TECHNICAL PAPERS445 1991 Soichiro Monda Lecture: Energy, Elliciency, and the Environment: Three Big Es 01

TransportationK. J. Springer

459 Natural Gas Fueling 01 a Caterpillar 3406 Diesel EngineG. E. Doughty, S. R. Bell, and K. C. Midkiff

466 Emission Reductions Through Precombustion Chamber Design in a Natural Gas, Lean BurnEngine

M. E. Grane and S. R. King

475 Effects 01 Spark Plug Number and Location in Natural Gas EnginesR. C. Meyer, D. P. Meyers, S. R. King, and W. E. Liss

480 Development 01 the Cooper.Bessemer CleanBurnTM Gas-Diesel (Dual.Fuel) EngineD. T. Blizzard, F. S. Schaub, and J. G. Smith

488 The Effect 01 Injection Timing, Enhanced Altercooling, and Low.Sullur, Low.Aromatic DieselFuel on Locomotive Exhaust Emissions

V. O. Markworth, S. G. Fritz, and G. R. Cataldi

496 Electrically Heated Catalysts lor Cold.Start Emission Contrai on Gasoline. and Methanol-Fueled Vehicles

M. J. Heimrich, S. Albu, and M. Ahuja

502 Coal.Fueled Diesel Engine Development Update at GE Transportation SystemsB. D. Hsu

509 Features and Perlormance Data 01 Cooper.Bessemer Coal.Fueled Six.Cylinder LSB EngineA. K. Rao, E. N. Balles, and R. P. Wilson, Jr.

515 Progress on the Investigation 01 Coal-Water Slurry Fuel Combustion in a Medium.SpeedDiesel Engine: ParI 5-Combustion Studies

B. D. Hsu, G. L. Conler, and Z. J. Shen

522 Injection Characteristics 01 Coal-Water Slurries in Medium.Speed Diesel EquipmentL. G. Dodge, T. J. Callahan, T. W. Ryan, 111, J. A. Schwalb, C. E. Benson, andR. P. Wilson, Jr.

528 Coal-Water Slurry Spray Characteristics 01 a Positive Displacement Fuellnjection SystemA. K. Seshadri, J. A. Caton, and K. D. Kihm

534 Implicit Numerical Model 01 a High.Pressure Injection System (92.ICE.3)A. E. Catania, C. Dongiovanni, and A. Mittica

544 Optimization 01 Injection Law lor Direct Injection Diesel Engine (92-ICE.4)M. Feola, P. Pelloni, G. Cantore, G. Bella, P. Gasoli, and G. Toderi

553 Variable Geometry and Waste.Gated Automotive Turbochargers: Measurements andComparison 01 Turbine Perlormance (92-ICE-9)

M. Capobianco and A. Gambarotta

561 Emissions From Heavy.Duty Trucks Converted to CNG (92.ICE.10)S. G. Fritz and R. I. Egbuonu

568 Progress in Diesel Engine Emissions Contrai (92.ICE.14)M. K. Khair

578 Application 01 Methods lor Determining Total Organic Contribution to Diesel Particulates(92.ICE-16)

M. S. Newkirk

590 Development 01 the Long-Stroke Version 01 the Mitsubishi SU Diesel Engine (92-ICE.17)Y. Nakamura, M. 110, and H. Arakawa

597 Catalytic Contrai 01 NO., CO, and NMHC Emissions From Stationary Diesel and Dual.FuelEngines (92.ICE.19)

R. W. Bittner and F. W. Aboujaoude

ANNOUNCEMENTS

567 Change 01 address lorm lor subscribers

602 Inlormation lor authors

Published Ouarterly by The American Society of Mechanical Engineers

VOLUME 114 . NUMBER 3 . JULY 1992

M. Capo bianco

A. GambarottaIn turbocharging automotive Diesel engines, an effective method to extend theturbine flow range and contrai the boost level is the use of a variable geometryturbine ( VGT): This technique con be very helpful to improve the transient responseof the engine and reduce exhaust emissions. In arder to compare the performanceof variable geometry and conventional waste-gated turbines, a thorough experimentalinvestigation was developed on a test facility at the Department of Energetic En-gineering of the University of Genoa (DINE). Two VG turbines were considered:a variable area turbine ( V A T) and a variable nozzle turbine ( VNT). The VG turbineswerecompared with afixed geometry waste-gated turbine in both steady and unsteadyflow conditions, referring to mass flow and efficiency characteristics.

Department 01 Energetic Engineering,University 01 Genoa,

Genoa, Italy

IntroductionIn recent years the application of exhaust turbocharging to

Diesel engines has become wider and wider, since it allows forhigh power-weight ratios, lowering specific fuel consump,tionand exhaust emissions. However turbocharger matching iS' nota simple task for the designer, especially in the case of auto-moti ve engines, which usually operate under transient condi-tions.

At present most automotive turbochargers afe fitted withfixed geometry nozzleless radiai turbines: In those cases, al-though the tlow range of a defined trim is fairly large, theturbine acts as a fixed restriction, and the pressure drop acrossit varies greatly with the mass tlow rate. Since the turbinepower increases with mass tlow and expansion ratio, the boostpressure curve rises notably with engine speed and load. Forautomotive applications, where the operating conditions oftenvary within a wide range, the engine may bave insufficientboost at low speeds and loads and be overboosted at highspeeds and loads (Watson and Janota, 1982).

In order to control the boost level, a bypass valve ("waste-gate") is usually adopted, but this results in a significant lossof efficiency when operating away from the design point sincea large amount of exhaust gas energy is lost. A more effectivemethod to govern the boost level is the use of a variable ge-ometry turbine (VGT): Changing a suitable controlling areaallows the expansion ratio to be varied, and therefore theturbine power output to be varied (Flaxington and Szczupak,1982). This technique, joined with a proper control strategy,

Contributed by the Internai Combustion Engine Division and presented atthe Energy-Sources Technology Conference and Exhibition, Houston, Texas.January 26-30, 1992. Manuscript received by the InternaI Combustion EngineDivision September 30,1991. Paper No. 92-ICE-9. Associate Technical Editor:J. A. Calano

can be very helpful to improve both steady-state performanceand transient response of the engine, and reduce emissions interms of exhaust smoke (Wallace et al., 1982; Watson andBanisoleman, 1986, 1988; Wallace et al., I 986a, 1986b; Ar-coumanis et al., 1990).

Several variable area systems bave been proposed by re-searchers and manufacturers (Flaxington and SzClupak, 1982;Hirhikawa et al., 1988; O'Connor and Smith, 1988); however,the most effective and reliable solutions seem to be based ona variation of the turbine housing throat area (variable areaturbine, V A T) or on the use of a variable geometry nozzleturbine (VNT).

As a first step of a specific research program on optimizationof turbocharger matching and control, a wide experimentalinvestigation has been developed at the Department of Ener-getic Engineering of the University of Genoa (DINE) (jointlywith the Centro Ricerche Fiat of Orbassano) in order to achievea better knowledge of steady and unsteady flow behavior ofradiai turbines fitted with different variable geometry devices,in comparison with a conventional waste-gate system.

Two Garrett automotive turbochargers were considered, fit-ted with the same turbine wheel but with different regulatingsystems: a variable area turbine (Garrett VA TO25) and a vari-able nozzle turbine (Garrett VNTO25). Steady and pulsatingflow performance was measured on the test facility of theDINE, with reference to different fixed settings of the variablegeometry devices, and was related to those previously obtainedon a quite similar fixed geometry waste-gated turbine (GarrettTBO25) (Capobianco et al., 1990; Capobianco and Gamba-rotta, 1990).

Experimental results afe presented and compared in the pa-per, emphasizing the influence of the control device setting onturbine steady flow performance with particular reference to

JUL Y 1992, Vol. 114 I 553Journal of Engineering for Gas Turbines and Power

Experimental Facility and Investigation ProgramMeasurements were performed on the DINE test rig, fully

described in previous papers (Capobianco et al., 1990; Ca-pobianco and Gambarotta, 1990), which is schematically shownin Fig. l. In the present layout it allows us to develop steadyand pulsating flow tests both on single devices and on complete. intake and exhaust systems of automotive engines.

Two separate feeding lines afe available: Compressed air issupplied by three screw compressors SC (with a total flow rateof 0.65 kg/s at a maximum pressure of 8 bar) or, alternatively,by a centrifugaI compressor DC (with a maximum delivery of2.2 kg/s and a compression ratio of 2.13). Pressure level ineach line can be controlled by a pro per regulating system.

The main feeding line is fitted with an electrical air heaterAH (which allows air temperature up to 400 K) and a pulsegenerator system. This device, developed by the authors, makespossible an effective control of pressure oscillation character-istics, namely frequency, amplitude, and mean value. Thisresult is obtained by mixing two flows, downstream of a smallreservoir R, in one of which pressure pulses afe generated bya rotating valve RV (composed of a cylindrical rotor with adiametral slot revolving inside a stator with inlet and outletports). Different flow area diagrams can be obtained by re-placing the rotor and the stator ports, while pulse frequencycan be adjusted through valve speed in the typical range ofhigh-speed multicylinder engines (10-250 Hz). Dedicated valvesallow the two flow components to mix properly, thus con-

trolling the oscillation parameters. Recent modifications madeit possible to test two entry devices (e.g., twin entry turbo-charger turbines): In this case both entries can be fed withcontrolled pulsating flows, while the phase between the pres-sure pulses can be easily modified (by changing the relativeangular position of the rotating valve rotors).

The experimental investigation on variable geometry tur-bochargers referred to in this paper was developed feeding theturbine through the main line, while the compressor was usedas a dynamometer. In order to vary the compressor power atconstant rotational speed, its inlet pressure was controlled bythe second feeding line (Fig. 1). By this technique it was possibleto investigate turbine characteristics up to higher values of theexpansion ratio, while lowest values were reached by a par-ticular device, which was mounted in piace of the originaicompressor housing (Capobianco and Gambarotta, 1990).

Turbocharger lubrication was ensured by a dedicated circuit:lnlet oil temperature and pressure were controlled and keptconstant at set levels.

Measurements were performed through an automatic dataacquisition system. Pressures and temperatures were evaluatedupstream and downstream of the turbine and the compressorby high frequency response strain gage transducers and byplatinum resistance thermometers. Turbocharger speed androtating valve frequency were measured by inductive probes.Turbine and compressor mass flow rate were evaluated by alaminar flow meter LM and a sharp edged orifice FM, re-spectively.

Data acquisition was governed by an IBM-A T computerthrough an HP-3497 A unito In the case of pulsating flow tests,a digital oscilloscope Gould 4020 (with 2 MHz sampling fre-quency and 4 kbytes internai memory) was used for recordingand storage of lime dependent pressure diagrams at the turbineinlet and exit. Dedicated software has been developed to con-trol the acquisition procedure, calculate turbocharger per-formance parameters, and automatically plot turbine andcompressor characteristics.

The investigation program was developed to analyze steadyand pulsating f'low behavior of turbocharger turbines fittedwith variable geometry devices, and compare their perform-ance with a conventional waste-gate system.

First the variable area turbine of a Garrett V A T025 turbo-charger was considered: In this case the turbine throat areacan be controlled by means of a single biade moving insidethe housing (Flaxington and Szczupak, 1982; Hirhikawa et al.,1988) (Fig. 20). Since the turbocharger was fitted with a waste-gale valve, this was kept closed in any operation condition.

A Garrett VNT025 turbocharger turbine with nozzle ringwas also considered: Its stator area can be varied by simul-taneously rotating twelve blades arranged around the periph-eral portion of the turbine housing (Flaxington and Szczupak,1982; Hirhikawa et al., 1988; O'Connor and Smith, 1988) (Fig.2b). No waste-gate valve was fitted on this turbocharger.

In order to compare properly the different variable geometrysystems, both turbines were quite similar (as regard wheels andrelated aerodynamic parameters, i.e., trim) to the waste-gatedfixed geometry turbine (Garrett TB025) tested in former in-

Nomenclature

f = frequencyK = ratio between average pulsat-

ing and steady fIow valuesM = mass fIow raten = rotational speedp = pressurep = powert = time

4 = turbine exitm = mechanicalM = mass flow rate

NSF = nonsteady flow conditionsSF = steady flow conditionsp = powert = turbineT = stagnation conditionsT = torque

T = temperature, pulse period6. = amplitudeE = expansion ratio'I = efficiency

'l/ = 'I,' '1m

T = torque

Subscripts3 = turbine inlet

554/ Vol. 114, JUL Y 1992 Transactions 01 the ASME

2.0

1.6

(a)

1.2

0.8

0.80

0.60

0.40

0.20-1.0 1.4 1.8 2.2 2.6

Fig. 3 Equivalence condition 01 mass flow characteristic curves andcorresponding turbine elliciency

vestigations (Capo bianco et al., 1990; Capo bianco and Garn-barotta, 1990).

The performance of variable geometry vubines was evalu-ated by measuring several constant speed characteristics atdifferent fixed settings of the regulating systems, expressedthrough the position of the relevant push rodo To define une-quivocally this pararneter and take up slack, the originai spring-. diaphragm actuator was replaced by a simple anchor plate,

which allowed the rod to lock in the considered positions. Inboth cases the setting of the regulating system was related tothe linear displacement of the push rod (measured with ref-erence to two gaging surfaces) and was defined as a percentageof the total displacement of the rod, ranging from O to 100percent when varying the flow area from the minimum to themaximum value.

1.0 1.4 1.8 2.2 2.6 3.0

Fig. 4 Steady mass flow characterlstics for the Garrell VATO25

Analysis and Comparison of Experimental Results

Steady Flow Tests. Steady flow performance of both tur-bines was evaluated with reference to constant speed charac-teristics (in term of turbine speed factor n/.JT;) at differentfixed settings of the variable geometry systems. As in previousinvestigations, four values of the speed factor were considered,ranging from 2500 to 5500 rpm /~.

In arder to account far significant positions of the variablegeometry devices, "equivalent" settings were defined far bothturbochargers by finding out the equality condition of massflow curves, assuming as a reference the fixed geometry turbine(Garrett TBO25) with waste-gate valve fully closed (Capo biancoet al., 1990; Capo bianco and Gambarotta, 1990). A very goodagreement was noticed at positions of 79.2 percent far theVA T and 56.8 percent far the VNT, respectively, as is shownin Fig. 3 far a speed factor of 3500 rpm /--IK. However thiscongruence was not found referring to the efficiency ,,/ (de-fined as turbine isentropic efficiency Il, multiplied by turbo-charger mechanical efficiency 11m, Fig. 3), which was alwayslower far the variable geometry turbines. Similar results wereobtained far different values of turbine speed factor. Thisparticular behavior will be analyzed later.

In arder to extend performance evaluation of the two tur-

bines, steady flow curves were also measured for differentsettings, equal to O, 25, 50, 100 percent for the V A T, and toO, 25, 45, 100 percent for the VNT. Mass flow and efficiencycharacteristics were defined for each setting. Figures 4 and 5show constant-speed steady mass flow characteristics for thetwo variable geometry turbines.

Mass flow rate varies in a wide range for both turbochargers:At a given pressure ratio the corresponding turndown is alwayshigher than 50 percent (while for the fixed geometry waste-gated turbine it was slightly lower).

Superimposition of characteristic curves shows a higher sen-

Journal of Engineering for Gas Turbines and Power JULY 1992, Vol. 114/555

2.0

1.6

1.2

0.4 I .

1.0 1.4 1.8 2.2 2.6 3.0

Fig. 5 Steady mass flow characteristics for the Garrett VNTO25

1.2

8

4

o

0.6o 20 40 60 80 100

Fig. 6 Mass flow rate versus VG setting !or the Garrett VATO25 andVNTO25

ever, in the case of the flXed geometry turbine, overall effi-ciency TI: drops when the control devi ce (i.e., the waste-gatevalve) is operated (Capo bianco et al., 1990).

Both for V A T and for VNT the highest efficiency was ob-tained at an intermediate setting (maximum values of TI( wereabout 0.55 and 0.60 for the variable area and variable nozzleturbine, respectively), while it decreased moving toward fullyopen and fully closed settings. A similar peculiar effect waspointed out by Wallace et al. (1982) and by Watson and Ban-isoleman (1986). This result may be due to the fact that forboth turbines the scroll geometry is optimized for a particularvelocity distribution achieved at an intermediate setting of thecontrol device (Hirhikawa et al., 1988), which is the designpoint of the turbine housing. Besides, referring to the variablenozzle turbine, the flow in the intermediate configuration maybe guided more closely by the moving vanes than in the fullyopen or fully closed setting (Wallace et al., 1982). The existenceof a definite design operating condition may be related to thenonlinear reaction forces on the push rod, which were observedduring experimental investigation.

In order to extend performance compari so n and confirmprevious remarks, optimum efficiency values for the differentregulating systems considered in the investigation were plottedversus nondimensional mass flow rate (referred to the maxi-mum value of each turbine) at constant turbocharger speed(Fig. 7). Results obtained for the two variable geometry andfor the fixed geometry turbine were not unexpected. The high-est efficiency is achieved by the fixed geometry turbine in anarrow operating range, while it drops at high mass flow rateswhen the waste-gate valve has to be opened. The variabl6 nozzleturbine shows better efficiency in a wider range than the vari-able area one, except for highest mass flow rates. This effectmay be related to different losses induced by the two controlsystems, with particular reference to the scroll design and toits interactions with the flow pattern (which depends on vari-able geometry settings) (Hirhikawa et al., 1988).

Experimental data obtained in steady flow operation rep-resent an important starting point to define optimized controlstrategies to take full advantage of variables geometry turbineson their whole operating range. These results will be very usefulto define proper schedules (e.g., the optimal setting for anyoperating condition, the initial opening condition and the rateof change of geometry setting), which will be tested and verifiedthrough direct on-engine measurements in future developmentsof the investigation.

sitivity (defined as the ratio of change induced in the output,i.e., mass flow rate, to the related input change, i.e., push rodposition) of the regulating systems in the field of lower valuesof turbine flow area, particularly for the variable nozzle turbine(Fig. 5). This effect is apparent in Fig. 6, in which mass flowrate is represented versus system setting for constant values ofturbine speed and expansion ratio. This peculiar behavior, notcompletely unexpected, may be caused by the geometri c char-acteristics of the variable geometry devices, and must be con-sidered in the design of the actuator system and in the definitionof proper control schedules in order to take full advantage oftije increased flow range (Watson and Banisoleman, 1986,1988; Wallace et al., 1986a, 1986b; Arcoumanis et al., 1990).

For both variable geometry turbines experimental resultsshowed a decrease of maximum efficiency 1// compared withthe fixed geometry values. This drawback may be explainedas a consequence of higher losses induced by variable geometrysystems: most likely incidence losses, increased surface friction(due to the additional walls), leakage losses through clearancesbetween vanes and contour walls and, especially in the case ofthe variable area turbine, wake losses behind guiding vane(Flaxington and Szczupak, 1982; Hirhikawa et al., 1988). How-

556/ Vol. 114, JULY 1992 Transactions of the ASME

that in several operating conditions pulse amplitude at theturbine exit is not negligible. This peculiar behavior should beconsidered in future investigations, and particularly in the de-velopment of turbocharged engine simulation models, whichtake account of wave propagation phenomena in the exhaustsystem.

Unsteady Flow Tests: Average Turbine Performance.Starting from unsteady tlow measurements, average perform-ance of both VG turbines was evaluated and compared withsteady tlow results. It is important to point out that the def-inition of suitable methods of comparison is a very difficulttask. This problem is substantial for several devices of the in-take and exhaust systems of automotive engines, since theyoften operate in unsteady tlow conditions. On the contrary,performance characteristics determined on the basis of steadytlow tests afe usually available and empiric correction factorsafe introduced to take into account tlow unsteadiness.

Different procedures for relating steady and average pul-sating performance of automotive turbocharger turbines havebeen presented and discussed in recent years in the open lit-erature (Wallace et al., 1969; Benson, 1974, Kosuge et al.,1976; Zinner, 1978; Shamshi, 1979; Winterbone et al., 1991).If a comparison between the two tlow conditions is accepted,performance analysis developed at the same level of turbinemean expansion ratio is an easy and quick procedure, since itrequires only limited experimental information (i.e., inlet andoutlet pressure diagrams for calculating turbine average ex-pansion ratio) for its application.

This method was applied by the authors in previous papers(Capobianco et al., 1990; Capobianco and Gambarotta, 1990),referring to the ratios KM and KT between mean pulsating massflow and torque and the corresponding steady flow values atthe same average expansion ratio. This procedure was followedin a first step of the present investigation, in order to try topoint out the intluence of pulse inlet amplitude and frequencyon the performance of the Garrett V A T and VNT. This waspossible since the experimental facility allowed tests with con-trolled values of the main parameters of the pulsating tlow atthe turbine inlet.

In order to avoid errors due to a possible nonuniform pres-sure distribution or to the different characteristics of measuringsystems, the same pressure tappings and transducers were usedboth in steady and pulsating flow conditions. Transducers weremounted close to the duct wall and an appropriate geometryof tappings was chosen to reduce the passage effects (Winter-bolle et al., 1991).

It is important to note that the same measuring and proc-essing procedure of pressure data should be used in steady andpulsating flow conditions with this arrangement: In fact, steadyflow pressure diagrams, especially at the turbine entry, wereaffected by high- frequency oscillations (the amplitude of whichproved to be not negligible) , probably due to the impellerblades. The use of a typical steady flow measuring process,instead of recording and integration of pressure signals, mightcause errors in the definition of pressure levels of the referencesteady flow characteristics.

Average mass flow (KM) and torque (KT) factors were cal-culated for the V A T and VNT in the considered operatingconditions. Figures 9 and IO show the effect of pulse amplitudeand frequency on the performance of the variable geometryturbines, respectively for mass flow and torque parameters.

The analysis of the results provided by this method of com-parison seems quite difficult: However, a generai trend to areduction of mass flow factors when increasing pulse ampli-rude, at constant frequency and mean inlet level, is apparentfor both the V A T and VNT. Similar behavior was observedalso in the case of the fixed-geometry turbine in a previousstage of the study and was confirmed by a subsequent exper-imental investigation, developed on the V A T, extended to a

Unsteady Flow Tests: Pressure Diagrams. In order to com-pare unsteady flow behavior of fixed and variable geometryturbines, measurements were developed with reference to the"equivalent" settings (79.2 percent for the V A T and 56.8 per-cent for the VNT) at a constant nondimensional speed (3500rpm/.JK). Amplitude, fr~quency, and mean value of pressurepulses at the turbine inlet were controlled through the pulsegenerator system. As in previous investigations (Capo biancoet al., 1990; Capo bianco and Gambarotta, 1990), with refer-ence to real on-engine conditions, three different levels of pulseamplitude élnd frequency (0.30, 0.50, 0.66 bar and 50, 86.6,105 Hz, respectively) were considered, for a constant meaninlet pressure of 1.4 bar.

Static pressure diagrams at the turbine inlet P3 (t) and outletP4(t) were recorded, while measured values of temperature,mass flow, and rotational speed were assumed as average levelsover the pulse period (due to the low response characteristicsof the relevant metering devices). For both turbines mean torquewas evaluated on thè basis of measurements on the coupledcompressor.

Experimental pressure diagrarns were compared with thoseobtained for the fixed geometry turbine: An example is re-ported in Fig. 8. As noticed in previous investigations, pulseshape at turbine inlet P3 (t) was mainly influenced by the forc-ing frequency: In the case of variable nozzle turbine differentpressure waveforms were often found, especially at higherpulse frequency.

Pressure signals at turbine exit P4(t) showed different al-terations for each turbocharger (Fig. 8). With reference to thevariable area turbine, a high frequency disturbance in exitpressurediagramp4(t) was found inevery operatingcondition:This result might be related to the wake flow induced by themoving vane (Hirhikawa et al., 1988) and to vibrational mo-tions of impeller blades, which may be increased due to asym-metri c entry conditions caused by variable area system(Flaxington and Szczupak, 1982).

In the case of the variable nozzle turbine, pulse shape mod-ifications with frequency afe apparent both for P3 (t) andp4(t):Higher amplitudes of P4 (t) diagrams were found in every testedcondition. These effects might be connected to the presenceof wakes induced by each\of the nozzle vanes, as theorized byO'Connor and Smith (1988), but they can hardly be explainedwithout careful measurements in the inside of the turbine hous-Ing.

On this subject it would be interesting to deepen the inves-tigation in order to analyze the influence of variable geometrysetting on the propagation of pressure waves through the tur-bocharger turbine. Unlike the results obtained on the fixed-geometry turbine, in this case pressure diagrams showed clearly

Journal of Engineering for Gas Turbines and Power JULY 1992, Vol. 114/557

05

00

0.95

0.90

0.850.3 0.5 0.7 0.3 0.5 0.7

1.05

1.00

0.95

0.90

0.8550 80 110 50 80 110

Fig. 10 Comparison between average pulsating and steady flow turbinetorque

Garrett VAT025 Setting 79.2%n~= 3500 rpm/VK; f = 86.67 Hz

P = 1.4 bar3

K=-.. at ~so- =é'so-

02

0.98

0.94

0.90

wider range of pulse amplitude values (Fig. Il). In the authors'opinion the link between mass flow factor and pu1se amplitudemight be explained as a quasi-steady flow effect related to thevariable slope of turbine steady flow curves (Capobianco andGambarotta, 1990).

As in the case of previous investigations performed on fixedgeometry turbines (Capo bianco et al., 1989; Capobianco andGambarotta, 1990), the influence of pulse frequency on turbineunsteady performance wasn't evident. This is probably relatedto the observed modifications of pressure diagrams with fre-quency, due to the wave action in the inlet and outlet ducts.

Values of KM and KT factors calculated far the VNT weregenerally lower than the corresponding ones far the V A T. Asa hypothesis, this may be related to the different pressurediagrams measured far the two variable geometry turbines,and particularly to the higher outlet amplitudes found far theVNT.

Torque factor levels (Fig. lO) resulted in the same range ofmass flow values, but any correlation with pulse amplitudeand frequency is difficult to find. This may be a limitation ofthe method of comparing turbine performance at the sameaverage expansion ratio.

More suitable procedures to relate steady and average pul-sating performance of intake and exhaust devices of auto-alotive engines bave to be developed: A dedicated study onthis subject, based on the comparison of steady and meanpulsating parameters at the same inlet energy level, is nowbeing developed at DINE, the results of which will be discussedin a future papero

In a first step of this investigation, turbine power in steadyand pulsating flow conditions was compared for the sameaverage nondimensional mass flow levels (related to the meaninlet energy flow rate). In the case of equal average temperatureat the turbine inlet, the ratio K;' between mean pulsating andsteady turbine power (expressed in nondimensional form) is

0.860.0 0.2 0.4 0.6 0.8 1.0

Fig.11 Influence 01 pulse amplilude on average mass flow performance01 VAT

558 I Voi 14. JUL Y 1992 Transactions 01 the ASME

Garrett VA TO25 Setting 79.2%n/l/T.:;= 3500 rpm/VK: f = 86.67 Hz

P3= 1.4 bar

~ lP3 ~P3

ot

~ NSF 'so-

28

20

1.12

04

0.96O.Q 0.2 0.4 0.6 0.8 1.0

Fig. 12 Comparison 01 VAT steady and pulsating power at constantaverage mass lIow rate

geometry systems were apparent. In both cases steady flowresults showed a wide mass flow operating range, with limitedeffects on efficiency (particularly for the VNT). Although peakefficiencies were lower in comparison with the fixed geometryturbine (because of higher losses induced by the control de-vices), better efficiency values were clearly achieved with ref-erence to the whole operating range.

For both turbines the existence of a definite design pointfor the housing was apparent (probably due to the influenceof the fixed scroll area distribution), since the highest effi-ciencies were obtained at intermediate settings of the variablegeometry systems. This peculiar effect seems to agree with theresults given by other authors (Wallace et al., 1982; Watsonand Banisoleman, 1986).

In addition, turbine unsteady flow behavior was investigatedby controlling frequency, amplitude, and mean value of inletpressure pulses. For an "equivalent mass flow setting" of thecontrol system, pressure diagrams at the turbine inlet and outletand average turbine performance were analyzed.

Pulse shape at the turbine inlet was mainly influenced bythe frequency, although somewhat affected by the variablegeometry system (particularly for the VNT). In both cases,however, the presence of moving vanes seemed to introducesignificant oscillations in the pressure diagrams at the turbineoutlet. This effect, which may be related to wake flows inducedby the vanes, doesn't agree with the common hypothesis ofconstant pressure downstream of a turbocharger turbine. Adeeper investigation on the influence of system setting on thepropagation of pressure waves through the turbine would behelpful in the development of theoretical simulation models.

A verage pulsating performance of the V A T and VNT wasfirst analyzed and compared with steady flow results at thesame mean expansion ratio. The link between turbine per-formance and pulse characteristics (amplitude and frequency)was not evident, except for the influence of pulse amplitudeon mass flow factor KM (which showed a quasi-steady trendto a reduction of KM when increasing pulse amplitude).

Calculated coefficients were generally lower for the VNTturbocharger: This is probably connected to the different shapeand amplitude of upstream and downstream pressure diagramsmeasured for the vaned turbine.

In order to overcome the outlined drawbacks, a dedicatedstudy on the methods for comparing steady and pulsating flowperformance of intake and exhaust devices of automotive en-gines has been started and is now in progress at DINE. Theresults of the first step of this study afe encouraging to a deeperfuture investigation.

The lise of proper contro l strategies to take CulI advantageof VGT benefits seems to be essential. In this direction enginesimulation models and experimental investigations afe usefultools for the design of the control systems in order to improvesteady and transient engine performance and reduce emissions.

This study represents an essential prerequisite to the pro-gression of deeper investigation on the optimization of VGT-engine matching. On the basis of experimental data presentedin the paper, a simulation model of a turbocharged Dieselengine is being developed by the authors: This will allow theanalysis with reference to engine performance to be extendedand, through dedicated experimental program, to deCine andverify optimized control systems and strategies.

re presentati ve of the ratio of turbine specific work in the twoflow conditions. Figure 12 shows the first results of this pro-cedure for the Garrett V A T turbocharger, referring to therelationship between the factor KI- and pulse amplitude at theturbine inlet, for constant values of other operating param-eters. The coefficient KI- proved to be generally above one,rising with pulse amplitude: This might be explained as anincrease of turbine specific work in pulsating flow conditions,related to flow unsteadiness.

These first results afe encouraging to further future devel-opment of the investigation on methods for comparing steadyand average pulsating performance of automotive engine de-vices.

Quasi-steady-flow (q.s.f.) calculations of average pulsatingturbine performance were also developed for both V A T andVNT. The relevant results generally confirmed the ones ob-tained in a previous stage of the investigation far the fixedgeometry Garrett TBO25 turbine (Capobianco and Gamba-rotta, 1990). Calculated performance resulted in good accord-ance with mean experimental values.

However, it is interesting to note that the differences betweenaverage q.s.f. levels and mean experimental pulsating per-formance resulted, in the case of small automotive turbo-charger turbines, in the same amount of the shifting betweenmeasured steady and pulsating flow values.

AcknowledgmentsThe authors would like to thank Garrett S.A.-France, and

particularly Mr. Luciano Bernardini, far the technical supportand assistance given during the development of the experi-mental activity.

The authors also wish to thank the Centro Ricerche Fiat,Orbassano, far their permission to publish this paper.

Conclusions and Future DevelopmentsA thorough experimental investigation has been developed

in order to define and analyze steady and unsteady flow be-havior of two different variable geometry turbocharger tur-bines (V A T and VNT), compared with a similar fixed geometryturbine.

As regards turbine performance, the benefits of variable

Journal of Engineering for Gas Turbines and Power JULY 1992, Vol. 114/559

ReferencesArcoumanis, C., Chan, S. H., and Bazari, Z., 1990, "Oplimisation of the

Transient Performance of a Turbocharged Diesel Engine Using Turbochargingand Fuel Injection Controls," Seminar on "Engine Transient Performance,"IMechE, London.

Benson, R. S., ]974, "Nonsteady Flow in a Turbocharger Nozzleless RadiaiGas Turbine," SAE Technical Paper No. 740739.

Capobianco, M., Gambarolta, A., and Cipolla, G., ]989, "Innuence of thePulsating Flow Operalion on Ihe Turbine Characteristics of a Small InternaiCombustion Engine Turbocharger," presented al the 2nd International Con-ference on Ihe Smallinternai Combustion Engine, Paper No. C372/019, IMechE,London.

Capobianco, M., and Gambarolta, A., 1990, "Unsteady Flow Performanceof Turbocharger Radiai Turbines," presented al the 41h International Confer-ence on Turbocharging and Turbochargers, Paper No. C405/017, IMechE, Lon-don.

Capobianco, M., Gambarotta, A., and Cipolla, G., 1990, "Effect of InletPulsaling Pressure Characteristics on Turbine Performance of an AulomotiveWastegated Turbocharger ," presented at the SAE International Congress andExposition, Paper No. 900359, Detroit.

Flaxington, D., and Szczupak, D. T., 1982, "Variable Area RadiallnnowTurbine," presented al Ihe 2nd International Conference on Turbocharging andTurbochargers, Paper No. C36/82, IMechE, London.

Hirhikawa, A., Okazaki, Y., and Busch, P., 1988, "DevelopmentsofVariableArea Radiai Turbines far Small Turbochargers," presented al the SAE Inter-nalional Congress and Exposition, Paper No. 880120, Detroit.

Kosuge, H., Yamanaka, N., Ariga, I., and Walanabe, I., 1976, "Performanceof Radiai Flow Turbines Under Pulsaling Flow Condilions," ASME JOURNALOF ENGINEERING FOR POWER, Vol. 98, No. I, pp. 53-59.

O'Connor, G. F., and Smith, M. M., 1988, "Variable Nozzle Turbochargersfar Passenger Car Applications," presented al the SAE International Congressand Exposilion, Paper No. 880121, Detroit.

Shamsi, S. S., 1979, "Eslimaling Ihe Innuence of Pulsaling Flow Conditionson Ihe Performance of a Turbine," SAE Technical Paper No. 790068.

Wallace, F. J., Adgey, J. M., and Blair, G. P., 1969, "PerformanceoflnwardRadiai Flow Turbines Under Non-sleady Flow Conditions," Proc. I. Mech. E.,Vol. 184, PI. I, No. IO.

Wallace, F. J., Baghery, A., and Ziriali, M. R., 1982, "Variable GeomelrYTurbocharging far Transporl Engines," presenled al Ihe 2nd Inlernalional Con-ference on Turbocharging and Turbochargers, Paper No. C38/82, IMechE,London.

Wallace, F. J., Roberls, E. W., and Howard, D., 1986a, "Variable GeomelryTurbocharging-Oplimizalion and Contrai Under Sleady SIate Condilions,"presemed al Ihe 3rd Imernalional Conference on Turbocharging and Turbo-chargers, Paper No. C97/86, IMechE, London.

Wallace, F. J., Anderson, K., and Howard, D., 1986b, "Variable GeomelrYTurbocharging-Comrol Under Transienl Condilions," presemed al Ihe 3rdImernalional Conference on Turbocharging and Turbochargers, Paper No. C98/86, IMechE, London.

Walson, N., and Janola, M. S., 1982, Turbochorging the InternolCombustionEngine, McMillan Press, London.

Walson, N., and Banisoleman, K., 1986, "Performance of a Highly RaledVehicle Diesel Engine Wilh a Variable GeomelrY Turbocharger ," presemed alIhe 3rd Imernalional Conference on Turbocharging and Turbochargers, PaperNo. C103/86, IMechE, London.

Walson, N., and Banisoleman, K., 1988, "A Variable-Geomelry Turbochar-ger Contrai Syslem far High OUlpUI Diesel Engines," presemed al Ihe SAEImernalional Congress and Exposilion, Paper No. 880118, Delroil.

Wimerbone, D. E., Nikpour, B., and Frosl, H., 1991, "A Comribulion IOthe Underslanding of Turbocharger Turbine Performance in Pulsating Flow,"presemed al Ihe Conference on Internai Combustion Engine Research, PaperNo. C433/011, IMechE, London.

Zinner, K., 1978, Superchorging 01 Internai Combustion Engines, Springer-Verlag, Berlin.

Transactions of the ASME560 I Vol. 114, JUL Y 1992