You are on page 1of 4

Material

E A

A MPTIAC

Dr. A.K. Kuruvilla


IIT Research Institute Huntsville, Alabama

LIFE PREDICTION AND PERFORMANCE ASSURANCE OF STRUCTURAL MATERIALS IN CORROSIVE ENVIRONMENTS

Introduction Reliable estimates indicate that corrosion of metals costs the US economy almost $300 billion per year or about 4% of the GNP. Corrosion is a pervasive problem that affects almost all sectors of the economy including transportation (bridges, automobiles, aircraft, ships, pipelines, etc.), energy (power plants, petrochemical industries etc.) and defense (naval applications, aging aircraft & weapon systems, etc.). Although the complete elimination of corrosion is virtually impossible, it is believed that a third of this cost is avoidable based on the present status of corrosion science, the state-of-the-art in industrial practice and the continual developments in associated technologies. Broader application of corrosion resistant materials, improvements in corrosion prevention practice, effective detection and repair and an investment in corrosion related research are the necessary ingredients of any program that targets minimization of corrosion related costs. Given the ubiquitous nature of corrosion, life prediction and performance assurance is vital to the cost-effective management of components and systems that encounter corrosion in service. Over - conservative design and operation has been the preferred approach to reducing failures and ensuring the designed performance in corrosive environments. However, the cost penalty associated with this approach is continually directing efforts toward more effective life prediction and performance assurance. The Essential Elements Any life prediction effort related to corrosion of structural components should comprise, for a minimum, certain essential elements: 1) 2) 3) 4) 5) 6) 7) Definition of the environment Definition of the material / microstructure Definition of the stress state Understanding of the mode of corrosive attack Definition of failure Field data (if available) Results of accelerated testing. The following is a brief overview/discussion of the influence of these factors on life predictive capability and assurance of trouble free operation. Environment: All corrosion involves reaction of the metal with the environment and hence, a definition of the chemical environment with the metal is of primary importance. Typically, the factors that control reactivity include the electrochemical potential, the pH, the passivating characteristics, temperature and the flow rate of the aqueous environment. The chemical species in the environment clearly affect most of these factors. In defining the environment, very often, the localized chemical characteristics are more important than the bulk chemistry. Therefore, all possible environmental variations must be considered when dealing with a structural component. For example, if the component design results in a crevice or an occluded cell, the local chemistry will be quite different from (and often more aggressive) that of the bulk chemistry. As a result, the corrosion characteristics will also be significantly different. The environment is typically described using a potential-pH diagram that is more commonly known as the Pourbaix diagram. These plots are ideal for displaying the potential corrosion regimes given the environmental conditions. It has been pointed out in the literature that it is also important to superimpose a passivating parameter on these plots in order to describe the effect of particular chemical species on the passivation of metal surfaces under a given set of conditions. Additionally, if a component experiences different environments at different points, the potential-pH range of exposure can be well described on such plots. Note that it is equally important to document changes in the environment with time, if any, to be able to predict the nature and rate of corrosion. Material / Microstructure: The reactivity of a metal to any given environment is dictated by its chemical composition. Since reactions occur at metal surfaces, the surface composition is often

AMPTIAC is a DoD Information Analysis Center Sponsored by the Defense Technical Information Center and Operated by IIT Research Institute

Material
E A

E
Prediction involves assessing the life of the component in the environment whereas assurance deals with the level of confidence with which performance is guaranteed through the design life of the component. The latter therefore involves inspection, monitoring and maintenance functions in addition to the initial prediction. The key to predicting life is to quantify the qualitative information as discussed in the previous section dealing with the environment, material etc. This is accomplished by developing accurate theoretical models that can be verified with real life data. Another way of predicting life is to use the accelerated test data or service data (if available) to generate empirical relationships with which the time to failure can be estimated. Theoretical Modeling The development of sound theoretical models of failure is a good starting point for life prediction. The accuracy of the model and its range of applicability are, therefore, extremely important. follows: tf = c-n, tf = geQ/RT, tf = a[H+]-m. (1) (2) (3) The better known dependencies of the time to failure are represented in Equations 1 - 3 as

extremely important. The effect of major alloying elements is usually pronounced in most alloy systems. However, minor alloying elements and impurities also have a significant effect on the reactivity of the metal surface. If the grain boundary composition is different from that of the bulk, then the nature of corrosion will be different at the grain boundary than in the matrix. In addition to the composition, other microstructural factors such as yield strength, amount of deformation, the composition and distribution of second phases and surface conditions are also significant to corrosion characteristics of the metallic material. Stress: Integral to the environment, stresses assist the degradation of metals under corrosive conditions. Loads such as bending, tension and torsion that can be analytically defined are often not the most critical contributors to corrosive damage. Instead, it is the stresses such as the residual stress and stresses resulting from the expansion of corrosion products that cause the most damage. The stress situation at the surface of the metal that encounters the corrosive environment must therefore be accurately defined in order to be able to predict the extent of corrosion. Mode of Corrosion: As the metal interacts with the environment, it results in a corrosion path. The morphology of this path is referred to as the corrosion mode. Understanding the mode of corrosion is vital to any corrosion life predictive capability. Examples of the modes of corrosion are general corrosion, pitting corrosion, intergranular corrosion and stress corrosion which can be either transgranular or intergranular in its mode of failure. The mode of corrosion can be superimposed on the potential-pH diagrams to generate maps that predict the type of corrosion for a given set of environmental conditions. Definition of Failure: Failure can be defined in several ways by establishing the necessary criteria under the given operating conditions. Failure data are often characterized using statistical distributions such as the Weibull distribution. Field data: If available, service data from the field can be extremely valuable to performance assurance and life prediction. Ideally, failure data could be used to generate the representative statistical distribution from which parameters indicative of the potential and probability for failure can be extracted. Accelerated Testing: Components are generally designed for a specific life span. However, the assurance of economical, reliable and safe performance for the design life and extended times require past performance data, conservative design and results of accelerated tests. Temperature, stress, ripple loading, thermal cycling, variation of the environmental composition etc. are some of the variables that can be employed in accelerated tests. The key is to identify and employ those variables that yield results in a short period of time that would correspond to real life conditions. Life Prediction and Performance Assurance

The combined equation that represents an overall dependency for timeto-failure would then be tf = d-n eQ/RT [H+]-m, where: tf

(4)

= = = = = = =

time to failure stress exponent of stress thermal activation energy absolute temperature gas constant hydrogen ion activity exponent of hydrogen ion activity

n Q T R m

[H+] =

c, g, a, d are constants. Equation 2 is the well-known Arrhenius form of the equation that represents the dependency of the rate of a chemical reaction (corrosion in this case) on temperature. In electrochemical reactions, the electrochemical potential is also included and the equation takes on the following form:

A MPTIAC
A D VA N C E D M AT E R I A L S
AND

PROCESSES TECHNOLOGY

Another example of modeling crack velocity under stress corrosion cracking is the following equation:

r = B exp Q* zFE RT
where: r B Q* z F

(5) CV =

Qf ct M

fzFd

(7)

where: = = = = = = = rate of electrochemical reaction electrochemical rate constant chemical activation energy in the electrochemical charge on the activated species Faradays constant transfer coefficient electrochemical potential CV Qf = = = = = = = = crack velocity charge density crack tip strain rate strain to rupture the crack tip film valency of the solvated species atomic weight of the metal Faradays constant density of the metal.

reaction

ct f
z M F d

Q* zFE = electrochemical activation energy. It is important to note that if there is a mechanically related process operating in conjunction with the electrochemical process, stress would also be a factor that influences the activation energy term. In such instances, corrosive degradation may not be adequately represented by mere addition of the stress and chemical reaction terms. The complexity of service conditions often precludes theoretical models from adequately representing actual performance. However, theoretical models serve as good starting points for analysis and life prediction.

There are several expressions available in the literature to calculate the crack tip strain rate under static, monotonically increasing or cyclic loading conditions. Models similar to those cited above have been derived for different modes of corrosive degradation such as stress corrosion cracking, corrosion, fatigue, etc. 2.) Statistical Concepts There are two approaches to applying statistics to corrosion related failure. In the first, corrosion probability concepts are applied to failure. For example, assuming a Poisson distribution for the generation of pits (in pitting corrosion), the probability of the formation of a pit can be computed as it varies with the surface area or time in the corrosive environment. Experimental results for pitting induced failure have been found to correlate well with theoretical predictions in some studies. In the second approach, the statistical distribution of service failure data can be used to generate parameters that are indicative of the potential for failure. The Weibull distribution, for example, which was developed for

Modeling Experimental Data: 1.) Empirical Relationships Experimental data (either from service or from accelerated testing) are typically modeled either empirically (with some basis in theoretical modeling) or statistically. An example of the former is the correlation of crack growth rate to crack tip strain rate under stress corrosion cracking, given by the following equation:

Vt = A (ct)n, where: Vt = = average crack velocity crack tip strain rate

(6)

characterizing fatigue data has been found to be useful for characterizing corrosion failure data as well. The effect of environmental variables on the parameters of the distribution can be modeled to enable performance assurance of the material under varying environmental conditions. A further enhancement to this approach is to compute the overall probability of failure of a component based on the probability of failure under different modes of corrosion that are operative simultaneously.

ct

A, n depend on the crack tip material and environmental compositions.

Summary There are numerous factors that affect the corrosion of structural components. A thorough understanding of the localized corrosion environment (at the corrosion location) and the material and its microstructure is necessary to begin addressing the problem of life prediction. A qualitative prediction may be readily made using a potential-pH diagram that incorporates passivation characteristics. Such maps can delineate the conditions under which the different modes of corrosion may be operative, thus enabling the requisite changes in the environment, the material or both to extend life or prevent premature failure. Quantitative predictions are largely based on either 1) theoretical modeling that is verified by experimental data, 2) empirical models that are derived from experimental data or field data or 3) statistical distributions of failure data and an understanding of the parameters that are extracted from such distributions. Although the overall approach to life prediction and performance assurance can be generalized, the availability of experimental data, models, service data, and understanding the precise degradation conditions dictates a situationspecific approach to this important problem. References
The following is a list of publications that discuss the various aspects of life prediction under a variety of corrosion conditions. [1] Life Prediction of Corrodible Structures, Vols. 1 and II, NACE, Houston, TX, 1994. [2] Materials Considerations in Service Life Predictions, R. P. Wei and D. G. Harlow, Applied Mech. Review, Vol. 46, may 1993. [3] Modeling and Life prediction of Stress Corrosion Cracking in Sensitized Stainless Steel in High Temperature Water, P. L. Andresen and F. P. Ford, Predictive Capabilities in Environmentally Assisted Cracking, ed. R. Rungta, ASME, NY, 1985, pp. 17-39. [4] Corrosion Damage Function Interface Between Corrosion Science and Engineering, D. D. Macdonald and M. Urquidi-Macdonald, Corrosion, Vol. 48, No. 5, 1992, pp 354-367. [5] A Dominant Flaw Probability Model for Corrosion and Corrosion Fatigue, D. G. Harlow and R. P. Wei, 12th Int. Corrosion Congress, NACE, Houston, TX, 1993. [6] Development and Use of a Predictive Model of Crack Propagation in 304/316L, A533/A508 and Inconel 600/182 Alloys in 288 oC Water, F. P. Ford and P. L. Andresen, Environmental Degradation of Materials in Nuclear Power Systems-Water Reactors, eds. G. J. Theus and J. R. Weeks, TMS, Warrendale, PA, 1988, pp. 789-800. [7] Application of Accelerated Tests to Service Life Prediction of Materials, ASTM STP 1194, eds. G. Cragnolino and N. Sridhar, American Society for Testing of Materials, Philadelphia, Pa, 1993. [8] Development and Application of Corrosion Mode Diagrams, R. W. Staehle, Fundamental Aspects of Stress corrosion Cracking, eds. S. M. Bruemmer, E. I. Meletis, R. H. Jones, W. W. Gerberich, F. P. Ford and R. W. Staehle, TMS, Warrendale, PA, 1992, pp. 457-491.

Note: A comprehensive report that details the state-of-the-art in this topic area is being developed for publication in the Fall of 1998.

A D VA N C E D M AT E R I A L S
EMAIL:

AND

PROCESSES TECHNOLOGY

A MPTIAC
PHONE: FA X : 315.339.7117 315.339.7107

http

amptiac@rome.iitri.com ://rome.iitri.com/amptiac

AMPTIAC is a DoD Information Analysis Center Sponsored by the Defense Technical Information Center and Operated by IIT Research Institute

You might also like