ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

ANN modelling of the steam reforming of naphthalene based on non-stoichiometric thermodynamic analysis

Chinenye Adaobi Igwegbe, Adewale George Adeniyi, and Joshua O. Ighalo

Department of Chemical Engineering, Nnamdi Azikiwe University, Awka, Nigeria

E-mail: oshea.ighalo@yahoo.com

Received: 10 November 2020  Accepted: 11 February 2021

Abstract:

Naphthalene is a major component of tar whose formation is a technical barrier in gasification systems. It can be used for hydrogen production via the steam reforming process. In this study, artificial neural network was used to model the steam reforming of naphthalene. The dataset will be developed by non-stoichiometric computation of the minimisation of Gibbs free energy method. The effect of temperature and steam-to-oil ratio (STOR) on the selectivity of hydrogen gas, carbon dioxide, carbon monoxide and methane in the product stream was investigated. Temperature and STOR increase favoured H₂ production in the steam reforming process. At the threshold of temperature > 600° C and STOR > 4 kg/kg, optimal H₂ selectivity is achieved. The coefficient of determination and root mean squared error for the model for all regimes (training, validation and testing) and all synthesis gas constituents was > 0.99 and < 1 mol%, respectively. Parity plots revealed that the predictions were accurate at both high and low levels of prediction. Paired samples correlation revealed a strong positive correlation between model predictions and the experimental values. The current approach is unfavourable in scenarios where quick predictions and preliminary estimations are required other investigations in product and process development.

Keywords: Artificial neural network; Energy; Hydrogen; Naphthalene; Steam reforming

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-021-01566-2

Chemical Papers 75 (7) 3363–3372 (2021)