An expert model for estimation of the performance of direct dimethyl ether synthesis from synthesis gas

Abstract

In this work, an artificial neural network (ANN) has been trained and tested for estimation of the performance of direct synthesis of dimethyl ether (DME) from synthesis gas. Yield and selectivity of DME production and also conversion of CO could be predicted when temperature and pressure of reactor and H 2 /CO molar ratio in feed have been specified. The results of ANN estimation for yield of DME, selectivity of DME and CO conversion are in very good agreement with experimental values. For this development, database was collected from our previous experiment. The accuracy and trend stability of the trained networks were tested against unseen data. Different training schemes for the back-propagation learning algorithm, such as: Scaled Conjugate Gradient (SCG), Levenberg-Marquardt (LM), Gradient Descent with Momentum (GDM), variable learning rate Back propagation (GDA) and Resilient back Propagation (RP) methods were used. The SCG algorithm with seven neurons in the hidden layer shows the best suitable algorithm with the minimum average absolute relative error 0.05231.Dans ce travail, une réseau de neurones artificiels (RNA) aété instruit et testé pour l'estimation du rendement de la synthèse directe d'oxyde de diméthyle (ODM)à partir de gaz de synthèse. On peut prédire le rendement et la sélectivité de la production de ODM et la conversion de CO lorsque la température et la pression du réacteur ainsi que le rapport molaire H2/CO de la charge de départ ontété spécifiés. Les résultats de l'estimation RNA du rendement de ODM, de la sélectivité de la production de ODM et la conversion de CO sont en bon accord avec les valeurs expérimentales. Pour ce dévelopment, une base de données de notre précédente expérimentation aété collectée. L'exactitude et la stabilité des tendances des réseaux instruits ontété testées par rapportà des données anonymes. Différents systèmes d'instruction pour l'algorithme de rétropropagation d'apprentissage ontété utilisés, par exemple, le gradient conjugué pondéré (SCG), Levenberg-Marquardt (LM), la descente de gradient avec momentum (GDM), la rétropropagationà taux d'apprentissage variable (GDA) et la rétropropagation résiliente (RP). L'algorithme SCG avec sept neurones dans la couche cachée se démontre commeétant l'algorithme le mieux approprié avec une erreur relative absolue moyenne minimale de 0,05231.

Recent Progress in Direct DME Synthesis and Potential of Bifunctional Catalysts

“…Recent approaches such as the use of artificial neural networks (ANNs) have been applied, due to their flexibility and robustness [167][168][169][170][171]. Studies regarding the application of ANNs to optimize process conditions and predict performance for the direct DME synthesis have also been reported [167,[172][173][174]. Fig.…”

Experimental methods in chemical engineering: Artificial neural networks–ANNs

“…Feed‐forward ANNs also predicted heavy oil production rates . Moradi and Parvizian estimated yield, selectivity, and conversion of synthesis gas to dimethyl ether based on temperature, pressure and /CO molar ratio in the feed with a feed‐forward ANN. They compared multiple BP algorithms, including Levenberg‐Marquardt, scaled conjugate gradient and gradient descent with momentum training methods.…”

Modeling the Direct Synthesis of Dimethyl Ether using Artificial Neural Networks

“…Using artificial neural networks (ANNs) is the most widespread machine learning approach for modeling complex phenomena due to their simple formulation, flexibility and robustness 1, 2. ANNs have proven to be suitable for creating predictive models for chemical engineering processes and several applications have been subject of research in the last decades such as the evaluation and modeling of complex kinetic data 3–6, catalyst design 7, 8, soft sensoring 1, 9, advanced process control 10, and others 11. Studies regarding the application of ANNs for the synthesis of dimethyl ether (DME) have been reported, e.g., for the screening of additives 7, 8, the optimization of temperature profiles in a temperature gradient reactor 12, and the modeling of the single process steps 13, 14.…”

“…Studies regarding the application of ANNs for the synthesis of dimethyl ether (DME) have been reported, e.g., for the screening of additives 7, 8, the optimization of temperature profiles in a temperature gradient reactor 12, and the modeling of the single process steps 13, 14. Furthermore, ANNs have been used for predicting the performance of the liquid phase direct synthesis of DME over CuO/ZnO/Al 2 O 3 and H‐ZSM‐5 catalysts 9. In this work, we used ANNs to model the direct synthesis of DME from CO 2 ‐rich synthesis gas over a mixed catalyst bed of commercial CuO/ZnO/Al 2 O 3 (CZA) and γ ‐Al 2 O 3 catalysts at high pressure.…”