Janet P. Buckingham

Stress Corrosion Cracking of Carbon Steel in Ethanol

Abstract This paper presents the results of a study on the effects of water, acetic acid (CH3COOH), oxygen, corrosion inhibitor, chloride, methanol (CH3OH), denaturant, and corrosion product on the stress corrosion cracking (SCC) of steel in ethanol (C2H5OH). The factor that was found to have the greatest effect on causing SCC was corrosion potential, as influenced by oxygen. The lower critical potential for SCC ranges from 25 mV vs. saturated calomel electrode (SCE) to 300 mVSCE, depending on the presence of chloride and methanol as impurities. Galvanic contact with precorroded steel appeared to exacerbate SCC by increasing the corrosion potential. Within the fuel ethanol specification limits, chloride had a less significant effect than oxygen. SCC was intergranular when the chloride concentration in ethanol (both laboratory and field samples) was low (less than 1 ppm) and it was transgranular when the chloride concentration was high (32 mg/L). A denaturant, a corrosion inhibitor, and acidity, within the...

PROBABILISTIC ANALYSES OF INDUSTRIAL GAS TURBINE DURABILITY

Probabilistic analyses of component durability have been developed for first stage buckets and nozzles in General Electric MS7001 B and MS7001 E industrial gas turbine engines. The analyses illustrate two different approaches to the development of probabilistic durability algorithms. The bucket algorithm is built around an existing thermal-mechanical fatigue life model which predicts average fatigue life as a function of local strains, dwell times, and constants in the fatigue model derived from laboratory testing. The fatigue model is linked to a fastprobability integration method which calculates the uncertainty or variability in the response variable (life or damage) resulting from uncertainties or variabilities in the physical input variables. The nozzle cracking algorithm, in contrast, is built around a large data base of crack length sums for each vane on seven different nozzles from the field. The crack length sums on nozzles with similar numbers of fired starts were described with standard normal distributions. Extreme value distributions were then calculated analytically to describe the crack length sums on the single most heavily cracked vane per nozzle for different probability levels.