A fuzzy logic model and a neuro-fuzzy system development on supercritical CO2 regeneration of Ni/Al2O3 catalysts. Issue 54 (December 2021)

• Record Type: Journal Article
• Authors: Ghadirinejad, Nickyar; Ghadiri Nejad, Mazyar; Alsaadi, Naif
• Is Part Of: Journal of CO₂ utilization. Issue 54(2021)
• Publication Date: 2021-12
• Subjects: Fuzzy logic modeling -- Neuro-fuzzy system -- Response surface methodology -- Supercritical carbon dioxide -- Regeneration of catalysts -- Midrex unit; Carbon dioxide -- Periodicals; Carbon dioxide -- Environmental aspects -- Periodicals; Carbon dioxide mitigation -- Periodicals; Carbon dioxide; Carbon dioxide -- Environmental aspects; Carbon dioxide mitigation; Periodicals; 628.53205
• DOI: 10.1016/j.jcou.2021.101706
• Languages: English
• ISSNs: 2212-9820

Abstract

Graphical abstract: Highlights: Deactivated rib ring nickel alumina (Ni/Al2 O3 ) catalysts are regenerated in a pilot plant supercritical CO2 extractor. A Mamdani-type fuzzy logic model is developed to elucidate the operational and to predict the outputs. The considered operational parameters are temperature, pressure, and flow rate of supercritical carbon dioxide. A Neuro-fuzzy system is developed to define and predict non-experienced data for the output variable. The Neuro-fuzzy system is validated by using the root mean square error (RMSE). The response surface methodology is used to find the optimal output value and the related input parameters values. Abstract: Catalyst regeneration is highly important in chemical processes in terms of process performance and economic efficiency. In this study, deactivated rib ring nickel alumina (Ni/Al2 O3 ) catalysts are regenerated in a pilot plant supercritical CO2 extractor. The catalysts are collected from the Iron Midrex reformer unit and are provided by Khouzestan Steel Company (KSC). A Mamdani-type fuzzy logic model and a Neuro-fuzzy system (FIS) model are developed to elucidate the operational parameters i.e. temperature, pressure, and flow rate of supercritical carbon dioxide. The models may define and predict non-experienced data for the output variable, which is the mass concentration of removed deactivating substances from the spent catalyst. Triangular architecture fuzzy models with seventeen rules for the first model, and fourteen rules as the training data for the latter one, are considered to develop the models. The proposed FIS model is validated by using the root mean square error (RMSE) and considering three of the experiments as the checking data. Moreover, the response surface methodology (RSM) is used to find the optimal output value and the related input parameters values. The results show that the proposed neuro-fuzzy model is valid with the RMSE of less than 1.3 %, and the optimum output value found by the RSM is 0.0767, when the temperature, pressure, and flow rate are 57.7, 182.5, and is 0.538, respectively.