Fractal-based variable drag model for porous-media tree representations Preprint: AIP/123-QED
T. Tokiwa, Y. Yin, R. Onishi · 2026
Read Fractal-based variable drag model for porous-media tree representations Preprint: AIP/123-QED by T. Tokiwa, Y. Yin, R. Onishi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-25096.txt.
Introduction
Takumi Tokiwa, Yuwei Yin, and Ryo Onishi propose a variable-drag model for porous-media tree representations. Tree geometry is too costly to resolve, so trees are modeled as porous media with a constant drag coefficient. The new model prescribes drag per cell from effective branching order and effective Reynolds number, capturing unresolved complexity and the flow regime. Flow simulations across grid resolutions and velocities test the model, judged by aerodynamic porosity. The model improves grid robustness against constant-drag and captures bulk drag variation without retuning.
Catalog record
| CREATOR | T. Tokiwa, Y. Yin, R. Onishi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36336 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.25096
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fractal-based variable drag model for porous-media tree representations Preprint: AIP/123-QED T. Tokiwa, Y. Yin, R. Onishi Abstract Accurate representation of trees is essential for predictive urban micrometeorological simulations. However, because explicitly resolving detailed tree geometry is computationally prohibitive, trees are commonly represented as porous media. Within this framework, conventional approaches typically prescribe a constant drag coefficient, even when spatially heterogeneous area-density distributions are introduced. This limits transferability across inflow conditions and can increase sensitivity to grid resolution, particularly in the grayscale regime where a tree is represented by only a few computational cells. In this study, we propose a fractal-based variable-drag framework for porous-media tree representations, in which the drag coefficient is prescribed cell-wise as C D = C D ( n eff , R e eff ) C_{D}=C_{D}(n_{\mathrm{eff}},Re_{\mathrm{eff}}) . Here, n eff n_{\mathrm{eff}} denotes the cell-effective branching order, which represents unresolved local morphological complexity, and R e eff Re_{\mathrm{eff}} denotes the cell-effective Reynolds number, which represents the local flow condition. The framework is assessed using steady Reynolds-averaged Navier–Stokes simulations of a porous fractal tree over systematic sweeps of grid resolution and inflow velocity. Model performance is evaluated primarily using aerodynamic porosity, which measures the bulk momentum attenuation induced by the tree. The proposed model produces a plausible aerodynamic response, including velocity deficit, bypass flow, and wake recovery around the porous tree region. Compared with conventional porous-tree models using constant drag coefficients, the proposed formulation improves robustness to grid resolution and captures the global inflow-velocity dependence of the bulk drag effect without empirical retuning. Notably, this whole-tree response is recovered through local cell-wise quantities, n eff n_{\mathrm{eff}} and R e eff Re_{\mathrm{eff}} . These results demonstrate that incorporating morphology- and flow-dependent drag into the cell-wise drag coefficient provides a practical route to improving porous-media tree modeling. Future work includes extension of the framework to unsteady simulations such as large-eddy simulation, and ultimately to multiple-tree and district-scale urban micrometeorological applications. I Introduction In recent years, accurate urban micrometeorological forecasting has become increasingly important for urban environmental assessment and planning, particularly in the context of accelerating urban population growth, intensifying heat-related risks, and the development of smart societies 32 ; 12 ; 26 ; 31 . Meeting this demand requires physically credible representations of the processes that shape the urban atmosphere, including the effects of vegetation. Trees play a vital role in regulating urban micrometeorology by modulating wind flow, temperature distribution, and pollutant dispersion 10 ; 14 ; 24 . These impacts are governed by aerodynamic interactions between the flow and the multi-scale architecture of trees, through which trees extract momentum and alter turbulence. Consequently, a reliable description of tree-induced aerodynamic resistance is a prerequisite for predictive simulations of urban ventilation, scalar transport, and related microclimatic conditions. For practical urban-scale computations, however, resolving detailed tree geometry is rarely feasible. Tree representations in CFD can be viewed as a resolution-dependent hierarchy: (i) explicit simulations resolve branch and trunk geometry and offer high-fidelity reference solutions, albeit at high computational cost 34 ; 33 ; (ii) implicit approaches parameterize vegetation effects as fully subgrid forcing 3 ; 16 ; and (iii) an increasingly common intermediate regime represents a tree with only a few grid cells, referred to here as grayscale simulations 21 ; 7 ; 8 . As computational resources improve, grayscale simulations are becoming more relevant for real-world applications, where fully resolved simulations remain too expensive while fully homogenized single-cell representations are overly restrictive. In both Reynolds-averaged Navier–Stokes (RANS) and large-eddy simulation (LES) frameworks, implicit and grayscale approaches typically model trees as porous media, i.e., distributed momentum sinks. In such models, the canopy effect is commonly expressed through a drag term proportional to a local area density and a drag coefficient, e.g., \mathbf{S}_{u}\propto-C_{D}\alpha|\mathbf{u}|\mathbf{u} , where \alpha(\mathbf{x}) denotes an area-density field such as plant or leaf area density (PAD/LAD). A common conventional practice is to prescribe spatially uniform area density over the tree cells together with a constant drag coefficient C_{D} , or to tune a constant C_{D} on a case-by-case basis, as in baseline porous-canopy implementations 18 ; 7 ; 8 . More recently, advanced conventional models have incorporated grid- or voxel-resolved PAD/LAD distributions derived from geometry or measurements to better represent spatial heterogeneity 21 ; 36 ; 25 , while still retaining a constant C_{D} . These developments have improved the geometric realism of porous-tree representations. However, the drag coefficient itself is still typically prescribed as a constant and therefore does not explicitly respond to local flow conditions or unresolved morphological complexity within each cell. This remaining limitation can reduce transferability across inflow conditions 17 ; 1 and can introduce sensitivity to grid resolution 6 ; 36 , particularly in the grayscale regime where the effective unresolved morphology represented within each cell changes with cell size. These issues motivate a drag formulation in which the cell-wise drag coefficient adapts to both morphology and flow regime. In this study, we propose a fractal-based variable-drag framework for porous-media tree representations in which the drag coefficient is prescribed as C_{D}=C_{D}(n_{\mathrm{eff}},Re_{\mathrm{eff}}) . Here, n_{\mathrm{eff}} is a cell-effective branching order that characterizes unresolved morphological complexity by exploiting self-similarity of the underlying fractal branching structure, and Re_{\mathrm{eff}} is an cell-effective local Reynolds number constructed from the local velocity regime and a morphology-based characteristic length. The mapping from (n_{\mathrm{eff}},Re_{\mathrm{eff}}) to C_{D} is obtained a priori using an analytical drag model for fractal trees (details in our previous work 29 ), enabling cell-wise updates of C_{D} during CFD simulations without ad hoc retuning. We validate the proposed framework using steady RANS simulations with systematic parameter sweeps in both grid resolution and inflow velocity. To isolate the bulk momentum attenuation attributable to the canopy forcing, we evaluate model performance primarily using aerodynamic porosity 11 ; 2 , defined as the ratio of leeward to windward plane-averaged streamwise velocity across the tree. We compare the proposed model against (i) a general conventional baseline using uniform PAD with constant C_{D} and (ii) an advanced conventional baseline using voxel-resolved PAD with constant C_{D} . The remainder of this paper is organized as follows. In Sec. II, we present the fractal-based variable-drag formulation, including the definitions of n_{\mathrm{eff}} and Re_{\mathrm{eff}} and the a priori construction of C_{D}(n_{\mathrm{eff}},Re_{\mathrm{eff}}) . In Sec. III, we describe the RANS methodology and computational setup, summarize the grid-resolution and inflow-velocity sweeps, and define the aerodynamic porosity metric used for evaluation. In Sec. IV, we present the results and discussion, focusing on qualitative flow behavior, resolution robustness, and Reynolds-number dependence compared with constant- C_{D} baselines. Finally, Sec. V concludes the paper and outlines future work. II Fractal-based variable-drag formulation II.1 Cell-wise descriptors: n_{\mathrm{eff}} and Re_{\mathrm{eff}} Effective branching order from fractal self-similarity. We define the cell-effective branching order n_{\mathrm{eff}} to characterize the unresolved, grid-dependent morphology represented within each computational cell. As sketched in Fig. 1 , changing grid size changes the portion of the tree contained in a cell and thus the effective unresolved geometry; fractal self-similarity allows this in-cell geometry to be interpreted in terms of an equivalent branching complexity, which we quantify by n_{\mathrm{eff}} . Figure 1 : Schematic illustration of grid-dependent in-cell morphology and its interpretation via fractal self-similarity. To infer n_{\mathrm{eff}} , we define a dimensionless