Strong lensing is a unique tool for investigating the inner structure of
galaxy clusters. Arc-like distorted images of distant galaxies can be used to
trace any kind of matter in the central regions of massive
structures. Numerous techniques have been developed at this
purpose. Unfortunately, the comparison between theoretical expectations and
observations is made difficult by the complexity of cluster
lenses. Substructures, asymmetries, ellipticity, dynamical activity are all
important factors which play a role in determining the strong lensing
properties of clusters. Moreover, interfacing observations and theoretical
predictions is difficult because of noises and practical issues which have
never been properly taken into account in theoretical studies. We propose a
way to close this gap, by making ray-tracing through numerically simulated
clusters, mimicking observational noises and including realistic modeling of
the background sources. We illustrate some ideas for addressing some issues
which are still under debate.