Modelling and optimization of modified anaerobic baffled reactor for recycled paper mill effluent treatment using response surface methodology and artificial neural network
Prestasi reaktor bersesekat anaerob terubahsuai (MABR) dalam rawatan air sisa kilang pembuatan kertas kitar semula telah dikaji dari segi penyingkiran permintaan oksigen kimia (COD), penyingkiran lignin dan kadar penghasilan metana (CH4) berkenaan dengan dua pembolehubah, iaitu influen COD (1,000-4,000 mg/L) dan masa tahanan hidraulik (HRT) (3-7 hari). Eksperimen telah dijalankan seperti yang diperlukan oleh rekaan pusat komposit dengan struktur berpusat muka (CCFD) dan data yang dikumpulkan talah dianalysis dan digunakan untuk mendapatkan model dengan pengaplikasi kaedah permukaan sambutan (RSM) dan rangkaian neural buatan (ANN). Ketepatan model RSM telah diuji dengan analisis varians (ANOVA) dan struktur model ANN telah dioptimum dengan min ralat kuasa dua yang paling rendah. Peramalan prestasi MABR menggunakan model RSM telah dikenal pasti lebih tepat berbanding dengan model ANN. Selain itu, kedua-dua model bersama-sama menunjukkan bahawa MABR beroperasi dengan influen COD yang tinggi and HRT yang rendah dapat mengakibatkan performasi MABR yang baik. Oleh itu, keadaan optimum untuk sistem rawatan ini telah dikenal pasti untuk mencapai efisiensi dan penghasilan metana yang lebih tinggi di samping merendahkan kos operasi.
_______________________________________________________________________________________________________
Performance of Modified Anaerobic Baffled Reactor (MABR) in recycled paper mill effluent (RPME) treatment was investigated in terms of chemical oxygen demand (COD) removal, lignin removal and methane (CH4) production rate with respect to two independent variables: feeding COD (1,000-4,000 mg/L) and hydraulic retention time (HRT) (3-7 days). Experiments were conducted as per central composite face-centered design (CCFD) and the data was analyzed and used for model building by adopting response surface methodology (RSM) and artificial neural network (ANN) approach. The precision of the RSM model was confirmed with analysis of variance (ANOVA) while the optimized architecture of ANN model was determined with least mean-squared-error (MSE). In comparison, RSM model turned out to be more accurate than ANN model in the prediction of MABR performance. However, both models agreed that MABR operating with high feeding COD and low HRT can result in better performance. Thus, performing optimization for the MABR, the optimal condition for the treatment system was determined in order to achieve high removal efficiency and methane production while minimizing the operating cost.
Modelling and optimization of modified anaerobic baffled reactor for recycled paper mill effluent treatment using response surface methodology and artificial neural network
Modelling and optimization of modified anaerobic baffled reactor for recycled paper mill effluent treatment using response surface methodology and artificial neural network / Lee Wen Jie