In polymer manufacturing, color index values determine product quality.

But no governing equations exist for predicting L*a*b* coordinates in real time.

Additives, residual catalysts, subtle variations in polymer microstructure are some of the factors that influence color in ways that first-principles models alone cannot capture.

SIMACRO collaborated with a polymer process technology company to address this. We built a hybrid architecture:

→ First-principles model recalibrated to current feedstock and equipment conditions
→ ML-based error correction bridging simulation-to-plant discrepancies
→ AI-driven soft-sensor data prediction of L*a*b* color indices in real time

As a result, quality management shifted from post-production lab analysis to proactive process control.

We are presenting this work at Aspen Technology#OPTIMIZE26 in Houston (May 11–14) and running an exhibition booth throughout the conference. If hybrid modeling for real-time quality prediction is on your radar, we would welcome the chance to connect.