In this paper, a biodiesel catalyst was prepared from the phosphogypsum (PG) by a hybrid approach through temperature swing acid leaching/crystallization steps followed by the subsequent fluidized bed calcination. The impacts of acid leaching and crystallization were extensively analyzed via a supervised machine learning approach using a limited number of experimental runs to find out the optimal condition. The determined optimal conditions are X1-95 (ºC), X2-30 (min), X3-30 (wt%-H2SO4), and corresponding validation experimental result (at the optimal condition setting) shows± 5% uncertainties. The prepared catalyst predominately contains CaSO4 (98 wt%) with the impurities less than 0.3 wt% (i.e., P2O5- and F-). The numbers of acid leaching cycles (up to 10 cycles) were investigated, and result indicates a good contaminates recovery (P2O5:1.8 g/100 g PG, Mg2+: 0.3 g/100 g PG, Al3+: 0.3 g/100 g PG, Fe3+: 0.1 g/100 g PG) in the leachate through the downstream solvent extraction. The catalytic conversion reaches about 50% with approximately± 5% deactivation when catalyst was reused at the same transesterification condition. The decreased binding energies of S2p (169.6 eV) and O1st (533 eV) in used catalyst indicate the deactivation of surface catalytic sites. The key properties of the prepared biodiesel are comparable to the American society for testing and materials (ASTM) standards.
- Machine learning optimization
ASJC Scopus subject areas
- Environmental Engineering
- Environmental Chemistry
- Chemical Engineering (all)
- Safety, Risk, Reliability and Quality