A Validated Macro-Scale CFD Thermal Framework for a Temperature-Controlled Coffee Fermentation Bioreactor: Integrated...

Temperature exerts a critical influence upon coffee fermentation, driving microbial metabolic pathways, acidification kinetics, and post-harvest bean quality. This study presents an integrated computational fluid dynamics (CFD)-experimental validation framework to analyze a temperature-controlled coffee fermentation bioreactor utilizing an external water-jacket system. To establish a computationally efficient engineering design tool, transient CFD simulations were executed utilizing a specialized macro-scale bulk fluid domain. Spatial discretization integrity was verified via a rigorous grid convergence index study, yielding a low numerical uncertainty (GCI₂₁ = 0.47%). Large-scale (25 kg) experimental trials conducted across three distinct thermal boundaries (17°C-20°C, 23°C-26°C, and 32°C-35°C) confirmed excellent thermal control stability, maintaining coefficients of variation below 5% and uniformity indices above 0.95. Macro-scale energy modeling reproduced the mean measured reactor temperature with a mean absolute percentage error (MAPE) of 8.38%, a mean absolute error (MAE) of 2.16°C, and a root mean square error (RMSE) of 2.24°C. It should be emphasized that the experimental validation was limited to the volume-averaged (bulk) reactor temperature at two operating conditions; the predicted spatial temperature distribution and transient flow field were not directly validated. This indicates that a simplified bulk-domain approach can approximate macro-scale thermal performance while avoiding costly multi-phase porous media overhead. Fermentation temperature strongly influenced processing kinetics and physical bean characteristics. Operating at 32°C-35°C maximized physical acceleration, driving a rapid pH reduction (to 4.14 within 12.0 h) and lowering residual mucilage to 11.2%. These effects, together with the significantly faster fermentation duration, were statistically significant (one-way ANOVA, p < 0.001). Conversely, the 23°C-26°C range produced a statistically significant non-monotonic acidification response (p < 0.001); we hypothesize that this reflects a shift in the dominant microbial pathway within the spontaneous mixed culture, although this interpretation was not confirmed by microbiological analysis. Bulk density and bean weight did not differ significantly among treatments (p = 0.25 and p = 0.91, respectively). These findings define a heat-transfer design space for temperature-regulated agricultural bioreactors that balances computational efficiency with physical processing predictability, within the bulk-thermal scope validated here.