Numerical simulation of pumping tests in heterogeneous aquifers using a Python-driven framework

 Accurate interpretation of pumping test data in stratified aquifers requires approaches that account for vertical heterogeneity, a factor often neglected in conventional analytical solutions. This study presents a Pythodriven axisymmetric numerical modelling framework, built using MODFLOW 6 and FloPy, to simulate both pumping and recovery phases in vertically heterogeneous confined aquifers. The model discretises the domain radially and vertically to allow layer-specific representation of hydraulic conductivity, while specific storage is assigned uniformly. An optimisation-based inverse modelling approach was used to estimate aquifer parameters by minimising the difference between observed and simulated drawdowns. Applied to a case study in Bahariya, Egypt, the results yielded hydraulic conductivity values consistent with the site’s stratigraphy – ranging from approximately 10−5 m∙d−1 in shale to over 27 m∙d−1 in limestone – and a specific storage of 4∙10−8 m−1. The simulated radius of influence was 133.67 m, and the root mean square error between the observed and simulated drawdown was 0.01 m. Sensitivity analysis demonstrated that vertical discretisation had the greatest influence on model accuracy, with coarser grids increasing residual error by nearly 90% and reducing the radius of influence by 9%. The temporal resolution had minimal impact on accuracy but significantly affected computation time.  This framework offers an open-source, automated, and script-based tool for simulating pumping tests in layered aquifer systems, enabling more reliable estimation of hydraulic parameters for both scientific and applied groundwater studies.