A #Bayeesian #model of #PeakCoal would probably best be done as a #FreeVariable for the peak year, and then a growth function and a decay function.
#Consumption then switches on that year.
A peak year not yet obtained could be estimated relative to a given sampled growth function.