Scattering, atmospheric corrosion

Furthermore impurities in the helium coolant, mainly the air constituents, can cause corrosion effects on the outside reformer tube walls which eventually change its properties. Measurements of impurity contents in Dragon and AVR revealed a large scattering of the data. Experimental results obtained within the Dragon project indicate a strong corrosion of aluminum and titanium, i.e., the formation of Cr-, Mn-, Si-, and Ti-oxide layers, and an increased corrosion rate in moist helium compared with a dry atmosphere [26]. 

Although the reason for the large seatter when metal levels were very low may be easy to imderstand, the reason for the same effect at high metal concentrations m be open to more than one explanation. The sensitivity of the solubility of zinc corrosion products to solution pH, combined with the open beaker configuration of the test that allows the loss of CO2 to the atmosphere, are sources of potential variability in the test procedure used. It has been shown also that devices used in the test to cover the beakers and to hold coupons in the test solution influence the diffusion rate of gases (CO2 and O2) from the atmosphere to the test solution and vice versa. The tightness of the system or its permeability to gas m vary from one laboratory to the other, and could explain part of the scatter observed between laboratories. 

The results are too scattered for close correlation to be made, but general guidance from the results is that (unless chloride is greater than 120 mg/m Cl ), zinc corrosion rate is a basic 0.5-2p.m/year plus 1 p,m for every 40 p.g/m SO2 in the atmosphere. Where chloride concentrations exceed 120 mg/m Cl , specific mass loss data must be provided for each site. 

Materials properties are one such area and, depending on upon the actual position of the subject tubes in the materials properties scatter-band, can have a significant effect on the failure probability versus time results. Prior to shutdown, the probabilistic approach deals with this by assuming a normal distribution centred on mean published standard materials properties. Whilst it is possible that actual tube materials properties are below the mean value, it should be borne in mind that all standard materials data have been derived from laboratory tests conducted in an air atmosphere. The test data, and hence the derived stress allowables, are thus intrinsically conservative with respect to true creep behaviour, particularly when the design procedure makes additional allowance for service oxidation and corrosion. 

If the atmosphere is clean but not dry and the humidity approaches 100%, a scattered pattern of corrosion spots eventually appears, but considerable areas of unaffected surface remain for a very long time. If, however, the surface becomes contaminated by corrosive dust or cathodic particles, the whole surface rapidly becomes covered with a grayish layer of corrosion product (Tawil, 1987). A clean, unprotected magnesium alloy surface exposed to indoor or outdoor atmospheres free from salt spray develops a gray Him that protects the metal from corrosion (Froats et al., 1987). The rates of corrosion and resistance to corrosion of magnesium alloys vary depending on alloy composition (Loose, 1946) as discussed previously. 

---