Dynamic Estimation of Overall Heat Transfer Coefficient in Temperature Oscillation Calorimetry Based on Co-Frequency Response Identification

Abstract

Accurate and real-time evaluation of the overall heat transfer coefficient (UA) is a critical prerequisite for reliable heat flow calculation and thermal risk assessment in automatic reaction calorimetry. While Temperature Oscillation Calorimetry (TOC) has been widely adopted for online (UA) estimation, the conventional phasor-based TOC method is highly susceptible to non-steady-state baseline drifts, often leading to severe estimation deviations or non-physical oscillations during strong exothermic processes. To overcome this limitation, this paper proposes a robust dynamic (UA) estimation method based on sliding-window co-frequency response identification combined with physical regularization constraints. The proposed method jointly utilizes two components to improve dynamic estimation. First, by establishing an identification relationship strictly at the fundamental excitation frequency, the co-frequency response identification effectively isolates non-steady-state baseline interferences. Second, the regularized physical constraints (including temporal continuity and calibration-based lower-bounds) suppress non-physical fluctuations and ensure globally optimal, physically plausible (UA) estimates. The method was experimentally validated using a controllable electric heating system under steady low-power, time-varying, and severe exothermic dynamic conditions, as well as static calibrations from 30°C to 90°C. Results demonstrate that the proposed method significantly suppresses transient jitter and non-physical fluctuations, maintaining a highly accurate tracking of (UA) and reaction heat release rate (Qr) even under severe thermal inertia and intense heat generation. This robust approach substantially improves the reliability of dynamic calorimetric data, laying a solid foundation for downstream kinetic parameter evaluation and chemical process safety analysis.