You are on page 1of 15

In-Cylinder Soot Deposition Rates Due to Thermophoresis in a Direct Injection Diesel

Engine
Author(s): Blake R. Suhre and David E. Foster
Source: SAE Transactions , 1992, Vol. 101, Section 3: JOURNAL OF ENGINES (1992), pp.
1648-1661
Published by: SAE International

Stable URL: https://www.jstor.org/stable/44611322

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

SAE International is collaborating with JSTOR to digitize, preserve and extend access to SAE
Transactions

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
921629

In-Cylinder Soot Deposition Rates Due


to Thermophoresis in a Direct
Injection Diesel Engine
Blake R. Suhre
Chrysler Motors Corp.

David E. Foster
University of Wisconsin-Madison

ABSTRACT thermophoresis is a main contributor to the soot deposits on


the in-cylinder surfaces.
An investigation of the mechanism causing in-
cylinder soot deposition in a direct injection diesel engineINTRODUCTION
was
carried out. First, an analytical study was undertaken to
determine which of following possible deposition The 1994 heavy-duty diesel emission standards w
mechanisms, thermophoresis, Brownian diffusion, turbulent require an extremely low particulate emission level (0.
diffusion, inertial impingement, or electrophoresis were gr/kWh). Looking beyond 1994, all indications are that
responsible for the deposition of the soot on the combustionrequirements will get even tighter; especially in large uib
chamber walls. Based on a series of numerical models areas. Many think meeting these standards without the u
comparing each mechanism under conditions typical of diesel an after-treatment device will be impossible. The problem
engine combustion, thermophoresis was singled out as thethat these devices are expensive and may be unreliab
most likely cause of in-cylinder soot deposition. Kittelson et al. (1990), Needham et al. (1989). Thus the
Second, an experiment was performed to test the a need to understand the in-cylinder processes resultin
hypothesis that the soot deposition was caused bysoot emissions so that an emission control method other than
thermophoresis. An optical probe wás designed to fit anafter-treatment may be devised.
access port in the cylinder head of a Cummins NH250 single In a recent paper by Kittelson, et al. (1990) a theory
cylinder test engine. The probe was designed such that awas developed to explain some recent findings in the area of
sapphire optical access window was positioned flush with the diesel particulate emissions. Kittelson asserts that a growing
combustion chamber surface when the probe was installed.body of evidence points to in-cylinder surfaces as one of the
The probe incorporated the capability of changing themajor factors contributing to particulate emissions. Recent
temperature of the window surface approximately 100°C experimental results such as Kittelson et al. (1990) and Du
when the engine was operating at: 1300 rpm, an equivalence and Kittelson (1983), were interpreted as soot first being
ratio of 0.3, and an intake pressure of 27¡psi. deposited on the in-cylinder walls during the combustion
A thermocouple was placed near the window to process and then being reentrained during exhaust blowdown.
measure the temperature. A fiber optic cable carried the Kittelson further claims that the principal mechanism driving
combustion radiation signal, transmittedlthrough the window,the deposition of the soot on the combustion chamber walls is
to a photodiode detector. The photodiode signal wasthermophoresis, while the subsequent reentrainment is due to
amplified and then recorded by a high speed data acquisition the high shear forces generated by the high fluid velocities
system. In the testing sequence, the engine was first motored present during exhaust blowdown. The results of his study
at 1300 rpm and then the fuel was turned on. The radiation suggest that between 20% and 45% of the total soot emitted
signal was then recorded for the next 1600 cycles and wasby the IDI diesel test engine was due to thermophoretic soot
observed to attenuate as soot accumulated on the window.deposition and subsequent reentrainment.
The transmittance of the soot layer on the window, x = I¿710 This theory suggests that a significant proportion of
was calculated by assigning the initial peak voltage radiation the soot emitted by a diesel engine may be protected from
signal recorded to I0 and the signal measured subsequent to oxidation under the thermal boundary layer. This could have
this to if. Using light scattering theory, the transmittance significant ramifications for some proposed methods of soot
was then applied to the calculation of the mass deposited on emission control. If a portion of the soot particles emitted by
the window. The results from the experimental study show a diesel engine are protected from oxidation, proposed
methods of controlling soot emissions through enhancing the
that the soot deposition rate was, on average, 46.9% higher
when the window was cooled versus uncooled. These results oxidation of gas phase soot during the expansion stroke may
be ineffective.
were consistant with the earlier predictions that
1648

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
Two main objectives were sought in this study. If this is the case, the concentration gradient exists entirely
First, an attempt was made to substantiate, analytically, within the laminar sublayer and thus Brownian diffusion
Kittelson's claim that thermophoresis is the predominant could only take place here.
mechanism in the in-cylinder soot deposition process. The
thermophoresis model used in this study employs some TURBULENT DIFFUSION - Turbulent Diffusion,
assumptions which differ from those made by Kittelson in his which is convection in a turbulent field, also follows Fick's
model. These differences will be demonstrated in a later law and is usually presented as an enhancement to Brownian
section. Second, an experiment was devised to test directly, diffusion:
in-cylinder, the results of the analytical study (i.e. is
thermophoresis the cause of in-cylinder soot deposition). It is
hoped that the results of this research will provide a more m . ^ . dc
complete and precise knowledge of the in-cylinder soot - = (D . ^ + £)- . (2)
deposition process.
A dy

where
POSSIBLE DEPOSITION MECHANISMS
£= turbulent "eddy" diffusion coefficient
THERMOPHORESIS - Thermophoresis is the
mechanism that causes particles, suspended in a gas, toIntravel
diesel combustion, turbulent diffusion would take place
during the early stages when soot is being produced at the
down a temperature gradient. This phenomena occurs
turbulent
because the more energetic gas molecules on the hot side, on flame front. Concentration gradients exist between
average, transfer more momentum to the particle than doregions and the rest of the chamber. Because of the
the flame
those on the cold side. Thus the net momentum transferhighly turbulent nature of the combustion, the turbulent
causes the particle to travel toward the cold side of the coefficient is much larger than the Brownian
diffusion
temperature gradient. This is precisely the situation diffusion
that coefficient. Turbulent diffusion should quickly
exists in the combustion chamber of a diesel engine. Becausethe soot to the laminar sublayer, at which point some
spread
the combustion chamber walls of a diesel engine other mass transfer mechanism would have to take over to
stay
deliver
relatively cool, typically 400 K to 500 K, while the bulk the soot to the wall. Thus while turbulent diffusion is
combustion gas temperature can be very hot, in some probably
cases the predominant soot transport mechanism up to the
laminar sublayer, it can not cause the soot to be deposited on
greater than 2500 K, a thermal boundary layer exists between
the walls.
the walls and the bulk gases. During combustion soot is
produced and convected throughout the core combustion
gases by the turbulent motion. As the soot comes in contact INERTIAL DEPOSITION - If the soot particles have
a high
with the thermal boundary layer it may be transported downenough mass to aerodynamic drag ratio (i.e. are large
the temperature gradient to the wall. enough) it would be possible for them to be thrown out of a
turbulent eddy and travel to the wall. On the other hand, if
BROWNIAN DIFFUSION - Brownian diffusion the mass to drag ratio is small (i.e. the particles are small) the
occurs when a suspended particle concentration gradient particles will follow the flow very closely and will not be able
exists in a gas. Brownian diffusion follows Fick's law which to break away from an eddy. A mathematical model was
states that the mass flux of a constituent per unit area is written, that will be presented later, to determine which is the
proportional to the concentration gradient as shown by case for typical soot particles produced in the combustion
Holman (1986): chamber.

ELECTROPHORESIS - Electrophoresis is the


m ^ dc process by which a particle carrying an unbalanced charge is

A~ dy (1) transported in an electric field. This process would require


that the particles acquire an uneven charge when or soon after
they are formed and that the combustion chamber walls cany
where
a static charge distribution. The soot particles might
ňl = mass flux per unit time [kg/s] conceivably be charged, for example, through a thermal
A = area [m^] ionization process. Again a model was constructed to test the
D = diffusion coefficient [m^/s] feasibility of this process occurring during diesel combustion.
c = mass concentration of suspended particulate
per unit volume [kg/nP] GRAVITATIONAL SEDIMENTATION - The final
y = length [m] aerosol mass transport mechanism which is sometimes
mentioned is gravitational sedimentation. For the case of
Assuming a turbulent hydrodynamic boundary layer exists diesel combustion it seems highly unlikely that this
over the surface of the combustion chamber, Brownian mechanism would have a significant effect on soot deposition.
diffusion would take place only in the laminar sublayer. First, because the particles are small, the turbulent fluid
Davies (1966) claims that turbulent convection causes the motion will overwhelm any effect due to gravity. Second, the
soot concentration to be uniform up to the laminar sublayer. sedimentation time for particles of this size and density tends
1649

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
to be on the order of minutes or even hours. Thus the time
scales involved in diesel combustion are too short. For these
reasons, gravitational sedimentation was not taken into
account in this study.

PROPOSED STUDY

MODELS - Before any experiments were run,


several models were constructed. An equation solver
developed at the University of Wisconsin-Madison by S. A.
Klein (1991) was used. This package can solve, not only
simultaneous algebraic equations, but also differential
equations.
The models were constructed to determine the
relative contributions of the possible mechanisms of in-
cylinder soot deposition in an attempt to verify Kittelson's
conclusions. From a large group of possible aerosol
deposition mechanisms, four were chosen for comparison:
thermophoresis, Brownian diflusion, electrophoresis, and
inertial impingement. From this group thermophoresis was
chosen as the only significant contributor. The
thermophoresis model was then compared, on an order of
magnitude basis, with the experimental results as a means of
validating the thermophoretic deposition theory.

EXPERIMENT - The purpose of the experimental


part of this study was to measure, directly, in-cylinder the
change in soot deposition rate due to a surface temperature
change. If the observed trends in deposition rate change were
in agreement with that predicted by the model, this would be
an indication that thermophoresis was the principal cause of
in-cylinder soot deposition. To accomplish this an engine
probe was built to allow optical access to the combustion Fig. 1. Probe
chamber. The probe utilized a flush mounted sapphire
window whose surface temperature could be controlled. The
probe is shown in figure 1. The window surface temperature /- Unu»«d Accms Port

was controlled by circulating water through the water jacket


shown in the diagram. The probe was positioned in the
cylinder head of the Cummins NH250 test engine as shown in
figure 2. ^^^^--Proa«u ro^Tron^ucor
The general procedure for testing was to install the /y >
/'' s
/' S 1' '^qlve
^ vi/ I ' ' IA.
' ''
clean probe and hook up all related plumbing, optics, and /' ' ^ A I ^ x 1 ' > I '
/ / / I 1 i i I X ' N ' '
data acquisition equipment. The engine was then brought to /
/
I
'
/
'
/
. ^ 'It ' . ' ' bottom
/ ' * ž* . ' ' 1 v / . ' ' v / '
speed by the dynamometer. After the engine was motoring at /
/ * f /
' f ž* i' 'N✓' 1
' / x /
' / ' v ^ / ģ
the desired speed, the fuel was turned on. The fuel flow rate '/
K
w
y'
w
'
»/
>
' ¡ -v v ^ ' / v ^ '
was kept as constant as possible to provide an equivalence Prob« Body - s. ' v

ratio of approximately <ļ> = 0.3. The attenuation of the End Cap- '
window - Intake^/
radiation signal was recorded as the window sooted up. The
rate of signal attenuation, in conjunction with light scattering ' 7NO<tv V v ' , /N^ s /'' /! '' ^
' ' 'vyCrjLA x v , i i 'i / * /
theoiy, was used to calculate an in-cylinder soot deposition ' I I Pfîu I I i i _ * -/ I /
' i * vArJ ' ' * _ * iV ' /
rate. ' i v v vArJ VisL/ J ' m iV L'' i /
- - - < bo wi.'c - ' xy/
B¿re

Fig. 2. Cylinder Head with Probe

1650

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
BASIC SOOT DEPOSITION MODEL diffusion and thermophoresis, an equation for particle
velocity (V) is obtained from the literature, Davies (1966) an
Kittelson
Convection (or, for turbulent flow, eddy diffusion) is et al. (1990). Thus all that is required is a soluti
simply the transport of particles with the flow of thetohost
a first order ordinary differential equation ( dt=l/V dy ).
fluid; in this case the combustion gases. Since the particlesIn the case of inertial deposition, a particle stopping
are so small, tens to hundreds of nanometers in diameter, they from a probable maximum in-cylinder velocity wa
distance
follow the flow very closely. Thus convection should be the
computed. Based on this result, a judgment was made on th
predominant mechanism of mass transfer, in the combustion
likelihood of inertia playing a part in soot deposition. T
chamber, up to a point very close to the wall. Very near the also provided an indication of how closely the particl
result
follow the flow.
wall, assuming a no slip condition exists, the hydrodynamic
boundary layer dictates that fluid velocities become very slow Finally, electrophoresis was considered. In the cas
and the velocity vector must point in a direction tangent to
of electrophoresis little has been mentioned in the literatur
the wall surface. From this point down to the wall surface,
about its possible effect on soot deposition. This may be du
convection can no longer be the predominant mechanism of fact that while the typical amount of charge carried by
to the
mass transport. Therefore, one or more of the other soot particle has been well documented by Kittelson, Pui an
mechanisms mentioned above, Brownian diffusion, inertial Moon (1986), Kittelson and Collings (1987), Collings et
impingement, electrophoresis, or thermophoresis, must be (1986), and Schweimer (1986), very little has been
responsible. documented about the possibility of an in-cylinder elect
From this discussion the basic overall process of in- field being set up during combustion. Here again
cylinder soot deposition begins to emerge. First, during calculation was made simply to get a qualitative assessme
combustion, soot is formed in the fuel rich regions of the of the magnitude of this mechanism. The "point" charge at
cylinder. This soot is convected by the highly turbulent flow the wall surface and the electric field strength required
in the "core" region (the region excluding the hydrodynamic produce a force acting on the particle equivalent to that
boundary layer). As the soot is churned and mixed in the thermophoresis was calculated. The values obtained for t
core region a portion of it will be transported into the equivalent charge and electric field strength should then te
turbulent hydrodynamic boundary layer where it may be us if electrophoretic deposition is likely. If it is not, th
"caught" in the laminar sublayer beneath the surface. mechanism can be neglected; but if on the other hand t
According to Davies (1966), if a steady turbulent numbers are reasonable, then a more sophisticated mod
boundary layer has developed, an aerosol will be transported must be constructed.
to the laminar sublayer such that the aerosol concentration is
uniform up to that point. In a diesel engine, time scales THERMOPHORESIS MODEL
dictate that a steady turbulent boundary layer probably will
not be formed and thus the concentration may not be uniform, The thermophoresis model is an assessment of t
but nonetheless turbulent diffusion is still the mechanism time it would take an individual particle to travel to
which transports the soot to the laminar sublayer. At this combustion chamber wall from the edge of the lami
point one or more of the mechanisms mentioned above will be sublayer (y+=5). It is assumed that the particles
responsible for the subsequent journey to the wall surface. convected up to the laminar sublayer by turbulent diffus
Thus it remains to be determined which process, or This model also assumes that there is no tangential par
combination of processes, mentioned above is responsible for velocity and that thermophoresis is the only mass transp
the transport of the soot over the final distance to the wall. mechanism acting on the soot particle. Other simplify
assumptions are: that the soot particles are spherical, tha
DETERMINATION OF THE PREDOMINANT SOOT combustion gases are air and obey the ideal gas law, and t
DEPOSITION MECHANISM
the temperature profile is represented by the cubic polyno
relationship normally reserved for laminar boundary laye
The objective of the analytical part of this research
These assumptions were made to simplify the computatio
was to compare the relative magnitudes of each of the fourand facilitate comparison with the other models.
mass transport processes to assess which is most likely The governing equations involve, first, the relation
responsible for in-cylinder soot deposition in diesel engines.
for thermophoretic force and velocity. These were derived
To accomplish this, simple mathematical models were Talbot et al. (1980) and used by Kittelson et al. (1990);
developed. First, thermophoresis and Brownian diffusionare valid for the entire range of soot particle sizes comm
were compared. A comparison was made based on the time observed. The equation for the velocity is derived by sim
(/) required for a particle to travel the length (y) ofequating
the the Stokes-Cunningham drag force (equation 3) w
laminar sublayer under typical combustion temperatures the
andthermophoretic force (equation 4).
pressures. The laminar sublayer thickness was calculated
based on a specified thermal boundary layer thickness and
mean fluid tangential velocity. The thermal boundary layer
thickness was assigned an arbitrary value in the range of P

thicknesses cited in the available literature, Lucht and Maris K

(1987), Lucht et al. (1991), Farrel and Verhoeven (1987),


Boggs and Borman. (1991). In both the case of the Brownian
1651

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
this assumption can be made with impunity. The diff
in deposition time calculated by the simplified model i
less than 0.01 percent compared to the exact solution.
k The equations used for thermophoretic forc
'2npvRC + CKn ultimately thermophoretic velocity were derived f
K (1 'dT exact formulas for both the continuum regime (K
F»= Brock (1962) and free path regime (Ã>i»l) by Wal
(1 + 3 CmKn) 1 + 2 - + 2CtKn and Schmidtt (1966). Since soot particle sizes can fa
L p intermediate region (AT«®1), the equation used
approximation that covers the entire range of po
where Knudsen number. Despite the use of this interp
p = dynamic viscosity of the gas equation, the results are never in error by more than
V = soot particle velocity percent which Kittelson et al. (1990) deemed suff
accurate.
R = soot particle radius
p = gas density The ensuing set of equations make up the turbulent
boundary layer part of the model, as described by Davies
Ru = ideal gas constant
A =1.20 (empirical constant) (1966). The first of the turbulence equations defines the
B =0.41 (empirical constant) nondimensionalized boundary layer length scale:
C = 0.88 (empirical constant)
v = kinetic viscosity of gas
y u,
kg = thermal conductivity of the gas y+~ -
y
kp = thermal conductivity of the particle
Cs = 1.17 (empirical constant)
Cf =2.18 (empirical constant) A value of _y+=5.0 defines the laminar sublayer thickness.
Cm = 1.14 (empirical constant) The next equation defines the friction velocity, ut:
T = gas temperature
normal to surface
Cc = Cunningham slip correction
= 1 +Kn(A+Be-c/Kn)

dT _ gas temperature gradient in the direction


dy The following equation calculates the wall shear stress and is
based on the pipe flow approximation:
Kn = Knudson Number = A
R

1 f -!
X = gas mean free path = Aí I n V-" f -!
PÍ%RJ
The smooth pipe friction factor, f is based on values taken
from the moody diagram. An equation for the Reynolds
The result of equating equations 3 and 4 is the number, also based on pipe flow, is used:
thermophoretic velocity, Vty, equation 5:

„ Bore ū
„ Re =

This is the famil


2 C,Ccv^- + CtKn , s point out that wh
kP (^MT , s not the same as th
is one of the fe
(1 + 3 CmKn) 1 + 2-^- + 2 CtKn available. Thus, w
L p should give values
all that was requ
time (0 the follow
Inherent in this equation is the assumption that the particle is
instantaneously accelerated to its terminal velocity. This is
obviously not possible, but the time required for the particle to
accelerate to terminal velocity is extremely small such that
1652

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
defines the pa
profile is defin
f° 1 »
t = - dy »
<v
2
which is equivalent to yu.
u =

f° ^ j. The foll
t = ] Jy*a*
Jy*a*f° ay
-r-4y j. velocity
F+, calcu

BROWNIAN DIFFUSION MODEL

As in the previous section, this model computes the


rM
M,
soot particle travel time in the laminar sublayer. The model
calculates the time required for a soot particle, starting at the
Finally, the differential equation shown next is solved:
surface of the laminar sublayer, to travel to the combustion
chamber wall if Brownian diffusion were the only mass
transport mechanism. This arrangement allows the diffusion
model to be compared directly to the thermophoresis model V =±
discussed previously. The same simplifying assumptions dt
made in the thermophoresis model are made here.
The governing equations for this model center and was done in the following fashion:
around the dimensionless particle diffusion velocity derived
by Davies (1966):

< JyaaH=y®y*=s
= f -dy dy

'dT ELECTROPHORESIS MODEL

K = - - - -Lxj - -
Recent work by Kittelson et al. (1986), Kittelso
14.5
14.5 . .. .. .
-In 1. ,
-In (l +-t=
+ ¿) 1, ,tan_/ -£=- -£=-2j-i) n
+- Collings
j=
al. (1987), Schweimer (1986), et al. (1986) in
6 , y¡3 { S J 6^3 j=
field of diesel engine particulates has shown soot particl
carry charges ranging from one to five units of elemen
with
charge per particle. Schweimer and Collings observed
positive charge on the particles, whereas Kittelson found
charges on each particle to be bipolar and net neu
Nevertheless, the effectiveness of the ion probes of Schw
and Collings suggest that, at least under some circumsta
a net positive charge might exist on the soot partic
2.9ļļ%
<i>= - 2.9ļļ% l-¡=
Because of this, it was felt necessary to calculate
magnitude of the electric field that would be requir
<7>
transport the charged particles to the surface in an amou
time comparable to that of thermophoresis. The
U * = ^7p <8> transport process involved here is known as electrophores
The electric field strength required to produ
force on the particle equivalent to the thermophoretic f
was calculated. The governing equation consists o
u ¡ is known as the friction velocity and is due to Von Karman
definition of an electric field:
(1930). The next equation, from Rosner (1986),

kT F
D =
£ = -Ł (9)
6njuRCc 9,

where E is the required electric field strength, F ¡y is the


equivalent thermophoretic force, and qp is the total charge on
1653

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
the particle. The results from a run using a thermophoretic
force of 1.0x10*1% (calculated with the thermophoresis r,dV
model), three charges per particle, and a distance from the F"=mV~r <13>
dy
wall of 6.5xl0"3 m, result in a required electric field strength
of E = 208,047.7 V/m. This shows that an extremely large
electric field would be required to generate a force equivalent with V = V¡ at y - ystart. Here F¿ is the force due to
to the thermophoretic force. To put this in perspective, the aerodynamic drag (Stoke's formula), m is the particle mass,
enormous electric field required to cause a spark to jump the and VdV/dy is simply the particle acceleration. Thus equation
13 is a form of Newton's second law. The solution to this
gap of a spark plug is only on the order of 50 times greater. It
then seems highly unlikely that electrophoresis has much of initial value problem is trivial if a simplifying assumption is
an impact on soot transport to the combustion chamber walls. made; that the dynamic viscosity, //, and the Cunningham
slip correction , Cc, are constant over the length of the
INERTIAL DEPOSITION MODEL particle travel from start to stop. In other words,
jł=/Ąstaft)=Constant and Cc=Cc(start)=Constant. This
This model computes the stopping distance for assumption
a soot is valid, it turns out, because the soot particles
travel
particle that is thrown into the laminar sublayer at only
a a small distance before stopping and thus the
change in ļi and Cc are negligible. To prove this a model
predetermined velocity. The only force acting on the particle,
usingisthe analytical solution, equation 12, was compared to
in this model, is aerodynamic drag. The aerodynamic drag
another model which solved the differential equation 13
computed with the use of the Stoke's drag equation modified
by the Cunningham slip correction for small particles. numerically. As expected the difference turned out to be
negligible.
The model first calculates the velocity component
normal to the wall based on a user input value for the swirl Using the same input values as in the previous
ratio, Rs, Heywood (1988); in this case Rs = 1.0. models
Wethe following values for the stopping position, y, and
distance
assumed the normal velocity was obtained from the following traveled, DistTrav, were determined.
equation and is based on our own assumptions:
y = 6.59 X 10"5 m
DistTrav = 3.3 x 10"^ m

F,=py^û>sin0+r (10) Next several different wall temperatures were investigated,


while the other variables were kept constant, and the results
are shown in the table below.
where a is the combustion gas angular velocity, 6 is the angle
the gas velocity vector makes with the wall and F is a
turbulent fluctuation velocity component in the direction
Wall Temp. (K)
normal to the wall. The V component is determined with the
use of the following equation:

800

Table 1. Temperature Dep


V'=V" (Stroke)^ (11)
These results show that
V" is the turbulent fluctuation velocity component as a
short distance in the lam
fraction of the mean piston speed and N is the engine speed in
that wall temperature h
RPM and stroke is the piston travel. Thus equation 8 particle will travel.
determines the particle velocity component normal to the wall Thus it seems improba
and contains two components; the first is swirl and the second
is a significant compon
turbulence. After the normal velocity component is
unless the laminar sublaye
calculated, an equation, which relates the particle position, y,
30 nm or less. This an
with the particle velocity, V, is used to determine the particle
expected for particles of
stopping distance: follow the combustion gas

CONCLUSIONS DRAWN FROM THE MODELS


m Cc V m Cc V,
y
o n ļiR 6 n n K
= Based on the results of the electrophoresis model, i
seems unlikely that this mechanism is a contributor to
cylinder soot deposition. Inertial impingement could b
factor if the laminar sublayer were extremely thin and/or
magnitude of the wall roughness were such that the parti
This equation is a solution to the following initial value
could be caught on these structures. Work carried out
problem: 1654

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
Boggs et al. (1991) indicates that this may occur in some Cylinder
cases. But the experimental results, which will be discussed Bore 139.7 mm (5.5 in)
later, indicate a strong dependence of the deposition rate on Stroke 152.4 mm (6.0 in)
wall temperature. The table of results from the inertial Compression Ratio 13.23
model, table 1, show very little temperature dependence. This Displacement 2.33 L (142 in^)
indicates that inertial impingement is probably not a major Fuel Injector
factor in in-cylinder soot deposition. Number of Orifices 8
With electrophoresis and inertial impingement Orifice Diameter 0.2 mm (0.008 in)
eliminated, thermophoresis and Brownian diffusion remain Spray Angle From Head 18°
the two possible means of soot deposition. The following
table (2) compares these two mechanisms at three points
during combustion: The first shortly after the start of Table 3. En
combustion, the second approximately midway through
combustion and the last shortly before the end of combustion. Cetane Number 48
Sulfur (% by weight) 0.03
Crank Temp. Pressure Deposition Deposition H/C Ratio 1.83
Angle (K) (kPa) Time Time API Gravity at 298 K
ATDC Thermo- Brownian
Comp. phoresis Diffusion Table 4. Fuel Characteristics
(deg)

-10 1740.53 6346.85 0.0399 8.702 RADIATION PROBE


4 2478.13 8968.47 0.0378 8.864
18 I 2376.22 1 7360 82 | 0.0302 1 8.743
A cross sectional view of the probe assembly is
shown in figure 1. The main functions of the probe are
position
Table 2. Comparison of Thermophoresis andthe sapphire window flush with the surface of the
Brownian
Diffusion combustion chamber and to allow for a significant change i
the window surface temperature. Two holes were drilled
RESULTS OF BASIC ANALYTICAL MODEL the probe to allow for positioning a pair of thermocoup
near the window surface. The probe also allows for th
From the comparison of the results obtained from thepositioning of a fiber optic bundle directly behind the
two models, it is obvious that thermophoresis is a more window. The probe end cap, the part surrounding the end o
significant factor than Brownian diffusion. Thermophoresisthe window which is exposed to the combustion, was ma
is shown to produce a deposition rate which is approximately from stainless steel because of its low thermal conductivity
three orders of magnitude higher than Brownian diffusion. This kept the temperature of the window surface as high a
Because of this result, the final analytical model used for the
possible when operating in the uncooled condition.
experimental study will consider thermophoresis to be the
MEASUREMENTS MADE
only deposition mechanism. Despite the difference in
approach from Kittelson et al. (1990), it is significant that the
same conclusion was drawn here. The engine probe, described in the previous section,
was used to measure the combustion radiation intensity at 65
ENGINE AND FUEL DELIVERY SYSTEM nm and the surface temperature near the window. A hig
speed PC based data acquisition system was used to record th
The engine that was used in the experiment is a radiation signal and probe surface temperature for 160
single cylinder version of the Cummins NH250, 6-cylinder, 4-consecutive cycles. The data were recorded at two cran
stroke, heavy-duty diesel engine. This engine has a quiescentangle degree intervals. This resolution was necessary to
combustion chamber of the Mexican hat type of bowl in
accommodate the relatively large number of cycles (160
recorded in an individual run.
piston design. The piston also has cutouts for clearance of the
four valves. The engine is of the direct-injection type and In addition to these measured parameters, several
uses a Cummins P-T injector. An Amoco low sulfur dieselother data were required for the analysis. Parameters such
flame temperatures and cylinder pressures have be
fuel was used for all of the tests run in this experiment. Table
3 contains a summary of the engine specifications and table measured
4 in this engine, under the same operating condition
lists the characteristics of the fuel used. by Fukuda et al. (1992).

DETERMINATION OF SOOT MASS DEPOSITION

The radiation signal data was used to calculate the


instantaneous mass of the soot on the window. To do this, the
transmittance of the soot layer on the window had to be
determined. The transmittance is the ratio of the radiation
1655

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
intensity, of the combustion, after passing through the soot I" L H
layer to the radiation intensity before passing through the soot
layer (I/- /I0). This is equivalent to the ratio of the peak
radiation intensity at a given sooted condition to the peak
radiation intensity when the window is clean. Since the
voltage signal measured is directly proportional to the
radiation intensity, the transmittance can be determined
Io # Io + dl
simply by the ratio of the photodiode voltages recorded at the
two conditions.
(ļx * NjCext
The equation used to determine the soot mass
deposited on the windows, mw, based on the transmittance of
This equation is applicable when the following criteria is met,
the soot layer is:
Bohren and Huffman (1983):

ln(r)ppA(/i4 +2n2k2 + 4n2 +k4 -4k2 +4) t (17)


tn

36mk
This constraint can be relaxed significantly if the contribution
where, to Cext by scattering is small; which is the case for soot
t = the soot layer transmittance particles.
Assuming Cext is constant over the entire path
pp = soot particle density
X = measured radiation wavelength length, L, the solution to equation 16 is:
n = real part of refraction index
k = imaginary part of refractive index
Aw = area of window

Also calculated was the mass per unit area, m, which


is simply the mass on the window, mw, divided by the which results in
window area, Aw:

(i '
ln(z)pJl(w4 +2n2k2 + 4n2 +k 4 -4k2 +4)
m = Ihvof = -NCmL
36 rnk

These equations were derived as follows. From at equivalently


Kerker (1969), the fundamental differential equation for the
attenuation due to extinction of a beam of light passing
through a dispersion of Rayleigh scatterers of equal size is:
T= ^ = exp (-NC^L)
■*0

-x=Arc-/
ax
"6> Substituting in the extinction efficiency, Qext = Cex/m2,
where a is the particle radius, the following expression is
where
obtained:

/ = intensity of light beam


N - particle number density t= expf-ß^ m2NL)
Cext = extinction cross section of the particles
= CSca + Cabs Then note:
X = distance traveled by the beam through the
particle cloud.

r= exp -Qtxt™ NLAW


V » >

Next substitute , n
particles on the win

1656

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
Next, 2710t//. is substituted for a:
(m 2 '
(m -Qextm
z- exp

a* j
^ T 'm2 - '}Zm
e-=-Imļ^Tī|- ^ T -
Now, note that the mass of soot on the window can be
Then, after expanding and reducing the imaginary terms:
expressed as, mw=ti(4/3wa^)pp. Then:

_ 48 mnk
^ _ " l(n4 +2 nV + 4»2 + k' - 4k
n= A (18)
- m2p Finally after substituting equation 23 into
3 p
solving for mw we arrive at equation 14:
Then substituting in the expression for x:

ln( % )p.À(n4 + 2n2k2 + 4 n2 + k4 - 4 k2


mw =

f-3 (I } 3>6m

X- exp (19)
t 4 KaPp J Note
dependent.
The resulting equation set allows the mass deposited
on the window to be calculated from measured or known
Next we must find an expression for Qext in terms of known
or measured quantities. From Kerker (1969) an expression values. For example, the transmittance, t, is calculated from
for the scattering and absorption efficiencies, assuming the radiation data taken in this experiment. The values for
Rayleigh scattering, is given: real and imaginary parts of the refractive index, n, k, are
based on the work of Lee and Tien (1981). The soot particle
density is based on the value obtained by Park and Appleton
(1973). The results from the use of this equation are shown
„ 8 4 iti2 - 1 2 in figures 3 and 4, and are discussed next.
(20)
SOOT DEPOSITION ANALYSIS

and
Figures 3 and 4 show an example of results from the
mass deposition calculations. The run numbers shown on the
graphs begin with either the letter "H" or "C." The H denote
the uncooled (hot) condition and C denotes a cooled run.
Qabs = -4 almi (21> Several values were required to calculate the mass.
l m +2J
First, was the soot particle density. A value of 1,800 kg/m3
where
was used; which was measured by Park and Appelton (1973).
Second, both the real and imaginary coefficients of th
a = dimensionless particle size parameter = liudX refractive index of the soot particles were needed. A value o
m = complex refractive index = n-ik.
1.90 was used for the real part and 0.45 for the imaginar
part (m=1.90-0.45i). These are values based on the work o
It can be seen that in the Rayleigh limit ( a < 0.3) Qscc/Qabs Lee and Tien (1981) and should be appropriate for th
«a3 « 1. Therefore, since Qext =Qsca +Qabs, Qext can temperature range measured near the window surface.
be approximated by Qabs. Then: Finally, a value for the radiation signal (in volts) produced b
the combustion when the window is clean (no soot deposited)
is required. This value is then used for I0 in the calculatio
of transmittance of the soot layer, x. The value used is on
(L * -4aīm|-^[m- +2
+2 Ji I (22) slightly higher than the highest peak radiation signa
recorded. The logic being that the window will hav
accumulated a small amount of soot as the fuel pressure
underwent its transient after the fuel pump was turned on.
Thus the radiation signal for the clean window would b
slightly higher than that indicated by the initial peak. Even i
the value chosen is incorrect only the absolute amount of soot
deposited will be predicted incorrectly. The calculated soo
1657

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
deposition rate, and thus the calculated change in deposition high and low surface temperatures are shown directly on the
rate between the two conditions, will not be affected. graphs (Figs. 3 and 4) and are summarized in table 3. As can
be seen, the correlation coefficients are very high, usually
well above 0.90, indicating that the data is indeed following a
linear relationship. The slope of the regression line
(parameter В on the plot) is the soot mass deposition rate in
units of kg/m^/cycle.
It is interesting to consider the reason for the linear
relationship between the deposited mass and the number of
cycles. If one considers the process for in-cylinder soot
deposition proposed by Kittelson et al. (1990), which states
that initially all the soot that is deposited stays through the
entire cycle, then, at least initially, a constant deposition rate
would be observed. This phenomena would result in the
linear relationship observed in the plots. It is also thought
that the amount of soot blowing off the window, during
blowdown, should increase over time as a steady-state level of
soot is reached. As can be seen in Fig. 4, the mass deposition
rate is starting to decrease by the later cycles. This may be a
result of soot starting to blow off the window.

Run Deposition Rate Change from run

C3-25 5.857 x 10"7


Fig. 3. Mass Deposition for Runs 3-26
H3-25 3.318 xlO'7 43.4 %
C3-26 1.078 x 10"6
H3-26 5.405 x 10"7 49.9 %
C3-30 1.082 x 10"0
H3-30 6.664 x 10"v 38.4 %
C3-31 1.818 x 10*6
H3-31 8.006 x 10"7 56.0 %
Average: С 1.141 x 10"6
H 5.848 x 10"7 46.9 %
Std. Deviation: С 5.080 x 10"7
H 1.993 x IO"7 7.67%

Table 5. Experimental Deposition Rate Result

As can be seen in table 5, a fair amount of v


in both the deposition rate and the percentag
deposition rate is observed. This is most lik
changes in operating conditions from one run
Despite the efforts made, it is difficult to ass
Fig. 4. Mass Deposition for Runs 3-3 1 equivalence ratio will be the same from one run t
because of the limitations of the fuel delivery sy
Equations 14 and 13 are only applicable when the The fuel lines between the pump and injector are
criteria NCextL« 1 is met but, may be relaxed in this case and could have been causing pressure fluctua
since Csca is proportionally small. Basically, this means that injector under these transient startup conditio
the soot loading on the window must be small. Because of Cummins PT type injector does not allow for a
this, a linear regression analysis was performed on only the manually controlling the injection event. Thu
first 600 cycles beyond the point at which the peak radiation control over fuel delivery might reduce the varia
value occurred. It should be noted that, in general, increasing the experiments.
or decreasing the number of cycles used did not have a
significant effect on the slope of the regression line. The
results of the regression analysis for two different runs at the
1658

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
PROPORTION OF EMITTED SOOT FROM IN- This result is surprising. The dep
CYLINDER SURFACES when integrated over the entire
higher, in the cooled condition
Kittelson et al. (1990) makes the claim that 20 % uncooled
to condition), than that
45% of the soot emitted by their test engine had come exhaust.
from One reason for this coul
the in-cylinder surfaces during the exhaust blowdown period.
combustion chamber has a soot de
Using the same assumptions made by Kittelson, that higher
the than the average depos
initial deposition rate to a clean surface gives achamber. good This might well be the
indication of the rate of soot being blown from the wallsone
once of the injector's eight orifi
a steady-state soot layer has been formed, one can calculate
window (see figure 2). Another ex
the maximum portion of soot being emitted that could soothave deposition rate to the wall
come from the walls. time. Perhaps the soot layer insu
To simplify the calculations, the in-cylinder head
that the surface temperature rise
and piston surfaces were considered flat. Also, the headthe
and thermophoretic effect to be re
piston crown areas were the only surfaces considered. The
exposed cylinder wall area, dining the period when CONCLUSIONS
soot is
most likely to be present, is small and was not considered.
Thus the area exposed to deposition was calculated as Based on the results of this study, it seems evident
follows: that thermophoresis must play the major role in so
deposition to the in-cylinder surfaces of a Diesel engin
Inertial deposition may also play a part but it is the opinion
the authors that it must be small in relation to
Bore = 0.1397/w
thermophoresis. It is also interesting to note the results of
comparison of in-cylinder deposition rates to particu
Exposed Area = (0. 1 397)2 x 2 = 0. 03 'm2 emission rates. The high deposition rates observed certainl
do not contradict the theory that in-cylinder surfaces pla
major role in particulate emissions.
The average wall deposition rate recorded experimentally was
1.141 x 10-7 kg/nrf/cycle for the cooled window condition
ACKNOWLEDGMENT
and 5.848 x 10"7 kg/m^/cycle for the uncooled condition.
Under the same operating conditions used in this study,
We thankfully acknowledge the financial assistance
Fukuda et al. (1992) measured an exhaust particulate
provided through the Engine Research Center which
emission rate of 10.5 mg/m^ of exhaust gas expanded to room
funded by the Army Research Office as the Cente
temperature and pressure. The flow rate under these
operating conditions was measured and found to be 2.81 Excellence for Advanbed Propulsion Systems. We must
express our gratitude for the help provided by Dale R. Tree
kg/min. The density of air at room temperature is 1.1774
the area of light scattering theory as it applies to die
kg/nP, Holman (1986). Next the flow rate is expressed in
combustion.
terms of nP/cycle:

REFERENCES

2 .SI-*?- x - 1-^1 x-!- = 0.00184-^- Alkidas, A.C., "Heat Release Studies in a Divided-Chambe
min 1.1774% 1300 cycle cycle Diesel Engine," ASME Technical Paper No. 85-
DGP-22, 1985.

Alkidas, A.C., and R.M. Cole, "Thermal Loading of


Now the in-cylinder deposition rate can be compared to the
particulate emission rate: Divided-Chamber Diesel Engine," SAE Technica
Paper No. 831325, 1983.

Boggs, D., G.L. Borman, "Calculation of Heat Flux Integr


Length Scales from Spatially-resolved Surface
In-cylinder deposition rate= Temperature Measurements in an Engine," SAE
Technical Paper No. 910721, 1991.
1.141xl0-6-
m • cycle cycle Bohren, C.F., D.R. Huffman, Absorption and Scattering
Particulate Emission Rate = Light by Small Particles, Wiley-Interscience, New
York, 1983.

1 0. 5^f x 0. 001 84-^- = 1. 932x1 O"8


m cycle cycle Brock, J.R., "On the Theory of Thermal Forces Acting on
Aerosol Particles," Journal of Colloid Science, Vol.
1659 17, pp. 768-780, 1962.

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
CoUings, N., N. Baker, W.G. Wolber, "Real-time Smoke Kishi, Y., H. Tohno, and M. Ara, "Characteristics and
Sensor for Diesel Engines," SAE Paper No. 860157, Combustibility of Particulate Matter," SAE Paper
1986. No. 920687, 1992.

Crump, J.G., J.H. Seinfeld, "Turbulent Deposition and Kittelson, D.B., J.L. Ambs, and H. Hadjkacem, "Particulate
Gravitational Sedimentation of an Aerosol in a Emissions from Diesel Engines: Influence of In-
Vessel of Arbitrary Shape," Journal of Aerosol Cylinder Surface," SAE Paper No. 900645, 1990.
Science, Vol. 12, No. 5, pp. 405-415, 1981.
Kittelson, D.B., D.Y.H. Pui, K.C. Moon, "Electrostatic
Davies, C.N., "Deposition From Moving Aerosols," Ch. XII Collection of Diesel Particles," SAE Paper No.
in Aerosol Science, C.N. Davies (Ed.), Academic 860009, 1986.
Press, London, 1966.
Kittelson, D.B., N. Codings, "Origin of the Response of
Doeblelin, E.O., Measurement Systems, Applications and Electrostatic Particle Probes," SAE Paper No.
Design, McGraw-Hill, New York, 1990. 870476, 1987.

Du, C.J., D.B.Kittelson, "Total Cylinder Sampling From a Klein, S.A. EES Manual, 1991.
Diesel Engine: Part III - Particle Measurements,"
SAE Paper No. 830243, 1983. Lee, S.C., and C.L. Tien, "Optical Constants of Soot In
Hydrocarbon Flames," Eighteenth Symposium
Eisner, A.D., D.E. Rosner, "Experimental Studies of Soot (International) on Combustion, The Combustion
Particle Thermophoresis in Nonisothermal Institute, 1981.
Combustion Gasses Using Thermocouple Response
Techniques," Vol. 61, pp. 153-166, 1985. Lucht, R.P., M.A. Maris, "CARS Measurement of
Temperature Profiles Near a Wall in an Internal
Epstein, P.S., "On the Resistance Experienced by Spheres in Combustion Engine," SAE Paper No. 870459, 1987.
Their Motion Through Gasses," Physical Review,
Vol. 23, Ser. 2, pp. 710-733, 1924. Lucht, R.P., D. Dunn-Rankin, T. Walter, T. Dreier, S.C.
Bopp, "Heat Transfer in Engines: Comparison of
Farrell, P.V., D.D. Verhoeven, "Heat Transfer Measurements CARS Thermal Boundary Layer Measurements and
in a Motored Engine Using Speckle Interferometry," Heat Flux Measurements," SAE Paper No. 910722,
SAE Paper No. 870456, 1987. 1991.

Ferguson, C.R., D.R. Tree, D.P. DeWitt, and S.A.H. Mackel, D.B., I.M. Kennedy, "Experimental and Numerical
Wahiduzzaman, "Design, Calibration and Error Investigation of Soot Deposition in Laminar
Analysis of Instrumentation for Heat Transfer Stagnation Point Boundary Layers," Twenty-third
Measurements in Internal Combustion Engines," in Symposium (International) on Combustion/The
Developments in Experimental Techniques in Heat Combustion Institute, pp. 1551-1557, 1990.
Transfer and Combustion, ASME, New York, pp.
67-82, 1987. Mohammad, I.S.A., "Simultaneous Pyrometer Measurements
Along Three Path Directions in an Open Chamber
Fukuda, M., D.R. Tree, D.E. Foster, and B.R. Suhre, "The Diesel," Ph.D. Thesis, University of Wisconsin-
Effect of Aromatic Content and Structure on Direct Madison, 1990.
Injection Diesel Particulates," SAE Paper No.
920110, 1992. Murray, R.G., "Performance and Emission Characteristics of
a Semi-Adiabatic Engine," ASME Technical Paper
Gafiney, J., R. Sapienza, T. Butcher, C. Krishna, W. Marlow, No. 80-DGP-44, American Society of Mechanical
and T. O'Hare, "Soot Reduction in Diesel Engines: A Engineers, New York, 1980.
Chemical Approach," Combustion Science and
Technology, v. 24, pp. 89-92, 1980. McDonald, J.R., "Construction and Testing Facility for Diesel
Engine Heat Transfer and Particulate Research," MS
Heywood, J.B., Internal Combustion Engine Fundamentals, Thesis, University of Wisconsin-Madison, 1984.
pp. 626-648, McGraw-Hill, New York, 1988.
Needham, J.R., D.M. Doyle, S.A. Faulkner, and H.D.
Holman, J.P., Heat Transfer, McGraw-Hill, New York, 1986. Freeman, "Technology for 1994," SAE Paper No.
891949, 1989.
Kerker, M., The Scattering of Light and Other
Electromagnetic Radiation, Academic Press, Park, C., and J.P. Appelton, "Shock Tube Measurements of
Orlando, 1969. Soot Oxidation Rates," Combustion and Flame, Vol.
1660
20, pp. 369-379, 1973

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms
Rosner, D.E., Transport Processes in Chemically Reacting
Flow Systems, Butterworths, Boston, 1986.

Schweimer, G.W., "Ion Probe in the Exhaust Manifold of


Diesel Engines," SAE Paper No. 860012, 1986.

Stevenson, R., "Morphology and Crystallography of Diesel


Particulate Emissions," Carbon, Vol. 20, pp. 359-
365, 1982.

Talbot, L., R.K. Cheng, R.W. Schefer, and D.R. Willis,


"Thermophoresis in a Heated Boundary Layer,"
Journal of Fluid Mechanics, v. 101, Part 4, pp. 737-
758, 1980.

Tree, D.R., Personal Communication, 1992.

Waldmann, L., K.H. Schmitt, "Thermophoresis and


Difiusiophoresis of aerosols," Ch. IV in Aerosol
Science, C.N. Davies (Ed.), Academic Press,
London, 1966.

Yan, J., "Analyses and In-Cylinder Measurements of Local


and Hemispherical Particulate Radiant Emissions
and Temperature in a Direct Injection Diesel
Engine," Ph.D. Thesis, University of Wisconsin-
Madison, 1988.

1661

This content downloaded from


109.207.199.235 on Mon, 27 Sep 2021 03:47:14 UTC
All use subject to https://about.jstor.org/terms

You might also like