Dr. Dionysios Angelidis, Featured Researcher
Dr. Angelidis’ research is primarily focused on the development of novel computational fluid dynamics (CFD) methods and codes which allow high fidelity calculations of turbulent flows; taking advantage of high-performance computing (HPC). He specializes in the development of new generationa adaptive mesh refinement (AMR) computational codes which enable multi-resolution scalable calculations with low computational cost. To efficiently tackle challenging real-life flow problems in scientific and engineering applications, his algorithms use state-of-the-art numerical methods and algorithms including: a) adaptive mesh refinement; b) immersed boundary methods; c) actuator line parameterization wind turbine rotors; d) large-eddy-simulations (LES) on locally refined grids, e) multi-phase flows using level-set methods (LSM) and f) fluid-structure-interaction modeling.
The major feature of the state-of-the-art immersed boundary - adaptive mesh refinement (IB-AMR) flow solver, developed in Dean Fotis Sotiropoulos' research group, is that even though the grid generation or regeneration algorithm may adopt a hierarchical / octree structure, the flow solver uses a fully unstructured data structure. Such an approach facilitates utilization of immersed boundary methods and satisfaction of the divergence free constraint of the continuity equation. A hybrid staggered/non-staggered grid approach is employed for discretizing the governing equations. The Navier-Stokes, continuity and scalar transport equations are discretized in space using three-point central finite differencing and integrated in time via a second-order accurate fractional step method. The level set equations are discretized with a third-order WENO scheme in space, and fourth-order Runge-Kutta scheme in time. An algebraic multigrid acceleration along with GMRES solver is used to solve the pressure Poisson equation. A matrix-free Newton-Krylov method is used for the filtered momentum equation, the scalar transport equations and the level set equation. The code is parallelized using MPI (Message Passing Interface). PETSc and Hypre libraries are employed for solving the discretized equations.
The two-phase flow version of the IB-AMR solver enable the performance of high-fidelity multi-resolution two-phase flow calculations on locally refined grids. For instance, the CFD code is being used to accurately capture the dynamics of the cavity formation and the water ejected as bodies hit the water; resolving, at the same time, the formidable range of temporal and spatial scales with low computational cost. Funded by the National Science Foundation (CBET-1509071) and in collaboration with Professor Eva Kanso’s group from the University of Southern California, Dr. Angelidis performs high-fidelity two-phase flow calculations on locally refined grids to investigate numerically the water entry of a V-shaped heavy wedge. The multi-resolution simulations enabled accurately capture the prevailing dynamics of the cavity formation and the water ejected as the body hits the water, in line with the corresponding experimental measurements (Figure 1).
In collaboration with Professor Alexandra Techet’s group and the Center of Ocean Engineering of the Massachusetts Institute of Technology (MIT), Dr. Angelidis, working in Dean Sotiropoulos’ group, uses and develops powerful computational codes to perform two-phase flow, fluid-structured-interaction (FSI) calculations on adaptively refined grids to investigate the hydrodynamics and aerodynamics of a jumping archer fish. Small scale archer fish, also known as Toxotes microlepis, are best known for spitting jets of water to capture prey, but also hunt by jumping out of the water to heights of up to 2.5 body lengths (typical lengths range from 7 to 11 cm). This study, coupled with high-speed imaging and particle image velocimetry (PIV) measurements will ultimately enable assessment of the role of jumping as a competitive foraging strategy. Figure 2 shows the simulated flow field after performing FSI calculations of an archer fish jumping out of the water. This study also reveals that the high-fidelity computational tools enable accurate and realistic two-phase biological-flow calculations.
Lastly, extracting energy from submerged tidal hydrokinetic turbines is an effective way to harness renewable energy. The major challenge in performing high fidelity calculations over an entire array of turbines is the multi-scale nature of the problem. In the recently completed project funded by NSF award IIP-1318201 entitled “The Roosevelt Island Tidal Energy (RITE) project”, Dean Sotiropoulos’ group simulated an array of hydrokinetic turbines in the East River using high-resolution bathymetry collected by Verdant Power Inc. (VPI). Essentially, during this project Dr. Angelidis further enhanced the capabilities of the IB-AMR code to perform high-performance computing (HPC) simulations with local mesh refinement, to improve both the design and the structural reliability of a new rotor design. In collaboration with Dr. Chawdhary, the local-mesh refinement CFD code was used to simulate an array of hydrokinetic turbines in the East River using high-resolution bathymetry collected by VPI. The RITE project became the world’s first grid-connected tidal power array and the first United States tidal power project to receive a Federal Energy Regulatory Commission (FERC) license. The IB-AMR code was applied to carry out the first-ever, turbulence resolving MHK array simulation for the RITE project using actual high-resolution bathymetry of the East River (Figure 3) and allowed the investigation of the wake characteristics and persistence of the VPI TriFrameTM system. Local mesh refinement around each turbine in the array enabled the resolution and mutual interactions of the coherent background induced by each one of the 30 turbines in the array while resolving the interactions of the entire MHK array with turbulence induced by the large-scale bathymetric features of the East River. The solution of the governing equations on locally refined grids enabled blade resolving calculations while also considering the geometrical details of the entire turbine.
Previous Featured Researchers
Dr. Leila Hajibabai, Spring 2017