{"title":"Predicting nickel catalyst deactivation in biogas steam and dry reforming for hydrogen production using machine learning","authors":"","doi":"10.1016/j.psep.2024.09.020","DOIUrl":null,"url":null,"abstract":"<div><div>This study employs Random Forests (RF) and Artificial Neural Networks (ANN) to model the transient behavior of Ni catalyst deactivation during steam and dry reforming of model biogas containing H<sub>2</sub>S, with a focus on hydrogen production. Deactivation, induced by carbon deposition and sulfur poisoning, is a complex and transient phenomenon demanding precise kinetic mechanisms for accurately predicting Ni catalyst behavior in biogas reforming. Black-box machine learning (ML) models are developed, incorporating catalyst properties, biogas composition, and operating conditions. Encompassing both dry and steam reforming, the ML models aim to predict catalyst behavior, expressed in terms of packed bed reactor exit mole fractions (H<sub>2</sub>, CO, CH<sub>4</sub>, and CO<sub>2</sub>) and conversions (CH<sub>4</sub> and CO<sub>2</sub>). The ML models are trained and tested across a temperature range of 700–900 C with 0–145 ppm of H<sub>2</sub>S in model biogas (CH<sub>4</sub>/CO<sub>2</sub> ratio varying from 1.0 to 2.0). RF outperforms the ANN across all performance metrics, including overall R<sup>2</sup> and root mean squared error (RMSE). The RF achieves a mean overall R<sup>2</sup> of 0.979, with training and testing RMSE equal to <span><math><mrow><mn>6.7</mn><mo>×</mo><msup><mrow><mn>10</mn></mrow><mrow><mo>−</mo><mn>3</mn></mrow></msup></mrow></math></span> and <span><math><mrow><mn>1.47</mn><mo>×</mo><msup><mrow><mn>10</mn></mrow><mrow><mo>−</mo><mn>2</mn></mrow></msup></mrow></math></span> respectively. In contrast, the ANN achieves a mean overall R<sup>2</sup> of 0.939, with training and testing RMSE equal to <span><math><mrow><mn>2.6</mn><mo>×</mo><msup><mrow><mn>10</mn></mrow><mrow><mo>−</mo><mn>2</mn></mrow></msup></mrow></math></span> and <span><math><mrow><mn>2.55</mn><mo>×</mo><msup><mrow><mn>10</mn></mrow><mrow><mo>−</mo><mn>2</mn></mrow></msup></mrow></math></span> respectively. Moreover, pre-trained RF models are validated with unseen data of dry reforming of biogas containing 30 ppm of H<sub>2</sub>S (25 data points). It is suggested that 35 % of this unseen experimental data is required to train the RF model for it to predict catalyst deactivation, achieving a validation R sufficiently<sup>2</sup>> 0.9. The mean overall R<sup>2</sup> values attained by the RF fine-tuned on 35 % of the unseen experiment data for both CH<sub>4</sub> and CO<sub>2</sub> conversions, as well as for all mole fraction predictions, are 0.952 and 0.948, respectively.</div></div>","PeriodicalId":20743,"journal":{"name":"Process Safety and Environmental Protection","volume":null,"pages":null},"PeriodicalIF":6.9000,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Process Safety and Environmental Protection","FirstCategoryId":"93","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S095758202401139X","RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENGINEERING, CHEMICAL","Score":null,"Total":0}
引用次数: 0
Abstract
This study employs Random Forests (RF) and Artificial Neural Networks (ANN) to model the transient behavior of Ni catalyst deactivation during steam and dry reforming of model biogas containing H2S, with a focus on hydrogen production. Deactivation, induced by carbon deposition and sulfur poisoning, is a complex and transient phenomenon demanding precise kinetic mechanisms for accurately predicting Ni catalyst behavior in biogas reforming. Black-box machine learning (ML) models are developed, incorporating catalyst properties, biogas composition, and operating conditions. Encompassing both dry and steam reforming, the ML models aim to predict catalyst behavior, expressed in terms of packed bed reactor exit mole fractions (H2, CO, CH4, and CO2) and conversions (CH4 and CO2). The ML models are trained and tested across a temperature range of 700–900 C with 0–145 ppm of H2S in model biogas (CH4/CO2 ratio varying from 1.0 to 2.0). RF outperforms the ANN across all performance metrics, including overall R2 and root mean squared error (RMSE). The RF achieves a mean overall R2 of 0.979, with training and testing RMSE equal to and respectively. In contrast, the ANN achieves a mean overall R2 of 0.939, with training and testing RMSE equal to and respectively. Moreover, pre-trained RF models are validated with unseen data of dry reforming of biogas containing 30 ppm of H2S (25 data points). It is suggested that 35 % of this unseen experimental data is required to train the RF model for it to predict catalyst deactivation, achieving a validation R sufficiently2> 0.9. The mean overall R2 values attained by the RF fine-tuned on 35 % of the unseen experiment data for both CH4 and CO2 conversions, as well as for all mole fraction predictions, are 0.952 and 0.948, respectively.
期刊介绍:
The Process Safety and Environmental Protection (PSEP) journal is a leading international publication that focuses on the publication of high-quality, original research papers in the field of engineering, specifically those related to the safety of industrial processes and environmental protection. The journal encourages submissions that present new developments in safety and environmental aspects, particularly those that show how research findings can be applied in process engineering design and practice.
PSEP is particularly interested in research that brings fresh perspectives to established engineering principles, identifies unsolved problems, or suggests directions for future research. The journal also values contributions that push the boundaries of traditional engineering and welcomes multidisciplinary papers.
PSEP's articles are abstracted and indexed by a range of databases and services, which helps to ensure that the journal's research is accessible and recognized in the academic and professional communities. These databases include ANTE, Chemical Abstracts, Chemical Hazards in Industry, Current Contents, Elsevier Engineering Information database, Pascal Francis, Web of Science, Scopus, Engineering Information Database EnCompass LIT (Elsevier), and INSPEC. This wide coverage facilitates the dissemination of the journal's content to a global audience interested in process safety and environmental engineering.