This study presents a numerical approach designed for material parameter identification for the coupled hydro-mechanical boundary value problem (BVP) of the piezocone test (CPTU) in normally and lightly overconsolidated clayey soils. The study is presented in two related papers and it explores the possibility of using neural networks (NNs) to solve the complex inverse problem of the penetration test, including partially drained conditions. It has been demonstrated that the development of NN-based inverse models can be based on training data sets that consist of pseudo-experimental measurements derived from numerical simulations of the piezocone test. The first paper presents the development of the FE model of the studied problem, which can be used to generate a training data population corresponding to typical piezocone measurements that are obtained for clayey soils. The paper contains a detailed description of the numerical model with a sensitivity analysis with respect to different model parameters including the effect of partial drainage. The analysis also includes the model verification by means of a comparative analysis with numerical models of penetration proposed in the literature, as well as experimental evidence. Finally, owing to the loss of accuracy observed when applying a 'rough' frictional interface in the Updated Lagrangian formulation, an equivalent semi-numerical model for the piezocone test is proposed, taking into account a possible occurrence of partial drainage during penetration. Copyright (C) 2010 John Wiley & Sons, Ltd.