Implementing a Risk- Based Main Replacement Strategy at EWEB

Transcription

1 Eugene Water & Electric Board Implementing a Risk- Based Main Replacement Strategy at EWEB Chris Irvin, Senior Engineer PNWS-AWWA 2017

2 Presentation Overview About EWEB Introduction to risk concepts and alternatives Developing a risk based model and evaluation tool [Risk] = [Probability] X [Consequence] Real world implementation of risk based model Lessons learned and areas for improvement

3 About EWEB EWEB serves Eugene, Oregon s third largest City Service Population:183,000 Average Daily Demand: 24 mgd Approximately 800 Miles of distribution pipeline 7 pressure levels between 607-ft to 1325-ft 25 upper level pump stations 23 Reservoirs with 70 Million Gallons of Storage

4 System Overview

5 Miles and Miles of Pipe Over 800 Miles! Approximately the distance of Eugene to Los Angeles

6 Average System Age 30 EWEB's Pipe Installation History Vs. Average Age Annual Pipe Install Sum of Age All Systems 45 Annual miles of Pipe Installed (Miles) Average System Age (Years) 0 0 Date (Years)

7 EWEB Pipe Material Polyvinyl Chloride Iron Pipe Steel Unknown Asbestos Concrete Concrete Cylinder High Density Polyethylene Ductile Iron Cast Iron Copper

8 Existing System Performance - Leak Rate EWEB Average System Age = 38 years EWEB Annual Leak Rate = 9 leaks/100 miles pipe AWWA Average = 25 leaks/100 miles pipe Annual Leak Rate Vs. Average System Age EWEB Water Distribution study, HDR

9 EWEB Leaks by Diameter EWEB's Pipe Leaks Vs. Diameter Overall leaks per mile Miles of pipeline per size 250 Leaks per Mile of Pipe Miles of Pipe < Diameter (Inches) 0

10 Condition of Water Pipelines The condition of buried water pipeline is usually unknown Age does not always indicate the need to replace a pipeline 76 year old 16-inch CIP - corrosion failure 75 years old 16-inch CIP Appears to be in good condition despite leak history

11 Age is a poor predictor of failure Constructing concrete cap over 20 DI installed in 1980 Saw-cutting new 12 thick reinforced concrete bus lane Repairing 20 inch leak weeks after bus lane built 36 year old ductile 20 inch, corrosion failure

12 EWEB Main Replacement Program Pre-2016 Based primarily on pipe age and material Compiled on a spreadsheet on one engineers computer Reactive and not proactive Replacement priorities questioned after multiple major 16-inch main breaks.

13 Management of Water Distribution System Assets Work and Asset Management (WAM) Software system implemented in 2014 No dedicated asset management staff Comprehensive asset management at utility moving slowly Simple risk based system created in the interim. Started with distribution mains

14 Stakeholder Team Assembled A main replacement prioritization strategy was developed collaboratively between water operations and engineering:

15 Overall Program Goals The Team developed program goals for the main replacement strategy: Prioritize main replacements using a consistent, clear and defensible methodology. Proactively address aging pipe and water main leakage in the most cost effective manner. Reduce overtime, damage to public facilities, impacts to the community, and frequency of unplanned water outages and boil water notices when a main break occurs. Maintain water to critical customers when a main break occurs. Maintain less than the national average leak rate of 25 annual leaks per hundred miles of pipeline. Protect water quality.

16 Main Replacement Prioritization Alternatives Considered Condition Assessment Field based evaluation often measuring pipe wall thickness or leaks Not cost effective for smaller diameter mains Service Life Analysis Estimate the useful life (service life) of pipelines Can consider a variety of factors like pipeline material, diameter, and install date Risk Model A risk model takes into account both the probability and the consequence of a pipeline failure.

17 Condition Assessment Appropriate for Transmission and possibly large Diameter distribution mains. Not cost effective for smaller distribution mains. Need a predictive tool that estimates condition for smaller pipelines.

18 Service Life Different methodologies are available and results vary HDR/EWEB analysis based on developing Break Frequency Curves for different pipe classes. Pipe Type EWEB 2015 Service Life, years (a) AWWA Service Life, years (b) Less than 4 Inch NA 4 Inch to 6 Inch CI Inch to 6 Inch DI NA Inch to 12 Inch CI a. EWEB analysis based on HDR methodology in 2015 b. AWWA Buried No Longer (2012)

19 Defining Risk Risk estimates total potential impacts to the utility [Risk] = [Probability] X [Consequence] Low Consequence High Consequence High Probability Low Probability

20 Risk is Relative Prioritize limited replacement and condition assessment resources on the assets that pose the most risk Focus of EWEB Risk Model Figure 7: Relative Risk Assessment for Water Mains David Spencer PE, HDR

21 Why use a risk based model? Advantages Considers consequences of a leak on EWEB and the community Desktop exercise that does not require field work. More defensible and responsible to public and ratepayers than a service life model Disadvantages More complex than a service life model Relies on good data to be useful Assumes condition of pipe from available data

22 Risk Model Development Surveyed stakeholders and customers. Developed a list of potential risk factors. Evaluated quality of existing data. Developed list of final risk factors. Weighted relative importance Stakeholder of each Team risk factor. Pilot studied risk model and developed final equations. Created and tested GIS model.

23 Final Risk Factors Probability Pipe Leak Count Pipe Material Pipe Diameter Pipe Install Date Pipe Leak Type Pipe Pressure Consequence Pipe Diameter Catastrophic Failing Pipe Material Road Type Location to Critical Infrastructure Water Service Area Location Customer Density/Valve Spacing Pipe Pressure

24 Risk Factor Weighting Example Category Weight Probability Factor Category 10 Pipe Leak Count Sub-Category Probability Factor Sub-Category Weight 10 Greater than 5 leaks per 500 feet 8 4 leaks per 500 feet 6 3 leaks per 500 feet 4 2 leaks per 500 feet 2 1 leak per 500 feet 0 No Leaks Category Weight 5 Probability Factor Category Leak Type (highest weight if more than one leak) Sub-Category Weight Probability Factor Sub-Category 3 Circumferential 8 Hole 6 Pinhole 10 Longitudinal 0 Type Not Available TTTTTTTTTT PPPPPPPPPPPPPPPPPPPPPP pppppp = SSSSSS[CCCCCCCCCCCCCCCC wwww XX SSSSSSSSSSSSSSSSSSSSSS wwww])

25 i Risk Model Tools Considered Risk Model add-on to existing hydraulic model Significant cost (>$20k) Limited number of Licenses Limited access to Data Third Party stand alone software Significant Cost Technical challenges of interfacing new software with existing databases Ongoing maintenance and licensing requirements GIS based model Would need to be developed and maintained in house Would work on existing software program pervasive in the utility Model could be available for all to use without special software or training Model would work in same environment as source data

26 GIS Based Tool Development GIS Based Tool chosen as most cost effective and versatile tool. Data model was designed and implemented in the ESRI/ArcGIS geodatabase. Model uses a mirrored and stripped down data set. Data fields were added for each risk factor. For non-spatial factors a SQL script was created for the calculations. For Spatial Factors, ArcGIS Model Builder was used to perform calculations. Final risk factors are symbolized using a color ramp with hotter colors indicating higher risk pipe. update gis.arcfm.wmainrisksstage set cf_pipe_types_prone_to_breaks = case material when 'AC' then 9 when 'CC' then 10 when 'CI' then 6 when 'IP' then 2 when 'PVC' then 6 end;

27 GIS Model Overview Pipe Risk Score Highest Risk Pipes in system generally have a Risk Score between The absolute highest risk pipe in the system has risk score of 100

28 Example Project: Polk Street Example 1: Polk Street 24 th to 25 th street Risk Index of 57.1% Eight! (8) Documented leaks Other Factors: Construction year (1962), Street Classification (Major Collector), Pipe size (6-inch)

29 Example Project: Elinor Street Planned Piping Connection Original Planned project was to connect Elinor St to University to improve fire flows and to add redundancy to a school, nursing home, and church. Risk index of 75, 16 th highest risk pipe in system. 6 documented Leaks Other factors: Year constructed (1952), type of leak (hole), Service Area (Hills)

30 Example Project: 8 th Street Pipe Built in leaks over 1500 feet Harder to estimate condition of large pipes due to less leak data Probability Index Consequence Index Risk Index Normalized Risk Factor Where do you stop? Calculation of risk data

31 Summary Summary: Model is only a tool! Allows EWEB to prioritize every pipe in Distribution System Entire utility has access to model to understand main replacement decisions Constantly searching out input to improve model

32 Lessons Learned Replacing higher risk pipe inherently leads to more difficult and expensive projects. Even though methodology is transparent, not everyone agrees with it. Operations and Engineering still have different perspectives and definitions of success. Need to continually ask for feedback and make improvements.

33 Management Feedback I personally think it is a great system. It is transparent and provides an objective way to select pipes for replacement. Wally McCullough, Water Engineering Supervisor We think it s a great process. my only issue is the perception for the field crews. We have mains that have had several leaks and are low risk. The crews expose and work on these breaks and voice their concerns on what they see. For us to tell them that this a low risk main and that it s cheaper to repair as needed can leave a bad impression on them. Jeremiah Hunt, Water Construction Supervisor

34 Thank you for your time! QUESTIONS?