1: \begin{abstract}

2: Statistical solutions, which are time-parameterized probability measures on spaces of square-integrable functions, have been established as a suitable framework for global solutions of incompressible Navier-Stokes equations (NSE).

3: We compute numerical approximations of statistical solutions of NSE on two-dimensional domains with \emph{non-periodic} boundary conditions and empirically investigate the convergence of these approximations and their observables. For the numerical solver, we use Monte Carlo sampling with an H(div)-FEM based deterministic solver.

4: Our numerical experiments for high Reynolds number turbulent flows demonstrate that the statistics and observables of the approximations converge. We also develop a \emph{novel} algorithm to compute \emph{structure functions} on unstructured meshes.

5: \end{abstract}