Scholarlink

<p>This study presents an advanced optimization of the Claus process in the SRU case study to enhance sulfur recovery efficiency and environmental compliance. Using Aspen HYSYS V12 with the Acid Gas property package, we simulated the process for a feed containing 38.93% H₂S, 54.16% CO₂, and trace hydrocarbons. Dynamic sensitivity analysis identified optimal conditions (205&#xa0;°C, 1.85 barg, O₂:H₂S ratio of 0.51:1), achieving a sulfur recovery efficiency of 96.7% (989.6&#xa0;kg/h), reducing energy consumption by 12% (-18.63 × 10⁶ kcal/h), and maintaining H₂S content below 3.3%, compliant with Euro VI standards, without a tail gas treatment unit. A Random Forest machine learning model, trained on 200 simulation points, predicted sulfur recovery and H₂S content across varied feed compositions (H₂S: 20–50%) with 95% accuracy against SRU case study operational data. The model leverages real data from three catalytic converters and precise Al₂O₃ catalyst kinetics (length 4.914&#xa0;cm, diameter 8.365&#xa0;cm, bulk density 3789&#xa0;kg/m³, particle size 6&#xa0;mm, porosity 0.859). This approach offers a cost-effective, sustainable framework for SRUs, eliminating the need for costly tail gas treatment while meeting stringent environmental regulations. Detailed codes, simulation inputs, and data sources are provided in the Supporting Information.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimization of the Claus Process Using Machine Learning and Real Operational Data: A Case Study in an SRU

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Amirfarzam Shokouhalaei

摘要

This study presents an advanced optimization of the Claus process in the SRU case study to enhance sulfur recovery efficiency and environmental compliance. Using Aspen HYSYS V12 with the Acid Gas property package, we simulated the process for a feed containing 38.93% H₂S, 54.16% CO₂, and trace hydrocarbons. Dynamic sensitivity analysis identified optimal conditions (205 °C, 1.85 barg, O₂:H₂S ratio of 0.51:1), achieving a sulfur recovery efficiency of 96.7% (989.6 kg/h), reducing energy consumption by 12% (-18.63 × 10⁶ kcal/h), and maintaining H₂S content below 3.3%, compliant with Euro VI standards, without a tail gas treatment unit. A Random Forest machine learning model, trained on 200 simulation points, predicted sulfur recovery and H₂S content across varied feed compositions (H₂S: 20–50%) with 95% accuracy against SRU case study operational data. The model leverages real data from three catalytic converters and precise Al₂O₃ catalyst kinetics (length 4.914 cm, diameter 8.365 cm, bulk density 3789 kg/m³, particle size 6 mm, porosity 0.859). This approach offers a cost-effective, sustainable framework for SRUs, eliminating the need for costly tail gas treatment while meeting stringent environmental regulations. Detailed codes, simulation inputs, and data sources are provided in the Supporting Information.