Optimal stopping time for thermal processing of canned foods under power failure events: Correction for real time implementation

Unexpected disturbances during the sustained (holding) phase of thermal sterilization complicate the decision of when a batch can be safely released without resorting to overly conservative margins. We present a model-based stopping rule that computes the minimal process extension required to reach the desired lethality at the product cold spot. The method combines a heat-transfer model with a lethality accumulator and an optimization procedure that evaluates the actual measured retort trajectory during and after the disturbance. To enable fast evaluation, a projection-based reduced-order model was developed using Proper Orthogonal Decomposition. The reduced model reproduced cold-spot temperature and final lethality within operational accuracy of full finite-element simulations while achieving approximately an order-of-magnitude (tenfold) reduction in computation time. Case studies on conduction-dominated, axisymmetric cans showed that the optimal stopping condition follows an approximately linear relation between cold-spot temperature and accumulated lethality, which collapsed into a universal non-dimensional rule with R 2 = 0 . 98 across different food products. These results demonstrate that the proposed approach enables safe corrections of thermal sterilization processes under power failure events. • Model-based rule determines optimal stopping time after retort failures. • Reduced-order POD model achieves tenfold faster simulation accuracy. • Linear relation found between cold-spot temperature and lethality. • Universal nondimensional stopping rule applies across food types. • Method enables safe, energy-efficient real-time process correction.