International journal of automotive technology, tập 11, số 6, 2010

The new cooling system provides more flexible control of the coolant flow rate and engine temperature, which previously relied strongly on engine driving conditions such as load and spee

ACTIVE COOLANT CONTROL STRATEGIES IN AUTOMOTIVE ENGINES

K B KIM 1) , K W CHOI 1) , K H LEE 1)* and K S LEE 2)

1)Department of Mechanical Engineering, Hanyang University, Gyeonggi 426-791, Korea

2)Department of Mechanical Engineering, Hanyang University, Seoul 133-070, Korea

(Received 12 August 2009; Revised 24 May 2010)

ABSTRACT− The coolant flow rate in conventional cooling systems in automotive engines is subject to the mechanical water pump speed, and high efficiency in terms of fuel economy and exhaust emission is not possible given this limitation A new technology must be developed for engine cooling systems The electronic water pump is used as a substitute for the mechanical water pump in new engine cooling systems The new cooling system provides more flexible control of the coolant flow rate and engine temperature, which previously relied strongly on engine driving conditions such as load and speed In this study, the feasibility of two new cooling strategies was investigated using a simulation model that was validated with temperatures measured in a diesel engine Results revealed that active coolant control using an electronic water pump and valves substantially contributed to a reduction of coolant warm-up time during cold engine starts Harmful emissions and fuel consumption are expected to decrease as a result of a reduction in warm-up time.

KEY WORDS : Flowmaster, Cooling, Warm-up, Emission

1 INTRODUCTION

Research is being conducted with the goal of fulfilling

stringent emission standards and improving the fuel economy

of automotive engines Research trends are two-fold: an

optimization of the combustion process and renovation of

the cooling system The homogeneous charge compression

ignition (HCCI) combustion technique has shown

promi-sing results in terms of near-zero NOx and PM emissions

A particular variation of this technique, premixed charge

compression ignition (PCCI) combustion, has recently drawn

substantial attention (Noguchi et al., 1979; Onishi et al.,

1979; Najt and Foster, 1983) Optimizing the geometry of

the high thermal-efficiency direct injection (DI) diesel engine

is also a strategy for better combustion performance and

lower emissions

Recent trends to increase the power output of an engine

with a turbocharger or supercharger are increasing the

demand on the capacity of engine cooling systems more

than ever Several efforts have been made to handle the

increased heat load on the system with innovative cooling

strategies such as reverse cooling, split cooling, and nucleate

boiling cooling (Ap et al., 2003, Ap and Tarquis, 2005;

Kruger et al., 2008), while reducing the cooling system

size and weight A heat accumulator can average out peak

heat loads so that an unnecessarily large cooling inventory

can be reduced in volume and weight (Vetrovec, 2008)

This system is also advantageous for reducing engine

warm-up time, contributing to a significant decrease in harmful

emissions during cold starts A number of researchers havealso proposed replacing the conventional water pump with

an electrically-driven pump that can actively control thecoolant flow rate based on the optimum driving temper-ature (Vegenas et al., 2004; Hnatczuk et al., 2000; Page et

al., 2005; Cho et al., 2005; Cho et al., 2007; Chalgren.,2004; Chanfreau et al., 2003; Kim et al., 2009) In a typicalautomotive engine cooling system, the water pump coupled

to the engine is driven by a crankshaft Thus, the pumpspeed and coolant flow rate are governed by the enginespeed, which gives rise to unavoidable parasitic losses Themechanical cooling pump cannot supply sufficient flow forcabin heating at idle conditions because of its low effici-ency A high rate of revolution of the crankshaft resultingfrom rapid acceleration is a direct cause of excessive enginecooling The pump must be independent of the engine foroptimal operation so that both the amount of heat released

by the engine to the coolant and the desired temperature ofthe engine can be controlled more precisely Application ofthis concept would reduce parasitic losses in engine powerand downsize the engine cooling system In addition, activecontrol of the coolant flow rate using an electronic waterpump has several advantages such as high coolant temper-ature control and fast warm-up and post-cooling, and it isconducive to a reduction in fuel consumption and harmfulemissions One concern with using an electrically-drivenpump is the high voltage requirement, around 42 V, to drivethe electric motor Fortunately, electronic pump and actuatortechnology has improved considerably, and research in thisarea is ongoing, so this difficulty is expected to be resolved

in the near future (Chalgren, 2003)

Two strategies for more flexible control of the coolant

temperature and flow rate using an electronic water pump

and valves were investigated in this study Similar strategies

have been studied to replace a wax-type thermostat with a

valve to control the coolant flowing through radiator more

efficiently However, there have been only rare efforts to

prevent the coolant from flowing through an oil cooler and

cabin heater during the warm-up period By reducing the

warm-up time, the proposed strategies are expected to

improve engine performance in terms of fuel savings and

emissions penalties With these goals in mind, a simulation

was carried out to explore how the innovated cooling

system could contribute to shortening the warm-up period

during a cold engine start

2 MODELING AND VALIDATION

To investigate the innovative cooling strategies, a diesel

engine and its cooling circuit were modeled using

one-dimensional CFD code software (Flowmaster 2) The

prototype used for modeling in this study was the 2.7L

HSDI diesel engine test rig, as shown in Figure 1, which

consists of a water cooler substituted for the radiator found

in real vehicles, a water pump, a wax-type thermostat, an

oil cooler, a heater core, and an emission sampling system

(Horiba MEXA-7100DEGR)

The specifications of the engine are also summarized in

Table 1, along with experimental conditions The coolanttemperature at the engine can be higher than the typicalboiling temperature of water at atmosphere because thepressure inside the coolant circuit is higher than ambientpressure All compartments in the test rig were designedwith components provided by the software, and the requiredinput data for each component were measured or provided

by manufacturers A schematic of the engine test rig and asimulation model corresponding to the test rig are shown inFigure 2

The simulation model was validated with the oil ature and coolant temperatures measured using a thermo-couple at four different positions in the cooling circuit.Figure 2 shows the thermocouple positions: at the engine,before and after the oil cooler, and after the cabin heater.This validation study was performed for the two enginespeeds of 2000 and 3000 rpm The temperatures measured

temper-at the four positions, and the oil tempertemper-ature temper-at these twospeeds fell within 10% of the error ranges

Figure 3 shows temperature results obtained from theexperiment and simulation at 3000 rpm for comparison ofthe two methods The simulation results were in excellentagreement with the experimental results measured at thefour positions but did not agree with the oil temperature.The overall deviations between the simulated and experi-

Table 1 Specifications of the engine employed in this study

Coolant temperature at engine 85~105°C

Figure 1 Engine test rig used in the experiment

Figure 2 (a) Schematics of the engine test rig and (b) asimulation model corresponding to the engine test rig used

in the experiment P, F, q, and ω in the model are controlicons The spots shown in (a) are the locations where thethermocouples are installed

mental results for the cases of 2000 and 3000 rpm werearound 9 and 4%, respectively, which provides confidence

in the simulation model The simulation errors for 2000and 3000 rpm are summarized in Table 2

3 EXPERIMENTAL RESULTS FOR EMISSIONS AND BSFC

A mechanically-driven water pump is coupled to the engine,which makes active control of the coolant flow rate im-possible An unnecessarily high amount of coolant removesheat from the engine, deteriorating combustion efficiency.Therefore, control of the coolant flow rate is crucial for

Figure 3 Comparison of oil and coolant temperatures

obtained by experiment and simulation for a pump speed

operation of the engine at the optimal temperature Figure 4

shows the effects of various coolant temperatures on

ex-haust emissions and fuel consumption The data were

obtained under steady state conditions while the coolant

temperature was increased from 90 to 105oC A cooler was

utilized to change and maintain the coolant temperature for

the experiment The most significant fact is that the

emi-ssions of THC and CO were reduced when the coolant

temperature was higher than 85oC Compared to the data at

85oC, THC and CO emissions at 105oC were reduced by

10% and 4%, respectively In addition, the BSFC was

re-duced by approximately 3% The higher coolant temperature

increases the in-cylinder temperature, leading to more

com-plete combustion of the fuel; thus, THC and CO emissions

were shown to decrease while the BSFC improved to some

degree However, NOx emissions were observed to increase

because the higher coolant temperature contributed to an

increase in the combustion temperature

Another significant fact demonstrated in Figure 4 is how

much the emissions and fuel consumption could be

improved using a coolant temperature control strategy

Frequent changes in the coolant temperature resulting from

passive control of the coolant flow rate are a direct cause of

deterioration of the emission and fuel consumption

charac-teristics of an engine Based on Figure 4, when the coolant

temperature is maintained at 95oC, the engine performance

in terms of emissions and BSFC is likely to be the optimum

The test rig was used to show the effect of the coolant

temperature on the emissions and fuel consumption and to

validate the simulation model with temperature

measure-ments However, it could not be used to show the reduction

in the warm-up time due to two electrically-controlled

valves because it is not equipped with the valves Building

a test rig with the valves will be a future study, and the

simulation was performed to investigate the effects of the

aforementioned strategies on the reduction in the warm-up

time

4 SIMULATION RESULTS AND DISCUSSION

Two engine cooling concepts with the goals of improving

fuel consumption and emissions were investigated using

the validated computational model of an engine cooling

system As discussed above, the coolant temperature must

increase as quickly as possible, especially during the cold

start engine warm-up period A low coolant temperature is

a direct cause of incomplete combustion, resulting in high

exhaust emissions and high fuel consumption Approaches

to decreasing the warm-up time begin with the idea of

reducing the coolant flow rate or quantity during the

warm-up period

A zero coolant flow rate can be achieved by pausing the

operation of the water pump The mechanical water pump

in a typical cooling system is coupled to the crank shaft by

belts and therefore subject to the engine operating speed

An electronic water pump is independent of engine

operation, which makes it possible to more flexibly controlthe coolant flow rate When the rotation speed of the waterpump was assigned to zero rpm in the simulation, theengine temperature increased quickly, as shown in Figure

5 The warm-up time was reduced by approximately 25%using this strategy

One problem with using the electrically-driven pump isits high voltage requirement, which is approximately 42 V,for driving the electric motor This voltage is achievable inhybrid or future electric cars but not in vehicles with atypical internal combustion engine From a system designand feasibility standpoint, preserving the passive coolingstrategy may be more practical at this time As an alter-native, a method to control the quantity of coolant byblocking the coolant flow pathway during the warm-upperiod was used This concept is illustrated in Figure 6 This idea is similar to a thermostat in a conventionalengine cooling system, but it is not identical The thermostatdoes not allow the coolant to flow through the radiator toavoid heat rejection; however, the coolant still passes throughthe oil cooler and heater core It slows down the enginewarm-up

Two electronic valves were placed in the coolant pathway,and temperature sensors measured the coolant temperature.Based on the data from the sensors, the electronic controlunit (ECU) determines the coolant path Initially Valve 1 inFigure 6 lets the coolant flow through the bypass until theengine temperature reaches a desirable level, normally 90oC.Valve 1 then blocks the coolant flow toward the bypass andsends it into the oil cooler Valve 2 can block the coolantflow through the heater core, eventually reducing heat lossthrough the cabin heater during the summer when theheater core is unnecessary

Figure 5 Effect of the zero flow strategy on coolanttemperature

Figure 6 Schematics of coolant path control strategy

In short, the two valves reduced the engine warm-up

period and established a steady coolant temperature once

the transient period was complete This approach was also

examined using the simulation model, and the warm-up

time of the system using the valve control strategy was

shorter than that of conventional systems The simulation

results are shown in Figure 7 When the valve between the

engine and the oil cooler (Valve 1) was closed, the

warm-up time was observed to decrease by approximately 25%

As discussed before, a fast warm-up will contribute to a

reduction in fuel consumption and harmful emissions

based on the results shown in Figure 4 A high engine

temperature is beneficial for the chemical reaction of fuel

in the combustion chamber, and it is accompanied by a

decrease in CO and THC emissions An increase in NOx

emissions is a concern based on the experimental results

However, the amount of NOx produced during the start-up

period might be small relative to the amount of NOx

emitted during normal driving The active cooling strategies

also have the advantage of a more prompt response to

sudden changes in engine temperature resulting from

accele-ration or load change Frequently undergoing temperature

changes leads to an increase in all harmful emissions

Furthermore, the strategy of active coolant flow control

using an electronic pump is expected to eliminate

post-cooling to prevent thermal soak, particularly in diesel engines,

which is also conducive to reduced emissions

5 CONCLUSIONS

In this study, the feasibility of novel cooling strategies that

use an electric water pump and electronic valves in the

automotive cooling circuit was investigated During operation

of an engine, the coolant quantity and flow rate were

actively controlled at a more optimal temperature using

electronic valves and an electric water pump, respectively

The principle conclusions of this study can be summarized

as follows:

(1) The zero-flow warm-up strategy helped the engine

temperature increase faster than the conventional

cool-ing strategy

(2) Controlling the flow path also shortened the warm-upperiod A significant reduction of the warm-up periodwas achieved by closing the valve between the engineand the oil cooler

(3) These two strategies show potential to reduce the all harmful emissions and fuel consumption and areexpected to satisfy the stringent emission regulations ofthe future

over-ACKNOWLEDGEMENT− This work was done as a part of Industry sources development project The authors acknowledge the financial support for this research project provided by Korean Ministry of Knowledge Economy

REFERENCES

Ap, N., Guerrero, P., Jouanny, P., Potier, M., Genoist, J andThuez, J L (2003). Ultimate Cooling New Cooling System Concept Using the Same Coolant to Cool All Vehicle Fluids. IMechE 661−674

Ap, N and Tarquis, M (2005) Innovative engine coolingsystems comparison SAE Paper No. 2005-01-1378.Chanfreau, M., Gessier, B., Farkh, A and Geels, P Y (2003).The need for an electrical water valve in a thermalmanagement intelligent system (THEMIS TM) SAE Paper No. 2003-01-0274

Chalgren, R D (2003) Development and verification of aheavy duty 42/14V electric powertrain cooling system

SAE Paper No. 2003-01-3416

Chalgren, R D (2004) Thermal comfort and engine

warm-up optimization of a low-flow advanced thermal ment system SAE Paper No. 2004-01-0047

manage-Cho, H., Jung, D and Assanis, D N (2005) Control strategy

of electric coolant pump for fuel economy improvement

Int J Automotive Technology 6, 3, 269−275

Cho, H., Jung, D., Filipi, Z S., Assanis, D N., Vanderslice,

J and Bryzik, W (2007) Application of controllableelectric coolant pump for fuel economy and cooling per-formance improvement J Engineering for Gas Turbines and Power, 129, 239−244

Hnatczuk, W., Michael, P., Bishop, L J and Goodell, J.(2000) Parasitic loss reduction for 21st century trucks

SAE Paper No. 2000-01-3423

Kim, K., Hwang, K., Lee, K and Lee, K (2009) gation of coolant flow distribution and the effects ofcavitation on water pump performance in an automotivecooling system Int J Energy Research, 33, 224−234 Kruger, U., Edwards, S., Pantow, E., Lutz, R., Dreisbach,

Investi-R and Glensvig, M (2008) High performance coolingand EGR systems as a contribution to meeting futureemission standards SAE Paper No. 2008-01-1199 Najt, P M and Foster, D E (1983) Compression-ignitedhomogeneous charge combustion SAE Paper No 830264 Noguchi, M., Tanaka, Y., Tanaka, T and Takeuchi, Y (1979)

A study on gasoline engine combustion by observationFigure 7 Effect of coolant path control strategy on reduc-

tion of the coolant warm-up period

of intermediate reactive products during combustion.

Onishi, S., Jo, S H., Shoda, K., Jo, P D and Katao, S

(1979) Active thermo-atmosphere combustion (ATAC)

- A new combustion process for internal combustion

engines SAE Paper No. 790501

Page, R W., Hnatczuk, W and Kozierowski, J (2005)

Thermal management for the 21st century-Improved

thermal control & fuel economy in an army mediumtactical vehicle SAE Paper No. 2005-01-2068

Vegenas, A., Hawley, J G., Brace, C J and Ward, M C.(2004) On-vehicle controllable cooling jets SAE Paper

No. 2004-01-0049

Vetrovec, J (2008) Engine cooling system with a heat loadaveraging capability SAE Paper No. 2008-01-1168

DEVELOPMENT OF AN ON-LINE MODEL TO PREDICT THE CYLINDER RESIDUAL GAS FRACTION BY USING THE MEASURED

IN-INTAKE/EXHAUST AND CYLINDER PRESSURES

S CHOI, M KI and K MIN *

School of Mechanical and Aerospace Engineering, Seoul National University, Seoul 151-742, Korea

(Received 4 September 2009; Revised 25 May 2010)

ABSTRACT− The in-cylinder RGF (residual gas fraction) of internal combustion engines for new combustion concepts, such

as CAI (controlled auto ignition) or HCCI (homogenous charged compression ignition), is a major parameter that affects the combustion characteristics Thus, measurement or prediction of the cycle-by-cycle RGF and investigation into the relation between the RGF and the combustion phenomena are critical issues However, on-line prediction of the cycle-by-cycle RGF during engine testing is not always practical due to the requirement of expensive, fast response exhaust-gas analyzers and/or theoretical models that are just too slow for application In this study, an on-line model that can predict the RGF of each engine cycle and cylinder during the experiment in the test cell has been developed This enhanced model can predict the in-cylinder charge conditions of each engine cycle during the test in three seconds by using the measured dynamic pressures of the intake, exhaust, and cylinder as the boundary conditions A Fortran77 code was generated to solve the 1-D MOC (method of characteristics) This code was linked to Labview DAQ as a form of DLL (dynamic link library) to obtain three boundary pressures for each cycle The model was verified at various speeds and valve timings under the CAI mode by comparing the results with those of the commercial code, GT-Power.

KEY WORDS : RGF (residual gas fraction), CAI (controlled auto ignition) combustion, Cyclic variation, MOC (method of characteristics)

1 INTRODUCTION

The residual gas fraction (RGF) level in a cylinder charge

influences the combustion process of internal combustion

(IC) engines In particular, in new combustion concepts

(such as CAI, or controlled auto ignition) that auto-ignite

fuels with high octane numbers, the RGF becomes more

important because the combustion is controlled by the

thermal energy of the high RGF up to 40% Though new

combustion concepts have several advantages, e.g., low

emission levels and high thermal efficiency (Koopmans et

al., 2003; Zhao et al., 2001; Stanglmaier et al., 1999),

combustion control is not easy because there is no direct

method of controlling combustion, such as fuel injection

timing and spark timing, which are used in diesel and SI

engines, respectively (Kaneko et al., 2001) Compressed

ignition (auto-ignition) is difficult because the ignition is

completely governed by chemical kinetics and is therefore

influenced by the fuel’s composition, the equivalence ratio,

and the thermodynamic state of the mixture (Kelly-Zion,

2000)

For these reasons, it is necessary to measure the charge

conditions cycle-by-cycle to know the variation of these

conditions with the control parameters, such as the speedand valve timing, and to study the relationship between thecharge conditions and the combustion phenomena If thecharge conditions, such as the RGF and the temperature,can be measured or predicted quickly during the enginetest, it can be helpful for engineers to understand the phen-omena at the developmental stage of new engines.Several methods have been developed to measure andpredict the in-cylinder RGF

With regard to experimental measurement methods (F.Galliot et al., 1990; Richard et al., 1999; F Schwarz et al.,2003), the in-cylinder RGF was obtained from an analysis

of the CO2 or NO/HC concentrations during intake, exhaust,

or in-cylinder gases These methods are real-time and rate but require fast response analyzers and special, fastgas-sampling techniques, which are very expensive.Gas exchange models (Heywood et al., 1993; Cho et al.,2001; Uwe et al., 2004; G Colin et al., 2007) and empiricalmodels (Amer, 2006) predict the in-cylinder RGF by usingsimple gas exchange equations and empirical equationsthat require some engine parameters (e.g., the compressionratio, valve overlap, volumetric efficiency, A/F, speed, EGR,etc.) These models can quickly predict the in-cylinder RGFwithout expensive equipment and can be used to controlthe engine; however, the empirical coefficients have to be

accu-*Corresponding author. e-mail: kdmin@snu.ac.kr

adjusted in accordance with the engine and the operational

conditions

Recently, an applied 1-D gas dynamics model was

intro-duced (Liu et al., 2006) By using the measured dynamic

pressures of the intake and exhaust ports as boundary

conditions, this model shortened the length of the pipes in

the computation range By solving 1-D gas dynamics (Liu

et al., 1996) and cylinder energy equations, it was possible

to predict the in-cylinder RGF much faster than

conv-entional 1-D gas dynamics models An on-line tool was

also developed with this model It predicts the in-cylinder

RGF of a single cylinder engine within 20 seconds per

engine cycle for 10 simulation cycles and within 5 seconds

for 2 simulation cycles

In this study, an on-line model was developed that can

predict the RGF for each cycle and the cylinder during the

engine experiment in the test cell The MOC (method of

characteristics) was embedded in the Fortran code as the

1-D gas dynamics solver Three measured dynamic pressures

for the intake, exhaust, and in-cylinder gases were applied

as the boundary conditions By utilizing this additional

measurement of the dynamic pressure of the cylinder, this

model no longer solves time-consuming cylinder energy

equations and therefore substantially reduces the

computa-tional time This new model is verified by comparing its

results with those from a commercial code (GT-Power) at

various speeds and valve timings The model can be applied

to both single cylinder and multi-cylinder engines for any

valve timing and lift An on-line model was developed by

inserting this model into a Labview DAQ system as a form

of DLL (dynamic link library) On-line engine tests were

performed for a single cylinder engine in the CAI mode

During the test, the cyclic in-cylinder charge conditions

were predicted by using the RGF model The investigation

of the cyclic variation at a high load CAI operation through

the on-line model is presented

2 RGF PREDICTION MODEL

2.1 Major Features and Assumptions

The RGF prediction model calculates the mass flow rate

through the intake and exhaust pipes and predicts the

cylinder charge conditions, such as the trapped mass,

in-cylinder temperature, and RGF This model consists of two

parts: a “pipe-flow solver” and a “cylinder-condition solver”

The pipe-flow solver is a 1-D gas dynamics model that

calculates the instantaneous mass flow rate through the

intake and exhaust system This model uses three points at

the ends of the intake and exhaust pipes as boundary

condi-tions; the instantaneous pressures of the intake, exhaust,

and cylinder are measured at these points Through this

feature, the flow calculation commences from the positions

of the intake and exhaust dynamic pressure transducers, as

shown in Figure 1 The calculation range and the solution

time can be minimized The gas flow dynamics in the pipe

are solved through the MOC (Benson, 1982)

By simply solving the ideal-gas state equation, the der-condition solver predicts the instantaneous in-cylindercharge conditions, such as the trapped mass, RGF, and temper-ature by the intake and exhaust mass flow rate (calculatedwith the pipe-flow solver), and measured cylinder pressure.Consequently, with this simplified feature, the in-cylindercharge condition can be calculated much faster than throughconventional 1-D gas dynamic models that solve computa-tionally expensive conservative equations

cylin-The code was generated by using Fortran 77 To simplifythe model, several assumptions were made as follows

• An ideal gas was assumed

• The gas intake was assumed to be a homogeneous ture

mix-• Perfectly burned products were assumed following bustion

com-• Mixing at the valve ports through back flow was ignored.Flow losses (through the change in area, friction, heattransfer, etc.) did not affect the mass-flow results for thesingle cylinder engine model due to the short length andsimple geometry of the pipes in the calculation range Forthis reason, an isentropic MOC solver was used in the singlecylinder engine model However, for a multi-cylinder enginemodel, flow losses are no longer negligible due to thecomplex geometries of the intake and exhaust in the MOCsolver; the losses were considered through changes in theRiemann variables

2.2 Simulation Details2.2.1 Gas dynamics simulation

To calculate the pipe flow through the intake and exhaustsystem, a 1-D MOC is applied to the model The pipe flowcan be assumed to be 1-D in the longitudinal direction.Through this 1-D flow and the isentropic assumption, thecontinuity and momentum equations are solved through theMOC along the pipe length at every time step By using theideal gas assumption, the flow condition can be charac-terized by the speed of sound (a) and the flow velocity (u);furthermore, the conservation equations are expressed interms of the change in a and u along the pipe (x) and overFigure 1 Three locations for the boundary-pressure inputsand the calculation range of the gas-flow solver

time (t) The MOC solves the conservation equations by

mapping a and u to non-dimensional physical variables, the

so-called Riemann variables The Riemann variables, λ and

β, are defined as in Equations (2) and (3) by using the

non-dimensional speed of sound (A), flow velocity (U), length

(X), and time (Z) These variables are rendered

dimension-less by using the reference speed of sound (aref) and the

length (Lref), as shown in Equation (1) λ and β are the

characteristics that represent the direction of the pressure

wave; they respectively speak to the forward and backward

directions κ is the specific heat ratio of gas

(1) (2) (3)The pipe is gridded along the length At each time step,

new Riemann variables are calculated for all grids along

the length by using the variables already determined in the

preceding time step

The details of the grid calculations are shown in

Equa-tion (4)~(5) and Figure 2 In EquaEqua-tion (4)~(5), λ I is the

forward characteristic and λ II is the backward characteristic

Figure 2 shows the mesh grid along the length and time

step The symbol, m, refers to the length of the grid, whereas

r indicates the time step grid The Riemann variables at

each pipe end, (λ I)r+1,1 and (λ II)r+1,m+1, are obtained from the

boundary conditions

(4)

(5)

When the MOC is applied for non-isentropic conditions,

the flow losses due to wall friction, wall heat transfer, andentropy change are considered by summing the variance inthe Riemann variable that is caused by these losses A newRiemann variable, which is called the ‘path line charac-teristic (AA),’ is defined to represent the entropy-level changethat is caused by the losses The total variance in theRiemann variable, λ (Equation 6), can be determined as thesum of the variance due to the entropy change (Equation7), wall friction (Equation 8), and wall heat transfer (Equa-tion 9) In the equations, the subscript P refers to the value

of the previous mesh, whereas R refers to the value of thepresent mesh

(6) (7)

(9)where,

ap-Inflow:

(10)Outflow: (11)

L ref - dX

( ) friction = − κ 1 ( – ) f∆X - xD ref

∆X

- λ P – β P

κ 1 – -

κ 1 – -3

f = τ w 1 2 -ρu 2

the length/time grid

Junction boundary The junction boundary condition is

applied at the points where the pipes divide or converge

Through the assumption of isentropic flow and equal

pre-ssure and density for adjacent pipe ends, the Riemann

vari-able is determined by the following equations (Equations

14 and 15) in a multiple-branch pipe system Fn denotes the

cross-sectional area of the nth pipe, and FT is the sum of the

cross-sectional areas of the pipes at the junction

(15)Valve boundary The valve boundary condition is ap-

plied to the valve at the border of the intake and exhaust

system that is connected to the cylinder From the pressure

ratio between the cylinder and the pipe, the flow condition

through the valve is determined When inflow occurs from

the pipe to the cylinder, the valve is treated as a nozzle;

therefore, the mass flow rate and the Riemann variable are

determined by the nozzle boundary In the case of outflow

from the cylinder to the pipe, the pressure drop, entropy

change, and mass flow rate across the valve are calculated

by using Equations (16)~(19) below As a boundary

pre-ssure, the measured cylinder dynamic pressure is input into

the model, as shown in Equation (20)

(16)Pressure drop: sonic flow

2.2.3 Cylinder conditions and RGF tracking

The cylinder conditions are determined by the ideal-gas

state equation, without solving the energy equation At each

time step, the trapped mass in the cylinder and the RGF are

calculated from the instantaneous mass flow rate through

the intake and exhaust valve Because the equivalence ratio

of the intake gas is known and perfect combustion is

assumed, the trapped mass can be calculated by simplyadding the mass flows to the trapped mass from the pre-ceding time step The RGF can also be determined from theinitial trapped mass before the opening of the intake valveand the mass that is induced through the intake valve Theresults of the trapped mass and the RGF tracking are shown

in Figure 3; the exhaust and intake mass flows are mined from the flow calculation The cylinder temperaturecan be determined at each time step by substituting thethree known variables, i.e., the measured cylinder pressure,the calculated volume, and the gas composition as obtainedfrom RGF, into the ideal-gas state equation

deter-2.2.4 Calculation processThe calculation commences from the IVC (intake valveclosed) timing The first step is to input the required dataand the initial conditions at the IVC The required data are

– 2

κ 1 – - B

–

π -

m · = κpref

a ref - FUA -2P

π α = λ in / p input

p ref -

• Cylinder dimensions: bore, stroke, compression ratio, etc

• Intake- and exhaust-valve data: valve lift profile, flow

coefficient curve, etc

• Dimensions of the intake and exhaust pipes from the

valve to the dynamic pressure transducers

• Operating conditions: speed, valve timing, etc

The required initial conditions are as follows

• Intake and exhaust system’s reference pressures and

temper-atures

• Initial intake and exhaust pressures

The following process is iterative and entails the

conver-gence of the cylinder mass By using the initial conditions,

the time step, pressure, and temperature of each pipe and

cylinder are calculated until the cylinder mass at the IVC

converges

Updating the exhaust reference conditions The energy

levels of the intake and exhaust systems are determined

from the reference temperature, pressure, and speed of

sound In accordance with the exhaust energy that varies

across time steps, the energy level of the exhaust system

has to be updated Figure 4 is a valve boundary a-s

dia-gram It shows the entropy change and the pressure drop

across the valve when outflow occurs Because the entropy

level of the exhaust system is a blue-dotted line that passes

through apipe, the reference speed of sound, aref, has to be

moved to the same entropy level, apref This process can be

expressed by Equation (21) C is the entropy change across

the valve, which is calculated by Equation (18) At every

time step, the reference speed of sound for the exhaust

system is updated

(21)

3 MODEL VALIDATION

Neither direct measurement of the instantaneous mass flow

rate nor a fast response gas analyzer and a sampling methodfor measuring the RGF of each cycle were available in thisstudy Hence, for validation, the calculated mass flow ratesand the RGF results from the newly developed model arecompared with those from a GT-Power model under thesame operating conditions

3.1 Model Validation with Regard to a Single CylinderResearch Engine

An equivalent GT-Power TPA (three pressure analysis) model,which has the same intake and exhaust configuration, isgenerated to validate the results of the RGF predictionmodel The TPA model is used for validation because italso requires the measured intake, exhaust, and cylinderdynamic pressures as input data similar to the developedmodel

An engine experiment was performed to obtain the modelinput data (geometries, dynamic pressures at three points,operating parameters, etc.) This engine has a dual CVVTsystem, which is operated by the hydraulic pressure that isdrawn from the engine’s oil pump The adjustable range of

Figure 4 a-s diagram of the exhaust valve boundary

Figure 5 Single cylinder dual CVVT engine

Figure 6 Cylinder pressure and valve actuation for CAIcombustion

the valve timing is 40oCA in both the intake and exhaust

valves Low-lift camshafts, which have a maximum lift of

2 mm, are installed at both the intake and exhaust (Figure

5) The NVO (negative valve overlap) strategy is used for

CAI combustion, as shown in Figure 6

Fuel is delivered to the intake port, and the intake air is

naturally aspirated In all cases, the throttle position is

main-tained as ‘wide open.’ A detailed description of the test

engine is given in Table 1 Three dynamic pressures, i.e.,

from the in-cylinder, exhaust, and intake, were measured to

analyze the combustion and to predict the RGF for each

cycle The cylinder pressure was measured with a Kistler

6041A water-cooling-type, relative-pressure transducer

The intake and exhaust pressures were measured by using

the Kistler 4045A absolute-pressure transducers The

pressure signals were recorded at increments of 1o in the

crank angle for 200 cycles A 200-cycle average of the

intake, exhaust, and cylinder dynamic pressures were used

in this validation process

As shown in Table 2, seven sets of speeds and valve

timings were selected to examine the model’s feasibility

with regard to the operating conditions In the condition

indicator, (1500/−10/20), (1500) represents the engine speed

in rpm, whereas (−10/20) indicates the timing-modification

values of the valve opening with reference to the standard

IVO/EVO (408/160), i.e., 10oCA retarded IVO and 20oCA

advanced EVO

3.2 Model Validation with Regard to a Multi-cylinder Engine

Because a multi-cylinder engine was not available for CAI

operation with low valve lift, the physical engine was

replaced by the GT-Power, four-cylinder full engine model

in the validation procedure The input data for the RGFmodel, which includes three points to measure the dynamicpressure, were obtained from this full engine model AnRGF model and a GT-Power TPA model that had the samegeometry as the full engine model were generated Theresults of the mass flow rate and the RGF were compared

As shown in Figure 6, validation was conducted at twodifferent cam profiles, which are a conventional cam pro-file with a maximum lift of 9 mm for SI engines and a low-lift cam profile with a maximum lift of 2 mm for CAIengines The engine speed was fixed at 1500 rpm To ex-amine the effect of valve timing, two different valve timingsfor the conventional cam profile and five different valvetimings for the low-lift cam profile were selected Thedetailed conditions are shown in Table 3

3.3 Validation ResultsThe calculated intake and exhaust mass flow rates andRGF under both the RGF prediction model and the GT-Power TPA model are compared for single and multi-cylinder engines The results of the comparison for RGFover seven different single cylinder engine conditions areshown in Figure 7 The RGF model shows reasonably goodagreement with the GT-Power model for the seven operat-ing conditions; the differences between the two sets ofresults are less than 4% Also, the RGF model will trace thevariation in the RGF with the speed and valve timing.When validating the multi-cylinder model, the effects of

Table 1 Single cylinder engine specifications

Valve duration IN/EX 172oCA/136oCA

Standard IVO/EVO 408oCA/160oCA

Table 2 Single cylinder model validation conditions

Low-liftCam

the two different cam profiles on the RGF prediction are

examined As shown in Figure 8, the RGF model will follow

fluctuations in the mass flow rate in both the exhaust and

intake flows In this figure, the solid lines indicate the

results of the RGF model, whereas the dotted lines refer to

the GT-Power results The red line denotes the exhaust

mass flow rate, and the blue line indicates the intake mass

flow rate through the valves The cumulative intake and

exhaust air masses are almost the same in both models

This means that the developed RGF model can trace real

flow phenomena and predict the mass flow rate under

various operating conditions With regard to RGF prediction,the RGF model still traces the variation in the RGF with thevalve timing On average, a deviation of about 5.5% isobserved when GT-Power is used, as shown in Figure 9.The simpler features of the RGF model enable the model

to predict the in-cylinder conditions (such as the RGF,temperature, and trapped mass) on the average three secondsper engine cycle (in a single core PC with a 2-GHz CPUclock and a 2-GB RAM) It only takes about 0.2 s for asimulation cycle, and it converges in 10~15 simulation cyclesunder moderate initial and boundary conditions In the fineinitial and boundary conditions, the calculation converges

in less than five simulation cycles; thus, this model canpredict the RGF in one second In contrast, known fast 1-Dgas dynamics models require a computational time that isbetween six to ten times the time for one simulation cycle.Thus, other models require around 2 seconds for a simu-lation cycle; even though the calculation converges in fivesimulation cycles, it takes about 10 seconds to obtain thesame results This means that the developed model is suit-able for on-line applications, where the quick output of in-cylinder conditions is required

4 ON-LINE RGF PREDICTION MODEL

An on-line RGF model that can predict the in-cylinder RGFduring an engine test has been developed by coupling theFortran RGF code and the Labview data acquisition system(DAQ)

4.1 Coupling the FORTRAN Code and LabviewThe Fortran 77 code and Labview VI are coupled via DLL(dynamic link library) The ready-made RGF code is modi-fied as a DLL and then inserted into Labview VI Duringthe engine test, the Labview DAQ system receives theintake, exhaust, and cylinder dynamic pressures as well asthe encoder signal In the meantime, Labview VI performsdata processing, such as finding the TDC, pegging therelative cylinder pressure for feeding dynamic pressure databased on the crank angle (CA) to the RGF code module,etc Some required data, such as the engine and pipe speci-fications and the initial conditions, are given by the userand transferred to the RGF code module by Labview as

Figure 8 Comparison of the intake and exhaust massflow

rates from the RGF model with the GT-Power model

(multi-cylinder model with a conventional cam)

Figure 9 Validation of the residual gas fraction

well After the calculation, the calculated mass flow and

RGF results are transferred to Labview Figure 10 and the

following numbered descriptions show the sequence and

inputs/outputs during the engine test and the RGF

predic-tion

4.2 Results of On-line Tests of the RGF Model

On-line tests that use the RGF model were performed in

the CAI mode by using the single cylinder engine

When the engine operates under comparatively high loads,

significant cyclic variation in the combustion is found The

peak pressure and start of the combustion (SOC) timing

fluctuate significantly and knock occurs frequently, as

shown in the upper plot in Figure 11; in this plot, 100

consecutive cycles of the measured pressures of (2000/0/

20) are superposed The right plot in Figure 11 is the

on-line, predicted result for the RGF during the same cycles

and shows that the RGF also fluctuates significantly as the

combustion fluctuates

Figure 12 is the correlation plot of the RGF, the trapped

mass, and the in-cylinder temperature after 100

consecu-tive cycles at the same condition, (2000/0/20) The results

show that these variables reflect a clear correlation for each

cycle As the RGF increases, the in-cylinder temperature

increases due to the hot residual gas, and the in-cylinder

trapped mass decreases due to the low charge efficiency

that is caused by the high charge temperature The RGF

fluctuates by more than 10%, and the temperature at the

TDC also varies over 800~1000 K These variables from the on-line RGF model give

im-portant information that can explain the cyclic variation It

is presumed that the combustion condition of the presentcycle affects the in-cylinder charge condition of the follow-ing cycle

5 CONCLUSIONS

An enhanced on-line RGF prediction model for single der and multi-cylinder engines is developed in this research.This on-line RGF prediction model has the followingcharacteristics

cylin-(1) By using the intake, exhaust, and cylinder dynamic ssures as boundary conditions, this model can predictthe mass flow rate and the RGF in single cylinder andmulti-cylinder engines without any change in the coeffi-cients It takes about 0.2 s for one simulation cycle, andthe calculation converges in 15 simulation cycles withmoderate initial and boundary conditions

pre-(2) The Fortran RGF code, which is used as the dynamics solver in this model, is validated by the well-matched GT-Power model at various speeds, valve lifts,and valve timings The results from the model arefound to be in good agreement with those from GT-Power

gas-(3) From the on-line RGF prediction during the engine test,this model can predict the cycle-by-cycle RGF in anaverage three seconds for any speed and valve timing

Figure 11 Cyclic variation of the RGF in a high load CAI

operation (2000/0/20, 100 cycles)

Figure 12 Correlation plot of the RGF-trapped T_TDC (2000/0/20, 100 cycles)

mass-(4) During the single cylinder engine test, the RGF, trapped

mass, and in-cylinder temperature of the subsequent

cycle are predicted by using the on-line RGF model

These variables indicate a clear correlation for every

cycle From these results, the cause of the cyclic

com-bustion variation at high load CAI operations can be

analyzed

ACKNOWLEDGEMENT− This study has been sponsored by

the Korea Automotive Technology Institute, which is directed by

the Hyundai Motor Company, and funded by the Ministry of

Commerce, Industry, and Energy of the Republic of Korea Also,

this research was supported (in part) by SNU-IAMD.

REFERENCES

Amer, A A (2006) A semi-empirical model for fast

resi-dual gas fraction estimation in gasoline engines SAE

Paper No 2006-01-3236

Benson, R S (1982) The Thermodynamics and Gas Dynamics

of Internal-Combustion Engines Clarendon Press Oxford

Cho, H., Lee, K., Lee, J., Yoo, J and Min, K (2001)

Mea-surements and modeling of residual gas fraction in SI

engines SAE Paper No 2001-01-1910

Colin, G., Giansetti, P., Chamaillard, Y and Higelin, P (2007)

In-cylinder mass estimation using cylinder pressure SAE

Paper No 2007-24-0049

Ford, R and Collings, N (1999) Measurement of residual

gas fraction using a fast response NO sensor SAE Paper

No 1999-01-0208

Galliot, F., Cheng, W K., Cheng, C O., Sztenderowicz, M.,

Heywood, J B and Collings, N (1990) In-cylinder

mea-surements of residual gas concentration in a

spark-ignition engine SAE Paper No 900485

Heywood, J B., Cheng, W K and Fox, J W (1993) A

model for predicting residual gas fraction in spark-ignitionengines SAE Paper No 931025

Kaneko, M., Morikawa, K., Itoh, J and Saishu, Y (2001).Study on homogeneous charge compression ignition gaso-line engine COMODIA 441-446

Kelly-Zion, P L (2000) A computational study on theeffect of fuel-type on ignition time in HCCI engines The

2000 Symp Int Combustion.Koehler, U and Bargende, M (2004) A model for a fastprediction of the in-cylinder residual gas mass SAE Paper

No 2004-01-3053

Koopmans, L., Ogink, R and Denbratt, I (2003) Direct line injection in the negative valve overlap of homogene-ous charged compression ignition engines SAE Paper

gaso-No 2003-01-1854

Liu, J P., Schorn, N., Schernus, C and Peng, L (1996).Comparison studies on the method of characteristics andfinite difference methods for one-dimensional gas flowthrough IC engine manifold SAE Paper No 960078.Liu, J P., Kleeberg, H., Tomazic, D., Ciaravino, J A andAmer, A A (2006) A model for on-line monitoring ofin-cylinder residual gas fraction (RGF) and mass flowrate

in gasoline engines SAE Paper No 2006-01-0656.Schwarz, F and Spicher, U (2003) Determination of resi-dual gas fraction in IC engines SAE Paper No 2003-01-3148

Stanglmaier, R H and Robets, C E (1999) Homogeneouscharged compression ignition (HCCI): Benefits, com-promise and future applications SAE Paper No 1999-01-3682

Zhao, H., Peng, Z., Williams, J and Ladommatos, N (2001).Understanding the effect of recycled burnt gases on thecontrolled auto ignition (CAI) combustion in four-strokegasoline engines SAE Paper No 2001-01-3607

EFFECT OF THE NUMBER OF FUEL INJECTOR HOLES

ON CHARACTERISTICS OF COMBUSTION AND EMISSIONS

IN A DIESEL ENGINE

B H LEE 1) , J H SONG 2) , Y J CHANG 2) and C H JEON 2)*

1)Department of Mechanical Engineering, Pusan National University, Busan 609-735, Korea

2)School of Mechanical Engineering, Pusan National University, Pusan Clean Coal Center, Busan 609-735, Korea

(Received 19 October 2009; Revised 27 March 2010)

ABSTRACT− The diesel combustion process is highly dependent on fuel injection parameters, and understanding fuel spray development is essential for proper control of the process One of the critical factors for controlling the rate of mixing of fuel and air is the number of injector holes in a diesel engine This study was intended to explore the behavior of the formation of spray mixtures, combustion, and emissions as a function of the number of injector hole changes; from this work, we propose

an optimal number of holes for superior emissions and engine performance in diesel engine applications The results show that increasing the number of holes significantly influences evaporation, atomization, and combustion However, when the number

of holes exceeds a certain threshold, there is an adverse effect on combustion and emissions due to a lack of the air entrainment required for the achievement of a stoichiometric mixture

KEY WORDS : Fuel injector, Hole number, Spray interaction, Combustion, Emissions

1 INTRODUCTION

Due to their relative simplicity, relatively low capital cost,

higher power density, and high efficiency capabilities, diesel

engines enjoy widespread popularity From small

single-cylinder generators to supertankers, diesel engines are

often the logical choice for use as prime movers, or are, at

least, considered viable alternatives However, the

tradi-tional diesel engine suffers from relatively high nitrogen

oxide (NOx) and particulate emissions Thus, there is an

increased focus in diesel-engine research on reduction of

such emissions

The mixing of injected fuel and ambient air in a

direct-injection diesel engine significantly influences the

heat-release rate and the formation and oxidation of pollutants

Hence, it is important to understand how fuel injection,

engine geometry, flow-field characteristics− and the

inter-action of all three components – may affect the mixing

process An understanding of these interactions is important

in the context of optimizing diesel engine performance to

maximize thermal efficiency and meet stringent regulations

on toxic pollutants

Engine designers have adopted several strategies to meet

these regulations In the automotive industry, for example,

the current particulate-reduction strategy is to use high

injection pressures with small-holed nozzles in the injector

tip

that the reductions made through the use of this strategy arethe result of increased mixing due to enhanced spray penet-ration, greater entrainment of air, and wall impingement.Nehmer and Reitz (1994) and Tow et al (1994) furthershowed that multiple injections have been used to reduceparticulate emissions Han et al (1996) showed this to bedue to the reduction in the amount of fuel in the rich zone atthe tip of the spray, which results in better mixing anduniform distribution of the fuel-air mixture

Another possible strategy is to employ alternate spraypatterns The variable parameters in this case include theangles between sprays in multi-hole injectors and the angle

of the spray with respect to the head Abraham et al (1999)showed that interactions between fuel jets in a multi-holeinjector and between the jets and the wall may affect thefuel-air mixing process in a direct-injection diesel engine.Through an experimental apparatus, Fujita et al (2007)reported the variation of multi-hole spray interference andmixture formation as a function of hole diameter, number

of holes, and swirl velocity Their results showed the effects

of the mixture-formation process during the ignition-delayperiod on the ignition and combustion of diesel sprays.Using a rapidly charging constant-volume combustionchamber, Mitroglou et al (2005, 2007) characterized theperformance of multi-hole injectors designed for use insecond-generation, direct-injection gasoline engines Theirresults showed that a comparison between two differentinjectors enables quantification of the influence of the

number of holes on the spray structure and spray tip

penetration, droplet velocity and droplet size distribution

for various injection types

The present study considers multi-hole injectors for diesel

engines A numerical simulation and several computational

models were employed to study effect of multi-hole injector

on fuel-air mixture formation, combustion, and formation

of pollutants when the number of holes is increased from a

baseline of six to ten Based on these results, we propose an

optimal number of holes for diesel engine applications

2 NUMERICAL MODELS AND METHODS

2.1 Diesel Engine Combustion Process

Timothy and Gardner (1992) and Deck (1997) showed that

the combustion process in diesel engines can be

sub-divided into three consecutive periods; 1) the ignition delay

period; 2) the pre-mixed combustion period; and 3) the

diffusion combustion period

The ignition delay period commences with the injection

of fuel and is usually defined in terms of either the initial

detection of combustion luminosity or the onset of a

com-bustion-induced rise of pressure in the cylinder Ignition

delay is widely considered to involve two components: a

physical delay and a chemical delay The length of the

delay period is of considerable importance since it

in-fluences the processes that follow ignition, including pollutant

formation The premixed combustion period is referred to

as the “uncontrolled” combustion phase, where the

com-bustion is determined by conditions that were previouslyestablished in the fuel spray at the end of the ignition delay

It is during this period that high rises in pressure and hightemperatures occur, leading to NOx formation

Any condition that results in sufficiently high peak cylinderpressures and temperatures will produce NOx emissions.The increased mixing of fuel and air during the ignition-delay period usually increases the rate of combustion andthe NOx level The diffusion combustion period followsthe pre-mixed combustion phase It is during this periodthat the combustion process is controlled by the rate ofinjection and mixing Since diesel combustion involvesheterogeneous mixtures of fuel and air, mixing plays a keyrole in the diesel combustion process and the consequentproduction of soot particles The soot is formed from theavailable carbon in the partially burned fuel, when fuel-richmixtures are subjected to high temperatures Therefore,soot emissions are generally associated with the main por-tion of the fuel that is injected during the diffusioncombustion period

To predict a number of phenomena described above fordiesel engine combustion, a numerical analysis was con-ducted using several models already available in Star-CDcode The analysis used the Reitz and Diwakar model(Reitz and Diwakar, 1986) as a droplet break-up model, theShell ignition model of Halstead et al (1977) as an ignitionmodel, the eddy break-up (EBU) model of Patterson et al.(1994) as a combustion model, an extended Zeldovichmechanism (Turns, 2000) as a NOx formation model, andthe model of Mauss et al (1988) as the soot formationmodel Further details of the models are given below.2.1.1 Droplet break-up model

The break-up model of droplets used in this study was theReitz and Diwakar model (Reitz and Diwaker, 1986), wherethe rate of change of the droplet diameter is expressed as

(1)where D d is the instantaneous droplet diameter and τ b is thecharacteristic time scale of the break-up process The dia-meter of the droplets changes continually until they asymp-totically approach a stable droplet diameter D d,stable It isassumed that no new droplets are created, but the number

of droplets is updated to conserve the mass Aerodynamicdroplet breakup is controlled by the Weber number, for whichtwo different modes (regimes) are identified: a bag breakupregime, which occurs at low droplet velocities, and a strippingbreakup regime that occurs at a higher droplet velocity

determin-ed by the Weber number W e:

(2)

dD d

dt - = − D d – D d stable ,

(1997)

where C b 1 is the critical Weber number, ρ is the density, u

is the gas velocity, u d is the droplet velocity and σ is the

surface tension The value of C b 1 is 6

The characteristic time scale of the bag breakup regime is

(3)where the model constant C b 2 is equal to be π

stripping breakup regime is

(4)where the Red is the droplet Reynolds number and the

model constant C s 1 is equal to 0.5 The characteristic time

scale for the regime is

(5)where the model constant C s 2 is a coefficient with a value

of 20

2.1.2 Ignition model

The Shell ignition model of Halstead et al (1977) is used

to predict the start of combustion through a simplified

8-step kinetic model between five species It cannot be

view-ed as a formally rview-educview-ed reaction mechanism with global

reactions between actual species; it rather represents a

virtual mechanism between generic species that attempts to

reflect the actual ignition behavior of hydrocarbon air

mix-tures, including multistage ignitions and cool flames The

reactions and species involved in this kinetic model are as

where RH is the hydrocarbon fuel (C n H 2 m), R * is the radical

formed from the fuel, B is the branching agent, Q is a labile

intermediate species, and P is oxidized products, consisting

of CO, CO2, and H2O in specified proportions The

ex-pressions for K q, K p, K b, K t, f 1, f 2, f 3, f 4, etc are those given

by Halstead et al (1977) In addition, the local

concent-rations of O2 and N2 are needed to compute the reaction

rates A chain-propagation cycle is formulated to describe

the history of the branching agent, Equations (7)~(11),

together with one initiation, Equation (6), and two

termin-ation reactions, Equtermin-ation (12)~(13)

2.1.3 Combustion modelThe combustion model considered in this study is the eddybreak-up (EBU) model of Patterson et al. (1994) Duringthe combustion process, the global single-step reaction offuel is assumed to form carbon dioxide and water vapor.The combustion rate is primarily governed by a turbulence-related time scale Specifically, the fuel consumption rate isexpressed on the base of a single step reaction as

(14)

dimensionless empirical coefficients, k is the turbulencekinetic energy, ε is the dissipation rate, and, accordingly,the turbulent time scale is k/ε, Y O,P,F is the mass fraction,

n O,P,F, the mole fraction, M O,P,F, the molecular fraction foroxidizer, products, and fuel This model has been and isstill widely used in engine simulations, primarily due to itssimplicity

2.1.4 NOx model

In a diesel engine, a high gas temperature and long sidence time at such temperatures may cause a greatamount of NO to be formed during the combustion process.Thus, the NOx model used in this study was the extendedZeldovich mechanism (Turns, 2000), whose reactions aregiven below The rate constants for these reactions havebeen documented in numerous experimental studies anddata obtained from these studies have been criticallyevaluated by Baulch et al. (1973)

re-k i,f = 1.8×1011 exp[−38,370/T] m3/(kgmol·s) (15)

k ii,f = 1.8×107 exp[−4,680/T] m3/(kgmol·s) (16)

k iii,f = 7.1×1010 exp[−450/T] m3/(kgmol·s) (17)

By assuming steady-state conditions for the N atoms, anequilibrium state for O atoms with O2,, and by makingsome simplifications, the following expressions for the NOreaction rate were found:

(18)where k i,f is the forward reaction coefficient of thegeneralized Arrhenius form. KC,i and KC,ii are the equili-brium constants for the reaction (15) and (16)

2.1.5 Soot modelThe soot model used in this study was the detailed kineticmodel of Mauss et al. (1988) This model includes gas-phase kinetics describing the chemical reactions on amolecular scale as well as particle dynamics to describe aseries of processes including particle inception, surface

growth, coagulation, and oxidation on a particle scale The

modeling of the soot/flow-field interaction is based on a

flamelet approach Source terms for the soot volume

frac-tion are taken from a flamelet library using a presumed

probability density function and integrated over the

frac-tion of the mixture space The transport equafrac-tion for the

soot mass fraction is given by

(19)where Y s is the soot mass fraction, µ t is the molecular

dynamic viscosity of the fluid, and µ j is fluid velocity

component in direction x j The Prandtl number for soot is

assumed to be 1.4 and the soot density, ρ s, is equal to 1860

kg/m3 The source term for the soot volume fraction is

given by

(20)where is the mean soot volume fraction and the

mean source terms are obtained from the flamelet library

mentioned earlier This is based on the solution of a

com-bustion system consisting of two oppositely reacting laminar

jets involving over 800 reactions and about 100 species for

n-heptane fuel

2.2 Computational Conditions and Parameters

The diesel-engine specifications considered in this

com-putation are listed in Table 1, while the engine operating

conditions are listed in Table 2

The major parameter to be varied was the number of

holes, which, in turn, affected the diameter of each hole

Depending on the number of holes, the hole diameter of the

nozzle was adjusted to keep the total discharged area of the

nozzle constant In all, three combinations of the number of

holes and corresponding hole diameters were considered:six holes (d=0.335 mm); eight holes (d=0.29 mm); and tenholes (d=0.259 mm) Figure 3 shows the mass flow rate per

a nozzle as a function of the number of holes The massflow rate was distributed across the nozzles so that the totalamount of fuel injected into the cylinder remained con-stant This configuration of mass flow rate was employed

to isolate the effects of the number of holes The anglebetween injectors in adjacent holes was assigned in relation

to the number of holes Hence, for six holes, the angle was

60 degrees (360/6); for eight holes, the angle was 45 grees (360/8); and for ten holes, the angle was 36 degrees(360/10) The values for the temperature and the pressure

de-at the initial condition for the transient analysis were

Injected fuel Normal dodecane (C12H26)

Total mass of fuel injected 456 mg/cycle

Start of fuel injection BTDC 6°

Number of holes to be varied 6, 8, 10

Figure 2 Geometry of the full scale engine, piston andinjector nozzle

Figure 3 Variations of the mass flow rate used for differenthole injectors

determined by running a one-dimensional boost model up

to 90° BTDC Figure 2 presents the geometry of the

full-scale engine, piston, and injector nozzle implemented in

this computation It should be noted that only the hole

diameter of the injector nozzle changed Other parameters

of the nozzle remain unchanged; for example, the aspect

ratio of L/D was 4 Normal dodecane (C12H26) was selected

as the fuel because its properties are similar to those of

diesel fuel

For validation of the full-scale CFD model employed in

this research, this result was compared with the result

obtained from the one-dimensional boost model and with

experimental result tested in the diesel engine The

com-parison shows a reasonable predictability for the 3-D model

in terms of the magnitude and phase of the in-cylinder

pressure at most stages (compression, ignition and

com-bustion, and power stroke after the combustion), and only a

slight difference near the early stage of injection

3 RESULTS AND DISCUSSION

3.1 Diesel Spray Characteristics

Figure 5 shows the temporal variation of the Sauter mean

diameter (SMD) of fuel droplets with crank angle degree

(CAD) The SMD represents the ratio of the total volume

to the total surface area of fuel droplets injected into the

cylinder With a numerical method, total number of

drop-lets, the mass and diameter of each droplet were calculated

right after the start of injection, for BTDC 20° to ATDC

40° From this raw data set, the temporal variation of the

SMD was further derived following a formula described

below:

(21)The SMD increased until BTDC reached 1°; thereafter,

it began to decrease due to the continuing atomization

caused by viscous friction with ambient air; it eventually

remains constant after reaching TDC The SMD decreased

as the number of holes increased, since atomization ispromoted by the higher friction dictated by the smaller holediameter To further understand the effect of injectionpressure on diesel spray characteristics under the conditionwhere the number of holes (and thus diameter) varies, it isnecessary to introduce the relation of velocity with thedifference in injection pressure and discharge coefficientderived from a modified Bernoulli equation that considersfriction

(22)where c is a discharge coefficient, ∆P is the pressuredifference between injection and the surrounding chargedair, and ρ is the flow density

In a real situation where the hole diameter of a nozzledecreases, an injection system needs to boost the injectionpressure to counteract the reduced effect of the dischargecoefficient (due to increased friction) to maintain a constantvelocity at each hole This behavior confirmed the asser-tion of Equation (22) regarding the potential pressurizedeffect on the character of spray atomization, which inevit-ably complicates the pure effect of the changing diameter

SMD=

droplets∑ d i3/

droplets∑ d i2

V=c 2∆P ρ -

Figure 4 Validation of two computational models

(one-dimensional boost model and three-(one-dimensional full scale

CFD model) with experimental results

Figure 5 Temporal variation of the SMD for differentnumbers of injector holes

Figure 6 Temporal variation of the total fuel mass in theliquid phase and rate of evaporation for different numbers

of injector holes

However, since numerical computations enable parametric

studies, it was possible to isolate the effect of changing

only the diameter on the pressure

Upon stabilization (as described in the preceding

para-graph), the corresponding values of the mean SMD for

three different numbers of hole were 34.4 mm (six holes),

31.9 mm (eight holes), and 27.1 mm (ten holes)

Figure 6 presents the total mass of fuel in liquid phase

and the evaporation rate as a function of crank angle For

all cases, evaporation surpassed 90% before ATDC 5°,

which indicates that the fuel evaporated in a short span of

time As the number of holes increases, the evaporation

rate increases because of the smaller droplets, which could

easily infer from the SMD result in Figure 5

3.2 Combustion Characteristics

As seen in Figure 7, local penetration of the liquid spray

decreased at ATDC 22o for increasing numbers of holes

Instead, an individual spray plume was likely to interact

with neighboring plumes due to the reduced spacing

bet-ween them Thus, the injected fuel vapor became more

dispersed at lower concentrations These results are

con-sistent with the experimental findings of Fujita et al (2007)

They also investigated the effect of hole number on spray

mixture formation with a constant total cross-sectional hole

area in a constant volume chamber As the number of holes

increased, an intense interaction of two adjacent sprays

resulted in a reduced degree of mixing with air and, in turn,

greatly influenced emission as well as combustion

charac-teristics Patrick and Rhim(2001) demonstrated the effect

of the number of injector holes on the entrained

charac-teristics of the surrounding air for a fuel spray jet tested in

a constant volume chamber using particle image

veloci-metry (PIV) Their results showed that the amount of

entrained air around a fuel jet decreases with increasing

number of holes, which is in line with our prediction

The temperature of the outer spray periphery where the

fuel vapor meets with the surrounding air at a particular

stoichiometric ratio is known to be the highest among allregions of spray Those reacting zone regions with thehighest temperatures are accordingly diminished withincreasing numbers of holes, while the fuel concentrationand temperature is more uniformly distributed across theentire regions Due to such temperature distributions, NOxemission is more uniformly distributed as the hole numberincreases

Therefore, it is expected that both NOx emission and theresultant temperature averaged over the combustion chamberarea gradually decrease as combustion proceeds

Figure 8 displays the temporal variation of the mixingcharacteristics for three different hole injectors The mixingdomain where the fuel vapor mixes with the ambient air isdivided into two regions: flammable and rich fraction Theflammable fraction is the ratio of fuel vapor volume whenthe equivalence ratio of the mixture is less than 1.5 to thetotal volume of fuel vapor The rich fraction is the fraction

of total fuel that has an equivalence ratio greater than 1.5.According to this definition, an increase in the rich fractionduring the fuel vapor development process indicates poorentrainment of the air surrounding air the fuel spray requir-

ed for subsequent mixing When ATDC=11° (fuel injection

is almost complete), the rich fraction of the entire fuelvapor was 16.3% for six holes, 18.3% for eight holes, and22.3% for ten holes, while the flammable fraction decreas-

ed from 83.6% to 77.7% as the hole number increased fromsix to ten

The high rich fraction for the ten-hole injection systemcorroborates the idea of that jet-jet interactions lead toworse mixing for the tested conditions as previously de-scribed This effect of hole number on mixture formation isconsistent with the experimental results of Abraham et al.(1999), where the excessive interaction between two jetsreduced the extent of mixing as the number of holesincreased beyond an optimal value

Figure 9 presents the local distribution of fuel vaporconcentration, temperature, and NOx at the timing ofATDC 10° in a vertical plane of combustion chamber This

Figure 7 Local distribution of fuel vapor concentration,

temperature (K), and NOx emission for ATDC 22° in a

horizontal plane 20 mm below a cylinder head

Figure 8 Temporal variation of two mass fractions of fuelvapor, flammable and rich fraction, for three different holeinjector

exhibits greatest wall impingement of six-hole injection

where liquid fuel spray is collided with piston bowl causing

the presence of cooling effect to those regions This

impin-gement occurring at such regions can lead to the reduction

of average temperature, which effect will be explained in

subsequent results

Figure 10 shows the pressure and the heat release rate as

function of crank angle During the initial period of

com-bustion when wall impingement is not fully initiated, the

heat release rate of six-hole injection should be higher than

that of other cases due to lesser uniform fuel-air

distribu-tion for enhanced combusdistribu-tion as shown in previous Figures

7 & 8 However, the heat release rate of six holes begins to

fall down below the others due to the strongest penetration

of liquid spray as soon as spray impingement on the bowl

develops appreciably around ATDC 10o As a result, heat

loss from the jet-wall impingement considerably decreased

the heat release produced during the period This

observa-tion is in a good agreement with the results reported by

Lijun and Abraham (2003), in which the impingement of

spray jet against the piston bowl decreased in cylinder

temperature and NO emission due to heat loss involved

during the process Those kinds of combustion

characteri-stics are also supported from the distribution of pressure inthe cylinder Meanwhile, the heat release rate and pressurefor ten holes are remarkably lower than those in other casesdespite of the improved evaporation and atomization seen

in Figures 5 and 6

It is because poorly distributed fuel-air mixture couldproduce the lowest average temperature, while heat lossfrom the wall impingement is not developed significantlydue to the lowest momentum among all cases as shown inthe prior Figure 9

Figure 11 shows the temporal variation of the averagedconcentration of remaining unburned fuel and the averagemean temperature with the crank angle This result con-firms that combustion does not proceed to a completeextent at the end of combustion with ten-hole injection due

to the lowest peak of mean temperature and highest centration of remaining unburned fuel during the com-bustion process This is readily expected from the worstmixing observed with ten-hole injection Again, the peaktemperature of eight-hole injection is apparently highestamong all injections, resulting from the competing effects

con-of temperature in response to both local distribution con-offuel-air mixture and wall impingement

3.3 NOx and Soot EmissionsFigure 12 shows a temporal variation of NOx and sootemissions, which is averaged over entire section of com-bustion chamber The formation of NOx is known to beaffected significantly by thermal NOx mechanism, which

is strongly dependent on temperature Upon the late stage

of combustion process, the quantitative amount of NOxreaches the highest with eight hole injection, which can bereadily understood by the temperature changes from twoeffects previously described Likewise, the amount of NOx

is observed to be the smallest with ten hole injection marily due to the reduced local temperature from the poormixture formation These results are consistent with theexperimental observations of Montgomery et al (1996)where the exhaust emissions were degraded by a presence

pri-of fuel rich zones developed from poor entrainment pri-of

Figure 9 Local distribution of fuel vapor concentration,

temperature (K), and NOx emission at the time of ATDC

10° in a vertical plane of combustion chamber

Figure 10 Temporal variation of pressure and its heat

release rate

Figure 11 Temporal variation of cylinder mean ature and unburned fuel concentration remaining in thecylinder

temper-surrounding air

In the case of six holes, considerable amount of soot is

generated at the outset of combustion and during entire

combustion

In the case of six holes, an increased amount of injected

fuel as seen from the Figure 3 is likely to produce high

possibility of unburned fuel to be concentrated on specific

region as fuel is injected across the combustion chamber

Furthermore, the heat loss augmented from wall

impin-gement, as already seen in the heat release rate, reduces the

cylinder temperature which could promote soot formation

At such conditions, soot is favorable to be continuously

formed until soot oxidations initiated However, the

eight-hole injection produces a slightly lower soot emission than

that for ten-hole injection As the injector hole increases

from eight to ten, the previous results involve the absence

of wall impingement which produces lesser chance for

fuel-air mixture to be locally rich particularly near piston

bowl Furthermore, high temperature reacting zones is less

pronounced or more dispersed over entire regions Such

conditions are indeed favorable for suppressing soot

formations, which is opposite to the results of injector hole

changed from eight to ten

Figure 13 compares the brake specific fuel consumption

(BSFC) computed for three injectors with different numbers

of holes; this parameter is inversely related to the efficiency

of combustion The BSFC is greatest for the case of 10holes, indicating the worst combustion efficiency for thiscase

This is consistent with previous findings that combustiondoes not occur to a sufficient extent in the cylinder Thesix-hole injection system produces a BSFC comparable tothat of the eight-hole system despite heat loss from wallimpingement The lowest BSFC (highest combustion effici-ency) observed with the eight-hole injection system hassignificant implications for determining the optimal holenumber for improved engine performance and lowest sootemissions

4 CONCLUSIONS

(1) The concentration of liquid fuel, penetration length, andspray angle injected from a single injector decreasesmainly because of the reduced momentum of the liquidfuel as the hole number increases Nonetheless, liquidspray atomization and evaporation is promoted by thesmaller droplet size

(2) The eight-hole injection geometry yields the best bustion behavior and engine performance (the lowestBSFC) among all tested injections, which results fromthe competing effects of temperature influenced by thelocal distribution of the fuel vapor-air mixture and wallimpingement

com-(3) Lower soot emissions were observed with the hole injection compared with the six-hole injection pri-marily due to the combined effects of reduced temper-ature and the absence of fuel-rich zones There is slightpenalty on NOx emissions with the eight-hole injectionsystem, but the emissions are not significantly differentfrom those with the six-hole system

eight-(4) The optimal number of injector holes in the particularengine used for this study is determined to be eight whenparameters such as engine performance, combustion,and overall emissions are considered

ACKNOWLEDGEMENT− This work was supported by the second phase of the Brain Korea 21 program in 2010

REFERENCES

Abraham, J., Khan, A and Magi, V (1999) Jet-jet and wall interactions of transient jets from multi-hole injectors

jet-SAE Paper No 1999-01-0513

Baulch, D L., Drysdall, D D., Horne, D G and Lloyd, A

C (1973) Evaluated Kinetic Data for High Temperature Reactions Butterworth Cambridge

Deck, J E (1997) A conceptual model of D.I dieselcombustion based on laser-sheet imaging SAE Paper

No 970873

Fujita, Y., Umemoto, K., Kidoguchi, Y and Omae, K.(2007) A study on multi-hole spray interference andmixture formation in diesel combustion JSAE Paper

Figure 12 Temporal variation of average NOx and soot

emissions

Figure 13 Comparison of the brake specific fuel

consump-tion (BSFC) for different numbers of injector holes

No 20075129.

Han, U A., Hampson, G and Reitz, R D (1996)

Mech-anism of soot and NOx emission from a D.I diesel

engine SAE Paper No 960633

Halstead, M P., Kirsch, L J and Quinn, C P (1977) The

auto ignition of hydrocarbon fuels at high temperatures

and pressures-fitting of a mathematical model

Com-bustion and Flame, 30, 45−60

Jeong, D Y and Lee, J T (2005) Analysis of the suitable

injection pressure for diesel injection with high pressure

Int J Automotive Technology 6, 2, 87−93.

Lee, S H., Jeong, D Y., Lee, J T., Ryou, H S and Hong,

K (2005) Investigation on spray characteristics under

ultra-high injection pressure conditions Int J Automotive

Technology 6, 2, 125−131.

Lijun, S and Abraham, J (2003) Influence of wall

impin-gement on the structure of reacting jets SAE Paper No.

2003-01-1042

Patterson, M A., Kong, S C., Hampson, G J and Reitz, R

D (1994) Modeling the effects of fuel injection

charac-teristics on diesel engine soot and NOx emission SAE

Paper No 940523

Mauss, F., Magnusson, I., Balthasar, M and Karlsson, A

(1998) Simulation of soot formation under diesel engine

conditions using detailed kinetic soot model SAE Paper

No 981022

Mitroglou, N., Nouri, J M., Yan, Y and Arcoumanis, C

(2005) Spray characteristics of a multi-hole injector for

direct-injection gasoline engines Int J Engine Research

7, 3, 255−270.

Mitroglou, N., Nouri, J M., Yan, Y and Arcoumanis, C.(2007) Spray structure generated by multi-hole injectionfor gasoline direct-injection engines SAE Paper No.2007-01-1417.

Montgomery, D T., Chan, M., Chang, C T., Farrell, P V.and Reitz, R D (1996) Effect of injector nozzle holesize and number of spray characteristics and the perfor-mance of a heavy duty D.I diesel engine SAE Paper

No 962002.

Nehmer, D A and Reitz, R D (1994) Measurement ofthe effect of injection rate and split injections on dieselengine soot and NOx Emissions SAE Paper No 940668.Patrick, V F and Rhim, D R (2001) Effect of gas densityand the number of injector holes on the air flow sur-rounding non-evaporating transient diesel sprays SAE Paper No 2001-01-0532

Reitz, R D and Diwakar, R (1986) Effect of dropletbreakup on fuel sprays SAE Paper No. 860469.Timothy, P and Gardner, T P (1992) Investigation of theeffects of engine design parameters on diesel combus-tion and emissions using taguchi methods SAE Paper

No 920116

Tow, T C., Pierpont, D A and Reitz, R D (1994) ing particulate and NOx emissions by using multipleinjections in a heavy duty D.I diesel engine SAE Paper

Reduc-No 940897

Turns, S R (2000) An Introduction to Combustion, cepts and Applications 2nd Edn. McGraw-Hill NewJersey

EFFECT OF VARIOUS LPG SUPPLY SYSTEMS ON EXHAUST PARTICLE

EMISSION IN SPARK-IGNITED COMBUSTION ENGINE

J W LEE 1)* , H S DO 2) , S I KWEON 3) , K K PARK 3) and J H HONG 3)

1)Department of Mechanical Engineering, Soongsil University, Seoul 156-743, Korea

2)Undergraduate School, Department of Mechanical Engineering, Soongsil University, Seoul 156-743, Korea

3)Transportation Pollution Research Center, National Institute of Environmental Research, Incheon 404-707, Korea

(Received 1 February 2010; Revised 11 June 2010)

ABSTRACT− The particle size distribution and particle number (PN) concentration emitted by internal combustion engine are

a subject of significant environmental concern because of their adverse health effects and environmental impact This subject has recently attracted the attention of the Particle Measurement Programme (PMP) In 2007, the UN-ECE GRPE PMP proposed a new method to measure particle emissions in the diluted exhaust of automotive engines and a regulation limit (<6.0×10 11 #/km, number of particles) The specific PN regulation of spark-ignited combustion engine will be regulated starting on September 1, 2014 (EURO 6) In this study, three types of LPG supply systems (a mixer system and a multi-point injection system with gas-phase or liquid-phase LPG fuel) were used for a comparison of the particulate emission characteristics, including the nano-sized particle number density Each of the three LPG vehicles with various LPG injection systems contained a multi-cylinder engine with same displacement volumes of 2,000 cm 3 and a three-way catalytic converter The test fuel that was used in this study for the spark-ignited combustion engine was n-butane basis LPG fuel, which is primarily used for taxi vehicles in Korea The characteristics of nano-particle size distribution and number concentration of particle sizes ranging from 20 to 1,000 nm (aerodynamic diameter) that were emitted from the three LPG vehicles with various LPG supply systems were investigated by using a condensation particle counter (CPC), which is recommended by the PMP under both the NEDC and FTP-75 test modes on a chassis dynamometer The experimental results indicate that the PN emission characteristics that were obtained by the CPC system using the PMP procedure are sufficiently reliable compared to other regulated emissions Additionally, the sources of PN emissions in ascending order of magnitude are as follows: mixer type, gas-phase LPG injection (LPGi) and liquid-phase LPG injection (LPLi) passenger vehicles The liquid-phase LPG injection system produced relatively large particle sizes and number concentrations compared to the gaseous system, regardless of the vehicle driving cycle This phenomenon can be explained by unburned micro-fuel droplets that were generated due to a relatively short homogeneous fuel-air mixture duration in the engine intake manifold Also the particle number emissions from the LPG vehicle were influenced by the vehicle driving cycle.

KEY WORDS : LPG (liquefied petroleum gas), Mixer system, Gas-phase LPG injection system, Liquid-phase LPG injection system, Nano-sized particle, PMP (particle measurement programme)

1 INTRODUCTION

The Environmental Protection Agency (EPA) defines

ex-haust particulate matters (PM) as the mass of material

collected by the filtration of exhaust from an automotive

engine at a temperature less than 50oC, and the PM

con-tinues to attract attention due to its toxic health effects on

humans and other living organisms (SAE Handbook, 1994;

HEI Report, 2002)

In particular, it is known that the nano-sized particles can

be transported into human organ systems via the lungs and

can penetrate into the blood and brain barriers and can

cause cancer (Donaldson et al., 1998) Recently, more

interest has increased for particl sizes less than 1 µm in

diameter, and number concentration of particles that

origin-ate due to the incomplete combustion of fuel and enginelubricant (Kittelson, 1998; Kasper et al., 2001; Lee et al.,2008; Myung et al., 2009) The total mass of particulatesemitted from an automotive engine may be of less concernthan the specific emissions of the finest particles below 100

nm in diameter Of course, these engine particles can becharacterized by parameters such as the total number con-centration, particle size distribution by number density andchemical composition among the pollutants In a particlesize distribution with many fine particles, the aerodynamicdiameter can be considered as the characteristic dimension

of the particle size (ACEA Report, 2002) The primaryconstituents of particulate are solid carbon clusters, solubleorganic fractions (SOF), sulfates and inorganic species,such as sulfur dioxide The SOF adheres to particulates atcertain exhaust gas temperatures Regarding their size, theparticulates can be grouped into the nucleation mode with

particle sizes of less than 50 nm and the accumulation

mode with particle sizes in the 50 to 250 nm diameter range

(Kittelson, 1998; Pagan, 1999) This particulate is known

to vary in its size and number density according to the

sampling method used, including the length of exhaust pipe,

connection condition of the exhaust to a dilution tunnel and

dilution ratio Currently, conventional gravimetric PM mass

analysis is the international standard for measuring the

amount of PM emitted by combustion engines

Additional-ly, the CVS (constant volume sampling) method based on

the entire flow of exhaust gas is used for the weight PM

mass measurement on a chassis dynamometer Recently,

much attention has been given to the activity of the Particle

Measurement Programme (PMP) for a new measuring

system and standard level to replace the current PM mass

measuring system, which is intended to evaluate

nano-particle emissions for applying EURO 5+ and 6 regulations

from investigation results by the inter-laboratory

round-robin test that was organized by the JRC (Joint Research

Center of European Commission) in 2005, which included

Korea, Japan, Germany, France, Italy and England In 2007,

the UN-ECE GRPE PMP proposed a method for

measur-ing particle emissions in the diluted exhaust of automotive

engines and the regulation limit (<6.0×1011 #/km, number

of particles) Because of the introduction of this PN

regu-lation in compression-ignited combustion engines initially

with diesel fuel from September 1, 2011 (EURO 5+), the

specific PN regulation regarding the spark-ignited

combus-tion engine will be regulated starting on September 1, 2014

(EURO 6) (EU Legislative Acts and Other Instruments,

2007)

Alternative automotive fuels with a lower carbon content

than gasoline and diesel fuel, such as liquefied petroleum

gas (LPG) with a mixture of propane (C3H8) and butane

(C4H10), have been widely investigated for reducing exhaust

emissions Because this fuel evaporates easily and is well

established worldwide Specially, over 2.32 million vehicles

operate on autogas in Korea with 1,648 LPG filling stations

in 2008 (Korea LPG Association’s Report, 2009) LPG fuel

primarily consists of simple hydrocarbon compounds, and

the emissions from LPG-driven vehicles contain lower

levels of hydrocarbon compounds, nitrogen oxides, sulfur

oxides, air toxins and particulates Fuel supply systems for

LPG engines range from a simple mixer type (so called

first-generation) to a second-generation feedback mixer

type and a third-generation gas or liquid-phase port injection

type Typically, the second-generation LPG supply method

has difficulty in accurately controlling the air-fuel ratio, an

inferior response, a low output power of the LPG engine

and the possibility of backfire Therefore, the

third-gene-ration type, which injects LPG fuel based on multi-point

port fuel injection with gas-phase or liquid-phase, has been

developed and commercialized for reducing the exhaust

emissions by accurately and rapidly controlling the air-fuel

ratio and enhancing the engine power by increasing the

volumetric efficiency due to the evaporation heat of the

LPG fuel The LPG MPI system is on the market for a widerange of vehicles including passenger cars and city buses inEuropean countries, especially in the Netherlands (Van derSteen et al., 1996) Previous studies (Goto et al., 2001; AliKhan, 2006; Watson and Phuong, 2007; Sierens, 1992; Lee

focused on exhaust emissions except for the particle numberdensity from LPG engines with various LPG injection types,and several studies have reported a benefit with respect totoxic hydrocarbon emissions and ozone formation due tothe composition and CO2 emissions levels of LPG fuelcompared to gasoline fuel Some of the studies that com-pared the engine performance of an LPG engine with aliquid-phase injection type optimized for a well-knownphysical principle to deliver LPG fuel into the engine inletmanifold in exactly the same way as gasoline injection(both of the fuels are liquid) Another major breakthrough

in LPG injection technology, the direct injection (DI) scheme

is currently in development but is not economically rated into production in LPG vehicles (Boretti, 2009; Parkand Han, 2004), and several studies have reported animprovement in engine performance and a reduction in theemission of carbon monoxide and hydrocarbons However,several of these studies were performed on LPG engineswith combustion systems and control technologies that wereoptimized on a test bed in terms of spark timing advanceand equivalence ratio for the best catalytic conversion effici-ency and control of exhaust temperature for a substantialreduction in pollutant emissions, not including nano-sizedparticle number emissions

integ-Although there are many detailed studies regarding motive LPG engines, comparisons among various LPG injec-tion types with respect to nano-sized particle emissionanalysis under PMP measuring guidance are scarce In thisstudy, the effect of three LPG supply systems on nano-particle number emissions (20~1,000 nm diameter) isinvestigated, and the particle emissions are quantitativelycompared by applying two vehicle driving cycles (theNEDC and FTP-75 test mode) on a chassis dynamometer.Each of the three LPG vehicles used in this investigationhave multi-cylinder with same displacement of 2,000 cm3

auto-and a three-way catalytic converter All of the vehicleswere equipped with the latest generation of combustionchamber design and each of the LPG fuel delivery systems

2 EXPERIMENTAL APPARATUS AND TEST PROCEDURE

A gaseous mixer system has the advantage of availability

in LPG engines, but it is not compatible with current tronic port fuel injection engines In particular, a stoichi-ometric air-fuel mixture is allowed to stagnate in the intakemanifold, and combustion in the intake system, which iscalled back-fire phenomenon, can be caused by the provid-

elec-ed heat In this study, this feelec-edback mixer and port fuelinjection (PFI) with gaseous or liquid-phase injection systems

are compared In a PFI system with LPG fuel, fuel does not

occupy a significant volume in the intake, and the LPG

vaporizes as it mixes with intake air, which absorbs energy

from the air, lowers the air temperature and ultimately

increases its density As shown in Figure 1, the

liquid-phase LPG injection (LPLi) system consists of a fuel tank,

fuel pump, bottom-feed type injector, pressure regulating

unit and high-pressure fuel rail

The three LPG supply systems are used in 2.0 L

pro-duction automotive LPG engines, and each engine was

assembled onto a spark ignition engine base and naturally

aspirated The detailed specifications of the three LPG

engines are listed in Table 1

The LPG fuel used in this study is shown in Table 2 Thebutane concentration in the LPG fuel blend was on avolume basis to meet Korea’s domestic clean fuel require-ments (Korea Institute of Petroleum Management’s Webpage) The test facility and particle measurement systemrecommended by PMP in 2005 were used in this investi-gation (see Figure 2) The particle number measurementswere performed by taking the sample out of the dilutiontunnel close to the sample probe for the gravimetric mea-surement This system has two dilution stages Samplesover 2.5 mm diameter were rejected by the cyclone andimmediately diluted in first-stage particle number diluter.The first-stage particle number diluter was designed to dilutethe particle number concentration, and output a dilutesample equal to 150oC±5oC to avoid particle nucleation.The particles were heated in an evaporation tube at aconstant temperature of 300oC to remove volatile particles(SOF) After the dilution was cooled by the second-stageparticle number, the samples were sent to the condensationparticle counter (CPC), which continuously measured theparticles number concentration (Lee et al., 2008; Kim et

al., 2005) Additionally, an engine exhaust particle sizer(EEPS) was used to measure the size distribution of theexhaust particle emissions in the sub-micrometer rangewith the fastest time resolution available In this study, achassis dynamometer was used to measure the exhaustemissions from the light-duty LPG vehicles The specifi-cations of the chassis dynamometer are shown in Table 3.For the engine exhaust measurement, exhaust gases passedthrough a full-flow type constant volume sampler (CVS),

Figure 1 Schematic diagram of the LPLi system with

third-generation LPG injection technology

Table 1 Specifications of the three LPG engines (vehicles) with the same displacement volume

Combustion type Naturally aspirated 4-stroke, Spark-ignition, Dual Overhead Camshaft

Fuel injection type LPG mixer type

•Requires a vaporizer

•Throttle body with mixer

Multi-point gaseous-phaseLPG PFI (LPGi)

•Fuel returnless type

•Pinjection= 0.3 bar

Multi-point liquid-phaseLPG PFI (LPLi)

•Fuel return type

•∆Pinjection= above 5 bar

where they were diluted with filtered ambient air using a

HEPA filter Gas samples were taken from each of the three

CVS bags The bag of samples were analyzed by regulated

emissions (THC, NOx, CO2), which were measured

continu-ously in the undiluted exhaust gas with an exhaust gas

analyzer The HC was sampled using heated lines at a

temperature of 200oC As a reference, although the

back-ground level of the particle number measured in the CVS

was not available, the following was performed before

each test: (1) CPC zero check, (2) CPC high response, and

(3) Leak integrity check

To obtain a correct measurement of the particulate

matter, the sampling temperature was always controlled

within 50oC by the dilution process The PM that emitted

from the LPG vehicle was collected using a teflon-coated

glass fiber filter and weighed using a micro-balance after at

least 8 hours of conditioning in the chamber, which was

always controlled within a specified temperature and

humi-dity Figure 3 shows the experimental set-up with the

parti-cle measuring instrument and test passenger LPG vehiparti-cle

on the chassis dynamometer Each of the three ignited multi-cylinders, which were naturally aspirated LPG-fueled vehicles with either mixer or PFI fuel (gaseous orliquid-phase) injection, was coupled to a chassis dynamo-meter These tests were performed under transient condi-tions for the NEDC (ECE15+EUDC) and FTP-75 test modes

spark-in 3 times, respectively, for ensurspark-ing the test repeatabilitywith a maximum of coefficient of variation (COV) within5% in this experiment

3 RESULTS AND DISCUSSION

Figure 4(a), (b) and (c) show the real-time transient particlenumber concentrations emitted from each test vehicle with

a LPLi system, LPGi system and mixer system during theFTP-75 mode, which is typically used for the emissioncertification of light-duty vehicle in Korea, etc The blackline in these figures indicates a trace of vehicle speedthroughout the operating time for the FTP-75 driving cycle.Before the vehicle test, the vehicle remained at a test

Table 2 Physical and chemical properties and quality specifications of automotive LPG fuel

Specific gravity Liquid(water=1, 20oC) 0.501 0.579 Sulfur content (wt ppm) Below 100

Gas (air=1, 15oC) 1.522 2.006

Heat of vaporization (kcal/kg) 101.8 92.1 Corrosion of copperplate(40oC, 1 h) Below 1

Vapor pressure (kg/cm3, 20oC) 8.35 2.10 Vapor pressure (40oC, MPa) Below 1.27

Calorific value (kcal/kg)(kcal/L) 12,0306,110 11,8406,910 Residue after vaporization(mL/100 mL) Below 0.05

Stoichiometric air-fuel ratio (kg/kg) 15.71 15.49 (*) corresponding period of this research

Permanent tractive force (N) 6,000

Peak tractive force (N) 10,000

Electric inertia simulation (kg) 454~545

Axle distance (mm) 2,100~4,700 Figure 2 Schematic diagram of the particle measurementsystem constructed from PMP’s guidelines

temperature of 20~30oC for at least six hours As

previous-ly mentioned, the vehicle emissions were sampled duringthe cycle according the “constant volume sampling” techni-que and expressed in mg/km for each of the pollutants.Particle emissions between 20~1,000 nm, regardless of thetype of fuel injection, were emitted at the cold state of theLPG engine until 200 seconds After the engine warmed

up, this peak phenomenon did not appear, except in theLPLi system The LPLi vehicle emitted a total particlenumber concentration of 8.86×1010 #/km Additionally, theLPGi vehicle showed a lower particle emission trend thanthe LPLi vehicle and emitted a total particle number con-centration of 4.88×1010 #/km The LPG vehicle with amixer system exhibited a lower particle emission with atotal particle number concentration of 1.81×1010 #/km Thelower particle emission of mixer LPG vehicle is due tomore homogeneous air-fuel mixture

To investigate the effect of vehicle driving cycle onparticle emission, the NEDC mode (known as the MVEG-

A cycle), which is used for emission certification of duty vehicle in Europe, was used and a comparison of theFigure 3 Experimental set-up with the particle measuring

light-instrument and LPG-fueled passenger vehicle

Figure 4 Comparison of real-time particle emissions for

various LPG injection types during the FTP-75 mode Figure 5 Comparison of real-time particle emissions forvarious LPG injection types during NEDC mode.

real-time particle emissions for three LPG injection types

during the NEDC mode is shown in Figure 5 Although the

maximum vehicle speed of the NEDC mode is higher than

that of the FTP-75 mode, due to the shorter drivingduration compared to the FTP-75 mode, an overall lowerparticle number density was maintained compared to theFTP-75 Under the NEDC driving condition, the LPLivehicle emitted 7.5×1010 #/km and the LPGi vehicle emitt-

ed 2.8×1010 #/km in total particle number concentration.Furthermore, in the case of the LPG vehicle with a mixersystem, the total particle number density of 7.43×109 #/kmwas significantly lower than that of the PFI type, regardless

of the vehicle driving conditions

The difference in particle number distributions of thethree LPG vehicles is due to the various types of fuel injec-tion systems in the combustion engines

Figure 6 shows the effect of the type of LPG injection onthe particle size distribution at two vehicle driving modes.The NEDC mode consists of four elementary urban cycles(ECE15) and an extra urban highway cycle (EUDC) Anelementary urban cycle consists of a cycle with a duration

of 195 sec., and an extra urban cycle represents a speed duration In the NEDC mode, the type of PFI systemwith LPG fuel had a larger effect on the particle numberconcentration than the other type, and the greatest differ-ence was observed during the ECE15 period (0~195 sec.)

high-in the NEDC mode

The particle number emissions from the LPG vehiclewith the LPLi, LPGi and mixer system were reduced by15.4%, 42.6% and 58.9%, respectively, depending on thevehicle driving cycle

The particle number densities shown in Figure 4 andFigure 5 were analyzed with respect to the vehicle speed inFigure 7 At the same vehicle speed, the particle numberdensity in the FTP-75 mode is generally higher than theparticle number density in the NEDC mode, except for thecase of the LPLi system in the NEDC mode Similar to theentire FTP-75 cycle with three segments (cold start phase,transient phase and hot start phase) and an average vehiclespeed of 34.1 km/h, the particle emission section with avehicle speed below 20 km/h produced the highest particledensity only in the case of the LPLi system, and a secondpeak of the particle density was generated between avehicle speed of 20 km/h and 70 km/h And a third smallpeak appeared in the particle emission section with avehicle speed above 70 km/h More particles were emittedduring severe acceleration and deceleration of the testvehicle during the FTP-75 driving cycle In the NEDCmode, which is a combination of the ECE15 cycle with anaverage vehicle speed of 18.7 km/h and the EUDC cyclewith an average vehicle speed of 62.6 km/h, higher particleemissions by only one peak particle density were produced

at vehicle speeds below 50 km/h From the above results,the sources of particle emissions in descending order ofmagnitude for the LPG vehicles are as follows: mixersystem, LPGi system and LPLi system

Figure 8 shows a comparison of the correlation betweenthe PM on a mass basis, and the PN concentration emittedfrom the LPG vehicle with various LPG injection systems

Figure 6 Effect of the type of LPG injection on the particle

number spectral density for two vehicle driving modes

Figure 7 Comparison of the particle number spectral

den-sity to the vehicle speed for the NEDC and FTP-75 test

modes

for two driving test modes In general, the PM formation

process is characterized by the maximum soot

concent-ration during combustion The LPLi combustion is likely to

be related to rich mixture regions due to the imperfect

mixing of the stoichiometrically proportioned fuel and air

streams that were supplied to the combustion chamber

inside of the engine cylinder Ultimately, the mixer

com-bustion produced the largest differences between the PM

and PN emissions, which demonstrates the advantages of a

homogeneous fuel and air mixture with respect to

parti-culate formation and emission

Figure 9 shows a comparison of the particle diameter

and the total particle number concentration during first 200

sec in the transient-state vehicle driving period for the

FTP-75 and NEDC modes for the LPG vehicle with the

LPLi system Most particles appeared to be under about

200 nm in particle size diameter

And a difference in the particle number distributionsbelow 200 nm in the particle diameter was observed bet-ween the FTP-75 and NEDC modes Additionally, all ofthe size distributions exhibit a clear accumulation modewith a geometric number mean diameter in the 50~80 nm

of the peak particle number concentration

Figure 10 shows a comparison of the regulated exhaustemissions (CO2, NOx, THC) with various LPG injectionsystems for the FTP-75 and NEDC test modes The LPLivehicle was substantially lower (15% lower in CO2 emi-ssion and 20% lower in NOx emission) than the mixervehicle in the FTP-75 mode All of the THC levels werenoticeably lower than the EURO 5 limit of 100 mg/km,although the total hydrocarbon emission from the LPLisystem was slightly larger due to the vehicle driving cycle

In Figure 6 and Figure 9, the LPLi vehicle shows thehighest number distribution of particles with a particle dia-meter below 200 nm, including most above 50 nm (whichrepresents accumulation mode particles) in particle size.The LPLi technology is essential to obtain lower CO2

emission while maintaining the higher engine performancebecause the LPG fuel in this system does not occupy asignificant amount of volume in the intake system Thisadvantage over gaseous systems, including LPGi and mixertypes, removes the need for a forced induction mechanism

to maintain engine performance However, the greatestproblem with this type of liquid-phase injection system isassociated with an increase in the nano-sized particle numberemissions

4 CONCLUSIONS

Three LPG-fueled passenger vehicles, mixer system, phase LPG injection (LPGi) and liquid-phase LPG injec-tion (LPLi) system, with the same displacement volume of2.0L were used to investigate in a comparison of vehicleexhaust emissions, especially particulates emitted under

gas-Figure 8 Comparison of PM and particle number

concent-ration emitted from LPG vehicles with various LPG

injec-tion systems

Figure 9 Comparison of particle diameter and total particle

number concentration for 200 sec of the FTP-75 and

NEDC modes for LPG vehicle with LPLi system

Figure 10 Comparison of regulated exhaust emissions (CO2,NOx, THC) with various LPG injection systems betweenFTP-75 and NEDC test modes

two different vehicle driving cycle (FTP-75 and NEDC

modes) The main results obtained by this study can be

summarized as follows:

(1) Despite the lower CO2 emission while maintaining higher

engine performance, the liquid-phase LPG injection

system produced relatively large particle sizes and a

large number concentration compared to gaseous system

strategies (LPGi and mixer type), regardless of the

vehicle driving cycle Therefore, the LPG mixer system

resulted in considerably less particle number (PN)

emissions, although it has problems in meeting future

NOx emission standards because of the difficulty in

precisely controlling the air-fuel ratio

(2) Compared to the effect of the LPG vehicle speed, the

total PN concentration is affected by the test vehicle

driving cycle, such as the FTP-75 and NEDC modes

However, the sources of particle emission in

descend-ing order of magnitude for LPG vehicles are as

follows: mixer system, LPGi system and LPLi system,

regardless of the vehicle driving cycle

(3) For the LPLi vehicle, the particle size distribution that

was observed under approximately 200 nm in particle

size diameter was nearly single modal in accumulation

and emitted a total particle number concentration of

8.86×1010 #/km Consequently, all of the LPG vehicles

with various LPG supply systems are capabile of

satis-fying the EURO 6 regulation level regarding PN

emi-ssion limits (<6.0×1011 #/km) in spark-ignited internal

combustion engines

ACKNOWLEDGEMENT− This research was conducted with

the sponsorship of the joint research project in National Institute

of Environmental Research under Ministry of the Environment of

Korea, 2009.

REFERENCES

ACEA Research Report (2002) Programme on emissions

of fine particles from passenger cars ACEA

Ali Khan, M (2006) Optimization of LPG throttle-body

injection system for improved efficiency emissions and

air-fuel mixing, FISITA Paper No 20068154

Boretti, A (2009) Development of a direct injection high

efficiency liquid phase LPG spark ignition engine SAE

Paper No 2009-01-1881

Donaldson, K., Liu, X Y and MacNee, W (1998)

Ultra-fine (Nanometer) particle mediated lung injury J Aerosol

Science,29, 553−560

EU Legislative Acts and Other Instruments (2007)

Regu-lation of the European Parliament and of the Council on

type approval of motor vehicles with respect to emissions

from light passenger and commercial vehicles (Euro 5

and Euro 6) and on access to vehicle repair and

main-tenance information EU Legislative Acts and Other

Instruments, PE-CONS 3602/07

Fuel quality standards for Liquefied Petroleum Gas in

Korea Korea Institute of Petroleum Management, (http:/

/www.kpetro.or.kr/eng/standards/standards_02.jsp)

Goto, S., Wakasa, R and Lee, D (2001) Research trends

of LPG-fueled engine system J Society of Automotive Engineers of Japan 55, 5, 30−37

Health Effects Institute (2002) Understanding the Health Effects of Components of the Particulate Matter Mix: Progress and Next Steps. Health Effects Institute.Kasper, M., Matter, U and Burtscher, H (2001) NanoMet:On-line characterization of nanoparticle size and com-position SAE Paper No 2001-01-1998

Kim, H S., Kim, S S., Cho, J M., Cho, G B., Sung, N W.and Jeong, Y I (2005) Measurement of size distribution

of diesel particles: Effects of instruments, dilution methodsand measuring positions Int J Automotive Technology

6, 2, 119−124

Kittelson, D B (1998) Engines and nanoparticles: A review

J Aerosol Science, 29, 575−588

Korea LPG Association (2009) Statistical Review of Global

LP Gas Korea LPG Association 2009-Report Lee, E J., Park, J W., Huh, K Y., Choi, J J and Bae, S D.(2003) Simulation of fuel/air mixture formation for heavyduty liquid phase LPG injection engines SAE Paper No.2003-01-0636

Lee, J W., Jeong, Y I., Jung, M W., Cha, K O., Kwon, S.I., Kim, J C and Park, S S (2008) Experimental investi-gation and comparison of nano-particle emission charac-teristics in light-duty vehicles for two different fuels Int.

J Automotive Technology 9, 4, 397−403

Mizushima, N., Sato, S., Yamamoto, T., Konagai, G., Ogawa,Y., Sawut, U., Kawayoko, K and Takigawa, B (2009).Combustion characteristics and performance increase of

an LPG-SI engine with liquid fuel injection system

Pagan, J (1999) Study of particle size distributions

emitt-ed by a diesel engine J Fuels and Lubricants (SAE Transactions) 108, 4, 557−562

Park, K and Han, S (2004) The ignition characteristics forthe direct liquid injection combustion in LPG engine

FISITA Paper No 20042357

SAE Handbook (1994) Chemical methods for the ment of nonregulated diesel emissions SAE 1994-Hand- book

measure-Sierens, R (1992) An experimental and theoretical study

of liquid LPG injection SAE Paper No 922363.Van der Steen, M., De Rijke, J and Seppen, J J (1996).Stoichiometric and lean burn heavy-duty gas engines: Adilemma between exhaust emissions and fuel consumption

TNO-Paper No VM9605

Watson, H C and Phuong, P X (2007) Why liquid phaseLPG port injection has superior power and efficiency togas phase port injection SAE Paper No 2007-01-3552

NEW DESIGN OF A ROAD PROFILER

M DEMI 1) , D DILIGENSKI 2)* , M DEMI 3) and I DEMI 1)

1)University of Kragujevac, Mechanical Engineering Faculty, 34000 Kragujevac, Serbia

2)Vinèa Institute of Nuclear Sciences, Department for Engines and Vehicles, P.O Box 522, 11001 Belgrade, Serbia

3)Ikarbus, Development Department, 11000 Belgrade, Serbia (Received 16 February 2009; Revised 27 April 2010)

ABSTRACT− Macro and micro road profiles are of significant importance for vehicular motion studies, reliable calculations

of vehicle system properties, and ensuring vehicular safety As such, road profiles should be considered carefully Macro profiles consider the spatial geometry of the road (curves, longitudinal and lateral slopes) while micro profiles consider roughness in longitudinal and lateral directions These profiles have random characteristics that can be quantified under on- road and off-road conditions using a road profiler This paper presents an analysis of a new concept for a universal profiler without gyroscopic stabilizers

KEY WORDS : Road, Profile, Profiler, Optimization

1 INTRODUCTION

The design of a motor vehicle should always be preceded

by a thorough analysis of mutual relations that occur in the

complex dynamic systems such as the

driver-vehicle-environ-ment (DVE) (Demi , 2008; Ha aturov et al., 1976)

Dis-agreement among any of the DVE links could lead to a

disturbance of the whole system with serious consequences

Therefore, it is necessary to analyze mutual links within the

DVE system in the early stages of vehicle design

Information that a driver receives from the environment

and the vehicle includes macro and micro road roughness,

vibration and subsystem noise, deviation of vehicle position

from the desired trajectory, and torque on the steering wheel

(Demi , 2008; Gillespie, 1992; Ha aturov et al., 1976;

Milliken and Milliken, 1995) This paper will consider the

possibility of experimentally determining road profile

para-meters

Macro road profiles (Demi , 2008; Demi and Diligenski,

2003; Ha aturov et al., 1976) take into consideration spatial

forms (curves, longitudinal, and lateral slopes), while micro

profiles deal with road unevenness Macro profiles depend

on the geographic location of the road (i.e., mountains or

flat-land) and the road category (the higher the category the

lower the intensity of curves and slopes in both directions)

Micro profiles are mainly determined by the road surface

type and its state (Braun, 1969, 1971; Demi , 1981, 1982,

2003, 2006; Fiala, 1967; Genta, 2003; Georg, 2004; Gillespie,

1992; Mitschke, 1957, 1972; Heidenreich and Hunt, 1998;

Rotenberg, 1972; Sayers and Karmihas, 1998; Ellis, 1994)

This paper will give a description of a road profiler capable

of recording both macro and micro on-road and off-roadprofiles simultaneously

During motion, positions of vehicle masses vary withinthe referent coordinate system The motions are influenced

by the road profile (an external factor) and by inertial characteristics, engine torque and vehicle speed(internal factors)

vibration-In the literature, there are descriptions of devices

design-ed for measuring macro road profiles basdesign-ed on classicalgeodetic measurements made with a gyroscope or GPS(Demi , 2006; Demi and Diligenski, 2006)

An alternative approach to road profile determination isneeded because of the following deficiencies of gyroscopi-cally stabilized platforms:

• A regular mechanical gyroscopically stabilized platform

is expensive compared to a set of linear accelerometers

• Rotating gimbals in a mechanical gyroscope produce tion and therefore errors caused by real precession

fric-• Drift characteristics of momentum wheel gyroscopes arestrongly affected by bearing torques Therefore, thesegyroscopes are often designed with innovative bearingtechnologies (e.g., gas, magnetic, or electrostatic bearings)

• Use of more expensive higher performance componentsquickly drives up the system costs Even if the most ex-pensive components are used, effects such as bearing sealfriction, cable wraps and base motion may still cause un-acceptable system jitter Use of second set of fine gimbalswill produce dramatic increases in platform stability but

at a price of at least double or that of the original

gimbal-ed system Furthermore, system size and weight areusually also doubled

• Base costs of laser ring gyroscopes are greater than basecosts of mechanical gyroscopes

c

*Corresponding author. e-mail: diligens@vinca.rs

• Laser ring gyroscopes are susceptible to an error known

as “Lock-In” when the rate of turn is very small

In practice, two systems for micro road profile

measure-ment have been produced, both using gyroscopes for

stabili-zation of accelerometers (Demi , 1981) Under large

vibra-tions and in the presence of dust this equipment tended to

become unstable, rendering the profile-meter useless without

corrections for the spatial position of the accelerometer

This vibration creates difficulties, particularly if the tests

are carried out on low quality roads In light of these

difficulties it will be beneficial to have a cheaper and more

reliable solution that is not based on gyroscopic equipment

The performed analyses resulted in the development of

profilers capable of simultaneous recording of macro and

micro profiles using devices for measurement of micro road

profile (Demi , 2003c), to be explained in what follows

Figure 1 shows a spatial macro profile and Figure 2

shows micro profiles in the longitudinal and lateral directions

Figure 1 shows the spatial geometry in the longitudinal

road direction (curves, acclivities and declivities, and lateral

slopes) Figure 2 shows that the road surface varies in both

road directions In the literature, terms like longitudinal

profile and lateral profile are often used (Sayers and

Karmihas, 1998)

Many researchers agree that the macro road profile

belongs to the group of stochastic processes requiring

stati-stical parameters for treatment A micro profile is identified

on the basis of micro roughness

2 METHOD

A large number of profilers have already been described in

the literature (Braun, 1969, 1971; Demi and Diligenski,

2003; Sayers and Karmihas, 1998; Simiæ et al., 1973;

Simiæ, 1988) During motion the device (vehicle) varies itsspatial position in an arbitrarily chosen fixed coordinatesystem (often called inertial or global) (Pars, 1971) Toanalyze profiler motion parameters, it is necessary to knowthe displacement and the velocity or acceleration of thedevice at characteristic points For simplicity, the device isoften regarded as a solid body in spatial motion To de-scribe the type of motion, it is necessary to know 6 gene-ralized coordinates: 3 translations and 3 rotations (Bendat,1998), shown in Figure 3 in a simplified form

The easiest way of performing the analysis is to observethe translational motions of the center of mass (point C,which should be as close as possible to the CG of themeasuring trailer in static position, Figure 3), which isdefined by coordinates Xc, Yc and Zc with reference to thefixed coordinate system OXYZ and 3 rotations around theaxes of the moving coordinate system (Cxyz, firmly con-nected to the vehicle) (Pars, 1971)

For vehicles, the mentioned angles of rotation are known

as roll angle (around the longitudinal axis), pitch angle(around the lateral axis) and yaw angle (around the verticalaxis) These terms will also be used in the case of theprofiler

For easier description of the measurement method of thedeveloped profiler, a brief review of the basic kinematicterms, in the general case of solid body motion, will begiven (Pars, 1971)

According to Figure 3, a position vector of an arbitrarilychosen point, M, on a solid body, in case of a generalspatial motion, is given by expression:

(1)where the distance between points M and C is a constantvalue

Speed and acceleration of point M can be calculated bymeans of the first and second derivative of Equation (1).The parameters of solid body motion, shown in Figure 3,can be determined by application of matrix calculus (Pars,1971) In that case, transformation matrices betweencoordinate systems OXYZ and Cxyz have to be known Thistheory is also well known and the problem is easily solved

by using the software package Mathematica 5.2 (WolframResearch, Champaign IL USA) For the calculation of the

c ó

c

ó

c ó

Figure 1 3-D view of macro road profile specified in two

measurement points, MP1 and MP2

Figure 2 Macro road profile in longitudinal and lateral

directions Figure 3 Coordinate system used for definition of spatialposition of a solid body

motion parameters in the general form, the software

pack-age NEWEUL, developed mainly for analyses of the solid

body dynamics, should be applied

The concept for the profiler without a gyroscope is

presented in Figure 4 A trailer without flexibly suspended

wheels is used as a basis of the device The tire diameter

and air pressure should be as close as possible to those of

the real vehicles

In accordance with the rules of NEWEUL software and

for description of the motion of an arbitrarily chosen point,

five coordinate systems have been introduced including:

the global-inertial system (OXYZ), which is also regarded as

the reference coordinate system; the moving coordinate

system (Cxyz), which is firmly connected to the vehicle

body; the coordinate system originated at the observed point,

which has axes parallel to those of the moving coordinate

system due to the firm link with the vehicle body; and two

auxiliary coordinate systems

Parameters of the motion of the observed point are

calculated with reference to the global coordinate system

(OXYZ) and can be projected to any other coordinate system,

including the moving coordinate system (Cxyz), by the use

of aforementioned software (Mathematica 5.2)

The aim of this paper is to determine the motion

para-meters of a device without using a gyroscopically stabilized

platform Because the spatial position is fully described by

six generalized coordinates, their values can be calculated

on the basis of the six known displacements, speeds or

accelerations of any point that belongs to the solid body

The easiest method is to record six projections of the

accelerations of these points

Selection of the number and position of measurement

points is arbitrary, but the simplest method is to choose two

characteristic points (1 and 2) on a specially designed rigid

frame that is firmly connected to the trailer

The accelerations of these two points are measured in three

directions (Figure 4) with reference to the moving

coordi-nate system, {a1,b1,c1} and {a2,b2,c2} To facilitate the

ana-lysis, it is convenient to locate these points symmetricallywith respect to point C (Figure 4, “P”) Values of a, b and cmay be chosen arbitrarily, but it is preferable that they are

as high as possible The components of the measuredaccelerations are also shown in Figure 4

Accelerations are recorded using either inductive or electric sensors Each of these sensors has advantages anddisadvantages, which are described in detail in (HBM, 1980;Jovanovi and Pijev evi , 2006; Regelungs und Messtechnik,2006) Inductive sensors can easily be calibrated, the lowerfrequency is equal to zero; however, they are sensitive toaccelerations that are not along the measurement axis andthey have a relatively low upper frequency limit Piezo-electric sensors are less sensitive to acceleration compo-nents not aligned with the measurement axis, but calib-ration is more complicated and the lower frequency is notzero These sensors are also less sensitive to shocks and aresuitable for applications as three-axis sensors

piezo-Because the sensors are installed on a rigid frame that isfirmly connected to the trailer body, they record accele-rations in directions of the axes of the moving coordinatesystem In practice, displacements of the observed pointsare calculated by double integration of the measuredaccelerations

Using NEWEUL software, the displacements of points 1and 2 are calculated in directions of the axes of the movingcoordinate system as follows:

(2)where

X C, Y C, Z C are unknown values of profiler C.G nates with respect to the fixed coordinate system, ϕ, θ, ψare unknown values of roll, pitch and yaw of the trailerwith respect to the moving coordinate system, a 1,2, b 1,2, c 1,2

coordi-are coordinates of measurement points 1 and 2 with respect

to the moving coordinate system, x 1,2, y 1,2, z 1,2 are knownvalues of displacements of points 1 and 2 in directions ofthe axes of the moving coordinate system, calculated bydouble integration of the measured accelerations

The accelerations of points 1 and 2 in the direction of aparticular axis are measured using appropriate sensors,whose output values are electrical voltages Between thesevalues and in the recommended frequency domain of sensorapplication, the following relations apply:

where

a x,y,z components of accelerations in x, y, z directions,

u x,y,z voltages proportional to the acceleration components

in x, y, z directions (the output voltages are usually low andhave to be amplified, therefore the voltages are recorded as

c

x 1 2 , =X C cosψ cosθ +Y C sinθ sinϕcosψ +

Y C sinψ cosϕ−Z C sinθcosϕcosψ +Z C sinψ sinϕ ±a 1 2 ,

y 1 2 , =−X C sinψ cosθ−Y C sinψ sinθsinϕ+

Y C cosψ cosϕ−Z C sinψ cosθcosϕ+Z C sinϕcosψ ±b 1 2 ,

z 1 2 , =X C sinθ−Y C sinϕcosθ +Z C cosθcosϕ ±c 1 2 ,

a x y z , , =k x y z , , u x y z , ,

Figure 4 Determination of the motion parameters of the

characteristic profiler points

the amplified output), and k x,y,z coefficients of

proportion-ality for the whole measurement chain, from the sensor to

the system data memory

As previously mentioned, the displacements of points 1

and 2 can be calculated by double integration of the

record-ed accelerations as follows:

(4)

Expression (2) consists of six equations with six

un-knowns (X C, Y C, Z C, ϕ, θ, ψ) and six known displacement

variables in directions of the axes of the moving coordinate

system (x 1, x 2, y 1, y 2, z 1, z 2) Because the coordinates of

points 1 and 2 (a 1,2, b 1,2,and c 1,2) are known, it is possible to

calculate profiler motion parameters on the basis of the

known motion components of points 1 and 2 obtained by

the measurement system

The equations in Expression (2) are transcendental,

coupled, and nonlinear They can be simplified by

lineari-zation for the case of small rotation angles However,

because yaw angle is non-negligible in service conditions,

this approach could not be applied Hence, this system of

equations has to be solved numerically

For this purpose, the method of linear iterations did not

give acceptable results due to poor convergence Therefore,

an optimization procedure, based on the application of

Hooke-Jeeves method (Bunday, 1984; Demi , 1990, 1991,

1994, 1996b, 1997), was adopted The squared difference

between the calculated displacements of points 1 and 2,

given by Expression (2), and the measured displacements,

given by Expression (4), was minimized More

specifi-cally, the criterion function is of the following form:

(5)where indices c and m indicate the calculated and measured

values of the corresponding displacement components,

respectively Minimization of Expression (5) was

perform-ed for each discrete value of the calculatperform-ed and measurperform-ed

displacement components (1−N), with the following

limit-ations:

(6)

A block diagram of the optimization procedure is shown

in Figure 5 The calculation can be performed at any point

for which data of the six measured displacements are

known The end of the iterative process for each point is

reached when the difference between successive values of

the criterion function becomes less than 10−20 In the

optimization phase, the method of an external penaltyfunction, explained in detail in (Bunday, 1984; Demi , 1990,

1991, 1994, 1996b, 1997), has been applied All the lations were performed using specially developed softwarewritten in Pascal

calcu-Due to a lack of experimental data, a time series based

on random numbers evenly distributed in the domain {0,1} was used for testing the developed method and theprincipal design of the profiler The applied time seriesconsisted of 4096 points and a sample increment of 0.02 senabled reliable calculation in the frequency domain from0.0122 to 25 Hz (Bendat and Piersol, 1980, 2000; Bendat,1998) An illustrative example for the roll is given inFigure 6 The test is rigorous because of intensive variation

of the simulated experimental data Estimation of thedeveloped procedure should be done by use of statisticalmethods Note that analyses involving more data points(and lower level of reliable frequency) showed the samebehavior This consistency is important for applications,

x 1 2 , =k x ∫∫a x 1 2 , dt

y 1 2 , =k y ∫∫a y 1 2 , dt

z 1 2 , =k z ∫∫a z 1 2 , dt

c ó

Figure 5 Block diagram of the developed procedure forcalculation of the profiler motion parameters

Figure 6 Illustration of the time series of body roll

where low frequencies are of major significance.

The correlation function, often used to obtain the relation

between two signals in the time domain, is described briefly

Let the signals x(t) and y(t) (Figure 7) be known The

cross-correlation function facilitates analysis of the

vari-ation of y(t) with respect to x(t) for various values of the

time delay parameter, τ The mathematical definition of the

cross-correlation function is given by expression (Bendat,

1998; Bendat and Piersol, 1980, 2000):

(7)

or in the discrete form:

(8)

In practice, one does not have an infinitely long signal;

thus, calculation errors appear According to Bendat (1998),

Bendat and Piersol (1980, 2000), the normalized standard

error is given by expression:

(9)where:

B is the frequency bandwidth, and T is the length of the

sample functions x(t) and y(t)

In the specific case of this manuscript, this error is 0.781,

which is acceptable

In the first stage of testing, cross-correlations were culated using ANALSIGDEM (Demi 2003a) An illustra-tive example is shown in Figure 8

cal-Analysis of cross-correlation functions of measured andcalculated variables indicates that they tend to decreasing

in time, confirming that the dependence of the observedpair of variables can be regarded as stationary, i.e there is

no increase of their difference in time

In the course of an experiment, the response of a dynamicsystem is often influenced by the errors due to noisepresent during the experiment Figure 9 shows a schematicdiagram of this effect:

The input to the system is denoted by x(t), the ideal,faultless, output of the system is v(t), r(t) is measurementerror, and y(t) is system output in the presence of an error

In cases like this, an estimate of the relation x(t) vs y(t) isobtained by using the ordinary coherence function The coherence function can be calculated on the basis ofthe complex characters auto-spectrum and cross-spectrum.There are several procedures for calculation of auto- andcross-spectra described in the literature The procedurebased on Wiener-Khinchine relations (Bendat, 1998, Bendatand Piersol, 1980, 2000) is presented as follows:

(10)where R xx, R yy, R xyare auto-correlation and cross-correlationfunctions, respectively; f is frequency; t is the time delay;and S xx, S yy, S xy are single sided auto-spectra and cross-spectrum, respectively The other symbols are known.The function of ordinary coherence is defined by theexpression

(11)Coherence measures the percentage of power in signal

“y” that is caused by (phase coherent with) the power insignal “x” (the input) Coherence is a dimensionless value,which varies from 0 to 1 If the coherence is 1, all thepower of the output signal is due to the input signal If thecoherence is 0, the input and output are completely random

-Figure 8 Test of the cross-correlation function and the

calculated vehicle roll

Figure 9.Schematic diagram of the response of a dynamicsystem in the presence of noise

with respect to one another.

Values of the coherence function less than one are

possible if some of the following situations occur:

A) Measurements are contaminated by non-correlated

noise

B) Measurements are contaminated by an additional

external signal source

C) The system is non-linear

D) Additional inputs are present in the system

E) Error leakage is not reduced with windowing

It is well known that the auto-spectra and cross-spectra

are calculated by averaging ensembles of functions (Bendat

1998, Bendat and Piersol, 1980, 2000) An infinitely long

signal is never available, so calculation of the coherence

function is always performed with a normalized standard

error defined by:

(12)where γ xy is the ordinary coherence function, and n d is the

number of ensemble averages (the number of realizations

per ensemble)

In the specific case of this manuscript, the normalized

standard error was approximately 0.00791, which could be

accepted (Bendat, 1998, Bendat and Piersol, 1980, 2000)

Given the character of the adopted test times, it has been

concluded that calculation of the coherence functions of the

measured and calculated variables should be performed

using DEMPARKOH (Demi , 2003b), shown in Figure 10

The data presented in Figure 10 shows that the

coher-ence functions, calculated with 1024 averages, reach very

high levels (above 0.9) This intensity indicates that the

measured and calculated values of displacements of points

1 and 2 have very similar frequency content This ment confirms a high reliability of the developed methodfor calculation of the profiler motion parameters.The mean values of the signal were not included in theanalysis This elision resulted in the appearance of higheramplitudes in the spectra near 0 Hz Therefore, the valuesaround 0 Hz in Figure 10should not be considered

measure-On the basis of the performed analyses, it can beconcluded that the developed method for measuring thespatial motion of a solid body without using expensivegyro-stabilized platforms can provide a basis for the design

of a profiler for identification of both macro and micro roadprofiles For this method, standard accelerometers and ap-propriate amplifiers are applied, which is far more conveni-ent for measurements under service conditions Consequently,the method is applicable for registration of macro andmicro profiles for both on-road and off-road conditions(with significant space angles) This result was not possible

to achieve with existing devices, with the partial exception

of GPS However, GPS cannot effectively detect microroad profiles

3 PRACTICAL APPLICATION OF THE METHOD

The analyses have shown that the applied procedure fordetermining spatial motion parameters of a solid body, aswell as the developed software, can be applied successfullyfor determination of the macro and micro on-road and off-road profiles

By solving Equation (2), the values of Xc, Yc, Zc, ϕ, ψand θ can be calculated and used to define the spatialposition of the device To determine a macro road profile, it

is necessary to use all of these values

For calculation of the micro profile parameters, thefollowing expressions can be used (Fiala, 1967):

(13) (14)where Z l, Z r are vertical displacements of the wheel centers,respectively, and l p is the device’s wheel track

Expressions (13) and (14) enable calculation of thevertical amplitudes of micro roughness in wheel centers, Z l

and Z r, on the basis of the known values of Z c and ϕ. It isimportant to emphasize the fact that the measuring trailercan identify micro roughness in wheel centers, which isclose to real vehicle conditions, as well as with the devicesalready in application In order to minimize the influence ofwheel stiffness and dimensions, it is necessary to usewheels and tire pressures that are similar to the majority ofvehicles on the road

Wheel dimensions (diameter and width) and stiffness of

ε = 2 1 γ ( – xy2)

γ xy n d

-c ó

Zc=Z l + Z r

2 -

ϕ =Z l – Z r

l p

-Figure 10 Coherence functions of the vehicle motion:

F10.1 translational motion (x-direction),

F10.2 translational motion (y-direction),

F10.3 translational motion (z-direction),

F10.4 rotational motion (roll),

F10.5 rotational motion (pitch),

F10.6 rotational motion (yaw)

the tires influence the measurement results For illustration,

Figure 11 shows the influence of the radius of a stiff wheel

on the ratio of the wheel center acceleration to the

acceleration due to micro roughness of the road, assuming

a constant roughness height (Mitschke, 1957; Fiala, 1967;

Sayers and Karmihas 1998; Simi , 1973)

From Figure 11 it can be seen that an increase in the

wheel radius increases the measurement error of the

acceleration due to micro roughness of the road

The influence of wheel elasticity and radius to the

mea-surement results is illustrated in Figure 12 (Mitschke, 1957;

Fiala, 1967; Sayers and Karmihas, 1998; Simi , 1973) By

analyzing the data in Figure 12, one can establish that both

the radius of the wheel and the radial elasticity of the wheel

influence the measurement results

4 CONCLUSIONS

Given the shortcomings of gyroscopic equipment, it would

be useful to develop a new method for measuring macroand micro road profiles In this study, an instrument of thiskind was not realized, but the feasibility of the method hasbeen verified by applying thorough tests More precisely,the concept of the instrument has been formulated, but theinstrument has not been developed For this reason, wecould not compare the characteristics of the proposed devicewith those of existing devices However, the application ofthe proposed concept will be simple because it is based onmeasurements of acceleration by commercially availableaccelerometers Moreover, it is expected that the measuringdevice will be less expensive than existing equipment.Based on the performed analyses, the following can beconcluded:

(1) The developed method enables indirect calculation ofthe six motion parameters of the device (three trans-lations and three rotations) The performed tests, whichwere based on dynamic simulation using random timeseries, have shown a high reliability of calculated devicemotion parameters Thus, the developed procedure can

be used for experimental determination of road profileswithout requiring a gyroscopically stabilized platform.(2) The suggested profiler concept retains the positive fea-tures of existing designs while eliminating the disadvant-ages The realization of this device will enable reliableassessment of road profiles in both on-road and off-road conditions

Ana-Bendat, J S and Piersol, A G (1980) Engineering cations of Correlation and Spectral Analysis John Wileyand Sons London

Appli-Braun, H (1971) Auslegung Elektro-hidraulischer unebenheits-simulation ATZ

Fahrbahn-Braun, H (1969) Untersuchungen von heten und Anwendungen der Ergebnisse Ph.D Disserta-tion, TU Braunschweig

Fahrbahnuneben-Bunday, P (1984) Basic Optimization Methods.woode Ballantyne Colchester and London

Spottis-Demi , M (1997) Identification of vibration parametersfor motor vehicles Veh Syst Dyn., 27, 65−88

Demi , M (1996) Optimization of characteristics of damping elements from aspect of oscillatory comfortand vehicle handling Int J Veh Des 17, 1, 76−91Demi , M (2003a) ANALSIGDEM-Software for signalanalysis, http://www.ptt.rs/korisnici/i/m/imizm034/

elasto-c ó

c ó

c ó c ó c ó

Figure 11 Measurement error vs wheel radius

Figure 12 Influence of the radius and radial elasticity of a

wheel on micro roughness measurements

Demi , M (2003b) DEMPARCOH: Software for partial

coherence function calculation http://www.ptt.rs/

korisnici/i/m/imizm034/index.swf

Demi , M (1994) Optimization of vehicles

elasto-damp-ing elements characteristics from the aspect of ride

com-fort Veh Syst Dyn., 23

Demi , M (1991) A contribution to optimization of vehicle

seats Int J Veh Des., 5/6, 618−629

Demi , M (1990) A contribution to the optimization of

the characteristics of elasto-damping elements of

passen-ger cars Veh Syst Dyn., 19, 3−18

Demi , M (1996) Analysis of influence of design

para-meters on steered wheels shimmy of heavy vehicles Veh.

Syst Dyn., 26, 343−379

Demi , M (1997) The optimization of vibration parameters

of motor vehicles Monograph, Mechanical Engineering

Faculty in Kragujevac (in Serbian).

Demi , M (2008) Cybernetic system:

Human-vehicle-environment Research Centre of Serbian Academy of

Sciences and Art and University of Kragujevac

Demi , M (1982) The experimental identification of

longitudinal micro profile unevenness in Yugoslavian

roads (in Serbian) MVM, 42

Demi , M (1981) A method of identification of

longi-tudinal unevenness of road micro profile from the aspect

of car vibration excitation (in Serbian) Symp Protection

from Noise and Vibration in Living and Working

Environ-ment, Belgrade

Demi , M and Diligenski, Dj (2003) The road surface

profile investigation perspective The 30' Session, Modern

Technologies in XXI Century, Bucharest, Romania, CD,

35-42

Demi , M (2003c) A contribution to design of road

profiler, Tehnika (Mašinstvo) 52, 4/5 (in Serbian)

Demi , M (2006) Dynamic excitations of a car, Vin a

Institute of Nuclear Sciences, Centre for Engines and

Vehicles (in Serbian)

Demi , M and Diligenski, Ð (2006) Satellite navigation

of vehicles Vin a Institute of Nuclear Sciences, Centre

for Engines and Vehicles (in Serbian)

Ellis, J R (1994) Vehicle Handling Dynamics Mechanical

Engineering Publications Limited London

Fiala, E (1967) Strassenwelligkheit, Fahrkomfort und

Radlastschwankung Strasse und Autobahn.Genta, A (2003) Motor Vehicle Dynamics Politecnika diTorino

Georg, R (2004). Vehicle Dynamics Lecture Notes hochschule Regensburg.

Fach-Gillespie, T (1992) Fundamental of Vehicle Dynamics.SAE.

Ha aturov, A A., Afanasev, V L., Vasiljev, V S., Goljdin,

G V., Dodonov, B M., Žigarev, V P., Koljcov, V I., Jurik,

V S and Jakovljev, E I (1976) Dinamika Sistemi: Doroga-Shina-Avtomobilj -Voditelj Mašinostr Moscow(in Russian)

Heidenreich, W and Hunt, L (1998) Pavement Smoothness Indices Oregon Department of Transportation

Hottinger Balldwin Messtechnik (1980) Elektrisches Messen Mechanischer Groessen.

Jovanovi , S and Pijev evi , D (2006) Comparison ofresponse of piezoelectric and inductive accelerometers

on mechanical shocks and post process displacements

Vojnoteh Glas 1/2006, 97-101 (in Serbian)

Miliken, W and Miliken, D (1995) Race Car Dynamics.SAE.

Mitschke, M (1972) Dynamik der Kraftfahrzeuge Springer.Mitschke, M (1957) Messung von Strassenunbenheiten und deren Einfluss auf Fahrzeug und Fahrbahn Ph.D.Dissertation TU Braunschweig

NEWEUL (2000) Manual TU Stuttgart

Pars, L (1971) A Treatise on Analytic Dynamics Heinemann.London

Regelungs und Messtechnik (2006) Dynamik Test Systems.Hamburg Germany

Rotenberg, V (1972) Vehicle Suspension Mašinostr Moscow(in Russian)

Sayers, M and Karmihas, S (1998) The Little Book of Profiling-Basic Information about Measuring and Inter- preting Road Profiles University of Michigan

Simi , D (1988) Motor Vehicle Dynamics. Nauèna Knjiga,Belgrade (in Serbian)

Simi , D., Savli , A and Barali , B (1973) Investigation and Classification of Roads from the Aspect of Longitu- dinal Profile Roughness Project, Mechanical EngineeringFaculty in Kragujevac (in Serbian)

Wolfram Research (2005) Mathematica 5.2. Champaign,