Our review illustrates how similar types of structural settings have been modeled using various designs. Comparison between these experimental set-ups reveals the strengths and limits of each design (Table 1) and illustrates the progressive improvement in designing models whose geometric and rheological properties are analogous to those of the geological prototypes.

In the above discussion, we have shown how the first set of models using a brittle layer resting directly above a basal sheet (Faugère et al., 1984; McClay and Ellis, 1987; Vendeville, 1991) controlled the location, orientation, and sense of asymmetry of faults in pull-apart basins or in the relay zones. The design by Sims et al. (1999) represents an improvement of this set-up by incorporating a basal viscous layer. This design, however still constrains too much fault location and orientation in the model. Although the third design by Le Calvez and Vendeville (1996) allowed to effectively insulate the brittle layer from the model base, it still does not permit significant fault-block vertical movements and rotations. The new set-up illustrated in this article represents only the latest step in the evolution of model design and will with no doubt be superceded by newer designs in the future. With this design, experimenters can confidently model the interaction of normal faults as they propagate along-strike during extension above a weaker viscous layer. The same design could easily be adapted to study fault interaction in different tectonic regimes, such as regional strike-slip or compressional deformation.