One other possibility is using a sample with a scribe on it. in such a way the corrosion procedure will be accelerated, but the result is not exactly correlated to the barrier property of the coatings.
Also in the case of testing intact coating, it will be really time consuming. Which one do you suggest? and why?