The aim of this study was to develop an automatic system based on selected parameters collected from the biological process to control a fluidized-bed reactor. In the first part of this study, identification of parameters sensitive to overloads was carried out with vinasses as wastewater. From these results, three parameters, considered as the most realistic for the monitoring of the reactor, were selected: pH in the liquid phase, the biogas production and its concentration of dihydrogen. These were used to develop an algorithm based on expert system principle. This automatic control system calculated the flow rate of the feeding pump in order to adjust continuously the load applied to the reactor. The system was used to monitor and control several kinds of overload in an anaerobic fluidized bed reactor, of laboratory and pilot scales. The strategy used was able to operate automatically the digesters.

A pilot fluidised bed reactor was subjected to shockloads using wine distillery wastewater. The parameters' response was recorded. Parameters analysed in the gaseous phase were : H2, CH4, CO2 and gas production; in the liquid phase : pH, temperature, TOC and VFA. Different shockloads were applied : short but intensive shockloading, long shockloadings; temperature variations; change of nature of substrate (wine distillery wastewater was replaced by glucose). Generally speaking, the parameters' response was short (less than one hour). The specific parameters of the gaseous phase and pH can be used for the optimal control of the fermentation process.

The start-up procedure of a laboratory fluidized-bed was monitored by an automatic control system. Twenty six days were necessary to increase the load from 1 to 35 kg COD/m3.d. The system took into account the pH of the liquid phase, the production of gas, and the concentration of hydrogen in the gas phase. These parameters were measured by on-line sensors. According to variations of these parameters, algorithm based on a principle very close to an expert system adjusted the flow-rate of the feeding pump. The automatic control system was tested in the following situations: perturbations in the process generated by large pH variations (± 0.2 to 0.7 pH unit) and organic overloads (from 50 to 100 kg COD/m3.d). After 44 hours of overloading, CH4 and CO2 concentrations in the gas phase were very useful to detect the deviation of the reactor.