Catalytic production of biodiesel from waste cooking oil in a two-phase oscillatory baffled reactor: Deactivation kinetics and ANN modeling study


Ali M. M., Gheni S. A., Ahmed S. M., Hmood H. M., Hassan A. A., Mohammed H. R., ...More

Energy Conversion and Management: X, vol.19, 2023 (ESCI) identifier

  • Publication Type: Article / Article
  • Volume: 19
  • Publication Date: 2023
  • Doi Number: 10.1016/j.ecmx.2023.100383
  • Journal Name: Energy Conversion and Management: X
  • Journal Indexes: Emerging Sources Citation Index (ESCI), Scopus
  • Keywords: ANN model, Biodiesel, Dolomite, Kinetics, Oscillatory baffled reactor, Transesterification
  • Ondokuz Mayıs University Affiliated: Yes

Abstract

As fossil fuels deplete day by day due to the growing demand in various sectors such as agriculture and transportation, the escalation of fuel prices in the international oil market, and increasing global warming, the search for alternative fuels has become urgent. The technology utilized and the cost of feedstock have the greatest impact on the cost of biodiesel production. In the present work, the goal was to produce high-quality biodiesel utilizing waste cooking oil as a feedstock and a natural catalyst in a continuous mode via Oscillatory Baffled Reactor (OBR). A transesterification catalyst based on the activation of dolomite rocks was prepared under a vacuum atmosphere. A kinetic study was conducted using oleic acid as a waste cooking oil model compound at 40, 50, and 60 °C in the OBR at low and high oscillation conditions and a reaction time range of 5–40 min. The study revealed that the transesterification reaction over the dolomite catalysts was the first order, and the activation energy was approximately 34 kJ/mol. The oscillation conditions do not affect the kinetics of the reaction. The OBR was used to evaluate the dolomite catalyst for the continuous slurry production of biodiesel via transesterification. The evaluation parameters were methanol: oil ratio of 6:1, temperature (50, 60, and 70 °C), residence time (5–40 min), the amplitude of oscillation (2, 4, 6, and 8 mm), and frequency of oscillation (1, 2, 3, 4 and 4.3 Hz). The results of the evaluation were used to generate an Artificial Neural Network (ANN) model to predict the continuous transesterification process in the OBR. The model was built with one hidden layer and 20 neurons. The simulated results were very close to the experimental results as a mean square error (MSE) of 0.0712 and a regression coefficient (R2) of 0.998 were obtained.