Overview - Transportation Research Boardonlinepubs.trb.org/onlinepubs/webinars/120919.pdfOverview 0...
Transcript of Overview - Transportation Research Boardonlinepubs.trb.org/onlinepubs/webinars/120919.pdfOverview 0...
Overview 0 Highlight sessions conducted at the 9th National
Conference on Transportation Asset Management, which took place on April 16-18, 2012, in San Diego, California
0 Four conference tracks: 0 Pavement and Bridges (24 presentations) 0 Beyond Pavement and Bridges (18) 0 Focus on Implementation (20) 0 Transit State of Good Repair (21)
0 Today is just a small taste of the conference experience
Objectives for Today 0 Enhance knowledge of all owners and operators of
pavements and bridge assets 0 Learn about successful asset management programs 0 Learn successful stakeholder communications
methods 0 Understand program development techniques 0 Learn to develop clear performance metrics
Presenters Today 0 Ron Hagquist, Operational Excellence Manager, Texas DOT 0 Use of Management Science Analytics for Asset Management at
Texas DOT 0 Sui Tan, Regional Streets and Locals Program Manager, California Metropolitan Transportation Commission 0 Performance-Based Approach to Funding Policy for Local
Streets and Roads 0 Eleni Bardaka, Graduate Assistant, Purdue University 0 Forecasting the Life of Asset Preservation Treatments: A
Comparative Evaluation of Alternative Tools and Techniques 0 Mohammadsaied Dehghani, Graduate Student, Virginia Tech 0 Corridor-Level Performance Measures to Support Resource
Allocation Strategies in Highways
Summary 0 Four presentations providing some highlights from
the 2012 National Transportation Asset Management Conference
0 Conference materials including presentations available from the TRB website
0 Planning has started for the 10th National Conference, most likely taking place in the spring of 2014 0 Pooled fund participation available
Questions? 0 Ron Hagquist, Operational Excellence Manager, Texas DOT 0 Use of Management Science Analytics for Asset Management at
Texas DOT 0 Sui Tan, Regional Streets and Locals Program Manager, California Metropolitan Transportation Commission 0 Performance-Based Approach to Funding Policy for Local
Streets and Roads 0 Eleni Bardaka, Graduate Assistant, Purdue University 0 Forecasting the Life of Asset Preservation Treatments: A
Comparative Evaluation of Alternative Tools and Techniques 0 Mohammadsaied Dehghani, Graduate Student, Virginia Tech 0 Corridor-Level Performance Measures to Support Resource
Allocation Strategies in Highways
Use of Management Science Analytics for Asset Management
At TxDOT
Ron Hagquist, TxDOT Operational Excellence Office Transportation Research Board Webinar September 19, 2012
ANALYTICS
Operations Research
Management Science
Determining Optimality & Efficiency Assessment
Data Analysis, Forecasting, &
Decision-Making Methods
DATA PROBLEMS
Data
Information
DECISION
Statistical Analysis
MS/OR* Decision Methods
Analytics Basic statistical analysis can often provide adequate basis for making decisions.
But sometimes the number of choices and mathematical considerations require structured decision methods and software.
* Management Science / Operations Research
Organizational Levels Of Analytics Capabilities
Transformed Experienced Aspirational
Prescribe actions Justify actions
Analytics used to:
………………………………………………
………………………………………………
………………………………………………
Guide actions
Organizational Analytics Capabilities
Well…
A key finding of the MIT Sloan School study was
that “top performing organizations use analytics
five times more than lower performers.”
-- Page 2, Figure 1 Analytics: the New Path to Value
Limited skill base in an organization
Lack of or limited understanding on how analytics can be used to improve an organization
Unclear ownership or governance of the data Lack of or limited sharing of data
A new administration, with an increased commitment to analytics for data-based decision making.
Creation of an Analytics Strategic Plan based on MIT study.
• Creation of in-house consulting group with operations research & management science capabilities.
• Creation of “Operations Research Analyst” job classification.
• Conducting tailored in-house OR/MS training for analysts.
A TxDOT project is currently a semifinalist for the INFORMS Innovations
in Analytics Award.
Increased emphasis on major analytics projects.
The Story of Analytics at TxDOT
8
“Gentlemen, we are out of money. We shall have to think. ” -Address to Parliament by Winston Churchill
So… how hard do WE have to think?
Needs & encumbrances A BASEBALL
Revenues A GOLF BALL
Challenge A TENNIS BALL
LESS:
EQUALS:
Stimulus funds AN AIR-RIFLE BB
Sports Analogies 2009 – 2030 Time Horizon
What to do ?. . . . . …. . find more money. . ?
$155
$27$12
$21$12
$5$0
$20
$40
$60
$80
$100
$120
$140
$160
$180
StandardRevenue
DoubleVehicle
RegistrationFees
Index the GasTax to CPI
End DPSTransfers
Raise theState GasTax by 5
Cents
Prop 12
Bill
ions
$ 77 B
Estimates Of Possible Revenue Enhancements
Possible Revenue Enhancements
What else to do? 1. Improve internal efficiencies 2. Improve allocation methods
ANALYTICS
Operations Research
Management Science
Determining Optimality & Efficiency Assessment
Data Analysis, Forecasting, &
Decision-Making Methods
DATA PROBLEMS
What else to do? 1. Improve internal efficiencies: Organizational (management audits, restructuring, modernization initiative) Operational (modernization initiative, operations research projects, Operational Excellence Office)
The benefit of ROW advance-purchase is avoiding property price escalation.
The benefits of accelerating project completion are [1] avoiding highway construction cost inflation and [2] earlier delivery of benefits to travelers.
Project Purpose: to develop a tool to find the economic balance between ROW advance purchase and ongoing project acceleration
Advance ROW Optimization Tool
www.library.ctr.utexas.edu/
District Number of Parcels
Difference in Mean Cost Due to Early
Acquisition
Mean Cost of Early Acquisition
Saving/Cost Ratio
Early Acquisition Parcel ID
Austin 20 $769,000 $181,000 4.25 5TE
Dallas 10 $44,653,000 $16,781,000 2.66 9
El Paso 19 $475,000 $475,000 1.00 3
Houston 28 $172,791,000 $183,057,000 0.94 2
Advance ROW Optimization Output
Equipment Replacement Optimization
TxDOT fleet: 17,000 vehicles Annual replacement budget: $50,000,000 Estimated annual savings = $2,500,000 - $5,000,000
Project Purpose: to develop a tool to find the minimum-cost vehicle-class replacement policy which can be updated as parameters change over time.
www. library.ctr.utexas.edu/
Year 1 Year 2 Year 3……………….Year 20
……………
…………… ……………
…………… ……………
…………… ……………
……………
Keep Replace
Keep Replace Keep
Replace
Solution Method: Dynamic Programming
Problem: What is the least-cost sequence of 200,000,000,000,000,000,000 K-R decisions ?
Project Evaluation Tool
the
A Benefit/Cost Toolkit for Anticipating & Evaluating the Impacts of Roadway Improvements
Examples of projects that PET can analyze:
1. Capacity Expansion & Tolling
2. Demand Management
3. Shoulder Lane Use
4. Speed Harmonization
5. Work Zone Phasing
6. Incident Management
7. Transit
Austin Network & Case Studies - $131 M
capacity expansion project in East Austin.
- 3 alternatives analyzed:
- Freeway upgrade;
- Tollway upgrade;
- TOD tolling
Project Financing Base Case: No Build
Freeway Upgrade
Tollway Upgrade
Tolling by Time of Day
NPV of New Tolling Revenues $0 -$88 K $118 M $142 M NPV of Initial and Future Project Costs $0 $123 M $138 M $138 M Project Financing Perspective NPV $0 -$123 M -$20 M $4.2 M Project Financing Perspective IRR N/A N/A 6.0% 8.2% Project Financing Perspective PP N/A N/A > 20 years 17.8 years
User Benefits Base Case: No Build
Freeway Upgrade
Tollway Upgrade
Tolling by Time of Day
Net Present Value $0 $690 M $376 M $265 M Internal Rate of Return N/A 60% 29% 21% Benefit / Cost Ratio N/A 6.6:1 3.7:1 2.9:1 Payback Period N/A 2.0 years 4.8 years 8.1 years
All PET files are available at www.caee.utexas.edu/prof/kockelman/PETwebsite
Management Science Applications for TxDOT
- Scoping Study -
Project Purpose: to conduct a management science “audit”
– to examine TxDOT business decisions which are amenable to
improvement with MS/OR methods – generating problem
statements for in-depth research.
www.library.ctr.utexas.edu/
WHY THIS PROJECT ?
1. Many operational problems at TxDOT are not unique to TxDOT, or even to transportation – they are problems common to many industries.
2. Only a small fraction of the universe of MS techniques have been applied to transportation.
3. Many of the successful decision-support projects at TxDOT have been due to the right department person finding the right researcher at the right time for the right problem – ad-hoc – how to systematize?
Process of mapping applications:
TxDOT OR/MS Experts
technique-driven
task-driven
Examples of Identified Analytics Applications • Forecasting – revenue; traffic; asset condition
• Multi-criteria prioritizing – research projects; hiring
• Location & scheduling optimization – emergency
response; fleet replacement; bridge & pavement management
• Project Selection – road capacity projects; bridge projects
• Multiple-objective funds allocation – MPO funds
allocations; capacity vs. maintenance allocations
• Life-cycle cost analysis – asset management
• Efficiency analysis – maintenance sections
“Gentlemen, we are out of money. We shall have to think. ” -Address to Parliament by Winston Churchill
What else to do? 2. Improve allocation methods
ANALYTICS
Operations Research
Management Science
Determining Optimality & Efficiency Assessment
Data Analysis, Forecasting, &
Decision-Making Methods
DATA PROBLEMS
Allocation Formula Advisory Project
Goal: To produce the best possible allocation formulas. Mission: To employ the best available expertise to review and improve TxDOT maintenance funds allocation formulas, and incorporating input from transportation stakeholders. Team: Texas Transportation Institute; Cambridge Systematics Inc.; expert panel; stakeholders; TxDOT Maintenance Division; and TxDOT Operational Excellence Office
Criteria Discipline Expert
Allocation Formula Advisory Project
Panel Selection Process:
When considering a system for allocating scarce transportation funds, there are six basic criteria:
1. Does this allocation produce the “most bang for the
buck”?
2. Does this allocation consider all relevant factors of
need?
3. Does this allocation make economic sense?
4. Do the formulas make mathematical sense (are they
specified correctly)?
5. Does the allocation consider fairness?
6. Does the allocation consider customer values?
Corresponding Mathematical Disciplines
1. Optimization – most bang for the buck
2. Decision Analysis – multiple considerations
3. Macroeconomics – “it’s the economy…”
4. Econometrics – getting the equations right
5. Fair Division – money is short
6. Market Analysis – knowing customer values
District Allocation Formulas
Customer Input –
Conjoint Analysis
TxCAP Road Quality
Index
Beginning Integration of Activities
Pavement Management
System
MAINTENANCE FORMULA PROJECT
35
Maximize customer value within a budget
A Report of Findings from the 2010 New Intelligent Enterprise Global Executive Study and Research Project
By IBM Analytics and the MIT Sloan School of Management
RECOMMENDATION 1: Focus on the biggest and highest value opportunities
TxDOT IMPLEMENTATION PRIORITIES:
1. Road maintenance optimization
2. Bridge maintenance optimization
3. Capacity project selection
4. Performance efficiency analysis
5. Workforce optimization
RECOMMENDATION 2: Within each opportunity, start with questions, not data
KEY QUESTIONS FOR TXDOT:
1. Within a budget, what is the least life-cycle cost of maintaining which roads, given traveler values?
2. Ditto for bridges.
3. Which capacity investment projects bring the greatest rates of return, and how to account for equity considerations?
4. How to systematically assess operational efficiency?
5. How to optimize personnel placement and training?
RECOMMENDATION 3: Embed insights to drive actions and deliver value
TRANSFORMING INSIGHTS INTO TXDOT ACTIONS: • Determine what pro forma information is needed in
dashboard form.
• Build pro forma dashboards linked to simulations of alternative scenarios.
• Quantify customer values and preferences in ways conducive to making decisions.
RECOMMENDATION 4: Keep existing capabilities while adding new ones
TXDOT ACTIONS TO AUGMENT ANALYTICS: • Provide training on basics of spreadsheet engineering,
data analysis, and optimal visual presentation of quantitative data.
• Provide training on intermediate and advanced analytics.
• Establish an Analytics career ladder supported by appropriate training.
• Provide briefings on the utility of analytics to managers.
RECOMMENDATION 5: Use an information agenda to plan for the future
A TXDOT INFORMATION AGENDA: • Chief Information Officer reconfigures agency data
systems to be integrated, consistent, and trustworthy.
• Align priority analytics activities to support strategic goals.
• Use a forward-looking information agenda to enable TxDOT to keep pace with advances in mathematical sciences and technology.
Analytics Career Ladder Analytics Training Analytics Competency Center
• Ad-hoc analytics projects
• Management Science Scoping Project
• Operational Excellence Office • Analytics Strategic Plan • In-house training • Analytics career track
Continued commitment to analytics for data-based decision making.
Carry out the Analytics Strategic Plan based on recommendations of MIT Sloan School Study, Analytics: The New Path to Value :
• Tailored in-house OR/MS training for analysts - enterprise-wide.
• Increased collaboration with Business and Industrial Engineering Schools
• Possible creation of departmental Analytics career path Increased emphasis on analytics projects
The Future of Analytics at TxDOT
Supporting Performance, Accountability, and Transparency at Regional Level
Sui Tan, MTC 9th National Conference on Transportation Asset
Management Webinar September 19, 2012
Business As Usual ???
• MAP- 21 focuses on performance-driven framework: Performance Accountability Transparency
How We Get into Performance
Unfair funding policy for local agencies
Population, mileage, and needs
“Worst first” practice prevalent
Conditions are facing steep decline
Gaps in full pavement management cycle
San Francisco Metropolitan Region
Population = 7.3 mil
9 counties
100 cities
42,500 lane-miles
1,500 miles of highway
23 transit agencies
7 toll bridges
One MPO: MTC
Bay Area Local Street and Road Conditions
Average PCI = 64
Reflect U.S. economic downturn
Closer to “tipping point”
60
62
64
66
68
70
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
San Francisco Bay Area Pavement Conditions
64 P
CI
Why are Local Streets and Roads a Regional Concern?
Supports All modes of
transportation
$50 billion replacement value
Conditions are facing steep decline
Escalating deferred maintenance jeopardizes funding for All transportation priorities
MTC’s StreetSaver®
Used by all 109 Bay Area jurisdictions; 300 nationwide
Developed 25 years ago
Designed specifically for local agencies
Minimize costs - maximize benefits
Purpose:
Document conditions & needs
Promote pavement preservation
Better Pavement Management in Bay Area
MTC is recognized by the FHWA as “one of the first regions in the country to implement a pavement management system— FHWA Office of Asset Management
Pothole Report: http://tinyurl.com/6t6y5lm
28-Year Needs Assessment
County Available Revenues Pavement Needs Non-Pavement
Needs Total Capital
Needs Total Remaining
Capital Needs
Alameda $ 2,148 $ 3,715 $ 4,082 $ 7,798 $ 5,650
Contra Costa $ 2,915 $ 3,111 $ 2,674 $ 5,786 $ 2,871
Marin $ 655 $ 865 $ 641 $ 1,506 $ 852
Napa $ 219 $ 1,087 $ 429 $ 1,516 $ 1,297
San Francisco $ 2,299 $ 2,416 $ 2,363 $ 4,778 $ 2,480
San Mateo $ 1,440 $ 1,929 $ 1,984 $ 3,913 $ 2,473
Santa Clara $ 3,374 $ 5,776 $ 5,118 $ 10,894 $ 7,520
Solano $ 488 $ 1,906 $ 1,289 $ 3,195 $ 2,707
Sonoma $ 994 $ 3,699 $ 1,319 $ 5,018 $ 4,023
REGION $ 14,531
$ 24,504
$ 19,899
$ 44,404
$ 29,872
($ in Billion)
Performance Measures
Deferred Maintenance Pavement Condition Index (PCI)
Remaining Service Life Threshold
Remaining Service Life Min % in Very Good Condition Max % in Poor Condition Pavement Maintenance Index Actual PM Work / PM Work Recommended
Performance Measures
Difficult to find a “One Size Fits All” performance measure
Guiding principles:
Measurable
Objective as possible
Can be fairly applied
Utilizes data widely available
Meaningful (promotes pavement preservation)
Performance Measure Criteria
Performance‐Based Funding Allocation Formula
Shortfall
Lane Miles
Pavement Preservation Index
Shifts practice from “worst first” to preventive maintenance
Replaces “Maintenance of Effort”
No advantage or disadvantage due to existing network features or budget
Promotes pavement preservation principles
Data from StreetSaver
Outcome-Driven Performance Measure
Your Tax Dollars At Work
Performance
Use of Asset Management
Promote sound pavement preservation
Environmental Sustainable technologies
Outcome driven – “worst first” to preventive maintenance
Accountability
Taxpayers know where the money is spent
Establish “maintenance of effort” for local agencies
Transparency
Report card on pavement conditions
Summary
Stretching Maintenance $$$
Environmental Sustainability
Performance Preservation
Contact
Sui Tan, P.E. StreetSaver Program Manager
Metropolitan Transportation Commission Oakland, California
(510) 817-5844 [email protected]
Forecasting the Life of Asset Preservation
Treatments: A Comparative Evaluation of Tools and Techniques
Eleni Bardaka, Purdue University Samuel Labi, Purdue University
John Haddock, Purdue University
TRB Webinar: Transportation Asset Management September 19th 2012
Outline
• Introduction • Objectives & Scope • Literature Review • Methodology • Application to PMS Practice • Conclusions
2
Asset Performance Predictions • Essential Input for: Preservation Treatment Evaluation
- Ascertain effectiveness of actions - Establish optimal preservation strategies that minimize asset
life-cycle cost - Re-Evaluate design guides & preservation manuals
Project-Level Planning & Decision-Making
Network-Level Physical and Fiscal Needs Assessment - Project prioritization & programming
Asset Valuation
3
Problem Statement • Asset Management (AM) needs appropriate preservation
modeling techniques to ascertain the reliability of system decisions and operation.
• Most techniques used in the past yield models that:
▫ Need further refinement to become more reliable
Or
▫ Are not practical for PMS use
4
Study Objectives • Develop an enhanced methodology for performance
prediction that duly considers: ▫ Nature of preservation data ▫ Reliability of predictions ▫ Purposes and integration requirements of AM
• Use the developed methodology to: ▫ Predict post-rehabilitation performance ▫ Estimate treatment effectiveness (service life) ▫ Assess future needs
5
Study Scope AM Component System: PMS
Preservation Type: Rehabilitation, Thin Overlay
Data for illustration and configuration of the proposed methodology:
Indiana Interstates, rehabilitated during 1996-2006
Treatments: 1. Thin HMA Overlay 2. Structural HMA Overlay 3. Functional HMA Overlay 4. Crack & Seat PCC, and HMA
Overlay 5. Repair PCC and HMA Overlay 6. Rubblize PCC and HMA Overlay 7. PCC Overlay on PCCP
6
Review of Past Practices and Available Techniques 1/3
Empirical Technique
Highway Agencies
Advantages for PMS Application
Disadvantages for PMS
Application Illinois DOT, Washington State DOT, Louisiana DOT, Utah DOT, Minnesota DOT
- Requires one explanatory variable
- Does not require
investigation of the relationship between deterioration and its causes
- Not always reliable
Markov Chains Arizona DOT, Michigan DOT, Ohio DOT, Georgia DOT
- Reliable representation of the network condition
- Can be integrated with Network Programming and Budgeting
- Inconvenient for project-level management
Techniques Widely Used by Highway Agencies
Refs: Bham et al, 2003; Butt et al, 1994; Chou et al, 2008; Golabi et al, 1982; Jiang et al, 1988; Li et al, 1996; Li et al, 2006; Madanat et al, 1995; MnDOT, 2011; Silva et al, 2000; Pierce et al, 2004; UDOT, 2009; Wang et al, 1992 & 1994; Wang et al, 2010; Washington et al, 2011.
7
Empirical Technique
Investigated for PMS By:
Advantages for PMS Application
Disadvantages for PMS Application
Multivariate Regression
Rajagopal and George, 1991; Sebaaly et al, 1995; Mohamad et al, 1997; AASHTO, 2008 (ME-PDG)
- Relatively easy to estimate - Results easy to interpret
- Requires multiple explanatory variables
Bayesian Regression
Hajek and Bradbury, 1996; George,2000
- Useful for agencies that face lack of field data
- Requires a source of “prior” information/knowledge along with actual performance data
Autoregressive Models
Abu-Lebdeh et al, 2003 - Suitable for short-term performance prediction
- Requires information on previous year’s performance
Neural Networks Ferregut et al, 1999; Lou et al, 2001; Yang et al, 2003; Pekcan et al, 2008
- Can more easily model nonlinear data and complex interactions
- The estimation procedure resembles a “black box”
- Long and complicated equat.
Survival Analysis Gharaibeh and Darter, 2003; NCHRP 08-71, 2011
- Flexible in terms of data requirements
- Difficult result interpretation - Used for service life only
Refs: Aguiar-Moya et al, 2011; Amador and Mrawira, 2011; Attoh-Okine, 1994; Kajner et al, 1996; Kargah-Ostadi et al, 2010; Labi and SInha, 2004; Morian et al, 2011; Paterson, 1987; Paterson and Chesher, 1986; Prozzi and Madanat, 2003; Prozzi and Madanat, 2010; Wang et al, 2005; Washington et al, 2011; Yang, 2007; Yu et al, 2008.
8
Review of Past Practices and Available Techniques 2/3
Techniques Investigated for Implementation in PMS
Empirical Technique
Pavement Research
Advantages for PMS Application
Disadvantages for PMS
Application Mixed Linear Models (One-Way Random Effects)
Madanat and Shin, 1998; Archilla, 2006; Yu et al, 2007; Chu and Durango-Cohen, 2008; Hong and Prozzi, 2010
- Account for unobserved variables
- Reliable “population-wide” and
in-sample predictions
- Flexible to accommodate complicated data structures
- In-Sample predictions require the estimation of the model in parallel
Seemingly Unrelated Equations (SURE)
Prozzi and Hong, 2008; Anastasopoulos et al, 2012
- Captures the correlation of the different performance indicators, and thus provides a more efficient estimation than single-equation approaches
- Requires multiple explanatory variables and at least two indicators of pavement performance
Refs: Hsiao, 1993; Littell et al, 2006; Washington et al, 2011.
9
Review of Past Practices and Available Techniques 3/3
Other Available Techniques
PMS Purposes and Requirements Related to Treatment Analysis
• Project-Level Planning & Decision-Making ▫ Short-term/Long-term performance prediction for a
treatment applied to a specific pavement section
• Network-Level Needs Assessment ▫ Remaining Service Life for past-rehabilitated pavements
• Treatment evaluation ▫ Long-term treatment effectiveness models or estimates,
such as the average service life
10
Characteristics of Pavement Rehabilitation Data
Rehabilitation Treatment X
Contract 1 Constructed in year t
on j road miles
Condition Measurements
for
Section 1 - year t - year t+1
- … - year T
…...
Condition Measurements
for
Section j - year t - year t+1
- … - year T*
…… Contract i
Constructed in year t’ on j’ road miles
11
Formulation Assumption
•
Contract i
• 1st Level
Pavement Section j • 2nd Level
Repeated Measures • 3rd Level
12
3-Level Nested Linear Model Case Study: Repair PCC and HMA Overlay, Indiana IS
Independent Variable Parameter Estimate t-Statistic Constant -162.85 -5.67
Treatment Age [years] 4.104 25.28
Commercial Vehicles [1000vehicles/day] 2.588 6.52
Log(Pre-Treatment IRI) 93.338 6.81
Model Statistics Contracts: 12, Pavement Sections: 72, Observations: 694
166.370 2.01
62.267 4.13
121.120 17.44
Restricted LL (convergence) - 2,719.9
Restricted LL (regression) -2,902.2
Performance Indicator: IRI
13
Applicability to PMS Practice
•
14
Functional Overlay Treatment Age
Average IRI Estimate Lower 95% Limit Upper 95% Limit
12 109.8 102.6 117.0 13 114.7 107.2 122.1 14 119.5 111.9 127.2 15 124.4 116.5 132.3 16 129.3 121.0 137.5 17 134.1 125.7 142.6 18 139.0 130.2 147.8
Service Life Estimate and Range
Range of Service Life
There is 95% certainty that, 12.6 years after the treatment application, the average treatment performance will be equal to or higher than the performance threshold.
There is 95% certainty that, 15.8 years after the treatment application, the average treatment performance will be lower or equal to the performance threshold.
15
Applicability to PMS Practice
Treatment Service Life (years) [Linear Regression]
Service Life (years) [Mixed Models]
Thin HMA Overlay 12 (11-13) 12 (9.6-13.4)
Structural HMA Overlay >40 24.5 (20 -31.4)
Functional HMA Overlay 18 (17-19) 12.7 (11.3-14)
Crack & Seat PCC, and HMA Overlay
23 (17-29) 10.7 (9.6-12)
Repair PCC and HMA Overlay 17 (16-18) 13 (11.2-14.8)
Rubblize PCC and HMA Overlay 31 (27-35) 17.8 (16-20)
Average Treatment Service Life Estimation: Interstate Pavements, Indiana
BLUP using Mixed Model Regression
17
•
Applicability to PMS Practice
Performance-Based Needs Assessment Framework
(Sinha et al, 2005)
Determine Remaining Service Life of Each Pavement Section in the Network
Assess Physical Needs for Each Pavement Section in the Network
Assess Monetary Needs for Each Pavement Section in the Network
Determine Inventory
Asset Performance Models
Establish Performance Thresholds
Develop Cost Models for Each Preservation Treatment
Define Horizon Period
Select First Year of Horizon Period
Repeat Analysis for All Years within the Horizon Period
The BLUP method, using à priori information, can offer reliable predictions for each section in the network of rehabilitated pavements.
These predictions could be used to more reliably assess the future preservation needs of these pavements.
The effect of increased prediction reliability on needs assessment was found higher for short-term monetary needs estimation (35% underestimation).
18
Applicability to PMS Practice
Conclusions • The developed framework for treatment analysis :
▫ Takes into account the challenging structure of pavement rehabilitation data
▫ Covers the purposes and requirements of PMS regarding treatment analysis:
Project-Level Planning & Decision-Making → “population-wide” predictions
Network-Level Needs Assessment → conditional predictions (BLUP method)
Treatment Evaluation → Service Life Estimates and Range
▫ Accounts for unobserved variables:
Among pavement sections that belong to the same contract (e.g. construction quality)
Among measurements that refer to the same pavement section (e.g. layer thicknesses)
• The presented model formulation:
▫ Can be used for continuous performance indicators ▫ Can be easily modified to incorporate non-linear functions
19
Forecasting the Life of Asset Preservation
Treatments: A Comparative Evaluation of Tools and Techniques
Eleni Bardaka, Purdue University Samuel Labi, Purdue University
John Haddock, Purdue University
TRB Webinar: Transportation Asset Management September 19th 2012
Supplementary Slides
Disclaimer
• This research is part of an ongoing project for the Indiana DOT (INDOT).
• The research results may or may not be implemented by INDOT until the research project is finalized.
Impact of Proposed Methodology on Future Network Needs Estimation
• Investigate if increased reliability of disaggregate performance predictions has an impact on the aggregate network monetary needs.
• Case Study: Indiana Interstates
Inventory: Indiana Interstate Pavements Rehabilitated
during the period 1996-2006
Horizon Period: 2010-2020
Prediction Methods: 1. Regression
2. Mixed Linear Models
Performance Indicator: International Roughness Index (IRI)
Performance Records Availability: 1995-2009
Cost Models: Average Treatment Cost
(INDOT Contracts Division)
Impact of Proposed Methodology on Future Network Needs Estimation
Future Physical Needs of Interstate Pavements in Indiana, Rehabilitated 1996-2006
Impact of Proposed Methodology on Future Network Needs Estimation • Future Monetary Needs of Rehabilitated Interstate Pavements in Indiana
BLUP vs. Population-wide predictions
Treatment Age
Pavement Performance After Rehabilitation, represented by IRI [in/mile]
Observed Linear Regression 3-Level Nested Linear Model
5 48
6 57
7 50
8 58
9 62
10 - 98 74
11 - 102 79
12 - 105 83
13 - 109 88
14 - 112 92
15 - 116 96
Corridor-Level Performance Measures to Support Cross-Asset Resource Allocation Strategies in Highway
Mohammadsaied Dehghani Ph.D. Candidate, Center for Sustainable Transportation Infrastructure, VTTI
Gerardo Flintsch Professor, The Via Department of Civil and Environmental Engineering Director, Center for Sustainable Transportation Infrastructure, VTTI
TRB Webinar: Transportation Asset Management: Current Trends in Pavement and Bridges
Outline
o Introduction
o Methodology
o Case study
o Applications
o Summary/conclusion
Introduction - What’s the Problem?
Each component contributes to the overall wellbeing of our road segment
Need to know the overall health Corridor → System
Objective
o Developing a framework to aggregate performance of roadway assets into overall roadway performance measures for cross-asset resource allocation,
tradeoff analysis, etc.
Background
o Performance measures already developed for individual assets
o Several studies trying to aggregate all different types of performance measures for one asset (e.g. COST method)
o Aggregating performance of multiple assets into corridor-level (system-level) performance measures Not yet explicitly addressed
First Attempt
Pavements:•Cracking•IRI•Rutting•FWD Data•Etc.
Structural
Functional
Environmental
Safety
Pavement Health Rating
Bridges:•Primary Members•Abutments•Bridge Deck•Etc.
Functional
Environmental
Bridge Health Rating
Facilities:•Toll Plazas•Weigh Stations
Safety Features:•Signs•Pavement Markings
Other
Serviceability
Functional
Serviceability
Functional
Facilities Health Rating
Safety Features Health Rating
Structural
Safety
Corridor Health Rating
Performance Indicators
•PI_IRI•PI_Rut•PI_Cr
Performance Indicators•PI_Girder•PI_Abut•PI_Deck
Performance Indicators
•PI_Toll
Performance Indicators
•PI_Sign•PI_Mark
Quality Measures
Performance Indicators
Health Indicators
Asset Health Ratings
PI_Cracking
PI_Primary members
Performance Indicators
PI_IRI
PI_Rutting
PI_Abutment and Piers
PI_Deck
PI_Other elements
Structural
Functional
Asset Health Indicators
Structural
Functional
Corridor Health
Indicators
Corridor Structural Indicator
Corridor Functional Indicator
Corridor Overall Health Rating
Pave
men
tsBr
idge
s
Quality Measures
• Cracking• IRI• Rutting• FWD Data• Etc
• Primary Members
• Abutments• Bridge
Deck• Other
PI_Sign
PI_Marking
Structural
Functional
Corridor Safety
Indicator
Safe
ty F
eatu
res
• Pavement signs
• Guard rails• Pavement
markings PI_Guardrail
Safety
Safety
Safety
Revised Methodology
Performance Indicators (PI)
o Value based on quality measures o Reflects remaining time until quality measure
exceeds acceptable limits o Scale 0 to 10 10 – “Like New” condition 0 – Quality measure at unacceptable level
o Calculation different for each asset Different quality measures available
Calculations
PI_Cracking
PI_Primary members
Performance Indicators
PI_IRI
PI_Rutting
PI_Abutment and Piers
PI_Deck
PI_Other elements
Structural
Functional
Asset Health Indicators
Structural
Functional
Corridor Health Indicators
Corridor Structural Indicator
Corridor Functional Indicator
Corridor Overall Health Rating
Pave
men
tsBr
idge
s
Quality Measures
• Cracking• IRI• Rutting• FWD Data• Etc
• Primary Members
• Abutments• Bridge
Deck• Other
PI_Sign
PI_Marking
Structural
Functional
Corridor Safety
Indicator
Safe
ty F
eatu
res
• Pavement signs
• Guard rails• Pavement
markings PI_Guardrail
Safety
Safety
Safety
kjpijw
jz
k: type of quality measure j: Health Indicator type i: Asset type
1
_ min(10,max(0, -0.05 12))PI IRI IRI= × +
10ff
f
WEQPI
TEQ
= ×
f fi iWEQ EQ α= ×∑
ik kj
kij
kjk
PI pAHI
p
×=
∑∑
2
3
4
,
∑∑ ×
=
iij
iijij
j w
wAHICHI
∑∑ ×
=
jj
jjj
z
zCHIOCHR
Case Study
Pavement (for every 0.1 mile) IRI Rut depth Longitudinal cracks Alligator cracks Transverse cracks
Bridges (five condition states) Primary members Deck Abutment and piers
I-81 North bound Mileage 50-100 Assets
Pavements Bridges
Measures Structural Functional
Calculations
Results - Pavements
0
2
4
6
8
10
50 60 70 80 90 100
Stru
ctur
al
0
2
4
6
8
10
50 60 70 80 90 100
Func
tiona
l
Milepost
Results - Bridges
0
2
4
6
8
10
50 55 64 68 78 86 94 100
Stru
ctur
al
0
2
4
6
8
10
50 55 64 68 78 86 94 100
Func
tiona
l
Mile
Results (cont.)
0 2 4 6 8
10
50 60 70 80 90 100
Stru
ctur
al
0 2 4 6 8
10
50 60 70 80 90 100
Func
tiona
l
Mile
0 2 4 6 8
10
50 55 64 68 78 86 94 100
Stru
ctur
al
0 2 4 6 8
10
50 55 64 68 78 86 94 100
Func
tiona
l
Mile
Pavements Bridges
0 2 4 6 8
10
50 60 70 80 90 100 Stru
ctur
al
Results (cont.)
Homogenous corridor health indicators
Continuous profiles for each indicator
0 2 4 6 8
10
50 60 70 80 90 100 Func
tiona
l
0 2 4 6 8
10
50 60 70 80 90 100
Ove
rall
Cor
ridor
He
alth
Rat
ing
Mile
7.8
7.2
7.6
Corridor
Applications - Resource Allocation
Common “quality” scale needed for cross-asset resource allocation
Application (Simplified Example)
Assets Pavements Bridges
Health Indicators Functional Structural
Pa
vem
ent
Treatment Type
Treatment Cost
($/Lane-mile)
Extended Life (Years)
Maximum Functional Gain (unit)
Maximum Structural Gain (unit)
Preventive 10,000 3 1.5 - Corrective 80,000 8 5 2 Restorative 200,000 12 8 5
Reconstruct/ Rehabilitate 500,000 20 10 10
Bri
dge
Treatment Cost ($/ft2)
Extended Life
(Years)
Maximum Functional Gain (unit)
Maximum Structural Gain (unit)
Epoxy Overlay 70 6 4 - Deck
Replacement 120 15 6 1.5
Heavy Rehab 370 25 8 5 Reconstruction 1000 40 10 10
Pavement Bridge Lane. Mile 4 - Area (ft2) - 2000
Functional Indicator 6 4
Structural Indicator 5 7
Required Budget ($) 1,500,000
Available Budget($) 1,000,000
Application (Simplified Example)
Scenario selected
Treatment applied based on budget
Performance averaged over 5 year analysis period Optimal allocation
10% 20% 30% 40% 50%
7 7.2 7.4 7.6 7.8
8 8.2 8.4 8.6 8.8
50% 60% 70% 80% 90%
Bridge Budget Share
Indi
cato
r
Pavement Budget Share
Functional Indicator
Overall Indicator
Structural Indicator
Future Research
Consider more assets
Consider other performance measures, particularly, safety, and environmental factors
Conclusion
o Proposed method for homogenous aggregation of performance measures
o For strategic level decision making
Comparing investment trade-offs
Support cross-asset resource allocation
Acknowledgement
VDOT Tanveer Chowdhury,
Raja Shekharan and William Duke (Office of Asset Management)
Richard Thompson (Office of Structure and Bridge)
Jeff Price (Operations Planning Division)
MP 50-100
Detailed information can be found in: Dehghanisanij, M., Flintsch, G.W. , and Verhoeven, J.G. (2012). Framework for Aggregating Highway Asset Performance Measures: Application to Cross-Asset Resource Allocation. . In Transportation Research Record: Journal of the Transportation Research Board, No. 2271, pp. 37-44. Verhoeven, J. and Flintsch, G. (2011). Generalized Framework for Developing a Corridor-Level Infrastructure Health Index. In Transportation Research Record: Journal of the Transportation Research Board, No. 2235, pp. 20–27.
Please contact me if you have questions: Mohammadsaied Dehghani Ph.D. Candidate, Center for Sustainable Transportation Infrastructure, VTTI [email protected] Phone: 540-808-8347
TRB Webinar: Transportation Asset Management: Current Trends in Pavement and Bridges
Summary 0 Four presentations providing some highlights from
the 2012 Transportation Asset Management Conference
0 Conference materials including presentations available from the TRB website
0 Planning has started for the 10th National Conference, most likely taking place in the spring of 2014 0 Pooled fund participation available
Questions? 0 Ron Hagquist, Texas Department of Transportation 0 Use of Management Science Analytics for Asset Management at
Texas DOT 0 Sui Tan, California Metropolitan Transportation Commission 0 Performance-Based Approach to Funding Policy for Local
Streets and Roads 0 Eleni Bardaka, Purdue University 0 Forecasting the Life of Asset Preservation Treatments: A
Comparative Evaluation of Alternative Tools and Techniques 0 Mohammadsaied Dehghani, Virginia Polytechnic Institute and State University 0 Corridor-Level Performance Measures to Support Resource
Allocation Strategies in Highways