Two-Stroke Reed FSI Modeling And Validation - CONVERGE CFD

Two-Stroke Reed FSI Modeling and ValidationPaulWesthoff, Jon Servias – BRP MPS David Rowinski - CSIConverge UGM 21051

Agenda1. Overview2. Converge CFD Model Setup3. Experimental Data4. CFD Results and Validation5. Future WorkConverge UGM 21052

Agenda1. Overview Background Motivation Approach2. Converge CFD Model3. Experimental Validation Data4. CFD Results5. Future WorkConverge UGM 21053

Background Work is Based on 2-stroke DI V6 outboard engine Crankcase scavenged with intake reed valves Reed motion responds pressure fluctuations whichcome from many sources in the system Gas flow path is very interactiveConverge UGM 21054

Motivation Accurate in-cylinder gas exchange and trapping predictions Current crankcase modeling techniques are dependent onexperimental data Opt 1: Pressure boundary condition to model filling of crankcase Opt 2: Reset crankcase pressure at assumed reed closure anddrive flow via piston motion Predictive air induction model in Converge that isinteractive with entire systemConverge UGM 21055

Approach Model the crankcase filling process Implement a reed motion UDF into Converge Run model at 3 WOT operating conditions (5500, 3000 and 2000 rpm) based on reed motion regime Obtain experimental validation data from fired engine Compare resultsConverge UGM 21056

Agenda1. Overview2. Converge CFD Model Setup UDF Resources3. Experimental Validation Data4. CFD Results5. Future WorkConverge UGM 21057

Model GeometryModel ExtentIntake PlenumSimplificationReed AssemblyModeling single cylinder of V6 engineConverge UGM 21058

Pressure Boundary ConditionsAmplitude significantrelative to reed dpPressure boundary conditions are still dependent on experimental dataCould couple to 1D gas dynamics code in futureConverge UGM 21059

Model SetupMoving ComponentsWith sealsFull model with moving reeds, piston, conrod, and crankshaftAll runs include spray and combustionConverge UGM 210510

Model SetupReed StopPetal tapers at rootEach petal moves independentlyConverge UGM 210511

Model SetupRegions upstream of each reedClose off based on min liftO.12 mm gap reed to blockTo minimize cell paringConverge UGM 210512

Mesh .5grid220.50.52.5grid340.50.54Grid3 used in this studyConverge UGM 210513

Mesh0.5mm at reedMeshing strategy 3 allowed for head room for AMRConverge UGM 210514

Model Resources Run time Run time 4 days/cycle Min number of cycles 3 12 days total run time Computer Resourses 32 cores (2 nodes x 16 cores) 3.3 GHz Xeon E5-2667v2 128 Gb memory per node 1 GbE interconnectSignificant run time for industry project useConverge UGM 210515

Reed Deflection Model Reed Motion is controlled by UDF Model is based on original work by G.P. Blair as adapted by Y. Zeng Reed is modeled as free vibration of cantilevered beamFlower plower dA 4 y 2 yEI 4 A 2 0 x txy ( x ) e i tFupper pupperdAyZeng, Y., Strauss, S., et al, Predicting and Optimizing Two-Stroke Engine Performance Using Multidimensional CFD, SAE 2004-32-0039Fleck, R., Blair, G. P., and Houston R. A. R., An Improved Model for Predicting Reed Valve Behavior in Two-Stoke Cycle Engine, SAE 871654Hinds, E. T., and Blair, G. P. , Unsteady Gas Flow through Reed Valve Induction Systems, SAE 780766Relatively simple model with proven resultsConverge UGM 210516

Reed Deflection Modely cosh i x cos i x sinh i l sin i l(sinh i x sin i x)cosh i l cos ilry i ( x) zi (t )i 1Damping factor – tunable variabled 2 zidzi 2 i2 zi Fii i2dtdt i ( i l ) 2EI Al portunities for additional tuningConverge UGM 210517

Agenda1. Overview2. Converge CFD Model3. Experimental Validation Data Reed lift Crankcase pressure4. CFD Results5. Future WorkConverge UGM 210518

Validation DataReed LiftCrankcase PressureStrain Gauged ReedsOptical MeasurementReed lift measured by two techniquesConverge UGM 210519

Strain Gauged Reed Petals – Reed Lift ValidationStrain gauge at root of petalEach reed petal calibrated for static tip deflectionHigher order bending modes give false readingConverge UGM 210520

Experimental Setup – High Speed Reed VideoMeasured Reed PetalOptical Access ReedsChallenging operating environment to make videoConverge UGM 210521

Reed Video – 5500 rpm1 opening and close event per cycleConverge UGM 210522

Reed Video – 3000 rpm2 opening per cycle – partial close betweenConverge UGM 210523

Reed Video – 2000 rpm3 open and close events per cycleConverge UGM 210524

Reed Tip Motion Tracking – 5500 rpmGood repeatability cycle to cycleConverge UGM 210525

Reed Tip Motion Tracking – 5500 rpmLow noise reed lift signalConverge UGM 210526

Reed Video from Intake SideStrain gauges on top reed petalsConverge UGM 210527

Crankcase Validation PressureDecreasing pressureDecreasing rpmCrankcase PressureMeasurement LocationLower CC pressure at lower rpm earlier reed openingConverge UGM 210528

Agenda1. Overview2. Converge CFD Model3. Experimental Validation Data4. CFD Results Comparison to experimental data CFD videos5. Future WorkConverge UGM 210529

Reed Lift Validation – 5500 rpmGood closing timingGood opening timingGood correlation in timing and magnitude to experimental dataConverge UGM 210530

Crankcase Pressure Validation – 5500 rpmGood correlation to experimental dataConverge UGM 210531

Reed DP and Lift – 5500 rpmCorresponds to openingReed motion responds to DP sign change but not every inflection in curveConverge UGM 210532

Reed Lift Validation – 3000 rpmNot Capturing ShapeEarly openingRelatively poor correlation with experimentCFD opening is early and missing partial closeConverge UGM 210533

Crankcase Pressure Validation – 3000 rpmRORCOver PredictingOver predicting peak crankcase pressureConverge UGM 210534

Reed DP and Lift – 3000 rpm DP early relative to openingReed DP is too early relative to experimental opening timeConverge UGM 210535

Reed Lift Damped and Undamped – 3000 rpmUndampedUndamped beam vibration model gave more pronounced double peakStill off in timingConverge UGM 210536

Reed Lift Validation - 2000 rpmGood closing timingGood opening timingCaptured 3 opening eventsLow in peak magnitudeConverge UGM 210537

Crankcase Pressure Validation – 2000 rpmRORCOver PredictingOver predicting peak crankcase pressureConverge UGM 210538

Reed DP and Lift – 2000 rpmCorresponds to openingCorresponds to closing startGood correlation between DP and –DP with opening and closingConverge UGM 210539

Plenum – Crankcase Press 2000 rpmPlenum – crankcase DPReed DP is much different than average plenum – crankcase pressureConverge UGM 210540

ATM vs Plenum BC – 2000 rpmATM BC closer to dataSensitivity to intake plenum pressure BC is modestConverge UGM 210541

Intake Tracer – 5500 rpmConverge UGM 210542

Pressure – 5500 rpmConverge UGM 210543

Velocity - 5500 rpmConverge UGM 210544

Pathlines – 5500 rpmConverge UGM 210545

Intake Tracer - 2000 rpmConverge UGM 210546

Pressure - 2000 rpmConverge UGM 210547

Velocity - 2000 rpmConverge UGM 210548

Pathlines – 2000 rpmConverge UGM 210549

Agenda1. Overview2. Converge CFD Model3. Experimental Validation Data4. CFD Results5. Future WorkConverge UGM 210550

Future Work Improve correlation at 3000 rpm Reed UDF tuning – damping factor, correction for neck down, reed stop Correlate to no plenum, no reed stop and motored data sets 1-D coupling to plenum, exhaust Reed design optimizationConverge UGM 210551

Thanks for Your AttentionConverge UGM 210552

Converge UGM 2105 2 Agenda 1. Overview 2. Converge CFD Model Setup 3. Experimental Data 4. CFD Results and Validation 5. Future Work