XZXutong Zhang

Xutong Zhang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xutong Zhang has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Simulation of Flow Around a Finite Rectangular Prism: Influence of Mesh, Model, and Subgrid Length Scale

    by Xutong Zhang, Maxime Savoie, Mark K. Quinn, Ben Parslew, Alistair Revell

    This study investigates the flow field around a finite rectangular prism using both experimental and computational methods, with a particular focus on the influence of the turbulence approach adopted, the mesh resolution employed, and different subgrid length scales. Ten turbulence modelling and simulation approaches, including both ‘scale-modelling’ Reynolds-Averaged Navier–Stokes (RANS) models and ‘scale-resolving’ Delayed Detached Eddy Simulation (DDES), were tested across six different mesh resolutions. A case with sharp corners allows the location of the flow separation to be fixed, which facilitates a focus on the separated flow region and, in this instance, the three-dimensional interaction of three such regions. The case, therefore, readily enables an assessment of the ‘grey-area’ issue, whereby some DDES methods demonstrate delayed activation of the scale-resolving model, impacting the size of flow recirculation. Experimental measurements were shown to agree well with reference data for the same geometry, after which particle image velocimetry (PIV) data were gathered to extend the reference dataset. Numerical predictions from the RANS models were generally quite reasonable but did not show improvement with further refinement, as one would expect, whereas DDES clearly demonstrated continuous improvement in predictive accuracy with progressive mesh refinement. The shear-layer-adapted (SLA) subgrid length scale (ΔSLA) displayed consistently superior performance compared to the more widely used length scale based on local cell volume, particularly for moderate mesh resolutions commonly employed in industrial settings with limited resources. In general, front-body separation and reattachment exhibited greater sensitivity to mesh refinement than wake resolution. Finally, in order to correlate the observed DDES mesh requirements with the observations from the converged RANS solutions, an approximation for the Taylor microscale was explored as a potential tool for mesh sizing.


  • Open Access

    Comparison of Lattice Boltzmann and Navier-Stokes for Zonal Turbulence Simulation of Urban Wind Flows

    by Marta Camps Santasmasas, Xutong Zhang, Ben Parslew, Gregory F. Lane-Serff, Joshua Millar, Alistair Revell

    In modelling turbulent flow around buildings, the computational domain needs to be much larger than the immediate neighbourhood of the building, resulting in computational costs that are excessive for many engineering applications. Two nested models are presented to solve this problem, with an outer domain calculated using a Reynolds Averaged Navier Stokes (RANS) solver in both cases. The inner region is calculated using large eddy simulation (LES) from both a lattice Boltzmann (LB) and a Navier Stokes (NS) based solver. The inner domains use the mean RANS velocity as boundary conditions for the top and the side boundaries and incorporate the RANS turbulence using a synthetic eddy method (SEM) at the inner domain inlet. Both models are tested using an atmospheric boundary layer flow around a rectangular building at ReH = 47,893, comparing the computational resources spent and validating the results with experimental measurements. The effect of the inlet turbulence, the size of the domain and the cell size are also investigated. Both LB and NS based simulations are able to capture the physics of the flow correctly and show good agreement with the experimental results. Both simulation frameworks were configured to run in a similar computational time, so as to compare the computational resources used. Due to the use of GPU programming, the approach based on LB was estimated to be 25 times cheaper than the NS simulation. Thus these results show that a nested LB-LES solver can run accurate wind flow calculations with consumer level/cloud based computational resources.