Surface Crack Growth Behavior of Pipeline Steel Under Disbonded Coating at Free Corrosion Potential in Near-Neutral pH Soil Environments

Abstract

Crack growth behavior of X65 pipeline steel at free corrosion potential in near-neutral pH soil environment under a CO₂ concentration gradient inside a disbonded coating was studied. Growth rates were found to be highest at the open mouth of the simulated disbondment where CO₂ concentrations, hence local hydrogen concentration in the local environment, was highest. Careful analysis of growth rate data using a corrosion-fatigue model of the form ΔK ^α/K ^β_max /f ^γ, where (1/f ^γ) models environmental contribution to growth, revealed that environmental contribution could vary by up to a factor of three. Such intense environmental contribution at the open mouth kept the crack tip atomically sharp despite the simultaneous occurrence of low-temperature creep and crack tip dissolution, which are the factors that blunt the crack tip. At other locations where environmental enhancement was lower, significant crack tip blunting attributed to both low-temperature creep and crack tip dissolution was observed. These factors both led to lower crack growth rates away from the open mouth.

Appendix

Conversion of the potential drop readings to actual crack sizes was done using a self-developed curve fitting method since none of the methods listed in the literature was found to work due to the unique geometry of the cracks. The method used is summarized below.

It is known that the potential drop is related to the crack area through the resistivity of the material:

$$ \rho = \frac{{RA^{*} }}{l}, $$

(1)

where R is the resistance of material, A ^* is the area of the un-cracked ligament, and l is the length between the signal wires. The voltage across the crack crevice, V, can be related to the applied current, I, which was kept constant, and the resistance of the material in the un-cracked ligament through the following relation:

$$ V \propto \frac{Il\rho }{{A^{*} }}. $$

(2)

For the specimen used in the current investigation, it is obvious that

$$ A^{*} = A - A_{\text{c}}, $$

(3)

where A is the cross-sectional area of gage area, and A _c is the instantaneous crack area. Therefore,

$$ V \propto \frac{Il\rho }{{A - A_{c} }}. $$

(4)

The initial and final crack sizes are known measurable values. Therefore, the measured potential drop could be related to the point to point crack area. Since ρ is a finite value at every point on the curve, the numerator in Eq. [4] could be treated as a constant so that

$$ V = \frac{k}{{A - A_{\text{c}} }}. $$

(5)

To convert the full potential drop data to actual cracks sizes, the following steps could be used:

• 1. On the original potential vs time plot, select a few points (up to 12 was used).

• 2. Record the potentials and times at these points. Convert the times to number of stress cycles (N) that elapsed up to that point during the test.

• 3. Make a plot of the inverse of the initial and final remaining crack ligaments to the corresponding potential drop values to obtain a linear relationship between V and A ^*.

• 4. Use the relationship obtained in step 3 to determine an approximate value for A ^* for each of the points selected in step 2.

• 5. From each value of A ^*, determine the value of A _c, the crack area.

• 6. A _c could then be related to the area of a semi-ellipse:

$$ A_{\text{c}} = \frac{\pi ac}{2}, $$

(6)

where a is the crack depth and c is the crack size measure along the major semi-axis at the sample surface.

• 7. Plot the initial and final values of ‘c’ against the initial and final values of ‘a’ to obtain a linear relationship between ‘c’ and ‘a’.

• 8. The relationship obtained could then be used to substitute for c in Eq. [6].

• 9. The resulting equation could then be solved for ‘a’ to obtain the crack depth at all the points selected in step 1.

• 10. Finally, the values of ‘c’ could then be obtained using the relationship obtained in step 7.

The crack growth data obtained by the potential drop method were used to obtain crack growth rates per cycle da/dN, where the da/dN is the slope of the crack size vs number of cycle plot. Also, corrosion-fatigue analysis was carried out using Chen’s corrosion-fatigue model to calculate the so-called combined factor $(\varDelta K)^{2} K_{\hbox{max} } /f^{\gamma }$. All data were finally presented as a da/dN vs combined factor plot. K _max was calculated using stress-intensity solution for a semielliptical surface flaw in a flat plate for a ≤ c (see Table A9.1, p. 434, in Reference 32). A value of 0.1 was assigned to γ as deduced by Chen and Sutherby[13].