1. ZETA-TECH Associates, Inc.
Use of Track Component Life Prediction Models in Infrastructure Management
Todd Euston, MCE
Project Manager
ZETA-TECH Associates, Inc.
900 Kings Highway North
Cherry Hill, NJ USA 08034
+1 (856)

2. Objectives of Maintenance Planning
What is in track now?
Track Measurement Systems
Exception Reports
Data Base
What will I need?
This Year (Short Term)
1 - 3 Years (Medium Term)
Years (Long Term)
Maintenance Requirement Forecasting
Quantity of Components (Rail, Ties, Ballast)
Budget
What should be done first?
Prioritization of needs
Ability to Expand/Contract Budget
Decision-making tools (What if…?)

3. Infrastructure Maintenance Management
Focus of this presentation is on Maintenance Planning
Component Maintenance Planning
Determine short, medium and long term component needs (quantity, budget)
Spot Maintenance
Production (Out-of-Face)
Minimize annual maintenance requirements
Medium to Long term planning capability

4. Maintenance Management
Rail
Rail Replacement Forecasting
Fatigue Life
Wear Life
Grinding Requirements/Planning
Rail Test Scheduling
Ties (Sleepers)
Spot Maintenance
Replacement Analysis
Degradation/Forecasting
Surfacing
Track Geometry Car Scheduling
Spot Maintenance Requirements
Forecasting Surfacing Cycles
Track System Approach
Resource Allocation Models
Prioritization based on Track Component Indices
Condition Based

5. Component Life Forecasting
Analysis current (and past) condition
Predict the rate of failure of key track components
Predict component replacement point (life)
Predict where and when key components must be replaced
Forecasting of annual track component replacement requirements
Forecasting of annual track component budgetary requirements
Allows for more accurate planning and scheduling of track maintenance activities

6. Basic Modeling Approach
Make use of Inspection Information Available
Example: Rail
Defects (UT)
Wear
Profile
Utilize This Information To Its Fullest Extent
Not Just For Exception Processing
Project from current conditions to future condition
Divide Track into Homogeneous Analysis Segments
Statistical And Engineering Analysis
Analyses Are Hierarchical To Allow For Incomplete Data
Presentation Of Results in Useable Format
Rail caused derailments are a major category of derailments in North America, often costing in excess of $200,000 on average.
In order to minimize these type of derailments, ultrasonic testing is the primary method for finding rail defects before they cause a derailment.
Traditional scheduling of test frequencies has been based on rules of thumb including rail age, annual traffic density, defect counts, and others.
A well defined risk-based methodology for scheduling tests has not traditionally been used.
BNSF has implemented such a risk-based methodology which accounts for several factors including:
allowable risk
rail age
rail usage, and
historic failure rate

7. Example Model: Rail Life Analysis (RailLife)
Analysis of short and long term rail replacement requirements
Analysis methodology addresses both rail wear and rail fatigue
Rail Wear Life Analysis
Rail Fatigue Life Analysis
Track segmentation consists of discrete homogeneous segments based on characteristics of the track, traffic, and maintenance history
Curves vs. tangents
Subdivided further based on:
- Rail installation date - Rail metallurgy
- Rail size - Traffic
Wear analysis performed on individual curve and tangent segments
Fatigue life analysis performed on segments of sufficient length
(2-10 miles) to provide enough data to perform a meaningful
analysis

8. Rail Life Model Part 1: Rail Fatigue Life Analysis
Uses Actual Railroad Data
Rail Installation
Defects (Service and Detected)
Tonnage
Track Layout
Forecasting of fatigue life of rails uses two parameter Weibull distribution approach
Actual defect history is used to project the future defect growth rate
When defect rate (in defect/mile/year or defects/rail/MGT) exceeds a predefined value the rail becomes a candidate for replacement
Forecast life is calculated in cumulative MGT
Calculated fatigue life together with the MGT history of rail segment is used to determine segment’s replacement year

9. Weibull Plot
The Weibull function plotted here on a log-log graph, shows the linear representation of increased defect probability with increased rail age.
Rail age is shown here as cumulative tonnage over the rail.

10. Defect Rate Plot
This graph shows the Weibull rate equation which clearly shows the accelerated increase of annual rail defects with increased rail age in cumulative MGT.

11. Rail Life Model Part 2: Rail Wear Life Analysis
Uses Actual Railroad Data
Rail Installation Data
Wear Measurements (Head and Gage): LaserRail, Orian
Tonnage
Track Layout
Track Segmented Into Logical Analysis and Replacement Units
Wear Degradation Modeled Using Regression Analysis and Actual Data
Standards Applied To Degradation Rate To Determine Forecast Replacement Date
Results Reported

12. Rail Wear Analysis
Methodology for forecasting the wear life of rail is a two part approach
Use measured rail wear values and corresponding tonnages to determine the rail wear rate relationships
Multivariate statistical analysis of wear data
Use engineering equations calibrated to railroad specific track and operating parameters to forecast the rail wear life (if insufficient data)/verify validity of statistical data
Wear life calculated based on wear rates for head and gage face wear of each segment
Using Railroad replacement standards for head and gage face wear
Remaining wear life of segment defined as minimum of head and gage face remaining wear lives
Replacement year due to wear is then determined from
remaining life in MGT and annual MGT

13. Wear Rate Plot

14. Combined Rail Life Analysis Reporting
Analysis of overall rail life takes part in two steps, fatigue life analysis and wear life analysis
Each rail segment is analyzed independently for both fatigue and wear, using the appropriate segmentation file
Combined report developed for determining the overall rail requirements for the segment
Overall rail life is determined for each homogeneous segment (wear and fatigue) defined from the input data
Smaller ( finer) segmentation used for spot replacement
Segments combined for production replacement
Wear and fatigue life files are overlaid and combined based on production requirements

15. Rail Wear Life Forecast: Spot (Curve Patch) Segment Results

16. Rail Fatigue Life Forecast: Production Segment Results
Division Track Prefix From MP To MP Rail Side Last 2 Years Length,
Defects Mile
V L
NA R
NA L
N L
NA L *
NA R
NA L
NA R
NA L
Rail miles to be replaced in
Summary <1.5 yrs (Red) yrs (Yellow) yrs (Green)
*includes short segment between two contiguous red segments

17. Combined Fatigue and Wear Replacement Forecast

18. RailLife Reporting Format
This graph shows the Weibull rate equation which clearly shows the accelerated increase of annual rail defects with increased rail age in cumulative MGT.

19. EXAMPLE: Tie Replacement Planning and Scheduling
Most tie inspection is visual, based on the evaluation of a tie inspector
Current railroad practice is to count number of “bad” ties per mile
New generation TieInspect unit allows for mapping 100% tie condition
Four classes; Very Bad, Bad, Marginal, Good
Records location and condition for each tie
Results used to calculate tie replacement requirements
New generation of “track strength” inspection systems allows for measurement of gage holding strength of tie
Used for locating “weak spots” in the track
Used for “mapping” track strength/gage

20. TieInspectTM
This screen shot shows a sample of the parametric analysis screen for a given segment of track.
This example shows that for a 60 mile segment of track with 50 MGT per year tested twice per year, and a risk factor of 0.1 defects per mile, as well as 19 service and 20 detected defects results in an increased test requirement of two and a half test per year or a test every 20 MGT.

21. TieInspectTM
This screen shot shows a sample of the parametric analysis screen for a given segment of track.
This example shows that for a 60 mile segment of track with 50 MGT per year tested twice per year, and a risk factor of 0.1 defects per mile, as well as 19 service and 20 detected defects results in an increased test requirement of two and a half test per year or a test every 20 MGT.

22. Detailed Tie Condition Analysis TieInspect
Immediate tie replacement analysis
Based on
Cluster Analysis
Safety Standards (FRA/other)
Maintenance Standards
Good/Marginal ties on each side of “Bad” tie
Maximum number of bad ties
Varies with Class/Speed/curvature
Generates detailed tie by tie replacement List
Provides field deployable information on ties to be replaced

23. Tie Degradation Modeling and Forecasting (TieLife)
Predicts tie requirements on a segment by segment basis
Analysis of tie requirements for “homogeneous segment”
Based on Current Tie Conditions
- TieInspect tie condition “map”
- Bad, marginal, good ties
Uses Tie Statistical failure analysis
“Forest Products” failure distribution
Short, Medium and Long Term Time Horizons

24. Tie Forecasting Report
Year by Year Forecast
Year Cum Ties
0 0
1 23
2 198
3 448
4 276
5 49
6 14
7 16
8 22
9 34
10 51
11 78
12 87
13 118
14 134
15 141
16 166
17 161
18 146
19 157
20 142
21 119
22 107
23 93
24 95
25 83
26 61
27 76
28 72
29 57
30 51
Next Tie Gang 2007
Second Tie Gang 2018

25. Tie Forecasting Report (700 Tie Threshold)

26. Tie Forecasting Report (1000 Tie Threshold)

27. EXAMPLE: Track Geometry Degradation (TrackLife)
Using Track Geometry Data
Analysis of multiple track geometry data measurements
Track Geometry Analysis
Calculation of TQIs
Surfacing Forecasting and Planning
Forecasting of surfacing due dates

29. Track Geometry Analysis
Calculation of TQIs
Calculate Individual TQIs
Calculate Total TQI
Weightings for Individual TQIs
TQIs used as basis for surfacing analysis
Used as basis for forecasting next surfacing dates and locations

30. Surfacing Forecasting and Planning
TQIs used to project rate of track geometry degradation
TrackLife calculates maintenance due dates
Based on railroad or FRA thresholds
Based on segment maintenance history
Maintenance activity can be derived from geometry data
Project Forward to Next Maintenance Date

32. Forecast Surfacing Gang Dates

33. Surfacing Forecast

34. Maintenance Management using Component Life Forecasting
Make use of new generation track inspection technology
Predict the rate of failure of key track components
Predict component replacement point (life)
Predict where and when key components must be replaced
Forecasting of track component replacement requirements (short, medium, long term)
Forecasting of budgetary requirements
Allows for more accurate planning and
scheduling of track maintenance activities
More efficient use of resources

35. Budget Forecast System Rail Requirements
Millions of
Dollars

36. Rail Maintenance Requirements Effect of “Deferred Maintenance”
Miles of Rail
No Deferred Maint.
Deferred Maint.

37. What If Capability Analysis of alternate maintenance options/strategies
System Rail Budget
Millions of Dollars
Add. Grinding Costs Profile Grinding Conv. Grinding

38. Risk Management
Use of inspection data to control safety/derailment risk
Risk management in scheduling inspections
Risk based Ultrasonic Rail Test scheduling to control broken rail derailments
Risk based Track Geometry scheduling to control track geometry related derailments
Risk based Track Strength test scheduling to control tie/fastener related derailments
Allows for definition of level of risk by line segment based on:
Type of traffic carried
Operating conditions
Potential for high cost accidents
Example: Rail Test Scheduling (RailTest)
Level of Risk for Bulk Commodity Freight Line 0.1
Level of Risk for freight Line with Limited Passenger Trains 0.05
Level of Risk for Passenger Lines

39. Summary
Forecasting modeling allows for the prediction of future (short, medium, and long term) component replacement requirements
Models forecast component replacement requirements and forecast of future replacement needs
Models forecast maintenance gang schedules and allow for developing uniform annual programs
Component life forecast information allows railroads to
project future capital requirements and plan future
budgets
Risk management used to control level of defects
Potential for component failure
Derailment