Combustion Analysis in PCCI Diesel Engines by Endoscopic and Pressure-Based Techniques

Combustion Analysis in PCCI Diesel Engines by Endoscopic and Pressure-Based Techniques A.E Catania 1, E. Spessa 1, G. Cipolla 2, A. Vassallo 2 1. IC Engines Advanced Laboratory Politecnico di Torino 2. General Motors Powertrain Europe 1. Abstract Endoscopic and pressure-based techniques were applied to the combustion diagnostics in a PCCI (Premixed Charge Compression Ignition) diesel engine featuring a low compression ratio (15.5:1). The pressure-based technique is based on an innovative premixed-diffusive multizone approach for heat release and emission formation analysis developed at Politecnico di Torino (PT). The combustion chamber is split into homogeneous zones (liquid fuel, unburned-gas/vapor-fuel rich mixture, unburned gas, premixed burned gas, diffusive burned gas zones), to which mass and energy conservation equations are applied. The diagnostic tool includes submodels for estimating NO, CO and PM formation. The two approaches were compared at a single engine operating point for different EGR rates. Endoscopic and pressure-based technique detected virtually the same combustion temperatures and similar trends of PM emissions. However, in addition to the endoscopic approach and thanks to the multizone diagnostic tool capability to take into account the burned gas expansion, NO x and CO levels were also evaluated, whereas PM trace could be extended to crank-angles at which extremely low combustion luminosity does not provide any more possibility for optical soot detection. 2. Engine specification and experimental setup The baseline engine is derived from the GM Powertrain 1.9l 4-cylinder in-line 4-valves-percylinder EU4 engine, whose features are listed in Table 1. A new combustion bowl prototype was manufactured ([1]) so as to obtain a compression ratio (CR) target of 15.5:1. The bowl prototype features a central-dome shape and is characterized by both a high K-factor for improving the air utilization at full load and a lower aspect ratio in order to tolerate advanced injection timings at partial load. The choice of the target compression ratio value has been done on the basis of previous investigations ([1]) looking at mid-term post EU 5 timeframe that assessed the possibility to still have a robust lowtemperature combustion ignition with such a CR value. Furthermore, the analyzed engine has been fitted with a glowing system with high protrusion metallic glow-plugs that offer an increased contact surface for igniting the diesel spray. All of the experimental tests on the reference engine as well as on piston and injector prototypes are carried out on a new AVL highdynamic test bed (Fig. 1) at ICE Advanced Laboratory of Politecnico di Torino. The test rig is equipped with an ELIN AVL APA 1 cradle-mounted AC dynamometer, capable of Engine type 1.9l EURO4 Displacement 191 cm 3 Bore x stroke 82.mm x 9.4mm Stroke-to-bore ratio 1.1 Compression ratio 17.5:1 Valves per cylinder 4 Turbocharger Fuel injection system Maximum power and torque Specific power and torque Single-stage with VGT Common Rail 2 nd gen. CRI2.2 16bar 11kW @ 4rpm 32Nm @ 2rpm 57.6kW 167Nm/l Table 1: Main engine specifications IV-6, 1

31st Meeting on Combustion realizing full four-quadrant operation with high speed and torque dynamics, including simulation of zero torque and gear shifting oscillations in the drivetrain. The main measuring systems includes the AVL KMA 4 for continuously metering the engine fuel consumption, the Pierburg AVL AMA 4 raw exhaust-gas analyzer (2 trains for THC, NO x, CO, CO 2, O 2 measurements upstream of and downstream form DOC; 1 train for CO 2 measurement in inlet manifold), and the heated AVL 715S smokemeter. A high-frequency KISTLER 653 CCSP piezo-electric transducer is installed on the engine cylinder head for taking the pressure time-history of Fig. 1: AVL high-dynamic test bed the gases in the combustion chamber of cylinder 4, whereas a highfrequency KISTLER 475A1 piezoresistive transducer is used to detect pressure level in the inlet runner of cylinder 4 for referencing in-cylinder pressure. An AVL 365C crank-shaft driven encoder generates the time base for an automatic 14-bit data-acquisition system based on AVL IndiModul 62 system, which can acquire up to 8 channels with a maximum frequency of 8 khz per channel. The engine was equipped with AVL Visioscope system for fuel spray and combustion imaging. 3. PT Multizone Premixed-Diffusion Combustion Model The cylinder content is split into homogenous zones. Between the injection (SOI) and combustion start (SOC), three zones are present: the liquid fuel zone containing the injected fuel (designated by the index f,l), the unburned gas zone (u), filled with a mixture of fresh air, EGR and residual gas, and the mixture zone (m), i.e., a rich mixture of fuel vapor and unburned gas. The model accounts for the fuel evaporation and transfer from liquid zone to mixture zone. At SOC an additional zone (bp) is formed, made up of burned gas from the premixed combustion in the mixture zone. The premixed combustion products complete their oxidation through a diffusion flame around the jet periphery, where the required oxygen is available. At the diffusive combustion start (SOC d ), a diffusive burned gas region is generated, which in turn is split into zones. These are generated at specific crank angles (), which are selected so as to divide the fuel mass that undergoes diffusive burning into a userdefined suitable number of equal portions. The last-generated zone (bd,n) is fed by the premixed burned-gas zone and by the unburned-gas zone, so as to reach stoichiometric conditions, whereas the other diffusive zones (bd,j) evolve at constant mass. From SOC d onward, combustion proceeds as a two-stage quasi-steady process, i.e., for each heat release is due to both premixed and diffusive contributions. The model performs measured in-cylinder pressure heat-release analysis applying the conservation principles of thermodynamics to each zone, the perfect gas law to each gaseous zone and a proper fuel evaporation model to the liquid zone. The new pressure-based combustion model is thoroughly described in [1]. The computed thermodynamic and thermochemical properties in the burned gas zones allowed the post-processing analysis of nitric IV-6, 2

Italian Section of the Combustion Institute Fig. 2: AVL VisioScope setup and schematics of in-cylinder angle of view. oxide (NO), particulate matter (PM) and carbon monoxide (CO) formation. 4. AVL Endoscopic technique The AVL VisioScope (Fig. 2) system has been applied for combustion analysis in the combustion chamber of the GMPT-E multicylinder engine. It features a cooled endoscope (Fig. 2a) which provides optical access to the interior area of the engine through the glowplug seat (Fig. 2b). A digital CCD camera transfers digital data straight to a PC, whereas synchronization with the engine is achieved via an AVL 365C encoder at a resolution of 1 crank-angle (CA) deg. A strobe connected to the Light Unit is used to facilitate the correction angle adjustment. In addition to fuel spray and combustion imaging, the system has been applied to the determination of instantaneous combustion temperatures and soot levels of the diesel flames based on the spectral flame temperature measurement technique using the AVL ThermoVision two colour method (for its detailed explanation, please refer to [2]). 5. Results and discussion Figures 3 and 4 show the images taken with AVL VisioScope at the four different CA indicated on the correspondent in-cylinder pressure ( Pressure, black solid line) and HRR time-histories reported to the bottom-right of each figure. The engine was operated at N = 15 rpm and bmep = 2 bar. Fig. 3 and 4 refers to EGR rate = % and 32.5%, respectively. HRR distributions were worked out by means of the PT multizone premixed-diffusion combustion model and show the contribution of the premixed burning ( HRR, Premixed, dash-dotted line) in addition to the global heat release rate time history ( HRR, red solid line). The four images in each figure show the pilot pulse (1), the main pulse (2) and the flame images for the crank angle at which maximum HRR (3) and maximum luminous emissions (4) occur. Figure 5 shows the diesel flame temperature (Fig. 5a) and the soot concentration (Fig. 5b) maps obtained by AVL ThermoVision two-colour method at the CA of Fig. 4.4. For temperature images, a frequency distribution of the temperature values in the image region was also generated for each CA. At each CA the temperature values of an image are sorted and divided into 1 ranges (j=1 1) and for each range the mean temperature values T j are worked out so that T 1 (CA) represents the mean of the highest 1% of the temperature values, T 2 (CA) the mean of the next highest 1% of the temperature values and T 1 (CA) the mean of the lowest 1% of the temperature values. Fig. 6 plots the time-histories of each T j (labeled as TV j in the legend) for the test case at EGR rate = 32.5% along with the correspondent ensemble-averaged mean temperature T m (labeled as TV Tm ), which is a characteristic temperature of the diffusive flame and of the sooting regions encircled by the IV-6, 3

31st Meeting on Combustion Fig. 3: VisioScope flame images along with correspondent experimental in-cylinder pressure and multizone HRR time-histories. N = 15 rpm, bmep = 2 bar, EGR = %. Fig. 4: VisioScope flame images along with correspondent experimental in-cylinder pressure and multizone HRR time-histories. N = 15 rpm, bmep = 2 bar, EGR = 32.5%. diffusive flame. Figure 7 shows a comparison amongst ThermoVision T m (labeled as TV Tm in Fig. 7) and burned gas temperatures of premixed (T bp : labeled as MZ, Premixed in Fig. 7) and diffusive (T bd,j : labeled as MZ j in Fig. 7) zones evaluated by means of multizone premixed-diffusion IV-6, 4

T [K] Italian Section of the Combustion Institute Fig. 5: Diesel flame temperature and soot concentration maps obtained by AVL ThermoVision two-colour method. N = 15 rpm, bmep = 2 bar, EGR = 32.5%. CA = 373 deg (Fig. 4.4). combustion model for the test case with EGR = % (Fig. 7a) and 32.5% (Fig. 7b). Multizone diagnostic model generates each zone bd,j at the temperature of diffusive flame, therefore T bd,j peaks are virtually equal to ThermoVision T m at the corresponding CA. Then, as CA increases the temporal evolution of each zone bd,j takes the burned gas expansion into account, thus T bd,j decreases with respect to ThermoVision T m. Figure 8 compares the soot levels worked out by multizone program ( MZ Soot [mg] ) to the corresponding trends evaluated by AVL ThermoVision ( TV Soot [a.u.]) for the test cases at EGR = % (Fig. 8a) and at EGR = 32.5% (Fig. 8b). The global trends evaluated by the two methods show a good agreement. It is worthwhile pointing out that no quantitative statement can be made about the correctness of the soot levels worked out by the two colour method, therefore TV Soot is expressed in arbitrary units (a.u.). On the other hand, based on a specifically developed calibration procedure ([1]) multizone approach is able to calculate quantitative soot levels by coupling Hiroyasu soot formation model and Nagle and Strickland-Constable soot oxidation model to the computed thermodynamic and thermochemical properties in the burned gas zones. Such quantitative soot levels are in line with the experimental outcomes, as is supported by Fig. 9 where experimental and multizonecomputed global soot levels are compared as a function of EGR rate at N = 15 rpm and bmep = 2 bar. Figures 1 and 11 reports the NO (Fig. 1) and CO (Fig. 11) time-histories calculated with the multizone diagnostic tool for the EGR = % (Figs. 1a, 11a) and EGR = 3 28 26 24 22 TV 1 TV 2 TV 3 TV 4 TV 5 TV 6 TV 7 TV 8 TV 9 TV 1 TV Tm 2 35 36 37 38 39 4 41 Fig. 6: Frequency distribution and mean value of temperature values at: N = 15 rpm, bmep = 2 bar, EGR = 32.5%. IV-6, 5

T [K] T [K] 31st Meeting on Combustion TV Tm MZ 1 MZ 2 MZ 3 MZ 4 MZ 5 MZ 6 MZ 7 MZ, Premixed HRR, Premixed HRR 3 5 3 5 24 18 12 6 4 3 2 1 HRR [J/deg] 24 18 12 6 4 3 2 1 HRR [J/deg] 35 36 37 38 39 4 41 32.5% (Figs. 1b, 11b) test cases. NO and CO time-histories are determined by applying the SEZM (Super Extended Zeldovich Model) and Bowman submodels, respectively, to the burned gas zones according to the procedure detailed in [1]. The thin lines refer to the NO and CO formation and decom position in diffusive ( Zone j ) and premixed ( Premixed ) zones, whereas the thick line indicates the resulting global NO and CO levels in the combustion chamber throughout the engine cycle. The black arrow to the righthand side indicates the value at exhaust valve opening worked out based on NO and CO engine-out measurements. Fig. 12 compares experimental and multizonecomputed global NO (Fig. 12a) and CO (Fig. 12b) levels as a function of EGR rate at N = 15 rpm and bmep = 2 bar. An excellent agreement between experimental and diagnostic-based results is apparent, thus supporting the capability of multizone approach to capture the trends and the levels of CO and NO emissions, with a generally SOOT [mg/kwh] 35 36 37 38 39 4 41 Fig. 7: Comparison amongst ThermoVision flame temperature (TV Tm) and PT multizone burned gas temperatures of premixed and diffusive regions for EGR = % and 32.5%. N = 15 rpm, bmep = 2 bar. TV, SOOT [a.u.] 2 16 12 8 4 TV, SOOT [a.u.] MZ, SOOT [mg] 35 36 37 38 39 4 41.1.8.6.4.2 MZ, SOOT [mg] TV, SOOT [a.u.] 2 16 12 8 4 35 36 37 38 39 4 41 35 3 25 TV, SOOT [a.u.] MZ, SOOT [mg] Fig. 8: Comparison amongst ThermoVision (TV Soot) and PT multizone (MZ Soot) soot levels for EGR = % and 32.5%. N = 15 rpm, bmep = 2 bar. SOOT SOOT exp.1.8.6.4.2 MZ, SOOT [mg] 2 1 2 3 4 5 EGR [%] Fig. 9: Experimental and multizone-computed global soot levels at different EGR rates. N = 15 rpm, bmep = 2 bar. IV-6, 6

NOx [ppm] CO [ppm] CO [ppm] CO [ppm] NOx [ppm] NOx [ppm] 18 15 12 9 6 3 zone 1 zone 2 zone 3 zone 4 zone 5 zone 6 zone 7 Premixed Global 35 36 37 38 39 4 41 Italian Section of the Combustion Institute 195 ppm 1 8 6 4 2 zone 1 zone 2 zone 3 zone 4 zone 5 zone 6 zone 7 Premixed Global 35 36 37 38 39 4 41 Fig. 1: NO time-histories calculated with the multizone diagnostic tool for the EGR = % and EGR = 32.5% test cases. N = 15 rpm, bmep = 2 bar. 12 1 8 6 4 2 zone 1 zone 2 zone 3 zone 4 zone 5 zone 6 zone 7 Premixed Global 35 36 37 38 39 4 41 713 ppm 1 8 6 4 2 zone 1 zone 2 zone 3 zone 4 zone 5 zone 6 zone 7 Premixed Global 35 36 37 38 39 4 41 Fig. 11: CO time-histories calculated with the multizone diagnostic tool for the EGR = % and EGR = 32.5% test cases. N = 15 rpm, bmep = 2 bar. 2 1 15 NOx NOx exp 9 8 CO CO exp 65 ppm 954 ppm 1 7 5 1 2 3 4 5 EGR [%] satisfactory accuracy degree at a very low computational cost. 6 1 2 3 4 5 EGR [%] Fig. 12: Experimental and multizone-computed global NO and CO levels at different EGR rates. N = 15 rpm, bmep = 2 bar. 6. Conclusions Pressure-based and endoscopic techniques showed to be powerful means to investigate cause and effect relations between injection rate, heat release, in-cylinder temperatures and pollutant emissions. In particular, AVL VisioScope system are very effective for spray and combustion imaging when a comparison amongst different hardware configuration (intake ports, injector nozzle, piston bowl, ) is required. By applying AVL ThermoVision to endoscopic measurement, diesel flame temperature and soot levels distributions can also be calculated. IV-6, 7

31st Meeting on Combustion PT multizone premixed-diffusion combustion model can be used to separate the premixed and diffusive burning effects on combustion parameters and emissions, based on in-cylinder pressure measurements. A very good accuracy degree can be achieved at a very low computational cost and by using the measuring instruments that are usually available for base engine development activities. The flame temperatures calculated by AVL VisioScope/ThermoVision and PT multizone diagnostic models shows a very good agreement. However, PT approach is also capable of evaluating burned gas temperatures during the expansion, thus allowing quantitative evaluation of NO and CO levels. With reference to soot level, both AVL VisioScope/ThermoVision and PT multizone techniques give virtually the same trends during combustion. However, soot levels worked out by the two-colour method are only expressed in arbitrary units, whereas PT multizone technique can calculate quantitative soot levels, which are in line with the experimental outcomes. 7. References 1. Baratta, M., Catania, A.E., Ferrari, A., Finesso, R., Spessa, E.: Innovative Multizone Premixed-Diffusion Combustion Model for Performance and Emission Analysis in Conventional and PCCI Diesel Engines Comodia 28, Japan, July, (28). 2. AVL Product Guide AT165E, Thermovision advanced, (24). IV-6, 8