Software sensors design and selection for the production of biodiesel from grease trap wastes

In this paper, a couple of software sensors were designed for CSTR processes and they were tested in the esterification of grease trap wastes which is a low-cost feedstock for biodiesel production. Both software sensors were designed for the estimation of the concentration of Free Fatty Acids (FFAs) from discrete CSTR's temperature measurements. One of these software sensors, called asymptotic observer, is recognized for its capacity to provide estimates without the knowledge of the process kinetics. It has been also designed an estimation algorithm known as reset fuzzy observer. This approach has been recently proposed and is able to update the estimated states at each instant when discrete measurements are available. Both observers were built regarding the structure of a validated dynamical model for the esterification of grease trap wastes and they were tested considering experimental data. The results show that the asymptotic observer offers a very poor performance when some process's disturbances (e.g. input concentration of FFAs, jacket temperature) are carried out, whereas the reset fuzzy observer is able to reconstruct satisfactorily the concentration of FFAs in the presence of such disturbances. Therefore, this fuzzy observer can be selected as reliable monitoring approach that could be further used as a key part of robust control schemes conceived for this specific biodiesel production process.