Corrosion protection values must be averaged harmonically
Various test methods are available for determining the corrosion protection of coatings on components. In principle, the results of such methods are more widely scattered than those of immediate measurement methods. Therefore, one always puts several components simultaneously into the salt spray test and similar corrosion tests, such as condensation, alternating climate, Kesternich and other tests. An average value is then often given as the result. What has not yet become generally known is that one does not have to form the arithmetic mean but the harmonic mean of corrosion protection values, provided that it is a kind of resistance value, i.e. basically the reciprocal of the corrosion variable.
For example, the resistance of a coating in the salt spray test is given as the duration in hours until certain corrosion phenomena appear. The longer it takes, the higher the resistance (= corrosion protection). If several components are now placed in the salt spray chamber, the result is basically a parallel circuit of corrosion resistances. The resistance value of a parallel circuit is the reciprocal of the sum of the reciprocals of the individual resistances. Now, the mean value of the individual resistors is the harmonic one, and it is obtained by multiplying the total resistance by the number of resistors connected in parallel.
Example: six galvanized and passivated parts were subjected to a salt spray test, and the duration until the first appearance of white rust has been noted:
A

B

C

D

E

F

72 h

60 h

96 h

72 h

120 h

96 h

Arithmetic mean:

(72 + 60 + 96 + 72 + 120 + 96)/6 = 86 h

Harmonic mean:

6/(1/72 + 1/60 + 1/96 + 1/72 + 1/120 + 1/96) = 81.5 h

The harmonic mean is always smaller than the arithmetic one. Outliers to high values have a smaller, and outliers to low values (up to short circuit) a larger impact on the result.
I had worked out a similar application case, where harmonic averaging also comes into play, here for coating resistances, for my Dissertation  3.8.2 Resistance and Roughness.
Copyright © Cyclaero Ltda.  rj  20181016 14:35:17
