10DBMC International Conference On Durability of Building Materials and Components LYON [France] 17-20 April 2005

Sustainable tools and methods for estimating building materials and components service life.

J. Hans, JL. Chevalier CSTB, 24 Joseph Fourier Street, F-38400 Saint Martin d'Hères, France [email protected]

TT4-90

ABSTRACT

The need of building materials and components service life assessment is strongly increasing in the actual building sector. This phenomenon is due to new techniques developments and new environmental requirement and principles. The behaviour of material performance within time in several environments has been largely studied, and good models contribute to a reliable evaluation of service life of materials when environmental and external solicitations are well defined. On the other hand, evaluation of components service life is not so well known. It can be explained by the complexity of the components and the lack of knowledge on the failure scenarios of these components. Furthermore, the feedback management is not efficient, and the knowledge is not enough collected and accessible. Our main interest is the development of tools and methods for service life assessment which remain perfectly suitable to the possible evolution of research and standardisation on the numerical characterisation of the different factors impacting the components service life. Our research and development activity is focused on three complementary tools to answer this task: (Failure Mode Effect and Criticality Analysis method; Product service life data base (including material service life data base) and associated tools compatible with ISO 15686; Data fusion). The purpose of this paper is the presentation of the efficiency and the complementarities of these tools and methods. Indeed, by the use of these tools, we are performing a multi scale analysis of building product and component (at the scale of the product itself (data base, data fusion and F.M.E.A.), at the scale of the component (data base, data fusion and F.M.E.A.) and of the material (data base, data fusion). Further more, data base and data fusion are much more efficient at the material scale, especially if the material is subjected to one identified load. The F.M.E.A. allows us to evaluate the service life of the product from the service life values of the different materials loaded by one unique identified solicitation. We can thus use this result of service life at the scale of the product to feed the product service life data base. This study highlight the need of performing F.M.E.A. analysis to obtain reference service life without performing long, complicated and onerous tests at the scale of the product, and to collect these data into a data base which take into account all the information accompanying the service life value. A project of F.M.E.A. software is on his way in order to facilitate the analysis.

KEYWORDS

Service life assessment, F.M.E.A., ISO 15686, data fusion

10DBMC International Conférence on Durability of Building Materials and Components LYON [France] 17-20 April 2005

1 INTRODUCTION

The aim of this article is to propose a methodology to estimate the service life of products and components by the use of several specific tools developed by the CSTB and suitable with ISO 15686 standards application. Indeed, the development of the ISO 15686 standard is generating new approach for estimating the service life of product and component. We will first shortly introduce this method. This approach is highlighting the lack of building products and components service life data. The second paragraph will describe the interest of using F.M.E.A. to collect reference service life data (RSL). Finally, we will present the tools developed in CSTB, and a general methodology which is using these tools to estimate service life data of building products in a specific case study. This methodology intends to be an operational tool for construction stakeholders, (Designers, managers, manufacturers…) compatible with recommendation of ISO 15686. Furthermore, according to the fact that knowledge in term of characterisation and quantification of factors impacting the service life is limited, this methodology try to propose a sustainable solution, which take into account the evolution of knowledge in this domain.

2 "FACTOR METHOD": ISO 15 686

2.1 Presentation of the method

The factor method is a simple system to estimate service life when there is limited knowledge of long- term performance of components. (ISO 15686-1, ISO 15686-2 and ISO DIS 15686-8) This method is using the following equation to estimate service life of building product and component:

ESLC = RSLC × FactorA× FactorB × FactorC × FactorD × FactorE × FactorF × FactorG (1)

Where:

RSLC is the reference service life of component ESLC is the estimated service life of component

The estimated service life of a component (ESLC) is a function of the reference service life (RSLC) and a number of factors:

A: Material/Component factor B: factor C: Workmanship factor D: Internal environment factor E: External environment factor F: In-use factor G: Maintenance factor

This formula acts as a reminder of what should be taken into account when estimating service life.

TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

2.2 Reference Service life

Our capability to Collect RSL is the key issue for the performance of the factor method. A step further in the factor method is the calibration of the influence of the factors on the ESL. At this stage the factor are supposed to be representative of the difference between the reference in use condition set and the condition of the case study. According to the fact that we can use several RSL for the same product, it is not possible to propose a unique quantified scale for each factor according to associated conditions. This highlight one of the interest of the notion of intrinsic reference service life (IRSL) describe in paragraph 4.2.

3 INTEREST OF MULTISCALE ANALYSIS TO COLLECT RSL

The behaviour of material performance within time under one identified sollicitation has been largely studied, and good models contribute to a reliable evaluation of service life of materials when environmental and external solicitations are well defined. This evaluation became more complex with the coupling of the external and environmental solicitations. Therefore there is still a strong need of experimentation in order to improve the models for complex and coupled cases study. If the elaboration of models seems to be possible at the scale of materials, it is not as simple at the product scale. The evaluation of products and components service life is a difficult task, due to the complexity of the components and the lack of knowledge on the failure scenarios of these components.

Concerning experimentation:

Taking into account that products have several functions, constitutive materials, and are submitted to complex environment, we do believe that realisation of accelerated ageing difficult is tricky. The size of the sample is also an obstacle for laboratory testing. The number of constitutive materials of the product create problem to accelerate ageing without changing any ageing mechanism

Concerning modelling:

Numerical modelling of product ageing assumes to be able at least to perform thermo- hydro-mechanical modelling of heterogeneous components. It also imply to model interaction between materials.

Furthermore, the feedback management is not efficient, and the knowledge is not enough collected and accessible.

According to this context, we have developed a tool which allows us to get service life information at the scale of the building product by using the one at the scale of the material. This tool is based on F.M.E.A. (Failure Mode and Effect Analysis) (Talon [2004], Lair [2002]). Using first structural and functional analysis and then a Failure mode and effect analysis, this analysis provides us all the potential failure scenarios of the product in use in its environment. Failure event graph is then drawn for a better exploitation of the result. (Figure 1). By the use of this graph we can define for each scenario, several stage of degradation implying a material under a single solicitation.

TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

Initial Step 1 Step 2 Step 3 Step 4 Failure of Failure of State componant product

- Deviation of temperature - Thermal shock - Shocks - Pressure - Wind

Cracking Breaking Failure Glazing of absorber of absorber of absorber

Liquids Pollutants Failure Vertebrates Corrosion Holing Failure of the Hail of coating of coating of coating product Coating (Absorber) Breaking of coating

Figure 1: Failure event graph used to present result of F.M.E.A. Illustration with two components (Glazing and Coating) of a solar panel. The determination of the kinetic of each of these degradations allows us to calculate the service life of each scenario. (Figure 2 and equation 2).

ni SL[][Scénario(i) = ∑ SLmat (mat( j), perf ( j)) ] (2) j=1

Where:

• i is the number of the studied scenario

• ni: successively degraded materials in scenario number i (stages of the scenario)

Considering the service life of each scenario, it is possible to obtain a service life of the product. Indeed, the characterisation of the criticality of each scenario allows us to select representative scenarios from the initial exhaustive list. Then, the determination of duration of these scenarios gather a service life data for the performance associated to the scenario (equation3):

SL[ product, perf ( j)] = min(SL[Scenarioselected (i( perf ( j)))]) (3)

If we consider the need for characterisation of durability and\or service life of the product. The task is largely simplified by the used of F.M.E.A.. Indeed, the challenge was to gather Service life data for complex products (several materials, complex geometry and environments, several possible uses and needs), it is now to use data of degradation kinetic of a material under one identified solicitation to provide the service life of this material under this solicitation according to the associated performance. The figure 2 illustrate how to get a service life value for a product in a specific case study, by staying at the scale of the product (factor method, ISO 15686), and by the use of F.M.E.A. which is using the information at the scale of the material.

It can be a suggestion to evaluate service life if the "factor method" is not applicable by lack of RSL data or knowledge on the factors.

TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

Figure 2: Methodology of service life estimation at both product and material scale. ISO 15 686 propose a direct methodology at the scale of the product (the "factor method") using the notion of RSL. The F.M.E.A. is using the service life data at the scale of the material to provide ESL which can be capitalize as a RSL for other case study.

4 PROPOSED TOOLS AND METHODOLOGIES TO ESTIMATE SERVICE LIFE OF PRODUCTS AND COMPONENTS

4.1 Reference Service Life Data base

The ISO 15 686 part 8 "Reference Service Life" provides guidance on the provision of reference service life (RSL) for use in the application of the "factor method". Methodology such as F.M.E.A. is recognised as a method to provide RSL data by the ISO 15686 standard. We have developed a data base which is able to collect RSL and its associated information (Source, factors A to G, data quality, references…). This SQL data base is accessible on the CSTB intranet, but could be proposed soon on a website. Some of the specificities of this data base are the following: - It includes fields for Numerical Unit Spread Assessment Pedigree (N.U.S.A.P.) in order to perform data fusion (Lair et al [2001]) - It includes fields to collect information on A to G factors according to the data providers knowledge, but also includes a second data base which proposes a factors format for the product (according to the state of the art, or standardization on that product), and the initial information on the factors can be updated according to these formats. Initial information is kept and the proposed format can be updated too.

4.2 Tools to optimize the factor method

4.2.1 Suggestion of Intrinsic Reference Service Life (IRSL) notion

We suggest to use the notion of Intrinsic Reference Service Life (I.R.S.L.).The idea is to artificially use the same "intrinsic reference in use conditions" for all the IRSL value. This is applicable if we are able to quantify the impact of the factors on the service life. When Collecting the RSL corresponding to the reference in use conditions, we back calculate an IRSL, corresponding to the following equation TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

4, where AR, BR,CR, DR, ER, FR and GR are the factor corresponding to the application of the factor method when replacing RSL by IRSL and ESL by RSL.

RSL IRSL = (4) AR × BR × CR × DR × ER × FR × GR

Making the assumption that the factors can be quantified, each value of collected RSL is then able to provide an IRSL value.

4.2.2 Interest

The interest of using the IRSL value is the efficiency of the data fusion tool (Lair et al [2001]) on all the calculate IRSL to get a single CIRSL on each product (Consolidated Intrinsic Reference Service Life)

Figure 3: Theoretical illustration of the interest of IRSL and data fusion. This sketch highlights the interest of the fusion on a IRSL set of data in comparison with a classic RSL set of data. The failure indicator curve is more exploitable with the IRSL set; indeed we observe that the failure indicator curve slope of the IRSL is nearly constant when the one of CIRSL is largely bigger between 60 and 80 years. The consensual value of a service life is thus easier to extract. To each service life, we assign a mass m (which is similar to a probability assignment) according to the quality of the basic durability data. Thus m is a quality indicator according to the origin of the data. Theoretical structures, data input, set-up conditions… are quality criteria (Funtowicz & Ravetz, [1990]) valued on a five level scale.

The concept of IRSL and CIRSL is compatible with the RSL data recommendations of ISO 15686-8.

4.3 General methodology for service life estimation

TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

The needs for estimating service life of building components and products can have several origins such as whole life costing analysis, life cycle analysis, Performance Based Building (PeBBu Network), PFI…. The tools developed (RSL data base, Data fusion and the analysis based on F.M.E.A. and associated tools) are complementary and supply a real scientific gain for the evaluation of service life by the use of the factor method.

The classical case study is the estimation of a product service life used in a constructed assets or a building where we precise information on the in use conditions as well as on material quality and conception level, protective coating etc… According to these data and to the specific functional requirements due to building context (Sustainable building, Whole life costing, Life cycle analysis…) we are able to characterize the functions of the product by leading Functional analysis or F.M.E.C.A. (which is integrating a functional analysis).

Figure 4: General methodology for service life estimation of a specific case study.

To estimate a service life of this product it is necessary to identify the "critical" function and his associate "critical" performance". Indeed, we are estimating the service life of a product for one given function the performance of which is the reason of the failure. At this stage we can start the process of service life estimation according to the "factor method".

This first way is corresponding to the case where the factors are not quantified and where no specific study is or has been led. One can see the interest of a reference service life data base and selection tool to choose the appropriate reference service life for the case study. Indeed, the "factor method" efficiency is strongly link to the ability of supplying a reference service life with reference in use conditions as close as possible to the one of the case study; in that case the corrective factor are less influential and the uncertainty on estimated service life value is smaller. But there is an interest to perform specific study on the product to determine and quantify the value of the factor impacting on the service life. The aim is to obtain a scale of every factor which is based on the intrinsic reference in use conditions. TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier

10DBMC International Conference on Durability of Building Materials and Components LYON [France] 17-20 April 2005

If such analysis is done, we had the opportunity to calculate intrinsic reference service life (IRSL) and to realize a more efficient data fusion in order to obtain a consolidated intrinsic reference service life (CIRSL). The CIRSL can be reused for any case study concerning this product and provides a calculation of the estimated service life without using subjective corrective factor. However this theoretical methodology needs to be validated by practical application

5 PERSPECTIVES AND CONCLUSION

Providing RSL data is the first step for service life planning. We are focusing our research in the use of data at the scale of the material to estimate service life of component or product. The F.M.E.A. is a very powerful tool to realize this task. Indeed, it explains the behaviour of products in their environment, by giving the detail of each step of each failure scenario. The scenario is then scattered in several well defined degradation.

6 GLOSSARY

ESLC: Estimated Service Life of Component RSLC: Reference Service Life of Component IRSL: Intrinsic Reference Service Life RSL: Reference Service Life CIRSL: Consolidated Intrinsic Reference Service Life FMEA: Failure Mode Effect Analysis FMECA: Failure Mode Effect and Criticality Analysis

7 BIBLIOGRAPHY

ISO 15686-1,2000, Building and constructed asset - Service life planning – Part 1: General principle ISO 15686-2,2001, Building and constructed asset – service life planning – Part 2: service life prediction procedures ISO DIS 15686 -8, Building and constructed asset – service life planning – Part 8: Reference service life

Talon, A. Boissier, D. Chevalier, J.-L. Hans, J. 2004, "A methodological and graphical decision tool for evaluating building component failure", Proc. CIB World Building Congress 2004, Toronto, Canada, 2 7 may 2004.

Lair, J., Chevalier J.L. (2002). "Failure Mode Effect and Criticality Analysis for risk analysis (design) and maintenance planning (exploitation)". 9th Durability of Building Materials and Components (9th DBMC) Brisbane, Australie – 17/21 Mars 2002.

Lair, J., Chevalier J.L., Rilling J. (2001). "Operational methods for implementing durability in service life planning framework. CIB World Building Congress". WELLINGTON, (April 2001).

Funtowicz, S. O. and Ravetz J.R. (1990). "Uncertainty and Quality in Science for Policy". Dordrecht (The Netherlands). Ed.: Kluwer Academic Publishers. 229 pages.

TT4-90, Sustainable tools and methods for estimating building materials and components service life., J.Hans, JL Chevalier