My experience with the k_rm (and the e50 in the clay models, for that matter) is that they do not affect the results very much. Typically, I find that changing the k_rm value affects the result far less than other uncertainties within the analysis. Hence, I often use the default values selected...
