Sharp interface approaches and deep learning techniques for multiphase flows
Graphical abstract
Introduction
Multiphase flows are ubiquitous in the physical and life sciences and their simulations are used in a wide array of applications. Examples where simulations are key include the study of cavitation in the naval industry, the understanding of solidification processes of most metals, the analysis of boiling flows in nuclear plants, the discovery of novel surfactant and liquid delivery into the lung, the prediction of porous media flows for oil and gas operations, and even in the art, where simulations are used to produce believable flows in computer graphics.
The simulation of multiphase flows presents several challenges. First, one must keep track of the interface between phases; this interface evolves in time and can experience changes in topology. Second, the large variations in some key physical quantities, say the fluid's density, velocity or the pressure, can only be represented as jump conditions at the mesoscopic and macroscopic scales so that numerical methods must enforce such a representation. Third, the multiscale nature of multiphase flows, where fine features develop locally in some fluid regions, and the limitation of computational resources demand that numerical methods on adaptive grids and parallel architecture be considered.
Early work on the treatment of multiphase flows can be traced back to the work of the immersed boundary method of Peskin [103], [102]. Although multiphase flows per se were not considered, the use of a discrete δ-function to solve blood flows in an immersed elastic interface paved the way to extending to arbitrary domains the discretization of partial differential equations in rectangular domains. Unverdi et al. used this formulation in tandem with the front-tracking method to introduce a multiphase incompressible flow solver. In this case, the interface between phases was allowed to undergo large deformations including changes in topology. In [129] and [19], the authors used the level-set method to capture the interface motion instead of the front tracking approach, easing the treatment of topological changes. However, even if the front tracking and the level-set methods are sharp interface representations as opposed to phase-field models for example, which represent the interface between phase as a numerical mushy zone, the use of a δ-function formulation smears the physical quantities across the interface. In this work, we focus on ‘sharp’ interface methods, i.e. schemes that numerically preserve the discontinuity in discontinuous quantities.
In what follows, we consider a computational domain , where and are separated by an interface Γ. The fluid's motion is modeled by the incompressible Navier–Stokes equations: where t is the time, is the velocity field, p is the pressure and f includes the external forces such as gravity. We consider here the case of a fluid with uniform viscosity μ and uniform density ρ in each subdomain and . Appropriate boundary conditions are imposed on Γ and on the boundary of the computational domain, ∂Ω. These will be discussed in details in the remaining of this manuscript.
The framework we use to solve those equations is the standard projection approach introduced by Chorin in 1967 [20], which consists of the following three steps. Compute an intermediate velocity field using the momentum equation (1) and ignoring the pressure gradient term, e.g. in the case where the temporal derivative is discretized with a forward Euler step: The second step is based on the Helmoltz–Hodge decomposition that decomposes a twice continuously differentiable bounded vector field into a divergence-free component and a curl-free component ∇Φ: where Φ is referred to as the Hodge variable. Imposing the incompressibility constraint on gives an equation for Φ: The boundary conditions and how to impose them will be described throughout this manuscript.
In section 2, we review the level-set and the particle-level-set methods; in section 3, we discuss the level-set method on Quad-/Oc-tree Cartesian grids in parallel as well as how to solve free surface flows; in section 4, we discuss numerical methods related to solving multiphase flows; section 5 is focused on free surface flows and section 6 discusses Machine and Deep Learning opportunity for multiphase flows. A short conclusion is given in section 7.
Section snippets
Surface tracking vs. capturing
Explicit representation of the interface between phases can be achieved with the use of massless particles (see e.g. front tracking [42], [43], [57], [58], [135]). Since the particles' coordinates can be updated with high accuracy, explicit methods provide very accurate interface evolutions. On the other hand, changes in the topology needs to be handled explicitly as well, which adds to the complexity of those methods. Implicit representation treat changes in topology in a straightforward
Adaptive Cartesian Oc-/Quad-tree grids
Adaptive grids are desired in problems where different length scales occur. This is the case of fluids in general (turbulence requires a grid size proportional to , with Re the Reynolds number) and multiphase flows in particular since the interface between phases can stretch into thin features. An example is given in Fig. 2, where a fluid with pushes another fluid with higher viscosity, inducing viscous fingering.
Adaptive grids based on Quad-/Oc-tree data structures are particularly
Poisson solver with jumps
Solutions to multiphase flow problems admit jumps across the interface between phases. In particular, the pressure jump is related to surface tension forces. In the context of the projection method outlined in section 1, this boundary condition occurs when solving for the Hodge variable Φ, i.e. one needs to solve a problem of the form: where is taken to be constant in each subdomain but experiences a discontinuity across the interface. The
Free surface flows
Free surface flows are considered in applications where the effect of the momentum of the gas is negligible. In this case, the step that has the largest impact on the accuracy is the pressure boundary treatment at the free surface. This boundary condition is applied when solving for Φ in equation (5), i.e. one solves the system: and controls the effect of surface tension through the boundary condition . In [40], a second-order accurate discretization of the Poisson
The data explosion
Data is becoming plentiful. Over the past several years, the digital data collectively generated by humans has doubled every two years or less [35]. Some of this data comes from physical sensors, such as those on General Electric's locomotives that generate 150,000 data points each minute,1 or the various cameras and sensors in forthcoming self-driving cars, estimated by Intel to be 4 TB of data per car per day.2
Conclusion
We have presented a review of sharp interface methods as well as a discussion of machine learning and deep learning techniques for the simulation of multiphase flows. We have also discussed the recent advances in adaptive grid refinement techniques and their parallel extensions.
Acknowledgements
The research of Frederic Gibou was supported by ONR under MURI award N00014-17-1-2676 and by ARO W911NF-16-1-0136. The research of David Hyde and Ron Fedkiw was supported by ONR N00014-13-1-0346, ONR N00014-17-1-2174, ARL AHPCRC W911NF-07-0027 and NSF CNS1409847.
References (145)
- et al.
A monolithic mass tracking formulation for bubbles in incompressible flow
J. Comput. Phys.
(2013) - et al.
The fast construction of extension velocities in level set methods
J. Comput. Phys.
(1999) - et al.
Development of an exchange-correlation functional with uncertainty quantification capabilities for density functional theory
J. Comput. Phys.
(2016) - et al.
Quantifying uncertainties in first-principles alloy thermodynamics using cluster expansions
J. Comput. Phys.
(2016) Computational methods in Lagrangian and Eulerian hydrocodes
Comput. Methods Appl. Mech. Eng.
(1992)- et al.
A zonal Galerkin-free POD model for incompressible flows
J. Comput. Phys.
(2018) - et al.
A nonlinear manifold-based reduced order model for multiscale analysis of heterogeneous hyperelastic materials
J. Comput. Phys.
(2016) - et al.
A sparse grid based method for generative dimensionality reduction of high-dimensional data
J. Comput. Phys.
(2016) - et al.
Nonlinear Laplacian spectral analysis of Rayleigh–Bénard convection
J. Comput. Phys.
(2016) - et al.
A numerical method for two-phase flow consisting of separate compressible and incompressible regions
J. Comput. Phys.
(2001)
Eulerian capturing methods based on a level set formulation for incompressible fluid interfaces
J. Comput. Phys.
A numerical method for solving incompressible viscous flow problems
J. Comput. Phys.
A parallel fast sweeping method for the eikonal equation
J. Comput. Phys.
An adaptive discretization of incompressible flow using a multitude of moving Cartesian grids
J. Comput. Phys.
A hybrid particle level set method for improved interface capturing
J. Comput. Phys.
A fast and accurate semi-Lagrangian particle level set method
Comput. Struct.
Design and optimization of artificial neural networks for the modelling of superconducting magnets operation in tokamak fusion reactors
J. Comput. Phys.
A level set based sharp interface method for the multiphase incompressible Navier–Stokes equations with phase change
J. Comput. Phys.
A fourth order accurate discretization for the Laplace and heat equations on arbitrary domains, with applications to the Stefan problem
J. Comput. Phys.
A review of level-set methods and some recent applications
J. Comput. Phys.
A second-order-accurate symmetric discretization of the Poisson equation on irregular domains
J. Comput. Phys.
Efficient symmetric positive definite second-order accurate monolithic solver for fluid/solid interactions
J. Comput. Phys.
Numerically stable fluid-structure interactions between compressible flow and solid structures
J. Comput. Phys.
Solving elliptic problems with discontinuities on irregular domains – the Voronoi interface method
J. Comput. Phys.
A stable projection method for the incompressible Navier–Stokes equations on arbitrary geometries and adaptive quad/octrees
J. Comput. Phys.
Uniformly high order accurate essentially non-oscillatory schemes, III
J. Comput. Phys.
Two coupled particle-finite volume methods using Delaunay–Voronoi meshes for the approximation of Vlasov–Poisson and Vlasov–Maxwell equations
J. Comput. Phys.
Volume of fluid (VOF) method for the dynamics of free boundaries
J. Comput. Phys.
A front tracking method for dendritic solidification
J. Comput. Phys.
Computations of boiling flows
Int. J. Multiph. Flow
Data-driven reduced order models for effective yield strength and partitioning of strain in multiphase materials
J. Comput. Phys.
A general CFD framework for fault-resistant simulations based on multi-resolution information fusion
J. Comput. Phys.
A resilient and efficient CFD framework: statistical learning tools for multi-fidelity and heterogeneous information fusion
J. Comput. Phys.
An unconditionally stable fully conservative semi-Lagrangian method
J. Comput. Phys.
On two-phase flow solvers in irregular domains with contact line
J. Comput. Phys.
A data-driven adaptive Reynolds-averaged Navier–Stokes k–ω model for turbulent flow
J. Comput. Phys.
Machine learning strategies for systems with invariance properties
J. Comput. Phys.
Efficient simulation of surface tension-dominated flows through enhanced interface geometry interrogation
J. Comput. Phys.
A boundary condition capturing method for Poisson's equation on irregular domains
J. Comput. Phys.
Weighted essentially non-oscillatory schemes
J. Comput. Phys.
Spatially adaptive techniques for level set methods and incompressible flow
Comput. Fluids
A review of predictive nonlinear theories for multiscale modeling of heterogeneous materials
J. Comput. Phys.
Motion of multiple junctions; a level set approach
J. Comput. Phys.
Flow regime identification methodology with neural networks and two-phase flow models
Nucl. Eng. Des.
Neural network modeling for near wall turbulent flow
J. Comput. Phys.
A second order accurate projection method for the incompressible Navier–Stokes equations on non-graded adaptive grids
J. Comput. Phys.
A second order accurate level set method on non-graded adaptive Cartesian grids
J. Comput. Phys.
A supra-convergent finite difference scheme for the variable coefficient Poisson equation on non-graded grids
J. Comput. Phys.
Parallel level-set methods on adaptive tree-based grids
J. Comput. Phys.
Level set methods: an overview and some recent results
J. Comput. Phys.
Cited by (66)
Physics-constrained neural networks as multi-material Riemann solvers for compressible two-gas simulations
2024, Journal of Computational ScienceNumerical modeling of liquid spills from the damaged container and collision of two rising bubbles in partially filled enclosure using modified Volume-Of-Fluid (VOF) method
2023, Engineering Analysis with Boundary ElementsMachine learning algorithms for three-dimensional mean-curvature computation in the level-set method
2023, Journal of Computational PhysicsA projection-based, semi-implicit time-stepping approach for the Cahn-Hilliard Navier-Stokes equations on adaptive octree meshes
2023, Journal of Computational PhysicsA finite element method for partially erodible bed evolution coupled with multiphase flows
2023, Computer Methods in Applied Mechanics and EngineeringA shallow Ritz method for elliptic problems with singular sources
2022, Journal of Computational PhysicsCitation Excerpt :Recently, Egan and Gibou [9] have developed a novel idea to extend the original GFM by recovering the convergence of the gradient so that second-order accuracy can be achieved without modifying the resultant linear system. Along with this GFM approach, many other numerical methods for solving elliptic problems with interfaces or in irregular domains [11,13,8,2,10,3] have been successfully developed to improve the overall accuracy in maximum norm. One should mention that the linear systems resulting from those above methods are often symmetric positive definite, which can be solved efficiently by using iterative methods.