SERVICE LIFE PREDICTION OF CONCRETE STRUCTURES

Mohamad Nagi and Robert Kilgour

GHD Global Pty Ltd, Dubai, U.A.E

Synopsis: Methodologies and computer models for predicting the

service life of reinforced concrete structure were developed over

the years. Methods were mainly used to define the remaining

service life of existing structures based on the durability

characteristics of concrete. Currently, these technologies were

adopted to define the service life at the design stage. In the

Arabian Peninsula and the Gulf region, authorities are requesting

extended service life (75 to 100 years) of their key infrastructures

such as bridges and towers with minimum maintenance and life

cycle cost. Since the Gulf is considered the most corrosive place

in the world, corrosion of reinforcement is the main durability

element controlling the service life of structures in this region.

Prediction models were used to assess the service life of reinforced

concrete towers and bridges and provide guidance to achieve such

targeted life. High performance concrete, corrosion resistant steel

and inhibiting admixtures are currently used in the region to

enhance concrete durability and extend the service life of

structures.

Keywords: Service life, diffusion coefficient, corrosion, concrete

durability

INTRODUCTION

During the first construction booming in the Gulf and Arabian

Peninsula in the early 1970’s, many concrete structures marine

structures were built based on foreign codes without paying

attention to the unique environment in the region. As a result, the

high temperature and severe environment have lead to a major

durability-related deterioration, in some of these structures within

10 to 15 years (1).

Currently, the Gulf and Arabian Peninsula region is on the top of

the world’s list in concrete construction and the daily consumption

of concrete is probably the highest in the world. From super tall

towers to marine, industrials and highways structures, reinforced

concrete stands as the main material used in construction.

In the last decade, Supplementary Cementing Materials (SCM)

such as fly ash, silica fume and ground granulated blast furnace

slag (GGBS) made their way to the Gulf and are commonly used to

produce high strength and high performance concretes.

Considering the high initial construction cost, Developers and

authorities are demanding much longer service life of their

structures (75 to 100 years or more) with minimum maintenance

and life cycle cost. Due to the harsh and severe environments in

the Gulf region, durability characteristics of concrete control its

service life. Production of durable and good quality concrete is the

key to extend the service life of the structures. Designers are now

looking into durability modeling to assess the service life of the

designed facilities.

SERVICE LIFE PREDICTION

In general, service life is the period of time during which a

structure meets or exceeds the minimum requirements set for it.

The requirements limiting the service life can be technical,

functional or economical. (2). The technical requirements are

related to the structural functions of the structure.

As mentioned above, the main deteriorating factors affecting the

service life of concrete structures are durability related ones.

Durability by definition is the ability of concrete to resist

weathering action, chemical attack, and abrasion while maintaining

its desired engineering properties. Concrete ingredients and their

proportions, and interaction among ingredients, curing and placing

of concrete control its ultimate durability (3). Alkali-aggregate

reaction, sulfate attack, and corrosion are the main factors affecting

the performance of concrete structures. In the Gulf region, while

other factors exist, the corrosion of reinforcing steel is the main

factor controlling the service life of reinforced concrete structures.

Mechanism of corrosion is well covered and understood. Steel

reinforcement is usually protected in concrete as far as the

passivated layer (protective iron oxide film) formed in the concrete

high-alkali environment is existed (4, 5). Whenever this layer is

damaged either due to carbonation or the presence of chloride ions,

and in the presence of oxygen, corrosion will start. The chloride-

induced corrosion is the common form of corrosion in the region.

The Gulf area is predominantly ex-seabed sand. There is a very

high chloride content in the sand and ground water. Salt content

can be several times the seawater combining with the high ambient

temperature and high humidity, making the Gulf one of the most

corrosive location in the world.

Service life prediction models

Service life models can be divided into two groups: deterministic

and probabilistic. Deterministic models are based on empirical

relationship, while probabilistic models are based on stochastic

behavior of structures (2,6). It is based on the idea that the service

life cannot be accurately predicted.

All models developed over the years are based on the idea that the

service life is the total of the initiation time and the propagation

time of corrosion. Figure 1 illustrates the principle of service life

analysis. It is assumed that corrosion is initiated when chloride

content at the level of reinforcing steel reaches the defined

corrosion threshold. The estimated time for corrosion initiation

can be calculated using Fick’s second law of diffusion, assuming

diffusion is the main mechanism of chloride ingress into concrete

DC/dt = D. d2C/dx

2

C Chloride content

T time

X depth (form exposed surface)

D apparent diffusion coefficient

The general solution of the above-mentioned equation is as

follows:

Dt

xERFCCCCC isstx

2,

with Cx,t = the chloride concentration at concrete depth x and

at time t,

Cs = the projected chloride concentration at the surface,

Ci = the initial chloride concentration,

D = the apparent chloride diffusion coefficient (m2/sec,

in.2/year), and

ERFC is an error function.

One of the earliest and simplest model for predicting service life of

structures was the one developed by Tutti in the early 1980’s (2,

8). The propagation time is considered to be a constant period of

time. Tutti’s model analysis was supported by experimental data.

A more complicated model combining both deterministic and

probabilistic models was developed by Gannon and others (9)

using Monte Carlo statistical simulation. The Monte Carlo

simulation is used to generate values for an equation whose

variables have a specified distribution. Variables such as

reinforcing cover and diffusion coefficient are used to solve the

diffusion equation.

Determination of Diffusion Coefficient.

As mentioned above, diffusion coefficient is a key factor in

predicting the time to corrosion. For existing structures, concrete

cores are taken and tested to establish chloride profile. Following

the procedures of ASTM C 1556 “Standard Test Method for

Determining the Apparent Chloride Diffusion Coefficient of

Cementitious Mixtures by Bulk Diffusion” the diffusion

coefficient can be determined. Ligozio and Nagi (10), as part of

their study to determine the remaining service life of Chamberlain

Bridge over the Mississippi river in the U.S. measured the

diffusion coefficient of concrete in different parts of the bridge.

The diffusion coefficient for the 50-year old concrete elements

ranged from 0.8 to 2 x10-12

m2/s.

For new constructions, concrete samples can be prepared and

tested in accordance to ASTM C 15556 or NT Build 443. These

tests require an average of two months to be completed. Recently,

a rapid test (NT Build 492) has been introduced in the region to

measure the chloride migration coefficient. The test is based on

applying an external electrical potential to force chloride ions into

the samples. It was reported by Tang (11) that this test has a good

correlation with the diffusion coefficient measured using the NT

Build 443.

Determination of Surface Chloride Concentration

The surface chloride concentration to which an element may be

exposed is not quantified. Codes often provide broad qualitative

exposure classifications such as submerged, spray or splash zones but

these do not provide adequate information to determine the surface

chloride level. For sulfate-bearing ground, a more quantitative

approach has been adopted. For example, AS 2159 refers to five

exposure classifications based on the actual level of sulfate present in

the ground. For coastal structures, Bamforth (12) suggested the

following number of exposure classes and associated surface chloride

concentrations .

PRE-CONSTRUCTION SERVICE LIFE ASSESSMENT

(CASE STUDY)

A large-scale reinforced concrete structure with nominal design

life of a 100 year is under construction in the Gulf region.

Assessment of the durability of the concrete elements of the

structures subjected to defined deterioration scenarios was required

prior to finalizing the design and commencing construction.

Recommendations were made for the mix designs, protective

measures and construction quality assurance.

The main concrete elements of the structure considered in the

durability design were:

Bored piles and piles caps

Retaining walls

Raft slab

Ground floor slabs

Exposed superstructures elements

The durability and serviceability of concrete in aggressive

environment is addressed at the design stage by the selection of

appropriate mix designs and the specification of additional

protective measures and construction quality assurance measures.

The deterioration scenarios assessed in the project were sulfate

attack, carbonation of concrete and chloride induced corrosion of

PC Blended Cement

Extreme exposure Csn = >0.75% >0.9%

Severe exposure Csn = 0.5% to 0.75% 0.6% to 0.9%

Moderate exposure Csn = 0.25% to 0.5% 0.3% to 0.6%

Mild exposure Csn = < 0.25% <0.30%

the reinforcement. A durability plan that outlines the requirements

for durability and the assessment of compliance with final

requirements was prepared. The assessment of service life based

on chloride-induced corrosion is presented in this paper.

Ground Conditions

Chemical testing undertaken on soil and ground water indicated

high level of chlorides, up to 21 g/L. Groundwater pH was

reported to be between 7.1 and 7.6. The structure foundations are

located below groundwater table.

Basis of Analysis

The study was based on the assumption that the diffusion process

governs chloride ingress into the concrete over the longer term.

The diffusion coefficient for concrete is generally influenced by

the permeability and porosity of the concrete, which in itself is

influenced by the cement content, the aggregate grading, the use of

cement replacements, the water cement ratio compaction and

curing. Data of measured diffusion coefficients for various

concrete mixes using blends of OPC with GGBS, and OPC with

PFA as well as ternary blends that include silica fume was used to

show the possible diffusion coefficient variability, prior to

measurement of diffusion coefficient of the trial mixes conducted

prior to construction.

Minimum Requirements for Atmospherically Exposed Concrete

The following requirements are based on the guidance from AS

5100.5 (2004) and assume the concrete will have a minimum

strength of 45 MPa. The requirements apply to formed slabs,

beams, walls and columns.

Exposure classification B2

Cover 55 mm

Probabilistic Corrosion Model

The acceptable degree of deterioration considered in the study of

the buried and atmospherically exposed elements is to avoid

spalling of concrete. In this case, corrosion initiation is allowed,

with progression of corrosion sufficient to just cause cracking of

the concrete after the required service life

The in-house probabilistic model determines the probability that

sufficient corrosion will occur to cause cracking of the concrete

over a range of diffusion coefficients, surface chloride levels,

concrete covers, activation thresholds and corrosion rates. The

reliability index is the number of standard deviations from the

mean of the failure equation:

P(t) = (- ) = {CCx(t) – Xc 0}

Where CCx(t) is the cumulative corrosion (in microns) at

reinforcement depth at time t

Xc is the amount of corrosion necessary to induce

cracking at the reinforcement depth

is the standard normal distribution function

is the reliability index (no. standard deviations from the

mean)

In the elements of many structures, loss of cover and minor section

loss do not in themselves constitute any significant loss of

structural capacity or serviceability. Significant loss of section

would be required for structural safety to be compromised.

Accordingly the limiting value of Beta for design does not need to

be the same as in structural considerations, where 3.8 is often used

(e.g. Eurocode).

The study considered a value of 1.65 (5% probability of cracking)

was an appropriate minimum value to adopt for a long service life

(50 years plus) for civil engineering structures. A reliability index

of 2.3 was used where cracking itself carried a safety risk (I.e., for

atmospherically exposed concrete n above pedestrian and vehicular

access ways.

Data Analyses

The probabilistic model assesses the risk of deterioration due to the

ingress of chlorides. The following scenarios (Table 1 and 2) are

considered for the various possible configurations of concrete mix

design, concrete cover to reinforcement as well as bar size (based

on the design development documentation). The diffusion

coefficients (determined by bulk diffusion tests such as those

described by NT Build 443 or ASTM C 1556) are assumed based

on what could be achieved with a carefully designed mix and good

batching and construction quality control. It is possible that these

mixes could achieve much lower diffusion coefficients.

Cover to the reinforcement is taken as the cover to the external

bars of the reinforcing cage –typically identified as

ligatures/stirrups/hoops depending on the element in question.

Main bars will have additional cover equivalent to at least the

diameter of the ligature/stirrup/hoop reinforcement.

For piles it is assumed that the coefficient of variability (CoV) of

the cover will be significant (the model assumes CoV is 25% as

evidenced on various foundation projects in the region.

The variability of cover for elements such as slabs, columns and

beams should be less (the model assumes CoV is 10%.

The model assumes a target cover of 100 mm for piles, 75 mm for

slabs (all faces exposed to soil).

In either case, normal code tolerances for cover to the

reinforcement should not be exceeded.

Table 3 and 4 summarise the outputs of the model. The acceptable

reliability index will be influenced by the structural impact of

cracking (if any) – typically a minimum reliability index of 1.65 is

considered acceptable, however higher values (> 2.3) may be more

appropriate where cracking and spalling presents a risk to safety.

Typical time to corrosion and time to cracking charts for the

selected rebar sizes and type of steel are shown in Figures 2 and 3

for GGBS and PFA concretes.

Actual cover required for these elements would be subject to the

type of formwork used and the application of coatings and

cladding systems. For inaccessible elements, a minimum of 65 mm

cover will be required to surfaces where CPF is not used or where

coatings are not applied. Where coatings are applied, they should

be able to provide the equivalent protection of at least 10 mm of

concrete and be maintained in accordance with manufacturer’s

recommendations.

The model is based on a service-ability limit state of sufficient

corrosion to just cause cracking, i.e. approximately 100 microns.

This amount of corrosion will not affect the structural performance

in any way. If the analysis had assumed a greater degree of

corrosion to cause say loss of 20% bar section, or more, which may

affect structural integrity, then a much higher reliability index

would be appropriate, however the model considers a conservative

service-ability limit state therefore we consider that the minimum

RI=1.65 is adequate (5% of the element affected), as achieving

such an amount of corrosion will not cause failure.

SUMMARY AND CONCLUSIONS

Service life prediction of concrete structures based on concrete

durability characteristics is well-established methodology.

Computers models were developed to predict remaining service

life of existing structures taking into consideration exposure

conditions, design issues (e.g., concrete cover) and concrete

properties. Diffusion coefficient of concrete is the essential

property used in the service life assessment.

In the Arabian Peninsula and Gulf region, due to harsh and severe

environments, chloride-induced corrosion of reinforcing steel

dominates the durability factors leading to deterioration of concrete

structures. Foundations, raft slabs and piles are the critical

elements of the structures since they are buried with contaminated

soil and ground water and cannot be repaired.

Due to the high initial cost of massive structures being built in the

Gulf region, developers and authorities are demanding extended

service life of the structures with minimum life cycle cost.

The authors used in-house probabilistic model to assess the service

life of large-scale tower to be built in the region at the design stage.

Using the recommended concrete mix design with the design-

related inputs, the structure could be expected reach the 100-year

design life with minimum repair or maintenance. The use of other

corrosion protection system such as ASTM A 1035 steel will add

assurance to the owner that structure will reach its service life even

if there were changes in the input items (e.g., covers) occur during

construction. Measurement of diffusion coefficient of concrete

trial mixes was added to the specification of the project. Authors

fond that rapid migration coefficient test can be used as indication

for assessing the finding of the model.

REFERENCES

1. “Guide to the Maintenance and Repair of Reinforced Concrete

Structures in the Arabian Peninsula,” Concrete Society,

presented at the Bahrain 6th

International Conference,

November 2000.

2. Vesikari, E, “Service Life of Concrete Structures with regard

to Corrosion of Reinforcement, ”Technical Research Centre of

Finland, ESPOO 1988.

3. Kosmatka and et.al, “Design and Control of Concrete

Mixtures,” Fourteenth Edition, Portland Cement Association,

Skokie, IL, U.S.A

4. Nagi, M and Whiting, D, “Corrosion of Prestressed

Reinforcing Steel in Concrete Bridges, State-of-the-Art,”

Concrete Bridges Aggressive Environments Symposium, SP

151, 1994 American Concrete Institute, Detroit, Michigan,

U.S.A.

5. Broomfield, J., “Corrosion of Steel in Concrete,” 1997 E &

FN Spon, U.K.

6. “Chloride Penetration into Concrete, State-of-the-Art,”

HETEK, Report No. 53, 1996, Road Diroctorate, Denmark.

7. Weyers, et al, “Concrete Bridge Protection and Rehabilitation:

Service Life Estimates,” SHRP-S-668, Transportation

Research Board, Wahsingtion, D.C., 1994

8. Tutti, K., “Corrosion of Steel in Concrete,” Swedish Cement

and Concrete Research Institute, Report No. 4-82, 1982.

9. Gannon, et.al, “Deterioration Model for Corrosion in Concrete

Using Monte Carlo Simulation,” Structural Engineering in the

21st Century. ASCE, 1999.

10. Ligozio, C. and Nagi, M., “Remaining Service Life

Evaluation, Chamberlain Bridge Substructures,” South

Dakota Department of Transportation, U.S.A. 2004

11. Tang, L and Sorensen, H.E., “Precision of the Nordic Test

Methods for Measuring the Chloride Diffusion/Migration

Coefficients of Concrete,” Materials and Structures, Vol. 34,

October 2001.

12. Bamforth, P.B and Pocock, D.C. (2000) “Design for durability

of reinforced concrete exposed to chlorides” Workshop on

Structures with Service life of 100 years- or more, Bahrain

TABLES

Table 1. Scenarios – Buried concrete

Scenario BG 1 BG 2

Mix 40% PFA + SF 70% GGBS + SF

Diffusion coefficient

(×10-12

m2/s)

1.5 2.0

CoV 15%

Cover (mm) 75, 100

Bar size (mm) 12, 16,

16 (ASTM A 1035 steel)a, 32

b

aAssumes ASTM A1035 steel – corrosion threshold is increased

bassumes main bars with additional cover due to tie bars (typically

T12)

Table 2. Scenarios – Atmospheric concrete

Scenario AC 3

(Exterior columns)

Mix 25% PFA + SF

Diffusion coefficient

(×10-12

m2/s)

2.0

CoV 15%

Cover (mm) 55, 65, 75

Bar size (mm) 16, 32

Table 3 – Calculated Reliability Index for Time to Cracking –

Buried elements

Scenario Mix basis Bar

diameter

(mm)

RI 100 y

(Piles –

100 mm

cover)

(wet)

RI 100 y

(Slabs,

Rafts – 75

mm cover)

(wet)

BG 1 40% PFA +

SF

D=1.5x10-

12m

2/s

12 3.1 2.6

16 2.7 2.1

16 (ASTM

A 1035

steel)

5.7 8.5

32 3.8 7.5 (100

mm cover)

BG-2 70% GGBS

+ SF

D=2.0x10-

12m

2/s

12 4.3 4.2

16 3.7 3.6

16 (ASTM

A 1035

steel)*

8.5 8.5

32 5.5 2.7

Table 4 - Calculated Reliability Index for Time to Cracking –

Atmospherically exposed elements

Scenario Bar

diameter

(mm)

RI 100 y

(55 mm

cover)

(dry)

D=2.0×

10-12

m2/s

RI 100 y

(65 mm

cover)

(dry)

D=2.0×

10-12

m2/s

RI 100 y

(75 mm

cover)

(dry)

D=2.0×

10-12

m2/s

AC 3 16 0.9 2.3 4.2

32 0.3 1.6 3.2

FIGURES

Fig. 1 - Principles of Concrete Service Life (Chloride-Induced

Corrosion)

Fig. 2 - Time to Corrosion and Time to Cracking (GGBS Mix)

Fig. 3 - Time to Corrosion and Time to Cracking (PFA Mix)