pith. sign in

arxiv: 2604.21858 · v1 · submitted 2026-04-23 · ⚛️ physics.flu-dyn · cs.NA· math.NA· physics.comp-ph

Meshless h-adaptive Solution for non-Newtonian Natural Convection in a Differentially Heated Cavity

Miha Rot , Gregor Kosec This is my paper

Pith reviewed 2026-05-09 20:36 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn cs.NAmath.NAphysics.comp-ph
keywords meshless methodsadaptive discretisationnon-Newtonian fluidnatural convectionde Vahl Davis benchmarkshear-thinningboundary layercomputational efficiency
0
0 comments X

The pith

Adaptive node density refinement in meshless methods improves efficiency for non-Newtonian natural convection in a differentially heated cavity.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that meshless methods can adapt the local density of computational nodes to the evolving solution, concentrating nodes where thin boundary layers form in shear-thinning non-Newtonian flows. This approach is tested on the classic de Vahl Davis benchmark of natural convection in a rectangular cavity. A reader would care because non-Newtonian fluids appear in many natural and engineering settings, such as blood flow, and uniform meshes waste effort on smooth regions while under-resolving sharp features. The work shows that adaptation reduces the total number of nodes and operations while preserving accuracy, and it explores how the refinement thresholds control the trade-off between cost and fidelity.

Core claim

Adaptively increasing the density of nodes in regions of high shear rate or steep temperature gradients allows meshless discretisation to solve the non-Newtonian natural convection problem with fewer total nodes than a uniform distribution, while the solution remains consistent with reference results for the de Vahl Davis test case.

What carries the argument

h-adaptive meshless node placement, in which local node density is increased or decreased according to intermediate solution features such as velocity gradients or boundary-layer thickness.

If this is right

  • Fewer nodes are required overall when refinement is confined to the vertical boundary layers, lowering memory and CPU cost for the same accuracy target.
  • The method can capture the sharper flow structures produced by shear-thinning viscosity without globally increasing node count.
  • Solution quality depends on the specific refinement thresholds; improper choices can either waste nodes in smooth regions or leave boundary layers under-resolved.
  • The approach is directly applicable to other cavity-driven flows where boundary layers dominate the transport.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same adaptive strategy could be transferred to three-dimensional or irregular geometries where uniform meshing becomes prohibitive.
  • Because node placement is decoupled from a fixed mesh, the method may simplify coupling to moving boundaries or free surfaces in non-Newtonian problems.
  • If the refinement criteria prove robust across a range of rheological parameters, the technique could serve as a building block for parameter studies of blood-like fluids in biomedical simulations.

Load-bearing premise

The chosen refinement criteria and parameter values keep the discretisation accurate inside the thin boundary layers of the non-Newtonian flow without introducing uncontrolled numerical errors.

What would settle it

Run the same non-Newtonian de Vahl Davis case with successively stricter refinement thresholds and compare the resulting Nusselt number and velocity extrema against a well-converged reference solution obtained on a uniformly fine mesh; any systematic drift in the reported quantities with changing adaptation parameters would falsify the claim of maintained accuracy.

Figures

Figures reproduced from arXiv: 2604.21858 by Gregor Kosec, Miha Rot.

Figure 1
Figure 1. Figure 1: Node distributions during the first four adaptivity steps for the de Vahl Davis case. Nodes marked for refinement are plotted in orange, nodes [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sketch of the geometry and boundary conditions for the two [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Velocity magnitude and temperature field for the de Vahl Davis [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Evolution of system observables for the de Vahl Davis case with [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: The final number of nodes in the discretisation as a function [PITH_FULL_IMAGE:figures/full_fig_p004_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Computational cost of different parts of the adaptive refinement [PITH_FULL_IMAGE:figures/full_fig_p005_9.png] view at source ↗
Figure 8
Figure 8. Figure 8: Convergence of the average Nusselt number on the cold boundary [PITH_FULL_IMAGE:figures/full_fig_p005_8.png] view at source ↗
Figure 11
Figure 11. Figure 11: Evolution of Nu throughout the solution procedure for the compared discretisation strategies. tive. All of the considered discretisations have the same minimum internodal spacing hmin = 0.00251 . Constant density discretisation has by far the most nodes and an order of magnitude longer computational time. Refined procedure results in approximately half the nodes of the adaptive due to the aforementioned i… view at source ↗
read the original abstract

One of the main challenges in numerically solving partial differential equations is finding a discretisation for the computational domain that balances the accurate representation of the underlying field with computational efficiency. Meshless methods approximate differential operators based on the values of the field in computational nodes, offering a natural approach to adaptivity. The density of computational nodes can either be increased to enhance accuracy or decreased to reduce the number of numerical operations, depending on the properties of the intermediate solution. In this paper, we utilise an adaptive discretisation approach for the numerical simulation of natural convection in non-Newtonian fluid flow. The shear-thinning behaviour is interesting both due to its numerous occurrences in nature, blood being a prime example, and due to its properties, as the decreasing viscosity with increasing shear rate results in sharper flow structures. We focus on the de Vahl Davis test case, a natural convection driven flow in a differentially heated rectangular cavity. The thin boundary layer flow along the vertical boundaries makes this an ideal test case for refinement. We demonstrate that adaptively refining the node density enhances computational efficiency and examine how the parameters for adaptive refinement affect the solution.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a meshless h-adaptive collocation method for natural convection of non-Newtonian (power-law) fluids in a differentially heated square cavity, using the de Vahl Davis benchmark. It claims that locally increasing node density in regions of high gradients (thin vertical boundary layers) improves computational efficiency relative to uniform discretizations, while exploring the sensitivity of results to the adaptivity parameters.

Significance. If the accuracy claims hold, the approach could provide a practical route to resolving sharp shear-thinning structures with fewer nodes than uniform meshes, which is relevant for applications such as blood flow or polymer processing. The work correctly identifies the de Vahl Davis cavity as a suitable test for adaptivity due to its boundary-layer character.

major comments (2)
  1. [§4] §4 (Results): no quantitative error metrics (L2 norms, wall Nusselt number deviations, or extrema of the stream function) are reported for the adaptive solutions relative to uniform high-density runs or literature benchmarks (de Vahl Davis for n=1; extensions for n<1). Without these, the central claim that adaptivity preserves fidelity while reducing cost cannot be verified, especially for shear-thinning cases where viscosity gradients are sensitive to local node spacing.
  2. [§3.2] §3.2 (Adaptive refinement criterion): the indicator used to trigger node insertion (presumably based on velocity or temperature gradients or residual) is not shown to be sufficient to control truncation error in the thin boundary layers for n<1; the paper examines parameter effects but does not demonstrate that the chosen thresholds avoid under-resolution that would bias Nusselt numbers.
minor comments (2)
  1. [§2] Notation for the power-law index n and the adaptive threshold parameters should be introduced once in §2 and used consistently; occasional re-definition in the results section reduces clarity.
  2. [Figures 4-6] Figure captions for the adaptive node distributions should include the final node count and the corresponding uniform-mesh equivalent for direct visual comparison of efficiency.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough and constructive review of our manuscript on the meshless h-adaptive method for non-Newtonian natural convection. The comments highlight important aspects for strengthening the validation of our approach. We address each major comment below and will revise the manuscript accordingly to provide the requested quantitative evidence and demonstrations.

read point-by-point responses

  1. Referee: §4 (Results): no quantitative error metrics (L2 norms, wall Nusselt number deviations, or extrema of the stream function) are reported for the adaptive solutions relative to uniform high-density runs or literature benchmarks (de Vahl Davis for n=1; extensions for n<1). Without these, the central claim that adaptivity preserves fidelity while reducing cost cannot be verified, especially for shear-thinning cases where viscosity gradients are sensitive to local node spacing.

    Authors: We agree that explicit quantitative error metrics are necessary to rigorously support the claim of maintained accuracy alongside efficiency gains. The current manuscript focuses on qualitative demonstrations of boundary-layer resolution and parameter sensitivity, along with computational cost reductions, but does not include direct L2 norm comparisons, Nusselt number deviations, or stream-function extrema against high-resolution uniform meshes or literature values. In the revised manuscript, we will add these metrics for both n=1 (benchmarking directly to de Vahl Davis) and representative n<1 cases, including tables of relative errors and convergence behavior. This will enable clear verification that adaptivity does not introduce bias in the sensitive shear-thinning regime. revision: yes

  2. Referee: §3.2 (Adaptive refinement criterion): the indicator used to trigger node insertion (presumably based on velocity or temperature gradients or residual) is not shown to be sufficient to control truncation error in the thin boundary layers for n<1; the paper examines parameter effects but does not demonstrate that the chosen thresholds avoid under-resolution that would bias Nusselt numbers.

    Authors: The refinement indicator is based on local velocity and temperature gradient magnitudes, with node insertion occurring above a user-specified threshold; the parameter study in §3.2 already varies this threshold and reports resulting changes in global quantities. We acknowledge, however, that this does not yet explicitly quantify truncation-error control or Nusselt-number bias for the thinner boundary layers at n<1. In the revision we will augment §3.2 with targeted convergence tests: for selected n<1 values we will compare Nusselt numbers obtained with the chosen thresholds against both finer adaptive and uniform reference solutions, demonstrating that the selected thresholds keep deviations below a specified tolerance and avoid under-resolution bias. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical demonstration of adaptive meshless solver on standard benchmark

full rationale

The paper applies an existing meshless collocation method with h-adaptivity to the de Vahl Davis natural convection benchmark for power-law fluids. No derivation chain is claimed; the work consists of numerical experiments showing efficiency gains and parameter sensitivity. No equations reduce a 'prediction' to a fitted input by construction, no self-citation is load-bearing for a uniqueness or ansatz result, and no renaming of known patterns occurs. The central claim (adaptive refinement improves efficiency while preserving accuracy) is supported by direct comparisons to reference solutions rather than self-referential definitions. This is a standard computational fluid dynamics demonstration paper whose content remains independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on standard assumptions of meshless collocation methods and h-refinement heuristics for PDEs; no explicit free parameters, new entities, or ad-hoc axioms are stated in the abstract.

axioms (1)
  • domain assumption Meshless methods can approximate differential operators from scattered node values with controllable accuracy.
    Invoked as the foundation for the adaptive discretization in the abstract.

pith-pipeline@v0.9.0 · 5509 in / 1033 out tokens · 27000 ms · 2026-05-09T20:36:05.476067+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 2 canonical work pages

  1. [1]

    A simple error estimator and adaptive procedure for practical engineering analysis,

    O. C. Zienkiewicz and J. Z. Zhu, “A simple error estimator and adaptive procedure for practical engineering analysis,”International Journal for Numerical Methods in Engineering, vol. 24, no. 2, pp. 337–357, 1987. 1Note that while according to Figure 8 this selection ofh min does not yet result in an accurate Nuvalue decreasing it further would lead to exc...

    1987

  2. [2]

    A review of some a posteriori error estimates for adaptive finite element methods,

    K. Segeth, “A review of some a posteriori error estimates for adaptive finite element methods,”Mathematics and Computers in Simulation, vol. 80, no. 8, pp. 1589–1600, 2010

    2010

  3. [3]

    A comparison of hp-adaptive strategies for elliptic partial differential equations,

    W. F. Mitchell and M. A. McClain, “A comparison of hp-adaptive strategies for elliptic partial differential equations,”ACM Transac- tions on Mathematical Software, vol. 41, no. 1, pp. 1–39, 2014

    2014

  4. [4]

    An efficient localized radial basis function meshless method for fluid flow and conjugate heat transfer,

    E. Divo and A. J. Kassab, “An efficient localized radial basis function meshless method for fluid flow and conjugate heat transfer,” ASME Journal of Heat Transfer, vol. 129, no. 2, pp. 124–136, 2007

    2007

  5. [5]

    Solution of incompressible fluid flow problems with heat transfer by means of an efficient RBF-FD mesh- less approach,

    R. Zamolo and E. Nobile, “Solution of incompressible fluid flow problems with heat transfer by means of an efficient RBF-FD mesh- less approach,”Numerical Heat Transfer, Part B: Fundamentals, vol. 75, no. 1, pp. 19–42, 2019

    2019

  6. [6]

    Adaptive radial basis function–generated fi- nite differences method for contact problems,

    J. Slak and G. Kosec, “Adaptive radial basis function–generated fi- nite differences method for contact problems,”International Journal for Numerical Methods in Engineering, vol. 119, no. 7, pp. 661– 686, 2019

    2019

  7. [7]

    Strong form mesh-freehp-adaptive solution of linear elasticity problem,

    M. Jan ˇciˇc and G. Kosec, “Strong form mesh-freehp-adaptive solution of linear elasticity problem,”Engineering with Computers, vol. 40, 2024

    2024

  8. [8]

    Improved finite difference method for phase-field modelling of dendritic solidification,

    T. Dobravec, B. Mavri ˇc, and B. Šarler, “Improved finite difference method for phase-field modelling of dendritic solidification,”Jour- nal of Computational Physics, vol. 553, p. 114716, May 2026

    2026

  9. [9]

    Adaptive rbf-fd meshless solution of 3d fluid flow and heat transfer problems,

    L. Bacer, R. Zamolo, D. Miotti, and E. Nobile, “Adaptive rbf-fd meshless solution of 3d fluid flow and heat transfer problems,” Engineering Analysis with Boundary Elements, vol. 179, p. 106367, 2025

    2025

  10. [10]

    Adaptivity in local kernel based methods for approx- imating the action of linear operators,

    J. A. Reeger, “Adaptivity in local kernel based methods for approx- imating the action of linear operators,”SIAM Journal on Scientific Computing, vol. 46, no. 4, pp. A2683–A2708, 2024

    2024

  11. [11]

    Refined radial basis function-generated finite difference analysis of non-newtonian natural convection,

    M. Rot and G. Kosec, “Refined radial basis function-generated finite difference analysis of non-newtonian natural convection,” Physics of Fluids, vol. 37, no. 3, p. 033130, 03 2025. [Online]. Available: https://doi.org/10.1063/5.0257896

    work page doi:10.1063/5.0257896 2025

  12. [12]

    On using radial basis functions in a “finite difference mode

    A. I. Tolstykh and D. A. Shirobokov, “On using radial basis functions in a “finite difference mode” with applications to elasticity problems,”Computational Mechanics, vol. 33, no. 1, pp. 68–79, Dec 2003. [Online]. Available: https://doi.org/10.1007/ s00466-003-0501-9

    2003

  13. [13]

    On generation of node distributions for mesh- less PDE discretizations,

    J. Slak and G. Kosec, “On generation of node distributions for mesh- less PDE discretizations,”SIAM Journal on Scientific Computing, vol. 41, no. 5, pp. A3202–A3229, Oct. 2019

    2019

  14. [14]

    Guidelines for rbf-fd discretization: Numerical experiments on the interplay of a multitude of parameter choices,

    S. Le Borne and W. Leinen, “Guidelines for rbf-fd discretization: Numerical experiments on the interplay of a multitude of parameter choices,”Journal of Scientific Computing, vol. 95, no. 1, p. 8, Feb 2023. [Online]. Available: https://doi.org/10.1007/ s10915-023-02123-7

    2023

  15. [15]

    Shepard’s method of

    W. J. Gordon and J. A. Wixom, “Shepard’s method of "metric interpolation" to bivariate and multivariate interpolation,” Mathematics of Computation, vol. 32, no. 141, pp. 253–264, 1978. [Online]. Available: http://www.jstor.org/stable/2006273

    work page arXiv 1978

  16. [16]

    A stable algorithm for divergence- free radial basis functions in the flat limit,

    K. P. Drake and G. B. Wright, “A stable algorithm for divergence- free radial basis functions in the flat limit,”Journal of Computa- tional Physics, vol. 417, p. 109595, Sep. 2020

    2020

  17. [17]

    Natural convection of air in a square cavity: a bench mark numerical solution,

    G. de Vahl Davis, “Natural convection of air in a square cavity: a bench mark numerical solution,”International Journal for Numer- ical Methods in Fluids, vol. 3, no. 3, pp. 249–264, 1983

    1983

  18. [18]

    A comprehensive review on the natural, forced, and mixed convection of non-newtonian fluids (nanofluids) inside different cavities,

    L. Yang and K. Du, “A comprehensive review on the natural, forced, and mixed convection of non-newtonian fluids (nanofluids) inside different cavities,”Journal of Thermal Analysis and Calorimetry, pp. 1–22, 2019

    2019

  19. [19]

    Artificial compressibility approaches in flux reconstruction for incompressible viscous flow simulations,

    W. Trojak, N. Vadlamani, J. Tyacke, F. Witherden, and A. Jameson, “Artificial compressibility approaches in flux reconstruction for incompressible viscous flow simulations,”Computers & Fluids, vol. 247, p. 105634, 2022

    2022

  20. [20]

    Laminar natural convection of power-law fluids in a square enclosure with differentially heated side walls subjected to constant temperatures,

    O. Turan, A. Sachdeva, N. Chakraborty, and R. J. Poole, “Laminar natural convection of power-law fluids in a square enclosure with differentially heated side walls subjected to constant temperatures,” Journal of Non-Newtonian Fluid Mechanics, vol. 166, no. 17-18, pp. 1049–1063, 2011

    2011

  21. [21]

    Transient buoyant convection of a power-law non-newtonian fluid in an enclosure,

    G. B. Kim, J. M. Hyun, and H. S. Kwak, “Transient buoyant convection of a power-law non-newtonian fluid in an enclosure,” International Journal of Heat and Mass Transfer, vol. 46, no. 19, pp. 3605–3617, 2003

    2003

  22. [22]

    Parallel domain discretization algorithm for rbf-fd and other meshless numerical methods for solving pdes,

    M. Depolli, J. Slak, and G. Kosec, “Parallel domain discretization algorithm for rbf-fd and other meshless numerical methods for solving pdes,”Computers & Structures, vol. 264, p. 106773, 2022

    2022