Numerical simulation of transient wall temperatures in diesel engines: Validation and potential analysis

Transcription

1 Numerical simulation of transient wall temperatures in diesel engines: Validation and potential analysis D. Mayer, Dr. F. Köpple, Matthias Mansbart Robert Bosch GmbH Dr. R. Fiereder, AVL Deutschland GmbH Prof. Dr. M. Bargende IVK, University of Stuttgart October 17 th 2018, Ulm

2 The presentation has been prepared and provided for informational purposes only and strictly on a non-reliance basis. The presentation does not include any legal advice. Information regarding legislative and other regulatory requirements should be checked carefully on an independent basis. 2

3 Agenda Motivation Introduction Complementary use of experiment and simulation Modeling approach for transient wall temperatures Excursion: Instantaneous Surface Temperature Measurements Validation: Comparison between experiment and simulation Potential analysis: Constant vs. transient wall temperatures Summary and outlook 3

4 Motivation

5 Average CO2 emissions (g/km) limit value curve 2021 limit value curve 2016 Motivation EU: CO2 Targets & OEM Status (2017) Monitoring of CO2 emissions from passenger cars new registration volume Average fleet mass 2016 based on [3] NEDC Split of Losses constant volume cycle Diesel 41.2% Alt. fuels 3.2% Petrol 56% Fleet average g/km brake efficiency friction calorics 12-14% Wall Heat Loss [2] g/km further CO 2 reduction required kg [1]:EEA Average mass (kg) piston 55% head 30% liner 15% Further CO2 reduction strategies for PC necessary to fulfill the future fleet CO2 limits. Wall heat losses are still a major limitation in thermal energy conversion 5 *based on 1D-CFD; 50% load

6 Introduction

7 ሶ Introduction Complementary use of experiment and simulation 7 Experiment provide data for validation & calibration of simulation models supply of boundary conditions Surface temperature single cyl. diesel test engine CR = 16:1 Displacement 0.5L External charging Pressure indication Mean flow motion Turbulence Unsteady Heat Flux Flame-wall interaction Crank Angle [CAD a.tdc] gas phase Derive and evaluate strategies for reducing wall heat losses T gas δ t 5 CAD 30 CAD w gas w=0 convection δ w ሶ Q = αa T gas T wall T wall 1D wall λ, ρ, c p heat conduction T ρc p t = λ 2 T x 2 + Q [4] [5] [6] [7] 3D-CFD 3D-Simulation quick and fast setting changes possible tempoal and spacial distributions plausiblity checks for experiments combustion simulation + copuling with transient heat conduction: Thin-wall y z x

8 Modeling approach for transient wall temperatures: 2 workflows

9 51.43 Modeling approach for transient wall temperatures Simplified CFD-Workflow sector mesh 1D-CFD-simulation injector pressure traces p cyl, p in, p exh injection rate BOSCH Virtual Injection (VIN) 3D-CFD-simulation nozzle noz.-file Mesh specifications: thermocouples 0.5mm 4-5 faces per thermocouple 0D/1D-CFD-simulation update T n once per cycle full cycle pressure trace analysis Thermal Solver 360 FEM-Modell p E, T E, w i T FEM,radial 3D-CFD-simulation high pressure stage sector mesh temporal temp. swing cell avg =0.3mm Model specifications Heat transfer Combustion Han-Reitz Σ= cells N layer =3 General Species: ERCmechanism + NOx N angular =69: 0.75 deg/cell 9 Temperature [K] y z x Thin-wall Turbulence Evaporation Spray model Breakup d Soot k-ζ-f model Abramzon Discrete droplet method Wave model Kinetic soot model

10 Modeling approach for transient wall temperatures Conjugate Heat Transfer 360 mesh 1D-CFD-simulation injector injection rate 3D-CFD-simulation nozzle Mesh specifications: A-A pressure traces p cyl, p in, p exh BOSCH Virtual Injection (VIN) noz.-file A A embedded structured grid 0D/1D-CFD-simulation full cycle p E, T E, w i 3D-CFD-simulation full cycle T fluid + HTC Model specifications Heat transfer Combustion Turbulence Han-Reitz General Species: ERCmechanism + NOx k-ζ-f model 360 mesh temporal temp. swing detailed FEM model at least 2 loops oil cooling simulation Evaporation Abramzon y z Spray model Discrete droplet method x Breakup Wave model T FEM,radial Soot Kinetic soot model 10 Thin-wall d [AVL Deutschland GmbH Dr. R. Fiereder]

11 Excursion: Instantaneous Surface Temperature Measurements

12 System 1 System 2 Excursion: Instantaneous Surface Temperature Measurements Data transmission Bottom View Continuous measurement via radio Rotor transmission antenna for B - B power supply Wireless power supply in spot mode 16 Bit A/D-converter Cylinder Head Unit 2 2 x 8-channel frequencymultiplexer Cylinder Liner Unit 1 TDC Electronic Unit sampling rate: 10 ks/s Analysis System Cylinder Block Piston BDC Stator Antenna Electronic Unit Electronic Unit Rotor antenna for power supply Evaluation Unit B B Receiving Antennas 1.8 MBps. 12 *telemetry system supported by Manner Sensortelemetrie GmbH Power Supply Crank Shaft [8] data radio transmission

13 Excursion: Instantaneous Surface Temperature Measurements Measurement technique Focus on two measurement sectors: Intake NiCr-Ni thermocouple Sector 1: IV swirl direction A #2 #3 #4 A #7 #6 #5 Focus on engine swirl direction Spray Axis IV Spray Targeting #8 2mm #7 2mm A - A #5 #6 Exhaust Thin film technique: 1 Glue Inconel 2 Ceramic Binder 3 Jacket 13 NiCr-wire MgOceramic MgOpowder Ni-wire Gold (0.15 µm) *thermocouples supplied by FKFS, University of Stuttgart Sector 2: EV Spray Axis #13 #10 B #16 #15 #12 #11 #9 Focus on spray axis direction. B EV B - B 4mm #11 Spray Targeting 4mm #9 #10 #8

14 Comparison Excursion: Instantaneous Surface Temperature Measurements Measurement strategy & heat flux calculation motored: motored: Surface Temperature Measurements motored: Clean Surface *Conditioning :Baseline *Re-Conditioning motored: if Baseline :Clean Reference :STM Soot Layer Correction Motored Post-Processing :x soot :motored meas. Surface temperature method (1D): average over 100 cycles t z zero crossing Tgas qሶ w x = 0, t = qሶ w,m + b meas i=1 iω 2 FFT FFT A i + B i cos iωt + B i A i sin iωt Corrected temperature traces will be used for comparison with calculation results. 14 *Conditioning O.P.: 2000rpm/2bar λ=5 damping and phase shift b meas = f(t wall ) [9] Evaluation of surface wall heat flux via Fast Fourier Transformation and exp. estimated thermal effusivity of TC. FFT: Fast Fourier Transformation

15 Comparison between experiment and simulation

16 combustion Δ18K steady temp. field Comparison between experiment and simulation Temperature swing methodology & analytical correction method Example #02 init: 482K compression CAD a. TDC initialization of thermal boundary condition T FEM,radial 1. Reuse of surface temperature method for corrected wall heat flux with thermal eff. of measurements : iω qሶ w x = 0, t = qሶ w,m + b meas A 2 i + B i cos iωt + B i A i sin iωt i=1 from measurement FFT FFT for calculated surface temperatures 2. Recalculation of surface temperatures Alternative: b Α = meas balsi Theoretically corrected values: Thin-wall parameters: d Tw 3mm n layer 30 compression 1.2 Initialization T FEM,radial 60% 60% In 3mm depth the temperature swing decays and a steady temperature field is gained. 16 [presentation AVL ISC 2017] Analytical correction method not used due to lower expected temperature swing and heat flux level. Raw data used! FFT: Fast Fourier Transformation

17 #16 #15 #12 #11 #10 #09 #07 #05 #06 AVG #14 cross section #8 #01 Mean Temperature [K] Comparison between experiment and simulation Mean temperatures: 2000rpm/8bar-Mi Simplified FEM-model (1D-Sim): EV IV Temperature [K] piston temperature section: overestimation Experimental Simplified FEM Detailed FEM Engine load: 8 bar MFB CAD a.tdc SOI -5.8 CAD Mi p intake 1586 mbar p rail 800 bar Speed 2000 min -1 Detailed FEM-model (AVL FIRE): EV IV EV IV EV IV 460 average #11 # mean surface temperatures component temperatures #11 #8 cross section Focusing on the thermocouples in direction of spray axis (system1 & system2). Characteristic temperature profile can be matched well in both methodologies. Weak overestimation at the bowl rim observable. 17 oil nozzle oil nozzle

18 Comparison between experiment and simulation Transient heat fluxes: 2000rpm/8bar, MI Experimental and numerical results: Engine load: 8 bar MFB CAD a.tdc SOI -5.8 CAD p intake 1586 mbar p rail 800 bar Speed 2000 min -1 Mi # CAD a. ITDC y-plus condition # CAD Sector-model: Full-model: 6.0 CAD a TDC 6.0 CAD 6.0 CAD 6.0 CAD Higher calculated wall heat fluxes in the central flame-wall impingement area compared to the measurement results. Obvious under-prediction of flame-wall contact in the squish area and at the bowl rim. 18

19 #12 #11 #07 #06 Mean Temperature [K] #12 #11 #07 #06 ΔT Combustion Comparison between experiment and simulation Transient wall temperatures: 2000rpm/8bar, MI Experimental and numerical results: Engine load: 8 bar MFB CAD a.tdc SOI -5.8 CAD p intake 1586 mbar p rail 800 bar Speed 2000 min -1 Mi Experimental Num: sector-model Num: full-model Experimental Simplified FEM Detailed FEM Same tendencies between sector-model and full-model observable. Temperature levels follow the trends captured by the FEM. But in general weaker resulting wall temperature swings compared to the measurement results. 19

20 Potential analysis: Constant vs. transient wall temperatures

21 Potential analysis: Constant vs. transient wall temperatures Influence on wall heat losses: 2000rpm/8bar, MI Tw=const. locally higher heatflux 15.0 CAD a. ITDC Mi 15.0 CAD 9.0 CAD temporal change Tw=trans. locally higher temperatures 15.0 CAD 9.0 CAD 9.0 CAD a. ITDC Wall Heat Distribution: Q w [J] % Liner Head Piston Tw=const. Tw=trans. Temporal change in wall temperatures influences the near-wall temperature distribution towards higher local temperature peaks. Influence on overall heat flux remains low due to the much higher overall temperature gradient between gas and the walls. 21

22 NOx emissions [%] Soot emissions [%] Potential analysis: Constant vs. transient wall temperatures Influence on emissions: 2000rpm/8bar, EGR-loop Tw=const. EGR= 0.1% Tw=trans EGR= 0.1% EGR= 0.1% normalized to max. values 15.0 CAD 15.0 CAD 15.0 CAD locally higher temperatures 15.0 CAD Exp. Sim.: Tw=const. Sim.: Tw=trans Exp. Sim.: Tw=const. Sim.: Tw=trans % 8.74% 20.67% EGR-RATE Higher near-wall temp. profile leads to higher local NOx-levels near piston surface and weaker soot formation in the piston bowl. Improved overall emission trend captured by CFD in combination with transient wall temperatures compared to engine results CAD 15.0 CAD 15.0 CAD 5.0 CAD a. TDC higher NOx 15.0 CAD 15.0 CAD lower soot 15.0 CAD 5.0 CAD

23 Summary and outlook

24 Summary and outlook 2 methodologies were shown to estimate the local wall temperature distribution and to calculate the temperature swing inside the piston walls. Using thin-walls -modeling in AVL FIRE can calculate the unsteady temperatures inside the walls. The resulting temperature swing seems to be under-predicted compared to the measurements. Weak difference observable between detailed FEM model via CHT and simplified FEM-model via separate cycle simulation. Higher calculated wall heat fluxes in the central flame-wall impingement area compared to the measurement results. The transient wall heat flux at the locations in the squish area and at the bowl bottom is underestimated. Use of transient surface temperatures leads to an change in the near-wall temperature profile as well as in local emission distribution (Soot/NOx). The results indicate improved emission trends compared to engine meas. Next steps: Focus on application-relevant operation points: Influence of injection pattern Influence of nozzle design influence of spray targeting Basis HE 0.75 mm HE k mm 24 Derive and evaluate strategies for reducing the wall heat losses.

25 Thank you.

26 Sources [1] Monitoring of CO2 emissions from passenger cars Regulation 443/2009 provided by European Environment Agency (EEA), 2016 [2] ACEA Report: Vehicles in use Europe 2017 [3] Heikes, H. ; Trzebiatowski, T.: Vergleich der innermotorischen und fahrzeugseitigen Verluste eines modernen Extrem-Downsizing-Ottomotors mit denen eines Downsizing- Dieselmotors auf Grundlage verschiedener Fahrzyklen. In: Wissenschaftssymposium Automobiltechnik, 2012, S [4] Bargende, M. : Ein Gleichungsansatz zu Berechnung der instationären Wandwärmeverluste im Hochdruckteil von Ottomotoren. Darmstadt, Technische Hochschule, Dissertation 1990 [5] Wimmer, Andreas. "Oberflächentemperaturaufnehmer zur experimentellen Bestimmung des instationären Wärmeüberganges in Verbrennungsmotoren," Ph.D. thesis, University of Graz, [6] Polifke, Wolfgang. Wärmeübertragung: Grundlagen, analytische und numerische Methoden. Pearson Deutschland GmbH, [7] Eiglmeier, Christian. Phänomenologische Modellbildung des gasseitigen Wandwärmeüberganges in Dieselmotoren. Diss. Dissertation, Universität Hannover, [8] Manner Sensortelemetrie GmbH, Instruction manual radio telemetry for piston usage,