Partially molten lower-crustal domains are key features of many large hot orogens, including the Altai tract of the CAOB, Tibet, and the Altiplano–Puna, where seismic imaging reveals extensive low-velocity, high-temperature regions interpreted as crustal mush zones. These regions form and evolve under the combined influence of crustal thickening, mantle–lithosphere removal, and thermally driven weakening, which together promote melt generation, rheological softening, and decoupling between crustal levels, thereby enabling lateral and vertical redistribution of material. Yet, the feedback between melt migration, deformation, and mechanically anisotropic, layered mush systems remains insufficiently quantified.In this contribution, we compare different types of models that approximate the crustal evolution of hot orogens: semi-scaled thermal paraffin-wax analogue models with visco-plastic finite-element numerical models that respect the laboratory scale and employ similar material properties. In addition, orogen-scale control numerical models were performed to mimic natural prototypes. The models were designed to explicitly address (i) crustal strength, shortening, and basal heating, and (ii) how these parameters govern internal deformation, melt coalescence, and plumbing-system architecture.Analogue experiments successfully reproduced melt migration and its coalescence in buckled and folded layers, suggesting that mechanically layered mush domains can localize melt along fold hinges, shear zones, and evolving permeability pathways, with potential transitions from dominantly lateral to buoyancy-driven melt transfer as deformation proceeds. However, the analogue models lack rigorous and precise quantification of the thermal field, and melt migration is tracked approximately, using image analysis based on DIC and X-ray CT-scanning methods. Furthermore, precise dynamical scaling is currently limited by the absence of a representative rheological law based on the complex rheometry of the paraffins used, restricting us to discrete viscosity measurements for different velocities and temperatures.The first step in the numerical approach was to reproduce the analogue model geometry and melt distribution. Simulations successfully mimic the development of melt-rich regions at both laboratory and orogen scales within a range of natural parameter values, but they do not reproduce the degree of melt coalescence within crustal layers that occurs naturally in the analogue models; nevertheless, these weak zones promote further strain distribution and strongly influence the overall crustal deformation response. An important advantage of the numerical approach is the well-constrained control on the thermal regime and temperature–melt evolution in the system. For the tested parameters, however, the numerical models yield a more homogeneously thickened crust without significant buckling of the lower crust, and a major limitation is the absence of porous melt flow, which may play a key role in melt coalescence and ascent.The discrepancy between analogue and numerical results suggests more complex rheological coupling between deformed and partially molten crustal layers than can be captured by either method alone. This motivates a re-evaluation of the dynamical and thermal scaling of analogue experiments and the implementation of porous flow and anisotropic permeability in future numerical models.