PSI - Issue 78

Stefano Ercolessi et al. / Procedia Structural Integrity 78 (2026) 1497–1504

1498

a standard sequential version for conventional analyses, as well as two parallel variants, that are OpenSeesSP and OpenSeesMP, designed to exploit the advantages of parallel implementations in order to handle large models or para metric analyses. These, features make OpenSees a versatile and a powerful tool for computational research. However, OpenSees lack both a built-in pre-processor and post-processor. As a results, particularly for complex or large-scale models, users often rely on third-party software to assist with the model setup and the results interpretation. While these tools are well-tested and widely adopted, often they are distributed under commercial licenses, which contrast with the open-source ethos underlying OpenSees itself. Moreover, several powerful open-source tools for pre- and post-processing are available and can be easily integrated in the common FEA framework employed by the users. Among, them noteworthy are Gmsh (Geuzaine (2009)) and ParaView (Ayachit et al. (2021)). Gmsh is an open-source 3D finite element mesh generator that also includes pre- and post- processing capabilities. It supports both structured and unstructured mesh generation. Its integration with OpenSees is facilitated by the C ++ and Python Application Programming Interfaces (API). ParaView, on the other hand, is an open-source cross-platform application for data analysis and visualization, capable of handling large-scale simulations results. It supports both interactive 3D vi sualization and automated batch processing, together with a client-server structure suitable for parallel processing, making it ideal to a range of computational scenarios. It can be integrated in the OpenSees workflows, exporting the simulation results outputs in the ParaView Data (PVD) format, which enables seamless import into ParaView. Sev eral common features facilitate the integration of the aforementioned software tools into a cohesive FEA framework. Among these, the cross-platform compatibility, the support for parallel computing paradigms, and the flexibility from a developer’s perspective are noteworthy. However, the integration of the pre-processing stage (e.g. Gmsh) with the OpenSees analysis environment is not straightforward. It requires an intermediate tool, which consist of a custom developed TCL script generator library. This library serves as a bridge between the geometric-mesh data and the OpenSees input format, and its development is essential to ensure seamless interoperability. Based on these premises, this paper introduces a complete open-source, multi-platform workflow for FEA implementations. The pre-processing step is based on the C ++ Gmsh API. A TCL generator capable to generate the required TCL scripts employed by the analysis software are developed. Finally, the post-processing step is performed reading the OpenSees output recorders in the ParaView framework. The presented work is structured as follows: Section 2 presents the proposed workflow in detail, outlining all the necessary steps for the three-software integration; Section 3 presents a practical application on a typical problem regarding the Soil-Structure Interaction (SSI) analysis; finally, conclusion and comments are provided. FEA, generally, can be seen as a multi-step process that begins with the geometry definition and ends with analysis results and visualization. Along this process, it is possible to identify three macro-steps in which the analysis can be decomposed. The first step, also known as pre-processing, considers the geometry definition and its discretization. Geometry definition can be performed by considering a CAD system. The discretized geometry is then managed by a finite element kernel, which can perform the desired numerical simulations storing all the desired outputs in an appropriate format. Thus, the post-processing phase is handled by a post-processor, which allows data visualization and interpretation. The proposed FEA workflow, aim to consider multi-platform open-source applications able to run according to the parallel computing paradigm. An overall illustration of the proposed workflow is depicted in Figure 1. In the following, are reported the fundamental steps associated with the proposed workflow. The pre-processing phase relies on C ++ Gmsh API, which allows the user to define the geometry of the model. Gmsh supports, two di ff erent CAD kernels, a built-in version and a the OpenCASCADE kernel. Moreover, it can be easily integrated in parallel workflows, since it is able to generate a global mesh and partitioning using tools such as METIS. Once that the geometry and hence the discretization is complete, is critical the definition of the so-called physical groups. These are user-defined labels assigned to geometric entities, such as points, lines etc, to indicate the geometry parts which have physical meaning in the simulation, such as load, boundary conditions, etc. The physical group definition does not a ff ect the mesh itself but allows the users to identify the critical entities and then query 2. The FEA workflow 2.1. Gmsh pre-processing