PLANT DATA ANALYSIS OF PRE-RINSE IN CLEANING-IN-PLACE (CIP) PROCEDURES FOR AN ULTRA-HIGH TEMPERATURE (UHT) UNIT OPERATION

Sammanfattning

CIP systems are critical for ensuring hygienic integrity in ultra-high temperature (UHT) food processing. However, conventional cleaning cycles are often rigidly timed and overdesigned, leading to unnecessary water, energy, and chemical use. This thesis presents a data-driven framework to evaluate and optimize the prerinse phase of UHT CIP cycles using pressure drop behavior as an alternative for soil removal efficiency. Using plant data, an empirical model for clean-system system resistance (Ksys) was developed based on sterilization-phase measurements. This model, along with theoretical formulations based on the Darcy–Weisbach equation and two friction factor estimators (Swamee–Jain and Colebrook–White), was used to calculate pressure drop during cleaning. However, due to underprediction and inconsistency in the theoretical methods, only the proprietary model was used for RSL calculations. RSL was computed as a dimensionless ratio of observed to clean-state pressure drop, and RSLdrop defined from the start of prerinse to the start of caustic, was used to assess cleaning effectiveness. Regression models were developed to relate RSLdrop to prerinse flow rate, temperature, and duration, using production-linked CIP cycles with ≥360 minutes of prior operation to ensure relevant fouling. These were done for 2 specific groups split based on prerinse durations: •For short prerinse durations (≤20 min), cleaning performance was highly sensitive to flow rate and temperature. Regression fits showed moderate predictability (R² = 0.82), but inconsistent RSLdrop behaviour suggested incomplete or unstable soil removal under suboptimal rinse exposure. •For extended prerinse durations (>20 min), models showed stronger consistency and higher accuracy (R² = 0.87), indicating that longer exposure times allow for more complete hydration, softening, and dislodgement of fouling layers.