Work

Development of Automated Platform for Rapid Kinetic Studies of Gas-Solid Reactions

Year: 2025
Type: dissertation
Abstract: Industry 4.0 has transformed chemical manufacturing, particularly in reactor design and process automation, with digitalisation playing a pivotal role in driving global interest. Kinetic models are crucial for reliable reactor design, yet their applicability can be limited by the uncertainty associated with model predictions within the design space. To address this, an automated flow micropacked bed catalytic reactor platform was developed for the rapid screening of kinetic models through pre-planned and optimal experiments. This is a gas-solid system, involving high temperature operation and can operate with very small amount of catalyst. The microreactors were fabricated using photolithography and deep reactive ion etching of a silicon wafer and were packed with 5 wt.% Pd/Al2O3 catalyst for total oxidation and 5 wt.% Pt/Al2O3 catalyst for partial oxidation of methane. During each experimental campaign, the platform automatically adjusted conditions, and product stream analysis was conducted via online gas chromatography. The system was monitored and controlled by LabVIEW, integrated with Python scripts for online experimental design and data analysis. The optimal design of experiments employed techniques such as model-based design of experiments (MBDoE) for model discrimination and parameter precision, along with explorative model-based design of experiments (eMBDoE). These methods were applied to the complete oxidation of methane using an automated micropacked bed reactor. By conducting automated experiments guided by MBDoE, a suitable kinetic model—the Mars-van Krevelen model—was identified with precise parameter estimates within two days. Furthermore, the eMBDoE approach significantly reduced model prediction variance and minimised experimental burden, achieving statistically satisfactory parameters with only five optimal experiments—far fewer than the 23 typically required by conventional MBDoE, thus reducing the experimental burden by 78%. A digital twin for the automated microreactor platform was successfully demonstrated. Physical experimental results were compared with virtual system predictions, validating the digital twin's accuracy and its ability to closely replicate the physical system's behaviour. This autonomous online model identification approach provides precise predictive information for model-based decision support in process design and operation for a single gas-solid reaction.
Language: English
Open Access status: green
OpenAlex ID: W7065139909
Cites: 212
Cited by:
Related to: 10