Computational fluid dynamics (CFD) of cerebral venous flows has become popular owing to the possibility of using local hemodynamics and hemoacoustics to help diagnose and plan treatments for venous diseases of the brain. Lumen geometries in low-pressure cerebral veins are different from those in cerebral arteries, often exhibiting fenestrations and flattened or triangular cross-sections, in addition to constrictions and expansions. These can challenge conventional size-based volume meshing strategies, and the ability to resolve non-laminar flows. Here we present a novel strategy leveraging estimation of length scales that could be present if flow were to become transitional or turbulent. Starting from the lumen geometry and flow rate boundary conditions, centerlines are used to determine local hydraulic diameters and cross-sectional mean velocities, from which flow length scales are approximated using conventional definitions for local Kolmogorov and Taylor microscales. By inspection of these scales, a user specifies minimum and maximum mesh edge lengths, which are then distributed along the model in proportion to the approximated local Taylor length scales. We demonstrate on three representative cases that this strategy avoids some of the pitfalls of conventional size-based strategies. An exemplary CFD mesh-refinement study shows convergence of high-frequency flow instabilities even starting from relatively coarse edge lengths near the lower bounds of the approximated Taylor length scales. Rational consideration of the length scales in a possibly non-laminar flow may provide a useful and replicable baseline for de-novo meshing of complicated or unfamiliar venous lumen geometries.