Experiment of Using Local Economic Potential Based Decisions for a Mixing Process by the Generic Bi-Layered Net Model

Abstract

Generic Bi-layered Net (Csukás and Bánkuti 2003a, 2003b) based software technology, developed for the Direct Computer Mapping of process models at the Institute of Mathematics and Information Technology, University of Kaposvár, has been applied for the dynamic scheduling of a fodder mixing plant. The studied system is a typical example for a process, consisting of a network, built from storage volumes, as well as from discrete and continuous transportations and transformations (purchase, processing, transport, sale). The actual example for the simulation based investigation has been developed for an agricultural farm (Csukás et al. 2001, Bóity et al. 2001), containing cultivation, fodder mixing and pig-breeding. The formerly developed simplified, simulation based planning model has been changed for a detailed dynamic scheduling one. In this application the Generic Bi-layered Net model is used for the description of a hybrid automaton (Bánkuti and Csukás, 2003). The detailed structural model and the respective dynamic database have been described. In addition to the balance calculation we prepared the model also for the detailed dynamic simulation of the measures of demand and costs, coming from the sales, as well as from the purchases and processing, respectively. In the (for the time being simplified and plausible) case study, we investigated the possibility of the cooperative local decisions about the alternative and/or concurrent elementary processes, based on the economic potential, calculated from the summarized changes in the cost and in the measure of demand, associated with the respective storage volumes. The „possibility space” of the model is determined by the alternative purchases and sales, as well as by the concurrent mixing receipts, realizable in the single mixing equipment. The complexity of the system is increased by the possible sales of the raw materials of own production and, by the consumption of mixed products from other firms. According to our experiences the method tends to select the locally advantageous suboptimal solutions for each subsystem. A natural cooperativity is evolving in the system, because the same storage volumes are connected with various elementary processes that results in the overlapping of the neighboring local subgoals. The essence of this kind of cooperative architecture is that the functionally connected parts of the process are forced to develop such suboptimal states that allow developing the suboptimal configuration also for their neighbors. Based on our investigation it is obvious that the Generic Bi-layered Net model gives an adequate representation of the logistical processes involving discrete and continuous elements, while the method makes possible the dynamic accounting of the costs and measures of demand, associated with the storage volumes. The main conclusion is that with the knowledge of the cost- and measure of demand modifying effects of the alternative and/or concurrent processes we can make suboptimal cooperative local decisions based on the comparison of the respective economic potentials.