A Master of Science thesis in Chemical Engineering by Noora Naif Darwish entitled, "Adsorption study of desulfurization of diesel oil using activated charcoal," submitted in May 2015. Thesis advisor is Dr. Zarook Shareefdeen. Soft and hard copy available.
This research investigates the use of adsorptive desulfurization for diesel oil using carbon-based adsorbents, and it evaluates the effect of the adsorptive desulfurization on the ignition quality of diesel fuel. Two of the adsorbents are commercial powdered activated carbons (PAC1 and PAC2); whereas, the third one is a granular activated carbon (GAC). The desulfurization process is investigated at different conditions of three factors: amount of sorbent material (3 wt. % - 10 wt. %), temperature (room temperature, 30, and 50 oC) and contact time (0.5 - 2 hrs). Equilibrium and kinetics studies of the adsorption process using the three adsorbents are considered. In addition, results from the experimental data found are analyzed using a two-level full factorial design and are correlated using artificial neural networks. This study shows that PAC1 and PAC2 have better sulfur removal affinity compared to the GAC. The adsorptive desulfurization of diesel fuel improved the ignition quality of the fuel significantly. The adsorption isotherms are determined using two isotherm models which are: Langmuir, and Freundlich. Results show that the adsorption behavior for both PAC1and PAC2 is described by Freundlich model at all temperatures. In the factorial experimental designs, two outputs are investigated: sulfur removal percentage and diesel index. According to the 22 and 23 designs, the amount of sorbent material shows a positive effect on the two response variables; whereas, the temperature has a varying effect on the two outputs. The predicted outputs are calculated using a regression model generated and compared with the actual experimental data. The predicted values show an excellent agreement with the experimental data. Finally, a feed-forward neural network with one hidden layers of size 15 is used to correlate the set of experimental data. Results show that the sulfur removal capacities for PAC1 and PAC2 can be correlated perfectly using Artificial Neural Networks (ANN).