The space environment modeling group at the University of Washington has over the last decade developed a three-dimensional multi-fluid code simulating the interaction of the terrestrial magnetosphere with the solar wind.
With the development of the multi-fluid model, the evolution of different sources of ions could be tracked and this work lead to the first three-dimensional identification of the geopause [Winglee, 1998], as well as a description of the relative roles of ionospheric and solar wind plasma in populating the magnetosphere [Winglee, 2000], the importance of ionospheric mass outflows in mass loading of the magnetosphere, the generation of the Harang discontinuity, and the cross polar cap potential [Winglee et al., 2002; 2004]. Subsequent modeling has demonstrated that the timing and spatial distribution of the heavy ion outflows from the model are consistent with IMAGE/HENA data [Winglee et al., 2005]
In addition to multi-fluid modeling of the terrestrial magnetosphere, single-particle tracking using time-dependent global magnetic and electric fields have been used by Cash et al.  to investigate the generation of the ring current from ionospheric outflows during an internally driven substorm, as well as modeling the 10 March 1998 storm to investigate storm time acceleration, injection and trapping mechanisms associated with the formation of the ring current.
Harnett et al.  used these high resolution simulations to model the 26 February 2008 substorm that was well observed by the THEMIS spacecraft. The interaction of fast flows at the inner edge of the plasma sheet is closely associated with auroral onset, indicating that the substorm is internally triggered. This multi-scale/multi-fluid model of the terrestrial magnetosphere has been the first to model both tailward and earthward moving flux ropes and to investigate the relative timing of substorm process [Winglee et al., 2009] and the influence of heavy ionospheric ions on substorm onset [Winglee and Harnett, 2011]
Multi-scale/multi-fluid simulations have been developed to provide the first high resolution (~120 km) simulations of the thin tail current sheet within the global framework of the Earth's magnetosphere [Harnett et al., 2006] and simulations of flux ropes at the magnetopause within a global magnetospheric model [Winglee et al., 2008]