2016 annual conference & exposition sun05 – … · 2016 annual conference & exposition ....

2016 Annual Conference & Exposition

SUN05 – Designing a Risk-based Pipeline R&R Planning Program Using a

Combination of Inspection and Analytical Approaches

June 19, 2016 Chicago, Illinois

Workshop Agenda

9:00 Welcome Annie Vanrenterghem Raven, infraPLAN 9:05 Introduction Kevin Campanella, Burgess & Niple 9:15 Introduction Questionnaire Kevin Campanella, Burgess & Niple 9:20 AM Framework and Data Kurt Vause, streamlineAM, LLC 9:40 Data Clean-up Exercises Annie Vanrenterghem Raven, infraPLAN 10:30 Break 10:40 Analytical Approach: From Basic to Advanced Approach Non-Inspected and Inspected Pipes Annie Vanrenterghem Raven, infraPLAN Celine Hyer, ARCADIS 11:40 Case Study: Columbus – From Basic to Advanced Approach Non-Inspected Pipes Kevin Campanella, Burgess & Niple noon Lunch 1:00 Q/A About Columbus Kevin Campanella, Burgess & Niple 1:10 Case Study: Dallas – From Basic to Advanced Approach Inspected Pipes Annie Vanrenterghem Raven, infraPLAN Celine Hyer, ARCADIS

1:40 Group Discussion: Where do you stand? Where do you want to go? Celine Hyer, ARCADIS Kurt Vause, streamlineAM, LLC 2:40 Break 2:50 Synthesis of Previous Discussion Kevin Campanella, Burgess & Niple 3:00 Mixing Analytical Approach and Inspection Program Annie Vanrenterghem Raven, infraPLAN 3:25 Case Study: AWWU –Mixing Analytical Approach with Inspection Program Kurt Vause, streamlineAM, LLC 3:55 Wrap-up Annie Vanrenterghem Raven, infraPLAN

Presenter Information

Annie Vanrenterghem Raven, infraPLAN [email protected]

Bio: For the last 20 years, Annie’s research and consulting work have focused on water and waste water infrastructure, addressing the optimal planning of short and long-term rehabilitation projects. She is now the Managing Director of infraPLAN, a firm she created in 2008 that develops analytical models translated in functional software, and provides consulting and training. The goal is to help utilities create comprehensive and Advanced Analytical Asset Management programs, and, ultimately, identify and justify their long-term investments, and short-term projects selection. Annie is a member of the AWWA Asset Management Committee’s Advisory Committee. She holds a Ph.D. in Civil Engineering, Polytechnic Institute of New York University, New York, NY.

Celine Hyer, ARCADIS [email protected]

Bio: Celine is the National Conveyance Market Sector Leader within ARCADIS and is located in their Tampa Florida office. She has a B.S. in Chemical Engineering and an M.S. in Engineering Management from Florida Institute of Technology. Celine has 25 years of experience in Engineering with 15 years that are directly related to asset management program implementations including advanced renewal and replacement planning for pipelines as well as treatment and pumping facilities. Celine currently serves as the Secretary of the AWWA Asset Management Committee and as the Chairwoman of the Industry Advisory Board for the Sustainable Water Infrastructure Center at Virginia Tech.

Kurt Vause, streamlineAM [email protected]

Bio: From 1998 - 2015, Kurt was Engineering Division Director of Anchorage Water Wastewater Utility. He was responsible for AWWU’s capital construction program, its Grants & Loans section, and the utility’s Strategic Asset Services and Planning sections. In 2016, he became Special Projects Director providing planning, integration, and execution of strategic utility initiatives. Kurt currently serves on American Water Works Association’s Water Utility Council, presently serving as the Regulatory Subcommittee chair and is Incoming Chair of the Council. In addition, he is a member and Vice-Chair of AWWA’s Asset Management Committee. He also served on the 2012 International Water Association - Water Supply Association of Australia Asset Management Best Practices Benchmarking Project Steering

mailto:[email protected]



Committee. He is co-founder of StreamlineAM, LLC, an Alaskan-based consulting service dedicated to utility management, asset management, and engineering for the water sector.

Kevin Campanella, BURGESS & NIPLE [email protected]

Bio: Kevin has been in the asset management services field for 14 years. In 2002, he moved to New Zealand to work with Meritec, joining some of the authors of the original International Infrastructure Management Manual. From 2003 to 2008, he worked for Brown and Caldwell, developing asset management solutions with the Columbus Division of Sewerage and Drainage to address their CMOM consent order. For seven years beginning in 2008, he led the comprehensive asset management program at the Columbus Department of Public Utilities as an Assistant Director. Kevin is a member of the AWWA Asset Management Committee’s Advisory Committee, is chair of the “Progress in Asset Management Survey” subcommittee, and is Vice Chair of the Ohio AWWA Asset Management Committee.


Kevin Campanella

Celine Hyer

Annie Vanrenterghem Raven

Kurt Vause

1

Designing a Risk-based Pipeline R&R Planning

Program using a Combination of Inspection and

Analytical Approaches

Our Speakers

2

Annie Vanrenterghem RaveninfraPLAN

[email protected]

Celine HyerARCADIS

[email protected]

Kurt VausestreamlineAM

[email protected]

Kevin CampanellaBURGESS & NIPLE

[email protected]

2

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20

Introduction Questionnaire

Kevin

9:20-9:40

AM Framework and Data

Kurt

9:40-10:30

Data clean-up Exercises

Annie et al

10:30-10:40

BREAK

10:40-11:40

Analytical Approach – From Basic to Advanced Approach

non-inspected and Inspected Pipes

Annie and Celine

11:40-12:00

Case Study: Columbus – From Basic to Advanced Approach

non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40

Case Study: Dallas – From Basic to Advanced Approach

Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion

Where do you stand? Where do you want to go?

Celine, Kevin, Kurt

2:40-2:50

BREAK

2:50-3:00

Synthesis of Previous Discussion

Kevin

3:00-3:25

Mixing Analytical Approaches and Inspection Program

Annie

3:25-3:55

Case Study: AWWU – From Basic to Advanced Approach

Mixing Analytical Approach with Inspection Program

Kurt

3:55-4:00 Wrap-up

Annie

3

12:00-1:00LUNCH

Workshop Agenda

3

44

5

6

7

8

9

10

11

What you will learn?

12

You will learn:

1. about approaches, from basic to advanced, to manage in a cost efficientway the planning of Inspection, Rehabilitation & Replacement (R&R) of water pipelines using utility-specific data whether they are inspected or not

2. about tradeoffs inherent to these approaches in terms of benefits(increasing accuracy of when and what to inspect, rehabilitate or replace; and avoiding risk of asset failure) and costs (data requirements and level of effort)

3. to identify, mine, clean-up, format and analyze the data needed to successfully use these advanced approaches

4. where your organization stands and where it wants to go

5. from experts who have been applying these approaches and, as a result, have specific case studies and lessons to share

Why is this an important topic?

• Linear buried water assets = 2/3 of the value of water infrastructure

• Substantial percentage is approaching end of Effective Useful Life (EUL)

• Discrepancy between projected cost of next CIP and R&R needs, and analytical justification or reliance on analytics, to optimize program

• $ will need to be -and can- be stretched using powerful and convincingammunitions!

• Physical Condition Assessment (PCA) still $$ and proportional to length reliance on analytics $ is not proportional to length

13

Why should you prefer advanced approaches?

• More accurate projections (but, not always longer EULs)

• Optimization of resources and savings (using risk-based approach)

• Possibility to explore many different long term planning options

• More granularity in results - larger choice of projects to prioritize

• More defensible, justifiable, credible budget projections

14

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

2014 2050 2100

Co

st $

M

Year

Industry-Assumed EULs

Cumulative Cost up to 2050 and 2100 based on Industry-Assumed EULs

15

System North East US with medium average break rate - 1,200 mi

$1.8B

$4B

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

2014 2050 2100

Co

st $

M

YearIndustry-Assumed EULs Utility- and Risk-Specific EULs

65% Reduction ($1.2B)

50% Reduction ($2B)

Results with Advanced ApproachUtility- and Risk-Specific EULs

16

A few words of wisdom about data • Data quality matters with any approach or any software, from basic to advanced.

• Even a simple break rate will be wrong if definition of breaks, number of breaks, length of pipes, and number of years are wrong.

• Precision is not so much the goal (but precise results are of course preferred!). We do not want wrong estimates. Which is what using generic and assumed data, andsimplistic approaches can lead to.

• It does not take more time or effort to collect the right (break and pipe) data in the right format.

• It does take time to clean-up old data (never collected for planning) but this is tobe done once for all.

• Once data is Okay (old one cleaned-up; new one properly restructured andformatted) why not use appropriate and advanced approaches that provide more reliable and defensible answers; that make requests more credible and easier tojustify?

17

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40

Analytical Approach – From Basic to Advanced –

Non-inspected and Inspected Pipes

Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion

Where do you stand? Celine

Where do you want to go? Kurt

2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25

Mixing Analytical Approach and Inspection Program

Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

18

12:00-1:00LUNCH

Workshop Agenda

Questionnaire

• Handout 1

• Presentation of questionnaire that will be used for Celine, Kevin and Kurt’ session in the afternoonWhere does your organization stand?

Where does your organization want to go?

• While attending the upcoming presentations, reflect on where your organization stands and where it would like to go; the data, tools, resources currently at your disposal (or not).

19

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

20

12:00-1:00LUNCH

Workshop Agenda

FrameworkFrom Basic to Advanced

21

Assumed

EULs

InventoryLong-Term

Plan

Basic Framework

22

Cost

R&R

COF

Weighted

Score

LOF

Weighted

Score

Risk Score and

Short-Term

Prioritization Plan

Answer Q1

How much money

is needed until

2050?

Answer Q2

What projects

should be

addressed first?

Advanced Framework

• Non-inspected pipes failure data

• Inspected pipes inspection records

23

Non-Inspected Pipes

24

Break Cons.

of

Failure

GIS Pipes

Failure

Forecasting

CMMS Aging

Function

EULs

Long –Term

R&R Plan

Criticality and

Risk Score

Likelihood

of

Failure

Asset

Management

Plan

Clean-up

Statistics

GIS Breaks

Short -Term

Prioritization

Plan

Utility Data

Software-

algorithm

Output - Input

Where is

info about

pipes

attributes?

Where is

info about

breaks?

How is future

behavior

predicted?

What are the

output results

from a failure

forecasting

model?

Answer Q1

How much money

is needed until

2050?

Answer Q2

What projects

should be

addressed first?

Inventory

Hydraulic

CriticalityCost

R&R

Other

Criteria

Inspected Pipes

25

GIS Pipes

Physical

Condition

Forecasting

CMMS

Likelihood

of

Failure

Clean-up

Statistics

Inspection records

Utility Data

Software-

algorithm

Output - Input

Where is

info about

pipes

attributes?

Where is info

about

inspection?

How is future

behavior

predicted?

Inspected Pipes

26

Cost

Failure Cons.

of

Failure

GIS Pipes

Physical

Condition

Forecasting

CMMS Aging

Function

EULs

Long –Term R&R

and Inspection

Plan

Criticality and

Risk Score

Likelihood

of

Failure

Asset

Management

Plan

Clean-up

Statistics

Inspection records

Short -Term

Prioritization

Plan

Utility Data

Software-

algorithm

Output - Input

Where is

info about

pipes

attributes?

Where is

info about

inspection

?

How is future

behavior

predicted?

Inventory

Hydraulic

CriticalityCost

R&R & Insp.

Other

Criteria

Key Elements of the AM Plan

27

Cost

Break

Cons.

of

Failure

GIS Pipes

Failure /PCA

Forecasting

CMMS Aging

Function

EULs

Long -Term

Plan

Criticality and

Risk Score

Likelihood

of

Failure

Asset

Management

Plan

clean-up

Statistics

GIS Breaks

Short -Term

Prioritization

Plan

Inventory

Hydraulic

CriticalityR&R

LTP STP

Aging

Curve

EUL

COF

LOF

Other

Criteria

Data

28

Pipe Data

• System level

Inventory from purchasing records

(pre GIS).

• Pipe level

Pipe Data Base (DB) or GIS.

Pipe-level statistics.

• Active (ACT) and Abandoned

(ABN) pipes

Better failure statistics

29

FEATID DOI DISTRICT DIAM MAT LENGTH LIFE STATUS

DOA SOIL ROAD

17608 1/1/1934 Ridgefield 2 SCI 0.0033 ABN 12/20/2003 1 2



10946 5/31/1949 Bridgeport 8 SCI 0.0967 ACT 2 2

10947 1/10/1961 Bridgeport 8 SCI 0.0681 ACT 2 3

10948 6/9/1905 Bridgeport 24 PCI 0.1992 ACT 1 3



10952 11/27/1963 Stratford 8 SCI 0.058 ACT 3 1

Non-Inspected PipesBreak Data

• Break DataLocation, Type, Date Of Break (DOB)

• Not geocoded/pre GIS– Break reports/CMMS– Address; no pipe ID; not connected to the pipe data

generic failure statistics aging function and Effective Useful Life (EUL) have to beassumed

• Geocoded/GIS– Same + Pipe ID

Pipe linked to YOI and Age Time-related statistics aging function and EUL computed

• Active (ACT) and Abandoned (ABN) better failure statistics, forecast and aging functions if ABN data is available.

30

BREAK DATA

PRE GISLOCATION MAT DIAM DATE OF BREAK

10 Main street CI 12 7-Jun-15

POST GIS101 CI 12 7-Jun-15

PIPE DATAPIPE ID MAT DIAM YOI LENGTH

101 CI 12 1900 200

Importance of Pipe and Break Information about Abandonment

Imagine worse pipes have been abandoned between 85 and 95 but abandonment data is not available.

we will predict behavior of all mains that are now between 65-75, when they are 85-95, based on best pipes; not whole array

optimistic EULs

under-estimated R&R predictions

Example:• pipes installed between 1920 and 1950 are now between 65 and 95 yrs old.

• pipes degrade same way (same cohort) but for each YOI there are better or worse pipes.

• we have 10 years of breaks 2005-2015.

• when pipes reach high break rate (between 85 and 95) they are replaced. Between 2005-2015 this applied to pipes installed between 1920 and 1930.

• by age of 85-95 best pipes are left.

• EUL 105 instead of 95

31

Break Rate based on Year of Installation (ACT Mains Only)

32

0

5

10

15

20

25

30

35

40

192

0

192

1

192

2

192

3

192

4

192

5

192

6

192

7

192

8

192

9

193

0

193

1

193

2

193

3

193

4

193

5

193

6

193

7

193

8

193

9

194

0

194

1

194

2

194

3

194

4

194

5

194

6

194

7

194

8

194

9

195

0

Bre

ak R

ate

(Nb

Bre

aks/

mi/

yr)

Year of Installation

Break Rate based on Year of Installation

Break Rate Based on Age at Year of Break

0

5

10

15

20

25

30

35

40

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

Bre

ak R

ate

(Nb

Bre

aks/

10

0 m

i/yr

)

Age at Year of Break

Break Rate based on Age at Year Of Break

Break Rate 1 - no info on RPLT Break Rate 2 - info on RPLT

Expon. (Break Rate 1 - no info on RPLT) Expon. (Break Rate 2 - info on RPLT)

95 105

20

Inspected Pipes Inspection Data

• Inspection Data

– Location, Type (technology, contractor),

Date of Inspection

• Data Base (address) or GIS (ID)

– Same issue as with non-inspected pipes

33

Major Differences between Break and Inspection Data

• Need break history for all pipes and all years during window of observation of breaks

• Not the case for inspection data: can be any year and any pipe

34

0

50

100

150

200

250

300

0

0.05

0.1

0.15

0.2

0.25

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

08

20

10

20

12

20

14

20

16

Len

gth

(m

i)

Bre

ak R

ate

(Nb

Bre

aks/

mi/

yr.)

YOB

Net Cumulative Length and Break Rate based on YOB

Break Rate Length Linear (Break Rate)

Zip Code Breaks 1990-2015 Length10014 200 10010015 250 10010016 300 10010017 150 10010018 220 10010019 400 10010020 2 100

PipeID Inspection Date Inspection Score101 1990 1101 2005 2101 2010 3102 2015 3103

Breaks missing in

certain areas

and at certain years

Pipes not all inspected and not every year

Main Issues with Pipe, Break and Inspection Data

• Issues

– Definition

– Incomplete

– Missing

– Assumed

– Incorrect

– Incoherent

– Questionable

– Limited

– Consistent

• How to identify and clean-up issues? What does it take?

– Manually or diagnostic /clean-up algorithms

– Assumptions/statistics

– clean-up may require going back to some historical records35

OBJECTID PlacementI PlacementS

Placement

D

Placement

M SourceScal UpdateId

UpdateDat

e

UpdateMet

h UpdateSour FeatureId DateInsta l Dis trict

19 WLS FLDSKETCH 10/17/2006 DIGITIZED NTS sdemezzo 7/20/2011 ORIGINAL ORIGINAL 292143 2109000



29 WLS FLDSKETCH 10/18/2006 DIGITIZED NTS ORIGINAL ORIGINAL 292207 1/1/2000 2109000




















57 NA LEGACY 10/21/2004 MIGRATED 500 WOOLPERT 7/7/2005 DIGITIZED CONVERSION 126231 1/1/2000 2109000

58 NA LEGACY 10/21/2004 MIGRATED 500 sdemezzo 7/20/2011 DIGITIZED CONVERSION 126234 2109000

59 WOOLPERT CONVERSION 7/7/2005 DIGITIZED UNK sdemezzo 7/20/2011 ORIGINAL ORIGINAL 306616 2109000

Data QualityPast and Future

• 2 levels of data:

– Past

– Future

• 2 Goals for data diagnostic :

– Identify issues and clean-up past data (eventually)

– Restructure future data (and never have to clean-up data again!)

36

Other Data

• Operations/hydraulic– hydraulic capacity

– fire flow

– pressure

– service points

– consumption

– water quality

– leaks

• Service – customers criticality

– complaints (shortage, water quality, frequency of construction projects)

– planned work for sewers or pavement

37

• Cost – repair (from basic to very advanced

if indirect and social costs included)

– rehabilitation

– replacement

• Environmental/location – soil

– traffic

– population density/construction

– sensitive targets (rail track, subway entrance, tunnel)

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

38

12:00-1:00LUNCH

Workshop Agenda

Real Life Data Data Clean-up Exercise

• Non-inspected– Pipe and break data

– Pipe and break data quality diagnostic

– Identify issues in sample datasets – handout 2

• Inspected– Pipe and inspection data

– Pipe and inspection data quality diagnostic

– Identify issues in sample datasets – handout 3

39

Non-inspected Pipe and Break Data

• 3,194 mi system in North East (10% rejected)

• 2,105 breaks ranging between 2007 and 2013 (13% rejected)

• ACT and ABN pipe data

40

Pipes Issues - Summary

41

SELECTION OR QUALITY ISSUESAbandoned before installed. DIAM < 4. Duplicate Feature ID. Early abandonment (< 3). L < 0.0002. No DIAM. No DOI. No MAT. Suspicious DOI or MAT.

No Issue 2,870.93 90%Issues 323.39 10%Total 3,194.32 1

Pipe Issues-

Breakdown

42

Issues Length (mi)Abandoned before installed. 0.14

DIAM < 4. 24.78

DIAM < 4. L < 0.0002. 0

DIAM < 4. Suspicious DOI or MAT. 0.01

Duplicate Feature ID. 4.89

Duplicate Feature ID. DIAM < 4. 0.65

Duplicate Feature ID. Early abandonment (< 3). 0.05

Early abandonment (< 3). 1.63

L < 0.0002. 0.01

No DIAM. 6.19

No DIAM. L < 0.0002. 0

No DOI. 100.34

No DOI. DIAM < 4. 16.24

No DOI. Duplicate Feature ID. 0.46

No DOI. Duplicate Feature ID. DIAM < 4. 0.46

No DOI. L < 0.0002. 0

No DOI. No DIAM. 0.22

No MAT. 1.42

No MAT. DIAM < 4. 0.05

No MAT. No DIAM. 0.06

No MAT. No DOI. 134.75

No MAT. No DOI. DIAM < 4. 12.53

No MAT. No DOI. Duplicate Feature ID. 0.12

No MAT. No DOI. Duplicate Feature ID. No DIAM. 0.25

No MAT. No DOI. L < 0.0002. 0

No MAT. No DOI. No DIAM. 7.57

Suspicious DOI or MAT. 10.54

No Issue 2,870.93

Total 3,194.32

Break Issues - Summary

43

TYPE BREAK NB BREAKS 2007-2013

Circumferential 1,067

Leak 236

Longitudinal 395

Third Party 70

UNK 337

TOTAL 2,105

STATUS BREAKS NB BREAKS % NB BREAKS

N 264 13.0%

Y 1,771 87.0%

TOTAL 2,035 100.0%

ISSUES BREAKS

Bad DOB: DOB < DOI.

Break after abandon.

Duplicate Feature ID. Same DOB. Duplicate Break ID.

Pipe N.

Same Feature ID. Same DOB. Different Break ID.

YOB = YOI.

Break Issues - Breakdown

44

ISSUES BREAKS NUMBER %

Bad DOB: DOB < DOI. 31 1.50%

Break after abandon. 1 0.00%

Duplicate Feature ID. Same DOB. Duplicate Break ID. 25 1.20%

OKAY 1,753 86.10%

Pipe N. 181 8.90%

Pipe N. Bad DOB: DOB < DOI. 1 0.00%

Pipe N. Break after abandon. 2 0.10%

Pipe not in pipe DB. 9 0.40%

Same Feature ID. Same DOB. Different Break ID. 18 0.90%

Same Feature ID. Same DOB. Different Break ID. Bad DOB: DOB < DOI. 2 0.10%

Same Feature ID. Same DOB. Different Break ID. Pipe N. 1 0.00%

YOB = YOI. 11 0.50%

TOTAL 2,035 100.00%

Identify issues in pipe and break sample datasets

• Take handout 2

45

Inspected Pipes - Inspection Data

• 2, 303 pipes for a total of 233.5 mi

• 1,348 inspections between 2011 and 2015; 4 technologies/contractors on 962 pipes inspected 1 to 5 times

• After clean-up of pipes 2,273 left (99%).

• After clean-up of inspections, and one inspection kept per pipe, 803 inspections left (60%)

• 84 mi of pipes with valid inspections kept for analysis = Analysis achieved with 36% of length inspected

46

Pipe Issues - Summary

47

SELECTION OR QUALITY ISSUESDiscrepancies between various sources of pipe data.Duplicate Feature ID. No DIAM. No DOI. No MAT. Suspicious DOI or MAT.

Inspection Issues - Summary

48

ISSUES INSPECTIONS

Early inspection.

Low score on older pipe.

High score on young pipe.

Duplicate Feature ID. Same DOINSP.

Inspection assigned to wrong pipe.

No pipe.

Discrepancy between pipe data and inspection data.Discrepancy between pipe attributes on multiple inspection records.

Incoherencies between scores on multiple inspections.

Suspicious length of inspection.

Change of pipe ID.

Split pipe.

Sonar results.

INSPECTION KEPT REMOVEDUNK 4CES 102 16ESG 436 53IMX 464RZR 261 12

Pipe and Inspection Issues Breakdown

49

STATUS PIPES KEPT FOR ANALYSISNUMBER OF PIPES

KEPT

STATUS PIPES REMOVED FROM ANALYSIS

NUMBER OF PIPES

REMOVED

1 INSPECTION KEPT. 600

1 INSPECTION REMOVED. 71NO INSPECTION.

6

SEVERAL INSPECTIONS (2-5). ONE INSPECTION LEFT.

264 2 INSPECTIONS. NO INSPECTIONS LEFT. 2

SEVERAL INSPECTIONS. NO INSPECTION LEFT.

2

RE-ASSIGNED. 1 INSPECTION LEFT. 1 RE-ASSIGNED. NO INSPECTION LEFT. 20

DUPLICATE. NO INSPECTION LEFT. 2

NO INSPECTION. 1,335

TOTAL NUMBER OF PIPES 2,273 30

Identify issues in pipe and inspection sample datasets

• Take handout 3

50

How do we fix past problems?

Combination of approaches that are:

• Logical: Using statistics; for example for missing MAT: study the distribution of MAT of pipes with all data based on location, YOI, DIAM

• Manual: Consult maps, design or break records, or field personnel

• Automatic: Algorithm run 1 check 1 fix 1 run 2, etc.

Work in progress:

• Set goals

• Create quality indicators

• Change processes and data structure

51

Break

52

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40

Analytical Approach – From Basic to Advanced


Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

53

12:00-1:00LUNCH

Workshop Agenda

Analytical ApproachesBasic - Advanced

• Level 1: Basic Approach – Minimum Data

• Level 2: Advanced Data and Approach

• Likelihood of Failure (LOF)

• Consequence of Failure (COF)

• Short-Term Prioritization Plan

• Aging Function/Effective Useful Life

• Long-Term Plan (LTP)

54

Level 1 - Basic• Approach

– LOF, COF, STP = weighted score

– EUL and aging function = drawn with assumed data

– LTP = use simple simulation tool (most simple BNL)

– Same for non-inspected and inspected pipes

• Plus

– Simple; little data needed: may be no knowledge of breaks,

or breaks not assigned to pipes; and no or little inspection yet.

– Most can be computed in Excel. GIS data not necessary.

– If GIS available, attributes and results can be visualized.

– EUL and LTP: can be done at cohort level with assumed EUL values.

• Minus

– Little differentiation

– Weights subjective (OK for COF and STP; serious problem for LOF)

55

Word of wisdomRegard Level 1 as a starting point; use it to better understand and improve data

quality and structure, and aim towards more advanced approach.

Likelihood of Failure (LOF)

56


• Approach: Weighted Score

⁻ Define LOF criteria

⁻ Assign weight (wi) to each criterion

⁻ Calculate LOF score = Sum (wi x LOFi)

• Comments

⁻ If breaks not assigned to pipe, number of

previous breaks not a criterion

⁻ Score difficult to compute especially if

multiple criteria

⁻ Estimates often quite off. Example:

Avg. Score = 2.6; Peak= 3; Weighted 3.5

Gap between Weighted and Avg.:

⁻ 6% off by 1 pt.

⁻ the rest is off by 2 pts or more

⁻ 13% off by 4 pts!

57

Example of Weighted LOF Scores at Cohort Level

0

50

100

150

200

250

300

350

-4 -3 -2 -1 0 1 2 3 4

Gap Between Weighted Score and Real Readings

Weighted - Peak Weighted - PACP

What data could be used to define Likelihood of Failure (LOF) score?• Operations/hydraulic

– hydraulic capacity– fire flow– pressure – service points– consumption– water quality– leaks

• Service – customers criticality– complaints (shortage, water quality,

frequency of construction projects)– planned work for sewers or

pavement


if indirect and social costs included)– rehabilitation– replacement

58

• Environmental/location – soil– traffic– population density/construction– sensitive targets (rail track, subway

entrance, tunnel)

• Pipes– material– diameter– year of Installation/age– year of Abandonment– Length

• Breaks– type– date– pipe

Weighted LOF - Example

• Take handout 4

59

Consequence of Failure (COF)

60


• Approaches: Weighted Scores

⁻ Define COF criteria

⁻ Assign weight (wi) to each criterion

⁻ Calculate COF score = Sum (wi x COFi)

• Comments:⁻ A lot of data to gather. Make it

incremental: start with simple criteria with limited data and build from there over time

⁻ Difficult to evaluate social and indirect impact

⁻ Little differentiation⁻ 1 x 100 = 10 x 10⁻ Weights are subjective but no

analytical way to come up with them (unlike LOF scores).

61

Triple Bottom Line Approach for Criteria Selection

66%

20%

14%

COF

1 - Low 2 - Medium 3 - High

What data could be used to define Consequence of Failure (COF) score?• Operations/hydraulic




pavement


if indirect and social costs included)– rehabilitation– replacement

62


entrance, tunnel)

• Pipes– material– diameter– year of installation– year of abandonment– Length


Weighted COF - Example

• Take handout 5

63

Short-Term Prioritization (STP)

64


• Approach: Weighted Scores

⁻ Calculate Risk = Likelihood of Failure (LOF) x


⁻ Define opportunistic (non-risk) criteria

⁻ Estimate weights

⁻ Establish priorities that integrate both risk-based

and non-risk criteria

• Comment

⁻ Quality depends on quality of COF and LOF

⁻ Treat risk and non risk criteria differently

65

Identify non-Risk Prioritization Criteria• Operations/hydraulic




pavement


if indirect and social costs included)– rehabilitation– replacement 66


entrance, tunnel)

• Pipe– material– diameter– year of Installation– year of Abandonment– Length


Assign Prioritization Scores

67

Take handout 6

Aging Function/Effective Useful Live (EUL)

68

Aging Function/Effective Useful Live (EUL)⁻ Represents the break rate/average inspection score based on age⁻ One curve per pipe group

69

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100 120

Bre

ak R

ate

Age

Break Rate based on Age

EUL =85 @ 40 brks/100 miles/yr

Aging Function/Effective Useful Live (EUL)Assumed

70

• Data Requirements/Tools: 2 points;EUL assumed, service level, Excel

• Plus: simple

• Minus: not precise; different utilitiesmay end up with same EULs.

Buried No Longer EUL Table

Service Level; assumed EUL

Current Break rate + age

Long-Term Plan (LTP)

71

Long-Term Plan Buried No Longer (BNL) - Level 1

• Approach– Determine R&R needs per year based on

assumed EULs

• Data and Tool requirements– Assumed EULs per cohort– Inventory – Cost data– Excel or BNL Tool

• Plus– Little data needed– Fast estimate

• Minus– Not utility-specific– Not linked to performance of system– Probably imprecise

72

Chart Legend

Water Mains

Sewers

WTP

WWTP

WW Lift Stations

Period

Average$/yr

Escalated*

* Average escalated annual R&R expenditure assuming

3% annual inflation

5 Year (09-13)

10 Year (09-18)

20 Year (09-28)

$1.1m

$1.1m

$1.4m

$/yr Non -

Escalated

$1.1m

$1.0m

$1.1m

$0

$500,000

$1,000,000

$1,500,000

$2,000,000

$2,500,000

$3,000,000

$3,500,000

$4,000,000

$4,500,000

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028

Water Line Assets $96,597 $99,495 $102,48 $105,55 $108,72 $111,98 $613,70 $632,11 $651,07 $670,60 $560,90 $577,73 $595,06 $612,92 $631,30 $650,24 $91,835 $94,590 $97,428 $100,35

Water Treatment Plant Assets $972,68 $391,60 $0 $9,976 $6,447 $36,314 $154,95 $566,21 $110,88 $66,096 $178,49 $445,50 $40,579 $43,899 $43,050 $1,044, $106,56 $141,12 $27,426 $2,072,

Water Pump Station Assets $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0

Water Miscellaneous Assets $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0

Sewer Line Assets $3,076 $3,168 $3,263 $3,361 $3,462 $3,566 $9,544 $9,830 $10,125 $10,429 $6,608 $6,807 $7,011 $7,221 $7,438 $7,661 $1,123, $1,156, $1,191, $1,227,

Sewer Treatment Plant Assets $28,461 $2,193, $0 $306,89 $19,139 $0 $107,50 $477,72 $826,73 $21,020 $21,169 $407,09 $210,29 $16,429 $217,68 $153,94 $15,224 $434,32 $69,480 $186,12

Sewer Pump Station Assets $731,21 $22,124 $44,627 $129,09 $149,08 $76,779 $17,099 $33,022 $195,23 $165,82 $589,37 $929,66 $209,27 $212,92 $341,14 $276,08 $63,193 $109,46 $134,08 $235,40

Sewer Miscellaneous Assets $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0

Average $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442, $1,442,

R&

R C

on

trib

uti

on

Fiscal Year

Annual R&R Contribution Requirement

Example LTP by pipe material

Level 2 - Advanced

73

• Approach

– LOF2: Failure/PC forecasting model (different for inspected and non-inspected pipes)

– COF2: Ranking tree or monetized

– STP2: LOF2 x COF2 or MCDMM

– EUL and aging function = drawn from statistics and forecasting model

– LTP = More advanced simulation tool (could be pipe-level and GIS-based)

• Plus

– No guessing of weights for LOF; model does it!

– Results at pipe level

– Better differentiation of LOF, COF, STP

– More utility-specific EULs and aging curves

• Minus

– Data needed at pipe level

– Data must be thoroughly cleaned-up (it could be a plus!)

– Minimum 5 yrs. of breaks; 10-25% of pipes inspected (depending)

– Software and expertise needed – look for interactive tool


74

Likelihood of Failure (LOF)• Approach: Multivariable Regression Model

– Run descriptive statistics, calibrate and validate model

– Use model that takes all failure/PCA factors into account simultaneously

• Data Requirements/Tools

⁻ Pipe and environmental data (active –ACT- and, for

non-inspected, if available, abandoned –ABN-)

⁻ Breaks/inspection scores assigned to pipes

⁻ Statistical model/software

• Plus

- Can take RPLT (non-inspected pipes) into account

• Minus

⁻ Preliminary statistics and expertise to calibrate

model

75

0

10

20

30

0

0.2

0.4

0.6

1941

1945

1949

1953

1957

1961

1965

1969

1973

1977

1981

1985

1989

1993

1997

2001

Len

gth

(mi)

Bre

ak R

ate

(Nu

mb

er

Bre

aks/

mi/

yr.)

YOI

CI - Length and Break Rate based on Year of Installation

Weighted Break Rate Length

0

50

100

150

200

0

0.1

0.2

0.3

0.4

0.5

7

11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71

Len

gth

(mi)

Bre

ak R

ate

(Nu

mb

er

Bre

aks/

mi/

yr)

Age

CI 1942-2015 - Length and Yearly Break Rate based on Age at Year of Break

0

0.25

0.5

0.75

1

1.25

1.5

0

0.5

1

1.5

2

2.5

3

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

Len

gth

(m

i)

Bre

ak R

ate

(Nu

mb

er

Bre

aks/

mi/

yr.)

YOA

DI ABN - Length and Weighted Average YearlyBreak Rate up to YOA based on YOA

-0.5

0.5

1.5

2.5

3.5

4.5

-10.0

10.0

30.0

50.0

70.0

90.0

Sco

re

Len

gth

(mi)

MAT7

Length of Inspected Pipes, Average Age and Structural score based on MAT6

LENGTH INSPECTED AVE INSP AGE AVE SCORE PEAK

0

20

40

60

80

100

1

2

3

4

5

12 18 24 27 30 33 36 39 42 45 48 51 54 60 66 72 78 90 96

Len

gth

(mi)

/Ave

rage

Age

at I

nsp

ecti

on

Sco

re

DIAM

Length of Pipes (All and Inspected), Average Age at Inspectionand Structural Scores based on DIAM

AVE SCORE FEB 2016 PEAK 2016

AVE AGE AT INSP LENGTH INSPECTED PIPES

Example Results Non-inspected Pipes

76

FeatID NB BRKS YOI DIAM SOIL L COMMENTS LOF

439406 0 1960 12 BAD 0.001 SAME CO-VARIATES 0.00012 SAME LOF 5020359 0 1960 12 BAD 0.0011 SAME CO-VARIATES 0.00012 SAME LOF 414765 1 1960 12 BAD 0.0314 MORE BREAKS 0.01604 HIGHER LOF 423809 0 1960 12 BAD 0.0319 0.00135 396706 2 1948 6 BAD 0.0612 OLDER 0.08651 HIGHER LOF 379035 2 1967 6 BAD 0.0643 0.04741 438274 1 1953 4 BAD 0.0048 SMALLER DIAM 0.00544 HIGHER LOF 448483 1 1954 8 BAD 0.0044 0.00384 389358 1 1972 12 BAD 0.1847 WORSE SOIL 0.05209 HIGHER LOF 433341 1 1973 12 GOOD 0.1924 0.01721 379182 0 1960 12 4 0.27 LONGER 0.00757 HIGHER LOF 447565 0 1960 12 4 0.0137 0.00071

LOF

Output Results/Proxies for LOF: Predicted Break Number per pipe and per year

Example Results - Inspected PipesAnalytical Basis

77

Time t in years

Survival

probability S(t)

1

00

1

2

3p3(t)

p2(t)

p1(t)

S23(t)

S12(t)

ID Year p1 p2 p3 p4 p5 Lengthaggregated state

01000000006M-Central 2016 0.12 0.01 0.35 0.01 0.50 0.215 3.765

LOF: Aggregated Score = 1 x 0.12 + 2 x 0.01 + 3 x 0.35 + 4 x 0.01 + 5 x 0.5 = 3.77

Output Results/Proxies for LOF: probability to be in a certain state per pipe and per year

Example ResultsInspected Pipes

78

YOI Score Pred.

Year

1 2 3 4 5 Aggregat

ed Score

1930 1 10002000010M-10000000040M 2016 0.9751 0.0230 0.0018 0 0 1.0267

1930 1 10002000010M-10000000040M 2100 0.3230 0.0856 0.5878 0.0024 0.0012 2.2737

1960 1 20000000150M-20000000130M 2016 0.9681 0.0286 0.0032 0 0 1.0351

1960 1 20000000150M-20000000130M 2100 0.1734 0.0608 0.7583 0.0043 0.0032 2.6032

1994 1 06000000390M-06000000386M 2016 0.9495 0.0409 0.0095 0.0000 0 1.06

1994 1 06000000390M-06000000386M 2100 0.0038 0.0041 0.9562 0.0147 0.0211 3.0453

1930 3 11000000180M-11000000170M 2016 0 0 0.9924 0.0070 0.0006 3.0082

1930 3 11000000180M-11000000170M 2100 0 0 0.2015 0.1459 0.6526 4.4511

1960 3 05000000810M-TMP1513519M 2016 0 0 0.9920 0.0072 0.0008 3.0089

1960 3 05000000810M-TMP1513519M 2100 0 0 0.0122 0.0400 0.9478 4.9357

1994 3 05000000210M-05000000200M 2016 0 0 0.987 0.0096 0.0027 3.0151

1994 3 05000000210M-05000000200M 2100 0 0 0 0 1 5

1930 3 08000000020M-08000000009M 2016 0.1008 0.0099 0.3679 0.0049 0.5165 3.8264

1930 NA 08000000020M-08000000009M 2100 0.0517 0.0078 0.2750 0.0385 0.6270 4.1814

1960 NA TMP1447319M-TMP1447314M 2016 0.1426 0.0168 0.3831 0.0401 0.4175 3.5732

1960 NA TMP1447319M-TMP1447314M 2100 0.0671 0.0090 0.3421 0.0223 0.5596 3.9983


79


• Approaches

1. Ranking Tree (see next slide)

2. Monetized

⁻ Same but criteria are monetized


⁻ Impact criteria data

⁻ Excel/GIS

• Plus⁻ Real cost of break

• Minus⁻ Can be difficult to put price tag even on

indirect cost

80

Example Monetized Criteria

Criteria Category Criteria Low Moderate High Very High

Asset Repair Costs <$20K $20K - <$100K $100K - <$500K >$500K

Emergency Repair Costs <$20K $20K - <$100K $100K - <$500K >$500K

Asset Replacement Costs <$20K $20K - <$100K $100K - <$500K >$500K

Property Damage <$20K $20K - <$100K $100K - <$500K >$500K

Operational Losses (lost revenue, exporting to other facilities) <$20K $20K - <$100K $100K - <$500K >$500K

Economic Administrative and Legal Costs of Damage Settlements <$20K $20K - <$100K $100K - <$500K >$500K

Consultant/Engineering Services <$20K $20K - <$100K $100K - <$500K >$500K

Permit Violation yes

Environmental Environmental Regulatory Fine for Spills or Releases <$20K $20K - <$100K $100K - <$500K >$500K

Disruption of Service yes

Social Safety- Public and CWW staff yes

Magnitude Ranges for Triple Bottom Line Analysis

do not apply to safety

do not apply to permit violations

do not apply to safety

Ranking TreeApproach

⁻ Define TBL criteria

⁻ Set queries in GIS

⁻ High score rules with modifiers

Plus

⁻ Better differentiation in scores

Minus

⁻ Difficult to evaluate social and indirect costs

81


82


• Approaches:

• 1. Risk Score = LOF2 x COF2

• 2. Multi-Criteria Decision Making Model - for example

- Electre-Tri

⁻ Variables, criteria, weights

⁻ Reference profiles


⁻ COF and LOF

⁻ Opportunistic criteria data

⁻ Model/Software• Plus

⁻ Simple input files in excel⁻ No 1 x 100 = 10 x 10

• Minus⁻ Need special software⁻ Pipes assigned to priority categories. Ranking

difficult.

83

- a3 candidate for R&R- a8 not candidate for R&R- Model sorts out other pipes with less certain status

Example Electre-Tri Reference Profile

Aging FunctionEffective Useful Live (EUL)

84

Aging FunctionEffective Useful Live (EUL)

• Approach

– 2 options:

1. Plot historical break rate/score based on age at Year of Break (YOB) and extrapolate

1. Plot Predicted rate/score (LOF outputs)

– Set service level(s)

– Estimate EUL(s) from Aging Curve

85

Extrapolation – Cohort Level

86

0

0.02

0.04

0.06

0.08

0.1

0.12

0

10

20

30

40

50

60B

reak

Rat

e (N

um

ber

Bre

aks/

mi/

yr)

Age

DI - Length and Yearly Break Rate based on Age at YOB

Break Rate Linear (Break Rate)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0

10

20

30

40

50

60

70

80

Bre

ak R

ate

(Nu

mb

er B

reak

s/m

i/yr

)

Age

CI 1942-2015 - Length and Yearly Break Rate based on Age at Year of Break

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1 11 21 31 41 51 61 71 81 91 101 111 121

Sco

re

Age

Score of Inspected Pipes based on Age

Non-inspected Pipes - PBNPipe or Cohort Level

0.000

0.025

0.050

0.075

0.100

0.125

0.150

0.175

0.200

32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 92 96 100

Bre

ak R

ate

(brk

/mi/

yr)

Age

DI - Predicted Break Rate of ACT Pipes by Age (computed with PBNs)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.55

0.60

51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99

Bre

ak R

ate

(brk

/mi/

yr)

Age

CI - Predicted Break Rate of ACT Pipes by Age (computed with PBNs)

59 86

0.40

0.25

Low COF BR= 0.25Medium COF BR = 0.4

Inspected Pipes - State ProbabilitiesPipe or Cohort Level

88

YOI COF Score Pred.

Year/

EUL

1 2 3 4 5 Aggregated

Score

1970 M 3 03000000130M-03000000120M2067/97

0.0000 0.0000 0.3032 0.1459 0.5509 4.2477

1942 M 3 12023000060M-12023000050M2082/140

0.0000 0.0000 0.3007 0.1484 0.5509 4.2501

1949 M 01000000006M-Central2071/122

0.0671 0.0082 0.3553 0.0183 0.5510 3.9780

1949 L 01000000010M-01000000006M2097/148

0.0523 0.0072 0.3074 0.0325 0.6007 4.1221

1994 L 2 20000000450M-20000000440M2096/102

0.0000 0.0000 0.2558 0.1385 0.6058 4.3500

1930 L 2 10000000070M-10000000060M2095/165

0.0000 0.0000 0.2513 0.1471 0.6016 4.3503

Medium COF 55% probability to be in state 5 118.5Low COF 60% probability to be in state 5 128.5

Long-Term Plan (LTP)

89

Long-Term Plan - Level 2 (Cohorts)

• Approach– Same as BNL but R&R needs linked to service level (break rate or score)

• Data and Tool requirements– Same but aging function and EULs needed– Simulation tool

• Plus– Versatile: allows using data at all levels

• Minus– Long-Term Plan (LTP) not integrated with Short-Term Prioritization Plan (STP)

90

Example LTP Simulation Model Output

Long-Term Plan - Level 2 (Pipes)

• Approach– Same as previously but at pipe level in

GIS– Develop Long-Term Plan (LTP) and

Short-Term Prioritization Plan (STP) jointly

• Data and Tool requirements– All data must be at pipe level– Simulation tool can be in GIS

• Plus– LTP and STP can be developed jointly– Priority projects can be visualized in

GIS • Minus

– STP limited to scoring

91

Example GIS-based and pipe-level LTP Model using output

Five Key TakeawaysTop five basic and essential recommendations:

• Move away from assumed data and scoring

• Work with pipe and break data at pipe level, ultimately within GIS

• Keep data about abandonment (ABN pipes)

• Build adequate data structure throughout organization from collection to analysis - Business process - Audit, audit, audit

• Set up data quality indicator and improvement plan and

present results with quality indicator

92

DataData Data Data Data

Conclusion

• Long-Term R&R Plan and Short-Term Prioritization Plan can be generated through approaches that vary from basic to advanced.

• More advanced approaches allow for more reliable, credible and optimized estimates, and eventually savings.

• More advanced approaches require more precise data and advanced algorithms and models.

93

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

94

12:00-1:00LUNCH

Workshop Agenda

Lunch

95

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-2:10

Group Discussion

Where do you stand? Where do you want to go?

Celine, Kevin, Kurt

2:10-2:20


Kevin

2:20-2:50

Mixing Analytical Approaches and Inspection Program

Annie

2:50-3:00

BREAK

3:00-3:25



Kurt

3:25-3:50


Inspected Pipes

Celine and Annie

3:50-4:00 Wrap-up

96

12:00-1:00LUNCH

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

97

12:00-1:00LUNCH

Workshop Agenda

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

98

12:00-1:00LUNCH

Workshop Agenda

Questionnaire

• Take handout 1

99

Break

100

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

101

12:00-1:00LUNCH

Workshop Agenda

AWWA Progress in AM Survey Results (2015)

• 550 Utilities Responded

• 59% serve less than 50,000 customers; 15% serve over 500,000

• 91% public; 9% private

• 85% retail; 45% wholesale

• 29% have part- or full-time AM staff

102

103

104

105

106

107

108

109

09:00-9:05WelcomeAnnie

9:05-9:15Introduction Kevin

9:15-9:20Introduction QuestionnaireKevin

9:20-9:40AM Framework and DataKurt

9:40-10:30Data Clean up ExercisesAnnie et al

10:30-10:40BREAK

10:40-11:40Analytical Approach – From Basic to Advanced Approach Non Inspected and Inspected PipesAnnie and Celine

11:40-12:00Case Study: Columbus – From Basic to Advanced ApproachNon Inspected PipesKevin

1:00-1:10Q/A about ColumbusKevin

1:10-1:40Case Study: Dallas – From Basic to Advanced Approach Inspected PipesCeline and Annie

1:40-2:40Group DiscussionWhere do you stand? CelineWhere do you want to go? Kurt

2:40-2:50BREAK

2:50-3:00Synthesis of Previous DiscussionKevin

3:00-3:25 Mixing Analytical Approach and Inspection ProgramAnnie

3:25-3:55Case Study: AWWU – From Basic to Advanced Approach Mixing Analytical Approach with Inspection ProgramKurt

3:55-4:00 Wrap-upAnnie

Workshop Agenda12:00-1:00LUNCH

EUL Estimation - ApproachesCost versus Precision

> >Lab

CouponLimited number

of pipes

PCAhydrant to hydrantAverage thickness

Limited length

Analytical ApproachesPipe, cohort

Whole system

110

PCA in a Cost Efficient AM Plan

• Example:

– 2,000 mi

– $ 0.8M budget allocated for PCA in 2016

– PCA $20,000/mi

– We can do 40 mi = 2% of system

• What is the purpose of the PCA?

• Can a certain purpose be accomplished with 2%?

• What purpose can be accomplished with 2%

• How much money is needed to accomplish a certain purpose ?

• What pipes should be inspected to accomplish a certain purpose with a given budget?

111

Why conducting PCA?

1. Justify investment: Evaluate PCA of pipes with high Risk Score to decide whether they should be replaced

2. Validate and improve aging curves and EULs generated using analytical approaches (basic or, preferably, advanced)

3. Generate aging curves and EULs for pipe groups with few breaks but high COF (like large pipesfor example)

Are we trying to generate results to answer a punctual question, e.g., ‘should pipe be replaced this year?’, or, for optimal cost efficiency when building an R&R & Inspection Plan?

112

Suggestion

Using a combination of analytical approaches along with PCA will lead

to more cost-efficiency but this requires a plan!

+ =Saving

113

Level of Cost of PCA

1. Risk scores have been produced using basic or advanced approaches; pipes have been prioritized for replacement based on their risk score.

PURPOSE OF PCA: justify R&R investment

PCA $ depends on length of high risk pipes (vice versa if PCA $ is limited, focus shouldbe on highest risk pipes)

114

Risk Score = COF x LOF

Based on purpose, data availability and type of analytical approach used performed

3 different situations

2. A lot of breaks; good LOF and aging curves have been generated using advanced analytical approaches

PURPOSE OF PCA: validate aging curves and analytical results

PCA $ depends on statistically significant size of the sample

3. No break; no inspection data; access to PCA/LOF forecasting model

PURPOSE OF PCA: conduct a PCA program on a sample of the pipes in order to extrapolate to overall group of “similar pipes” and have PCA/LOF score for every pipe and in the future

PCA $ depends on statistically significant size of the sample

115

ZscoreStdDev

margin of errorpopulation size

Level of Cost of PCA (Cont’d)

Sample size out of a certain cohort’s population?

Intuitively (and mathematically) sample size depends on:

• How much, when conducting multiple measurements on a population (of individuals who behave “similarly” within a certain range), the results vary (average variation is the standard deviation) - If we want StdDev to be small we need to get a population with very similar degradation pattern (descriptive statistics will help do that).

• How much + or – error we can tolerate in the measurement (margin of error/confidence interval).

• How confident (90%? 95%, 99%?) we want to be that the actual mean of the future measurement on the sample would fall within the confidence interval (expressed by the Z score - published statistical tables).

• Population from which sample is extracted. A population twice as big does not require a sample twice bigger.

116

Example

Level of

confidenceZ score StdDev

Margin of error+/-

Sample size

95% 1.96 0.5 15.2% 40

90% 1.64 0.5 12.7% 40

• Fixed:- StdDev for a certain cohort and PCA technology (here 50% = .5) - Budget: we only have money for 40 mi out of a 1,000-mi population of pipes.

• If we want:- 95% confidence margin of error will be 15.2%.- 90% confidence margin of error will go down to 12.7%

117

The smaller the population of reference the larger the proportion between sample size and that population. For previous example for same statistical performance (Confidence and Error):- If population of reference is 1,000 miles, sample size must be around 40 mi- If population of reference is 100 miles, sample size must be around 29 mi- If population of reference is of any size above 1,000 mi, the requested sample size

remains close to 40 mi

PopulationConfidence Interval

StdDevMargin of Error

Sample Size

100 95% 0.50 0.153 29.3

500 95% 0.50 0.153 38

1,000 95% 0.50 0.153 39.5

2,000 95% 0.50 0.153 40.2

3,000 95% 0.50 0.153 40.5

4,000 95% 0.50 0.153 40.6

5,000 95% 0.50 0.153 40.7

Looks positive but in reality it is difficult to have small StdDev with very large population.

So solution is not to shoot for large cohorts

118

Example of a cost efficient plan

– 2,000 mi

– $ 800,000 budget allocated for PCA in 2016 and every year

– PCA $20,000/mi

– Budget for 40 mi/yr. = 2% of system

– Challenge is to get good estimate of StdDev which depends on coherence of the cohort. The fewer the number of breaks, the more difficult it is to get smaller StdDev.

119

AC Pipes2,000 miles / 850 breaks / 4 brk-100 mi-yr.

Type I | AC (<=1956)306 brks / 12 mi / 254 mi)

10.2 brk-100 mi-yr

Type || AC (>=1957)566 brks / 26 mi / 1,874 mi

3.2 brk-100 mi-yr

Small<=6154 brks

6.4 mi73 mi

19.2 b-r

Med 8-14154 brks

5.5 mi108 mi

11.2 b-rate

Large >=160 brks0 mi4 mi

0 b-rate

Small<=6194 brks

8.8 mi688 mi4.8 b-r

Med 8-14350 brks15.5 mi

1,342 mi2 b-rate

Large >=1622 brks0.3 mi188 mi

0.8 b-rate

First Year

• Use PCA to estimate the Likelihood Of Failure (LOF) and aging curve of the pipes that belong to a cohort with few breaks (below, the large pipes) for which statistical analysis is not possible.

• If we spend our $0.8M on PCA for large pipes (pop of 192 miles) we could have a sample size of 40 mi. Given a StdDev of 0.5 we could have a margin of error of 13.8% and a confidence interval of 95%. We can then use PCA forecasting model to compute LOF and risk score of every large pipe and select priority pipes for inspection in subsequent years.

• On medium and small size pipes there are typically enough breaks to generate LOF, agingcurves and EUL out of failure statistics.

Small<=6154 brks

6.4 mi73 mi

19.2 b-r

Med 8-14154 brks

5.5 mi108 mi

11.2 b-rate


0 b-rate

Small<=6194 brks

8.8 mi688 mi4.8 b-r

Med 8-14350 brks15.5 mi

1,342 mi2 b-rate


0.8 b-rate

120

Subsequent years• Inspect large pipes as determined by risk score.

• In addition, based on remaining budget, inspect small and medium pipes in order of prioritydetermined by risk score. If we consider 4 different cohorts (73, 108, 688 and 1,342 mi) and we want same stat performance (StdDev = 0.5 TBC-large cohorts may have a bigger stdDev, margin of error = 10%, confidence interval = 95%) we will need to conduct PCA on 41 (57%) + 51 (47%) + 84(12%) + 90 (7%) = 266 (12% total) miles; which could be achieved in 6.5 years at a cost of $ 5.3 M.

• Then PCA failure forecasting model could be used for small and medium pipes to obtain LOF (and risk score) for each pipe and each year.

• Using LOF and risk scores as well as cost data, decide at what point pipes should be re-inspected or just rehabilitated or replaced.

121

Small<=6154 brks

6.4 mi73 mi

19.2 b-r

Med 8-14154 brks

5.5 mi108 mi

11.2 b-rate


0 b-rate

Small<=6194 brks

8.8 mi688 mi4.8 b-r

Med 8-14350 brks15.5 mi

1,342 mi2 b-rate


0.8 b-rate

Conclusion

This requires:

- planning ahead including purpose of the PCA program

- Adoption of advanced analytical approaches

+ =Saving

122

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

123

12:00-1:00LUNCH

Workshop Agenda

09:00-9:05Welcome

Annie

9:05-9:15

Introduction

Kevin

9:15-9:20


Kevin

9:20-9:40


Kurt

9:40-10:30


Annie et al

10:30-10:40

BREAK

10:40-11:40



Annie and Celine

11:40-12:00


Non-inspected Pipes

Kevin

1:00-1:10

Q/A about Columbus

Kevin

1:10-1:40


Inspected Pipes

Celine and Annie

1:40-2:40

Group Discussion



2:40-2:50

BREAK

2:50-3:00


Kevin

3:00-3:25


Annie

3:25-3:55



Kurt

3:55-4:00 Wrap-up

Annie

124

12:00-1:00LUNCH

Workshop Agenda

2016 annual conference & exposition sun05 – … · 2016 annual conference & exposition ....

Documents