Sharp-interface Level-set Method Research Articles

A dual grid-based deep reinforcement learning and computational fluid dynamics method for flow control and its application to nucleate pool boiling

This paper presents a proposition of a dual grid (DG)-based coupled deep reinforcement learning (DRL) and computational fluid dynamics (CFD) method for active flow control. The DG-DRL-CFD method uses a dual-resolution grid, with a coarser grid for the training phase and a finer grid for the testing phase of the DRL-CFD method. Further, after a validation study for our DRL-CFD-based parallelized in-house solvers, this paper presents a performance study for a periodic suction/ejection-based drag reduction for a cylinder in a channel-confined flow. Using an immersed boundary method for CFD on a Cartesian grid, this novel DG-DRL-CFD method is shown to result in almost same accuracy (within 1%) in a substantially reduced computational time as compared to the traditional DRL-CFD method (on the finer uniform grid). Finally, using a sharp interface level set method for an axisymmetric two-phase flow, this paper presents an application of the DR-DRL-CFD method for an oscillating base plate-based enhancement of heat transfer during nucleate pool boiling. As compared to our recent CFD study-based parametric study for the nucleate pool boiling problem, the present DG-DRL-CFD method leads to a profile and frequency of plate-oscillation that results in a larger value of average Nusselt number Nuavg for the plate. The coupled DRL-CFD method leads to plate oscillation profile giving higher enhancement in Nuavg, compared to an open-loop control strategy that involves a parametric sweep involving multiple simulations.

On ghost fluid method-based sharp interface level set method on a co-located grid and its comparison with balanced force-based diffuse interface method

This paper, on Computational multi-Fluid Dynamics (CmFD) development, presents an extension of the Ghost Fluid Method (GFM)-based Sharp Interface Level Set Method (SI-LSM) (originally proposed on a staggered grid) for a co-located grid system. Further, this paper presents a comparative study for the GFM-based SI-LSM and a balanced force method (BFM)-based Diffuse Interface Level Set Method (DI-LSM). The BFM-based DI-LSM avoids a pressure-interfacial force decoupling, during the implementation of the MIM, by a proper-discretization of the surface tension term; not needed for the single-phase flow. Whereas, the GFM-based SI-LSM implicitly couples two immiscible, incompressible fluids via interface jump condition, which makes the Momentum Interpolation Method (MIM) for the two-fluid flow same as that for a single-fluid flow. A unified, for both SI-LSM and DI-LSM, mathematical and numerical formulations are presented. Order of accuracy is demonstrated in-between first-order and second-order; similar to that reported earlier for the various CmFD methods. Finally, validation and relative-performance study are presented for four test cases: static droplet, dam break, drop coalescence, and Rayleigh-Taylor instability. For the static droplet, the relative-performance of the present SI-LSM as compared to the DI-LSM is found similar to that in the literature for SI-VOF (Volume of Fluid) method as compared to DI-VOF method. For the various test cases, superior performance is demonstrated for the SI-LSM as compared to DI-LSM. This paper presents a broader perspective to the MIM for a single-versus-two fluid-flow and diffuse-versus-sharp approaches; and a novel GFM-based SI-LSM on a co-located grid.

Adaptive interface-Mesh un-Refinement (AiMuR) based sharp-interface level-set-method for two-phase flow

Adaptive interface-Mesh un-Refinement (AiMuR) based Sharp-Interface Level-Set-Method (SI-LSM) is proposed for both uniform and non-uniform Cartesian-Grid. The AiMuR involves interface location based dynamic un-refinement (with merging of the four control volumes) of the Cartesian grid away from the interface. The un-refinement is proposed for the interface solver only. A detailed numerical methodology is presented for the AiMuR and ghost-fluid method based SI-LSM. Advantage of the novel as compared to the traditional SI-LSM is demonstrated with a detailed qualitative as well as quantitative performance study, involving the SI-LSMs on both coarse grid and fine grid, for three sufficiently different two-phase flow problems: dam break, breakup of a liquid jet and drop coalescence. A superior performance of AiMuR based SI-LSM is demonstrated - the AiMuR on a coarser non-uniform grid ( $$NU_{c}^{AiMuR}$$ ) is almost as accurate as the traditional SI-LSM on a uniform fine grid ( $$U_{f}$$ ) and takes a computational time almost same as that by the traditional SI-LSM on a uniform coarse grid ( $$U_{c}$$ ). The AMuR is different from the existing Adaptive Mesh Refinement (AMR) as the former involves only mesh un-refinement while the later involves both refinement and un-refinement of the mesh. Moreover, the proposed computational development is significant since the present adaptive un-refinement strategy is much simpler to implement as compared to that for the commonly used adaptive refinement strategies. The proposed numerical development can be extended to various other multi-physics, multi-disciplinary and multi-scale problems involving interfaces.

PERFORMANCE OF SHARP-VERSUS-DIFFUSE INTERFACE-BASED LEVEL SET METHOD ON A STAGGERED-VERSUS-CO-LOCATED GRID FOR CMFD

The present work is on comparison of computational performance for various level set methods (LSMs): diffuse interface level set method on staggered grid (DI-LSM<sub>stag</sub>), sharp-interface level set method on staggered grid (SI-LSM<sub>stag</sub>), diffuse interface level set method on co-located grid (DI-LSM<sub>col</sub>), and sharp interface level set method on co-located grid (SI-LSM<sub>col</sub>). Even though the implementations of the diffuse and sharp interface (DI and SI) approaches on staggered grid are straightforward, an additional pressure-interfacial force balance needs to be ensured on the co-located grid. This is established here with balanced force method (BFM) for the DI-LSM and ghost fluid method (GFM) for the SI-LSM. Computational performances of these LSMs are presented for a variety of computational multi-fluid dynamics (CMFD) problems: static drop, dam break, rising bubble, falling droplet, and droplet coalescence. Greater accuracy is found with SI-LSMs for the static drop, dam break, and rising bubble, whereas for the other problems, both SI-LSM and DI-LSM result in almost the same accuracy. Smaller computational time is taken by the SI-LSM for rising bubble and falling droplet, and by DI-LSM for the dam break and droplet coalescence. Comparing between grid systems, co-located grid resulted in greater accuracy for all the problems except falling droplet, for which both grid systems resulted in similar accuracy, whereas, a smaller computational time is taken by the co-located grid for rising bubble and falling droplet, and by the staggered grid for dam break and droplet coalescence. Overall, SI-LSM on the co-located grid shows better results with a slight increase in computational time as compared to the other LSMs, and is a suitable alternative to the staggered grid.

Investigation of cavitation bubble dynamics near a solid wall by high-resolution numerical simulation

We investigate dynamics of a single cavitation bubble in the vicinity of a horizontal wall throughout expansion and collapse using a sharp–interface level-set method. The numerical scheme is based on a finite-volume formulation with low-dissipation high-order reconstruction schemes. Viscosity and surface tension are taken into account. The simulations are conducted in three-dimensional axi-symmetric space. A wide range of initial bubble wall standoff distances is covered. We focus, however, on the near-wall region where the distance between the bubble and the wall is small. We reproduce three jetting regimes: needle, mixed, and regular jets. The needle jets impose a significant load on the solid wall, exceeding the force induced by the collapse of the pierced torus bubble. For intermediate standoff distances, the large delay time between jet impact and torus bubble collapse leads to a significant decrease in the imposed maximum wall pressure. A liquid film between bubble and wall is observed whenever the bubble is initially detached from the wall. Its thickness increases linearly for very small standoff distances and growths exponentially for intermediate distances leading to a significant increase in wall-normal bubble expansion and bubble asymmetry. For configurations where the torus bubble after jet impact reaches maximum size, the collapse time of the cavitation bubble also is maximal, leading to a plateau in the overall prolongation of the cycle time of the bubble. Once the initial bubble is attached to the solid wall, a significant drop of all macroscopic time and length scales toward a hemispherical evolution is observed.

JAX-Fluids: A fully-differentiable high-order computational fluid dynamics solver for compressible two-phase flows

Physical systems are governed by partial differential equations (PDEs). The Navier-Stokes equations describe fluid flows and are representative of nonlinear physical systems with complex spatio-temporal interactions. Fluid flows are omnipresent in nature and engineering applications, and their accurate simulation is essential for providing insights into these processes. While PDEs are typically solved with numerical methods, the recent success of machine learning (ML) has shown that ML methods can provide novel avenues of finding solutions to PDEs. ML is becoming more and more present in computational fluid dynamics (CFD). However, up to this date, there does not exist a general-purpose ML-CFD package which provides 1) powerful state-of-the-art numerical methods, 2) seamless hybridization of ML with CFD, and 3) automatic differentiation (AD) capabilities. AD in particular is essential to ML-CFD research as it provides gradient information and enables optimization of preexisting and novel CFD models. In this work, we propose JAX-Fluids: a comprehensive fully-differentiable CFD Python solver for compressible two-phase flows. JAX-Fluids is intended for ML-supported CFD research. The framework allows the simulation of complex fluid dynamics with phenomena like three-dimensional turbulence, compressibility effects, and two-phase flows. Written entirely in JAX, it is straightforward to include existing ML models into the proposed framework. Furthermore, JAX-Fluids enables end-to-end optimization. I.e., ML models can be optimized with gradients that are backpropagated through the entire CFD algorithm, and therefore contain not only information of the underlying PDE but also of the applied numerical methods. We believe that a Python package like JAX-Fluids is crucial to facilitate research at the intersection of ML and CFD and may pave the way for an era of differentiable fluid dynamics. Program summaryProgram title: JAX-FluidsCPC Library link to program files:https://doi.org/10.17632/pzvkwn5s6p.1Developer's repository link:https://github.com/tumaer/JAXFLUIDSCode Ocean capsule:https://codeocean.com/capsule/6819679Licensing provisions: GNU GPLv3Programming language: PythonSupplementary material: Source code; Examples; Videos: Moving solid bodies, Taylor-Green vortex, Rising bubble, Shock-bubble interaction.Nature of problem: The compressible Navier-Stokes equations describe continuum-scale fluid flows. These flows often involve highly complex flow phenomena such as shocks, material interfaces, and turbulence. The intrinsic nonlinear dynamics render the numerical simulation of the these equations challenging. Machine learning provides novel avenues for describing partial differential equations. Machine learning models rely on gradient information provided by automatic differentiation and are often implemented in Python. In contrast, existing high-performance computational fluid dynamics codes are typically written in Fortran or C++ and do not offer inherent automatic differentiation capabilities. These discrepancies hinder the advance of machine-learning-supported computational fluid dynamics. Up to this day, a general-purpose fully-differentiable computational fluid dynamics solver for compressible two-phase flows is missing.Solution method: We introduce JAX-Fluids: a general-purpose three-dimensional fully-differentiable computational fluid dynamics solver for compressible two-phase flows. JAX-Fluids is a simulation framework intended for machine-learning-supported computational fluid dynamics research. Our framework is written entirely in JAX, a high-performance numerical computing library with automatic differentiation capabilities. We have used an object-oriented programming style and a modular design philosophy. This allows the straightforward exchange of numerical methods. We provide a wide variety of state-of-the-art high-order computational methods for compressible flows. The modularity of our framework additionally facilitates the integration of custom subroutines. We use the sharp-interface level-set method to model two-phase flows. The software package can easily be installed as a Python package. We have build the source code around the JAX NumPy API. This makes JAX-Fluids accessible and performant. JAX-Fluids runs on CPUs, GPUs, and TPUs. We use HDF5 in combination with XDMF for writing output quantities. The Python packages Haiku and Optax are used for implementation and training of machine learning methods.Additional comments including restrictions and unusual features: JAX-Fluids relies on open-source third-party Python libraries. These are automatically installed. In the current version, JAX-Fluids only runs on a single accelerator (CPU/GPU/TPU). Future versions will include support for parallel execution. JAX-Fluids has been tested on Linux and macOS operating systems.

Jetting mechanisms in bubble-pair interactions

Jetting mechanisms in cavitation bubbles play a crucial role in the destructive forces of cavitation. Depending on the application, these forces can have desirable effects like in medical treatments or catastrophic effects like in the erosion of ship propellers. Still today, thorough understanding of all details in complex bubble collapse scenarios is lacking. Hence, in this work, we numerically investigate the jetting mechanisms for air bubble pairs in water following a recent experimental setup. We apply a finite-volume approach with fifth-order low-dissipation shock-capturing weighted essentially non-oscillatory reconstruction. The interface interaction is described by a conservative sharp-interface level-set method. For time integration, a third-order total-variation-diminishing Runge–Kutta scheme is employed. Complementing experimental observations, our simulations reveal the presence of dominating gas jets and new types of jetting mechanisms.

ALPACA - a level-set based sharp-interface multiresolution solver for conservation laws

ALPACA is a simulation environment for simulating hyperbolic and (incompletely) parabolic conservation laws with multiple distinct and immiscible phases. As prominent example, consider the compressible Navier-Stokes equations (NSE). Solutions to these equations give insight and understanding of many important engineering applications. Numerical simulations of nonlinear parabolic systems of equations are very challenging for their complex nonlinear dynamics including the propagations of discontinuities such as shocks and phase interfaces. Accurate predictions require high temporal and spatial resolutions for such multi-scale problems. We utilize low dissipation high-resolution methods to capture the dynamics inside the separate phases. Their interaction is modeled by a sharp-interface level-set method with conservative interface-interaction. This allows to accurately locate the interface position and to easily prescribe arbitrary coupling conditions. We tackle the resulting immense computational loads by using a block-based multiresolution (MR) algorithm and adaptive local time stepping. The level-set treatment is integrated into the MR algorithm with little overhead by employing a smart tagging system and adaptive storage of the fluid data in the MR nodes. We embed these methods in a C++20 object-oriented modular framework using state-of-the-art programming paradigms. Furthermore, our implementation is capable to exploit the multiple levels of parallelism in modern high-performance computing (HPC) systems efficiently. We demonstrate the capabilities of our framework by simulating a variety of compressible multi-phase flow problems. Problem-sizes are of O(1010) effective degree of freedom (DOFs). By the use of MR, we typically achieve memory and compute compressions of >90%. We demonstrate near-optimal parallel performance for scaling runs using O(104) cores, regardless of the employed numerical models. Program summaryProgram Title: ALPACA - Adaptive Level-set Parallel Code AlpacaCPC Library link to program files:https://doi.org/10.17632/5zr3sg83ct.1Developer's repository link:https://gitlab.lrz.de/nanoshock/ALPACALicensing provisions: GPLv3Programming language: C++20Supplementary material: Code: Copy of the git repository, Videos: Air-helium shock bubble interaction (front and back view), Air-R22 shock bubble interaction, Three bubble shock interface interaction.Nature of problem: Numerical simulation of conservation laws such as the compressible Navier-Stokes equation with several interacting gaseous and liquid phases remains challenging even today. These flows often involve singularities such as shocks and interfaces as well as instabilities driven by their interaction. The inherent highly nonlinear dynamics of those systems leads to a broad range of temporal and spatial scales that have to be resolved. There exists a variety of mutually exclusive numerical models that are able to tackle these challenges. Those, however, are computationally expensive and require computational power that is offered only by large-scale state-of-the-art distributed-memory machines.Solution method: We have developed a modular simulation environment for conservation laws allowing exchange of numerical methods without loss of parallel performance. Computational efficiency is enhanced by employing multi-resolution schemes with adaptive local-time stepping. Our block-based implementation of these schemes is parallelized using the Message Passing Interface and exploits vectorization capabilities of the compute hardware. To simulate distinct and immiscible phases inside the computational domain, we utilize a sharp-interface level-set method. The level-set method allows to accurately locate the interface position and to easily prescribe arbitrary coupling conditions. The narrow-band approach reduces the computational load of the level-set method. We extend this method by a smart tagging system that exploits the block-based nature of the algorithm and further reduces the computational load. The simulation framework is written in modern C++20 and provides a Python interface for integration in Machine Learning and Uncertainty Quantification toolchains. We use parallel HDF5 in combination with XDMF to output field quantities. CMake is used as build system. We have completely annotated the source code using doxygen-style comments, allowing automated documentation generation in different formats. The source code is equipped with CI/CD-automated unit tests and an exhaustive integration test suite.Additional comments including restrictions and unusual features: ALPACA relies on open-source third-party libraries for input and testing. These are packaged as git submodules and automatically integrated. ALPACA is tested on Linux and macOS.

Numerical and analytical study of laminar film condensation in upward and downward vapor flows

Laminar film condensation in upward and downward vapor flows is numerically investigated by using a sharp-interface level-set method to track the condensate film surface and accurately calculating the phase-change mass flux under the saturation temperature condition at the interface. An analytical model for steady laminar film condensation in upward as well as downward vapor flows is developed to validate the present numerical results. As the vapor velocity increases, the condensation rate is observed to decrease in upward vapor flows whereas it increases in downward vapor flows. The effects of vapor velocity and wall temperature on laminar film condensation in upward and downward vapor flows are investigated.

Level-set based numerical simulation of film condensation in a vertical downward channel flow

Numerical simulation of film condensation in a vertical downward channel flow is performed by employing the sharp-interface level-set method which can accurately implement the saturation temperature and phase-change mass flux conditions at the liquid-vapor interface. An analytical model for the internal film condensation is also developed by including the effect of vapor flow in the classical Nusselt model. The numerical results for film condensation in the laminar regime show good agreement with the analytical solutions. The level-set method is applied to investigate the effects of vapor velocity and wall temperature on the condensate film thickness and heat flux in the laminar and wavy regimes.

On comparison of the sharp-interface and diffuse-interface level set methods for 2D capillary or/and gravity induced flows

The present work is on comparison of numerical-methodology as well as performance of the two types of level set methods (LSMs): sharp-interface and diffused-interface. The numerical-methodology along with mathematical-formulation are presented in-detail for relatively recent ghost fluid method based sharp-interface level-set-method (SI-LSM); and compared with the methodology for the traditional diffuse-interface level-set-method (DI-LSM) as well as an improved diffuse-interface dual-grid-level-set-method (DI-DGLSM). Two different types of surface models are considered: continuum surface force (CSF) and sharp surface force (SSF) model; CSF model for the DI-LSM and SSF model for the SI-LSM. The SI-LSM considers the physically realistic sudden variation of the thermo-physical property across the sharp-interface while DI-LSM considers a smoothened value of the properties across the numerically diffused-interface. For solving the pressure Poisson equation in the SI-LSM, a finite volume method based generic formulation is proposed and its implementation-details are presented. For the SI-LSM as compared to DI-LSM and DI-DGLSM, the relative performance study is presented on four reasonably different two-phase problems: surface-tension model induced unphysical flow for a static water droplet, collapse of a water-column in air, falling of a water-droplet in air, and coalescence of an ethanol-droplet over a pool of ethanol in air. For the performance study on various problems, a comparison of the results obtained by the three types of LSMs (SI-LSM, DI-LSM and DI-DGLSM) and various other numerical methods in the literature are presented. SI-LSM as compared to DI-LSM and DI-DGLSM is shown to substantially reduce the unphysical spurious velocity and results in better accuracy on the same grid size.

A sharp-interface level-set method for compressible bubble growth with phase change

A sharp-interface level-set method is presented for simulating the growth and collapse of a compressible vapor bubble. The interface tracking method is extended to include the effects of bubble compressibility and liquid-vapor phase change by incorporating the ghost fluid method to efficiently implement the matching conditions of velocity, stress and temperature at the interface. The numerical results for one-dimensional compressible flows and spherical bubble growth show good agreement with the exact solutions. The level-set method is applied to investigate the effects of phase change, ambient temperature and wall on the compressible bubble growth and collapse.

Multiscale level-set method for accurate modeling of immiscible two-phase flow with deposited thin films on solid surfaces

We developed a multiscale sharp-interface level-set method for immiscible two-phase flow with a pre-existing thin film on solid surfaces. The lubrication approximation theory is used to model the thin-film equation efficiently. The incompressible Navier–Stokes, level-set, and thin-film evolution equations are coupled sequentially to capture the dynamics occurring at multiple length scales. The Hamilton–Jacobi level-set reinitialization is employed to construct the signed-distance function, which takes into account the deposited thin-film on the solid surface. The proposed multiscale method is validated and shown to match the augmented Young–Laplace equation for a static meniscus in a capillary tube. Viscous bending of the advancing interface over the precursor film is captured by the proposed level-set method and agrees with the Cox–Voinov theory. The advancing bubble surrounded by a wetting film inside a capillary tube is considered, and the predicted film thickness compares well with both theory and experiments. We also demonstrate that the multiscale level-set approach can model immiscible two-phase flow with a capillary number as low as 10−6.

Direct numerical simulation of 3D particle motion in an evaporating liquid film

A direct numerical simulation method is developed for 3D particle motion in liquid film evaporation. The liquid-gas and fluid-solid interfaces are tracked by a sharp-interface Level-set (LS) method, which includes the effects of evaporation, contact line and solid particles. The LS method is validated through simulation of the interaction between two particles falling in a single-phase fluid. The LS based DNS method is applied to computation of the particle motion in liquid film evaporation to investigate the particle-interface and particle-particle interactions.

Stationary contact line formation and particle deposition in dip coating

ABSTRACTThe stationary contact line formation and particle deposition in the evaporation regime of dip coating is investigated numerically by solving the conservation equations of mass, momentum, energy, vapor mass fraction, and particle concentration. A sharp-interface level-set method for tracking the liquid–gas interface is extended to include the effect of phase change and to treat the particle deposition as well as the liquid–gas–solid contact line on a moving substrate. The computations demonstrate that the particle deposition occurs spontaneously near the stationary contact line and the deposition thickness depends on the deposition rate constant and the substrate withdrawal velocity.

Numerical simulation of stick-slip contact line motion and particle line formation in dip coating

Numerical simulation is performed for stick-slip contact line motion and particle line formation in dip coating. The liquid–gas interface as well as the liquid–gas–solid contact line is tracked by a sharp-interface level-set method, which is extended to include the effect of phase change, to implement the contact angle model for the stick-slip contact line motion, and to treat the particle deposition on a moving substrate. The computations demonstrate that the particle line formation is directly related to the stick-slip contact line motion. The effects of initial particle concentration, substrate withdrawal velocity, substrate temperature and contact angle on the particle line formation in dip coating are investigated.

On sharp-interface level-set method for heat and/or mass transfer induced Stefan problem

A detailed sharp interface level set method (SI-LSM) based CFD development methodology as well as its implementation and application to the various types of Stefan problem is not found and reported here. The methodology and code-validation are presented in-detail for heat and/or mass transfer induced Stefan flow, 2D Cartesian as well as axi-symmetric coordinate system and Dirichlet as well as jump boundary conditions at the interface. For the Dirichlet BC, a higher order extrapolation methodology is presented – for the computation of ghost fluid value across the interface. Whereas, for the jump BC, a generic finite volume method based discretized equation is proposed – consisting of additional source term and modification in diffusion coefficient for the control-volumes adjoining the interface. An excellent agreement between the present and published results are shown separately for Stefan flow problems induced by heat transfer (melting as well as solidification), mass transfer (isothermal evaporation) and heat as well as mass transfer (non-isothermal evaporation). Finally, a performance study is presented demonstrating a superior performance of the present sharp as compared to diffused interface level set method. The performance study is also presented for two different methods of computation of normal gradient across the interface, with regard to numerical oscillations in the results.The present work – involving localized fluid flow and heat as well as mass transfer in the two-phase domain – will be extremely useful to CFD developer on multi-phase flow.

Numerical simulation of liquid film formation and evaporation in dip coating

Numerical simulation is performed for liquid film formation and evaporation on a moving vertical plate, which occurs in dip coating. The liquid–gas interface is tracked by a sharp-interface level-set (LS) method, which is modified to include the effect of phase change at the interface. Numerical techniques for the liquid–gas–solid contact line on a moving plate and the conservation of particle concentration are incorporated into the LS method. The flow and thermal characteristics associated with the dip coating are investigated by solving the conservation equations of mass, momentum, energy, vapor mass fraction, and particle concentration. The effects of plate velocity and temperature on the dip coating process are investigated.

A Level-Set Method for the Direct Numerical Simulation of Particle Motion in Droplet Evaporation

A sharp-interface level-set (LS) method is presented for direct numerical simulation (DNS) of particle motion in droplet evaporation. The LS formulation for liquid–gas flows is extended to liquid–gas–solid flows by treating the moving solid region as a high-viscosity fluid phase. The evaporation effect is accurately implemented by imposing the coupled temperature and vapor fraction conditions at the interface. The LS method is tested through computations of particle sedimentation in single-phase and two-phase fluids. The DNS of particle motion in droplet evaporation demonstrates the pinning phenomena of the liquid–gas–solid contact line.

Numerical simulation of droplet evaporation between two circular plates

Numerical simulation is performed for droplet evaporation between two circular plates. The flow and thermal characteristics of the droplet evaporation are numerically investigated by solving the conservation equations of mass, momentum, energy and mass fraction in the liquid and gas phases. The liquid-gas interface is tracked by a sharp-interface level-set method which is modified to include the effects of evaporation at the liquid-gas interface and contact angle hysteresis at the liquid-gas-solid contact line. An analytical model to predict the droplet evaporation is also developed by simplifying the mass and vapor fraction equations in the gas phase. The numerical results demonstrate that the 1-D analytical prediction is not applicable to the high rate evaporation process. The effects of plate gap and receding contact angle on the droplet evaporation are also quantified.