This article examines how soil corrosiveness and pipeline corrosion rate data support integrity management programs. Pipeline integrity management programs address a wide range of threats to safe operation. Among these threats is external corrosion of buried steel pipelines. While external coatings and cathodic protection (CP) are typically used to mitigate external corrosion, corrosion can result from the combined presence of degraded coatings and inadequate CP.
External corrosion can be detected through a variety of assessment methods, including inline inspection (ILI) and external corrosion direct assessment (ECDA). The detection of corrosion provides pipeline operators with information regarding the areas vulnerable to eventual failure. However, it is the rate of corrosion that will influence how quickly those flaws can grow to critical size and cause failure. Estimating the rate of corrosion provides valuable input to several aspects of pipeline integrity management including the following applications:
- NACE Standard RP0502 (Anon. "NACE RP0502-2008 Standard recommended practice -- Pipeline External Corrosion Direct Assessment Methodology" NACE International, March 20, 2008) includes a requirement that the rate of corrosion be considered when setting re-inspection intervals for ECDA.
- Corrosion rate can be considered in prioritizing features detected during indirect inspections (i.e., above ground surveys) performed during ECDA. For example, among features having similar characteristics based on indirect inspections, those features located where the expected corrosion rates are higher can be prioritized for quicker direct examination.
- ILI results may show the presence of hundreds of external metal loss features. Often, many of those features have similar estimated failure pressures. Estimates of the expected corrosion rate can be used to prioritize those features for mitigation since higher corrosion rates will increase flaw severity and decrease the failure pressure more quickly.
SELECTING A METHOD
Having established the value of knowing the corrosion rate, a user is then faced with selection of an appropriate method of estimating a site-specific corrosion rate. When multiple high-resolution ILI inspections have been made over a period of years, the apparent change in anomaly size can be used to estimate the corrosion rate. In the absence of data from multiple ILI runs, the corrosion rate can be estimated based on the use of any of the following methods:
- Buried corrosion coupons,
- Corrosion probes, and
- Corrosion rate modeling using data derived from soil analysis.
Various models and procedures have been described that relate soil characteristics to the estimated corrosiveness of the soil. Soil corrosiveness models generally fall into two groups that include 1) methods that result in either relative ranking (i.e., "A" is more corrosive than "B") or qualitative estimates of corrosion (i.e., the expected corrosion is "moderate"), or 2) models that generate estimates of quantitative corrosion rates.
Models that result in qualitative descriptions of the corrosiveness can sometimes be useful in prioritizing locations for direct examination, but only if the qualitative results sufficiently discriminate among the various locations. For example, if the results all fall into the "mildly corrosive" category, which location should be inspected first? Quantitative estimates of corrosion are required if the data is to be used to set re-inspection intervals based on estimates of remaining safe life of the pipeline segment.
Perhaps the most basic soil corrosiveness model is the reliance upon soil resistivity as an indicator of soil corrosiveness. The rule of thumb is that corrosiveness generally decreases as soil resistivity increases. However, several technical resources show that the relationship is not consistent. For example, industry experts and researchers have noted that, "It has been difficult to find a statistically meaningful correlation between soil resistivity and corrosion rates in the studies examined." (Norin, M., Vinka, T.G., "Corrosion of Carbon Steel in Filling Material in an Urban Environment", Materials and Corrosion 2003, 54, No. 59, 2003); "Soil Resistivity is not a clear parameter for rapid assessments of soil corrosivity."
"No obvious systematic trends can be found (based on coupon weight loss). Resistivity of the soil may give some indication of the corrosivity, but may just as well lead to misinterpretation." (Nielson, L., "Microbial Corrosion and Cracking in Steel -- Results from A Screening of Soil resistivity Using Weight Loss Coupons" (Report 2 of 4) project ENS 1313/95, Technical University of Denmark, 1998); and "Soil resistivity is by no means the only parameter affecting the risk of corrosion damage. A high soil resistivity alone will not guarantee absence of serious corrosion." (www.corrosionsource.com/technicallibrary/corrdoctors/Modules/SoilCorrosion/Variables.htm).
Graphing resistivity vs. corrosion rate data shows that while a trend exists toward lower corrosion rate as resistivity increases, there is significant scatter in the data that prevents defensible estimates of corrosiveness to be consistently made on the basis of soil resistivity alone.
The ability to predict soil corrosiveness is improved by the use of multi-parameter models in which several soil characteristics are considered. Common examples include the AWWA model (Anon. "ANSI/AWWA C-105/A21.5 American National Standard for Polyethylene Encasement for Ductile Iron Pipe Systems" American Water Works association, Dec. 1, 2005) for predicting the need for various corrosion control options on ductile iron piping; and the Caltrans model (Anon. California Test 653 "Method for Estimating the Service Life of Steel Culverts," State of California Department of Transportation Engineering Services Center, Sacramento, California, November 1999) for estimating the lifetime of galvanized steel culverts.
The AWWA model assigns points to nine soil characteristics, depending upon the extent to which they are present. The points assigned to each characteristic are summed and the total score is a unitless number used to determine the expected corrosion severity and need for mitigation. In comparison, the Caltrans model uses a graphical approach to illustrate how the combined effect of soil resistivity and pH affect the expected life of a culvert of specified initial thickness. That model can be considered to be a quantitative model since the results are in units of years and when combined with the initial thickness, can be used to estimate average long-term corrosion rate.
A drawback of most of the soil corrosiveness models is that they do not address the beneficial effect of cathodic protection, and only indicate the inherent corrosiveness of the soil. Therefore, they may greatly overestimate the actual external corrosion rate, leading to overly conservative estimates of remaining life.
DEVELOPING A QUANTITATIVE MODEL
For application in oil and gas pipeline integrity management programs, the preferred soil corrosiveness model would be one which:
- Provides quantitative estimates of corrosion rates to enable predictions of remaining life to be made.
- Considers multiple soil attributes to address the complex interaction of various soil attributes.
- Provides a means to illustrate the beneficial effect of cathodic polarization on the estimated rate of external corrosion of the pipeline.
A review of the literature showed that none of the available soil corrosion models met all of those requirements. Therefore, Structural Integrity Associates, Inc. (SI) began searching for suitable empirical datasets from which a model could be developed.
The report for PRCO project PR-208-163 (Barlo, T., "Field Testing the Criteria for Cathodic Protection of Buried Pipelines," PRCI project PR-208-163 final report, Pipeline Research Council International, February 1994) contained results of an assessment of the effect of cathodic polarization on full-size pipe samples buried in 15 different soils in the U.S., Canada, and Australia. At each location replicate pipe samples were cathodically polarized to various pipe-to-soil potentials for five years. The testing also included samples with no cathodic polarization (i.e., at the free corrosion potential). Seventeen soil attributes were measured at each location. At the conclusion of the testing, the samples were excavated and recovered, and the pitting corrosion on each sample was quantified.
SI staff performed 130,000 regression analyses on 42 of the datasets from the report to determine the soil and cathodic protection attributes which had the most significant relationship to the reported pitting rate. Eventually, three similar models were developed to accommodate client datasets that were lacking information regarding one attribute or another. The three models are included in a proprietary software package named "SoilPro."
The three models were tested by comparing the predicted corrosion rates to the reported corrosion rates for 14 datasets from the PRCI report that were not used in the model development. Table 2 lists the soil and cathodic protection attributes used in each model and the resulting standard error for each of the models over the indicated ranges of corrosion rates, The accuracy of the model, as determined by comparing predicted to actual pitting rates is illustrated graphically in Figure 1.
Output from the SoilPro model includes:
- The calculated estimate of corrosion pitting rate for each of the three models for which the appropriate input data is entered.
- The ability to compare any two soil samples to each other and determine the percent confidence level associated with the conclusion that the corrosiveness of the first sample is greater than the second. This capability allows the amount of uncertainty in the ranking of the relative corrosiveness of two locations to be quantified.
- Graphs illustrating relationships between any selected soil attribute and the predicted corrosion rate. This capability is useful for showing that there is no consistent relationship to any single soil attribute.
- The ability to illustrate how hypothetical changes in the CP will impact the estimated pitting rate. In some cases relatively large decreases in expected corrosion rate result from minor improvements in the cathodic polarization.
GENERATING THE INPUT DATA
Input data is derived from two sources:
- Measurement of cathodic protection potentials, including an "instant off" potential at the location of interest, and preferably, the free corrosion potential derived from unpolarized steel. An unprotected casing or other unprotected steel structure (not galvanized steel) can be used to estimate the free corrosion potential.
- Measurement of soil attributes derived from laboratory analysis of soil samples.
The laboratory measurements include measurement of soil resistivity, pH, moisture content, and various soluble anions and cations. The soil particle size distribution is measured using the procedure in ASTM Standard D422 (Anon., ASTM D422-63, 2007, "Standard test Method for Particle-Size Analysis of Soils" ASTM, October 15-2007), including the use of a hydrometer to measure silt and clay contents. The mass of soil required for testing is somewhat dependent upon the soil particle size, but generally the chemical analyses can be performed on a sample volume of about 1 to 2 pints (0.5-IL) and for soils with relatively low gravel content an additional gallon (3.8L) is typically more than enough for particle size analysis.
Soil samples should be removed from a location close to the pipe surface and should be stored in a sealed container that prevents contamination and desiccation of the soil. Either a sturdy plastic bag (8 mil or 0.2 mm thickness) with the closed seal reinforced by shipping tape, or a plastic container with tight fitting lid may be used. In either case, the bag or container should be completely filled or air squeezed out before sealing. Containers should be labeled to clearly indicate the location from which the contents were obtained. Samples should be packed to withstand the rigors of shipping. Guidance on sample labeling, packaging and shipping requirements can often be obtained from the analytical lab.
Ideally, seasonal variations are addressed by taking samples and measuring CP characteristics during the dry and wet seasons of the year.
BENEFITS OF THE MODEL
Pipeline operators have used the results of SoilPro to prioritize locations for direct examination following indirect examination in ECDA projects. Often, the results have verified that existing CP levels are sufficient to control corrosion at locations of coating damage, even if the inherent soil corrosiveness is high. As a result, reinspection intervals can often be lengthened appreciably compared to the use of the default corrosion rates described in NACE Standard RP0502. We have also demonstrated the benefit of using select backfill by showing that even many years after installation, the characteristics of the backfill result in lower anticipated pitting corrosion rates than the native soil.
Bill Amend
