Forest fires are time evolving disasters that consume environmental and financial resources, endangering the rescue units that try to mitigate them. As such, a simulator that can predict the fire propagation is essential to locate control systems, in order to reduce the loss of natural resources without risking the firefighters. This paper proposes a simulator based on discrete representation of the selected areas where the velocity of fire propagation between neighbors depends on variables associated with locative or climatological characteristics. We consider discrete classification based on the spread in each direction. To validate our simulator we considered two scenarios: a theoretical area to test the algorithm considering complete information characteristics and a forest area near Bogota, Colombia. The results show realistic propagation patterns compared to region real past forest fire events. For a better prediction we need more reliable data and relate the fire to both location and weather characteristics.