Complex Physics Modelling with OpenFOAM · 2018-10-02 · Complex Physics Modelling with OpenFOAM...

Complex Physics Modelling withOpenFOAM

Open Source CFD in Industry and Academia

Prof. Hrvoje Jasak

Wikki Ltd. United Kingdom

University of Zagreb, Croatia

Tsinghua University, Beijing, 17 September 2018

Complex Physics Modelling with OpenFOAM – p. 1

Outline

Objective

• Review the design modern numerical simulation / CFD software development and

capabilities

• Highlight applications of OpenFOAM in the automotive industry

• Review examples of electrochemistry simulations: PEM/SOFC fuel cells

Topics

1. Design and limitations of contemporary CFD software

2. Model representation through equation mimicking

3. Object orientation, generic programming and library design

4. Some practical applications: Automotive CFD

5. Modelling complex physics with OpenFOAM: detailed modelling of flow, heat

transfer and electrochemistry in fuel cells

6. Why consider OpenFOAM

7. Summary


Background

State of the Art in CFD Software Use

• Numerical modelling is a part of product design

◦ Improvements in computer performance: CPU speed and memory

◦ Improved physical modelling and numerics, covering more physics

◦ Sufficient validation and experience with modelling capabilities

• Two-fold requirements in CFD development

◦ Integration into the CAD-based design process

◦ Quick and reliable implementation of new and complex physical models

• Complex geometry support, high-performance computing, automatic meshing,

dynamic mesh capabilities etc. needed across the spectrum

• Opening new areas of CFD simulation

◦ Non-traditional physics: complex heat and mass transfer models,

electromagnetics, fluid-structure interaction, electrochemistry

◦ New solution techniques, eg. flow and shape optimisation; robust design,

uncertainty propagation, adjoint sensitivity, optimisation under uncertainty


Design of Modern Solvers

Physics and Numerics

• A single discretisation method (FVM, FEM), parallelism and vectorisation

• User-defined modifications inefficient and limiting: proprietary software

• Model-to-model interaction matrix is becoming increasingly complex

Software Organisation

• Functional approach: centralised data and multiple functions operating on it

• Monolithic implementation and integrated software: single executable for all cases

• A CFD software is a large project: order of 1-2 million lines of source code

• Complex solver-to-solver interaction or embedding virtually impossible: two

solvers, solver and mesh generator, embedding in a CAD environment

Consequences

• Difficulties in development, maintenance and support

• Some new simulation techniques cannot be accommodated at all: sensitivity

• In spite of the fact the all components are present, it is extremely difficult to

implement “non-traditional” models

Change of paradigm has started from approx. 2010


Object Orientation

Object-Oriented Software: Create a Language Suitable for the Problem

• Analysis of numerical simulation software through object orientation:

“Recognise main objects from the numerical modelling viewpoint”

• Objects consist of data they encapsulate and functions which operate on the data

Example: Sparse Matrix Class

• Data members: protected and managed

◦ Sparse addressing pattern (CR format, arrow format)

◦ Diagonal coefficients, off-diagonal coefficients

• Operations on matrices or data members: Public interface

◦ Matrix algebra operations: +,−, ∗, /,

◦ Matrix-vector product, transpose, triple product, under-relaxation

• Actual data layout and functionality is important only internally: efficiency

Example: Linear Equation Solver

• Operate on a system of linear equations [A][x] = [b] to obtain [x]

• It is irrelevant how the matrix was assembled or what shall be done with solution

• Ultimately, even the solver algorithm is not of interest: all we want is new x!

• Gauss-Seidel, AMG, direct solver: all answer to the same interface


Object Orientation

Implementation of Complex Physics Models

• Flexible handling of arbitrary equations sets is needed

• Natural language of continuum mechanics: partial differential equations

• Example: turbulence kinetic energy equation

∂k

∂t+∇•(uk)−∇•[(ν + νt)∇k] = νt

[

1

2(∇u+∇uT )

]2

−ǫo

kok

• Objective: Represent differential equations in their natural language

solve

(

fvm::ddt(k)

+ fvm::div(phi, k)

- fvm::laplacian(nu() + nut, k)

== nut*magSqr(symm(fvc::grad(U)))

- fvm::Sp(epsilon/k, k)

);

• Correspondence between the implementation and the original equation is clear


Object Orientation

Object Oriented Analysis

• Main object = operator, eg. time derivative, convection, diffusion, gradient

• How do we represent an operator in CFD?

◦ A field: explicit evaluation

◦ A matrix: implicit algorithm

• . . . but we need many other components to assemble the equation

◦ Representation of space and time: computational domain

◦ Scalars, vectors and tensors

◦ Field representation and field algebra

◦ Initial and boundary condition

◦ Systems of linear equations and solvers

◦ Discretisation methods

• The above does not complete the system

◦ Physical models, (eg. turbulence) and model-to-model interaction

◦ Pre- and post-processing and related utilities

◦ Top-level solvers


Object Orientation

Application Development in OpenFOAM

• Custom-written top-level solvers are written for each class of physics

• Solvers are optimised for efficiency and storage, re-using basic components

• Writing top-level code is very similar to manipulating the equations

• Ultimate user-coding capabilities: components can be re-used to handle most

problems in computational continuum mechanics

Layered Development

• Design encourages code re-use: developing shared tools

• Classes and functional components developed and tested in isolation

◦ Vectors, tensors and field algebra

◦ Mesh handling, refinement, mesh motion, topological changes

◦ Discretisation, boundary conditions

◦ Matrices and linear solver technology

◦ Physics by segment in library form

• Library level mesh, pre-, post- and data handling utilities

• Model-to-model interaction handled through common interfaces

• New components do not disturb existing code: fewer bugs


Object Orientation and User

How Does Object Orientation Impact the CFD Software User?

• The view above is relevant for software developer and model implementation

• . . . but user requirements are different

◦ Existing applications across CFD modelling areas

◦ Solver customisation and extension: “User Coding”

◦ Vertically integrated custom solver design: naval hydrodynamics,

turbomachinery, external aerodynamics, fuel cell design etc

◦ Automated execution, hiding of solver parameters

◦ Validation and verification

User-Specific CFD Solution Framework

1. Deploy ready-to-use OpenFOAM solvers to conventional applications:

aerodynamics, HVAC, marine hydrodynamics, turbomachinery, combustion

2. Based on existing components and building blocks to create a custom CFD solver

and work-flow: mesh generation, post processing and data analysis

3. For new complex physics CFD: develop and simulate new models:

eg. plasma physics in welding applications, catalytic chemistry, fluid-solid

interaction etc.


External Aerodynamics Simulations


External Aerodynamics Simulations

DrivAer Geometry: External Aerodynamics, Coupled Solver, 13.2M Cells


Automotive Applications

Modelling Diesel Particular Filters

• Steady-state compressible flow through thin porous layers

• Detailed 3-D meshing of channels expensive: mesh resolution requirements

• . . . but due to flow non-uniformity, channel-scale simulations cannot provide the

answer: not all channels are equally loaded and energy equation is solved globally

• Solution: multi-scale filter model

◦ Each channel is one cell thick: (1-D) simulation

◦ Porosity is a face property; flow friction is a volumetric sink

◦ Porous faces are assigned time-dependent filtration efficiency due to soot

deposition, affecting the flow and species distribution

• Automatic meshing tool for the monolith, unstructured mesh for inlet and outlet


Automotive Applications

Spray, Wall Film and Combustion Simulations in Internal Combustion Engines

• Complete simulation of spray injection, evaporation, wall film and combustion in a

GDI engine. Mesh motion and topological changes as shown before

• Basic flow solver with automatic mesh motion and topological changes

• Full suite of Diesel spray modelling using Lagrangian modelling framework

• Spray-wall interaction; liquid film modelling

• Simulation of pollutants: NOx, unburnt hydrocarbons

• Mesh sensitivity of spray penetration: solved with adaptive refinement


Internal Combustion Engine

Internal Combustion Engine Simulations

• Diesel combustion, 1/8 cylinder sector

• RANS, k − ǫ turbulence model, simplified

5-species chemistry and 1 reaction,

Chalmers PaSR combustion model

• Temperature on the cutting plane, spray

droplets coloured with temperature

Cold Flow with Valves

• Parallel valves

• Exhaust and intakestroke

Two-Stroke Engine

• Topological

changes: cell

layering and sliding

• Simulating exhaust

gas recirculation


Fuel Injection Flash Boiling

Flash-Boiling Flows

• The fundamental difference between flash boiling and cavitation is that the process

has a higher saturation pressure and temperature: higher density

• Enthalpy required for phase change is provided by inter−phase heat transfer

• Jakob number: ratio of sensible heat available to amount of energy required for

phase change

Ja =ρlcp∆T

ρvhfg

• Equilibrium models are successful for cavitation since Ja is large and timescale of

heat transfer is small. Flash boiling represents a finite rate heat transfer process:

Homogeneous Relaxation Model (HRM)

Dx

Dt=

x− x

Θ; Θ = Θ0ǫ

−0.54φ1.76

x is the quality (mass fraction), relaxing to the equilibrium x over a time scale Θ

• The timescale Θ is obtained from empirical relationship: Downar–Zapolski [1996].

ǫ is the void fraction and φ is the non−dimensional pressure.



Flash-Boiling Flows: Numerical Method

• Conservation of Mass∂ρ

∂t+∇ · (φvρ) = 0

• Conservation of Momentum

(∂ρU0)

∂t+∇ ·

(

φU0)

= −∇pn +∇ ·(

µ∇U0)

• Pressure Equation

1

ρ

∂ρ

∂p

∣

∣

∣

x,h

(

∂(ρpk+1)

∂t+∇ · (ρUpk+1)

)

+ ρ∇ ·φ∗− ρ∇

1

ap∇pk+1

+M(

pk)

+∂M

∂p

(

pk+1− pk

)

= 0

The HRM model term is denoted as M(= DxDt

). The superscripts k and k + 1 are

the corrector steps for the pressure equation.



Asymmetric Fuel Injector Nozzle-Design from Bosch GmbH.


Liquid Film Model

Finite Area Method: Simulating Surface Phenomena

• Finite Area Method: discretised equations on a curved surface in 3-D

• Surface is discretised using polygonal faces

• Discretisation takes into account surface curvature. A level of smoothness isassumed in calculation of curvature terms

Equation Set of a Thin Liquid Film Model

• Continuity equation

∫

Sw

∂h

∂tdS +

∮

∂Sw

m•v h dL =

∫

Sw

mS

ρLdS

n

m

m h

v

nmd,i

Sw


Liquid Film Model

Equation Set of a Thin Liquid Film Model

• Momentum equation

∫

Sw

∂hv

∂tdS +

∮

∂Sw

m•(hvv +C) dL

=1

ρL

∫

Sw

(τ fs − τw) dS +

∫

Sw

hgt dS −1

ρL

∫

Sw

h∇spL dS +1

ρL

∫

Sw

Sv dS

n

m

m

n

v

v

pL

gtgn

vd,i

md,i pL = pg + pd + pσ + ph

pd,i =ρ(vd,i)

2

n

2

pσ = −σ∇s •(∇sh)

ph = −ρLn•gh

Sv =∑

imd,i (vd,i)tdt dS

τfsh τw

Sw


Liquid Film Model

Model Validation: Flow Down an Inclined Plane with a Hump, Polygonal Mesh

0.005

0.01

0.015

0.02

0.025

0.03

-1 -0.5 0 0.5 1

Ave

rage

vel

octy

, m/s

z, m

Average velocity at outlet edge

Numerical solutionExact solution for plate


Liquid Film Model

Roof and Down-Pipe System Performance Analysis by Arup: London Aquatic Centre


Liquid Film Model



Liquid Film Model

Liquid Film Simulations: Avalanche Modelling and Vehicle Soiling: Rauter, 2017


Liquid Film Model

Volume-Surface-Lagrangian Simulation

• Main coupling challenge is to implement

all components side-by-side and control

their interaction

• Lagrangian tracking uses an ODE solver:

block coupling at matrix level is not

needed or cannot be used as before

• Close coupling is achieved by sub-cycling

or iterations over the block system for

each time-step

• If the model-to-model coupling fails,

options on improving the stability are

considerably limited

• Engine wall film simulation: courtesy of Po-

litecnico di Milano


Non-Linear Solid Mechanics

Two Decades of Solid Mechanics Development Bear Fruit

• Extensive development of FVM solid mechanics: non-linear structural problems

• Geometric non-linearity: large deformation

• Material non-linearity: plastic hardening, visco-elasto-plasticity

• Modelling of contact (linear and non-linear) with complex friction models

• New generation of mixed mode lubricated contact modelling


Fuel Cell Modelling

The MUSIC Project: A Multi-Scale Integrated Fuel Cell Project

• Fuel cells: potential source clean energy solutions. To build economically viable

and reliable fuel cells: need good models

• Proprietary software impedes progress: open source solution and shared

knowledge on physics and modelling

• MUSIC: An international collaboration to develop a better design tool to build

better fuel cells. Share the implementation, and its development, but hide the

actual fuel cell design

• Partners/financial support:

◦ Natural Science and Engineering Research Council, Canada

◦ Forschungszentrum Jülich GmbH, Germany

◦ Lund University, Sweden

◦ . . . and looking for further partners!

• Both SOFC and PEM cells considered; possibly include battery modelling

• Multi-scale modelling of physical processes

◦ Large (50-cell) stacks: rate equations in place of diffusion terms

◦ Single cells and small stacks: detailed CFD simulation

◦ Micro-scale simulations for effective properties and cell-level parameters


Fuel Cell Modelling


• Cell-level flow simulation: Júlich F-design SOFC Fuel cell


Fuel Cell Modelling


• Cell-level electrochemistry simulation: Júlich F-design SOFC Fuel cell


Fuel Cell Modelling

The MUSIC Project: Micro-Scale Simulations

• First used to compute effective diffusion and thermal conductivities in porous

media; currently used to evaluate detailed kinetics on micro-scale

• Geometry constructed either by dropping spheres or by detailed tomography

reconstruction of real materials

• Energy production occurs at the linear junction of triple-phase boundaries

• Local species values determined and mapped to line junction

• Electrochemical equations solved to determine local current production

• Linear current mapped back to adjacent cell volumes as source/sink terms


Fuel Cell Modelling

The MUSIC Project: Micro-Scale Work

• Results: ionic potential, electron potential, concentration of O2


Why Consider OpenFOAM

Why Consider OpenFOAM

• Open architecture

◦ Access to complete source: no secret modelling tricks, no cutting corners

◦ Both community-based and professional support available

◦ Common platform for new R&D projects: shipping results of research into the

hands of a customer with no delay

• Low-cost CFD

◦ No license cost, portable to any computing platform (IBM Blue Gene)

◦ Efficient on massively parallel computers, portable to new comms protocols

• Problem-independent numerics and discretisation

◦ Tackle non-standard continuum mechanics problem: looking beyond the

capabilities of commercial CFD

• Efficient environment for complex physics problems

◦ Tackling difficult physics is made easier through equation mimicking

◦ Utility-level tools readily available: parallelism, moving mesh

◦ Track record in non-linear and strongly coupled problems

◦ Excellent piece of C++ and software engineering! Decent piece of CFD


Summary

Summary

• OpenFOAM is a free software, available to all at no charge: GNU Public License

• Object-oriented approach facilitates model implementation

• Equation mimicking opens new grounds in Computational Continuum Mechanics

• Extensive capabilities already implemented; open design for easy customisation

• Solvers are validated in detail and match the efficiency of commercial codes

OpenFOAM in Research

• Open Source tools are ideal for a research environment: industrial partner gains

access not only to physical model equations but also to a validated implementation

• Deployment of results of research is faster and more reliable

• Uni Zagreb is the right partner: combination of in-depth knowledge of numerical

simulation tools, physical modelling and CFD methods

OpenFOAM in a Commercial Environment

• Deployment of Open Source CFD tools requires close collaboration: validation,

support, custom simulations, process integration and tool development

• Costs are reduced and simulation capability is expanded

• Wikki provides in-depth knowledge of methodology and industrial experience


About Me

Hrvoje Jasak

• First degree: mechanical engineering, University of Zagreb, Croatia 1988-1992

• PhD, Imperial College London 1993-1996

• Senior development engineer, CD-adapco (Siemens), 1996-2000

• Technical director, Nabla Ltd. 2000-2004

• Consultant on CFD software, numerics and modelling, ANSYS Fluent 2000-2008

Current Work

• Director, Wikki Ltd: UK-based CFD company 2004-

• Professor, University of Zagreb, Croatia 2007-

• Mercator Fellow, TU Darmstadt, 2016-

• Various software development and commercial support projects based on

OpenFOAM with consultants and large industrial partners

• Coordinating open source OpenFOAM development to allow contributions from the

public domain developers

• OpenFOAM workshops, lectures and seminars, visiting professorships (TU Delft,

Chalmers University, Uni Zaragoza, USM Chile and others)


Complex Physics Modelling with OpenFOAM · 2018-10-02 · Complex Physics Modelling with OpenFOAM...

Documents

Transcript of Complex Physics Modelling with OpenFOAM · 2018-10-02 · Complex Physics Modelling with OpenFOAM...