Development of A Clean Diesel Combustion System by Engine Testing and CFD-Simulation 1 1 1 2 3 3 3 3 J.Weber *, G. Thuir ,H. Schwab S. Saeki , G. Kotnik , K. Wieser , P. Gutmann , P. Matthis 1 DENSO Automotive Dtl. GmbH, Aachen Engineering Center, Germany 2 DENSO Cooperation, Japan 3 AVL List GmbH, Austria Future European market trends favor system solutions with low fuel consumption and low raw emissions to reduce the amount of exhaust gas aftertreatment. On this market, the challenge is to deliver a system concept and demonstrate its technical advantage in the competition. The optimization of the combustion system within the engine boundaries of engine friction, turbo-charger and gas exchange for low emis- sions, noise and fuel consumption with a target of high power density is complex. Hence the engine test- ing becomes time- and cost intensive even though state-of-the art tools as Design of Experiments and Model Based Calibration methods are applied. Therefore the optimization of the piston bowl design and selection of nozzle parameters e.g. spray cone angle, no. of holes is evaluated by the usage of CFD si- mulation. Although this combined approach of simulation and testing has limited prediction, the new com- bustion system achieves the emission targets with the given fuel consumption penalty. Introduction pressure system with DENSO’s Piezo injector The market demands for the next legislation G3P. limit of EU6 are quite challenging from system 2. Adaptation of the air path by an additional point of view. Low fuel consumption and low raw Low-Pressure-Loop (LPL) EGR system to demon- emissions are a necessity to get customer accep- strate EU6 emission levels tance from environmental and system cost point of 3. Change to a high performance turbo- view. The key to control the combustion process is charger with adaptation of the bowl change to the injection system to phase the combustion in achieve EU6 emissions with an increased power time and space and the air-path management for density. intake temperature and oxygen content control [1]. The design of the combustion chamber is the In this study, an existing 2.0l, 4 cylinder EU4 major focus in this study. LIEF measurements engine, CR=16, is used to demonstrate the capa- indicate that the spray of the G3P injector has a bilities of DENSO’s Engine Management System. leaner distribution than the baseline injector as The engine configuration can be viewed in Fig. 1. seen in Fig. 2. Moreover, the spray penetrates deeper into the piston bowl due to an increased rail pressure in comparison to the baseline. Both fea- tures have to be addressed by the design of the combustion bowl chamber. Thus a re-design of the bowl-chamber is necessary and will be supported by CFD simulations. The CFD code FIRE from AVL was used in this study. The spray model is the well known Discrete Droplet Model (DDM). Base- line engine data was used to calibrate the spray model parameters due to the limitations of this approach [3]. With regard to combustion, the ECFM-3Z model [4] is applied. In the following, a Fig. 1 Test-engine with replaced Engine Management new piston bowl was developed by a combination System (EMS) of CFD and engine testing. The objective of this study is to demonstrate EU6 emissions and to increase the power density from 55 to >60 kW/l by downsizing: two engine versions A and B are existing. The higher boost pressure of version B compared to A is beneficial to increase the engine power density [2]. Three steps have been applied to this base engine: 1. Change of the baseline series 1600 bar Fig. 2 Equivalence ratio distribution between baseline rd Piezo to 3 generation 2000 bar common rail series engine injector and G3P. * Corresponding author: J.Weber@denso-auto.de DENSO Automotive Dtl. GmbH, Aachen Engineering Center, Wegberg, Germany full load. Fig. 6 shows the Filter Smoke Number Development of a New Piston Bowl (FSN) at rated engine conditions for various nozzle An initial bowl design denoted as piston 1 was tip protrusions, hydraulic Flow Rates (HFR) and proposed based on CFD calculations as shown in no. of holes. The rated power is only limited by the Fig. 3. The potential for this bowl is indicated by a turbine temperature. Furthermore, the nozzle tip better soot oxidation among the baseline and other protrusion (NTP) was fixed to 2.9 mm, the no. of bowl proposals. holes to 8 and the HFR to 750 cm³/min which re- sults in a hole size diameter of 121µm. Overall, a power density of 63 kW/l can be achieved. 1.6 NTP 1.9, HFR 750, 8-hole 1.4 NTP 2.4, HFR 750, 8-hole NTP 2.9, HFR 750, 8-hole 1.2 NTP 2.4, HFR 800, 8-hole NTP 2.4, HFR 750, 7-hole 1.0 FSN [-] Fig. 3 Piston bowl designs 0.8 0.6 In a second step, nozzle parameters as spray 0.4 cone angle and no. of holes were optimised by 0.2 CFD to define a nozzle matrix since the spray-bowl 0.0 interaction is one parameter to control the soot 85 90 95 100 105 110 115 Effective Power [kW] 120 125 130 formation [5]. The CFD simulation predicts a better Fig. 6 Selection of NTP, HFR and no. of holes emissions performance of the new piston 1 bowl with an increased no. of holes from 6 to 8 and an For the high load emission points e.g. at engine increased spray cone angle from 150° to 159° as it speed of 2250 rpm and BMEP of 8 bar, only LPL- is seen in Fig. 4 and Fig. 5 at a higher part-load EGR was used. The NOx-soot trade-off is influ- emission Mode point (engine speed of 2250 rpm, enced by the cooling efficiency (Fig. 7). Increasing BMEP of 8 bar). the efficiency from 54% to 85% reduces tic from 65°C to 40°C. DOE optimum, 65deg downstr. Intercooler 40deg downstr. Intercooler Railpressure variation Target 9 8 7 Particulates 6 [g/h] 5 4 3 2 1 0 Fig. 4 Variation of no. of holes 270 260 [g/(kW*h)] BSFC 250 240 94 92 Noise [dB] 90 88 Fig. 5 Variation of spray cone angle 86 84 Engine Testing of Piston Bowl 1 0 5 10 15 20 25 30 35 40 45 The piston 1 design and nozzle samples were NOx Emission [g/h] manufactured and evaluated by engine testing. Fig. 7 Effects of intake charge cooling and rail pressure The testing procedure includes a calibration pro- variation on performance cedure in all emission mode points by Design of Experiments (DoE) and Model Based Calibration The effect of intake charge cooling can be (MBC) methods as well as manual calibration at viewed in the combustion analysis from Fig. 8. The heat release by the early double pilot injection is not changed but the ignition of the late main injec- tion is retarded. The premixed combustion is in- creased as the higher peak in ROHR indicates and less diffusion controlled combustion of rich areas occurs so that soot emissions are reduced. 100 90 67 °C Cylinder Pressure 80 70 40 °C 60 Fig. 10 Post-processed distribution of fuel vapor be- [bar] 50 40 tween bowl and squish area 30 20 10 The engine testing of piston 2 bowl 0 120 design indicated that the SCA of 159° has to be 100 [J/DegCA] 80 decreased to 155° due to an increase in soot ROHR 60 40 emissions. In order to address the penalty in noise 20 0 caused by decreased intake charge temperature a -50 -40 -30 -20 -10 0 Crank Angle [deg] 10 20 30 40 second option is to advance the pilot injections Fig. 8 Effect of intake charge cooling closer to the main (Fig. 11) which follows a more effective pilot combustion. A major challenge is to reduce the NOx-soot trade-off under a penalty in BSFC and noise. The retarded combustion shows a higher noise and lower soot level. If the rail pressure is additionally reduced, the soot emission benefit from the cooled intake charge is converted into a combustion noise benefit. Evaluation of Piston 2 Bowl Design In a second step, the piston 1 design was slightly changed to address the robustness sensi- tivity on injector production tolerances on the Fig. 11 Pilot timing effect on the noise model spray-bowl intersection and as shown in Fig. 3 and to improve the thermal robustness. A more effective pilot injection will shorten the The evaluation of piston 2 design included both, ignition delay of the main injection. Therefore less simulation and engine testing in a simultaneous time is available to homogenise the mixture and process. The CFD simulation of piston 2 bowl de- less premixed combustion will decrease the noise sign shows that the fuel vapor is pushed from the but vice versa more diffusive burning of rich mix- piston bowl into the squish area (Fig. 9). The mix- ture increases the soot emissions as it can be ob- ture in the bowl becomes leaner (Fig. 10) but air- served off-line from the MBC in Fig. 12. The soot excess is still available in the squish area. advantage of the decreased intake charge temper- ature can be changed into a noise benefit at a constant rail pressure level. Fig. 9 Comparison of equivalence ratio distribution be- Fig. 12 Pilot timing effect on the soot model tween piston 1 and 2 design The final engine performance is demonstrated in Fig. 13. The emissions as well as the BSFC tar- get can be achieved. An additional measurement showed that any further noise reduction would violate the BSFC penalty. This limitation is inherent to the system boundaries of the engine configura- system. Especially the performance of the turbo- tion. The high performance turbo-charger of engine charger has to be selected carefully. The higher B requires a higher back-pressure at the end of the specific power at rated conditions requires a higher expansion stroke compared to engine A and in- boost pressure but will violate the constraint in creases the pumping losses. BSFC on part-load conditions which is transferred into a violation of the constraint in noise level. A Additional measurement higher fun-to-drive pays back immediately by an Bowl 1, Calib. for 67deg, SCA 159deg acceptance of a higher noise level or by usage of a Bowl 2, Calib. for 40deg, SCA 159deg Bowl 2, Calib. for 40deg, SCA 155deg Target two-stage turbo-charger and increased system 9 8 cost. 7 Particulates Opt. System Bowl 2 Tailpipe (estimated) 6 Target Tailpipe (estimated) [g/h] 5 Baseline (bowl&TC) EU 6 Limit 4 30 3 Particulates raw 2 25 [mg/km] 1 20 0 15 275 10 BSFC [g/kWh] 270 5 265 0 260 3000 255 CO raw [mg/km] 2500 250 2000 245 92 1500 91 1000 90 500 89 Noise [dB] 0 88 87 5.1 Fuel Consumption 86 5.0 85 4.9 [l/100km] 84 4.8 0 5 10 15 20 25 30 35 40 45 4.7 NOx Emission [g/h] 4.6 Fig. 13 Engine performance of piston 2 bowl design 4.5 4.4 0 50 100 150 200 250 300 350 400 450 Vehicle emissions are estimated for a NEDC in NOx +HC [mg/km] Fig. 14 from four emission mode points. Overall, Fig. 14 Vehicle estimation EU6 emissions are achieved with the current con- figuration. If noise and BSFC shall be furthermore References reduced, the system configuration has to be changed. Either the low performance turbo-charger [1] L. M. Pickett and D. L. Siebers, “Non-Sooting, Low can be used if a lower power density is accepted Flame Temperature Mixing-Controlled DI Diesel or a two-stage turbo-charger with a better perfor- Combustion”, Paper No. SAE 2004-01-1399, 2004 mance at part-load conditions could be considered [2] S. Koidl, J. Hammer, „Challenges on Common Rail Diesel Injection Systems in Changing Surroun- but increase the system costs. th dings“, Engine Combustion Processes (8 Cong- ress), Berichte zur Energie- und Verfahrenstech- Summary and Conclusion nik, Munich, 2007 Engine development to meet new legislation [3] J. Weber, „Optimization Methods for the Mixture limits is to be considered as a system optimization Formation and Combustion Process in Diesel process within the given boundaries. This process Engines“, Ph.D. thesis, RWTH Aachen, 2008 was accomplished on a series production engine [4] O. Colin and A.. Benkenida, „The 3-Zones Ex- to demonstrate EU6 emissions with a high power tended Coherent Flame Model (ECFM3Z) for density target including the FIS and air- Computing Premixed / Diffusion Combustion”, Oil & Gas Science and Technology, Vol. 59, No. 6, pp. management system. 593-609 , 2004 The combined usage of CFD and engine testing [5] A. Weigand, F. Atzler, O. Kastner, T. Schulze, U. enables a pre-selection of nozzle parameters and Leuteritz, H. Zellbeck, A. Müller, D. Eckardt, „In- definition of bowl shape geometry which must be fluence of Vertical Spray Position on Diesel Com- adapted to the individual spray characteristics. bustion Process“, Engine Combustion Processes th Real engine testing is still mandatory and can- (8 Congress), Berichte zur Energie- und Verfah- not be omitted. The final calibration of the engine renstechnik, Munich, 2007 testing by DoE and MBC is including high rates of cooled EGR to shift the combustion towards lower temperatures and better homogenisation of the spray. The BSFC depends on the air-management