Sei sulla pagina 1di 17

2001-01-1235

A Simplified Model for the Spatial Distribution of Temperature in a Motored DI Diesel Engine
D. T. Hountalas and E. G. Pariotis
National Technical University of Athens Mechanical Engineering Department

Copyright 2001,Society of Automotive Engineers, Inc ABSTRACT The purpose of this paper is to present an alternative method to predict the temperature and flow field in a motored internal combustion engine with bowl in piston. For the fluid flow it is used a phenomenological model which is coupled to a computational fluid dynamic method to solve the energy conservation equation and therefore the temperature field. The proposed method has the advantage of simplicity and low computational time. The computational procedure solves the energy conservation equation by a finite volume method, using a simplified air motion model (estimating axial and radial velocities) to calculate the flow field. The finite volume discretization employs the implicit temporal and hybrid central upwind spatial differencing. The grid used contracts and expands following the piston motion, and the number of nodes in the direction of piston motion vary depending on the crank angle. The mean cylinder pressure, the local temperature distribution and the flow field are calculated at each crank angle. Experiments have been conducted in our Laboratory, on a DI diesel engine with bowl in piston at various speeds and the experimental compression curve is compared with the theoretical one. A very good agreement between the predicted and the experimental cylinder pressure is observed. The results obtained provide information concerning the distribution of gas temperature and gas velocity. The current model can be used either to examine the combustion mechanism in homogeneous charge engines or it can be combined with a jet model to develop a sophisticated but always simple model for the air-fuel mixing mechanism. This allow us to examine the combustion and pollutant formation mechanisms on an engine cycle basis, which is extremely difficult when using sophisticated CFD models. Thus the current proposal seems to be a compromise between detailed CFD models and sophisticated multi-zone phenomenological ones, offering the advantage of low computational time and examining in a more fundamental way compared to the phenomenological models the various processes taking place inside the engine cylinder. In the present work are presented the results of the first step which is necessary before expanding the analysis to describe the fired part of the engine as well. INTRODUCTION Computer modeling of the various processes taking place inside the combustion chamber of a diesel engine has advanced considerably and its contribution to the fundamental understanding of these processes is essential. Depending on the objectives, various computer models have been developed that can predict the flow field inside the engine, the spray behavior, the fuel-air mixing process, the heat transfer, the combustion mechanism and the formation of pollutant emissions. The purpose of the modeling defines which of these models should be used, in order to model the various processes effectively. There are two main categories of computer models. The first one is based on empirical and phenomenological relations [1-4], whereas the second one is based on computational fluid dynamics [5-9]. The phenomenological models have the advantage of low computational time and simplicity, but do not offer enough information about the mixing and combustion mechanism. On the other hand using a multidimensional CFD model the local characteristics of the various processes can be predicted reliably, but these models are very complex and require great computational time and even these have limitations in their application. In this study a new simplified model has been developed to predict the spatial gas temperature distribution and the flow field in a motored diesel engine. The present model offers a compromise between the simple phenomenological models and the complex but more detailed CFD ones. To check and validate the various sub *Numbers in brackets designate references at the end of the paper.

models involved it was decided to describe initially the motored operation and in the future the fuel injection and combustion mechanisms will be modeled. The cylinder is divided in finite volumes (cells) where the energy conservation equation is numerically solved to calculate the temperature distribution. For the solution of the energy equation, the flow field has to be known. In this study, contrary to what usually CFD models do, the flow field is computed by a newly developed phenomenological gas motion model. In this way, the required computational time is limited, while the model gives more information compared to existing phenomenological models, but it cannot replace CFD models for fundamental studies. At first an attempt was made to use the axial and radial velocities calculated by a widely used and simple phenomenological gas motion model which is presented in Appendix A. Given that this model gives the values of the radial and axial velocities normal to the surfaces f and fch (Fig. A1) a linear distribution of these velocities had to be assumed to obtain the velocity field. Applying this method, it was observed that the spatial pressure distribution inside the cylinder was greatly non-uniform. High local pressure differences were observed, which are not acceptable, given that the pressure field inside the cylinder is nearly uniform [1]. To overcome this problem and to avoid using a CFD model it was decided to use the criterion of uniform pressure field to calculate the local values of the radial and axial velocities. Following this method a velocity field is predicted at each time step. Moreover the energy equation is solved using this simply calculated velocity field and the spatial temperature distribution is obtained. It has to be mentioned that the proposed gas motion model can be applied only for the closed part of the engine cycle. Furthermore, it should be mentioned that at multi-zone phenomenological models the problem of maintaining the pressure uniform is encountered by a simpler but less reliable way. Mass transfer between the zones is neglected and the volume of each zone is computed thermodynamically to maintain the pressure uniform. It is obvious that this is not the case and mixing between the zones should occur. The problem is to define how this mass exchange should take place. This is the final target of the present work. The application of the model is made on a high-speed small DI diesel engine. A comparison between computed and experimental data for cylinder pressure is made, and the effects of engine variables such as piston cavity diameter and engine speed on the velocity and temperature distribution are examined. The main purpose is to have a first evaluation of the model against conventional engine data i.e. cylinder pressure and then couple it to an existing jet formation and combustion model for diesel engines [2,3] to simulate the combustion, air-fuel mixing and pollutant formation mechanism. In this way we wish to develop an intermediate tool offering the benefits of phenomenological models while examining in more detail the fundamentals of the mixing and combustion mechanism. Thus the purpose of the present work is not to replace CFD models since its use is strictly limited to a certain part of the engine cycle.

BRIEF DESCRIPTION OF THE NEW MODEL As far as the computer model is concerned, a simple flow diagram is given in Fig.1. At each time step the cylinder is divided in finite volumes (computational cells). The gas temperature and velocity is defined at each node of the computational cells. The model consists of two submodels that are related and solved simultaneously through an iterative procedure. The flow model, which is the first, calculates the flow field, based on the assumption that the pressure should be practically uniform inside the cylinder at each time step [1,10] using an empirical method. The spatial distribution of the density is calculated solving the equation of mass conservation at each computational cell. On the other hand to predict the temperature field, the energy conservation equation is numerically solved by the finite volume technique using the velocities and the densities at each computational cell estimated from the flow model. At the beginning of each time step it is assumed that the gas velocity relative to the grid is zero. A first estimation of the temperature field is provided by solving the energy conservation equation and the static pressure at each computational cell is calculated from the perfect gas state equation. This pressure field is obviously non-uniform, given that the mass transfer between the computational cells has been neglected. Taking into account that the pressure field should be nearly uniform, a new velocity field is determined. The energy conservation equation is solved again in this newly developed velocity field and the former procedure iterates until the temperature field has converged and the pressure field is nearly uniform. Using this method for each time step the velocities, temperatures and densities are calculated at each computational cell, while the time required compared to CFD models is relatively low. COMPUTATIONAL DOMAIN The engine considered in this study has a piston cavity centered in the cylinder bore. Thus the domain where the computations are made is symmetric and the computation can be confined to the quarter cylindrical coordinates (r,,z) as shown in Fig.2. The area inside the cylinder is divided into cylindrical computational cells as shown in Fig.3. The number of cells in the r and direction is constant, in contrast to the number of cells outside the piston bowl in the z direction, which is variable depending on piston position. The number of cells inside the piston bowl in direction z is constant. The grid follows the piston motion contracting and expanding [7,11,12,13]. The axial grid velocity inside the piston bowl is equal to the piston velocity, in contrast to the axial velocity of the grid in the area between the top of the piston and the cylinder head, which is given from EQ(1). In the radial direction the cylinder is divided into an annular and inner volume. Thus the number of nodes, referring to the r direction, in the annular volume may differ from the corresponding one in the inner volume.

Read Initial Values Calculate coefficients Ai,i=e,w,n,s,t,b, Su,Sp, etc Advance time t=t+dt Iterate on temperature field using line iteration and finite difference technique. (Energy conservation equation) Pressure distribution calculation using the equation of state. Estimation of the velocity field. Calculation of the mean Pressure and Temperature. Test: Has the Temperature field converged? YES

NO

Grid Generation

Set the relative to the grid velocity of the gas equal to zero

Test: Has required time interval been covered?

NO

Calculation of the density distribution (Mass conservation equation)

YES END

Fig. 1 Flow diagram of the computational procedure.

r direction

direction

E e P
z direction

N n

w W

Fig.2 Computational domain

r direction

r S

E z
Fig. 3. Computational cell and coordinates

w grid

if z > z piston w piston = z w piston if z z piston z piston

(1)

In the present study the grid used has: seven cells in the r-direction inside the piston cavity, seven cells in the rdirection outside the piston cavity, six cells in the z direction inside the piston bowl, and twelve to five cells in the z-direction out of the piston bowl. At present, given that there is no fuel injection, the quarter is not divided in the direction. Reducing the number of cells during piston movement towards TDC, a significant decrease of the required computational time is observed. Moreover this technique provides a more uniform grid since the computational cells in the piston bowl retain their shape in contrast to the ones outside the bowl which contract and expand [13]. TEMPERATURE FIELD ENERGY CONSERVATION EQUATION The energy conservation equation is described in terms of the cylindrical coordinates [11], which expand and contract with piston motion as follows:
1 z piston z piston h t

Heat transfer rate through S= the cylinder boundaries Power due to pressure + variation S = S convection + S pressure

(4)

The heat exchanged between cylinder walls and the gas computational cells by convection is defined by : (T Tcell ) W (5) Sconvection = h A wall m3 Vcell The convection heat transfer coefficient is obtained from the following correlation: k (6) h = c1 Re c2 Pr c3 l char where : Re = w char
Pr = cp
lchar

(7)

) + 1 ( u r h ) + 1 ( v h )
r r r h r

h r ( w h ) 1 r 1 + = + z r r r h z + +S z

(2)

Substituting specific enthalpy with c p T and


with

k into EQ(2) we obtain the following equation: cp


z piston T t

1 z piston

) + 1 ( u r T ) + 1 ( v T )
r r r

(8) k and c1,c2,c3 : constants. In this study c1=0.30, c2=0.80, c3=0.33 In the Reynolds number expression, lchar is the characteristic length and wchar is the characteristic velocity. As far as the energy source rate due to pressure changes is concerned, it is defined as: P S pressure = W/m 3 (9) t It is assumed that the gas is ideal which means that the pressure at each computational cell can be defined by, P = R T (10) The density of the gas at each computational cell is calculated from the gas motion model described later on. Solving EQ(2) using the finite volume technique at every computational cell, the spatial distribution of temperature is obtained at each crank angle. Consequently the local pressures can be computed from the perfect gas state equation EQ(10). The calculation of the spatial distribution of pressure is essential for the gas motion model, as explained later on.

k T k T r r ( w T ) 1 c p r 1 c p + + = r r r z k T c p z S + + z cp

(3)

Velocities u, v, and w are obtained from the gas motion model described later on. The volumetric source rate S is defined by the following expression:

NEWLY PROPOSED SIMPLIFIED GAS MOTION MODEL The definition of the velocity field inside the cylinder is very important given that in the energy conservation equation the convection term depends on the magnitude and direction of the local velocity field. However the solution of the momentum equation needs special treatment given that the convective terms contain non-linear quantities, and that the momentum equation (containing the local pressure) and the continuity equation are intricately coupled. Although there have been proposed many computational fluid dynamic methods for the calculation of the velocity field inside the cylinder of an internal combustion engine [5-9], they have in common that they are complicated and extremely time consuming. On the other hand they appear

to be quite accurate. To overcome the disadvantage of high computational time with some sacrifice to accuracy, at first it was decided to use the flow field that was estimated from a simple phenomenological model [14], like the one presented in details in Appendix A, and solve the energy conservation equation, using the implicit finite volume method. However, great variations of local pressure inside the cylinder at each crank angle were observed, which were not acceptable, given that the pressure is practically uniform at each crank angle. Based on the assumption that the gas pressure is practically uniform inside the engine cylinder, a new simplified gas motion model has been developed. The velocity field is obtained following an iterative procedure. First, at each crank angle the energy equation is solved, and a spatial distribution of the temperature is obtained assuming that the velocity of the gas relative to the grid is zero. Then the pressure distribution is obtained using the perfect gas state equation. Given that the pressure must practically be uniform, an amount of mass dmcell should be transferred to each computational cell through its boundaries from the neighboring cells to eliminate the pressure difference and make it practically equal to the mean pressure of the cylinder. The required mass is defined from, P Vcell dm cell = mean m cell (11) R Tcell The total transferred mass dmcell for all computational cells at each crank angle should be equal to zero, given that the total mass of the gas inside the cylinder is constant. Thus the mean cylinder pressure can be estimated from, m cell R

r direction Start

z direction Finish direction 1

Fig. 4a Sweeping direction 1 for the estimation of the velocity field.


r direction Finish

i, j,k

Pmean =

Vcell T i, j, k cell

(12)

z direction

In this way the amount of mass dmcell transferred to each computational cell is estimated. This amount of mass has to be taken from the neighboring cells. The amount of mass transferred between the neighboring cells depends on the local pressure differences. Sweeping the computational cells, the velocities of the gas at the boundaries of each cell are calculated. Two sweeping directions are considered as shown in Fig. 4a,b. The process for velocity estimation is described for sweeping direction 1, given that the procedure for sweeping direction 2 is similar. The velocity boundary conditions are defined by the following equations: u = 0 on the solid boundaries u = 0 on the symmetry axis (13) w = 0 on the solid boundaries v = 0 at each computational cell Moreover the radial velocity of each computational cell on its North boundary should be equal to the South radial velocity of the neighbor cell located on its North EQ(14).

Start direction 2

Fig. 4b Sweeping direction 2 for the estimation of the velocity field. Similarly the axial velocity of each computational cell on its Bottom boundary should be equal to the Top axial velocity of its neighbor cell located on its Bottom EQ(15). (u n )i, j,k = (u s )i+1, j,k (14)

(w b )i, j,k = (w t )i, j-1,k

(15)

Thus, for each computational cell, there are two unknown velocities that have to be calculated and these are the radial velocity on the south boundary and the axial on the Top boundary. The amount of mass that will be transferred through the south and top boundary of the computational cell is defined from: dmst,cell = dm cell w b b A b dt + (16) + u n n A n dt

The fraction of mass dmst,cell transferred through the south or top boundaries depends on the pressure of the neighbor cells on the south and top direction. Quantities dpmean,s and dpmean,t are computed for the south and the top neighbor cells as follows, dp mean,s = Pcell,s Pmean Eq(17)
dp mean,t = Pcell,t Pmean

The source term S is a function of the dependent variable Tp and is approximated by, SVcell = Su + SP TP
= + 1 t + dt z piston P dt r dz dr d + t t c p t 1 t + dt z piston h A (Twall TP ) dt r dz dr d t t c p Vcell

(22)

Eq(18)

The contribution of the cells located at the top and south direction, to the dmst,cell mass transfer is assumed to be analog to the dpmean,s and dpmean,t, in case they are positive. If the pressure of the top and south located cells is lower than the mean cylinder pressure, then the contribution of these cells to the mass transfer dmst,cell, is assumed to be reversibly analog to dpmean,s and dpmean,t. If the south and the top located cells have opposite signs then it is decided that only one of them will contribute to the required mass transfer depending on the sign of the dmst,cell. In this way the mass transferred from the south and the top boundaries of the computational cell is defined and consequently the velocities relative to the grid at the boundaries of the cell are calculated. This procedure is repeated for all computational cells and the velocity field is defined. At each iteration the total mass transferred to each computational cell from the beginning is computed from: dm total,cell = dm total,cell + dm cell (19) With the new velocity field the mass conservation equation is solved for each computational cell to determine the local densities, m + dm total,cell (20) cell = cell,old Vcell and the energy conservation equation is then solved, to determine a new temperature field. The final solution of the velocity and the temperature field is achieved when the temperature field has converged. It has to be mentioned that the velocity field is obtained following the two sweeping directions of computational cells alternatively through the iterations to limit the effect of the sweeping direction on the flow field calculations even though no serious differences have been observed.
SOLUTION OF THE ENERGY CONSERVATION EQUATION The energy conservation equation is solved using the fully implicit finite volume method. The discretized equation has the following form, PTP = W TW + E TE + N TN + STS (21) + T TT + BTB + 0 P T 0 P + Su

Coefficients i , i=E,W,N,S,T,B used in EQ(21) for the hybrid differencing scheme are given in the following Table : Table 1. Neighbor coefficients for the hybrid differencing sheme. I Di Fi i
E

ke Ae c p ()PE

( v )e Ae

kw Aw cp ()WP kn An c p (r )PN ks As cp (r )SP


kt At c p (z )PT kb Ab c p (z )BP

( v )w A w

D w (Pe w ) + max(F ,0 ) z p,m w D n (Pe n ) + max(- F ,0 ) z p, m n D s (Pe s ) + max(F ,0) z p,m s D t (Pe t ) + max(- F ,0) z p, m t

D e (Pe e ) + max(- F ,0 ) z p, m e

( u )n A n
( u )s As ( w )t A t ( w )b A b

D b (Pe b ) + max(F ,0) z p,m b

In the previous expressions values Pei, i=E,W,N,S,T,B, are calculated from the following formulae: F (23) Pe i = i , where i = e, w, n, s, t, b Di Function A(Pei) where i=e,w,n,s,t,b is defined by the following equation for the hybrid spatial differencing scheme, A(Pei ) = max (0,1 - 0,5 Pei ) (24) It was decided to use the hybrid-differencing scheme due to its ability to exploit the advantages of the upwind and the central differencing scheme. It switches to the upwind differencing when the central differencing produces inaccurate results at high Peclet numbers. The scheme is fully conservative and since the coefficients are always positive it is unconditionally bounded [15]. As far as the temporal differencing is concerned it is decided to apply the fully implicit method due to its unconditionally stable behavior for any time step. However, the accuracy of the scheme is only first-order in time, which is the reason for

where P = W + E + N + S + T + B + 0 p SP with and


0P = P 0Vcell z piston, mean t 1 t + dt z piston, mean = z piston dt . t t

selecting a rather small time step. In this study the time increment is equivalent to 0.5 degree crank angle. Discretising the energy equation results to a system of linear algebraic equations that are solved by the tri-diagonal matrix algorithm (TDMA) which is applied iteratively, in a line by line fashion. In this study the Top-Bottom sweep direction has been applied for the line by line solution of the system.
EXPERIMENTAL FACILITIES AND PROCEDURE To have a first evaluation of the present model an experimental investigation has been conducted on a single cylinder, Lister LV1, direct injection diesel engine located at the authors laboratory. The results of this investigation are used to calibrate and evaluate the model as far as the prediction of the cylinder pressure is concerned. We decided to do this comparison since a fine prediction of the cylinder pressure during compression will reveal if our computational method is in the correct direction so that we can then expand it into the combustion and expansion stroke. The technical data of the engine are given in Table 2. It is a naturally aspirated, air-cooled, four-stroke engine, with a bowl-in-piston combustion chamber and its speed range is 1000-3000 rpm. Table 2. Engine basic design data, Lister LV1 high speed diesel engine Type Single Cylinder, 4-Stroke,DI Bore 85.73mm Stroke 82.55mm Connecting Rod Length 148.59mm Compression Ratio 18 Inlet Valve Opening 15oCA before TDC Inlet Valve Closure 41oCA after BDC Exhaust Valve Opening 41oCA before BDC Exhaust Valve Closure 15oCA after TDC

Table 3. Cases examined 1500 2000 rpm rpm d/D = 44%, X X Bowl Height=20mm d/D = 54%, X X Bowl Height=13mm d/D = 64%, X X Bowl Height=9,1mm

2500 rpm X X X

maintained constant at 18. The compression stroke starts at 41 CA deg, were the air temperature is Tair=47 K and the pressure is Pair=1,1 bar according to available engine data. EVALUATION OF THE PROPOSED MODEL Mean Cylinder Pressure - The proposed model consists of two main sub-models which compute the temperature and velocity fields. The temperature field is used to calculate the mean cylinder temperature at each time step and consequently the mean cylinder pressure can be determined by means of the ideal gas law. To determine the overall validity of the proposed method, the computed cylinder compression pressure diagram is compared with the compression part of the measured one at 20% of full load, and for d/D=54% which is the standard piston cavity/bore ratio, for various engine speeds. In Figs. 5a,b 7a,b are given the computed and measured cylinder pressure diagrams. As mentioned the comparison between the computed and the measured pressure values is made only for the compression stroke i.e. from the inlet valve closure up to fuel injection. Also an enlarged diagram is shown, referring to the crank angle interval from 120 to 170
60
1500 rpm, d/D=54%

Cylinder Pressure (bar)

50 40 30 20 10 0 40 80

Calculated Measured

Experimental results have been obtained at 1500, 2000, 2500 rpm engine speed. The comparison between the experimental and the calculated data focuses on the cylinder pressure diagram using the part from inlet valve closure up to fuel injection. In the near future an effort will be made to present results obtained by a CFD model in order to compare the temperature and velocity fields which are predicted.
RESULTS AND DISCUSSION TEST CASES EXAMINED Computations are made for various piston bowl configurations at three different engine speeds under motoring conditions in an effort to determine the effect of these parameters on the spatial distribution of gas temperature and mainly the flow field. In Table 3, the various combinations examined are given. The cylinder wall and piston crown temperatures are fixed at 400 K for the 2000 and 2500 rpm engine speed and 375 K for 1500 rpm engine speed. The compression ratio

120 160 200 240 280 320

Crank Angle deg.


Fig. 5a Computed and Measured Mean Cylinder Pressure diagram for 1500 rpm engine speed and d/D=54% degrees which offers a closer view of the pressure diagram. Comparing the diagrams, a very good agreement is observed which is quite encouraging considering that a

40

1500 rpm, d/D=54%

40

2000 rpm, d/D=54% Measured

Cylinder Pressure (bar)

30

Cylinder Pressure (bar)

Measured Calculated

30

Calculated

20

20

10

10

0 120 130 140 150 160 170

0 120 130 140 150 160 170

Crank Angle deg.


Fig. 5b Enlarged Computed and Measured Mean Cylinder Pressure diagram for 1500 rpm engine speed and d/D=54%
60
2000 rpm,d/D=54%

Crank Angle deg.


Fig. 6b Enlarged Computed and Measured Mean Cylinder Pressure diagram for 2000 rpm engine speed and d/D=54%
60
2500 rpm,d/D=54% Calculated Measured

Cylinder Pressure (bar)

40 30 20 10 0 40 80

Measured

Cylinder Pressure (bar)

50

Calculated

50 40 30 20 10 0

120 160 200 240 280 320

40

80

120 160 200 240 280 320

Crank Angle deg.


Fig. 6a Computed and Measured Mean Cylinder Pressure diagram for 2000 rpm engine speed and d/D=54% complicated method is used to simulate the compression stroke. Evaluation of the proposed Flow Model - To evaluate qualitatively the flow field obtained from this new method a comparison is made with the one obtained using a well proven simple phenomenological model [14], which is presented in Appendix A. In the simple phenomenological model it is assumed that the gas density is uniform and that the gas axial and radial velocity components in the annular volume Vr and in the piston bowl Vch are zero, respectively. To compare the two gas motion models, we compare gas axial w0 EQ(A10) and radial wr EQ(A18) velocities computed from the simple phenomenological model, with the corresponding mean gas velocities obtained from the newly developed gas motion model, at the corresponding

Crank Angle deg.


Fig. 7a Computed and Measured Mean Cylinder Pressure diagram for 2500 rpm engine speed and d/D=54% surfaces f and fch shown in Fig. A1. The mean gas velocities U, W through surfaces f and fch respectively are computed from the following equations:
u s i, j i, j s U=
r= d 2

f
z = z piston

(25)

W=

w t i, j i, j A t A f ch

(26)

40

Cylinder Pressure (bar)

Measured

(Axial velocity) / (Mean Wp)

2500 rpm, d/D=54%

2
Wo/Wp,m Wm/Wp,m

30

Calculated

20

10

-1

0 120 130 140 150 160 170

-2 40 80 120 160 200 240 280 320

Crank Angle deg.


Fig. 7b Enlarged Computed and Measured Mean Cylinder Pressure diagram for 2500 rpm engine speed and d/D=54% The results are shown in Figs. 8a,b - 10a,b for 1500 rpm engine speed and for various piston bowl geometry conditions. As observed the two gas motion models provide similar gas velocities revealing that the newly proposed method is in the correct direction, which is encouraging.
6 5 4 3 2 1 0 -1 -2 -3 -4 -5 40 80
Wr/Wp,m Um/Wp,m

Crank Angle Degree


Fig. 8b Comparison between the Axial Gas Velocity computed by the simple phenomenological model (Wo) and the one obtained by the proposed gas motion model (Wm) at 1500 rpm and d/D=44%.

(Radial velocity) / (Mean Wp)

4 3 2 1 0 -1 -2 -3 -4 40 80
Wr/Wp,m Um/Wp,m

(Radial velocity) / (Mean Wp)

120 160 200 240 280 320

Crank Angle Degree


Fig. 9a Comparison between the Radial Gas Velocity computed by the simple phenomenological model (Wr) and the one obtained by the proposed gas motion model (Um) at 1500 rpm and d/D=54%. respectively using 20 deg intervals. As observed the gas motion model provides a realistic view of the in-cylinder flow field, compared to the simple phenomenological model described by detail in Appendix A. To make some statements about the flow field, it is decided to divide the cylinder volume into three regions. The first one called (A) is the annular cylindrical region of the cylinder, the second one called (B) is the inner cylindrical region above the piston bowl, and the third one, called (C) is the region inside the piston bowl. It is observed that at 140 CA deg

120 160 200 240 280 320

Crank Angle Degree


Fig. 8a Comparison between the Radial Gas Velocity computed by the simple phenomenological model (Wr) and the one obtained by the proposed gas model motion (Um) at 1500 rpm and d/D=44%. FLOW FIELD In the following paragraphs the computed flow field and the effect of piston bowl geometry and engine speed on it, is examined. Computed Velocity Field - The flow field computed from the gas motion model in the case of 2000 rpm engine speed, and the standard piston bowl dimensions, is shown in Figs 11a-e for 140 to 220 crank angle degrees

1
Wm/Wp,m

(Axial velocity) / (Mean Wp)

(Axial velocity) / (Mean Wp)

Wo/Wp,m

Wo/Wp,m Wm/Wp,m

-1

-2 40 80 120 160 200 240 280 320

-1 40 80 120 160 200 240 280 320

Crank Angle Degree


Fig. 9b Comparison between the Axial Gas Velocity computed by the simple phenomenological model (Wo) and the one obtained by the proposed gas motion model (Wm) at 1500 rpm and d/D=54%.

Crank Angle Degree


Fig. 10b Comparison between the Axial Gas Velocity computed by the simple phenomenological model (Wo) and the one obtained by the proposed gas motion model (Wm) at 1500 rpm and d/D=64%.

(Radial velocity) / (Mean Wp)

3 2 1 0
-0.02 -0.00

-0.01

-0.01 142.87 m/sec

-1
0.00 0.01 0.01 0.02 0.02

0.015 m/sec

0.03

0.03

0.04

-2 -3 40 80

Wr/Wp,m Um/Wp,m

Fig. 11a Velocity field at 140 CA deg, 2000 rpm engine speed and d/D=54%

120 160 200 240 280 320

Crank Angle Degree


Fig. 10a Comparison between the Radial Gas Velocity computed by the simple phenomenological model (Wr) and the one obtained by the proposed gas motion model (Um) at 1500 rpm and d/D=64%. (Fig. 11a) vortex flows are developed inside regions A, B, and C. The gas has a rotating motion and is driven into the piston bowl. Moreover gas is going out of the piston bowl close to the cylinder symmetry axis. As compression continues on, we observe that at 160 CA deg (Fig. 11b) the vortex flow is been restricted in region C, and its direction has changed compared to the corresponding one at 140 CA deg. Moreover the magnitude of the velocity vectors is increased, especially the ones at the piston crown in region A, and at the cylinder symmetry axis inside the piston bowl.

-0.00 142.87 m/sec -0.01 0.015 m/sec -0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

Fig. 11b Velocity field at 160 CA deg, 2000 rpm engine speed and d/D=54%

-0.00

142.87 m/sec

-0.01

0.015 m/sec

0.00

0.01

0.01

0.02

0.02

0.03

0.03

0.04

Fig. 11c Velocity field at 180 CA deg, 2000 rpm engine speed and d/D=54%

-0.00

-0.00

-0.01

-0.01

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

-0.01

Fig. 11d Velocity field at 200 CA deg, 2000 rpm engine speed and d/D=54%

-0.02 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

-0.00

Fig. 14 Velocity field at 160 CA deg, 2000 rpm engine speed and d/D=44%

-0.01
-0.00 142.87 m/sec

-0.01
-0.01 0.015 m/sec 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

-0.02

0.00

0.01

0.01

0.02

0.02

0.03

0.03

0.04

Fig. 11e Velocity field at 220 deg, 2000 rpm engine speed and d/D=54%

Fig. 15 Velocity field at 160 deg, 2000 rpm engine speed and d/D=64% The majority of gas mass flows from the piston bowl to region B through the surface near the cylinder axis, and the surface of the piston bowl near region A. Moreover it is observed that in regions A and B, the fluid particles located near the plane defined by the piston crown have higher velocities. Finally as expansion continues on at 220 CA deg (Fig. 11e) the velocity field is analogous to the velocity field obtained at 140 CA deg., but as observed the magnitude of the gas velocity has been slightly decreased compared to 140 CA deg (symmetrical piston position). Moreover it should be noticed that the magnitude of fluid velocities at the boundaries between regions A-B, and B-C are higher compared to the ones computed from the simple phenomenological model. This can be attributed to the fact that the gas flow from region A to B and from B to C occurs mainly through a fraction of surfaces f and fch respectively as shown in Fig. A1 of the Appendix A. Effect of engine speed on the estimated velocity field In Figs. 12, 11b and 13 are given the computed flow fields for 1500, 2000 and 2500 rpm engine speed respectively, for the standard piston bowl geometry d/D=54%. As observed the maximum speed is augmentative as the engine speed is increased. Moreover the flow fields at 2000 and 2500 rpm engine speed are quite similar. Effect of bowl geometry on the estimated velocity field To investigate the effect of piston bowl geometry on the predicted flow field we compare Figs. 14, 11b and 15 corresponding to d/D=44%, 54%, 64% respectively at 160 CA deg and 2000 rpm engine speed. The flow field for d/D=54% is similar to that at d/D=64%, in contrast to the one computed at d/D=44% where the vortex flows that are generated have opposite directions. Moreover at d/D=44%,

-0.00 142.87 m/sec -0.01 0.015 m/sec -0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

Fig. 12 Velocity field at 160 CA deg, 1500 rpm engine speed and d/D=54%

-0.00

-0.01

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

Fig. 13 Velocity field at 160 deg, 2500 rpm engine speed and d/D=54% Relatively increased velocities are observed in region B while the gas is still driven into the piston bowl. At TDC (Fig. 11c) a significant decrease at the magnitude of the velocity vectors is observed which is in accordance to what the simple phenomenological model predicts. During expansion at 200 CA deg (Fig. 11d) the flow has changed direction and the gas moves from the piston bowl to the regions B and A. The flow field is analogous to the one obtained at 160 CA deg but towards the opposite direction.

in region B and close to the piston cavity radius, a vortex has been generated which does not exist at d/D=54% and d/D=64%. Comparing the axial and radial gas velocities computed from the phenomenological model presented in Appendix A, at 2000 rpm engine speed for the various piston bowl geometries respectively (Figs. 16a,b), it is observed that the decrease of the ratio d/D, causes an increase of both
5

effect of bowl geometry on the velocity field is estimated at least qualitatively in a correct manner by this newly developed model, despite the preliminary nature of the present work. ESTIMATION OF THE TEMPERATURE FIELD The estimated temperature field and how it is affected by the engine speed and piston bowl geometry, is investigated in the following paragraphs. It should be mentioned that the absolute temperature differences are relatively small. Spatial Temperature Distribution - As far as the spatial temperature distribution is concerned, in Figs. 17a-f are given the predicted temperature fields at 2000 rpm engine speed for the standard piston bowl geometry (d/D=54%), at various crank angles. At 140 CA deg. (Fig. 17a) the temperature inside the cylinder has raised above the cylinder wall temperature, and heat is transferred from the gas to the cylinder walls. Due to this the gas temperature near the cylinder walls is lower than the one in the center of the cylinder. As compression continues at 160 CA deg. (Fig. 17b) the cylinder temperature rises further, and the warmest fraction of the gas is restricted towards the center of the piston bowl. Around TDC (180 CA deg -Fig. 17c) the gas temperature has reached its maximum value, and the maximum temperature difference is observed inside the cylinder. The heat transfer through the boundaries of region A is higher than the one through regions B and C and for this reason, the lowest gas temperatures are observed in region A. Inside the piston bowl and near the cylinder symmetry axis, which is assumed to be an adiabatic boundary, the gas temperature is maximized. At 200 CA deg. (Fig. 17d) the gas temperature has decreased and region A continues to have the lowest temperatures compared to the regions B and C. The hottest fraction of the gas which exists in the piston bowl has moved, drifted by the reversed gas motion, from the cylinder bowl to regions B and C. At 220 CA deg. (Fig. 17e) the vortex flow which has been developed in region B affects the temperature distribution inside the hot region. Finally at 240 CA deg (Fig. 17f) the gas temperature is further reduced and in some regions is lower than the cylinders wall temperature, thus a reversed heat flux from the cylinder walls to the gas occurrs.

(Radial Velocity)/(Mean Wp)

4 3 2 1 0 -1 -2 -3 -4 -5 40 80
2000 rpm d/D=44% d/D=54% d/D=64%

120 160 200 240 280 320

Crank Angle Degree


Fig. 16a Mean Radial Velocity computed by the Phenomenological model at 2000 rpm engine speed and d/D=44%, 54%, 64%
2
2000 rpm d/D=44%

(Axial Velocity)/(Mean Wp)

d/D=54% d/D=64%

-1

-2 40 80 120 160 200 240 280 320

Crank Angle Degree


Fig. 16b Mean Axial Velocity computed by the Phenomenological model at 2000 rpm engine speed and d/D=44%, 54%, 64% radial and axial gas velocities. Examining Figs 14, 11b, and 15 the former observation can be verified by the computed flow field with the proposed gas motion model. In case of d/D=44% the flow from region A to B is more intensive compared to what occurs at d/D=54%, 64%. Moreover the increase of the axial gas velocity as d/D decreases can be attributed to the fact that the area fch through which the gas mass has to flow is decreased. Thus it appears that the
-0.00

584 579
-0.01

574 569 564

-0.01

-0.02

559
0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

554

Fig. 17a Temperature field at 140 CA deg, 2000 rpm engine speed and d/D=54%

747 739 731


-0.00

-0.01

415
-0.01

410 405 400 395 390 385 380 375 370 365 360 355 350 345 340

723 715

-0.02

-0.01

707 699
-0.02

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

691 683 675


-0.03

Fig. 17b Temperature field at 160 CA deg, 2000 rpm engine speed and d/D=54%
850 835 820 805 790
-0.00

-0.03

-0.04 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

Fig. 17f Temperature field at 240 CA deg, 2000 rpm engine speed and d/D=54%

775 760 745 730 715


0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

-0.01

700 685 670 655 640

Fig. 17c Temperature field at 180 CA deg, 2000 rpm engine speed and d/D=54%
690 670 650 630
-0.01

-0.00

610 590

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

570 550 530

Generally, a gradually developed temperature field from the cylinder walls to the inner volume, is observed. Moreover it can be noticed that in the region near the cylinder symmetry axis, the temperature field is not uniform, although the boundary condition is that the symmetry axis is adiabatic. This is due to the fact that convection has a significant contribution to the heat transfer inside the cylinder, and examining the velocity field, the spatial distribution of the temperature can be explained. Effect of piston bowl shape on the Temperature Distribution - Comparing the spatial distribution of temperature at 2000 rpm engine speed and 160 CA deg for different piston bowl geometry as shown in Figs. 18, 17b, and 19, we observe that the temperature spatial distribution is similar although the various temperature contour lines encompass different areas.

Fig. 17d Temperature field at 200 CA deg, 2000 rpm engine speed and d/D=54%

510
-0.00

742 734
-0.00

500 490

726 718

-0.01

480 470
-0.01

710 702

-0.01

460 450
-0.01

-0.02

440 430 420


0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

694
-0.02 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

686 678 670

410

Fig. 17e Temperature field at 220 CA deg, 2000 rpm engine speed and d/D=54%

Fig. 18 Temperature field at 160 CA deg, 2000 rpm engine speed and d/D=44%

750 740 730 720


-0.00 -0.00

745 739 733 727 721 715 709 703


-0.01

710 700

697 691 685

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

690 680 670 660 650


-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

679 673 667 661 655

Fig. 19 Temperature field at 160 CA deg, 2000 rpm engine speed and d/D=64% Effect of engine speed on the Temperature Distribution Examining the temperature distributions for engine speeds 1500, 2000 and 2500 rpm, for d/D=54% and 160 CA deg. shown in Figs 20, 17b, 21, no serious effect of engine speed on temperature distribution can be observed. The temperature distribution and the absolute values of temperature are similar. It should be noticed that the slightly lower values of temperatures for the case of 1500 rpm engine speed is attributed to the lower Temperature of the cylinder walls. From the previous results it is revealed that the new computational method developed in the present work is in the correct direction. The results obtained especially for temperature distribution appear logical taking into account both experimental data and also data obtained from phenomenological modeling [16-18]. It is also encouraging that the prediction of the cylinder pressure is quite accurate for motoring conditions and for all operating speeds examined without making any adjustment. Finally using the new approach we are in position without using sophisticated

Fig. 21 Temperature field at 160 CA deg, 2500 rpm engine speed and d/D=54% We will be then in position to describe in a more fundamental way the fuel air mixing mechanism. Even though the accuracy of the proposed model will probably be less, compared to CFD ones, we will be in position to consider at least qualitatively for the effect of combustion chamber geometry on the combustion mechanism Existing phenomenological models do not appear capable to consider for this effect without the use of adjustable parameters.
CONCLUSIONS In the present work a simplified model for the prediction of the flow field and the spatial distribution of gas temperature has been developed for motored DI Diesel Engines. From the application of the model on a DI diesel engine located at the authors laboratory, it is revealed that the developed model predicts adequately well the engine cylinder pressure diagram. Special emphasis is given on the modeling of the flow field inside the cylinder using a newly developed simple model very efficient in time. The results obtained from this newly developed gas motion model are in accordance to the overall ones obtained from a well tested phenomenological model. The proposed gas motion model seems to offer the ability to investigate in a more fundamental way the effect of engine design parameters on the in-cylinder flow field. Combining the sophisticated phenomenological gas motion model which predicts the flow field with a simple fluid dynamic one, we manage to offer information concerning the temperature distribution and the heat transfer mechanism. On the other hand it is quite encouraging that the proposed model manages to predict the effect of engine speed and piston bowl geometry on both the flow and temperature field. This conclusion is based on known mean data values for motored engines either experimental or computational derived from tested phenomenological models. In the near future comparison will be made to the results obtained by a CFD model in order to have a more detailed evaluation of the proposed model Another advantage of the proposed method is the relatively low computational time required for a complete run from compression start to expansion end which is for a time increment equivalent to 0.5 CA deg., about 14 min. on a Pentium III 450 MHz PC. However it should be

741 733 725 717


-0.00

709 701

-0.01

693 685 677

-0.01 0.00 0.01 0.01 0.02 0.02 0.03 0.03 0.04

669 661 653 645

Fig. 20 Temperature field at 160 CA deg, 1500 rpm engine speed and d/D=54% CFD codes to examine the temperature distribution inside the combustion chamber of DI diesel engines and have a realistic prediction of the in-cylinder gas flow field and its variation with engine speed and piston bowl geometry. This information is of great importance when coupled to existing phenomenological models.

mentioned that the proposed gas flow model can be applied only for the closed part of the cycle. Currently following the preliminary results obtained from this new type of modeling an effort is conducted to combine it with a jet model to simulate the air-fuel mixing and combustion mechanism inside the combustion chamber of modern DI diesel engines. The challenge is promising and it seems that the present model may offer great assistance during the task of providing a simple but detailed model capable to consider the effect of engine speed and bowl geometry on the processes occurring inside the combustion chamber of DI diesel engines.
APPENDIX A SIMPLE PHENOMENOLOGICAL DETERMINATION OF THE AXIAL wo AND RADIAL wr VELOCITY OF THE GAS - As the piston moves towards the TDC, it squishes the air into the chamber bowl [14]. Relations for the axial and radial velocities of the developed flow are derived here under the fundamental assumption that the air density in the entire space above the piston crown is uniform. Moreover, the air is assumed to be an ideal gas.
D

V Vx

f Vr

Wr

Wo Vch r-z plane view

fch

Fr

Fig. A1 Graphical representation of the quantities necessary for the determination of axial and radial velocities. The instantaneous volume of the space above the piston crown shown in Fig. A1 is: V = Vch + Vx + V (A1) The compression ratio is defined as : V + Vs + V (A2) = ch Vch + V The instantaneous piston stroke is : s x = R cr () (A3) () is where Rcr is the crank radius, and the function expressed as follows :
() = 1 cos + L R cr R cr 1 1 L
2 sin 2

(A4)

The piston speed is


ds x = R cr () dt

(A5)

where is the angular crankshaft speed and () , R () = sin + cr sin 2 (A6) 2 L Making use of the above equations and after an appropriate mathematical rearrangement it follows that () 1 dV (A7) = 2 V d + () -1 When both valves are closed it is assumed that the mass of air inside the cylinder remains constant: m=const. The density of the air above the piston crown can be calculated as a function of crank angle from the following equation: m 1 d () = (A8) = 2 V dt + () -1 Axial Velocity - The mass of air contained in the chamber is, m ch = Vch (A9) As piston moves towards TDC the air is squished into the piston bowl, changing the mass mch contained in it. The air mass flow rate through area fch is given from the following expression : dm ch d ( A -8) f ch w o = = Vch dt dt (A10) V () w o = ch 2 f ch + () -1 Radial Velocity- The mass of air contained inside volume Vr is : m r = Vr (A11) As piston moves, the air is squished and flows through area f. The air radial velocity wr is derived from the mass flow rate through area f as follows: d r dm r dV = r + Vr = f wr dt dt dt (A12) 1 d 1 dVr wr = + Vr f dt f dt Replacing the derivative with respect to time by that with respect to the following expression for the radial velocity is derived: dV V dV (A13) wr = r r f d V d where : Vr = Fr (s x + ) (A14) and Fr is the area of the piston upper surface (except the combustion chamber area) as shown in Fig. A1. From Fig. A1 it follows that: Vx + V = F(s x + ) (A15) V + V (s x + ) = x F where F is the cross-section of the cylinder. Substituting EQ(A15) into EQ(14) it follows that :

Sx

Vr =

Fr (Vx + V) F

(A16)

wpiston wr z zpiston (r) (z) ()

and consequently dVr Fr dVx Fr dV = = (A17) d F d F d Substituting EQ(A17) and EQ(A16) into EQ(A13) ,and taking into account EQ(A7) we finally end up to: V F () (A18) w r = ch r f F 2 + () -1
NOMENCLATURE A :Area of a computational cell which heat is in contact to the solid wall A(P) :Function of the Spatial discretisation scheme :Specific heat of the gas cp d :Piston bowl diameter D :Cylinder bore :Mass which need to be tranfered to a dmcell computational cell in order to have a pressure equal to the mean pressure :Mass which is transfered to each dmst,cell computational cell through the South and the Top cell walls h :Specific enthalpy of the gas k :Conduction heat transfer coefficient :Characteristic length lchar L :Connecting rod length :Mass of the gas which is contained in a mcell computational cell P :Pressure :Mean pressure of the gas in the cylinder Pmean r :Radial direction R :Gas constant :Crank radius Rcr Rm :Universal gas constant S :Volumetric source rate Sconvection :Volumetric source rate due to heat transfer through the cylinder walls :Volumetric source rate due to the change of Spressure the pressure with respect to time t :Time T :Gas Temperature :Gas Temperature of a computational cell Tcell :Temperature of the cylinder boundaries Twall u :Radial component of gas velocity v :Circumferential component of gas velocity :Volume of clearance between piston and V head :Chamber volume Vch :Volume of a computational cell Vcell :Stroke volume Vs w :Axial component of gas velocity :Characteristic velocity wchar :Axial velocity of grid lines wgrid :Axial velocity of the gas computed by the wo phenomenological gas motion model

:Axial velocity of the piston :Radial velocity of the gas computed by the phenomenological gas motion model :Axial direction :Distance between the gas face of the cylinder head and the piston top :Distance between the nodes of two neighbor cells in radial direction :Distance between the nodes of two neighbor cells in axial direction :Distance between the nodes of two neighbor cells in circumferential direction

Greek Symbols :Diffusion coefficient :Circumferential direction :Dynamic viscocity :Gas density :Crank angle degree :Angular crankshaft speed Subscripts P e w t b n s

:Central point of the computational cell :East side face of a control volume :West side face of a control volume :Top side face of a control volume :Bottom side face of a control volume :North side face of a control volume :South side face of a control volume

Dimensionless Groups Re :Reynolds number Pe :Peclet number Pr :Prandtl number REFERENCES 1. Fitzgeorge, D. and Allison, J.L., Air Swirl in a RoadVehicle Diesel Engine, Proc. Instn. Mech. Engrs. (A.D.), No 4,pp. 151, 1962-63 2. Kouremenos, D.A., Rakopoulos, C.D., and Hountalas, D.T., Multi-Zone Combustion Modeling for the Prediction of Pollutants Emissions and Performance of DI Diesel Engines, SAE Transactions, paper No 970635, 1997. 3. Rakopoulos, C.D., and Hountalas, D.T., Development and validation of a 3-D multi-zone combustion model for the prediction of DI diesel engines performance and pollutants emissions, SAE Transactions, paper No 981021, 1998. 4. Payri, F., Benajes, J., and Tinaut, F.V., A Phenomenological Combustion Model for Direct Injection, Compression Ignition Engines, Applied Math. Modeling, Vol. 12, pp.293-304, 1988. 5. Ahmadi-Befrui, B., Gosman, A.D., Issa, R.I., Watkins, A.P., EPISO Implicit non-iterative solution procedure for the calculation of flows in reciprocating engine chambers, Computer Methods in Applied

6. 7. 8.

9. 10. 11.

12.

13. 14. 15. 16.

17.

18.

Mechanics and Engineering, Vol. 79, pp.249-279, 1990. Diwakar, R., et al., Inviscid Solutions of the Flowfield in an Internal Combustion Engine, AIAA Journal, Vol. 14, No. 12, pp. 1667-1668, 1976. Gosman, A.D., et al., Axisymmetric Flow in a Motored Reciprocating Engine, Proc. Inst. Mech. Engrs, Vol. 192, No. 11, pp. 213-223, 1978. Ramos, J.I., Humphrey. A.C., and Sirignano, W.A., Numerical Prediction of Axisymmetric Laminar and Turbulent Flows in Motored, Reciprocating Internal Combustion Engines, SAE Transactions, paper No 790356, 1979. Griffin, M.D., et al., Computational Fluid Dynamics Applied to Flows in an Internal Combustion Engine, AIAA, paper 78-57. Yang, J., Pierce, P., Martin, J.K., and Foster, D.E, Heat Transfer Predictions and Experiments in a Motored Engine, SAE paper No 881314, 1988. Nishida, K., and Hiroyasu, H., Simplified ThreeDimensional Modeling of Mixture Formation and Combustion in a D.I. Diesel Engine, SAE Transactions, paper No 890269, 1989. Theodorakakos, A., Bergeles, G., "Predictions of the In-Cylinder Fluid Flow of a Motored Internal Combustion Engine, Journal Entropie, No174/175, pp.7-14, 1993. Theodorakakos, A., Bergeles, G., Numerical investigation of the flow inside a 4-X IC model diesel engine, Journal Entropie, No 200, pp 53-63, 1996. Kowalewicz, A., Combustion Systems of High-Speed Piston I.C. Engines, Elsevier, Vol. 3, 1984. Patankar, S. V., Numerical Heat Transfer and Fluid Flow, Hemisphere Publishing Corporation, Taylor & Francis Group, New York Bahram, K., Multidimensional In-Cylinder Flow Calculations and Flow Visualization in a Motored Engine, Journal of Fluids Engineering, Vol 117, 1995. Payri, F., Desantes, M., Pastor, J.V., LDV measurements of the flow inside the combustion chamber of a 4-valve D.I. diesel engine with axisymmetric piston bowls, Experiments in Fluids, Vol. 22, pp. 118-128, 1996. Khalighi, B., Multi-dimensional in-cylinder flow calculation visualizations in a motored engine, J. Fluids Eng., Vol. 117, pp. 282-288, 1995.

Potrebbero piacerti anche