Controle nebuloso ("fuzzy”) aplicado à combustão em caldeiras a bagaço de cana

Abstract

In the sugarcane industry, boilers are widely used for cogeneration. This term designates the simultaneous generation of more than one type of energy, namely thermal and electrical, from the same fuel source. This industry uses mainly sugarcane bagasse as the primary fuel source followed by other types of biomass, such as cane straw, woodchips and corn stover. According to the Brazilian Sugarcane Industry Union (UNICA) and the National Electric Energy Agency (ANEEL), the installed capacity for electricity supply from biomass represents 9% of the Brazilian energy matrix. Fuel price has a major impact on the industry’s results, due to the current scenario of the energy market. Therefore, optimizing the combustion is an important lever to reduce this cost. Real-time measurement of oxygen content in the boilers’ flue gas is one of the requirements for this optimization. Due to the capital cost and the requirements for personnel qualification for maintenance and calibration, sugarcane mills rarely apply such analyzers in closed loop. In this work, a fuzzy control was implemented in the combustion air flowrate of a large boiler in the Tereos group, to maintain the oxygen content in the combustion products within the optimum range. A tool was used to identify the process dynamics and to tune the control loop in real time. After downloading the new strategy in the factory’s control system and tuning the controller, the oxygen control loop started operation in automatic mode. After twenty-four hours of continuous operation, regulatory and servo controls were able to, respectively, keep the desired setpoint with low variability (± 1,2% variation in excess air coefficient) and respond to setpoint changes with good response speed (1,5-20 minutes) and overshoot (8-56%). Results reached so far demonstrate the applicability of fuzzy control for combustion in boilers on sugarcane bagasse, with 40% less variability than the PID controller, and 69% less variability compared to operation in manual mode.