Fuzzy Control, die multivariable Kennfeldregelung Ein gelungenes Modell zur Optimierung des MVA-Betriebes Ergebnisse des VGB-Forschungsprojektes 173

Wastes requiring special monitoring are commonly combusted in plants with relatively small throughput. The operators of such plants either apply special procedures for a narrow range of wastes or processes where a number of wastes are combusted according to demand. These wastes have highly varying combustion characteristics and a broad range of calorific values. Combustion menus can only be drafted according to energy input in order to safeguard the furnace output. The amount of solid, liquid and gaseous fuels in the entire energy input is subordinated under this criterion. Thus, waste input does not permit any prediction about the combustion characteristics (release of volatile matter, ignition, solids burn-out, gas burn-out and slag behaviour). Fuzzy control is able to supplement process control that orients at measuring values by operators' experience. The optimization of the combustion process is of great importance and thus the related improvement of burn-out of waste gases and solid combustion residues. One important pre-requisite for minimum emissions is furnace control that can balance irregularities of the combustion process. The technical basics (hardware, software and engineering) for the application of fuzzy control were created by procuring a fuzzy programmable controller (or lately modular design (PS416)) able to serve as master device. The Institute IEC of the Freiberg Bergakademie was involved in these considerations. The decisive aspect for applying fuzzy control was the idea to use own plant knowledge. Besides, the idea was convincing to have an own tailor-made furnace control system. Additionally, fuzzy control has proven well for complicated and complex processes where all important process technical data have to be combined for making a decision. The installed fuzzy controller is deciding by weighing synchronously four to five measured process data. This is quite close to the averaging of human thinking. Within seconds the controller reverts to 1000 rules, about 40 of which contribute to the result of the controller. Own station personnel was able to develop, maintain and document fuzzy control. This demonstrates that specially in smaller decentral thermal plants, where the specific staff costs are generally high than in central waste treatment plants, solutions are looked for which support to reduce the cost for external services.