Experimental investigation on particle pair dispersion in homogeneous isotropic turbulence

Turbulence is one of the main drivers of how substances spread in air and water. It influences the dispersion of pollutants, nutrients, droplets and bubbles, and even the transport of microplastics and other contaminants in natural and engineered environments, from the atmosphere to the ocean, to rivers, industrial pipes and mixing reactors. One of the main drivers of this process is pair dispersion: how the distance between two tracer particles carried by the flow grows over time. Understanding this seemingly simple problem matters because it constrains how we model mixing, predict contamination patterns, and estimate how fast turbulence dilutes or concentrates what it transports. Classical turbulence theories propose that pair separation follows different behaviors as time progresses, from an early stage strongly influenced by the initial distance between the tracers to a later stage where turbulence dominates and separation accelerates significantly. In practice, however, observing a clear and extended “universal” behavior in experiments is difficult. Finite turbulence intensity, limited observation volumes, and experimental limitations can mask the expected scaling trends, while rare events can contribute disproportionately to the overall statistics. In this thesis, we investigate turbulent pair dispersion using three-dimensional Lagrangian particle tracking in a laboratory flow designed to approximate homogeneous isotropic turbulence. Large datasets collected over multiple experimental runs and a range of initial separations are used to quantify both average dispersion and its variability. Beyond mean trends, we analyze conditional statistics to isolate the role of initial conditions and examine probability distributions to capture intermittent and extreme separation events. Finally, we introduce a complementary geometrical diagnostic based on the relative orientation of particle separation and relative velocity, which has been suggested as a robust indicator of inertial-range dynamics when direct scaling is difficult to observe. Overall, the results provide an experimental assessment of when Richardson-like behavior can be identified in realistic laboratory conditions and clarify how initial separation, intermittency and geometrical alignment shape turbulent relative dispersion.