Beyond ideal membranes: Multiphysics modelling of vacuum-driven graphene oxide dehumidification systems

Space cooling accounts for a growing share of global electricity demand, with latent loads dominating in humid climates. Idealized models predict membrane dehumidification can raise moisture removal efficiency from roughly 5 to 55 kg kWh −1 relative to vapor compression, but assume infinite selectivity and negligible polarization. This study relaxes those assumptions using a three-dimensional finite-element multiphysics model of a pilot-scale graphene oxide system, in which a 0.33 m 2 dehumidification module is coupled to a 0.35 m 2 downstream moisture-rejection module. Simulations spanned selectivities of 100 to 2000 and permeate pressures of 0.1 to 3 kPa at a water permeance of 1 × 10 −5 mol m −2 s −1 Pa −1 . Because higher selectivity suppresses the transmembrane driving force, a selectivity of 100 outperformed 2000 by 16% at 1 kPa, though at twice the pumping speed. Deeper vacuum yielded diminishing returns. Lowering the permeate pressure from 1 kPa to 100 Pa improved moisture removal by only 37% while pumping speed rose roughly sevenfold, even at high selectivity. In the rejection module, coefficient of performance and second-law efficiency peaked at about 11 and 36% under ideal compression, but only near 2 × 10 −3 m 3 /s, and decayed rapidly as throughput increased. With commercially available pumps, the coefficient of performance fell below 0.5, indicating that vacuum pump losses, rather than membrane selectivity or permeance, govern practical viability at the investigated scale. By resolving module hydrodynamics while closing the system energy balance with real pump characteristics, the present framework reveals performance penalties that neither module-scale nor system-scale analyses capture alone.