Tags: code-mphi/ECOGEN
Tags
Version 5.0 # Major points - Possibility to use second-order numerical scheme on unstructured meshes. Although still in beta. - Acoustic waves can now be modelled through source terms (sinusoidal or Gaussian pulses). This follows the work of _Maeda & Colonius_ [https://doi.org/10.1016/j.wavemoti.2017.08.004](https://doi.org/10.1016/j.wavemoti.2017.08.004) for single-phase flows and has been extended here for multiphase systems of equations. - For developers: Improvement of the CI now including code coverage, clang-formatting, better artefact management, use of address sanitizer and pre-commit. - Clang-format applied to the whole code. # Minor points - Gnuplot scripts added for pdf generation of results with GNU output format - Add velocity in x-direction to the 1D output using Gnuplot. - Correction of translation: restart -> resume. - The CFL length estimation has been improved for quadrangle elements. - Possibility to add record frequency for cuts. - Correction of memory errors (deallocations). - Possibility to compile the code using `cmake` instead of `makefile` (for advanced users). - Update of documentation and test cases, as usual. - Use `VTK` key instead of `XML` one for `outputMode` tag of `main.xml` file. - Modify the deepness of all 1D and 2D test cases to unity. This was made to avoid possible issues during the visualisation of the results with tools such as Paraview. # Fixes - Update names of manuals to be coherent with input files from the test cases. - Update Cv from air EOS. - Fix bug and variables names update on `ModUEq::solveRIemannInletInjTemp`. - Fix EOS assignment in boundary conditions and add that the EOS key must be present in `dataFluid` tag of boundary conditions (must be consistent with model EOS). - Fix `Euler` test cases missing EOS key in `dataFluid` tag of boundary conditions. - Fix memory errors during termination of the execution when EOS is missing. - Fix unstructured-mesh file parsing using msh v4 format (working now, although only for serial simulations). - Fix wrong elapsed time from simulations (was using CPU time, now uses wall time).
Version 4.0 **Major points:** * Removed the version numbers of the input file names (break compatibility with previous version). * Possibility to use finite pressure relaxation on UEq model for more than two phases. More information in: Schmidmayer _et al._ [https://doi.org/10.1002/fld.5142](https://doi.org/10.1002/fld.5142). * Possibility to initialize an unstructured simulation with the result of a previous simulation performed on a similar mesh and/or a different number of CPUs. This feature is particularly useful to fasten steady state convergence on a fine mesh using coarse mesh results. More information in: Cazé _et al._ (2023). Modelling and simulation of the cavitation phenomenon in space-engine turbopumps. [http://dx.doi.org/10.2139/ssrn.4479910](http://dx.doi.org/10.2139/ssrn.4479910). * PUEq phase-change model (handled through PTMu relaxation) does not require mass fraction threshold anymore to trigger mass transfer. Also more information in: Cazé _et al._ (2023). [http://dx.doi.org/10.2139/ssrn.4479910](http://dx.doi.org/10.2139/ssrn.4479910). * Add Moving Reference Frame coupling of a rotating region with a static one. Also more information in: Cazé _et al._ (2023). [http://dx.doi.org/10.2139/ssrn.4479910](http://dx.doi.org/10.2139/ssrn.4479910). * Renamed boundary condition names (break compatibility with previous version). * Immersed boundaries can be added in a Cartesian mesh domain (physicalEntity = -1). Similarly to: Trummler _et al._ [https://doi.org/10.1017/jfm.2020.432](https://doi.org/10.1017/jfm.2020.432). **Minor points:** * Option to record boundary quantities such as pressure forces and shear stress (useful for aerodynamic coefficient computation). * Possibility to display cells' reference length on XML output with unstructured mesh. * Add a tutorial on mesh mapping and low-Mach preconditionning options. * Add scripts related to droplet shock-induced cavitation. More information in: Biasiori-Poulanges & Schmidmayer (2023) [https://doi.org/10.1063/5.0127105](https://doi.org/10.1063/5.0127105) and Schmidmayer & Biasiori-Poulanges (2023) [https://doi.org/10.1063/5.0151404](https://doi.org/10.1063/5.0151404). * Improve variable name style for Gnuplot output. * Increase code coverage by nonreg. * Updated AMR refinement criteria to match the different modelling. * Always a little more source code translation from French to English. * A wall is now considered as a symmetry BC if viscosity is ignored. * Update of documentation. **Fixes:** * Fix the getTemperature() and copyPhase() methods for multiphase models. * Fix restart bug when using alphaNull on PUEq model. * Fix bug on simulation progress when using iteration control mode. * Fix a MPI data type that caused a crash during MPI_Allreduce on some compilers. * Minor fix for Euler-Korteweg model. * Correction on UEq BC for volume fraction flux, using sM instead of uStar to respect transport equation. * Update non-regression scripts to make them compatible with OS X. * Correction for phase change with a second-order method.
Version 3.1 Major points: * Add velocity-equilibrium model. Read the following paper for more information: K. Schmidmayer, J. Cazé, F. Petitpas, E. Daniel, N. Favrie. "Modelling interactions between waves and diffused interfaces". 2021. * Add velocity-equilibrium model based on total-energy equations. Read the previous paper for more information. * Possibility to use another gradient method (Green-Gauss) which is compatible with unstructured meshes. Previously, finite differences were the only available. * Possibility to use low-Mach preconditioning technique for Euler & pressure-velocity-equilibrium models. * Option to use 1D with smooth cross-section-variation scheme (useful for nozzles). * Add models augmented Euler-Korteweg and defocusing non-linear Schrödinger as well as Van der Waals equation of state. Mainly experimental work. Minor points: * Possibility to reduce output variable number for UEq and PUEq models. * Restarting computation possible for all models for XML format output. * Possibility to record flux on boundary (massflow, total enthalpy) and/or flow quantities (density, pressure, velocity) in agreement with Moving Reference Frame method. * Possibility to set precision for specific outputs (probes, cuts, global quantities, boundary flux/quantities). * Possibility to run test cases from a different folder. * Minor modification of axi-symmetric computations for second-order simulations. * Better handling of pressure relaxations when Newton-Raphson method does not converge. * Update of documentation regarding test cases. Fixes: * Fix recording of probes with computational restart. * Fix bug on unstructured meshes using hexaedron and prisms. * Correction on NASG EOS to allow null volume fractions. * Fix bug GNU dataset output precision. * Fix bug AMR parallel when large domains were used * Fix bug AMR parallel on printing of load.