Study on the optimization of lychee quick-freezing process parameters and damage assessment model based on response surface methodology

To address low-temperature damage in lychee during liquid nitrogen quick-freezing, this study proposed a Box-Behnken optimization method and developed a thermo-mechanical model to reduce and predict damage under various conditions. Guiwei lychee was used in this study as preliminary experiments revealed that it exhibited the highest cracking rate compared to other lychee varieties. The optimal freezing times and pretreatment methods were assessed and determined at different freezing temperatures. Box-Behnken response surface optimization identified the optimal parameters as 60-min pretreatment, -52 °C freezing temperature, and 812 rpm fan speed. Under these conditions, the cracking rate was 32% and the freezing time was 20.6 min, with absolute errors of 8.9% and 9.8% compared to the predicted results. In addition, physical parameters such as moisture content, hardness, and fluid velocity of lychee were measured, and the porosity of each structure and the convective heat transfer coefficient at the fluid–solid interface were calculated. A damage model coupling solid mechanics and heat transfer was developed, with mechanical parameters (elastic modulus of 16.74 MPa and tensile strength of 0.611 MPa) determined using a central composite design. The objective weighting method (CRITIC) predicted a cracking rate of 56.03% at -40 °C and 1000 rpm, representing an absolute error of 7.76% compared to the experimental value of 52%. These findings provide a theoretical basis and practical reference for addressing freezing damage in the food industry. • Our 60-minute pretreatment reduces cracking by 8% vs. 120-minutes in literature • Peel and kernel moisture cannot be neglected in establishing thermophysical property equations • The lychee freezing model demonstrates good agreement with actual measurements • High correlation exists between predicted model and experimental data (R=0.9983) • A Biot number greater than 0.1 reveals a significant temperature gradient • Damage assessment model based on CCD demonstrates good predictive efficiency