This thesis investigates laminar–turbulent transitions in branching transport networks within a unified graph-based hydraulic framework. The central objective is to understand how regime changes, controlled by local Reynolds numbers, propagate through network architectures and how they affect global transport performance. Two complementary classes of networks are considered. First, optimized transport networks (OTNs) are obtained by minimizing a global cost functional balancing hydraulic dissipation and maintenance expenditure. The constitutive pressure–flow relation allows for fully laminar, fully turbulent and mixed configurations, where the exponent is assigned edgewise according to a Reynolds threshold. Numerical experiments show that the onset of turbulence is hierarchical: it consistently appears on high-flow, large-radius edges and propagates downstream as demand increases.