ICE Roadmap Japanese STAR Conference Richard Johns Introduction - PowerPoint PPT Presentation

ICE Roadmap Japanese STAR Conference Richard Johns

Introduction  Top-Level Roadmap  STAR-CCM+ and Internal Combustion Engines  Modeling Improvements and Research Support  Sprays  LES  Chemistry  Meshing  Summary

Top-Level Roadmap for In-Cylinder CFD Will b ill be available ailable and d maintai intained ned STAR STAR-CD CD for as lo long as is is requi quired red ICE Model dels Knowled wledge Experience perience Will b ill bec ecome ome the STAR STAR-CCM CCM+ suc uccessor sor in in-cylinde ylinder New w Meshing shing Tech chno nologies logies code de Model del and d Best st Prac actice ice Refinemen inements

Advantage of a Integrated STAR-CCM+ Solution Fuel el Inje ject ction ion and d Fuel el Sy System tem Optim imization ization Superc ercharg harging ing Intak ake/ e/Ex Exhaust haust System ems In In-Cylinder Cylinder 1D 1D-3D coup upling ling Af Aftertreatm rtreatmen ent Engine ine CHT T & St Struc uctural ural Analysis lysis Pis iston n undercrown dercrown cooling oling Cran ankcase/ kcase/oil oil splas lash/ h/ bear arings/br ings/breather eather sys ystem em

The Development and Adoption of Mathematical Models CD- adapco does not choose to be an “Inventor of Models”” • We see our tasks as: •  Model adopter  Implementation - Refinement, Generalization and Industrialization  Validation  Development of Recommended Usage & Best Practices  Dissemination  Benchmarking  Support  Further refinement in the light of experience

Engineering-Level vs High Fidelity Models Engineering ineering-Le Level vel spr pray ay-break breakup up model del Hig igh-Fidelit Fidelity y spr pray ay-break reakup up model odel Why do we support High-Fidelity Models? •  We learn a lot about fundamental physics  If we can’t solve the problem in a HF model we certainly won’t solve it in a EL model  Our objective is to deliver both HF and EL models

University Research & “Research Clubs” We actively support and fund a number of universities • around the world where there are strong ICE Research groups We support and engage with many others – the following • slides are not a comprehensive list – just a brief summary We also members of and/or participate in various research • “clubs”:  FVV, Germany Support of industry-initiated university projects   DERC – Direct-Injection Engine Research Consortium Univ of Wisconsin research club focused on  experimental and theoretical ICE fundamentals

University Research University of Modena, Italy Principal Investigator: Prof S Fontanesi Development and Application of LES and improved RANS models for flow, mixture distribution and knock University of Connecticut, US Principal Investigators: Professors Tianfeng Lu and Zhuyin Ren Chemistry mechanisms, solver speedup, reduced mechanisms, and mechanism reduction tools

University Research Penn State University, US Principal Investigator: Prof D Haworth Validation of in-cylinder Flow and Turbulence with experimental data, soot model development Doshisha University, Japan, Principal Investigator: Prof Senda Spray-Wall impingement modeling. fuel injection, spray heat & mass transfer, flash boiling

University Research University of Darmstadt, Germany Principal Investigator: Prof Janicka Validation of Flow and Turbulence with in-cylinder measurements Universi Un sity ty of Vien Vienna Principal Investigator: Prof Lauer Modeling of multicomponent fuels, wall effects, autoignition in gasoline engines

University Research Seoul National University (SNUAL) Principal Investigator: Prof Min Combustion and emissions modeling: flamelets, dual-fuel, level-set, knock ETH Zurich (Technical University) Principal Investigator: Dr Yuri Wright Combustion and emissions modeling: CMC, level-set, fuels, LES and DES in-cylinder flows

Fuel Injection and Sprays  Current injection systems and operating conditions can lead to complex spray phenomena Pin inj = 2 20 Mpa Pamb mb = 1 120 0 kPa Tin inj = 100 0 C Pamb mb = 4 40 kPa Ref: Parrish ish & Zink, , GM R R&D, Ilass 2008 2008

Non-flashing vs flashing analysis Non-flas lashing hing Fla lash shing ing Y is is th the mass ss frac action ion of n non-conde condensib nsible le gas as Courtesy of Prof David Schmidt, University of Massachusetts

Pintle nozzle – jet-string breakup Pin intle le Noz ozzle zle

LES/VOF calculation of jet-string breakup ATOMIC 5 deg sector

Spray Modeling Strategy  Support of HF and EL projects to understand better and improve physics modeling  Develop numerical and meshing technologies to support new Engineering Level models that can be embodied into in-cylinder calculations

In-Cylinder LES, mixture formation and Cyclic Variability  Ensemble or cycle-averaged CFD and measurements are an approximation to reality

In-Cylinder LES, mixture formation and Cyclic Variability  Small-scale information lost in the averaging process

LES – Intake take Flow low Struct ructure ure

Cy Cycled led-Averag veraged ed LES

LES – multicycle flame development

LES - 3D Results Insight:

Prediction of COV

In-Cylinder LES, mixture formation and Cyclic Variability  The Challenges:  Can we predict this successfully using High-Fidelity modeling?  Can we derive an Engineering-Level model from the lessons learned from a High-Fidelity model?  Are these modeling additions to our code or different ways of processing existing solutions?  Are there other flow features, such as instabilities, that we ignore at our peril?  Is there a first-mover advantage for an OEM in being an early- adopter of high-end technology?

Combustion Chemistry  Strengthened Team:  Graham Goldin  Karin Fröjd  DARS v2.10 release:  ECFM and PVM library generation incl dual-fuel  Extended Range Soot library  University Collaborations  Complex fuel chemistry (DME, biodiesel etc)  Soot modeling  Dual Fuel

Meshing Technology  Meshing is important!  Automatic o Efficiency & consistency in a production environment o Optimization and automated shape-change  Accurate o The ability to capture boundary details and small-scale phenomena and gradients, for example, sprays, spark ignition etc  Robust & Fast o Must be successful 99% of the time and add minimal overhead

Meshing Technology  We have/will have 3 meshing options: 1) Existing template/trimmed meshing – es-ice today, being developed for STAR-CCM+ 2) XMesher – based on multiple morphed meshes and solution mapping – STAR-CD and STAR-CCM+ 3) Overset Mesh – “Attachment” of individual meshes to components moving through a stationary background mesh – STAR-CCM+ only

Overset Mesh

XMesher – Automated Meshing for ICE  Developed over the past ~3 years  Stage 1:  Fully automatic, full-cycle meshing  Tested successfully on ~30 different real engine cases  Close comparison of results with es-ice meshing  Will be released in 2015  Stage 2  Local mesh refinement linked to specific events  Embedded spray-adapted meshing

XMesher Solution Strategy Prism Layers Solution mapped Constrained Polyhedra to next mesh Core Cartesian Mesh Aligned valve-seat mesh Mesh Mesh morphed morphed in - time in + time Mesh generated at this time Morphing Solution Continues Solution Mapping

XMesher

Illustration of XMesher process Initial Meshes are inserted automatically

Illustration of XMesher process Meshes are morphed with piston and valve motions, tested and then additional meshes inserted as required

Illustration of XMesher process This cycle is repeated until meshes for the entire calculation have been generated.

Performance  Concurrent meshing/morphing is used. Typical total time ~ ½ to 1 hour Tim ime (mins mins) 90 90 60 60 30 30 2 4 8 # cores res

Summary of Test Cases Case Total Mesh Number Base Mesh Max No. of Min No. Generation of Grids (mm) cells of cells time (mins) #1 2v gasoline 22 19 1 666945 498419 #2 4v gasoline 33 20 2 691707 485946 #3 4v diesel 60 23 2 1994121 1397606 #4 4v gasoline 42 16 1 751447 538127 #5 4v gasoline 37 17 2 1473225 777593 #6 4v gasoline 30 18 1 895494 720668 +24 Furt rthe her r Ca Cases

Cell Count through an engine cycle 1.5M 1.2M

Vector Field

XMesher – Stage 2  Objective is:  Physics-dependent local refinement  General spray-oriented meshing, including the possibility if meshing inside the injector if required  In prototype stage now

Physics-driven Locally-refined meshing

Physics-driven Locally-refined meshing

General Spray-Adapted Mesh

General Spray-Adapted Mesh