Abstract: Dynamic desorption of water molecule from the capillary wick comprising highly porous nickel-powder sintered and coated on the inner surfaces of loop heat pipe(LHP),an advanced thermal control device in spacecraft and submarine,was empirically approximated,mathematically formulated with dynamic/neural-network/grey-forecasting models based on the experimental data measured and/or reported in literature,theoretically analyzed and numerically simulated to predict the service lifetime of LHP. The simulated results show that when it comes to time-evolution of water desorption rate/quantity,a descending order can be listed:2^nd-order dynamic model(124.90×10^-6 and 150 d) > neural-network model(109.60×10^-6 and 75 d) > 1^st-order dynamic model(81.49×10^-6 and 30 d) > grey-forecasting model(59.06×10^-6 and 24 d).We suggest that neural-network model may best describe the dynamic water-desorption characteristics from capillary wick in LHP.