Tu-CHS-12 - A Physics-Informed Inverse Problem for Tracking Gas Chromatography Column Degradation

Event

EUROSENSORS 2026
2026-09-06 - 2026-09-09
Zurich

Band

Poster

Chapter

Chemical Sensors

Author(s)

A. Riaud, B. Toth, Y. Maret, K. Hencken - ABB Corporate Research Switzerland,Baden-Dättwil (Switzerland)

Pages

441 - 442

DOI

10.5162/eurosensors2026/Tu-CHS-12

ISBN

978-3-910600-12-6

Price

free

Abstract

Many sensing systems require inferring a hidden system state from a distorted waveform. In process gas chromatography, column degradation changes the shape of chromatographic peaks, but the underly-ing physics is nonlinear and difficult to invert. We cast column health estimation as an inverse problem and solve it with a simple physics informed model. The model links peak asymmetry to Langmuir ad-sorption and converts the inferred loading behavior into a column health metric. On two years of indus-trial data from six instruments, this metric tracked aging and separated degraded columns early enough to support predictive maintenance.