Development of Correlations for Overall Gas Hold-up, Volumetric Mass Transfer Coefficient, and Effective Interfacial Area in Bubble Column Reactors Using Hybrid Genetic Algorithm-Support Vector Regres

详细信息    查看全文

• 作者：Prashant P. Gupta ; Shamel S. Merchant ; Akshay U. Bhat ; Ankit B. Gandhi ; Sunil S. Bhagwat ; Jyeshtharaj B. Joshi ; Valadi K. Jayaraman ; Bhaskar D. Kulkarni
• 刊名：Industrial & Engineering Chemistry Research
• 出版年：2009
• 出版时间：November 4, 2009
• 年：2009
• 卷：48
• 期：21
• 页码：9631-9654
• 全文大小：683K
• 年卷期：v.48,no.21(November 4, 2009)
• ISSN：1520-5045

文摘

The objective of this study was to develop hybrid genetic algorithm−support vector regression (GA-SVR)-based correlations for overall gas hold-up (b>Gb>), volumetric mass-transfer coefficient (kb>Lb>a), and effective interfacial area (a) in bubble column reactors for gas−liquid systems employing viscous Newtonian and non-Newtonian systems as the liquid phase. The hybrid GA-SVR is a novel technique based on the feature generation approach using genetic algorithm (GA). In the present study, GA has been used for nonlinear rescaling of attributes. These, exponentially scaled, are eventually subjected to SVR training. The technique is an extension of conventional SVR technique, showing relatively enhanced results. For this purpose an extensive literature search was done. From the published literature, 1629 data points for viscous Newtonian and 845 data points for viscous non-Newtonian systems for b>Gb>, 500 data points for viscous Newtonian and 556 data points for viscous non-Newtonian systems for kb>Lb>a, and 208 data points for viscous non-Newtonian systems for a, respectively, were collected. These data sets were collected spanning the years 1965−2007. Correlations were developed after taking into account all the parameters affecting b>Gb>, kb>Lb>a, and a such as column and sparger geometry, gas−liquid properties, operating temperature, pressure, and superficial gas and liquid velocities. The correlations thus developed gave prediction accuracies of 0.994 and 0.999 and average absolute relative errors (AARE) of 3.75 and 1.65% for viscous Newtonian and non-Newtonian systems for b>Gb>, prediction accuracies of 0.983 and 0.998 and AARE of 8.62 and 1.91% for viscous Newtonian and non-Newtonian systems for kb>Lb>a, and prediction accuracy of 0.999 and AARE of 1% for viscous non-Newtonian systems for a, respectively. These correlations also showed much improved results when compared with all the existing correlations proposed in literature. To facilitate their usage, all the hybrid GA-SVR-based correlations have been uploaded on the web link b/UICT-NCL" class="extLink">http://www.esnips.com/web/UICT-NCL.