A Deep Neural Network for Oxidative Coupling of Methane Trained on High Throughput Experimental Data

Document Type



The oxidative coupling of methane (OCM) is a chemical process that converts methane (the main component of natural gas) into higher hydrocarbons, such as ethylene and propylene. OCM has significant importance in the chemical industry as it offers a potentially cleaner and more economical way to produce valuable chemicals from natural gas. Methane is an abundant and inexpensive feedstock, and OCM offers a way to convert it into higher value products without the need for expensive and energy-intensive processes. Overall, OCM has a significant potential to promote sustainability by reducing greenhouse gas emissions, conserving resources, promoting circular economy, and improving energy efficiency.

In this work, we develop a deep neural network model of the reaction rate of oxidative coupling of methane from published high throughput experimental catalysis data. The neural network is formulated so that the rate model satisfies the plug flow reactor design equation and is utilized to investigate the reactant and product composition variation within the reactor for the reference catalyst Mn−Na2WO4/SiO2 at different temperatures. Additionally, the model is employed to identify new catalysts and combinations of known catalysts that could enhance the yield and selectivity relative to the reference catalyst. The study reveals that methane is predominantly converted in the first half of the catalyst bed, while the second part consolidates the products, resulting in an increased ethylene to ethane ratio. The screening study of over 3,400 combinations of pairs of previously studied catalysts indicates that a reactor configuration comprising two sequential catalyst beds leads to synergistic effects and increased yield of C2 products compared to the reference catalyst at identical conditions and contact time. An expanded screening study of 7,400 combinations, including previously studied metals with several new permutations, identifies multiple catalyst choices with enhanced yields of C2 products. This study highlights the value of learning a deep neural network model of the instantaneous reaction rate directly from high-throughput data and represents a first step in constraining a data-driven reaction model to satisfy domain information. This thesis contributes to the advancement of the understanding of the oxidative coupling of methane process and provides insights that could be beneficial in the development of new catalysts and optimization of the reactor design.

First Page


Last Page


Publication Date



Thesis submitted to the Deanship of Graduate and Postdoctoral Studies

In partial fulfillment of the requirements for the M.Sc degree in Machine Learning

Advisors: Dr. Martin Takac, Dr. Karthik Nandakumar

Online access available for MBZUAI patrons