Grain boundaries formed during chemical vapour deposition (CVD) of graphene govern film integrity and electronic performance, yet the coupled roles of surface diffusion, attachment kinetics and stochastic multi-nucleation in setting the grain-boundary (GB) network remain difficult to predict. Motivated by scalable, CVD-enabled graphene manufacturing [1], we introduce a transport-coupled, free-energy-based multiphase phase-field formulation that recasts island evolution as coupled PDEs on a fixed Eulerian domain, enabling repeated nucleation, growth, impingement and coalescence without front tracking. The model couples anisotropic Allen–Cahn dynamics (six-fold interfacial anisotropy, ϵa = 0.04) to a surface-carbon reaction–diffusion field. Stochastic nucleation is supersaturation-dependent and progressively suppressed with coverage, and polycrystalline films are represented by one order parameter per in-plane orientation with a monolayer sum-to-one constraint and non-overlap penalty, following multiphase phase-field concepts for polycrystalline growth [2]. Using finite-element discretisation (continuous P1, N = 400 per side on L = 200) with backward-Euler/Crank–Nicolson time integration, we compute process–microstructure maps in the (influx F, diffusivity D) plane (explored up to F = 0.09 and D = 60) and analyse GB-network connectivity and percolation at fixed coverage. The simulations recover diffusion-limited faceting and attachment-limited uniform edge advance consistent with prior phase-field graphene studies [3] and reported temperature-dependent CVD kinetics on Cu [4]. During impingement, overlapping depletion halos focus flux into narrowing inter-island gaps, accelerating gap closure and fixing the emerging GB morphology. These results connect controllable process knobs (F, D, nucleation scheduling and seeding orientation) to microstructural quality, supporting strategies to promote larger grains and reduce transport-limiting GBs [5] in wafer-scale integration workflows [6].