Grzegorz P. Filip, Ph.D.

Seakeeping and Wave Loads in Extreme Marine Environments

This project focuses on the performance assessment of vessels operating in extreme marine environments. Because extreme events are stochastic processes, long-exposure-time windows are necessary to observe the desired extreme and thus the use of high-fidelity CFD tools is an expensive or even an impossible proposition. To alleviate this limitation, the design-loads generator (DLG) is used to combine information about the dynamical system and the operating environment to construct deterministic extreme events within relatively short time windows.

A series of long-crested and short-crested extreme wave environments

The information necessary to generate the extreme responses can be obtained from regular-wave seakeeping analysis and the associated response amplitude operators (RAO). Two sample snapshots of regular-wave seakeeping simulations are shown below.

Seakeeping simulation in long-crested head waves

Frictional Drag Reduction Using Air-Layer Injection

The dynamics of the air-layer injected below a simplified ship hull are studied using numerical and experimental methods. The goal of the project is the accurately predict the air cavity topology and its influence on the frictional drag component. The cavity topology is simulated using large-eddy simulations (LES) and the water speed, the air injection rate, and the turbulent boundary layer thickness are varied to determine each parameter's impact on the flow.

Air cavity topology and ship hull pressure contours as a function of the flow parameters

The numerical simulations accurately predict the mean and the instantaneous flow features and thus these simulations allow for a deeper insight into this complex flow. A sample snapshot of the air cavity near the injector is compared to the experimental images below.

Details of the air cavity near the injector

Improving Off-Design Performance of a Vessel Through Surrogate Modeling

A novel methodology based on design space exploration and exploitation through surrogate modeling is used to retrofit the bulbous bow of a modern container ship. The retrofit aims to improve the performance of the vessel and to reduce its greenhouse gas emissions in the off-design conditions associated with the slow-steaming operating profile.

Bulbous bow design candidates and the associated power score surrogate model

The design evaluation is performed at the full scale of the vessel using state-of-the-art computational fluid dynamics (CFD) solvers. The resultant power consumption predictions are used to generated power score surrogate models in MATLAB and to generate a series of new design candidates in Rhinoceros 3D with improved performance characteristics. The surrogate models produce a new bulb design that reduces the overall power requirements and thus the vessel's fuel consumption by 7% across the entire operating profile.

Free-surface elevation near the bow of the original bulb design compared against the H1-C1 design candidate

Vehicle Drag Reduction Using Vortex Generators

Cutting-edge CFD tools are used to examine the impact of vortex generators on the rear separation region of ground vehicles. Relatively small geometric components placed near the separation region are found to strongly influence the pressure gradient and thus to affect the drag coefficient of the vehicle. The drag reduction achieved through this approach translates directly into fuel consumption savings.

Wake and near-body fow of a simplified ground vehicle and the effect of vortex generators on the rear separation region

The future of the project is focused on an automated adjoint-based optimization of the vortex generators for real vehicles. The design evaluation is based on the CFD tools combined with modern turbulence models that are necessary to resolve the complex turbulent flow field around the vehicle geometry. Several of these large-eddy simulation (LES) and detached-eddy simulation (DES) models are examined for the simplified ground vehicle flow and summarized in Fig. 4. The importance of selecting an appropriate turbulence model and a sufficiently resolved computational grid are clearly demonstrated by comparing the family of force coefficients against the experimental measurements.

Force coefficients associated with various components of the simplied ground vehicle

Spilling and Plunging Breaking Waves

The ultimate goal of this project is to simulate high Reynolds number marine flows characterized by breaking waves using high fidelity large-eddy simulation (LES) techniques. The presence of a complex unsteady fluid interface in such flows poses LES modeling challenges which are to be addressed by a priori testing of direct numerical simulations of less complex multiphase flows. The wave breaking process and the importance of the air-water interface resolution are studied by directly solving the unsteady multiphase Navier-Stokes equations. The focus is on simulating two and three-dimensional spilling and plunging breaking waves in order to determine the smallest turbulent scales present during these dynamic processes. The influence of the thickness of the interface transition region on the solution and the energy dissipation rate are also studied on multiple levels of grid refinement.

Local dissipation rate in water at four instances in time

Total mechanical energy in water (left) and air (right) as a function of time and varying resolution of the air-water interface

Numerical simulation of a plunging breaking wave

Tension Leg Platform Design for Deep Water Floating Offshore Wind Turbines

The purpose of this work is to examine the use of computational fluid dynamics (CFD) in the design process of a floating offshore wind turbine. Select hydrodynamic parameters of the tension-leg platform (TLP) model (1/50th scale) based on the PelaStar design from the Glosten Associates are computed and compared against recent experimental data. The hydrodynamic parameters used for the numerical tool validation are the heave and pitch viscous damping coefficients which influence the structural response of the TLP. The viscous damping coefficients are typically found using model-scale experiments which are expensive and ideally replaced by CFD in the near future.

The PetaStar offshore wind turbine design and design alternatives for shallow and deep water regions

Schematic of the TLP design and the associated pitch decay time series as a function of mooring material

The Effects of Streamlined Rigging on Sailboat Performance

The influence of two cross-section shapes of the rigging on the performance of an IMS40 sailing yacht were investigated. A modified NACA foil and a circular section were selected for the comparison, and both shapes were analyzed with numerical simulations and physical experiments. The physical experiments were conducted inside of a low-Mach number wind tunnel and a water tow-tank at the University of Michigan using a custom designed and manufactured 2-DOF dynamometer. The change in sailboat performance due to the different aerodynamic and material properties of the rigging were determined. A velocity prediction program (VPP) was used to quantify the performance change by comparing the time required to sail two different race courses.

Physical model tests of the streamlined rigging in the low-Mach wind tunnel (left) and towing tank (right)

Numerical grid at 45 degrees angle of attack (left) and the drive and heeling force coefficients of the streamlined foil (SCR) and standard cylindrical rod from experiments (right)

Vortex Induced Vibrations Aquatic Clean Energy (VIVACE) Converter

The novel VIVACE converter was originated by Professor Michael Bernitsas in an effort to extract clean energy from water currents and rivers. The flow of water past circular cylinders results in a dynamic process known as vortex-induced vibrations (VIV) where the shed vorticies in the wake of the cylinders force the cylinders to move. A power take-off system and a series of springs bound the motion to a linear oscillation as described in Fig. 12 below.

Description of the VIVACE converter concept and one of the model-scale experiments performed at the Marine Renewable Energy Laboratory

Visualized vortex shedding behind the cylinder from physical experiments (left) and numerical simulation (right)

The operating profiles for various water speeds and converter configurations have been quantified experimentally and using numerical simulation techniques. Several prototypes were built at the Marine Renewable Energy Laboratory and tested inside of the low-turbulence water channel (LTWC) and the tow tank. Currently, small-scale prototypes are being evaluated in several rivers and large grid-scale deployment is expected to generated energy at a competitive rate of $0.10/kWh.

Flow Around a Vertical Circular Cylinder

In this work, the flow around a vertical-oriented free-surface-piercing cylinder is simulated with high resolution in both the space and time coordinates. The spectrum of time-advancement and turbulence-modeling strategies to solve appropriate equations for this flow is broad. The OpenFOAM CFD toolkit is used as the platform for the development of our simulation tool. In this case, the fractional step (FSM) time advancement algorithm is benchmarked against the library-standard PISO-type method. Also, several different sub-grid stress models, including a localized dynamic Smagorinsky model and a localized dynamic kinetic energy transport model are studied.

Time-averaged and instantaneous free-surface elevation at Fn = 0.80

Time-averaged streamwise velocity slices as a function of depth

DARPA SUBOFF Large-Eddy Simulation Benchmark

Numerical prediction of marine hydrodynamic problems with an emphasis on high efficiency and accuracy has become an important capability of modern naval research facilities. One of the must common numerical tools, the Reynolds-Averaged Navier-Stokes (RANS) methods, offer the benefits of relatively small computational costs and a long history of application in the aeronautical and marine industries. These methods rely on turbulence modeling throughout the computational domain and only resolve large-scale time-averaged flow unsteadiness. Unlike RANS, large-eddy simulation (LES) and detached-eddy simulation (DES) methods resolve a significant portion of the turbulent flow and model only the smallest-scale turbulence. In principle, this resolution allows for a better prediction of the unsteady-flow features such as complex wakes and vortex structures commonly found in real high Reynolds number flows.

Turbulent structures colored by mean pressure (left) and velocity slices along the length of the SUBOFF AFF8 geometry (right)

Flow Past a Square Cylinder - Validation Study

Reynolds-Averaged Navier-Stokes (RANS) solvers are today's industry standard and have been applied to a wide range of flows. The time averaging inherent to RANS allows for significant computational time savings but largely removes the instantaneous effects of unsteady structures such as shed vortices. Large eddy simulations (LES) are thought to be better suited for unsteady vortex shedding flows at the price of increased computational time. In LES, the energy-containing vortex structures are resolved using the numerical grid, while the small, ideally isotropic structures are modeled using a subgrid-stress model such as the Smagorinsky model.

In this study of a flow past a square cylinder, the implementation of the dynamic version of the Smagorinsky model inside of the open-source CFD toolkit OpenFOAM is validated. In Fig. 17, the mean axial velocity and two RMS velocity components are compared to experimental results of Lyn et al. (1995) and LES results of Kim (2004).

Velocity components for a flow past a square cylinder at Re = 22,000

Favorable agreement has been found between our LES results and those of Lyn et al. The present results also show better correlation with the experimental results especially in the far region of the wake.