressure Validation for 2 Pumps – BS1500 (Region of Peel) ...... 112
Figure 6-7: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 1 .. 117
Figure 6-8: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 2 .. 119
Figure 6-9: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 3 .. 119
Figure 6-10: Model v. Field Comparison (2 Pump to 1 Pump) – La Caldera PS (Mexico City) ....... 122
Figure 6-11: Model v. Field Comparison (1 Pump to 0 Pumps) – La Caldera PS (Mexico City) ..... 122
Figure 6-12: Model v. Field Comparison (2 Pumps to 0 Pumps) – La Caldera PS (Mexico City).... 123
Figure 6-13: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 1 ..................... 125
Figure 6-14: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 2 ..................... 126
Figure 6-15: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 3 ..................... 127
Figure 6-16: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 4 ..................... 128
Figure 7-1: Single Transient Event Properties & Parameters ................................................................. 135
Figure 7-2: Multiple Transient Event Properties & Parameters ............................................................. 138
Figure 7-3: Transient Risk Index Schematic ............................................................................................. 141
Figure 7-4: Sample Preliminary TRI Graph for the Lakeview Zone 1 and 2 Systems ....................... 144
x
Abbreviations
AC Air Chamber
ADD Average Day Demand
ARV Air Release Valve
ANSI American National Standards Institute
ASCE American Society of Civil Engineers
ASME American Society of Mechanical Engineers
ASTM American Society for Testing and Materials
AV Air Valve
AWWA American Water Works Association
AWWARF American Water Works Association Research Foundation
BPS Booster Pump Station
CAV Combination Air Valve
CPP Concrete Pressure Pipe
DI Ductile Iron Pipe
DWF Dry Weather Flow
EPS Extended Period Simulation
FSI Fluid Structure Interaction
GIS Geographic Information System
GMT Greenwich Mean Time
GPS Global Positioning System
HAC Hydro-Pneumatic Air Chamber
HB Horizontal Bend
HDPE High Density Polyethylene Pipe
HGL Hydraulic Grade Line
HLPS High Lift Pump Station
HP High Point
I&I Inflow and Infiltration
LAN Local Area Network
LLPS Low Lift Pump Station
LP Low Point
MDD Maximum Day Demand
xi
MHD Minimum Hour Demand
MLD Mega Litres Per Day
MOC Method of Characteristics
MRI Moment of Rotational Inertia
O&M Operations and Maintenance
OWWA Ontario Water Works Association
PDA Personal Digital Assistant
PHD Peak Hour Demand
PVC Polyvinyl Chloride Pipe
PS Pump Station
Pv Full Vacuum Pressure Limit
Pw Working Pressure
RPM Revolutions per Minute
SAV Surge Anticipation Valve
SCADA Supervisory Control and Data Acquisition
SD Standard Deviation
SPS Sewage Pumping Station
SRV Surge Relief Valve
TDH Total Dynamic Head
TP-1 Transient Pressure Monitor Unit
TransAM Transient Analysis Model
TRI Transient Risk Index
USB Universal Serial Bus
UPS Uninterruptible Power Supply
VFD Variable Frequency Drive
VB Vertical Bend
Vp Vapour Pressure
WCM Wave Characteristics Method
WDS Water Distribution System
WTP Water Treatment Plant
WSP Water Supply Plant
WWF Wet Weather Flow
WWTP Wastewater Treatment Plant
1
Chapter 1 Introduction
1.1 Background and Context
The title of this thesis, “Assessing the Hydraulic Transient Performance of Water and Wastewater
Systems Using Field and Numerical Modeling Data” is rather long and appears to be
comprehensive. The reason for this is primarily the fact that the document covers a wide range of
issues pertaining to hydraulic transients in water and wastewater systems. Nonetheless, the core and
unifying theme of this work is the benefit of, and the need for, an increase in hydraulic transient
based field work. The goal of such field work is to fill-in the gaps that currently exist in numerical
modeling, and to subsequently and concurrently improve the long-term performance of water and
wastewater systems.
The content of this work derives from a strong academic interest and influence, but also from direct
industry experience. The goal of the thesis is to demonstrate the current shortcomings in the
industry’s approach to addressing transients, and to blend the current (and often practical) needs of
industry with recent technological advances and academic research. This thesis aims to provide
context for these issues, as well as to establish a road map for several key and important topics
within the realm of numerical and field hydraulic transient data. This thesis does not aim to provide
an extensive discussion and an answer to all of the individual topics, but rather aims to show how
each topic plays an integral role in the analysis process and the assessment of overall system
performance.
1.2 Significance of Topic
The topic of hydraulic transients falls within the field of hydraulics and water resources, which in
turn falls within the more general field of environmental engineering. From this hierarchal point of
view, environmental engineering explicitly focuses on improving the welfare of society in the areas
pertaining to the environment and the planet. Water and wastewater are essential engineering
components of the present society, and the branch of hydraulics is an important field that addresses
the physical challenges of engineering such systems.
2
While technically and academically challenging, hydraulic transients comprise a small proportion of
all hydraulic based studies. This is mostly due to their rare and relatively minimal impact (in
comparison to other tasks) on the overall analysis and design time and cost of an engineering
project. However, from the point of view of a water and wastewater system’s performance,
efficiency and safety, and from the point of view of the potential cost implications to both a system
owner and society, hydraulic transients are extremely important and often overlooked.
The task of performing a hydraulic transient analysis as part of all hydraulic system designs has only
recently been widely accepted, but with this acceptance also comes an increase in physical
misunderstanding, malpractice and carelessness. The sole reliance by engineers on hydraulic models
built by others has proven to be dangerous, and the lack of attention to both the details of analysis
as well as to the real life (i.e., in-situ) systems is troublesome. The focus of this thesis is to address
many of these troubling issues and to provide context on why they are important and on how they
can be improved. To that end, the overall importance of the topic should be relatively clear.
1.3 Organization
As noted earlier, this thesis is a collection of important and relatively specialized topics on hydraulic
transients; topics that focus on assessing the hydraulic transient performance of water and
wastewater systems. Each of the chapters in this document aims to provide a practical discussion of
the relevant topic, such that the advances in technology and academia can actually be readily
implemented in the industry.
Chapter 2 overviews hydraulic transients and hydraulic transient analysis and then moves towards a
brief discussion of the different types of hydraulic systems. The chapter provides context on this
topic, by considering the risks, protection options and analysis approaches.
Chapter 3 provides a critical assessment of current practice, through the focus on the shortcomings
of numerical models, assumptions, and analysis/design approaches. The chapter reaffirms the
importance of field observations.
Chapter 4 takes a brief aside from assessing hydraulic transients, and continues along the path of
field investigation to bring light to the water quality concerns that are associated with negative
3
pressures. This chapter summarizes a field study regarding the pathogen intrusion risks in air valve
chambers during negative transient events.
Chapter 5 directs its focus on the core of the thesis topic – the advances and uses of transient
pressure monitoring technology. The chapter ultimately summarizes the transient pressure data and
statistics acquired through several case studies.
Chapter 6 proceeds to expand on the previous chapter by introducing and comparing numerical
modeling results to the actual field results, thereby leading to the idea and practice of model
validation and/or calibration. The chapter relies on several case studies that bring light to some of
the main issues in the validation of numerical transient models.
Chapter 7 develops the methodology for a preliminary transient risk index; an index that can be used
to assess the hydraulic transient risk of a water (or wastewater) system using continuous transient
pressure monitoring. It provides the necessary transient pressure recording logic and transient event
definitions, as well as the type of improvements required to formally implement such an approach.
Chapter 8 summarizes the conclusions from the thesis and the research to-date, and reaffirms the
contributions of this work towards both research and industry practice.
4
Chapter 2 Overview of Hydraulic Transients
2.1 Definition and Overview
Hydraulic transients have traditionally been referred to as water hammer and in some cases as
pressure or hydraulic surges. Presently, these non-steady pressure events are more often than not
referred to as hydraulic transients, and the act of performing a representative analysis (and design) is
therefore referred to as a hydraulic transient analysis, or in a simpler form, a transient analysis. Such
an analysis is typically performed in order to determine the overall risk due to transients and to
provide recommendations on how this overall risk can be minimized and/or possibly eliminated.
A hydraulic transient analysis can be performed on any fluid system, but it is the fluid and the type
of system that determines the appropriate analysis method. Some examples of fluid systems that
often require a transient analysis include:
• Water supply and transmission systems;
• Water distribution networks;
• Pressurized wastewater systems (e.g., forcemains);
• Wastewater effluent systems;
• Combined (surcharged) sewer systems;
• Oil and gas pipelines;
• Hydroelectric plants;
• Open channel flow systems (e.g., rivers and streams);
• Jet fuel distribution systems;
• Steam, heating and cooling pipelines;
• Well and groundwater systems.
This document focuses on the impact and application of hydraulic transients in water and
wastewater systems, and in particular, to large scale municipal systems. According to Chaudhry
(1987) and to McInnis and et al. (2004), a non trivial physical distinction can be made between three
types of water and wastewater systems: closed conduit or pressure pipe systems, open channel
5
systems with a free surface, and combined systems with regions of free surface and pressurized flow
(i.e., mixed systems). While the latter two are indeed fascinating and challenging topics, this
document focuses on closed conduit pressurized water and wastewater systems because of their
widespread existence and unique physical challenges/attributes.
In the context of closed pressurized systems, hydraulic transients are the means by which pressures
and flows are adjusted. In basic terms, transients mark transitions between steady state conditions,
and such transitions are the source of the special attention. The transition from a steady state
condition (i.e., a condition in which the key parameters do not change significantly over time) can be
smooth or rough, predictable or unexpected, and most importantly, it can subject a system to a high
level of stress and an increased risk of failure. In fluid systems, these changing parameters (or
variables) are mostly flow and pressure. As a general principle, the rate that a fluid enters a pipe
segment must be equal to the rate that it leaves this same segment, or else an accumulation of fluid
would result. Furthermore, steady state condition is achieved by the equilibrium of three key
physical quantities: force, momentum, and energy.
During a transient event the fluid inflow or outflow in some part of the system is disturbed via a
change in, or addition of a boundary condition, and this leads to a local imbalance that allows mass
to accumulate or deplete. Such an imbalance in pressurized flow can only be accommodated
through two components: fluid compressibility and pipe extension. However, since fluids like water
are not easily compressed, and since pipelines do not easily expand, this imbalance typically results in
large pressure forces. The pressure forces are translated into pressure waves, and these pressure
waves propagate through the system. The rate of pressure wave propagation is rather quick,
typically between about 250 and 1250 m/s. This fluid and system property is often termed
wavespeed or wave celerity, and its exact value depends on the compressibility of the fluid and the
elastic properties of the pipe. The resulting transient pressures, which can be both positive
(upsurge) or negative (downsurge) are then superimposed on the existing steady state system
pressures. If the system is not capable of handling these pressure magnitudes or changes, then it is
subjected to a higher risk of failure. Eventually, the energy (strain and kinetic) resulting from the
pressure waves dissipates and decays due to the friction in the system, and ultimately establishes a
new steady state condition.
6
2.2 Type of Systems
The following subsections further define and distinguish three types of important closed pressure
systems that are often the subject of a transient analysis. The systems types are:
1. Water Supply and Transmission Systems;
2. Water Distribution Networks; and
3. Wastewater Forcemain Systems.
The goal of these subsections is to briefly illustrate the hydraulic differences between these systems,
and to bring attention to important considerations for a hydraulic transient analysis and design.
2.2.1 Water Supply and Transmission Systems
Water supply and transmission systems have traditionally been a key area of focus for hydraulic
engineers, through the typical tasks of investigation, analysis and design. Most great urban areas
were developed in proximity to a water source such as a lake, river, groundwater aquifer, and/or
ocean. However, due to constraints such as unfavourable topology, distance, etc., the water supply
sources and the consumers often required a strong engineering link; a link achieved through
dedicated water transmission systems. Water transmission systems have become even more
prevalent and important in the modern era, as existing and new communities search for additional
water sources.
Water transmission systems generally comprise long and large conduits that transfer raw or treated
water from the source location, to another location closer to the population or urban area. In such a
case, the source location may be an actual water source, or an intermediate reservoir, water
treatment plant (WTP), etc. Similarly, the ultimate discharge point can be a reservoir, WTP,
distribution system, pump station, etc. The transmission conduits have historically been dedicated
conveyance systems, and therefore without many connections or interconnections. These
conveyance systems can include: pumped pipeline systems, pressurized gravity systems, or free
surface gravity conveyance systems. These large diameter conduits are typically referred to as
transmission mains, feedermains, or pipelines, rather than watermains – a term which is typically
reserved for small diameter water pipes only.
7
Water supply and transmission systems are designed to overcome any elevation differences (i.e.,
static head) and head losses (e.g., due to appurtenances and friction). In such systems, the head loss
due to the friction of the pipe often governs the hydraulics. The limiting design of a dedicated
transmission system is typically the maximum velocity of the fluid (e.g., about 2 m/s), and this
limitation therefore often leads to a significant size requirement. As an example, the regional area of
greater Mexico City, which boasts a population of nearly 20 million, partially relies on a water
transmission system by the name of the Cutzamala Aqueduct. This system, which is pictured in
Figure 2-1, comprises two parallel 2500 mm gravity pipelines, and conveys water over a distance of
150 km from the Cutzamala River Basin to Mexico City. This large scale water transmission system
is operated within a flow range of 10 to 20 m3/s – or in other words, a large quantity of water.
Figure 2-1: Sample Water Transmission System (Mexico City)
The nature and risk of hydraulic transients in water transmission systems is system specific, but the
general characteristics depend on whether the water is pumped or delivered via gravity. In general,
pumped systems are subject to the risk of pump station based flow adjustments, including pump
trips, power failures, valve operations, etc. Gravity systems are mostly at risk from poor valve
operational protocols and control. The key distinction between a pure or dedicated water
transmission system and some of the other types of water and wastewater systems is the continuity
of the transient induced pressure wave. Pressure waves travel back and forth along the dedicated
conveyance line(s) with little or no fragmentation and/or dissipation. The transient pressure waves
8
(and therefore their magnitudes) are relatively easy to predict if the other system properties (e.g.,
friction factors, flow rates, etc.) are known with some level of certainty.
A subset of a water transmission system is a water collection system. Such a system collects raw
water from multiple wells and well fields and conveys it to a holding reservoir or WTP. While these
systems typically contain many branched transmission lines, they are less isolated, and therefore the
typical transient conditions and risks are slightly different. The difference comes in the form of the
potential pressure wave origin (i.e., due to multiple well pumps), pressure wave propagation
direction, and the amplification potential due to the presence of branched dead ends.
Transmission systems are not only present in rural and inter-basin transfer areas such as the one
shown in the previous Mexico City example, but also in dense urban areas such as first world
metropolitan centres. In these systems, dedicated transmission pipelines are used to transfer treated
water from a lower pressure zone to a higher pressure zone and from one pump station to the next.
In such a case, the transmission pipelines are typically of the pumped variety, and at times can be
interconnected to the local water distribution networks.
2.2.2 Water Distribution Networks
Water distribution systems and water transmission/supply systems have a lot of properties in
common. First, they obviously convey water that is ultimately directed for consumption. Second,
both of these systems are exposed to the risk of hydraulic transients, even if transients are explicitly
considered and designed for. On the other hand, these two types of water systems also have
significant and distinctive differences when it comes to both their physical characteristics and the
response to such hydraulic transient based risks.
Water distribution systems are typically defined as smaller scale potable water systems that comprise
a network of pipes. Due to the health requirements arising from direct consumption, the potable
nature of the fluid media necessitates that these types of systems be of the pressurized variety. In
most distribution systems the requirements of maintaining a certain level of pressure necessitates the
reliance on pumped systems. The reliance on pumps acts to introduce additional boundary
conditions, which if altered, can act to introduce additional transients and risks to the system.
9
Water distribution systems typically contain pump stations or booster pump stations, more
complicated networks (loops or branches) of smaller diameter pipe (e.g., 400 mm or less), service
connections to ultimate consumers (e.g., households, industry, etc.), storage elements (e.g., tanks or
reservoirs), and additional appurtenances (e.g., flow control valves, fire hydrants, etc.). Figure 2-2
presents a typical layout of a part of a water distribution system in a large metropolitan area.
Figure 2-2: Sample Water Distribution Network (Region of York)
The increased complexity of a distribution system makes the hydraulic transient response much
different, and more difficult to ascertain or predict directly. Unlike transmission systems that
contain a small number of pipes and boundary conditions (and therefore a relatively small number
of possible sources of transients), distribution systems are by their very nature hydraulically complex.
In these systems, transient risks can arise from a variety of sources, including those pertaining to
supply, conveyance and demand. Fortunately, the complexity of these systems is often beneficial
when it comes to their transient based response. This is because complex pipe networks act to both
fragment and dissipate transient pressure waves, through both additional frictional sources and flow
paths. The overall character and magnitude of water distribution system transients is dependent on
a host of factors, including the amount of storage, the interconnectivity of the network and the
system demand relative to the supply and conveyance capacity. Overall, while these types of systems
are similar to water transmission systems, they should undoubtedly be treated differently from the
point of view of both hydraulics and hydraulic transients.
10
2.2.3 Wastewater Forcemain Systems
Wastewater systems come in many forms and varieties, and therefore their hydraulic properties can
significantly vary. Wastewater operations and flows range from those that are driven by gravity in
open channels to those that are fully pressurized in closed conduits, and with intermediaries which
are referred to as mixed flow regimes. While the latter two of these are hydraulically interesting,
their hydraulic transient response is significantly different than those of water transmission and
distribution systems, and will therefore not be considered as part of this thesis.
Pressurized wastewater systems can be closed conduit or open channel, the former of which is more
commonly referred to as wastewater (or sewage/sanitary) forcemain systems. These systems are
hydraulically similar to water transmission systems in that they typically comprise one or two
dedicated pipelines (forcemains) that convey the fluid from a pump station to a downstream
receiving location such as a sanitary sewer manhole, outlet, treatment plant or another pump station.
These systems can often include a limited number of pump stations and branch forcemains. Most
wastewater forcemain systems are installed in areas in which the topography precludes the use of
gravity driven sanitary sewers, and in many instances, these pumped systems work against a relatively
low static head and mostly act to overcome the frictional head loss.
The risk due to hydraulic transients in wastewater forcemain systems is relatively similar to those of
water transmission systems. These risks mostly arise due to changes in boundary conditions such as
those arising from pump failures or line filling, or due to other unplanned events such as forcemain
breaks or premature air valve closure. For additional detail on wastewater forcemain design
considerations, please refer to Section 3.3.3.
2.3 Risks Due to Transients
2.3.1 Background
There are two main reasons why system owners should care about hydraulic transients. First, the
magnitude of the high and low pressures associated with transient events is often large enough to
cause serious damage to system components, devices and pipeline segments. Second, but just as
important, these transient pressure fluctuations can be controlled and managed by reasonable and
11
strategic investment in a surge protection system and through logical and controlled system
operation.
Any hydraulic transient analysis and design, whether achieved via modeling or field work, should
always consider both the positive (upsurge) and negative (downsurge) pressure potential. Only a few
surge protection devices can actually mitigate both of these transient pressure risks, and therefore an
understanding of their limitations is required. The positive pressure concerns have traditionally been
the dominant concern because of the obvious connection between unexpectedly high pressures and
pipe/component failure. However, negative transient pressures, especially those that are sub-
atmospheric (i.e., partial or full vacuum), should also be avoided in order to:
i) Minimize the added stress on the pipes, fittings, and devices;
ii) Reduce the magnitude of any reflected positive pressures (i.e., return upsurges);
iii) Prevent the formation of vapour cavities that can collapse and lead to significant
positive pressure transients;
iv) Minimize fatigue stresses on the pipe that can lower the ultimate strength of the
pipe and allow future upsurges to cause a break;
v) Prevent the possibility of an ‘intrusion event’ that can lead to a public health risk
and/or to the drawing in of contaminated water and/or soil particles;
vi) Prevent a partial or full pipe collapse due to vacuum conditions that would lead
to significant leakage, service disruption and/or loss of capacity;
vii) Prevent the potential wear and tear on the inside mortar and liners of pipes, as
well as on gasket and other joint elements; and
viii) Minimize the propagation of small cracks or breaks that can increase leakage and
therefore reduce the economic performance and overall efficiency of the system.
Negative transient pressures are often difficult to completely eliminate, but in most water and
wastewater systems their duration should be reduced and minimized. In that sense, it is often the
product of the negative pressure magnitude and its duration that should be minimized.
Transient events in typical water and wastewater systems occur as a result of any flow change or
disruption, and these changes are often brought on due to a change in a boundary condition. The
12
following is a partial list of events that can induce transient pressure fluctuations and subject a
system to a high stress (and potential failure):
• Power failure event that causes a pump to trip (i.e., turn off);
• Power failure event that causes an automated valve to suddenly close or open;
• Rapid change in the water level of a storage element such as a reservoir or tank;
• Pipeline (watermain, forcemain, etc.) break or failure;
• Improper operation of a fire hydrant;
• Mechanical failure of a pump or valve;
• Rapid expulsion of air through an air valve, or, indeed, any other valve;
• Rapid valve closure or opening;
• Constant speed pump start-up against a fully open valve;
• Premature or rapid check valve closure (i.e., slam)
• Repetitive valve cycling (i.e., chattering);
• Rapid filling or draining of a pipeline;
• Surge protection device failure; and
• Excessive fluid discharge via a surge device.
Managing the risk of hydraulic transients requires assessing the threat and assigning appropriate
protection measures. It requires the analysis of all the different events that can cause transients, and
the estimation of the consequences of their failure. The following subsection outlines the traditional
methodology for such an assessment, while the subsequent chapters further elaborate on some of
the shortcomings of a typical hydraulic transient analysis.
2.3.2 Traditional Risk Analysis
In its simplified mathematical form risk is a product of the probability of a negative (from the
perspective of the system) event occurring, and the subsequent consequences of such an event.
Despite its mathematical nature, risk is often subjectively assessed and defined. Given the number
of factors involved in a total or performance based failure of water and wastewater systems, it is
often quite difficult to isolate a single causal event and equally as difficult to directly link specific
events to consequences. Therefore, the term risk itself only provides an indication of possibilities
13
rather than a certainty in the outcome. Rowe (1979) defines risk as the “potential for the realization
of unwanted consequences from impending events”.
From the point of view of hydraulic transients, risk is similarly often subjectively described due to
the uncertainties in quantitatively predicting the probability of occurrence and the impact of the
consequence. In other words, the potential worst-case consequence of a specific transient event is
understood, but the probability of such an event occurring and the likelihood that it will combine
with other factors to yield the worst-case consequence, is difficult to quantitatively predict. As such,
traditional transient risk assessments often take the form of scenarios or risk-consequence matrices.
Although risk assessment of hydraulic transients is informally performed via a transient analysis, its
traditional methodology is scenario and surrogate based. Consideration is rarely given to a
comprehensive risk analysis of all transient events, and how such a combination can lead to a system
or performance failure. Thorley (2004) outlined a twelve step systematic methodology for assessing
and managing transient risks in hydraulic systems. This type of methodology must act to address
questions such as:
1. What impending events could give rise to these situations?
2. What is the probability that they will occur?
3. Does it matter?
4. If so, what can be done about them?
The problem with the above questions (and associated qualitative methodology) is two-fold. First,
questions 2 and 3 can rarely be answered in a quantitative manner, as the frequency of occurrence
(i.e., loading) and the consequence are not only difficult to determine, but are also rarely a function
of only hydraulic transients. Second, the methodology is driven by a single worst-case event, and
disregards the joint impact of many different transient (and non-transient) events and/or causes.
Other authors such as Lee et al. (2009), Fleming et al. (2005), and Kirmeyer et al. (2001), have
focused on specific risks from hydraulic transients such as intrusion during low pressure events. In
such cases, the specific risks were indeed correlated to the probability of occurrence and the
consequence. Furthermore, the risks were attributed to a few system parameters and/or metrics.
The best quantitative risk analysis in this field of study can likely be attributed to the traditional
14
fatigue analysis of plastic pipes (e.g., PVC, HDPE, etc.). In such an analysis (and resulting design
standard), the consequence in the form of a plastic pipe failure is directly tied to the frequency and
magnitude of pressure loading (both normal operating and transient). In other words, the
acceptable loading conditions are defined by the magnitude and number of pressure cycles (see
Bowman 1990 and Jeffrey et. al 2004). In the realm of CPP pipe, risk assessments of pipe failures
can directly be tied to the operating pressure regime, the frequency and magnitude of transient
pressures, and the current condition of the pipe (e.g., with the assistance of a wire break condition
assessment and analysis). Zarghamee and Fok (1990), and Zarghamee et al. (2003), developed a
finite element model that correlates a variety of the above mentioned external parameters, and
provides a quantitative risk analysis via potential failure curves and limit states. Chapter 7 describes
some of these quantitative risk analysis methodologies in more detail, and ultimately proposes a field
data derived and index based transient risk assessment of existing systems.
While the difficulty of performing a transient risk assessment has been briefly discussed above, it is
equally, if not more important to address Thorley’s last question on what can be done to minimize
any such risk. Once an analysis is performed and some form of transient risk has been identified,
the next step is to make design and/or operational modifications in order to reduce the consequence
of any such event(s). The following section provides a discussion on the typical options for
transient protection, both in the form of devices/equipment, and specific design strategies.
2.4 Transient Protection Options and/or Equipment
The direct (or traditional engineering) approach to hydraulic transient protection is commonly
understood from a physical perspective as it relies on the design and installation of dedicated surge
protection equipment and/or devices. However for the most part, the current practice relies on
rules-of-thumb and is based on a narrow selection of options. A proper transient protection design
should consider all possible options, and these options should then be narrowed down to a few that
are applicable in the system. The few remaining options should then be analyzed in detail across
several criteria. This section summarizes a partial list of typical surge protection options and
approaches that should at least be preliminarily considered during a design and/or upgrade of a
water (and to some extent wastewater) system. As noted in the previous section, not all equipment
and approaches target the same (or all) hydraulic transient concerns, and hence there exists a clear
trade-off between performance, applicability and ultimately cost.
15
1. Do-Nothing (i.e., dealing with the consequences as they occur)
This strategy requires no initial or dedicated investment in surge protection or control, but may
accumulate greater costs in maintenance and repair of system components damaged from transient
events. Since every pipeline system is unique it may be the case that for some systems a do-nothing
approach is valid given the nature of the system.
2. Improved Operation and Training (i.e., learning to use the system more gently)
Includes practices such as slowly opening and closing valves and highly controlled ramp-ups and
shutdowns of pumps. Transients arise from imbalances between flow entering and leaving a pipe
segment. Any system operation that quickly changes flow conditions can induce a transient. Gentle
system operations allow flow to gradually change in the pipeline thereby minimizing any imbalance
in flow. This is particularly relevant to filling, draining and valve operations.
3. Increased Pipe Strength and Rating
Increased pipe rating is a protection strategy that is relevant to proposed systems and systems where
a pipe break has occurred and needs to be repaired. As the name suggests, this strategy involves
choosing a pipe with a working pressure rating strong enough to safely withstand the maximum
theoretical transient pressures. Even though most pipes have an ultimate strength greater than the
rated working strength, it is best practice not to encroach upon this built-in factor of safety.
Implementation of the strategy assumes that the transients will be handled by the pipe only.
4. Less Rigid Pipes (Lower Wave Celerity)
The maximum and minimum theoretical change in head produced by a sudden transient event is
directly proportional to the wavespeed. The wavespeed is the sonic velocity of a liquid flowing
through a pipe and is a function of the pipe wall elasticity. Rigid pipes have greater wavespeeds than
less rigid pipes. As such, the theoretical maximum and minimum transient pressures can be greater
for rigid pipes than for flexible pipes if all else is the same. However, reality is often more complex
than this simple rule would imply.
5. Surge/Pressure Relief Valves
These types of valves are designed to automatically open at a predetermined high pressure setting.
Springs, counterweights or automatic hydraulic pilots can be used to set the opening pressure. The
16
release of water (to the atmosphere, suction line or well/reservoir) can alleviate excessively high
pressures in a pipeline.
6. Surge Anticipation Valves
Surge anticipation valves function the same way as surge relief valves except that they are designed
to open at a predefined low pressure setting in anticipation of a returning upsurge. If the valve cycle
of opening and closing are properly set, the valve will be completely open in time for the transient
upsurge, and then close slowly when the pressure drops to normal. This type of valve must be used
with caution because they have been known to make low pressures situations worse.
7. Rate of Rise Surge Anticipation Valves
These types of valves are often confused with the surge anticipation valves because their goal is to
anticipate an upsurge. These valves are designed to sense a rapid increase in pressure as the surge
wave returns and to open fully with no stroke limiters on the valve opening. It must be ensured that
the valves open fast enough on rate of rise to be fully open when upsurge arrives.
8. Automatic Air Release (Outflow) Valves
Automatic air release valves are placed at points in the pipeline where air can accumulate, be it
significant high points or rapid changes in pipe slope. They are designed to automatically open and
release trapped air, and to close again once the pipe is full of water. Even though they do not
necessarily offer direct transient protection they can have an impact on transient performance. If
the air is expelled too quickly a phenomena known as “air slam” may occur as the water column
rejoins.
9. Automatic Air Vacuum or Vacuum Breaker (Inflow) Valves
An air vacuum valve opens automatically when the internal pipe pressure drops to atmospheric
pressure. This valve allows air to enter the pipe thereby providing protection against possible pipe
collapse caused by negative pressures.
10. Automatic Combination Air Valves
Combination air valves contain both an air release component and an air vacuum component. This
type of valve can expel trapped air in a pipeline, as well as to allow the air to both enter a pipeline if
17
the pressures drop below atmospheric and to leave once the pressures rise above atmospheric.
These valves come in many configurations, including single and dual body designs.
11. Automatic 3-Stage or Non-Slam Air Valves
Three stage air valves operate similarly to conventional air valves except that air is released through
two increasingly smaller orifices. This three stage release of air is designed to reduce the occurrence
of “air slam” in the valve and/or pipeline. These valves include a high capacity air release and
vacuum breaking capability together with a specifically designed anti-shock orifice to provide
controlled air release during the critical moments prior to the elimination of the remaining air
pocket. Variations on the non-slam design exist.
12. Conventional Check Valves
Check valves are designed to allow flow to travel in one direction only. The simplest forms include
a flap that closes under its own weight. Conventionally, they are primarily placed downstream of a
pump to prevent backflow when the pump is not in service. One benefit of a check valve is that it
prevents line draining. From a transient perspective this is beneficial because it can reduce transients
at pump start-up.
13. Dampened Check Valves
Dampened check valves are also known as slow closing or surge check valves. When a pump fails
and flow reverses, a properly functioning check valve will close to prevent backflow. If the closure
is too abrupt, then the valve slams, therefore creating another possibly dangerous transient event for
both the system and the valve itself. The goal of these types of valves is to minimize or eliminate
this undesirable event. These valves come in many forms including spring-loaded, tilted disk and
swing flex.
14. Dashpot Controls
Dashpots are the mechanisms that control check valve closure and prevent slamming. They usually
consist of a rod and hydraulic cylinder. An air and oil reservoir is utilized to provide return force for
the rod when the valve is closing. This ensures a more controlled valve closure through the
provision of a cushion.
18
15. Pressure Regulating Valves
Not to be confused with pressure reducing, relief or sustaining valves, these types of valves are used
to throttle the flow during pump start-ups. The goal of a pressure regulating valve is minimize the
operation of a pump at a point on the pump curve corresponding to a high flow and a low (or zero)
head. In doing so, these valves allow the flow to establish and therefore minimize the transients
associated with events such as line filling.
16. Valve By-Pass Lines
Large in-line valves are often difficult to control especially when they are almost closed. Such
problems can be minimized by the addition of one or two smaller valves mounted in a parallel by-
pass orientation. This strategy has proven very positive during line filling and system restarts.
17. Pump By-Pass Lines
Pump bypass lines are designed to prevent the build-up of high pressures on the suction side of a
pump and cavitation on the discharge side during a downsurge. They are useful in low-head systems
where the pump suction line operates under positive pressures. Bypass lines are installed parallel to
the pump connecting the suction line to the discharge line. They contain a check valve that only
permits water to flow from the suction side of the pump to the discharge side. The check valve is
activated when the suction head exceeds the discharge head thereby providing a source of water to
mitigate low pressure events on the discharge side or high pressure events on the suction side.
18. Increased Pump Inertia
Pump inertia is a measure of the resistance a pump has to a change in the rate of rotation. The
greater the inertia the longer it will take for a pump to come to a complete stop once the power
supply has been cut off. By increasing the inertia of a pump the pump run down time is extended.
This is advantageous from a transient perspective because water is being pushed through the system
(albeit at a decreasing rate) even after a pump trips. This volume of water can greatly decrease the
initial downsurge and the resultant upsurge. Increasing the pump inertia is most easily done through
the initial design stage in which the pumps are sized by taking into account the inertia.
19. Flywheels
In the context of transients, a flywheel is a weight added between the motor and pump to increase
the mass being rotated and thus the pump inertia. The increased inertia increases the rate of pump
19
slowdown and therefore the magnitude of the resulting transient. Flywheels are good alternatives
for increasing inertia when the actual inertia of the designed pumps cannot be increased. Flywheels
are more easily installed with non-submersible pumps. Some manufacturers are hesitant to warranty
a pump if a flywheel in installed.
20. Electronic Capacitors
Electronic capacitors are made to drive the motor of a pump in the event of a power failure, thereby
limiting surge pressures. These devices essentially act as "electronic flywheels". This proprietary
technology is currently relatively rare and unproven.
21. Variable Speed or Frequency Drives
When power is available, pumps with variable frequency drives can be ramped up and shut down in
a slow and controlled manner. This feature reduces the instantaneous change in flow conditions
whenever a pump is turned on or off. This reduces the imbalance in flow and can decrease the
severity of transients at start-up and shutdown. These types of drives may also have additional
benefits with respect to efficiency, energy consumption and resonance/vibration. However, such
devices do not necessarily mitigate the surge pressures associates with power failures.
22. Soft-Start Starters
Soft starters reduce the load and torque in the power train of the pump motor during start-up.
These are superficially similar to variable speed drives in that they reduce the mechanical stress on
the motor and shaft, as well as the electro-dynamic stresses on the electrical accessories. Unlike
variable speed drives, soft starters cannot vary the speed of the motor once the motor is up to
speed.
23. Backup Power
Backup power can be used to maintain pump operation in the event of a primary power failure.
This strategy improves system reliability by avoiding down time following a power failure. Backup
power is typically achieved by the use of a second parallel power source, such as a diesel generator.
24. Alternate Pump Drive System
Alternate pump drive systems attempt to minimize the risk of a multiple and concurrent failure of
pumps during a power failure event. Unlike in the case of backup power, this strategy requires that
20
at least one pump is primarily powered by an alternate drive and/or power source, such as natural
gas or diesel. This strategy is typically used during peak hour operating conditions for a specific
peak hour pump.
25. Open Surge Tanks
Open surge tanks are protection devices that can relieve both excess and minimum transient
pressures. The simplest form of open surge tank is a vertical standpipe connected to a pipeline.
When pressure in the pipeline increases, the water level in the surge tank increases and when
pressures in the pipeline decrease the surge tank provides a supply of water to reduce the minimum
pressures. Transient pressures are dampened out by fluid friction as the water level in the tank
fluctuates up and down. Because they are open, this type of surge tank must be designed sufficiently
tall so that it will not overflow.
26. Traditional Air Chambers (Hydro-Pneumatic Pressure Vessel/Tank)
Traditional air chambers represent a modification to open surge tanks, and can come in several
different configurations including vertical, horizontal and spherical. In high head systems the height
requirement of an open surge tank may be prohibitive from a construction, safety or economic point
of view. An air chamber is basically a closed top surge tank with a volume of both air and water. In
the traditional air chamber, the air is usually kept under pressure by means of a compressor. During
an upsurge the air acts as a cushion and absorbs some of the excess pressures. During a down surge
the tank provides a supply of water to mitigate low pressures in the pipeline. Air chambers must be
designed such that the tank will not empty during a transient event and suck air into the system.
27. Bladder Type Air Chambers (Hydro-Pneumatic Pressure Vessel/Tank)
A bladder type air chamber functions in a similar manner to a traditional air chamber with
compressor. This tank, however, is equipped with a bladder that is pressurized such that under
normal operating conditions it maintains the desired air volume. In this type of air chamber the air
and water phases are separated and the air does not have to be refilled.
28. One-Way (Feed) Tanks
One-way surge tank or feed tank can be of the open or closed type surge vessel previously discussed.
The purpose of this device is limited compared to other surge tanks in that it only functions to
minimize the low pressure characteristic of a transient downsurge. A check valve only allows flow
21
to travel from the tank into the pipeline during low pressure events. As such, this vessel does not
protect against transient upsurges.
29. Storage Tanks or Reservoirs
Storage tanks buffer a pipeline from an imbalance in supply and demand. In an emergency situation
when supply is cut off (i.e., power failure) a storage tank can provide a supply of water. Elevated
storage tanks especially can help maintain positive system pressures in the pipeline downstream of
their location.
30. Specific Device Strategies
These types of strategies can be uniquely employed at specific pump stations, storage tanks or on
unusual transmission lines. For example, a check valve can be located between pressure zones to
provide a backup or emergency supply of water, or to prevent a high point from draining during
power failure. The check valve could be designed to be closed under normal operations. If pressure
should drop in the higher pressure zone however, the check valve would open and permit water to
flow from the low-pressure zone into the high-pressure zone. As another example, a check valve
can be employed as a control measure allowing for redundancy in air chamber operation and
protection across different pressure zones at the same pump station.
31. Route Alignment
System topology often governs the ultimate alignment and profile of main transmission lines, and
this initial design decision is often the cause of many future transient problems. For example, a
decision to tunnel rather than to build across a hill may prove to be beneficial with respect to the
future transient response, because this might eliminate the troublesome system high point that
would almost always have problems with negative pressures.
32. Specific Pressure and Zone Based Strategies
These types of strategies are more easily employed during the master planning and system design
stages. The decisions on zone boundaries and on where to put important boundary conditions such
as pump stations or tanks have a great impact on the response of transients. For example, one
strategy might be to equip an elevated tank in higher zone with a pressure reducing valve in order to
eliminate the need for a lower zone transmission line. Another strategy may be to add an additional
booster pump station or to change the design location of another.
22
33. Distribution System Interconnection Strategies
This strategy relies on the fact that long transmission mains are extremely susceptible to transient
pressure fluctuations. Its goal is to interconnect the transmission system into the distribution
system, therefore allowing for additional sources of pressure wave dissipation. It also can be used to
enable bypass from one zone to another to relieve short-term downsurge conditions, providing
sufficient positive pressure head is available.
34. Shock, Vibration and Resonance Strategies
These specific strategies deal with complications that may or may not arise from transient flow
conditions. Some of these strategies may include: base isolation of pumps, extra lengths of pipe,
changes in natural frequency of devices, and thrust blocking and/or rubber pad absorption of
undesired loads.
The above detailed list of transient protection devices and risk mitigation approaches essentially
represent the options that an engineer, designer, and/or system owner have to consider. However,
in order to identify the risk and then subsequently determine the appropriate solution from this (and
additional) list of options, the same individual(s) must first determine the type of analysis approach.
2.5 Hydraulic Transient Analysis Approaches
The previous section provides an overview on what hydraulic transients are, as well as outlines some
of the resulting risks and protection options. The next topic that needs to be addressed is the one
pertaining to the available options for performing a hydraulic transient analysis. The following
subsections outline typical analysis approaches that have been previously, and are currently, being
used in practice. While the analysis approaches are not the main focus of this thesis, they are still
briefly described in order to provide context for the following chapter on the shortcomings of
current analyses.
2.5.1 Do-Nothing
Through historical practice, the hydraulic transient analysis of a water and wastewater system could
easily have been cast as a decision making task. In other words, the question of whether or not to
consider and address hydraulic transient concerns has traditionally been made directly or implicitly.
23
To that end, most rational decision making (and therefore rational analysis and design) typically
considers the option of doing nothing. In the realm of transient analysis, the do-nothing option is
defined as not considering the impacts of hydraulic transients but also as not doing anything to
protect against their risks. While the first of these is strongly fading away due to better education
and engineering standards, the second is still prevalent due to economic, social and professional
limitations.
The do-nothing transient analysis approach essentially states that transients are likely not significant
in the particular system at hand, and that if something does happen as a result of transients, it will
then be addressed accordingly. The do-nothing approach is focused on the prospect of short-term
cost savings in both the engineering professional services and in the capital or construction cost of
potential protection options. The resolution of any resulting problem(s) may simply be in the form
of a physical repair that returns the system to its status quo, or may be in the form of an alternative
analysis approach that leads to a different solution. While the do-nothing approach to hydraulic
transients is not recommended for most water and wastewater systems, it does serve a purpose in
some trivial cases. The question that needs to be addressed is whether or not the individual(s)
making the decision for this approach are knowledgeable and experienced enough to do so, or
whether or not the decision is based on a preconceived bias or lack of understanding.
2.5.2 Rule-of-Thumb and Standard Convention
The next hydraulic transient analysis approach can be described as a step-up from the do-nothing
approach. This approach of using standard convention and rules-of-thumb for the analysis and
design of water and wastewater systems essentially accepts the fact that hydraulic transients can pose
a risk, but also assumes that such a risk is quite predictable and constant for most systems. The
standard conventional rules and rules-of-thumb aim to quantify both the risk (e.g., magnitude and
characteristic of a transient event) and the nature and size of the required protection. The following
is a sample and partial list of common rules-of-thumb that are often employed in practice in one
form or another. (A detailed discussion on why these may not apply to all systems is outside of the
focus of this thesis and is therefore not provided here, however, a partial discussion is provided in
Karney and McInnis, 1990.)
24
• The risk due to excessive transient pressures can be estimated by the simple
Joukowsky expression (∆H = a∆v/g) relating the pressure fluctuation to a sudden
change in velocity.
• Flexible pipe should be chosen over rigid pipe due to its greater expansion and its
lower wavespeed.
• For rigid pipe such as CPP, the transient pressures should be assumed as 40% above
the maximum operating pressures.
• Higher wavespeed values should always be used for predicting a transient response
because they are more conservative.
• Air valves can be sized on conservative maximum flow rates that are used to represent
filling and draining scenarios. Simple equations relating the required valve size as a
function of the flow rate can be summarized in simple graphs or slide rules.
• One size of air valve is best for an entire system.
• All surge protection devices should be designed for peak flow/operating conditions.
• The shorter the pipe length the lower the risk due to transients.
• Maximize the number of system appurtenances such as elbows and tees and lower the
overall risk due to transients.
• Minimize the fluid velocity and therefore lower the risk due to transients.
• Interconnect systems (e.g., into distribution networks and loops) and eliminate the
need for additional transient analysis.
• Install SRVs at all pump stations, and size the SRVs as a fraction of the discharge
piping size.
• SRVs automatically act to limit the maximum surge pressures to the level of the
chosen high pressure set-point.
• Doubling the number and size of any surge protection device in the system will act to
lower the risk of hydraulic transients.
• HAC volumes can simply be designed as a percentage of the total fluid volume in the
system.
• If an option exists, always choose the larger surge protection option.
25
As is the case with the do-nothing approach, the standard convention and rule-of-thumb approach
attempts to apply rules that are often applicable in simple systems to other more complex hydraulic
systems. As Karney and McInnis (1990) note: “Traditional wisdom for identifying worst-case
scenarios is based on elementary equations, rules-of-thumb, or common sense; in other words,
simple relations that may have little or no bearing on the performance of more complex systems”.
The true problem with the rule-of-thumb approach to transients is that no two hydraulic systems are
the same, as every system is partially unique. For example, consider the implication of one of these
rules in a water system with storage, and in the same water system without storage (i.e., a closed
system). The transient response between the two systems is undoubtedly different. The reason that
this approach is still commonly used is because it is geared at the root sense of engineering.
Engineering has long been a profession of logic, common sense, and prescriptive analysis and
design. It is the prescriptive nature of these rules that attracts their end users, believers and
practitioners.
2.5.3 Numerical Modeling
Numerical hydraulic modeling rose to prominence as a result of both the increasing complexity of
the required hand calculations, as well as the unwillingness to experiment on real-life systems in the
field. As Walski (2006) noted, “engineers throughout the early 20th century were able to design and
analyze the hydraulics of a functioning water distribution system using a combination of
simplifications, rule-of-thumb, and conservatism. The ability of engineers to construct systems
exceeded the profession’s ability to analyze them.” Concurrent with the increasing complexity of
hydraulic systems was the advancement in technology, especially in the powers of computing. The
field of numerical hydraulic modeling was thus born and it has never looked back since.
In the recent decade, numerical modeling has been regarded as the standard convention for
hydraulic analysis and design. It relies on the construction and utilization of a computer model; a
model which seeks to portray the physical system through a mathematical representation of its key
properties and parameters. In the realm of water or wastewater systems, the numerical models
encode system properties such as length, connectivity, roughness, demand, boundary conditions,
etc., and are ultimately then used to solve an analysis, design, or optimization objective. The
complexity of models can vary through their detail, physics, and overall assumptions.
26
Typical hydraulic transient models (i.e., full water hammer models) represent a single (albeit a more
complex) type of a numerical hydraulic model. These models are more complex than steady state,
EPS, quasi-steady and rigid column models, but in practice they are frequently not employed to their
full mathematical and physical potential. More specifically, traditional numerical transient models
are typically (but not always) limited to single phase, one-dimensional flow, steady friction, constant
wavespeed, and do not take into account fluid structure interaction (FSI). The reason for this is
simple – in most systems, the additional accuracy does not warrant the additional complexity. In
other words, we can often make the same type and level of decision using a simpler model.
Traditional hydraulic transient models typically fall within the time domain analysis spectrum and
rely on unidirectional analysis. These models are mathematically based on two quasi-linear
hyperbolic partial differential equations known as the dynamic and continuity equations. The
numerical solutions to these equations generally fall into the following broad categories: method of
characteristics (MOC), wave characteristics method (WCM), finite difference techniques and finite
element methods (McInnis et al., 2004). For a summary chart of traditional transient analysis
solution approaches, please refer to Figure 2-3. The choice of solution method strongly depends on
the complexity of the system (and therefore the accuracy required), the computing time, and the
experience required. Nonetheless, in most water and wastewater applications, the MOC is the most
widely accepted solution approach; a solution approach that is incorporated into several commercial
(and private) transient software analysis packages. The main rival to the MOC models are the WCM
models; models based on a method that tracks the event or disturbance based on wave propagation
mechanics (i.e., almost, but not quite algebraic water hammer), rather than through a time-space
grid. The commercial software packages that implement the WCM are sold on the notion of a faster
computing time for water systems consisting of large distribution networks.
Numerical transient modeling is an effective means by which a system can be analyzed and/or
designed without inflicting a physical change. In this common transient analysis approach, the
engineer can look at various design options and alternatives. Numerical models enable a proper
sensitivity analyses, future condition representation, and much easier review than any other transient
analysis approach. These are just some of the reasons why numerical transient analysis has become
the industry norm.
27
Figure 2-3: Traditional Transient Analysis Solution Approaches
The above advantages of this approach are unfortunately also accompanied by some often
unrealized limitations. First, numerical transient models are highly theoretical, and if not validated
or calibrated, they may represent a different reality than the one intended. Second, numerical
models and software packages are increasingly becoming more and more user friendly, to the point
28
where any amateur engineer (or even technician) can start to feel comfortable in performing such
analysis and design. In other words, numerical transient models can bring comfort and prescriptive
simplicity to unqualified individuals, and therefore possibly make the end objective equally as
troubling as the previously discussed analysis approaches (i.e., do-nothing and standard convention).
Some shortcomings of the current numerical modeling approach to transient analysis and design are
further discussed in Chapter 3.
2.5.4 Field Data and Investigation
As noted earlier, the primary concern with the standard numerical modeling approach to transient
analysis is the theoretical (and often academic) nature of these models. A possible solution to this
dilemma is to rely on actual field data analysis. Before the age of numerical models, difficult
hydraulic systems were often troubleshot in the field using either a trial and error approach, or with
the implementation of a physical model. Ironically enough, it was the risks arising from field
troubleshooting and the cost of physical modeling that partly gave rise to the numerical analysis
approach that is now the norm. The industry simply needed an analysis tool that reduced the time
and cost commitment, while also reducing the potential liability associated with poor analysis/design
and the poor results that were often attributed to field work.
The question that should be asked at this point is simple: Why is this field approach to transients still
considered if the numerical analysis approach has been shown to be widely accepted? The answer
partly lies in the previously mentioned shortcoming of the numerical approach, but also in the
advancement of field sensor technology. The modern field sensor technology (such as high
frequency transient pressure monitors) allows the system owners and designers to listen to their
existing systems and to optimize their newly designed and constructed systems. Transient pressure
monitors can gauge the pulse of the system by continuously monitoring transient pressures at key
system locations. The continuous (or even limited) transient pressure monitoring can bring insights
on the level and frequency of most hydraulic transient risks, while at the same time act to provide
clear answers to design uncertainties such as surge protection size, physical parameters (e.g.,
wavespeed or friction), etc.
The field data analysis and investigation approach to transient analysis cannot solve all of the design
problems, but sometimes it can more directly and decisively solve some. More importantly, it can
29
close the theoretical limitation loop for all design problems. For an expanded discussion of the field
work approach, including the type of field pressure monitors and sample output, please refer to
Chapter 5 and Chapter 6.
2.5.5 Hybrid
The previous sections outline and describe four (4) distinct types of transient analysis approaches:
do-nothing, rule-of-thumb, numerical modeling, and field work. All of these approaches have been
shown to be effective in many types of engineering problems, but none of these can be classified as
a fool-proof solution method. The numerical modeling approach is currently the most widely used
approach, as it is the most effective in bridging the gap between doing nothing and doing everything
at the lowest cost. However, numerical modeling has not completely eliminated the need for the
other three approaches. The do-nothing approach is still commonly used for simple and/or older
systems, in which the cost of repair is just part of the business. The rule-of-thumb and standard
convention approach has a wide use that is similar to the do-nothing approach, and quite applicable
to certain types of systems. More importantly, it is often an excellent method for preliminary
analysis and/or screening of alternatives.
The field work approach has recently experienced a resurgence in its application due to the
advancements in, and availability of, modern sensor technology. This resurgence has brought light
to some of limitations and risks of numerical modeling, and is attempting to substitute parts of the
modeling process with field based studies and reviews. While it will likely never (nor should it ever)
replace the numerical modeling approach, it can act to supplement and confirm the theoretical
predictions and assumptions of the actual system.
The obvious limitations of the four analysis approaches can be avoided if these approaches are
combined into one type of transient analysis. Such a transient analysis can be classified as a
“Hybrid” method, in that its goal is to draw the best from all approaches. A hybrid transient
analysis is nothing new, as engineers have often combined parts or all of the approaches to solve
transient related problems. However, such an approach is currently not promoted, as the industry
mostly relies on numerical modeling outputs. Any good engineer or designer should be able to use
all tools that they have at their disposal when it comes to a hydraulic transient analysis. Therefore,
such an engineer should consider implementing the do-nothing approach for certain systems. They
30
should also use rules-of-thumb and standard convention for preliminary reviews and initial design,
and as part of numerical and field work based estimations and design. Lastly, the engineer should
also supplement most numerical based analyses with field based confirmations, and vice versa. Such
an integrated approach can, and will, lead to the design of more cost effective and reliable water and
wastewater systems.
2.6 Summary
The focus of this chapter is to provide a widespread background introduction to hydraulic transients
and the current variance in the industry’s approach to considering these important hydraulic events.
The chapter initially provides a definition of hydraulic transients, and establishes the link between
these events and the risk to water and wastewater systems. The chapter then proceeds to define and
discuss the properties of three (3) key types of common fluid systems: water transmission, water
distribution, and wastewater forcemains. The hydraulic transient conditions and risks in these
systems are identified in order to explain and support one of the general themes of this document –
that no two systems are the same and that a design and/or analysis of any such system must
therefore consider a variety of transient analysis approaches.
The chapter proceeds in identifying the types of risks, while at the same discussing the general
concept of risk and providing a general understanding how this risk has traditionally been identified
and dealt with. In essence, it provides a critical assessment of the qualitative nature of traditional
transient risk identification and mitigation. The chapter then presents a long list of transient
protection devices and system design approaches, in order to not only provide an indication of the
complexity of system protection options, but also as a means of providing context to the decision
making process.
The chapter concludes with a discussion of the different types of transient analysis approaches that
can (and often are) considered, and identifies five (5) of these which are most commonly employed
in the industry. The central focus of this discussion is the current trend towards numerical
modeling, and how this trend can be both dangerous and narrow minded. The chapter essentially
establishes the need for a hybrid (or multi-faceted) approach to hydraulic transient analysis; one that
relies not only on theoretical numerical modeling, but also on the actual in-situ field performance of
the system and the engineering expertise of the analyst.
31
Chapter 3 Current Analysis Shortcomings
The focus of this chapter is to demonstrate a few of the most direct and often overlooked
shortcomings in the analysis and design for hydraulic transient conditions, including some of which
are typically rooted at the steady state level. With that in mind, it should be noted that this chapter
does not attempt to provide a deep theoretical analysis of current steady state analysis techniques;
techniques which have been the subject of numerous studies and publications over the past few
decades. The overall goal of this chapter is to identify key analysis topics (from the point of view of
transients) that the industry has still not grasped or given much thought to.
3.1 Steady State Models
The design task of performing a hydraulic transient analysis more often than not relies on the use of
an existing or newly constructed numerical hydraulic model. Most water and wastewater system
owners such as municipalities or utilities have progressed to the point at which they currently own
and maintain a numerical hydraulic model of their water and wastewater systems. The type and
complexity of the models can significantly vary from one owner to another, but more often than not
(especially for large water systems) the numerical models are derived from an asset management
frame of mind. In other words, the hydraulic models are constructed in great detail using
commercial software for reasons other than those pertaining to pure hydraulics. The commercially
available software packages are always evolving and their capabilities are significantly increasing
through their GIS, CAD and geospatial connectivity. These extensive capabilities lead to large
system wide asset management (or component based) numerical models; models which are then also
used for the hydraulic analysis and design tasks.
The existing complex numerical hydraulic models are usually of the steady state or extended period
simulation (EPS) variety. These models can vary (or include options) for: the design year or year of
operation, actual (current) or predicted (future) system demands and flow rates, daily variations in
hydraulic conditions (e.g., reservoir levels, diurnal demand curves), etc. As a result, steady state
models can be so complex as to contain as many as 100,000 nodes (i.e., junctions) and links (i.e.,
pipe segments) for a variety of possible operating scenarios. The important question to be asked is,
are these models suitable for a hydraulic transient analysis?
32
In order to answer the above question, one must first look at the limitations of the steady state
models. Steady state model limitations can be broken down into two categories: assumptions and
complexity. In the first category, steady state models undoubtedly contain assumptions on both the
system’s physical condition, and hydraulic condition and capability. The physical realities assume
that the model comprises all proper components, and that these components are properly
represented in the numerical model. Examples of physical properties which are typically estimated
and assumed can include some of the following:
• Exact location and length of all pipe segments;
• The type, location, and physical properties of system appurtenances such as flow
control valves, valve chambers and interconnections, etc.;
• Actual pump curves and pump properties;
• Type of pipe, year of installation, and physical condition;
• Open or closed connection points within the system; and
• Location and allocation nature of demand points.
If physical assumptions such as those listed above are not true, or even just significantly different,
then the results predicted by the steady state model may not be accurate enough. If the physical
system properties and the steady state results are not accurate, then a hydraulic transient analysis
based on such a model may be improperly used to make an important design decision. (As is the
case with any analysis or design, the uncertainty can more often than not be accounted for through a
proper sensitivity analysis of the critical assumptions and parameters).
The second group of assumptions under the first category are those pertaining to the system
hydraulics, and these can include some of the following:
• Pipe friction factors such as Hazen-Williams or Darcy-Weisbach;
• Actual (or lack there of) minor head loss coefficients for appurtenances;
• Total flow rates from source locations (e.g., pump stations);
• Flow rate contribution from all different sources;
• Actual system demands (see following section) and their diurnal timing;
• Fluid levels in storage elements;
33
• Pump sequencing and operational logic; and
• System flow and pressure control logic.
As is the case for the physical system assumptions, these assumptions can also significantly affect the
predicted steady state results, and therefore any additional analysis or design – such as a
supplemental transient analysis. The second category of steady state model limitations pertains to
the previously noted idea of complexity. The complexity in the steady state models that are derived
from asset management purposes arises due to multiple reasons, including:
• The complexity arising from the sheer size of the model, including the number of
nodes, links, boundary conditions, etc. The use of such a model for a transient
analysis becomes prohibitive due to either model run-time and/or the reality of
properly interpreting (and actually believing) the results.
• The complexity arising from the deterministic nature and the automatic spatial
allocation of demands, including those demands of insignificant magnitude.
• The complexity arising from the type of demand, and the various peaking factors and
diurnal demand curves.
• The simplification of an EPS model to a single period model, which requires the
assumption of the most common operating and demand conditions.
• Complexities arising from multiple sub-systems, including combined pressure zones
and the lack of pressure zone delineation.
• Complexity arising from too much detail, including the presence of service connection
and intermediate nodes that provide no additional information with respect to
elevation, demand, or simply – hydraulics.
• The complexity arising from successive model updates and changes, including those
pertaining to physical system changes, operational changes, future design years, etc.
The master planning and decision making approach that originally drives the creation
of such models is prone to inadequate model maintenance, validation, and calibration.
More often than not, the models contain system components that should not be there,
or model components that are significantly different.
34
• The complexity arising from the lack of model change tracking, both within the model
itself, as well as by the modeller (owner or consultant). In many cases, the individuals
that create the models are not the model end users.
In general, the proliferation of numerical steady state hydraulic models for water and wastewater
systems has come to the point where the models themselves essentially make the decisions. More
often than not, a modeller is inclined to say something along the lines of “But the model says, or the
model predicts…,” etc. The irony here is that hydraulic models can essentially predict anything that
the user wants them to predict, and therefore the sole reliance on the model outputs without the
consideration of both the physics, and the previous model developments and limitations, is quite
dangerous.
In the end, the lesson that should be taken from this section is that steady state models undoubtedly
have their limitations. Steady state models are not all the same, and are not fit for every purpose.
Most steady state hydraulic models are of the simplest physical form, and lack the physical
understanding and developmental progress that has been extensively presented in academic
publications. Many steady state hydraulic models are rarely if ever properly (or at all) validated or
calibrated with field results, and are then unfortunately often used as the stepping stone for a
hydraulic transient analysis. Therefore to answer the question that was posed earlier (i.e., Are these
models suitable for a hydraulic transient analysis?) – the answer is…not always. Hydraulic transient
models must be fit-for-purpose, and therefore the typical steady state models must be properly
adjusted and validated. The following section takes a step back to look at one of the most important
steady state model assumptions – the deterministic demand.
3.2 Average and Maximum Day Demands
The traditional design of water supply, treatment and distribution systems has been accomplished
using a multi-step approach; one that has often relied on a single objective optimization of a system.
As with most circulation and consumption systems, the design is driven by a mix of input and
output characteristics. In water systems, the output properties (from the point of view of the
system) are most often the current and/or future/potential water demands; outputs that essentially
determine the required design inputs such as water supply and storage. The analysis and design of
such systems is therefore highly dependent on quantifying the expected water demands. In
35
traditional (and most commonly) applied analysis and design, such demands are deterministically
quantified across the life of the system. While many studies (e.g., Lansey et al. 1989 and Goulter
1992) have demonstrated the negative impact of a deterministic demand assumption, and while
many other studies (e.g., Babayan et al. 2005, and Giustolisi et al. 2009) have presented various
theoretical approaches (e.g., genetic algorithms, Monte Carlo simulations, etc.) to stochastic and
multi-objective based design of water (and wastewater) systems, the fact still remains that most of
these systems are still designed based on deterministic constraints such as demand. In his Ph.D.
dissertation on multi-objective stochastic design, Filion (2006) provides an excellent literature review
of both the risks and potential solution approaches to dealing with the demand variable (among
many of the other analysis and/or design variables).
In reality (i.e., outside of the academic and progressive realm), the current and still unchanged design
philosophy is to design water systems by taking into account the peak conditions and constraints
during the life of the system; constraints such as forecasted water demands. One of the most
important demands and forecasts is that of the average base case, which is referred to as average day
demand (ADD). ADD is the primary demand component in design and is defined as “the total
annual quantity of water production for an agency or municipality divided by 365” (Ysusi, 2000). All
other variations in demand (e.g., maximum week, maximum day, maximum hour, etc.) can be
derived in the same manner or represented as a percentage of the ADD. For example, maximum
day demand (MDD) is defined in a similar manner, but it incorporates the maximum 24-hour period
in a given year. The meaning and application of the ADD term seems simplistic, but its origin (and
hence complete justification) is relatively unknown. Similarly, the interpretation of the MDD term is
conceptually clear, but the factors that feed this interpretation are more complicated. Both of these
quantities are system and time specific, and are influenced by expected and/or random events. In
other words, a water demand is not actually a deterministic variable; rather it is just assumed to be.
All widely accepted water system design handbooks refer to the ADD term without giving a single
line or thought as to its origin. The references acknowledge that this quantity is highly important
and that it is part of the first step of determining the rate of water consumption. The well published
1954 Water Supply and Wastewater Disposal (and its 1966 successor) by Fair and Geyer, may be one
of the earliest books which does not explicitly refer to the above term. The authors initially note
that “the capacity and of individual system components is set by what is expected of them”. This
statement, be it vague, at least gives partial justification. However, the book does refer to a
36
synonymous “average daily rate”, meaning that the use of the ADD term in design has been around
for quite some time.
Modern day water system design references typically only define the ADD quantity (along with the
other permutation quantities), and in some instances also elaborate on how (but not why) it is used.
A relevant partial list of such references includes the following:
• Water Distribution Systems Handbook (1999), ed. by Larry W. Mays.
• Urban Water Demand Management and Planning (1998), by Duane D. Baumann, John J.
Boland, and W. Michael Hanemann.
• Advanced Water Distribution Modeling and Management (2003), by Haestad Methods.
• Introduction to Urban Water Distribution (2006), by Nemanja Trifunovic.
• History of Water Distribution (2006), by Thomas M. Walski (AWWA Journal).
• Water Encyclopedia: 5 Volume Set (2005), ed. by Jay H. Lehr and Jack Keeley.
• Toronto's Water Efficiency Plan (2002), by Toronto Water.
• Distribution System Water Storage Tank Sizing: How Large is Too Large? (2006), by Baxter
& Woodman Consulting Engineers (ISAWWA-IWEA 2006 Joint Water Conference).
In order to determine an ADD for the system one must evaluate all respective water system uses
that contribute to the output. This quantity includes direct residential and industry consumption,
unaccounted water (such as leakage), fire fighting and other emergency requirements. The
quantification of leakage is highly dependent on metering or audits, or else its magnitude is merely
an educated estimate based on available data. Fire demands are derived from highly sensitive
surveys performed by the insurance industry. They are also often expressed as a percentage of the
MDD. Both of these factors essentially confirm the fact that this quantity comprises a level of
uncertainty (and therefore inherent risk), and that its deterministic use for design is merely a form of
simplification.
Many different techniques can be used to determine actual (albeit still deterministic) consumption
rates, and some of these include: analysis of historical data (billing or metering records), similarity
comparisons with other municipalities, unit consumption demands, extrapolation from regional
models/studies, and small scale surveys of individual and/or industry use. If historical data
37
(computerized or manual) is unavailable, unreliable or incomplete, then the most common practice
is to apply standard per capita consumption while at the same time taking into account different land
use patterns. This approach is simplistic in that it depends on population forecasting and on
predetermined unit consumption rates. Present population census data is fairly reliable, but future
population forecasting is inherently probabilistic in nature. It is therefore highly influenced by
random future events and factors that bring uncertainty to any resulting prediction. The unit
consumption rates are also only averages, and can drastically vary not only on the regional scale but
also on the local scale.
Unlike the ADD, the MDD is used as a design quantity for the purpose of meeting changes in
demand variations. This demand is intended to take into account that one specific day, or period of
days, when the demand is much higher than the average. For example, in North America this
demand may take into account that one hot day in July when many residents are watering their lawns
and yards. As conceptually simple as it may seem, such a typical day does not exist. In reality, years
do not simply repeat themselves, and this is now clearly evident with the ongoing effects of climate
change. As a result, such an estimate should be based on a random variable – that is, as a probability
of occurrence. This estimate is significantly influenced by events such as the one described above,
but also by events beyond a municipality’s or utility’s control or thought. These unexpected events,
whether government policy changes or unusual weather patterns, may also have a systematic effect
on the MDD quantity; an effect that this estimate is not built to take into account.
If historical data is available, its statistical analysis undoubtedly also yields a probabilistic output for
the ADD and the MDD. Unfortunately in both cases (historical or non-historical) and for both
demands, this quantity is always presented as a single deterministic value; a value that is then further
used to determine all other inputs characteristics. For example, a design flow may be prescribed as a
maximum of a) the MDD plus fire flow demand or b) the peak hour demand (PHD). If the
probabilistic MDD is added to the probabilistic fire flow, one cannot be fully certain of the variance
and risk contained within the ultimate design value. Similarly, the PHD also falls in this category as
it is either: a) derived from the same data as the ADD or b) a percentage of the ADD (via peaking
factors). The later case does often include a range, but is still dependent on the original ADD input.
The random nature of these quantities is almost always overlooked for the purpose of simplifying
the task at hand. For example, it is not unusual for a basic design handbook to note something
similar to the statement below:
38
"The daily, weekly, and annual cycles are never repeated in exactly the same way;
however, for design purposes a sufficient accuracy is achieved if it is assumed that all
water needs are satisfied in a similar schedule during one day, week, or year”
(Trifunovic, 2006).
Even though these concepts are definitely not new, the unfortunate truth is that most water systems
are still designed to meet regulations and guidelines; guidelines which still simplify the demand to a
single deterministic value as a function of time (i.e., system expansion or year).
While this variable is actually quite important in the overall analysis and design, it is often simply
overlooked as a fixed design quantity. The term “average day” is almost benign in that its history is
never discussed. It has been an idea and/or methodology that is rarely disputed or questioned, but
more importantly it has been used as a basic quantity without an associated level of variance (and
therefore probability). A lot of information and assumptions go into determining the ADD, yet its
final end use is extremely simplistic. Its justification may be simple or it may be complex, but it is
certainly not freely discussed or questioned. Similarly, the idea of MDD is conceptually simple as it
is supposed to take into account the system stresses that are beyond the expectations contained
within the average day. However, this quantity too is influenced by a variety of probabilistic and
unpredictable factors and events, and is therefore statistically vague. Without basing the design
approach on a stochastic framework, one must at least understand that the ADD and MDD (and all
demands derived from that point on) are probabilistic and hence that they carry an associated level
of uncertainty. As the previous statement indirectly implies, these simplifications are useful if and
only if, they are accurate and representative estimates of the true random nature of demand. In the
end, the key test is the sensitivity of decisions made with these assumptions. In the context of this
thesis, these assumptions feed the decisions that are derived from a subsequent hydraulic transient
analysis and design.
3.3 Current Design Approach
As discussed earlier, traditional engineering design has often relied on worst-case design loading;
loading which is typically set out by a regulated design code, guideline, or practice. Research,
innovation and technological improvements have made engineering design easier, but the basic
theory and design approach has unfortunately not changed much. Whether we are talking about
39
designing a bridge or a water transmission system, the least variable (i.e., the most restricted) aspect
of the design comes through the determination of loads. Once a general design concept has been
agreed upon, the detailed design is typically highly dependent on a pre-specified set of deterministic
loads such as weight, seismic quake, wind, pressure, flow, etc.
3.3.1 Steady State Design Loads
The design of a water system, whether it consists of components pertaining to supply, transmission
or distribution, has long suffered through the question of what loading is appropriate. Nowhere is
this clearer than in North America, where water is treated as a right and where as a result, a water
system’s required capacity is typically determined by a relatively arbitrary worst-case design load; a
load driven by a deterministic assumption or forecast of demand.
The determination of the worst-case load typically first requires the quantification of an average or
typical load (as discussed in an earlier section). Long-term population and industry forecasting is
used to predict the size of a future service area, and these extrapolations are then typically multiplied
by a per capita or per industry consumption rate. A standard consumption rate in North America
may be 300 to 400 litres per capita per day. Even though this approach can rely on historical
analysis of data, its ultimate application is purely deterministic. When the service areas are multiplied
by the deterministic consumption rates, a rather benign quantity is born: the ADD. This baseline
load directly feeds into all other design loads, including the previously discussed MDD, PHD, and
minimum hour demand (MHD). Once cast, it is a combination of these loads that ultimately
determines the size of the water system components such as water treatment plants, feedermains,
pump stations, reservoirs, and ultimately for the purpose of this thesis – the surge protection.
However, these design loads are also by their very nature variable, and therefore the conventional
design of a water system is left without any other option than to use a worst-case variation of this
load. With the diurnal nature of demand (and therefore pressure), the difference between PHD and
MHD can be many fold. Even more troubling is that original design loads are rarely confirmed or
monitored post construction and/or operation, and lessons are infrequently learned. As a result, the
conservative load based design of water systems has yielded a sleuth of under-used system
components that are more difficult and costly to operate. (A good example of such components
comes in the form of dedicated transient protection equipment, which is typically designed via
numerical hydraulic transient analysis techniques.) Filion (2006) summarized the worst-case
40
deterministic demand design shortcoming best through a synonymous explanation of a “longitudinal
analysis”; a term best explained via the following simple statement:
“Since a major goal in design is to provide an acceptable level of service throughout the life
of a system, it may be more useful to estimate the hydraulic reliability of a network over its
entire design life rather than the end of it”.
3.3.2 Transient Modeling Approach
The field of hydraulic transients has long been considered as a “black box” of a topic. The
traditional and true expertise in this field has often been limited to a few specialists and within other
fluid systems, and its importance is only now truly gaining widespread traction in standard water
transmission and distribution systems. The increased requirement for the considerations of
transients in design has quickly brought on the need for engineers to develop this expertise in order
to be competitive. At the same time, the advances in, and acceptance of, computer modeling has
never been higher. With these two driving factors, the field of transient analysis has quickly
progressed to the development of an unwritten design strategy – the worst-case scenario design with
the aid of a computer model.
A transient analysis of a water system requires the understanding of the system characteristics; an
understanding that is nowadays always summarized within a comprehensive steady state computer
model. It is the combination of this steady state hydraulic model and the previously discussed
deterministic design loads that drives the decision making and design process for water systems.
The proliferation of hydraulic modeling arose from the need to “predict” the future without
intruding on, or risking the performance of, a real life system. The advances in research and in
technology have spawned an entire branch of hydraulics which is simply referred to as modeling.
Whether or not we are talking about models for asset management, master planning, computational
fluid dynamics, transient analysis or system optimization, the purpose of any model is still only to
mimic nature through mathematics and to be able to do it in a non-intrusive manner. Every model
should have its own purpose and goal, and nowhere is this clearer than in hydraulic transient
modeling. However, this important prerequisite is quickly fading because the scale and complexity
of hydraulic models is rapidly increasing and because the North American industry is again quickly
shifting towards rehabilitation and optimization of existing infrastructure.
41
A standard transient analysis typically relies on a previously constructed steady state model that
contains all system characteristics pertaining to pipes, elevations, and boundary conditions. This is
probably the most overlooked step in a standard transient analysis. Given the traditional
deterministic nature of the steady state model outputs, the subsequent transient model inputs are
therefore themselves also assumed to be deterministic. Very little thought is actually given to the
uncertainties and variability of steady state modeling parameters such as roughness, leakage, demand,
reservoir levels, etc. Uncertainty in water distribution system parameters is usually ignored and the
most likely parameter value is selected and used in the analysis and design (Grayman, 2005). The
best one can hope is that the steady state model has been calibrated or at least partially calibrated for
a current design year. These steady state model parameters in turn affect other more critical
hydraulic transient values such as the initial velocity.
All transient numerical modeling packages solve a set of two governing non-linear partial differential
equations. The two most accepted and used procedures for solving the equations are the before
mentioned MOC and WCM models. Depending on the code, the steady state model might have to
be simplified in order to reduce the run-time of the transient model simulation. For example, a
boundary intensive WCM solution typically requires orders of magnitude fewer calculations and is
often much quicker in solving large systems with many pipe lengths and nodes (Wood et al., 2005).
This type of transient analysis approach relies on the pure belief that computer power can overcome
anything, and often leads to individuals performing detailed transient analysis on asset management
models in excess of 30,000 links and nodes. The ultimate question comes down to whether or not
the results are believable and/or realistic when it comes to actual in-situ (or future) field conditions.
The model simplification process is typically referred to as skeletonization for surge purposes; the
rules for which are relatively unclear and often contradictory of the general hydraulic equivalency
theory. Skeletonization can be performed manually or through the aid of computer algorithms,
however this theory is predicted based on steady state equilibrium and does not consider the
implications of pressure wave interactions throughout the system and at important boundary
conditions (Jung et al., 2007). Furthermore, it is difficult to simplify a model for all the worst-case
transient conditions, especially if the simplification precedes any analysis (Karney and McInnis,
1990). Once the steady state model for the surge analysis is finalized, assumptions and estimates on
several key transient related parameters have to be made. These may relate to: pump run times,
pump and motor inertias, acoustic wavespeeds, vapour pressures, valve characteristics, etc. These
42
parameters are often uncertain and are also used as surrogates for other more variable and non-
modelled parameters such as air content, true energy dissipation, unsteady friction, sediment
content, etc. Even if all of the system characteristics and parameters are determined and/or
assumed, the final question for a transient analysis still comes down to what to design the system
for, and at what load?
Unlike steady state loading that is in the form of system demand and therefore supply, transient
loads (or events) arise in different forms through different sources, and it is therefore very difficult
to consider all potential risks. Nevertheless, a power failure event that impacts the flow conditions
has often (but not always) been found to be the most consequential of all transient events. Some
will argue that this event has achieved such a lofty status only since it is more likely to occur, but also
since its physics are easier to model within a numerical software package.
Power failures must be considered inevitable over the long life of a water and wastewater system,
because they are by their very nature unpredictable events. Thus the consideration of a power
failure precludes a more direct avoidance of its occurrence and substitutes as a worst-case surrogate
for potentially less damaging routine or non-routine pump and valve operations. Nevertheless,
power failures can occur at almost any operating condition, for which the variables may include: the
magnitude of the system demand and/or flow, the water level in reservoirs and tanks, the number of
pump stations operating, the number of pumps operating per pump station, and the status of
dedicated surge protection equipment. As systems and their representative models increase in size
and complexity, the number of design scenarios that arise from these numerous variables quickly
becomes overwhelming. As a result and an example, a computer model aided transient analysis may
typically consider a combination of scenarios based on the following options (i.e., variables):
i) Global system power failure or local pump station power failure;
ii) PHD or MHD (i.e., reservoir/tank filling); and
iii) Current or future year system characteristics (e.g., system layout, flow rates, etc.).
These transient analysis design variables lead to eight distinct scenarios and therefore to at least eight
possible design loads. The number of permutations can significantly increase depending on the
nature of the change in system characteristics. For example, if an analysis must consider a capital
43
construction and phasing plan consisting of three new watermains, then the analysis may have to be
repeated for each phase of construction and therefore operation.
The design year and system demand typically determines how many pump stations and individual
pumps would operate during such a scenario, and therefore establish the conditions during which
the surrogate power failure event can occur. The ultimate surge protection recommendations and
design are then made on the worst-case level of risk from the scenarios considered. To get to this
ultimate design load, several predictions and assumptions have already been made on steady state
system demand, system characteristics, steady state system parameters, transient parameters and
transient events. Since real systems contain many unrepresented dissipation mechanisms, this
approach can lead to conservative and over designed systems which are rarely monitored or learned
from. In the end, a transient analysis using numerical modeling is only as good as its model inputs;
inputs which are truly not deterministic and therefore highly variable. Unfortunately, most practical
design procedures are not of an academic nature, and cannot afford the budget or the time to
consider all likely events. In other words, a comprehensive sensitivity analysis is typically not an
option. Lastly (and as is discussed in Chapter 7), even if the loads for a hydraulic transient analysis
can be determined, the industry stills struggles with quantifying the frequency of this loading.
Conventional detailed design is often thought of as being relatively restrictive, in that the decisions
made at this stage have significantly less impact than those made earlier on in the process.
Nevertheless, a detailed design that is driven by deterministic or worst-case loads can also have a
significant impact on the future cost and the performance of any water (or wastewater) system. The
deterministic design of these systems and the subsequent worst-case load design for transient
conditions have a potential to significantly increase the long-term cost through inadequate,
inefficient and/or highly conservative system components. Furthermore, these designs are rarely, if
ever confirmed or monitored. Instead, a surrogate in the form of computer aided hydraulic
modeling is seen as the optimal solution for design and rehabilitation. With the complexity of the
systems and their respective hydraulic models increasing, a new and complimentary approach for
transient analysis and design of existing water system components is needed. The most logical base
for the new approach is to learn from actual system evolution, and to use this knowledge to
determine future levels of risk.
44
3.3.3 Wastewater Design Considerations
The previous subsections explicitly focused on some of the shortcomings of the current hydraulic
transient analysis and design practice in municipal water systems. Since pressurized wastewater
systems typically also fall in the realm of standard hydraulic transient analysis, they too require
dedicated attention. Wastewater systems are quite similar to water systems in that their fluid media
is after all, mostly water. The difference in the fluid media is primarily through the gas and sediment
content. However, the difference in the design and analysis of wastewater systems is based on other
system properties such as layout, operating methodology, and most importantly, design loads.
Unlike in the case of water systems, the operation of wastewater systems is not driven by the end
user or by downstream system demands, but rather through the supply, or more properly referred to
as the sewage “generation rate”. The generation rate consists of the actual sewage produced within
the upstream drainage area, and the external inflow and infiltration (I & I) that results from wet
weather flow (WWF) conditions. As a result, pumped wastewater systems are designed on the basic
premise that the pump station storage should be not exceeded. In other words, the pump station is
designed to convey the always changing inflow in order to ensure a maximum utilization of its
storage element’s volume. Therefore, actual wastewater pump station flow rates (and thus the
system flow rates) are always changing. The result is that sewage pump stations almost always
operate intermittently, and that the peak flow rates are only achieved during WWF conditions. In
other words, most wastewater systems are not optimally designed for regular (DWF) conditions, bur
rather for the peak (WWF) events.
Pressurized (i.e., pumped) wastewater systems are typically much simpler than water distribution
systems in that they are more often than not comprise a single pump station and a single (or
twinned) forcemain. The forcemain(s) typically discharge to a gravity sewer or outfall, via either an
open atmosphere manhole or a vortex suppressor. In the latter case, back pressure on the system
from the downstream end is greater.
Systems such as the ones described above are typically designed on two steady state level criteria.
The pump station itself is designed for a maximum discharge capacity that is governed by the
estimated peak factored WWF inflow rate. The forcemain itself is designed on the same possible
peak flow rate vis-à-vis minimum and maximum velocities in the pipe (variable throughout design
45
standards, but typically between 0.6 m/s and 2 m/s). These steady state design criteria are then
extended to the transient analysis; an analysis that then determines the protection requirements via
the worst-case discharge load. The worst-case system load is typically modelled in such a way that
the discharge (i.e., the flow) rate is maximized. A peak flow rate and therefore velocity usually (but
not always) yields the greatest potential transient magnitudes, if a power failure event scenario is the
design event of choice. In order to achieve the maximum flow rate, the models are typically set with
the maximum number of pumps in operation and with the lowest system head losses. In order to
achieve the minimal system head loss, minor head loss is ignored and the friction head loss terms
such as the Hazen Williams C-factors are set at values equivalent for brand new pipes (e.g., 140 or
150). The shortcoming in this approach is that most peak flows are rarely, if ever, achieved within
the first part of the system’s life. When (or if ever) the peak design loads are actually achieved, the
system has aged significantly, and the friction in the pipe (especially in true sewage fluid media) has
increased. The conclusion here is that, most hydraulic transient analysis loads (and therefore
designs) are highly conservative, especially if one also considers the actual likelihood of a
hypothetical design scenario, such as the following:
A power failure of all pumps will occur when the system is newly constructed, when the
friction losses are minimized, and when the peak design WWF inflow rate is maintained
for a sufficient period of time such that the pump station outflow (i.e., discharge) rate is
equal to the peak inflow rate.
In addition to the conservative design loads, the current practice for numerical transient analysis of
wastewater systems typically assumes that the sewage fluid is essentially identical to pure water.
While mostly true, such an assumption is a poor one in the context of composition, sediments,
wavespeed, vapour pressure, and negative pressure potential.
Besides H20, wastewater also contains gases (e.g., air and hydrogen sulphide) and sediments that can
not only affect the capacity of the forcemain, but also its transient behaviour and performance. The
most direct transient impact of additional gases is the inherent reduction in the wavespeed. As
shown in Figure 3-1 below, the addition of air into a pressurized system can have a dramatic impact
on the assumed wavespeed.
46
If not properly considered, a typical and reasonable assumption for the value of wavespeed that is
based on the pipe material (e.g., 1000 m/s for rigid pipe and 350 m/s for flexible pipe) can be very
dangerous on two fronts. First, it can drastically impact the predicted maximum and minimum
transient pressure envelopes, as a lower wavespeed significantly affects the pressure wave
characteristic. Second, in most cases a lower wavespeed would also yield a longer transient pressure
oscillation time and therefore it can significantly affect design and operational protocols that are
based on timing. For example, if a standby period following a power failure is predicted based on a
typical working assumption such as 10 L/a, where “L” is the length of the forcemain and “a” is the
wavespeed, then a lower in-situ value for “a” can significantly underestimate such a prescribed
standby time. A detailed numerical and physical example of this notion is presented in Chapter 6.
Figure 3-1: Wave Velocity v. Air Content (Wylie and Streeter, 1993)
The other mentioned assumption pertains to the vapour pressure (i.e., the boiling point) of the fluid.
While this parameter is dependent on both elevation and temperature, it is also dependent on the
physical composition of the fluid. For a wastewater system, the typically assumed value of 0.4 m
H20 may not be suitable, and therefore this typically requires a proper sensitivity analysis. An
example of such an analysis for a wastewater system is shown in Figure 3-2.
47
Figure 3-2: Sample Vapour Pressure Sensitivity Analysis for a Wastewater System
In this case, the maximum transient pressures at certain points in the system were significantly
dependent on the vapour pressure when it approached a relatively high vale of 5 m H20. While it is
a distraction to this thesis to consider this system in detail, these numerical results do imply that an
important sensitivity exits in practice that is typically neglected in analysis. In summary, the assumed
value for the vapour pressure can potentially affect the predicted transient pressure envelopes, and
therefore a distinction between the modeling of a water and wastewater system must be made on
this account.
3.4 Lack of Confirmation and Monitoring
Chapter 2 establishes an introduction on the brief history of numerical hydraulic transient modeling
and also provides insights into some numerical modeling limitations, risks and shortcomings. The
key point raised is the often theoretical nature of such an exercise, and therefore the need for
improved field data derived validation, confirmation, and monitoring. This section provides specific
examples of some of these shortcomings and risks, in order to justify the need for more transient
field pressure monitoring. The section is broken down into two separate subsections: i)
confirmation, validation and calibration, and ii) performance monitoring.
48
3.4.1 Confirmation, Validation and Calibration
Of what use is a model if it does not represent reality, or, at best, is blindly assumed to do so? While
the answer should be obvious, the (unfortunate) reality is that uncalibrated theoretical models are
often used in most detailed transient analyses and designs. As the previous sections have shown, a
numerical transient analysis is typically at the mercy of someone else’s (e.g., an owner’s) hydraulic
steady state model; a model that is typically not calibrated and may also be out of date. The direct
result of this fact is a poor starting base for a transient model. Nonetheless, for the purpose of this
section, the steady state model limitations are set aside and attention is given to additional and
subsequent transient model limitations.
Ideally, all theoretical models such as hydraulic transient models should be calibrated to match the
actual field conditions. In simple terms, model calibration can be defined as model adjustments that
result in a better match between actual and predicted system performance. In reality, the calibration
process is often difficult to achieve due to limitations in time, money and the quality of data.
Furthermore, the calibration of a hydraulic model is actually not a simple process and is often open
to error in application and judgement. As Walski et al. (2000) noted in the Water Distribution
Modeling Handbook, the subjective adjustment of system parameters to match a calibration
objective can lead to models being improperly calibrated. From the point of view of a transient
analysis, the word calibration is therefore quite misleading. Such an exercise is almost always
difficult to perform, and the end objective is almost always difficult to achieve. Comprehensive
hydraulic transient calibration techniques, including those such as genetic algorithms and inverse
transients, are still considered as academic and are limited to simple systems. Why then is the word
“calibration” always used to describe the action of correlating field data to numerical modeling data?
The answer is simple – there is a reassurance in seeing or hearing the word “calibration” before or
after the words “numerical modeling”. The inherent assumption here is the obvious belief that
models will likely always continue to just be theoretical and mathematical representations of reality.
In the realm of hydraulic transient analysis, the process of calibration should be subdivided into
three (3) distinct objective levels: calibration, validation and confirmation. The term “calibration”
should be reserved for the traditional comprehensive exercise in which the steady state and transient
conditions are completely matched to a wide range of system performance criteria and operating
conditions. The term “validation” should be used to refer to basic transient model adjustments such
49
as those pertaining to key parameters such as roughness, wavespeed, moment of inertia, etc. The
term “confirmation” should simply be used to refer to field observations for the purpose of
assessing transient performance. In summary, the following field and numerical data comparison
levels should be adopted:
Confirmation – the use of field data to access risk and performance,
while using the model as a starting base.
Validation – the adjustment of specific and sensitive model
parameters for the purpose of ensuring some
agreement between the model and the field data.
Calibration – the use of specific comprehensive techniques(s) to
adjust model performance across a wide range of
operating conditions.
The full calibration level is very difficult to achieve and is almost always not warranted for most
analysis and design. However, the validation level of field work should always be set as the target
for any detailed transient analysis and design, in which the chosen primary approach is theoretical
numerical modeling. Unfortunately, this type of model validation is rarely performed and important
system decisions are solely made without the physical understanding and connection to the actual
system. To that end, the omission of the numerical and field data validation step in a transient
analysis can lead to a variety of errors and risks, including those pertaining to inadequate design and
extremely conservative design. A partial list of some potential errors and risks is summarized below:
• Actual versus model predicted steady state operating conditions (i.e., flows and pressures);
• Actual versus assumed acoustic wavespeed values, especially in systems containing a
significant amount of air (e.g., wastewater);
• Actual versus model predicted energy dissipation mechanisms and levels, including actual
contributions from interconnected systems (e.g., distribution network);
• Actual versus model predicted, calculated, or manufacturer prescribed pump and motor
moment of inertia values;
• Effects of non-steady friction on transient response;
Complexity & Accuracy
50
• Actual versus model predicted or assumed valve timing and operation (e.g., SRV closure
times and set-points);
• Actual versus model predicted or assumed pump control timing (e.g., VFD ramp-up times
or pump control valve closure times);
• Actual versus model predicted transient event durations and their impact on design
recommendations (e.g., required post-event standby timing);
• Actual versus model predicted positive and negative transient pressure envelopes;
• Nature and impact of check valve slam;
• Actual versus model predicted negative pressure durations; and
• Much more.
The above list of considerations for a transient analysis illustrates the importance of performing
supplementary field work up to a minimum level required for validation. Without this, the
numerical modeling approach to hydraulic transient analysis is incomplete and risky. Chapter 6
provides explicit examples of numerical and field data validation using real life systems.
3.4.2 Performance Monitoring
This second subsection continues with the goal of confirming the need for transient based field
investigation and correlation. While the first subsection focuses on the traditional confirmation of
numerical models, this subsection focuses on the actual confirmation of system performance. Once
a system has been analyzed and designed, and once recommendations have been implemented, the
traditional approach has been to react only if a significant problem arises. In other words, the
hydraulic transient performance of systems is rarely monitored in the long-term, and previous
decisions are rarely, if ever, re-evaluated in light of future in-situ conditions.
Surge protection devices, such as those described in Chapter 2, are designed based on information
available at the time of analysis. Since all hydraulic systems are dynamic, such assumptions may no
longer hold true in the future as the system conditions evolve. As a result, it is logical to assume that
the adequacy of the original surge protection should be periodically revisited as the system ages.
Surge protection devices, equipment and strategies can comprise a significant capital cost in the
original system design, and therefore they should always be monitored for their actual in-the-field
51
performance. Unfortunately, this is currently rarely done on a long-term performance assessment
basis. Primary surge protection equipment such as HACs or SRVs are inspected and maintained
based on prescribed intervals, but these maintenance plans are only geared towards ensuring that the
equipment is operational. In practice, little is actually done to monitor the performance of
prescribed transient protection and this is mostly due to the short-term frame of mind approach that
is often adopted by the owner, engineer and operator. Just like in the case of numerical model
validation, hydraulic transient field investigation strategies such as continuous pressure monitoring
can be (and should be) used to access the long-term performance of both the prescribed surge
protection, and the entire system. Examples of the possible benefits of continuous transient based
performance monitoring are as follows:
• Assessment of pre and post surge protection transient conditions;
• Assessment of equipment/device degradation and transient protection performance;
• New equipment calibration during commissioning stage;
• Recording and review of unexpected transient events and risks;
• Review of the impact of rare operational events protocols such as draining or filling;
• Determination of faulty surge protection equipment (e.g., non-performing air valve);
• Assessment of the quality of long-term air management;
• Recording of specific transient event frequency (e.g., power failures);
• Assessment of operational protocols following specific transient events;
• Information for improved valve timing;
• Long-term deterioration of pump performance, including vibration, trip frequency and cavitation;
• Frequency of unexpected transient events (e.g., power failure);
• Typical pump operating schedules;
• The nature and timing of the pump discharge valve operation;
• General operational protocols, including valve operation and pump changes;
• Steady state pressure changes (e.g., difference between pump(s) on and off);
• Long-term changes in the extent of transient wave energy dissipation across system;
• Long-term calibration of changing numerical models; and
• Much more.
52
The above list of possible benefits for transient based performance monitoring illustrates the
importance of revisiting original designs and the ease by which system problems can be identified
before they pose a more significant risk to a system. The second half of Chapter 5 provides a few
sample statistics that can be derived from continuous transient pressure monitoring and Chapter 7
proposes a preliminary transient risk assessment methodology that is also derived from such an
exercise. The following chapter takes a brief aside towards the need for more field work by looking
into potential water quality issues arising from poorly performing transient protection systems.
3.5 Summary
The main purpose of this chapter is to expand on the previously established understanding of
hydraulic transients and hydraulic transient analysis approaches. The chapter begins by discussing
how numerical transient modeling is significantly dependent on the ever expanding field of hydraulic
steady state modeling. This discussion provides examples of the key assumptions and parameters of
steady state modeling and demonstrates how some of these could significantly impact any transient
analysis considerations that subsequently follow.
The chapter proceeds to provide a discussion on two of the most often overlooked hydraulic
assumptions and design considerations; the concepts of system demand and design loads. The
discussion centers on the industry’s general acceptance of vague and arbitrary demand quantities and
terms, and how these quantities are typically then used in combination with scenario based transient
loading to conduct a hydraulic transient (i.e., risk) analysis. The chapter concludes with a critical
discussion on the lack of actual system (i.e., in-situ) monitoring, both from the point of numerical
model calibration and overall system performance and risk assessment. In summary, the chapter
establishes the overall need for an increase in field based consideration of hydraulic transient
performance and risk for water and wastewater systems.
53
Chapter 4 Water Quality Issues
4.1 Background
The advances in the general knowledge and/or awareness of transients, as well as the advances in
field based analysis, has given rise to a variety of important topics. One of these interesting topics,
albeit non-hydraulic in nature, is that of water quality during low pressure events. In this area of
study, hydraulic transients present themselves as a risk that is not in the form of excessive system
pressures. However, this chapter of the thesis provides a brief aside and illustrates how transients
can pose a different type of risk in the form of potable water contamination.
Legislation must be both proactive and active in the prevention and minimization of the risk that is
associated with the contamination of potable water in transmission and distribution systems. One
of the often unconsidered and/or discarded water quality concerns is that pertaining to pathogen
intrusion during negative transient pressure events. A negative pressure within a pipe is defined as a
pressure that is below atmospheric or zero gauge. During such an event, the pipe and the fluid
inside the pipe is exposed to the risk of outside contamination; contamination that comes in the
form of pathogen intrusion. Pathogen intrusion has been extensively studied and legislated, but the
risk still exists due to the uncertainties pertaining to chlorine residuals. An excellent study into and
recommendations against general pathogen intrusion in a distribution system is presented by
Kirmeyer et al. in the 2001 AWWARF project and publication no. 436.
The risk for pathogen intrusion during negative pressure conditions can, and has previously been
introduced through hydraulic transient events. As is shown through modeling and field
observations in the subsequently documented chapters, routine and non-routine operations of water
systems can continuously subject the potable water to low and negative pressures. As a result, the
overall risk of pathogen intrusion is simply increased during a low pressure transient event.
Several studies and publications have looked into the potential routes of pathogen intrusion during
transient events. One of the most extensive reviews of such conditions has been performed over
the years by the Ecole Poly Technique in Montreal. Using the isolated water network of the City of
Laval, the researchers conducted several transient field tests and water sampling in conjunction with
54
several other organizations, including AWWARF. The best analysis of the risks of pathogen
intrusion during transient events is presented in a 2007 Ph.D. dissertation by Marie-Claude Besner.
As a quick background summary, the two most critical routes for pathogen intrusion during negative
pressure events are through:
1. Soil contamination via a high groundwater table; and
2. Contamination through flooded air valve chambers.
As part of transient pressure monitoring that was conducted in the Region of Peel (see Chapters 6
and 7 for more detail), a subset study was also conducted in order to determine the nature and risk
of contamination through flooded air valve chambers. Air valves are important system devices that
are unfortunately often misunderstood, misrepresented, and therefore misused (Radulj, 2007). The
purpose of an air valve in a water system can be three-fold: to routinely vent air during regular
system operation, to exchange air during transient conditions, and to expel air during filling. All
three of these air valve operating conditions require that the pipe and system be exposed to the
environment. As a general note, most air valves are typically installed within dedicated valve
chambers, and in most cases at locations at which the pipe profile is a high point.
The previously mentioned risk of contamination is only present in the case where the pressures in
the pipe are lower than that of the external atmosphere. In other words, the risks exist only if the
internal pressures are negative. Flooded air valve chambers provide an easy pathway for
contaminated water to enter the potable water system during the period of time at which the internal
pressures are negative. The level of risk is a function of the magnitude of this negative pressure
(e.g., partial vacuum to full vacuum), its duration, and the frequency of occurrence. This overall risk
usually comes in the form of pathogen intrusion.
Unlike the slow process of soil contamination via a high groundwater table, pathogen intrusion
through flooded air valve chambers can be and has been shown to be quick. The severity (i.e.,
frequency and concentration) of the pathogen determines the risk of the contamination, because
most potable water systems in North America are operated with a certain level of chlorine residual.
If the pathogen concentration is significant, the system is at a higher risk of intrusion. The case
study presented at the end of this chapter analyzes several flooded air valve chamber water samples.
The pathogens considered in this analysis are briefly discussed in the following section.
55
4.2 Important Pathogens
The following section provides a brief description of the likely and potential pathogens that can
pose a risk to a water system through flooded air valve chambers. The description of the following
organisms is derived from a variety of sources, including Viessman et al. (2005), MWH (2005) and
Fraser et al. (2010): Total Coliforms, Escherichia Coli, Enterococci and Clostridium Perfringens.
Total Coliform
Total coliform refers to coliform bacteria from three origins: feces, soil and other origin. Coliform
bacteria are indicator organisms in public water supplies and are evidence that the water supply has
been contaminated by human or animal (i.e., warm-blooded) feces. Coliform bacteria are used as an
indicator organism because laboratory analysis for pathogens is difficult to perform and for some
pathogens it is impossible to perform. If a sample of water tests positive for total coliform it is
subsequently tested for the presence of fecal coliform and Escherichia coli.
Escherichia Coli (E. coli)
E. coli is a bacterial pathogen that belongs to the Enterobacteriaceae family. Escherichia coli live in
the intestinal tract of warm-blooded animals and are normally beneficial to the human body because
it suppresses the growth of harmful bacteria and even produces vitamins. There are varieties of E.
coli, however, that are pathogenic in human beings. E. coli can enter a water system through runoff
containing animal feces, and evidence suggests that E. coli may potentially also survive and grow in
distribution system biofilms.
Enterococci
Enterococci are bacteria that are found in the intestines of both humans and animals and they
generally do not grow in the environment except in tropical climates. They are a normal part of the
intestinal flora but are also the cause of some serious infections. Enterococci bacteria are anaerobic
organisms that can grow in many different environments and have been shown to survive longer
than E. coli.
56
Clostridium Perfringens
Clostridium Perfringens is a micro organism that grows both on food and in the environment. The
organism produces a toxin that if ingested can produce non-inflammatory gastroenteritis. Because
their spores persist for long periods in the environment, C. perfringens can result in false positives
and may be less suitable as an indicator of recent fecal contamination.
4.3 Case Study and Discussion
With the above concerns in mind, a water sampling and testing study was undertaken in the Region
of Peel in order to determine the nature and risk of pathogen intrusion. Several air valve chamber
locations were chosen in consultation with the operations staff and with previous condition
inspection reports. The sampling locations were chosen from a master list of problematic air valve
chamber locations. This master list was narrowed down to a shorter list of accessible chambers (i.e.,
chambers with easy access, not at a busy road, etc.); most of which were then visited and inspected
prior to the sampling. The goal of this selection process was to find air valve chambers that were
often flooded.
In total, standing water was sampled and tested at ten (10) air valve chambers in the Region of Peel,
and these locations are shown in Figure 4-1. Given the time constraint associated with the sampling,
shipping, and testing, the majority of the chosen chamber locations were concentrated in the East
side of the Peel transmission line system. The sampling took place on September 29 and 30 of 2008,
and six (6) 1 L sample bottles were collected at each of the locations. The weather conditions during
the first day of sampling were dry, but the day did follow a week of rain. The second day was
extremely wet, but most of the samples were taken in the early morning and prior to the rainfall.
The collected samples were kept cold and shipped via same day delivery to the Institut Armand-
Frappier (INRS) in Laval for testing. This is where the previously mentioned studies originated, and
hence why their valuable experience was drawn upon. The samples were all tested for the four (4)
previously discussed pathogens. The water quality test results are shown in Table 4-1, and are
compared with a similar study in the City of Laval (courtesy of Marie-Claude Besner) in Figure 4-2.
57
Figure 4-1: Air Valve Chamber Water Sampling Locations
58
Table 4-1: Air Valve Chamber Water Sampling Test Results
Figure 4-2: Air Valve Chamber Water Sampling Test Result Comparison
Sample IDTotal
ColiformsE. coli Enterococci
Clostridium
perfringens
HN2100 Chamber # 1A 16,500 50 3,110 6
BS1500 Chamber # 2 1,100 <5 395 <2
BS1500 Chamber # 8A 1,950 255 38 <2
BS1500 Chamber # 8 165,000 <5 10 <2
EB1050 Chamber # 16 12,000 <5 6 <2
BS1200 Chamber # 15 650 <5 6 <2
NB900 Chamber # R1 21,500 5 170 <2
NB900 Chamber # 8B 450,000 135 1,240 4
NB900 Chamber # 9B <5* <5* 120* <2*
SV1050 Chamber # 13 1,365,000 4,500 4,260 2
- All results are in cfu/100ml
*Turbid sample, possible interference
59
The above results are typical of very dirty storm water runoff; runoff that may or may not contain
animal fecal matter. The tested pathogens are good indicators of fecal coliforms and therefore may
not be the best indicators for other organisms and viruses. Expert review of the above results has
preliminarily concluded that the risk does exist, but that more comprehensive studies should be
performed. This has, and will continue to bring the risk attention to water system legislators. It
should not be a surprise if local authorities impose their own legislation that directs that all air valve
chambers be maintained in a dry state.
The selected pathogens would most likely be disinfected by the chlorine residual, if proper chlorine
residual is present and if mixing is sufficient in the pipe. However, if these pathogens are indeed
present in these concentrations, then other more dangerous pathogens may be as well. Some of
these other pathogens cannot be disinfected by chlorine residuals. Furthermore, if these other
pathogens have this same intrusion pathway (i.e., through air valves during negative pressures), then
this water quality risk is undoubtedly present. The above nature (i.e., locations and timing) of
sampling is in no way scientific, comprehensive or conclusive, and therefore it may not be indicative
of the risk across this system, or other systems. Many factors such as weather, location, chamber
condition can affect the test results. As a result, a more comprehensive investigation is highly
recommended in order to remove any liability associated with not being proactive, but more
importantly, to protect the public and those dependant on the municipal water supply.
As additional proof of the inherent risk and uncertainty, these results can also be statistically
compared to the previously mentioned Laval study, and to typical raw water and wastewater results.
These comparisons (again courtesy of Marie-Claude Besner) are presented in the following figures,
and are broken down by pathogen. The high variability in the Peel results is indicative of the
relatively small sample size, and the larger size and complexity of the system.
4.4 Summary
This chapter provides a brief aside to the transient analysis theme of the thesis, but does so with a
transient risk and field work related topic pertaining to pathogen intrusion into potable water
systems through flooded air valve chambers. This general topic of pathogen intrusion during low
pressure transient events has been at the forefront of current research and this chapter demonstrates
through a case study in the Region of Peel that the typical risk of intrusion through air valves is
60
indeed quantifiable, albeit at the same time relatively mild due to the surface runoff characteristics of
the standing water.
Figure 4-3: Comparison of Total Coliform Concentrations
Figure 4-4: Comparison of E. coli Concentrations
T ota l col i form s
M edian
25%-75%
Non-Outl ier Range T CLaval T CPeel
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
To
tao
l co
lifo
rms (
log
CF
U/1
00
ml)
E. co l i
M edian
25%-75%
Non-Outl ier Range ECLaval ECPeel ECRaw ECWW
0
1
2
3
4
5
6
7
8
E.
co
li (lo
g C
FU
/10
0m
l)
61
Figure 4-5: Comparison of Enterococci Concentrations
Figure 4-6: Comparison of Clostridium Perfringens Concentrations
En teroc oc ci
M edian
25%-75%
Non-Outl ier Range EntLaval
EntPeel
EntRaw
EntWW
0
1
2
3
4
5
6
7E
nte
roco
cci (lo
g C
FU
/10
0m
l)
Clostrid ium perfringens
M edian
25%-75%
Non-Outl ier Range CPLaval CPPeel CPRaw CPWW
0
1
2
3
4
5
C.
pe
rfrin
ge
ns (
log
CF
U/1
00
ml)
62
Chapter 5 Advances in Field Work
5.1 Traditional Field Work Approach
The general acceptance of, and the movement towards numerical hydraulic modeling has been partly
due to the advances in the accuracy, computational speed, and overall user friendliness of numerical
software packages. Another key and important driving force towards this change has been the
traditional difficulty (and at times the cost), of performing the desired and/or required field work.
Field work has generally been difficult to perform due to several reasons. First, it always has and
still does, require proper planning and scheduling. The scheduling task requires the co-operation of
multiple parties, including the system owner, the system operator, and other consultants and/or
researchers. Without this, field work is typically either rushed or significantly delayed, and if delayed,
it is typically always hurried during the most critical stage. Second, field work is always at the mercy
of the actual in-situ conditions. For example, if the work is being performed outside it is likely to be
affected by the weather. Weather conditions such as extremely low temperatures or excessive
rainfall can make even a simple data logger installation difficult. Similarly, even if the work is
performed indoors (e.g., within a pump station), it is always dependent on the physical and operating
conditions of that day. Physical conditions may include the lack of access, power source, pressure
taps, and dry space. The operating conditions can dictate what can and cannot be done. For
example, only a certain number of pumps may be operated when the water level in a downstream
reservoir is high. All of these general constraints have made the historical and the current field work
practice challenging.
Traditional or historical transient pressure monitoring relied on a variety of instruments that were
not only bulky, but also complicated. Some examples of these include transducers whose signal was
captured by an electronic transfer box and then recorded by a chart reader. While accurate, a device
such as this was not easily portable or installed. Furthermore, every individual component requires
proper and individual calibration. Lastly, all of the individual components also require the
transmitted signals to be individually synchronized and correlated by the receiving units.
63
Advances in technology, and more specifically in electronics, has clearly made a positive impact
towards the reduction in the size of these units. The modern pressure monitoring units typically
comprise three key components: a pressure transducer, a data logger and a power source. The
pressure transducers are also often referred to as electronic pressure gauges, pressure sensors,
pressure transmitters, etc. However, these pressure transducers are specifically designed for the
recording of transient pressures, in that their speed of recording (typically expressed as a frequency
in Hz) is higher than normal. Nonetheless, the modern sensors and loggers have allowed for a
much simpler and less invasive installation within a water (or wastewater) system, and have therefore
re-enabled the previously forgotten option of assessing the transient pressure conditions of a system
through the use of field work. The following section describes and compares the commercially
available modern transient pressure monitoring equipment, for the purpose of identifying a
technology capable of assisting in the long-term performance evaluation of fluid systems.
5.2 Modern Pressure Monitors and Comparison
The transient pressure monitor market is not significant in size, and therefore most technological
advances in this area have been as a result of either research or individual built-for-purpose
application of a proprietary idea. The advances in these technologies have arisen via different and
generally more profitable industries, including heating, cooling, and transportation. These other
industries have traditionally experienced the need for high frequency and real time data.
Nonetheless, there currently exist several prominent instrumentation manufacturers that officially
produce and market transient pressure monitors for a wide range of industries. Table 5-1 and Table
5-2 present a two-page comparison summary of the most popular transient pressure monitors that
are currently available on the market for the general use within the water and wastewater industry.
The comparison is made based on the goal of continuous long-term pressure monitoring.
The comparison tables include the following pressure monitoring units, each comprising a column:
Pipetech TP-1 (Pipetech), Omega CP-PRTRANS (Omega), Radcom RDL1071 L/3 (Radcom),
Telog HPR 31i (Telog), Madgetech PRTRANS1000 (Madgetech) and Cla-Val X142PT (Cla-Val).
While marketed differently, the Omega and Madgetech units are actually the same, and therefore will
only be referred as Omega from here on. The following paragraphs briefly describe and discuss
each of the pressure monitor features (i.e., the rows), and ultimately attempt to determine which unit
is best for what type of application.
64
Table 5-1: Pressure Monitor Comparison Summary
65
Table 5-2: Pressure Monitor Comparison Summary (cont’d)
66
Manufacturer
The manufacturers of transient pressure monitors range from small research companies to large
scale water and wastewater equipment providers and instrumentation companies. While all of the
manufacturers have distributors around the world, all but Radcom (U.K.) are actually headquartered
in the U.S.A. There is no clear advantage with respect to name or location.
Base Price
The base price comparison includes the minimum number of components required to properly
install and run the equipment. This price in Canadian dollars typically includes the pressure
transducer or sensor, data logger, power source, data download (e.g., cable) and software. The
Omega and Cla-Val monitors come in under the $1,000 mark, while the Radcom and Telog loggers
are essentially priced in the $2,000 to $3,000 range. The Pipetech unit clearly does not win in the
price category, with its $6,000+ price point. One point worth noting is that the listed prices are
those as quoted by local distributors. As a result, the actual price from the original manufacturer
(i.e., without the middle man mark-up) would likely be less. Nonetheless, the price advantage is to
Omega and Cla-Val.
Varieties, Options and Installation
The purpose of this category is two fold. First, it is to list and/or acknowledge additional (i.e., non-
basic) options and models. Second, its intent is to also list unique and important benefits and
complications for actual field installations. Most units come with an optional submersible sensor,
while only some of the units come with submersible data loggers. This option is important for field
installations within chambers that are, or can potentially become, flooded. The Pipetech unit has an
advantage with the additional options, but due to its bulky size, it is also more difficult to install and
store. The Telog unit has the advantage with its easy hydrant connector feature.
Sensor Information
While most of the units can probably be reconfigured with any pressure sensor or gauge, this
category compares the basic sensor that is provided with the unit. The sensor construction is highly
variable and can include ceramics, silicon, and strain gauges. While the analogue gauge for the
Radcom is out of date, there is no clear advantage for the rest. A proper comparison requires a
long-term performance analysis.
67
Sensor Range
A variety of pressure sensor ranges can be purchased for most units, and these should typically be
chosen for the specific application. For example, since the accuracy is based on the full pressure
range, a high pressure range sensor may not be accurate enough for low pressure applications. The
key aspect of this category is whether or not the sensors can record negative (i.e., sub-atmospheric)
pressures. The advantage in this case is to the Pipetech, Telog and Cla-Val units.
Sensor Accuracy
As partially noted above, the accuracy of the pressure recording is dependent on the range of the
sensor. The accuracy of the sensor is typically given as a percentage (plus or minus) of the full scale
or range. Therefore a typical 0.25% value for the sensor accuracy is less important for smaller
pressure ranges. With its lower 1% accuracy, the Cla-Val unit is at a disadvantage to the others;
other units between which there is no clear distinction.
Data Logger Features
This category examines additional data logger features that are not separately analyzed. While not all
of these are listed, the Pipetech, Telog and Cla-Val have inherent positive features that make them
slightly better than the other two.
Data Download
The recorded and stored data is typically downloaded with a wired connection; a connection which
every one of the units is capable of. However, the Pipetech also has a wireless download feature via
a portable PDA that gives it a partial advantage over the others. The advantage is only partial,
because the wired connection option for the Pipetech is via a difficult computer LAN connection.
The Omega and Cla-Val units have an advantage with the easy to use USB connection and
download, which is more modern and faster than the serial connection of the other two units.
Software and Data Type
All software is capable of properly opening and displaying the data. However, every one of the
software packages have unique issues that make them difficult to use at times. Nonetheless, the
advantage in this category goes to the Pipetech unit because the software converts the stored data to
a database (rather than text file) format. Databases are excellent ways of storing a large number of
68
data points. While such a system may not be the simplest for short duration recordings, it is indeed
the best for long-term continuous pressure monitoring.
Pressure Units
The pressure readings are recorded as signal outputs from the transducers, and the data loggers then
store the data in different forms. In the end, the most important aspect of this category is whether
or not the software package can display a variety of units. The Pipetech unit is at a clear
disadvantage with its psi (gauge) only display, while the Radcom and Telog units have the advantage
with the user defined unit feature which allows the data to be presented in any pressure unit.
Recording Frequency
The frequency of recording is what distinguishes transient pressure monitors from the regular and
standard pressure transducers. High frequency recording is required in order to properly capture the
properties of the transient pressure waves and events. If the frequency is not high enough, then the
resulting pressure profiles will not be detailed enough, and would possibly also miss recording the
maximum and minimum transient pressures. The recording frequency should be distinguished from
the sampling frequency (or rate). While a monitor can sample at a high frequency, it may not
actually be able to record or store the data at such a rate. With respect to absolute magnitudes, the
Pipetech and Omega units have a clear advantage because of their 100 Hz (i.e., 100 times per
second) recording frequency. Similarly, the Cla-Val unit is limited to only 8 Hz (8 per second). The
last point worth mentioning is that a high frequency such as 100 Hz is not required for most water
and wastewater applications, since a lower frequency such as 20 Hz can typically provide the same
level of detail. Furthermore, certain intermediate frequency ranges can be subject to electrical
interference from other sources at the location of installation. Such things as fluorescent lighting
and pump vibrations can interfere with the pressure transducer signals and cause erroneous readings.
High Frequency Control
The most important feature for a transient pressure monitor during long-term transient pressure
monitoring is the capability of only recording the transient pressures at the high frequency. While
continuous high frequency recording is neat and interesting, the quantity of the resulting data is
excessive. This leads to extremely large data files, and longer processing and analysis times. As a
result, a transient pressure monitor should ideally be able to record at two different rates: slowly
during periods of low pressure fluctuation (i.e., steady state) and fast during periods in which the
69
pressure is unsteady (i.e., transient). This category evaluates the frequency control logic. The clear
winner is the Pipetech unit for its moving window statistical analysis approach. The frequency
control for most of the other units is based on a pressure band, range, or window; set points which
are all quantified on a single starting value of pressure. The following two sections break this
frequency control capability down to start and stop record triggers.
Start Record Trigger
The start record trigger is the type of control logic that tells the data logger when to record at the
desired high frequency. As mentioned earlier, the logger should only record at high frequency
during the period of the transient event. Most start record triggers are based on absolute pressure
differences (negative, positive or both). However, the different monitors determine the absolute
pressure difference start points on a wide range of characteristics, including: current pressure values,
range of previous pressure values, specific number of previous pressure samples, and preset time
period of recording. Some of the units also contain start record triggers based on pressure bands.
Nonetheless, the Pipetech unit is the clear winner in this category because of its statistical approach
that allows the user to set the start record triggers based on both absolute pressure and standard
deviation. These metrics are calculated based on a user prescribed number of previous samples.
Stop Record Trigger
While most units are capable of reasonably distinguishing between a ”regular” pressure and a
pressure that meets some of the start record trigger criteria, most of the units are actually not
properly capable of stopping the high frequency of recording at the end of the transient event. For
example, the Cla-Val and Radcom units simply continue to record at the high frequency once the
start record trigger has been met and maintained. The absolute pressure value stop record trigger
for the Omega unit does not take into account that future background (or steady state) pressures
may not actual be the same as those during the initial stage. While, an improvement, the predefined
time period based stop record trigger for the Telog unit requires the user to understand the duration
of a transient event prior to witnessing it. As a result, the Pipetech unit also wins in this category
due to its standard deviation based stop record trigger. Nonetheless, even this trigger is not
foolproof in that it is dependent on the other previously mentioned statistical settings. A standard
deviation that is based on a set of high frequency recordings may be subject to premature stoppage
at a point at which the transient pressures are actually in the vicinity of the background pressure
70
magnitudes (i.e., of same magnitude but still transient in nature. (What is really needed here is a
standard deviation based stop record trigger that also takes into account a minimum duration.)
Other Set Points
This category examines other record features that do not necessarily fall within the start or stop
control logic. In addition to the Pipetech unit, the winners in this category are also all the other
units except for Radcom. The most useful feature is that of a timed and/or delayed start for the
recording.
Power Supply
The properties of the prescribed power supply determine the length of time for which the unit can
be installed in the field. The specific Lithium Ion batteries for the Omega, Telog and Cla-Val units
are not only portable and built-in, but also last for a significant duration. Most of these can last in
excess of a year with a medium frequency range of recording. Units such as the Cla-Val are easily
charged via an USB connection. The Pipetech unit is at a significant disadvantage due to its
significant power requirement. While it is typically powered via DC power, its bulky backup battery
options are limited in their recording duration.
The above comparison summary has served to provide an understanding on the features that the
modern suite of transient pressure monitors is manufactured with. While an ideal unit does not yet
exist, each of the listed units has their own advantages and disadvantages. Some units are better
suited for short-term transient pressure tests, while others for long-term continuous transient
pressure monitoring. Some units are easier to install while others are easier to calibrate and/or
maintain. Nonetheless, for the sake of comparison, Table 5-3 ranks the five (5) different transient
pressure monitor units based on the previously discussed and compared categories. The ranking is
based on the single purpose of continuous pressure monitoring and is rather subjective through
direct and anecdotal experience of the author.
The units with the best feature in each category was assigned a value of three, the worst with a value
of one, with those in between being assigned an intermediate value of two. The rankings are in no
way scientific since the categories were rather arbitrary and at times repetitive, and because the
categories were equally weighed. Nonetheless, even with its excessive cost, bulky size and significant
power requirement, the Pipetech TP-1 can be said to be the best commercially available monitor for
71
long-term continuous transient pressure monitoring of water and wastewater systems. On the other
hand, the Radcom RDL1071 L/3 significantly trails behind the rest of the units. Overall, if the data
control logic and features of the Pipetech TP-1 could be combined with the ease of use features
from the Cla-Val or Telog units, the industry would then have close to an ideal device for
continuous transient pressure monitoring. (It is likely that such a device already exists, but that it is
not commercially available.)
As a cautionary note, the author admits to a partial (yet unintentional bias) that derives from
different degrees of direct and indirect experience with the five listed units. As a result, the author
accepts no responsibility for the subjective ranking of these units at this time. Nonetheless, while
different, each unit can and does serve a purpose in some form of transient pressure monitoring.
For example, if the investigation goal is to record transient pressures during a quick field test, then a
transient pressure monitor capable of recording at high frequency for the entire time may be all that
is required. In summary, this comparison and ranking is made solely on the requirement for long-
term transient pressure monitoring.
Table 5-3: Transient Pressure Monitor Ranking for Long-Term Applications
72
5.3 Pipetech TP-1 Transient Pressure Monitor
As previously discussed, the Pipetech TP-1 Transient Pressure Monitor, from herein only referred to
as TP-1, is the best commercially available unit on the market for the purpose of continuous long-
term transient pressure monitoring of water and wastewater systems. This section describes the
recording control logic and features of the TP-1 device in order to provide a better context for the
long-term transient pressure monitoring data, analysis, and discussion that follow.
The TP-1 device and associated equipment is specifically designed to provide detailed transient
pressure profile information, including maximum and minimum pressure magnitudes, event
duration, and event character (including phase, etc.). The TP-1 equipment comprises several key
components that are required for a typical installation within a pump station:
1. High-Frequency Pressure Transducer (Sensor) - capable of detecting pressures
between -14.7 psi (-10.4 m H20) vacuum and 500 psi (351 m H20), at rates of up to
approximately 100 Hz. The sensor connects to the TP-1 logger through a wired
connection cable.
2. TP-1 Smart Data Logger and Control Box - capable of monitoring the pressures
measured by the sensor while only recording when prescribed (minimum) statistical
start and stop triggers are met. It also allows for wired data download using a laptop,
wireless data download using a PDA, or external remote access via the internet.
3. Deep Charge Battery and/or UPS - provides power to the control box, even during a
system power failure. Two batteries can be connected in parallel to provide extended
and/or sole power for a remote installation such as a transmission line chamber.
4. PDA Unit - for changing record settings and time, and for downloading the data from
the control box.
5. Wi-Fi Antenna - for transmitting the signal to the PDA.
6. Manual Pressure Gauge - for calibrating the TP-1 unit.
7. 2-in-1 Modem and Router – for remote access connection via the internet.
73
The above discussed components of the TP-1 monitors are shown below through sample field
installation photographs.
Outdoor (Left) and Indoor (Right)
100 Hz Pressure Transducer
Manual Pressure Gauge
Back-Up Battery Power
TP-1 Control Box
Figure 5-1: TP-1 Transient Pressure Monitor Components
Figure 5-2: Typical TP-1 Installation in the Field
74
The simplest (and often most beneficial) installation of a TP-1 is achieved at a water or wastewater
pump station. The overall benefits of an indoor installation are significant, and they include: easy
access to a primary power source, ease of entry, protection against unfavourable weather and field
conditions (e.g., frost or flooding), etc. Within a pump station, a TP-1 is typically best installed on
either the common discharge and/or suction header pipe. These common pressure locations are
typically key for determining the timing, magnitude and frequency of transient events caused by
pump operations, pump changes, and power failures. For example, a common discharge header
location is better than an individual pump discharge pipe location because it is downstream of the
check valve(s). As a result, such a location enables the recording of all transient pressures in the
system, even if a specific pump is not in operation.
In an ideal installation, the pressure transducer should be installed at the bottom (or side) of the
pipe, where gasses such as air cannot build up and subsequently affect the transient pressure
measurements through a dampening effect. If such a tap location is unavailable, the air in the
transducer connection line must regularly be bled out. This requires that the transducer not be
connected in a direct loop, in which the air cannot be manually released. In other words, a T-
connection such as that shown in one of the previous figures is best.
Most sensors are factory pre-calibrated, while others must be re-calibrated following the installation
especially if previously used. Upon a successful installation, the following steps are taken for the
quick calibration of each unit:
1. Isolate the transducer using the fittings and release all pressure.
2. Set the TP-1 pressure in the monitor to zero using the PDA.
3. Reintroduce the transducer to the actual system pressure.
4. Bleed all air from the fittings.
5. Using a manual gauge (or regular digital gauge if installed) as a benchmark,
adjust the TP-1 pressure accordingly in the PDA.
6. Synchronize the TP-1 time with the preset time on the PDA.
A true (i.e., initial) calibration of any pressure transducer should expose the transducer to a wide
range of pressures, including the minimum and maximum pressures that each unit is capable of
75
recording. This is typically best achieved using a hydraulic hand pump, and through a step-wise
calibration procedure, as shown in a pressure versus time plot in Figure 5-3.
Figure 5-3: Sample Calibration Pressure Profile
Each pressure transducer is connected to the TP-1 control box, which dynamically calculates the
average pressure detected by the sensor and then uses this information to determine the rate at
which it logs the readings. The following parameters are the user-defined control features, and are
illustrated through the PDA’s input interface in Figure 5-4:
• Start Record: indicates whether the TP-1 should record at the high-frequency rate as
soon as it starts (manual) or only when a deviation from the background pressure is
experienced (auto);
• High Frequency (msec): the rate at which transient events are recorded;
• Start Threshold (SDx10): indicates the amount that the pressure can deviate from the
background level standard deviation before being considered as a transient pressure
(and therefore event);
• Absolute Difference (psi): indicates the minimum change in pressure required for the
consideration as a transient pressure (and therefore event);
76
• Stop Record: indicates whether recording of transient events stops after a defined
interval (manual) or is controlled by threshold and absolute difference values (auto);
• Manual Record Time (sec): indicates how long transient events are recorded for, if
the Stop Record is set to manual;
• Stop Threshold (SDx10): works
similarly to the start threshold, but
indicates when the high frequency
recording should stop;
• Background Sample Rate (msec):
indicates how frequently the
background pressure is calculated;
• Background Record Rate (sec):
indicates the rate at which the samples
are recorded while the pressure is
within the background levels; and
• Number of Samples Averaged:
determines the sensitivity of the
background pressure to normal
fluctuations.
The proper selection and calibration of the above settings and parameters is crucial because it can:
1. Eliminate random pressure noise associated with steady state (background)
pressures; pressures that are always changing;
2. Ensure that all significant transient events are recorded;
3. Ensure that all transient events are recorded from the beginning to the end;
4. Eliminate the need to always record at the highest frequency and sample rate,
thereby reducing the size of the data file(s); and
5. Minimize the amount of power being drained from the battery supply (if solely
battery powered).
Figure 5-4: TP-1 Pressure Monitoring Equipment Settings
77
The following table summarizes two samples of the initial settings that were used for two pump
station installations in the Region of Durham. These settings were ultimately determined through
trial and error during the first week of installation and were subsequently adjusted on an individual
installation basis.
Table 5-4: Sample TP-1 Pressure Record Settings
Setting/Parameter Ajax WSP Harwood PS
Start Record Mode Auto Auto
Start High Frequency Rate (msec) 100 100
Start Threshold (SD X 10) 10 10
Absolute Difference (psi) 6 5
Stop Record Mode Auto Auto
Manual Record Time (sec) 180 5
Stop Threshold (SD X 10) 2 2
Background Sample Rate (msec) 500 500
Background Record Rate (sec) 120 120
Samples Averaged 50 50
Once the TP-1 units are calibrated, and after the data has been recorded and downloaded, it must be
analyzed. In order to analyze the data, a graphical software package by the name of QAnalyze is
used to produce graphs of Pressure (psi) versus Time (date and time). The blue data is classified as
steady state (i.e., background) pressures and the red data is classified as transient (i.e., unsteady)
pressures. The above calibration settings are the main variables (albeit, user defined) that determine
the type of pressure and therefore the colour of the plot. More specifically, if the settings are too
sensitive then the data will be mostly red (transient). If the data is not sensitive enough, then the
data will mostly be blue (background). The key is to prescribe the settings so that all significant
transient events are recorded in full, and so that minor noise in steady state operation (i.e., plus or
minus a few psi) is not recorded. The following two figures show the difference between sensitive
and less sensitive settings, for two different locations.
78
Figure 5-5: Pressure v. Time Plots for Different TP-1 Settings - Sensitive (above), Less Sensitive (Below)
Similarly, the scales for the Pressure and Time axes can be manually adjusted (i.e., zoomed in and
out), as to allow for a more detailed representation of any specific transient event. For example, the
following four figures show the same event, but in progressively greater detail (i.e., shorter time
frame and smaller scale).
79
- Longest Time Period (Top Left) to Shortest Time Period (Bottom Right)
5.4 Sample Data
This section aims to provide actual sample field data consisting of transient pressure monitoring
results. All sample pressure series graphs shown in this section were acquired in the field using the
previously discussed Pipetech TP-1 transient pressure monitoring technology. The graphs present
the recorded pressure in the units of psi, as per the recording logic described in the previous section.
This section is divided into the following three (3) subsections:
i) Long-Term Histories – Samples of transient pressure recordings over several
days and/or months;
ii) Individual Events – Samples of transient pressures recordings during individual
routine or non-routine events; and
iii) Unique Events – Samples of transient pressure recordings during unique system
events.
Figure 5-6: Pressure v. Time Plots for the Same Event but with Different Time Scales
80
5.4.1 Long-Term Histories
Long-term transient pressure histories can provide excellent insights on the daily, monthly and yearly
operational patterns of water and wastewater systems. The TP-1 pressure monitoring equipment is
ideally suited to record transient pressure histories by actively delineating between steady state and
transient pressures, thereby conserving the recording memory. Transient pressure history plots (via
the inherent transient pressure database recordings) can be used to assess the true long-term
performance of systems by recording a variety of different types of events, including routine or non-
routine, planned or unplanned, and frequent or infrequent. In the end such data can be used to
complete a transient risk assessment, and as discussed in Chapter 7, can also be instrumental in a
proposed metric such as the Transient Risk Index (TRI).
Figure 5-7 presents a 5-month transient pressure history at Hanlan PS in the Region of Peel,
Ontario, Canada. Hanlan PS is an older station in a pressure zone that is currently experiencing a
high rate of water demand growth. The combination of its age, increased flow output, transmission
system connectivity, and lack of dedicated transient protection, makes this station a high transient
risk location within the Peel system. This can easily be observed in the 5-month pressure history; a
history that contains a significant number of transient events (shown in red). The normal (i.e.,
steady state) operating pressures at the discharge point typically varies between 55 psi and 65 psi
(depending on the number of pumps in operation), but in this time period the positive transient
pressures often exceeded 80 psi and the negative transient pressures often reached a significant level
and duration of partial vacuum conditions. Hanlan PS is currently in the process of a capacity and
equipment upgrade, one which will also see the addition of a significant HAC volume. The addition
of HACs will act to control the rate, magnitude, and frequency of transient pressure fluctuations,
and therefore the transient pressure monitoring will be able to provide a clear comparison of pre
and post upgrade conditions.
Figure 5-8 presents a sample 4-month transient pressure history at another key pump station in the
Region of Peel, called Lakeview. The Zone 2 system of the HLPS at the Lakeview WTP is the
source of the highest pressure and flow rate in the Peel system, and would therefore theoretically be
at a high risk due to transients. Fortunately, the station is protected by HACs with a 400 cu. m
volume, which act to control and dampen any significant pressure fluctuations arising from transient
events such as power failures, demand changes, and routine pump operation. When compared to
81
the Hanlan PS plot in Figure 5-7, the sample transient pressure history at Lakeview can be classified
as being less severe, especially if one considers the more complex hydraulic conditions and the fact
that y-axis scales for the two plots shown are actually different. The Lakeview HLPS Zone 2 system
typically operates at normal pressure range of 160 to 170 psi and it rarely experiences any significant
positive or negative transient pressure fluctuations.
Figure 5-9 presents a shorter 1-month pressure history at a much smaller pump station in the City of
Ottawa, called Glen Cairn. Glen Cairn PS feeds a relatively smaller (and lower demand) pressure
zone than either Hanlan or Lakeview; a zone which is also located at the end of the City’s hydraulic
system. In combination with its direct connection to the distribution system, the lower flow output
of this station generally places it in a lower transient risk category. The pressure history at Glen
Cairn is significantly different than those of Hanlan or Lakeview, thereby further confirming the
assertion that no two hydraulic systems are the same. If they were, then their typical transient
pressure response over the long-term would be more similar.
Figure 5-7: Sample Transient Pressure History – Hanlan PS (Region of Peel)
82
Figure 5-8: Sample Transient Pressure History – Lakeview WTP (Region of Peel)
Figure 5-9: Sample Transient Pressure History – Glen Cairn PS (City of Ottawa)
83
Figure 5-10 and Figure 5-11 present a transient pressure history at the La Caldera PS in Mexico City,
Mexico. La Caldera is a large pump station that conveys pumped well field water from a receiving
reservoir over a large hill and to the distribution system. Its single pipeline and high static head
place it a high risk to transient pressures. The first figure presents a short 1-month pressure history,
and the second figure provides a sample 2-day pressure history. The majority of the frequent and
similar transient pressure events are a result of operator initiated pump changes. The inflow to the
station is highly variable due to an unreliable well field supply. When combined with the single fixed
speed pump size, this unsteady inflow makes for a difficult operation; an operation that requires
repetitive cycling between 1 and 2 pumps in order to ensure that the reservoir does not overflow
and/or that it does not completely drain. The larger transient events are due to frequent power
failures, and these events can yield positive transient pressures that are 60% greater than the normal
operating pressures. Transient field pressure monitoring such as this can not only yield valuable
information on the magnitude and frequency of all transient pressures, but it can also provide
insights into unique and unexpected events such as check valve failures, etc. (please refer to the
following subsections).
Figure 5-10: Sample Transient Pressure History – La Caldera PS (Mexico City) – 2 Months
84
Figure 5-11: Sample Transient Pressure History – La Caldera PS (Mexico City) – 2 Days
Figure 5-12: Sample Transient Pressure History – Mixquic Well No. 13 (Mexico City)
85
Figure 5-12 presents a 1-month pressure history at Well Pump No. 13 in the Mixquic well field of
Mexico City. A pressure monitoring location at a well field can often record a variety of transient
events; events whose origin is often difficult to determine due to the sheer number of well pumps in
the system. A transient event induced at one well location will have a different pressure profile than
an event induced at a different well, but nonetheless, all transient events will be felt across the entire
well field due to its hydraulic interconnectivity. Transient pressures at this location (i.e., at Mixquic
Well No. 13) are often amplified due to its location at an upstream dead-end of a well water
collector branch pipeline. For more detailed and sample transient events at this particular location,
please refer to the following subsections.
Figure 5-13 presents a 3-month transient pressure history at a key location in the Region of Durham,
Ontario, Canada. The pump station at the main Ajax WSP water supply source is directly connected
to the distribution system. As discussed in Chapter 2, distribution systems can act to de-fragment
and partially dissipate transient pressure wave energy, thereby minimizing the magnitudes of any
transient events. As shown in the Ajax WSP figure, the discharge pressures are relatively stable and
typically range between 100 psi and 120 psi in accordance to the number of pumps in operation and
the flow rate (both of which are a function of system demand). Figure 5-14 shows a sample 3-day
pressure history at the same location. This type of pressure history is synonymous with a diurnal
demand pattern of water distribution systems in that the demand changes induce a change in the
operation of the supply system. This in turn leads to a pump change; a pump change which is
achieved by a transient pressure event that moves the system operation from one steady state to
another.
86
Figure 5-13: Sample Transient Pressure History – Ajax WSP (Region of Durham) – 3 Months
Figure 5-14: Sample Transient Pressure History – Ajax WSP (Region of Durham) – 3 Days
87
5.4.2 Individual Events
Long-term transient pressure histories comprise extended durations of normal (i.e., steady state)
operating periods, and depending on the location, a certain number of transient pressure
fluctuations referred to as transient events. This subsection provides sample field pressure profiles
for a variety of individual transient events.
Figure 5-15 presents a sample pressure profile for a typical controlled pump start-up, as recorded at
the Ajax WSP in the Region of Durham, Ontario, Canada. In this case, a constant speed pump was
started against a mostly closed discharge control valve (e.g., butterfly or gate valve). This type of
start-up subsequently relies on the slow and continuous opening of the discharge valve and enables a
smooth transition from one operating condition to the next. As shown in the graph, this transition
is smooth due to the length of time (> 1 minute) required to establish the new flow rate. The end
result is the avoidance of the rapid pressure change (i.e., a transient event) that is typically associated
with an uncontrolled pump start-up. Transient field pressure monitoring can therefore be used to
adequately test and configure pump start-up protocols by adjusting the speed and duration of the
discharge valve opening.
Figure 5-16 presents a transient pressure profile for a typical pump shutdown at the same Ajax WSP
location. The pump shutdown is similar to the previous pump start-up in that it relies on the use of
the discharge control valve. In this case, the transient event comprises two distinct phases: a smooth
pressure reduction followed by an unsteady pressure fluctuation consisting of a rapid pressure drop
prior to the final valve closure. Since the last 10% of a valve closure is the most significant, the
current valve closure settings at this location are clearly not optimized. The field transient pressure
monitoring can be used to identify and improve such operating protocols, thereby minimizing the
frequency and magnitude of more routine pump stops.
Figure 5-17 presents a transient pressure profile for another pump shutdown event. In this case, the
pressures are recorded at a high point of the La Caldera pipeline in Mexico City, and the routine
pump shutdown is not aided by discharge valve control. As a result, the initial downsurge pressure
wave is more pronounced and it is also observed at the downstream location.
88
Figure 5-15: Typical Pump Start-Up – Ajax WSP (Region of Durham)
Figure 5-16: Typical Pump Shut-Off – Ajax WSP (Region of Durham)
89
Figure 5-17: Typical Pump Shut-Off at High Point – La Caldera Pipeline (Mexico City)
Figure 5-18: Typical Pump Switch – Harwood PS (Region of Durham)
90
Figure 5-18 presents a pressure profile for a routine pump switch at the Harwood PS in the Region
of Durham. This controlled operation consists of a routine pump shutdown immediately followed
by a routine start-up of a different pump. The smooth and curved nature of the transient pressure
change is indicative of a sufficient valve closure and opening duration, and is a trademark of good
pump control. Figure 5-19 and Figure 5-20 present a transient pressure profile for a power failure
induced pump trip at two different types of systems. The first pressure profile was recorded at the
pump station of the Ajax WSP in the Region of Durham; a station that is directly tied into the
distribution system. The second pressure profile was recorded at Hanlan PS in the Region of Peel; a
station that comprises a significant transmission only system. In addition to the obviously
significant magnitudes of the power failure induced downsurge events, the two figures provide a
clear example of how different the positive upsurge reflections can be. In the Ajax WSP case, the
reflected upsurge is essentially non-existent – primarily due to the added benefit of the distribution
system based pressure wave fragmentation and energy dissipation. In the Hanlan PS case, the lack
of distribution system attenuation leads to a more pronounced upsurge event; an event that can
impose a more severe consequence. In both cases, the transient pressure monitoring identified the
presence and character of the power failure induced transient risk.
Figure 5-21 presents a transient pressure profile for a sequence of planned emergency pump trips at
the La Caldera PS in Mexico City. The planned pump trip sequence was implemented as part of
planned transient field tests in order to determine the risk and magnitude of power failure induced
transient events. The first transient event is a result of a single pump trip from 2 pumps down to 1
pump. The second transient event is a result of a single pump trip from 1 pump down to 0 pumps.
As expected, the second transient event is more pronounced in both the magnitude and duration
than the first, and this is due to the range of the flow rate change. In this particular case, the first
pump change reduced the flow rate from approximately 800 L/s down to 500 L/s, and the second
pump change further reduced the flow rate from 500 L/s to 0 L/s. The complete stoppage in the
flow requires a longer period of pressure adjustment, thereby inducing a longer duration of pressure
wave propagation. In this case, transient pressure monitoring provided insights into both the
character of a pump trip induced transient event(s), as well as the duration of pressure fluctuation
following such an event. This type of data can also be used to estimate the transient event standby
period and the actual in-situ wavespeed. The almost perfect cyclic and attenuating nature of the
pressure fluctuation is characteristic of a simple transmission type water system.
91
Figure 5-19: Power Failure Event without Upsurge – Ajax WSP (Region of Durham)
Figure 5-20: Power Failure Event with Upsurge – Hanlan PS (Region of Peel)
92
Figure 5-21: Multi Pump Shut-Off, From 2 to 1 to 0 Pumps – La Caldera PS (Mexico City)
5.4.3 Unique Events
The previous section provides examples of typical transient pressure events in water systems,
including both planned and unplanned, and routine and non-routine events. For example, while a
power failure induced transient event can be rare and is not considered routine; it is rarely unique or
unexpected. Power failures are by their very nature unpredictable, but likely. This section provides
examples of unique transient events or overall system changes that induce unique transient pressure
responses.
Figure 5-22 presents a 1-month transient pressure history at the La Caldera PS in Mexico City. In
this pressure history, a long duration pressure change is clearly evident. The system depressurization
is due to unplanned system maintenance that resulted in the complete shutdown and dewatering of
the hydraulic system. The long-term continuous transient pressure monitoring was able to record
this unplanned event, and was subsequently used to review the draining and filling protocols
(pressure details for which are not shown).
93
Figure 5-22: Complete System Shutdown – La Caldera PS (Mexico City)
Figure 5-23: Single Well Shutdown – Mixquic Well No. 4 (Mexico City)
94
Figure 5-23 presents a transient pressure history for an unknown well shutdown at the Mixquic well
field in Mexico City. In this case, the entire well field was taken out of service in order to
troubleshoot its operation. The lack of pressure (i.e., 0 psi gauge) downstream of the check valve
pump location indicates that the well line was indeed depressurized. Figure 5-24 shows the pressure
profile for an unexpected but progressive check valve failure at the previously discussed La Caldera
PS. The check valve began leaking in the early hours of the morning, prior to finally bursting and
flooding the facility. The complete system shutdown and restart took place in excess of 1.5 hours.
The check valve failure was progressive and not catastrophic, and the continuously increasing
leakage acted to dissipate the self-induced transient pressures. This is an indirect example of how
system based leaks act as natural pressure relief points.
Figure 5-25 presents a discharge pressure history at the North Richmond Hill PS, located in the
Region of York, Ontario, Canada. The pressure history illustrates a clear change in the typical pump
station operating pressure profile at the 1/3 mark of the graph. Pump schedules and modes of
operation were investigated, but these indicated that no significant change took place at the pump
station. Upon a detailed system investigation, it was determined that at this exact recorded point in
time, a pressure zone boundary PRV was adjusted in the downstream distribution system. The
adjustment of the PRV induced a significant change in the system curve, and therefore shifted the
operating point of the constant speed pumps at the station. The end result was an obviously more
significant transient response during routine pump operations. This required the revision of the
pump control logic, including the discharge valve opening and closing durations. A change such as
this would likely not have been observed without the continuous transient pressure monitoring.
Figure 5-26 presents a discharge pressure history at the Montreal Road PS, in the City of Ottawa,
Ontario, Canada. In addition to the obvious transient pressure fluctuations, this plot illustrates a
significant change in the operating mode. The middle section corresponds to a one month period in
which the mode of operation was changed from VFD control to pressure control. In the pressure
control mode, a constant speed pump (which is equivalent to a VFD controlled pump at max speed)
is used to feed a closed pressure zone. The evident response is characterized by a higher discharge
pressure and a greater fluctuation in the steady state pressures. The wider range of fluctuation is a
direct result in the loss of precise control that a VFD typically provides; control which is required to
accommodate diurnal demand changes in the closed (i.e., without storage) pressure zone.
95
Figure 5-24: Progressive Check Valve Failure – La Caldera PS (Mexico City)
Figure 5-25: Distribution System PRV Adjustment – North Richmond Hill PS (Region of York)
96
Figure 5-26: Pump Control Logic Change – Montreal Road PS (City of Ottawa)
5.5 Sample Statistics
5.5.1 Event Summaries
Continuous transient pressure monitoring over a long-term period can yield a long list of transient
events that represent a host of planned and unplanned operation activities. For example, one year
of pressure monitoring may provide between 500 and 2000 events that would subsequently need to
be reviewed, and if found to be serious and/or questionable, would also have to be correlated to
system operations via logbook and SCADA records. Table 5-5 presents a summary list of all
unplanned (and possibly unique) recorded transient events at the La Caldera PS in Mexico City, over
a period of approximately 5-months. The table lists the specific events along with their date and
time of occurrence, as well as four (4) key pressure profile characteristics: original steady state
pressure, minimum transient pressure, maximum transient pressure, and final steady state pressure.
97
A list such as this one needs to be compiled and then correlated to the pump station operations and
the activity during those days and times.
Table 5-6 presents a similar transient event summary list, as recorded at the Tlahuac and Mixquic
well fields in Mexico City. Every significant event (i.e., an event that is large in magnitude or
duration or simply unique) is presented, along with its minimum and maximum transient pressures
at six (6) different pressure recording locations. The locations represent different well stations
across the well fields. If the hydraulic transient behaviour of the system is at least partly known (i.e.,
via experience or via modeling), a simultaneous event comparison at different locations can provide
a significant amount of information on the source, duration, and dissipation of transient events.
Table 5-7 presents a short list of the most critical transient events that were recorded in the Region
of Peel’s water system during the summer of 2008. The table not only provides additional
information such as the total event duration and the duration of negative pressures (if applicable),
but it also correlates the specific events to SCADA records and system operation. For example, the
transient event that dropped the pressures at the ST1500 Chamber 9 location to -8 psi for a duration
of 13 seconds was caused by a simultaneous trip of three operating pumps.
98
Table 5-5: Unplanned Transient Event Summary – La Caldera PS (Mexico City)
* All pressures are in units of kg/cm2
99
Table 5-6: Transient Event Pressure Summary – Tlahuac/Mixquic Well Systems (Mexico City)
100
Table 5-7: Detailed Event Summary – Region of Peel System
5.5.2 General Event Statistics
In addition to identifying the details (e.g., properties and causes) of individual transient events, long-
term transient pressure monitoring can provide useful and interesting statistics on the types and
properties of all transient events recorded. Table 5-8 provides a simple list of event statistics at
eight (8) well field locations in Mexico City. While the sample size (i.e., duration of pressure
recording) is relatively short, the simple statistics do provide insights into the different levels of
pressures experienced in the system, as well the frequency of the formation of negative pressures.
Table 5-9 presents statistics for 5-months of transient pressure monitoring at the La Caldera PS in
Mexico City. Standard statistical parameters such as the mean and standard deviation are not only
useful towards determining the level of risk that hydraulic transients impose on the system, but can
also be used in the future in order to continuously track the evolution of the system. In other
words, this particular system can be further monitored indefinitely in order to determine the change
No. TP1 LocationSource
StationDate Time
Average
Pressure
(psi)
Min
Pressure
(psi)
Max
Pressure
(psi)
Total
Event
Duration
Negative
Pressure
Duration
Detailed Operation and
Information
5 Streetsville PS Streetsville July 15 11:25 AM 66 22 78 3:23 N/A
6MV900
Chamber 8Streetsville July 15 11:26 AM 34 5 47 3:23 N/A
16HG1500
Chamber 11Lorne Park July 22 2:54 PM 40 10 51 0:09 N/A
1430 - # 2,4,5 - OFF;
1455 - total shutdown;
23ST1500
Chamber 9Herridge July 31 9:11 AM 16 -8 26 1:13 0:13
BEFORE - #1, 6, 8, 9 ON;
0910 - #1 (23 MLD) OFF, #8 (36
MLD) OFF, #9 (36 MLD) OFF;
0915 - #6 (68 MLD) OFF; #2 (23
MLD) ON, #5 (68 MLD) ON
27Beckett's
Sproule PS
Beckett's
SprouleAug 7 12:54 AM 62 7 73 1:22 N/A
BEFORE - #1 (27 MLD) ON, #2
(46 MLD) ON, #14 (90 MLD) ON;
#11 (90 MLD) ON
0055 - #11 OFF, #4 (46 MLD) ON
28 Hanlan PS Hanlan Aug 7 1:20 PM 59 -6 80 2:43 0:11
29BS1500
Chamber 8Hanlan Aug 7 1:21 PM 29 -24 81 2:43 0:40
33 Silverthorne PS Silverthorne Aug 17 4:34 AM 65 15 74 2:20 N/A0400-0455 - Failure, Gen's ON; #
1,3,4,5,6 TRIP
BEFORE - #3 (45 MLD) - ON
1125 - #6 (55 MLD) - OFF
AFTER - #7 (90 MLD) - ON
BEFORE - #1 (23 MLD) ON, #3
(35 MLD) ON, #5 (110 MLD) ON,
#9 (136 MLD) ON, #10 (136 MLD)
ON;
1425 - # 3, 5, 9, 10 FAIL / OFF, #
6 (126 MLD) ON, #7 (110 MLD)
ON but FAIL
101
in such statistical parameters; a change that can possibly indicate system deterioration or a shift in
operational protocols. In the short-term, such statistics can be used to find answers to specific
transient pressure related questions. For example, why was the mean number of events per day in
July significantly less than in the rest of the months? This is counter intuitive because July is
typically a peak demand month in the Northern hemisphere, and therefore an initial hypothesis may
be that the system was limited in its operation, thereby reducing the likelihood of a transient event
occurring. (The answer in this case was actually that the performance of the upstream well field
pumps was consistent, and as a result the inflow to the La Caldera pump station was steadier,
meaning that the number of emergency pump operations was minimal.)
Table 5-8: General Short-Term Statistics – Tlahuac & Mixquic Wells (Mexico City)
Location ID
# of Days of Data
Average Pressure (kg/cm2)
Maximum Pressure (kg/cm2)
Minimum Pressure (kg/cm2)
Negative Pressure
Count
T1 25 4.5 8.2 -0.1 7
T6 31 2.5 4.3 -0.2 7
T14 32 2.1 3.7 -0.2 11
M6 5 2.6 7.8 0 3
M13 36 2.4 5.1 0.5 0
SC4 35 1.8 4.9 0.3 0
NT1 4 2.2 3.8 -0.1 1
NT2 2 2.2 2.7 1.1 0
Table 5-9: General Long-Term Statistics – La Caldera PS (Mexico City)
102
Table 5-10 presents similar long-term (albeit) simple transient event statistics, as recorded in the
Region of Peel’s water system. The interesting point in this table is the duration of pressure
monitoring at some of the locations. For example, the 500+ days of continuous pressure
monitoring at the Lakeview WTP can be seen as a more meaningful indicator of the absolute
transient pressure magnitudes, as well as the frequency of negative pressure events.
Table 5-10: General Long-Term Statistics – Region of Peel System
Table 5-11 extends the previous Lakeview WTP discussion through a more detailed statistical
consideration of transient events in the first year of continuous transient pressure monitoring. The
more complex list of statistics is aimed at answering additional questions, such as:
• How close are the maximum transient pressures to the pipe pressure rating and how
often are the most severe of these pressures experienced?
LocationNo. of
Days
Average
Pressure
(psi)
Min
Pressure
(psi)
Max
Pressure
(psi)
No. of
Negative
Events
Streetsville LLPS 25 62 14 85 0
MV900 Chamber 8 20 35 -15 53 1
HG1500 Chamber 11 20 40 9 57 0
BS1500 Chamber 8 56 30 -24 81 6
Beckett's Sproule LLPS 36 62 4 79 0
Silverthorne LLPS 36 60 15 76 0
ST1500 Chamber 9 36 17 -8 32 2
Lakeview LLPS Z1 500+ 107 102 134 0
Lakeview LLPS Z2 500+ 170 151 197 0
Hanlan LLPS 200+ 57 -6 79 6
103
• Are the maximum transient pressures increasing through time and are they
encroaching on the maximum pipe pressure rating? (i.e., is the additional structural
“capacity” decreasing?)
• Is there a correlation between the magnitude and/or the frequency of transient events
with the day of the week? (e.g., is the system at a higher risk during the week rather
over the weekend?).
• Is one pressure zone performing better than the other, and if yes, is it due to the
difference in the size and volume of HAC protection?
Table 5-11: Detailed Long-Term Statistics – Lakeview WTP (Region of Peel)
Figure 5-27 expands on the discussion of structural “capacity” at the Lakeview WTP by plotting a
curve that takes the form of a probability distribution function. Site 1 and Site 2 refer to the two
pressure zones, with Site 1 referring to the high pressure Zone 2 and Site 2 to the low pressure Zone
1. The y-axis represents the percentage of days of occurrence and the x-axis the maximum daily
pressure (transient or steady) as a percentage of the maximum pipe pressure rating. For example, if
the maximum recorded pressure in a single day was 180 psi, this pressure would account for 85% of
104
the pipe pressure rating and therefore provide only a 15% buffer in structural capacity (not
considering any safety factors, etc.). The cumulative nature of the curves allows for a simple, but
visual comparison of the two pressure zones. For example, the break point in the curve of
approximately 12 % for Site 1 (Zone 2) is significantly lower than the 40% for Site 2 (Zone 1). This
illustrates that Zone 1 operates much closer to the ultimate operating threshold.
Figure 5-27: Sample Cumulative Pressure Distribution – Lakeview WTP (Region of Peel)
While a discussion such as the one above is still preliminary and general and prone to the errors and
variability associated with a small sample size, it can still provide key insights into both the transient
response of the system and the ultimate level of the hydraulic transient induced risk. The larger the
sample size (i.e., the duration of transient pressure monitoring) the more meaningful the statistics
become and the more complex the questions that can be asked. This type of methodology is the
building block for a statistic based risk assessment and this is further discussed in Chapter 7.
However, Chapter 6 first takes a step aside from field based transient pressure statistics and looks at
an alternate use of such data. In particular it addresses the important topic of numerical model
validation through the use of specific case studies in which the field transient pressure data was used
to adjust model assumptions and parameters.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0% 10% 20% 30% 40% 50%
Percentage of Days
Maximum Daily Pressure as a Percentage of Pipe Rated Pressure
Site 1
Site 2
105
5.6 Summary
The main purpose of this chapter is to demonstrate the capability and benefits of using field work
and actual system data to assist in the decision making process and with the hydraulic transient
analysis task. The chapter initially provides a brief overview of the traditional approach to transient
field work and then moves on to describe the trends and changes through a discussion of modern
technology. As part the discussion, six (6) different commercially available high frequency transient
pressure monitors are compared across a wide range of performance criteria, including cost, control
logic, etc. While the various transient pressure monitors are all relatively different and some are
shown to be significantly better than others for some of the criteria, only one of these, the Pipetech
TP-1 is demonstrated to be capable of properly recording transient pressures for a long period of
time (e.g., months).
The second half of the chapter provides a comprehensive summary, including actual real life
examples, of the Pipetech TP-1 monitoring capabilities. The summary examines a variety of topics,
including calibration, parameter adjustments for recording, sample long-term transient pressure data
and statistics. Many of the small case studies in this chapter provide indications of why this type of
work is beneficial, and therefore presents how and why it should be used as a tool for field based
transient risk assessments.
106
Chapter 6 Validation of Numerical Models
This chapter provides examples of numerical hydraulic transient model validation using the
previously discussed high frequency pressure recordings from the field. The nature and importance
of transient model validation is presented through a series of different case studies, in which the
field data is used to validate and adjust the initial modeling results; results which are then
subsequently used to make other more difficult transient analysis and design decisions. In essence,
the field data was used to validate that the theoretical mathematical model actually represented
reality. The overall goal of this chapter is to reaffirm the importance of field work for a typical
transient analysis by showing how computer models can sometimes be inadequate, conservative, and
incomplete, but also how they can sometimes be complimentary to the proposed field work.
All field data was acquired through planned transient field tests or long-term transient pressure
monitoring, in which unique or planned transient events were observed. For example, as part of a
planned field test program, power failure induced trips of one or more operating pumps were
simulated at a specific pump station. The transient pressures were typically recorded at the discharge
header and/or at another point downstream (e.g., forcemain, transmission main, etc.), using one of
the transient pressure monitors described in Chapter 5. The numerical modeling data was simulated
using one of two different transient software packages: TransAM or H20Surge. The transient
models were developed from either existing steady state models or from steady state models
specifically developed for the system being considered.
6.1 Case Study I: Region of Peel
6.1.1 Background
The Region of Peel is a large and growing regional municipality with a current population in excess
of a million. The Region operates the water (and wastewater) system, including supply, transmission
and distribution for three local municipalities. The potable water is treated from a surface water
source and distributed along a significant distance using a partially separated transmission and
distribution pipe network. The transmission system carries the water from one pump station to
another, and each pump station also distributes the water locally within the specific pressure zones.
107
The detailed hydraulic model for the system contains tens of thousands of links broken down and
isolated into a number of pressure zones. A typical pressure zone may have between 1000 and 3000
links. The steady state model is partially calibrated for the current year and is also expanded for a
future planning year.
As part of a larger transient system study, the Region commissioned specific transient field tests and
pressure monitoring at a few of its pump stations and feedermains in pressure zone no. 3. The
resulting field data and numerical model analysis and validation were performed using the WCM
based H20Surge software package, and the results are presented below. The modeling results
presented for this case study are those pertaining to the original (unadjusted) model, in order to
illustrate the risk of not validating or adjusting a theoretical model.
6.1.2 Analysis
As part of this case study, individual transient events from specific transient field tests and
continuous pressure monitoring in pressure zone no. 3 were chosen for the initial validation and/or
confirmation of the numerical model. The chosen events were recorded at either Streetsville PS or
Hanlan PS. Many of the events observed during the continuous pressure monitoring were often of
the same source (e.g., a single pump trip) and many could not be subsequently correlated to a cause
and/or operational activity. As a result, three (3) unique, known, and predicable events were chosen
and these are briefly described below:
Event No. 1: Single Pump Trip at Streetsville LLPS
The results of this transient event and the numerical model validation are shown in Figure 6-1 and
Figure 6-2. The first figure shows the pressure profile as recorded at the source (Streetsville PS) and
the second figure shows the pressure profile as recorded at a downstream feedermain high point
location (MV900 Chamber 8). This event was caused by a planned transient field test. Before the
test, the only operating LL pump was pump no. 6; a pump with a rated capacity of 55 MLD. At
11:05 AM pump no. 6 was tripped, thereby simulating a power failure event during a single pump
operation. At 11:17 another LL pump, pump no. 8, was started.
At the source location, the single pump trip resulted in a significant downsurge with a range of 45
psi and a subsequent reflected upsurge with an ultimate peak that was 15 psi above the original
108
operating pressure. The initial transient pressure wave travelled downstream and was also recorded
at the second valve chamber location. At this location, the transient pressure range actually
increased, with the peak pressure being 20 psi greater than the original operating pressure and the
minimum pressures reaching full vacuum conditions. The second of these is primarily due to the
higher elevation and therefore the lower initial operating pressures. While the original (unadjusted)
model performed quite well for the source location, its results for the downstream location were
significantly off in both the transient range/magnitude and overall phase. There are numerous
factors at play in this case, including the lack of steady state model calibration, the improper
assumption of key transient parameters, and the realistic impact of system devices such as air valves.
Event No. 2: Single Pump Trip at Hanlan LLPS
The results of this transient event and the numerical model validation are shown in Figure 6-3 and
Figure 6-4. The first figure shows the pressure profile as recorded at the source (Hanlan PS) and the
second figure shows the pressure profile as recorded at a downstream feedermain high point
location (BS1500 Chamber 8). This event also occurred during a planned transient field test. Before
the test, two large LL pumps were in operation, namely pumps no. 9 and no. 10; each of 136 MLD
rated capacity. At 11:14 AM pump no. 9 was routinely shut-off (i.e., in a controlled manner), and at
11:16 AM pump no. 10 was tripped, thereby simulating a power failure event during a single pump
operation. Five minutes later, two other LL pumps were turned on.
The single large pump failure event at Hanlan LLPS induced negative pressures across the system,
including pressures in the full vacuum range at the downstream chamber location. The initial part of
the field versus model pressure profiles is quite similar, especially when one considers the maximum
and minimum pressures. However, in this section of the graph the model predicted a quicker and
longer duration downsurge event, likely owing to the conservative pump inertia assumptions. The
more evident dissimilarity between the two data sets is the excessive model noise in the second part
of the plots. The model continuously predicted re-occurring negative pressures, rapid pressure
fluctuations, and clearly unrealized positive upsurge pressures (e.g., up to a peak of 110 psi). For
example, the maximum upsurge pressures never exceeded the initial operating pressures due to real
system dissipation. The clear difference in the transient pressure wave phase also shows how
complex the wave interaction in this system is, and how models can overestimate transient risks;
109
risks which a real system can often minimize. The follow up subsection further discusses the general
trends in this overall case study.
Event No. 3: Multiple Pump Trip at Hanlan LLPS
The results of this transient event and the numerical model validation are shown in Figure 6-5 and
Figure 6-6. The first figure shows the pressure profile as recorded at the source (Hanlan PS) and the
second figure shows the pressure profile as recorded at a downstream feedermain high point
location (BS1500 Chamber 8). This unplanned event was a direct result of a mechanical problem
with a large pump. Before the event, the operating LL pumps (and rated capacity) were: no. 5 (110
MLD), no. 9 (136 MLD) and no. 10 (136 MLD). At 2:25 PM these three LL pumps simultaneously
tripped. Shortly after, another two LL pumps, pumps no. 6 and no. 7 were started, but no. 7 failed.
The multiple pump failure event at Hanlan LLPS induced negative pressures across the system,
including pressures in the partial to full vacuum range at the station and at the downstream chamber
location. The plots also bring light to a possible pressure recording malfunction, especially if one
considers the -20 psi (or so) pressure recorded at the downstream location. While extended and
severe magnitude full vacuum conditions (i.e., less than 14.7 psi) are possible in very clean water,
they are highly unlikely in this type of potable water system, especially with the typical presence of
air and some nucleation sites.
The transient response and dissimilarities between the field data and modeling data are quite similar
to the previous event case. However, the multiple pump failure is evidently more severe, especially
in the subsequent upsurge reflections and possible vapour cavity formation and collapse. The
maximum pressures at both the locations were in the 80 psi range, and the sharp pressure rise in the
second graph is indicative of a vapour cavity collapse at or near the high point. The other clear
observations again pertain to the overestimation of the nature and duration of the transient
pressures. The model predicted significant pressure fluctuations, while the system indicated
relatively rapid pressure energy attenuation. Additional insights into this type of model behaviour
are provided in the following subsection.
110
Figure 6-1: Model v. Field Event Pressure Validation – Streetsville PS (Region of Peel)
Figure 6-2: Model v. Field Event Pressure Validation – MV900 Chamber 8 (Region of Peel)
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50
Pre
ssu
re (
psi
)
Time (seconds)
Streetsville LLPS - Single Pump (#6) Trip
Model
Field
-20.0
-10.0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
0 10 20 30 40 50
Pre
ssu
re (
psi
)
Time (seconds)
MV900 Chamber 8 - Single Pump Trip (#6) at Streetsville LLPS
Model
Field
111
Figure 6-3: Model v. Field Event Pressure Validation – Hanlan PS (Region of Peel)
Figure 6-4: Model v. Field Event Pressure Validation – BS1500 Chamber 8 (Region of Peel)
-20.00
0.00
20.00
40.00
60.00
80.00
100.00
120.00
0 10 20 30 40 50
Pre
ssu
re (
psi
)
Time (seconds)
Hanlan LLPS - Single Pump Trip (#10)
Model
Field
-20
-10
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50
Pre
ssu
re (
psi
)
Time (seconds)
BS1500 Chamber 8 - Single Pump Trip (#10) at Hanlan LLPS
Model
Field
112
Figure 6-5: Model v. Field Event Pressure Validation for 2 Pumps – Hanlan PS (Region of Peel)
Figure 6-6: Model v. Field Event Pressure Validation for 2 Pumps – BS1500 (Region of Peel)
113
6.1.3 Discussion
The previously described field and model (unadjusted) validation plots are synchronized with respect
to the starting time, and are compared for a time period of 50 to 100 seconds. In general, the
unadjusted transient model is not significantly off from reality, but it does have a variety of concerns
from both ends of the traditional “conservatism” spectrum. The differences and similarities
between the model and field event plots, along with the inherent uncertainties and conclusions are
summarized below:
• The steady state pressures for some events and locations are different, thereby
confirming that the steady state model needs to be re-calibrated. A complete calibration
or even a good validation of a transient model requires a good calibration of the steady
state model. If the steady state model cannot predict the routine pressures, then the
transient model results cannot be taken by their absolute values. For example, in the
second sample event, the discrepancy in pressures is a direct result of the initial model
pressures being much greater than the field pressures.
• In most cases, the field pressures at downstream locations are lower than what the
model predicted. This may be due to many things in the field, including: higher pipe
roughness or reduced diameters, longer lengths of pipe, service connections, status
and/or operation of control, air valves, etc.
• Model validations are also strongly dependent on the presence and performance of
devices that are designed to, or simply can and do impact the surge conditions. The
model runs assumed that the existing pump station surge protection was working and
that the existing downstream air valves were not. Such an assumption cannot be proven
without a field inspection, but it can impact the results. For example, an air valve can
introduce a significant amount of air into the system, and this air can rapidly be expelled
thereby inducing a significantly higher pressure rise than predicted (or observed if vice
versa). Furthermore, the surge protection at the two stations consists of SRVs, and in
the model runs these devices did not act because the maximum pressure set points were
never reached.
• In most cases, the field and model pressures are quite close for the initial source
locations (i.e., at the pump stations).
114
• The results for the Streetsville event show that the field and model pressures are quite
similar in terms of relative and absolute magnitudes. The two are slightly different
because they are out of phase in either the horizontal or vertical axis. Nonetheless, the
downsurge and initial reflections (i.e., the first 20 seconds) are quite synchronized. The
duration of low pressures caused by the downsurge are very close, and therefore provide
comfort to specific model assumptions such as inertia, layout, etc.
• The remainder of the plots (i.e., after the first 20 seconds) are out of sync because the
field pressures show that the subsequent reflections actually occurred much quicker.
This is always a good indication of a different than assumed wavespeed, different pipe
length or a different/additional major boundary condition within the system. Overall,
the Streetsville event is well predicted by the model.
• The two Hanlan events bring light to the problems and uncertainties of numerical
transient modeling. Assuming that the steady state model is reasonably close and that
the field results are indeed trustworthy, the numerical model is shown to be highly
conservative in many situations. This is the case for most models in that they cannot
predict the system uncertainties which tend to reduce the impact of transients. For
example, small leaks within a system or transmission through boundary condition
components can act to reduce the energy of the pressure waves, but these types of
things are rarely modelled or considered.
• The four Hanlan plots clearly establish that actual system conditions allow for a
significant attenuation of transient pressures; additional attenuation that the model does
not predict. The system has been shown to dissipate the energy and to return to a
steady state much more quickly.
• The high frequency of the model data is unrealistic since it is being compared to the
field data which is recorded at 100 Hz (i.e., 100 times per second). These things
contribute to the model wearing off in the later stages.
• The opening and closing of check valves between numerical time steps is not something
that is easily predicted, or physically certain. Such things can lead to a numerical
instability that can produce excessive pressure oscillations such as those shown in the
two Hanlan events.
• The specific H20Surge transient modeling software (and its WCM solution approach) is
generally stable when the system unsteadiness arises mostly from inertial effects. As
115
high frequency is introduced, the frictional approach of the “wave characteristic”
approach tends to cause wave reflections which react to other devices (such as those
previously described), thereby leading to rapid negative and positive fluctuations that are
characteristic of the “hashing” in an unstable numerical approach.
• The Hanlan events also show that the pump model assumptions for this station need to
be adjusted. This is shown by the difference in the steepness and length of the initial
downsurge events. The actual pump rundown time and inertia are probably higher than
those assumed and modelled.
• The model also limits the negative pressures to full vacuum of -14.7 psi gauge, which as
previously discussed is not always a good assumption. Even though the field results
below -14.7 psi cannot be completely trusted because they exceed the pressure
transducer range, the possibility of reaching such pressure values cannot be precluded.
A potable water system with few nucleation sites and very tight linings is more than
capable of maintaining full vacuum conditions, particularly for a short duration. When
the model limits the minimum pressures, it can often introduce a change in the
predicted positive pressures. This is similar to shifting the pressure envelope upwards.
Nevertheless, the key point for decision making is the fact that negative pressures in the
full vacuum range were predicted and have been shown to occur in the Peel system.
The above case study explanations of the model validation are often true for most systems, in that
most models cannot be expected to produce identical results. However, since the Peel transient
model results are very similar to the observed field results in both the event form and magnitude, the
model users eventually felt a greater sense of confidence in using this model, especially for the
ultimate task of performing a long-term transient planning level assessment. The following section
aims provides a more detailed field and model transient pressure validation, through a case study in
which the initial model was subsequently adjusted based on the field data.
6.2 Case Study II: Region of Durham
6.2.1 Background
The Region of Durham is another growing regional municipality that provides water and wastewater
services to a high number of customers in four (4) larger municipalities (and several small ones).
116
The City of Ajax (and its water system) is one of these four municipalities, and it receives its potable
water from a surface water source via the Ajax WSP. Unlike in the case of the two specific pump
stations in the Region of Peel, the pump station at the Ajax WSP actually pumps treated water
directly into the distribution system. The Region of Durham commissioned a field transient
pressure monitoring study in order to determine the actual risk of transients, but also to validate the
transient model which would be used to make predictions (i.e., budget allocation) for the required
surge protection 30 years down the line. A high frequency TP-1 monitor was installed at the pump
station discharge header of the Ajax WSP, and naturally occurring transient events were recorded
and subsequently correlated to SCADA and system operation. The resulting field data and
numerical model analysis and validation were performed using the MOC based TransAM software
package, and the results are presented below.
6.2.2 Analysis
This subsection provides specific examples of how the raw transient field pressure data was used to
validate and adjust the numerical transient model. Operational and SCADA records could not
confirm the nature of several of the recorded transient events, and therefore the modeling exercise
was partially used to confirm the type and characteristics of a few pump shutdown events. Figure
6-7 presents a series of pressure profiles pertaining to a couple of pump shutdown events at the Ajax
WSP. The purple series pertains to a controlled shutdown of a single small pump (i.e., from 1
pump to 0 pumps). As expected, the routine shutdown (achieved via discharge valve control) of a
large pump yielded a continuous and smooth pressure reduction and did not produce a severe
transient pressure fluctuation. The green and the blue pressure series pertain to two independent
power failure events, of a small and a large pump, respectively. The typical downsurge pressure
profile was more significant for the larger pump failure (blue series) due to the greater change in
flow rate. The last series (red) presents an unadjusted model representation of a single large pump
power failure. While the ultimate minimum and maximum model predicted transient pressures are
essentially the same as those observed in the field for an actual large pump failure, the pressure
profile properties of the event are significantly different. The model under predicted the magnitude
of the initial downsurge and over predicted the magnitude and occurrence of subsequent negative
and positive wave reflections. If the model was not adjusted, its sample results would at first glance
appear to resemble a single small pump power failure, which in itself would make any subsequent
model predictions less than conservative. As a quick aside, all four series in this plot also show how
117
a system that is connected to the distribution network can minimize any upsurge reflections. In this
case, neither one of the initial (or subsequent) upsurges ever come even close to exceeding the initial
operating pressure. This is quite different than the previous case study; a study that was based on a
water transmission type of system.
Figure 6-7: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 1
The hydraulic transient model was to be used for future year planning decisions and needed to be
adjusted in order to better represent the actual field conditions. Figure 6-8 presents partial results of
the model adjustment exercise, using the single large pump power failure transient event discussed
above. Such a task requires a sensitivity analysis of many individual model, hydraulic, and system
parameters, both on an individual and combined level. The blue series represents the field data and
the red series represents the original (unadjusted) model data. The purple and green series
represent the modeling results for two of the more influential model changes, namely the addition of
an SAV and the inclusion of additional air in the system. The SAV was originally excluded from the
model because it was assumed to act much like an SRV. An SRV is used to reduce positive transient
pressures, and it should rarely, if ever, affect the negative pressures. An SAV can and often does,
impact the downsurge pressures as it is set to anticipate the upsurge through opening at the onset of
a pressure drop.
0
20
40
60
80
100
120
0 20 40 60 80 100 120 140
Pressure (psi)
Time (sec)
Routine Shutdown Nov 20 (Large Pump)
Power Failure Feb 11 (Small Pump)
Power Failure Nov 2 (Large Pump)
Model Power Failure (Large Pump)
118
As shown in the plot, the inclusion and modeling of the SAV significantly improved the model
predicted response of the initial downsurge. From a physical point of view this is expected, as the
SAV can (and in this case did) discharge a significant amount of fluid immediately following the
initial pressure drop, thereby further exacerbating the downsurge conditions. The second model
adjustment in the form of additional air in the distribution system (achieved by the addition of small
air pockets in the model via air valves and/or air chambers) acted to dampen the pressure
fluctuations resulting from any subsequent pressure wave reflections. From a physical point of view
this is also expected, as air pockets can (and often do) act as “cushions” in the system. While neither
one of the two simple model adjustments can be said to resemble the actual field results, their
individual attributes can be combined with additional parameter adjustments (see next figure and
discussion) in order to more precisely validate the transient model. This additional exercise was not
performed in this case due to uncertainties in the steady state model parameters. In fact the most
obvious difference between the field and model plots is in fact the steady state pressure following
the transient event. (It is likely that the steady state model under predicted the back pressure in the
system through low modelled reservoir levels, thereby showing lower operating pressures at the
source when the pumps are not in operation.) Nonetheless, the point of this figure is to illustrate
two simple examples of how model assumptions need to be tested and how subsequent adjustments
can target specific discrepancies in a typical transient event profile. Furthermore, in this particular
case the model adjustments acted to confirm the activation and subsequent negative impact of the
SAV at the Ajax WSP; knowledge which was later used to make appropriate design decisions.
The last plot, Figure 6-9, presents the results of a simple sensitivity analysis of the key system
wavespeed assumption that was originally made in the transient model. The original model
assumption of 1000 m/s (red series) was derived with the knowledge that the majority of the key
pipes in the system were rigid (e.g., CPP, DI, etc.). The green and purple series present the
modelled system response with a reduction of wavespeed; more precisely with wavespeed values of
750 m/s and 300 m/s. The 300 m/s wavespeed range is more typical of a flexible pipe system (e.g.,
PVC, HDPE, etc.), but can also be used to represent the presence of additional air (see previous
discussion). In order to more completely validate the model, this parameter adjustment would have
to be combined with the other two key parameters.
119
Figure 6-8: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 2
Figure 6-9: Model v. Field Event Pressure Validation – Ajax WSP (Region of Durham) – Part 3
20
30
40
50
60
70
80
90
100
110
120
0 25 50 75 100 125 150 175 200
Pressure (psi)
Time (sec)
Ajax WSP - Large Pump Failure - Model AdjustmentsField
Model - Unadjusted
Model - 250 mm SAV
Model - Distributed Air
20
30
40
50
60
70
80
90
100
110
120
0 25 50 75 100 125 150 175 200
Pressure (psi)
Time (sec)
Ajax WSP - Large Pump Failure - Wavespeed ImpactField
Model: a=1000 m/s
Model: a=750 m/s
Model: a=300 m/s
120
6.3 Case Study III: Mexico City 6.3.1 Background
The water supply to the greater Mexico City comprises a variety of sources, including the
conveyance from many well fields. At the outskirts of the city, the Tlahuac, Santa Catarina and
Mixquic well fields contribute water to a single pipeline and this pipeline then conveys this water to
the La Caldera Reservoir. At La Caldera, a set of three HL pumps convey treated water over a large
hill and into a distribution system. The utility commissioned a study to look into the transient
conditions at the La Caldera PS and pipeline; a study in which transient pressure monitoring would
be used to validate a numerical transient model. The numerical transient model would then be used
to evaluate future transient conditions following a planned upgrade, and to ultimately determine the
required surge protection. Unlike in the Peel and Durham case studies, this system is much simpler
from the point of view of hydraulics. At the time of analysis, the system consisted of three (3)
800HP vertical turbine pumps, a single 1200 mm diameter CPP pipeline with a length of 450 m, and
no additional system connections. A TP-1 monitor was installed at the discharge header, and the
resulting field data was used to validate a numerical model. The numerical analysis and validation
was performed using the MOC based TransAM software package, and the results are presented
below.
6.3.2 Analysis
The continuous transient pressure monitoring at the La Caldera PS recorded a variety of power
failure events; events that caused the pump(s) to trip and therefore induce a transient event. With
the properties of the rigid CPP pipeline, the relative short length (450 m), and the high static head
(80 m), the transient response following a pump trip can be classified as potentially severe. Figure
6-10 presents a model versus field transient pressure comparison for a single pump trip, from 2
pumps down to 1 pump. The unadjusted (i.e., as originally assumed) model results are generally in
good agreement with the field data. The slight differences in the minimum and maximum values are
likely due to several of the original steady state and physical system assumptions, including the
precise lengths, elevations, and pipe roughness.
121
Figure 6-11 presents another model validation plot for a single pump failure event. However in this
case the pump change is from 1 pump to 0 pumps. Such a change is more significant because of the
inherent grater change in flow (and therefore velocity) that is a general property of parallel pump
operation. In other words, the reduction in flow to zero is much more significant. Several
interesting observations can be made from this plot. First, the transient response in this system is
relatively simple and predictable. A pump trip leads to a downsurge, and this downsurge propagates
through the system and returns as an upsurge. This process repeats until the pressure wave is
dissipated, and this process is relatively smooth, symmetric, and quite predictable. Second, the
unadjusted model results generally conform to the field results, but are out of phase and actually
predict intermediate reflections that are not present in the real system. The potential cause of this
discrepancy could be the representation of the pump station check valves. In the field, the check
valves for the non-operating pumps were found to be quite leaky. In the model, the non operational
pump check valves were initially modelled as dead ends, thereby creating additional reflection points.
Lastly, the model results slightly underestimate the minimum transient pressures. This is likely due
to assumptions on actual operating conditions (e.g., flow and head), as well as pump inertia values.
Nonetheless, the overall magnitudes, profile, and dissipation mechanisms in the two data sets do
confirm that the model is quite reasonable and that it can be used to further analyze future
conditions.
Figure 6-12 presents the pressure profile for a more severe two pump trip at the La Caldera PS.
While the modeling results again suffer with respect to the phase of the pressure wave and the
intermediate reflections (note initial rise due to a check valve slam in the model), the maximum and
minimum transient pressure magnitudes generally do match those observed in the field. The model
versus field validation provides and confirms several important insights into the overall transient
response in the system. The clear risk in this system pertains to the maximum upsurge pressures;
pressures which have been found to be twice that of the operating pressure, and greater than the
rated pressure of the pipe. Furthermore, the field data also confirms the quick nature of the
transient response via a calculated wavespeed in excess of 1100 m/s. This typical wavespeed value
confirms that any air in the system is indeed being discharged at the top of the hill, rather than being
trapped along the pipeline. In the end, it is information such as this that can be used to better refine
the model assumptions in order to more precisely predict a system response that cannot be
simulated. In this case, the utility relied on the model to predict a variety of potential transient
pressure conditions when three (3) pumps are in operation.
122
Figure 6-10: Model v. Field Comparison (2 Pump to 1 Pump) – La Caldera PS (Mexico City)
Figure 6-11: Model v. Field Comparison (1 Pump to 0 Pumps) – La Caldera PS (Mexico City)
Model Validation - La Caldera Transient Pump Tests
0
20
40
60
80
100
120
0 2 4 6 8 10 12 14 16
Time (sec)
Pressure (m)
Field: Single Pump Trip (Two Pumps On)
Model Pump Trip
Model Validation - La Caldera Transient Pump Tests
0
20
40
60
80
100
120
0 5 10 15 20 25
Time (sec)
Pressure (m)
Field: Single Pump Trip (One Pump On)
Model Pump Trip
123
Figure 6-12: Model v. Field Comparison (2 Pumps to 0 Pumps) – La Caldera PS (Mexico City)
6.4 Case Study IV: Wastewater Forcemain
6.4.1 Background
The previous three case studies considered the validation of numerical models for simple and
complex potable water systems. This particular case study examines a single wastewater forcemain
in Ontario, Canada; a 650 mm diameter CPP forcemain, with a total length of approximately 1.5 km.
The forcemain is relatively flat in profile, fed from a single pump station, and discharges the fluid to
a gravity system via a bulkhead and manhole. The utility commissioned a field based transient
assessment in order to determine the potential risk of failure; failure potential that was primarily
perceived to be due to the age of the pipe, but also due to criticality and location of this asset. A
TP-1 monitor was installed at the discharge header of the pump station, and the resulting field data
was initially used to assess the in-situ risk from hydraulic transients. The field data was then
subsequently used to validate a numerical model, a model that could then be used to assist in specific
Model Validation - La Caldera Transient Pump Tests
0
20
40
60
80
100
120
140
0 5 10 15 20 25
Time (sec)
Pressure (m)
Field: Double Pump Trip (Two Pumps On)
Model Pump Trip
124
design decisions. The numerical analysis and validation was performed using the MOC based
TransAM software package and the results are presented below.
6.4.2 Analysis
The following figures present the results of a model validation and adjustment exercise, in which the
field data from a simulated power failure of a single pump is compared to the incrementally adjusted
model results. In all of the figures, the red series represents the recorded field pressure history at
the discharge header of the pump station.
Figure 6-13 presents a field versus model sensitivity analysis of the important wavespeed parameter.
The blue series represents the original modeling assumption of 1000 m/s – a value that is very
typical of a rigid CPP pipe. The evident discrepancy between the unadjusted model with a
wavespeed of 1000 m/s (blue series) and the actual field data (red series) can be described as both
significantly out of phase and not attenuating. The reasonably assumed wavespeed is indeed too
high and it leads the model to predict a much faster (and less attenuated) wave reflection period,
thereby ultimately leading to more pressure cycles. As the model wavespeed is incrementally
decreased to 300 m/s (orange series) and 150 m/s (green series), the pressure wave phase becomes
more in sync with the field data. From a physical perspective, a wavespeed in the 150 m/s range is
completely justified since this is close to the measured field value. In other words, the green series
is almost in phase with the red series. Such a low wavespeed is more characteristic of a flexible pipe,
but can also result from high air content (see previous discussion in Chapter 3). The amount of air
in this system is therefore significant, and predominantly due to it being a wastewater system, but
also because of the forcemain’s flat profile and lack of air valves.
While the wavespeed adjustment exercise essentially corrected the phase of the transient pressure
wave, it still did not account for the true pressure wave attenuation in the system. The amount of
energy dissipation in the real system is significantly increased by the previously discussed air content
(especially if at the downstream discharge end), but the numerical model cannot adequately take that
into account, especially if the air forms locally and in pockets. Nonetheless, the difference in
pressure wave attenuation can also be explained by how the model accounts for friction. Such a
discrepancy is characteristic of a water hammer model that employs a steady state friction formula to
compute friction loss during transient flow conditions (Axworthy and Chabot, 2004). Figure 6-14
125
presents the validation results when an unsteady friction model is implemented. Maintaining a
constant wavespeed of 150 m/s and accounting for unsteady friction (blue series) leads to a
significantly better representation of the true pressure wave dissipation in this wastewater system.
While not perfect, these two concurrent transient model adjustments significantly improve the
model and field data correlation.
Figure 6-13: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 1
The previous two model adjustments predominantly accounted for the pressure wave phase and
dissipation. However, the model still under-predicts the first two positive pressure wave amplitudes.
In addition to the important physical system properties (e.g., length, diameter, etc.) and steady state
model assumptions (e.g., friction, flow, head, etc.), the transient model for a wastewater system can
also be adjusted to better account for vapour pressure, check valve slam, and air cavity formation
and collapse. The next pair of figures present a few alternative results of such an additional model
adjustment exercise; an exercise performed through several sensitivity analyses and concurrent trial
and error adjustments. Figure 6-15 shows the comparison results with the following model (blue
series) adjustments:
126
• Vapour pressure (see Chapter 3) is increased to a very high value of 7 m;
• Wavespeed is set at 175 m/s;
• MRI is significantly reduced to 1 kg-m2 to take into account the new (and less
heavy) type of pump; and
• Unsteady friction assumption is preserved.
With the above transient parameter adjustments, the model better approximates the initial positive
upsurge value, while still preserving the pressure wave phase and dissipation mechanism.
Nonetheless, the model still under predicts the maximum transient pressure by almost 40% (12 psi).
From a physical perspective, the large upsurge magnitude recorded by the field pressure monitor is
likely due to a vapour cavity collapse. The field test observation records noted a significant air
presence and a loud and evident activation (i.e., hissing) of two CAVs within the station, following a
pump trip. As such, it is likely that the field data recorded a significant air pocket collapse following
the first return upsurge. This can further be adjusted in the model through the addition of the
proper air valves; air valves that would then be simulated to rapidly expel the air that they initially
allow to enter the system.
Figure 6-14: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 2
127
Figure 6-15: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 3
Figure 6-16 presents results for a few additional and incremental model adjustments. It includes two
series (green and blue) that fine-tune the three previous parameters of wavespeed, MRI, and Vp.
More interestingly, these two model results also include an adjustment to the representation of the
check valve(s) at the pump station. In these cases, the check valve properties were changed such
that a partial reverse flow through the check valve(s) was possible, thereby increasing the magnitude
of the check valve slam and the subsequent rise in pressure.
Overall, the above discussed model adjustments act to significantly improve the original model
performance. While still not perfect, the model nonetheless is more representative and therefore
can better be used to make additional analysis and design decisions for this system. A complete
calibration is essentially impossible to achieve due to the number of variables. Furthermore, it
requires additional field data sets for confirmation. Otherwise, one can easily over compensate
through specific parameter adjustments, thereby making the model less representative as a whole.
128
Figure 6-16: Model v. Field Comparison – Wastewater Forcemain (Ontario) – Part 4
6.5 Summary
Building upon the previous chapter’s examination of the benefits of using transient field work for
making important design and analysis decisions, this chapter provides a clear and useful link between
field data and numerical modeling. The resulting numerical model validation discussion in this
chapter essentially forms the basis of the previously proposed hybrid method for a transient analysis.
The benefits of combining field work with numerical model are presented through four (4) unique
case studies of actual (i.e., in-situ) systems. The case studies are carefully selected not only to
provide examples of the risks of solely relying on numerical modeling, but also to illustrate the
difference between a variety of system types (as is discussed in Chapter 2). While the numerical
validation examples are quite technical in nature, the overall chapter premise and conclusion is that
field data and modeling data must be combined, such that they can collectively be used to perform
comprehensive transient analysis and design.
129
Chapter 7 Transient Risk Index
The previous chapters describe the importance of transient pressure monitoring and show how the
data acquired from such field work can be used to assist in the analysis and design of systems, as
well as to validate numerical transient models. The TP-1 equipment is shown to be the best for
long-term transient pressure monitoring, and this continuous pressure monitoring was anecdotally
shown to be useful for assessing the hydraulic performance of water and wastewater systems. This
chapter proposes that such long-term transient pressure data be used in a more direct and
statistically derived transient risk assessment. More specifically, it proposes the use of a Transient
Risk Index (TRI) in order to benchmark and compare the overall transient risk between similar
hydraulic systems.
7.1 Purpose and Background
7.1.1 The Nature of System Failure
The difficult task in conducting a transient risk assessment is estimating a realistic likelihood of
occurrence of an event, and the likely magnitude and timing of any resulting consequence(s).
Hydraulic transients are typically brought on by local changes, but their impact (and therefore risk) is
often felt across the entire hydraulic system. Furthermore, transient events present one form of risk;
a risk that is further compounded by other system risks, including those potentially arising from
poor design, improper construction, environmental conditions, poor operation, aging infrastructure,
and emergency events. As a result, it is not only an individual transient event that can directly lead
to a system or performance failure, but also the joint effect of multiple transient events in
combination with other system events and/or risks. Of course, the actual cause of a joint system or
performance failure is often difficult to ascertain immediately following the occurrence.
Furthermore, and possibly more importantly, the probability (or even, the frequency) of occurrence
and the timing of any resulting consequence are equally as difficult to predict for a joint system
failure. As an example, consider the 1979 near catastrophic failure at the Three Mile Island nuclear
plant, in which an unlikely and unpredictable series of events lead to a system failure. In this case, a
filter blockage caused a moisture leak, which tripped valves and shut down the flow of cold water
into the plant's steam generator. The primary backup system was not functioning on that day, and
the secondary backup relief valves got stuck and failed to properly close. Lastly, the primary backup
130
system failure alarm was partially obstructed and the gauge for the relief valves malfunctioned.
Gladwell (1996) summarized the near catastrophic system failure as “a major accident caused by five
discrete events”, and the culprit as the way these “minor events unexpectedly interacted to create a
major problem”. Water and wastewater systems are often subjected to severe single event type of
risks, but also to risks arising from joint or series events such as the ones in the Three Mile Island
case.
The simplest and most direct synonymous quantitative application of the above idea is the case in
which a pipe break is solely caused by internal pipe pressure changes. Water and wastewater pipes
are typically designed to withstand a normal operating (often called “working”) internal pressure, but
also a short-lived transient pressure. The following subsection provides a brief summary on the
design of pipes for internal pressure in order to establish the general background understanding for
the purpose of the proposed TRI.
7.1.2 Pressure Class Design for Isolated Transient Events
The methodology for determining the required pressure rating for a pipe directly accounts for both
operating pressures and transient pressures. The current approach for pressurized pipe design relies
on the concept of pressure class. Pipe pressure class design is a performance based specification
that gives pipes a rating based on their ability to withstand internal pressures. It provides standard
criteria for manufacturing, for design, and for comparison across different material types with the
same performance expectations. In steel, DI, and CPP, the pressure rating is directly based on the
wall thickness of the pipe and the strength of the material. With a given design pressure that a pipe
will likely experience during operation, the designer can solve for a minimum thickness or simply
pick from a list of available pipes with specific pressure ratings. Each of the pipe pressure classes
includes an additional allowance for short-lived transient pressures. This “surge allowance” is
usually given as a percentage of the working pressure (Pw); a summary example of which is shown in
Table 7-1 (Mielke, 2004).
131
Table 7-1: Sample Pipe Pressure Class Properties (Mielke, 2004)
For thermoplastic pipes, the pipe’s rating or pressure class is determined using design equations that
factor in the material’s strength, pipe wall thickness, and diameter. The strength of PVC and
HDPE pipes is usually determined through ASTM testing procedures (ASTM D1598, ASTM
D2837, ASTM D638); procedures that subject the pipe to an internal pressure and extrapolate the
strength after 100,000 hours. In determining a thermoplastic pipe’s pressure class, standards such as
AWWA C900 incorporate a surge allowance equivalent to a 2 ft/s instantaneous stoppage of water
in the pipe. However, most thermoplastic standards require testing the pipe up to 2 to 5 times its
pressure rating; a test that in most cases covers common isolated transients as well. Lastly, the
transient pressure build-up in thermoplastic pipes is considered lower than for other materials due to
the material’s behaviour under stress.
7.1.3 Pressure Class Design for Cyclic Loading
The risk of failure from cyclic surge pressures is usually not fully accounted for in most hydraulic
transient analysis of water and wastewater systems due to the lack of guidance available from many
standards (Murray et al., 2004). Since the number of cycles required for pipe failure is often quite
large, system designers tend to ignore the potential risk of fatigue failure resulting from repetitive
pressure cycling. As noted earlier, a failure that occurs due to pressure cycling will likely result due
to a series of events, and therefore inadequate design will rarely be proven as the sole cause of the
failure.
The standards that do cover fatigue from cyclical pressure loading often take a different approach.
The British and Australian codes explicitly mention cyclic loading from transient pressures. A
132
number of studies (Jeffrey et. al. 2004, Kirby 1980, Bowman 1990, Zarghamee 1990, and Marshall
1998) have been conducted on thermoplastic pipes to determine their behaviour under cyclic
loading. These studies subjected pipes to a large number of cyclic loads and observed for signs of
failure. As a result, a number of failure curves have been produced from such studies, and these
curves typically show the amplitude of the pressure loadings and the number of cycles for failure.
An example of fatigue design load factors from the Plastics Industry Pipe Association of Australia is
shown in Table 7-2.
Table 7-2: Fatigue Load Factors for PE Pipe (PIPA, 2002)
7.1.4 Importance of Risk Assessment
The above pipe design approaches essentially look at the difference in the risk of failure due to both
overall internal pressure and internal pressure fluctuations. In a general sense, transient risks can be
divided into two broad categories:
1. Those associated with catastrophic failure from significant isolated surge events or a
catastrophic combination of many other joint but discrete events; and
2. Those associated with long-term fatigue or cycling failure from repetitive and frequent
pressure variations or combination of other joint discrete events.
While both of these transient induced failures are form of system risk, only the first is typically
considered in traditional transient analysis, especially if the analysis is driven by numerical modeling.
As discussed in Chapter 2, hydraulic transient analysis is typically performed based on anticipated
133
worst-case scenarios; scenarios which are deemed to present the risk in category no. 1. The main
obstacle to a comprehensive transient risk assessment is the difficulty of addressing the statistics –
the probability of occurrence, the frequency of occurrence, and the magnitude of the consequence.
As a result, most transient analyses are rather quite qualitative in nature. This is where the
continuous transient pressure monitoring of actual systems can indeed be used to bridge the gap
between qualitative or subjective analysis and actual quantitative (or frequency based) assessments
that rely on acquired data. The objective of the subsequently discussed methodology is to address
the composite transient induced system risk; a risk that includes events that range from high
consequence but infrequent to those that are lower consequence but frequent.
7.2 Requirements, Parameters, and Definitions
In order to derive a useful and statistic based methodology for a transient risk assessment using field
pressure data (an example of which is the proposed TRI), several system and analysis requirements
must first be established. The requirements must differentiate between different types of systems
and must establish clear statistical definitions of parameters that are to be considered. The
parameter definitions must then be used to establish a consistent methodology for acquiring and
analyzing pressure data and for determining composite parameters than can be compared across
different systems.
7.2.1 System and External Influences
Hydraulic transient events and overall transient risks are unique and different for every system.
Unlike other areas of study such as rainfall and runoff analysis, travel time analysis, structural load
analysis, etc., the number of externalities that can significantly undermine a statistical analysis
approach for hydraulic transients is much greater. The occurrence, risk, and consequence of
transient events are influenced by a host of system properties and external factors, including, but not
limited to the following:
1. Type of fluid (e.g., potable water, wastewater, etc.);
2. Type of system (e.g., pressurized wastewater, water transmission, water distribution,
gravity, etc.);
3. Method and nature of system operation (e.g., pump changes, valve operations, etc.);
134
4. System condition and location (e.g., strength of pipe and equipment, construction
quality, age of equipment, elevation, etc.);
5. Additional (non-transient) risks (e.g., soil aggressiveness, quality of water source,
awareness of transients, etc.);
6. Type and nature of surge protection; and
7. Multiple transient event occurrence and interference.
The above factors can significantly influence any statistical approach that is aimed towards
comparing the risk of one system to another. This is especially the case for the proposed TRI; the
methodology for which is described in the following section. Nonetheless, since this type of risk
assessment methodology is still in the preliminary stage, it is useful to make some general system
assumptions in order to develop the proper definitions. These definitions are based on a typical
system with the following properties:
1. A pressurized water or wastewater system in which transient events are induced by
routine and non-routine operations (e.g., valves, pumps, etc.).
2. Prior to any transient event, the pressure is relatively constant and can be defined as a
steady state starting point.
3. A transient event is defined as any pressure fluctuation that is quantifiably different
(i.e., meets minimum criteria) than the steady state pressure.
4. A typical transient event consists of a either an initial downsurge or an upsurge, and is
followed by period of pressure oscillation.
5. Any subsequent transient event that is an automatic reaction to an initial event is
considered as a separate and unique event.
6. Following the transient event, the system re-stabilizes at a new steady state that may or
may not be equal to the original steady state.
With the following basic and general system properties in place, the next step is to define specific
parameters. The following subsections present two important categories of parameter definitions:
1. Single Transient Event Parameters – used for defining statistical properties of any one
transient event.
135
2. Multiple Transient Event Parameters – used for delineating between multiple transient
events and for defining field pressure monitoring and recording parameters.
7.2.2 Single Transient Event Parameters
Continuous transient pressure monitoring typically yields a significant amount of data; data that
consists of both steady state pressure fluctuations and transient events. In order to derive a TRI (or
any similar metric), specific parameters must first be defined for a “typical” transient event. Figure
7-1 presents a graphical illustration of a “typical” transient event, including important event
properties and parameters.
Figure 7-1: Single Transient Event Properties & Parameters
136
Table 7-3: Single Transient Event Parameter Definitions
Notation Name Definition
Po Initial (Pre-Event) Steady State Pressure
Constant (or relatively constant) average background pressure before a single transient event.
PMIN Minimum Transient Event Pressure Magnitude
The absolute minimum pressure (crest) during a single transient event.
PMAX Maximum Transient Event Pressure Magnitude
The absolute maximum pressure (crest) during a single transient event.
PV Full Vacuum Pressure Full vacuum pressure limitation of -10.4 m H20 gauge or -14.7 psi gauge.
PF Final (Post-Event) Steady State Pressure
Constant (or relatively constant) average background pressure after a single transient event.
∆PMIN Transient Low Pressure Range
Maximum low pressure range (i.e., downsurge amplitude) during a single transient event.
∆PMAX Transient High Pressure Range
Maximum high (pressure range (i.e., upsurge amplitude) during a single transient event.
∆PSS Steady State Pressure Change
Difference between the initial and final steady state pressure magnitudes.
TEVENT Transient Event Duration Duration of transient event, from initial to final steady state pressure equilibriums.
TPERIOD Pressure Wave Period Average pressure wave period or cycle (i.e., duration between wave crests or wave troughs).
TNEGATIVE Negative Pressure Duration Summation of negative pressure (i.e., below atmospheric pressure or 0 gauge) durations for a single transient event.
With the above single event parameters and definitions in place, several additional, useful, and often
obvious relationships can be derived. These include the following:
1. ∆PMIN = Po - PMIN
2. ∆PMAX = PMAX - Po
3. ∆PSS = Po - PF
4. TPERIOD = 4L/a where L is the system length and a the wavespeed
5. TEVENT = nTPERIOD = n4L/a where is n the number of cycles during the event
With these definitions in place, simple event driven statistics such as the mean, standard deviation,
correlation coefficients, etc., can be calculated (see sample data in Chapter 5.5).
137
7.2.3 Multiple Transient Event Parameters
The previously discussed parameters and definitions are relatively easy to understand and calculate
when the background (steady state) pressures are mostly constant and when a transient event is
relatively clear. The difficulty in the above approach arises when the background pressures are more
variable (i.e., moderately transient) and when multiple transient events occur within a short period of
time. If the transient pressure monitor in the field is set to continuously record at a high frequency
(e.g., 100 Hz), the resulting sample data set would render the single transient event definition, and
the associated parameters, much less clear. In such a case, a different and more rigorous
methodology would need to be established in order to statistically define what a transient event is.
Fortunately, this thesis has thus far shown the benefits of, and relied on, the use of the Pipetech TP-
1 transient pressure monitor. This technology is capable of recording transient events at high
frequency and background (steady state) pressures at a lower frequency; thereby not only minimizing
the size of the data set, but also inherently defining a single transient event through user initiated
recording parameters. As Chapter 5 has shown, the three (3) key parameters for defining a transient
event using the TP-1 transient pressure monitor (Pipetech, 2008) from an average background
pressure are as follows:
1. Start Record Standard Deviation (SSTART);
2. Start Record Absolute Pressure Change (∆PSTART); and
3. Stop Record Standard Deviation (SSTOP).
The above TP-1 parameters typically have to be adjusted to match specific pressures and the
operating philosophy of a particular system, but once in place these parameters do an excellent job
of delineating between a background pressure and a transient event. As a result, the traditional
event definition conundrum of any long-term statistical analysis is addressed in advance of the data
analysis stage. While this is beneficial for the ultimate end goal, the proper selection of the above
recording parameters is extremely important. The following TRI methodology directly assumes that
this critical step has been optimized.
While a TP-1 monitor can easily distinguish a transient event from a background pressure, it often
has trouble with:
1. Recording the entire transient event as a single event; and
138
2. Recording the complete transient event.
The problem in both of these two cases is the issue of the pressure remaining relatively constant for
a short period of time during an actual transient event. From a physical point of view, one can
understand and accept that transient events can indeed contain plateau shaped crests (positive or
negative), or periods in which the relative constant pressure may actually resemble a new steady
state. Figure 7-2 presents a sample but generic pressure profile for a multiple transient event trace.
Figure 7-2: Multiple Transient Event Properties & Parameters
The blue data points represent low frequency recording during a relatively steady background
pressure period, while the red data points represent high frequency recording of transient pressures.
From a physical point of view, this sample pressure trace comprises two transient events: event no. 1
and event no. 2. Event no. 1 is a typical downsurge event that resembles a power failure. Following
this event, the system (background or steady state) pressure settles at a new equilibrium level. Event
no. 2 is an upsurge event that eventually raises the system background pressure. However, due to
the plateau in the low pressure crest of this second event, this event would likely be recorded as two
139
separate events in the field. In other words, the TP-1 record parameters can mistake a short period
of constant pressure as new steady state equilibrium, thereby fragmenting a single transient event
into two (or more) events. This problem is unfortunately rooted back to the previously discussed
definition of a transient event, and more importantly requires a rigid definition that can delineate
between multiple transient pressure traces.
In their probabilistic analysis of the rainfall-runoff phenomenon, Adams and Papa (2000) applied
the concept of a statistical (and user defined) inter-event time in order to delineate between rainfall
events. Since such a user defined parameter can significantly impact any resulting statistical analysis,
the authors noted three (3) potential (but previously published) methods for determining a suitable
inter-event time for a data set. These methods include the following:
1. Autocorrelation Analysis – correlation of data in one point in time with another point
in time via a minimum lag time parameter.
2. Probability Density Analysis – correlation of inter-event times with a density
distribution such as the exponential, and a subsequent selection that yields a
coefficient of variation equal to unity.
3. Event Number Analysis – correlation between inter-event times and the resulting
number of events in a data set, and a subsequent selection based on minimal additional
impact on the number of events.
While an inter-event time definition is statistically powerful, and while the above inter-event time
selection methods can indeed be used, such a methodology is currently not necessarily required in
order to deal with the multiple event dilemma. In fact, due to the reliance on the recording
technology for the definition of a transient event, it is much simpler to account for multiple events
by a post data processing algorithm adjustment that is similar to the inter-event time definition.
Based on an analysis of multiple data sets (for different systems and different recording parameters),
the unintended fragmentation of a single transient event into multiple events can be eliminated by
introducing an inter-event time (TIE) that is correlated to a number of background recordings (see
Figure 7-2), such that:
TIE = m TIB
140
where m is the number of background recordings during the perceived inter-event
time period and TIB is the preset time period between background recordings (i.e.,
inverse of the background record frequency)
In most of the data sets analyzed to date, a value of 2 or 3 for m is sufficient for ensuring that a post
data processing algorithm properly accounts for “fictitious” multiple transient events. In other
words, such a post data processing algorithm essentially prescribes a minimum number of
background recordings (and when combined with the frequency of recording – the overall duration)
prior to a transient event being classified as complete, thereby ensuring a more accurate
representation of any one transient event.
7.3 Methodology
With fine-tuned TP-1 (or equivalent) pressure recording parameters and rigid event parameters a
data processing algorithm can be used to extract all transient events from a continuous pressure
monitoring data set. Samples of such transient event summaries have previously been shown in
Chapter 5.5. These transient event summaries can then be used to calculate individual and
combined statistical event parameters, such as those previously outlined in Table 7-3. This is best
achieved through a simple database query of the transient event summary. The overall goal of this
type of analysis would be to associate a frequency to the event, and therefore establish a true metric
of risk. (To date, hydraulic transient “risk” assessments still lack the characterization of the
frequency component and are therefore mostly qualitative assessments of the expected
consequences).
With all of the definitions, assumptions and processed event data in place, a preliminary transient
risk assessment methodology such as the proposed Transient Risk Index (TRI) can finally be
compiled. Figure 7-3 presents a generic graphical representation of a possible TRI. The goal of a
TRI is to assess the cumulative risk of all transient events for a pressure monitoring period and to
then ideally extend this to the complete lifespan of a system. The basic notion here is that small
magnitude transient events (both positive and negative) occur frequently but with lower
consequence, and that large magnitude transient events (both positive and negative) occur
infrequently but with high consequence. Unlike in the case of a traditional risk analysis, it is NOT
only the rare and high consequence event that presents a risk to the system, but actually it is the
141
combined effect of ALL events. While the TRI is likely not the only viable metric or approach,
system indices such as these are very simple to understand, benchmark, and compare, and are
therefore quite useful for understanding the transient performance (and therefore the risk) of a
system.
The TRI would be plotted on a both a positive and negative y-axis constituting an absolute or
normalized maximum and minimum transient pressure, and on an x-axis that constitutes an event
return period (TR). The negative pressure axis would ideally be limited to the full vacuum limit due
to inherent assumption that traditional water and wastewater systems likely cannot sustain such a
condition. (This statement does not imply that water is not capable of sustaining full vacuum
pressures.)
Figure 7-3: Transient Risk Index Schematic
142
The event return period (or recurrence interval) for the x-axis can be calculated by initially compiling
a list, plot, or histogram of maximum and minimum transient pressures. The data processing
algorithm would produce an occurrence (i.e., event magnitude) count for a pre-selected pressure
gradation (e.g., 1 psi, 2 psi, 3 psi, etc. range), and this event magnitude count would then be ranked
for an m number of different pressure gradations. Several statistical analysis approaches can be
taken from this point on, but at this stage it is likely the simplest to define an event return period
such that:
where m is the rank of the pressure gradation count and n is the data
set recording or analysis period (in days)
The figure would essentially yield two curves that start from the origin: a positive transient pressure
risk limit curve and a negative transient pressure limit curve. The shape of the positive and negative
risk curves have yet to be officially determined, but should likely have a shape in which the transient
pressure magnitudes increase for a larger value of the return period. With a large enough sample
size (i.e., long enough duration of pressure monitoring), a specific function (e.g., exponential) could
eventually be mapped to the curves such that:
and
With the above two continuous functions in place, a TRI can then be calculated for the positive risk
index and for the negative risk index. The proposed notation for such indices is as follows:
TRI+= Positive Transient Risk Index
TRI- = Negative Transient Risk Index
TRI+ would simply be equal to the area under the positive risk curve and above the x-axis, and the
TRI- would be equal to the area above the negative risk curve and below the x-axis. The generic
equations for the two TRIs metrics would therefore be as follows:
0
( )
T
MAX RTRI P f T
+
+
= =∫
1R
nT
m
+=
( )MAX RP f T= ( )MIN RP f T=
143
0
( )
T
MIN RTRI P f T
−
−
= =∫
The limits of integration for the risk indices would be that of the maximum return period from the
data set (denoted as T+ and T- in the above equations), and the units for the indices would be in the
form of pressure multiplied by the return period, such as psi-days or m(H20)-days. However, the
units would often be omitted in order to uphold the true sense of an index value.
The proposed TRI methodology is intended to provide a simple and easy method for determining
the degree of the overall transient pressure risk for water and wastewater systems. More
importantly, it is intended to provide a simple link and understanding between continuous transient
pressure monitoring and a risk assessment. Having said that, the concept, methodology and
definitions for the TRI are still in their preliminary stage, and the TRI may end up simply being used
as a stepping stone for more comprehensive (yet still quantitative in nature) hydraulic transient risk
assessments.
7.4 Sample Results
As noted earlier, the TRI concept is still in its early stage of development and therefore any sample
results must be considered as trial, and even as slightly premature. The short duration of less than a
year for a few of the system monitoring locations does not provide a large enough of a sample data
set. Furthermore, the data processing algorithm still requires additional flexibility and adjustments
for the gradation of the recorded transient events and pressure data. Nonetheless, Figure 7-4
presents two preliminary data sets for the previously discussed Lakeview Zone 1 and Zone 2 water
distribution systems in the Region of Peel.
144
Figure 7-4: Sample Preliminary TRI Graph for the Lakeview Zone 1 and 2 Systems
The absolute positive transient pressure event magnitude is plotted against the return period, for a
pressure gradation of 2.5 psi. Due to the existing surge protection at the Lakeview WTP, the
transient pressure magnitudes are well controlled and therefore the sample data set is dominated by
absolute pressure in the 160 to 180 psi range. Furthermore, due to the current limitations of the
data processing algorithm, the y-axis only plots the absolute pressure value and not the transient
pressure range – a range which must be calculated based on the starting operating (i.e., steady state)
pressure. As of a result of both of these limitations, a risk curve and the calculation of the TRI (i.e.,
the area under the curve) are not provided at this time. Nonetheless, the sample figure re-establishes
the general TRI approach using actual field data.
7.5 Discussion
The previous chapters demonstrate the need for a field data based transient risk assessment and the
previous sections in this chapter provide the initial definitions and methodology for one sample
approach – the TRI. While general rules for the pressure recording and the definitions for a
145
transient event (both single and multiple) are established, there still exist a lot of questions and
uncertainties as to what the best method to calculate such a risk assessment metric actually is. The
proposed TRI is simply one alternative; an alternative that frankly still needs to be completed before
it could gain widespread acceptance. There are many details that must be considered, and many
options and/or variances that can be used. This section provides a brief and itemized discussion of
several important considerations, and provides recommendations for the improvement and
implementation of this concept and methodology.
• Long-term data on the order of years should be acquired for a variety of systems, and the
data should be processed for the purpose of developing better TRI curves and values.
• A sensitivity analysis of the user defined TP-1 settings and parameters should be performed
in order to establish a rigorous approach for acquiring the transient pressure data.
• The TRI methodology should investigate the impacts of, and how to deal with, the
interference of actual multiple transient events. For example, how should the TRI metric
account for two transient events that superimpose across a certain period in time?
• More rigorous guidelines should be established for the pressure gradations (i.e., y-axis), and a
sensitivity analysis of the gradation parameter should be performed.
• A theoretical (i.e., non-recording logic driven) definition of a transient event should be
derived and compared to the results achieved from the data post-processing. In other
words, what actually constitutes a transient event? Are not all pressure recordings transient
depending on the scale being considered?
• Using long-term processed data, several TRI curves should be plotted and the sample values
calculated. The goal would be to determine the true nature and profile of the risk curves.
• The TRI curves and values should be compared across a range of different system types,
including but not limited to the differences between: water and wastewater, transmission and
distribution, high head and low head pumping, with and without surge protection, etc.
• TRI curves are likely to be fragmented (i.e., discretized) for most systems, and therefore the
continuous function approach may need to be reconsidered.
• The addition of another minimum pressure curve should be considered to account for the
fact that not all minimum pressures are negative. In other words, the methodology should
consider adjusting the x-axis to match an average operating pressure range rather than the
value of atmospheric pressure (i.e., 0 gauge).
146
• The methodology should consider the potential benefits and differences of changing the
concept of both the x and y axes. More specifically, the value and/or usefulness of the
metric should be evaluated by changing the y-axis to either a normalized (i.e., ratio based)
version of transient pressure or a transient pressure range (rather than absolute value).
Similarly, the benefits of changing the x-axis from a traditional return period to a simpler
count (or histogram) should also be investigated.
• The proposed TRI methodology should consider and investigate the maximum limit of
integration. In other words, should it be limited to the maximum recorded pressure?
Maximum pipe or device pressure rating? Infinity? etc.
• Initial TRI system values should be used to set benchmarks and to compare not only the risk
between different systems but also the change in the risk across the life of one system.
• Once a rigid transient pressure recording and TRI methodology are established, the sample
TRI curves should be fitted with probability distribution curves. The type of curve would
likely depend on the properties of the system, but could be quite useful for not only
calculating the TRI values but also for establishing closed-form relationships between
transient pressure magnitudes and event occurrence.
• The resulting TRI curves should be correlated to system strength, such as pipe rating or
fatigue curves, in order to present a true transient load versus system strength analysis.
• The TRI methodology should be used to assess and adjust for different system components
(e.g., pipe, valve, pump, etc.) and failure types (intrusion, pipe rupture due to high pressure,
fatigue failure, etc.). Furthermore, the data and subsequent analysis should be used in
conjunction with specific (albeit still to be properly quantified in this context) risk measures
such as reliability, resilience and range of damages.
• The TRI methodology should be extended and applied to different transient event
parameters and not just pressure. Such parameters to consider include: event duration,
negative pressure duration, number of cycles, and transient wave decay.
• While the TRI should be accepted as an initial starting point, additional consideration and
investigation should be given to other (i.e., non TRI) approaches for making beneficial use
of long-term transient pressure data. Any such approach should be aimed at correlating the
frequency of a transient event to the overall risk to the system. Such options may include a
moving window analysis, transient scenario analysis using stochastic events, random event
analysis, choice-constraint analysis, analysis of frequency of loading, etc. For example, can
147
the long-term transient pressure monitoring results be used within a Monte Carlo simulation
to determine a metric such as an annual expected damage? Similarly, can the long-term
transient pressure monitoring results be used to determine the frequency of loading (e.g.,
power failure events)?
• Academia and industry should begin to promote (as well to develop) frequency based risk
assessment methodologies in the realm of hydraulic transients. Decisions must eventually be
correlated to quantifiable metrics in order to truly understand the hydraulic transient
performance of the systems in question.
As demonstrated by the above long list of considerations and recommendations, the current state of
the proposed TRI (or any other quantitative transient risk assessment derived from long-term field
pressure data) is limited and uncertain, but also open for significant discussion. However, with the
proper definitions and additional (i.e., more comprehensive) field data, the answers for many of
these questions and concerns will slowly take form. In the end, the goal of any such analysis would
be to actually quantify the overall in-situ risk due to transients for a variety of traditional water and
wastewater systems.
7.6 Summary
The purpose of this chapter is to establish the need and methodology for a quantitative risk
assessment of transient events in water and wastewater systems. The chapter initially provides
background information on system failure types, and establishes the general idea that a transient risk
is not confined to a single rare event of high consequence, but to the combined impact of all events.
With that in mind, the chapter proposes a trial risk assessment methodology relying on an index-
based metric. Sample guidelines are provided for the recording of transient pressures and for the
definition of both single and multiple transient event parameters. Initial sample results are provided
under the knowledge that several externalities currently play a key role, and that many limitations
currently still exist. The chapter concludes with a discussion and recommendation of items that
should be considered in order to move this quantitative risk assessment approach forward.
148
Chapter 8 Summary and Conclusions
Water and wastewater systems are an integral part of the overall infrastructure network and they play
a significant role in the welfare and progress of society. The analysis and design of such systems has
been at the forefront of Civil Engineering for centuries, and still continues to attract innovations and
intellectual thought in both academia and industry. The purpose of this thesis is to bring specific
attention to a small, often minimized, but quite important segment of water and wastewater system
design – the task of performing a hydraulic transient analysis. The overriding theme of this thesis is
the need to increase the use of field based hydraulic transient assessments as a means of
supplementing the now default standard of numerical modeling.
This thesis document provides the basic background for understanding what hydraulic transients
are, why they are important, and what should be done about them. More specifically, the thesis
discusses the nature of, and difference between, hydraulic transients in the context of three common
types of pressurized municipal systems: water transmission, water distribution and wastewater
forcemain. As a means of illustrating the risk and subsequent prevention/mitigation, a long list and
discussion of potential transient protection options and alternatives is presented.
The thesis document identifies the traditional and current methods for conducting a hydraulic
transient analysis, and proposes a hybrid alternative in which actual in-situ field assessment data can
be used in combination with numerical modeling and standard convention to yield the desired
output – a well designed and protected water and wastewater system. The thesis document also
provides a focused critical discussion of some of the shortcomings in the current approach; one in
which numerical models are typically used to assess the hydraulic transient risk. The discussion
includes consideration of steady state model assumptions, the non-deterministic nature of water
demand, and the traditional worst-case nature of transient loading design. Furthermore, this
discussion also identifies the key differences between water and wastewater transient analyses and
the re-emphasizes the overall need to improve the in-situ performance monitoring of real systems.
Following a brief aside on the field based quantification of the water quality risk from pathogen
intrusion via flooded air valve chambers during low pressure transient events, the thesis document
moves on to provide a comparison and assessment of modern transient pressure monitoring
149
technology. The document provides a detailed and comprehensive comparison of six (6)
commercially available transient pressure monitors on the criteria of long-term field based transient
pressure recording. With this objective in mind, a single technology is shown to demonstrate the
greatest current potential, and the details of this technology, including actual case studies and
examples, are then presented in order to establish the capability of a field driven transient risk
assessment. The long-term transient pressure data is supplemented by data derived from transient
field tests in order to illustrate the difference between, and the need for, the proposed numerical and
field validation approach. This objective is also achieved through four (4) distinct case studies;
studies which cover a wide range of system types and transient based system variables.
The last section of the thesis provides an introduction to one option by which the long-term
transient pressure data for a system could be used to evaluate the overall risk due to hydraulic
transients. The approach is established as a means by which the shortcomings in the theoretical
assumptions could be overcome via a quantitative analysis in the form of a Transient Risk Index
(TRI). The proposed TRI methodology first establishes the required definitions for both the
recording logic and the general properties of a transient event. It then proceeds to illustrate how
such an index-based metric could actually be calculated and how this overall approach can be
beneficially implemented, used, and improved.
While a significant portion of the thesis document cannot be considered as completely innovative
from an academic point of view, it is the author’s strong belief that the nature and synthesis of the
information is indeed an indication of a significant shortcoming in the current (and the direction of
the future) industry practice with respect to hydraulic transient analysis. Even with that important
point in mind, the thesis document does provide several key and additional contributions to both
academia and industry, and these include the following:
1. The identification of current hydraulic transient analysis shortcomings and the need
for a revised hybrid approach that relies on field work such as transient pressure
monitoring to compliment and validate traditional numerical hydraulic transient
models.
150
2. A comparison and identification of suitable technology for assessing the long-term
hydraulic transient performance of water and wastewater systems in the field, including
the detailed approach and sample recording logic for quick implementation.
3. The identification of the benefits and the establishment of the initial framework for
both short-term and long-term hydraulic transient pressure monitoring and risk
assessment. Inherent in this is the establishment of the need for a revised and
quantifiable form of a hydraulic transient assessment.
4. The establishment of the preliminary methodology for an index based hydraulic
transient risk assessment that relies on data derived from long-term continuous
transient pressure monitoring.
The last chapter of the thesis provides a discussion of the uncertainties and current limitations
of the proposed transient risk assessment approach. As part of the discussion, a list of future
work considerations is provided such that the TRI can be improved, tested, and implemented.
In the end, the goal of any such assessment methodology is to make good use of valuable in-
situ data for the purpose of ultimately improving the performance, efficiency, reliability and
safety of the system at hand. It is exactly this theme – the need to close the loop between
transient analysis and actual system performance – that is the overall purpose and focus of this
thesis document.
151
References
Adams, B. J., and Papa, F. (2000). Urban Stormwater Management Planning with Analytical Probabilistic Models. John Wiley & Sons, Inc., New York, U.S.A. ANSI/AWWA. (1997). “C900 Standard: Polyvinyl Chloride (PVC) Pressure Pipe and Fabricated Fittings, 4 in. through 48 in. for Water Distribution,” American Water Works Association, Denver, Colorado, U.S.A. Axworthy D. H., and Chabot, N. (2004). “Pressure Transients in a Canadian Sewage Force Main.” Canadian Journal of Civil Engineering, 31:1039. Babayan, A. V., Kapelan, Z., Savic, D. A., and Walters, G. A. (2005). “Least Cost Design of Robust Water Distribution Networks Under Demand Uncertainty.” Journal of Water Resources Planning and Management, 131:5:375(8). Benjamin, J., and Cornell, C. A. (1970). Probability, Statistics, and Decisions for Civil Engineers. McGraw-Hill, New York, U.S.A. Bergant, A., Simpson, A. R., and Tijsseling A. S. (2006). “Water Hammer with Water Column Separation: A Historical Review.” Journal of Fluids and Structure, 22:135. Bergant, A., and Simpson A. R. (1999). “Pipeline Column Separation Flow Regimes.” Journal of Hydraulic Engineering, ASCE, 125:8:835. Besner, M. C. (2007). “Risk Evaluation of Drinking Water Distribution System Contamination Due to Operation and Maintenance Activities.” Ph.D. Dissertation, Ecole Polytechnique de Montreal, Montreal, Quebec, Canada. Bolous, P. F., Lansey, K. E., and Karney, B. W. (2006). Comprehensive Water Distribution Systems Analysis Handbook for Engineers and Planners, Second Edition. MWH Soft, Pasadena, California, U.S.A. Bolous, P. F., Karney, B. W., Wood, D. J., and Lingireddy S. (2005). "Hydraulic Transient Guidelines for Protecting Water Distribution Systems." Journal of AWWA, 97:5:111. Bowman, J. A. (1990). “The Fatigue Response of Polyvinyl Chloride and Polyethylene Pipe Systems”. In Buried Plastic Pipe Technology, Buczala and Cassady, eds., ASTM, Baltimore, Maryland, U.S.A. Chaudhry, H. M. (1987). Applied Hydraulic Transients. Van Nostrand Reinhold, New York, U.S.A. Elliot, R. C., and Axworthy D. H. (2009). “Special Considerations in Pressure Surge Analysis and Control for Wastewater Systems.” Proceedings of the 33rd IAHR Congress: Water Engineering for a Sustainable Environment, Vancouver, Canada; August 2009. Filion, Y. R. (2006). “Multi-Objective Stochastic Design of Water Distribution Systems.” Ph.D. Dissertation, University of Toronto, Toronto, Ontario, Canada.
152
Fleming, K. K., Gullick, R. W., Dugandzic, J. P., and LeChevallier, M. W. (2006). Susceptibility of Potable Water Distribution Systems to Negative Pressure Transients. AWWARF, Denver, Colorado, U.S.A. Fraser, S. L., Lim, J., and Donskey, C. J. (2010). Enterococcal Infections. Emedicine Publication, January 7. Retrieved from: http://emedicine.medscape.com/article/216993-overview Giustolisi, O., Laucelli, D., and Colombo, A. (2009). “Deterministic Versus Stochastic Design of Water Distribution Networks.” Journal of Water Resources Planning and Management, 135:2:117(11). Gladwell, M. (1996). “Blowup.” The New Yorker, January 22, 1996; New York, U.S.A. Goulter, I. (1992). “Systems Analysis in Water-Distribution Network Design: From Theory to Practice.” Journal of Water Resources Planning and Management, 118:3:1238(11). Grayman, W. M. (2005). “Incorporating Uncertainty and Variability in Engineering Analysis.” Journal of Water Resources Planning and Management, ASCE, 131:3:158. Jeffrey, J. D., Moser, A. P., and Folkman, S. L. (2004). Long-Term Cyclic Testing of PVC Pipe. Utah State University, Logan, Utah, U.S.A. Jung, B. S., Bolous, P. F., and Wood, D. J. (2007). "Pitfalls of Water Distribution Model Skeletonization for Surge Analysis." Journal of AWWA, 99:2:87. Jung, B. S., Filion, Y. R, Adams, B. J., and Karney, B. W. (2010) “Multi-Objective Optimization with Analytical Probabilistic Assessment to Design Branched Pipeline Systems for Fire Flow Protection”, Under Preparation. Jung, B. S., and Karney, B. W. (2009). "Systematic Surge Protection for Worst-Case Transient Loadings in Water Distribution Systems." Journal of Hydraulic Engineering, ASCE, 135:3:218. Jung, B. S., and Karney, B. W. (2008). "Systematic Exploration of Pipeline Network Calibration Using Transients." Journal of Hydraulic Research, IAHR, 46:1:129. Jung, B. S., and Karney, B. W. (2005). “The Search for the Worst-case Transient Loadings in Water Distribution Systems.” Proceedings of IAHR Conference, Seoul, Korea; September 2005. Jung, B. S., Karney, B. W., Bolous, P. F., and Wood, D. J. (2007). "The Need for Comprehensive Transient Analysis of Distribution Systems." Journal of AWWA, 99:1:112. Karney, B. W., and McInnis, D. (1990). “Transient Analysis of Water Distribution Systems.” Journal of AWWA, 82:7:62. Karney, B. W., and Radulj, D. (2009). “Transient Field Monitoring as a Key Driver for Decision Making and Design.” Proceedings of the 33rd IAHR Congress: Water Engineering for a Sustainable Environment, Vancouver, British Columbia, Canada; August 2009.
153
Karney, B. W., and Radulj, D. (2008). “Assessing the Performance of a Water Transmission System Using an Inverse Transient Method.” Ontario Water Works Association Joint Annual Conference, London, Ontario, Canada; April 2008. Kirby, P. C. (1980). “Surge and Fatigue in Unplasticised PVC Sewer Rising Mains, Plastics and Rubber.” In Plastics and Rubber Materials and Applications. Water Research Centre, 5:1:78(5), London, United Kingdom. Kirmeyer, G., Friedman, M., Martel, K., Howie, D., Lechevallier, M., Abbaszadegan, M., Karim, M., and Funk, J. (2001). “Pathogen Intrusion into the Distribution System.” AWWARF. Denver, Colorado, U.S.A. Lansey, K. E., Ning Duan Mays, L. W., and Yeou-Kung, T. (1989). “Water Distribution System Design Under Uncertainties.” Journal of Water Resources Planning and Management, 115:5:630(16). Lauchlan, C. S., Escarameia, M., May, R. W. P., Burrows, R., and Gahan, C. (2005). “Air in Pipelines: A Literature Review.” Report SR 649, Rev. 2, HR Wallingford Ltd., United Kingdom. Lee, J., Lohani V. K., Dietrich, M., and Loganathan, G. V. (2009). “Low Pressure Propagation at Service Lines”, World Environmental and Water Resources Congress, ASCE. Kansas City, Missouri, U.S.A.; May 2009. Lehr, J. H., and Keeley, J. (2005). Water Encyclopedia: 5 Volume Set. John Wiley & Sons, New York, U.S.A. Lingireddy, S., Wood, D. J., and Zloczower, N. (2004). “Pressure Surges in Pipeline Systems Resulting from Air Releases.” Journal of AWWA, 96:7:88. Marshall, G. P., Brogden, S., and Shepherd, M. A. (1998). Evaluation of Surge and Fatigue Resistance of Poly (Vinyl-Chloride) and Polyethylene Pipeline Materials for use in the UK Water Industry. Pipeline Developments Ltd., Salford, United Kingdom. McInnis, D. A., Karney, B. W., and Axworthy D. H. (2004). TransAM Reference Manual. HydraTek Solutions Inc., Ajax, Ontario, Canada. McKay, M. D., Conover, W. J., and Beckman, R. J. (1979). “A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code.” Technometrics, 21:2:239(7). Ministry of the Environment (1985). Guidelines for the Design of Water Distribution Systems. Environmental Approvals and Project Engineering Branch, Province of Ontario, Toronto, Ontario. Ministry of the Environment (2008). Design Guidelines for Drinking Water Systems. Province of Ontario, Canada. Ministry of the Environment (2008). Design Guidelines for Sewage Works. Province of Ontario, Canada.
154
Mielke, D. R. (2004). “A Guide for the Design of Water Transmission Pipelines.” ASCE, Proceedings of the Pipeline Division Specialty Congress, San Diego, California, August. Murray, S. L., Lecina, S., Thollet, J., and Clark, P. B. (2004). “Surge in Water Systems: Some Issues Facing the Designer.” Proceedings of the 9th International Conference on Pressure Surges, BHR Group, Chester, United Kingdom; March 2004. MWH (2005). Water Treatment: Principles and Design. 2nd Edition, John Wiley & Sons, Hoboken, New Jersey, U.S.A. National Research Council (2006). Drinking Water Distribution Systems: Assessing and Reducing Risks. The National Academic Press, Washington, D.C., U.S.A. National Research Council (2003). “Water Quality in Water Distribution Systems: A Best Practice by the National Guide to Sustainable Municipal Infrastructure.” Issue 10, Ottawa, Ontario, Canada. Opitz, E. M., Langowski, J. F., Dziegielewski, B., Hannah-Sommers, N. A., Willett, J. S., and Hauer, R. J. (1998). "Forecasting Urban Water Use: Models and Application." Urban Water Demand Management and Planning, Baumann D., Boland, J. and Hanemann, W. H., eds., McGraw Hill. New York, U.S.A. O’Rourke, C. E. (1940). General Engineering Handbook. 2nd Edition, McGraw-Hill Book Company, New York, U.S.A. Pipetech International (2008). Transient Pressure Monitoring System Manual. Phoenix, Arizona, U.S.A. Plastics Industry Pipe Association of Australia Limited (2002). Polyethylene Pressure Pipes Design for Dynamic Stresses. Industry Guideline Document Issue 5, Chatswood, NSW, Australia. Radulj, D. (2007). “The Role of Air Valves in Water Distribution Systems.” B.A.Sc. Thesis, University of Toronto, Toronto, Ontario, Canada. Rowe, W. D. (1979). Introduction to Risk Assessment. In Energy Risk Management, G.T. Goodman and W.D. Rowe, eds. Academic. Press, London, United Kingdom, p. 7-19. Trifunovic, N. (2006). Introduction to Urban Water Distribution. Taylor & Francis, Netherlands. Uni-Bell PVC Pipe Association (2001). Handbook of PVC Pipe Design & Construction. Fourth Edition, Dallas, Texas, U.S.A. Viessman, W., and Hammer, M. J. (2005). Water Supply and Pollution Control. 7th Edition, Pearson Prentice Hall, Upper Saddle River, New Jersey, U.S.A. Walski, T. M. (2006). “History of Water Distribution.” Journal of AWWA, 98:3:3. Walski, T. M. (2003). Advanced Water Distribution Modeling and Management. 1st Edition, Haestad Methods Inc., Connecticut, U.S.A.
155
Walski, T. M. (2001). “The Wrong Paradigm - Why Water Distribution Doesn’t Work?” Journal of Water Resources Planning and Management, 127:4:203(3). Walski, T.M., Chase, D. V., and Savic, D. (2001). Water Distribution Modeling. 1st Edition, Haestad Methods Inc., Connecticut, U.S.A. Wood, D. J., Lingireddy, S., Bolous, P. F., Karney, B. W., and McPherson, D.L. (2005). "Numerical Methods for Modeling Transient Flow in Distribution Systems." Journal of AWWA, 97:7:104. Wylie, B. E., and Streeter, V. L. (1993). Fluid Transients. FEB Books, Ann Arbor, Michigan, U.S.A. Ysusi, M. A. (2000). "System Design: An Overview." Water Distribution Systems Handbook, Mays, L. W., ed., McGraw-Hill, New York, U.S.A. Zarghamee, M. S., Eggers D. W., Ojdrovic, R. P., and Rose B. (2003). “Risk Analysis of Prestressed Concrete Cylinder Pipe with Broken Wires.” Proceedings of ASCE Specialty Conference Pipelines 2003, Baltimore, Maryland, U.S.A. Zarghamee, M. S., and Fok, K. (1990). “Analysis of Prestressed Concrete Pipe under Combined Loads.” Journal of Structural Engineering, 116:7:2022.