Indian Roads Congress

Special Publication 60

AN APPROACH
DOCUMENT FOR
ASSESSMENT OF
REMAINING LIFE
OF CONCRETE
BRIDGES

NEW DELHI 2002

 Digitized by the Internet Archive
in 2014

https://archive.org/details/govlawircy2002sp60_0
 IRC:SP:60-2002

Indian Roads Congress

Special Publication 60

AN APPROACH
DOCUMENT FOR
ASSESSMENT OF
REMAINING LIFE
OF CONCRETE
BRIDGES

Published by
THE INDIAN ROADS CONGRESS

Copies can be had from

The Secretary, Indian Roads Congress
Jamnagar House, Shahjahan Road,
New Delhi'110011

NEW DELHI 2002

Rs 200/-
price .

(Plus Packing& Postage)

IRC:SP:60-2002

First Published : November, 2002

Reprinted : June, 2009

(Rights of Publication and Translation are Reserved)

Printed at Abhinav Prints, Delhi- 110041

(500 copies)
 IRC:SP:60-2002

CONTENTS
Page

Personnel of Bridges Specifications ... (l)tO

and Standards Committee (iv)

Background 1

1 . Introduction 2

2. Degradation Causing Factors, Deterioration 8

Processes and Damage Modes

3. Deterioration Rates ... 14

4. Methodologies for Life Predictions ... 29

5. General Procedure for Life Assessment ... 49

6. Action Plan ... 53

7. References ... 56
IRC:SP:60-2002

FIGURES
Page
1. Deterioration and maintenance life cycle 6

2. Degradation cycle due to corrosion 6

3. Simplified deterioration models 12

4. Schematic representation of the process of carbonation 4

within a concrete cover

5. Probability distribution function of service life due to 17

carbonation

6. Probability density function of service life due to 17

carbonation

7. Concrete cover vs. time to start carbonation induced 18

corrosion

8. Determination of service life with respect to corrosion of 19

reinforcement

9. Typical plot of polarisation resistance 23

10. Relationship between cover, diffusion coefficient(D), 27

chloride content(c ), and time of initiation
s

11. Deck condition rating vs. time 28

12. Variable amplitude strees history 33

13. 50% probability S-N curve 35

14. Typical Markov chain process 42

15. Increase of failure probability (Illustrative presentation) 47

44
16. Cross-section of a column (degradation both in concrete 61
and steel)

17. Cross section of a beam44 62

18. Reductions in material cross-sections and compressive 63

capacity of a column
19. Reduction in the bending capacity of a beam 44 64

20. Chloride measurement on a bridge deck 66

Appendix : Numerical illustrations from literature

 IRC:SP:60-2002

PERSONNEL OF THE BRIDGES SPECIFICATIONS AND

STANDARDS COMMITTEE
(As on 14.7.2001)

1. S.C Sharma Director General (Road Dev.) & Addl.

(Convenor) Secretary to the Govt, of India, Ministry of
Road Transport & Highways, Transport
Bhawan, New Delhi- 1 10001
2. N.K. Sinha Member (Technical), National Highways
(Co-Convenor) Authority of India, Plot No. G-5/6, Sector,
Dwarka, New Delhi- 1 10045
3. The Chief Engineer (V. Velayutham), Ministry ofRoad Transport
(B) S&R & Highways, Transport Bhawan, New
(Member-Secy) Delhi-1 10001

Members
K.N. Agrawal Chief Engineer NDZ-I, CPWD, Nirman Bhavan,
New Delhi-1 10011
CR. Alimchandani Chairman & Managing Director, STUP
Consultants Ltd., 1004-5, Raheja Chambers, 213,
Nariman Point, Mumbai-400021
D.S. Batra Consulting Engineer, Sir Owen Williams
Innovestment Ltd., Innovestment House, 1072,
Sector-37, Noida-20 1 303

S.S. Chakrabbrty Managing Director, Consulting Engg. Services

(I) Ltd., 57, Nehru Place, New Delhi- 1 1 00 1

C.V.Kand Consultant, E-2/136, Mahavir Nagar,

Bhopal-462016

D.K. Kanhere Chief Engineer, Block No. A-8, Building No. 12,
Haji Ali Officer's Qtrs., Mahalaxmi,
Mumbai-400034

10. KrishanKant Chief General Manager, National Highways

Authority of India, Plot No.G-5/6, Sector-10,
Dwarka, New Delhi- 1 1 0045

ADG (B) being not in position, the meeting was presided by Shri S.C. Sharma, DG (RD) &
Addl. Secretary to the Govt, of India, MORT&H.

(i)
IRC:SP:60-2002

11. Ninan Koshi DG(RD) & Addl. Secy., MOST (Retd.), 56,
Nalanda Apartments, Vikashpuri,
New Delhi- 110018
12. Dr. R. Kapoor Director, Unitech India Ltd., Gurgaon

13. Vijay Kumar Managing Director, UP State Bridge Corporation

Bhavan, 16, Madan
Ltd., Setu Mohan Malviya
Marg, Lucknow-226001

14. N.V. Merani Principal Secy., Maharashtra PWD (Retd.),

A-47/1344, Adarsh Nagar, Worli, Mumbai-400025

15. M.K. Mukherjee 40/182, C.R. Park, New Delhi- 1 10019

16. A.D. Narain DG(RD) & Addl Secy., MOST (Retd.), B-186,
Sector 26, NOIDA-20 1301
17. M.V.B. Rao Area Coordinator Bridge & Instrumentation
Engineering, Central Road Research Institute,

P.O. CRRI, New Delhi- 1 10020

18. Dr. T.N. Subba Rao Chairman, Construma Consultancy (P) Ltd., 2nd
Floor, Pinky Plaza, Mumbai-400052

19. D. Sreerama Murthy Chief Engineer (Retd.) H.No. 8-3-1 158, Gulmarg
Enclave, Flat No. 203, Srinagar Colony,
Hyderabad

20. A. Ramakrishna President (Operations) & Dy. Managing Director,

Larsen & Toubro Ltd., ECC Constn. Group,
Mount Ponnamallee Road, Mannapakkam, P.O.
Box No. 979, Chennai-600089
21. S.A.Reddi Dy. Managing Director,Gammon India Ltd.,
Gammon House, Prabhadevi, Mumbai-400025
22. Ramani Sarmah Secretary to the Govt, of Meghalaya, Public
Works Department, Lower Lachumiere,
Shillong-793001

23. N.C. Saxena Executive Director, Intercontinental Consultants

& Technocrats Pvt. Ltd., A-l 1, Green Park, New
Delhi-1 10016

24. G. Sharan Chief Engineer, Ministry of Road Transport &

Highways, Transport Bhawan,New Delhi-1 10001

(«)
 IRC:SP:60-2002

25. S.R. Tambe Secretary, Maharashtra PWD (Retd.),72, Pranit

J. Palkar Marg, Opp. Podar Hospital, Worli,
Mumbai-400025
26. Dr. M.G. Tamhankar BH-1/44, Kendriya Vihar, Sector- 1 1 ,
Kharghar,
NaviMumbai-410210
27. Mahesh Tandon Managing Director, Tandon Consultants (P)
Ltd., 17, Link Road, Jangpura Extn., New
Delhi- 110014

28. P.B.Vijay DG (Works), CPWD (Retd.), A-39/B, DDA Flats,

Munirka, New Delhi- 1 1 0062
29. The Chief Engineer (S.K. De), M.P. Public Works Department, 'D'
(NH) Wing, 1st Floor, Satpura Bhavan, Bhopal-462004

30. The Principal Secy, to (H.P. Jamdar), R&B Department, Block No. 14,
the Govt, of Gujarat 2nd Floor, New Sachivalaya,
Gandhinagar-382010

31. The Chief EngineeifNH) (S.Rakshit), Public Works Deptt., Writers'

Building, Block 'G' 4th Floor, Kolkata-70000

32. The Chief EngineeifNH) (S.S. Lai), U.P. Public Works Deptt.,
Lucknow-226001

33. The Chief EngineeitNH) Punjab P.W.D., B&R Branch, Patiala- 147001
34. The Chief Engineer(R) (Jai Prakash), Ministry of Road Transport &
S&R, T&T Highways, Transport Bhavan, New Delhi- 1 10001

35. The Engineer-in-Chief K.R. Circle, Bangalore-5 60001

(NH)

36. The Director (S. Saravanavel), Highways Research Station,

P.B. No. 2371, 76, Sardar Patel Road,
Chennai-600025

38. The Director & Head Bureau of Indian Standards, Manak Bhavan, 9,
(Civil Engg.) Bahadurshah Zafar Marg, New Delhi- 1 10002

37. The Dy. Director (B.K. Basu, VSM, SC) Dy. Director General
General (Bridges), Border Roads Directorate, Seema
Sadak Bhawan, Naraina, Delhi Cantt,
New Delhi-1 10010
39. The Director, RDSO (Vijay Nathawat) Director (Bridges & Structure)
Research, Design & Standards Organisation,
Lucknow-226001

(iii)
IRC:SP:60-2002

40. The Addl. Director (Krishan Kumar), Addl. Director General,

CPWD CPWD, Central Design Orgn., Nirman Bhavan,
New Delhi- 10011 1

Ex-Officio Members
41. President, A.B. Pawar
Indian Roads Congress Secretary (Works), Maharashtra P.W.D.
Mantralaya, Mumbai-400032
42. DG(RD) S.C. Sharrna, D.G.(RD) & Addl. Secy., Ministry
of Road Transport & Highways, Transport
Bhavan, New Delhi- 1 1 000

43. Secretary, G. Sharan, Chief Engineer, Ministry of Road

Indian Roads Congress Transport & Highways, Transport Bhawan,
New Delhi-1 10001

Corresponding Members

44. M.K. Agarwal Engineer-in-Chief (Retd.), H.No.40, Sector 16,

Panchkula-134113

45. Dr.V.K.Raina B-13, Sector-14, Noida-201301

46. Shitala Sharan Chief Consultant, Consulting Engg. Services

(I) Ltd. 57, Nehru Place, New Delhi-1 10019
47. S.P.Khedkar Hindustan Constn. Co. Ltd., Hincon House, Lai
Bahadur Shastri Marg, Vikhroli (W),
Mumbai-400083
48. The Technical Director (H. Guha Viswas), Simplex Concrete Piles (I)
Pvt. Ltd., Vaikunt, 2nd Floor, 82, Nehru Place,
New Delhi-1 10019

(iv)
 IRC:SP:60-2002

AN APPROACH DOCUMENT FOR

ASSESSMENT OF REMAINING LIFE OF
CONCRETE BRIDGES
BACKGROUND

The Bridge Maintenance and Rehabilitation Committee

(B-9) in its meeting held on 4.7.97, constituted a Sub-group
consisting of Dr. M.G. Tamhankar (Convenor), S/Shri S.S.
Chakraborty, Ajit Singh, M.V.B. Rao, A.K. Harit and Mahesh
Tandon as members for preparing draft report on Assessment
of Remaining Life of Concrete Bridges. The draft document
prepared by its Convenor, Dr. M.G. Tamhankar, was discussed
by the Sub-group and in the B-9 Committee in its number of
meetings. The final draft was approved by the B-9 Committee
in its meeting held at Mumbai on 20.12.99. The personnel of
B-9 Committee which approved this document is given below:

A.G. Borkar .. Convenor

CE(R&B), NH, Hyderabad .. Co-Convenor
(D. Sree Rama Murthy)
D.K. Kanhere .. Member- Secretary

MEMBERS
Ashok Kumar Basa Dr. T.N.SubbaRao
P.C. Bhasin M.V.B. Rao
S.S. Chakraborty S.A. Reddi
A.K. Harit Dr. N.S. Rengaswamy
S.G. Joglekar Ajit Singh

C.V.Kand Gurdip Singh

P.Y. Manjure Dr.M.G. Tamhankar
N.V. Merani Mahesh Tandon

1
IRC:SP:60-2002

Ex-Officio Members

President, IRC DG(RD)

(K.B. Rajoria) (Prafulla Kumar)

Secretary, IRC
(S.C. Shanna)

Corresponding Members

M.C. Bhide Y.G. Patwardhan

M.P. Gaj apathy Rao S.R. Tambe
N.G.Thatte

The Bridges and Standards Committee

Specifications
during its meeting held on 14th July, 2001 approved the
document for printing as a Special Publication of the IRC
subject to certain modifications in light of the comments made
by the members. The draft was approved by the Executive
Committee in its meeting held on the 16th December, 2001 and
later by the Council of the IRC in its 164th meeting held at

Kochi (Cochin) on the 8th January, 2002.

1. INTRODUCTION
With ever increasing stock of distressed bridges and
dwindling resource position for maintaining the same, the need
of assessment of remaining useful life of bridges in Bridge
Management Systems cannot be over emphasised. It has been
recognised that despite its crucial role the life assessment has
eluded explicit modelling and numerical evaluation of its

'absolute' value due enormous complexities in the degradation

mechanism, and lack of data-base on material degradation and
bridge performance. The problem has got further compounded
due to uncertainties associated with material properties,
construction details, exposure conditions, deterioration rates
and the quality of maintenance.

2
 IRC:SP:60-2002

Therefore, to predict as to how long the given bridge will

last is perhaps beyond the present level of development of the
subject. At best one can try to estimate whether the given
bridge can reach a specific age, with the help of probability
distribution function of service life. This is also possible if the
acceptable degree of failure probability can be pre-determined.
24
It has been emphasised that our design philosophy
comprises a number of essential and interacting elements, such as:

a behavioural model,
criteria defining satisfactory performance,
loads under which these criteria should be satisfied,
relevant characteristic material properties, which should reliably
be achievable in the construction process,

factors or margins to take account of vagaries and variability

in the system.

But the present day predictive models, however, consider

the first three of the above elements reasonably accurately. The
relationship between materials forming the structure and the
overall response of the structure is often so complicated that
this indeed poses a question as to whether the service life

prediction can ever be generalised.

However, given its crucial role in distressed or rehabilitated

bridges, it becomes imperative to make a beginning in this
direction. It is necessary to create awareness to keep track of
performance against the assumed expectations in general and
to look into specific issues, such as:

existing load carrying capacity,

cause of distress, if already noticed,
degree of aggressiveness of the environment,
risk of future damage,
rate of deterioration, and
the effect of repairs conducted from time to time on the rates
of deterioration.

3
IRC:SP:60-2002

There is also an urgent need to document data, wherever

available, on degradation causing factors, deterioration
processes, damage modes, and With these
deterioration rates.
objectives in the background, document aims at presenting
this

the existing status of the subject by synthesising the available

knowledge in the fields of material degradation and life
assessment to initiate further steps and make a beginning towards
introducing the concept of life prediction.

Given the developmental stage of this subject, the

document is prepared as an approach document only, to enable
engineers to appreciate the need of the subject, to understand
present status of the knowledge and its and to
limitations,
introduce road map for future course of actions. Obviously the
subject has not attained the level where it can be enforced as a
part of normal design procedure.

1.1. Durability Concept

Durability is the property expressing the ability of the

structure/component/product to maintain the required performance
level over at least a specified time under the influence of the
degradation factors. The minimum acceptable values for
performance (or maximum acceptable values for degradation) are
called durability limit states. The limit-state is a performance
requirement critical to the service life, which can be set with regard
to either the ultimate limit or the serviceability limit. A new feature
here is the incorporation of time into the design problems. It,

thereby, allows the possibility of treating degradation of materials

as an essential part of structural calculations. Traditionally, the
durability design is based on implicit rules for materials, material
compositions, working conditions, structural dimensions, etc.

(e.g., minimum concrete cover, maximum water/cement ratio,

minimum cement content, crack width limitations, cement type,

coatings on concrete, categorisation of exposure conditions).

4
 lRC:SP:60-2002

1.2. Service Life

The concept of service life can be approached from at

least three different aspects: (i) technical, (ii) functional and
(iii) economic. The service life of a bridge span, component or
its constituting materials is defined as the period of time after
installation during which its essential properties meet or exceed
minimum acceptable values, when routinely maintained. It is

the period during which no excessive expenditure is required in

operation, repair. Each component has an
maintenance or
expected service life. The exact definition of service life is
obscured by the maintenance routines performed during the
service life. Maintenance can influence the length of service
life and hence, the definition of service life should include

"when routinely maintained". Some components, such as,

bearings, expansion joints, wearing coat, require periodic
maintenance or replacement, whereas, the main structural
components may also require periodic inspection and preventive
maintenance but are expected to perform their expected functions
during their service life. Required service life imposed by

general rules, the client or the owner of the structure is called

the target service life. Fig. 1 shows a generalised pattern of
deterioration and maintenance cycles in the life of a bridge.
Considering the corrosion-induced deterioration, life of the
structure can be split into following four periods, Fig. 2.

T, - Unto whim tho uxtmml ugem C0 2,

uhltuiMua) hwtuuua
active
T 2
- time when the effect of CCX/chlorides reaches steel zone
(initiation time)
T 3
- time when first visible sign of damage due to corrosion is

noticed (e.g., crack, staining)

T4 - time when damage reaches serious level affecting structural
safety (e.g., wider cracks, spalling), viz., end of the operational
period
T 5
- time when structure reaches failure stage.

5
IRC:SP:60-2002

YEARS

1,3
Fig. 1. Deterioration and maintenance life cycle

According to the Danish Road Directorate ,

T4 , the
operational period should be a minimum 100 years and T 3, a
minimum of 50 years.

LOAD
CARRYING
CAPACITY
CRITICALLY
AFFECTED
CEFSL)

T2 T3
TIME (YEARS)

Fig. 2. Degradation cycle due to corrosion

1.3. Design Life

Distinction needs to be made between Service Life and

Design Life. The design life is a period considered (possibly
longer than the service life) that will give sufficiently high
probability of the structure achieving the required service life.

The design would depend upon the importance of the

life

structure. It would ensure required safety against falling below

the target service life. The design life would influence the

6
 IRC:SP:60-2002

specifications and detailing of structural components.

Various service/design life periods are prescribed in the

literature according to importance, e.g.,

Studies for a major suspension bridge over the Straits of

Messina have indicated that the conventional service life of
such structures should be 200 years.
In the U.K., according to BS:5400, it is assumed to be 120
years.
For the construction of structures for Oresund Link between
2
Denmark and Sweden , the owners have specified a service
life of 100 years, out of which the first 50 years must be totally

free of maintenance, while minor concrete repairs will be

acceptable after the first 50 years.
Different components can have different design lives. Figures
can vary from bridge to bridge as well as with wide range of
structural systems and material qualities of components. Typical
figures for different components are indicated in the following:

Foundation Same as design life of the bridge

Piers & abutments Same as design life of the bridge with

periodic minor concrete repairs (e.g.,
after every 20 years)

Bearings Varying depending upon type and

quality

Steel main members Same as design life of the bridge with

periodic minor painting (e.g., after

every 6 years) and major painting (e.g.

after every 12 years)

Concrete decking Same as design life of the bridge

Wearing coat Varying depending upon type (e.g.,

15 years)

Expansion joints Varying depending upon type and

quality (e.g., 15 years)

7
IRC:SP:60-2002

Euro-code (CEN-1994) presents the

1 classification for
44
the target service life as shown below :

Class Target Service Example

Life (Years)

1. 1-5 Temporary structures

2. 25 Replaceable structural parts,

_ e.g., bearings

3. 50 Building structures and other

common structures

4. 100 Monumental building structures,

bridges and other civil engineering
structures

In fact, it is not feasible to design a bridge for a very specific

life period. The available experience of traditional materials is

often inappropriate and many modern high technology products

are relatively untried. Hence, the anticipated durability can only
be an estimate, its prediction being subject to many variables. In
general, the designer has in mind a useful life for the bridge
alongwith reasonable level of maintenance from time to time.

2. DEGRADATION CAUSING FACTORS,

DETERIORATION PROCESSES AND
DAMAGE MODES
The process in which the resistance (and thereby
performance) decreases with time is called degradation process.
Materials and components have finite service lives since they
gradually undergo chemical, physical or mechanical changes
resulting in degradation and reduction in their ability to perform
as required. Degradation processes are numerous. The type and
the rates of deterioration processes determine the resistance and
rigidity of the materials, sections and the elements constituting
the bridge structure.

8
 IRC:SP:60-2002

2.1. Degradation Causing Factors

Commonly recognised degradation causing factors can

broadly be categorised into the following:

(i) Water

Liquid (e.g., rain water/stagnating/trapped/splashing zone, etc.)

Solid (e.g., snow, freezing of water)
Vapour (e.g., air humidity, moisture in pores)

(ii) Aggressive air constituents

Carbon dioxide (leading to carbonation)

Chlorides (promoting corrosion)
Sulphates (leading to expansive reaction with cement)
Acids (dissolving cement)
Alkalies (leading to expansive reaction with aggregates)
Ozone
Ammonia (leading to disintegration of concrete)
High voltage transmission lines (stray currents)

(iii) Biological agents

Micro-organism
Fungi

(iv) Temperature

Absolute value
Gradient
Cycle
Solar radiation

(v) Foundation condition

Scour around piers and foundations

Sub-soil strata leading to settlement, tilt, etc.

(vi) Time dependent material characteristics

Creep in concrete
Relaxation in steel

9
IRC:SP:60-2002

(vii) Traffic

Axle loads (leading to abrasion, impact or exceeding design

load limits)
Frequency (leading to fatigue)

(viii) Other non-continuous actions

Earthquake
Cyclone
Collision with vehicle, barge, etc.
Vandalism

Climatic influences on bridge components can be further

sub-divided into macro, meso and micro climates, corresponding
to country, site and specific element of the structure respectively.
Depending upon the nature and duration of the degradation causing
factors and the type of the bridge, it will lead to partial or complete
loss of serviceability or of its mere aesthetic appearance.

2.2. Degradation Processes/Mechanisms

Bridge component fails, if component

the strength of that
is no longer sufficient to resist the actual load effects. The

resistance of the component is affected due to various

degradation processes to which the material of the bridge
component is exposed. The sequence of chemical or physical
changes that lead to detrimental changes in one or more
properties of structural materials or components when exposed
to one or a combination of degradation factors, is called
Degradation mechanism. The degradation processes follow
different patterns as listed below:

(i) Degradation processes progressing linearly with time, e.g.,

corrosion process over years, wear and tear of the deck.

(ii) Degradation processes slowing down with time, e.g.,
carbonation in concrete.
(iii) Exponentially accelerating degradation process, e.g., fatigue.

(iv) Degradation process non-continuous in time, e.g., collision,

earthquake, etc.

10
 IRC:SP:60-2002

(v) Two-stage mechanism of degradation process, e.g., degradation

of protection layer first as in the case of cover on reinforcement,
or as in the case of coating on a steel member.

Fig. 3 depicts the above patterns of degradation. Some of

the commonly observed deterioration processes are the
following:

2.2.1. Fatigue : It is defined as the tendency of the

material to break under repeated cyclic stresses considerably
below the ultimate tensile stress. Cyclic loading induces or
propagates the pre-existing cracks in the bridge. Movement of
frequently occurring heavily loaded commercial vehicles can
cause fatigue damage in road bridges. The fatigue action
comprises three phases: (a) crack nucleation in which invisible
changes occur, (b) crack propagation in which micro-cracks
grow to visible macroscopic dimensions, and (c) instant failure.
2.2.2. Corrosion of steel : Bridges in a coastal belt
have shown premature distress due to corrosion of steel arising
from salinity in the air. Corrosion of steel in concrete girder
can be faster than that in the steel in the deck slab due to
crowding of reinforcement in girders especially at the lap joints
with accompanying honey-combing.

Corrosion is the mechanism, which is in most cases an

important parameter to influence the load-carrying capacity
and the service life of the bridge. It can lead to: (a) loss in steel

integrity, (b) possible loss in mechanical properties of steel,

strength and ductility, (c) spitting and spalling of the concrete
cover and possible loss of effective concrete cross-section,
(d) loss of bond between steel and concrete in case of cracks
running parallel to the steel, and (e) loss of prestress in the case
of corrosion of pre-stressing steel. Pre-stressing steel apart
from uniform corrosion may also suffer from pitting corrosion,
crevice corrosion, stress corrosion and hydrogen embrittlement.

11
IRC:SP:60-2002

• CORROSION IN STEEL

d • io * k.i • WEAR AND TEAR OF DECK

• CARBONATION IN CONCRETE
d » kt *
1
• CHLORIDE INGRESS IN
CONCRETE

• FATIGUE

• ALKALI-AGGREGATE REACTION

• FROST ATTACK

DEGRADATION IN
TWO STAGE • REINFORCEMENT COVERED IN
MECHANISM CONCRETE
_ • COATED STRUCTURAL STEEL

• COLLISION
1

j— ^NDN-CC
NON-CONTINUOUS . EARTHQUAKE

4, 8
Fig. 3. Simplified deterioration models

2.2.3. Disintegration, cracking, delamination of

concrete :These are the commonly observed distresses in
concrete leading to loss of stiffness, composite section, etc.

12
 IRC:SP:60-2002

2.2.4. Combination effect (Synergy) : In actual service,

degradation factors may interact to increase the rate of degradation.
Synergistic actions are in fact difficult to simulate or to account for,
e.g., the effect of corrosion induced defects on the fatigue strength of
steel bridges, or influence of temperature on creep in concrete bridges.

In the presence of aggressive environment causing

corrosion of embedded steel as well as deterioration of concrete,
the fatigue life of the structural component can get reduced
considerably. The phenomenon is termed as corrosion fatigue.
It is reported that fatigue lives of beams exposed to sea-water
are fifty per cent less than those in air.

Initiation and later the development of corrosion can also

get accelerated by alkali-aggregate reaction (AAR) and frost
impact. Furthermore, AAR and frost taking place at the same
time may accelerate each other39 .

2.3. Damage Modes

The degradation causing have different effects on
factors
the structure, leading to different damage modes. A distinction
has to be made between the disintegration at the micro level in
the material and the collapse mode of the bridge from the
structural consideration. Degradation mechanism can be
discovered at an early stage in some situations by measuring -
potential differences, settlement, sagging and strains, density

and thickness of concrete cover, depth of carbonation and

chloride penetration, etc. But the severity of the damages needs
to be examined from their range of influence, e.g., (i) loss of
section, (ii) limited to failure of structural element, (iii) failure
of the entire structure. In this exercise, weightage needs to be
given to the degree of risk involved. Risk is defined as the
product of failure consequence and failure probability. A minor
fault in a critical part of the bridge may signify a greater risk
than a more extended fault in a less critical area.

13
IRC:SP:60-2002

3. DETERIORATION RATES
The correlation between the attacking agent and the
quantum of deterioration with time is the first requirement in
prediction of the remaining life. However, there are considerable
modelling deterioration or in collection of past
difficulties in
data on performance leading to extrapolation of available stray
information. Despite the on going research world wide,
exhaustive analytical models to predict rates of deterioration in
different components of the bridge are not yet available for
their application in practical cases. Models are also not
available to consider interaction of aggressive actions that can
act simultaneously. Deterioration rates suggested by various
researchers are discussed in the following paras.

3.1. Deterioration Rate Due to Carbonation

Carbonation is the reaction of C0 2

in the air with hydrated

minerals in concrete, leading to lowering of the pH value in the

carbonated zone. The protective film on the surface of the steel
lying in the carbonated zone is destroyed and the propagation
of corrosion starts. Time for carbonation front to reach
reinforcement leading to initiation of corrosion is called initiation
time. The depends upon cover thickness, cement, curing,
rate
density, water cement ratio, etc. Carbonation penetration below
the concrete surface progresses as the square-root of time.
Thus, if concrete cover is doubled, corrosion of reinforcement
is delayed by four times. By the same token errors in fixing
reinforcement which may reduce cover can affect the steel
44
earlier than expected. Expressed mathematically ,

M (d) = Kt"2

Where,

\i (d) is the mean of the depth of carbonation at time t (mm)

Kc is the carbonation rate factor (mm/year*) and
t is the time (or age in years)

14
 IRC:SP:60-2002

The initiation time of corrosion t0 can hence be expressed as

t
0 = [*kc y

Where,

d is the concrete cover

The carbonation rate factor Kc depends upon the strength

and composition of the concrete, and on environmental factors,
like, humidity, temperature, etc. The depth of carbonation is

assumed normally distributed and the coefficient of variation

(ratio of standard deviation to mean) is assumed to be constant.
Fig. 4 shows the process of carbonation in the concrete cover.

Fig. 4. Schematic representation of the process of carbonation within

a concrete cover

The part of distribution of carbonation depth that exceeds the

thickness of the concrete cover shows the failure probability.
There are several formulae for modelling of carbonation rate

15
IRC:SP:60-2002
44
Kc as follows

(0 K c
= {CCTV C „ a (f k + m
Where,

Cenv is the environmental coefficient

C ajr
is the coefficient of air content
fk is the characteristic strength of concrete (MPa)
a, b, constants (depending upon the binding agent)

Suggested values for C env C , air

, a and b are

Cenv = 1 (structures sheltered from rain)

= 0.5 (structures exposed to rain)
C ajr
= 1 (not air entrained)
= 0.7 (air entrained)
a = 1 800, b = - 1 .7 (for Portland Cement)
a = 360, b = - 1 .2 (for Portland Cement with 28 per cent

fly ash or 70 per cent blast furnace slag)

(ii) K {
[2D(C 1
-C )]/a}'
2
/2 '

Where,

= the amount of alkaline substance in concrete

= the effective diffusion coefficient for C0 2
at a

given moisture distribution in the pores (m 3 /S)

= concentration difference of C0 2
between air and
the carbonation front (Kg/m 3 )

(iii) K 0A
[64 k ] / C05
Where,

K = oxygen permeability of concrete at 60 per cent RH

c = alkaline content in the cement

As a typical example for the following data, the distribution

Sanction of probability of failure due to carbonation, and probability
density function are presented in Figs. 5 and 6 (refer para 4.4.2).

Data:
Structure sheltered from rain ( C =1.0)

16
 IRC:SP:60-2002

1 1 1 1 1
T I i

10 20 30 40 50 60 70 80 90 100
Time (Years)

Fig. 6. Probability density function of service life due to carbonation

Concrete made with portland cement with no air entrainment

(C air =
v 1)1
i

Characteristic compressive strength of concrete 30 MPa (f =

k

30).
Thickness of cover 25 mm D
( = 25).
Coefficient for variance for carbonation depth (v = 0.6)

.17
IRC:SP:60-2002

Coefficient for variance for concrete cover (v = 0.2)

Constants a=1800, b= -1.7 (applicable for portland cement)

3
Fig. 7 is the result of studies indicating the age to start

corrosion due to carbonation in various grades of concrete. Yet

7
another study observes that carbonation and chlorides can
penetrate to the interior of concrete at a lower rate than would be
given by a square-root of time function. This means that if the

concrete cover is halved, the critical state for incipient danger of

corrosion will be reached in less than a quarter of the time.

Cover (mm)
3,7
Fig. 7. Concrete cover vs. time to start carbonation induced corrosion

3.2. Deterioration Rate Due to Corrosion

Corrosion of reinforcement in reinforced concrete members

can lead to rust strains on the concrete surface, cracks in the cover,
spalling of concrete and the loss of bond between steel and
concrete. The action is complex since apart from general uniform
corrosion, there can be localised corrosion with pit depths as much
as five to ten times deeper than the average corrosion penetration.

18
 IRC:SP:60-2002

Concrete protects steel physically and chemically. It provides

physical barrier against agents that promote corrosion, e.g., water,
oxygen, chlorides. The chemical effect of concrete is attributed
to its alkalinity, causing oxide layer to form on the steel surface.
The deterioration due to corrosion consists of two time segments:
(i) Initiation Time to i.e., the time required for external agents to

de-passivate the protective oxide film on steel bars, and (ii)

Propagation time t. i.e., corrosion action leading to rust formation,
reduction in bar diameter, cracking of concrete, loss of bond, etc.
(Fig. 8). As a conservative approach it is advisable to limit the
service life of the member corresponding to initiation time only.
Governing deterioration rates in such cases would be those due to
actions of causing de-passivation of steel.

LIMIT STATE S

to
Time

Fig. 8. Determination of service life with respect to corrosion of

reinforcement

Propagation time begins when passive film is destroyed

as a result of falling pH due to carbonation, or as a result of
chloride content rising above the threshold close to the
reinforcement. Some guiding expressions are available in the
literature to determine the propagation time. However, it should

19
 IRC:SP:60-2002

be noted that the rate of corrosion of steel in cracks or in the

presence of chlorides is not yet fully understood.

Several techniques are being explored to detect presence

of corrosion activity. Techniques involving visual inspection,
detection of laminations, measurement of corrosion potentials
or the extraction of physical samples for laboratory analysis is

generally both time-consuming and expensive. Rapid and non-

contact type techniques, such as, impulse radar and infrared
thermography may prove useful. But although these techniques
would detect delamination that may be due to corrosion of
reinforcement, they do not detect corrosion directly. Quantitative
assessment of corrosion activity in concrete is still in the
research stage. Following are the formulae for estimating
propagation of corrosion :

(i) f,=43aax
r

(ARmax based on critical threshold value of the structural capacity of

the member)

Where,

t, = the propagation time of corrosion (years)

ARmax = the maximum loss of radius of the steel bar

r = the rate of corrosion (fim/year)

80C '

(
(based on cracking of the concrete cover)

Where, j s ,
••

j^oi /:. •
.1. ) CI ;f« ;
'
'

f ' < .
;

C= the thickness of the concrete cover (mm)

D = the diameter of the reinforcement bar (mm)

20
 IRC:SP:60-2002

Rates of Corrosion

The rate of corrosion in concrete required in the above

expressions depends strongly on the environmental factors, such
as, temperature and humidity, besides chloride contents. Influence
of temperature is considered by the following expressions:

r = CT r
o

Where,

CT = the temperature coefficient and,

r
o
= the rate of corrosion at + 20°C
(values of CT recommended for European
Countries vary between 0.21 and 0.73)

Influence of relative humidity in carbonated and chloride-

44
contaminated concrete is illustrated in the following :

Rate of corrosion in
Relative Humidity Carbonated Chloride
(%) concrete contaminated
(|um/year) concrete
(urn/year)

99 2 34

95 (exposed to rain situation) 50 122

90 (sheltered from rain 12 98

situation)
85 3 78

80 1 61

75 0.1 47

70 0 36

60 0 19

50 0 9

The above values are approximate averages based on the

experimental data reported by Tuutti.

21
IRC:SP:60-2002

Yet another study recommends the following values for

the mean corrosion rates in the calculations: 44

Action RH Corrosion rate

(%) (fim/year)

Carbonation only 90 - 98 5 - 10
<85 <2

Chloride 100 <10

contamination

80-95 50-100
<70 <2

iii) Linear Polarisation Resistance (LPR) technique46

This method developed for on-site study of corrosion

rates of steel in concrete, is based on the experimentally
observed assumption that the polarisation curve for a few mV
around the corrosion potential obeys quasi-linear relationship.
The slope of this curve is the polarisation resistance Rp where

R =
AV
P
—
asAV->0
AI

From this slope the corrosion rate is determined using

relationship

i
corr
= B/R
p

where i
corr
is the corrosion current or corrosion rate (uA/
vr

cm 2
) and B is a constant the value of which lies between 13 and
52 mV. B equal to 40 is considered to be adequate in many cases.
A typical plot of linear polarisation resistance curve is shown in
Fig. 9.

22
 IRC:SP:60-2002

+002 Slope AV/AI

Polarisation
Resistance

>
F CORR S
5
*->
o
Ph

Current I

Fig. 9. Typical plot of polarisation resistance46

Having obtained i
corr
the rate of degradation in
reinforcement bar due to corrosion can be expressed as,

<p=<p
T t ' i
- 0.023 i
corr
«

Where, </> t
is the reinforcement bar diameter at time t

(mm), <j> is the

x
initial diameter of reinforcement bar (mm), i
corr
2
t is the corrosion current or corrosion rate (|iA/cm ), t is the
time after beginning of the propagation period (years), 0.023 is

the conversion factor of jiA/cm into mm/year. From above

expression, it can be deduced that the corrosion current of 1 [iAJ
cm2 measured by
, resistance polarisation can be expected to
10
lead to corrosion penetration into iron of 11.6 jum/year .

The above formula for degradation in reinforcement

bar has been further modified by multiplying i
corr
t by two
more parameters 'W and t
'
'a' to include localised corrosion
effect, since pitting corrosion can affect smaller diameter

23
IRC:SP:60-2002

bars significantly. Here 'W t

f

is the wetness period or effective

time of corrosion and 'a' is the concentration factor due to
localised corrosion (varying from 2 to 10).

In order to determine the impact of this loss of steel section

on the structure, it is necessary to distinguish between the
serviceability and the ultimate limit states. While the loss of
diameter of steel section would affect the ultimate flexural strength,
much before that cracking of concrete cover due to corrosion
would determine the permissible threshold value for corrosion

penetration from serviceability and durability considerations. A

value of about 50 fim is generally considered the threshold value
1

for corrosion penetration to cause cracking of concrete. Based on

this value and the deduction arrived at from the above corrosion
rate (i.e., i
corr
of 1 |iA/cm2 leading to 1 1 .6 fim corrosion penetration
per year), following guidelines are available :

i
coiT
< 0.22 fiA/cm 2 no corrosion damage exected.
0.22 < i
cwr
< 1 .08 fiA/cm 2 corrosion damage possible in the range
of 10-15 years
1.08 < i
corr
< 10.8 (iiA/cm
2
corrosion damage expected in 2-10 years
Kon
> 10.8 nA/cm 2
corrosion damage expected in less than
2 years.

Above guidelines need to be revised with more experience

and as the data becomes available. For example, with respect to
cracking of the cover which results from the expansive character
of the iron oxides, it has been reported that around 10-50
microns of corrosion penetration enough to produce visible
is

(0.05 mm wide) cracks. However, accuracy of prediction would

depend upon the estimation of the value i corr Further, the .

technique may be suitable in RC components but may not help

in detecting damage to prestressed steel. For prestressing steel
in view of their smaller diameters, initiation time alone should
be considered to determine the onset of corrosion.

24 /
 IRC:SP:60-2002

3.3. Deterioration Rate Due to Ingress of Chloride

Because of exposure to the salt water in the coastal

regions, concrete bridges are contaminated with chlorides. The
presence of chlorides (resulting in the loss of alkaline
environment), oxygen and water lead to corrosion of embedded
steel. Chloride ions diffuse through the porous concrete and
reach the corrosion threshold value. The resultant cracking of
concrete cover allows intrusion of chlorides and oxygen at a
much faster rate, thus accelerating the corrosion process. Existing
chloride in concrete is not considered in the following expression.
As a result of chloride penetration a gradient is set up near the
concrete surface. The time at which the critical chloride content
reaches the steel surface and de-passivates it, is regarded as the
initiation time of corrosion.

Estimation of service life in this case is based on FICK's

second law on diffusion of chloride ions through porous materials
18
such as, concrete . To model the chloride transport process in
a porous material, it is assumed that a saturated condition exists
and that, FICK's law applies, even though corrosion does not
occur when the material is continuously saturated because of
lack of oxygen for cathodic reactions. In reality, there is a
combination of exposure conditions. Further, one of the
assumptions in the derivation of FICK's 2 nd law is that the
porous mediumhomogeneous, which is not the case for
is

concrete. It is also assumed that the medium is non-reactive

and non-absorptive, and that does not hold for concrete either.
Chloride ions can be physically absorbed on to the surface of
the pores and chemically combined to the aluminates.
Furthermore, experimental testing in concrete specimens has
shown that the diffusion coefficient varies with time, solution
type and concentration. The time dependence is, in part, a

direct consequence of the continuing cement hydration reactions

IRC:SP:60-2002

and pore blocking. Despite the differences between the

assumptions, FICK's law still provides the only way available
to model chloride diffusion into concrete . It is expressed as

C(X, t) = C0 {1 - erf [(X/2)/(VD e

.t)]}

Where,

C (X, t) = 3
chloride concentration (kg/m ) at depth X
at time t

C = 3
equilibrium chloride concentration (kg/m )
(assumed 1.3 cm below the surface)

erf = error function

D c
= chloride diffusion constant (property of
concrete cmVyear)

t = time in years (for initiation time,

X = cover of concrete)

X = depth atwhich chloride concentration is required

(i.e., depth where steel is located)

Thus, the initiation time is calculated using above

expression on the basis of chloride measurements carried out
on the bridge, and/or from measurements of carbonation. Instead
of modelling chloride ingress gradient by error function, the
44
formula is simplified by using a parabolic function as follows :

C(X,t) = C (

2(3D c t) 2 J

As an illustration, Fig. 10 gives the theoretical service life

as a function of D c
(Diffusion coefficient) and cover for C s

(chloride content as per cent by weight of dry concrete)= 0.2 per

2
cent and 0.1 per cent Diffusion coefficient . D c
is calculated as
5
equal to 5000 * (W/C) mm/year. Limited data is available since

26
 IRC:SP:60-2002

it is on laboratory experiments or from practical data

either based
collected from existing structures of approximately
same age
group (20-30 years old) only. General validity of the model
for
use in long term predictions has, therefore, not been confirmed
in
practice. Sensitivity of parameters C and D can be
c
seen from
Fig. 10 2 If C is not 0.2 per cent as estimated but 0.4
.
s per cent, the
service life is theoretically reduced from 40 to 30 years. If D is
not 40 mm/year but say 80, service life is theoretically reduced
from 40 to 28 years 2 .

10 80 30 %0 & 60 7ft CO VER (mm)

20 40 60 80 li0 l&Q l40 D<mm /YEAR) 2

li SI 02 65 <W 05 06 0."7 Cs%

Fig. 10. Relationship between cover, diffusion coefficient (D), chloride

2
content (C.) and time of initiation

Estimation of service life in the case of chloride attack on

piers of five different bridges is illustrated
39
, (Appendix).

3.4. Scale for Deterioration Rate

Deterioration rates discussed earlier relate to micro-level

degradation inside the body of the structure. Different ratings

27
IRC:SP:60-2002

for damages or for performance of the bridge component have

been proposed in the literature.

RILEM draft recommendations for damage classification

41
of concrete structures indicate ratings for different visible
10
damages. ECB bulletin classifies different levels of
deterioration states arising from damages due to corrosion
based on external signs, such as, rust spots, cracks, cover

spalling and reduction in the reinforcement cross-section.

Under the FHWA inspection procedure condition of bridge

9

elements, such as, deck, superstructure or substructure is evaluated

on a scale of 9 to 0 (9 being the perfect condition and 0 the worst).

Thus, when the bridge element rating changes from one condition
rating to another, it can only change in integer values, such as, 1

The data for condition rating versus time, therefore, do not

2, etc.
yield a smooth curve when plotted (Fig. 11). Unless the

LEGEND :-

..... . - GENERAL PATTERN

RC DECK
PC DECK

1-
, ~| 1 1 1
1 r
20 30 40 50 60 70
JO
TIME (YEARS)
9
Fig. 11. Deck condition rating vs time

28
 IRC:SP:60-2002

maintenance or rehabilitation is performed the element condition

rating is expected to remain unchanged or to drop downward on
any subsequent inspection, which
is normally conducted once in

every two years by trained technicians. Several studies

incorporating the above condition rating scale have been reported
from the U.S. Although the observations from these studies cannot
be directly extended in a generalised manner, a few of the
observations extracted below can be of interest.

A study 9 has indicated that the structural condition of the

deck would deteriorate at the rate of 0.094 per year during the
first ten years and thereafter at 0.025 per year. This implies that
the average condition would never fall below a condition rating
of 6 until after 60 years.

Another study 9 indicated that the deck would deteriorate

slightly faster with age than the superstructure or substructure.
The study estimated the average deterioration of decks to be
about 1 point in 8 years and that of both superstructure and
substructure to be about 1 point in 10 years.
22
In yet another study^ ,
bridge superstructure deterioration
was found to be a convex function, with superstructure condition
ratings deteriorating more slowly as the bridge ages. Fig. 1

indicates the typical deterioration curves for bridge decks of

RCC and PSC 9 .

4. METHODOLOGIES FOR LIFE PREDICTIONS

In the case of a bridge under investigation, a clear
understanding of the agencies which can cause decay and the
manner and extent to which they will have an effect, singly or in
combination becomes a pre-requisite. Obviously, fewer the 'ageing
causes' (e.g., only carbon dioxide or fatigue), easier becomes the
handling of life estimation. Synergistic effects are complicated

29
IRC:SP:60-2002

and unpredictable. The situation becomes worse if accidental

actions are also to be considered along with environmental
aggressivity Further for prediction of remaining
. life, in the absence
of data on construction defects, it may become necessary to assume
that the bridge has been designed and constructed as per the known
provisions in codes and specifications.

To determine service life, it is necessary to model bridge

deterioration process. A number of methods have been proposed,
although most of them are still in the developmental stage.
They are summarised in the following:

4.1. Estimates Based on Experience

This method relies primarily on engineering judgement

and past experience of the investigator.

4.2. A Comparative Approach

For estimating life of specific material, deductions are

made from the performance of similar quality materials in
similar exposure conditions. One approach is, therefore, to

record condition states of a particular bridge element from

similar bridges of different age groups, so that a graph of
degradation pattern against time can be plotted. This would
help in estimating rate of degradation and thereby to predict the
remaining life of the element in question.

The following are some of the potential determinants of

22
bridge superstructure deterioration .

AGE
• Average daily total volume of traffic (ADT)
Average daily total volume of truck traffic (ADTT)
• Bridge structural material
• Structural type
• Span length

30
 IRC:SP:60-2002

• Maintenance level
• Environment

This approach, therefore, assumes that above data in addition

to the periodic condition rating assigned by the evaluator to the
bridge components, such as, deck, superstructure, sub-structure,
etc. is available. Regression analysis is used in such cases to
estimate parameters, which describe a functional relationship
between empirically measured sets of dependent and independent
variables. Here, condition rating becomes the dependent variable
and the factors leading to degradation, such as, age of concrete,
live load intensity, etc., are the independent variables.

For example, one such regression based equation obtained

in one study 12 where AGE and ADT were considered the likely
primary determinants of superstructure deterioration, is:

Superstructure = 9.0 - 0.674 (log AGE) - 0.005 (log ADT)

condition rating (for steel)
9.0 - 0.444 (log AGE) - 0.024 (log ADT)
(for prestressed concrete)

Where, AGE is the age of the bridge in years, and ADT

is the average daily traffic. (The ADT is calculated as per item
29 of the National Bridge Inventory of the Federal Highway
22
Administration ). With the help of such expressions, the age
corresponding to the lower end of the rating scale would
indicate the threshold of the service life.

Problems in comparative approach

(i) One has to be cautious in utilising data from different bridges.

Sometimes the failures/degradations are caused due to incorrect
design, poor construction work or faulty maintenance. The field
data from such sites may generate important information on the
quality of design, construction and maintenance but does not
necessarily give the required feedback of information on
performance of materials.

31
IRC:SP:60-2002

(ii) In regression analysis finer the homogeneity of the bridge

group, better could be the accuracy of the regression equation.

(iii) Accuracy of the regression equations hinges upon accuracy of

by the investigator
the initial data, in particular, the rating assigned
after the inspection. This in turn would depend upon the NDT,
instrumentation and judgement of the investigator.

(iv) The problem arises with many of the materials used today,
such as, admixtures, protective coatings which have a relatively
short performance history.

Reliable data can, therefore, be generated from field

performance only if the data stems from well planned systematic
inspections of the state of thoroughly characterised existing
bridges in thoroughly characterised environments.

4.3. Accelerated Testing

Simulated testing, properly correlated with service

conditions can enable life predictions with reasonable confidence.
Accuracy in prediction would logically depend on how accurately
the environmental agencies that produce changes in the materials
are defined and simulated in the tests. It is mainly because of lack
of knowledge of the quantitative levels of the agencies likely to
cause deterioration that researchers are tempted to test samples
exposed in the field. This, however, being an extremely slow
process, one resorts to an accelerated version in the laboratory.
Here an assumption is made that the number of cycles in the
accelerated ageing test have some kind of relationship to the life

time of that material in actual condition. Before designing the

accelerated laboratory test, it is, therefore, necessary to understand
degradation mechanism so that in the test only those factors of
degradation are incorporated which are most influential.

Fatigue testing of structural steel of railway bridges in the

laboratories is a typical example of accelerated testing 11 . The

32
 IRC:SP:60-2002

prediction of fatigue life in these bridges is based either on (a)

cumulative damage assessment based on S-N curves or (b)
cumulative crack increments based on Fracture Mechanics. Both
approaches require assessment of cumulative fatigue damage
because fatigue behaviour under constant amplitude loading can
be markedly different than that under variable amplitude loading.

The prediction of fatigue life through S-N curves comprises

following activities:

In-situ strain measurements at critical locations

Theoretical corroborative analysis

Live load surveys
Dynamic signal analysis
Preparation of stress histograms
Fatigue tests in the laboratory
Plotting of S-N curves
Estimation of rate of fatigue damage
Assessment of remaining life

4.3.1. Stress-history : This involves instrumentation

of the bridge and measurement of strain-time history at critical

sections, Fig. 12. Corroboration of stress intensities through

analytical modelling is desirable. Traffic surveys and old records
help to estimate the number of cycles of different stress ranges

to which the bridge components are already subjected. Possibility

of future increase in the traffic intensity also needs to be
considered.

Time

Fig. 12. Variable amplitude stress - history

33
IRC:SP:60-2002

4.3.2. Stress-histograms : Stress-history traces from the

field are analysed to construct stress-histograms showing the
number of cycles by using one of the
for different stress ranges
counting methods, such as, Rain-flow counting, or any other
suitable method The choice depends upon the type of signal to
.

be analysed. In these methods, time element is not taken into

account. FFT random dynamic signal on
analysers process a
automatic basis. Similarly by making use of analogue to digital
converters and dedicated software, dynamic signals are analysed
to obtain stress ranges and their corresponding number of
occurrences.

4.3.3. Fatigue testing in laboratory : Fatigue tests are

conducted on representative samples (about 6 to 8) from the bridge
for each stress range. Programmable loading facilitates better

simulation of the actual load. Tests are continued till the failure of
the specimen or upto two million cycles. Few specimens are also
tested to assess the physical and chemical properties of the
material, so that fatigue test data from similar materials can also
be taken into consideration. If the number of specimens tested is
small, S-N curve is plotted through test points taking geometric
mean of all the test results for each stress-range and assuming
that the curve follows a straight line between N = 10 and N = 10
4 6

cycles. If median value of test results for each stress-range is

plotted against the number of cycles to failure, then this
relationship would provide 50 per cent probability of fatigue failure
for the member, Fig. 1 3 This means that at particular stress range
.

value SI, 50 per cent of the samples will have failed before
reaching the cycles N.. For more exact determination of the P-S-
N curves where P is the probability of the specimen's survival, a
large number of specimens need to be tested.

4.3.4. Estimation of damage accumulation and

remaining life: Although, most fatigue tests are conducted at

34
 TRC:SP:60-2002

LOG CYCLES
"
Fig. 13. 50 per cent probability S-N curve

constant amplitude of cyclic stress, in reality the bridge

components receive a load spectrum, i.e., the load and cyclic
stress vary in some fashion under service conditions. To consider
this feature, it is assumed in Miner's hypothesis that the damage
by fatigue action accumulates before failure.

As per Palmgren-Miner's hypothesis, if N. cycles of constant

amplitude stress cause failure, then n. cycles of the same stress
range use up a fraction of (a / N.) of the life. Failure would occur
when the sum of used life fractions D f
reaches unity, i.e.,

D =2
r
(n. / N.) = 1

Numerous variable amplitude loading tests have shown that

sum of the average cycles at failure, called the damage sum at failure
D may deviate considerably from unity.
f ,
The main reason for the
deviation of the damage sum (D f ) from unity is that fatigue damage
is assumed to be a linear function of 2 n./N. =1, irrespective of the

35
IRC:SP:60-2002

loading sequence and stress level. But non-linear damage

accumulation, effect of residual stresses and some other interaction
effects can result in conservative and non-conservative deviations
from Miner's fatigue criterion and in different values of the damage
sum. Therefore, researchers recommend that 50 per cent probability
of fatigue failure curves (S-N curves) and Palmgren-Miner's
summation of 0.3 in place of 1

Miner's damage accumulation index (D.) caused per day

or per vehicle is computed from the equation, D. = 2 (n./ N),

where n. is the number of stress cycles actually applied and N.

are the number of stress cycles to cause failure. Here the
number of cycles n. for each stress range level is obtained from
the stress-histograms of the component and the corresponding
values of N. are obtained from the S-N curve. The remaining
fatigue life is computed by dividing-D (Miner's damage index)
with fatigue damage caused per day (D.).

D is taken equal to one at failure, if the S-N curve for 1

per cent probability of fatigue failure is used for predicting

partial damage per day. Test results have, however, shown that
the sum of cycle ratios (n/N) defers widely from the value of
unity. In case S-N curves with 50 per cent probability of fatigue
failure are used to compute damage per day. Damage Index D
is taken as 0.3 at failure. Fatigue behaviour forms the distinct
branch of research with a large specialised literature

Problems in accelerated testing

The difficulties experienced in utilising the accelerated

tests are the following:

(i) Along with the accelerated test, it is necessary to check how

well the results of this accelerated ageing test compare with
those from an in-service exposure. But this comparison is

seldom definitive, more often the correlation is only marginal.

36
 IRC:SP:60-2002

(ii) Most of the test methods that have been developed for
generating service life data focus upon climatic agents as the
factors causing degradation. These climatic agents are difficult
to quantify and to incorporate meaningfully into accelerated
tests.

(iii) Short term accelerated ageing tests are usually designed to

evaluate the effect of a small number of degradation factors.
Although such test results may be useful for ranking of
materials, they are only of limited value for predicting service
life, unless the degradation factors studied are those that are
responsible for all of the in-service degradation.

(iv) It takes a long time to obtain results from field exposure tests
unless the property changes leading to upgradation are detectable
at eerly stages in the exposure.

(v) Ex posure conditions cannot be controlled and the intensities of

weathering factors are seldom measured, particularly at the

micro-environmental level. It is difficult, therefore, from field

exposure tests to identify mechanisms of degradation and to

isolate the effects of various degradation factors.

(vi) Accelerated testing is satisfactory provided the artificial

environment does not induce forms of degradations which do

not occur in the service environment.

A typical case of service life prediction

17
of concrete
coating used to restrict ingress of chlorides based on laboratory
tests is illustrated in the Appendix.

4.4. Mathematical and Simulation Modelling

Models used to consider effect of degradation can be

deterministic or stochastic. Deterministic models do not consider
scatter of concerned values. The model yields only one value
of degradation, performance or service life that is often the
mean value. In some cases, however, deterministic models are
formulated to give a specific fractile value instead of the mean.
In many cases, however, the deterministic models are insufficient

37
IRC:SP:60-2002

to evaluate the risk of not reaching a definite figure of service

life. In structural designs, therefore, stochastic approach is

considered essential as the scatter due to degradation can be

wide and the degree of risk may be high. In stochastic modelling
for assessing remaining life, certain minimum reliability or

failure probability is assumed. In stochastic design, deterministic

design models are normally used for evaluating the mean
degradation. To evaluate the standard deviation, a constant
coefficient of variation is given:

Thus,

o= v.u

Where,
o = the standard deviation of degradation
\i= the mean of degradation, and
v = the coefficient of variation

Standard deviation can also be obtained analytically as

follows:

Where,

a (t )
L
= the standard deviation of service life distribution (t )
L
a (x.) = the standard deviation of parameter x.

X. = one of the n parameters in the service life model (t )

L

An example to calculate service life is illustrated in the

Appendix.

In the exercise for life prediction, one not only considers

the target service life but also assumes the value of maximum
allowable probability of not reaching the target service life. It

is called the probability of failure. Hence, the probability of

failure can be defined as the probability of exceeding or falling

38
 IRC:SP:60-2002

below a certain limit state, which may be an ultimate limit state

or a serviceability limit state.

4.4.1. Discrete time markov chain process : Markov

chain {X n n n > 0,
: 1, 2 ...} is a discrete time stochastic
process, which is used to model the deterioration process as a
decay of condition ratings over time. A stochastic process is an
indexed set of variables which evolve randomly over time.
Stochastic methods rely on the analysis of performance data
without consideration of the mechanisms involved. Here X n
is

the bridge condition-state at the n* time point.

The method simulates a natural deterioration process starting

with perfect condition and proceeding with gradual and random
degradation. However, unlike other probability methods, no
parameters or assumptions for the type of distribution are needed.
The state space S in the present case consists of condition ratings
of the bridge components at any time described by a number from
a pre-selected discrete numerical scale (say 1 to 5 with 1 and 5
being the initial and the worst states respectively).

In this process, the probability that an element transitions

from condition state i to another condition state j does not depend
how the element arrived at the i th state. Probability of transition
from state i to state j at time point n is denoted by P. (n)

Expressed analytically P.. (n) = P{Xn+1 = j |

X n
= i}. These
transition probabilities are given in a transition probability
matrix [P] which is expressed as

Pm P.2 Pi M - 1 Pi M

P 21 ?22 P2.A/ - 1 P2.JM

[P] Pm P 32 P3.W - 1 P3.M

Pm 2

39
IRC:SP:60-2002

Where,

p.. > 0 for all i, j e S,

^jes
P lJ for all i € S and S = { 1, 2, M }

M being the total number of degradation condition stages.

Matrix [P] has rows and columns equal to assumed discrete
condition, states (M).

Initially, it is necessary to assign p.. values, e.g., 0. 1 , 0.2, 0.7

depending upon the probability of the element transitioning from
its initial state, say 1 to 2, 3 or will continue to remain in 1 itself.

Existing bridge deterioration models used in the Markov

chain theory invariably assume that

(i) the condition of structure cannot be improved during the

process, and
(ii) the condition can either remain same or shift to the next state
within one transition of 1 year step.

With the first assumption all transition probabilities below

p 2l condition
the diagonal probabilities are zero, since, e.g., in
state 2 cannot transit backwards to condition state 1, i.e., from

existing state to improved state. Due to the second assumption,

all probabilities above those next to diagonal ones are zero. As
the structure must remain at the same state or drop to the next
one only within one year step and cannot jump to any other
lower state, sum of probabilities of remaining (diagonal
elements) and dropping to the next state (elements next to the
diagonal ones) must be 1. Thus, the probabilities of dropping to
the next state can be calculated by subtracting the diagonal
probability values from 1. Consequently, only the diagonal
probabilities of the matrices are unknown parameters.

40
 IRC:SP:60-2002

Thus, the transition matrix [P] can be expressed as

p, 1-p, . 0 0
0 0 0
[p]=
0 0 Pm-, 1-Pm
0 0 0 0

Where, p., i = 1 to M-l represents the probability of

remaining in the i
th
state in the next transition. As M is the
worst state in a state space of 1, 2, M and cannot further
deteriorate, PMbe equal to one. p
will have to to p have
numerical values between 0 and 1.

The expected condition of the bridge at a future time n or

conversely, the expected time to reach any specific future state,
can be calculated by using the following relationship:

|jr<
n
>| = |
n < 0)
|
: [P]
n

Where, 7t(n) is the state probability vector at any time n

consisting of the probability mass function (pmf) of X or the
n
degradation index distribution vector for n th year, n (0) is the
n th
initial state probability vector and [P] is the n power of the
transition probability matrix [P]. Vector matrix [tz] has elements
equal to the assumed condition states (M). Thus, changes after
n years can be predicted by multiplying the initial degradation
index distribution 7t
io)
by the transition matrix n times. The
process is depicted in Fig. 14. If we assume structure initially
(o)
in best condition then 7t would be [1 0 0 ...0] meaning that
structure is at degradation index indicating the best stage.

The Markov chain theory assumes that transition

probabilities depend upon the current state irrespective of age
of the bridge, thus only one transition matrix [P] is used for the
whole life span. As a modification, some BMS softwares use

41
IRC:SP:60-2002

El
P.,
1 1
Q to lo lol ASSUMED CONDITION STATES 5
YEAR 0 0 P22 P23
jrth) = jr«>).[P]h
0 0 P33
0 P45

0 P55

Ml 13

YEAR (I)

,(n-l)

I I

YEAR
s 1

(n-1)
i
DO

53
YEAR
Fig. 14. Typical markov chain process 44

different transition matrices for each age group, which means

the stochastic nature of the deterioration process would depend
upon both the current states as well as the ages of the bridges.

4.4.1.1. Estimation of transition probabilities : Two

methods 'Frequency approach' and 'Regression approach'
-

have been suggested to estimate transition probabilities.

In the Frequency approach, P.. is calculated as

P«=^ i,j = l, 2, 3..., M

Where, n.. is the number of bridges originally in state i
which have moved to state j in one step, and n., is the total
number of bridges in state i before the transition. This approach

42
 IRC:SP:60-2002

would require at least two sets of inspection data pertaining to

two different points in time.

In Regression approach, only one set of bridge data is needed.

A regression function is first obtained by regressing condition
ratings on ages. Transition probabilities are then estimated by fitting
the regression function with the transition matrix.

Here degradation index distribution for each year jr

(t)
is
obtained by multiplying the degradation index distribution of
the previous year by the transition matrix [P].
(0 (t - n
••
k l
= k l*[P]

The mean of the degradation index distribution at each

year E(t, P) is obtained by multiplying the scale index vector
R= |
0,1,2,3 .. M by the degradation index distribution vector
|

(Vector multiplication)

.•• E(t, P) = |*«| x |R|

This mean of the degradation

compared with the E is
available reference degradation curve for that particular year. The
probabilities p in the transition matrix are selected by minimising
the sum of yearly deviations between the reference degradation
curve and the Markov estimation for the degradation curve, i.e.,
N
min SUMD = £|s(t)-E(t,P)|
t=i

Where,

SUMD = the sum of deviations at each year

N = the number of years within service life
P = the transition matrix with unknown probability
elements p.
s(t) = the value of the degradation index from the reference
degradation curve at year t

E (t, P) = the mean of the degradation index distribution

calculated by the Markov chain method at year t.

43
IRC:SP:60-2002

It has been observed that the tractability of the Markov

chain mean curve is best with respect to square root pattern of
reference degradation curve.

4.4.1.2. Problems in application : In the absence of any

more accurate method, Discrete Time Markov Chain
other
(DTMC) has been gaining entry into the Bridge Management
System software packages. A network optimisation system for
bridge improvement and maintenance PONTIS of FHWA
assumes transition probabilities to depend upon the current
states and not on the ages of the bridges and hence uses only
19
one transition matrix for the whole life sjpan BRIDGIT is yet .

another software developed in the U.S. There are, however, .

19
several issues connected with this method which need to be
examined by applying the technique to practical problems:

(i) Suitability of Markov Chain theory in bridge deterioration

modelling

This can be confirmed only from the accuracy achieved in

prediction of remaining life. Presently, structural reliability
based methods are the only alternative option available.
Confirmation from the actual performance is a slow process,
till such time it is necessary to keep applying the DTMC on a
variety of bridges with whatever initial data is available.

(h) Suitability of condition rating as the bridge performance index

Although, it is relatively simpler it has, however, been observed

that condition rating in numeric scale is not adequate as
performance index. Condition rating does not reflect the
structural integrity of the bridge. It has been suggested that
bridge deterioration modelling should include load rating.

(iii) Method of estimating the transition probabilities p

u

This difficulty is expected in the beginning as no field data are

available. However, based on instrumentation, non-destructive
testing and visual inspection an experienced bridge engineer

44
 IRC:SP:60-2002

can assign number representing condition state after each

inspection. With the availability of condition states from the
past inspections, the transition probability matrix can be further
improved.

Readers are advised to refer the modern text books on

38,42,44
probabilistic methods for detailed treatment of the method .

4.4.2. Reliability based methods : Decisions to continue

service and/or to perform maintenance need to be supported by
quantitative evidence that the strength is sufficient to withstand
future extreme events within the proposed service period with
an acceptable level of reliability. Structural reliability methods
provide the basis for this evaluation. Statistical simulation is

used in stochastic methods to estimate the likelihood of a

bridge component being in a certain state of condition at some
point of time in future. In contrast to stochastic methods,
reliability based methods do require that the mechanism of
degradation is understood. As the service life is often represented
by a failure distribution, and since the variable conditions exist
in both the loading and resistance characteristics of any bridge,
it is claimed that reliability theory can be an effecti ve modelling
tool. It involves a systematic probabilistic procedure for
quantitative prediction of service lives of materials and
components. Main features of this approach are given below:

Structural loads, engineering material properties and

strength degradation random function of time,
mechanisms are
and, therefore, the performance of the bridge is also of random
variable nature depending upon material characteristics,
workmanship, maintenance standard, etc. The margin of safety
M(t) at any time t can be expressed as:

M (t) =R (t) - S (t), in R (t)

and S (t) are resistance
which
and structural actions respectively. The approach is based on

45
IRC:SP:60-2002

the criterion that the probability of the resistance of the structure

being smaller than the load within the target service life (t ) is
g
smaller than a certain maximum acceptable failure probability

Mathematically expressed,

P{failure} tg = P{R-S<0} tg
< P fmax

Where, P { failure } = the probability of failure of the structure

within tg. The problem can be solved if the distribution of the
load and the resistance are known and the P.fmax is defined.

If R (r) and S(r) are instantaneous physical values of

resistance and the load at the moment r, the failure probability
Pf in a life time t can be expressed as:

P f (t) = P{R(r) < S (r)} for all r < t

As R and S are stochastic quantities with time-dependent

or constant density distributions, P^t) can be expressed as:

P{R(t)< S(t)}

It will be seen from Fig. 15 that the failure probability

increases with time. At t = 0, the density distributions of load
and resistance are far apart and the failure probability is small.
With time, the distributions approach each other, forming an
overlapping area of increasing size. This overlapping area
illustrates the failure probability.

Loads (S) and resistance (R) are generally assumed as

normally distributed. Considering continuous distributions, the
failure probability Pf at certain moment of time x can be
determined using the following convolution integral:
00

Pf(t)= jFR (s)f s (s)ds

46
 IRC:SP:60-2002

Where,

F R (s) = the distribution function of R

f (s) = the probability density function of S, and,
s

s = the common quantity or measure of R and S

TIME
44
Fig. 15. The increase of failure probability (illustrative presentation )

In this case, the failure probability can be determined

using the test index 6

Where,

/AR,t]-rtS,t]

Here and a denote the mean and standard deviation

\x

respectively. In structural design, the index B is referred to as

safety index or reliability index. Very often R or S is constant
in which case p

~ ' ^ constant)
a[s t]

47
IRC:SP:60-2002

The failure probabilities corresponding to p are available

as tables. To obtain the distribution of service, the failure
probabilities have to be solved with several values of t (e.g.,

10, 20, etc. years).

As an example the case of carbonation of cc icrete is

44
considered . Failure is assumed to occur when carbonation
depth exceeds the depth of reinforcement. Here S is the time
related to carbonation process and R the constant concrete
cover is also considered as stochastic quantity. For the numerical
data given in section 3.1, the test index P can be expressed as

I
7 2
25-1800(30+8)"' 1
p
_
\2

0.6xi800(30+8r't +(0.2x25)'

The corresponding probability distribution function and

probability density function are given in Figs. 5 and 6.

For service life prediction and reliability assessment, one

is more interested in the satisfactory performance over some
period of time, say (0,t). The probability that the structure
would survive during interval (0,t) is defined by reliability

function L (0,t), in other words, it is the probability that the

time to reach failure exceeds the time t, i.e.,

L(t) = P[T>t]

where,

t > 0

Where, T is the non-negative random variable representing the

time to failure. For detailed understanding of the procedure,
9 ' 3M4,45
attention is drawn to specialised literature >

48
 IRC:SP:60-2002

Fracture mechanics approach": Techniques based

4.4.3.

on fracture mechanics have also been used in modelling creep,

fatigue or fracture degradation of metals. As an alternative to
cumulative damage assessment based on S-N curves, fracture
mechanics approach based on cumulative crack increments has
also been attempted. Flaw size is an additional variable in this
approach, and fracture toughness replaces strength as the relevant
material property. This approach appears to be more useful in
fatigue related studies, since the measure of damage, i.e., the
crack length can be physically assessed whereas S-N curve
approach does not allow for direct measurement of damage and
damage increments.

5. GENERAL PROCEDURE FOR LIFE ASSESSMENT

Procedures are available in the literature to predict the
2
service life in the case of building materials and components' .

In fact there cannot be one general procedure for all types of

bridges and environments. Considering the history and the
environmental conditions of the bridge in question, the procedure
will require modification. The deterioration and actual damage
levels differ from structure to structure, making it impossible to
lay down a strict framework of the procedure.
Before predicting the remaining life, it is essential to fix
performance criteria for the bridges and in the case of reliability
based approach, the acceptable level of probability of failure.

Factors that would affect the safety and serviceability of the

bridge, inspite of periodic routine maintenance, are required to
estimate the remaining life. A concrete bridge component
reaches the end of its functional service life (Fig. I) when the
level of physical damage warrants not just repair but
15
rehabilitationof the component There are several examples
.

of wear and tear affecting the riding quality, or of malfunctioning

of bearings, expansion joints, etc. These are the results of lack

49
IRC:SP:60-2002

of maintenance and cannot be the governing considerations to

determine the remaining life.

5.1. Broad Steps

Based on the various aspects discussed in the earlier

sections, a comprehensive procedure to assess the remaining
life of an existing bridge is summarised below. All the listed

activities may not be necessary in all the situations. However,

for the sake of comprehensiveness, they are listed together.

Step 1: Defining of minimum acceptable performance

limits, e.g., span/deflection ratio, deflection recovery under
loads, sag, crack widths, riding quality, vibration levels,
acceptable probability of failure, etc.

Step 2: Characterisation of existing status of the bridge: Status

of constituting materials, component, etc. based on condition survey,
i.e., visual inspection for cracks, chloride content, carbonation level,
porosity, corrosion mapping, hammer tapping, cover scan, strength
related properties through hammer, UPV and core tests, petrography,
radiography, endoscopy, etc., if situation demands, condition state

rating (component wise), installation of strain, rotation, deflection

and temperature sensors (for periodic evaluation) followed by static

load test and dynamic signature tests.

Step 3: Analysis of the environment (Macro, meso and

micro level)

Chloride ion contents from concrete samples

Quality of water below the bridge
Daily and seasonal temperature changes at the site, within the
box (in the case of closed sections), and within the concrete
Relative humidity level
Air pollution data
Live load induced stress histograms
Water discharge, scour level
Drainage arrangement at the deck level

50
 IRC:SP:60-2002

Step 4: Identification of critical degradation causing agents

and their rates, e.g., Carbonation induced corrosion, Chloride
ingress-induced corrosion, Live load induced fatigue, wear-
tear, abrasion, scour.

To determine deterioration rate ideally one would require

past periodic data on the performance and access to the accelerated
testing facility. But this would rarely be available. In the absence
of the same, the identification of potential degradation agents and
their rates will largely depend upon investigator's engineering
judgement and past experience. While selecting the degradation
rate, it should be kept into consideration that some prestressed
concrete bridges have suffered premature deterioration 52 53 Guess .

work will have to be made based on the data available from the
in-situ tests, quantification of deterioration vs. time rates based
on field data from similar cases, empirical formulae, e.g., on
corrosion, carbonation, chloride diffusion discussed in section
3.0, results from accelerated tests, etc.

Maintenance of bridges should be need based addressing

the causes of damage.

Influence of repairs or the level of maintenance measures

to be undertaken in future may also have bearing on the rate of
deterioration. However, for conservative estimate, this may
either be ignored or suitable allowance be given.

Step 5: Identification of structural failure modes : Scanning

of critical sections/zones based on Design calculations,
Construction data, Condition survey, Observation during load
test, Maintenance record, etc.

Conversion of deterioration processes into structural failure

modes, and estimation of critical structural parameters, such as,
Loss of prestress, Loss of bond and area of steel, Loss of

51
IRC:SP:60-2002

composite action, Change in stiffness- E value, Excessive

geometrical changes, e.g., settlement, sag, etc.

Step 6: Prediction of remaining life based on most critical

deterioration rate and corresponding structural failure mode.

No unique procedure is feasible for all types of situations.

Available approaches include,

• Estimation based on engineering judgement

• Regression function to model deterioration based on
performance of bridges having similar materials in similar
exposure conditions.
• Probabilistic approaches, such as, Markov chain process,
structural reliability approach.

As the entire procedure becomes qualitative based on

several assumptions, it is desirable to supplement the prediction
with more than one approaches.

5.2. Limitations

It is to be appreciated that the calculations of remaining

life involves many uncertainties, e.g.,

range and accuracy of in-situ investigation procedures

estimated deterioration rates
selection of critical damage mode which would become the
cause of failure in future
assumption of environmental conditions and the level of repair/
maintenance during the remaining life, since the long term
performance of the bridges would depend upon their periodic
maintenance.

In all above considerations, sound engineering judgement

and past experience of the investigator plays a crucial role. The
service life determination helps to forecast the time frame for
possible repair or rehabilitation in future. Therefore, from
practical point of view, even a broad range of time period in
years would be useful.

52
 IRC:SP:60-2002

The prediction of residual

service life, where long periods
(e.g., 50 100 years) are involved or where no visible signs of
-

distress are apparent, can at times be difficult. But in such cases,

the situation also allows more inspections to apply mid-course
corrections. From the present status of the knowledge and
technology, it appears possible to carry out a qualified service life

prediction. But in damaged structures, there will always be a need

for more intensive inspections during the predicted period, no
matter how advanced the service life modelling is.

6. ACTION PLAN
The status of life assessment of bridges world wide is at

such a stage that while its importance is being recognised,

scientific data on material degradation and bridge performance
is lacking. As a consequence, the available analytical tools for
life prediction have remained untested. Therefore, in order to
create awareness to monitor bridge perfonnance during its

service life and to develop data bank several steps need to be

initiated in Highway Departments and Research Institutions.

It is necessary to characterise degradation agents,

understand damage modes, develop accelerated test procedures
and generate data, document field performances, report bridges
in distress to the concerned authorities, test the available and
newly emerging mathematical models, and above all an
institutionalised arrangement to overview and co-ordinate these
tasks on a long term basis. The ultimate objective is to introduce
achievements of material research of concrete structures into
the art of structural design involving 'time' dimension. Specific
tasks requiring actions are the following:

(a) To develop effective mechanism for reporting data on actual

in-service performance of bridges including lessons from cases
where failures have clearly occurred.

53
IRC:SP:60-2002

(b) Identification and grouping of bridges based on bridge geometry

and exposure to aggressive environment (e.g., aggressive
climate, heavy traffic, scour, etc.)

(c) Guidelines for minimum expected performance levels.

Structurally acceptable limits, such as, on deflection, crack

widths, riding quality, vibration level prescribed in the codes
of practice, provide one set of threshold values. However, the
limiting values indicating the end of the functional service life
have bearing on techno-economic considerations. A bridge
component reaches the end of its functional life when the level
of damage warrants not just repair but rehabilitation of the
component. Unlike the end of the structural service life, which
often can be objectively defined on the basis of readily
observable distress signs, the end of functional service life, is
ultimately a matter of opinion.

(d) To develop data bank for documentation of periodically

collected data from bridges and arrangement for drawing of
statistical conclusions.

(This feedback would form the source and the data base
required to assign transition probabilities in Markov chain
process as well as to develop regression expressions connecting
condition ratings with the potential degradation agents.)

(e) Collection of data on the environmental factors causing

degradation and improved methods to measure the intensities
of the factors.

(f) To develop improved knowledge of the mechanisms by which

constituting materials (including coatings/sealers) degrade.

This would involve laboratory tests under controlled conditions

on deterioration rates, such as, in corrosion-affected

to collect data
passive and prestressed steel, rate of carbonation in concrete,
fatigue accumulation under programmable loading. Deterioration
rates referred in earlier sections relate to specific case studies.
The scope would increase as more and more forms of deterioration
processes are identified. This would also lead to establishing
laboratory infrastructure for testing, e.g., weathering chambers,
programmable pulsating loading facility, etc.

This should also cover identification of degradation phenomena

acting concurrently and development of methods to simulate

54
 IRC:SP:60-2002

or account for synergism amongst degradation factors. This is

essential since it has been the experience that failure - mild or

severe - generally will be due to combination of factors, many
of which are not amenable to mathematical modelling.

To develop instrumentation based performance monitoring

procedures incorporating time saving, easy to handle improved
tools and methods for measuring degradation in-situ and a
mobile laboratory operated by trained engineers to evaluate the
condition of the structure on site during testing.

Testing efficacy of different service models

Research programme to analyze concrete structures subjected

to time-related degradation processes need to be planned. Not
only are 'first level' cross-sectional studies but 'second level'
related to the behaviour of the component and 'third level'
related to the responseof the whole structural assembly also
need to be included in such a programme. It is necessary to
understand accuracy of different life prediction methods, as
well as the difficulties in their applications to field examples.
This is true particularly in the cases of Discrete Time Markov

Chain (including its procedure to evolve elements of Transition

Probability Matrix) and structural reliability methods.
Application of these techniques to a large number of existing
and new bridges where in-situ test data on a periodic basis is
available, needs to be undertaken.

Development of durability design philosophy

Control of the durability is gaining increasing importance in

the design of concrete bridges. It is necessary to introduce into
conventional structural design, structural concept of targeted
service life, general theory of structural reliability and available
calculation models for the commonest degradation processes.
Concept of durability at the design stage would enable fixing
durability related parameters, such as, concrete cover, properties
of materials, amount of reinforcement and dimension of the
members taking into account the actual degradation process.
Further the introduction of 'time' element in the design,
availability of periodically updated degradation rates during
service lifeand the predefined targeted service life, would all
help to plan the scope and schedule of maintenance measures.

55
IRC:SP:60-2002

It has to be appreciated that estimation of remaining life

is not a straight-jacket exercise but is an inter-disciplinary

science covering as divergent subjects as material science,
electro-chemistry, fracture mechanics, fatigue, statistics,

probability theories, instrumentation, besides failure mechanisms

of structural engineering. In an effort to expose the readers to
the recent developments in all the related disciplines, it is likely
that the document would give the impression of being less
user-friendly and wanting in precise steps to be followedby the
prospective user in handling the practical cases.

However, it will be realised that the life prediction is one

of today's front line areas of research and given the complexities
involved, every investigation has to be bridge-specific. While it

was necessary to touch upon various related subjects, being an

approach document, any attempt towards spelling out precise
step-by-step procedure would have been inconsistent with the
level of development of the subject. Need in fact is to draw up
a long term plan and co-ordinated efforts from bridge owners,
designer, academic and research establishments, so that at least
the qualitative estimation of health and the remaining life enters
the Bridge Maintenance Systems in future. The purpose of this
document would be served if it triggers action in this direction.

7. REFERENCES
1. T.P. Tassios, "A Seismic Design and the Lifetime Concept", The
Design Life of Structures, Blackie, 1992.

2. C.F. Henriksen, L. Ladefoged and N.Thaulow, "Concrete Specifications

for New Bridges", Bridge Management 3, E & FN SPON 1996.

3. H.P. Webber, "The Client's view-the Public Sector", The Design Life
of Structures, Blackie, 1992.

4. A. Van der Toorn, "The Maintenance of Civil Engineering Structures",

Heron, Vol.39, No.2, 1994.

56
 IRC:SP:60-2002

5. K.H. White, "A Performance Approach to Design", The Design Life

of Structures, Blackie, 1992.

6. G.W. Jordan, "Factors Affecting the Prediction of Design Life of

Structures in the Nuclear Processing Industry", The Design Life
of Structures, Blackie, 1992.

7. CEB Durable Concrete Structures - Design Guide, Thomas Telford

Ltd. 1992.

8. D.G. Manning, "Design Life of Concrete Highway Structures - The

North American Scene", The Design Life of Structures, Blackie,
1992.

9. I. J. A. Rahim and D.W. Johnson, "Bridge Element Deterioration

Rates", Transportation Research Record No. 1490, NRC, Washington
1995.

10. J.A. Gonzalez et. al., "Some Questions on the Corrosion of Steel in
Concrete -Part II", Materials and Structures, Vol.29, March 1996.

1 1 . G.P. Garg, "Assessment of Fatigue Life of Steel Girder Bridges", An

Approach Paper, R.D.S.O., Lucknow, October 1995.

12. L.W. Masters and E.Brandt, "Prediction of Service Life of Building

Materials and Components", CIB W80/RILEM 71 -PSL-Finai Report.

13. P.C. Das, "Bridge Management Objectives and Methodologies", Bridge

Management 3, E & FN SPON, 1996.

14. J. Jubb, "Overview of Service Life Prediction-materials Issues

including QA and Certification". The Design Life of Structures,

Blackie, 1992.

15. M.G. Fitch, R.E.Weyers and S.D.Johnson, "Determination of End of

Functional Service Life for Concrete Bridge Decks", Transportation
Research Record No.1490, NRC, Washington 1995.

1 6. "Service Life Estimates-concrete Bridge Protection and Rehabilitation:

Chemical and Physical Techniques", SHRP-S-668 NRC, Washington
1994.

57
17. J.Zemajtis and R.E. Weyers, "Service Life Evaluation of Concrete
Surface Coatings", Transportation Research Record No. 1490, NRC,
Washington 1995.

1 8. R.E. Weyers, J. Zemajtis and R.O.Drumm, "Service Lives of Concrete

Sealers", Transportation Research Record No. 1490, NRC, Washington
1995.

19. S.K. NG and F. Moses, "Prediction of Bridge Service Life Using

Time-dependent Reliability Analysis", Bridge Management 3, E &
FN SPON 1996.

20. S.A. Barakat and D.M. Frangopol, "Towards Lifetime-oriented

Reliability Based Design of Bridge Systems", Restructuring America
and Beyond, pp. 591-594.

21. B. Ellingwood, "Time-dependent Reliability and Service Life

Predictions ofAging Concrete Structure", Restructuring America
and Beyond, pp. 591-594.

22. D. Veshosky, C.R.Beidieman, G.W.Buetow and M.Demir,

"Comparative Analysis of Bridge Superstructure Deterioration",
Journal of Structural Engineering, ASCE, Vol.120, No.7, July 1994.

23. M. Maage et. al, "Service Life Prediction- an Overview", Concrete

Structures Exposed to Marine Environment", ACI Materials Journal,
Vol. 93, No.6, Nov.-Dec.1996.

24. G. Sommerville, "Service Life Prediction-an Overview", Concrete

International, Vol.14, No.ll, Nov. 1992.

25. P.D. Zuraski and J.E. Johnson, "Research on the Remaining Life in
Steel Bridges", 4th ASCE Speciality Conference on Probabilistic
Mechanics and Structural Reliability, ASCE, 1984.

26. M. Matsumoto, N. Shiraishi, H. Miyake and T. Takanashi, "Evaluation

of Residual Life of Steel Bridges Concerning Corrosion Deterioration",
Structural Safety and Reliability, ICOSSAR 93, Vol.2, 1994.

27. H. Wawk, "BRIDGIT Deterioration Models", Transportation Research

Record No. 1490.

28. M.G.Tamhankar et. al., "Fatigue Tests on Concrete Hinges", The

Indian Concrete Journal, Vol.50, No.2, February, 1976.

58
 IRC:SP:60-2002

29. P.C. Bhasin, D.S.P. Rao, M.G. Tamhankar and S.P. Sharma, "Fatigue
Behaviour of Composite Steel-concrete Girder", IABSE Colloquim
on Fatigue in Steel and Concrete Structures, Lussane, March, 1982.

30. "Guidelines for Inspection and Maintenance of Bridges", IRC Special

Publication 35, 1990.

3 1 . "Guidelines for Evaluation of Load Carrying Capacity of Bridges",

IRC Special Publication 37, 1991.

32. Sonali Ranade and D.V. Reddy, "Computer Modelling for Chloride
Diffusion and Corrosion Fatigue of Reinforced Concrete Beams in
the Marine Environment", Corrosion and Corrosion Protection of
Steel in Concrete, Sheffield Academic Press, Vol.1, pp.648-657.

33. P. Schiessl, "Zur Frage der zulaessigen Rissbreite und der erfordlichen
Betondeckung im Stahlbetonbau unter besonderer Berucksichtigung
der Karbonatisierung des Betons", DAfS, Heft 255, Berlin 1976.

34. K. Tutti, "Corrosion of Steel in Concrete", Swedish Cement and

Concrete Research Institute, Stockholm, 1982.

35. C. Andrade, "Advances in on-site Electrochemical Measurement of

Reinforcement Corrosion and Their Use for Predicting Residual
Life", 13th ICC Proceedings, Paper 003, 1996.

36. J.Rodriguez, L.M. Ortega, J. Casal, J.M.Diez, Conference on Concrete

Repair, Rehabilitation and Protection, Dundee U.K., E & FN Spon,
1996 pp. 65-77.

37. N.S. Rengaswamy, Central Electrochemical Research Institute,

Karaikudi, Personal Correspondence on Corrosion Aspects, Sept. 1997.

38. V.G. Kulkarni, "Modelling and Analysis of Stochastic Systems",

Chapman and Hall, 1995.

39. C. F. Henriksen "Optimisation of Maintenance of Concrete Structures

through Service Life Calculations", Bridge Management 3,. Ed. J.E.
Harding et. al, 1996.

40. R. Wilson, "The Durability of Concrete Bridges", Construction

Repairs: Concrete Repairs 5.

41. "Draft Recommendation for Damage of Concrete

Classification
Structures", RILEM Recommendations, Material and Structures, 1994.
IRC:SP:60-2002

42. H. Subbarao and T.N. Subbarao, "An Approach Document on Whole

Life Costing of Bridges", Paper No.439, Journal of Indian Roads
Congress, Vol.57, December 1996.

43. M.G. Tamhankar, "Assessment of Remaining Life of Bridges- An

Approach Document", Paper No.447, Journal of Indian Roads
Congress, Vol.58, December 1997.

44. Sarja and E. Vesikari, "Durability Design of Concrete Structures",

RILEM Report 14, E & FN SPON, 1996.

45. Milik Tichy, "Applied Methods of Structural Reliability", Kluwer

Aca Publishers, 1993.

46. K. Mani et. al., "An Approach to Investigation of Corrosion -

Affected Concrete Structures and Some Data from Case Studies",

The Indian Concrete Journal, June, 1998.

47. M. Prezzi, P. Geyskens and P.J.M. Monteiro, "Reliability Approach

to Service LifePrediction of Concrete Exposed to Marine
Environment", ACI Materials Journal, No v. -Dec. 1998.

48. — - "Investigation on the Assessment of Residual Fatigue Life of

Early Steel Girder Bridges", RDSO Civil Engineering Report No. C
- 245.

49. J.W. Fisher, "Fatigue and Fracture in Steel Bridges - Case Studies",
John Wiley & Sons, New York, 1984.

50. Buch, "Fatigue Strength Calculations", Material Science Survey No.

6, Transtech Publications Ltd. Switzerland 1988.

51. K.R.Trethewey and J.Chamberlain, Corrosion for Science and

Longman Group Ltd., 1995.
Engineering.,

52. R.J. Woodward and D.L.S. Wilson, "Deformation of Segmental Post-

tensioned Precast Bridges as a Result of Corrosion of the Tendons",
Proceedings of Institution of Civil Engineers, Part 1, 1991, 90, April
pp. 397-419.

53. B. Massicotte et. al., "Strengthening of a Long Span Prestressed

Segmental Box Girder Bridge" PCI Journal, May- June, 1994, pp. 52-
80.

60
 IRC:SP:60-2002

Appendix
NUMERICAL ILLUSTRATIONS FROM LITERATURE
Being a newly emerging area of R&D, there are not many
reported case studies of practical applications. Three numerical
illustrations from the literature are, however, presented in the
following, primarily to bring out the status of the subject. They
cover the following cases:

RC column and beam under degradation of both concrete and

steel .

39
Bridges with piers under chloride attack using in-situ tests .

• ^Concrete coatings used to restrict ingress of chlorides, using

accelerated tests .

I. Service Life Predictions in RC Column and Beam 44

Cross-sections of a simple axially loaded square column

and a beam are considered to illustrate life estimation under
degradation in both concrete and steel (Figs. 16 and 17).

Fig. 16. Cross-section of a column 44 (degradation both in concrete

and steel)

61
IRC:SP:60-2002

Fig. 17. Cross-section of a beam 44

RC Column
Load bearing capacity

R d = A (t)~+A (t)^
c s
yc y*

Where, A (t)
c
and A (t)
s
are areas of cross-section of concrete
and steel respectively as function of time t, and hence can be
expressed as:

A c (t)=(b 0 -2c'(t)) 2

A s (t)=4^-(D„~2d'(t))
4

Where c' = degradation model of concrete expressing the depth of

deterioration of concrete and
d' = degradation model of steel expressing the depth of
corrosion in reinforcement

b. and D are the initial width of column and diameter of

steel respectively.
 lRC:SP:60-2002

Load bearing capacity at any time t R (t) can be expressed as:

d

f
Rd(t) = Rdo - 4 boC (t)— + * Cbd'(t)-^
Yc ys

Where, Rdo is the initial capacity.

Depending upon whether c'(t) and d'(t) are linear or

accelerating/retarding models, the load bearing capacity will

also show similar trend. Fig. 18 shows the reduction in A A,

and Rd expressed in per cent for a typical axially loaded

column when C and d' are assumed to be linear with time.
Calculations have been done for

29 40 fib S 100120 140 160 180 200 220 240 260 260 300
TIME (YEARS)
Fig. 18. Reductions in material cross-sections and compressive
capacity of a column 44

63
IRC:SP:60-2002

RC beam
Load bearing capacity of beam is similarly determined by
calculating the moment of resistance. For more details see Ref. .

Fig. 19 shows the reduction in A s

,
z, R ds
and R dc
expressed
in per cent for a typical beam with following dimensions and

0 40 80 120 160 800 240 260

TIME (YEARS)

Fig. 19. Reduction in the bending capacity of a beam 44

with constant degradation rate c and d\ Here

f

R ds
and R dc
are
the moments of resistance corresponding to under-reinforced
and over-reinforced section, respectively, and z is the internal
lever arm of the moment
bo = mm 400
d0 - 700 mm
D =25 mm
N s
= 3

f
c
= 40 MPa
f = 400 MPa
E = 200000 MPa

64
 IRC:SP:60-2002

E = 9500(f ck +8) J N/mm ;

Yc - 1.5

Ys = 1.15

II. Service Life Calculation in the Case of Chloride

Attack39

The Danish Bridge Maintenance System comprises

modules where service life calculations are necessary for the
optimum use of the system. To meet the requirement of the
Danish Road Directorate, Henriksen of RH&H Consult has
reported a test-cum-service life calculation methodology using
services of a mobile laboratory to evaluate the condition of the
structure on-site during testing. The test results are used to
determine the actual position of the structure on the service life

deterioration vs. age curve and obtain the four points T T T4

2, 3 ,

and T on
5
the time scale (Fig. 2). The table below shows the
examples of service life calculations in the case of chloride
attack on the piers of 5 different bridges. The initiation time

Bridge Cs (%) D (mm /y)2

Time
No. Ground Ground TA * T3 * T4 * T5 *
2

level level

70.0146 0.08 19 OO QO CO CO

70.0182 0.07 110 2000 2010 2015 2025

70.0186 0.13 36 1988 1988 2003 2018
70.0036 0.08 79 1989 1999 2004 2014
70.0047 0.06 16 00 CO CO CO

* with critical chloride level at 0.05 per cent of mass of dry concrete
weight.

Cs - Surface chloride coefficient

D - Diffusion coefficient
1993 - Time of investigation

65
IRC:SP:60-2002

(T 2 ) is calculated using FICK's 2nd Law on the basis of

chloride measurements carried out on the structure. The results
of chloride measurements are used for fitting the curve to
FICK's 2nd Law (Fig. 20).

LEGEND
0.16
Position 1
i
\

Position 2
a s 0.13
u
2
O
u
\i

0.05
VA

0 30 60 90 120

DEPTH (mm)
Fig. 20, Chloride measurements on a bridge deck

The rate of deterioration of structure used for evaluating

points T 3 ,
T 4 and T 5
is estimated on the basis of measurement
on the structure of the electrical resistance, the moisture content,
the porosity of the concrete and the annual average temperature
and moisture variation. These are the primary parameters for
calculating the deterioration of concrete.

The time periods are intended to enable plan the bridge

repairing priorities aimed at optimising available maintenance
funds. For more details see Ref. 39.

III. Life Prediction of Concrete Coating Based on

Accelerated Tests 17

Service life prediction of concrete coating used to restrict

66
 IRC:SP:60-2002

ingress of chlorides from salt-water environment based on

accelerated tests in the laboratory is described in the following.
The example illustrates the application of accelerated test as

well as simulation of field condition.

Series of tests are performed in the laboratory on specially

designed surface coated concrete specimens immersed in sodium
chloride for a long period (one year in the present case).

Due to the presence of coating, surface chloride

concentration is not constant. The equilibrium chloride
concentration (kg/m ) assumed 1.3 cm below the surface is

hence treated as surface concentration. The regression analysis

carried out on the test result values of concentration at 1.3 cm
from the surface shows that it varies as square root of time.
1

(=k t ) where k is the chloride ingress rate through coated or

uncoated layer.

The solution to the problem of semi-infinite medium

whose surface concentration varies in proportion to the function
of time (square root) is obtained by Laplace Transform of
diffusion equation, as:

2 -
= kVt
—D
x
4
c .t
r
xvrc (,
( erf(x) \
C(x, t) e x

2VDct I 2a/d7FJ

Where,

k = coating characteristic constant

t = time
x = depth

C = Chloride concentration

D = diffusion constant

The tests results generate data to know the 'k' values for

67
 IRC:SP:60-2002

the coated as well as the uncoated surfaces. Since these

values
are valid for the laboratory scale tests, they have to be
modified
for their actual application in the field. Hence,

^
Kmodified =
Kcoating
X Kfield
Kc oncrete

Here K coatin g
and Concrete denote the regression based values
for K coefficients obtained in the tests on coated and uncoated
specimens. K is the value calculated from the field data.
field

The following data is considered:

C(x,t) = 0.71 kg/m 3 (chloride threshold value)

x = 4.1 cm (location of steel)

D - 0.84 cmVyear

t = 50 years (desired corrosion protection period)

surface
- 8.9 kg/m 3 (Surface chloride concentration related to
severe condition)

K„a„„g = 1-928

^concrete
- 7.829 (obtained on regressing the test data)

Using above equation and above data with all quantities

except 't' being known, the protection time provided by the
specific coating can be calculated. [In the present case t

equalled 29 years Refer [17, 18, 47] for more details.

68
I