(521ek) Quantum Computing Assisted Data-Driven Modeling for Yield Prediction of Naphtha Cracking Process

Naphtha cracking process is essential to produce elementary chemicals such as ethylene and propylene. The yield of naphtha cracking process is hard to be estimated and optimized due to its complex relations of various operating conditions. Although many theoretical and mathematical methods have been tried to solve the above problem, it needs lots of equations for reactions and time for calculation. Recently, there has been considerable interest in quantum computing (QC) and machine learning methods. This paper aims to investigate the application of quantum neural networks (QNNs) to predict product yields of naphtha cracking furnace, and discuss the impacts of the number of features and layers on the yield prediction performance. Their designs, implementation and results are stated as following steps. First, coil outlet temperature (COT) and compositions of major components in naphtha including normal paraffins and aromatics were selected as the features due to their considerable impacts on the product yield. Second, three QNN models was developed depending on the number of layers (three, four, and five). Lastly, the models were evaluated with the coefficient of determinant (R²) and compared with each other to select the optimal QNN-based model. Additionally, this paper describes some challenges of state-of-the-art quantum computers and their great potential to impact the field of chemical engineering.