Innovative energy management using time-series forecasting for atmospheric water harvesting in a semi-arid climate, combining a hybrid cascade cycle and AI

This study presents an innovative energy management strategy for atmospheric water harvesting (AWH) in semi-arid climates, combining a hybrid cascade refrigeration system with predictive artificial intelligence. The system integrates a vapor compression cycle (R1234yf), selected for its low global warming potential, and an absorption cycle (LiBr/H2O). A comprehensive multiphysics model has been developed, coupling mass and energy balances of the cascade cycle with heat and mass transfer equations in the evaporator, explicitly distinguishing sensible cooling from latent condensation. System performance is quantified through three complementary indicators: the refrigeration cycle coefficient of performance (COPFr), the overall AWH system coefficient of performance (COPAWH) incorporating recovered heat, and the specific energy consumption (Espec), which serves as the primary optimization objective. To address climate variability, a dual forecasting methodology combining Prophet and Long Short-Term Memory (LSTM) networks is implemented to predict dew point temperature (Tdew) and specific humidity (ωair) over a five-year period in Benguerir, Morocco. The LSTM model demonstrates superior predictive accuracy (R2>0.9) with well-calibrated prediction intervals. The forecasted climate data are integrated into a hierarchical control architecture that separates offline computation, where long-term projections generate a performance map, from online execution, where 48-hour forecasts enable instantaneous retrieval of optimal setpoints for evaporation temperature (TEv) and air mass flow rate (m˙air). A daily resetting mechanism ensures continuous adaptation to field conditions. The analysis explicitly accounts for the fundamental condensation condition TEv Suggested Citation Remaidi, Mohammed & Elfethi, Najoua & Ennawaoui, Amine & Mastouri, Hicham & Ennawaoui, Chouaib, 2026. "Innovative energy management using time-series forecasting for atmospheric water harvesting in a semi-arid climate, combining a hybrid cascade cycle and AI," Applied Energy, Elsevier, vol. 416(C). Handle: RePEc:eee:appene:v:416:y:2026:i:c:s0306261926006069 DOI: 10.1016/j.apenergy.2026.127954 as HTML HTML with abstract plain text plain text with abstract BibTeX RIS (EndNote, RefMan, ProCite) ReDIF JSON