Computational fluid dynamics and multiphysics modeling
Title: Computational fluid dynamics and multiphysics modeling
DNr: LiU-compute-2024-20
Project Type: LiU Compute
Principal Investigator: Magnus Andersson <>
Affiliation: Linköpings universitet
Duration: 2024-06-10 – 2025-07-01
Classification: 20306


This application involves several sub-areas within the field of Computational Mechanics and Engineering, dealing with numerical simulations of fluid flows; often coupled with other types of physics such as heat transfer, structural dynamics, fluid-structure interaction, chemical reactions, multiphase, etc. at different spatio-temporal scales. A short summary of the different areas (projects) is given below. Cavitation-induced erosion in oil-hydraulic systems is a huge concern in many fields, reducing the life expectancy of the components, and is essential to minimize. To better understand how and where these hazardous events may occur in realistic systems requires sufficient modeling strategies at multiple tiers (from the fully/sub system to benchmark/unit problems). This calls for very highly resolved multiphase CFD simulations, including cavitation and erosion modelling, coupled with co-simulation frameworks. Additive manufacturing makes it possible to produce complex lattice structures with high quality, which may be a paradigm shift in application designs for many industries, where rapid design-space exploration tools for multi-physics problems are missing. Here, we seek to develop design tools based on topology and shape optimization for generating ultra-lightweight multifunctional solid-lattice designs for problem involving fluid flows and heat transfer, where high-fidelity (large/multi-scale) multi-physics simulations play a key role; both for e.g. metamodel development and validation of final designs. The production of biogas plays an important role in achieving sustainable development goals, but where current processes lack optimal efficiency. Here we focus on investigating different mixing processing using low/high-fidelity large-scale, multiphase CFD simulations coupled with compartmental modeling, which will be used to link CFD with kinetic models for investigating parameters that are affected by mixing. The design of components such as gas turbine blades or hydrokinetic turbines downstream hydropower plants requires processes that are highly iterative, involving complex physics and a large parameter space. This makes large-scale experiments and CFD methods impractical and the need for faster and more robust approaches for design performance analysis. This project seeks to investigate the possibility of using a CFD-driven machine learning framework (Physics-Informed Neural Networks), to reduce computational times, improve design efficiency, and accelerate the innovation cycle for real-world flow-related multiphysics problems. This will involve generation of large sets of CFD data used for the training phase, as well as for validating the predictability of the AI model. This project concerns patient-specific models of cardiovascular flow (heart as well as blood vessels). Apart from an increased understanding of the normal and abnormal blood flow in the human body we target intervention planning as well as follow-up and diagnostic aid for different reconstruction procedures. To establish such capability, a thorough understanding of normal flow conditions is required. We utilize the basic principles of fluid dynamics as well as the modelling and simulation capabilities from computational engineering and high-performance computing in combination with modern imaging modalities and image processing. With the introduction of very high-resolution CT (photon counting CT, PCCT) the need for HPC is increased significantly.