https://doi.pangaea.de/10.1594/PANGAEA.997505Zamani Asl, Mohammad MehdiMohammad MehdiZamani Asl0000-0002-1322-6258Center of Applied Space Technology and MicrogravityAvila, MarcMarcAvila0000-0001-5988-1090Center of Applied Space Technology and MicrogravityPANGAEA2026turbulence simulationturbulence theoryturbulent mixingBinary ObjectBinary Object (File Size)FigureTitleFile nameFile formatDescriptionNumerical simulatedDataset10.1017/jfm.2026.12098168 data pointstext/tab-separated-valuesCreative Commons Attribution 4.0 InternationalDespite its widespread use as a model configuration for investigating mixing processes in chemical engineering, the turbulent regime of the T-shaped mixer remains relatively unexplored, in contrast to its well-documented transitional behaviour. To address this gap, we performed direct numerical simulations at Reynolds numbers up to Re = 2000 with a Schmidt number of Sc = 1 with spectral-element solver Nek5000. The results reveal a streamwise evolution characterised by two distinct regions before the flow approaches a fully developed duct-flow state. In the vicinity of the junction, the dynamics are dominated by a jet-like structure that displays approximate self-similarity resembling that of transitional planar jets. Downstream of this region, turbulence undergoes a decay process in which turbulent kinetic energy, dissipation, and scalar variance exhibit power-law scaling. Interestingly, the velocity-related decay exponents are characteristic of turbulence confined within bounded domains, whereas scalar variance decays at a rate consistent with unbounded turbulent flows. This difference can be explained by the spatial development of the mixing process: although turbulent fluctuations have spread across the entire channel cross-section by the end of the jet region, scalar homogenisation continues to advance laterally from the channel centre toward the walls within the decay region. In addition, the mean scalar distribution retains an error-function shape in the transverse direction, reminiscent of laminar mixing. The resulting streamwise development of the mean mixing state is described using an effective diffusion coefficient that varies along the channel.

This dataset provides all the post-processed data necessary to reproduce the figures in the manuscript (doi:10.1017/jfm.2026.12098). Further, original Nek5000 binary output files (.f) can be used to extract simulation data at arbitrary spatial coordinates within the T-mixer for the averaged TKE quantities and scalar variance. Instantaneous velocity and scalar fields of the simulations are provided that allow to restart all simulations in Nek5000.