The process chemometrics group develops chemometric methods for the monitoring of key variables and the modeling of reaction systems using non-specific, highly redundant measurements.
For the monitoring of key variables, calibration and classification models are typically built off-line using a database consisting of (i) non-specific measurements (infrared (IR), near-infrared (NIR), UV, Raman, mass spectroscopy, nuclear magnetic resonance, magnetic resonance imaging, etc.), and (ii) the corresponding specific measurements of key variables such as product concentrations or product identity. These models require adaptation due to drifts and measurement noise. Indeed, drifts, which result from changes in the process, the measurement device, or its environment, can lead to bias in on-line prediction of the key variables, whereas measurement noise leads to variance in the prediction error. We investigate update schemes for such history-based models in order to keep them alive on-line despite drift and significant measurement noise. Reaction systems such as hydrogenations, oxidations, and fermentations are used in the production of various important chemical and pharmaceutical products. Modeling of such reaction systems is an essential task during the process development stage with the main goals being (i) a better understanding of the reaction system through model-based experimental analysis, and (ii) process development and long-term performance improvement. Advanced instruments such as calorimeters and spectrometers (IR, NIR, Raman, mass spectroscopy, etc.) enable the indirect measurement of key variables such as the heat flow and the concentrations of initial and final products in-situ or at-line. We develop process chemometric methods for the incremental modeling of reaction systems and the kinetic modeling of microreactors using multi-sensor data.
Keywords : Data analysis; modeling; gas-liquid reaction systems; homogeneous reaction systems; spectroscopic data; calorimetric data; multi-sensor data;
Keywords : Data analysis; modeling; gas-liquid reaction systems; homogeneous reaction systems; inlets; outlets; spectroscopic data; calorimetric data; multi-sensor data;
Gas-liquid reaction systems such as hydroformylation, alkylation, carboxylation, polymerisation, hydrometallurgy and fermentation are key in producing industrial chemical, pharmaceutical and biochemical products. The proposed research deals with the development of a modeling framework for gas-liquid reaction systems and involves an incremental modeling approach using multi-sensor data. [More]
Keywords :Spectral measurements; multivariate calibration; drift invariance; orthogonal projection;
Two fundamental differences between drift correction methods such as generalized least squares (GLS) and dynamic orthogonal projection (DOP) concern (1) how to estimate the drift components using reference/replicate measurements, and (2) whether calibration data lying in these components be scaled down partially or fully, e.g. through orthogonal projection. In this project, the second difference is investigated using real-life data. [More]