You are on page 1of 5

Interrelation among Service Life, Reliability Index,

and Costs of Concrete Structures Subjected


to Aggressive Exposure
Břetislav Teplý 1

Abstract: The combined impact of the service life and relevant reliability level of structures on the financial and environmental aspects of
infrastructure building has not yet been given full consideration. The decision of the client regarding the optimal service life and associated
level of reliability is needed. A suitable approach is presented in the context of limit states associated with durability. The reliability assess-
ment is effectively performed by employing stochastic methods. Predictive mathematical models for material degradation are utilized. Some
user-friendly tools and numerical studies for concrete structures show how the necessary data about service life and reliability level can be
Downloaded from ascelibrary.org by New York University on 05/10/15. For personal use only.

gained; in this way good design practice in the construction industry and effective decision-making processes are supported. Two numerical
examples are presented, as follows: (1) an analysis of initiation period with consideration given to the carbonation effect (a compromise
between the service life and the reliability index is studied), and (2) an example of a concrete structure manufactured from concretes of
different quality and exposed to chloride ion ingress. DOI: 10.1061/(ASCE)CF.1943-5509.0000476. © 2014 American Society of Civil
Engineers.
Author keywords: Service life; Reliability; Lifecycle costs; Limit states; Concrete structures; Degradation models.

Introduction 3. The LCC, LCA, and LCM are hence also influenced by the
level of reliability.
One of major goals of infrastructure building policies is the optimal This paper is aimed at underlining the role of mutual impacts
allocation of available economic resources. When attempting to and the dependency between the design service life and the relevant
achieve this, the ability to make a reliable comparison of the life- reliability level of structures, or between the target reliability and
cycle costs for different construction, technological, and material the assessment or prediction of the relevant service life with special
scenarios brings the necessary clarity to the decision-making pro- consideration given to concrete structures. Illustration of previously
cess. In this context the important topics are e.g., the probabilistic mentioned points 2 and 3 is thus provided.
vulnerability assessment of civil infrastructure systems, lifecycle The cost-reliability interaction in LCC has also been addressed
costing (LCC), lifecycle assessment (LCA), and lifecycle manage- by Kong and Frangopol (2004), who introduced so-called cost
ment (LCM). Cost-benefit issues are considered together with functions. The proposed method is rather general, aimed at the cost
sustainability and environmentally friendly policies (Frangopol efficiency of lifecycle maintenance solutions for deteriorating
et al. 2012; Thoft-Christensen 2012). The durability issue has structures. Its limitation lies in the utilization of subjective assump-
gained considerable attention together with performance-based ap- tions and engineering judgments.
proaches, sustainability, and cradle-to-grave thinking.
The construction industry is based on client-driven and user-
driven complete lifecycle processes. The cost reduction in the Reliability and Durability
overall value chain may result in the increased competitiveness
of building firms. This creates the need for performance-based A general definition of structural reliability is introduced in ISO
and consequence-based design, and also the modeling and realistic 2394 (ISO 1998) and EN 1990:2002 (EN 2002). The verification
prediction of the deterioration of aging structures with the relevant of a structure with respect to its reliability, i.e., to a particular limit
consideration of uncertainties; these are the prerequisites for an state (LS), is carried out via estimation of the probability of the
effective decision-making process. occurrence of failure Pf in a specified reference period. This prob-
Considering the previously stated points, three facts and conse- ability is usually expressed as
quences may be stressed, as follows:
1. Reliability and durability are crucial performance characteris- Pf ¼ PfR ≤ Sg ð1Þ
tics of a structure;
2. The service life of a structure may be considerably influenced where S = effect of action; and R = resistance. Both are random
by the target reliability level; and variables (or, in a more general sense, random fields). The general
1
limit condition for both the ultimate limit state (ULS) and service-
Professor, Faculty of Civil Engineering, Brno Univ. of Technology, ability limit state (SLS) reads
Veveří 95, 602 00 Brno, Czech Republic. E-mail: teply.b@fce.vutbr.cz
Note. This manuscript was submitted on December 9, 2012; approved
on May 21, 2013; published online on May 23, 2013. Discussion period P f < Pd ð2Þ
open until October 9, 2014; separate discussions must be submitted for in-
dividual papers. This paper is part of the Journal of Performance of Con- where Pd = design (acceptable, target) probability value. The index
structed Facilities, © ASCE, ISSN 0887-3828/04014003(5)/$25.00. of reliability β is alternatively utilized instead of Pf in practice; the

© ASCE 04014003-1 J. Perform. Constr. Facil.

J. Perform. Constr. Facil. 2014.28.


target value then becomes β d . The formula for the transformation where td = design life; and tS = service life, which can be deter-
reads mined as the sum of two service-life predictors

β ¼ −Φ−1 Pf ð3Þ tS ¼ ti þ tp ð5Þ

where Φð·Þ is the standard normal probability distribution function. where ti = time of the initiation of reinforcement corrosion;
Generally, both S and R may change in the course of time, and and tp = time period after corrosion initiation (the propagation
hence Pf and β are time dependent. This should be stressed period).
especially in the context of limit states (LSs) associated with Considering the durability issue, a special category of LS
durability per ISO 13823 (ISO 2008) and International Federation has been introduced which is intended to prevent the onset of
for Structural Concrete (2010). Thus, for the reliability assessment deterioration or allow only a limited range of degradation effects.
of newly designed as well as existing structures the full probabi- This category represents a set of limit states which may be called
listic safety format has to be employed. Moreover, this approach durability limit states (DLSs) per ISO 13823 (ISO 2008) and
is the only one which provides quantitative information about International Federation for Structural Concrete (2010). The DLS
the safety level. The development of applications using stochastic may be formally regarded as belonging in the SLS category as usu-
methods is inevitable and the considerable uncertainties associated ally the depassivation of reinforcement, i.e., the initiation period, is
with parameters governing deterioration processes highlight the considered to be a limit.
need for the use of probabilistic methods. Being random quantities, the relevant values of variables S
Reliability level, limit state definition, service life, and cost- and R used in Eq. (1) have to be assessed via the utilization of
Downloaded from ascelibrary.org by New York University on 05/10/15. For personal use only.

saving results are mutually related; some elements of whole life a suitable degradation model, or by field or laboratory investiga-
costs are service-life dependant, e.g., inspection and maintenance tions. In the former case effective probabilistic software tools
costs change with service life length. Durability and its reliability should be utilized.
implications therefore need to be addressed during the design pro- The problem of degradation in RC structures is broad and com-
cess due to their pronounced economic and sustainability impacts; plex. Different aggressive agents affect the materials used in the
the agreement or decision of the client should be a fundamental structure during its service life and a variety of degradation effects
part of that process. This is not yet a view commonly held by en- take place governed by many factors depending on the type of
gineers, even though these ideas are clearly expressed in Section 3 degradation mechanism, exposure, and material used; it would be
of International Federation for Structural Concrete (2010), as fol- beyond the scope of this paper to list all of them. For example, the
lows: “Specifying performance requirements and the associated main factors affecting concrete carbonation are the type and content
constraints of service life and reliability creates an initial bridge of binder, the water-binder ratio, the degree of hydration, the con-
between the needs of the stakeholders and the design or the asses- centration of CO2 , the relative humidity, and temperature.
sment. : : : The specified (design) service life or the residual service A great many works have been published in last 2 decades about
life are related to the required service life as given by the stakehold- the durability assessment of RC structures (Stewart and Rosowsky
ers and to other implications of the service criteria agreement 1997; DuraCrete 1999; Biondini and Frangopol 2008; Vořechovská
e.g., with regard to structural analysis, maintenance and quality et al. 2010; Teplý et al. 2010a, b).
management.”
Consequently, the target level of reliability in the context of
durability should be discussed with the client or left to the client’s Effect of Reliability Index Value
decision, as it should depend on a balance between the consequen-
ces of failure and the costs of safety measures per ISO 2394 (ISO As stated in the basic codes for structural design, i.e., in ISO 2394
1998) and EN 1990:2002 (EN 2002). (ISO 1998) and EN 1990:2002 (EN 2002), the recommended reli-
ability index value for an irreversible state (i.e., SLS) is 1.5, which
is relevant to a 50-year design service life. The values of 0.8 < β d ≤
Durability of Concrete Structures 1.8 for the DLS are still under discussion, e.g., the recommendation
of International Federation for Structural Concrete (2010) for the
RC is the most widely used building material, and its durability β d is the value 1.3.
features have to be understood and assessed. Concrete is composed Reliability level, limit state definition, target service life, and
of cement, water, sand, and aggregates; cement being the most cost-saving results are mutually related; some elements of whole
pollutive element as its production is responsible for about 80% life costs and benefits are service-life dependent. Those are espe-
of the concrete industry’s CO2 emissions (Habert 2012). The cially the following four points:
most effective way to reduce environmental impact is to replace 1. Operating costs (such as insurance, energy, water, sewage, and
portland cement with supplementary cementitious materials facility management), all during the design service life td ;
(SCMs), which may have considerable durability implications 2. Maintenance costs (including inspections, i.e., actions planned
(Chromá et al. 2007). during the design service life td );
Owing to either carbonation of RC or the ingress of chlorides 3. Repair and/or degraded element replacement costs, including
into such concrete, the depassivation of reinforcing steel occurs any possible loss due to the discontinuation of operations;
(the initiation period); this may be followed by steel corrosion these effects come with a certain probability (associated with
(the propagation period; Tuutti 1982). With the focus on these peri- the relevant limit state and target value of the reliability index);
ods, the modelling of relevant LSs within the probabilistic approach this should be taken into account when making decisions; and
for the RC durability issue is described briefly next. 4. Expected benefits (for the owner, user, and society), all during
Eq. (1) may also be expressed by means of the service life the design service life td .
format as To summarize, the interdependence of service life, reliability
level, costs, benefits, and environmental stress should be recog-
Pf ¼ Pðts ≤ td Þ ≤ Pd ð4Þ nized and balanced well-optimized, i.e., the target is the inclusion

© ASCE 04014003-2 J. Perform. Constr. Facil.

J. Perform. Constr. Facil. 2014.28.


of long-term financial implications in the evaluation of design, con-
struction, and real estate decisions.
As mentioned previously, the agreement or decision of the client
regarding the demanded/optimal service life and associated level
of reliability (i.e., maximum failure probability or corresponding
reliability index) are needed. Regarding this point, the client also
has to agree with chosen serviceability criteria in the context of the
deterioration process. The designer performs the modeling and the
assessment of the relevant limit states, related reliability indices,
and consequently the service life.

Tools and Examples

Qualified decision making with consideration given to all the


consequences and uncertainties discussed previously is rather a
complex task, and one which is strongly case-dependent, so it
is not feasible to provide a complete realistic example in this Fig. 1. Service life versus reliability index, influenced by the carbona-
paper. Instead, some studies showing the effect of reliability level tion effect for different concrete cover thicknesses a; also, the level of
Downloaded from ascelibrary.org by New York University on 05/10/15. For personal use only.

on service life or other implications for concrete structures are β ¼ 1.5 is shown
detailed next.
A brief introduction to the software tools (both developed in
cooperation with the writer) utilized for these studies is as follows:
components and for ambient air parameters) are not repeated in this
• Rovnaník (2008) is a user-friendly tool which utilizes three
models of the concrete carbonation process. It considers input paper to save space; moreover, their details are not relevant to the
data as random variables (e.g., the probability density function, goal of this paper. Fig. 1 shows the results of the stochastic analysis,
mean value, and standard deviation) and performs stochastic which shows the tradeoff between service life and the reliability
analysis. It is freely accessible, offering two options, as follows: index. The level of the reliability index recommended by EN
(1) service life assessment, which provides the statistical evalua- 1990:2002 (EN 2002) for the SLS is also highlighted. For example,
tion of initiation period (considered to be equal to service life) for a ¼ 30 mm and β ¼ 1.5 the corresponding tS ¼ 42 years,
and optionally provides the service life corresponding to the whereas for β ¼ 0.8 the tS is approximately 60 years, i.e., the de-
target value of the reliability index β d ; and (2) concrete cover cision to use a lower level of reliability leads to an increase in the
assessment, which provides the statistical evaluation of concrete service life prognosis of more than 40%. Evidently, such a choice
cover. The reliability index β relevant to the required value of may have remarkable financial consequences in terms of the impact
concrete cover can be gained. on the costs and benefits related to the expected service life; see
• Feasible Reliability Engineering Tool—Degradation FReET-D points 1, 2, and 4 in the Effect of Reliability Index Value section.
(Novák et al. 2010) is a module for the statistical, sensitivity, and Such a type of study carried out during the early phases of the
reliability assessment of degradation effects in reinforced and design process may support effective decision making and help
prestressed concrete structures, currently and their modifica- to optimize lifecycle costs.
tions. It serves for assessing the potential degradation of newly Regarding Study 2, to present the effect of material quality an
designed as well as existing concrete structures in tasks such as example of a concrete structure exposed to deicing salts or the ef-
the assessment of service life and the level of relevant reliability. fect of a coastal environment is analyzed for different concrete
The user may create different limit conditions. For more details qualities, represented in this paper by the diffusion coefficient
see Novák et al. (2010) and Teplý et al. (2007). Having a broad- D of concrete. According to Duprat (2007), the values D ¼
er choice of models is useful, e.g., due to problems with the 1 × 10−12 and 2 × 10−12 represent excellent and good quality con-
availability of statistical data for the input variables of some cretes, D ¼ 4 × 10−12 is for ordinary quality concrete, and D ¼
models. The comparison and verification of several models 7 × 10−12 (m2 =s) is for poor quality concrete, with a lognormal
of concrete carbonation and chloride ingress into concrete uti- probability distribution and a coefficient of variation of 0.7 (high
lized in FReET-D are presented in Teplý et al. (2012), together uncertainty). The mean concentration of Cl− on the nearest surface
with examples of detailed input and output data. CS ¼ 2 (weight %=c) is considered (one-bounded normal distribu-
Regarding Study 1, utilizing Rovnaník (2008), an analysis of tion with a coefficient of variation of 0.75). Using the critical Cl−
initiation period with consideration given to the carbonation effect concentration Ccr ¼ 0.6 (weight %=c; Duprat 2007), Eq. (4) was
was performed in Teplý et al. (2010a). In this paper some of those assessed for concrete cover a ¼ 50 mm. The most frequently ref-
results are extended to illustrate the interdependence of the service erenced model, applied e.g., by Collepardi et al. (1972), see also
life (defined as the period before the commencement of reinforce- Tikalsky (2005), was employed using FReET-D.
ment corrosion) and the level of the related reliability; the techno- Fig. 2 presents the results of the stochastic analysis, the depend-
logical and/or design quality of a structure are also assessed in this ence of service life versus the reliability index. For example, for
way. Eq. (4) was formulated and evaluated employing a model de- β ¼ 0.8 the tS ranges from 2–18 years depending on concrete qual-
veloped by Morinaga (Papadakis et al. 1992), for a concrete of low ity; for the higher reliability requirement β ¼ 1.8 the corresponding
quality (compressive strength 23 MPa) in a typical city environ- service life is predicted to be much lower. It is expected that the
ment (mean ambient content of CO2 ¼ 800 mg=m3 ) and for differ- chosen variant will have relevant consequences in terms of costs
ent values of concrete cover thickness (a ¼ 25–45 mm). Detailed and benefits; these are not specified in this paper because of
information about the model and some other input data (statistical case-specific dependences on many factors (structure type, loading
characteristics and types of probability distributions, for concrete and exposure conditions, maintenance, and others).

© ASCE 04014003-3 J. Perform. Constr. Facil.

J. Perform. Constr. Facil. 2014.28.


5
CS = surface chloride concentration;
4 D = diffusion coefficient;
P = probability;
[-]
3
Pd = design probability of failure;
Reliability index

2
Pf = probability of failure;
1 R = resistance;
0
S = effect of action;
td = design life;
-1 ti = time of initiation of corrosion;
-2 tp = corrosion propagation period;
ts = service life;
-3
0 20 40 60 80 100 120 β = reliability index; and
Time [years] Φ = standard normal probability distribution function.
(EN 1990) a = 25 mm a = 30 mm
a = 35 mm a = 40 mm a = 45 mm

Fig. 2. Service life versus reliability index, influenced by chloride ion References
ingress for concretes of different diffusion coefficient values
Biondini, F. A., and Frangopol, D. M. (2008). “Probabilistic limit analysis
Downloaded from ascelibrary.org by New York University on 05/10/15. For personal use only.

and lifetime prediction of concrete structures.” J. Struct. Infrastr. Eng.,


399–412.
Conclusions Chromá, M., Rovnaník, P., and Teplý, B. (2007). “Carbonation modelling
and reliability analysis of RC structures made from blended cements.”
This paper discusses the effect of and compromise between the Proc., Int. RILEM Workshop on Performance Based Evaluation and
service life and the time-varying reliability level of deteriorating Indicators for Concrete Durability, Vol. 47, V. Baroghel-Bouny,
structures on whole-life costs and benefits. Due to the large uncer- et al., eds., Bagneux, France, 319–325.
tainties associated with the prediction of deterioration processes the Collepardi, M., Marcialis, A., and Turrizuani, R. (1972). “Penetration of
application of stochastic methods and the employment of analytical chloride ions into cement pastes and concrete.” J. Am. Ceram. Soc.,
models are recommended. 55(10), 534–535.
The presented approach for the assessment of a degrading con- Duprat, F. (2007). “Reliability of RC beams under chloride-ingress.”
crete structure is formed by the following steps for each of the rel- Constr. Build. Mater., 21(8), 1605–1616.
evant design variants: (1) the gathering and sorting of information DuraCrete. (1999). “General guidelines for durability design and redesign.”
European Union–Brite-EuRam III Task 7 Rep., COWI Consulting
about the structure, (2) decision making regarding the exposure
Engineers and Planners, Lyngby, Denmark.
condition (choice of limit state and choice of model), (3) the European Committee for Standardization. (2002). “Basis of structural
obtaining of relevant input data values including their statistical design.” EN 1990:2002, Brussels, Belgium.
parameters (statistical and reliability analysis, and service life Frangopol, D. M., Saydam, D., and Kim, S. (2012). “Maintenance, man-
and reliability index assessment, and (4) the evaluation of results agement, life-cycle design and performance of structures and infrastruc-
and related costs (appropriate decision making). Final decision ture: A brief review.” J. Struct. Infrastruct. Eng., 1–25.
making follows in cases when there is more than one design variant. Habert, G. (2013). “A method for allocation according to the economic
The prerequisite for such a decision-making process is that behaviour in the EU-ETS for by-products used in cement industry.”
the demanded reliability level and required service life should Int. J. Life Cycle Assess., 18(1), 113–126.
be agreed on or decided by the client. This option may have pro- International Federation for Structural Concrete. (2010). “Draft model
code.” Bulletins No. 55 and 56, Lausanne, Switzerland.
nounced financial and sustainability impacts.
ISO. (1998). “General principles on reliability for structures.” ISO 2394,
Two studies focused on concrete structures are presented show- Geneva.
ing the mutual impacts of service life and reliability level. ISO. (2008). “General principles on the design of structures for durability.”
The presented approach may be applied when searching for op- ISO 13823, Geneva.
timum lifecycle cost solutions, in the advanced design of structures Kong, J. S., and Frangopol, D. M. (2004). “Cost-reliability interaction in
(making decisions about alternatives), or in cost-effective decision- life-cycle cost optimization of deteriorated structures.” J. Struct. Eng.,
making processes concerning maintenance planning (including 10.1061/(ASCE)0733-9445(2004)130:11(1704), 1704–1712.
inspection planning). It may also be of use, e.g., in the drafting Novák, D., et al. (2010). “FReET-D.” 〈http://www.freet.cz〉 (Apr. 12, 2012).
or assessment of public procurements and public private partner- Papadakis, V. G., Fardis, M. N., and Vayenas, C. G. (1992). “Effect of
ship projects. composition, environmental factors and cement-lime mortar coating
on concrete carbonation.” Mater. Struct., 25(5), 293–304.
Rovnaník, P., et al. (2008). “RC lifetime.” 〈http://www.freet.cz〉 (Apr. 12,
2012).
Acknowledgments
Stewart, M. G., and Rosowsky, D. V. (1997).“Corrosion, deterioration, and
time-dependent reliability of reinforced concrete bridge decks.” Struc-
The financial support of the Technology Agency of the Czech
tural Engineering Research Rep. STR-97-01, Clemson Univ., Clemson,
Republic, Project No. P105/10/1156 COMOCOS–Complex SC.
Modeling of Concrete Structures, is acknowledged. Teplý, B., Chromá, M., and Rovnaník, P. (2010a). “Durability assessment
of concrete structures: Reinforcement depassivation due to carbona-
tion.” J. Struct. Infrastruct. Eng., 317–327.
Notation Teplý, B., Chromá, M., Rovnaník, P., and Novák, D. (2013). “Role of
modeling in probabilistic durability assessment of concrete structures.”
The following symbols are used in this paper: Life-cycle and sustainability of civil infrastructure systems, A. Strauss,
a = reinforcement cover thickness; D. M. Frangopol, and K. Bergmeister, eds., Taylor & Francis Group,
Ccr = critical chloride concentration; London, 876–882.

© ASCE 04014003-4 J. Perform. Constr. Facil.

J. Perform. Constr. Facil. 2014.28.


Teplý, B., Matesová, D., Chromá, M., and Rovnaník, P. (2007). “Stochastic in concrete exposed to de-icing salts.” Proc., Int. Congress Global
degradation models for durability limit state evaluation: SARA–Part VI.” Construction: Ultimate Concrete Opportunities, Concrete for Trans-
Proc., Int. Conf. on Structural Health Monitoring of Intelligent portation Infrastructure, R. K. Dhir, M. J. McCarthy, and S. Caliskan,
Infrastructure (CD-ROM), B. Bakht and A. Mufti, eds., Vancouver. eds., Dundee, 251–258.
Teplý, B., Vořechovská, D., and Keršner, Z. (2010b). “Performance-based Tuutti, K. (1982). Corrosion of steel in concrete, Swedish Cement and
design of concrete structures: Durability aspects.” Struct. Eng. Mech., Concrete Research Institute, Stockholm, Sweden.
35(4), 535–538. Vořechovská, D., Teplý, B., and Chromá, M. (2010). “Probabilistic
Thoft-Christensen, P. (2012). “Infrastructure and life-cycle cost-benefit assessment of concrete structure durability under reinforcement corro-
analysis.” J. Struct. Infrastruct. Eng., 507–516. sion attack.” J. Perform. Constr. Facil., 10.1061/(ASCE)CF.1943-5509
Tikalsky, P. J. (2005). “Monte Carlo simulation of chloride diffusion .0000130, 571–579.
Downloaded from ascelibrary.org by New York University on 05/10/15. For personal use only.

© ASCE 04014003-5 J. Perform. Constr. Facil.

J. Perform. Constr. Facil. 2014.28.