LRFD Metal Loss and Service-Life Strength Reduction Factors for Metal-Reinforced Systems (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

51 This appendix describes the history and pertinent details of metal loss models that have been proposed for estimating sac- rificial steel requirements for MSE reinforcements. Most state highway agencies use some form of the AASHTO specifica- tions for the design of MSE walls. Therefore, the AASHTO metal loss model is used in this study to compute nominal sacrificial steel requirements that serve as a basis for calibra- tion of resistance factors for LRFD. This appendix will describe earlier metal loss models and corresponding data sources leading to the development and adoption of the AASHTO model. These models include the Darbin/Romanoff Models and the Stuttgart Model. Galvanized steel reinforcements are most often employed for the construction of MSE walls, particularly with respect to transportation-related projects. However, the behavior of plain steel is also of interest, as this is compared or related to the loss of steel after the zinc is depleted from the surface. Fur- thermore, the AASHTO metal loss model only considers the use of galvanized reinforcements, and other models need to be identified to consider the behavior of plain steel (i.e., not galvanized). Metal loss models used to estimate sacrificial steel require- ments for MSE are empirical, and therefore it is important to describe the data sources considered in their develop- ment. Although fill characteristics are important consider- ations, these models do not explicitly relate fill characteristics in terms of their electrochemical properties to corrosion rates. In general, the models consider the effects of time on corrosion rates and apply to particular ranges of fill char- acteristics. Care must be exercised when using these mod- els to be sure that fill materials have electrochemical properties within the range for which the models are intended. Metal loss models that may be applied to fill materials, that do not necessarily meet AASHTO require- ments, are also identified. Darbin/Romanoff Model Romanoff (1957) describes 47 years of data collected by the U.S. National Bureau of Standards (NBS) from extensive monitoring of metal samples buried in situ. In general, the corrosion rate was observed to be greatest during the first few years of burial, subsequently decaying to a steady but signifi- cantly lower rate. Romanoff suggested the following expo- nential equation to predict the amount of general corrosion at some time (t) after burial: In Equation (A-1) x is the loss of thickness in the material at time, t, and K and n are parameters that are soil and site depend- ent. In Equation (A-1) lower case “x” describes metal loss that may include zinc and steel for galvanized elements. Capital “X” is used for other metal loss equations in this appendix to denote loss of steel subsequent to depletion of zinc. Comprehensive as it was, less than 10% of those data from the NBS study came from free-draining granular soils such as those used in MSE walls, and even less of these data came from galvanized steels. Darbin et al. (1988) addressed this shortcom- ing during a 20-year study not only to evaluate the corrosion of metallic earth reinforcements in typical MSE wall backfill, but also to identify the soil parameters that determine the kinetics of the corrosion process. Using the form of Equation (A-1), Darbin et al. (1988) proposed that metal loss of galva- nized steel could be described with a constant exponent “n” equal to 0.65, and coefficient “K” depending on soil aggressive- ness (K = 25 μm/year for soils with resistivity ρ ≥ 1,000 Ω-cm and K = 20 μm/year for soils with ρ ≥ 3,000 Ω-cm). Maximum corrosion rates and loss of reinforcement ten- sile strength from corrosion may be estimated by multiplying the general corrosion rate obtained from Equation (A-1) by a factor of 2 (Elias, 1990). This factor is applied to the metal x Ktn= ( )A-1 A P P E N D I X A Details of Metal Loss Models

loss of base steel subsequent to depletion of zinc from the sur- face. Thus, a factor of 2 is applied to the Darbin Model to con- sider strength loss as follows: for galvanized elements: In Equation (A-2), X is loss of steel (base metal) in units of μm, and tf is service life in years. Loss of base steel occurs sub- sequent to depletion of the zinc coating, and zi is the initial zinc thickness. Equation (A-2) is applicable to the range of fill conditions representative of MSE wall construction that exhibit ρmin greater than 1,000 Ω-cm. Elias (1990) proposed the following model for plain steel reinforcements which also has the same form as the original equation proposed by Romanoff [Eq. (A-1)]: Data reviewed for Equation (A-3) are based on the NBS data set for plain steel and include a wide range of fill conditions, many not meeting the stringent electrochemical requirements for MSE fills. However, given the scatter inherent to measure- ments of fill properties and corrosion rates for plain steel, Equation (A-3) is used as a conservative estimate of metal loss in fills that meet MSE fill requirements, but not by a wide mar- gin. Equation (A-3) also includes a factor of 2 to consider the maximum metal loss. Although corrosion rates for both galvanized steel and car- bon steel clearly vary exponentially with respect to time, sim- ple models involving linear extrapolation have been proposed and are considered valid (Elias, 1990) over the limited time frame from which metal loss measurements of earth re- inforcements were available (<20 years). Given this limited time frame, most observations of metal loss for galvanized reinforcements are observations of the loss of the zinc coating, not the carbon steel (i.e., steel was not exposed during the monitoring period). The following models, including the Stuttgart, Caltrans, and AASHTO models, are linearized forms of the Romanoff/Darbin equation. Stuttgart Model Rehm (1980) proposed an alternative piecewise linear model for describing metal loss. The longevity of the zinc coat- ing is considered using a bilinear model such that the rate of X yr t fμ μ m m A-3( ) = ×80 0 8. ( ) if then m m t z X yr f i> ⎛⎝⎜ ⎞⎠⎟ ( ) = ⎛ ⎝⎜ ⎞ ⎠⎟ ×25 50 1 54. μ μ t yr z t z f i f i 0 65 1 2 25 . . ( ) ( ) − × ( ) ≤ ⎛⎝⎜ ⎞⎠⎟ μm A-2 if 54 0then mX μ( ) = zinc consumption is greatest during the first 2 to 4 years, fol- lowed by a significantly reduced rate. Therefore, the reduced rate considers passivation of zinc that occurs in backfill soils typical of MSE wall construction. Steel consumption is con- sidered to begin after the zinc layer is consumed, but at a rate observed from samples of plain steel appropriate to the age of the reinforcements (i.e., rate of corrosion for steel that has been in service for more than 2 years). This model was the basis for the sacrificial steel requirements for galvanized rein- forcements recommended by Task Force 27 (1990); AASHTO (2002a) later adopted a different, more conservative, piece- wise linear model as described later in this appendix. Metal loss models are proposed considering galvanized or plain steel reinforcements and fill materials with low and high salt contents. Low salt contents are described as materials with 4.5<pH<9.5, ρ > 1,000 Ω-cm, chloride content less than 50 ppm, and sulfate content less than 200 ppm. For this con- dition, the Stuttgart model is as follows: for galvanized elements: for plain steel elements: For fill materials that are saturated with chloride or sul- fate concentrations greater than the threshold values, the Stuttgart model is as follows: for galvanized elements: for plain steel: AASHTO Model According to AASHTO, MSE fill must comply with the fol- lowing electrochemical criteria: • pH = 5 to 10 • Resistivity ≥ 3,000 Ω-cm, X yr yr t yr yr fμ μ μ m m m A-7( ) = × + −( )×80 2 2 12 ( ) X yr t yr z yr f iμ μ μμm m m m ( ) = × − − −( ) ⎛ ⎝ ⎜⎜⎜ ⎞ ⎠ ⎟⎟12 3 51 2 ⎟ yr ( )A-6 X yr yr t yr yr fμ μ μ m m m A-5( ) = × + −( )×45 2 2 9 ( ) X yr t yr z yr f iμ μ μμm m m m ( ) = × − − −( ) ⎛ ⎝ ⎜⎜⎜ ⎞ ⎠ ⎟⎟⎟ 9 2 12 2 yr ( )A-4 52

• Chlorides ≤ 100 ppm, • Sulfates ≤ 200 ppm, and • Organic content ≤ 1%. The fill requirements are intended to control corrosion potential with fills that are between noncorrosive and “mildly” corrosive. The AASHTO metal loss model defines the follow- ing rates at which first zinc, then steel, will be lost from the MSE reinforcement section: • Loss of zinc (first 2 years): 15 μm/yr; • Loss of zinc (to depletion): 4 μm/yr; and • Loss of steel (after zinc depletion): 12 μm/yr. Using the AASHTO Model the steel loss per side (X) in μm/yr for a given service life, tf, and initial thickness of zinc coating, zi, is computed as Both the Darbin/Romanoff and the Stuttgart Models con- tribute to the basis of the AASHTO Model. Table A-1 identi- fies sources of data associated with each. These data sources include the NBS studies from metal samples that were buried within a wide range of fill conditions at sites located through- out the United States; and from carefully controlled labora- tory tests conducted in France, specifically with regards to MSE reinforcements. Laboratory studies included electro- chemical test cells and burial boxes. The electrochemical test cells were assembled using relatively small (compared to bur- ial boxes) plastic tubes containing specimens of reinforce- ment surrounded by soil. Electrodes were sealed into the ends of the tubes, serving as reference and counter electrodes, to facilitate measurements of corrosion rates at frequent inter- vals. Compared to the electrochemical test cells, burial boxes incorporated representative specimens of MSE reinforce- ments and conditions that more closely resemble field instal- lations. The burial boxes employed weight loss measurements X yr t yr z yr f iμ μ μμm m m m ( ) = × − − −( ) ⎛ ⎝ ⎜⎜⎜ ⎞ ⎠ ⎟⎟12 2 30 4 ⎟ yr ( )A-8 that could be related to metal loss, and corresponding corro- sion rates averaged over longer time intervals. Thus,theAASHTOmodelconsiders a variety of data sources, each with its own set of strengths and limitations. However, results from these different data sources compare reasonably well. The outstanding limitations of each data source involve a lack of data to document the corrosion/metal loss of the base steel subsequent to depletion of zinc from the surface. Thus, similar to the Stuttgart model, the AASHTO model considers steel consumption to begin after the zinc layer is consumed, but at a rate observed from samples of plain steel appropriate to the age of the reinforcements. Figure A-1 illustrates the comparison between the Stuttgart and AASHTO models. The corrosion rates for the zinc coating during the first 2 years of service and for steel subsequent to depletion of zinc roughly correspond to the Stuttgart model that applies to higher salt contents (e.g., chlorides in excess of 50 ppm). However, the corrosion rate for zinc after 2 years in service of 4 μm/year is twice the value of 2 μm/year from the Stuttgart Model. The rationale for the use of the higher corro- sion rate may be understood by examining the comparison with the Darbin model as depicted in Table A-2. 53 MODEL DATA SOURCE MAXIMUM Stuttgart NBS NO Darbin Controlled conditions - electrochemical test specimens & samples buried in soil box YES AASHTO Stuttgart & Darbin YES Table A-1. Summary of data sources for metal loss models for MSE reinforcements. STUTTGART CR = 17 μm/yr for t < 3 CR = 2 μm/yr for t > 3 PLAIN STEEL CR = 80 μm/yr for t < 2 CR = 12 μm/yr for t > 2 AASHTO CR = 4 μm/yr for 2 < t < 16 CR = 15 μm/yr for t < 2 CR = 12 μm/yr for t >16 ? with zi = 86 μm Factor of 2 ?? ZINC Figure A-1. Comparison of AASHTO and Stuttgart Models.

Table A-2 indicates that metal losses computed with the AASHTO model compare reasonably well with the Darbin model. The differences depend on time (service life), and the two models render nearly the same metal loss when t = 64 years. Apparently, using a higher corrosion rate for zinc of 4 μm/year renders metal loss consistent with the Darbin model, which directly considers applying a factor of 2 to consider maximum metal loss. Note that although the AASHTO specifications require fill with ρ > 3,000 Ω-cm, the basis of the models (both Stuttgart and Darbin) are referred to fill with ρ > 1,000 Ω-cm. Thus, the AASHTO model and corresponding specifications for fill are conservative. Caltrans-Interim Design Guide Based on the results from limited field studies, Caltrans (Jackura et al., 1987) has proposed design guidance for a wider range of reinforced fill conditions than those considered by AASHTO. Higher rates of metal loss are specified for comput- ing sacrificial steel requirements when reinforced fills that are more aggressive relative to corrosion are considered during design. These metal loss rates are based on limited data col- lected from MSE wall sites in California (Jackura et al., 1987), and use data available from the earlier NBS studies. Interim design guidance considered fill properties that include mini- mum resistivity more than 1,000 Ω-cm. However, current specifications used by Caltrans (2009) do not allow use of reinforced fill with minimum resistivity less than 2000 Ω-cm. The steel loss, X, for design life, tf, is described by the Caltrans interim model as: where C is the time for zinc depletion (years) assuming an ini- tial zinc thickness of 86 μm and K (μm/yr) is the corrosion rate of the base steel. Table A-3 provides values for C and K as functions of fill conditions. Current specifications used by Caltrans (2009) do not allow use of reinforced fill with minimum resistivity less than 2,000 Ω-cm. Specifically, the current Caltrans specification allows for backfill with a resistivity greater than 2,000 Ω-cm, a pH between 5.5 and 10, and maximum chloride and sulfate concentrations of 250 ppm and 500 ppm, respectively. Califor- nia considers these conditions by using a higher rate of metal loss in determining sacrificial steel and reducing the design life of the MSE wall to 50 years. Caltrans assumes that the zinc coating provides 10 years of service life for the specified mini- mum coating thickness of 2 oz/ft2 (86 μm per side). This is less than the 16 years of zinc life inherent to the AASHTO metal loss model. A corrosion rate of 1.10 mils/yr (28 μm/yr) is con- sidered to affect the base steel after the zinc has been consumed X t C yrs K yr fμ μ m m A-9( ) = − ( )( )× ( ) 54 Table A-3. Summary of parameters for Caltrans-Interim guidelines (Jackura et al., 1987). Darbin Model X (μm) = 50t0.65 - 2 (zi) (Includes Factor of 2) AASHTO Model X (μm) = (t-C) x 12 (μm/yr) (C= 16 years with zi = 86 μm) Computed Metal Loss X (μm) t (yrs) DARBIN AASHTO 50 464 408 64 574 576 75 655 708 100 825 1,008 Table A-2. Comparison of AASHTO and Darbin models. Fill Type K(m/yr) C (years) Neutral & Alkaline 28 10 Acidic 33 10 Corrosive 71 6 Select Granular 13 30 Notes: Neutral and alkaline: minimum resistivity > 1,000 Ω-cm and pH > 7. Acidic: minimum resistivity > 1,000 Ω-cm and pH < 7. Corrosive: minimum resistivity < 1,000 Ω-cm. Select granular soils are clean, free draining gravels with less than 5% fines and minimum resistivity > 1,000 Ω-cm.

and used to compute the sacrificial steel requirements. These corrosion rates account for the potential for localized corro- sion and pitting; that is, a factor of 2 relating the loss of tensile strength to idealized uniform corrosion rates is included. Caltrans specifications provide incentives to use select granular fill, which is a better quality fill with less than 5% fines and with plasticity index (PI) <6. Caltrans reduces the steel corrosion rate to 13 μm/yr for backfill meeting addi- tional requirements for select granular fill. For select granu- lar fill, lower resistivity and higher salt concentrations are allowed, but the allowable fines content is less compared to current AASHTO requirements. 55