Flow and transport that involve multiple fluid phases and components in porous materials are often controlled by thermodynamic phase change behaviors such as evaporation and condensation. In a nanometer-scale pore space, the phase behavior of a multicomponent fluid will deviate from that in a larger pore space (i.e., on the order pf micrometer or greater)—the pressure and temperature that the fluid begins to evaporate or condensate can differ significantly in the nanopores. Additionally, natural nanoporous materials such as shale rocks often contain a huge number of interconnected pores whose sizes span from nanometers to sub-micrometers. Such complex pore structures further modify the pore size-dependent phase behavior in the nanoporous material. The above so-called nanoconfined phase change behavior—commonly observed during oil and gas recovery from shale formations—has posed significant challenges for accurate prediction of hydrocarbon production. While the phase behavior in a single nanopore has been extensively studied by molecular-level theories, the new molecular-level understanding not yet been incorporated in Darcy-scale continuum models. To bridge this knowledge gap, we develop a novel pore-network-based upscaling framework to understand and quantify how nanoconfined phase change behaviors of multicomponent fluids manifest in shale rocks that have complex nanopore structures. The upscaling framework allows us to derive new constitutive relationships for Darcy-scale continuum models that fully account for the interactions between phase change behaviors, two-phase flow, multicomponent transport, and the multiscale nanopore structures. These newly derived constitutive relationships are then used for quantitative predictions of hydrocarbon production from shale formations.