Software sensor for process monitoring of second-generation biofuel production from sugarcane bagasse hydrolysate
Tuesday, April 29, 2014
Exhibit/Poster Hall, lower level (Hilton Clearwater Beach)
William E. Herrera1, Victor Coelho Geraldo1, Henrique Real Guimarăes1, Elmer Ccopa Rivera2 and Rubens Maciel Filho3, (1)School of Chemical Engineering – University of Campinas, and Brazilian Bioethanol Science and Technology Laboratory -CTBE-CNPEM, Campinas, Brazil, (2)Brazilian Bioethanol Science and Technology Laboratory - Brazilian Center for Research in Energy and Materials (CTBE/CNPEM), Campinas, Brazil, (3)Department of Process and Product Development, School of Chemical Engineering, University of Campinas - Unicamp, Campinas, Brazil
Brazil as the largest producer of ethanol from sugarcane has an interesting environment since the first generation mills may be used to accommodate second generation bioethanol process, because the raw material is already available and the existing facilities may be shared, reducing the investments. However, the efficiency of fermentation by Saccharomyces cerevisiae using lignocellulosic feedstock depends on the fermentability of sugars from the hydrolysate which may be affected by inhibitors that are byproducts of the pre-treatment and hydrolysis process with impact on the kinetics, ethanol produced and productivity affecting economically the whole process. Thus, there is a demand for the development of models and procedures to describe the process in such way that process optimization, control and improved operation techniques may be used. In this work it was developed a Software Sensor based on Artificial Neural Networks (ANN) to infer the concentration of substrate, cells and ethanol from secondary measurements of pH, turbidity, CO2 flow rate and temperature. Experiments performed in the temperature range of 30 - 38oC were used to develop the software sensors. The raw material for the fermentation is a mixture of 75% hydrolyzed sugarcane bagasse and 25% sugarcane molasses. The software sensor was written using the commercial, equation-oriented process simulator Aspen Custom Modeler® and hence can easily be integrated with commercial process simulation packages. This study was completed comparing the Software Sensor prediction with a phenomenological model.