Replication package: face averaging, grid resolution and dimensionality in nonlinear diffusion drying of cylinders

This deposit provides replication materials for the manuscript “Under what conditions can a simplified drying model predict the endpoint reliably? Face averaging, grid resolution and dimensionality in nonlinear diffusion drying of cylinders”.

The deposit contains:

• 01-external-comparison.zip: a frozen dataset of 110 points digitised from a published mean-moisture-ratio figure, per-point quality information, analytical and finite-volume models, calibration and holdout results, threshold-crossing times, sensitivity analyses using two estimated missing-marker coordinates, an independent analytical check, and figure-generation code.

• 02-reproducibility-package.zip: the main-study solvers, case specifications, cached numerical results, plotting scripts, and instructions for obtaining and importing the external benchmark inputs.

The official benchmark workbooks and the published source image are not redistributed. Main solver recomputation requires obtaining the official 2026 China Undergraduate Mathematical Contest in Modeling, Problem A materials from:
https://www.mcm.edu.cn/html_cn/node/27b6e148f8113f09b0269f64a02629fb.html

The external comparison uses measurements digitised from Fig. 9(a) of Singh and Talukdar (2019), International Journal of Thermal Sciences, 142:176–187:
https://doi.org/10.1016/j.ijthermalsci.2019.04.019

The external analysis can run from the included frozen CSV without repeating digitisation. Repeating digitisation requires obtaining the source image separately.

Dependencies include Python, NumPy, SciPy, Matplotlib, Pillow and openpyxl. Full external-figure export additionally requires PyMuPDF and Poppler. Archive README files explain input preparation, cache behaviour and supported reproduction commands.

The main study concerns numerical verification of a prescribed benchmark. The external comparison assesses mean moisture ratios and threshold crossings of an auxiliary model; it does not experimentally validate the wettest-point drying endpoint. Sensitivity analyses using estimated marker coordinates do not recover the missing observations.

Python source code and software configuration files are licensed under the MIT License. Numerical data, figures and documentation are licensed under CC BY 4.0. See LICENSING.md within each archive.