Analisis Model Matematika Newton dan Page pada Pengeringan Lapis Tipis Pisang Candi (Musa paradisiacal Fa Corniculata) Menggunakan Oven Konveksi

Abstract

Candi banana (Musa paradisiaca Fa Corniculata) is a local horticultural commodity prone to rapid post-harvest decay due to its high initial moisture content of 68–78%. Diversifying it into unripe banana flour extends shelf life while leveraging its high resistant starch content of 76–84%, which provides significant functional health benefits. The transition from conventional open-sun drying to controlled convection oven drying at the small and medium enterprise (UMKM) scale necessitates precise mathematical modeling to optimize operational parameters and ensure consistent product quality. This study aims to analyze the thin-layer drying characteristics of 3 mm thick unripe Candi banana slices (150 g) using a convection oven at 50°C, 60°C, and 70°C, and to comparatively evaluate the accuracy of the Newton and Page mathematical models in predicting the moisture reduction rate. Each temperature treatment was conducted with four replications (three observation replicates and one validation replicate), with mass reduction observed at 15 fixed time intervals from 0 to 1590 minutes until equilibrium moisture content (<12% db, per SNI) was reached. Drying kinetics were evaluated based on the Moisture Ratio (MR) and validated using the Coefficient of Determination (R²), Root Mean Square Error (RMSE), and visual parity plots with a ±10% tolerance limit. The results indicated that the water loss rate is directly proportional to the drying temperature, with 70°C generating the fastest drying completion near minute 300, attributed to high thermal energy and a wider partial vapor pressure difference (driving force) arising from reduced relative humidity at elevated temperatures. Modeling evaluation demonstrated that the Page model is significantly more accurate and superior to the Newton model. The Page model consistently yielded high R² values (0.989–0.994) and low RMSE values (0.032–0.050), with all data points falling within the ±10% error tolerance band in visual parity plot validation. In contrast, the Newton model exhibited significant underprediction deviations, particularly at 70°C, with R² values as low as 0.951 and RMSE values reaching 0.117. In conclusion, the Page model is recommended as the most accurate mathematical instrument for predicting the thin-layer drying kinetics of unripe Candi banana slices using a convection oven at 50–70°C.