International journal of automotive technology, tập 11, số 3, 2010

Copyright © 2010 KSAE 1229−9138/2010/052−01 International Journal of Automotive Technology, Vol 11, No 3, pp 289−306 (2010) DOI 10.1007/s12239−010−0037−x NUMERICAL INVESTIGATION OF TURBOLAG REDUCTION IN HD CNG ENGINES BY MEANS OF EXHAUST VALVE VARIABLE ACTUATION AND SPARK TIMING CONTROL M BARATTA , E SPESSA and P MAIRONE 1) 1)* 2) IC Engines Advanced Laboratory, Politecnico di Torino, Torino 10129, Italy Centro Ricerche Fiat, Orbassano 10043, Italy 1) 2) (Received 17 October 2008; Revised August 2009) ABSTRACT−Turbocharging port-injected Natural Gas (NG) engines allows them to recover gaseous-fuel related power gap with respect to gasoline engines However, turbolag reduction is necessary to achieve high performance during engine transient operations and to improve vehicle fun-to-drive characteristics Significant support for the study of turbocharged Compressed Natural Gas (CNG) engines and guidelines for the turbo-matching process can be provided by 1-D numerical simulation tools However, 1-D models are predictive only when a careful tuning procedure is set-up and carried out on the basis of the experimental data In this paper, a 1-D model of a Heavy-Duty (HD) turbocharged CNG engine was set up in the GT-POWER (Gamma Technologies Inc., Westmont, IL, US) environment to simulate transient operations and to evaluate the turbolag An extensive experimental activity was carried out to provide experimental data for model tuning The model buildup and tuning processes are described in detail with specific reference to the turbocharger model, whose correct calibration is a key factor in accounting for the effects of turbine flow pulsations The second part of the paper focuses on the evaluation of different strategies for turbolag reduction, namely, exhaust valve variable actuation and spark timing control Such strategies were aimed at increasing the engine exhaust-gas power transferred to the turbine, thus reducing the time required to accelerate the turbocharger group The effects of these strategies were examined for tip-in maneuvers at a fixed engine speed Depending on the engine speed and the applied turbolag reduction strategy, turbolag reductions from 70% to 10% were achieved KEY WORDS : Turbocharging, Turbolag, 1-D simulation NOMENCLATURE p : in-cylinder pressure at EVO P : prelift (of the exhaust valve) PFP : peak firing pressure PR : pressure ratio/turbine pressure ratio SOC : start of combustion ST : spark timing t : time T : temperature U : rotor blade tip speed UEGO: universal exhaust gas oxygen VVA : variable valve actuation WC : Wiebe constant WG : waste gate WOT : wide open throttle x : burned mass fraction γ : ratio of specific heats (of air) η : efficiency θ : crank angle τ : duration of the transient A : advance of EVO bmep : brake mean effective pressure BSR : blade speed ratio c : air specific heat at constant pressure C : engine brake torque CA : crank angle CNG : compressed natural gas C : isentropic gas velocity E : Wiebe exponent E-EVO : early exhaust valve opening EVO : exhaust valve opening HRR : heat release rate IC : inter-cooler L : lift (of the exhaust valve) : mass flow rate m· MAP : manifold absolute pressure N : engine speed n : turbocharger shaft speed NG : natural gas p s b SUPERSCRIPTS O *Corresponding author e-mail: ezio.spessa@polito.it 289 : total conditions 290 ' – M BARATTA, E SPESSA and P MAIRONE : burned gas : time-averaged value SUBSCRIPTS air c cmp exh in max red trb : air : combustion : compressor : exhaust gases : inlet : maximum value : reduced : turbine should be paid to the optimization of engine behavior under severe transient operations The introduction of a turbocharger strongly increases the complexity of the engine system and of the design process In particular, the problem of matching the engine with the turbocharger arises Although the final setup has to be defined through experimental analysis, a great deal of information about turbo-matching can be derived from numerical simulation based on 1-D fluid-dynamics codes These simulations allow engines to be studied under a wide range of operating conditions with limited cost penalties and are extremely useful for addressing the engine optimization process (Bush , 2000; Sammut and Alkidas, 2007) 1-D simulations are also widely used for valve lift selection and timing, intake and exhaust manifold layout optimization, and valve dimensioning (Westin and Ångström, 2003; Galindo , 2004, 2006) For turbocharged engines, it has been found that if simulation models are not properly tuned, calculation outcomes are likely to be quite different from the experimental results Such discrepancies are the result of turbine and compressor map quality as well as of errors in accounting for pulsating-flow effects on the turbine performance Turbine maps are typically measured under steady-state operations In order to overcome such discrepancies, a reliable procedure for correcting turbine steady-state maps is required (Westin and Ångström, 2003; Westin , 2004; Winkler et al ' INTRODUCTION NG-fuelled engines have recently emerged as a promising solution for the transportation sector in industrialized countries, thanks to the intrinsic environmental features of NG and to the favorable geopolitical distribution of reservoirs (d’Ambrosio , 2006) The application of NG engines is most advantageous for public urban transportation Any limitations to the vehicle’s operating range, due to the storage of fuel in a gaseous state, can be overcome by scheduling refueling stops at stations that are directly operated by the transportation providers The gaseous state of the fuel also reduces the engine power output (Kato , 1999; Zhang , 1998) However, that gap can be recovered by turbocharging (d’Ambrosio , 2006), as in the new-generation high-performance NG buses which exploit the high knock resistance of methane In contrast, the turbolag phenomenon is one of the major concerns regarding these engines due to driver perception of the vehicle’s performance Turbolag introduces a delay in the torque response under severe tip-in maneuvers The delay is due to the time required to increase the pressure in the intake manifold, which is influenced by the acceleration time of the turbocharger shaft Hence, particular attention et al et al et al et al et al ' et al Table Test engine characteristics Number of cylinders (in line) Number of valves (per cylinder) Bore 115 mm Stroke 125 mm Displacement 7.78 dm3 Compression ratio 11:1 Turbocharged Yes Intercooler Yes Figure Test engine: (a) Intake manifold and ports; (b) Manifold inlet, injectors and rail NUMERICAL INVESTIGATION OF TURBOLAG REDUCTION IN HD CNG ENGINES BY MEANS 291 the 1-D GT-POWER model is presented and its calibration procedure discussed, with particular reference to the turbocharger sub-model under both steady-state and transient working conditions In the second part, the tuned model is applied to the evaluation of different strategies for turbolag reduction, namely, exhaust valve variable actuation and spark timing control Such strategies were aimed at increasing the engine exhaust-gas power transferred to the turbine, thus reducing the time required to accelerate the turbocharger group TEST ENGINE AND EXPERIMENTAL SETUP Figure Engine performance at WOT Each quantity is normalized to its maximum value and Ångström, 2007) This paper can be divided into two parts In the first part, ' The test engine was developed at Fiat Research Centre for application to urban buses The major engine characteristics are reported in Table 1, and an engine schematic is provided in Figure Figure shows the nominal performance of the engine at Wide Open Throttle (WOT) The engine head Figure GT-Power engine model Figure Raw turbine performance maps Each quantity is normalized to a specific reference value 292 M BARATTA, E SPESSA and P MAIRONE features a spherical bowl-in-piston combustion chamber with a compression ratio (CR) of 11:1, four valves per cylinder, and one centrally-located spark plug The engine is boosted by a turbocharger with a twinentry turbine A closed-loop controller for the air temperature at the intercooler (IC) outlet is used During the dynomometer tests, the following cycle-averaged quantities were acquired: engine speed; torque; engine-out emissions downstream from the catalyst; temperatures at the exhaust ports; temperature and pressure values at compressor inlet, compressor outlet, IC outlet, intake manifold, intake ducts, turbine entries, locations upstream and downstream from the catalytic converter The in-chamber pressure timehistory was also acquired by means of a piezoelectric trans- ducer installed in the first cylinder In-cylinder pressure traces were referenced based on the intake absolute pressure measured by a piezoresistive transducer in the inlet manifold Finally, the engine was equipped with two airfuel ratio ‘NGK’ UEGO sensors (one for rich mixtures and the other for lean mixtures) in the exhaust system and with a pressure sensor in the injection rail ENGINE MODEL IN GT-POWER The engine was modeled with GT-POWER v6.2 build #3, a 1-D simulation tool licensed by Gamma Technologies, Inc (Westmont, IL, US) The GT-POWER model map is shown in Figure Figure 3(a) shows the cylinders, intake and Figure Turbine performance maps: (a), (b) Mass flow and efficiency fit versus data as functions of BSR; (c), (d) Mass flow and efficiency fit versus data as functions of PR; (e), (f) Final turbine maps, including the extrapolated range of reduced speed and PR NUMERICAL INVESTIGATION OF TURBOLAG REDUCTION IN HD CNG ENGINES BY MEANS exhaust systems, and Figures 3(b), (c) show the compressor-IC-intake manifold system and the twin-entry turbine, respectively As can be seen in Figure 3(b), the intake manifold has a central entry Between the intercooler and the manifold there is a by-pass duct whose aperture is pneumatically controlled in the production engine to limit the engine back-pressure at partial loads 3.1 Pipe and Flowsplit Submodels GT-POWER solves the inviscid form of the conservation laws of mass, momentum and energy With reference to pipes, these equations are discretized using a 1-D approach and a finite volume technique Pressure losses due to friction are computed automatically by the code, taking the Reynolds number and the surface roughness of the walls into account The modeled global heat exchange coefficient was proportional to friction using the Colburn analogy In some cases, it may be necessary to tune friction and heat transfer coefficients on the basis of experimental data regarding gas pressure and temperatures at relevant points Flow-splits were specifically designed (Gamma Technologies, 2006) to account for the conservation of momentum in three dimensions, even though the code is otherwise one-dimensional It is important to correctly specify the flow-split parameters (expansion diameter, characteristic length and orientation) to correctly reproduce wave phenomena and friction without using friction multipliers that are too far from unity 3.2 Turbocharger Submodel The turbocharger sub-model is a critical part of the overall engine model The approach followed in GT-POWER is to include turbocharger performance data in the form of lookup tables, which are processed by the software to obtain interpolated maps The quality of the final maps is highly dependent on the amount and type of experimental data, which are usually measured in a flow rig under steady-state conditions As an example, Figure shows the raw turbine map data, in terms of the reduced mass flow rate (left graph) and efficiency (right graph) versus pressure ratio (PR) for different speed lines (each colored line represents a different reduced speed nred) Each quantity has been normalized to its maximum value In GT-POWER, the performance tables are preprocessed to create internal maps that define the performance of the turbine and compressor in a wide range of operating conditions In particular, the turbine data, the quality of which is critical for turbocharged engine simulation (Westin and Ångström, 2003; Westin , 2004; Winkler Ångström, 2007), are preprocessed by the software (Gamma Technologies, 2006) based on well-known characteristics of turbines regarding efficiency, reduced mass flow rates and blade speed ratio (BSR) For a fixedgeometry turbine, efficiency and reduced mass flow rate should lie on specific trend lines when plotted against BSR, provided that each quantity is normalized to its value at the 293 maximum efficiency point on the actual speed line The application of this approach to the test engine is shown in Figures 5(a), (b) The agreement between the experimental data and the fit curves of Figures 5(a), (b) supports the accuracy of the procedure in a wide range of BSR values Figures 5(c), (d) show a comparison of the experimental data (symbols) from turbine performance maps to the correspondent preprocessed values (lines) obtained from the fit in Figures 5(a), (b) This highlights the capability of accurately reproducing the whole set of experimental turbine data with the exception of a couple of efficiency values at low PR and at high nred (Figure 5(d)) Figures 5(e), (f) show the complete extent of the mass flow (Figure 5(e)) and efficiency (Figure 5(f)) maps, which are obtained based on the fit curves, including the extrapolated ranges of nred and PR These plots are a graphical representation of the maps internally used by GT-POWER for the present application 3.3 Combustion Submodel The instantaneous value of the burned mass fraction b was modeled by means of a Wiebe function: E+1 (1) xb ( θ ) = ηc [ – exp( ( WC ) ( θ – SOC ) ) ] where ηc is the combustion efficiency, WC is the Wiebe constant, SOC is the crank angle at the start of combustion, θ is the instantaneous crank angle and is the Wiebe exponent The combustion model was applied using the two-zone thermodynamic approach of GT-POWER (Gamma Technologies, 2006) x E ' et al ' Figure Time histories of cylinder pressure, normalized to a specific value, at the indicated operating conditions 294 M BARATTA, E SPESSA and P MAIRONE The parameters in Equation (1) can be extracted by the heat-release analysis of experimental in-cylinder pressure time-histories To that end, a large number of experiments were carried out on the engine test rig under steady-state operating conditions at different values of engine speed (N) and brake mean effective pressure (bmep) for nominal spark timing (ST) operation Heat-release analysis was carried out with the specific tools embedded in GT-POWER so that the same thermodynamic and chemistry routines were applied for both diagnostic and prediction purposes For nominal ST operations, the obtained Wiebe parameters were organized in look-up tables as functions of N and bmep Such look-up tables were then used as inputs for the predictive model Figure provides an example of the experimental (diamonds) and simulated (solid line) pressure traces, both normalized to a specific reference value MODEL CALIBRATION It is generally accepted that 1-D models need to be carefully calibrated in order to provide accurate results (Westin and Ångström, 2003), especially in turbocharged engine applications Such a calibration is usually carried out by tuning the model so that it accurately reproduces experimental measurements taken under selected steady-state operating conditions In order to achieve the correct values of compressor and turbine operating points, it is necessary to adjust the turbine efficiency so as to match turbine and compressor cycleaveraged powers (Iwasaki , 1994, Westin and Ångström, 2002): ' ' et al ∫ Tcycle m· air ⋅ - ⋅ cpTin cmp ⋅ ( PRcmp ηcmp ∫ γ −1/ γ , Tcycle [ – 1) m· exh ⋅ ηtrb ⋅ cp′Tin trb ⋅ ( – PRcmp 1− , γ ′/ γ ′ dt = )] (2) dt where mair is the mass flow rate through the compressor, mexh is the mass flow rate of the exhaust gases through the turbine, p is the specific heat at constant pressure of air, γ is the ratio of specific heats of air, γ ' is the ratio of specific heats of exhaust gases, in,cmp is the total gas temperature at the compressor inlet, cmp is the pressure ratio across the compressor, and η cmp and ηtrb are the efficiencies of compressor and turbine, respectively All of the above quantities are instantaneous and the power balance is made with reference to a complete engine cycle The motivation for adjusting steady-state turbine efficiency is mainly related to the pulsating flow to which the turbine is exposed in the engine installation (Westin , 2004, Winkler and Ångström, 2007, Westin, 2005, Rakopoulos and Giakoumis, 2006) More specifically, under engine operations, the fraction of exhaust-gas energy that is available at the turbine inlet can be different from that under steady-state conditions (Baines, 2005) In addition, the extent of the pulsating flow-field · · c T PR ' et al requires that turbine maps cover a wide operating range with respect to both N and PR Flow pulsations are also present on the compressor side but they are much less significant 4.1 Model Calibration Results – Steady-State Figure provides the results of the model calibration procedure at three different loads (WOT, 25% and 4.2% of the maximum torque) and four engine speeds The following engine quantities are reported: Manifold Absolute Pressure (MAP; Figure 7(a)), boost pressure (Figure 7(b)), pressure ( in,trb) and temperature ( in,trb) at the first turbine inlet (Figures 7(c), (d)), pressure ( out,trb) at the turbine outlet (Figure 7(e)), compressor mass-flow rate (Figure 7(f)), peak firing pressure (PFP; Figure 7(g)), and engine brake torque (Figure 7(h)) Each quantity has been normalized with respect to a specific value The model is generally well calibrated in all the tested cases (Figure 7) With reference to the whole intake system and the portion of the exhaust ports within the cylinder head, the wall temperatures at the fluid side were set to specific values, which were selected based on the outcomes of the experimental tests For the pipes downstream from the exhaust ports, the GT-POWER Wall Temperature Solver was activated and the external temperature was set equal to the value in the cell cabinet Intake and exhaust ports were modeled as straight pipes, and therefore heat-transfer multipliers were introduced to account for bends, roughness, additional surface area and turbulence caused by the valves and stems (Gamma Technologies, 2006) There was no need to set heat-transfer multipliers elsewhere in the intake system or to add friction multipliers to the model because pressures and temperatures in the engine manifolds and ports were well reproduced (Figures 7(a), (b), (d)) The agreement between simulated and experimental values of PFP (Figure 7(g)) and engine brake torque (Figure 7(h)) demonstrate the accuracy of the combustion and engine friction sub-models, respectively As suggested by Westin and Ångström (2003) and by Gamma Technologies (2006), the above calibration was made with reference to a simplified model, which was obtained by removing both the turbocharger group and the IC, and by setting pressure, temperature and fluid composition at the domain boundaries to their experimentally measured values The first variable tuned during the calibration of the complete model, including the turbocharger and IC, was the turbine outlet pressure because it directly influenced the turbine power (Equation (2)) The pressure drop across the catalyst was simulated through a Multiple-Pipe object in which the diameter and the length of each pipe were based on the geometric characteristics of the catalyst The friction multiplier of the Multiple-Pipe object was set for each operating point in order to match the experimental values for pressure at the turbine outlet (Figure 7(e)) p T p ' NUMERICAL INVESTIGATION OF TURBOLAG REDUCTION IN HD CNG ENGINES BY MEANS 295 Figure Model results under steady-state working conditions, as functions of engine speed: (a) Manifold Absolute Pressure; (b) Boost pressure; (c) Pressure at turbine inlet (cylinder side); (d) Temperature at turbine inlet; (e) Pressure at turbine outlet; (f) Air mass-flow rate; (g) Peak Firing Pressure (cylinder 1); (h) Engine brake torque – Each quantity is normalized to a specific value For the calibration of the turbine efficiency, it is worthwhile making reference to the WOT conditions in Figure The first three experimental points on each WOT curve (engine speeds between 0.35 and 0.5 on the normalized scale) were characterized by closed waste-gate (WG) valve, whereas the WG valve was partially open for the remaining two points on each curve For closed WG operations, the turbine efficiency multiplier was selected to match the measured turbocharger-shaft speed In the wastegated cases, both shaft speed and mass-flow rate across the waste-gate valve should be matched However, the latter quantity was not measured in the experimental tests Therefore, as suggested by Westin and Ångström (2003), the multiplier for turbine efficiency was chosen so as to match the turbo speed and the pressure at turbine inlet (Figure 7(c)) Differences between simulated (solid line) and experi' 296 M BARATTA, E SPESSA and P MAIRONE mental (circles) data are consistent with the uncertainty of the pressure measurements The results of this tuning procedure are reported in Figure 8, where the ratio between the resultant apparent turbine efficiency under pulse-flow conditions (ηtrb,apparent) and the correspondent steady-state efficiency from turbine maps (ηtrb,steady) are plotted as functions of cycle-averaged turbine PR ηtrb,apparent corresponds to ηtrb in Equation (2) Figure was organized in a look-up table as a function of PR and included in the model 4.2 Model Calibration Results – Transients Two load steps at different constant engine speeds ( / max =0.55 and 0.75) were considered For both load steps, the throttle was opened abruptly and the torque varied from about 4.2% load to the steady-state values at WOT Before applying the model to the transient simulations, the following changes were made: • the Wall Temperature Solver was activated in the pipes between the compressor and the intercooler in order to accurately simulate the temperature time-history at the compressor outlet; • the catalyst friction multiplier was organized in a look-up table as a function of the mass flow and was included in the model The calibration results for transient operations are shown in Figure for / max =0.55 The model was well calibrated Not only are the asymptotic values well reproduced but also the simulated slopes occurring during the transient are comparable to the experimental ones However, some discrepancies are observed in the time-histories of the temperature at the turbine inlet (Figure 9(d)) and the brake torque (Figure 9(f)) The main differences between simulated and experimental in, trb time-histories are that: • the simulated asymptotic value at the end of the transient is higher than the experimental one This can be ascribed to an underestimation of the measured gas temperature, which is due to the heat transferred from the thermocouple to the pipe walls by both radiation and conduction through thermocouple stem (Westin and Ångström, 2003; Westin, 2005) To reduce the heat transfer by conduction, the thermoN N N N T ' Figure Ratio of apparent turbine efficiency under pulsating flow conditions to steady-state flow efficiency Cycleaveraged PR is normalized to a specific value couple should be immersed as far as possible into the pipe To reduce the measurement error due to radiation, proper radiation shields should be used (Doebelin, 1990) In the considered experimental setup, no shielded thermocouples were used, and the engine test-bench layout limited the insertion length of the probe to about 10~15 times the probe diameter, which is generally reported to be insufficient (Ehrlich, 1998; Westin, 2005) Hence, an underestimation of the gas temperature had to be taken into account • Due to thermocouple thermal inertia, the slopes of simulated and measured temperature rise are different Fast temperature oscillations, such as those calculated during the first transient phase, cannot be measured by the thermocouple (Westin and Ångstrưm, 2003) • The computed gas temperature before the tip-in event is lower than that measured during the experiment Under such a partial load, the turbocharger group produces virtually no boost, which in turn has no practical influence on the transient simulation Therefore, the calibration of exhaust-pipe heat-transfer multiplier and wall temperature were not performed at this operating condition, likely contributing to the observed difference in gas temperatures Experimental temperature measures can also be affected by both uncertainty in the thermocouple position and high temperature gradients in the exhaust manifold (Westin and Ångström, 2003) The slight difference between the calculated and the measured brake torque at the transient end (Figure 9(f)) can be primarily ascribed to an underestimation of the gas pressure contribution to the friction mean effective pressure under full-load operations ' ' TECHNIQUES FOR TURBOLAG REDUCTION The described GT-POWER engine model was applied to the analysis of turbolag during tip-in events at constant engine speeds Two strategies were investigated: • Early-Exhaust Valve Opening-Variable Valve Actuation (E-EVO-VVA): immediately after the tip-in event, Exhaust Valve Opening (EVO) was advanced for fixed exhaust valve closing and a different profile for exhaust valve lift was actuated After a selected number of engine cycles had elapsed, EVO and valve lift were switched back to their baseline values and profiles • Combustion Retard (ComR): immediately after the tip-in event a retard in ST was set Then, after a selected number of engine cycles, ST was switched back to its baseline value Both strategies determined a higher enthalpy drop across the turbine and a consequent increase in the turbine power This in turn caused faster turbo-shaft accelerations However, such approaches might reduce piston work during the expansion stroke As such, the trade-off between turbo shaft acceleration and reduced piston work has to be analy- NUMERICAL INVESTIGATION OF TURBOLAG REDUCTION IN HD CNG ENGINES BY MEANS 297 Figure Model results under transient working conditions (load step at constant engine speed – N = 0.55 Nmax): (a) Manifold absolute pressure; (b) Boost pressure; (c) Pressure at turbine inlet (cylinder side); (d) Temperature at turbine inlet (cylinder side); (e) Exhaust mass-flow rate; (f) Engine brake torque Each quantity is normalized to a specific value zed as a function of EVO advances, valve lift profiles, EVO and valve lift switch-back timings, ST retards and ST switch-back timings A combination of the strategies introduced above was also investigated by advancing EVO and by setting a retard in ST at the same time The resultant strategy will be referred to as Combined in the following sections Figure 10 E-EVO-VVA lift profiles: (a) Full Lift; (b) Prelift Dashed lines indicate the baseline lift profile 5.1 E-EVO-VVA Strategy Figure 10 shows the investigated lift profiles of the exhaust valve A suitable advance (A) in EVO along with the new lift profiles (solid lines in Figure 10) were set immediately after throttle-valve step-opening, whereas the lift profile was switched back to its baseline value (dashed lines) after a specific number of engine cycles Preliminary analyses indicated that the effects of the Prelift (Figure 10(b)) profile with P/Lmax ≥ 0.3 on turbolag were equivalent to those of a Full Lift (Figure 10(a)) profile with the same EVO advance In addition, a Prelift profile can be realized by EFFECTS OF STACK ARRAY ORIENTATION ON FUEL CELL EFFICIENCY FOR AUXILIARY POWER UNIT 431 produced during the run was purged from the stack before initiating the next run Data analysis was performed in three phases In the first phase, net and gross data files were aligned by synchronizing all of the set-point transitions In the next phase, the demand for net efficiency, maximum power and auxiliary power was calculated Breaking down the auxiliary power demand into individual component demands required the use of Regression Analysis In the final phase, ANOVA was used to test the effect of the control variables on net efficiency (J Neter, , 1996) ANOVA calculates an F distribution as defined below: et al Mean_Square factors -below: = -Mean (1) _ Square errors The value reflects the significance of the effect of variable on net efficiency To perform ANOVA analysis, a statistical package program (SAS) was used All calculations were performed using a 95% level of confidence Each fuel cell stack has its own unique polarization curve, which indicates the amount of power the stack will produce for a given voltage or current By joining stacks into arrays, particular parameters of the polarization curves are bound together; the series arrays bind the net currents, and the parallel arrays bind the net voltages The Nexas used in this study were well matched but not perfectly matched As a result, differing polarization curves introduced unequal load sharing The load sharing for each array was quantified by calculating the load share ratio as follows: F F Σ ( A /B) Σ(B/B) - : n n (2) where A and B designate the power output of the stronger and weaker stacks at a particular time period, respectively; n represents the total number of samples taken during the time period (EG&G Technical Services Inc., 2002) RESULTS AND DISCUSSION 3.1 Maximum Power of Dual Stack Arrays Although an individual Nexa power module is able to produce 1500 W, they not maintain that capability while in array configurations Both series and parallel arrays failed under a 3000 W load because of a sustained low purge cell voltage or an over-current condition When connected to 2900 W loads, all of the series runs were completed successfully, and all of the parallel runs failed because of an over-current Table provides a typical example of a series stack failure caused by a sustained low purge cell voltage during a 3000 W test The currents and voltages were within reason, but the purge cell voltage combined for the series stack remained below 1.10 V for longer than 20 sec In applications not using the NexaMon program to monitor the purge cell voltage, the operator would get no warning Table Typical examples of a series stack failure during a 3000 W set-point Power Current Voltage PCV Failure (W) (A) (V) (V) Nexa-570 1509 55.9 27.0 1.12 No Nexa-592 1487 55.9 26.6 1.09 Yes Series array 2996 55.9 53.6 N/A Yes Table Typical examples of an imminent (0.06 s) failure during a parallel FC array test Power Current Voltage PCV Failure (W) (A) (V) (V) Nexa-570 1514 64.81 23.36 1.11 No Nexa-592 1390 59.52 23.36 1.00 No Parallel array 2904 124.3 23.36 N/A No that a stack failure due to low purge cell voltage is about to occur Even with such monitoring, the only way to avert such a failure would be to reduce the load on the FC array by an amount sufficient to raise the purge cell voltage above 1.10 V Table illustrates a typical parallel test that showed that the Nexas could provide 2900 W, but only for a fraction of a second Providing the power during the overshoot period required the stronger stack to produce current in excess of the 60 A maximum limit Had the maximum current limit been exceeded for longer than sec, the Nexa would have shutdown due to an over-current failure Unlike the failure of the series array above, a failure of a parallel array could be predicted using commonly available current and voltage sensors As was the case with the series array, the only way to avert operational stack failure was to reduce the load on the FC array 3.2 Effects of Dual Stack Array Orientation on Net Efficiency Using the maximum sustainable power level, the randomized load schedule given in Table was used to test both the series and parallel arrays Although neither array experienced an operational stack failure during the preliminary testing, one failure did occur during the randomized load-level testing Representative power curves for the parallel and series arrangements are given in Figures and 3, respectively Overall, two dynamics occur during the operation of both arrays First, there is a fairly constant load sharing ratio, where the Nexa with the higher polarization curve produces more of the power Second, the fluctuating partial pressure of hydrogen within the two Nexas causes an instantaneous load sharing ratio to vary with time A summary of all the parallel and series run results are given in Tables and 5, respectively 432 K.-S CHOI et al Figure Representative power curves and load sharing for the parallel arrangement Figure Representative power curves and load sharing for the series arrangement Despite appearing essentially identical, as shown in Tables and 5, there is a statistical difference in net efficiency between the two array arrangements, as shown in Tables 6, 7, and The F-Test results in Tables 6, 7, and show that the difference in net efficiency between the parallel and series Table Summary and net efficiency data of all of the parallel run results Successfully completed runs Failed run Run P1 P2 P3 Nexa N-570 N-592 Parallel N-570 N-592 Parallel N-570 N-592 Average gross current (A) 23.9 23.11 47.01 23.78 23.23 47.01 23.18 22.28 Average net power (W) 661.35 640.82 1303.12 654.14 648.66 1302.81 640.2 617.86 Run time (s) 870 870 749 ∆H (W, HHV) 1,664 1,609 3,273 1,656 1,617 3,273 1,614 1,551 Net efficiency (%, HHV) 39.74 39.83 39.81 39.5 40.12 39.8 39.67 39.84 Parallel 45.47 1259.18 Table Summary and net efficiency data of all of the series run results Successfully completed runs Run S1 S2 Nexa N-570 N-592 Series N-570 N-592 Series N-570 Average gross current (A) 23.2 23.05 23.2 23.1 22.97 23.1 23.33 Average net power (W) 656.5 647.12 1303.61 656.01 647.61 1303.62 655.55 Run time (s) 870 870 870 ∆H (W, HHV) 1,615 1,605 3,230 1,608 1,599 3,216 1,624 Net efficiency (%, HHV) 40.65 40.32 40.36 40.8 40.5 40.54 40.37 S3 N-592 23.21 648.05 Series 23.33 1303.6 1,616 40.1 3,248 40.14 Table Table-System Net Efficiency: dependent variable: system, number of variables Source DF Sum of squares Mean square F value 0.4411067 0.4411067 22.62 Model error 0.0779906 0.0194977 corrected total 0.5190973 R-Square Coeff var Root MSE System mean 0.849757 0.348453 0.139634 40.07257 Source DF Type I SS Mean square F value Arrangement 0.4411067 0.4411067 22.62 3,166 39.77 Probability>F 0.0089 Probability>F 0.0089 EFFECTS OF STACK ARRAY ORIENTATION ON FUEL CELL EFFICIENCY FOR AUXILIARY POWER UNIT 433 Table Table-NEXA 570 Net Efficiency: dependent variable: NEXA 570, number of variables Source DF Sum of squares Mean square F value 1.9021912 1.9021912 27.98 Model error 0.2719629 0.0679907 corrected total 2.1741541 R-square Coeff var Root MSE System mean 0.874911 0.648541 0.26075 40.20568 Source DF Type I SS Mean Square F Value Arrangement 1.9021912 1.9021912 27.98 Probability>F 0.0061 Probability>F 0.0061 Table ANOVA Table-NEXA 592 Net Efficiency: dependent variable: NEXA 592, number of variables Source DF Sum of squares Mean square F value Probability>F 0.0781146 0.0781146 4.16 0.1109 Model error 0.0781146 0.0187624 corrected total 0.1531644 R-square Coeff var Root MSE System mean 0.510005 0.342138 0.136976 40.03532 Source DF Type I SS Mean square F value Probability>F Arrangement 0.0781146 0.0781146 4.16 0.1109 arrays was more significant for the Nexa-570 than for the Nexa-592 This is a consequence of the overall load sharing between the two Nexas The average load sharing ratios for the series and parallel arrays were 1.02:1 and 1.09:1, based on Tables and 3, respectively The discrepancy explains why the series array was able to sustain a 2900 W load, while the parallel array was only able to sustain a 2800 W load The difference in maximum sustainable load is due to the different parameter bounds in each array orientation With current as the limiting factor at high power production levels, binding the currents together in the series array ensures that each stack reaches the maximum current limit at the same time This forced the weaker stack to produce its fair share of the work In contrast, binding voltages together will force the stack with the stronger polarization curve to produce more power because it will produce more current at a given voltage Thus, at maximum powers, the stronger stack in a parallel array will approach the maximum current limit before the weaker one While the polarization curves had the strongest influence on the overall load sharing, there were some other factors This is evidenced by the variability in the instantaneous load sharing, as indicated by the rough appearance of the curves in Figures and This roughness is a result of the change in partial pressure of hydrogen at the anodes of the Nexas, which in turn affects the stack voltage The frequency of the voltage oscillation is dependent on the level of power production and the frequency of the anode purges At higher levels of power production the period of purge oscillation is fairly high, on the order of 30 sec; at lower levels the period of the purge oscillation is about The stack power for a given load level, at the right side of the curves, is scattered because almost the entire voltage oscillation occurs during the sampling period This makes those portions of the curves appear rougher At lower power levels, only a portion of the voltage oscillation occurred during the sampling period This makes that section of the curve appear smoother The oscillations occurred asynchronously for series arrays and synchronously for parallel arrays, which is why the effect was more pronounced in Figure The effective array voltage oscillation for the parallel array is almost twice as much as it is for the series array One of the potential concerns for running mismatched fuel cells in parallel is the likelihood of reverse currents from the stronger stack into the weaker one As soon as two stacks with differing open circuit voltages are connected in parallel a current loop is created, even before a load is applied The amount of current flow can be found if both the polarization curve of the stronger stack and the resulting system voltage are known In this study, one Nexa’s average VOC was 1.4 V higher than the other When connected in parallel, the voltage of the stronger stack dropped by 0.7 V, the voltage of the weaker one increased by 0.7 V, and a negative current of about 2.8 A began flowing into the weaker stack With a VOC of 42.0 V, 117.6 W of power was dissipated by the weaker stack We not know how this condition would affect the lifespan of the weaker stack 434 K.-S CHOI if it were to occur over a long period of time CONCLUSIONS The sensitivity of an FC system to dual-stack parallel and series array operation was investigated experimentally Although an individual Nexa FC is able to produce 1500 W, two Nexas in an array configuration could not produce 3000 W because of unequal load sharing With an overall load share ratio of 1.02:1, the series array reliably produced 2900 W of power, while with an overall load share ratio of 1.09:1, the parallel array reliably produced only 2800 W of power We have shown that array orientation significantly affects both the system net efficiency and the individual stack net efficiency The system net efficiency is lower for the parallel arrangement than for the series arrangement because connecting the stacks in parallel equalizes the stack voltages The weaker stack depresses the polarization curve of the stronger stack, while the stronger stack boosts the polarization curve of the weaker stack Finally, even though the Nexas used in this study were well matched, arranging them in parallel produced a reverse current between the two stacks The power flowing between the two stacks was approximately 118W in the open circuit The effects of this reverse current on the lifespan of the weaker stack should be investigated in future studies ACKNOWLEDGEMENT−This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No 2009-0080496) REFERENCES Ballard Power Systems (2004) Nexa Power Module Specification Sheet [online] Ballard Power System Inc et al Available from: http://www.ballard.com/resources/ powergen/NexaSpecSheet.pdf [accessed 11 Dec 2006] Brodrick, C J (2002) Evaluation of fuel cell auxiliary power units for heavy-duty diesel trucks Transportation Research-Part D 7, 4, 303−315 Brodrick, C J (2000) Demonstration of a proton exchange membrane fuel cell as an auxiliary power source for heavy trucks SAE Paper No 2000-01-3488 Cacciola, G., Antonucci, V and Freni, S (2001) Technology up date and new strategies on fuel cells J Power Sources 100, 1−2, 67−79 EG&G Technical Services, Inc (2002) Fuel Cells: A Handbook 6th Edn US Department of Energy Illinois He, B., Ouyang, M and Lu, L (2005) Modeling and PI control of diesel APU for series hybrid electric vehicles Int J Automotive Technology 7, 1, 91−99 Neter, J., Kutner, M H., Nachtsheim, C J and Wasserman, W (1996) Applied Linear Statistical Models 4th Edn McGraw-Hill Boston Qi, Z and Kaufman, A (2002) PEM fuel cell stacks operated under dry-reactant conditions J Power Sources 109, 2, 469−476 Read, C J., Jan, H J., Thijssen, S and Carlson, E J (2001) Fuel cell auxiliary power systems: Design and cost implications SAE Paper No 2001-01-0536 Venturi, M and Martin, A (2001) Liquid-fueled APU fuel cell system for truck application SAE Paper No 200101-2716 Venturi, M., Kallio, E., Smith, S and Baker, J (2003) Recent results on liquid-fuelled APU for truck application SAE Paper No 2003-01-0266 Zizelman, J., Shaffer, S and Mukerjee, S (2002) Solid oxide fuel cell auxiliary power unit - A development update SAE Paper No 2002-01-0411 Copyright © 2010 KSAE 1229−9138/2010/052−17 International Journal of Automotive Technology, Vol 11, No 3, pp 435−440 (2010) DOI 10.1007/s12239−010−0053−x EFFECTS OF INJECTION PARAMETERS ON THE SPRAY CHARACTERISTICS OF SWIRL AND SLIT INJECTORS USING THE MIE-SCATTERING METHOD C H LEE , K H LEE and K B LIM 1) 2)* 3) Technology Institute R&D Engineering, STX Engine Co Ltd., 80 Seongsan-dong, Changwon-si, Gyeongnam 641-315, Korea Department of Mechanical Engineering, Hanyang University, Gyeongi 425-791, Korea Department of Mechanical Design Engineering, Hanbat National University, Daejon 305-719, Korea 1) 2) 3) (Received 16 January 2008; Revised 16 September 2009) ABSTRACT−We investigated the effects of injection parameters such as injection pressure, ambient pressure, and ambient temperature on spray characteristics We calculated the turbulence occurring point (tc), defined as the time required to generate a vortex, and the deceleration point (tb), defined as the time when spray penetration begins to decelerate, to elucidate the breakup mechanism of the test injectors The spray velocity coefficient (Cv) was obtained to evaluate the spray characteristics As the ambient pressure increases in the case of a slit injector, Cv decreases We investigated the effects of nozzle tip shape according to injection pressure, ambient pressure, and fuel properties on spray characteristics and provide a Cv value of 0.38 for the swirl injector with a spray angle of 60o and the slit injector under atmospheric conditions The value of Cv in the case of a slit injector was reduced by increasing the ambient pressure Our results suggest that Cv of a swirl injector is constant regardless of changes in ambient pressure, injection pressure, and fuel properties On the other hand, Cv of a slit injector is altered by changes in ambient pressure KEY WORDS : Turbulence occurring point (t ), Deceleration point (t ), Velocity coefficient (C ), Direct injection spark b ignition (DISI) c NOMECLATURE tb tc S Vb t Pinj Pa Ta Vinj ∆P ρf C v v engine, as is the emission of toxic gases such as CO, HC, and NO The Direct Injection Spark Ignition (DISI) engine was recently developed as a next-generation engine that can satisfy Super Ultra Low Emission Vehicle (SULEV) regulations and reduce fuel consumption Therefore, many researchers have studied the high-pressure, vortex-type injector that is one of the key parts of the DISI engine (Tanaka et al., 2003; Lee et al., 2003, 2001) The preparation of a stratified mixture in a specified area of the combustion chamber within a limited time interval requires an optimum combination of three factors: gas flow, fuel spray, and combustion chamber configuration Visualization or quantitative measurement of the mixture formation process in the combustion chamber is a very effective way to optimize the operating conditions, and such techniques are also effective in shortening the development period of the combustion concept for a DISI engine The third generation of spray-guided DISI engines is now on the verge of being released We investigated these engines and identified the essential aspects of the sprayguided mechanism by applying quantitative methods for measuring mixture concentration Stratified charge combustion involves localized mixture formation; therefore, we traced mixture behavior in the x : deceleration point [msec] : turbulent occurring point [msec] : spray penetration [mm] : penetration velocity before breakup [m/s] : time after injection start [msec] : injection pressure [MPa] : ambient pressure [MPa] : ambient temperature [K] : injection velocity [m/s] : injection pressure difference [MPa] : density of liquid fuel [kg/m ] : velocity coefficient INTRODUCTION As environmental problems caused by vehicle exhaust emissions become increasingly severe, exhaust emission standards and fuel economy regulations become more stringent The emission of CO gas (one of the main causes of global warming) is a major drawback of the gasoline *Corresponding author e-mail: hylee@hanyang.ac.kr 435 436 C H LEE, K H LEE and K B LIM formation process to determine the best way to achieve combustion In addition, changes in the air-fuel ratio near the spark plug gap were measured continuously by means of an infrared absorption technique developed at Nissan as an onboard measurement method With these data, we performed quantitative analysis of the stable combustion region by manipulating the ignition timing and the fuel injection timing, which are the most important aspects of mixture formation in third-generation spray-guided DISI engines Many researchers have investigated the spray characteristics of fuel injectors Zhao (1996) and Yamaguchi (1996) visualized the spray process of the DISI injector by using two-dimensional Mie-scattering and Laser-Induced Exiplex Fluorescence (LIEF) methods and also measured the Sauter Mean Diameter (SMD) (Shelby , 1998; Melton, 1983, 1984; Senda , 1997; Fujimoto , 1997) Zhao (2005) performed a numerical and experimental analysis of the fuel distribution characteristics of a high-pressure DISI injector using the Mie-scattering method Detailed information about the spray structure and characteristics of DISI injectors is limited, however, due to experimental difficulties We investigated the spray behavior of swirl and slit injectors according to injection pressure, ambient pressure and temperature, and fuel properties The relationships of b and c were examined under ambient temperature and pressure conditions By analyzing the coefficient velocities of test injectors, we determined the effects of injection parameters such as ambient temperature, injection pressure, and fuel properties on injector performance et al et al et al et et al Figure Experimental apparatus al et al t t EXPERIMENTAL APPARATUS AND PROCEDURE 2.1 Experimental Apparatus Figure describes the experimental apparatus system used to obtain the Mie-scattering images employed in this study to analyze the spray characteristics of DISI injectors We obtained spray images under different ambient pressure and temperature conditions in a high-pressure simulated engine combustion chamber The internal diameter and height of the chamber were 100 mm and 300 mm, with a total volume of 2355 cc The chamber was constructed of duralumin, which has excellent durability and thermal diffusivity A 4-kW heater attached to the chamber wall controlled the initial temperature A Nd:YAG (continuum sulite-II) laser with an intensity and wavelength of 200 mJ/ pulse and 532 nm was used as a light source The sheetbeam was 0.5 mm thick The exposure timings, injection signals, and laser timings were controlled by the LabVIEW program manufactured by N.I Ltd When the pumping frequency is 10 Hz in the Nd:YAG laser, the energy level and intensity maintain constant values Figure shows a system diagram of the optimized pulse generation and a timing chart for synchronizing the camera and laser signal Figure Schematic diagram of the optimized pulse generation and timing chart Table Experimental conditions Ambient temperature 283 K, 373 K, 432 K Ambient pressure [MPa] 0.1, 0.5, and 1.0 Injector Swirl , Slit Fuel Gasoline, n-Octane, n-Heptane Spray Inj Press [MPa] 5, 8, 10, 12 Injection period msec Table Specifications of test injectors Injector hole Inj rate θ0, α0 Type (spray angle) (mm) (mg) o Swirl (Figure 3(a)) θ0 =60 0.5 10 h1=1.1 Slit (Figure 3(b)) α0 =60o 11.3 h2=0.15 Tables and show the experimental conditions and specifications of the test injectors, respectively Figure shows a comparison of sectional configurations for two injectors Table shows the properties of the fuel used in this experiment 2.2 Measurement of Spray Penetration The spray characteristics of interest were the spray penetration length and spray cone angle These were measured directly from spray images, as shown in Figure The spray penetration length is defined as the distance from the EFFECTS OF INJECTION PARAMETERS ON THE SPRAY CHARACTERISTICS OF SWIRL AND SLIT INJECTORS 437 Figure Definition of spray penetration and turbulence occurring point (t ) c Figure Spray behavior according to ambient pressure with a swirl injector (Pinj: MPa, ambient temperature: 283 K) Figure Comparison of sectional configurations for test injectors nozzle exit to the farthest position of the spray in the axial direction The spray angle is calculated by measuring the angle formed by two straight lines drawn from the nozzle tip to the outer periphery of the spray The turbulence occurring point (t ) is defined as the time necessary to generate a vortex c RESULTS AND INVESTIGATION 3.1 Spray Characteristics of Swirl and Slit Injectors Figure displays the visualization results obtained under different ambient pressures using the system shown in Figure At an early stage of injection, the spray from a swirl injector under normal atmospheric conditions did not form a swirling conical shape Instead, a bulk of liquid consisting of large droplets and ligaments emerged from the nozzle hole However, a conical hollow-cone spray formed as the spray developed, and the slug of liquid lost its form 1.407 msec after the start of injection At the same time, a vortex became evident at the outer edges of the spray The vortex moved downstream as the spray developed The image taken at 2.34 ms shows droplets at the tip as they begin to turn upwards and towards the outside as they were entrained in the outside vortex This recirculation zone suppressed the spray cone angle The vortex flow also tended to carry smaller drops, and as a result, the vortex Table Properties of test fuels Boiling point Vapor press Latent heat of vaporiza- Surface tension Density Critical temp Critical press [K] [MPa] [K] [MPa] tion at 0.1 MPa [kJ/kg] [mN/m] [kg/m3] n-octane 399 0.0109 301.26 21.3 703 568.8 2.49 n-heptane 372 0.0067 320 20.1 678 540 2.74 Gasoline 320~470 320 25.1 746 438 C H LEE, K H LEE and K B LIM the side view shows that a vortex was generated in the middle part of the spray at 0.737 ms and 1.407 ms This vortex increased as the injection pressure increased The front view reveals that the fuel density of the spray section was heterogeneous, caused by heterogeneous scattering, which occurs when a liquid column is not equally distributed in the section of spray Figure Spray behavior according to ambient temperature with a swirl injector (Pa: 1.0 MPa, differential injection pressure: ∆P (P − Pa)=5 MPa) inj 3.2 Characteristics of Spray Penetration 3.2.1 Determination of deceleration point (tb) In order to find the deceleration point (tb), which is defined as the time when the spray begins to decelerate, the penetration length (S) is plotted on a logarithmic chart against the time after the initial injection Results of the chart are shown in Figure and Figure The gradient of each data line is constant at 45 for some time after the initial injection, and it changes to around 22.5 at t = tb An arrow is used to indicate the turning point Figures and show that the penetration length (S) increases proportionally with time (t) when t < tb However, S increases nearly proportionally to t when t > tb In addition, the S curve at 0.1 o o 1/2 Figure Spray behavior according to injection pressure at the front and side views with a slit injector (T =283 K, Pa=0.1 MPa) a cloud was seen at the spray tip in the final spray image Increasing the ambient pressure caused the spray penetration and angle to decrease It appears that spray penetration reduced the fuel momentum due to an increase in the ambient air density within the pressure chamber Figure shows the spray behavior observed at different ambient temperatures at an ambient pressure of less than MPa Vortex formation was reduced by increasing the ambient temperature, which suggests that an increase in ambient temperature will decrease the spray penetration In the last stage of the spray, the vortex magnitude appears to diminish due to the vaporization of spray droplets in the vortex cloud region Figure provides front and side views of spray behavior for a slit injector observed at different injection pressures These visualization results reveal that increases in spray penetration were proportional to increases in injection pressure Major differences between two injection pressures were not observed in the side view (Figure 7(b)) However, Figure Spray penetration at different ambient pressures when ambient temperature=283 K Figure Spray penetration at different ambient temperatures when ambient pressure=1.0 MPa EFFECTS OF INJECTION PARAMETERS ON THE SPRAY CHARACTERISTICS OF SWIRL AND SLIT INJECTORS MPa of ambient pressure does not coincide with the 22.5 line, due to the reacceleration phenomenon explained in the next section This result demonstrates that the injection duration time (namely, the injection quantity) has no effect on spray behavior, including tb, because the injection rate is kept constant, as shown in Table From Figure 8, we found that tb occurs earlier with greater ambient pressure because the spray penetration velocity is reduced due to the increase in ambient pressure The difference between the injection pressure and ambient pressure is kept constant at MPa The relationships of S and t at different ambient temperatures are shown in Figure When the ambient temperature increases, the spray penetration length does not increase when t ≥ tb but does increase when t ≤ tb Since the droplet velocity is influenced by the vaporization characteristics resulting from increases in the ambient temperature, the spray penetration length increases slightly At an ambient temperature of 373 K, tb increased, but it decreased at an ambient temperature of 432 K because the fuel went from the liquid phase to the vapor phase A system using the Mie-scattering method provides only a weak signal in cases of high room temperatures 439 o 3.2.2 Correlation of tb and tc The correlation of tb and tc at different ambient pressures and temperatures is shown in Figure 10 When the ambient pressure is held at 0.1 MPa, increases in the ambient temperature lead to smaller values of tb and tc, while the turbulence occurring point tc is more strongly affected than tb as the ambient pressure increases Therefore, increasing the ambient density can force the turbulence occurring point tc to occur earlier The correlation of tb and tc under different ambient pressures and fuel properties is shown in Figure 11 The surface tension of the fuel has the strongest effect on tb and tc, as shown in Table As the surface tension becomes Figure 11 Correlation of tb and tc at different ambient pressures and with various fuels small, the state phase quickly changes, due to the effects of the ambient temperature When the ambient temperature increases, tb and tc decrease These results suggest that tb and tc occur latest for n-heptane and earliest for gasoline 3.3 Analysis of the Velocity Coefficient The penetration velocity before break-up is constant, as discussed in the previous section The penetration velocity, Vb, is equal to the injection velocity Vinj Therefore, S = V ⋅ t = C ⎛ -2 ⋅ ∆ P -⎞ ⋅ t , (1) 1/2 b V ⎝ ρf ⎠ where ∆P denotes the injection pressure difference [Pa], ρf is the density of the liquid fuel, and C is the velocity coefficient The value of C is determined by the formula CV = Vb /(2·∆P/ρf ) , where the velocity obtained by measuring the gradient of the S-t curve is used for Vb, and v v 1/2 Table Determination of C values Conditions C Range Ave P =5 MPa Swirl =0.1, 0.5, 1.0 MPa injector 0.33~0.41 0.37 PTa=283, 373, 432 K (60 ) Three fuels P =10 MPa MPa 0.36~0.41 0.385 Pa=0.1 Ta=283, 373, 432 K Slit Three fuels injector Pinj=10 MPa (60 ) MPa 0.28~0.33 0.31 Pa=0.5 Ta=283, 373, 432 K Three fuels v v inj a o inj o Figure 10 Correlation of tb and tc at different ambient pressures and temperatures 440 C H LEE, K H LEE and K B LIM the value Pinj − Pa is used for ∆P Values of C for various conditions are shown in Table C for a swirl injector with a spray angle of 60 and a slit injector, under a Pa of 0.1 MPa, has a value near 0.38 This value agrees with the value reported by Hiroyasu et al (1980) for diesel fuel The value of C for a slit injector is reduced by increasing the ambient pressure In addition, we found that C is different for the two injector types, and the C of a slit injector is affected by changing the ambient pressure v v o v v v and spray angle in diesel Trans JSAE, , 5−11 (in jananess) Lee, C H and Lee, K H (2003) New technology of the mixture formation for the spark ignited direct injection gasoline engine J Korean Society of Automotive Engineers , , 21−28 Lee, C S., Chon, M S and Park, Y C (2001) Spray structure of high pressure gasoline injector in a gasoline direct injection engine Int J Automotive Technology , , 165−170 Melton, L A (1983) Spectrally separated fluorescence emissions for diesel fuel droplets and vapor Applied Optics, , 2620−2624 Melton, L A and Verdieck, J F (1984) Vapor/liquid visualization in fuel sprays 20th Int Symp Combustion (Combustion Institute), 1283−1290 Senda, J., Kanda, T., Kobayashi, M and Fujimoto, H (1997) Quantitative analysis of fuel vapor concentration in diesel spray by exciplex fluorescence method SAE Paper No 970796, 1012−1024 Shelby, M H., van der Wege, B A and Hochgreb, S (1998) Early spray development in gasoline directinjected spark ignition engines SAE Paper No 980160, 67−84 Tanaka, Y., Takano, T., Sami, H., Sakai, K and Osumi, N (2003) Analysis on behavior of swirl nozzle spray and slit nozzle spray in correlation to DI gasoline combustion SAE Paper No 2003-01-0058, 1−20 Yamauchi, T and Wakisaka, T (1996) Computation of the hollow-cone sprays from a high-pressure swirl injector for a gasoline direct-injection SI engine SAE Paper No 962016 Zhao, F.-Q., Lai, M.-C and Harrington, D L (1996) A review mixture preparation and combustion control strategies for spark-ignition direct-injection gasoline engines Progress in Tech., , 21−64 Zhao, F.-Q., Yoo, J.-H and Lai, M.-C (1996) Spray dynamics of high pressure fuel injectors for DI gasoline engines SAE Paper No 961925, 1924−1953 21 25 4 CONCLUSIONS (1) The spray penetration and angle decreased with increased ambient pressure because these quantities are affected by the reduction of fuel momentum caused by increases in the ambient air density (2) At t ≤ tb , S increased proportionally to time t At t ≥ tb , S increased nearly proportionally to t (3) CV of swirl injectors and CV of slit injectors have different inlet configurations This result agrees with the report by Hiroyasu et al (1980) for diesel fuel (4) For swirl injectors, the spray velocity coefficient is constant with changing ambient temperature, injection pressure, ambient pressure, and fuel species However, the spray velocity coefficient of slit injectors is affected by changes in ambient pressure 1/2 ACKNOWLEDGEMENT−This work was supported in part by the Industrial Core technology development project of Ministry of Knowledge Economy in Republic of Korea The authors appreciate this financial support REFERENCES Fujimoto, H., Kusano, S and Senda, J (1997) Distribution of vapor concentration in a diesel spray impinging on a flat wall by means of exciplex fluorescence method − In case of high injection pressure SAE Paper No 9702916, 133−144 Hiroyasu, H and Ari, M (1980) Fuel spray penetration 22 91 Copyright © 2010 KSAE 1229−9138/2010/052−18 International Journal of Automotive Technology, Vol 11, No 3, pp 441−445 (2010) DOI 10.1007/s12239−010−0054−9 ANALYSIS OF NECK FRACTURES FROM FRONTAL COLLISIONS AT LOW SPEEDS S.-J PARK , S.-W CHAE and E.-S KIM 1) 1)* 2) Department of Mechanical Engineering, Korea University, Seoul 136-713, Korea National Institute of Scientific Investigation, Seoul 158-707, Korea 1) 2) (Received 23 March 2009; Revised 12 October 2009) ABSTRACT−Neck fracture is a major cause of death in traffic accidents This pattern of injury normally occurs in a frontal collision or overturn of a vehicle This study investigates the case of a neck fracture from a low-speed collision In the examined case, the passenger in the front seat of the car fractured his neck and died He did not have his seatbelt on when the vehicle slipped on a frozen road surface on a downward slope of a hill and impacted into the shoulder of the road at low speed In this type of collision, an occupant’s body will be impacted by the windshield or other interior trim of the car However, in this case, rather unusually, neither body tissue nor fiber remained although the collision involved a broken windshield Thus, the reason for the passenger death was unidentified This study applied the computer simulation package Madymo for analyzing the accident The result of the simulation was that the passenger, who did not wear a seatbelt, moved forward due to inertia The upper part of the passenger then rotated and lifted when the knee contacted with the dashboard By evaluating the structural deformation of the vehicle at the front, we deduced that the collision velocity was 30 km/h Through a computational experiment that was undertaken using Madymo 7.0, NIC was estimated to be 240 m2/s2 This result far exceeded the threshold for neck injuries In particular, in comparison with whiplash injuries, when the passenger's head directly impacts the roof following a rear-end collision, the bending moment through hyperextension of the neck is greatly increased In this study, we concluded that the manner of death was the hyperextension of the neck, as the passenger’s head contacted the roof from underneath KEY WORDS : Vehicle collision, Madymo, Low speed, Neck injury, Traffic accidents INTRODUCTION cervical vertebrae Some reasonable standards have been reported McConnell performed an experiment with seven volunteers and reported that when the rear-end collision was simulated at about 10.9 km/h, the driver was inflicted with a sprain in the cervical vertebrae (McConnell , 1995) Through a BioRID (biofidelic rear impact dummy), Zuby determined the limits of the flexion, the extension angle, and the bending moment at 11 km/h The results obtained by Davis did not exceed 80 and 57 N·m of IARVs (Injury Assessment Reference Values) and discounted the possibility of injury through hyperextension Because the NIC (Neck Injury Criterion) was lower than 15, it was not an important factor at a velocity of 11 km/h (Zuby , 1995) Croft (2002) reported that volunteers complained of pain during the 10-day in-sled test at a NIC value of 11.8 (Croft , 2002) Kaneko quoted NIC in a collision through the following equations: et al Regardless of whether or not a car passenger wears a seatbelt, the cervical vertebrae are the most frequently injured parts of the body in traffic or rear-end collisions In the case examined in this study, the passenger's neck fracture resulted from collision after the car slid at a low speed The case report is as follows While the wife of the passenger was driving him to the hospital to take him for treatment for dyspnea, a chronic disease, her vehicle slipped over an ‘S’-shaped, one-way, downhill road because of icy road conditions and impacted into the left side of the slope Whether or not the slight damage to the front side of the vehicle from the collision ultimately caused the death of the passenger in this case is unknown The criminal investigation section suspected that the driver had intended to camouflage the collision to conceal a homicide in which her husband had been murdered before the collision elsewhere There has been extensive research to clarify the relationship between collisions and neck fractures In particular, volunteers have participated in experiments that examine the sprain of * Corresponding author et al et al o et al et al et al et al NIC = 0.2 × arel( t) + (vrel(t) ) (1) (2) arel(t) = aT – aHead , vrel(t) = vT – vHead In other words, Kaneko referred to NIC as the square of the difference in the velocity and acceleration between the head and the torso and that neck fracture 1 et al e-mail: swchae@korea.ac.kr 441 442 S.-J PARK, S.-W CHAE and E.-S KIM would result when the NIC value exceeded 15 (Kaneko , 2004) However, the methods of previous studies are flawed in that they cannot test the NIC values at higher speeds because of the possibility of neck injury to the volunteers Yoganandan determined the degree of fracture by impacting the neck of a corpse at 6.6 m/s (Yoganandan , 2000) Nightingale obtained the limit of fracture from an experiment on the neck of a corpse The fracture limit with respect to bending was 23.7±3.4 N·m in the flexion direction and 43.3±9.3 N·m in the extension direction from O to C2; furthermore, the upper cervical vertebrae were stronger than the lower cervical vertebrae However, because this experiment was conducted without muscles, these results are somewhat difficult to apply to actual injury analysis (Nightingale , 2002) Recently, international agencies, such as EuroNCAP, have noted that the possibility of injury over the AIS level is likely to be 20% when the moment that acts near the neck exceeds 57 N·m In particular, the result of neck fracture was indicated by experiments that made use of corpses Although many studies have been carried out by automobile companies or universities to reduce the incidence of injuries from accidents, the mechanism of injury is still unclear because these studies investigate the reconstruction of the scene of a real accident By using computer simulation of nine cases of death by collision, Balažic found that when the passenger seated behind the driver does not wear a seatbelt, the possibility of injury to the driver is very high (Balažic , 2006) O'Riordain compared the results obtained from an experiment that used a falling human body at various poses and positions with those from a computer simulation through MADYMO (O'Riordain , 2003) Computer simulations have been used not only for R&D for vehicles but also the reconstruction of accident scenes Cory developed a simulation system for performing in situ surface tests to assess the potential severity of head impacts from falls This assessment was carried out by computer simulation Especially in forensic science, this method of computer simulation plays an important role because we cannot observe the accident or the crime The simulation is useful for reconstructing the crime scene of either a suicide or a murder, defining injuries to the cervical vertebrae, and addressing the perceptual problems of hit-and-run drivers (Cory and Jones, 2006; Lezeau and Nazat, 2004; Ermenc and Prijon, 2005) This case focuses on the clarification of the neck-injury mechanism For low-speed collisions, the neck fracture can be predicted by using MADYMO et al et al et al et al t et al particularly discerned Considering the fracture and the critical reaction in the neck, the most likely cause of death was deduced to be damage to the nervous system in the neck due to multiple fractures of the cervical spine This conclusion was supported by the absence of evidence of trauma, disease, or toxicosis Indications from the condition of the vehicle, the scene of the accident, and the corpse were that the vehicle slid to the left side of the curved line to follow the ‘S’-shaped road The indications of collision of the car were that the front bumper was detached and the understructure of the radiator was slightly deformed However, there was no significant fracture or deformation to the main structure Moreover, no mark was found on the inside of the driver's seat and the passenger seat was pulled back, as shown in Figure In the result of the autopsy for the passenger, fractures were found in the upper (C2-C3), middle (C4C5), and lower (C6-C7) parts of the neck, as shown in Figure Bleeding was identified from the site of the fracture The car crashed on the left side of the hill because of unstable steering conditions on the icy road After slipping to the left side of the road, the vehicle could not return to the correct direction Considering the road shape and condition, it seemed that the vehicle was moving at a et al et al et al et al Figure Collision at the left side of the hill at the end of a downhill road REVIEW OF THE ACCIDENT The scene of the accident was an ‘S’-shaped downhill road The vehicle slipped on the icy road and impacted the left side of the hill The last position of the collision is shown by the arrow in Figure A piece of the front bumper fell off at this location Trauma of the knees and shanks was not Figure Condition inside the vehicle, in which the passenger's seat was pulled down ANALYSIS OF NECK FRACTURES FROM FRONTAL COLLISIONS AT LOW SPEEDS 443 Figure Neck fracture revealed by autopsy low velocity Moreover, the vehicle may have crashed at 30 km/h based on the detached bumper and the marginal deformation of the substructure of the radiator Considering the damage to the vehicle, the car was predicted to have crashed at about 30 km/h However, the exact velocity could not be obtained In any event, given that the passenger did not wear a seatbelt, it was necessary to infer the passenger's behavior from computer simulation and to ascertain the probable cause of full fracture to the neck and of the resulting death APPLICATION TO MADYMO SIMULATION In Figure 2, the seatback is reclined by about 50 degrees, and there is a 30-cm interval between the dashboard and the front part of the seat cushion The distance between the front part of the seat cushion and the ceiling of the vehicle was about 80 cm Figure 4(a) shows the application of computer simulation wherein MADYMO is used with reference to these measurements Madymo 7.0 was used to analyze this accident The car model was remade to fit the appropriate properties (mass, size, etc.) The dummy model was BioRID II, which was 174 cm high in Madymo 7.0 (TNO Automotive, 2008) The elastic properties of the vehicle model were controlled to validate the deformation of the bumper, which was the colliding part of the vehicle The deformation was controlled to 30 cm at 30 km/h in the following equation (Equation (3)), which was presented by Campbell (1974) V = 3.0 + 1.35(Crash) (3) V: collision velocity (mph), Crash: deformation (inch) The weight of the vehicle was 1500 kg, the length of the wheelbase was 2.7 m, and the interval of the tread was 1.6 m In this simulation, we applied these fundamental specifications of the vehicle We then observed how the upper part of the human model behaved in the simulation, where the car speed was varied from km/h to 30 km/h in increments of 2.5 km/h We also examined how the neck was burdened at the instant of the crash Figure Simulation result showing extreme folding of the neck of the human model under the ceiling at 30 km/h RESULTS AND DISCUSSION According to the results of MADYMO, the passenger behavior at 30 km/h is shown in Figure At the moment of collision, the upper part of the human model moved to the anterior seat, as shown in Figure 4(b), and the knee portion came into contact with the dashboard Subsequently, the lower part did not move to the anterior, and the upper part leapt to the ceiling, as shown in Figure 4(c) and Figure At a speed of 10 km/h, although the knee contacted with the frontal dashboard, the effect of the leaping upper body was smaller than at 30 km/h At speeds of 20 km/h and greater, the neck bumps against the ceiling and is significantly pulled back This situation is different from a whiplash injury that results from a momentary impact with the back of the seat in a collision However, whether or not the impact is momentary is 444 S.-J PARK, S.-W CHAE and E.-S KIM uncertain because the upper part of the passenger moves to the anterior and the neck part folds under the ceiling Therefore, the bending moment that is applied near the neck is the most important factor with regard to the Injury Assessment Values Figure presents the bending moment with regard to the velocity If we compare these values with the criteria of EuroNCAP for neck injury, which corresponds to a risk of 20% that the injury is greater than AIS 3, namely, an NIC of 15, we find that the estimated values are much greater than the stipulated values The bending moment increased at speeds above about 12.5 km/h, exceeding 57 N·m, which is the limit of injury in AIS3 The maximum NIC with respect to the changing velocity in Equation (1) is shown in Figure NIC greatly increased at a speed of 12.5 km/h That speed corresponded to the impact moment between the passenger's head and the roof inside the vehicle, which resulted in hyperextension of the neck NIC exceeded the value of the threshold, viz., 15 m2/ s2, which means that the head collided directly with the roof In that event, the passenger could have been injured much more severely than in the case of whiplash injuries during rear impact at low speeds CONCLUSION Figure Maximum injury of the neck at 30 km/h Figure Variation of the bending moment with speed Figure Variation of NIC with speed Computer simulation must include certain assumption when used to reconstruct real situations, such as murders or fatal injuries For example, the characteristics of human models in simulations are not the same as those of cadavers, and it is difficult to represent the clothing or interior of the vehicle Although computer simulations have limitations in reconstructing reality, they can approximately deduce passenger behavior and the cause or the possibility of collision In the result from reconstruction through computer simulation, when the vehicle in question impacted the side of a road in a downhill slide at low velocity, the passenger's upper body was shown to have rotated to the anterior The top portion of the head may have come in contact with the ceiling, and the head might have bent back Therefore, a neck injury could have resulted In the computer simulation, even at 15 km/h, the head part could contact the underside of the ceiling, and the neck was greatly bent at 25 km/h The value of NIC was 240 m2/s2, which was much greater than the threshold of 15 m2/s2 This result shows that the possibility of neck injury is much higher than the case of whiplash injury when the head directly impacts the roof The driver, who recognized the moment of collision, might have resisted the force of impact However, the unconscious passenger, who was seated next to the driver, may have suffered an injury that resulted from the contact of his neck with the ceiling Even at low speed, when the passenger did not wear the seatbelt and reclined his seat and the car collided against an object in front, the upper part of the passenger moved to the anterior and rotated through the collision; as a result, he might have suffered a neck injury In the present case, the passenger in the vehicle did not wear a seatbelt and had laid back his seat We may deduce that during the collision, through inertia, his upper portion moved to the anterior He could have become erect through the contact of his knee with the dashboard, following which, his head would have hit the ceiling or the front-upper end of the interior Based ANALYSIS OF NECK FRACTURES FROM FRONTAL COLLISIONS AT LOW SPEEDS on this deduction, we can conclude that the passenger was injured by hyperextension of the neck The suspicions about his wife were withdrawn, and she was cleared of all charges Normally, every passenger thinks that the seatbelt should be fastened when the car is driven However, passengers may assume that a reclined seat is safer than a seat in the upright position and often not fasten their seatbelts Unfortunately, seatbelts are practically designed to constrain the passenger when the seatback is in the upright position; the seatbelt cannot adhere to the passenger’s body Therefore, a seatbelt system should be developed for ensuring a recumbent passenger’s safety ACKNOWLEDGEMENT−This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (No 2009-0063173) REFERENCES Balažic, J., Prebil, I and Certanc, N (2006) Computer simulation of the accident with nine victims Forensic Science Int., 156, 161−165 Campbell, K L (1974) Energy basis for collision severity SAE Paper No 740565 Cory, C Z and Jones, M D (2006) Development of a simulation system for performing in situ surface tests to access the potential severity of head impacts from alleged childhood short falls Forensic Science Int., 163, 102−114 Croft, A C., Herring, P., Freeman, M D and Haneline, M T (2002) The neck injury criterion: future considerations Accident Analysis and Prevention, 34, 247−255 445 Ermenc, B and Prijon, T (2005) Suicide, accident? The importance of the scene investigation Forensic Science Int 147S S21-S24 Kaneko, N., Wakamatsu, M., Kukushima, M and Ogawa, S (2004) Study of bioRID II sled testing and MADYMO simulation to seek the optimized seat characteristics to reduce whiplash injury SAE Paper No 2004-01-0336 Lezeau, T and Nazat, D (2004) Crime scene analysis: Daily approach in a French research section Forensic Science Int.146S, S29-S30 McConnell, W E., Howard, R P., Poppel, J V., Krause, R., Guzman, H M., Bomar, J B., Raddin, J H., Benedict, J V and Hatsell, C P (1995) Human head and neck kinematics after low velocity rear-end impacts-understanding “Whiplash” SAE Paper No 952724 Nightingale, R W., Winkelstein, B A., Knaub, K E., Richardson, W J., Luck, J F and Myers, B S (2002) Comparative strengths and structural properties of the upper and lower cervical spine in flexion and extension J Biomechanics, 35, 725−732 O'Riordain, K., Thomas, P M., Phillips, J P and Gilchrist, M D (2003) Reconstruction of real world head injury accident resulting from falls using multibody dynamics Clinical Biomechanics, 18, 590−600 TNO Automotive (2008) MADYMO Model Manual Version 7.0 Yoganandan, N., Pinter F A., Stemper, B D., Schlick, M B., Philippens, M and Wismans, J (2000) Biomechanics of human occupants in simulated rear crashes: Documentation of neck injuries and comparison of injury criteria SAE Paper No 2000-01-SC14 Zuby, D S., Vann, D T., Lund, A K and Morris, C R (1995) Crash test evaluation of whiplash injury risk SAE Paper No 95SC17