Computers & Fluids最新文献

{"title":"Towards a fully well-balanced and entropy-stable scheme for the Euler equations with gravity: General equations of state","authors":"Victor Michel-Dansac,&nbsp;Andrea Thomann","doi":"10.1016/j.compfluid.2025.106853","DOIUrl":"10.1016/j.compfluid.2025.106853","url":null,"abstract":"<div><div>The present work concerns the derivation of a fully well-balanced Godunov-type finite volume scheme for the Euler equations with a gravitational potential based on an approximate Riemann solver in a one-dimensional framework. It is an extension to general equations of states of the entropy-stable and fully well-balanced scheme for ideal gases recently forwarded in Berthon et al., (2025). A second-order extension preserving the properties of the first-order scheme is given. The scheme is provably entropy-stable and positivity-preserving for all thermodynamic variables. Numerical test cases illustrate the performance and entropy stability of the new scheme, using six different equations of state as examples, four analytic and two tabulated ones.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"303 ","pages":"Article 106853"},"PeriodicalIF":3.0,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145204237","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MLD-PINN: A multi-level datasets training method in Physics-Informed Neural Networks","authors":"Yao-Hsuan Tsai ,&nbsp;Hsiao-Tung Juan ,&nbsp;Pao-Hsiung Chiu ,&nbsp;Chao-An Lin","doi":"10.1016/j.compfluid.2025.106849","DOIUrl":"10.1016/j.compfluid.2025.106849","url":null,"abstract":"<div><div>Physics-Informed Neural Networks (PINNs) have emerged as a promising methodology for solving partial differential equations (PDEs), gaining significant attention in computer science and various physics-related fields. Despite demonstrating the ability to incorporate physical laws for versatile applications, PINNs still struggle with challenging problems that are stiff to solve and/or have high-frequency components in their solutions, resulting in accuracy and convergence issues. These problems not only increase computational costs but may also lead to accuracy loss or solution divergence in the worst-case scenario. In this study, we introduce a novel PINN framework, dubbed MLD-PINN, to mitigate the above-mentioned problems. Inspired by the multigrid method in the CFD community, the underlying idea of our approach is to efficiently remove different frequency errors by training with different levels of training samples. This provides a simpler way to improve training accuracy without spending time fine-tuning neural network structures, loss weights, or hyperparameters. To demonstrate the efficacy of our approach, we first investigate a canonical 1D ODE with high-frequency components and a 2D convection–diffusion equation using a V-cycle training strategy. Finally, we apply our method to the classical benchmark problem of steady lid-driven cavity flows at different Reynolds numbers (Re) to examine its applicability and efficacy for problems involving multiple modes of high and low frequencies. Through various training sequence modes, our predictions achieve 30% to 60% accuracy improvement. We also investigate the synergy between our method and transfer learning techniques for more challenging problems (i.e., higher <span><math><mrow><mi>R</mi><mi>e</mi></mrow></math></span> cases). The present results reveal that our framework can produce good predictions even for the case of <span><math><mrow><mi>R</mi><mi>e</mi><mo>=</mo><mn>5000</mn></mrow></math></span>, demonstrating its ability to solve complex high-frequency PDEs.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"303 ","pages":"Article 106849"},"PeriodicalIF":3.0,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145219066","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Invariant control strategies for active flow control using graph neural networks","authors":"Marius Kurz ,&nbsp;Rohan Kaushik ,&nbsp;Marcel Blind ,&nbsp;Patrick Kopper ,&nbsp;Anna Schwarz ,&nbsp;Felix Rodach ,&nbsp;Andrea Beck","doi":"10.1016/j.compfluid.2025.106854","DOIUrl":"10.1016/j.compfluid.2025.106854","url":null,"abstract":"<div><div>Reinforcement learning (RL) has recently gained traction for active flow control tasks, with initial applications exploring drag mitigation via flow field augmentation around a two-dimensional cylinder. RL has since been extended to more complex turbulent flows and has shown significant potential in learning complex control strategies. However, such applications remain computationally challenging owing to its sample inefficiency and associated simulation costs. This fact is worsened by the lack of generalization capabilities of these trained policy networks, often being implicitly tied to the input configurations of their training conditions. In this work, we propose the use of graph neural networks (GNNs) to address this particular limitation, effectively increasing the range of applicability and getting more <em>value</em> out of the upfront RL training cost. GNNs can naturally process unstructured, three-dimensional flow data, preserving spatial relationships without the constraints of a Cartesian grid. Additionally, they incorporate rotational, reflectional, and permutation invariance into the learned control policies, thus improving generalization and thereby removing the shortcomings of commonly used convolutional neural networks (CNNs) or multilayer perceptron (MLP) architectures. To demonstrate the effectiveness of this approach, we revisit the well-established two-dimensional cylinder benchmark problem for active flow control. The RL training is implemented using Relexi, a high-performance RL framework, with flow simulations conducted in parallel using the high-order discontinuous Galerkin framework FLEXI. Our results show that GNN-based control policies achieve comparable performance to existing methods while benefiting from improved generalization properties. This work establishes GNNs as a promising architecture for RL-based flow control and highlights the capabilities of Relexi and FLEXI for large-scale RL applications in fluid dynamics.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"303 ","pages":"Article 106854"},"PeriodicalIF":3.0,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145218983","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-reflecting boundary conditions for updated Lagrangian hydrodynamics simulations","authors":"N. Therme,&nbsp;S. Guisset","doi":"10.1016/j.compfluid.2025.106848","DOIUrl":"10.1016/j.compfluid.2025.106848","url":null,"abstract":"<div><div>This paper presents a non-reflecting boundary condition (NRBC) strategy specifically designed for compressible flow simulations within an updated Lagrangian hydrodynamic framework. The method is grounded in the characteristic decomposition of the linearized Euler equations, extending the approach pioneered by Giles (Giles, 1990). By analyzing the local eigenstructure of the system, incoming and outgoing waves are precisely identified at the domain boundaries. This identification enables the selective suppression of incoming waves, effectively preventing unwanted reflections. A key advantage of this NRBC approach is its full compatibility with all standard updated Lagrangian schemes, whether staggered or cell-centered. The practical implementation of the NRBC is discussed in detail. Numerical experiments conducted on one-dimensional and two-dimensional problems demonstrate the method’s effectiveness in absorbing outgoing disturbances, confirming its non-reflecting properties.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106848"},"PeriodicalIF":3.0,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217603","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Constructing the body source for high-order lattice Boltzmann method","authors":"Shunyang Li ,&nbsp;Li Wan ,&nbsp;Nan Gui ,&nbsp;Xingtuan Yang ,&nbsp;Jiyuan Tu ,&nbsp;Shengyao Jiang","doi":"10.1016/j.compfluid.2025.106851","DOIUrl":"10.1016/j.compfluid.2025.106851","url":null,"abstract":"<div><div>This paper presents a novel strategy for constructing body source terms in the high-order lattice Boltzmann method (LBM), designed to efficiently introduce various physical phenomena by modifying the non-equilibrium distribution function. The source term, expressed as a Hermite polynomial, provides a flexible framework for simulating complex fluid flows. Three typical source terms are given: a body force source for gravity-driven flows, a thermal dissipation source for controlling the Prandtl number, and a pressure tensor source for modeling multiphase flows. Chapman-Enskog analysis confirms that the source terms recover the expected macroscopic equations. Notably, the proposed strategy eliminates the need for explicit construction of the collision operator, a challenge in conventional approaches for handling diverse physical scenarios. Furthermore, the method is compatible with the traditional BGK model, ensuring its applicability to various high-order lattices. The model’s accuracy and versatility are validated through a series of benchmark tests, showing excellent agreement with existing literature results.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106851"},"PeriodicalIF":3.0,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel numerical method for applications of aero-icing predictions","authors":"B. An ,&nbsp;C. Xi ,&nbsp;F. Mellibovsky ,&nbsp;J.M. Bergada ,&nbsp;D. Li ,&nbsp;W.M. Sang","doi":"10.1016/j.compfluid.2025.106850","DOIUrl":"10.1016/j.compfluid.2025.106850","url":null,"abstract":"<div><div>We present a new numerical algorithm for predictions of water droplets accumulation responsible for in-flight ice accretion. Unlike the traditional Lagrangian and Eulerian methodologies, the new approach is based on the single-phase multi-component lattice Boltzmann method (SPMC-LBM), which is a mesoscopic algorithm that focuses on the movement of air particles and water droplets in gas phase (vapour particles). The trajectory and impingement of the vapour particles is taken as approximately equivalent to that of water droplets. The ice growth process is simulated numerically by employing the classic Messinger’s model. The tree grid structure is used for local grid refinement and to improve the computational efficiency and robustness. To simulate water collection processes, we propose a novel approach for treating vapour particles at curved boundaries. Additionally, we develop a nondimensionalization method to convert physical diffusion coefficients into lattice diffusion coefficients, effectively capturing diffusion effects in multi-component mixed flows. For the heavier component in the mixed flow, a Lagrangian 9-bit interpolation scheme is employed for the supplement of the streaming process of the distribution functions. The numerical results show this novel method agrees well with experimental data, having a potential for development to allow tackling three dimensional geometries and complicated icing conditions. The adoption of a tree grid substantially enhances the mesh generation process. Notice as well it is an improvement for LBM related applications, since it is for the first time that LBM has been employed in the predictions on aero-icing problems.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106850"},"PeriodicalIF":3.0,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimized workload distribution for GPU-accelerated combustion simulations in heterogeneous CPU–GPU architectures","authors":"Álvaro Moure,&nbsp;Anurag Surapaneni,&nbsp;Daniel Mira","doi":"10.1016/j.compfluid.2025.106846","DOIUrl":"10.1016/j.compfluid.2025.106846","url":null,"abstract":"<div><div>This work presents a set of workload distribution algorithms designed to optimize the hybrid use of CPUs and GPUs in reacting flow simulations on heterogeneous High-Performance Computing (HPC) systems. The algorithms extend advanced computational software originally developed for CPUs to hybrid CPU–GPU environments. Unlike GPU-exclusive software, hybrid codes require specialized orchestration to maximize GPU utilization while minimizing CPU idle time. Combustion simulations are computationally demanding due to the evaluation of non-linear source terms and the transport of large number of PDEs with strong imbalanced MPI workloads, so it requires highly efficient codes with advanced parallel algorithms. Algorithms based on different MPI-GPU mapping roles defined to maximize chemistry batch size while reducing GPU communication overhead are proposed to accelerate combustion simulations using heterogeneous HPC systems. These approaches offload the expensive chemical integration step to the GPUs, while the transport remains on the CPUs using an operator splitting technique. Stiff chemical integration is GPU-accelerated with <span>ChemInt</span>, a newly developed CPU/GPU-compatible <span>C++</span>/<span>CUDA</span> library designed for coupling with CPU-based CFD codes. A comparison of the different approaches is presented and discussed demonstrating performance improvements of more than threefold over CPU-only executions.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106846"},"PeriodicalIF":3.0,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145155923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Density-based topology optimization strategy for optimal design of uniform flow manifolds","authors":"Sanjay Vermani ,&nbsp;Nitish Anand","doi":"10.1016/j.compfluid.2025.106847","DOIUrl":"10.1016/j.compfluid.2025.106847","url":null,"abstract":"<div><div>Flow manifolds are devices that distribute or collect fluid across multiple channels, playing a crucial role in the performance of many fluid and energy systems. However, designing efficient manifolds that ensure uniform flow distribution remains challenging, especially for multi-channel three-dimensional manifolds. This study presents a scalable topology optimization framework for systematically designing multi-channel flow manifolds. The proposed method builds on the conventional density-based topology optimization formulation by introducing a flow maldistribution coefficient as an explicit constraint. This novel approach was implemented using the incompressible Navier–Stokes flow solver available in the open-source CFD suite SU2. The performance of the proposed method was benchmarked against two established topology optimization strategies using an exemplary planar z-type flow manifold, where both the inlet and outlet manifolds were designed simultaneously. The results demonstrate that the proposed method achieves flow uniformity comparable to established approaches while significantly reducing computational costs. Furthermore, when applied to large-scale three-dimensional problems, the proposed method produces feasible designs that achieve uniform flow distribution and exhibit innovative geometrical features. These results highlight the robustness and scalability of the proposed method.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106847"},"PeriodicalIF":3.0,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A 3D Eulerian meshless conservative level set method for two-phase flows with complex geometries","authors":"Anand S. Bharadwaj ,&nbsp;B. Premachandran","doi":"10.1016/j.compfluid.2025.106839","DOIUrl":"10.1016/j.compfluid.2025.106839","url":null,"abstract":"<div><div>In this paper, we develop a 3D Eulerian meshless method based on the conservative level set method targeted to solve two-phase flows with complex geometries. The method combines the advantages of Eulerian methods and meshless methods. Being an Eulerian method, it does not require neighbourhood estimation every time step. At the same time, being a meshless method, it does not require mesh connectivity between points in the domain and consequently, alleviates the difficulty of mesh-generation, makes point cloud adaptation and simulation with complex geometries relatively straight forward. In this method, we use a point cloud generating algorithm as a part of the fluid-solver, which can be used to generate or change the point cloud whenever there are any geometric changes in the domain. The meshless method is based on the Generalized Finite Difference Method (GFDM), which uses differential operators that are derived from a least-squares error minimization procedure. The advection equation of the volume fraction <span><math><mi>α</mi></math></span> uses the Directional Flux Error Minimization (DFEM) scheme. The 5th order WENO scheme is used for the flux-reconstruction in the advection equation. The interface sharpening step is performed at regular intervals to ensure that the sharpness of the interface is retained, thus, reducing the mass losses associated with the dissipative errors in the advection step. To further improve the accuracy, we propose the adaptation of the point cloud in the vicinity of the interface using the convolution of the volume fraction function (<span><math><mi>α</mi></math></span>). The method is validated using benchmark test cases. Additionally, some flow problems involving complex geometries are presented — flow through a porous cavity with uniform and randomly distributed obstacles and the flow of molten metal in the casting of a helical bevel gear.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106839"},"PeriodicalIF":3.0,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Phase-field modeling of melting processes in viscoelastic materials","authors":"Hongfei Xie,&nbsp;Ying Meng,&nbsp;Kang Luo,&nbsp;Hongliang Yi","doi":"10.1016/j.compfluid.2025.106845","DOIUrl":"10.1016/j.compfluid.2025.106845","url":null,"abstract":"<div><div>In this paper, the solid–liquid phase change problems of viscoelastic materials are solved using the spectral element method (SEM). The phase-field method is employed to simulate the melting dynamics, while the log-conformation reformulation (LCR) method is utilized to address the high Weissenberg number problem (HWNP). The reliability of the proposed numerical model is systematically validated through benchmark comparisons with literature data and finite volume method (FVM) simulations. Subsequently, a comprehensive analysis is conducted to investigate the influence of viscoelastic effects on the melting dynamics in a cavity under different Rayleigh and Weissenberg numbers. The computational results demonstrate that the viscoelastic effect significantly enhances heat transfer efficiency and accelerates the melting process. This enhancement can be attributed to the viscoelastic effects that effectively reduce flow resistance while enhancing the average kinetic energy of convective flows.</div></div>","PeriodicalId":287,"journal":{"name":"Computers & Fluids","volume":"302 ","pages":"Article 106845"},"PeriodicalIF":3.0,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145155924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}