This paper studies the problem related to the variation of measured properties of a bituminous coal, fired in a Colombian electric power plant. The strategy was to implement a DOE methodology to find the causes of these variations, changing the storage conditions and relating it to the weather and different ways of handling the samples. The study contemplates the design, deployment and analysis of three experiments in order to trace the influence of such parameters as time and deplete of storage, weather evolution prior to consumption andsome sampling procedures and in this way explain the divergence of coal properties between the reception in the yards and its medium grinding just before firing. The results had shown that stocking time in yards and the automatic system´s performance, as the main cause of the problem.