It has been noted that weathering steel does not appear to form a protective patina in the presence of chlorides, especially de-icing salt [Albrecht & Naeemi 1984, Albrecht & Hall 2003, Cook 2005]. This is due in part to the hydroscopic qualities of the salt; it attracts water, and so keeps the steel moist for longer periods of time and preventing the occurrence of dry conditions necessary for the patina formation. The other factor is the presence of chloride ions, Cl - . These negatively charged ions increase the negative potential of the steel, which accelerates the rate of corrosion of the metal (this is also an effect of the SO4 -2 , though to a lesser degree). Also, in the presence of chloride ions the corrosion reaction results in relatively large amounts of -FeOOH. This oxide does not convert to form FeOx(OH)3-2x, which is the main component of the protective oxide layer of weathering steel
Another composite model recommended in [McCuen & Albrecht 1994] is the powerpower model, similar to the power-linear model, but in this case both equations are power equations. In both of these cases, numeric fitting to data points is recommended, as opposed to the logarithmic approach of [Townsend & Zoccola 1982]. Finally, in , McCuen and Albrecht modify their power-power model to account for the variable alloy content of weathering steel, specifically for the metals: copper, chromium, phosphorus, silicon, and nickel. This equation compares favourably with regards to the Legault and Leckie  equations, but no comparison is made between this and the logarithmic power model. Although the merits of each of the mentioned models can be argued, Equation 2.3, the logarithmic power model, is among the simplest of these and is commonly applied to structural problems resulting from corrosion attack. It should be noted that none of the models appear to have been experimentally verified for their validity over the long term.