Process Analytical Technology Selection via Predictive Modeling and Human-informed Decision Criteria: Enabling At-line Quality Control in pMDI manufacturing

Industrial quality assessment often relies on time-consuming offline laboratory methods that delay operational response and limit timely process decisions. At-line process analytical technology (PAT) can reduce this delay by bringing measurements closer to manufacturing operations. Handheld spectroscopy is well suited to this purpose because it enables rapid, non-destructive measurements with minimal sample preparation. Its deployment, however, requires selecting both a device and a measurement protocol. Selection based primarily on predictive performance may overlook cost, measurement time, operator effort, and other implementation requirements. This study introduces a human-in-the-loop framework integrating predictive protocol optimization with stakeholderdefined multi-criteria analysis. Unlike existing approaches that treat predictive performance and operational requirements separately, the framework integrates both into configuration selection and subsequently validates the selected configuration under routine manufacturing conditions. The framework was demonstrated for predicting the color and viscosity of polymeric methylene diphenyl diisocyanate (pMDI). Six handheld spectroscopic devices were evaluated under controlled variations in lighting and container geometry as key factors in the protocol. SCiO combined with support vector regression (SVR) provided the strongest predictions for both properties. The analytic hierarchy process integrated model-derived performance with stakeholder assessments of cost, time, and operator effort. Following production-floor implementation, the selected configuration yielded R2 = 0.91 and RMSEP = 0.7 for color, and R2 = 0.87 and RMSEP = 39.46 cSt for viscosity. This framework provides a transferable approach for selecting and implementing PAT configurations that meet the technical and operational requirements for industrial adoption.

Authors

Institutions

Publication Details

Journal
ChemRxiv
Published
2026-11-12
DOI
https://doi.org/10.26434/chemrxiv.15012067/v1
Primary Topic
Spectroscopy and Chemometric Analyses
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Process Analytical Technology Selection via Predictive Modeling and Human-informed Decision Criteria: Enabling At-line Quality Control in pMDI manufacturing

Jeroen Jasper Jansen, Sin Yong Teng, Gerjen Herman Tinnevelt, Diogo Barros Goncalves et al.
ChemRxiv
Spectroscopy and Chemometric Analyses
preprint

Process Analytical Technology Selection via Predictive Modeling and Human-informed Decision Criteria: Enabling At-line Quality Control in pMDI manufacturing

Jeroen Jasper Jansen, Sin Yong Teng, Gerjen Herman Tinnevelt, Diogo Barros Goncalves, Sanwouly Marlene Yao, Ivo Jacobs
preprint en

Abstract

Industrial quality assessment often relies on time-consuming offline laboratory methods that delay operational response and limit timely process decisions. At-line process analytical technology (PAT) can reduce this delay by bringing measurements closer to manufacturing operations. Handheld spectroscopy is well suited to this purpose because it enables rapid, non-destructive measurements with minimal sample preparation. Its deployment, however, requires selecting both a device and a measurement protocol. Selection based primarily on predictive performance may overlook cost, measurement time, operator effort, and other implementation requirements. This study introduces a human-in-the-loop framework integrating predictive protocol optimization with stakeholderdefined multi-criteria analysis. Unlike existing approaches that treat predictive performance and operational requirements separately, the framework integrates both into configuration selection and subsequently validates the selected configuration under routine manufacturing conditions. The framework was demonstrated for predicting the color and viscosity of polymeric methylene diphenyl diisocyanate (pMDI). Six handheld spectroscopic devices were evaluated under controlled variations in lighting and container geometry as key factors in the protocol. SCiO combined with support vector regression (SVR) provided the strongest predictions for both properties. The analytic hierarchy process integrated model-derived performance with stakeholder assessments of cost, time, and operator effort. Following production-floor implementation, the selected configuration yielded R2 = 0.91 and RMSEP = 0.7 for color, and R2 = 0.87 and RMSEP = 39.46 cSt for viscosity. This framework provides a transferable approach for selecting and implementing PAT configurations that meet the technical and operational requirements for industrial adoption.

ChemRxiv
Radboud University Nijmegen (NL), Huntsman (United States) (US), Maastricht University (NL)
Ministerie van Economische Zaken en Klimaat
Industry, innovation and infrastructure
Spectroscopy and Chemometric Analyses
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon