Catalyst design using artificial intelligence: SO₂ to SO₃ case study

Abstract

Catalyst design is key to the improvement of chemical process efficiency. The required work for the development of new catalysts can be supported through the proper application of artificial intelligence to identify optimal compositions. A generic methodology for the application of machine learning to catalysis research is therefore outlined in this work. The catalytic oxidation of SO2 was used to exemplify the first iteration of this methodology. 1784 data points from 31 published papers were compiled into a databank. The inlet SO2 concentration ranged from 0 to 66 mol%. An artificial neural network (ANN) was trained on the databank in order to predict SO2 conversion based on the catalyst composition and the reactor operating conditions (temperature, pressure, catalyst mass: volumetric flowrate ratio (w/v), and feed composition). The model achieved a root‐mean‐square error of 6.6%. A preliminary screening step identified 3:1 V‐Mg/SiO2 catalysts as exhibiting high conversion at 648 K. A multi‐objective optimization was then performed on a single catalyst to identify solutions exhibiting high conversion and high productivity at 648 K while minimizing the catalyst cost. The optimal solution was predicted to be a 2.9 wt% V/0.2 wt% Mg/SiO2 catalyst operating at a w/v of 7.49 kg‐cat · s/m3 STP, achieving 100% SO2 conversion with a material cost among the bottom third of cost values. Artificial intelligence can then be employed to extract useful knowledge from published catalytic data and orient future search for novel catalyst development.

Infochemistry and the Future of Chemical Information Processing

“…Machine learning is increasingly being used to predict the catalytic activity of materials and to model catalytic processes (191). In particular, Chen et al (192) proposed using a neural network to analyze density functional theory data on the catalytic reduction of CO 2 to CO.…”

Artificial neural networks and genetic algorithms: An efficient modelling and optimization methodology for active chlorine production using the electrolysis process

“…[7,16,17] Various types of problems in science can be cast in the form of such pattern-matching, and among the methods within machine learning tools, ANNs are one of the most effective methods. [18][19][20] Some recent publications illustrate successful application of ANN models in various electrochemical processes. [21][22][23][24][25] GAs belong to the category of evolutionary algorithms that are used for the optimization of objective (fitness) functions by means of parameter space coding.…”

Genetic‐based multi‐objective optimization of alkylation process by a hybrid model of statistical and artificial intelligence approaches

“…These models are used in different fields of science and engineering to describe the input-output relations. [7][8][9][10][11][12] The ability to figure out the complicated input-output relations distinguishes it from traditional methods. Another goal is to investigate the influences of several operational parameters, including temperature, length of reactor, and pressure on the performance of the alkylation process for cumene production.…”