You are on page 1of 35

HDM-4 Calibration

Reliability of Results Depends On:

• How well the available data represent the


real conditions to HDM
• How well the model’s predictions fit the
real behaviour and respond to prevailing
conditions

2
How Credible are HDM-4 Outputs?

• Depends on Level of Calibration (controls


bias)
• Depends on accuracy and reliability of input
data (asset & fleet characteristics, conditions,
usage)
• HDM-4 has proved suitable in a range of
countries
• As with any model, need to carefully check
output with good judgement

3
Approach to Calibration

• Input data
 Must have a correct interpretation of the
input data requirements
 Have a quality of input data appropriate for
the desired reliability of results
• Calibration
 Adjust model parameters to enhance the
accuracy of its representation of local
conditions

3
4
Data & Calibration

• Need to appreciate importance of data over


calibration
• If input data are wrong why worry about
calibration?

Calibration

'The Depth of the Sea and


Data
the Height of the Waves'

5
Calibration Focus

• Road User Effects


 Predict the correct magnitude of costs and
relativity of components - data
 Predict sensitivity to changing conditions -
calibration
• Pavement Deterioration & Works Effects
 Reflect local pavement deterioration rates and
sensitivity to factors
 Represent maintenance effects

6
Estimating Calibration Coefficients

Model

Defect (%)
Extent of
We attempt to
minimize the
"mistake"

Actual
deterioration

Time

Un-calibrated Calibrated

Progression
Predicted
Progression
Predicted

Actual Progression Actual Progression


7
Hierarchy of Effort

Time Re quire d

Experimental
Ye ars
Surveys and
Research

M onths
Field Surveys

We e ks
Desk Studies

Re source s
Limite d M ode rate Significant Re quire d

General Planning Project A ppraisal Research and


Quick Prioritisation Detailed Feasibility Development
Preliminary Screening

Coarse Estimates Reliable Estimates

8
Calibration Levels

• Level 1: Basic Application


 Addresses most critical parameters
 ‘Desk Study’
• Level 2: Calibration
 Measures key parameters
 Conducts limited field surveys
• Level 3: Adaptation
 Major field surveys to requantify
relationships
 Long-term monitoring
9
Level 1 - Basic Application

• Required for ALL HDM analyses

• Once-off ‘set-up’ investment for the model

• Mainly based on secondary sources

• Assumes most of HDM default values are


appropriate

10
Level 2 - Calibration

• Makes measurements to verify and adjust


predictions to local conditions

• Requires moderate data collection and


moderate precision

• Adjustments entered as input data, typically


no software changes

11
Level 3 - Adaptation

• Comprises
 Structured research, medium term
 Advanced data collection, long term
• Evaluates trends and interactions by
observing performance over long time period
• May lead to alternative local
relationships/models

12
Important Considerations

• Calibrate over full range of values likely to be


encountered
• Have sufficient data to detect the nature of
bias and level of precision
• High correlation (r^2) does not always mean
high accuracy: can still have significant bias
• Primary aim: minimize bias (mean observed
values / mean predicted values)

13
Bias and Precision
Low Bias Low Bias
A High Precision
B Low Precision

Data
Predicted

Predicted
Data

Observed = Predicted Observed = Predicted

Observ ed Observ ed

High Bias High Bias


C High Precision
D Low Precision

Observed = Predicted Observed = Predicted


Predicted

Predicted

Data

Data

Observ ed Observ ed

14
Calibration Adjustments

A Rotation B Translation

Trans lation

R otation
Observ ed = Predicted Observ ed = Predicted
Predicted

Predicted
D ata D ata

Observed Observed

Trans lation
Rotation and
C Translation
R otation

Observ ed = Predicted
Predicted

D ata

Observed

15
Correction Factors

• Used to correct for bias


• Two types of factors
 Rotation (CF = Observed/Predicted)
 Translation (CF = Observed - Predicted)
• Rotation factors adjust the slope
• Translation factors shift the predictions
vertically

16
HDM-4 Road Deterioration
Calibration Factors

All relationships have a calibration


factor - ‘K’ factor

Used to adjust
predicted to observed

17
Typical Relationship

Initiation of Cracking

ICA = Kcia{a0 exp[a1SNP + a2(YE4/SNP2)]}

Calibration Model
Factor Coefficients

18
Road Deterioration Calibration
Factors
Calibration Deterioration
Factor Model
Kddf Drainage Factor
Kcia All Structural Cracking - Initiation
Kciw Wide Structural Cracking - Initiation
Kcpa All Structural Cracking - Progression
Kcpw Wide Structural Cracking - Progression
Kcit Transverse Thermal Cracking - Initiation
Kcpt Transverse Thermal Cracking - Progression
Krid Rutting - Initial Densification
Krst Rutting - Structural Deterioration
Krpd Rutting - Plastic Deformation
Krsw Rutting - Surface Wear
Kvi Ravelling - Initiation
Kvp Ravelling - Progression
Kpi Pothole - Initiation
Kpp Pothole - Progression
Keb Edge Break
Kgm Roughness - Environmental Coefficient
Kgp Roughness - Progression
Ktd Texture Depth - Progression
Ksfc Skid Resistance
Ksfcs Skid Resistance - Speed Effects
19
Cracking Initiation Calibration

Crack Initiation

100
90
Percent Area of Cracking

80
70
60
50
40
30
20
10
0
0 5 10 15 20
Years

Kci = 1.00 Kci = 1.80 Kci = 0.55

20
Cracking Progression Calibration

Crack Progression

100
Percent Area of Cracking

90
80
70
60
50
40
30
20
10
0
0 5 10 15 20
Years

Kcp = 1.0 Kcp = 2.0 Kcp = 0.4

21
Road Deterioration Calibration (1)

• Simulation of Past Since Construction


 take sample of roads with historical data
(traffic, design, etc.)
 simulate with HDM-4 the deterioration from
construction time to current age
 compare the simulated results with actual
road condition at current age
 deal with the uncertainty regarding the
road conditon at construction time

22
Road Deterioration Calibration (2)

• Simulation from Two Points in Time


 take sample of roads with road condition data
available for two years (e.g. roughness
measurements surveyed in two different
years)
 simulate with HDM-4 the deterioration from
the first year to the second year
 compare the simulated results with the actual
road condition at the second year

23
Kazakhstan Calibration Example
Without Calibration Scenario
14.00
Observed Roughness Values (IRI, m/km)

Roughness Environmental Factor = 1.0


12.00 Cracking Initiation Factor = 1.0

10.00

8.00

6.00

4.00
Bias = Mean Observed / Mean
Predicted = 1.14
2.00

0.00 With Calibration Scenario


0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 14.00
Predicted Roughness Values (IRI, m/km)

Observed Roughness Values (IRI, m/km)


Roughness Environmental Factor = 1.5
12.00 Cracking Initiation Factor = 0.6

10.00

8.00

Roughness 6.00

surveys 4.00
Bias = Mean Observed / Mean
three years 2.00
Predicted = 1.03

apart
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
Predicted Roughness Values (IRI, m/km)
24
Road Deterioration Calibration (3)

• Controlled Studies
 collect
detailed data over time on traffic,
roughness, deflections, condition, rut
depths, etc.
 sections must be continually monitored
 long-term (5 year) commitment to quality
data collection

25
What to Focus On?

• HDM-III has about 80+ data items and model


parameters; HDM-4 has more.

• Sensitivity of each item has been classified by


sensitivity tests

• Simplify effort for less-sensitive items

26
Sensitivity Classes

Impact Sensitivity Impact


Class Elasticity
High S-I >0.50
Medium S-II 0.20-0.50
Low S-III 0.05-0.20
Negligible S-IV <0.05

27
Sensitivity Classes
Sensitivity Impact Parameter Important for Parameter Important for
2/ 3/ 4/
Class Elasticity Total VOC VOC Savings
S-I > 0.50 kp - parts model exponent kp - parts model exponent
New Vehicle Price New Vehicle Price
CSPQI - parts model
roughness term
C0SP - parts model constant
term
S-II 0.20 - 0.50 Roughness E0 - speed bias correction
E0 - speed bias correction ARVMAX - max. rectified
Average Service Life Average velocity
Annual Utilisation CLPC - labour model exponent
Vehicle Weight
S-III 0.05 - 0.20 Aerodynamic Drag Coefficient Beta - speed exponent
Beta - speed exponent Vehicle Age in km
BW - speed width effect C0LH - labour model constant
Calibrated Engine Speed term
CLPC - labour model exponent Labour Cost
C0SP - parts model constant Hourly Utilisation Ratio
term BW - speed width effects
CSPQI - parts model Number of tires per Vehicle
roughness term New tire Cost
Crew/Cargo/Passenger Cost Lubricants Cost
Desired Speed Crew/Cargo/Passenger Cost
Driving Power Vehicle Weight
Energy Efficiency Factors Number of Passengers
Fuel Cost
Hourly Utilisation Ratio
Interest Rate
Projected Frontal Area
S-IV <0.05 All Other Variables All Other Variables
28
Sensitivity Classes
Sensitivity Impact Parameter Outcomes Most Impacted
Class Elasticity Pavement Resurfacing Economic
Performance and Surface Return on
Distress Maintenance
S-I > 0.50 Structural Number 2/   
Modified Structural Number2/   
Traffic Volume 
Deflection3/   
Roughness  
S-II 0.20 - Annual Loading   
0.50 Age  
All cracking area  
Wide cracking area  
Roughness-environment factor  
Cracking initiation factor   
Cracking progression factor 
S-III 0.05 - Subgrade CBR (with SN) 
0.20 Surface thickness (with SN)  
Heavy axles volume  
Potholing area  
Rut depth mean 
Rut depth standard deviation 
Rut depth progression factor 
Roughness general factor  
S-IV < 0.05 Deflection (with SNC) 
Subgrade compaction  
Rainfall (with Kge) 
Ravelling area 
Ravelling factor 

29
Information Quality Levels
HIGH LEVEL DATA

System Performance
IQL-5 Performance
Monitoring

Planning and
IQL-4 Structure Condition Performance
Evaluation
Programme
Frictio Analysis or
IQL-3 Ride Distress
n Detailed
Planning
Project Level or
Detailed
IQL-2 Programme
Project
IQL-1 Detail or
Research
LOW LEVEL DATA

30
Information Quality Levels

• IQL-1: Fundamental Research


 many attributes measured/identified
• IQL-2: Project Level
 detail typical for design
• IQL-3: Programming Level
 few attributes, network level
• IQL-4: Planning
 key management attributes
• IQL-5: Key Performance Indicators
31
Adapting Local Data
Road Condition

IQL-2 IQL-2B IQL-3 IQL-4


Lane roughness (m/km IRI) Roughness (6 ranges) Ride quality (class)
All Cracks Area (% area) Cracking (score, or
Universal Cracking
Index, UCI)
Wide Cracks Area (% area)
Transverse thermal cracks
(no./km)
Ravelled Area (% area)
Potholes Number Disintegration (score) Surface Distress Index Pavement Condition
(units/lane-km) (SDI) (class)
Edge-break area (m2/km)
Patched Area (% area)
Rut Depth Mean (mm) Deformation (score)
Rut Depth Standard Dev.
(mm)
Macro-texture depth (mm) Surface texture (class)
Skid Resistance (SF50) Friction (class) Surface Friction (class)

32
Time Spend on Different
Phases of Analysis

Model
Calibration
10%

Establishing
Verification of
Reliable Input Running HDM-4
Output
Data 10%
20%
40%
Treatments,
Triggers and
Resets
20%

33
Can We Believe HDM-4 Output?

• Yes, if sufficiently calibrated


• HDM-4 has proved suitable in a range of
countries
• As with any model, need to carefully scrutinize
output against judgement
• If unexpected predictions occur, check:
 Data used
 Calibration extent
 Check judgment of the expert

23
34
For Further Information

• A guide to
calibration and
adaptation
• Reports on various
HDM calibrations
from:
www.lpcb.org

35