The DYNAMELT project proposes an experimental and numerical study of the melting dynamics in multiphase, eutectic and peritectic alloys (multiphase melting). It aims at a first comprehensive investigation of pattern formation and microstructure selection during melting and melting/solidification (M/S) processes involving two solid phases. By combining experimental and numerical methods, it takes advantage of the winning methodological strategy in solidification science. Fundamental and engineering-oriented aspects of interdisciplinary interest for nonlinear physics and materials science will be addressed. Main objectives are 1-to unveil elementary mechanisms during early stages of multiphase melting, 2-to analyse coupled-melting patterns in peritectic and eutectic alloys, and, on this new basis, 3- to cast light on the memory effects during M/S cycles. Despite an apparent similarity, asymmetries exist between melting and solidification concerning nucleation and migration of interfaces, as well as solute redistribution. One striking example is the markedly low chemical diffusivity in solid phases that causes the freezing of microsegregation patterns. These inhomogeneous microstructures affect the partial melting dynamics and M/S cycles, and thus change materials properties on a large scale. The project addresses multiphase melting by controlled experiments employing temperature gradients, thin films as well as bulk samples with tailored microstructures, along with phase-field numerical simulations considering moving boundary conditions and diffusive couplings of interface motion on the relevant scaling lengths. Focus will be made on in situ diagnostics to investigate hitherto unresolved dynamic processes occurring during melting of multiphase alloys. The work program is based on a well-established expertise of the French-German consortium in the science of solidification and melting, with synergic interaction between experiments and modeling.