BRIDGE MANAGEMENT SYSTEMS &

IMPLEMENTATION – AN OVERVIEW

Professor Yew-Chaye Loo AM, Chair, Board of Directors, SIAMA

SIAMA+AIT Workshop on Effective BMS & BDC 24-26 June 2015, Bangkok

ACKNOWLEDGEMENTS

ASIAN INSTITUTE OF TECHNOLOGY

BANGKOK, THAILAND

SIAMA R&D Pty Ltd

GOLD COAST (AUSTRALIA) & NANJING (CHINA)

2

CONTENTS

Background: BAM & BMS

Backward Prediction Method (BPM)

Integrated Deterioration Prediction Method

Case studies & comparisons

Conclusion

3

BACKGROUND

Why Bridge Asset Management (BAM)?

What is Bridge Management System (BMS)?

4

5Bridge Statistics of AUSTRALIA

National Highways (2003) Arterial roads (2003) Local roads (1996)

Construction Era No. of bridges Total

areas (m2)

No. of bridges Total

areas (m2)

No. of bridges Total

Areas (m2)

Post – 1976 1,556 1,092,063 3,877 2,706,935 4,194 533,461

1948 – 1976 1,026 496,611 5,430 2,184,182 4,374 552,162

Pre - 1948 154 48,178 1,805 572,239 14,652 1,283,792

Total 2,746 1,636,852 11,112 5,463,356 23,220 2,369,415

Bridges by road type and age – Australia, 2003 (RoadFact 2005, Austroads)

AUSTRALIAN TRANSPORT INFRASTRUCTURE STATISTICS

800,000 km of public roads (Road Facts 2005)

37,000 bridges (Road Facts 2005)

Bridge maintenance, repair & rehabilitation (MR&R) costs

A$380M in 2003 (National Transport Commission Annual Reports)

A$1,600M in 2012 (i.e. 320% increase in 10 years)

6

News reports on badly deteriorated bridges in NSW 7

8

9

Bridge Asset Management (BAM)?

To determine and implement the best possible strategy that ensures an adequate level of safety at the lowest possible life-cycle cost (Frangopol et al., 2000)

Bridge Management System (BMS)?

A systematic and scientific way to obtain the best strategies for MR&R

10

11Background (1/6)

To determine and implement the best possible strategy that ensures an adequate level of safety at the lowest

possible life-cycle cost.

Bridge Asset Management

(Frangopol et al., 2000)

(Das, 1998; FHWA, 1996)

Bridge

Agency

Increasing LCC

Must be durable

Must be safe

Limited funds

Large bridge network

Make right decisions

Required systematic approaches by using BMS

Keep up-to-date bridge info.

Effective use of funds

Long-term planning

Cost effective operation

Bridge

Asset

Management

Figure 1. Necessity for BMS Implementation

Bridge Management System (BMS)

BMS History in USA

12

1982 1991 20101968

Federal Highway Administration (FHWA) =>

National Bridge Inventory (NBI) (Czepiel, 1995)

Research on computer-

based BMS

Release of commercial BMS

software PONTIS ver.1

PONTIS ver.4.x

13

What does BMS do? Three main tasks!

Collection of

inventory data

Inspection data

Import/export

data

Analytical

Process

Reports on

optimal MR&R

strategies

MR&R work by

bridge authorities

* MR&R = Maintenance, Repair and Rehabilitation

Traditional DETERIORATION analysis

AI-based DETERIORATION analyses

MR&R strategies development

Cost analysis

Optimisation

| Griffith School of Engineering 16

Validation of Back Prediction Methodology

Validation of the BPM

Using two different datasets from Maryland

Department of Transportation (Maryland DoT), USA.

- National Bridge Inventory (NBI)

- Element-level BMS condition ratings.

Historical non-bridge factors were also obtained from

those relevant information authorities.

Two different test methods are established using NBI.

1.0

0.8

0.6

0.4

0.2

0.0

0.9

0.7

0.5

0.3

0.1

Element-level condition

ratings

Predefined

Scale

BPM inputs

CS1

CS5

CS3

CS2

CS4

8

6

4

2

0

9

7

5

3

1

NBI

Scales of bridge samples and BPM inputs

Excellent

condition

Poor

condition

| Griffith School of Engineering 18

Current BMS Issues

Shortcomings related to the use of BMS software:

commercial BMS installed only since 1991;

even early users would have only 6 to 7 inspection records on average;

roughly 60% of BMS analyses/processes rely heavily on bridge condition records/ratings (from new);

Availability: USA – 1991/2009; NSW RTA -1995/2009; QDMR – 2000/2009;

with inadequate condition-rating records, BMS outcomes/forecasts will be unreliable.

Based on few sets of condition ratings, the main difficulty faced by current deterioration modeling techniques

is the lack of usable condition rating datasets.

This in turn leads to an unreliable prediction of future bridge condition ratings.

| Griffith School of Engineering 19

Integrated Deterioration Prediction Method (Schematic)

20

1. Full set of inspection records

2. Element types, material types, traffic volumes

and construction eras etc.

3. Automatically select the appropriate

deterioration model (i.e. time or state-based)

4. Generate a probability density function of

time (using Kaplan & Meier method - stochastic)

5. Generate long-term performance curves(using Elman Neural Network technique - ENN)

1. Available inspection records (via BPM)

2. Categorisation process

3. Selection process

4. Time-based model(i.e. time required for a

change of condition states

(CSs) or a fixed level of

deterioration

5. State-based model(i.e. CS changes over a

given time interval)

Long-term Bridge Performance Prediction (1/2) 21

0

1

2

3

4

5

6

7

8

9

2005 2010 2015 2020 2025 2030 2035 2040

Co

nd

itio

n R

atin

gPrediction Year

ID 1xxx090

Proposed method

Markovian-based procedure

Comparisons of long-term deterioration predictions (see graph for an example – Integrated method’s forecasts more meaningful)

Case Studies

National Bridge Inventory (NBI)

datasets

Total of 40 bridges (464 bridge substructure inspection records)

315 records used as input data

and remaining 149 records used

for validation

Prediction comparisons(2/2)22

Comparisons of average prediction errors - the proposed method is superior to

the standard Markovian-based procedure

0

1

2

3

4

5

6

7

8

9

10

1xxx72

0

2xxx17

0

1xxx35

0

1xxx09

0

1xxx64

0

1xxx61

0

1xxx20

2

1xxx57

0

1xxx93

0

1xxx16

0

100

0xxx

100

4xxx

100

6xxx

101

4xxx

101

6xxx

101

61xx

101

7xxx

102

8xxx

103

8xxx

103

4xxx

103

xxx2

104

xxx9

105

xxx9

107

xxx0

107

4xx0

107

7xx0

550

xxx9

103

xxx1

103

4xx1

104

4xx9

1x0x2

x0

1x4x1

x1

5x2x4

x0

1x7x4

x1

1x4x1

x0

1x9x1

x9

2x2x2

x0

Pre

dic

tio

n e

rro

r (%

)

Bridge ID

The proposed method Standard Markovian-based procedure

23Evaluation of Condition Status

Concrete

material

Steel material

Other material

1. Cracking;

2. Corrosion;

3. Spalling;

4. Carbonation;

5. Surface defect;

6. Alkali-silica reaction (ASR);

7. Delamination.

1. Cracking;

2. Corrosion;

3. Permanent deformation;

4. Loose connection

1. Cracking;

2. Splitting; Spalling; Disintegration

3. Loose mortar and stones.

Image

Processing

Technology

‣ Image Processing Technology - to improve the accuracy, consistency and

reliability of analysing and evaluating structural defects.

0.

1

Crack

Width (mm) 0.

3

0.

6

2C, Barriers 2C, Barriers

CS

1

CS2 CS3 CS4Condition

States

CS 3

CS 3CS 1 CS 2

Image Processing

2

4

‣Example: Crack detection

CONCLUSIONS & ON-GOING RESEARCH

Bridge deterioration is a big problem & bridge safety is a real concern

Reliable and sophisticated BMSs are urgently needed

AI-based deterioration models superior to traditional techniques

SIAMA’s continuing R+D programs

Smart inspection and data collection techniques (speedy)

Automated inspection data generation processes (incorporating pattern

recognition)

Commercialisation to-date

Collaboration with two Chinese bridge authorities for the planned

management of > 18,500 bridges

Three patents pending

25

26

PATENT Pending 1: LOCAL POSITIONING SYSTEM FOR AN UNMANNED AERIAL VEHICLE

PATENT Pending 2: AN UNMANNED GROUND VEHICLE FOR INSPECTING CONFINED

INFRASTRUCTURE

PATENT Pending 3: IMAGE PROCESSING TECHNIQUES FOR IDENTIFYING

INFRASTRUCTURE CONDITION STATUS

27

THANK YOU!