Corrosion tests probability sampling

It is seen in Table IV that both pilot plant samples (finished samples plus additives) passed the copper corrosion test, as did most of the additive-free refinery samples. However, all fourteen of the additive-containing refinery samples failed. These fourteen samples were all finished sairples and contained AO-30. Since the six DFM samples which failed the test did not contain FSII, it is assumed that this additive did not play a part. Since the two pilot plant samples (containing AO-29) passed the test, and the fourteen refinery samples (containing AO-30) failed, questions are raised concerning the anti-oxidant additives which were used. These results need to be related to the sequence of operations in the refinery process and to the nature of the additives which were used. It seems probable that either there was a problem with the acid treatment process by which a corrosive species was produced which ended up in the finished samples, or that the additives used may have been contaminated. Both aspects may be involved. 

In view of possible or probable variations in mechanical properties among different specimens of the same metal cut from different sheets or other pieces, or even from different sections of the same sheet or piece, it is necessary to pay careful attention to the initial sampling of stock to be used for control, as well as exposure, specimens. An interesting case in which several of these considerations were involved was provided by the long-time atmospheric exposure tests of non-ferrous metals carried out by Subcommittee VI of ASTM Committee B-3 on Corrosion of Non-Ferrous Metals and Alloys in which changes in tensile properties were used as one of the means of measuring the extent of corrosion. 

In all probability, any serious prediction or evaluation of an actual recirculating cooling water, based on an index or model calculated from an analysis of the cooling water, is of dubious validity, as problems of scale and corrosion do not originate at the precise time of sampling and testing. They have already occurred. 

---