u. J.

mecti. Sei. I'crg.uuon

PrC$:-'. U)j5.

Yo1. 17, pp. fli-1~L

----

Print-cd in Ureat Brltain

o
\

J,

A SIMULATION MODEL INCLUDING INTAKE AND EXHAUST

SYSTEMS FOH A SINGLE CYLINDEH FOUH-STROKE

SPARK IGNITION ENGINE
R.~.

HEC;SO);,*

W. J. D. Ax x.v x n ] and P. ('.

R<\lmAH;"

*J\l('l"hanical

El1l-'iIH'f'rinl-' DCjll1rt 1I1<'lIt. U 11ivr-rxi t v of Manchest Cl' Institute

Tc('IIlIo!ogy,

:-lackYillL- :)(n'l't"

]'I1111)(;IH's('I'; .uu l t I'cllut.ion

lTl1iY(~1'~it.\.

(Ren'il'ed

~lillHll('.'-d('!',

1;, Xl u rch UJ7-!.

nlld

i/l

raised

CYCLE

of Science alld
Unit" Manche-ster

H('s('nreh

I\~ngland
form

1\) A aqust

1974)

Summary-A
comprehensive
snuu lu.tion mode-l is presented
for 11 spark igniLion engine
including intake and exhaust S)'stClllS. The power cycle simulation
requires only one
empirical factor to correct for t.urhulont speed of the fiamr- front in order to complete
thc ovclo culcu lat.iou ineludillg .:\(). The ox huust, pipe gas dvmuuics include chemical
reactions along path lines. Calculations
are presented which compare well with experimcnt.al results.
The model predict ions compare Ia vourably with previous work.
~()TATION

a
Ilq

u,
aref

A
Aa
C,)
C,.
J)

f
F
fI

f.:
J\.

speed

of sound

Anrurnd const ant 1'0], ('otl,y,'l'lin' hca.t transfer

s]lppd of sound aft,'r iscn trop: change of statc to n-ference
roforoncc ~pced of sound
nou-dimensionul
speed of souru! (u /1{ rerl
non-dimonsionu.l

(1.,("-,111,1'1)

specific lu-a.t at, const unt pl'<'~sun'

specific heat at constant vulurmdiumoter
friction factor T,,) ~p/l~
area
(:ihh~ ff'l'(' l'llI'l'g.\spcL'ilie hOILt.rat ill (C '" (',.)

SlH~('ifi('

rl111('t

jelll

thermal conductivitv
Iif fillid
1\,. rate constant for ('(pml iou (;)
X.v
equifibrium l")Il~Lmt
'In
muss
n rot.at iuna.l spoed
jJ

}J,ef

Pl'(\SSUl"U

rr-Iorenco prossurr-

'Iralt-

(l
T

u
Bmo!

.n"
l{,.
t
/'i

T
u

"I' 1,,':J.i t 1':1.11,;1'",1)('1' unit i iuu- 1"'1' un i t llta"~

total h(%t, Ilu x

I"'('SSIII'<' rut io }'i}',..'f "I' r.uliu
gas ('onst;tnt,
universal gas c-onstunt
H(')-nolds numbe r
one-way equilibrium
rn.t o for reaction (I)
t.irno
tcmporat.uro for ith ZOJll'
gas temperature
particle velocity or specific internal energy
non-dimensional
particle velocity (v/a,,,,.) or internal

97

or

fillid

energy

pressure

R. S.

98
U,
'U(i,ij

V
IV
;1:

X
Xi
-"fer
h
Zl,'oeff

\V. J. D.

BENSON,

ANNAND

and P. C.

BARUAH

internal energy coefficient for ith term

internal e-nergy coefficient for ith term and jth species
volume or velocity
work
distance
non-dimensional length (xjXrer)
molar fraction of 'it.h species
l'"f,'r(lIH'"

1(1l\g(,h

non-d imonaioun.l
Lime [afcl(tf,l'l'I,d]
vis.-osit.y cootficiont,
(Tank n.ngln
lt.icuiaun
vurin.hlo _! + 1(I. - 1)/:211 '
Rii-uiann
v.uiu.hlo A - [(I. ,- I )/:2J I'

stoir-hiomot.ric coefficient

or

ill('I'('llH'nt

Ct

u u-rr-nu-n t of' Y,

p
TI/1

density
wal! slu-m- si I'('ss
equivuh-ncc
viscosit y

rn,t io

Suffi xes

st,a.gnati<lJl

burned l'ro<iu("(.01' backward

"
e

f
1Il
'111.
11
1/1

out
p
/("

condit

ion

cy l incle r

equilibrium or exhaust.
forward
inlet
incoming
unburnf
mixt ure
pt(\-jo\l:-:';
('lll'l't'll1.

t imc- st('P
t i n u: :--;1('1)

outgoing
product
wltll
I ~Tl{OD

lTTIOX

SHHTLATlON
models eau hn of grea.t. f\.ssistn.nc(' to the engine designers if they
give (1, good rcproscntat.ion of t.ho engine svstr-m. Their economic value is in
the reduction in time and costs for the development of new engines and their
technical value is in t.h idout.ifleat.ion of area,,; which require specific attention
as the design study evolves. :-;i mulut.ion models for complete engine systems
including intake and exhaust systems have been successfully developed for
diesel engines.l- 2 On the other hand, models for spark igllition engines have
been confined to the powcr cycle, in particular with the identification of the
parameters which inf uence the ex ha.ust em issions. 3 It is well known that
the quantity of residual gases in the cylinder at the commencement of the
compression stroke influences the nitric oxide emissions, and by artificially
increasing the res id unls through CX]W.lISj, gas rcoirculation the nitric oxide
emissions can be reduoed.! Qualitative, and to a certain extent quantitative,
work on hydrocarbon emissions indicates that the unburnecl hydrocarbons
leave t.ho cylinder ill t.wo peaks, ono at. t.ho beginning of the exhaust stroke
during blowdown and t.ho other at the end of the ex haust stroke." 5 About

A model for spark ignition engine with intake and exhaust systems

99

half the total mass emissions comes from each peak. At present, the model
simulations
for both nitric oxide and hydrocarbon
emissions neglect the
interaction
of the gas dynamics
in the intake and exhaust system with the
cylinder conditions.
Since it would a,ppear that control of exhaust emissions
can he at.tempted
through the desigll of the intake and exhaust system, it is
clear that a suitable model to simulate t.}wl'e svstoms is desirable.
In this paper
a model is presented for a single cylinder four-stroke
spark ignition engine. In
a second paper the model is applied to a crankcase
scavenged
two-stroke
engine." The power cycle will be first described followed by the gas exchange
process.
Nitric oxide emissions only will he examined.
The model will be
compared with experimental
results.
PO'YER

I
1

r
:1
S

e
.e

,t
y
le
e,
1S
;:0

.It

CYCLE

MODEL

A number of power cycle models have been suggested. That of Lavoie et al.' is
generally recognized as giving the closest approximation to the combustion process in lot
spark ignition engine. However, t ho model depends either on a known pressure-time
diagram or a semi-empirical burning law which would also include the ignition delay.
Lavoie et al:" considered a temperature gradient in the cylinder related to the burning rate.
In this paper we propose to use the chemical react ion equations suggested by Lavoie but
the cylinder volume is restricted to two zones, a burnt and an unburnt zone. The two
zones are separated by the flame front. Furthermore, we propose to use a simple method
for predicting the ignition time for the commencement. of combustion. With these two
modifications to Lavoie's method we avoid the empirical burning law and simplify the
computational procedure. However, the new method introduces a new variable, namely,
the flame propagation speed. It is to be hoped that introduction of this variable may
enable a more rigorous prediction of the combustion process than a burning law.
Since the rates of the energy producing reactions in the burned gas zone are so fast
that the burned gases are close to thermodynamic equilibrium, we assume that the pressure
u.ru] j,<'Il'l"'l'lLjlll'!\
ill j,bo IHII'Il(ld gllH ('1L1lh" ('(1I1I1'1I1,('dfrom oqu ilibrium
t.hormodynumies.
The nitric oxide couccntrat.ions, only, arc oornput.ed from the rate equations. Other species
concentrations are calculated by numerical solution of the equilibrium equations based
on a modification of Vickland's method." The volume of the burnt gas zone during the
combustion process is determined by methods reported by Annand.s- Contrary to Lavoie
et al,' the analysis includes heat transfer during the combustion process.
Since equilibrium calculations are iterative, it is necessary to make close estimates of
the product temperature if rapid convergence is to be achieved. Some of the techniques
will be described. These are based on numerous calculations covering normal engine
conditions.
Before discussing the cycle calculation, we will briefly review the model for the
mechanism of flame propagation and ignition, 'Ne will then review the equilibrium
thermodynamics and the rate kinetics.
MECHANISM

OF FLAME

PROPAGATION

Despite the excellent qualitative work, particularly from combustion photography,

there still remains uncertaint.y about the fundamental details of the structure and
behaviour of the combustion zone in a spark ignition engine, One area in which quantitative
data cannot yet be reliably predicted is the mechanism of the flame propagation.
The theory of laminar flame propagation through a homogeneous air fuel mixture is
well established. Tho rate of propagation depends partly on the physical properties
(specific heat, thermal conductivity, mass diffusivity) of the substance making up the
unburned mixture and products and partly or the chemical nature of the reaction and the
relation between tho controlling react.ion ra.tos and temperature.
Spalding et al,lo have
shown that the propagation speed can be calculated if the reaction can be fully described .
However, under turbulent conditions the flame front is modified and the propagation

H. S.

100

BEKSON,

w.

J. D.

ANNAND

and P. C.

BARuAR

speed increases with increasing turbulence. Until recently it was held that the increase of
flame speed was duo to the wrinlding of the flame front (which increases its a1'ea)l1 in
combination with enhancement of heat a.nd muss transfer rates, and the turbulent propagation speed 1tt would be expressed as a simple function of the larninar propagation
speed 11./ and the r.m.s. turbulent velocity It'. This has been found to be true only for Iow
intensity of turbulence, '\'hen u' becomes comparable with u/ it appears that the flame
front is broken up and replaced by a much th ickor zone containing small volumes of
unburned , burning and burned. material stirred together by turbulence. Sokolik et aU2
and Chomiak+' have presented some experimental evidence for this. Howe and Shipman-"
have proposed an elaborate statistical model charact.criz.ing the burning zone as spherical
parcels of unburned gas dispersed through burned gas with burning at periphery of the
parcels, At the moment, there is no precise relation between turbulent and laminar
flame speeds for t.urbulence of high intensity.
In their computer simulation work, Phillips and Orman" reviewed a number of
theories for laminar flame propagation and t.hoir final choice was for the thermal theory of
Mallard and Le Chatelior.!" Bailey" at.t emptod to use the Mallard and Le Chatelier
expression for his work on combust ion limit at.ions ut' gaseous fuels in reciprocating engines.
However, some difficulty was encountered relating to the adjustment of temperatures.
This was attributed to the exclusion of a pressure term in the expression. Kuehl-" used
measurements made on a fif,t burnt'!' to develop an empirical expression for propane-air
mixture which correlated experimental flame speed data over a wide range of pressures
and temperatures,
This was successfully used by Bailey Y Kuehl's expression for Iaminar
flame speed u/ in a propane air mixture is
_

u/ whoro

'I;"

'/',,1\1'0

[
((lO'ITb)

1087 x 10r.
+ (900/7',,))H38

-0,09876

(1 )

th" (.<'lll])(lI'ltt
lll't's of' un lmrncd IIwl j)lu'nl'd urixt.urox,

!{';

11/

iR t.ho lamina!'

flame speed, em/sec; (met T' is the pressllro in. Hg.

In the present model Kuehls expression for Iaminar

flame speed is used as the basis

of the calculut.ion. To allow for turbulence a multiplying factor is used. This factor will
be somo funct.ion of tho t.urbu lcnco k-vol ill l.ho yliudorund
the r-ngino speed, The turbu
lent flume speed is then given by
(2)

whcW.ff is a flume factor dotcrminod by OIW of two met.hods. The best approximation of
the value of .If can be obtained by carrying out a number of calculat.ions with different
values of.JJ and comparing the computed pressure-time diagrams, and the experimental
pressure-time diagram. When the peak pressures in both the diagrams are matched the
value of'/f it>obtained. The other approximate method is followed in the case when no
experimental diagram is at hand. In this If is adjusted unt.il the computed pressure
diagram gives a symmetrical burning time about T.D.C, Having obtained the flame
speed it is assumed that the flame propagates spherically from the spark plug.
IGNITION

AND INITIATION
COMBUSTION

OF TWO
SPACE

ZONES

IN

In t.ho model wo shall use, ignition of the fuel is said to take place when a finite volume
of the burnt mixture exceeds 0001 times the total cylinder volume T~(swept volume plus
clearance volume). At the nominal ignition time the mixture (at state Pm' tm) is assumed
to burn adiabat.ice.lly at constant volume to the product temperature, tp. For this purpose
an intornal cnergy bnlanco is made for unit rnass of the mixture. The first estimate.
of tho product tcmpcrut.uro tp is bused on the following expressious :
t. = t",+2500xrf;x!;

1'or1",1'0

und
t" = t",
where

i. is the

+ 21)00

x rf; x f,

70()(cp - IO)!i

for cl> > 1 '0,

fraction fresh mixt.uro and cl> is the eqnivalencc ratio.

A model for spark ignit.ion engine with intake and exhaust systems
f
n

,.
n

101

An energy balance is made and tp adjusted until the specific internal energy of the
products = the specific internal enArgy of the reactants, and tp determined. The flame
speed (Ut) is calculated from equations (1) and (2) where Tb = tp and Tu = tm The volume
of the burnt mixturo I'~ is given by

le
)f
12
14

al

of
of
er
3S.

where r = u,(6ex/360n)
is the radius of the flame front from the spark plug, 6ex is the
angle increment and n is the engine speed, rev/see.
For initiat.ion of combust.ionF, is set equal to 0001~~and the angle (6ex)deIIlY calculated
from the above expression, 1f the time from the nominal spark angle is less than (6ex)delay
the mixture is consi.lcrcd not to have been burnt and the compression process carries on
for the next timo increment <lex. The process ontinues for as many time intervals as
lWl'l'sslLry unti! t j", t ,,111,1 H.lIg1u [ro ru t.lu- nm niuul spar-k t-iming i.; grout,or than (6ex)deluy
when combustion is Raid to have commenced.
After the combustion of the small nucleus of fuel-air mixture the combustion chamber
is subd ividcd in t o t.wo ZOlH''l; a burnt zo no, suffix p, and an unburnt zone, suffix m. The
process is initiated ill tlll'l'e 8t(']1S. Referring t o Fig. 1, we show the three steps, A, Band C.

-:::.::::.----'

--

- -

3S.

od

Heat loss o.;

mp, /p;'p;'vp"

mp,lp2,lIp2

3.11'

.ar

mml:m
fm I
pi
v I

(1)

ll;fl_~~
LLJ '-- --'
mm2

tm

vm

(b)

(0)

Initial

isis

state

Final
volume

v2

(c)

2
2

v2,p2

Equilibration
of
pressure

Transform
at fixed
volume

FIG. 1. Basis of the first combustion step calculation.

vill

bu
Step A
(2)

The temperature

t;.. and

pressure p' are given by

i of
'ont
ital
the
no

(3)
(4)

HI1'('

vmc

Step 13. Appearance

o] flame nucleus

t;

During this process t;" and p' are kept unchanged. Using
as the product temperature
forad iabutic constant volume combustion from
we can write

t;..

(5)
.une
plus
mea
pose
mate

To obtain the mass of the products mv at this stage, we assume a slight increase in density of
the products compared with the density of the mixture:
(6)

where m is the total mass and r, the flame front radius. The second term in equation (6)
is the slight increase in density which has been found from a large number of calculations
to give a good starting point for the calculation of this stage of the process.
Now

(7)

g.

102

s.

\V. J. D.

BEl'SON,

ANl\AND

and P. C.

BAllUAH

and
(tl)
;;0

that the total internal energy is then

(9)

,..,'Irp ('.

ICqllililJ/'alio!L

oJ

1)/'('88111'('

This step is assumed to bo an ad ia.bat ir constu nt. vol umo

energy is

prO(;('SK.

The total intorna.l

(10)

Balancing the t.otal int.ornu.l energy before and after step C

(11)

which, after rearrangement,

m

becomes

nI'1

t:1I/

t'

l1l

(tm, /.'",

1) =

'/Ill'

t'p'

(.'CP

h- ~t~")
.
\

(12)

'
1)1

As before
~)
(

,k",-J "".

(13)

JJ

and

I". = (p")(kP-J)II.'P
,~p"

(14)

(1 fi)

Combining equations

(12), (14) and (15) and letting

and

we have

o-

= A

which can be solved for 0 by an iterative

(1 - EO'),

technique.

THERMODY='J
Eqn'il-ib1'i-mn

(16)

AMIC:::;

thermodspiam i~8

The following species are considered to be present in the product in the cylinder and
in the exhaust gases. They are referred to by the number in brackets appearing against
their names.
(1) H20,

(2) H2,

(3) OH,

(4-) H,

(5) N2,

(6) NO,

(7) N,

(8) CO~.

(9) CO.

(10) O2,

(11) 0,

(12) AR.

Argon is separated off from atmospheric nit.roger. in order to determine the NO concentration. The composition is calcu lated in terms of molar fractions of these species denoted
by Xi' 'I'ho [\1('1 is writ tr-n ItS C"" Hl\y Oox. and one mole of the total products arises from
et moles of fuel plus (l/cp) t.irncs t.he st.oichiomot.ric quant.ity of air (cp = equivaJence ratio).

A model for spark ignition engine with intake and exhaust systems
The equilibrium

distribution

of these species can bo fully described

(1)

~H2----+

(2)

~()2

( :l)

)-

~~2

2HzCJ

(-I)

H/J-----+

(5)

('02 + H2 ----'?

(iI)

by the following

11,

(17)

0,

(I H)

i\",

(In)

2H2 + O2,

(20)

OH +~H2'

(~l )

J-I/) +CO,

(22)

H2+NO.

H2O+1N2

(7)

103

rcuctions H.n'l t lu- same HS those used bv Vic-khmd et al."

The equilibrium constant K p for the stoichiomet.ric reaction between the substances
.I.B,C,D,
TIU'l.;p

"Hit

bp cx presse.l as!"
(24)

where v is t.he stoichiometric coefficient X is the molar fraction and p is the total pressure.
Using this expression of equilibrium const.an ts for reactions given in equations (17)-(23),
c-an be written respectively as
(I)

X4NX2,

]{~, =

J(p)

J(p) XllNXlO,

(2)

u;

(:1)

]{v, = J(p)

U)

1\ v,

}I.\

(5)

K;

,/(p)

(6)

J\",

= /).\ 9/X

(7)

J\",

(25)
(26)
(27)

.Y,NXo

(~l:i)

1\1//)2,

X3/(b

JXJ,

(29)

8'

(30)

,'(p) X o/(/J JX 5)'

Pi)

where

b = X1/X2
The equilibrium

constants
In K.

for these reactions are given by!"

(]{p)

[I( v,,(T)

Rm,)] T react

I ( v,,(T)

)
] _ t:.Ho
Rmol T prod
Rmo] T .

(32)

The specific Gibbs Iunction is given by the expression

6)

g(J.')
lllllol

= a (l-.lnT)-b

1'-S!.T2_~T3_~'1'4_k

(I

:3

4.

Url'

(:33)

where ay, by. eg, d,p e" and kg are constants, the values of which can be obtained from
ref. (HI).
We have 1:3 unknowns, the 12 species fractions and the total number of mols, so we
need 6 more equations for the solution. Onc of these is
nd

(8)

1St

IX; = 1.

The remainder are the atomic mass balances for argon, carbon, hydrogen, oxygen and
nitrogen.
The method of solution of equa.t.ions (17)-(34) is outlined in Appendix 1.

ented

BASIC

RATE

The following basi rea.et.ion i,,; considorcd

'0111

io).

A+H-~

Kr
J..~. iI

KINETICS
:

C+D,

(35)

104

R. S.

BENSON,

\V. J. D.

ANNAND

and P. C.

BAIWAH

where I(, and Kt> represent forward and backward rate constants.

(A)/(A),

C~,

= {J,

(B)/(ll),

(0)/(0),

y,

Using

(D)/(D),

= 8,

where the suffix e represents oquil ibrium, for the considered reaction, then for
t lu- ron\'HnJ

mt.f' = 1\,(.4) (In = rx{J](,(A)" (fJ)"

unu
the reverse rate = Kb(O) (D) = y8Kb(0),
At. equilibrium

(D),.

tho forwu n l ra.to is equal to the reverse rate,

](t(A),

(E),

= ](b(O),

(D),

R,.

Therefore,
the net rate = 1\/V-J.) (E) - 1\.b(O) (D) = (rx{3-y8) R,.

x o KI~ETrCS
]t, is now recognized that. the formation of NO in an engine combustion chamber is a
non-equilibrium llIofesl>.In an earlier work Newha.I120assumed t.hat the initial conditions
for t.ho expansion calculation are given by ohornical equilibrium distribution of species
corresponding to 11 combination of temperature and pressure representative of the engine
subsequent to flame propagation.
'I'ho assumption is based on two facts, firstly, the
pressure and temperature at the initial stage of expansion are at a maxima and the reaction
rates are at a. maxima, secondly, owing to the kinemat ics of the engine mechanism the
piston velocit.y and hence the expansion rate is very low at the beginning of the expansion
stroke.
In t.his quant.it at.ivo t.hcorotical unalysis rate expressions for 16 elementary chemical
reactions were combined to give la coupled noril inear differential equations representing
13 chemical spocics. 'I'lu-so equations wero numerically integrated along with a coupled
energy equation start.ing at the initial point of expansion where the chemical specios were
determined with equilibrium calculations.
The time variations of the rate-controlled
concentrations of the species from their equilibrium concentrations were shown.
In the present paper no such assumption is used and the reaction kinetics of nitric
oxide are applied from the start of combustion.
The rate kinetic model for NO is based on the theory developed by Lavoie et al?
with some modifications.
The governing equations for the mechanism of NO formation are

( 1)

N+NO

------" N +O,
2

Xlt = 31 X 1010x

..,....----_.

(2)

N+02

NO+O,

](21

= 64

(3)

N+OH

NO+H,

1(3[

N2+OH,

X., =

(4)

H+N20

--~

(5)

o + N20

:;:=----=::::. N 2 + O2,

(6)

0+N2O

(7)

N2O+M

Kof

NO+NO,

KGt

N2+O+M,

1(71

el-160fT);

(36)

106 x '1' x el-3125fT);

(37)

42

1010;

(38)

:30X 1010x

el-5350fT);

(39)

= 32 X 1012x el-18,900fT);

(40)

=
=

(41)

J(Sf;

1012x

el-3O,500fT).

(42)

The above rate constants are given in m /kg mole per sec.
At the temperatures and densities in an internal combustion engine it is a reasonable
first approximat ion to [1,8S1l11W that thc volume of t.he reaction zone is negligible and that
the gas within it cyl inder consists of a burned fraction at thermal equilibrium plus an
unburned fraction frozen at original condition. We will consider the process occurring
in t.ho burned gaR behind the reaction zone only. The rates of energy producing reactions
in a flame are sufficiently fast l;l() t.hut. t lio burned gases are close to thermodynamic
equilibrium. It can bo ussumod. therefore, that H and OH are at equilibrium concentrations, and that 0 iHin equilibrium with 02' The rate equations for NO, Nand NzO
can then be developed. These are developed in Appendix Il; it is there shown that, the
3

A model for spark ignition engine with intake and exhaust systems
rate equation

2.. ~
Vdt

for NO is the most important

((NO) J') = 9(11

2) (
O:e

105

and is given by

RI

1+O:e[1>\((1?2+R3)]

R6
1+[R6((R4+R5+R7)]'

(43)

where O:c denotes (NO)((NO)e'

R is the "one-way" equilibrium rate for the ith reaction,
l' is the volume of burnt gat;; and the suffix e denotes the equilibrium condition. The
kinetic rates for Nand N20 formation are much faster, and may be assumed to equilibrate
instant.aneously, giving values for eN) and (N20).
Along with (NO) determined by
integration of the rate equat.ion, a total nitrogen balance can be formed to yield (Nz) at
each instant..
'Ye now turn to the cycle calculation.

CYCLE
The cycle of events between air valve closing and exhaust valve opening comprises:
(1) Compression stroke.
(2) Ignition and propagation of flame front.
(3) Expansion: (a) two zone; (b) full products.

is a
ions
cies
gine

the
t.ion
the
.sion
nical
it.ing
rpled
were
oUed
,it.ric

(1) Compression

stroke

The compression process st.arts at the j ral'P<,d condition. The state of the gas at this
point is derived by using a perfect. mixing model for fresh charge and residuals from the
previous cycle. During compression any reaction is neglected.
The first law of thermodynamics is
(44)

The equation of state is

pT'
A combination

of equations

mRT.

(45)

(44) and (45) after rearrangement

gives

dp
= [_(l+R)pdl~+(R)dQJ/V
do:
O;
uex
O; do:

t al. 7

(46)

and
dT = T
do:

(36)
(37)

V dex

+~ dP).

(47)

p dex

For the unburned mixture,

dp)

dt", _
(~
V !.
do: - tm V do: + p do: .

(38)
(39)
(40)

(2.. dF

(48)

The heat transfer rate from gas to wall is given by Annands

heat transfer"

equation for convective

(41)
(49)

(42)

The work is
enable

.d that
.lus an
rurr ing
actions
;rnamlC
.oncenld N20
nat the

(50)

As the compression process continues

following general expression

the variables

are incremented

by using the

(51)

where x is any variable.

The numerical method used for this purpose is the Runge-Kutt.a

method.

R. S.

106

BENSOJ:-;, \\".

(2) Lqnltion. and propaqtttion

.r. D.

ANNA:sn

u.ncl P. C.

B,\Hl.'AH

of jlurne [ront

till' provess [01' usr-c-rtu.ining the ignition of the fucl-u.ir

mixture. Unce ignition has taken place we consider the flame to be propagated spherica.lly
from the spark plug.
The \OIU.IlH'S of t he burnt and. unburnt
regions
are computed
by
r.ho methods
described els('whnJ'(,.21
F(JI' t.h is 1HU'j>0::;C,a not.e must,
be kept not only of the
\-rH have

alrr-adv

conside-red

YOlUllH' agaill::;t t.he Ham!' rtul ius but al~o the flame front urea awl the Ham('
with r.ho cyl indi-r wa.l! and piston.
\\-'hen the flame i", propagated and t ho
zones set up by tl\l' mothods
dl'sl'.ri be( I ('Hrlipr, t hr eaiellllLtion
1)J'()(('('d" in a stop-by-stop
mannor
using HlIIlgl1-Kllt,(a
illtt'gnt-t ion,
onflumed

contact

(3a)

:U'C'lt

E,l'p([l'Ision uith t uo

Z01/f8

'1'11(' major tt,;slln1l'titlIlS

HI'(':
(H,) the original
l'ha,rgc- is htllllogelH'()(l>;:
(b) the pressure
ut any time is uniform
throughout
the cylinder:
(c) till' volume
orcupiod
by t.lu- fla.mr- roar-t ion Z()(W is negligib!:
(d) tlu: burned gas i" u.t full t.lH'\J)1tltlynami( "'1"ilibrilUll
ex('('pt fr (lit' nitrog(,1I spcei('~;
(e) the unburned
gas is froze-n at its original
coraposit ion :
(f) both burned and unburn.-d
gasps hav unifor-m loc-al "pt'cifi(' hont s ,
(g) t lun- is 110 lu-n.t. t rn.nstr-:: hpt \\'('(,11 burru-d "IHt unburnc-d ZOllt'S.
It can be SPC'1l that this mode! is cx trornc ly simplified
but, t'x]wJ'it'Ill:E' shows that till'
assumptions arc \'ery we ll jnst ifiod.
Spark

Total

plug

volume
Pressure

Fro.
Consider-ing

Fig,

2, the

total

= P

2. Cornbust ion zouc shape.

internal

energy

for the

system

is
(5~)

The

first law of thermodynamics

dQ

dV

da = do:

is

clW

(53)

+ "'J;; ,

dV

(.,)-1-)

do:

and

53
Therefore,

equation

(52) becomes

.
elm,.",
(./l'v-Il''')-I-+
III",v
{(Y

dl",.
,dl"
cl V dQ
-1-+nl),C"
-I +P-[---l
"lCX
uo;
ca

111<0:

o.

(55)

A model for spark ignilion engine wit h intake and exhaust systems

lr
!:"
'~

cl]'

cH;"

.n;

dex

dex

dex

-=--+-,

('.ally
.nd difl"!'t'ntiating t ho r-quation

(56)

(pi' = milT),

ofstan-

~ = (~_~)
<.!!iI"+'III,,, j(,~ ~~+1'II'p!~
dex
Ill"
'Ill '"
dex
Ji
dex
p

~:iv::~

107

dt,p_~
dJ!,
dex
]I <lex

(!i7)

:\I'I'I,\'ing tlu- f'ir~LIn\\' 10 It un it IImR~ of Ill",

1 (iq",

dll",

,lex
\\"11"1"

4)", i~ IlI'ILI,

I rnll~r(,l' I () I

,HIt tht' nrix t uro 'tllti

\,'t,

,'nil

'11/.".

cl I:;"
dex '

~-}J

h,' uuburmx l mi xt.uro ussumed uniformly dist.ributod t.hroughor tho mixture.

I';" is t ho ~p,,('ifi(' volume

l'ljllltlion
(:,)S) as

write

,,(_1 _dl". -.. _1 tiP)

, dl", _
1 d(.!",
(.,.- ----pi
(la
m III (la:

(!in)

p (la

t 111 (lex

11/

III

,md after

(fitJ)

using t.hr- usual gas state rt-lat ionshi ps. we have

(60)

From equations (!i5), (57) and (60) we can obtain separate equations for clpfdex, clt,,,/dCl.
and dip/cb:
From equation (60),
dt",

I;"

dp

+ 'In",

dex = m", CP dex

From equation
dt"=_P
dex
FmIl1

oquation

(61)

(57),
__

'IlL"

1
clQ",
Cp", '{fc; .

[dl' _(R"tp_Rmf.",)dll/p_Rm,r~"dP_

R"

da:

}J

pC"".

da:

})

Em dQm+.!::dP].
pC"", da:
p do;

tlex

(62)

(55),

X tl'll1"

(C.,,,._ Cv" Rm)

C"m

da:

The heat Iransfor terms ure determined

For burnt. mixture

() q [(

R"

dQ", _ dQ .

o.;

(6:3)

dex

from the following relations,

r(

L>'

I{p

)~;7 (/." - tw)

(64)

where
( I{u)"

PI'

[)I ;'i.<tOIl

OppfJ-1)

arul

0'7

fkp

For unburnt. mixture

{/" [(",(Hp)?}

(/'" - Iw)

(65)

where
(He)". = pm

Dl;>istoll

l.Lm

The piston speed Vpistoll

number expressions.

and

I{
m

and cylinder bore D ar used

= Cv", fkm
0'7
f1R

the parameters

in the Reynolds

R. S. BE~soN, W. J. D.

108

AN~AND

and P. C. BARuAH

The pressure changes dp/c1excan be obtained directly from equation (63) since all the
terms arc known. The temperature changes dt",/dex and dtp/dex can then be dircct ly
evaluated from equations (61) and (62). respectively.
This completes the calculation for the two zone expansion for the time ex QCA. We
then calculate the new variables x at the next. time step by the Rungo-Kut.ta method
using the general expression
d.i'

t'n+1 =

where :V is any variable, the suffix n

increment, in crank angle.

-r"+-l C:.a:
lex

+ 1 is

the new time stop and t:;ex

IS

the time step

Time step
In the Runge-Kutta
method the crank angle increment C:.ex is the parameter which
decides the accuracy of solution and the computing time. If C:.ex is large the results will be
inaccurate and if it is too small, t.lio computing time will be higher. To make a compromise,
a test was developed to maintain the right level of accuracy while adjusting the value of C:.o:.
In this method all the varia.blcs (except the variables heat transfer and work output)
during the expansion stroke arc tostcd to sco if the increment of any of them during a
time stop (C:.ex) has d()('I'clt,wd by Jll<H'O t.hun j,O pm cent. of the varie.Llo itself. If this ifl
found to occur the time st op is halved and the calculation repeated. In the actual
computing, with an average value of 6a = 0-1-0-5 CA during the expansion stroke,
reasonable result.s are obtained. It. is to be not.ed that during the two zone period the time
step is divided into Et number of sub steps.
Termination

of combustion

As the combustion process continues the product volume increases at a rate determined
by the flame speed and correspondingly mixture volume decreases by the same amount.
The completion of combustion is detected by the current mixture volume. Theoretically,
combustion should terminate when
O. In the numerical method it is assumed to
terminate at the beginning of the time step in which the current value of Vm is just negative.

~:n~

(3b) Expansion

with [ull product

Once the combustion is complete the variables are organized to calculate for single
zone only. The Runge-Kutta
method is used. Throughout the expansion calculation a
check is kept, to see if NO is frozen when the NO rate kinetic calculations are by-passed.
This completes the power cycle. V,Te now turn to the gas exchange process.

GAS EXCHANGE

PR.OCESS

We will first consider the gas dynamics in the pipe then examine the cylinder boundary
conditions between pipe and cylinder. Finally, we examine the cylinder thermodynamics.
Pipe
Benson et al.23 developed methods for studying non-steady flows in gases with variable
composition and specific heats. These methods gave comparable results with the normal
method for gases of constant specific heats;" We propose, therefore, to use the normal
methods? with modifications to allow for the calculation of the gas composition along
the path lines, and for the variable specifie heat due to temperature and composition.
We will start by reviewing the basic cquat.ions then proceed to the method of solution
with variable gas composition.
For one-dimensional non-steady flow with friction and heat transfer the basic
equat.ions'" are
Continuity:
(66)

A model for spark ignition engine with intake and exhaust systems

109

Momentum:

op

211
811
1
-+'1t-;;-+--;;-+'
(.1:
P (,.l;

ct

F'

(67)

where F' is wall friction defined by

F' = ~~~

2 III I

J)

The first law of thormodynamics

qpP

x.
t)
a

is
sI
:e,

oe

d.l;

[(pF d.r)

711)

-1-2'

'2'pn

(encrgy equation) is

(Cv

'1' + U;2)]

+ :v

[(puF)

(Cv

'1' +~+

'1;2)] dx,

(68)

where q represents rate of heat transfer per unit. time and per unit mass of fluid.
Using continuity and momentum equations and simplifying, the energy equation is
obtained in the form
0
8
(69)
(k-l) p(q+u.F') =
+u ",p -a2
+u

of
ot

(;

(!
,f).
cA

uX

uX

The heat transfer rate q may either be through the walls or be the result of longitudinal
heat conduction, In the event of a chemical reaction with change of state these will be
represented by It heat addition equal in magnitude to the constant volume heat of reaction
of the actual process."
The equations of continuity, energy and momentum are solved by the method on
characteristics.
For the continuity, energy and momentum equations this gives the
following charucteriet.ica:
dx

ed
,to

anc I j =

dt = u a

(70)

(characterist.ice)

and

ly,

dz
dt = u

to

(path lines),

,'C'.

Combining the continuity,

we obtain
.gle
a

dp
du
dtpadt-(k-l)p

:1

energy and momentwn

(4j

equations with the characteristics

a2 pu dF 4j pau2
+-yj'dxD-2-fUj=O,

U31.1,)

q+D2GZi

which is the compatibility condition along dxfdt = u a.

Combining the continuity, energy and momentum equations
obtain
dp a 2 dp
(41
u
dtdt-(k-l)p
q+1t D2GZi
= 0,

2 U)

.ary
iics.

lines, we
(73)

.:\=A+lc-1U,
2
fJ=A_k-1

and

rtion

A=.:\+fJ
.
2'

oasic
we have the characteristic
Direction condition:
(66)

with path

(72)

which is the compability condition along dxfdt = u:

Defining the Riernann variablea= as

able
'mal
'mal
long
11,

u=

.:\-fJ
k-l'

solutions in the Iorm :

clX _ U
dZ -

+ '

(74)

.!
'()

110

R. i::l.BE.'>SON, W. J. D.

and P. C.

ANNAND

BAHUAH

Compatibil ity condition:

d,\

_"-1

:2

Al.j ~dZ"".1
F dS

(L..j,,_~~
".-1"

2!-Y:reIU2

IT)

x ( 1-(1\-1)-

...f

I /..,'I

lJ

:2

,
(k-l)2 qX'r'" 1 I'
dZ+---,--<Z.
a#ef A

(7;';)

I)in'('( inll ('tllldi(i(lll:

I' -.1.

(11i)

Compat.ibility cond ition:

df3=_k-lA[;~.IP
lZ
~
2
F' dX ( + A A
x

( 1-l-(k-1)

k-12fXretU2~
DIU

+:2

U)'

.-1

r:
(/,:-1)2 qXref 1 r.
dZ+--)--3- -(lZ.

:.

(jeef

(77)

Path Iine charactorist.io :

u.

(78)

Compat.ibility condition:
Aa (gXref 2fXrd I U31)dZ.
-3-+-J)
.
arc(

_ k-1

( lA -

--)-

-:-12

.:.J

(79)

A numerical method for solving these equations has been described by Benson et al.22,
It has been tested extensively on various engines with good success, I, 2
The heat transfer term ill the '\, f3 characteristics is
(ri'I\)hcat

trnusf'er

= (k-l)2qXrcl~dZ
2

3--

(lrer

25

..

A comput.ational scheme using these terms for wall heat transfer in both subsonic
and supersonic flows using Reynolds analogy is given by Benson.P" In the case of chemical
reaction the heat of reaction is included in g.
q =

gheat transfer

+ qrenct.ioll'

A detailed method for including heat of reaction in the method of characteristics

soh'ing n, problem of It flame t.ubo has bee-n gi\'('1l by l\enson.24

for

Patt, lines
'I'h identity of a fluid eloment iHpreserved alt>ng a path lino. I n the present calculation
tho put.h line composit.ion is dofinr-d by t.ho n io la.r (',mcent.mt.ioll of the 12 species listed
earlier. Chemical reactions are considered along the path lines. At each time step the
composition is calculated from the t.horrnal quilibuium equations (17)-(34), with the
~O, N20 and N kinet.ically controlled through the expressions (36)-(43). From the
chemical reaction data the heat of reaction can be ovaluated. In the calculations described
in t.his paper no heat uf reuction was included, but. it will become important in thermal
reactors.
From a knowledge of the chemical composit iou and the temperature
the internal
energy of the mixture and the gal'; constant R can be obtained whence the ratio of the
specific heats can be determined for ('uch pa.th lino. The following procedure is used.
Initial values of the species are entered at, t.imo t = 0 from the release conditions of the
first power cycle.

II

III

A model for spark ignition engine with intake and exhaust systems
Considering a path line, Fig. 3, the pressure and temperature
expressions

. _ (1I+f3)2kl(h-ll
.) 4

Ji -

~.

are calculated

from the

(80)

Prel,

(81)

From Jl and T t,hp quilibrium composit.ion is calculated from equations (17) to (34)
using the same met.hocl as in tho l'yliwlel' a.lcu la.t.ions, and from the equilibrium values of
(~O), .. (N)" (N20)c the kinet ica lly controlled amounts of NO, Nand
N20 are evaluated
as ill t he cylinder culoulat.ious from cquat ion (:\6) to (43). In practice, the equilibrium
calculut.ions and the rate alcula.t ions a ro ol'g[l,nized as subroutines which can be entered
eit.hei-from the cyl inrler or the pipe ca.iculat ions. "re now have all the species, the pressure
and temporat.urc : we call calculate, therefore. the ratio of the specific heats, k, in the
following manner.
i)
Enter species trom
pr evious time steo

Z+LZ

Interpolote

atAandB
P,'T'

LZ

Boundary

Boundary

25

I
-.-l __

_____

~~~
Aa,

'"

P,T

[NO] Along
IS

:1. Cn,lraiIalion

1<'1(:.

ThL' spl'l'itil'

illh'nml

lIllorgy
'll

'Ill

OU)

+ 1/2 T+
a

1<z

Xn

U(2,tl))

+ 2CL x"

V(3,"))

171

mal

-nal
t hrL
t.he

u" T2+U T3+Ur. '['''

+ ')~'U3

'" + 3
J.

it,

7'

+ 3(2:

:t:n

whore n is the 1I111nl)('r of 'lpe"il's awl /I is t.he internal

Molecular weight of mixture
where

the
the
bed

pat.h line species.

(82)

111

7"- + 4
. it5 7'3 .

(83)

of species

c, = CL

t.hc

po+hune

controlled

is

C v = ( 8T
For a mixture

ion
ted

01"

rate

alld

cal

for

I
I
I

9)

.uc

,-

-~----- A\

r-I

3)

species

is t.he molecular

weight.

.1'

2:

1./.(4,n))

1'2

+ 4(2:

:lOn 1./.(5,n))

1'3,

(84)

energy of the mixture.

(85)

i!1; Xi'

is t.ho species fraction,

is the number

of species,
(86)

and

(87)
The effects oJ ariable gas COli/position and variable specific heat are identified by k in
the path. line and characteristic cO'llljJatibil'ity equation (75, (77) and (79). In the boundary
ca.lculat.ions we iruludo the offoct nf variable composition in the following manner.
The boundary species are entered OIl the boundary path lines depending on the direction
of flow. The general approach at this stage is to enter the species coming from the boundary
on to the boundary path line if there is flow into the pipe. Owing to the nature of calculation, t.ho values >LW PlltL'l'c(1 from the previous time step. 1 flow is leaving the pipe
the species are interpolated between the inside and outside path lines (Fig. 3). \;I,'e will
consider three cases: (a) cylinder, (b) junction and (c) nozzle and open end.

R. S.

112

BENSON,

\'1. J. D.

and P. C.

ANNAND

BAJWAH

(a.) Cyl-inder boundary

In this case when there is outflow from tho cylinder into the pipe the pipe end boundary
species are filled in with the cylinder species. If there is inflow from the pipe into the
cylinder the boundary species are obtained by the int.orpolat.ion of the inside and outside
path lines at the pipe end.
(b) Junction

boundary

The junction boundary uses a constant pressure pcrfoct mixing model. The principle
applied here is that only the boundary species flowing into the junction will participate
in mixing (Fig. 4). The resultant mixture species will be entered for the pipe ends where
flow is moving away from the junction. The pipe ends, where flow is into the junction,
will retain t heir own incoming species.

- ve flow adjust
outgoing species

FIG, 4. Junction boundary with species.

The flow direction and the mass flow parameters are obtained using the conventional
expressions. Using suffix nn to denote a pipe end at the junction, we can write:
Velocity parameter:
(88)

Mass flow parameter:

w - (,.\. -)
n rr

1Illln

t
'''01l.,11I

r".!!..F
A2
l1n

(89)

nn

Considering positive flows only, represented

concentration for species i

by W~" (Fig. 4), we write the average

(90)

For pipe ends having negative flow

(X;)nn

= (Xdav

A similar procedure will be applied for NO, Rand

are not included in the junct.ion boundary.

(91)

k. The equilibrium and rate equations

(c) Nozzle and open end boundary

In this boundary when flow is going out, of the pipe into the ambient condition the
species at the pipe end are obtained by the interpolation of the outside and inside path
lines existing at the pipe end. When there is inflow into the pipe the values from the
previous time step are used, to fix the pipe end species at the current time step. This is
an approximation, in actual practice a slight amount of ambient species may enter into the
pipe. However, it is expected that mass blown back will be small and this approximation
will have negligible effect, on the prediction of NO. EoI' engines with excessive back flow
the calculation will have to be modified to allow for the mixing of air and gas.
To obtain the integrated mass flow for any gas ;'( we use the expression

X - Im{X,)

Im

'

where Xi is the mass/mass concentration and. rn is the total mass flow rate,

(92)

A model for spark i(!nilion engine wit.h intake and exhaust systems

11:\

CylinrlcT
In t his model wo assume that the composition of the products of combustion are frozen
.tt release (e.v.o) arul rcmain so during tlw gas exchange process in the cylinder. When the
gases are in the pipes the residence time will be longer than in the cylinder and the
equilibrium and rate equations 0.1'(' used. The value of le in the cylinder will depend,
therefore, on the tcmperat uro in the c.ylin<iL'runt.il the air valve opens. At this point wc
aSSIU11t1 for a four-stroke engine that, the incoming fuel-air mixture mixes with the products
of combustion but there is 110 chemica.l reaction. We will now require to evaluate continuously the composition of the cylinder contents to determine le.
For flow through the inlet and exhaust valves, we assume a quasi-steady theory.
'I'hr- Riernann variable '\1\ lLIHI the entropy level Aa a.t a t.imo step are known from the
previous time step, They have, howr-vcr, to bo modified for inflow to the pipe for the
cu rrvnt. timo stop. All outline lJl'()('edlll'l' for the boundary orul it.ions for the valves
follows. \\'0 will then describe the t.hormodynamic relations for the gas exchange process
ill the cylinder.
Flou: through a valve
Outflow.

The boundary conditions at a given time Z will be dependent not only on

the valve but also on the conditions immediately downstream of the
valve. The Riemann variable Ai" is modified by the entropy Aa" of the gas entering the
pipe. The following expression is used :
t he flow through

'\"

1\1111

+ (A/Aa)

(Aac-Aun),

(93)

where \n c is the corrected Riemann variable, Ain n is the uncorrected Riemann variable,
Anc is the entropy of the gas leaving the cylinder, A"n is the entropy of the gas in the pipe
and (A/Aa) is the value of (A/Aa) eit.her at the previous time step of the ourrent time step
(usually taken as (Ain c +Aout c}/2Aac from the previous iteration).
,al

The flow .through the valve is based on the constant pressure model.vforms of the boundary equations are, using the parameters,

U-~
-

3S)

30)

21

The general

ac

and
(J.: -I) U2
-:-:(1--=-'t;-:'( le":'"---':-I
}:--'U=2=)
_,,_1

C=

90)

(94)

S u bsonic flow in va,lve

(::)
(1)

Sonic boundary

.ons

( le2 4_ 1 C)'

Choked flow in valve

the
rat h
the
.is is
,the
.tion
flow

(95)

(2)

Pp) _
- if} -(Pc
k+ 1

> ,/,2

sonic flow.

'f' '

(;-+11/2lk-1I

le l---U
2

Ell/ropy

(P

\ \') ("'
re ) Ik-11/2k
A ac-_ (A' Il1 c + /I.out
2
Pp

_ (3-1.)'
1.+1

AOIILe -

\ll('+

J(4(k-l}
(k-i-l)

re

Pc

A,-

1-

(96)

2)

-,
U

t') Ik-11/2k

(97)

(9S)

(3-lc)2]
lc+l

Aine

(99)

The computational procedure for a given value of le is to first estimate the entropy
level A ",t' Let Anc = Aa""t and calculate lilt: e from (93), then evaluate !tout c from (99),
Whence U can be determined from
(92)

U =

Aout

-Aill

(le - I)

(100)

R.

114

s.

BENSON,

\N. J. D.

and P. C.

ANNAND

BARUAH

T1H'1l whether the flow is choked or nut. t.lu- l'l'l's:;lll'e ratio (p,,/pJ is determined from
either (07) or (05). TI1(\ entropy Aa, is now evaluated from (98) and compared with the
c-st.imutc-d va.lue of /1 "e-t.' I f the two agree within the accurar-y required, then the calculation
is l'()}llpll'l ('; i i' not, t.lun tilt' c'al('liln,(ioll
is l'l 'Jl' ,,"t,od w i I h a.not.hor est imn.t.o
A"
unt.il t.lio required
l1l'l'uJ'Ul'Y
is o bt.a.i nr-d.
(If

Lnjlou',
t hruut

FoJ' inflov,:

jll'l'SSlIJ'C

In'

assumo that

i" ('l]u,),1 10 the

hc Hovv i" isr-nt rop i to t h throat

('.vlindl'r

PI'!'>'slll,('.

'I'Iio

boru u lu.ry

and that t.ho

ond it iona al'(' IlO\V

1'l'j>l'C'sc'lltl'dby till' expression

(10] )

for su bsonie flow where

A*

out -

It, is not, us uu.l to huvo

we can determine A~"
fully determined.

Cylinder

/\""1

_,.1/1

{Jirl.r)Ck-11/2k
\

Pc

r-ho kr-d inflow t.Ju'ollgh it \.;\.1\'(', From the vah ros of]J.., Aill and A.,
Equation (l01) is t.hen so lvcd for A * and then A~lIt whence Aout is

tlierniodijnamics

The combustion products, suffix 'jJ, are oonsidi-red t.o have a fixed composition at release,
This remains constunt until the air valve opens, a.v.o., when the air-fuel mixture
enters the cylinder.
The air-fuel mixture mixes with the products uiithout chemical
reaction to give the final products. suffix g.
The molecular weight of the fuel ail' mixt.nre '11/ "'m' and the products m are obtained
from the expression (85). The specific heats at constant volume for the air fuel mixture
C,.", and 1110 products 0"" arc. ol it.a.inod frorn .-xprossion (S4). Now
Cl.V.O.

UJ"

R =

HIII"i.
1nw

Ii ;

JI"lItli

(10:2)

=
'lJ1w}I

n.:

1))/11

(103)

'11/111111

})/II

C
and the numbe-r of mols

JlI'

+ It

(}I'iII

O"J)+H,p

(104)

Ill'

(105)

'}}I('/II

(106)

~\III

III

X,,=

Will

Ul(.-nll'm

(107)
1}J1.

}J

where ?ne is the total mass in t.he cyl inclcr and 1ll,. is t.ho mass of air-fuel
total IIln.SR ill the r-vlindcr
11/, = '11/ i" giv('1l ill'
'"
"
(J)u

'Htlly

The "l)('cific hoat s

'Lt,

constant.

Xm

1nll:m

+ ~\{ ','11v

Pl'l'>"SIll'l' Hlld

'P

will

The
(108)

= 'Inc

\'(1;11111('

mixture.

be
(109)

(110)

A model for spark ignition engine wit.h intake and exhaust systems
and viscosity

. _ (..{Ym Z('oeffm

:.JI.-oellq

();

/.L ==

,,rnl'/'m
,I'll/Will

+ X.pZ('oetTp
+ ....
y p \hn wp)

tr)

J11'l

115

(Ill)
,

(112)

Zl'oett'j/ X 110.645,

Spocific heat. ratio:

( 113)

mass flows into and out, of t l: (,.ylilld('l'(dlll";(le< an. l dm,,fde<) are controlled
by the
boundary condit ions at t lu- vaivos.
'llu- ('ylilldl'l' pressure (Pc) is calcula.ted by the first
lu.xv <Iftl\('rlll(l(I)'lH\'llil'~ ill till' form

The

I)
1;. <lp,.
---=
1.-<-1 <lex

whrn there is outflow

k,.

d 1;.

t hrough

dill i
( -. de<

led
llre

103)

104)
105)

106)

(l07)
The
(108)

(109)

(110)

elm"

dQ

d '

term on the right-hand

(114)

side

i;; llegative .

The mass balance in the cy linrler is

do:

ue
cal

(J~,

ho ill let, vu.lvo, the second

dmr
SE',

dill,

1.',.-1 cdex

o'

is

of,

-+--------+k,-1 do, kc-1 dex

---}I

dmi

llm.

~-d;;'

= (1IIe) " +
(dme/ch) I:;.e<, (Pc)"+J = (Pc)" + (dPe/do:) /:;.0: and (T ),,+l = (Pc ~/RmJ"+J'
The heat, transfer rate is calculated using Aunand:s convective equation (65)9 with
tnL = t" and
CPm = Cpu
The speed of sound a." is determined from the kev and R Cv and the cylinder temperature.
Thus, the basic wave action and gas exchange computer programs developed for
compression ignition engines can be modified by the adjustments indicated, In essence this
includes the addition to the path line compatibility condition Aa' the concentration (X,.)
of each of the species. These in turn lead to the updating of le for each path line as a
function of p, T' and Xi' The computer logic is more complex and the chemical reaction
(,denlat ions al()ng path lint's Hr<'t.inu- (()Il~l(Jllillg.In pru.ct.ico, below certain fixed tompornt ures the constituents arc almost constant.
111 these cases the computer program by-pusses
tho equ ilibr-iurn t.hermodynarnie expression and the rate equations.
(tncln+l

TERTS

ON SI1I1ULATION

MODEL

'rho model was dl'v('lo]ll,d in t.wo ]larts n.ru I then brought together and tested as an
integrated whole. The power cycle was developed and tested against work carried out on
iI small gas engine on which cngino rosearr-h had previously been carried out on combustion
limit.ations of gaseous fuels.!? The, gas exchange process had been fully developed for
compression ignition engines. This program was modified to include the chemical reactions
along path lines as wel! as the gas exchange process in the cylinder to allow for 12 species.
The small gas engine was set up as shown in Fig. 5. No carburetter was fitted, the fuel
and air being mixed in the chamber shown. The inlet pipe was 0600 m long, The exhaust
pipe was 0346 m long with a sampling station at. 0219 m from the engine. The receiver
>tt.t-he out.let end of t.ho exhaust pipe was connected to the laboratory exhaust system,
thoro was therefore a back pressure on t.he RyRt,em. The engine bore was 0105111, the
st.roke 0152 m and the compression rat.io 77. Tests were carried out at constant speed
of 920 rpm and variable air-fuel ratios, as po.rt of an investigation of sampling techniques, 28
\Ve only report here the NO results obtained with the N.D. Infra-red Analyser.
The purpose of the calculat.ion tests at. this stage were
(i) To establish whether the simplified 1110(lelfor the closed cycle gave reasonable
prediction of NO in exhaust pipe.
(ii) To establish whether t.ho simplified model for the open cycle gave reasonable
prediction of t.he effects of the gas exchange process on NO in the exhaust gases.
In the engine under test, the valve overlap was 18 CA (EVe 372, AVO 354).

116

R. S. BENsoN, W. ,T. D. ANNANDand P. C. BARuAH

If the model is to give a good simulation of the engine under test the number of
estimated variables in the input data should be a minimum. The flame factor ff is one
unknown, which must be based on previous experience. However, to reduce the computational time a good ost.imat.c of the rosiduuls will help. It should be pointed out that the
calculation wi ll automatically adjust. the i-csiduuls. In order to test how sensitive the

Sampling station

Cylinder

/@

CD

Junction
"'-. Mixing chamber
Receiver

Pipe No.

Length
(m)

IT
ID

Diameter
(m)

0219

0044

0[27

0044

0600

0044

FIG. 5. Four-stroke cycle engine configuration.

60r-------------------------,

::t

30-

'"

"0

x
o

.~

...

Fraction

residual

004--007
-_._Flome factor:

40

14
Air fuel ratio

FIG. 6. Effect of fraction residual on nitric oxide prediction.

prediction of NO is to the residuals a number of power cycles were calculated with
the flame factor set at 40 based on the previous work of Bailey-? on the same engine. The
results are shown in Fig. 6. It is quite clear that the peak NO is strongly influenced by
the residuals. This is not unexpected since NO formation rate is dependent on the
maximum temperature and this is influenced by the internal energy of the inert gases.
The greater the quantity of inert gases the lower the temperature rise during combustion.
The flame factor ff of 40 was based on Bailey's work-? which did not include wave
action calculations. We therefore carried out a number of calculations to establish the

A model for spark ignition enginc with intake and exhaust systems
)f
le
11.-

ne

he

117

influence of this factor on the NO prediction when wave action is included and the back
pl'C'>ismein the receiver set to the experimental value. The initial value of the residuals
was selected at 0055 and.u set at 3,0, 3'15, 3'5 and 40. The calculation was carried out
for some 8 consecutive cycles (16 revolutions) for each value of if. A typical result is
shown in Fig, 7. It. will be seen that the residuals soon settle to a constant value and at the

'"

- 25!0

!----

n
E

~
- 2490 -'0"
Q)

CL

- 4200
'OE
-, .:.1000

20.
~ 0.

<,

04

-I

02

E
E"
0'"

Q)

Q)

IJ..~

- 0060~

-I
2

-I

+-"0

u ,-

- 0056

I
3

I
4

o '"

l~

No of computed cycles
Equivalence ratio "0'9

Flame f cctor

>

3,15

FIG. 7. Number of cycles to steady state.

I with
The
ced by
)Il the
gases.
ustion.
3 wave
ish the
3.

eighth cycle the NO emission at the sampling point is constant. In Fig. 8, the NO results
at the eighth cycle at the sampling point are shown. Here it is clearly seen that the factor
,ff has a strong influence on the results. By comparing the results at an equivalence
ratio of 09 with the experiments, a flame factor of 315 was selected as the base for the
speed of 920 rpm. Calculations were carried out over an equivalence range of 08-1,1. The
results for the eighth cycle at the sampling point are shown in Fig. 9. On the same graph
the experimental results are also shown. It will be seen that there is good agreement
between experiments and calculations. It may be deduced, therefore, that the integrated
model gives good predictions of the NO emissions. The mass flow predictions were also
very good. However, no measurements were made of cylinder pressures or power so we
are not able to compare these results. A further paper will examine these parameters."
In Fig. 10 the calculated influence of receiver back pressure on the emissions is shown.
It will be seen that emissions are reduced as the back pressure is increased. This is due
to the increase in residuals in the cylinder which reduces the peak temperature. All the
results show that the calculated emissions at the sampling point for NO were very nearly
the same as those calculated in the cylinder. This was due to the short distance between
tho sampling point and t.J1O cylinder and the> low temperatures.
Details of the calculation (Fig. 11) show trends which have been reported by other
workers. This gives confidence in the modelling. The trend of flame speed with air fuel
ratio (Fig. 9) agrees with a number of published resulte.P The increase in flame radius
with time (Fig. l1(a)) agrees with results given by Bailey?" and Iinuma.s" The variation
of flame speed with angle (Fig. l1(b)) agrees with test results of Iinuma.P'' The prediction
of maximum NO and its relationship to the time for freezing (Fig. l1(c)) agrees well with
the work of Lavoie et al.'
9

118

R. S.

BENSON,

W. J. D.

and P. C.

ANNAND

BARUAH

The method is now being extended to four-cylinder four-stroke engines and to twostroke crankcase compression engines. It is hoped to report later on the four-cylinder
work, the results from the first stage of the crankcase compression engines are given in
another paper."

Equivalence ratio = 09
Computed points'

0,6,0

"Q
E
a.
a.

..----1\--__

---

6-

__ 6-6

-0

'"'"~
Q)

G:

90L---=--3LO----.....J3-.5-----4L.;.O
Flame factor
FIG.

8. Influence of flame factor.

a.
E

ID
...

o:
e
Q)

a.

",

E
a.

a.

...::>
~

ID

a.
E

Q)

ID

13

I-

11

.;

"

o
+-

III

Q)

I
I

-0

a.

I
I -I

'"

III

E
.'2

[NO] Computed
Experiment

u,

Air fuel ratio

14

FIG.

:2

9. Influence of air-fuel ratio.

Flame factor 3'15

A model for spark ignition engine with intake and exhaust systems

119

0-

.er
in

2500
2480

1-0

12

Back pressure.
FIG.

bar

10. Influence of back pressure.

(cl

_-----'''12700 ~
c:

.s
+o u

a.

\ \

'"'"o

"., '".

c
E ~

-e
x
o

+- "
:J:.o
Q.

o
U

.0

\\

-e

AI

E E

~
o

\~.

\\
/

"-

".,'"

<,

'"

20 -

0.

.;:
+-

.x:

'"

time

0.

--NO
--Temp
----Press

10

from ignition,

Equivalence
ratio 0-9
Flame factor 3-15
10

35

60

Crank ongle,

85

er:

110 135

ATDC
6th eye le

--

Computed

flame

---

Computed

prod. fraction

14

speed

12
10

E
0

+-

tcc
0

;:

Two lone
termination

~
c

0
705

720

15

30

~ ~

Equiva lence rotio = Og

,,'.

.
.,'"

Cylinder

temperature

0>

"

a.

+-

IJ..
-'"

I-

!"

'1>

c.

"

--------45

!38

360

572

Aye

EVO

Cronk

angle,

Cronk

ATDC
FIG.

11. Performance prediction.

ongle,

ATDC

120

H. :-:;.BE:SSON, \Y. J. D. A:SNA:SD and P. C. BARUAH

CONCL

USIONS

A comprehensive simulation model of 11 four-stroke cycle spark ignition

engine is presented. The model combines a full power cycle simulation with
the prediction of NO emissions with a comprehensive gas dynamic model for
the cylinder and ducts allowing for chemical reactions in the exhaust pipe.
Calculations are presented comparing the predicted NO with test results from
a. single cylinder cngine. It. is shown that good agreement is obtained between
the predicted and measured NO over an equivalence range of 08--11 when the
flame speed is corrected at the equivuleucc (00) corresponding to the peak NO.
The NO emission is sensitive to the residuals and the exhaust back pressure
both of which are automatically included in the overall simulation model.
Further work is required to obtain a better quantitative understanding of
the factors which influence the flame speed. In the meantime the selection of
an empirical factor jf which when multiplied by the Iarninar flame speed gives
the turbulent flame speed is suggested.
The program written from this sim ulatiou model call be used to assist in
the design of iutake and exhaust manifolds as well as to study many problems
such as location of thermal reactors, catalytic devices, exhaust gas recirculating
valves; effects of misfire and maldistribution of fuel-air mixture. Later papers
on these applications will be presented as the research progresses.
Acknowledgements-The
authors
wish to acknowledge
with thanks the assistance
of
Mr. 1. Milne who made available the results of his experiments
on the National
Gas
Engine, l\Ir. J. Nicholson for his assistance in the computing work and Dr. B. J. Tyler
for his advico on NO measurements.
The work has been carried out in both of the Mechanical Engineering
Departments
of
t.ho Univorsity of Manchester and U.M.l.~.T. wit.h t.heir funds. It. is now boing sponsored
by t.ho t)"iOlll'tJ HO"O>tl'l'.h Council.
All tho calculat.ions were carried out. on tho V.M.R.C.C. C.D.C 7600 computers.
Wo
acknowledge
with thanks the assistance of the Director and his staff.
REFEHENCES
1. It. S. BEXSON, S.A.E. Auto. Engng Coruj., Detroit, Michigan, Paper No. 710173 (1973).
2. H. S. BENS ON and P. C. BAIWAH, S.A.E. Combined Vehicle & Powerplant Meetings,
Chicago, Paper No. 730GG7 (1973).
3. J. B. HEYWOOD, S. M. MATHEWS and B. OWEN, S. A. E. Paper No. 710011 (1971).
4. J. B. HEYWOOD and J'. C. KEcR, EnviTOII. Se'i. Tech. 7,216 (1973).
t!fl It. J. TABACZYNSKI, J. B. HEYWOOn and J. C. KECK, S.A.E. Paper No. 720112 (1972).
n. S. BENS ON, P. C. BAIWAH and B. WHELAN, 1. l.1ech. Engng.
7 G. A. LAVOIE, J. B. HEYWOOD and J. C. Kacx, Combustion Sci. Tech. I, 313 (1970).
. ,. C. W. VrCKLAND, F. M. STJtANOE, H. A. BELL and E. S. STARKMAN, S.A.E.
'Prams,
70, 785 (1!l(2).
9. W. J. D. ANNAND, Froe. 1. Mech. E. 177,973 (1963).
10. D. B. SPALDING, P. L. STEPHENSON and H. G. 'I'AYLOR, Combustion Flame 17, 55 (1971).
11. K. AKI'J.'A, i.. Chem, Engng 11, 739 (1971).
12. A. S. SOKOLIIC,V. P. KARPov and E. S. SE:\1El,OV, Combustion, Explosion Shock Waves
3, a6 (1967).
.
13. J. CHOMIAK, Combustion Flame 15, 319 (1970j.
14. N. M. HOWE and C. W. SHIPMAN, Tenth Si/mp. (Int.) on Combustion, p. 1139 (1964).
15. H. A. PHILLIPS and P. L. ORMAN, Advances in Automobiie
Enqineerinq,
Vol. 4,
Pergamon Press, Oxford (19G6),
16. E. MALLARD and H. Le CHATELIER., Annales de Mines 4, 274 (1883).

\ '\a..>'1r
(j.f.1'''
\stI',
~
.

A model for spark ignition engine wit.h intake and exhaust systems

h
ir

"
:n
m

1e

J.
re

e1.
of
of

121

17. A. C. BAILEY, Ph.D. thesis, University of Manchester (1971).

IS. D. K. KUEHL, Eighth Symp. (Int.) on Combustion (1962).
Thermodynamics.
Commonwealth and Inter19. R. S. BENsoN, Acit-allced Enqineerinq
national Library.
20. H. K. NEWHALL,TU'elfth Symp. (IlIt.) on Combustion (196S).
... 21. W. J. D. ANNAND,.1. Mech. Ellgng Sci. 12, 146 (1970).
Int. J. mecli. Sci. 6, 117 (1964).
22. R. S. BENsoN, R. D. GARGand D. VVOOLLATT,
1'roe. I. Mech. E. 184,276 (1969-70).
23. R. S. BENSON. A. w. n.nd D. VVOOLIoN!"]',
24. H,. S. BENSON, Report. No. 1 for C.E.U.B., U.M.l.S.T. (1.970).
25. R. S. BENSON,Lnt.. J. mech. Sci. 14, 635 (1972).
26. K JENNY, Doctoral thesis, KT.H., Zurich (1949).
ami K. GALLOWAY,1'1'oe. I. Mech. E. 183, 2:')3 (UJ68-G9).
27. R. S. BJ<;NSON
2S. 1. L. MILNE, M.Sc. diss., U.M.LS.T. (1971).
29. \V. J'. D. ANNAND,U.N.LC.E.G. Discussion Draft (1971).
30. K. IINUMA,Jap, A'lIto. Res. t-. Tech. Memo, No. 2 (I U71).

Tes
APPENDlX

in
111S

ing
.ers

NUMEJUCAL
The equilibrium

PROCEDUH,E
EQUILIBRIUM

equations

(25)-(31)

FOR THE SOLUTION

EQUATIONS

can be rewritten

OF

as:-

of
Gas
'yler

(3)

ts of

(5)

=
Kb =
K; =
u, =
K. =

(fl)

11., =

[(Po'

(A6)

Kg =

]C",Ny,

(A7)

10.

(AS)

(1)
(2)

(4)

01'00

Wo

(7)
(S)

K,

IX;

Kp.l~p,

(AI)

Kp.l~p,

(A2)

R.p.l~P,

(A3)

K".Ip,

(A4)

R.p.l~P,

(A5)

The atomic balances are

973).
'i1lg8,

,971 ).
1972).

1970).
fral1s.

1971).
Waves

(1964).
iT01. 4,

(9)

X12 = av,

argon

(10)

carbon

(Il)

hydrogen

(12)

oxygen

(l3)

nitrogen

= aw,

'l,X.1+2X2+X3+X4

= ax,

(All)

2Xs+Xo+2X,o+Xu

= ay,

(A12)

az,

(A13)

X8+X9

""1

+X:1+-"6+

(A9)
(AIO)

2XS+X6+X?

where v, 10, X, Y and z are the number of atoms of A, C, H, 0 and N corresponding to

1 11101eof fuel plus the ail', I mole of product. a.risos from a mole of fuel plus IN stoichiometric quantity of air.
The procedure of solution of rhese 13 equal ions is based on Vickland 8 modified to
reduce computational t.ime.
The calculation starts with the preparation of the equilibrium constants in terms of
](0' Kb, K.c, Kd, Kt, K, and Hg
For one particular state (p, '1') these are fixed. Values of
a and bare selected and all tho values of X', are evaluated and the equations (AS) and
(Al:2) are checked wit.h theso calculated values for balance. If they do not balance within
a stipulated accuracy a Newtorr-Raphsou
adjustment is made to a and b and calculations
repeated.
From a large number of calculat.ions the following are suitable values for a and b for
the first estimate

R. S. BENSON, \V. J. D. ANNAND and P. C. BARUAH

122
Falue of a
When

rp ~

1'0,
13
a = [en + 0'5hy

+ l'863(2en + 05hy

- ox)!4>] exp (013T/1000)

(AI4)

rp < 10,

and when

13
a = [0'25hy+

2'363(2ca+

+ 0'50x]

0'5hy-ox)Jrp

exp (0,131'/1000)

(AI5)

Value oj b
'When temperature

7' (in OK) > 3000

(A16)

b = exp(10,3-(3'1-0'1710gP)7'/1000)
and when 1'.,; 3000
b = exp ( - 9'0

+ 0'5

log P

+ 30000/7').

(AI7)

Adjust b for the following ease

BX

= 20 - 9'0 log.p.

If BX>35,
B = exp (exp (3'5)

+ 0'25!og

(AIS)

P)

If BX.,;35,
B = exp [exp (20-9'0 log

rp) + 025 log P]

(A19)

If b B, then use
(A20)

b = B,

where ea is the number of carbon atoms in the fuel, hy is the number of hydrogen atoms
in the fuel, ox is the number of oxygen atoms in the fuel, .p is the equivalence ratio of
fuel (actual fuel-ail' ratio/theoretical
fuel-air ratio), P is the pressure in atmospheres and
7' is the temperature in OK.
Having fixed the values of a and b, X2 can be calculated by using the equilibrium
constants of equations (25) and (29) in equation (All) which will be
2bX2+2X2+K.b~X2+Ka~X2

= ax,

(A21)

Then follow the calculat.ion of Xl> X3 and X . From (27) and (31) inserted in (AI3), we
have
(A22)
whence Xs followed by Xs and X7

A combination

of equations

(30) and (AlO) gives

(A23)

whence Xg, XD and XlO are evaluated from equation (28) and Xll from equation (26).
Xl2 is given directly by equa.tion (A9). These values are now tested in equations (AS) and
(A12), neither of which has so far been used. If they do not balance within a stipulated
accuracy the calculations are repeated with new values of a and b adjusted by the NewtonRaphson method, in the following procedure.
Equation (AS) is written as
(A24)
Equation

(AI2) is written as
er =

The right-hand
write

Xl +X3+ 2Xs +X6 + 2X10 +Xll +Xo

y

sides of the above equations

--a.

(A25)

are functions of a and b only, therefore,

we

er = Pta, b),

(A26)

es = G(a, b).

(A27)

A model for spark ignition engine with intake and exhaust systems

123

The required solution is obtained when

The solution starts with approximate

er = 0,

(A28)

es = O.

(A29)

values of er and es where

bn),

(A30)

es = G(an, bn).

(A3I)

er = F(an,

The new values of the variables are adjusted

for the next step by

(A32)
(A33)

i)

where
D.a _
-

7)

er (G/ob) (es) -es (G/ob) (er)

J(an, bn)

t,b = er - (%a)
(%b)

(er) t,a
(er)

and

8)

J(a,,, bn) = oa (er) ob (es) - ob (er) oa (es).

The above procedure will be repeated for new values of

within stipulated precision is obtained.

9)

an+!

and bn+! until the solution

0)

ns
of
~d

APPENDIX
RATE

im

FORMATION

II

FOR

NITRIC

OXIDE

The rate formation for nitric oxide given by equation (43) is derived as follows.
Let Kir denote the forward rate constant for the ith reaction, Kib the backward rate
constunt for t.he ith react.ion and R, the "one-way" equilibrium rate for the ith reaction.
Thon

~ I)

we
22)

where the suffix e represents equilibrium

23)

and V is the volume of burned gas.

of (I/V) (d/dt) NO) V)

From equation (36) the net rate is

Calculation

~6).
md
ted

-KI/N)

(NO)+Klb(N2)

(0) = -cx,{3, K1r(N),

(NO),+K1b(N2),

(0),.

But

m-

Klr(N).

(0), = R"

(NO), = K,b(N2).

so that the net rate becomes - ex,(3. RI + RI'

Using the similar terms for equations (37), (38) and (41), involving NO, we finally have

.24)

I cl

TT

rlt NO)

V)

of (I/V)

(d/dt) N) V)

-ex.({3. RI +R2 +R3\-

2cx.R.) +Rl +j3,(Rz +Rs) + 2y,

e;

(A34)

.25)
Calculation

we

Considering equations

(36)-(38)

and using similar procedure we have

l26)
~ ~ N) V)

~27)

,~

_.

"

,-......,.

-(3.(CY., RI +R2 +Rs) +R, +cxe(Rz +Rs).

-~

-.

(A35)

--

'

'"

.'"Ir

124

R. S.

Calculation

BENSON,

W. J. D.

ANNAND

and P. C.

BARUAH

oj (1/V) (d/dt) ((N20) V)

Considering equations

(39)-(42), we have
F) = -y,(R4+R5+RS+R7)+R4+R,+a;R6+R7'

-VI~(("N20)
Clt

(A36)

It has been found? t.hat the relaxation times for equation (A35) and equation (A:36)
arc several orders of magnitude shorter than that of equation (A34). Therefore, steady
stn,te Inl1y bo assumed for (N) and (N20) which moans tho right-hand sides of equations
(A35) and (A36) can be set equal to zero.
Then from equation (A35)

fJ,

= RL +cxe(R,z+Ra)

.
and from equation

(A37)

cx.R,+l1z+Ra

(A36)
y, =

Using these values of

fJ.

R 4 + R 6 + a; Ro + R,
R4+R,+R6+B,
.

and y. in equation

~.:!.((NO)V)=2(1-CX;)(
Velt

R,
l+cx,[Rd(RdRa)]

which is the final rate equation for NO.

(A3S)

(A34) we have

.+

R6
1+[Rs/(R4+R5+R7)]

),

(A39)