Objective

Applying spectral data analysis and real-time analytics to improve process monitoring and operational efficiency

Background

As outlined in the first case study, BLX Chemicals, a leading producer of high-quality resin, made significant progress improving both the process stability and capability. Using a process model, better operating conditions (process settings) could be implemented leading to better purity, which is considered as the critical quality attribute.

The primary objective of this second project is the elimination of the offline purity measurement, leading to a more efficient production process. As a result, we hope to replace the offline purity monitoring step with online monitoring based on spectral (NIR) data. Near-infrared (NIR) analysis is a fast, nondestructive technique that measures how a material absorbs light in the near-infrared region to determine its chemical and physical properties. In offline measurements, samples are taken to a laboratory instrument for a detailed analysis. In online measurements, NIR sensors are integrated directly into the production process, enabling continuous, real-time monitoring and rapid adjustments and making NIR a powerful tool for improving quality, efficiency, and process control.

The Task

• Explore the spectral NIR data.
• Build and assess purity-NIR models:
• DOE data: Build a first model (PLS).
• With more data: Explore modeling options and select a purity-NIR model.