System dynamics (SD) is one of the most suitable methods available to simulate and quantify the behaviour of complex systems due to it non-linear, multivariable, information feedback and temporally changing characteristics. This paper has improved on Khan et al.'s established model (see Khan et al. (2009) ‘Analyzing complex behavior of hydrological systems through a system dynamics approach’, Environmental Modeling & Software 24, 1363–1372) by adding water production functions for three main crops (winter wheat, summer maize and cotton) based on soil water balance at the field level. Conclusions based on the simulation were as follows. (1) The model could simulate the dynamics of water balance components of winter wheat and summer maize relatively accurately through testing and validation. (2) Two to four irrigations were needed for the water-use requirement for winter wheat and summer maize when irrigation level was 75 mm each time. Cotton did not need to be irrigated except for the addition of 60 mm pre-sowing water. (3) Exploiting groundwater as far as possible and taking measures to reduce soil evaporation and at the same keeping irrigation unchanged was one of the best ways for sustainable utilization of water resources in the upper reaches of the Panzhuang District. (4) Water-use efficiencies were consistent with the regions' measured results showing that the model could simulate the water cycle in the field comparatively accurately.